1 2018-09-26 Uros Bizjak <ubizjak@gmail.com>
4 * config/i386/i386.h (NUM_MODES_FOR_MODE_SWITCHING): Update
5 for removed I387_MASK_PM entity.
7 2018-09-26 Jeff Law <law@redhat.com>
10 2018-09-26 Alexey Neyman <stilor@att.net>
12 * graphite.h: Include <isl/id.h> and <isl/space.h>; these
13 headers are no longer pulled in by <isl/val.h>.
15 2018-09-26 Richard Biener <rguenther@suse.de>
18 * dwarf2out.c (gen_lexical_block_die): Do not equate inline
19 or concrete instance DIE to the tree. Create abstract origin
20 attributes also for concrete instances.
22 2018-09-26 Alexey Neyman <stilor@att.net>
24 * graphite.h: Include <isl/id.h> and <isl/space.h>; these
25 headers are no longer pulled in by <isl/val.h>.
27 2018-09-26 Matthew Malcomson <matthew.malcomson@arm.com>
29 * config/arm/arm.c (arm_split_compare_and_swap, arm_split_atomic_op):
30 Use new helper functions.
31 * config/arm/sync.md (atomic_load<mode>, atomic_store<mode>):
32 Use new helper functions.
33 * config/arm/aarch-common-protos.h (aarch_mm_needs_acquire,
34 aarch_mm_needs_release): New declarations.
35 * config/arm/aarch-common.c (aarch_mm_needs_acquire,
36 aarch_mm_needs_release): New.
38 2018-09-26 Eric Botcazou <ebotcazou@adacore.com>
40 * config/arm/arm.c (arm_reorg): Skip Thumb reorg pass for thunks.
41 (arm32_output_mi_thunk): Deal with long calls.
43 2018-09-26 Richard Biener <rguenther@suse.de>
47 * tree-inline.c (expand_call_inline): When the location
48 of the call is UNKNOWN_LOCATION use DECL_SOURCE_LOCATION
49 or BUILTINS_LOCATION for the BLOCK_SOURCE_LOCATION of
50 the inserted BLOCK to make inlined_function_outer_scope_p
52 * dwarf2out.c (add_call_src_coords_attributes): Do not add
53 coords for reserved locations.
55 2018-09-25 Segher Boessenkool <segher@kernel.crashing.org>
57 * config/rs6000/rs6000.md (*movcc_internal1): Use set_attr_alternative.
58 (*call_indirect_nonlocal_sysv<mode>): Ditto.
59 (*call_value_indirect_nonlocal_sysv<mode>): Ditto.
60 (*sibcall_nonlocal_sysv<mode>): Ditto.
61 (*sibcall_value_nonlocal_sysv<mode>): Ditto.
63 (<bd>tf_<mode>): Ditto.
65 2018-09-25 Segher Boessenkool <segher@kernel.crashing.org>
67 * config/rs6000/altivec.md (*altivec_mov<mode>): Write the output
68 control string as a list of templates instead of as C code.
69 (*altivec_movti): Ditto.
70 * config/rs6000/darwin.md (movdf_low_di): Ditto.
72 2018-09-25 Jim Wilson <jimw@sifive.com>
74 * config/riscv/riscv.c (riscv_split_symbol): Mark auipc label as weak
75 when target symbol is weak.
77 2018-09-25 Bernd Edlinger <bernd.edlinger@hotmail.de>
80 * builtins.c (unterminated_array): Simplify.
81 * expr.c (string_constant): Handle SSA_NAME. Add more exceptions
82 where pointer arithmetic is safe.
84 2018-09-25 Segher Boessenkool <segher@kernel.crashing.org>
87 * config/rs6000/altivec.md (altivec_vspltb): Use
88 const_0_to_15_operand instead of u5bit_cint_operand.
89 (*altivec_vspltb_internal): Ditto.
90 (altivec_vspltb_direct): Ditto.
91 (altivec_vsplth): Use const_0_to_7_operand instead of
93 (*altivec_vsplth_internal): Ditto.
94 (altivec_vsplth_direct): Ditto.
95 (altivec_vspltw): Use const_0_to_3_operand instead of
97 (*altivec_vspltw_internal): Ditto.
98 (altivec_vspltw_direct): Ditto.
99 (altivec_vspltsf): Ditto.
100 (*altivec_vspltsf_internal): Ditto.
101 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Handle the
102 various splats with the proper size immediate. Reorder the various
103 cases by ascending size of immediate, and put all such together.
105 2018-09-25 Richard Biener <rguenther@suse.de>
108 * dwarf2out.c (add_AT_external_die_ref): Remove now redundant
110 (maybe_create_die_with_external_ref): Do not create
111 DW_TAG_imported_unit here.
112 (add_abstract_origin_attribute): Handle external BLOCK refs.
113 (dwarf2out_abstract_function): Simplify LTO case.
114 (dwarf2out_early_finish): Create DW_TAG_imported_unit explicitely
115 rather than using maybe_create_die_with_external_ref.
117 2018-09-25 Uros Bizjak <ubizjak@gmail.com>
120 * config/i386/i386.md (frndintxf2_mask_pm): Remove.
121 (frndintxf2_mask_pm_i387): Ditto.
122 (nearbyintxf2): Rewrite expander pattern to match rintxf2.
123 Enable for !flag_trapping_math.
124 (nearbyint<mode>2): Enable x87 modes for !flag_trapping_math.
125 Enable SSE modes for TARGET_SSE4_1 and expand them with round insn.
126 Change operand 1 predicate to nonimmediate_operand.
127 (attr "i387_cw"): Remove mask_pm.
128 * config/i386/i386.h (enum ix86_stack_slot): Remove SLOT_CW_MASK_PM.
129 (enum ix86_entity): Remove I387_MASK_PM.
130 * config/i386/i386.c (ix86_i387_mode_needed): Do not
132 (ix86_mode_needed): Ditto.
133 (ix86_mode_after): Ditto.
134 (ix86_mode_entry): Ditto.
135 (ix86_mode_exit): Ditto.
136 (emit_i387_cw_initialization): Do not handle I387_CW_MASK_PM.
138 2018-09-25 Jakub Jelinek <jakub@redhat.com>
140 * vr-values.c (vr_values::vr_values): Initialize to_remove_edges and
141 to_update_switch_stmts to vNULL instead of calling create on them
144 2018-09-25 Richard Biener <rguenther@suse.de>
146 PR tree-optimization/87402
147 * tree-ssa-sccvn.c (SSA_VISITED): Remove unused function.
148 (visit_phi): Re-instantiate handling of supposed to be VARYING
149 but non-VARYING backedge value.
151 2018-09-25 Richard Biener <rguenther@suse.de>
154 * dwarf2out.c (struct sym_off_pair): New.
155 (external_die_map): New global.
156 (lookup_decl_die): When in LTO create DIEs lazily from the
158 (lookup_block_die): New function, create DIEs lazily in LTO.
159 (equate_block_to_die): New function.
160 (dwarf2out_die_ref_for_decl): During WPA get the association
161 from the external DIE map.
162 (dwarf2out_register_external_die): Record mapping into the
164 (maybe_create_die_with_external_ref): New function split out from
165 DIE generation part of old dwarf2out_register_external_die.
166 (add_abstract_origin_attribute): Do not return the DIE. When
167 in LTO reference externals directly.
168 (dwarf2out_abstract_function): When in LTO ignore calls for
169 decls with external DIEs (already present abstract instances).
170 (gen_call_site_die): Adjust.
171 (add_high_low_attributes): Likewise.
172 (gen_lexical_block_die): Likewise.
173 (gen_inlined_subroutine_die): Likewie.
174 (gen_block_die): Likewise.
175 (dwarf2out_inline_entry): Likewise.
176 (dwarf2out_early_finish): In LTRANS phase create DW_TAG_imported_unit
179 2018-09-25 Martin Liska <mliska@suse.cz>
181 * ipa-fnsummary.c (estimate_node_size_and_time): Scale by two
182 integers and not by a float value.
184 2018-09-25 Martin Liska <mliska@suse.cz>
187 * dbgcnt.c (dbg_cnt_process_single_pair): Return false
189 * dumpfile.c (dump_enable_all): Remove extra parenthesis.
190 * gcov-tool.c: Declare the function with ATTRIBUTE_NORETURN.
191 * godump.c (go_format_type): Remove extra parenthesis.
193 2018-09-25 Martin Liska <mliska@suse.cz>
195 * alias.c (set_dest_equal_p): Remove unused function.
196 * config/i386/i386.c (def_builtin_pure2): Likewise.
197 * diagnostic-show-locus.c (class layout): Remove
199 (layout::layout): Likewise here.
200 * dump-context.h (class temp_dump_context): Likewise.
201 * dwarf2out.c (add_AT_fde_ref): Remove unused function.
202 (add_AT_loclistsptr): Likewise.
203 (add_AT_offset): Likewise.
204 (get_AT_hi_pc): Likewise.
205 (is_comdat_die): Likewise.
206 (type_is_enum): Likewise.
208 (add_AT_vms_delta): Likewise.
209 (is_class_die): Likewise.
210 * edit-context.c (class line_event): Remove unused field.
211 * graphite-sese-to-poly.c (tree_int_to_gmp): Remove
213 * ipa-cp.c (ipa_get_vr_lat): Likewise.
214 * lra-constraints.c (ok_for_index_p_nonstrict): Likewise.
215 (ok_for_base_p_nonstrict): Likewise.
216 * tree-chrec.c (is_not_constant_evolution): Likewise.
217 (chrec_fold_poly_cst): Likewise.
218 * tree-if-conv.c (has_pred_critical_p): Likewise.
219 * tree-ssa-coalesce.c (print_exprs): Likewise.
220 * tree-ssa-pre.c (bitmap_set_contains_expr): Likewise.
221 * tree-ssa-uninit.c (is_and_or_or_p): Likewise.
222 * tree-vrp.c (value_ranges_intersect_p): Likewise.
223 (value_range_nonnegative_p): Likewise.
225 2018-09-25 Martin Liska <mliska@suse.cz>
227 * config/powerpcspe/powerpcspe.c (rs6000_output_function_epilogue):
228 Do not handle "GNU Pascal".
229 * config/rs6000/rs6000.c (rs6000_output_function_epilogue):
231 * config/sparc/sparc.c (sparc_pass_by_reference): Remove Pascal
232 from documentation. Likewise.
233 * dbxout.c (dbxout_range_type): Likewise.
234 * doc/cpp.texi: Likewise.
235 * doc/extend.texi: Likewise.
236 * doc/frontends.texi: Likewise.
237 * doc/invoke.texi: Remove Pascal entry.
238 * tree.def (CLEANUP_POINT_EXPR): Likewise.
239 * doc/rtl.texi (MODE_FUNCTION): Remove not used entry.
241 2018-09-25 Martin Liska <mliska@suse.cz>
244 * doc/invoke.texi: Document all parameters and remove default
247 2018-09-25 Ilya Leoshkevich <iii@linux.ibm.com>
250 * rtl.c (rtx_code_size): Take into account that EQ_ATTR_ALT
251 contains HOST_WIDE_INTs when computing its size.
253 2018-09-24 Jim Wilson <jimw@sifive.com>
256 * config/riscv/riscv.h (STACK_BOUNDARY): Test riscv_abi == ABI_ILP32E
258 (ABI_STACK_BOUNDARY, MAX_ARGS_IN_REGISTERS): Likewise.
260 2018-09-24 Andrew Pinski <apinski@marvell.com>
262 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Don't
263 access prev before checking it for NULLness in the
264 AARCH64_FUSE_CMP_BRANCH case.
266 2018-09-24 H.J. Lu <hongjiu.lu@intel.com>
269 * config/i386/i386.c (rest_of_insert_endbranch): Set
270 endbr_queued_at_entrance to true and don't insert ENDBR if
271 x86_function_profiler will be called.
272 (x86_function_profiler): Insert ENDBR if endbr_queued_at_entrance
274 * config/i386/i386.h (machine_function): Add
275 endbr_queued_at_entrance.
277 2018-09-24 Ilya Leoshkevich <iii@linux.ibm.com>
279 * genattrtab.c (mk_attr_alt): Use alternative_mask.
280 (attr_rtx_1): Adjust caching to match the new EQ_ATTR_ALT field
282 (check_attr_test): Use alternative_mask.
283 (get_attr_value): Likewise.
284 (compute_alternative_mask): Use alternative_mask and XWINT.
285 (make_alternative_compare): Use alternative_mask.
286 (attr_alt_subset_p): Use XWINT.
287 (attr_alt_subset_of_compl_p): Likewise.
288 (attr_alt_intersection): Use alternative_mask and XWINT.
289 (attr_alt_union): Likewise.
290 (attr_alt_complement): Use HOST_WIDE_INT and XWINT.
291 (mk_attr_alt): Use alternative_mask and HOST_WIDE_INT.
292 (simplify_test_exp): Use alternative_mask and XWINT.
293 (write_test_expr): Use alternative_mask and XWINT, adjust bit
294 number calculation to support 64 bits. Generate code that
296 (main): Use alternative_mask.
297 * rtl.def (EQ_ATTR_ALT): Change field types from ii to ww.
299 2018-09-24 Ilya Leoshkevich <iii@linux.ibm.com>
302 * config/s390/s390.c (s390_emit_epilogue): Do not use PARALLEL
303 RETURN+USE when returning via %r14.
305 2018-09-24 Martin Liska <mliska@suse.cz>
307 * gcov.c (output_lines): Print colorization legend
308 for both flag_use_colors and flag_use_hotness_colors.
311 2018-09-24 Martin Liska <mliska@suse.cz>
313 * coverage.c (get_coverage_counts): Use warning_at
314 with current_function_decl location. Use %qD in warning
317 2018-09-24 Martin Liska <mliska@suse.cz>
319 * memory-block.h (memory_block_pool::release): Annotate with
320 valgrind that the memory is not accessible.
322 2018-09-24 Martin Liska <mliska@suse.cz>
325 * asan.c: Make asan_handled_variables extern.
327 * cfgexpand.c (expand_stack_vars): Make sure
328 a representative is unpoison if another
329 variable in the partition is handled by
330 use-after-scope sanitization.
332 2018-09-24 Richard Biener <rguenther@suse.de>
334 PR tree-optimization/63155
335 * tree-ssa-propagate.c (add_ssa_edge): Avoid adding PHIs to
336 the worklist when the edge of the respective argument isn't
339 2018-09-23 Uros Bizjak <ubizjak@gmail.com>
341 * config/i386/i386.h (enum reg_class): Rename MASK_REGS to
342 ALL_MASK_REGS and MASK_EVEX_REGS to MASK_REGS.
343 (MASK_CLASS_P): Update for rename.
344 (MAYBE_MASK_CLASS_P): Ditto.
345 (REG_CLASS_NAMES): Update.
346 (REG_CLASS_CONTENT): Update.
347 * config/i386/i386.c (regclass_map): Update for MASK_REG
348 and ALL_MASK_REGS rename.
349 * config/i386/constraints.md (Yk): Update for rename.
352 2018-09-23 Uros Bizjak <ubizjak@gmail.com>
354 * config/i386/i386.h (enum reg_class): Remove
355 EVEX_SSE_REGS and MOD4_SSE_REGS.
356 (REG_CLASS_NAMES): Update.
357 (REG_CLASS_CONTENT): Update.
358 * config/i386/i386.c (regclass_map): Declare AVX-512 SSE
359 registers as ALL_SSE_REGS.
360 (ix86_additional_allocno_class_p): Remove.
361 (TARGET_ADDITIONAL_ALLOCNO_CLASS_P): Remove.
362 (ix86_register_priority): Lower priority of EVEX SSE registers.
363 Use IN_RANGE macro where appropriate.
364 (ix86_hard_regno_mode_ok): Merge AVX-5124FMAPS and
365 AVX-5124VNNIW checks.
366 (ix86_modes_tieable_p): Tie 512-bit SSE modes.
367 * config/i386/sse.md (avx5124fmaddps_4fmaddps)
368 (avx5124fmaddps_4fmaddps_mask, avx5124fmaddps_4fmaddps_maskz)
369 (avx5124fmaddps_4fmaddss, avx5124fmaddps_4fmaddss_mask)
370 (avx5124fmaddps_4fmaddss_maskz, avx5124fmaddps_4fnmaddps)
371 (avx5124fmaddps_4fnmaddps_mask, avx5124fmaddps_4fnmaddps_maskz)
372 (avx5124fmaddps_4fnmaddss, avx5124fmaddps_4fnmaddss_mask)
373 (avx5124fmaddps_4fnmaddss_maskz, avx5124vnniw_vp4dpwssd)
374 (avx5124vnniw_vp4dpwssd_mask, avx5124vnniw_vp4dpwssd_maskz)
375 (avx5124vnniw_vp4dpwssds, avx5124vnniw_vp4dpwssds_mask)
376 (avx5124vnniw_vp4dpwssds_maskz): Use "v" instead of "Yh" constraint.
377 * config/i386/constraints.md (Yh): Remove.
379 2018-09-23 Uros Bizjak <ubizjak@gmail.com>
381 * config/i386/i386.c (regclass_map): Declare integer REX registers
384 2018-09-23 Gerald Pfeifer <gerald@pfeifer.com>
386 * doc/service.texi (Service): Switch the fsf.org link to https.
388 2018-09-22 Chung-Ju Wu <jasonwucj@gmail.com>
391 * config/nds32/nds32.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
392 Define to speculation_safe_value_not_needed.
394 2018-09-21 Florian Weimer <fweimer@redhat.com>
397 * doc/extend.texi (Common Function Attributes): Mention that
398 noreturn suppresses tail call optimization.
400 2018-09-21 Jeff Law <law@redhat.com>
402 * gimple-ssa-evrp.c (evrp_dom_walker::cleanup): Call
403 vr_values::cleanup_edges_and_switches.
404 * tree-vrp.c (to_remove_edges, to_update_switch_stmts): Moved into
406 (identify_jump_threads): Remove EDGE_IGNORE handling.
407 (execute_vrp): Move handling of to_remove_edges and
408 to_update_switch_stmts into vr_values class member functions.
409 * tree-vrp.h (switch_update, to_remove_edges): Remove declarations.
410 (to_update_switch_stmts): Likewise.
411 * vr-values.c: Include cfghooks.h.
412 (vr_values::vr_values): Initialize to_remove_edges and
413 to_update_switch_stmts.
414 (vr_values::~vr_values): Verify to_remove_edges and
415 to_update_switch_stmts are empty.
416 (vr_values::simplify_switch_using_ranges): Set EDGE_IGNORE as needed.
417 (vr_values::cleanup_edges_and_switches): New member function.
418 * vr-values.h (vr_values): Add cleanup_edges_and_switches member
419 function. Add new data members.
421 2018-09-21 David Malcolm <dmalcolm@redhat.com>
423 PR tree-optimization/87309
424 * dumpfile.c (dump_context::begin_scope): Filter the dump_loc
425 calls with pflags and alt_flags.
426 (selftest::test_capture_of_dump_calls): Add test of interaction of
427 MSG_OPTIMIZED_LOCATIONS with AUTO_DUMP_SCOPE.
429 2018-09-21 Olivier Hainque <hainque@adacore.com>
431 * config.gcc: Factorize and comment inclusion of vxworks-dummy.h.
433 2018-09-21 Olivier Hainque <hainque@adacore.com>
435 * config/vxworks.h (CLEAR_INSN_CACHE): #define to 1.
437 2018-09-21 Olivier Hainque <hainque@adacore.com>
439 * config/vxworks.h (VXWORKS_LIBGCC_SPEC): Remove -lc_internal.
440 Merge block comment with the one ahead of VXWORKS_LIBS_RTP. Then:
441 (VXWORKS_LIBS_RTP): Minor reordering.
443 2018-09-21 Olivier Hainque <hainque@adacore.com>
445 * config/vxworks.h (STARTFILE_PREFIX_SPEC): Define.
446 (VXWORKS_LIBS_DIR_RTP): Remove definition and use.
448 2018-09-21 Olivier Hainque <hainque@adacore.com>
450 * config/vxworks.h (SIZE_TYPE): Account for TARGET_VXWORKS64.
451 (PTRDIFF_TYPE): Likewise.
453 2018-09-21 Olivier Hainque <hainque@adacore.com>
455 * config.gcc: Enforce def of TARGET_VXWORKS64 to 1 from
456 triplet, similar to support for VxWorks7.
457 * config/vxworks-dummy.h: Provide a default definition
458 of TARGET_VXWORKS64 to 0.
460 2018-09-21 Olivier Hainque <hainque@adacore.com>
462 * config/vxworks.h (TARGET_VXWORKS7): Move default definition ...
463 * config/vxworks-dummy.h: here.
465 2018-09-21 Olivier Hainque <hainque@adacore.com>
467 * config.gcc: Prepend vxworks-dummy.h to tm_file for powerpc*
469 2018-09-21 Shaokun Zhang <zhangshaokun@hisilicon.com>
470 Bo Zhou <zbo.zhou@hisilicon.com>
472 * config/aarch64/aarch64-cores.def (tsv110): New CPU.
473 * config/aarch64/aarch64-tune.md: Regenerated.
474 * doc/invoke.texi (AArch64 Options/-mtune): Add "tsv110".
475 * config/aarch64/aarch64.c (tsv110_tunings): New tuning table.
476 * config/aarch64/aarch64-cost-tables.h: Add "tsv110" extra costs.
478 2018-09-21 Andrew Stubbs <ams@codesourcery.com>
479 Julian Brown <julian@codesourcery.com>
481 * builtins.c (get_builtin_sync_mem): Handle address spaces.
483 2018-09-21 Eric Botcazou <ebotcazou@adacore.com>
485 * config/rs6000/rs6000.c (rs6000_function_ok_for_sibcall): Return false
486 if the call takes a static chain.
488 2018-09-21 Martin Liska <mliska@suse.cz>
490 * auto-profile.c (autofdo_source_profile::read): Do not
492 (read_profile): Do not add working sets.
493 (read_autofdo_file): Remove sum_all.
494 (afdo_callsite_hot_enough_for_early_inline): Remove const
496 * coverage.c (struct counts_entry): Remove gcov_summary.
497 (read_counts_file): Read new GCOV_TAG_OBJECT_SUMMARY,
498 do not support GCOV_TAG_PROGRAM_SUMMARY.
499 (get_coverage_counts): Remove summary and expected
501 * coverage.h (get_coverage_counts): Likewise.
502 * doc/gcov-dump.texi: Remove -w option.
503 * gcov-dump.c (dump_working_sets): Remove.
504 (main): Do not support '-w' option.
505 (print_usage): Likewise.
506 (tag_summary): Likewise.
507 * gcov-io.c (gcov_write_summary): Do not dump
509 (gcov_read_summary): Likewise.
510 (gcov_histo_index): Remove.
511 (gcov_histogram_merge): Likewise.
512 (compute_working_sets): Likewise.
513 * gcov-io.h (GCOV_TAG_OBJECT_SUMMARY): Mark
515 (GCOV_TAG_PROGRAM_SUMMARY): Mark it obsolete.
516 (GCOV_TAG_SUMMARY_LENGTH): Adjust.
517 (GCOV_HISTOGRAM_SIZE): Remove.
518 (GCOV_HISTOGRAM_BITVECTOR_SIZE): Likewise.
519 (struct gcov_summary): Simplify rapidly just
520 to runs and sum_max fields.
521 (gcov_histo_index): Remove.
522 (NUM_GCOV_WORKING_SETS): Likewise.
523 (compute_working_sets): Likewise.
524 * gcov-tool.c (print_overlap_usage_message): Remove
526 * gcov.c (read_count_file): Read GCOV_TAG_OBJECT_SUMMARY.
527 (output_lines): Remove program related line.
528 * ipa-profile.c (ipa_profile): Do not consider GCOV histogram.
529 * lto-cgraph.c (output_profile_summary): Do not stream GCOV
531 (input_profile_summary): Do not read it.
532 (merge_profile_summaries): And do not merge it.
533 (input_symtab): Do not call removed function.
534 * modulo-sched.c (sms_schedule): Do not print sum_max.
535 * params.def (HOT_BB_COUNT_FRACTION): Reincarnate param that was
536 removed when histogram method was invented.
537 (HOT_BB_COUNT_WS_PERMILLE): Mention that it's used only in LTO
539 * postreload-gcse.c (eliminate_partially_redundant_load): Fix
541 * predict.c (get_hot_bb_threshold): Use HOT_BB_COUNT_FRACTION
542 and dump selected value.
543 * profile.c (add_working_set): Remove.
544 (get_working_sets): Likewise.
545 (find_working_set): Likewise.
546 (get_exec_counts): Do not work with working sets.
547 (read_profile_edge_counts): Do not inform as sum_max is removed.
548 (compute_branch_probabilities): Likewise.
549 (compute_value_histograms): Remove argument for call of
551 * profile.h: Do not make gcov_summary const.
553 2018-09-21 Monk Chiang <sh.chiang04@gmail.com>
555 * config.gcc (nds32*-*-*): Set TARGET_DEFAULT_TLSDESC_TRAMPOLINE=0.
557 2018-09-21 Eric Botcazou <ebotcazou@adacore.com>
559 PR tree-optimization/86990
560 * gimple-ssa-store-merging.c (imm_store_chain_info:coalesce_immediate):
561 Check that the entire merged store group is made of constants only for
564 2018-09-20 Allan Sandfeld Jensen <allan.jensen@qt.io>
566 * gcc.c (LINK_COMMAND_SPEC): Handle -r like -nostdlib.
567 (VTABLE_VERIFICATION_SPEC): Likewise.
568 (SANITIZER_EARLY_SPEC): Likewise.
569 (SANITIZER_SPEC): Likewise.
570 * config/darwin.h (LINK_COMMAND_SPEC): Likewise.
571 * doc/invoke.texi (Link Options): Document -r.
573 2018-09-20 Richard Biener <rguenther@suse.de>
576 * gimplify.c (gimplify_expr): Retain alignment of
577 addressable lvalue in dereference.
579 2018-09-20 Alexandre Oliva <aoliva@redhat.com>
582 * configure.ac: Check for .loc is_stmt support.
583 * configure, config.in: Rebuilt.
584 * dwarf2out.c (dwarf2out_source_line): Skip is_stmt
587 2018-09-20 Segher Boessenkool <segher@kernel.crashing.org>
589 * config/rs6000/rs6000.opt (misel=no, misel=yes): Delete.
590 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -misel=yes and
593 2018-09-20 Segher Boessenkool <segher@kernel.crashing.org>
595 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete
597 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete
600 2018-09-20 Marek Polacek <polacek@redhat.com>
602 * doc/invoke.texi: Add -Wno-init-list-lifetime to C++ Language Options.
604 2018-09-20 Richard Sandiford <richard.sandiford@arm.com>
606 PR tree-optimization/87288
607 * tree-vect-loop.c (vect_analyze_loop_2): Take PEELING_FOR_GAPS
608 into account when determining PEELING_FOR_NITERS.
610 2018-09-20 Richard Sandiford <richard.sandiford@arm.com>
612 PR tree-optimization/86877
613 * tree-vect-loop.c (vect_analyze_loop_2): Call
614 vect_verify_datarefs_alignment.
616 2018-09-19 Marek Polacek <polacek@redhat.com>
618 * doc/invoke.texi: Document -Wclass-conversion.
620 2018-09-19 John David Anglin <danglin@gcc.gnu.org>
622 * config/pa/pa.c (pa_adjust_priority): Delete.
623 (TARGET_SCHED_ADJUST_PRIORITY): Delete define.
625 * config/pa/pa.md (atomic_storeqi): Restore deleted expander.
626 (atomic_storehi): Likewise.
627 (atomic_storesi): Likewise.
628 (atomic_loaddi): Restore compare and swap exchange loop code.
630 2018-09-19 Segher Boessenkool <segher@kernel.crashing.org>
632 PR rtl-optimization/86902
633 * combine.c (try_combine): When changing the CC mode used, don't change
634 an unrelated mode in other_insn to that new CC mode.
636 2018-09-19 David Malcolm <dmalcolm@redhat.com>
638 * tree-data-ref.c (runtime_alias_check_p): Use formatted printing
639 with %T in place of calls to dump_generic_expr.
640 (prune_runtime_alias_test_list): Likewise.
641 (create_runtime_alias_checks): Likewise.
642 * tree-vect-data-refs.c (vect_check_nonzero_value): Likewise.
643 (vect_analyze_data_ref_dependence): Likewise.
644 (vect_slp_analyze_data_ref_dependence): Likewise.
645 (vect_record_base_alignment): Likewise. Use %G in place of call
647 (vect_compute_data_ref_alignment): Likewise.
648 (verify_data_ref_alignment): Likewise.
649 (vect_find_same_alignment_drs): Likewise.
650 (vect_analyze_group_access_1): Likewise.
651 (vect_analyze_data_ref_accesses): Likewise.
652 (dependence_distance_ge_vf): Likewise.
653 (dump_lower_bound): Likewise.
654 (vect_prune_runtime_alias_test_list): Likewise.
655 (vect_find_stmt_data_reference): Likewise.
656 (vect_analyze_data_refs): Likewise.
657 (vect_create_addr_base_for_vector_ref): Likewise.
658 (vect_create_data_ref_ptr): Likewise.
659 * tree-vect-loop-manip.c (vect_set_loop_condition): Likewise.
660 (vect_can_advance_ivs_p): Likewise.
661 (vect_update_ivs_after_vectorizer): Likewise.
662 (vect_gen_prolog_loop_niters): Likewise.
663 (vect_prepare_for_masked_peels): Likewise.
664 * tree-vect-loop.c (vect_determine_vf_for_stmt): Likewise.
665 (vect_determine_vectorization_factor): Likewise.
666 (vect_is_simple_iv_evolution): Likewise.
667 (vect_analyze_scalar_cycles_1): Likewise.
668 (vect_analyze_loop_operations): Likewise.
669 (report_vect_op): Likewise.
670 (vect_is_slp_reduction): Likewise.
671 (check_reduction_path): Likewise.
672 (vect_is_simple_reduction): Likewise.
673 (vect_create_epilog_for_reduction): Likewise.
674 (vect_finalize_reduction:): Likewise.
675 (vectorizable_induction): Likewise.
676 (vect_transform_loop_stmt): Likewise.
677 (vect_transform_loop): Likewise.
678 (optimize_mask_stores): Likewise.
679 * tree-vect-patterns.c (vect_pattern_detected): Likewise.
680 (vect_split_statement): Likewise.
681 (vect_recog_over_widening_pattern): Likewise.
682 (vect_recog_average_pattern): Likewise.
683 (vect_determine_min_output_precision_1): Likewise.
684 (vect_determine_precisions_from_range): Likewise.
685 (vect_determine_precisions_from_users): Likewise.
686 (vect_mark_pattern_stmts): Likewise.
687 (vect_pattern_recog_1): Likewise.
688 * tree-vect-slp.c (vect_get_and_check_slp_defs): Likewise.
689 (vect_record_max_nunits): Likewise.
690 (vect_build_slp_tree_1): Likewise.
691 (vect_build_slp_tree_2): Likewise.
692 (vect_print_slp_tree): Likewise.
693 (vect_analyze_slp_instance): Likewise.
694 (vect_detect_hybrid_slp_stmts): Likewise.
695 (vect_detect_hybrid_slp_1): Likewise.
696 (vect_slp_analyze_operations): Likewise.
697 (vect_slp_analyze_bb_1): Likewise.
698 (vect_transform_slp_perm_load): Likewise.
699 (vect_schedule_slp_instance): Likewise.
700 * tree-vect-stmts.c (vect_mark_relevant): Likewise.
701 (vect_mark_stmts_to_be_vectorized): Likewise.
702 (vect_init_vector_1): Likewise.
703 (vect_get_vec_def_for_operand): Likewise.
704 (vect_finish_stmt_generation_1): Likewise.
705 (vect_check_load_store_mask): Likewise.
706 (vectorizable_call): Likewise.
707 (vectorizable_conversion): Likewise.
708 (vectorizable_operation): Likewise.
709 (vectorizable_load): Likewise.
710 (vect_analyze_stmt): Likewise.
711 (vect_is_simple_use): Likewise.
712 (vect_get_vector_types_for_stmt): Likewise.
713 (vect_get_mask_type_for_stmt): Likewise.
714 * tree-vectorizer.c (increase_alignment): Likewise.
716 2018-09-19 Andrew Stubbs <ams@codesourcery.com>
718 * doc/rtl.texi: Adjust vec_select description.
719 * simplify-rtx.c (simplify_binary_operation_1): Allow VEC_SELECT to use
720 non-constant selectors.
722 2018-09-19 Matthew Malcomson <matthew.malcomson@arm.com>
724 * config/aarch64/aarch64-protos.h
725 (aarch64_offset_9bit_signed_unscaled_p): New declaration.
726 * config/aarch64/aarch64.md (arches): New "rcpc8_4" attribute value.
727 (arch_enabled): Add check for "rcpc8_4" attribute value of "arch".
728 * config/aarch64/aarch64.h (AARCH64_FL_RCPC8_4): New bitfield.
729 (AARCH64_FL_FOR_ARCH8_4): Include AARCH64_FL_RCPC8_4.
730 (AARCH64_FL_PROFILE): Move index so flags are ordered.
731 (AARCH64_ISA_RCPC8_4): New flag.
732 * config/aarch64/aarch64.c (offset_9bit_signed_unscaled_p): Renamed
733 to aarch64_offset_9bit_signed_unscaled_p.
734 * config/aarch64/atomics.md (atomic_store<mode>): Allow offset
736 * config/aarch64/constraints.md (Ust): New constraint.
737 * config/aarch64/predicates.md.
738 (aarch64_9bit_offset_memory_operand): New predicate.
739 (aarch64_rcpc_memory_operand): New predicate.
741 2018-09-19 Eric Botcazou <ebotcazou@adacore.com>
743 PR rtl-optimization/87361
744 * rtlanal.c (nonzero_bits1): Revert accidental change.
746 2018-09-19 Richard Biener <rguenther@suse.de>
748 PR tree-optimization/87349
749 PR tree-optimization/87342
750 * tree-ssa-sccvn.c (do_rpo_vn): Iterate max_rpo computation.
752 2018-09-18 Marek Polacek <polacek@redhat.com>
754 P1064R0 - Allowing Virtual Function Calls in Constant Expressions
755 * gimple-fold.c (gimple_get_virt_method_for_vtable): Adjust assert.
757 2018-09-18 Segher Boessenkool <segher@kernel.crashing.org>
759 * config/rs6000/rs6000.md: Remove old "Cygnus sibcall" comment.
761 2018-09-18 Segher Boessenkool <segher@kernel.crashing.org>
763 PR rtl-optimization/86882
764 * rtlanal.c (reg_overlap_mentioned_p): Handle CLOBBER.
766 2018-09-18 Uros Bizjak <ubizjak@gmail.com>
768 * config/i386/i386.md (*<code>extend<mode>xf2): Macroize insn from
769 *<code>extendsfxf2 and *<code>extenddfxf2 using MODEF mode iterator.
771 2018-09-18 Jonathan Wakely <jwakely@redhat.com>
774 * doc/invoke.texi (Link Options): Fix formatting and grammar.
776 2018-09-18 Richard Biener <rguenther@suse.de>
779 * tree-ssa-coalesce.c (tree_int_map_hasher): Remove.
780 (compute_samebase_partition_bases): Likewise.
781 (coalesce_ssa_name): Always use compute_optimized_partition_bases.
782 (gimple_can_coalesce_p): Simplify.
784 2018-09-18 Hans-Peter Nilsson <hp@bitrange.com>
786 Handle a library implementation of ffs calling __builtin_ffs.
787 * config/mmix/mmix.c (TARGET_INIT_LIBFUNCS): Override with...
788 (mmix_init_libfuncs): New function: make __builtin_ffs expand
791 2018-09-17 David Malcolm <dmalcolm@redhat.com>
793 * diagnostic-show-locus.c (class layout_range): Add field
795 (layout_range::layout_range): Add "original_idx" param and use it
796 to initialize new field.
797 (make_range): Use 0 for original_idx.
798 (layout::layout): Pass in index to calls to
799 maybe_add_location_range.
800 (layout::maybe_add_location_range): Add param "original_idx" and
801 pass it on to layout_range.
802 (layout::print_any_labels): Pass on range->m_original_idx to
804 (gcc_rich_location::add_location_if_nearby): Use 0 for
806 * gcc-rich-location.h (text_range_label::get_text): Update for new
808 (range_label_for_type_mismatch::get_text): Likewise.
810 2018-09-17 Uros Bizjak <ubizjak@gmail.com>
812 * config/i386/i386.c (ix86_emit_i387_log1p): Emit fldln2 earlier.
814 2018-09-17 David Malcolm <dmalcolm@redhat.com>
816 * gimple-ssa-sprintf.c (fmtwarn): Update for introduction of
817 format_string_diagnostic_t.
818 (fmtwarn_n): Likewise.
819 * substring-locations.c
820 (format_string_diagnostic_t::format_string_diagnostic_t) New ctor.
821 (format_warning_n_va): Convert to...
822 (format_string_diagnostic_t::emit_warning_n_va): ...this.
823 (format_warning_va): Convert to...
824 (format_string_diagnostic_t::emit_warning_va): ...this.
825 (format_warning_at_substring): Convert to...
826 (format_string_diagnostic_t::emit_warning): ...this.
827 (format_warning_at_substring_n): Convert to...
828 (format_string_diagnostic_t::emit_warning_n): ...this.
829 * substring-locations.h (class format_string_diagnostic_t): New
831 (format_warning_va): Convert to
832 format_string_diagnostic_t::emit_warning_va.
833 (format_warning_n_va): Convert to
834 format_string_diagnostic_t::emit_warning_n_va.
835 (format_warning_at_substring): Convert to
836 format_string_diagnostic_t::emit_warning.
837 (format_warning_at_substring_n): Convert to
838 format_string_diagnostic_t::emit_warning_n.
840 2018-09-17 Cesar Philippidis <cesar@codesourcery.com>
841 Bernd Schmidt <bernds_cb1@t-online.de>
843 * config/nvptx/nvptx.md (atomic_fetch_<logic><mode>): Enable with
846 2018-09-17 Uros Bizjak <ubizjak@gmail.com>
848 * config/i386/i386.md (truncxf<mode>2_i387_noop_unspec): Change
849 operand 0 predicate to nonimmediate operand.
850 (rint<mode>2_frndint): Remove insn pattern.
851 (rint<mode>2): Change operand 1 predicate to general_operand.
852 Extend operand 1 to XFmode and generate rintxf2 insn.
853 (frndintxf2_<rounding>): Rename from frndint<mode>2_<rounding>.
854 Do not use X87MODEF mode macro.
855 (frndintxf2_<rounding>_i387): Rename from
856 frndint<mode>2_<rounding>_i387. Do not use X87MODEF mode macro.
857 (<rounding_insn><mode>2): For non-SSE modes, extend operand 1
858 to XFmode and generate significandxf3 insn.
860 2018-09-17 Richard Biener <rguenther@suse.de>
862 PR tree-optimization/87328
863 * tree-ssa-sccvn.c (process_bb): Remove assertion about not
864 visiting unexecutable backedges when not iterating.
865 (do_rpo_vn): Mark all edges not executable even when not
868 2018-09-17 Martin Jambor <mjambor@suse.cz>
871 * doc/invoke.texi (Warning Options): Likewise.
873 2018-09-17 Richard Biener <rguenther@suse.de>
875 PR tree-optimization/87301
876 * tree-ssa-sccvn.c (eliminate_dom_walker::eliminate_cleanup): Properly
877 clean EH info from leftover copy assignments.
879 2018-09-17 Martin Liska <mliska@suse.cz>
881 PR gcov-profile/85871
882 * gcov.c (output_intermediate_file): Fix out of bounds
885 2018-09-17 Vineet Gupta <vgupta@synopsys.com>
887 * config/arc/arc.c: Object attributes for core4 not reflected
889 * config/arc/arc.h: Don't restrict DBNZ to core3 (core4 includes
892 2018-09-17 Alexey Brodkin <abrodkin@synopsys.com>
894 * config/arc/linux.h (LINK_EH_SPEC): Add missing space.
896 2018-09-17 Martin Liska <mliska@suse.cz>
898 * doc/gcov.texi: Document new option --use-hotness-colors.
899 * gcov.c (struct source_info): Declare new field.
900 (source_info::source_info): Set default for maximum_count.
901 (print_usage): Add new -q option.
902 (process_args): Process it.
903 (accumulate_line_info): Save src->maximum_count.
904 (output_line_beginning): Make color line number if
905 flag_use_hotness_colors is set.
906 (output_line_details): Pass default argument value.
907 (output_lines): Pass src->maximum_count.
909 2018-09-17 Martin Liska <mliska@suse.cz>
911 * common/config/i386/i386-common.c (ix86_get_valid_option_values):
912 Use processor_names table.
913 * config/i386/i386.c (ix86_default_align): Use
914 processor_cost_table for alignment values.
915 (ix86_option_override_internal): Use processor_names.
916 (ix86_function_specific_print): Likewise.
917 * config/i386/i386.h (struct processor_costs):
918 Add alignment values.
919 (struct ptt): Remove and replace with const char *.
920 * config/i386/x86-tune-costs.h (struct processor_costs):
921 Declare default alignments for all costs.
923 2018-09-17 Aldy Hernandez <aldyh@redhat.com>
925 * tree-vrp.c (extract_range_from_unary_expr): Do not special case
926 symbolics or VR_VARYING ranges for ABS_EXPR.
927 * wide-int-range.cc (wide_int_range_abs): Return positive numbers
928 when range will wrap.
930 2018-09-15 Eric Botcazou <ebotcazou@adacore.com>
933 * cfgexpand.c (expand_gimple_basic_block): Be prepared for a BARRIER
934 before and after a JUMP_TABLE_DATA.
936 2018-09-14 John David Anglin <danglin@gcc.gnu.org>
939 * dojump.c (do_compare_and_jump): Canonicalize function pointers
940 when one operand is a function pointer. Use POINTER_TYPE_P and
941 FUNC_OR_METHOD_TYPE_P.
942 * expr.c (do_store_flag): Use POINTER_TYPE_P and FUNC_OR_METHOD_TYPE_P.
943 * fold-const.c (build_range_check): Likewise.
944 * match.pd (simple_comparison): Likewise.
946 2018-09-14 David Malcolm <dmalcolm@redhat.com>
949 * spellcheck.c (get_edit_distance_cutoff): New function.
950 (selftest::test_edit_distance_unit_test_oneway): Rename to...
951 (selftest::test_get_edit_distance_one_way): ...this.
952 (selftest::test_get_edit_distance_unit): Rename to...
953 (selftest::test_get_edit_distance_both_ways): ...this.
954 (selftest::test_edit_distances): Move tests to this new function,
955 and test some more pairs of strings. Update for above renaming.
956 (selftest::get_old_cutoff): New function.
957 (selftest::test_get_edit_distance_cutoff): New function.
958 (selftest::assert_suggested_for): New function.
959 (ASSERT_SUGGESTED_FOR): New macro.
960 (selftest::assert_not_suggested_for): New function.
961 (ASSERT_NOT_SUGGESTED_FOR): New macro.
962 (selftest::test_suggestions): New function.
963 (selftest::spellcheck_c_tests): Move test_get_edit_distance_unit
964 tests to selftest::test_edit_distances and call it. Add calls to
965 selftest::test_get_edit_distance_cutoff and
966 selftest::test_suggestions.
967 * spellcheck.h (get_edit_distance_cutoff): New function declaration.
968 (best_match::consider): Replace hard-coded cutoff calculation with
970 (best_match::get_cutoff): New declaration.
971 (best_match::get_best_meaningful_candidate): Likewise.
973 2018-09-14 Bernd Edlinger <bernd.edlinger@hotmail.de>
975 * builtins.c (fold_builtin_strlen): Remove TODO comment.
977 2018-09-14 Bernd Edlinger <bernd.edlinger@hotmail.de>
980 2018-07-30 Bernd Edlinger <bernd.edlinger@hotmail.de>
982 * tree-ssa-forwprop.c (simplify_builtin_call): Don't create a not NUL
983 terminated string literal.
985 2018-09-14 Martin Sebor <msebor@redhat.com>
987 * builtins.c (unterminated_array): Handle ARRAY_REF.
988 (expand_builtin_stpcpy_1): Detect unterminated char arrays.
989 * builtins.h (unterminated_array): Declare extern.
990 * gimple-fold.c (gimple_fold_builtin_stpcpy): Detect unterminated
992 (gimple_fold_builtin_sprintf): Propagate NO_WARNING to transformed
995 2018-09-14 Martin Sebor <msebor@redhat.com>
996 Jeff Law <law@redhat.com>
998 * builtins.c (unterminated_array): New.
999 (expand_builtin_strcpy): Adjust.
1000 (expand_builtin_strcpy_args): Detect unterminated arrays.
1001 * gimple-fold.c (get_maxval_strlen): Add argument. Detect
1002 unterminated arrays.
1003 * gimple-fold.h (get_maxval_strlen): Add argument.
1004 (gimple_fold_builtin_strcpy): Detec unterminated arrays.
1006 * gimple-fold.c (get_range_strlen): Add argument.
1007 (get_maxval_strlen): Adjust.
1008 * gimple-fold.h (get_range_strlen): Add argument.
1010 2018-09-14 Wei Xiao <wei3.xiao@intel.com>
1012 * config/i386/movdirintrin.h: Fix copyright year.
1014 2018-09-14 Uros Bizjak <ubizjak@gmail.com>
1016 * reg-stack.c: Include regs.h.
1017 (replace_reg): Assert that mode is MODE_FLOAT or MODE_COMPLEX_FLOAT.
1018 (emit_pop_insn): Default pop insn mode to the reg_raw_mode of
1019 FIRST_STACK_REG, not DFmode.
1020 (emit_swap_insn): Default swap insn mode to the reg_raw_mode of
1021 FIRST_STACK_REG, not XFmode. Explicitly construct swap RTX.
1022 (change stack): Default register mode to the reg_raw_mode of
1023 FIRST_STACK_REG, not DFmode.
1024 * config/i386/i386.md (*swap<mode>): Remove insn pattern.
1025 (*swapxf): Rename from swapxf.
1027 2018-09-14 Carl Love <cel@us.ibm.com>
1029 * config/rs6000/emmintrin.h: Add _MM_SHUFFLE2.
1030 * config/rs6000/xmmintrin.h: Add _MM_SHUFFLE.
1032 2018-09-14 Segher Boessenkool <segher@kernel.crashing.org>
1035 * config/rs6000/rs6000.md (*mov<mode>_hardfloat64): Add Z to the Y
1038 2018-09-14 Sam Tebbs <sam.tebbs@arm.com>
1041 * config/aarch64/aarch64.md (*aarch64_bfxilsi_uxtw): Define.
1043 2018-09-14 Jason Merrill <jason@redhat.com>
1045 Fix --enable-gather-detailed-mem-stats.
1046 * hash-table.c (hash_table_usage): Change from variable to function.
1047 * hash-table.h: Adjust.
1048 * Makefile.in: Add missing dependencies on hash-table.h.
1050 2018-09-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1052 PR tree-optimization/87259
1054 (pass_cse_reciprocals::execute): Run optimize_recip_sqrt after
1055 execute_cse_reciprocals_1 has tried transforming.
1057 2018-09-14 Aldy Hernandez <aldyh@redhat.com>
1059 * tree-vrp.c (extract_range_from_binary_expr_1): Normalize
1060 VR_VARYING for PLUS/MINUS_EXPR.
1062 2018-09-14 Ilya Leoshkevich <iii@linux.ibm.com>
1064 * config/s390/s390-passes.def (INSERT_PASS_BEFORE): Improve
1067 2018-09-14 Richard Biener <rguenther@suse.de>
1070 * tree-ssa-coalesce.c (ssa_conflicts_merge): Remove conflict
1071 bits for the merged partition.
1073 2018-09-13 Martin Sebor <msebor@redhat.com>
1074 Bernd Edlinger <bernd.edlinger@hotmail.de>
1076 * builtins.h (c_srlen): Add argument.
1077 * builtins.c (warn_string_no_nul): New function.
1078 (c_strlen): Add argument and use it. Update recursive calls.
1079 Pass DECL argument to string_constant to get info on non
1080 terminated strings. Update *NONSTR as needed.
1081 (fold_builtin_strlen): Add argument to calls to c_strlen.
1082 Warn for unterminated arrays.
1083 (warn_string_no_null): Add prototype.
1084 * expr.c (string_constant): Update arguments. Update recursive
1085 calls appropriately. Detect missing NUL terminator and outermost
1086 declaration its missing in.
1087 Improve checks for arrays with nonzero lower bound or elements
1088 that are not a single byte. Simplify offset computation.
1089 Simplify checks for non-NUL terminated strings.
1090 * gimple-fold.c (get_range_strlen): Add argument to c_strlen call.
1091 * gimple-ssa-sprintf.c (get_string_length): Remove unnecessary code.
1093 2018-09-13 Bernd Edlinger <bernd.edlinger@hotmail.de>
1095 * builtins.c (c_strlen): Handle not zero terminated STRING_CSTs
1097 * fold-const.c (c_getstr): Fix function comment. Remove unused third
1098 argument. Fix range checks.
1099 * fold-const.h (c_getstr): Adjust protoype.
1100 * gimple-fold.c (gimple_fold_builtin_memory_op): Avoid folding when
1101 string is constant but contains no NUL byte.
1103 * expr.c (string_constant): Adjust function comment.
1104 Remove bogus check for zero termination.
1106 * fold-const.c (c_getstr): Clamp STRING_LENGTH to STRING_SIZE.
1108 * varasm.c (compare_constant): Compare type size of STRING_CSTs.
1109 (get_constant_size): Don't make STRING_CSTs larger than they are.
1110 (check_string_literal): New check function for STRING_CSTs.
1111 (output_constant): Use it.
1113 2018-09-13 Eric Botcazou <ebotcazou@adacore.com>
1116 * config/visium/visium.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Define.
1118 2018-09-13 Richard Biener <rguenther@suse.de>
1120 PR tree-optimization/87263
1121 * tree-ssa-sccvn.c (visit_phi): Revert some earlier changes.
1122 (struct unwind_state): Add max_rpo field.
1123 (do_rpo_vn): Allow up-to-date loop state to be used when not iterating.
1124 Compute max_rpo, the max RPO number a block can be backwards reached
1125 from. Re-write non-iterating mode to a RPO ordered worklist approach,
1126 separating it from the iterating mode.
1128 2018-09-13 Vlad Lazar <vlad.lazar@arm.com>
1130 * haifa-sched.c (rank_for_schedule): Schedule by INSN_COST.
1131 (rfs_decision): New scheduling decision.
1133 2018-09-13 Richard Biener <rguenther@suse.de>
1136 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Fix assert.
1137 (vn_nary_op_insert_pieces_predicated): Do not write useless
1138 valid_dominated_by_p entry outside of the allocated storage.
1140 2018-09-13 Omar Sandoval <osandov@osandov.com>
1141 Tom de Vries <tdevries@suse.de>
1144 * dwarf2out.c (is_c): New function.
1145 (add_subscript_info): Add DW_AT_count of 0 for C zero-length arrays.
1147 2018-09-13 Sam Tebbs <sam.tebbs@arm.com>
1150 * config/aarch64/aarch64.md (*aarch64_bfxil):
1152 * config/aarch64/constraints.md (Ulc): Define.
1153 * config/aarch64/aarch64-protos.h (aarch64_high_bits_all_ones_p):
1155 * config/aarch64/aarch64.c (aarch64_high_bits_all_ones_p):
1158 2018-09-13 Vlad Lazar <vlad.lazar@arm.com>
1160 * config/aarch64/aarch64.h (TARGET_COMPUTE_FRAME_LAYOUT): Define.
1161 * config/aarch64/aarch64.c (aarch64_expand_prologue): Remove
1162 aarch64_layout_frame call.
1163 (aarch64_expand_epilogue): Likewise.
1164 (aarch64_initial_elimination_offset): Likewise.
1165 (aarch64_get_separate_components): Likewise.
1166 (aarch64_use_return_insn_p): Likewise.
1167 (aarch64_layout_frame): Remove unneeded check.
1169 2018-09-13 Jakub Jelinek <jakub@redhat.com>
1171 * configure.ac: Only append
1172 " : (reconfigured) $TOPLEVEL_CONFIGURE_ARGUMENTS" to
1173 gcc_config_arguments if it was never reconfigured or last reconfigure
1174 was with different arguments.
1175 * configure: Regenerated.
1177 2018-09-13 Jakub Jelinek <jakub@redhat.com>
1178 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1181 * expr.c (maybe_optimize_pow2p_mod_cmp): New function.
1182 (maybe_optimize_mod_cmp): Use it if integer_pow2p treeop1.
1184 2018-09-13 Jakub Jelinek <jakub@redhat.com>
1186 PR tree-optimization/87287
1187 * fold-const.c (fold_binary_loc) <case EQ_EXPR>: Move signed modulo
1188 X % C == 0 to X % (unsigned) C == 0 optimization to ...
1189 * match.pd (X % C == 0): ... here. New optimization.
1191 2018-09-12 Jakub Jelinek <jakub@redhat.com>
1194 * expr.h (maybe_optimize_mod_cmp): Declare.
1195 * expr.c (mod_inv): New function.
1196 (maybe_optimize_mod_cmp): New function.
1197 (do_store_flag): Use it.
1198 * cfgexpand.c (expand_gimple_cond): Likewise.
1200 2018-09-09 Cesar Philippidis <cesar@codesourcery.com>
1201 Julian Brown <julian@codesourcery.com>
1204 * gimplify.c (gimplify_scan_omp_clauses): Set
1205 target_firstprivatize_array_bases in OpenACC parallel and kernels
1206 region contexts. Remove GOMP_MAP_FIRSTPRIVATE_REFERENCE clauses from
1207 OpenACC data regions.
1209 2018-09-12 Uros Bizjak <ubizjak@gmail.com>
1211 * config/i386/i386.md (sqrt_extend<mode>xf3_i387): Remove.
1212 (sqrt<mode>2): Extend operand 1 to XFmode and generate sqrtxf3 insn.
1214 2018-09-12 Richard Biener <rguenther@suse.de>
1216 PR tree-optimization/87280
1217 * tree-ssa-sccvn.c (process_bb): Handle the case of executable
1218 edge but unreachable target.
1219 (do_rpo_vn): For conservatively handling a PHI only mark
1220 the backedge executable but not the block reachable.
1222 2018-09-12 Richard Biener <rguenther@suse.de>
1224 PR tree-optimization/87266
1225 * tree-ssa-sccvn.c (do_rpo_vn): Always iterate to not yet
1228 2018-09-12 Andreas Krebbel <krebbel@linux.ibm.com>
1230 * config/s390/s390.md (PFPO_RND_MODE_DFP, PFPO_RND_MODE_BFP): New
1232 ("trunc<BFP:mode><DFP_ALL:mode>2")
1233 ("trunc<DFP_ALL:mode><BFP:mode>2")
1234 ("extend<BFP:mode><DFP_ALL:mode>2")
1235 ("extend<DFP_ALL:mode><BFP:mode>2"): Set proper rounding mode
1236 according to the target operand type.
1238 2018-09-12 Jakub Jelinek <jakub@redhat.com>
1239 Andreas Krebbel <krebbel@linux.ibm.com>
1241 PR tree-optimization/86844
1242 * gimple-ssa-store-merging.c
1243 (imm_store_chain_info::coalesce_immediate): For overlapping stores, if
1244 there are any overlapping stores in between them, make sure they are
1245 also coalesced or we give up completely.
1247 2018-09-12 Jakub Jelinek <jakub@redhat.com>
1250 * fold-const.c (fold_ternary_loc) <case COND_EXPR>: Verify also that
1251 BIT_AND_EXPR's second operand is a power of two. Formatting fix.
1253 2018-09-12 Tom de Vries <tdevries@suse.de>
1255 * common.opt (gdescribe-dies): Add option.
1256 * dwarf2out.c (add_name_and_src_coords_attributes): Add description
1257 attribute for artifical and nameless decls.
1258 (dwarf2out_register_external_die): Add description attribute to
1259 external reference die.
1260 (add_desc_attribute): New functions.
1261 (gen_subprogram_die): Add description attribute to
1262 DW_TAG_call_site_parameter.
1263 * tree-pretty-print.c (print_generic_expr_to_str): New function.
1264 * tree-pretty-print.h (print_generic_expr_to_str): Declare.
1265 * doc/invoke.texi (@item Debugging Options): Add -gdescribe-dies and
1267 (@item -gdescribe-dies): Add.
1269 2018-09-12 Aldy Hernandez <aldyh@redhat.com>
1271 * tree-vrp.c (vrp_shift_undefined_p): Remove.
1272 (extract_range_from_binary_expr_1: Call
1273 wide_int_range_shift_undefined_p instead of vrp_shift_undefined_p.
1274 * wide-int-range.h (wide_int_range_shift_undefined_p): Do not
1277 2018-09-12 Aldy Hernandez <aldyh@redhat.com>
1279 * gimple-ssa-warn-alloca.c
1280 (alloca_type_and_limit::alloca_type_and_limit): Initialize limit
1281 field for ALLOCA_BOUND_*_LARGE.
1283 2018-09-11 Nathan Sidwell <nathan@acm.org>
1285 * gcc.c (load_specs, execute, run_attempt): Use %qs not '%s'.
1287 2018-09-11 Uros Bizjak <ubizjak@gmail.com>
1289 * reg-stack.c (subst_asm_stack_regs): Call replace_reg also
1290 for clobbers. Remove obsolete comment.
1292 2018-09-11 Uros Bizjak <ubizjak@gmail.com>
1294 * config/i386/i386.md (define_attr "type"): Remove mpxmov, mpxmk,
1295 mpxchk, mpxld and mpxst types.
1296 (define_attr length_immediate): Remove all processing of mpx types.
1297 (define_attr prefix_0f): Ditto.
1298 (define_attr memory): Ditto.
1300 2018-09-11 Uros Bizjak <ubizjak@gmail.com>
1302 * config/i386/i386.md (fyl2x_extend<mode>xf3_i387): Remove.
1303 (log<mode>2): Change operand 1 predicate to general_operand.
1304 Extend operand 1 to XFmode and generate logxf3 insn.
1305 (log10<mode>2): Change operand 1 predicate to general_operand.
1306 Extend operand 1 to XFmode and generate log10xf3 insn.
1307 (log2<mode>2): Change operand 1 predicate to general_operand.
1308 Extend operand 1 to XFmode and generate log2xf3 insn.
1309 (fyl2xp1_extend<mode>xf3_i387): Remove.
1310 (log1p<mode>2): Change operand 1 predicate to general_operand.
1311 Extend operand 1 to XFmode and generate log1pxf3 insn.
1312 (fxtract_extend<mode>xf3_i387): Remove.
1313 (logb<mode>2): Change operand 1 predicate to general_operand.
1314 Extend operand 1 to XFmode and generate logbxf3 insn.
1315 (ilogb<mode>2): Change operand 1 predicate to general_operand.
1316 Extend operand 1 to XFmode and generate fxtractxf3_i387 insn.
1317 (significand<mode>2): Change operand 1 predicate to general_operand.
1318 Extend operand 1 to XFmode and generate significandxf3 insn.
1320 2018-09-11 Nathan Sidwell <nathan@acm.org>
1322 * gcc.c (perror_with_name, pfatal_with_name): Delete.
1323 (load_specs): Use fatal_error.
1324 (DELETE_IF_ORDINARY, process_command): Use error.
1325 (execute, run_attempt): Use fatal_error.
1327 2018-09-11 Andrew Stubbs <ams@codesourcery.com>
1329 * diagnostic-core.h (sorry_at): New prototype.
1330 * diagnostic.c (sorry_at): New function.
1332 2018-09-11 Aldy Hernandez <aldyh@redhat.com>
1334 * tree-vrp.c (extract_range_from_binary_expr_1): Treat all divisions
1335 by zero as VR_UNDEFINED.
1337 2018-09-10 Uros Bizjak <ubizjak@gmail.com>
1339 * config/i386/i386.md (<sincos>xf2): Rename from *<sincos>xf2_i387.
1340 (*<sincos>_extend<mode>xf2_i387): Remove insn pattern.
1341 (<sincos>mode2): New expander.
1342 (sincos_extend<mode>xf3_i387): Remove insn pattern.
1343 (sincos -> sin, cos splitters): Remove splitter patterns.
1344 (sincos<mode>3): Change operand 2 predicate to general_operand.
1345 Extend operand 2 to XFmode and generate sincosxf3 insn.
1346 (fptanxf4_i387): Change mode of operands 0 and 3 to SFmode.
1347 Change operand 3 predicate to const1_operand.
1348 (fptan_extend<mode>xf4_i387): Remove insn pattern.
1349 (tanxf2): Update operands in the call to fptanxf4_i387.
1350 (tan<mode>2): Change operand 1 predicate to general_operand.
1351 Extend operand 1 to XFmode and generate tanxf3 insn.
1352 (atan2xf3): Rename from *fpatanxf3_i387.
1353 (fpatan_extend<mode>xf3_i387): Remove insn pattern.
1354 (atan2xf3): Remove expander.
1355 (atan2<mode<3): Change operand 1 and 2 predicates to general_operand.
1356 Extend operands 1 and 2 to XFmode and generate atan2xf3 insn.
1357 (atan<mode>2): Change operand 1 predicate to general_operand.
1358 Extend operand 1 to XFmode and generate atanxf3 insn.
1360 2018-09-10 Uros Bizjak <ubizjak@gmail.com>
1362 * config/i386/i386.md (x87/SSE constant load splitter): Use
1363 memory_operand instead of nonimmediate_operand for input operand
1366 2018-09-09 Uros Bizjak <ubizjak@gmail.com>
1368 * config/i386/i386.md (float partial SSE register stall splitter): Move
1369 splitter near its instruction pattern.
1370 (float_extend partial SSE register stall splitter): Ditto.
1371 (float_truncate partial SSE register stall splitter): Ditto.
1373 2018-09-09 Hans-Peter Nilsson <hp@bitrange.com>
1376 * config/mmix/mmix.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Redefine
1377 to speculation_safe_value_not_needed.
1380 * config/mmix/mmix.c (mmix_assemble_integer): Handle byte-size
1381 non-CONST_INT rtx:es using assemble_integer_with_op ".byte".
1382 (MMIX_CFUN_NEEDS_SAVED_EH_RETURN_ADDRESS): Don't call
1383 leaf_function_p, instead use has_hard_reg_initial_val.
1385 2018-09-09 Nathan Sidwell <nathan@acm.org>
1387 * gcc.h (pfatal_with_name): Don't declare here.
1388 * gcc.c (pfatal_with_name): Make static.
1390 2018-09-09 Xianmiao Qu <xianmiao_qu@c-sky.com>
1392 * config/csky/csky.md (*cskyv2_adddi3, *cskyv2_subdi3): Add
1395 2018-09-08 John David Anglin <danglin@gcc.gnu.org>
1397 PR rtl-optimization/85458
1398 * sel-sched.c (sel_target_adjust_priority): Allow backend adjust
1399 priority hook to reduce the priority of EXPR.
1401 2018-09-07 Uros Bizjak <ubizjak@gmail.com>
1403 * config/i386/i386.md (float<SWI48x:mode><MODEF:mode>2) Enable
1404 DImode for x87 on 32bit targets. Conditionally disable x87 modes
1405 with X87_ENABLE_FLOAT. Remove preparation code.
1406 (*float<SWI48:mode><MODEF:mode>2): Rename from
1407 *float<SWI48:mode><MODEF:mode>2_mixed. Handle x87, SSE and mixed
1408 math using "enabled" attribute.
1409 (*floatdi<MODEF:mode>2_i387): Rename from
1410 *float<SWI48x:mode><MODEF:mode>2_i387. Handle only DImode and
1411 enable for 32bit targets only.
1412 (floatdi<X87MODEF:mode>2_i387_with_xmm pre-reload splitter): New
1414 (floatdi<X87MODEF:mode>2_i387_with_xmm): Use register_operand
1415 as operand 1 predicate. Rewrite as define_insn_and_split.
1416 (floatdi<X87MODEF:mode>2_i387_with_xmm memory input splitter): Remove.
1418 2018-09-06 Uros Bizjak <ubizjak@gmail.com>
1420 * reg-stack.c (get_true_reg) <case FLOAT_TRUNCATE>: Reorder
1421 to fallthru to FLOAT case.
1423 2018-09-06 Will Schmidt <will_schmidt@vnet.ibm.com>
1426 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Update logic
1427 around folding of vec_sl to handle out of range shift values.
1429 2018-09-06 Uros Bizjak <ubizjak@gmail.com>
1431 * config/i386/i386.md (fix_trunc<mode>_fisttp_i387_1): Remove.
1432 Update callers to gen_fix_trunc<mode>_i387_fisttp
1433 (fix_trunc<mode>_i387_fisttp): Change operand 0 predicate to
1434 nonimmediate_operand.
1435 (fix_trunc<mode>_i387_fisttp_with_temp): Remove insn pattern
1436 and corresponding splitters.
1437 (*fix_trunc<mode>_i387_1): Always expand with fix_trunc<mode>_i387.
1438 (fix_truncdi_i387): Change operand 0 predicate to nonimmediate_operand.
1439 (fix_truncdi_i387_with_temp): Remove insn pattern
1440 and corresponding splitters.
1441 (fix_trunc<mode>_i387): Change operand 0 predicate to
1442 nonimmediate_operand.
1443 (fix_trunc<mode>_i387_with_temp): Remove insn pattern
1444 and corresponding splitters.
1445 (*fistdi2_1): Remove.
1447 (fistdi2_with_temp): Remove insn pattern and corresponding splitters.
1448 (lrintxfdi2): Remove expander. Reimplement as define_insn.
1449 (*fist<mode>2_1): Remove.
1450 (fist<mode>2): Ditto.
1451 (fist<mode>2_with_temp): Remove insn pattern and corresponding
1453 (lrintxf<mode>2): Remove expander. Reimplement as define_insn.
1454 (*fist<mode>2_<rounding>_1): Always expand with fist<mode>2_<rounding>.
1455 (fistdi2_<rounding>): Change operand 0 predicate to
1456 nonimmediate_operand.
1457 (fistdi2_<rounding>_with_temp): Remove insn pattern
1458 and corresponding splitters.
1459 (fist<mode>2_<rounding>): Change operand 0 predicate to
1460 nonimmediate_operand.
1461 (fist<mode>2_<rounding>_with_temp): Remove insn pattern
1462 and corresponding splitters.
1464 (*fixuns_trunc<mode>si2_avx512f_zext): Depend on TARGET_SSE_MATH.
1466 2018-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
1468 * varasm.c (output_constructor_regular_field): Check TYPE_SIZE_UNIT of
1471 2018-09-06 Will Schmidt <will_schmidt@vnet.ibm.com>
1473 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
1474 early gimple folding of vec_splat().
1475 * tree-vect-generic.c: Remove static from tree_vec_extract() definition.
1476 * gimple-fold.h: Add an extern define for tree_vec_extract().
1478 2018-09-06 Will Schmidt <will_schmidt@vnet.ibm.com>
1480 * config/rs6000/rs6000.c (fold_mergehl_helper): Add types_compatible_p
1481 wrappers around TREE_TYPE comparisons.
1483 2018-09-06 Ilya Leoshkevich <iii@linux.ibm.com>
1486 * config/s390/predicates.md: Add nonsym_memory_operand.
1487 * config/s390/s390.c (s390_legitimize_cs_operand): If operand
1488 contains a SYMBOL_REF, load it into an intermediate pseudo.
1489 (s390_emit_compare_and_swap): Legitimize operand.
1490 * config/s390/s390.md: Use the new nonsym_memory_operand
1491 with UNSPECV_CAS patterns.
1493 2018-09-06 Ilya Leoshkevich <iii@linux.ibm.com>
1496 * config/s390/s390-passes.def: New file.
1497 * config/s390/s390-protos.h (class rtl_opt_pass): Add forward
1499 (make_pass_s390_early_mach): Add declaration.
1500 * config/s390/s390.c (make_pass_s390_early_mach):
1501 (s390_option_override): Remove dynamic registration.
1502 * config/s390/t-s390: Add s390-passes.def.
1504 2018-09-06 Ilya Leoshkevich <iii@linux.ibm.com>
1506 * config/s390/s390.c (s390_decompose_constant_pool_ref):
1507 Remove UNSPEC_LTREL_BASE check.
1508 (annotate_constant_pool_refs): Likewise.
1509 (find_constant_pool_ref): Likewise.
1510 (find_ltrel_base): Removed.
1511 (replace_ltrel_base): Removed.
1512 (s390_mainpool_finish): Remove replace_ltrel_base call.
1513 (s390_chunkify_start): Remove pending LTREL_BASE logic.
1514 (s390_chunkify_finish): Remove replace_ltrel_base call.
1515 * config/s390/s390.md: Remove UNSPEC_LTREL_BASE.
1517 2018-09-06 Hans-Peter Nilsson <hp@axis.com>
1520 * config/cris/cris.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Redefine
1521 to speculation_safe_value_not_needed.
1523 2018-09-05 Cesar Philippidis <cesar@codesourcery.com>
1524 Bernd Schmidt <bernds_cb1@t-online.de>
1526 * config/nvptx/nvptx-opts.h: New file.
1527 * config/nvptx/nvptx.c (nvptx_file_start): Print the correct .target.
1528 * config/nvptx/nvptx.h: Include "nvptx-opts.h".
1530 (TARGET_SM35): New macro.
1531 * config/nvptx/nvptx.md (atomic_fetch_<logic><mode>): Enable with the
1533 * config/nvptx/nvptx.opt (ptx_isa, sm_30, sm_35): New enum and its
1535 (misa=): New option.
1536 * doc/invoke.texi (Nvidia PTX Options): Document -misa.
1538 2018-09-05 Uros Bizjak <ubizjak@gmail.com>
1540 * config/i386/i386.md (truncdfsf2): Remove expander.
1541 (truncdfsf2_with_temp): Ditto.
1542 (truncxf<mode>2): Ditto.
1543 (*truncdfsf_fast_mixed): Remove insn pattern.
1544 (*truncdfsf_fast_i387): Ditto.
1545 (*truncdfsf_mixed): Ditto.
1546 (*truncdfsf_i387): Ditto.
1547 (*truncdfsf2_i387_1): Ditto.
1548 (*truncxfsf2_mixed): Ditto.
1549 (*truncxfdf2_mixed): Ditto.
1550 (*truncxf<mode>2_i387_noop): Ditto. Update callers
1551 to call gen_truncxf<mode>2 instead.
1552 (*truncxf<mode>2_i387): Remove.
1553 (reg->reg splitters): Remove splitter pattern.
1554 (reg->mem splitters): Ditto.
1556 (truncdfsf2): New insn pattern.
1557 (truncxf<mode>2): Ditto.
1559 2018-09-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1561 * tree-ssa-math-opts.c (is_mult_by): New function.
1562 (is_square_of): Use the above.
1563 (optimize_recip_sqrt): New function.
1564 (pass_cse_reciprocals::execute): Use the above.
1566 2018-09-05 Richard Biener <rguenther@suse.de>
1569 * tree-ssa-sccvn.c (rpo_elim::eliminate_push_avail): Make sure
1570 to zero-init the emplaced vec.
1572 2018-09-05 Martin Liska <mliska@suse.cz>
1574 PR tree-optimization/87205
1575 * tree-switch-conversion.c (pass_lower_switch::execute):
1576 Group cases for switch statements.
1578 2018-09-05 Richard Biener <rguenther@suse.de>
1580 PR tree-optimization/87217
1581 * tree-ssa-sccvn.c (vuse_valueize): New.
1582 (vn_reference_lookup_pieces): Use it.
1583 (vn_reference_lookup): Likewise.
1585 2018-09-05 Nathan Sidwell <nathan@acm.org>
1588 * stor-layout.c (place_field): Scan forwards to check last
1589 bitfield when ms_bitfield_placement is in effect.
1591 2018-09-05 Richard Biener <rguenther@suse.de>
1594 * tree-vect-stmts.c (vectorizable_simd_clone_call): Fix bogus
1597 2018-09-05 Siddhesh Poyarekar <siddhesh@sourceware.org>
1598 Kugan Vivekanandarajah <kugan.vivekanandarajah@linaro.org>
1600 * config/aarch64/falkor-tag-collision-avoidance.c: New file.
1601 * config.gcc (extra_objs): Build it.
1602 * config/aarch64/t-aarch64 (falkor-tag-collision-avoidance.o):
1604 * config/aarch64/aarch64-passes.def
1605 (pass_tag_collision_avoidance): New pass.
1606 * config/aarch64/aarch64.c (qdf24xx_tunings): Add
1607 AARCH64_EXTRA_TUNE_RENAME_LOAD_REGS to tuning_flags.
1608 (aarch64_classify_address): Remove static qualifier.
1609 (aarch64_address_info, aarch64_address_type): Move to...
1610 * config/aarch64/aarch64-protos.h: ... here.
1611 (make_pass_tag_collision_avoidance): New function.
1612 * config/aarch64/aarch64-tuning-flags.def (rename_load_regs):
1615 2018-09-05 Martin Liska <mliska@suse.cz>
1617 * doc/gcov.texi: Update documentation of humar
1619 * gcov.c (format_count): Print one decimal place, it provides
1620 more fine number of situations like '1G' vs. '1.4G'.
1622 2018-09-05 Martin Liska <mliska@suse.cz>
1625 * config/rs6000/rs6000.opt: Mark the option as Deprecated.
1626 * optc-gen.awk: Allow 'Var' for Deprecated options in order
1627 to generate a MASK value.
1629 2018-09-04 H.J. Lu <hongjiu.lu@intel.com>
1632 * dwarf2out.c (based_loc_descr): Allow hard frame pointer even
1633 if frame pointer isn't used.
1634 (compute_frame_pointer_to_fb_displacement): Likewise.
1636 2018-09-04 Jakub Jelinek <jakub@redhat.com>
1639 * common/config/i386/i386-common.c (OPTION_MASK_ISA_XSAVEOPT_SET,
1640 OPTION_MASK_ISA_XSAVES_SET, OPTION_MASK_ISA_XSAVEC_SET): Use
1641 OPTION_MASK_ISA_XSAVE_SET instead of OPTION_MASK_ISA_XSAVE.
1642 (OPTION_MASK_ISA_XSAVE_UNSET): Add OPTION_MASK_ISA_XSAVES_UNSET
1643 and OPTION_MASK_ISA_XSAVEC_UNSET.
1645 2018-09-04 Max Filippov <jcmvbkbc@gmail.com>
1647 * config/xtensa/xtensa.c (xtensa_expand_atomic): Reorder AND and
1648 XOR operations in NAND case.
1650 2018-09-04 Aldy Hernandez <aldyh@redhat.com>
1652 * wide-int-range.cc (wide_int_range_convert): New.
1653 * wide-int-range.h (wide_int_range_convert): New.
1654 * tree-vrp.c (extract_range_from_unary_expr): Abstract wide int
1655 code into wide_int_range_convert.
1656 (extract_range_into_wide_ints): Do not munge anti range constants
1657 into the entire domain. Just return the range back.
1659 2018-09-04 Martin Liska <mliska@suse.cz>
1661 * genmatch.c (output_line_directive): Add new argument
1663 (dt_simplify::gen_1): Encapsulate dump within __builtin_expect.
1665 2018-09-04 Jonathan Wakely <jwakely@redhat.com>
1667 * doc/invoke.texi (Option Summary): Add whitespace.
1669 * doc/invoke.texi (Option Summary): Add -Waligned-new.
1671 2018-09-04 Richard Biener <rguenther@suse.de>
1673 PR tree-optimization/87211
1674 * tree-ssa-sccvn.c (visit_phi): When value-numbering to a
1675 backedge value we're supposed to treat as VARYING also number
1676 the PHI to VARYING in case it got a different value-number already.
1678 2018-09-04 Aldy Hernandez <aldyh@redhat.com>
1680 * tree-vrp.c (vrp_can_optimize_bit_op): Remove.
1681 (extract_range_from_binary_expr_1): Do not call
1682 vrp_can_optimize_bit_op.
1683 * wide-int-range.cc (wide_int_range_can_optimize_bit_op): Make
1685 (wide_int_range_get_mask_and_bounds): New.
1686 (wide_int_range_optimize_bit_op): New.
1687 (wide_int_range_bit_ior): Call wide_int_range_optimize_bit_op.
1688 (wide_int_range_bit_and): Same.
1689 * wide-int-range.h (wide_int_range_can_optimize_bit_op): Remove.
1690 (wide_int_range_optimize_bit_op): New.
1691 (wide_int_range_get_mask_and_bounds): New.
1693 2018-09-04 Richard Biener <rguenther@suse.de>
1695 PR tree-optimization/87176
1696 * tree-ssa-sccvn.c (visit_phi): Remove redundant allsame
1697 variable. When value-numbering a virtual PHI node make sure
1698 to not value-number to the backedge value.
1700 2018-09-04 Jonathan Wakely <jwakely@redhat.com>
1702 * doc/extend.texi (Long Long, Hex Floats): Document support for
1703 long long and hex floats in more recent versions of ISO C++.
1705 2018-09-03 Richard Biener <rguenther@suse.de>
1707 PR tree-optimization/87177
1708 * tree-ssa-sccvn.c (vuse_ssa_val): Revert previous change, keep
1711 2018-09-03 Alexander Monakov <amonakov@ispras.ru>
1713 * bb-reorder.c (edge_order): Convert to C-qsort-style
1714 tri-state comparator.
1715 (reorder_basic_blocks_simple): Change std::stable_sort to
1718 2018-09-03 Alexander Monakov <amonakov@ispras.ru>
1720 * tree-loop-distribution.c (offset_cmp): Convert to C-qsort-style
1721 tri-state comparator.
1722 (fuse_memset_builtins): Change std::stable_sort to gcc_stablesort.
1724 2018-09-03 Alexander Monakov <amonakov@ispras.ru>
1726 * sort.cc (struct sort_ctx): New field 'nlim'. Use it...
1727 (mergesort): ... here as maximum count for using netsort.
1728 (gcc_qsort): Set nlim to 3 if stable sort is requested.
1729 (gcc_stablesort): New.
1730 * system.h (gcc_stablesort): Declare.
1732 2018-09-03 Alexander Monakov <amonakov@ispras.ru>
1734 * sort.cc (gcc_qsort) [CHECKING_P]: Call qsort_chk.
1735 * system.h (qsort): Always redirect to gcc_qsort. Update comment.
1736 * vec.c (qsort_chk): Do not call gcc_qsort. Update comment.
1738 2018-09-03 Segher Boessenkool <segher@kernel.crashing.org>
1740 * config/rs6000/rs6000.md (*mov<mode>_hardfloat32): Remove %U from the
1741 lxsdx and stxsdx alternatives.
1742 (*mov<mode>_hardfloat64): Ditto.
1743 * config/rs6000/vsx.md (*vsx_extract_<mode>_store): Ditto.
1745 2018-09-03 Richard Biener <rguenther@suse.de>
1747 PR tree-optimization/87200
1748 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Valueize a
1751 2018-09-03 Martin Liska <mliska@suse.cz>
1753 PR tree-optimization/87201
1754 * tree-switch-conversion.c (switch_decision_tree::balance_case_nodes):
1755 Fix parenthesis in an expression.
1757 2018-09-03 Richard Biener <rguenther@suse.de>
1759 PR tree-optimization/87197
1760 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Mark the new def
1761 visited. CSE the VN_INFO hashtable lookup.
1763 PR tree-optimization/87169
1764 * tree-ssa-sccvn.c (do_rpo_vn): When marking loops for not
1765 iterating make sure there's no extra backedges from irreducible
1766 regions feeding the header. Mark the destination block
1769 2018-09-03 Martin Liska <mliska@suse.cz>
1772 * common/common-target.def: Add TARGET_GET_VALID_OPTION_VALUES.
1773 * common/common-targhooks.c (default_get_valid_option_values):
1775 * common/common-targhooks.h (default_get_valid_option_values):
1777 * common/config/i386/i386-common.c: Move processor_target_table
1779 (ix86_get_valid_option_values): New function.
1780 (TARGET_GET_VALID_OPTION_VALUES): New macro.
1781 * config/i386/i386.c (struct ptt): Move to i386-common.c.
1782 (PTA_*): Move all defined masks into i386-common.c.
1783 (ix86_function_specific_restore): Use new processor_cost_table.
1784 * config/i386/i386.h (struct ptt): Moved from i386.c.
1785 (struct pta): Likewise.
1786 * doc/tm.texi: Document new TARGET_GET_VALID_OPTION_VALUES.
1787 * doc/tm.texi.in: Likewise.
1788 * opt-suggestions.c (option_proposer::suggest_option):
1789 Pass prefix to build_option_suggestions.
1790 (option_proposer::get_completions): Likewise.
1791 (option_proposer::build_option_suggestions): Use the new target
1793 * opts.c (struct option_help_tuple): New struct.
1794 (print_filtered_help): Use the new target hook.
1796 2018-09-03 Martin Liska <mliska@suse.cz>
1799 * predict.c (set_even_probabilities): Add likely_edges
1800 argument and handle cases where we have precisely one
1802 (combine_predictions_for_bb): Catch also likely_edges.
1803 (tree_predict_by_opcode): Handle gswitch statements.
1804 * tree-cfg.h (find_case_label_for_value): New declaration.
1805 (find_taken_edge_switch_expr): Likewise.
1806 * tree-switch-conversion.c (switch_decision_tree::balance_case_nodes):
1807 Find pivot in decision tree based on probabily, not by number of
1810 2018-09-02 Gerald Pfeifer <gerald@pfeifer.com>
1812 * doc/standards.texi (Standards): Update Objective-C reference.
1814 2018-09-01 Gerald Pfeifer <gerald@pfeifer.com>
1816 * doc/install.texi (Prerequisites): Update link for MPC.
1818 2018-09-01 Michael Matz <matz@suse.de>
1820 PR tree-optimization/87074
1821 * gimple-loop-jam.c (unroll_jam_possible_p): Check loop exit
1822 PHIs for outer-loop uses.
1824 2018-09-01 Gerald Pfeifer <gerald@pfeifer.com>
1826 * doc/generic.texi (OpenMP): Adjust link to openmp.org.
1827 * doc/invoke.texi (C Dialect Options): Ditto.
1829 2018-09-01 Gerald Pfeifer <gerald@pfeifer.com>
1831 * doc/install.texi (Prerequisites): Adjust link mpfr.org.
1833 2018-08-31 Richard Biener <rguenther@suse.de>
1835 PR tree-optimization/87168
1836 * tree-ssa-sccvn.c (SSA_VAL): Add visited output parameter.
1837 (rpo_elim::eliminate_avail): When OP was not visited it must
1840 2018-08-31 David Malcolm <dmalcolm@redhat.com>
1842 * tree-vrp.c (copy_value_range): Convert param "from" from
1843 "value_range *" to "const value_range *".
1844 (range_is_null): Likewise for param "vr".
1845 (range_int_cst_p): Likewise.
1846 (range_int_cst_singleton_p): Likewise.
1847 (symbolic_range_p): Likewise.
1848 (value_ranges_intersect_p): Likewise for both params.
1849 (value_range_nonnegative_p): Likewise for param "vr".
1850 (value_range_constant_singleton): Likewise.
1851 (vrp_set_zero_nonzero_bits): Likewise for param "ar".
1852 (extract_range_into_wide_ints): Likewise for param "vr".
1853 (extract_range_from_multiplicative_op): Likewise for params "vr0"
1855 (vrp_can_optimize_bit_op): Likewise.
1856 (extract_range_from_binary_expr_1): Likewise for params "vr0_" and
1858 (extract_range_from_unary_expr): Likewise.
1859 (debug_value_range): Likewise for param "vr".
1860 (value_range::dump): Add "const" qualifier.
1861 (vrp_prop::check_array_ref): Convert local "vr" from
1862 "value_range *" to "const value_range *".
1863 (vrp_prop::check_mem_ref): Likewise.
1864 (vrp_prop::visit_stmt): Likewise for local "old_vr".
1865 (vrp_intersect_ranges_1): Likewise for param "vr_1".
1866 (vrp_intersect_ranges): Likewise.
1867 (simplify_stmt_for_jump_threading): Likewise for local "vr".
1868 (vrp_prop::vrp_finalize): Likewise.
1869 * tree-vrp.h (value_range::dump): Add "const" qualifier.
1870 (vrp_intersect_ranges): Add "const" qualifier to params as above.
1871 (extract_range_from_unary_expr): Likewise.
1872 (value_range_constant_singleton): Likewise.
1873 (symbolic_range_p): Likewise.
1874 (copy_value_range): Likewise.
1875 (extract_range_from_binary_expr_1): Likewise.
1876 (range_int_cst_p): Likewise.
1877 (vrp_set_zero_nonzero_bits): Likewise.
1878 (range_int_cst_singleton_p): Likewise.
1880 2018-08-31 Vlad Lazar <vlad.lazar@arm.com>
1882 * config/aarch64/arm_neon.h (vabsd_s64): New.
1883 (vnegd_s64): Likewise.
1885 2018-08-31 Martin Jambor <mjambor@suse.cz>
1887 * ipa-cp.c (estimate_local_effects): Replace wrong MAX with MIN.
1889 2018-08-31 Martin Liska <mliska@suse.cz>
1891 * ipa-icf.c (sem_item::add_type): Use
1892 sem_item::m_type_hash_cache.
1893 * ipa-icf.h: Move the cache from sem_item_optimizer
1896 2018-08-31 Nathan Sidwell <nathan@acm.org>
1898 * doc/extend.texi (Backwards Compatibility): Remove implicit
1899 extern C leeway of () being (...).
1901 2018-08-31 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1903 * ipa-inline.c (can_inline_edge_by_limits_p): Fix typos in comment.
1905 2018-08-31 Segher Boessenkool <segher@kernel.crashing.org>
1909 * config/rs6000/rs6000.md (lround<mode>di2): Gate on TARGET_FPRND.
1911 2018-08-31 Jakub Jelinek <jakub@redhat.com>
1914 * expmed.c (expand_mult_const): Use immed_wide_int_const instead of
1915 gen_int_mode. Formatting fixes.
1917 2018-08-30 Sandra Loosemore <sandra@codesourcery.com>
1919 * target.def (custom_function_descriptors): Improve documentation.
1920 * doc/tm.texi.in (Trampolines): Expand discussion of function
1921 descriptors and move TARGET_CUSTOM_FUNCTION_DESCRIPTORS to the
1922 beginning of the section.
1923 * doc/tm.texi: Regenerated.
1925 2018-08-30 Jose E. Marchesi <jose.marchesi@oracle.com>
1927 * cfg.h (class auto_edge_flag): Spell out the template-id of the
1928 base class in the initializer list. This is a workaround for
1929 building with older GCC.
1930 (class auto_bb_flag): Likewise.
1932 2018-08-30 Aaron Sawdey <acsawdey@linux.ibm.com>
1934 * config/rs6000/altivec.md (altivec_eq<mode>): Remove star.
1935 (altivec_vcmpequ<VI_char>_p): Remove star.
1936 * config/rs6000/rs6000-string.c (do_load_for_compare): Support
1938 (expand_strncmp_vec_sequence): New function.
1939 (emit_final_str_compare_vec): New function.
1940 (expand_strn_compare): Add support for vector strncmp.
1941 * config/rs6000/rs6000.opt (-mstring-compare-inline-limit): Change
1942 length specification to bytes.
1943 * config/rs6000/vsx.md (vsx_ld_elemrev_v16qi_internal): Remove star.
1944 (vcmpnezb_p): New pattern.
1945 * doc/invoke.texi (RS/6000 and PowerPC Options): Update documentation
1946 for option -mstring-compare-inline-limit.
1948 2018-08-30 Thiago Macieira <thiago.macieira@intel.com>
1950 * config/i386/i386.c (PTA_WESTMERE): Remove PTA_AES.
1951 (PTA_SKYLAKE): Add PTA_AES.
1952 (PTA_GOLDMONT): Likewise.
1954 2018-08-29 Jan Hubicka <jh@suse.cz>
1957 * lto-opts.c (lto_write_options): Always stream PIC/PIE mode.
1958 * lto-wrapper.c (merge_and_complain): Fix merging of PIC/PIE.
1960 2018-08-29 Jan Hubicka <jh@suse.cz>
1962 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not follow
1964 (hash_tree): Do not visit TYPE_STUB_DECL.
1965 * tree-streamer-out.c (write_ts_type_common_tree_pointers): Do not
1966 stream TYPE_STUB_DECL.
1967 * tree-streamer-in.c (lto_input_ts_type_common_tree_pointers): Likewise.
1968 * ipa-utils.h (type_with_linkage_p): Do not rely on TYPE_STUB_DECL
1969 after free_lang_data.
1970 (type_in_anonymous_namespace_p): Likewise.
1972 2018-08-29 Jan Hubicka <jh@suse.cz>
1974 * sreal.h (SREAL_PART_BITS): Change to 31; remove seemingly unnecessary
1975 comment that it has to be even number.
1976 (class sreal): Change m_sig type to int32_t.
1977 * sreal.c (sreal::dump, sreal::to_int, opreator+, operator-): Use
1978 int64_t for temporary calculations.
1979 (sreal_verify_basics): Drop one bit from minimum and maximum.
1981 2018-08-30 Richard Biener <rguenther@suse.de>
1983 PR tree-optimization/87147
1984 * tree-ssa-sccvn.c (SSA_VISITED): New function.
1985 (visit_phi): When the degenerate result is from the backedge and
1986 we didn't visit its definition yet drop to VARYING.
1987 (do_rpo_vn): Properly mark blocks with incoming backedges as executable.
1989 2018-08-29 Jan Hubicka <jh@suse.cz>
1991 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not walk
1993 (hash_tree): Likewise.
1995 2018-08-29 Jan Hubicka <jh@suse.cz>
1997 * tree.c (find_decls_types_r): Walk also TYPE_NEXT_PTR_TO
1998 and TYPE_NEXT_REF_TO.
2000 2018-08-29 Jan Hubicka <jh@suse.cz>
2002 * sreal.h (SREAL_PART_BITS): Change to 31; remove seemingly unnecessary
2003 comment that it has to be even number.
2004 (class sreal): Change m_sig type to int32_t.
2005 * sreal.c (sreal::dump, sreal::to_int, opreator+, operator-): Use
2006 int64_t for temporary calculations.
2007 (sreal_verify_basics): Drop one bit from minimum and maximum.
2009 2018-08-30 Tamar Christina <tamar.christina@arm.com>
2011 * config/aarch64/aarch64.c (aarch64_expand_movmem): Set TImode max.
2013 2018-08-30 Vlad Lazar <vlad.lazar@arm.com>
2016 * expmed.c (canonicalize_comparison): Use wi::sub instead of wi::add
2017 if to_add is negative.
2019 2018-08-29 Bernd Edlinger <bernd.edlinger@hotmail.de>
2022 * builtins.c (c_strlen): Improve range checks.
2024 2018-08-29 Martin Sebor <msebor@redhat.com>
2025 Jeff Law <law@redhat.com>
2027 PR tree-optimization/86714
2028 PR tree-optimization/86711
2029 * builtins.c (c_strlen): Add arguments to call to string_constant.
2030 * expr.c (string_constant): Add argument. Detect missing nul
2031 terminator and outermost declaration it's missing in.
2032 * expr.h (string_constant): Add argument.
2033 * fold-const.c (read_from_constant_string): Add arguments to call to
2035 (c_getstr): Likewise.
2036 * tree-ssa-forwprop.c (simplify_builtin_call): Likewise.
2038 * tree-ssa-strlen.c (get_stridx): Likewise.
2040 2018-08-29 Jan Hubicka <jh@suse.cz>
2042 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers):
2043 Do not stream DECL_VINDEX.
2044 * tree-streamer-out.c (write_ts_function_decl_tree_pointers): Likewise.
2045 * tree.c (free_lang_data_in_decl): Clear DECL_VINDEX.
2046 (decl_function_context): Use DECL_VIRTUAL_P rather than DECL_VINDEX.
2048 2018-08-29 Richard Biener <rguenther@suse.de>
2050 * tree-ssa-sccvn.c (vuse_ssa_val): Return NULL for unvisited
2051 virtual operands that are not default defs to honor region
2053 (rpo_vn_valueize): Remove ineffective code here.
2055 2018-08-29 Richard Biener <rguenther@suse.de>
2057 PR tree-optimization/87132
2058 * tree-ssa-alias.c (get_continuation_for_phi): Do not translate
2059 when skipping defs reachable over backedges.
2061 2018-08-29 Richard Biener <rguenther@suse.de>
2063 * tree-core.h: Document use of deprecated_flag in SSA_NAME.
2064 * tree.h (SSA_NAME_POINTS_TO_READONLY_MEMORY): Define.
2065 * tree-into-ssa.c (pass_build_ssa::execute): Initialize
2066 function parameters SSA_NAME_POINTS_TO_READONLY_MEMORY from fnspec.
2067 * tree-ssa-sccvn.c (const_parms, init_const_parms): Remove.
2068 (vn_reference_lookup_3): Remove use of const_parms.
2069 (free_rpo_vn): Do not free const_parms.
2070 (do_rpo_vn): Do not call init_const_parms.
2071 * tree-ssa-alias.c (refs_may_alias_p_1): Honor
2072 SSA_NAME_POINTS_TO_READONLY_MEMORY.
2073 (call_may_clobber_ref_p_1): Likewise.
2075 2018-08-29 Alexander Monakov <amonakov@ispras.ru>
2078 * invoke.texi (Optimization Options): List -ftree-scev-cprop.
2080 (-ftree-scev-cprop): Document.
2082 2018-08-29 Jan Hubicka <jh@suse.cz>
2084 * sreal.h (normalize, normalize_up, normalize_down): Add new_sig/new_exp
2086 (sreal constructor): Update.
2087 * sreal.c (sreal:operator+, sreal:operator-, sreal:operator*,
2088 sreal:operator/): Update.
2090 2018-08-29 Martin Liska <mliska@suse.cz>
2092 * tree-switch-conversion.c (switch_conversion::expand):
2093 Strenghten assumption about gswitch statements.
2095 2018-08-29 Richard Biener <rguenther@suse.de>
2097 PR tree-optimization/87117
2098 * tree-ssa-sccvn.c (eliminate_dom_walker::eliminate_stmt): Only
2099 re-value-number released SSA VDEFs.
2101 2018-08-29 Richard Biener <rguenther@suse.de>
2103 PR tree-optimization/87126
2104 * tree-ssa-sccvn.c (vn_reference_insert): Remove assert.
2106 2018-08-28 Jim Wilson <jimw@sifive.com>
2108 * config/riscv/pic.md: Rewrite.
2109 * config/riscv/riscv.c (riscv_address_insns): Return cost of 3 for
2111 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Delete.
2112 (SOFTF, default_load, softload, softstore): New.
2114 2018-08-28 Jeff Law <law@redhat.com>
2116 * fold-const.c (fold_binary_loc): Remove recently added assert.
2118 2018-08-28 Joern Rennecke <joern.rennecke@riscy-ip.com>
2120 * genpreds.c (write_predicate_subfunction): Also add ATTRIBUTE_UNUSED
2121 to OP parmeter of generated function.
2123 2018-08-28 MCC CS <deswurstes@users.noreply.github.com>
2125 PR tree-optimization/87009
2126 * match.pd: Add boolean optimizations.
2128 2018-08-28 Martin Sebor <msebor@redhat.com>
2131 * calls.c (alloc_max_size): Treat HOST_WIDE_INT special.
2132 * gimple-ssa-warn-alloca.c (adjusted_warn_limit): New function.
2133 (pass_walloca::gate): Use it.
2134 (alloca_call_type): Same.
2135 (pass_walloca::execute): Same.
2136 * stor-layout.c (layout_decl): Treat HOST_WIDE_INT special.
2138 2018-08-28 David Malcolm <dmalcolm@redhat.com>
2140 * dumpfile.h (ATTRIBUTE_GCC_DUMP_PRINTF): Change version check on
2141 GCC_VERSION for usage of "__gcc_dump_printf__" format from
2144 2018-08-28 Richard Biener <rguenther@suse.de>
2146 PR tree-optimization/87124
2147 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Guard against
2148 constants before looking up avail.
2150 2018-08-28 Jakub Jelinek <jakub@redhat.com>
2153 * calls.c (maybe_warn_nonstring_arg): Punt early if
2154 warn_stringop_overflow is zero. Don't call get_range_strlen
2155 on 3rd argument, keep iterating until lenrng[1] is INTEGER_CST.
2156 Swap comparison operands to have constants on rhs. Only use
2157 lenrng[1] if non-NULL and INTEGER_CST. Don't uselessly
2158 increment lenrng[0].
2160 2018-08-28 Richard Sandiford <richard.sandiford@arm.com>
2162 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Fix unguarded
2163 use of tree_to_shwi. Remove duplicated test for the size being
2164 a whole number of bytes.
2166 2018-08-28 Richard Biener <rguenther@suse.de>
2168 PR tree-optimization/87117
2169 * tree-ssa-sccvn.c (eliminate_dom_walker::eliminate_cleanup):
2170 Handle removed stmt without LHS (GIMPLE_NOP).
2172 2018-08-28 Richard Biener <rguenther@suse.de>
2174 PR tree-optimization/87117
2175 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Exclude
2176 void which is is_gimple_reg_type by checking for COMPLETE_TYPE_P.
2178 2018-08-28 Richard Biener <rguenther@suse.de>
2180 PR tree-optimization/87117
2181 * tree-ssa-pre.c (compute_avail): Do not make expressions
2182 with predicated values available.
2183 (get_expr_value_id): Assert we do not run into predicated value
2186 2018-08-28 Richard Biener <rguenther@suse.de>
2188 PR tree-optimization/87117
2189 * tree-ssa-operands.c (add_stmt_operand): STRING_CST may
2190 get virtual operands.
2191 (get_expr_operands): Handle STRING_CST like other decls.
2193 2018-08-28 Martin Liska <mliska@suse.cz>
2195 * tree.h: Update documentation of fndecl_built_in_p
2199 2018-08-27 Jeff Law <law@redhat.com>
2200 PR tree-optimization/87110
2201 * tree-ssa-dse.c (compute_trims): Handle non-constant
2204 2018-08-27 Martin Sebor <msebor@redhat.com>
2206 PR tree-optimization/86914
2207 * tree-ssa-strlen.c (maybe_set_strlen_range): Avoid MEM_REF.
2209 2018-08-27 Martin Sebor <msebor@redhat.com>
2211 PR tree-optimization/87112
2212 * builtins.c (expand_builtin_strnlen): Convert c_strlen result to
2213 the type of the bound argument.
2215 2018-08-27 Jeff Law <law@redhat.com>
2217 * tree-ssa-dse.c (compute_trims): Handle case where the reference's
2218 type does not have a TYPE_SIZE_UNIT.
2220 2018-08-27 Steve Ellcey <sellcey@cavium.com>
2222 * config/aarch64/aarch64-speculation.cc: Replace include of cfg.h
2223 with include of backend.h.
2225 2018-08-27 Richard Biener <rguenther@suse.de>
2227 PR tree-optimization/86927
2228 * tree-vect-loop.c (vect_create_epilog_for_reduction): Properly
2229 use const cond reduction code.
2231 2018-08-27 Alexander Monakov <amonakov@ispras.ru>
2233 PR tree-optimization/85758
2234 * match.pd ((X & Y) ^ Y): Add :s qualifier to inner expression.
2236 2018-08-27 David Malcolm <dmalcolm@redhat.com>
2239 * diagnostic-show-locus.c (class layout_range): Update for
2240 conversion of show_caret_p to a tri-state.
2241 (layout_range::layout_range): Likewise.
2242 (make_range): Likewise.
2243 (layout::maybe_add_location_range): Likewise.
2244 (layout::should_print_annotation_line_p): Don't show annotation
2245 lines for ranges that are SHOW_LINES_WITHOUT_RANGE.
2246 (layout::get_state_at_point): Update for conversion of
2247 show_caret_p to a tri-state. Bail out early for
2248 SHOW_LINES_WITHOUT_RANGE, so that such ranges don't affect
2249 underlining or source colorization.
2250 (gcc_rich_location::add_location_if_nearby): Update for conversion
2251 of show_caret_p to a tri-state.
2252 (selftest::test_one_liner_multiple_carets_and_ranges): Likewise.
2253 (selftest::test_one_liner_fixit_replace_equal_secondary_range):
2255 (selftest::test_one_liner_labels): Likewise.
2256 * gcc-rich-location.c (gcc_rich_location::add_expr): Update for
2257 conversion of show_caret_p to a tri-state.
2258 * pretty-print.c (text_info::set_location): Likewise.
2259 * pretty-print.h (text_info::set_location): Likewise.
2260 * substring-locations.c (format_warning_n_va): Likewise.
2261 * tree-diagnostic.c (default_tree_printer): Likewise.
2262 * tree-pretty-print.c (newline_and_indent): Likewise.
2264 2018-08-27 David Malcolm <dmalcolm@redhat.com>
2267 * diagnostic-show-locus.c (get_line_span_for_fixit_hint): Show the
2268 line above for line-insertion fix-it hints.
2269 (selftest::test_fixit_insert_containing_newline): Update the
2270 expected results, and add a test with line-numbering enabled.
2272 2018-08-27 Martin Liska <mliska@suse.cz>
2275 * sanopt.c (sanitize_rewrite_addressable_params): Ignore
2276 params with DECL_HAS_VALUE_EXPR_P.
2278 2018-08-27 Martin Liska <mliska@suse.cz>
2280 * config/i386/i386.c (ix86_expand_set_or_movmem): Dump
2281 selected expansion strategy.
2283 2018-08-27 Martin Liska <mliska@suse.cz>
2285 * builtins.h (is_builtin_fn): Remove and fndecl_built_in_p.
2286 * builtins.c (is_builtin_fn): Likewise.
2287 * attribs.c (diag_attr_exclusions): Use new function
2288 fndecl_built_in_p and remove check for FUNCTION_DECL if
2290 (builtin_mathfn_code): Likewise.
2291 (fold_builtin_expect): Likewise.
2292 (fold_call_expr): Likewise.
2293 (fold_builtin_call_array): Likewise.
2294 (fold_call_stmt): Likewise.
2295 (set_builtin_user_assembler_name): Likewise.
2296 (is_simple_builtin): Likewise.
2297 * calls.c (gimple_alloca_call_p): Likewise.
2298 (maybe_warn_nonstring_arg): Likewise.
2299 * cfgexpand.c (expand_call_stmt): Likewise.
2300 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Likewise.
2301 (cgraph_edge::verify_corresponds_to_fndecl): Likewise.
2302 (cgraph_node::verify_node): Likewise.
2303 * cgraphclones.c (build_function_decl_skip_args): Likewise.
2304 (cgraph_node::create_clone): Likewise.
2305 * config/arm/arm.c (arm_insert_attributes): Likewise.
2306 * config/i386/i386.c (ix86_gimple_fold_builtin): Likewise.
2307 * dse.c (scan_insn): Likewise.
2308 * expr.c (expand_expr_real_1): Likewise.
2309 * fold-const.c (operand_equal_p): Likewise.
2310 (fold_binary_loc): Likewise.
2311 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
2312 * gimple-low.c (lower_stmt): Likewise.
2313 * gimple-pretty-print.c (dump_gimple_call): Likewise.
2314 * gimple-ssa-warn-restrict.c (wrestrict_dom_walker::check_call):
2316 * gimple.c (gimple_build_call_from_tree): Likewise.
2317 (gimple_call_builtin_p): Likewise.
2318 (gimple_call_combined_fn): Likewise.
2319 * gimplify.c (gimplify_call_expr): Likewise.
2320 (gimple_boolify): Likewise.
2321 (gimplify_modify_expr): Likewise.
2322 (gimplify_addr_expr): Likewise.
2323 * hsa-gen.c (gen_hsa_insns_for_call): Likewise.
2324 * ipa-cp.c (determine_versionability): Likewise.
2325 * ipa-fnsummary.c (compute_fn_summary): Likewise.
2326 * ipa-param-manipulation.c (ipa_modify_formal_parameters): Likewise.
2327 * ipa-split.c (visit_bb): Likewise.
2328 (split_function): Likewise.
2329 * ipa-visibility.c (cgraph_externally_visible_p): Likewise.
2330 * lto-cgraph.c (input_node): Likewise.
2331 * lto-streamer-out.c (write_symbol): Likewise.
2332 * omp-low.c (setjmp_or_longjmp_p): Likewise.
2333 (lower_omp_1): Likewise.
2334 * predict.c (strip_predict_hints): Likewise.
2335 * print-tree.c (print_node): Likewise.
2336 * symtab.c (symtab_node::output_to_lto_symbol_table_p): Likewise.
2337 * trans-mem.c (is_tm_irrevocable): Likewise.
2338 (is_tm_load): Likewise.
2339 (is_tm_simple_load): Likewise.
2340 (is_tm_store): Likewise.
2341 (is_tm_simple_store): Likewise.
2342 (is_tm_abort): Likewise.
2343 (tm_region_init_1): Likewise.
2344 * tree-call-cdce.c (gen_shrink_wrap_conditions): Likewise.
2345 * tree-cfg.c (verify_gimple_call): Likewise.
2346 (move_stmt_r): Likewise.
2347 (stmt_can_terminate_bb_p): Likewise.
2348 * tree-eh.c (lower_eh_constructs_2): Likewise.
2349 * tree-if-conv.c (if_convertible_stmt_p): Likewise.
2350 * tree-inline.c (remap_gimple_stmt): Likewise.
2351 (copy_bb): Likewise.
2352 (estimate_num_insns): Likewise.
2353 (fold_marked_statements): Likewise.
2354 * tree-sra.c (scan_function): Likewise.
2355 * tree-ssa-ccp.c (surely_varying_stmt_p): Likewise.
2356 (optimize_stack_restore): Likewise.
2357 (pass_fold_builtins::execute): Likewise.
2358 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
2359 (mark_all_reaching_defs_necessary_1): Likewise.
2360 * tree-ssa-dom.c (dom_opt_dom_walker::optimize_stmt): Likewise.
2361 * tree-ssa-forwprop.c (simplify_builtin_call): Likewise.
2362 (pass_forwprop::execute): Likewise.
2363 * tree-ssa-loop-im.c (stmt_cost): Likewise.
2364 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
2365 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Likewise.
2366 * tree-ssa-strlen.c (get_string_length): Likewise.
2367 * tree-ssa-structalias.c (handle_lhs_call): Likewise.
2368 (find_func_aliases_for_call): Likewise.
2369 * tree-ssa-ter.c (find_replaceable_in_bb): Likewise.
2370 * tree-stdarg.c (optimize_va_list_gpr_fpr_size): Likewise.
2371 * tree-tailcall.c (find_tail_calls): Likewise.
2372 * tree.c (need_assembler_name_p): Likewise.
2373 (free_lang_data_in_decl): Likewise.
2374 (get_call_combined_fn): Likewise.
2375 * ubsan.c (is_ubsan_builtin_p): Likewise.
2376 * varasm.c (incorporeal_function_p): Likewise.
2377 * tree.h (DECL_BUILT_IN): Remove and replace with
2379 (DECL_BUILT_IN_P): Transfort to fndecl_built_in_p.
2380 (fndecl_built_in_p): New.
2382 2018-08-27 Martin Liska <mliska@suse.cz>
2384 PR tree-optimization/86847
2385 * tree-switch-conversion.c (switch_decision_tree::dump_case_nodes):
2386 Dump also subtree probability.
2387 (switch_decision_tree::do_jump_if_equal): New function.
2388 (switch_decision_tree::emit_case_nodes): Handle special
2389 situations in balanced tree that can be emitted much simpler.
2390 Fix calculation of probabilities that happen in tree expansion.
2391 * tree-switch-conversion.h (struct cluster): Add
2393 (struct simple_cluster): Likewise.
2394 (struct case_tree_node): Add new function has_child.
2395 (do_jump_if_equal): New.
2397 2018-08-27 Martin Liska <mliska@suse.cz>
2399 * tree-switch-conversion.c (bit_test_cluster::find_bit_tests):
2400 Add new argument to bit_test_cluster constructor.
2401 (bit_test_cluster::emit): Set bits really number of values
2403 (bit_test_cluster::hoist_edge_and_branch_if_true): Add
2404 probability argument.
2405 * tree-switch-conversion.h (struct bit_test_cluster):
2406 Add m_handles_entire_switch.
2408 2018-08-27 Martin Liska <mliska@suse.cz>
2410 PR tree-optimization/86702
2411 * tree-switch-conversion.c (jump_table_cluster::emit):
2412 Make probabilities even for values in jump table
2413 according to number of cases handled.
2414 (switch_decision_tree::compute_cases_per_edge): Pass
2415 argument to reset_out_edges_aux function.
2416 (switch_decision_tree::analyze_switch_statement): Likewise.
2417 * tree-switch-conversion.h (switch_decision_tree::reset_out_edges_aux):
2420 2018-08-27 Martin Liska <mliska@suse.cz>
2422 * cfgexpand.c (expand_asm_stmt): Use label_to_block and pass
2423 cfun argument explicitly.
2424 * gimple-pretty-print.c (dump_gimple_switch): Likewise.
2425 * hsa-gen.c (gen_hsa_insns_for_switch_stmt): Use new
2426 function gimple_switch_default_bb.
2427 (convert_switch_statements):
2429 * ipa-fnsummary.c (set_switch_stmt_execution_predicate):
2430 * stmt.c (label_to_block_fn): Use label_to_block and pass
2431 cfun argument explicitly and use gimple_switch_label_bb.
2432 (expand_case): Likewise.
2433 * tree-cfg.c (lower_phi_internal_fn): Use label_to_block and pass
2434 cfun argument explicitly. Likewise.
2435 (make_edges_bb): Likewise.
2436 (make_cond_expr_edges): Likewise.
2437 (get_cases_for_edge): Likewise.
2438 (make_gimple_switch_edges): Likewise.
2439 (label_to_block_fn): Likewise.
2440 (label_to_block): Likewise.
2441 (make_goto_expr_edges): Likewise.
2442 (make_gimple_asm_edges): Likewise.
2443 (main_block_label): Likewise.
2444 (group_case_labels_stmt): Likewise.
2445 (find_taken_edge_computed_goto): Likewise.
2446 (find_taken_edge_switch_expr): Likewise.
2447 (gimple_verify_flow_info): Likewise.
2448 (gimple_redirect_edge_and_branch): Likewise.
2449 (gimple_switch_label_bb): New function.
2450 (gimple_switch_default_bb): Likewise.
2451 (gimple_switch_edge): Likewise.
2452 (gimple_switch_default_edge): Likewise.
2453 * tree-cfg.h (label_to_block_fn): Remove and replace ...
2454 (label_to_block): ... with this.
2455 (gimple_switch_label_bb): New.
2456 (gimple_switch_default_bb): Likewise.
2457 (gimple_switch_edge): Likewise.
2458 (gimple_switch_default_edge): Likewise.
2459 * tree-cfgcleanup.c (convert_single_case_switch): Use
2460 new gimple functions and pass new argument to label_to_block.
2461 (cleanup_control_flow_bb):
2462 * tree-eh.c (make_eh_dispatch_edges): Use label_to_block and pass
2463 cfun argument explicitly.
2464 (make_eh_edges): Likewise.
2465 (redirect_eh_dispatch_edge): Likewise.
2466 (lower_resx): Likewise.
2467 (lower_eh_dispatch): Likewise.
2468 (maybe_remove_unreachable_handlers): Likewise.
2469 (unsplit_eh): Likewise.
2470 (cleanup_empty_eh): Likewise.
2471 (verify_eh_edges): Likewise.
2472 (verify_eh_dispatch_edge): Likewise.
2473 * tree-ssa-dom.c (record_edge_info): Likewise.
2474 * tree-ssa-forwprop.c (simplify_gimple_switch_label_vec): Likewise.
2475 * tree-ssa-threadedge.c (thread_around_empty_blocks): Likewise.
2476 (thread_through_normal_block): Likewise.
2477 * tree-ssa-uncprop.c (associate_equivalences_with_edges): Likewise.
2478 * tree-ssa-uninit.c (convert_control_dep_chain_into_preds):
2479 * tree-switch-conversion.c (switch_conversion::collect): Use new
2481 (switch_conversion::check_final_bb): Likewise.
2482 (switch_conversion::gather_default_values): Pass new argument
2484 (switch_conversion::build_constructors): Likewise.
2485 (switch_decision_tree::compute_cases_per_edge): Use new
2486 gimple_switch_edge function.
2487 (switch_decision_tree::analyze_switch_statement): Pass new argument
2489 (switch_decision_tree::try_switch_expansion): Use
2490 gimple_switch_default_edge.
2491 * tree-vrp.c (find_switch_asserts): Pass new argument
2493 * vr-values.c (vr_values::vrp_visit_switch_stmt): Likewise.
2494 (vr_values::simplify_switch_using_ranges): Likewise.
2496 2018-08-27 Richard Biener <rguenther@suse.de>
2498 * cfganal.h (rev_post_order_and_mark_dfs_back_seme): Declare.
2499 * cfganal.c (rev_post_order_and_mark_dfs_back_seme): New function.
2501 * tree-ssa-sccvn.h (struct vn_pval): New structure.
2502 (struct vn_nary_op_s): Add unwind_to member. Add
2503 predicated_values flag and put result into a union together
2504 with a linked list of vn_pval.
2505 (struct vn_ssa_aux): Add name member to make maintaining
2506 a map of SSA name to vn_ssa_aux possible. Remove no longer
2507 needed info, dfsnum, low, visited, on_sccstack, use_processed
2508 and range_info_anti_range_p members.
2509 (run_scc_vn, vn_eliminate, free_scc_vn, vn_valueize): Remove.
2510 (do_rpo_vn, run_rpo_vn, eliminate_with_rpo_vn, free_rpo_vn):
2512 (vn_valueize): New global.
2513 (vn_context_bb): Likewise.
2514 (VN_INFO_RANGE_INFO, VN_INFO_ANTI_RANGE_P, VN_INFO_RANGE_TYPE,
2515 VN_INFO_PTR_INFO): Remove.
2516 * tree-ssa-sccvn.c: ... (rewrite)
2517 (pass_fre::execute): For -O2+ initialize loops and run
2518 RPO VN in optimistic mode (iterating). For -O1 and -Og
2519 run RPO VN in non-optimistic mode.
2520 * params.def (PARAM_SCCVN_MAX_SCC_SIZE): Remove.
2521 (PARAM_RPO_VN_MAX_LOOP_DEPTH): Add.
2522 * doc/invoke.texi (sccvn-max-scc-size): Remove.
2523 (rpo-vn-max-loop-depth): Document.
2524 * tree-ssa-alias.c (walk_non_aliased_vuses): Stop walking
2525 when valuezing the VUSE signals we walked out of the region.
2526 * tree-ssa-pre.c (phi_translate_1): Ignore predicated values.
2527 (phi_translate): Set VN context block to use for availability
2529 (compute_avail): Likewise.
2530 (pre_valueize): New function.
2531 (pass_pre::execute): Adjust to the RPO VN API.
2533 * tree-ssa-loop-ivcanon.c: Include tree-ssa-sccvn.h.
2534 (propagate_constants_for_unrolling): Remove.
2535 (tree_unroll_loops_completely): Perform value-numbering
2536 on the unrolled bodies loop parent.
2538 2018-08-27 Richard Biener <rguenther@suse.de>
2540 * tree-ssa-pre.c (compute_antic): Re-use inverted postorder
2541 for partial antic compute.
2543 2018-08-27 Jakub Jelinek <jakub@redhat.com>
2545 PR rtl-optimization/87065
2546 * combine.c (simplify_if_then_else): Formatting fix.
2547 (if_then_else_cond): Guard MULT optimization with SCALAR_INT_MODE_P
2549 (known_cond): Don't return const_true_rtx for vector modes. Use
2550 CONST0_RTX instead of const0_rtx. Formatting fixes.
2552 2018-08-27 Martin Liska <mliska@suse.cz>
2554 PR gcov-profile/87069
2555 * gcov.c (process_file): Record files already processed
2556 and warn about a file being processed multiple times.
2558 2018-08-27 Martin Liska <mliska@suse.cz>
2561 * config/aarch64/aarch64.c (aarch64_override_options_internal):
2562 Set default values for x_aarch64_*_string strings.
2563 * config/aarch64/aarch64.opt: Remove --{march,mcpu,mtune}==
2564 prefix. For -mabi do not print '=ABI' in help and use
2565 <option_value> format for -msve-vector-bits and -moverride
2568 2018-08-26 Jeff Law <law@redhat.com>
2570 * config/mips/frame-header-opt.c: Include "backend.h" rather than
2573 2018-08-26 Marek Polacek <polacek@redhat.com>
2575 PR c++/87029, Implement -Wredundant-move.
2576 * doc/invoke.texi: Document -Wredundant-move.
2578 2018-08-25 Martin Sebor <msebor@redhat.com>
2580 PR tree-optimization/87059
2581 * builtins.c (expand_builtin_strncmp): Convert MIN_EXPR operand
2582 to the same type as the other.
2583 * fold-const.c (fold_binary_loc): Assert expectation.
2585 2018-08-25 Iain Sandoe <iain@sandoe.co.uk>
2587 * config/darwin.c (machopic_legitimize_pic_address): Clean up
2588 extraneous parentheses, dead code section and formatting.
2590 2018-08-24 David Malcolm <dmalcolm@redhat.com>
2593 * diagnostic-show-locus.c (layout::layout): Ensure the margin is
2594 wide enough for jumps in the line-numbering to be visible.
2595 (layout::print_gap_in_line_numbering): New member function.
2596 (layout::calculate_line_spans): When using line numbering, merge
2597 line spans that are only 1 line apart.
2598 (diagnostic_show_locus): When printing line numbers, show gaps in
2599 line numbering directly, rather than printing headers.
2600 (selftest::test_diagnostic_show_locus_fixit_lines): Add test of
2601 line-numbering with multiple line spans.
2602 (selftest::test_fixit_insert_containing_newline_2): Add test of
2603 line-numbering, in which the spans are close enough to be merged.
2605 2018-08-24 Aldy Hernandez <aldyh@redhat.com>
2607 * gimple-ssa-evrp-analyze.c (set_ssa_range_info): Pass value_range
2608 to range_includes_zero_p. Do not special case VR_ANTI_RANGE.
2609 * tree-vrp.c (range_is_nonnull): Remove.
2610 (range_includes_zero_p): Accept value_range instead of min/max.
2611 (extract_range_from_binary_expr_1): Do not early bail on
2613 Use range_includes_zero_p instead of range_is_nonnull.
2614 (extract_range_from_unary_expr): Use range_includes_zero_p instead
2615 of range_is_nonnull.
2616 (vrp_meet_1): Pass value_range to range_includes_zero_p. Do not
2617 special case VR_ANTI_RANGE.
2618 (vrp_finalize): Same.
2619 * tree-vrp.h (range_includes_zero_p): Pass value_range as argument
2621 (range_is_nonnull): Remove.
2622 * vr-values.c (vrp_stmt_computes_nonzero): Use
2623 range_includes_zero_p instead of range_is_nonnull.
2624 (extract_range_basic): Pass value_range to range_includes_zero_p
2625 instead of range_is_nonnull.
2627 2018-08-24 Uros Bizjak <ubizjak@gmail.com>
2629 * emit-rtl.c (init_emit_once): Do not emit MODE_POINTER_BOUNDS RTXes.
2630 * emit-rtl.h (rtl_data): Remove return_bnd.
2631 * explow.c (trunc_int_for_mode): Do not handle POINTER_BOUNDS_MODE_P.
2632 * function.c (diddle_return_value): Do not handle crtl->return_bnd.
2633 * genmodes.c (complete_mode): Do not handle MODE_POINTER_BOUNDS.
2634 (POINTER_BOUNDS_MODE): Remove definition.
2635 (make_pointer_bounds_mode): Remove.
2636 (get_mode_class): Do not handle MODE_POINTER_BOUNDS.
2637 * machmode.h (POINTER_BOUNDS_MODE_P): Remove definition.
2638 (scalare_mode::includes_p): Do not handle MODE_POINTER_BOUNDS.
2639 * mode-classes.def: Do not define MODE_POINTER_BOUNDS.
2640 * stor-layout.c (int_mode_for_mode): Do not handle MODE_POINTER_BOUNDS.
2641 * tree-core.h (enum tree_index): Remove TI_POINTER_BOUNDS_TYPE.
2642 * varasm.c (output_constant_pool_2): Do not handle MODE_POINTER_BOUNDS.
2644 * config/i386/i386-modes.def (BND32, BND64): Remove.
2645 * config/i386/i386.c (dbx_register_map): Remove bound registers.
2646 (dbx64_register_map): Ditto.
2647 (svr4_dbx_register_map): Ditto.
2648 (indirect_thunk_bnd_needed): Remove.
2649 (indirect_thunks_bnd_used): Ditto.
2650 (indirect_return_bnd_needed): Ditto.
2651 (indirect_return_via_cx_bnd): Ditto.
2652 (enum indirect_thunk_prefix): Remove indirect_thunk_prefix_bnd.
2653 (indirect_thunk_name): Remove handling of indirect_thunk_prefix_bnd.
2654 (output_indirect_thunk): Ditto. Remove need_prefix argument.
2655 (output_indirect_thunk_function): Remove handling of
2656 indirect_return_bnd_needed, indirect_return_via_cx_bnd,
2657 indirect_thunk_bnd_needed and indirect_thunks_bnd_used variables.
2658 (ix86_save_reg): Remove handling of crtl->return_bnd.
2659 (ix86_legitimate_constant_p): Remove handling of POINTER_BOUNDS_MODE_P.
2660 (ix86_print_operand_address_as): Remove handling of UNSPEC_BNDMK_ADDR
2661 and UNSPEC_BNDLX_ADDR.
2662 (ix86_output_indirect_branch_via_reg): Remove handling of
2663 indirect_thunk_prefix_bnd.
2664 (ix86_output_indirect_branch_via_push): Ditto.
2665 (ix86_output_function_return): Ditto.
2666 (ix86_output_indirect_function_return): Ditto.
2667 (avoid_func_arg_motion): Do not handle UNSPEC_BNDSTX.
2668 * config/i386/i386.h (FIXED_REGISTERS): Remove bound registers.
2669 (CALL_USED_REGISTERS): Ditto.
2670 (REG_ALLOC_ORDER): Update for removal of bound registers.
2671 (HI_REGISTER_NAMES): Ditto.
2672 * config/i386/i386.md (UNSPEC_BNDMK, UNSPEC_BNDMK_ADDR, UNSPEC_BNDSTX)
2673 (UNSPEC_BNDLDX, UNSPEC_BNDLDX_ADDR, UNSPEC_BNDCL, UNSPEC_BNDCU)
2674 (UNSPEC_BNDCN, UNSPEC_MPX_FENCE): Remove.
2675 (BND0_REG, BND1_REG, BND2_REG, BND3_REG): Remove
2676 (FIRST_PSEUDO_REG): Update.
2677 (BND): Remove mode iterator.
2678 * config/i386/predicates.md (bnd_mem_operator): Remove.
2680 2018-08-24 Richard Sandiford <richard.sandiford@arm.com>
2682 * tree-vect-stmts.c (vectorizable_bswap): Handle variable-length
2685 2018-08-24 Richard Sandiford <richard.sandiford@arm.com>
2687 * tree-vect-slp.c (vect_transform_slp_perm_load): Separate out
2688 the case in which the permute needs only a single element and
2689 repeats for every vector of the result. Extend that case to
2690 handle variable-length vectors.
2691 * tree-vect-stmts.c (vectorizable_load): Update accordingly.
2693 2018-08-24 H.J. Lu <hongjiu.lu@intel.com>
2696 * dwarf2out.c (save_macinfo_strings): Call set_indirect_string
2697 on DW_MACINFO_start_file for -gsplit-dwarf -g3.
2699 2018-08-24 Richard Biener <rguenther@suse.de>
2701 * cfg.h (struct control_flow_graph): Add edge_flags_allocated and
2702 bb_flags_allocated members.
2703 (auto_flag): New RAII class for allocating flags.
2704 (auto_edge_flag): New RAII class for allocating edge flags.
2705 (auto_bb_flag): New RAII class for allocating bb flags.
2706 * cfgloop.c (verify_loop_structure): Allocate temporary edge
2708 * cfganal.c (dfs_enumerate_from): Remove use of visited sbitmap
2709 in favor of temporarily allocated BB flag.
2710 * hsa-brig.c: Re-order includes.
2711 * hsa-dump.c: Likewise.
2712 * hsa-regalloc.c: Likewise.
2713 * print-rtl.c: Likewise.
2714 * profile-count.c: Likewise.
2716 2018-08-24 Segher Boessenkool <segher@kernel.crashing.org>
2719 * config/rs6000/rs6000.c (toc_relative_expr_p): Check that the base is
2722 2018-08-24 Aldy Hernandez <aldyh@redhat.com>
2725 * wide-int-range.cc (wide_int_range_div): Do not ignore result
2726 from wide_int_range_multiplicative_op.
2728 2018-08-23 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
2730 * tree-vect-data-refs.c (vect_grouped_store_supported): Fix typo
2733 2018-08-23 Giuliano Belinassi <giuliano.belinassi@usp.br>
2735 * genmatch.c (parser::parse_operation): Fix typo 'exapnded'
2738 2018-08-23 Alexander Monakov <amonakov@ispras.ru>
2740 * tree-scalar-evolution.c (final_value_replacement_loop): Dump
2741 full GENERIC expression used for replacement.
2743 2018-08-23 Aldy Hernandez <aldyh@redhat.com>
2745 * tree-vrp.c (abs_extent_range): Remove.
2746 (extract_range_into_wide_ints): Pass wide ints by reference.
2747 (extract_range_from_binary_expr_1): Rewrite the *DIV_EXPR code.
2748 Pass wide ints by reference in all calls to
2749 extract_range_into_wide_ints.
2750 * wide-int-range.cc (wide_int_range_div): New.
2751 * wide-int-range.h (wide_int_range_div): New.
2752 (wide_int_range_includes_zero_p): New.
2753 (wide_int_range_zero_p): New.
2755 2018-08-23 Matthew Malcomson <matthew.malcomson@arm.com>
2757 * config/aarch64/aarch64.md (arches): New enum.
2758 (arch): New enum attr.
2759 (arch_enabled): New attr.
2760 (enabled): Now uses arch_enabled only.
2761 (simd, sve, fp16): Removed attribute.
2762 (fp): Attr now defined in terms of 'arch'.
2763 (*mov<mode>_aarch64, *movsi_aarch64, *movdi_aarch64, *movti_aarch64,
2764 *movhf_aarch64, <optab><fcvt_target><GPF:mode>2,
2765 <FCVT_F2FIXED:fcvt_fixed_insn><GPF:mode>3,
2766 <FCVT_FIXED2F:fcvt_fixed_insn><GPI:mode>3): Merge 'fp' and 'simd'
2767 attributes into 'arch'.
2768 (*movsf_aarch64, *movdf_aarch64, *movtf_aarch64, *add<mode>3_aarch64,
2769 subdi3, neg<mode>2, <optab><mode>3, one_cmpl<mode>2,
2770 *<NLOGICAL:optab>_one_cmpl<mode>3, *xor_one_cmpl<mode>3,
2771 *aarch64_ashl_sisd_or_int_<mode>3, *aarch64_lshr_sisd_or_int_<mode>3,
2772 *aarch64_ashr_sisd_or_int_<mode>3, *aarch64_sisd_ushl): Convert use of
2773 'simd' attribute into 'arch'.
2774 (load_pair_sw_<SX:mode><SX2:mode>, load_pair_dw_<DX:mode><DX2:mode>,
2775 store_pair_sw_<SX:mode><SX2:mode>, store_pair_dw_<DX:mode><DX2:mode>):
2776 Convert use of 'fp' attribute to 'arch'.
2777 * config/aarch64/aarch64-simd.md (move_lo_quad_internal_<mode>,
2778 move_lo_quad_internal_<mode>): (different modes) Merge 'fp' and 'simd'
2780 (move_lo_quad_internal_be_<mode>, move_lo_quad_internal_be_<mode>):
2781 (different modes) Merge 'fp' and 'simd' into 'arch'.
2782 (*aarch64_combinez<mode>, *aarch64_combinez_be<mode>): Merge 'fp' and
2785 2018-08-23 Segher Boessenkool <segher@kernel.crashing.org>
2787 PR rtl-optimization/87026
2788 * expmed.c (canonicalize_comparison): If we can no longer create
2789 pseudoregisters, don't.
2791 2018-08-23 Richard Earnshaw <rearnsha@arm.com>
2794 * config/arm/arm-protos.h (arm_emit_speculation_barrier): New
2796 * config/arm/arm.c (speculation_barrier_libfunc): New static
2798 (arm_init_libfuncs): Initialize it.
2799 (arm_emit_speculation_barrier): New function.
2800 * config/arm/arm.md (speculation_barrier): Call
2801 arm_emit_speculation_barrier for architectures that do not have
2803 (speculation_barrier_insn): Only match on Armv7 or later.
2805 2018-08-23 Richard Biener <rguenther@suse.de>
2808 * tree-inline.c (copy_bb): Drop unused __builtin_va_arg_pack_len
2811 2018-08-23 Richard Sandiford <richard.sandiford@arm.com>
2813 * config/aarch64/aarch64.c (aarch64_evpc_sve_tbl): Fix handling
2814 of single-vector TBLs.
2815 (aarch64_vectorize_vec_perm_const): Set one_vector_p when only
2818 2018-08-23 Richard Sandiford <richard.sandiford@arm.com>
2821 * config/aarch64/aarch64.c (aarch64_expand_vec_perm_const_1): Fix
2822 aarch64_evpc_tbl guard.
2824 2018-08-22 Bernd Edlinger <bernd.edlinger@hotmail.de>
2826 * tree-ssa-dse.c (compute_trims): Avoid folding away undefined
2829 2018-08-22 Martin Sebor <msebor@redhat.com>
2832 * tree-pretty-print.c (pretty_print_string): Add argument.
2833 (dump_generic_node): Call to pretty_print_string with string size.
2835 2018-08-22 Segher Boessenkool <segher@kernel.crashing.org>
2837 PR rtl-optimization/86771
2838 * combine.c (try_combine): Do not allow splitting a resulting PARALLEL
2839 of two SETs into those two SETs, one to be placed at i2, if that SETs
2840 destination is modified between i2 and i3.
2842 2018-08-22 Richard Sandiford <richard.sandiford@arm.com>
2844 PR tree-optimization/86725
2845 * tree-vect-loop.c (vect_inner_phi_in_double_reduction_p): New
2847 (vect_analyze_scalar_cycles_1): Check it.
2849 2018-08-22 Richard Sandiford <richard.sandiford@arm.com>
2851 PR tree-optimization/86725
2852 * tree-vect-loop.c (vect_is_simple_reduction): When treating
2853 an outer loop phi as a double reduction, make sure that the
2854 single user of the phi result is an inner loop phi.
2856 2018-08-22 Richard Sandiford <richard.sandiford@arm.com>
2858 * tree-vect-data-refs.c (vect_analyze_group_access_1): Convert
2859 grouped stores with gaps to a strided group.
2861 2018-08-22 Richard Sandiford <richard.sandiford@arm.com>
2863 * tree-vect-stmts.c (get_group_load_store_type)
2864 (get_load_store_type): Only test STMT_VINFO_STRIDED_P for the
2865 first statement in a group.
2867 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
2869 * config/darwin.h (LINK_COMMAND_SPEC_A): Sync LTO options with
2870 the sequence used in gcc/gcc.c.
2872 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
2875 * gcc-ar.c (main): Don’t try to invoke the plug-in if we’re not
2878 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
2880 * config/darwin10.h (LINK_GCC_C_SEQUENCE_SPEC): Adjust to use the
2881 Darwin10-specific unwinder-shim.
2882 * config/darwin12.h (LINK_GCC_C_SEQUENCE_SPEC): Remove.
2883 * config/rs6000/darwin.h (DARWIN_CRT1_SPEC, DARWIN_DYLIB1_SPEC):
2884 New to cater for Darwin10 Rosetta.
2886 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
2888 * config/i386/i386.c (ix86_output_addr_diff_elt): Move the MACH-O
2889 specific test before the one for HAVE_AS_GOTOFF_IN_DATA.
2891 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
2896 * gcc/dwarf2out.c (FUNC_SECOND_SECT_LABEL): New.
2897 (dwarf2out_switch_text_section): Generate a local label for the
2898 second function sub-section and apply it as the second FDE start
2900 * gcc/final.c (final_scan_insn_1): Emit second FDE label after the
2901 second sub-section start.
2903 2018-08-22 Richard Biener <rguenther@suse.de>
2905 PR tree-optimization/86988
2906 * tree-vrp.c (vrp_prop::check_mem_ref): Bail out on VLAs.
2908 2018-08-22 Richard Biener <rguenther@suse.de>
2910 PR tree-optimization/86945
2911 * tree-cfg.c (generate_range_test): Use unsigned arithmetic.
2913 2018-08-22 Alexandre Oliva <oliva@adacore.com>
2915 * config/rs6000/rs6000.c (SMALL_DATA_RELOC, SMALL_DATA_REG): Add
2916 a comment about how uses of r2 for .sdata2 come about.
2918 2018-08-22 Alexandre Oliva <aoliva@redhat.com>
2920 * tree-ssa-reassoc.c (is_reassociable_op): Fix cut&pasto.
2922 2018-08-21 Marek Polacek <polacek@redhat.com>
2924 PR c++/86981, Implement -Wpessimizing-move.
2925 * doc/invoke.texi: Document -Wpessimizing-move.
2927 2018-08-21 Jan Hubicka <jh@suse.cz>
2929 * tree.c (find_decls_types_r): Do not check for redundant typedefs.
2930 * tree.h (is_redundant_typedef): Remove.
2931 * dwarf2out.c (is_redundant_typedef): Turn into static function.
2933 2018-08-21 Jan Hubicka <jh@suse.cz>
2935 * tree.c (free_lang_data_in_decl): Remove types from DECL_CONTEXT
2938 2018-08-21 Tamar Christina <tamar.christina@arm.com>
2940 * expmed.c (extract_low_bits): Reject invalid subregs early.
2942 2018-08-21 Bernd Edlinger <bernd.edlinger@hotmail.de>
2945 * tree-ssa-strlen.c (adjust_last_stmt): Avoid folding away undefined
2948 2018-08-21 Rasmus Villemoes <rv@rasmusvillemoes.dk>
2950 * config/vxworks.h: Guard vxworks_asm_out_constructor and
2951 vxworks_asm_out_destructor by !HAVE_INITFINI_ARRAY_SUPPORT
2952 * config/vxworks.c: Likewise.
2954 2018-08-21 Rasmus Villemoes <rv@rasmusvillemoes.dk>
2956 * config/vxworks.c: Set targetm.have_ctors_dtors
2957 if HAVE_INITFINI_ARRAY_SUPPORT.
2958 * config/vxworks.h: Set SUPPORTS_INIT_PRIORITY
2959 if HAVE_INITFINI_ARRAY_SUPPORT.
2961 2018-08-21 Rasmus Villemoes <rv@rasmusvillemoes.dk>
2963 * config/vxworks.h: Add $(WIND_BASE)/target/h/wrn/coreip to
2964 default search path for VxWorks < 7.
2966 2018-08-21 Bernd Edlinger <bernd.edlinger@hotmail.de>
2968 * gimple-ssa-sprintf.c (decl_constant_value): Remove.
2969 (get_format_string): Refer to c_getstr.
2971 2018-08-21 Tom de Vries <tdevries@suse.de>
2973 * cgraph.h (debuginfo_early_init, debuginfo_init, debuginfo_fini)
2974 (debuginfo_start, debuginfo_stop, debuginfo_early_start)
2975 (debuginfo_early_stop): Declare.
2976 * cgraphunit.c (debuginfo_early_init, debuginfo_init, debuginfo_fini)
2977 (debuginfo_start, debuginfo_stop, debuginfo_early_start)
2978 (debuginfo_early_stop): New function.
2979 (symbol_table::finalize_compilation_unit): Call debuginfo_early_start
2980 and debuginfo_early_stop.
2981 * dwarf2out.c (dwarf2out_finish, dwarf2out_early_finish): Dump dwarf.
2982 * toplev.c (compile_file): Call debuginfo_start and debuginfo_stop.
2983 (general_init): Call debuginfo_early_init.
2984 (finalize): Call debuginfo_fini.
2985 (do_compile): Call debuginfo_init.
2986 * doc/invoke.texi (@gccoptlist): Add -fdump-debug and
2988 (@item -fdump-debug, @item -fdump-earlydebug): Add.
2990 2018-08-21 Tom de Vries <tdevries@suse.de>
2992 * dwarf2out.c (print_dw_val, print_loc_descr, print_die): Handle
2993 flag_dump_noaddr and flag_dump_unnumbered.
2995 2018-08-21 Aldy Hernandez <aldyh@redhat.com>
2997 * wide-int-range.cc (wide_int_range_abs): New.
2998 (wide_int_range_order_set): Rename from wide_int_range_min_max.
2999 * wide-int-range.h (wide_int_range_abs): New.
3000 (wide_int_range_min_max): New.
3001 * tree-vrp.c (extract_range_from_unary_expr): Rewrite ABS_EXPR
3002 case to call wide_int_range_abs.
3003 Rewrite MIN/MAX_EXPR to call wide_int_range_min_max.
3004 (extract_range_from_abs_expr): Delete.
3006 2018-08-20 Michael Meissner <meissner@linux.ibm.com>
3009 * config/rs6000/rs6000.md (extendsi<mode>2): Change constraints
3010 from 'Y' to 'YZ' to enable the LWAX instruction to be generated
3013 2018-08-20 Nathan Sidwell <nathan@acm.org>
3014 Jeff Law <law@redhat.com>
3016 * config/s390/s390-c (s390_macro_to_expand): Use cpp_macro_p.
3017 * config/spu/spu-c.c (spu_macro_to_expand): Likewise.
3019 2018-08-20 David Malcolm <dmalcolm@redhat.com>
3022 * attribs.c (diag_attr_exclusions): Add auto_diagnostic_group instance.
3023 (decl_attributes): Likewise.
3024 * calls.c (maybe_warn_nonstring_arg): Add auto_diagnostic_group
3026 * cgraphunit.c (maybe_diag_incompatible_alias): Likewise.
3027 * diagnostic-core.h (class auto_diagnostic_group): New class.
3028 * diagnostic.c (diagnostic_initialize): Initialize the new fields.
3029 (diagnostic_report_diagnostic): Handle the first diagnostics within
3031 (emit_diagnostic): Add auto_diagnostic_group instance.
3033 (inform_n): Likewise.
3034 (warning): Likewise.
3035 (warning_at): Likewise.
3036 (warning_n): Likewise.
3037 (pedwarn): Likewise.
3038 (permerror): Likewise.
3040 (error_n): Likewise.
3041 (error_at): Likewise.
3043 (fatal_error): Likewise.
3044 (internal_error): Likewise.
3045 (internal_error_no_backtrace): Likewise.
3046 (auto_diagnostic_group::auto_diagnostic_group): New ctor.
3047 (auto_diagnostic_group::~auto_diagnostic_group): New dtor.
3048 * diagnostic.h (struct diagnostic_context): Add fields
3049 "diagnostic_group_nesting_depth",
3050 "diagnostic_group_emission_count", "begin_group_cb",
3052 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior):
3053 Add auto_diagnostic_group instance(s).
3054 (find_explicit_erroneous_behavior): Likewise.
3055 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Likewise.
3056 * gimple-ssa-warn-restrict.c (maybe_diag_offset_bounds): Likewise.
3057 * gimplify.c (warn_implicit_fallthrough_r): Likewise.
3058 (gimplify_va_arg_expr): Likewise.
3059 * hsa-gen.c (HSA_SORRY_ATV): Likewise.
3060 (HSA_SORRY_AT): Likewise.
3061 * ipa-devirt.c (compare_virtual_tables): Likewise.
3062 (warn_odr): Likewise.
3063 * multiple_target.c (expand_target_clones): Likewise.
3064 * opts-common.c (cmdline_handle_error): Likewise.
3065 * reginfo.c (globalize_reg): Likewise.
3066 * substring-locations.c (format_warning_n_va): Likewise.
3067 * tree-inline.c (expand_call_inline): Likewise.
3068 * tree-ssa-ccp.c (pass_post_ipa_warn::execute): Likewise.
3069 * tree-ssa-loop-niter.c
3070 (do_warn_aggressive_loop_optimizations): Likewise.
3071 * tree-ssa-uninit.c (warn_uninit): Likewise.
3072 * tree.c (warn_deprecated_use): Likewise.
3074 2018-08-20 H.J. Lu <hongjiu.lu@intel.com>
3077 * config/i386/i386.md (eh_return): Always update EH return
3078 address in word_mode.
3080 2018-08-20 Chung-Lin Tang <cltang@codesourcery.com>
3082 * targhooks.c (std_gimplify_va_arg_expr): Properly handle case of when
3083 TARGET_SPLIT_COMPLEX_ARG is defined.
3085 2018-08-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
3087 * expr.c (store_field): Change gcc_assert to gcc_checking_assert.
3089 2018-08-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
3092 * expr.c (expand_assignment): Assert that bitpos is positive.
3093 (store_field): Likewise
3094 (expand_expr_real_1): Make sure that bitpos is positive.
3095 * config/alpha/alpha.h (CONSTANT_ADDRESS_P): Avoid signed
3098 2018-08-20 Nathan Sidwell <nathan@acm.org>
3100 * Makefile.in (CPP_ID_DATA_H): Delete.
3101 (CPP_INTERNAL_H): Don't add it.
3102 (GTFILES): Replace CPP_ID_DATA_H with CPPLIB_H.
3103 * gengtype.c (open_base_files): Replace cpp-id-data.h with cpplib.h
3105 2018-08-20 Richard Biener <rguenther@suse.de>
3107 PR tree-optimization/78655
3108 * tree-vrp.c (extract_range_from_binary_expr_1): Make
3109 pointer + offset nonnull if either operand is nonnull work.
3111 2018-08-20 Tom de Vries <tdevries@suse.de>
3113 * dwarf2out.c (add_scalar_info): Don't add reference to existing die
3114 unless the referenced die describes the added property using
3115 DW_AT_location or DW_AT_const_value. Fall back to exprloc case.
3116 Otherwise, add a DW_AT_location to the referenced die.
3118 2018-08-19 Uros Bizjak <ubizjak@gmail.com>
3121 * config/i386/i386.c (ix86_rtx_costs) [case SET]: Check source for
3122 register_operand when calling ix86_set_reg_reg_cost.
3123 [case CONST_INT, case CONST, case LABEL_REF, case SYMBOL_REF]:
3124 Set *total to 0 for operands that satisfy x86_64_immediate_operand
3125 predicate and to 1 otherwise.
3127 2018-08-18 Iain Sandoe <iain@sandoe.co.uk>
3129 * config/darwin.c (darwin_override_options): If -gsplit-dwarf is set,
3130 emit a diagnostic that it is not supported and reset the option.
3131 * config/darwin.h (DRIVER_SELF_SPECS): Note that gsplit-dwarf is not
3132 supported and consume the option. (ASM_FINAL_SPEC): New.
3134 2018-08-17 Segher Boessenkool <segher@kernel.crashing.org>
3136 * doc/md.texi (Patterns): Use @ref instead of @xref in the middle of
3139 2018-08-17 Sandra Loosemore <sandra@codesourcery.com>
3141 C-SKY port: Documentation
3143 * doc/extend.texi (C-SKY Function Attributes): New section.
3144 * doc/invoke.texi (Option Summary): Add C-SKY options.
3145 (C-SKY Options): New section.
3146 * doc/md.texi (Machine Constraints): Document C-SKY constraints.
3148 2018-08-17 Jojo <jijie_rong@c-sky.com>
3149 Huibin Wang <huibin_wang@c-sky.com>
3150 Sandra Loosemore <sandra@codesourcery.com>
3151 Chung-Lin Tang <cltang@codesourcery.com>
3153 C-SKY port: Backend implementation
3155 * config/csky/*: New.
3156 * common/config/csky/*: New.
3158 2018-08-17 Jojo <jijie_rong@c-sky.com>
3159 Huibin Wang <huibin_wang@c-sky.com>
3160 Sandra Loosemore <sandra@codesourcery.com>
3161 Chung-Lin Tang <cltang@codesourcery.com>
3162 Andrew Jenner <andrew@codesourcery.com>
3164 C-SKY port: Configury
3166 * config.gcc (csky-*-*): New.
3167 * configure.ac: Add csky to targets for dwarf2 debug_line support.
3168 * configure: Regenerated.
3170 2018-08-17 David Malcolm <dmalcolm@redhat.com>
3172 * dump-context.h: Include "dumpfile.h".
3173 (dump_context::dump_printf_va): Convert final param from va_list
3174 to va_list *. Convert from ATTRIBUTE_PRINTF to
3175 ATTRIBUTE_GCC_DUMP_PRINTF.
3176 (dump_context::dump_printf_loc_va): Likewise.
3177 * dumpfile.c: Include "stringpool.h".
3178 (make_item_for_dump_printf_va): Delete.
3179 (make_item_for_dump_printf): Delete.
3180 (class dump_pretty_printer): New class.
3181 (dump_pretty_printer::dump_pretty_printer): New ctor.
3182 (dump_pretty_printer::emit_items): New member function.
3183 (dump_pretty_printer::emit_any_pending_textual_chunks): New member
3185 (dump_pretty_printer::emit_item): New member function.
3186 (dump_pretty_printer::stash_item): New member function.
3187 (dump_pretty_printer::format_decoder_cb): New member function.
3188 (dump_pretty_printer::decode_format): New member function.
3189 (dump_context::dump_printf_va): Reimplement in terms of
3190 dump_pretty_printer.
3191 (dump_context::dump_printf_loc_va): Convert final param from va_list
3193 (dump_context::begin_scope): Reimplement call to
3194 make_item_for_dump_printf.
3195 (dump_printf): Update for change to dump_printf_va.
3196 (dump_printf_loc): Likewise.
3197 (selftest::test_capture_of_dump_calls): Convert "stmt" from
3198 greturn * to gimple *. Add a test_decl. Add tests of dump_printf
3199 with %T, %E, and %G.
3200 * dumpfile.h (ATTRIBUTE_GCC_DUMP_PRINTF): New macro.
3201 (dump_printf): Replace ATTRIBUTE_PRINTF_2 with
3202 ATTRIBUTE_GCC_DUMP_PRINTF (2, 3).
3203 (dump_printf_loc): Replace ATTRIBUTE_PRINTF_3 with
3204 ATTRIBUTE_GCC_DUMP_PRINTF (3, 0).
3205 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Convert
3206 use of HOST_WIDE_INT_PRINT_DEC on unsigned HOST_WIDE_INT "count"
3207 within a dump_printf_loc call to "%wu".
3208 (vector_alignment_reachable_p): Merge two dump_printf[_loc] calls,
3209 converting a use of HOST_WIDE_INT_PRINT_DEC to "%wd". Add a
3210 missing space after "=".
3211 * tree-vect-loop.c (vect_analyze_loop_2) Within a dump_printf
3212 call, convert use of HOST_WIDE_INT_PRINT_DEC to "%wd".
3213 * tree-vect-slp.c (vect_slp_bb): Within a dump_printf_loc call,
3214 convert use of HOST_WIDE_INT_PRINT_UNSIGNED to "%wu".
3215 * tree-vectorizer.c (try_vectorize_loop_1): Likewise. Remove
3216 duplicate "vectorized" from message.
3218 2018-08-17 Szabolcs Nagy <szabolcs.nagy@arm.com>
3220 * config/arm/arm-builtins.c (arm_init_simd_builtin_types): Clear
3221 polyNxK_t element's TYPE_STRING_FLAG.
3223 2018-08-17 Segher Boessenkool <segher@kernel.crashing.org>
3225 * config/rs6000/rs6000.md (*cbranch, *creturn): Name these patterns
3226 (they were unnamed before). Fix comments.
3228 2018-08-17 Nathan Sidwell <nathan@acm.org>
3230 * cppbuiltin.c: Include "cpplib.h", not "cpp-id-data.h".
3232 2018-08-17 Richard Biener <rguenther@suse.de>
3234 PR tree-optimization/86841
3235 * wide-int-range.cc (wide_int_range_lshift): Use to_uhwi.
3237 2018-08-17 Martin Liska <mliska@suse.cz>
3239 * common.opt: Remove Warn, Init and Report for options with
3240 Ignore/Deprecated flag. Warning is done automatically for
3242 * config/i386/i386.opt: Likewise.
3243 * config/ia64/ia64.opt: Likewise.
3244 * config/rs6000/rs6000.opt: Likewise.
3245 * cppbuiltin.c (define_builtin_macros_for_compilation_flags):
3246 Remove usage of flag_check_pointer_bounds.
3247 * lto-wrapper.c (merge_and_complain): Do not handle
3248 OPT_fcheck_pointer_bounds.
3249 (append_compiler_options): Likewise.
3250 * opt-functions.awk: Do not handle Deprecated.
3251 * optc-gen.awk: Check that Var, Report and Init are not
3252 used for an option with Ignore/Deprecated flag.
3253 * opts-common.c (decode_cmdline_option): Do not report
3255 (read_cmdline_option): Report warning for OPT_SPECIAL_deprecated
3257 * opts.h (struct cl_option): Remove cl_deprecated flag.
3258 (CL_ERR_DEPRECATED): Remove error enum value.
3260 2018-08-17 Richard Biener <rguenther@suse.de>
3263 * tree-inline.c (copy_bb): When inlining __builtin_va_arg_pack_len ()
3264 across a va-arg-pack using call adjust its return value accordingly.
3266 2018-08-16 Martin Sebor <msebor@redhat.com>
3268 PR tree-optimization/86853
3269 * gimple-ssa-sprintf.c (struct format_result): Rename member.
3270 (struct fmtresult): Add member and initialize it in ctors.
3271 (format_character): Handle %C. Extend range to NUL. Set MAYFAIL.
3272 (format_string): Handle %S the same as %ls. Set MAYFAIL.
3273 (format_directive): Set POSUNDER4K when MAYFAIL is set.
3274 (parse_directive): Handle %C same as %c.
3275 (sprintf_dom_walker::compute_format_length): Adjust.
3276 (is_call_safe): Adjust.
3278 2018-08-16 Bernd Edlinger <bernd.edlinger@hotmail.de>
3280 * builtins.c (c_strlen): Add new parameter eltsize. Use it
3281 for determining how to count the elements.
3282 * builtins.h (c_strlen): Adjust prototype.
3283 * expr.c (string_constant): Add new parameter mem_size.
3284 Set *mem_size appropriately.
3285 * expr.h (string_constant): Adjust protoype.
3286 * gimple-fold.c (get_range_strlen): Add new parameter eltsize.
3287 * gimple-fold.h (get_range_strlen): Adjust prototype.
3288 * gimple-ssa-sprintf.c (get_string_length): Add new parameter eltsize.
3289 (format_string): Call get_string_length with eltsize.
3291 2018-08-16 David Malcolm <dmalcolm@redhat.com>
3293 * diagnostic.c (default_diagnostic_start_span_fn): Call pp_string
3294 to emit the span, rather than setting it as the prefix.
3296 2018-08-16 David Malcolm <dmalcolm@redhat.com>
3298 * diagnostic-show-locus.c (layout::start_annotation_line): Add
3299 "margin_char" parameter, defaulting to space. Use it in place
3300 of pp_space for the initial part of the margin.
3301 (layout::print_leading_fixits): Use '+' when filling the margin
3302 of line-insertion fix-it hints.
3304 2018-08-16 Segher Boessenkool <segher@kernel.crashing.org>
3306 * config/rs6000/rs6000.md (two unnamed define_insn and define_split):
3309 2018-08-16 Segher Boessenkool <segher@kernel.crashing.org>
3311 * config/rs6000/altivec.md: Don't set length attribute to the default
3313 * config/rs6000/darwin.md: Ditto.
3314 * config/rs6000/dfp.md: Ditto.
3315 * config/rs6000/htm.md: Ditto.
3316 * config/rs6000/rs6000.md: Ditto.
3317 * config/rs6000/sync.md: Ditto.
3318 * config/rs6000/vsx.md: Ditto.
3320 2018-08-16 Segher Boessenkool <segher@kernel.crashing.org>
3322 * config/rs6000/altivec.md: Don't set length attribute to the default
3323 value, for branch instructions.
3324 * config/rs6000/darwin.md: Ditto.
3325 * config/rs6000/rs6000.md: Ditto.
3327 2018-08-16 Segher Boessenkool <segher@kernel.crashing.org>
3329 * config/rs6000/rs6000.md (length): Always define as const_int 4.
3330 (unnamed conditional branch define_insn): Set length to 4 or 8
3331 depending on offset.
3332 (<bd>_<mode>): Similar, for alternative 0.
3333 (<bd>tf_<mode>): Ditto.
3335 2018-08-16 Tamar Christina <tamar.christina@arm.com>
3337 * expr.c (copy_blkmode_to_reg): Perform larger copies when safe.
3339 2018-08-16 Matthew Malcomson <matthew.malcomson@arm.com>
3341 * doc/rtl.texi: Replace old RTX class names with new names.
3344 2018-08-16 Vlad Lazar <vlad.lazar@arm.com>
3346 * expmed.h (canonicalize_comparison): New declaration.
3347 * expmed.c (canonicalize_comparison, equivalent_cmp_code): New function.
3348 * expmed.c (emit_store_flag_1): Add call to canonicalize_comparison.
3349 * optabs.c (prepare_cmp_insn): Likewise.
3350 * rtl.h (unsigned_condition_p): New function which checks if a
3351 comparison operator is unsigned.
3353 2018-08-16 Nathan Sidwell <nathan@acm.org>
3355 * config/rs6000/rs6000-c.c (rs6000_macro_to_expend): Use cpp_macro_p.
3356 * config/powerpcspc/powerpcspe-c.c (rs6000_macro_to_expend): Likewise.
3358 2018-08-16 Tamar Christina <tamar.christina@arm.com>
3361 * config/arm/arm.c (arm_can_change_mode_class): Disallow subreg.
3362 * config/arm/neon.md (movv4hf, movv8hf): Refactored to..
3363 (mov<mov>): ..this and enable unconditionally.
3365 2018-08-16 Tamar Christina <tamar.christina@arm.com>
3367 * config/arm/neon.md (*neon_mov<mode>): Remove reg-to-reg alternative.
3369 2018-08-16 Sam Tebbs <sam.tebbs@arm.com>
3371 * config/aarch64/aarch64.opt (mlow-precision-recip-sqrt)
3372 (mlow-precision-sqrt, mlow-precision-div, mverbose-cost-dump): Replace
3373 "Common" with "Target".
3375 2018-08-15 Uros Bizjak <ubizjak@gmail.com>
3377 * config/i386/i386.opt (mmitigate-rop): Mark as deprecated.
3378 * doc/invoke.texi (mmitigate-rop): Remove.
3379 * config/i386/i386.c: Do not include "regrename.h".
3380 (ix86_rop_should_change_byte_p, reg_encoded_number)
3381 (ix86_get_modrm_for_rop, set_rop_modrm_reg_bits, ix86_mitigate_rop):
3383 (ix86_reorg): Remove call to ix86_mitigate_rop.
3384 * config/i386/i386.md (attr "modrm_class"): Remove.
3385 (cmp<mode>_ccno_1, mov<mode>_xor, movstrict<mode>_xor)
3386 (x86_mov<mode>cc_0_m1. x86_mov<mode>cc_0_m1_se)
3387 (x86_mov<mode>cc_0_m1_neg): Remove modrm_class attribute override.
3389 2018-08-15 Will Schmidt <will_schmidt@vnet.ibm.com>
3391 * config/rs6000/rs600.c (rs6000_gimple_fold_builtin): Add entries to
3392 allow folding of mergeh() and mergel() for the float and double types.
3393 (fold_mergehl_helper): Rework to handle building a permute tree
3396 2018-08-15 Uros Bizjak <ubizjak@gmail.com>
3398 * config/i386/i386.c (expand_vec_perm_movs): Enable V4SFmode
3401 2018-08-15 David Malcolm <dmalcolm@redhat.com>
3403 * common.opt (fdiagnostics-show-labels): New option.
3404 * diagnostic-show-locus.c (class layout_range): Add field
3406 (class layout): Add field "m_show_labels_p".
3407 (layout_range::layout_range): Add param "label" and use it to
3409 (make_range): Pass in NULL for new "label" param of layout_range's
3411 (layout::layout): Initialize m_show_labels_p.
3412 (layout::maybe_add_location_range): Pass in loc_range->m_label
3413 when constructing layout_range instances.
3414 (struct line_label): New struct.
3415 (layout::print_any_labels): New member function.
3416 (layout::print_line): Call it if label-printing is enabled.
3417 (selftest::test_one_liner_labels): New test.
3418 (selftest::test_diagnostic_show_locus_one_liner): Call it.
3419 * diagnostic.c (diagnostic_initialize): Initialize
3420 context->show_labels_p.
3421 * diagnostic.h (struct diagnostic_context): Add field
3423 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
3424 -fno-diagnostics-show-labels.
3425 * dwarf2out.c (gen_producer_string): Add
3426 OPT_fdiagnostics_show_labels to the ignored options.
3427 * gcc-rich-location.c (gcc_rich_location::add_expr): Add "label"
3429 (gcc_rich_location::maybe_add_expr): Likewise.
3430 * gcc-rich-location.h (gcc_rich_location::gcc_rich_location): Add
3431 label" param, defaulting to NULL.
3432 (gcc_rich_location::add_expr): Add "label" param.
3433 (gcc_rich_location::maybe_add_expr): Likewise.
3434 (class text_range_label): New class.
3435 (class range_label_for_type_mismatch): New class.
3436 * gimple-ssa-sprintf.c (fmtwarn): Pass NULL for new label params
3437 of format_warning_va.
3438 (fmtwarn_n): Likewise for new params of format_warning_n_va.
3439 * lto-wrapper.c (merge_and_complain): Add
3440 OPT_fdiagnostics_show_labels to the "pick one setting" options.
3441 (append_compiler_options): Likewise to the dropped options.
3442 (append_diag_options): Likewise to the passed-on options.
3443 * opts.c (common_handle_option): Handle the new option.
3444 * selftest-diagnostic.c
3445 (test_diagnostic_context::test_diagnostic_context): Enable
3447 * substring-locations.c: Include "gcc-rich-location.h".
3448 (format_warning_n_va): Add "fmt_label" and "param_label" params
3449 and use them as appropriate.
3450 (format_warning_va): Add "fmt_label" and "param_label" params,
3451 passing them on to format_warning_n_va.
3452 (format_warning_at_substring): Likewise.
3453 (format_warning_at_substring_n): Likewise.
3454 * substring-locations.h (format_warning_va): Add "fmt_label" and
3455 "param_label" params.
3456 (format_warning_n_va): Likewise.
3457 (format_warning_at_substring): Likewise.
3458 (format_warning_at_substring_n): Likewise.
3459 * toplev.c (general_init): Initialize global_dc->show_labels_p.
3461 2018-08-15 Qing Zhao <qing.zhao@oracle.com>
3464 * builtins.c (expand_builtin_memcmp): Do not expand the call
3465 when overflow is detected.
3467 2018-08-15 Martin Sebor <msebor@redhat.com>
3469 PR tree-optimization/71625
3470 * config/aarch64/aarch64-builtins.c
3471 (aarch64_init_simd_builtin_types): Clear Poly8_t's TYPE_STRING_FLAG.
3473 2018-08-15 Ilya Leoshkevich <iii@linux.ibm.com>
3475 * config/s390/s390.c (s390_reorg): Remove loop.
3477 2018-08-15 Iain Sandoe <iain@sandoe.co.uk>
3480 (darwin_function_switched_text_sections): Delete.
3481 * gcc/config/darwin.h
3482 (TARGET_ASM_FUNCTION_SWITCHED_TEXT_SECTIONS): Likewise.
3484 2018-08-15 Iain Sandoe <iain@sandoe.co.uk>
3487 * config/darwin.h: (DEBUG_STR_OFFSETS_SECTION, DEBUG_LOCLISTS_SECTION,
3488 DEBUG_RNGLISTS_SECTION) new macros. (DEBUG_PUBNAMES_SECTION,
3489 DEBUG_PUBTYPES_SECTION) update to include GNU variant.
3491 2018-08-15 Martin Liska <mliska@suse.cz>
3493 PR tree-optimization/86925
3494 * predict.c (expr_expected_value_1): When taking
3495 later predictor, assign also probability.
3496 Use fold_build2_initializer_loc in order to fold
3497 the expression in -frounding-math.
3499 2018-08-14 Allan Sandfeld Jensen <allan.jensen@qt.io>
3501 * config/i386/i386.c (expand_vec_perm_movs): New method matching movs
3503 (expand_vec_perm_1): Try the new method.
3505 2018-08-14 Ilya Leoshkevich <iii@linux.ibm.com>
3508 * lra-lives.c (remove_some_program_points_and_update_live_ranges):
3509 Check whether lra_live_max_point is 0 before dividing.
3511 2018-08-14 Martin Sebor <msebor@redhat.com>
3513 PR tree-optimization/86650
3514 * tree-vrp.c (vrp_prop::check_array_ref): Print an inform message.
3515 (vrp_prop::check_mem_ref): Same.
3517 2018-08-13 Liu Hao <lh_mouse@126.com>
3519 * pretty-print.c (eat_esc_sequence): Swap the foreground and
3520 background colors if the COMMON_LVB_REVERSE_VIDEO flag is set,
3521 and clear it thereafter, as it only works for DBCS.
3523 2018-08-13 Liu Hao <lh_mouse@126.com>
3525 * pretty-print.c (mingw_ansi_fputs): Do not call _close() on the
3526 handle returned by _get_osf_handle().
3528 2018-08-13 Will Schmidt <will_schmidt@vnet.ibm.com>
3530 * gcc/config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support
3531 for folding vec_perm.
3533 2018-08-13 Will Schmidt <will_schmidt@vnet.ibm.com>
3535 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin):
3536 Add support for gimple-folding of vec_pack() and vec_unpack()
3539 2018-08-13 Will Schmidt <will_schmidt@vnet.ibm.com>
3541 * config/rs6000/rs6000.c (rs6000_builtin_valid_without_lhs): Add
3542 vec_xst variants to the list.
3543 (rs6000_gimple_fold_builtin): Add support for folding unaligned
3544 vector loads and stores.
3546 2018-08-13 David Edelsohn <dje.gcc@gmail.com>
3548 * config.gcc (rs6000-ibm-aix4.x): Delete.
3549 (rs6000-ibm-aix5.1): Delete.
3550 (rs6000-ibm-aix5.2): Delete.
3551 (rs6000-ibm-aix5.3): Delete.
3552 * config/rs6000/aix43.h: Delete.
3553 * config/rs6000/aix51.h: Delete.
3554 * config/rs6000/aix52.h: Delete.
3555 * config/rs6000/t-aix43: Delete.
3557 2018-08-13 Ilya Leoshkevich <iii@linux.ibm.com>
3559 * config/s390/s390.c (s390_decompose_constant_pool_ref):
3561 (s390_decompose_address): Factor out constant pool ref
3564 2018-08-12 Chung-Ju Wu <jasonwucj@gmail.com>
3566 * config/nds32/nds32-predicates.c
3567 (nds32_can_use_bclr_p): Change return type as bool.
3568 (nds32_can_use_bset_p): Ditto.
3569 (nds32_can_use_btgl_p): Ditto.
3570 (nds32_can_use_bitci_p): Ditto.
3571 * config/nds32/nds32-protos.h
3572 (nds32_can_use_bclr_p): Change declaration.
3573 (nds32_can_use_bset_p): Ditto.
3574 (nds32_can_use_btgl_p): Ditto.
3575 (nds32_can_use_bitci_p): Ditto.
3577 2018-08-12 Chung-Ju Wu <jasonwucj@gmail.com>
3579 * config/nds32/nds32.c (nds32_expand_prologue, nds32_expand_epilogue):
3580 Support -msched-prolog-epilog option.
3581 * config/nds32/nds32.opt (msched-prolog-epilog): New option.
3583 2018-08-12 Chung-Ju Wu <jasonwucj@gmail.com>
3585 * common/config/nds32/nds32-common.c
3586 (nds32_option_optimization_table): Enalbe -malways-align.
3588 2018-08-12 Chung-Ju Wu <jasonwucj@gmail.com>
3590 * config.gcc (nds32*): Add nds32_isr.h and nds32_init.inc in
3592 * common/config/nds32/nds32-common.c (nds32_handle_option): Handle
3593 OPT_misr_secure_ case.
3594 * config/nds32/nds32-isr.c: Implementation of backward compatibility.
3595 * config/nds32/nds32-protos.h (nds32_isr_function_critical_p): New.
3596 * config/nds32/nds32.c (nds32_attribute_table): Add critical and
3598 * config/nds32/nds32.h (nds32_isr_nested_type): Add NDS32_CRITICAL.
3599 (nds32_isr_info): New field security_level.
3600 (TARGET_ISR_VECTOR_SIZE_4_BYTE): New macro.
3601 * config/nds32/nds32.md (return_internal): Consider critical attribute.
3602 * config/nds32/nds32.opt (misr-secure): New option.
3603 * config/nds32/nds32_init.inc: New file.
3604 * config/nds32/nds32_isr.h: New file.
3606 2018-08-11 John David Anglin <danglin@gcc.gnu.org>
3608 * config/pa/pa.md (UNSPEC_MEMORY_BARRIER): New unspec enum.
3609 Update comment for atomic instructions.
3610 (atomic_storeqi, atomic_storehi, atomic_storesi, atomic_storesf,
3611 atomic_loaddf, atomic_loaddf_1, atomic_storedf, atomic_storedf_1):
3613 (atomic_loaddi): Revise fence expansion to only emit fence prior to
3614 load for __ATOMIC_SEQ_CST model.
3615 (atomic_loaddi_1): Remove float register target.
3616 (atomic_storedi): Handle CONST_INT values.
3617 (atomic_storedi_1): Remove float register source. Add special case
3619 (memory_barrier): New expander and insn.
3621 2018-08-11 Jakub Jelinek <jakub@redhat.com>
3623 PR tree-optimization/86835
3624 * tree-ssa-math-opts.c (insert_reciprocals): Even when inserting
3625 new_stmt after def_gsi, make sure to insert new_square_stmt after
3626 that stmt, not 2 stmts before it.
3628 2018-08-10 Alexander Monakov <amonakov@ispras.ru>
3631 * config/i386/i386.md (<s>mul<mode>3_highpart): Use DWIH mode iterator
3634 2018-08-10 Martin Liska <mliska@suse.cz>
3637 * builtin-types.def (BT_FN_LONG_LONG_LONG_DOUBLE): Add new
3639 * builtins.c (expand_builtin_expect_with_probability):
3641 (expand_builtin_expect_with_probability): New function.
3642 (build_builtin_expect_predicate): Add new argumnet probability
3643 for BUILT_IN_EXPECT_WITH_PROBABILITY.
3644 (fold_builtin_expect):
3647 * builtins.def (BUILT_IN_EXPECT_WITH_PROBABILITY):
3648 * builtins.h (fold_builtin_expect): Set new argument.
3649 * doc/extend.texi: Document __builtin_expect_with_probability.
3650 * doc/invoke.texi: Likewise.
3651 * gimple-fold.c (gimple_fold_call): Pass new argument.
3652 * ipa-fnsummary.c (find_foldable_builtin_expect): Handle
3653 also BUILT_IN_EXPECT_WITH_PROBABILITY.
3654 * predict.c (get_predictor_value): New function.
3655 (expr_expected_value): Add new argument probability. Assume
3656 that predictor and probability are always non-null.
3657 (expr_expected_value_1): Likewise. For __builtin_expect and
3658 __builtin_expect_with_probability set probability. Handle
3659 combination in binary expressions.
3660 (tree_predict_by_opcode): Simplify code by simply calling
3661 get_predictor_value.
3662 (pass_strip_predict_hints::execute): Add handling of
3663 BUILT_IN_EXPECT_WITH_PROBABILITY.
3664 * predict.def (PRED_BUILTIN_EXPECT_WITH_PROBABILITY): Add
3666 * tree.h (DECL_BUILT_IN_P): New function.
3668 2018-08-10 Martin Liska <mliska@suse.cz>
3670 PR tree-optimization/85799
3671 * passes.def: Add argument for pass_strip_predict_hints.
3672 * predict.c (class pass_strip_predict_hints): Add new argument
3674 (strip_predictor_early): New function.
3675 (pass_strip_predict_hints::execute): Call the function to
3677 (strip_predict_hints): New function.
3678 * predict.def: Fix comment.
3680 2018-08-10 Thomas Preud'homme <thomas.preudhomme@linaro.org>
3682 * Makefile.in: Clarify which tm.texi to copy over to assert the
3683 right to grant a GFDL license for all.
3685 2018-08-09 Jeff Law <law@redhat.com>
3687 * config/m68k/m68k.c (m68k_adjust_decorated_operand): Remove
3690 2018-08-09 Andreas Schwab <schwab@linux-m68k.org>
3692 * config/m68k/m68k-protos.h (m68k_final_prescan_insn): Remove
3695 2018-08-09 Richard Sandiford <richard.sandiford@arm.com>
3697 * tree-vect-loop.c (vectorizable_reduction): Allow inner-loop
3698 reductions for variable-length vectors.
3700 2018-08-09 David Malcolm <dmalcolm@redhat.com>
3703 * common.opt (fdiagnostics-show-line-numbers): New option.
3704 * diagnostic-show-locus.c (class layout): Add fields
3705 "m_show_line_numbers_p" and "m_linenum_width";
3706 (num_digits): New function.
3707 (test_num_digits): New function.
3708 (layout::layout): Initialize new fields. Update m_x_offset
3709 logic to handle any left margin.
3710 (layout::print_source_line): Print line number when requested.
3711 (layout::start_annotation_line): New member function.
3712 (layout::print_annotation_line): Call it.
3713 (layout::print_leading_fixits): Likewise.
3714 (layout::print_trailing_fixits): Likewise. Update calls to
3715 move_to_column for new parameter.
3716 (layout::get_x_bound_for_row): Add "add_left_margin" param and use
3717 it to potentially call start_annotation_line.
3718 (layout::show_ruler): Call start_annotation_line.
3719 (selftest::test_line_numbers_multiline_range): New selftest.
3720 (selftest::diagnostic_show_locus_c_tests): Call test_num_digits
3721 and selftest::test_line_numbers_multiline_range.
3722 * diagnostic.c (diagnostic_initialize): Initialize
3723 show_line_numbers_p.
3724 * diagnostic.h (struct diagnostic_context): Add field
3725 "show_line_numbers_p".
3726 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
3727 -fno-diagnostics-show-line-numbers.
3728 * dwarf2out.c (gen_producer_string): Add
3729 OPT_fdiagnostics_show_line_numbers to the ignored options.
3730 * lto-wrapper.c (merge_and_complain): Likewise to the "pick
3731 one setting" options.
3732 (append_compiler_options): Likewise to the dropped options.
3733 (append_diag_options): Likewise to the passed-on options.
3734 * opts.c (common_handle_option): Handle the new option.
3735 * toplev.c (general_init): Set up global_dc->show_line_numbers_p.
3737 2018-08-09 Kelvin Nilsen <kelvin@gcc.gnu.org>
3739 * doc/extend.texi (PowerPC AltiVec Built-in Functions Available on
3740 ISA 2.07): Correct spelling of bcdsub to be __builtin_bcdsub. Add
3741 third argument of type "const signed char" to descriptions of
3742 __builtin_bcdadd, __builtin_bcdadd_lt, __builtin_bcdadd_eq,
3743 __builtin_bcdadd_gt, __builtin_bcdadd_ov, __builtin_bcdsub,
3744 __builtin_bcdsub_lt, __builtin_bcdsub_eq, __builtin_bcdsub_gt,
3745 __builtin_bcdsub_ov functions.
3747 2018-08-09 Richard Sandiford <richard.sandiford@arm.com>
3749 PR tree-optimization/86858
3750 * tree-vect-loop.c (vect_is_simple_reduction): Restore
3751 flow_bb_inside_loop_p calls.
3753 2018-08-09 Richard Sandiford <richard.sandiford@arm.com>
3755 PR tree-optimization/86871
3756 * tree-vect-stmts.c (vect_transform_stmt): Use gimple_get_lhs
3757 instead of gimple_assign_lhs.
3759 2018-08-09 Richard Earnshaw <rearnsha@arm.com>
3762 * config/aarch64/aarch64.md (add<mode>3_carryinC_zero): Add missing
3763 register constraint to operand 0.
3764 (add<mode>3_carryinC): Likewise.
3765 (add<mode>3_carryinV_zero, add<mode>3_carryinV): Likewise.
3767 2018-08-09 Martin Liska <mliska@suse.cz>
3770 * common.opt: Remove extra line.
3772 2018-08-09 Martin Liska <mliska@suse.cz>
3774 * params.def (PARAM_ALIGN_LOOP_ITERATIONS): Remove double dots
3775 at the end of a line, make first letter capital and end up
3776 a sentence with a dot.
3777 (PARAM_LOOP_INTERCHANGE_STRIDE_RATIO): Likewise.
3778 (PARAM_LOOP_BLOCK_TILE_SIZE): Likewise.
3779 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Likewise.
3780 (PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP): Likewise.
3781 (PARAM_MAX_ISL_OPERATIONS): Likewise.
3782 (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): Likewise.
3783 (PARAM_PROFILE_FUNC_INTERNAL_ID): Likewise.
3784 (PARAM_INDIR_CALL_TOPN_PROFILE): Likewise.
3785 (PARAM_SLP_MAX_INSNS_IN_BB): Likewise.
3786 (PARAM_IPA_CP_EVAL_THRESHOLD): Likewise.
3787 (PARAM_IPA_CP_RECURSION_PENALTY): Likewise.
3788 (PARAM_IPA_CP_SINGLE_CALL_PENALTY): Likewise.
3789 (PARAM_IPA_CP_LOOP_HINT_BONUS): Likewise.
3790 (PARAM_IPA_CP_ARRAY_INDEX_HINT_BONUS): Likewise.
3791 (PARAM_TREE_REASSOC_WIDTH): Likewise.
3792 (PARAM_HSA_GEN_DEBUG_STORES): Likewise.
3793 (PARAM_MAX_SPECULATIVE_DEVIRT_MAYDEFS): Likewise.
3794 (PARAM_MAX_VRP_SWITCH_ASSERTIONS): Likewise.
3796 2018-08-09 Andreas Krebbel <krebbel@linux.ibm.com>
3799 * config/s390/s390.c (s390_option_override_internal): Reduce the
3800 stack-clash-protection-probe-interval param if it would be too big
3803 2018-08-08 Andreas Schwab <schwab@linux-m68k.org>
3806 * config/m68k/m68k.h (FINAL_PRESCAN_INSN): Don't define.
3807 * config/m68k/m68k.c (handle_move_double): Don't call
3808 m68k_final_prescan_insn.
3809 (m68k_adjust_decorated_operand): Renamed from
3810 m68k_final_prescan_insn, remove first and third operand and
3812 (print_operand): Call it.
3813 (print_operand_address): Call it.
3815 2018-08-08 Nathan Sidwell <nathan@acm.org>
3817 * diagnostic.c (diagnostic_report_current_module): Use
3818 linemap_included_from & linemap_included_from_linemap.
3820 2018-08-08 Hongbo Zhang <hongbo.zhang@linaro.org>
3822 * config/aarch64/aarch64-cores.def: Add phecda core.
3823 * config/aarch64/aarch64-tune.md: Regenerate.
3824 * doc/invoke.texi: Add phecda core.
3826 2018-08-08 Andreas Krebbel <krebbel@linux.ibm.com>
3829 * config/s390/constraints.md ("NxHD0", "NxSD0"): New constraint
3831 * config/s390/s390.md ("movti"): Add more alternatives for
3832 constant to GPR copies.
3834 2018-08-08 Andreas Krebbel <krebbel@linux.ibm.com>
3836 * config/s390/s390.c: Fix whitespace damage throughout the file.
3837 * config/s390/s390.h: Likewise.
3838 * config/s390/tpf.h: Likewise.
3840 2018-08-08 Ilya Leoshkevich <iii@linux.ibm.com>
3842 * config/s390/s390.c (s390_loadrelative_operand_p):
3843 Remove TARGET_CPU_ZARCH usages.
3844 (s390_rtx_costs): Likewise.
3845 (s390_legitimate_constant_p): Likewise.
3846 (s390_cannot_force_const_mem): Likewise.
3847 (legitimate_reload_constant_p): Likewise.
3848 (s390_preferred_reload_class): Likewise.
3849 (legitimize_pic_address): Likewise.
3850 (legitimize_tls_address): Likewise.
3851 (s390_split_branches): Removed.
3852 (s390_add_execute): Removed.
3853 (s390_dump_pool): Remove TARGET_CPU_ZARCH usages.
3854 (s390_mainpool_start): Likewise.
3855 (s390_mainpool_finish): Likewise.
3856 (s390_mainpool_cancel): Removed.
3857 (s390_chunkify_start): Remove TARGET_CPU_ZARCH usages.
3858 (s390_chunkify_cancel): Likewise.
3859 (s390_return_addr_rtx): Likewise.
3860 (s390_register_info): Remove split_branches_pending_p uages.
3861 (s390_optimize_register_info): Likewise.
3862 (s390_init_frame_layout): Remove TARGET_CPU_ZARCH and
3863 split_branches_pending_p usages.
3864 (s390_can_eliminate): Remove TARGET_CPU_ZARCH usages.
3865 (s390_load_got): Likewise.
3866 (s390_expand_split_stack_prologue): Likewise.
3867 (output_asm_nops): Likewise.
3868 (s390_function_profiler): Likewise.
3869 (s390_emit_call): Likewise.
3870 (s390_conditional_register_usage): Likewise.
3871 (s390_optimize_prologue): Likewise.
3872 (s390_reorg): Remove TARGET_CPU_ZARCH and
3873 split_branches_pending_p usages.
3874 (s390_option_override_internal): Remove TARGET_CPU_ZARCH
3876 (s390_output_indirect_thunk_function): Likewise.
3877 * config/s390/s390.h (TARGET_CPU_ZARCH): Removed.
3878 (TARGET_CPU_ZARCH_P): Removed.
3879 (struct machine_function): Remove split_branches_pending_p.
3880 * config/s390/s390.md: Remove TARGET_CPU_ZARCH usages.
3882 2018-08-08 Ilya Leoshkevich <iii@linux.ibm.com>
3884 * common/config/s390/s390-common.c (processor_flags_table):
3886 * config.gcc: Remove with_arch/with_tune support.
3887 * config/s390/2064.md: Remove cpu attribute comparisons.
3888 * config/s390/driver-native.c (s390_host_detect_local_cpu):
3890 * config/s390/linux.h (ASM_SPEC):
3891 Remove -march support.
3892 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal):
3893 Use a table to get an arch level.
3894 * config/s390/s390-opts.h (enum processor_type):
3896 * config/s390/s390.c
3897 (processor_table): Remove entries, add arch_level values.
3898 (s390_issue_rate): Remove cases.
3899 (s390_option_override): Adjust
3900 s390_option_override_internal() call.
3901 (s390_option_override_internal): Remove deprecation warning.
3902 (s390_valid_target_attribute_tree): Adjust
3903 s390_option_override_internal() call.
3904 * config/s390/s390.h (struct s390_processor):
3905 Share with s390-c.c, add arch_level field.
3906 * config/s390/s390.md:
3907 Remove occurrences in cpu attribute.
3908 * config/s390/s390.opt: Remove -march/-mtune support.
3909 * config/s390/tpf.h (ASM_SPEC): Remove -march support.
3910 * doc/invoke.texi: Remove deprecation warning.
3912 2018-08-08 Luis Machado <luis.machado@linaro.org>
3914 * config/aarch64/aarch64.c (qdf24xx_vector_cost): New static
3916 (qdf24xx_tunings): Set vector cost structure to
3917 qdf24xx_vector_cost.
3919 * config/aarch64/aarch64.c (qdf24xx_addrcost_table)
3920 <register_sextend>: Set to 3.
3922 2018-08-07 Richard Sandiford <richard.sandiford@arm.com>
3925 * config/aarch64/iterators.md (FRECP, frecp_suffix): Delete.
3926 * config/aarch64/aarch64-simd.md
3927 (aarch64_frecp<FRECP:frecp_suffix><mode>): Fold FRECPE into...
3928 (@aarch64_frecpe<mode>): ...here and the move FRECPX to...
3929 (aarch64_frecpx<mode>): ...this new pattern.
3930 * config/aarch64/aarch64-simd-builtins.def: Remove comment
3931 about aarch64_frecp<FRECP:frecp_suffix><mode>.
3933 2018-08-07 Martin Liska <mliska@suse.cz>
3936 * predict.c (expr_expected_value_1): Handle DECL_IS_MALLOC,
3937 BUILT_IN_REALLOC and DECL_IS_OPERATOR_NEW.
3938 * predict.def (PRED_MALLOC_NONNULL): New predictor.
3939 * doc/extend.texi: Document that malloc attribute adds
3942 2018-08-06 John David Anglin <danglin@gcc.gnu.org>
3945 * config/pa/pa.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
3946 Define to speculation_safe_value_not_needed.
3948 2018-08-06 Jeff Law <law@redhat.com>
3950 * tree-ssa-dom.c (dom_opt_dom_walker::optimize_stmt): Pass down
3951 the vr_values instance to cprop_into_stmt.
3952 (cprop_into_stmt): Pass vr_values instance down to cprop_operand.
3953 (cprop_operand): Also query EVRP to determine if OP is a constant.
3955 2018-08-06 Nathan Sidwell <nathan@acm.org>
3957 * diagnostic.c (diagnostic_report_current_module): Reroll
3958 included-at loop. Translate text.
3960 2018-08-06 David Malcolm <dmalcolm@redhat.com>
3962 * function-tests.c (selftest::test_expansion_to_rtl): Call
3963 free_after_compilation.
3965 2018-08-06 Alan Hayward <alan.hayward@arm.com>
3967 * config/aarch64/aarch64.md: Add clobber highs to tls_desc.
3969 2018-08-06 Andreas Krebbel <krebbel@linux.ibm.com>
3971 * config/s390/s390.c (s390_loop_unroll_adjust): Prevent small
3972 loops with memory block operations from getting unrolled.
3974 2018-08-06 Ulrich Weigand <uweigand@de.ibm.com>
3977 * config/spu/spu.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
3978 Define to speculation_safe_value_not_needed.
3980 2018-08-06 Jeff Law <law@redhat.com>
3982 * reload1.c (forget_old_reloads_1): Adjust CLOBBER_HIGH
3985 2018-08-06 Jozef Lawrynowicz <jozef.l@mittosystems.com>
3988 * gcc/tree.c (build_common_tree_nodes): Initialize integer_types array
3989 with all enabled __intN types.
3991 * gcc/testsuite/gcc.target/msp430/pr86662.c: New test.
3993 2018-08-06 Alan Hayward <alan.hayward@arm.com>
3995 * alias.c (record_set): Check for clobber high.
3996 * cfgexpand.c (expand_gimple_stmt): Likewise.
3997 * combine-stack-adj.c (single_set_for_csa): Likewise.
3998 * combine.c (find_single_use_1): Likewise.
3999 (set_nonzero_bits_and_sign_copies): Likewise.
4000 (get_combine_src_dest): Likewise.
4001 (is_parallel_of_n_reg_sets): Likewise.
4002 (try_combine): Likewise.
4003 (record_dead_and_set_regs_1): Likewise.
4004 (reg_dead_at_p_1): Likewise.
4005 (reg_dead_at_p): Likewise.
4006 * dce.c (deletable_insn_p): Likewise.
4007 (mark_nonreg_stores_1): Likewise.
4008 (mark_nonreg_stores_2): Likewise.
4009 * df-scan.c (df_find_hard_reg_defs): Likewise.
4010 (df_uses_record): Likewise.
4011 (df_get_call_refs): Likewise.
4012 * dwarf2out.c (mem_loc_descriptor): Likewise.
4013 * haifa-sched.c (haifa_classify_rtx): Likewise.
4014 * ira-build.c (create_insn_allocnos): Likewise.
4015 * ira-costs.c (scan_one_insn): Likewise.
4016 * ira.c (equiv_init_movable_p): Likewise.
4017 (rtx_moveable_p): Likewise.
4018 (interesting_dest_for_shprep): Likewise.
4019 * jump.c (mark_jump_label_1): Likewise.
4020 * postreload-gcse.c (record_opr_changes): Likewise.
4021 * postreload.c (reload_cse_simplify): Likewise.
4022 (struct reg_use): Add source expr.
4023 (reload_combine): Check for clobber high.
4024 (reload_combine_note_use): Likewise.
4025 (reload_cse_move2add): Likewise.
4026 (move2add_note_store): Likewise.
4027 * print-rtl.c (print_pattern): Likewise.
4028 * recog.c (decode_asm_operands): Likewise.
4029 (store_data_bypass_p): Likewise.
4030 (if_test_bypass_p): Likewise.
4031 * regcprop.c (kill_clobbered_value): Likewise.
4032 (kill_set_value): Likewise.
4033 * reginfo.c (reg_scan_mark_refs): Likewise.
4034 * reload1.c (maybe_fix_stack_asms): Likewise.
4035 (eliminate_regs_1): Likewise.
4036 (elimination_effects): Likewise.
4037 (mark_not_eliminable): Likewise.
4038 (scan_paradoxical_subregs): Likewise.
4039 (forget_old_reloads_1): Likewise.
4040 * reorg.c (find_end_label): Likewise.
4041 (try_merge_delay_insns): Likewise.
4042 (redundant_insn): Likewise.
4043 (own_thread_p): Likewise.
4044 (fill_simple_delay_slots): Likewise.
4045 (fill_slots_from_thread): Likewise.
4046 (dbr_schedule): Likewise.
4047 * resource.c (update_live_status): Likewise.
4048 (mark_referenced_resources): Likewise.
4049 (mark_set_resources): Likewise.
4050 * rtl.c (copy_rtx): Likewise.
4051 * rtlanal.c (reg_referenced_p): Likewise.
4052 (single_set_2): Likewise.
4053 (noop_move_p): Likewise.
4054 (note_stores): Likewise.
4055 * sched-deps.c (sched_analyze_reg): Likewise.
4056 (sched_analyze_insn): Likewise.
4058 2018-08-06 Alan Hayward <alan.hayward@arm.com>
4060 * cse.c (invalidate_reg): New function extracted from...
4061 (invalidate): ...here.
4062 (canonicalize_insn): Check for clobber high.
4063 (invalidate_from_clobbers): invalidate clobber highs.
4064 (invalidate_from_sets_and_clobbers): Likewise.
4065 (count_reg_usage): Check for clobber high.
4066 (insn_live_p): Likewise.
4067 * cselib.c (cselib_expand_value_rtx_1):Likewise.
4068 (cselib_invalidate_regno): Check for clobber in setter.
4069 (cselib_invalidate_rtx): Pass through setter.
4070 (cselib_invalidate_rtx_note_stores):
4071 (cselib_process_insn): Check for clobber high.
4072 * cselib.h (cselib_invalidate_rtx): Add operand.
4074 2018-08-06 Alan Hayward <alan.hayward@arm.com>
4076 * lra-eliminations.c (lra_eliminate_regs_1): Check for clobber high.
4077 (mark_not_eliminable): Likewise.
4078 * lra-int.h (struct lra_insn_reg): Add clobber high marker.
4079 * lra-lives.c (process_bb_lives): Check for clobber high.
4080 * lra.c (new_insn_reg): Remember clobber highs.
4081 (collect_non_operand_hard_regs): Check for clobber high.
4082 (lra_set_insn_recog_data): Likewise.
4083 (add_regs_to_insn_regno_info): Likewise.
4084 (lra_update_insn_regno_info): Likewise.
4086 2018-08-06 Alan Hayward <alan.hayward@arm.com>
4088 * rtl.h (reg_is_clobbered_by_clobber_high): Add declarations.
4089 * rtlanal.c (reg_is_clobbered_by_clobber_high): Add function.
4091 2018-08-06 Alan Hayward <alan.hayward@arm.com>
4093 * emit-rtl.c (verify_rtx_sharing): Check for CLOBBER_HIGH.
4094 (copy_insn_1): Likewise.
4095 (gen_hard_reg_clobber_high): New gen function.
4096 * genconfig.c (walk_insn_part): Check for CLOBBER_HIGH.
4097 * genemit.c (gen_exp): Likewise.
4098 (gen_emit_seq): Pass through info.
4099 (gen_insn): Check for CLOBBER_HIGH.
4100 (gen_expand): Pass through info.
4101 (gen_split): Likewise.
4102 (output_add_clobbers): Likewise.
4103 * genrecog.c (validate_pattern): Check for CLOBBER_HIGH.
4104 (remove_clobbers): Likewise.
4105 * rtl.h (gen_hard_reg_clobber_high): New declaration.
4107 2018-08-06 Alan Hayward <alan.hayward@arm.com>
4109 * doc/rtl.texi (clobber_high): Add.
4110 (parallel): Add in clobber high
4111 * rtl.c (rtl_check_failed_code3): Add function.
4112 * rtl.def (CLOBBER_HIGH): Add expression.
4113 * rtl.h (RTL_CHECKC3): Add macro.
4114 (rtl_check_failed_code3): Add declaration.
4115 (XC3EXP): Add macro.
4117 2018-08-05 H.J. Lu <hongjiu.lu@intel.com>
4120 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Set
4121 cfun->machine->max_used_stack_alignment if needed.
4123 2018-08-04 Martin Sebor <msebor@redhat.com>
4125 PR tree-optimization/86571
4126 * gimple-ssa-sprintf.c (format_floating): Extend upper bound of
4129 2018-08-03 Sandra Loosemore <sandra@codesourcery.com>
4132 * config/nios2/nios2.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
4135 2018-08-03 Jeff Law <law@redhat.com>
4138 * config/mn10300/mn10300.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
4139 Define to speculation_safe_value_not_needed.
4141 2018-08-03 David Malcolm <dmalcolm@redhat.com>
4143 * doc/gcov.texi (-x): Remove duplicate "to".
4144 * doc/invoke.texi (-Wnoexcept-type): Remove duplicate "calls".
4145 (-Wif-not-aligned): Remove duplicate "is".
4146 (-flto): Remove duplicate "the".
4147 (MicroBlaze Options): In examples of "-mcpu=cpu-type", remove
4148 duplicate "v5.00.b".
4149 (MSP430 Options): Remove duplicate "and" from the description
4150 of "-mgprel-sec=regexp".
4151 (x86 Options): Remove duplicate copies of "vmldLog102" and
4152 vmlsLog104 from description of "-mveclibabi=type".
4154 2018-08-03 Richard Sandiford <richard.sandiford@arm.com>
4156 * internal-fn.h (first_commutative_argument): Declare.
4157 * internal-fn.c (first_commutative_argument): New function.
4158 * tree-vect-slp.c (vect_get_and_check_slp_defs): Remove extra
4159 restrictions for pattern statements. Use first_commutative_argument
4160 to look for commutative operands in calls to internal functions.
4162 2018-08-03 Aldy Hernandez <aldyh@redhat.com>
4164 * Makefile.in (wide-int-range.o): New.
4165 * tree-vrp.c: Move all the wide_int_* functions to...
4166 * wide-int-range.cc: ...here.
4167 * tree-vrp.h: Move all the wide_int_* prototypes to...
4168 * wide-int-range.h: ...here.
4170 2018-08-03 Tom de Vries <tdevries@suse.de>
4172 * common/config/nvptx/nvptx-common.c (nvptx_except_unwind_info): Return
4174 * config/nvptx/nvptx.c (TARGET_ASM_BYTE_OP): Remove define.
4175 * except.c (output_function_exception_table): Do early exit if
4176 targetm_common.except_unwind_info (&global_options) == UI_NONE.
4178 2018-08-03 Martin Liska <mliska@suse.cz>
4180 * predict.c (dump_prediction): Change to 2 digits
4183 2018-08-03 Siddhesh Poyarekar <siddhesh@sourceware.org>
4185 * config/aarch64/falkor.md (falkor_am_1_vxvy_vxvy): Move
4187 (falkor_am_1_gtov_gtov): ... a new insn reservation.
4189 2018-07-19 Ilya Leoshkevich <iii@linux.ibm.com>
4191 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Replace > with >=.
4192 * df-problems.c (df_remove_dead_eq_notes): Replace > with >=.
4193 * dwarf2out.c (mem_loc_descriptor): Replace > with >=.
4194 * lra-constraints.c (spill_hard_reg_in_range): Replace <= with <.
4195 * lra-remat.c (call_used_input_regno_present_p): Replace <= with <.
4197 2018-08-02 David Malcolm <dmalcolm@redhat.com>
4199 * diagnostic-show-locus.c (diagnostic_show_locus): Use
4200 pp_take_prefix when saving the existing prefix.
4201 * diagnostic.c (diagnostic_append_note): Likewise.
4202 * langhooks.c (lhd_print_error_function): Likewise.
4203 * pretty-print.c (pp_set_prefix): Drop the "const" from "prefix"
4204 param's type. Free the existing prefix.
4205 (pp_take_prefix): New function.
4206 (pretty_printer::pretty_printer): Drop the prefix parameter.
4207 Rename the length parameter to match the comment.
4208 (pretty_printer::~pretty_printer): Free the prefix.
4209 * pretty-print.h (pretty_printer::pretty_printer): Drop the prefix
4211 (struct pretty_printer): Drop the "const" from "prefix" field's
4212 type and clarify memory management.
4213 (pp_set_prefix): Drop the "const" from the 2nd param.
4214 (pp_take_prefix): New decl.
4216 2018-08-02 Aaron Sawdey <acsawdey@linux.ibm.com>
4218 * config/rs6000/rs6000-string.c (select_block_compare_mode): Move test
4219 for word_mode_ok here instead of passing as argument.
4220 (expand_block_compare): Change select_block_compare_mode() call.
4221 (expand_strncmp_gpr_sequence): New function.
4222 (expand_strn_compare): Make use of expand_strncmp_gpr_sequence.
4224 2018-08-02 Jeff Law <law@redhat.com>
4227 * config/m68k/m68k.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
4228 Define to speculation_safe_value_not_needed.
4231 * config/h8300/h8300.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
4232 Define to speculation_safe_value_not_needed.
4234 2018-08-02 Tom de Vries <tdevries@suse.de>
4237 * common/config/nvptx/nvptx-common.c (nvptx_except_unwind_info): New
4238 function. Return UI_TARGET unconditionally.
4239 (TARGET_EXCEPT_UNWIND_INFO): Redefine to nvptx_except_unwind_info.
4240 * config/nvptx/nvptx.c (TARGET_ASM_BYTE_OP): Emit commented out '.byte'.
4242 2018-08-02 Richard Sandiford <richard.sandiford@arm.com>
4244 * genemit.c (print_overload_test): Fix typo.
4246 2018-08-02 Richard Biener <rguenther@suse.de>
4248 PR tree-optimization/86816
4249 * tree-ssa-tail-merge.c (tail_merge_valueize): New function
4250 which checks for value availability before querying it.
4251 (gvn_uses_equal): Use it.
4252 (same_succ_hash): Likewise.
4253 (gimple_equal_p): Likewise.
4255 2018-08-02 Nick Clifton <nickc@redhat.com>
4258 * config/stormy16/stormy16.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
4259 Define to speculation_safe_value_not_needed.
4262 * config/v850/v850.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
4263 Define to speculation_safe_value_not_needed.
4266 * config/v850/v850.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
4267 Define to speculation_safe_value_not_needed.
4270 * config/rx/rx.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
4271 Define to speculation_safe_value_not_needed.
4274 * config/msp430/msp430.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
4275 Define to speculation_safe_value_not_needed.
4278 * config/mcore/mcore.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
4279 Define to speculation_safe_value_not_needed.
4282 * config/m32r/m32r.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
4283 Define to speculation_safe_value_not_needed.
4286 * config/iq2000/iq2000.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
4287 Define to speculation_safe_value_not_needed.
4290 * config/frv/frv.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Define to
4291 speculation_safe_value_not_needed.
4294 * config/fr30/fr30.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Define
4295 to speculation_safe_value_not_needed.
4297 2018-08-02 Richard Sandiford <richard.sandiford@arm.com>
4299 * doc/md.texi: Expand the documentation of instruction names
4300 to mention port-local uses. Document '@' in pattern names.
4301 * read-md.h (overloaded_instance, overloaded_name): New structs.
4303 (md_reader::handle_overloaded_name): New member function.
4304 (md_reader::get_overloads): Likewise.
4305 (md_reader::m_first_overload): New member variable.
4306 (md_reader::m_next_overload_ptr): Likewise.
4307 (md_reader::m_overloads_htab): Likewise.
4308 * read-md.c (md_reader::md_reader): Initialize m_first_overload,
4309 m_next_overload_ptr and m_overloads_htab.
4310 * read-rtl.c (iterator_group): Add "type" and "get_c_token" fields.
4311 (get_mode_token, get_code_token, get_int_token): New functions.
4312 (map_attr_string): Add an optional argument that passes back
4313 the associated iterator.
4314 (overloaded_name_hash, overloaded_name_eq_p, named_rtx_p):
4315 (md_reader::handle_overloaded_name, add_overload_instance): New
4317 (apply_iterators): Handle '@' names. Report an error if '@'
4318 is used without iterators.
4319 (initialize_iterators): Initialize the new iterator_group fields.
4320 * genopinit.c (handle_overloaded_code_for)
4321 (handle_overloaded_gen): New functions.
4322 (main): Use them to print declarations of maybe_code_for_* and
4323 maybe_gen_* functions, and inline definitions of code_for_* and gen_*.
4324 * genemit.c (print_overload_arguments, print_overload_test)
4325 (handle_overloaded_code_for, handle_overloaded_gen): New functions.
4326 (main): Use it to print definitions of maybe_code_for_* and
4327 maybe_gen_* functions.
4328 * config/aarch64/aarch64.c (aarch64_split_128bit_move): Use
4329 gen_aarch64_mov{low,high}_di and gen_aarch64_movdi_{low,high}
4330 instead of explicit mode checks.
4331 (aarch64_split_simd_combine): Likewise gen_aarch64_simd_combine.
4332 (aarch64_split_simd_move): Likewise gen_aarch64_split_simd_mov.
4333 (aarch64_emit_load_exclusive): Likewise gen_aarch64_load_exclusive.
4334 (aarch64_emit_store_exclusive): Likewise gen_aarch64_store_exclusive.
4335 (aarch64_expand_compare_and_swap): Likewise
4336 gen_aarch64_compare_and_swap and gen_aarch64_compare_and_swap_lse
4337 (aarch64_gen_atomic_cas): Likewise gen_aarch64_atomic_cas.
4338 (aarch64_emit_atomic_swap): Likewise gen_aarch64_atomic_swp.
4339 (aarch64_constant_pool_reload_icode): Delete.
4340 (aarch64_secondary_reload): Use code_for_aarch64_reload_movcp
4341 instead of aarch64_constant_pool_reload_icode. Use
4342 code_for_aarch64_reload_mov instead of explicit mode checks.
4343 (rsqrte_type, get_rsqrte_type, rsqrts_type, get_rsqrts_type): Delete.
4344 (aarch64_emit_approx_sqrt): Use gen_aarch64_rsqrte instead of
4345 get_rsqrte_type and gen_aarch64_rsqrts instead of gen_rqrts_type.
4346 (recpe_type, get_recpe_type, recps_type, get_recps_type): Delete.
4347 (aarch64_emit_approx_div): Use gen_aarch64_frecpe instead of
4348 get_recpe_type and gen_aarch64_frecps instead of get_recps_type.
4349 (aarch64_atomic_load_op_code): Delete.
4350 (aarch64_emit_atomic_load_op): Likewise.
4351 (aarch64_gen_atomic_ldop): Use UNSPECV_ATOMIC_* instead of
4352 aarch64_atomic_load_op_code. Use gen_aarch64_atomic_load
4353 instead of aarch64_emit_atomic_load_op.
4354 * config/aarch64/aarch64.md (aarch64_reload_movcp<GPF_TF:mode><P:mode>)
4355 (aarch64_reload_movcp<VALL:mode><P:mode>, aarch64_reload_mov<mode>)
4356 (aarch64_movdi_<mode>low, aarch64_movdi_<mode>high)
4357 (aarch64_mov<mode>high_di, aarch64_mov<mode>low_di): Add a '@'
4358 character before the pattern name.
4359 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
4360 (aarch64_rsqrte<mode>, aarch64_rsqrts<mode>)
4361 (aarch64_simd_combine<mode>, aarch64_frecpe<mode>)
4362 (aarch64_frecps<mode>): Likewise.
4363 * config/aarch64/atomics.md (atomic_compare_and_swap<mode>)
4364 (aarch64_compare_and_swap<mode>, aarch64_compare_and_swap<mode>_lse)
4365 (aarch64_load_exclusive<mode>, aarch64_store_exclusive<mode>)
4366 (aarch64_atomic_swp<mode>, aarch64_atomic_cas<mode>)
4367 (aarch64_atomic_load<atomic_ldop><mode>): Likewise.
4369 2018-08-02 Richard Sandiford <richard.sandiford@arm.com>
4371 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
4372 Allow HFmode constants if TARGET_FP_F16INST.
4374 2018-08-02 Jackson Woodruff <jackson.woodruff@arm.com>
4377 * config/aarch64/aarch64.c (aarch64_operands_adjust_ok_for_ldpstp):
4378 No longer check last store for clobber of address register.
4380 2018-08-02 Martin Liska <mliska@suse.cz>
4382 PR gcov-profile/86817
4383 * gcov.c (process_all_functions): New function.
4385 (process_file): Move functions processing to
4386 process_all_functions.
4388 2018-08-02 David Malcolm <dmalcolm@redhat.com>
4390 * dumpfile.c (dump_user_location_t::dump_user_location_t): Add
4391 "const" to the "gimple *" and "rtx_insn *" parameters.
4392 * dumpfile.h (dump_user_location_t::dump_user_location_t):
4394 (dump_location_t::dump_location_t): Likewise.
4396 2018-08-01 Martin Sebor <msebor@redhat.com>
4398 PR tree-optimization/86650
4399 * gimple-pretty-print.c (percent_G_format): Accept a "gimple *"
4400 rather than a "gcall *". Directly pass the data of interest
4401 to percent_K_format, rather than building a temporary CALL_EXPR
4403 * gimple-fold.c (gimple_fold_builtin_strncpy): Adjust.
4404 (gimple_fold_builtin_strncat): Adjust.
4405 * gimple-ssa-warn-restrict.h (check_bounds_or_overlap): Replace
4406 gcall* argument with gimple*.
4407 * gimple-ssa-warn-restrict.c (check_call): Same.
4408 (wrestrict_dom_walker::before_dom_children): Same.
4409 (builtin_access::builtin_access): Same.
4410 (check_bounds_or_overlap): Same
4411 (maybe_diag_overlap): Same.
4412 (maybe_diag_offset_bounds): Same.
4413 * tree-diagnostic.c (default_tree_printer): Move usage of
4414 EXPR_LOCATION (t) and TREE_BLOCK (t) from within percent_K_format
4416 * tree-pretty-print.c (percent_K_format): Add argument.
4417 * tree-pretty-print.h: Add argument.
4418 * tree-ssa-ccp.c (pass_post_ipa_warn::execute): Adjust.
4419 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Adjust.
4420 (maybe_diag_stxncpy_trunc): Same.
4421 (handle_builtin_stxncpy): Same.
4422 (handle_builtin_strcat): Same.
4424 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
4426 * match.pd: Optimise pointer range checks.
4428 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
4430 PR tree-optimization/86758
4431 * tree-vect-stmts.c (vectorizable_simd_clone_call): Don't try
4432 to remove pattern statements.
4434 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
4436 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Use the
4437 result of dfs_enumerate_from when constructing stmt_vec_infos,
4438 instead of additionally calling get_loop_body.
4440 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
4442 * tree-vectorizer.h (vect_create_data_ref_ptr): Remove inv_p
4444 * tree-vect-data-refs.c (vect_create_data_ref_ptr): Likewise.
4445 When creating an iv, assert that the step is not known to be zero.
4446 (vect_setup_realignment): Update call accordingly.
4447 * tree-vect-stmts.c (vectorizable_store): Likewise.
4448 (vectorizable_load): Likewise. Handle VMAT_INVARIANT separately.
4450 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
4452 * tree-vectorizer.h (vect_stmt_to_vectorize): New function.
4453 * tree-vect-loop.c (vect_update_vf_for_slp): Use it.
4454 (vectorizable_reduction): Likewise.
4455 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
4456 (vect_detect_hybrid_slp_stmts): Likewise.
4457 * tree-vect-stmts.c (vect_is_simple_use): Likewise.
4459 2018-08-01 Aldy Hernandez <aldyh@redhat.com>
4461 * tree-vrp (zero_nonzero_bits_from_bounds): Rename to...
4462 (wide_int_set_zero_nonzero_bits): ...this.
4463 (zero_nonzero_bits_from_vr): Rename to...
4464 (vrp_set_zero_nonzero_bits): ...this.
4465 (extract_range_from_multiplicative_op_1): Abstract wide int
4467 (wide_int_range_multiplicative_op): ...here.
4468 (extract_range_from_binary_expr_1): Extract wide int binary
4469 operations into their own functions.
4470 (wide_int_range_lshift): New.
4471 (wide_int_range_can_optimize_bit_op): New.
4472 (wide_int_range_shift_undefined_p): New.
4473 (wide_int_range_bit_xor): New.
4474 (wide_int_range_bit_ior): New.
4475 (wide_int_range_bit_and): New.
4476 (wide_int_range_trunc_mod): New.
4477 (extract_range_into_wide_ints): New.
4478 (vrp_shift_undefined_p): New.
4479 (extract_range_from_multiplicative_op): New.
4480 (vrp_can_optimize_bit_op): New.
4481 * tree-vrp.h (value_range::dump): New.
4482 (wide_int_range_multiplicative_op): New.
4483 (wide_int_range_lshift):New.
4484 (wide_int_range_shift_undefined_p): New.
4485 (wide_int_range_bit_xor): New.
4486 (wide_int_range_bit_ior): New.
4487 (wide_int_range_bit_and): New.
4488 (wide_int_range_trunc_mod): New.
4489 (zero_nonzero_bits_from_bounds): Rename to...
4490 (wide_int_set_zero_nonzero_bits): ...this.
4491 (zero_nonzero_bits_from_vr): Rename to...
4492 (vrp_set_zero_nonzero_bits): ...this.
4493 (range_easy_mask_min_max): Rename to...
4494 (wide_int_range_can_optimize_bit_op): this.
4495 * vr-values.c (simplify_bit_ops_using_ranges): Rename
4496 zero_nonzero_bits_from_vr into vrp_set_zero_nonzero_bits.
4498 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
4500 * tree-vectorizer.h (vect_orig_stmt): New function.
4501 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Use it.
4502 * tree-vect-loop.c (vect_model_reduction_cost): Likewise.
4503 (vect_create_epilog_for_reduction): Likewise.
4504 (vectorizable_live_operation): Likewise.
4505 * tree-vect-slp.c (vect_find_last_scalar_stmt_in_slp): Likewise.
4506 (vect_detect_hybrid_slp_stmts, vect_schedule_slp): Likewise.
4507 * tree-vect-stmts.c (vectorizable_call): Likewise.
4508 (vectorizable_simd_clone_call, vect_remove_stores): Likewise.
4510 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
4512 * tree-vectorizer.h (vect_transform_stmt): Remove grouped_store
4514 * tree-vect-stmts.c (vect_transform_stmt): Likewise.
4515 * tree-vect-loop.c (vect_transform_loop_stmt): Update call accordingly.
4516 (vect_transform_loop): Likewise.
4517 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
4519 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
4521 * tree-vectorizer.h (vect_schedule_slp): Return void.
4522 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
4523 (vect_schedule_slp): Likewise.
4525 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
4527 * tree-vect-loop.c (vect_transform_loop_stmt): Remove slp_scheduled
4529 (vect_transform_loop): Update calls accordingly. Schedule SLP
4530 instances before the main loop, if any exist.
4532 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
4534 PR tree-optimization/86749
4535 * tree-vect-patterns.c (vect_determine_min_output_precision_1):
4536 If the lhs is used in a COND_EXPR, check that it is being used
4537 as the "then" or "else" value.
4539 2018-08-01 Tom de Vries <tdevries@suse.de>
4542 * config/nvptx/nvptx.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Define to
4543 speculation_safe_value_not_needed.
4545 2018-08-01 Richard Biener <rguenther@suse.de>
4547 * tree-ssa-sccvn.c (visit_phi): Compare invariant addresses
4550 2018-08-01 Martin Liska <mliska@suse.cz>
4552 * value-prof.c (gimple_divmod_fixed_value_transform): Unify
4553 format how successful transformation is dumped.
4554 (gimple_mod_pow2_value_transform): Likewise.
4555 (gimple_mod_subtract_transform): Likewise.
4556 (gimple_stringops_transform): Likewise.
4558 2018-08-01 Martin Liska <mliska@suse.cz>
4561 * value-prof.c (interesting_stringop_to_profile_p):
4562 Simplify the code and add BUILT_IN_MEMMOVE.
4563 (gimple_stringops_transform): Likewise.
4565 2018-08-01 Sam Tebbs <sam.tebbs@arm.com>
4567 * config/aarch64/aarch64-simd.md
4568 (*aarch64_get_lane_zero_extendsi<mode>): Rename to...
4569 (*aarch64_get_lane_zero_extend<GPI:mode><VDQQH:mode>): ... This and
4570 use GPI iterator instead of SI mode.
4572 2018-08-01 Richard Earnshaw <rearnsha@arm.com>
4574 * config/rs6000/rs6000.md (speculation_barrier): Renamed from
4575 rs6000_speculation_barrier.
4576 * config/rs6000/rs6000.c (rs6000_expand_builtin): Adjust for
4577 new barrier pattern name.
4579 2018-08-01 Richard Earnshaw <rearnsha@arm.com>
4581 * config/i386/i386.md (unspecv): Add UNSPECV_SPECULATION_BARRIER.
4582 (speculation_barrier): New insn.
4584 2018-08-01 Richard Biener <rguenther@suse.de>
4587 * graphite.h: Include isl/id.h and isl/space.h to allow build
4590 2018-08-01 Jan Willem Jagersma <jwjagersma@gmail.com>
4593 * dwarf2out.c (dwarf2out_early_finish): Do not generate assembly in LTO
4594 mode for COFF targets.
4595 * defaults.h (TARGET_COFF): Define.
4596 * config/i386/djgpp.h (TARGET_ASM_LTO_START, TARGET_ASM_LTO_END,
4597 TARGET_COFF): Define.
4598 (i386_djgpp_asm_lto_start, i386_djgpp_asm_lto_end): Declare.
4599 * config/i386/djgpp.c (saved_debug_info_level): New static variable.
4600 (i386_djgpp_asm_lto_start, i386_djgpp_asm_lto_end): New functions.
4602 2018-07-31 Alexandre Oliva <oliva@adacore.com>
4604 * gimple-streamer-in.c (input_bb): Restore BB discriminator.
4605 * gimple-streamer-out.c (output_bb): Save it.
4606 * lto-streamer-in.c (input_struct_function_base): Restore
4607 instance discriminator if available. Create map on demand.
4608 * lto-streamer-out.c (output_struct_function_base): Save it if
4610 * final.c (decl_to_instance_map): Document LTO strategy.
4612 2018-07-31 Alexandre Oliva <oliva@adacore.com>
4613 Olivier Hainque <hainque@adacore.com>
4615 * debug.h (decl_to_instance_map_t): New type.
4616 (decl_to_instance_map): Declare.
4617 (maybe_create_decl_to_instance_map): New inline function.
4618 * final.c (bb_discriminator, last_bb_discriminator): New statics,
4619 to track basic block discriminators.
4620 (final_start_function_1): Initialize them.
4621 (final_scan_insn_1): On NOTE_INSN_BASIC_BLOCK, track
4623 (decl_to_instance_map): New variable.
4624 (map_decl_to_instance, maybe_set_discriminator): New functions.
4625 (notice_source_line): Set discriminator.
4627 2018-07-31 Ian Lance Taylor <iant@golang.org>
4629 * targhooks.c (default_have_speculation_safe_value): Add
4632 2018-07-31 David Malcolm <dmalcolm@redhat.com>
4634 * dump-context.h: Include "pretty-print.h".
4635 (dump_context::refresh_dumps_are_enabled): New decl.
4636 (dump_context::emit_item): New decl.
4637 (class dump_context): Add fields "m_test_pp" and
4639 (temp_dump_context::temp_dump_context): Add param "test_pp_flags".
4640 (temp_dump_context::get_dumped_text): New decl.
4641 (class temp_dump_context): Add field "m_pp".
4642 * dumpfile.c (refresh_dumps_are_enabled): Convert to...
4643 (dump_context::refresh_dumps_are_enabled): ...and add a test for
4645 (set_dump_file): Update for above change.
4646 (set_alt_dump_file): Likewise.
4647 (dump_loc): New overload, taking a pretty_printer *.
4648 (dump_context::dump_loc): Call end_any_optinfo. Dump the location
4649 to any test pretty-printer.
4650 (make_item_for_dump_gimple_stmt): New function, adapted from
4651 optinfo::add_gimple_stmt.
4652 (dump_context::dump_gimple_stmt): Call it, and use the result,
4653 eliminating the direct usage of dump_file and alt_dump_file in
4654 favor of indirectly using them via emit_item.
4655 (make_item_for_dump_gimple_expr): New function, adapted from
4656 optinfo::add_gimple_expr.
4657 (dump_context::dump_gimple_expr): Call it, and use the result,
4658 eliminating the direct usage of dump_file and alt_dump_file in
4659 favor of indirectly using them via emit_item.
4660 (make_item_for_dump_generic_expr): New function, adapted from
4662 (dump_context::dump_generic_expr): Call it, and use the result,
4663 eliminating the direct usage of dump_file and alt_dump_file in
4664 favor of indirectly using them via emit_item.
4665 (make_item_for_dump_printf_va): New function, adapted from
4666 optinfo::add_printf_va.
4667 (make_item_for_dump_printf): New function.
4668 (dump_context::dump_printf_va): Call make_item_for_dump_printf_va,
4669 and use the result, eliminating the direct usage of dump_file and
4670 alt_dump_file in favor of indirectly using them via emit_item.
4671 (make_item_for_dump_dec): New function.
4672 (dump_context::dump_dec): Call it, and use the result,
4673 eliminating the direct usage of dump_file and alt_dump_file in
4674 favor of indirectly using them via emit_item.
4675 (make_item_for_dump_symtab_node): New function, adapted from
4676 optinfo::add_symtab_node.
4677 (dump_context::dump_symtab_node): Call it, and use the result,
4678 eliminating the direct usage of dump_file and alt_dump_file in
4679 favor of indirectly using them via emit_item.
4680 (dump_context::begin_scope): Reimplement, avoiding direct usage
4681 of dump_file and alt_dump_file in favor of indirectly using them
4683 (dump_context::emit_item): New member function.
4684 (temp_dump_context::temp_dump_context): Add param "test_pp_flags".
4685 Set up test pretty-printer on the underlying context. Call
4686 refresh_dumps_are_enabled.
4687 (temp_dump_context::~temp_dump_context): Call
4688 refresh_dumps_are_enabled.
4689 (temp_dump_context::get_dumped_text): New member function.
4690 (selftest::verify_dumped_text): New function.
4691 (ASSERT_DUMPED_TEXT_EQ): New macro.
4692 (selftest::test_capture_of_dump_calls): Run all tests twice, with
4693 and then without optinfo enabled. Add uses of
4694 ASSERT_DUMPED_TEXT_EQ to all tests. Add test of nested scopes.
4695 * dumpfile.h: Update comment for the dump_* API.
4696 * optinfo-emit-json.cc
4697 (selftest::test_building_json_from_dump_calls): Update for new
4698 param for temp_dump_context ctor.
4699 * optinfo.cc (optinfo_item::optinfo_item): Remove "owned" param
4700 and "m_owned" field.
4701 (optinfo_item::~optinfo_item): Likewise.
4702 (optinfo::add_item): New member function.
4703 (optinfo::emit): Update comment.
4704 (optinfo::add_string): Delete.
4705 (optinfo::add_printf): Delete.
4706 (optinfo::add_printf_va): Delete.
4707 (optinfo::add_gimple_stmt): Delete.
4708 (optinfo::add_gimple_expr): Delete.
4709 (optinfo::add_tree): Delete.
4710 (optinfo::add_symtab_node): Delete.
4711 (optinfo::add_dec): Delete.
4712 * optinfo.h (class dump_context): New forward decl.
4713 (optinfo::add_item): New decl.
4714 (optinfo::add_string): Delete.
4715 (optinfo::add_printf): Delete.
4716 (optinfo::add_printf_va): Delete.
4717 (optinfo::add_gimple_stmt): Delete.
4718 (optinfo::add_gimple_expr): Delete.
4719 (optinfo::add_tree): Delete.
4720 (optinfo::add_symtab_node): Delete.
4721 (optinfo::add_dec): Delete.
4722 (optinfo::add_poly_int): Delete.
4723 (optinfo_item::optinfo_item): Remove "owned" param.
4724 (class optinfo_item): Remove field "m_owned".
4726 2018-07-31 Jozef Lawrynowicz <jozef.l@mittosystems.com>
4729 * gcc/cfgexpand.c (set_parm_rtl): Use the alignment of Pmode when
4730 MAX_SUPPORTED_STACK_ALIGNMENT would otherwise be exceeded by the
4731 requested variable alignment.
4732 (expand_one_ssa_partition): Likewise.
4733 (expand_one_var): Likewise.
4735 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
4737 * config/pdp11/pdp11.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Redefine
4738 to speculation_safe_value_not_needed.
4740 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
4742 * targhooks.h (speculation_safe_value_not_needed): New prototype.
4743 * targhooks.c (speculation_safe_value_not_needed): New function.
4744 * target.def (have_speculation_safe_value): Update documentation.
4745 * doc/tm.texi: Regenerated.
4747 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
4749 * config/aarch64/iterators.md (ALLI_TI): New iterator.
4750 * config/aarch64/aarch64.md (despeculate_copy<ALLI_TI:mode>): New
4752 (despeculate_copy<ALLI:mode>_insn): New insn.
4753 (despeculate_copyti_insn): New insn.
4754 (despeculate_simple<ALLI:mode>): New insn
4755 (despeculate_simpleti): New insn.
4756 * config/aarch64/aarch64.c (aarch64_speculation_safe_value): New
4758 (TARGET_SPECULATION_SAFE_VALUE): Redefine to
4759 aarch64_speculation_safe_value.
4760 (aarch64_print_operand): Handle const0_rtx in modifier 'H'.
4762 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
4764 * config/aarch64/aarch64-speculation.cc: New file.
4765 * config/aarch64/aarch64-passes.def (pass_track_speculation): Add
4766 before pass_reorder_blocks.
4767 * config/aarch64/aarch64-protos.h (make_pass_track_speculation): Add
4769 * config/aarch64/aarch64.c (aarch64_conditional_register_usage): Fix
4770 X14 and X15 when tracking speculation.
4771 * config/aarch64/aarch64.md (register name constants): Add
4772 SPECULATION_TRACKER_REGNUM and SPECULATION_SCRATCH_REGNUM.
4773 (unspec): Add UNSPEC_SPECULATION_TRACKER.
4774 (speculation_barrier): New insn attribute.
4775 (cmp<mode>): Allow SP in comparisons.
4776 (speculation_tracker): New insn.
4777 (speculation_barrier): Add speculation_barrier attribute.
4778 * config/aarch64/t-aarch64: Add make rule for aarch64-speculation.o.
4779 * config.gcc (aarch64*-*-*): Add aarch64-speculation.o to extra_objs.
4780 * doc/invoke.texi (AArch64 Options): Document -mtrack-speculation.
4782 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
4784 * config/aarch64/aarch64.md (cb<optab><mode>1): Disable when
4785 aarch64_track_speculation is true.
4786 (tb<optab><mode>1): Likewise.
4787 * config/aarch64/aarch64.c (aarch64_split_compare_regs): Do not
4788 generate CB[N]Z when tracking speculation.
4789 (aarch64_split_compare_and_swap): Likewise.
4790 (aarch64_split_atomic_op): Likewise.
4792 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
4794 * config/aarch64/aarch64.opt (mtrack-speculation): New target option.
4796 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
4798 * config/aarch64.md (unspecv): Add UNSPECV_SPECULAION_BARRIER.
4799 (speculation_barrier): New insn.
4801 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
4803 * config/arm/unspecs.md (unspecv): Add VUNSPEC_SPECULATION_BARRIER.
4804 * config/arm/arm.md (speculation_barrier): New expand.
4805 (speculation_barrier_insn): New pattern.
4807 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
4809 * builtin-types.def (BT_FN_PTR_PTR_VAR): New function type.
4810 (BT_FN_I1_I1_VAR, BT_FN_I2_I2_VAR, BT_FN_I4_I4_VAR): Likewise.
4811 (BT_FN_I8_I8_VAR, BT_FN_I16_I16_VAR): Likewise.
4812 * builtin-attrs.def (ATTR_NOVOPS_NOTHROW_LEAF_LIST): New attribute
4814 * builtins.def (BUILT_IN_SPECULATION_SAFE_VALUE_N): New builtin.
4815 (BUILT_IN_SPECULATION_SAFE_VALUE_PTR): New internal builtin.
4816 (BUILT_IN_SPECULATION_SAFE_VALUE_1): Likewise.
4817 (BUILT_IN_SPECULATION_SAFE_VALUE_2): Likewise.
4818 (BUILT_IN_SPECULATION_SAFE_VALUE_4): Likewise.
4819 (BUILT_IN_SPECULATION_SAFE_VALUE_8): Likewise.
4820 (BUILT_IN_SPECULATION_SAFE_VALUE_16): Likewise.
4821 * builtins.c (expand_speculation_safe_value): New function.
4822 (expand_builtin): Call it.
4823 * doc/cpp.texi: Document predefine __HAVE_SPECULATION_SAFE_VALUE.
4824 * doc/extend.texi: Document __builtin_speculation_safe_value.
4825 * doc/md.texi: Document "speculation_barrier" pattern.
4826 * doc/tm.texi.in: Pull in TARGET_SPECULATION_SAFE_VALUE and
4827 TARGET_HAVE_SPECULATION_SAFE_VALUE.
4828 * doc/tm.texi: Regenerated.
4829 * target.def (have_speculation_safe_value, speculation_safe_value): New
4831 * targhooks.c (default_have_speculation_safe_value): New function.
4832 (default_speculation_safe_value): New function.
4833 * targhooks.h (default_have_speculation_safe_value): Add prototype.
4834 (default_speculation_safe_value): Add prototype.
4836 2018-07-31 David Malcolm <dmalcolm@redhat.com>
4838 * dump-context.h (dump_context::dump_loc): New decl.
4839 * dumpfile.c (dump_context::dump_loc): New member function.
4840 (dump_context::dump_gimple_stmt_loc): Reimplement using dump_loc
4841 and dump_gimple_stmt.
4842 (dump_context::dump_gimple_expr_loc): Likewise, using
4844 (dump_context::dump_generic_expr_loc): Likewise, using
4846 (dump_context::dump_printf_loc_va): Likewise, using
4848 (dump_context::begin_scope): Explicitly using the global function
4849 "dump_loc", rather than the member function.
4851 2018-07-31 Martin Sebor <msebor@redhat.com>
4853 PR tree-optimization/86741
4854 * tree-vrp.c (vrp_prop::check_mem_ref): Avoid incomplete types.
4856 2018-07-31 Andreas Krebbel <krebbel@linux.ibm.com>
4858 * config/s390/s390.c (s390_expand_setmem): Make the unrolling to
4859 depend on whether prefetch instructions will be emitted or not.
4860 Use TARGET_SETMEM_PFD for checking whether prefetch instructions
4861 will be emitted or not.
4862 * config/s390/s390.h (TARGET_SETMEM_PREFETCH_DISTANCE)
4863 (TARGET_SETMEM_PFD): New macros.
4865 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4867 * tree-vectorizer.h (stmt_vec_info): Turn back into a typedef.
4868 (NULL_STMT_VEC_INFO): Delete.
4869 (stmt_vec_info::operator*): Likewise.
4870 (stmt_vec_info::operator gimple *): Likewise.
4871 * tree-vect-loop.c (vectorizable_reduction): Use NULL instead
4872 of NULL_STMT_VEC_INFO.
4873 * tree-vect-patterns.c (vect_init_pattern_stmt): Likewise.
4874 (vect_reassociating_reduction_p): Likewise.
4875 * tree-vect-stmts.c (vect_build_gather_load_calls): Likewise.
4876 (vectorizable_store): Likewise.
4877 * tree-vectorizer.c (vec_info::set_vinfo_for_stmt): Likewise.
4878 (vec_info::free_stmt_vec_infos): Likewise.
4880 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4882 * tree-vectorizer.h (vect_stmt_in_region_p): Delete.
4883 * tree-vectorizer.c (vect_stmt_in_region_p): Likewise.
4885 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4887 * tree-vectorizer.h (vec_info::new_vinfo_for_stmt)
4888 (vec_info::set_vinfo_for_stmt, vec_info::free_stmt_vec_infos)
4889 (vec_info::free_stmt_vec_info): New private member functions.
4890 (set_stmt_vec_info_vec, free_stmt_vec_infos, vinfo_for_stmt)
4891 (set_vinfo_for_stmt, new_stmt_vec_info, free_stmt_vec_info): Delete.
4892 * tree-parloops.c (gather_scalar_reductions): Remove calls to
4893 set_stmt_vec_info_vec and free_stmt_vec_infos.
4894 * tree-vect-loop.c (_loop_vec_info): Remove call to
4895 set_stmt_vec_info_vec.
4896 * tree-vect-stmts.c (new_stmt_vec_info, set_stmt_vec_info_vec)
4897 (free_stmt_vec_infos, free_stmt_vec_info): Delete in favor of...
4898 * tree-vectorizer.c (vec_info::new_stmt_vec_info)
4899 (vec_info::set_vinfo_for_stmt, vec_info::free_stmt_vec_infos)
4900 (vec_info::free_stmt_vec_info): ...these new functions. Remove
4901 assignments in {vec_info::,}new_stmt_vec_info that are redundant
4902 with the clearing in the xcalloc.
4903 (stmt_vec_info_vec): Delete.
4904 (vec_info::vec_info): Don't call set_stmt_vec_info_vec.
4905 (vectorize_loops): Likewise.
4906 (vec_info::~vec_info): Remove argument from call to
4907 free_stmt_vec_infos.
4908 (vec_info::add_stmt): Remove vinfo argument from call to
4911 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4913 * tree-vectorizer.h (free_stmt_vec_info): Take a stmt_vec_info
4914 rather than a gimple stmt.
4915 * tree-vect-stmts.c (free_stmt_vec_info): Likewise. Don't free
4916 information for pattern statements when passed the original
4917 statement; instead wait to be passed the pattern statement itself.
4918 Don't call set_vinfo_for_stmt here.
4919 (free_stmt_vec_infos): Update call to free_stmt_vec_info.
4920 * tree-vect-loop.c (_loop_vec_info::~loop_vec_info): Don't free
4921 stmt_vec_infos here.
4922 * tree-vect-slp.c (_bb_vec_info::~bb_vec_info): Likewise.
4923 * tree-vectorizer.c (vec_info::remove_stmt): Nullify the statement's
4924 stmt_vec_infos entry.
4926 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4928 * tree-vectorizer.h (vec_info::replace_stmt): Declare.
4929 * tree-vectorizer.c (vec_info::replace_stmt): New function.
4930 * tree-vect-slp.c (vect_remove_slp_scalar_calls): Use it.
4931 * tree-vect-stmts.c (vectorizable_call): Likewise.
4932 (vectorizable_simd_clone_call): Likewise.
4934 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4936 * tree-vectorizer.h (vec_info::remove_stmt): Declare.
4937 * tree-vectorizer.c (vec_info::remove_stmt): New function.
4938 * tree-vect-loop-manip.c (vect_set_loop_condition): Use it.
4939 * tree-vect-loop.c (vect_transform_loop): Likewise.
4940 * tree-vect-slp.c (vect_schedule_slp): Likewise.
4941 * tree-vect-stmts.c (vect_remove_stores): Likewise.
4943 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4945 * tree-vectorizer.h (vec_info::lookup_dr): New member function.
4946 (vect_dr_stmt): Delete.
4947 * tree-vectorizer.c (vec_info::lookup_dr): New function.
4948 * tree-vect-loop-manip.c (vect_update_inits_of_drs): Use it instead
4950 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr)
4951 (vect_analyze_data_ref_dependence, vect_record_base_alignments)
4952 (vect_verify_datarefs_alignment, vect_peeling_supportable)
4953 (vect_analyze_data_ref_accesses, vect_prune_runtime_alias_test_list)
4954 (vect_analyze_data_refs): Likewise.
4955 (vect_slp_analyze_data_ref_dependence): Likewise. Take a vec_info
4957 (vect_find_same_alignment_drs): Likewise.
4958 (vect_slp_analyze_node_dependences): Update calls accordingly.
4959 (vect_analyze_data_refs_alignment): Likewise. Use vec_info::lookup_dr
4960 instead of DR_VECT_AUX.
4961 (vect_get_peeling_costs_all_drs): Take a loop_vec_info instead
4962 of a vector data references. Use vec_info::lookup_dr instead of
4964 (vect_peeling_hash_get_lowest_cost): Update calls accordingly.
4965 (vect_enhance_data_refs_alignment): Likewise. Use vec_info::lookup_dr
4966 instead of DR_VECT_AUX.
4968 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4970 * tree-vectorizer.h (_loop_vec_info::unaligned_dr): Change to
4972 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Update
4974 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
4975 * tree-vect-loop-manip.c (get_misalign_in_elems): Likewise.
4976 (vect_gen_prolog_loop_niters): Likewise.
4978 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4980 * tree-vectorizer.h (set_dr_misalignment, dr_misalignment)
4981 (DR_TARGET_ALIGNMENT, aligned_access_p, known_alignment_for_access_p)
4982 (vect_known_alignment_in_bytes, vect_dr_behavior)
4983 (vect_get_scalar_dr_size): Take references as dr_vec_infos
4984 instead of data_references. Update calls to other routines for
4985 which the same change has been made.
4986 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Take
4987 dr_vec_infos instead of stmt_vec_infos.
4988 (vect_analyze_data_ref_dependence): Update call accordingly.
4989 (vect_slp_analyze_data_ref_dependence)
4990 (vect_record_base_alignments): Use DR_VECT_AUX.
4991 (vect_calculate_target_alignment, vect_compute_data_ref_alignment)
4992 (vect_update_misalignment_for_peel, verify_data_ref_alignment)
4993 (vector_alignment_reachable_p, vect_get_data_access_cost)
4994 (vect_peeling_supportable, vect_analyze_group_access_1)
4995 (vect_analyze_group_access, vect_analyze_data_ref_access)
4996 (vect_vfa_segment_size, vect_vfa_access_size, vect_vfa_align)
4997 (vect_compile_time_alias, vect_small_gap_p)
4998 (vectorizable_with_step_bound_p, vect_duplicate_ssa_name_ptr_info):
4999 (vect_supportable_dr_alignment): Take references as dr_vec_infos
5000 instead of data_references. Update calls to other routines for
5001 which the same change has been made.
5002 (vect_verify_datarefs_alignment, vect_get_peeling_costs_all_drs)
5003 (vect_find_same_alignment_drs, vect_analyze_data_refs_alignment)
5004 (vect_slp_analyze_and_verify_node_alignment)
5005 (vect_analyze_data_ref_accesses, vect_prune_runtime_alias_test_list)
5006 (vect_create_addr_base_for_vector_ref, vect_create_data_ref_ptr)
5007 (vect_setup_realignment): Use dr_vec_infos. Update calls after
5009 (_vect_peel_info::dr): Replace with...
5010 (_vect_peel_info::dr_info): ...this new field.
5011 (vect_peeling_hash_get_most_frequent)
5012 (vect_peeling_hash_choose_best_peeling): Update accordingly.
5013 (vect_peeling_hash_get_lowest_cost):
5014 (vect_enhance_data_refs_alignment): Likewise. Update calls to other
5015 routines for which the same change has been made.
5016 (vect_peeling_hash_insert): Likewise. Take a dr_vec_info instead of a
5018 * tree-vect-loop-manip.c (get_misalign_in_elems)
5019 (vect_gen_prolog_loop_niters): Use dr_vec_infos. Update calls after
5021 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
5022 * tree-vect-stmts.c (vect_get_store_cost, vect_get_load_cost)
5023 (vect_truncate_gather_scatter_offset, compare_step_with_zero)
5024 (get_group_load_store_type, get_negative_load_store_type)
5025 (vect_get_data_ptr_increment, vectorizable_store)
5026 (vectorizable_load): Likewise.
5027 (ensure_base_align): Take a dr_vec_info instead of a data_reference.
5028 Update calls to other routines for which the same change has been made.
5030 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
5032 * tree-vectorizer.h (vec_info::move_dr): New member function.
5033 (dataref_aux): Rename to...
5034 (dr_vec_info): ...this and add "dr" and "stmt" fields.
5035 (_stmt_vec_info::dr_aux): Update accordingly.
5036 (_stmt_vec_info::data_ref_info): Delete.
5037 (STMT_VINFO_GROUPED_ACCESS, DR_GROUP_FIRST_ELEMENT)
5038 (DR_GROUP_NEXT_ELEMENT, DR_GROUP_SIZE, DR_GROUP_STORE_COUNT)
5039 (DR_GROUP_GAP, DR_GROUP_SAME_DR_STMT, REDUC_GROUP_FIRST_ELEMENT):
5040 (REDUC_GROUP_NEXT_ELEMENT, REDUC_GROUP_SIZE): Use dr_aux.dr instead
5042 (STMT_VINFO_DATA_REF): Likewise. Turn into an lvalue.
5043 (STMT_VINFO_DR_INFO): New macro.
5044 (DR_VECT_AUX): Use STMT_VINFO_DR_INKFO and vect_dr_stmt.
5045 (set_dr_misalignment): Update after rename of dataref_aux.
5046 (vect_dr_stmt): Move earlier in file. Return dr_aux.stmt.
5047 * tree-vect-stmts.c (new_stmt_vec_info): Remove redundant
5048 initialization of STMT_VINFO_DATA_REF.
5049 * tree-vectorizer.c (vec_info::move_dr): New function.
5050 * tree-vect-patterns.c (vect_recog_bool_pattern)
5051 (vect_recog_mask_conversion_pattern)
5052 (vect_recog_gather_scatter_pattern): Use it.
5053 * tree-vect-data-refs.c (vect_analyze_data_refs): Initialize
5054 the "dr" and "stmt" fields of dr_vec_info instead of
5055 STMT_VINFO_DATA_REF.
5057 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
5059 * tree-vectorizer.h (_stmt_vec_info::pattern_stmt_p): New field.
5060 (is_pattern_stmt_p): Use it.
5061 * tree-vect-patterns.c (vect_init_pattern_stmt): Set pattern_stmt_p
5062 on pattern statements.
5064 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
5066 * tree-vect-patterns.c (vect_mark_pattern_stmts): Take the
5067 original stmt as a stmt_vec_info rather than a gimple stmt.
5068 (vect_pattern_recog_1): Take the statement directly as a
5069 stmt_vec_info, rather than via a gimple_stmt_iterator.
5070 Update call to vect_mark_pattern_stmts.
5071 (vect_pattern_recog): Update calls accordingly.
5073 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
5075 * tree-vectorizer.h (vect_get_vec_defs_for_stmt_copy)
5076 (vect_get_vec_def_for_stmt_copy): Take a vec_info rather than
5077 a vect_def_type for the first argument.
5078 * tree-vect-stmts.c (vect_get_vec_defs_for_stmt_copy): Likewise.
5079 (vect_get_vec_def_for_stmt_copy): Likewise. Return the original
5080 operand if it isn't defined by a vectorized statement.
5081 (vect_build_gather_load_calls): Remove the mask_dt argument and
5082 update calls to vect_get_vec_def_for_stmt_copy.
5083 (vectorizable_bswap): Likewise the dt argument.
5084 (vectorizable_call): Update calls to vectorizable_bswap and
5085 vect_get_vec_def_for_stmt_copy.
5086 (vectorizable_simd_clone_call, vectorizable_assignment)
5087 (vectorizable_shift, vectorizable_operation, vectorizable_condition)
5088 (vectorizable_comparison): Update calls to
5089 vect_get_vec_def_for_stmt_copy.
5090 (vectorizable_store): Likewise. Remove now-unnecessary calls to
5092 (vect_get_loop_based_defs): Remove dt argument and update call
5093 to vect_get_vec_def_for_stmt_copy.
5094 (vectorizable_conversion): Update calls to vect_get_loop_based_defs
5095 and vect_get_vec_def_for_stmt_copy.
5096 (vectorizable_load): Update calls to vect_build_gather_load_calls
5097 and vect_get_vec_def_for_stmt_copy.
5098 * tree-vect-loop.c (vect_create_epilog_for_reduction)
5099 (vectorizable_reduction, vectorizable_live_operation): Update calls
5100 to vect_get_vec_def_for_stmt_copy.
5102 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
5104 * tree-vect-data-refs.c (vect_record_base_alignment): Replace vec_info
5105 and gimple stmt arguments with a stmt_vec_info.
5106 (vect_record_base_alignments): Update calls accordingly.
5107 * tree-vect-slp.c (vect_record_max_nunits): Replace vec_info
5108 and gimple stmt arguments with a stmt_vec_info.
5109 (vect_build_slp_tree_1): Remove vinfo argument and update call
5110 to vect_record_max_nunits.
5111 (vect_build_slp_tree_2): Update calls to vect_build_slp_tree_1
5112 and vect_record_max_nunits.
5114 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
5116 * tree-vectorizer.h (nested_in_vect_loop_p): Move further down
5117 file and take a stmt_vec_info instead of a gimple stmt.
5118 (supportable_widening_operation, vect_finish_replace_stmt)
5119 (vect_finish_stmt_generation, vect_get_store_rhs)
5120 (vect_get_vec_def_for_operand_1, vect_get_vec_def_for_operand)
5121 (vect_get_vec_defs, vect_init_vector, vect_transform_stmt)
5122 (vect_remove_stores, vect_analyze_stmt, vectorizable_condition)
5123 (vect_get_smallest_scalar_type, vect_check_gather_scatter)
5124 (vect_create_data_ref_ptr, bump_vector_ptr)
5125 (vect_permute_store_chain, vect_setup_realignment)
5126 (vect_transform_grouped_load, vect_record_grouped_load_vectors)
5127 (vect_create_addr_base_for_vector_ref, vectorizable_live_operation)
5128 (vectorizable_reduction, vectorizable_induction)
5129 (get_initial_def_for_reduction, is_simple_and_all_uses_invariant)
5130 (vect_get_place_in_interleaving_chain): Take stmt_vec_infos rather
5131 than gimple stmts as arguments.
5132 * tree-vect-data-refs.c (vect_get_smallest_scalar_type)
5133 (vect_preserves_scalar_order_p, vect_slp_analyze_node_dependences)
5134 (can_group_stmts_p, vect_check_gather_scatter)
5135 (vect_create_addr_base_for_vector_ref, vect_create_data_ref_ptr)
5136 (bump_vector_ptr, vect_permute_store_chain, vect_setup_realignment)
5137 (vect_permute_load_chain, vect_shift_permute_load_chain)
5138 (vect_transform_grouped_load)
5139 (vect_record_grouped_load_vectors): Likewise.
5140 * tree-vect-loop.c (vect_fixup_reduc_chain)
5141 (get_initial_def_for_reduction, vect_create_epilog_for_reduction)
5142 (vectorize_fold_left_reduction, is_nonwrapping_integer_induction)
5143 (vectorizable_reduction, vectorizable_induction)
5144 (vectorizable_live_operation, vect_loop_kill_debug_uses): Likewise.
5145 * tree-vect-patterns.c (type_conversion_p, adjust_bool_stmts)
5146 (vect_get_load_store_mask): Likewise.
5147 * tree-vect-slp.c (vect_get_place_in_interleaving_chain)
5148 (vect_analyze_slp_instance, vect_mask_constant_operand_p): Likewise.
5149 * tree-vect-stmts.c (vect_mark_relevant)
5150 (is_simple_and_all_uses_invariant)
5151 (exist_non_indexing_operands_for_use_p, process_use)
5152 (vect_init_vector_1, vect_init_vector, vect_get_vec_def_for_operand_1)
5153 (vect_get_vec_def_for_operand, vect_get_vec_defs)
5154 (vect_finish_stmt_generation_1, vect_finish_replace_stmt)
5155 (vect_finish_stmt_generation, vect_truncate_gather_scatter_offset)
5156 (compare_step_with_zero, vect_get_store_rhs, get_group_load_store_type)
5157 (get_negative_load_store_type, get_load_store_type)
5158 (vect_check_load_store_mask, vect_check_store_rhs)
5159 (vect_build_gather_load_calls, vect_get_strided_load_store_ops)
5160 (vectorizable_bswap, vectorizable_call, vectorizable_simd_clone_call)
5161 (vect_create_vectorized_demotion_stmts, vectorizable_conversion)
5162 (vectorizable_assignment, vectorizable_shift, vectorizable_operation)
5163 (get_group_alias_ptr_type, vectorizable_store, hoist_defs_of_uses)
5164 (vectorizable_load, vectorizable_condition, vectorizable_comparison)
5165 (vect_analyze_stmt, vect_transform_stmt, vect_remove_stores)
5166 (supportable_widening_operation): Likewise.
5168 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
5170 * tree-vect-data-refs.c (vect_describe_gather_scatter_call): Take
5171 a stmt_vec_info instead of a gcall.
5172 (vect_check_gather_scatter): Update call accordingly.
5173 * tree-vect-loop-manip.c (iv_phi_p): Take a stmt_vec_info instead
5175 (vect_can_advance_ivs_p, vect_update_ivs_after_vectorizer)
5176 (slpeel_update_phi_nodes_for_loops):): Update calls accordingly.
5177 * tree-vect-loop.c (vect_transform_loop_stmt): Take a stmt_vec_info
5178 instead of a gimple stmt.
5179 (vect_transform_loop): Update calls accordingly.
5180 * tree-vect-slp.c (vect_split_slp_store_group): Take and return
5181 stmt_vec_infos instead of gimple stmts.
5182 (vect_analyze_slp_instance): Update use accordingly.
5183 * tree-vect-stmts.c (read_vector_array, write_vector_array)
5184 (vect_clobber_variable, vect_stmt_relevant_p, permute_vec_elements)
5185 (vect_use_strided_gather_scatters_p, vect_build_all_ones_mask)
5186 (vect_build_zero_merge_argument, vect_get_gather_scatter_ops)
5187 (vect_gen_widened_results_half, vect_get_loop_based_defs)
5188 (vect_create_vectorized_promotion_stmts, can_vectorize_live_stmts):
5189 Take a stmt_vec_info instead of a gimple stmt and pass stmt_vec_infos
5190 down to subroutines.
5192 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
5194 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Change the type
5195 of the worklist from a vector of gimple stmts to a vector of
5197 * tree-vect-stmts.c (vect_mark_relevant, process_use)
5198 (vect_mark_stmts_to_be_vectorized): Likewise
5200 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
5202 * tree-vect-loop.c (vect_analyze_loop_operations): Look up the
5203 statement before passing it to vect_analyze_stmt.
5204 (vect_create_epilog_for_reduction): Use a stmt_vec_info to walk
5205 the chain of phi vector definitions. Track the exit phi via its
5207 (vectorizable_reduction): Set cond_stmt_vinfo directly from the
5208 STMT_VINFO_REDUC_DEF.
5209 * tree-vect-slp.c (vect_get_place_in_interleaving_chain): Use
5210 stmt_vec_infos to handle the statement chains.
5211 (vect_get_slp_defs): Record the first statement in the node
5212 using a stmt_vec_info.
5213 * tree-vect-stmts.c (vect_mark_stmts_to_be_vectorized): Look up
5214 statements here and pass their stmt_vec_info down to subroutines.
5215 (vect_init_vector_1): Hoist call to vinfo_for_stmt and pass it
5216 down to vect_finish_stmt_generation.
5217 (vect_init_vector, vect_get_vec_defs, vect_finish_replace_stmt)
5218 (vect_finish_stmt_generation): Call vinfo_for_stmt and pass
5219 stmt_vec_infos to subroutines.
5220 (vect_remove_stores): Use stmt_vec_infos to handle the statement
5223 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
5225 * tree-vect-data-refs.c (vect_slp_analyze_node_dependences):
5226 (vect_check_gather_scatter, vect_create_data_ref_ptr, bump_vector_ptr)
5227 (vect_permute_store_chain, vect_setup_realignment)
5228 (vect_permute_load_chain, vect_shift_permute_load_chain)
5229 (vect_transform_grouped_load): Use stmt_vec_info rather than gimple
5230 stmts internally, and when passing values to other vectorizer routines.
5231 * tree-vect-loop-manip.c (vect_can_advance_ivs_p): Likewise.
5232 * tree-vect-loop.c (vect_analyze_scalar_cycles_1)
5233 (vect_analyze_loop_operations, get_initial_def_for_reduction)
5234 (vect_create_epilog_for_reduction, vectorize_fold_left_reduction)
5235 (vectorizable_reduction, vectorizable_induction)
5236 (vectorizable_live_operation, vect_transform_loop_stmt)
5237 (vect_transform_loop): Likewise.
5238 * tree-vect-patterns.c (vect_reassociating_reduction_p)
5239 (vect_recog_widen_op_pattern, vect_recog_mixed_size_cond_pattern)
5240 (vect_recog_bool_pattern, vect_recog_gather_scatter_pattern): Likewise.
5241 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
5242 (vect_slp_analyze_node_operations_1): Likewise.
5243 * tree-vect-stmts.c (vect_mark_relevant, process_use)
5244 (exist_non_indexing_operands_for_use_p, vect_init_vector_1)
5245 (vect_mark_stmts_to_be_vectorized, vect_get_vec_def_for_operand)
5246 (vect_finish_stmt_generation_1, get_group_load_store_type)
5247 (get_load_store_type, vect_build_gather_load_calls)
5248 (vectorizable_bswap, vectorizable_call, vectorizable_simd_clone_call)
5249 (vect_create_vectorized_demotion_stmts, vectorizable_conversion)
5250 (vectorizable_assignment, vectorizable_shift, vectorizable_operation)
5251 (vectorizable_store, vectorizable_load, vectorizable_condition)
5252 (vectorizable_comparison, vect_analyze_stmt, vect_transform_stmt)
5253 (supportable_widening_operation): Likewise.
5254 (vect_get_vector_types_for_stmt): Likewise.
5255 * tree-vectorizer.h (vect_dr_behavior): Likewise.
5257 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
5259 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence)
5260 (vect_slp_analyze_node_dependences, vect_analyze_data_ref_accesses)
5261 (vect_permute_store_chain, vect_permute_load_chain)
5262 (vect_shift_permute_load_chain, vect_transform_grouped_load): Avoid
5263 repeated stmt_vec_info lookups.
5264 * tree-vect-loop-manip.c (vect_can_advance_ivs_p): Likewise.
5265 (vect_update_ivs_after_vectorizer): Likewise.
5266 * tree-vect-loop.c (vect_is_simple_reduction): Likewise.
5267 (vect_create_epilog_for_reduction, vectorizable_reduction): Likewise.
5268 * tree-vect-patterns.c (adjust_bool_stmts): Likewise.
5269 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
5270 (vect_bb_slp_scalar_cost): Likewise.
5271 * tree-vect-stmts.c (get_group_alias_ptr_type): Likewise.
5273 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
5275 * tree-vect-data-refs.c (vect_check_gather_scatter): Pass the
5276 gcall rather than the generic gimple stmt to gimple_call_internal_fn.
5277 (vect_get_smallest_scalar_type, can_group_stmts_p): Use dyn_cast
5278 to get gassigns and gcalls, rather than operating on generc gimple
5280 * tree-vect-stmts.c (exist_non_indexing_operands_for_use_p)
5281 (vect_mark_stmts_to_be_vectorized, vectorizable_store)
5282 (vectorizable_load, vect_analyze_stmt): Likewise.
5283 * tree-vect-loop.c (vectorizable_reduction): Likewise gphi.
5285 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
5287 * tree-vectorizer.h (get_earlier_stmt, get_later_stmt): Take and
5288 return stmt_vec_infos rather than gimple stmts. Do not accept
5290 (vect_find_last_scalar_stmt_in_slp): Return a stmt_vec_info instead
5292 * tree-vect-slp.c (vect_find_last_scalar_stmt_in_slp): Likewise.
5293 Update use of get_later_stmt.
5294 (vect_get_constant_vectors): Update call accordingly.
5295 (vect_schedule_slp_instance): Likewise
5296 * tree-vect-data-refs.c (vect_slp_analyze_node_dependences): Likewise.
5297 (vect_slp_analyze_instance_dependence): Likewise.
5298 (vect_preserves_scalar_order_p): Update use of get_earlier_stmt.
5300 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
5302 * tree-vectorizer.h (stmt_info_for_cost::stmt): Replace with...
5303 (stmt_info_for_cost::stmt_info): ...this new field.
5304 (add_stmt_costs): Update accordingly.
5305 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost)
5306 (vect_get_known_peeling_cost): Likewise.
5307 (vect_estimate_min_profitable_iters): Likewise.
5308 * tree-vect-stmts.c (record_stmt_cost): Likewise.
5310 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
5312 * tree-vectorizer.h (_loop_vec_info::may_misalign_stmts): Change
5313 from an auto_vec<gimple *> to an auto_vec<stmt_vec_info>.
5314 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Update
5316 * tree-vect-loop-manip.c (vect_create_cond_for_align_checks): Likewise.
5318 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
5320 * tree-vectorizer.h (_stmt_vec_info::same_dr_stmt): Change from
5321 a gimple stmt to a stmt_vec_info.
5322 * tree-vect-stmts.c (vectorizable_load): Update accordingly.
5324 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
5326 * tree-vectorizer.h (vec_info::grouped_stores): Change from
5327 an auto_vec<gimple *> to an auto_vec<stmt_vec_info>.
5328 (_loop_vec_info::reduction_chains): Likewise.
5329 * tree-vect-loop.c (vect_fixup_scalar_cycles_with_patterns): Update
5331 * tree-vect-slp.c (vect_analyze_slp): Likewise.
5333 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
5335 * tree-vectorizer.h (_stmt_vec_info::first_element): Change from
5336 a gimple stmt to a stmt_vec_info.
5337 (_stmt_vec_info::next_element): Likewise.
5338 * tree-vect-data-refs.c (vect_update_misalignment_for_peel)
5339 (vect_slp_analyze_and_verify_node_alignment)
5340 (vect_analyze_group_access_1, vect_analyze_group_access)
5341 (vect_small_gap_p, vect_prune_runtime_alias_test_list)
5342 (vect_create_data_ref_ptr, vect_record_grouped_load_vectors)
5343 (vect_supportable_dr_alignment): Update accordingly.
5344 * tree-vect-loop.c (vect_fixup_reduc_chain): Likewise.
5345 (vect_fixup_scalar_cycles_with_patterns, vect_is_slp_reduction)
5346 (vect_is_simple_reduction, vectorizable_reduction): Likewise.
5347 * tree-vect-patterns.c (vect_reassociating_reduction_p): Likewise.
5348 * tree-vect-slp.c (vect_build_slp_tree_1)
5349 (vect_attempt_slp_rearrange_stmts, vect_supported_load_permutation_p)
5350 (vect_split_slp_store_group, vect_analyze_slp_instance)
5351 (vect_analyze_slp, vect_transform_slp_perm_load): Likewise.
5352 * tree-vect-stmts.c (vect_model_store_cost, vect_model_load_cost)
5353 (get_group_load_store_type, get_load_store_type)
5354 (get_group_alias_ptr_type, vectorizable_store, vectorizable_load)
5355 (vect_transform_stmt, vect_remove_stores): Likewise.
5357 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
5359 * tree-vectorizer.h (vect_dr_stmt): Return a stmt_vec_info rather
5361 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence)
5362 (vect_slp_analyze_data_ref_dependence, vect_record_base_alignments)
5363 (vect_calculate_target_alignmentm, vect_compute_data_ref_alignment)
5364 (vect_update_misalignment_for_peel, vect_verify_datarefs_alignment)
5365 (vector_alignment_reachable_p, vect_get_data_access_cost)
5366 (vect_get_peeling_costs_all_drs, vect_peeling_hash_get_lowest_cost)
5367 (vect_peeling_supportable, vect_enhance_data_refs_alignment)
5368 (vect_find_same_alignment_drs, vect_analyze_data_refs_alignment)
5369 (vect_analyze_group_access_1, vect_analyze_group_access)
5370 (vect_analyze_data_ref_access, vect_analyze_data_ref_accesses)
5371 (vect_vfa_access_size, vect_small_gap_p, vect_analyze_data_refs)
5372 (vect_supportable_dr_alignment): Remove vinfo_for_stmt from the
5373 result of vect_dr_stmt and use the stmt_vec_info instead of
5374 the associated gimple stmt.
5375 * tree-vect-loop-manip.c (get_misalign_in_elems): Likewise.
5376 (vect_gen_prolog_loop_niters): Likewise.
5377 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
5379 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
5381 * tree-vectorizer.h (_slp_tree::stmts): Change from a vec<gimple *>
5382 to a vec<stmt_vec_info>.
5383 * tree-vect-slp.c (vect_free_slp_tree): Update accordingly.
5384 (vect_create_new_slp_node): Take a vec<gimple *> instead of a
5386 (_slp_oprnd_info::def_stmts): Change from a vec<gimple *>
5387 to a vec<stmt_vec_info>.
5388 (bst_traits::value_type, bst_traits::value_type): Likewise.
5389 (bst_traits::hash): Update accordingly.
5390 (vect_get_and_check_slp_defs): Change the stmts parameter from
5391 a vec<gimple *> to a vec<stmt_vec_info>.
5392 (vect_two_operations_perm_ok_p, vect_build_slp_tree_1): Likewise.
5393 (vect_build_slp_tree): Likewise.
5394 (vect_build_slp_tree_2): Likewise. Update uses of
5395 SLP_TREE_SCALAR_STMTS.
5396 (vect_print_slp_tree): Update uses of SLP_TREE_SCALAR_STMTS.
5397 (vect_mark_slp_stmts, vect_mark_slp_stmts_relevant)
5398 (vect_slp_rearrange_stmts, vect_attempt_slp_rearrange_stmts)
5399 (vect_supported_load_permutation_p, vect_find_last_scalar_stmt_in_slp)
5400 (vect_detect_hybrid_slp_stmts, vect_slp_analyze_node_operations_1)
5401 (vect_slp_analyze_node_operations, vect_slp_analyze_operations)
5402 (vect_bb_slp_scalar_cost, vect_slp_analyze_bb_1)
5403 (vect_get_constant_vectors, vect_get_slp_defs)
5404 (vect_transform_slp_perm_load, vect_schedule_slp_instance)
5405 (vect_remove_slp_scalar_calls, vect_schedule_slp): Likewise.
5406 (vect_analyze_slp_instance): Build up a vec of stmt_vec_infos
5407 instead of gimple stmts.
5408 * tree-vect-data-refs.c (vect_slp_analyze_node_dependences): Change
5409 the stores parameter for a vec<gimple *> to a vec<stmt_vec_info>.
5410 (vect_slp_analyze_instance_dependence): Update uses of
5411 SLP_TREE_SCALAR_STMTS.
5412 (vect_slp_analyze_and_verify_node_alignment): Likewise.
5413 (vect_slp_analyze_and_verify_instance_alignment): Likewise.
5414 * tree-vect-loop.c (neutral_op_for_slp_reduction): Likewise.
5415 (get_initial_defs_for_reduction): Likewise.
5416 (vect_create_epilog_for_reduction): Likewise.
5417 (vectorize_fold_left_reduction): Likewise.
5418 * tree-vect-stmts.c (vect_prologue_cost_for_slp_op): Likewise.
5419 (vect_model_simple_cost, vectorizable_shift, vectorizable_load)
5420 (can_vectorize_live_stmts): Likewise.
5422 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
5424 * tree-vectorizer.h (_loop_vec_info::reductions): Change from an
5425 auto_vec<gimple *> to an auto_vec<stmt_vec_info>.
5426 (vect_force_simple_reduction): Take and return stmt_vec_infos rather
5428 * tree-parloops.c (valid_reduction_p): Take a stmt_vec_info instead
5430 (gather_scalar_reductions): Update after above interface changes.
5431 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Likewise.
5432 (vect_is_simple_reduction): Take and return stmt_vec_infos rather
5434 (vect_force_simple_reduction): Likewise.
5435 * tree-vect-patterns.c (vect_pattern_recog_1): Update use of
5436 LOOP_VINFO_REDUCTIONS.
5437 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
5439 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
5441 * tree-vectorizer.h (_stmt_vec_info::reduc_def): Change from
5442 a gimple stmt to a stmt_vec_info.
5443 * tree-vect-loop.c (vect_active_double_reduction_p)
5444 (vect_force_simple_reduction, vectorizable_reduction): Update
5447 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
5449 * tree-vectorizer.h (_slp_tree::vec_stmts): Change from a
5450 vec<gimple *> to a vec<stmt_vec_info>.
5451 * tree-vect-loop.c (vect_create_epilog_for_reduction): Change
5452 the reduction_phis argument from a vec<gimple *> to a
5454 (vectorizable_reduction): Likewise the phis local variable that
5455 is passed to vect_create_epilog_for_reduction. Update for new type
5456 of SLP_TREE_VEC_STMTS.
5457 (vectorizable_induction): Update for new type of SLP_TREE_VEC_STMTS.
5458 (vectorizable_live_operation): Likewise.
5459 * tree-vect-slp.c (vect_get_slp_vect_defs): Likewise.
5460 (vect_transform_slp_perm_load, vect_schedule_slp_instance): Likewise.
5462 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
5464 * tree-vectorizer.h (_stmt_vec_info::vectorized_stmt): Change from
5465 a gimple stmt to a stmt_vec_info.
5466 (vectorizable_condition, vectorizable_live_operation)
5467 (vectorizable_reduction, vectorizable_induction): Pass back the
5468 vectorized statement as a stmt_vec_info.
5469 * tree-vect-data-refs.c (vect_record_grouped_load_vectors): Update
5470 use of STMT_VINFO_VEC_STMT.
5471 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise,
5472 accumulating the inner phis that feed the STMT_VINFO_VEC_STMT
5473 as stmt_vec_infos rather than gimple stmts.
5474 (vectorize_fold_left_reduction): Change vec_stmt from a gimple stmt
5476 (vectorizable_live_operation): Likewise.
5477 (vectorizable_reduction, vectorizable_induction): Likewise,
5478 updating use of STMT_VINFO_VEC_STMT.
5479 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Update use
5480 of STMT_VINFO_VEC_STMT.
5481 (vect_build_gather_load_calls, vectorizable_bswap, vectorizable_call)
5482 (vectorizable_simd_clone_call, vectorizable_conversion)
5483 (vectorizable_assignment, vectorizable_shift, vectorizable_operation)
5484 (vectorizable_store, vectorizable_load, vectorizable_condition)
5485 (vectorizable_comparison, can_vectorize_live_stmts): Change vec_stmt
5486 from a gimple stmt to a stmt_vec_info.
5487 (vect_transform_stmt): Update use of STMT_VINFO_VEC_STMT. Pass a
5488 pointer to a stmt_vec_info to the vectorizable_* routines.
5490 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
5492 * tree-vectorizer.h (_stmt_vec_info::related_stmt): Change from
5493 a gimple stmt to a stmt_vec_info.
5494 (is_pattern_stmt_p): Update accordingly.
5495 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Likewise.
5496 (vect_record_grouped_load_vectors): Likewise.
5497 * tree-vect-loop.c (vect_determine_vf_for_stmt): Likewise.
5498 (vect_fixup_reduc_chain, vect_update_vf_for_slp): Likewise.
5499 (vect_model_reduction_cost): Likewise.
5500 (vect_create_epilog_for_reduction): Likewise.
5501 (vectorizable_reduction, vectorizable_induction): Likewise.
5502 * tree-vect-patterns.c (vect_init_pattern_stmt): Likewise.
5503 Return the stmt_vec_info for the pattern statement.
5504 (vect_set_pattern_stmt): Update use of STMT_VINFO_RELATED_STMT.
5505 (vect_split_statement, vect_mark_pattern_stmts): Likewise.
5506 * tree-vect-slp.c (vect_detect_hybrid_slp_stmts): Likewise.
5507 (vect_detect_hybrid_slp, vect_get_slp_defs): Likewise.
5508 * tree-vect-stmts.c (vect_mark_relevant): Likewise.
5509 (vect_get_vec_def_for_operand_1, vectorizable_call): Likewise.
5510 (vectorizable_simd_clone_call, vect_analyze_stmt, new_stmt_vec_info)
5511 (free_stmt_vec_info, vect_is_simple_use): Likewise.
5513 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
5515 * tree-vectorizer.h (vect_finish_replace_stmt): Return a stmt_vec_info
5516 (vect_finish_stmt_generation): Likewise.
5517 * tree-vect-stmts.c (vect_finish_stmt_generation_1): Likewise.
5518 (vect_finish_replace_stmt, vect_finish_stmt_generation): Likewise.
5519 (vect_build_gather_load_calls): Use the return value of the above
5520 functions instead of a separate call to vinfo_for_stmt. Use narrow
5521 scopes for the input gimple stmt and wider scopes for the associated
5522 stmt_vec_info. Use vec_info::lookup_def when setting these
5523 stmt_vec_infos from an SSA_NAME definition.
5524 (vectorizable_bswap, vectorizable_call, vectorizable_simd_clone_call)
5525 (vect_create_vectorized_demotion_stmts, vectorizable_conversion)
5526 (vectorizable_assignment, vectorizable_shift, vectorizable_operation)
5527 (vectorizable_store, vectorizable_load, vectorizable_condition)
5528 (vectorizable_comparison): Likewise.
5529 * tree-vect-loop.c (vectorize_fold_left_reduction): Likewise.
5530 (vectorizable_reduction): Likewise.
5532 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
5534 * tree-vectorizer.h (vect_is_simple_use): Add an optional
5535 stmt_vec_info * parameter before the optional gimple **.
5536 * tree-vect-stmts.c (vect_is_simple_use): Likewise.
5537 (process_use, vect_get_vec_def_for_operand_1): Update callers.
5538 (vect_get_vec_def_for_operand, vectorizable_shift): Likewise.
5539 * tree-vect-loop.c (vectorizable_reduction): Likewise.
5540 (vectorizable_live_operation): Likewise.
5541 * tree-vect-patterns.c (type_conversion_p): Likewise.
5542 (vect_look_through_possible_promotion): Likewise.
5543 (vect_recog_rotate_pattern): Likewise.
5544 * tree-vect-slp.c (vect_get_and_check_slp_defs): Likewise.
5546 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
5548 * tree-vectorizer.h (stmt_vec_info): Temporarily change from
5549 a typedef to a wrapper class.
5550 (NULL_STMT_VEC_INFO): New macro.
5551 (vec_info::stmt_infos): Change to vec<stmt_vec_info>.
5552 (stmt_vec_info::operator*): New function.
5553 (stmt_vec_info::operator gimple *): Likewise.
5554 (set_vinfo_for_stmt): Use NULL_STMT_VEC_INFO.
5555 (add_stmt_costs): Likewise.
5556 * tree-vect-loop-manip.c (iv_phi_p): Likewise.
5557 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost)
5558 (vect_get_known_peeling_cost): Likewise.
5559 (vect_estimate_min_profitable_iters): Likewise.
5560 * tree-vect-patterns.c (vect_init_pattern_stmt): Likewise.
5561 * tree-vect-slp.c (vect_remove_slp_scalar_calls): Likewise.
5562 * tree-vect-stmts.c (vect_build_gather_load_calls): Likewise.
5563 (vectorizable_store, free_stmt_vec_infos): Likewise.
5564 (new_stmt_vec_info): Change return type of xcalloc to
5567 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
5569 * tree-vectorizer.h (vec_info::lookup_single_use): Declare.
5570 * tree-vectorizer.c (vec_info::lookup_single_use): New function.
5571 * tree-vect-loop.c (vectorizable_reduction): Use it instead of
5572 a single_imm_use-based sequence.
5573 * tree-vect-stmts.c (supportable_widening_operation): Likewise.
5575 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
5577 * tree-vectorizer.h (vec_info::lookup_def): Declare.
5578 * tree-vectorizer.c (vec_info::lookup_def): New function.
5579 * tree-vect-patterns.c (vect_get_internal_def): Use it.
5580 (vect_widened_op_tree): Likewise.
5581 * tree-vect-stmts.c (vect_is_simple_use): Likewise.
5582 * tree-vect-loop.c (vect_analyze_loop_operations): Likewise.
5583 (vectorizable_reduction): Likewise.
5584 (vect_valid_reduction_input_p): Take a stmt_vec_info instead
5586 (vect_is_slp_reduction): Update calls accordingly. Use
5587 vec_info::lookup_def.
5588 (vect_is_simple_reduction): Likewise
5589 * tree-vect-slp.c (vect_detect_hybrid_slp_1): Use vec_info::lookup_def.
5591 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
5593 * tree-vectorizer.h (vec_info::lookup_stmt): Declare.
5594 * tree-vectorizer.c (vec_info::lookup_stmt): New function.
5595 * tree-vect-loop.c (vect_determine_vf_for_stmt): Use it instead
5597 (vect_determine_vectorization_factor, vect_analyze_scalar_cycles_1)
5598 (vect_compute_single_scalar_iteration_cost, vect_analyze_loop_form)
5599 (vect_update_vf_for_slp, vect_analyze_loop_operations)
5600 (vect_is_slp_reduction, vectorizable_induction)
5601 (vect_transform_loop_stmt, vect_transform_loop): Likewise.
5602 * tree-vect-patterns.c (vect_init_pattern_stmt):
5603 (vect_determine_min_output_precision_1, vect_determine_precisions)
5604 (vect_pattern_recog): Likewise.
5605 * tree-vect-stmts.c (vect_analyze_stmt, vect_transform_stmt): Likewise.
5606 * config/powerpcspe/powerpcspe.c (rs6000_density_test): Likewise.
5607 * config/rs6000/rs6000.c (rs6000_density_test): Likewise.
5608 * tree-vect-slp.c (vect_detect_hybrid_slp_stmts): Likewise.
5609 (vect_detect_hybrid_slp_1, vect_detect_hybrid_slp_2)
5610 (vect_detect_hybrid_slp): Likewise. Change the walk_stmt_info
5611 info field from a loop to a loop_vec_info.
5613 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
5615 * tree-vectorizer.h (stmt_vec_info): Move typedef earlier in file.
5616 (vec_info::add_stmt): Declare.
5617 * tree-vectorizer.c (vec_info::add_stmt): New function.
5618 * tree-vect-data-refs.c (vect_create_data_ref_ptr): Use it.
5619 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Likewise.
5620 (vect_create_epilog_for_reduction, vectorizable_reduction): Likewise.
5621 (vectorizable_induction): Likewise.
5622 * tree-vect-slp.c (_bb_vec_info::_bb_vec_info): Likewise.
5623 * tree-vect-stmts.c (vect_finish_stmt_generation_1): Likewise.
5624 (vectorizable_simd_clone_call, vectorizable_store): Likewise.
5625 (vectorizable_load): Likewise.
5626 * tree-vect-patterns.c (vect_init_pattern_stmt): Likewise.
5627 (vect_recog_bool_pattern, vect_recog_mask_conversion_pattern)
5628 (vect_recog_gather_scatter_pattern): Likewise.
5629 (append_pattern_def_seq): Likewise. Remove a check that is
5630 performed by add_stmt itself.
5632 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
5634 * tree-vect-loop.c (vectorizable_reduction): Fix an instance in
5635 which make_ssa_name was called with new_stmt before new_stmt
5638 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
5640 * tree-vect-loop.c (vect_valid_reduction_input_p): New function,
5642 (vect_is_slp_reduction): ...here...
5643 (vect_is_simple_reduction): ...and here. Remove repetition of tests
5644 that are already known to be false.
5646 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
5648 * tree-vectorizer.h (vect_free_slp_instance): Add a final_p parameter.
5649 * tree-vect-slp.c (vect_free_slp_tree): Likewise. Don't update
5650 STMT_VINFO_NUM_SLP_USES when it's true.
5651 (vect_free_slp_instance): Add a final_p parameter and pass it to
5653 (vect_build_slp_tree_2): Update call to vect_free_slp_instance.
5654 (vect_analyze_slp_instance): Likewise.
5655 (vect_slp_analyze_operations): Likewise.
5656 (vect_slp_analyze_bb_1): Likewise.
5657 * tree-vectorizer.c (vec_info): Likewise.
5658 * tree-vect-loop.c (vect_transform_loop): Likewise.
5660 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
5662 * tree-vect-loop.c (vectorizable_reduction): Assert that the
5663 function is not called for second and subsequent members of
5666 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
5668 * tree-vect-loop.c (get_initial_def_for_reduction): Move special
5669 cases for nested loops from here to ...
5670 (vect_create_epilog_for_reduction): ...here. Only call
5671 vect_is_simple_use for inner-loop reductions.
5673 2018-07-31 Martin Liska <mliska@suse.cz>
5675 PR gcov-profile/85338
5676 PR gcov-profile/85350
5677 PR gcov-profile/85372
5678 * profile.c (struct location_triplet): New.
5679 (struct location_triplet_hash): Likewise.
5680 (output_location): Do not output a BB that
5681 is already recorded for a line.
5682 (branch_prob): Use streamed_locations.
5684 2018-07-31 Martin Liska <mliska@suse.cz>
5686 PR gcov-profile/85370
5687 * coverage.c (coverage_begin_function): Do not mark target
5688 clones as artificial functions.
5690 2018-07-31 Martin Liska <mliska@suse.cz>
5692 PR gcov-profile/83813
5693 PR gcov-profile/84758
5694 PR gcov-profile/85217
5695 PR gcov-profile/85332
5696 * profile.c (branch_prob): Do not record GOTO expressions
5697 for GIMPLE statements which locations are already streamed.
5699 2018-07-31 Olivier Hainque <hainque@adacore.com>
5701 * gcc.c (handle_spec_function): Accept a soft_matched_part
5702 argument, as do_spec_1. Pass it down to ...
5703 (eval_spec_function): Accept a soft_matched_part argument,
5704 and pass it down to ...
5705 (do_spec_2): Accept a soft_matched_part argument, and pass
5706 it down to do_spec_1.
5707 (do_spec_1): Pass soft_matched_part to handle_spec_function.
5708 (handle_braces): Update call to handle_spec_function.
5709 (driver::set_up_specs): Update calls to do_spec_2.
5710 (compare_debug_dump_opt_spec_function): Likewise.
5711 (compare_debug_self_opt_spec_function): Likewise.
5713 2018-07-31 Olivier Hainque <hainque@adacore.com>
5715 * common.opt (nolibc): New option.
5716 * doc/invoke.texi (Link Options): Document it.
5717 * gcc.c (LINK_GCC_C_SEQUENCE_SPEC): Honor nolibc.
5718 * config/alpha/linux.h: Likewise.
5719 * config/arc/elf.h: Likewise.
5720 * config/arm/uclinux-elf.h: Likewise.
5721 * config/arm/unknown-elf.h: Likewise.
5722 * config/avr/avrlibc.h: Likewise.
5723 * config/bfin/bfin.h: Likewise.
5724 * config/bfin/linux.h: Likewise.
5725 * config/bfin/uclinux.h: Likewise.
5726 * config/darwin.h: Likewise.
5727 * config/darwin10.h: Likewise.
5728 * config/darwin12.h: Likewise.
5729 * config/gnu-user.h: Likewise.
5730 * config/lm32/uclinux-elf.h: Likewise.
5731 * config/pa/pa-hpux11.h: Likewise.
5732 * config/pa/pa64-hpux.h: Likewise.
5733 * config/sparc/sparc.h: Likewise.
5735 2018-07-31 Olivier Hainque <hainque@adacore.com>
5737 * gcc.c (getenv_spec_function): Prepend '/' to value for allowed
5738 undefined variables.
5740 2018-07-30 Segher Boessenkool <segher@kernel.crashing.org>
5743 * config/arm/arm.c (arm_block_set_aligned_vect): Use gen_int_mode
5746 2018-07-30 Bernd Edlinger <bernd.edlinger@hotmail.de>
5748 * tree-ssa-forwprop.c (simplify_builtin_call): Don't create a not NUL
5749 terminated string literal.
5751 2018-07-30 Segher Boessenkool <segher@kernel.crashing.org>
5753 PR rtl-optimization/85160
5754 * combine.c (is_just_move): New function.
5755 (try_combine): Allow combining two instructions into two if neither of
5756 the original instructions was a move.
5758 2018-07-30 Alexander Monakov <amonakov@ispras.ru>
5761 * doc/extend.texi (Global Register Variables): Discourage use of type
5763 (Local Register Variables): Likewise.
5765 2018-07-30 Richard Sandiford <richard.sandiford@arm.com>
5767 PR tree-optimization/86506
5768 * hwint.h (ceil_log2): Resync with hwint.c implementation.
5770 2018-07-30 Ilya Leoshkevich <iii@linux.ibm.com>
5773 * lra-constraints.c (spill_hard_reg_in_range): When selecting the
5774 hard_regno, make sure no insn between `from` and `to` clobbers it.
5776 2018-07-30 Cesar Philippidis <cesar@codesourcery.com>
5777 Tom de Vries <tdevries@suse.de>
5779 * config/nvptx/nvptx.c (PTX_GANG_DEFAULT): Rename to ...
5780 (PTX_DEFAULT_RUNTIME_DIM): ... this.
5781 (nvptx_goacc_validate_dims): Set default worker and gang dims to
5782 PTX_DEFAULT_RUNTIME_DIM.
5783 (nvptx_dim_limit): Ignore GOMP_DIM_WORKER.
5785 2018-07-29 John David Anglin <danglin@gcc.gnu.org>
5787 * config/pa/pa.c (pa_output_addr_vec): Align address table.
5788 * config/pa/pa.h (JUMP_TABLES_IN_TEXT_SECTION): Revise comment.
5789 * config/pa/pa32-linux.h (JUMP_TABLES_IN_TEXT_SECTION): Define.
5791 2018-07-27 Michael Meissner <meissner@linux.ibm.com>
5793 * config/rs6000/constraints.md (wG constraint): Delete, no longer
5795 * config/rs6000/predicates.md (p9_fusion_reg_operand): Rename
5796 predicate to reflect toc fusion has been deleted.
5797 (toc_fusion_mem_raw): Delete, no longer used.
5798 (toc_fusion_mem_wrapped): Likewise.
5799 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete toc
5801 * config/rs6000/rs6000-protos.h (fusion_wrap_memory_address):
5802 Delete, no longer used.
5803 * config/rs6000/rs6000.c (struct rs6000_reg_addr): Delete fields
5804 meant to be used for toc fusion.
5805 (rs6000_debug_print_mode): Delete toc fusion debugging.
5806 (rs6000_debug_reg_global): Likewise.
5807 (rs6000_init_hard_regno_mode_ok): Delete setting up fields for toc
5808 fusion and secondary reload support that were never used.
5809 (rs6000_option_override_internal): Delete TOC fusion, that was only
5810 partially defined, and it did not work unless you also used the
5812 (rs6000_legitimate_address_p): Delete TOC fusion support.
5813 (rs6000_opt_masks): Likewise.
5814 (fusion_wrap_memory_address): Delete function, no longer used.
5815 (fusion_split_address); Delete TOC fusion support.
5816 * config/rs6000/rs6000.h (TARGET_TOC_FUSION_INT): Delete, no
5817 longer used with toc fusion being deleted.
5818 (TARGET_TOC_FUSION_FP): Likewise.
5819 * config/rs6000/rs6000.md (UNSPEC_FUSION_ADDIS): Delete TOC fusion
5821 (toc fusion spliter): Delete TOC fusion support.
5822 (toc_fusionload_<mode>): Likewise.
5823 (toc_fusionload_di): Likewise.
5824 (fusion_gpr_load_<mode>): Delete generator function, this insn no
5825 longer needs to be named. Rename predicate to delete TOC fusion.
5826 (fusion_gpr_<P:mode>_<GPR_FUSION:mode>_load): Likewise.
5827 (fusion_gpr_<P:mode>_<GPR_FUSION:mode>_store): Likewise.
5828 (fusion_vsx_<P:mode>_<GPR_FUSION:mode>_load): Likewise.
5829 (fusion_vsx_<P:mode>_<GPR_FUSION:mode>_store): Likewise.
5830 (p9 fusion peephole2s): Rename predicate to delete TOC fusion.
5832 2018-07-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
5834 * doc/extend.texi (Basic PowerPC Built-in Functions Available on
5835 ISA 2.05): Replace __uint128_t with __uint128 and __int128_t with
5836 __int128 in built-in function prototypes.
5837 (PowerPC AltiVec Built-in Functions on ISA 2.07): Likewise.
5838 (PowerPC AltiVec Built-in Functions on ISA 3.0): Likewise.
5840 2018-07-27 Martin Sebor <msebor@redhat.com>
5842 PR tree-optimization/86696
5843 * tree-ssa-strlen.c (get_min_string_length): Handle all integer
5844 types, including enums.
5845 (handle_char_store): Be prepared for the above function to fail.
5847 2018-07-26 Qing Zhao <qing.zhao@oracle.com>
5849 * builtins.c (inline_expand_builtin_string_cmp): Disable inlining
5850 when optimization level is lower than 2 or optimize for size.
5852 2018-07-26 Martin Sebor <msebor@redhat.com>
5854 PR tree-optimization/86043
5855 PR tree-optimization/86042
5856 * tree-ssa-strlen.c (handle_builtin_memcpy): Handle strict overlaps.
5857 (get_string_cst_length): Rename...
5858 (get_min_string_length): ...to this. Add argument.
5859 (handle_char_store): Extend to handle multi-character stores by
5861 * tree.c (initializer_zerop): Use new argument. Handle MEM_REF.
5862 * tree.h (initializer_zerop): Add argument.
5864 2018-07-26 Jakub Jelinek <jakub@redhat.com>
5867 * omp-low.c (scan_sharing_clauses): Don't ignore map clauses for
5868 declare target to variables if they have always,{to,from,tofrom} map
5871 2018-07-26 Martin Liska <mliska@suse.cz>
5874 * lto-wrapper.c: Add linker_output as prefix
5875 for ltrans_output_file.
5877 2018-07-26 Segher Boessenkool <segher@kernel.crashing.org>
5879 PR rtl-optimization/85805
5880 * combine.c (reg_nonzero_bits_for_combine): Only use the last set
5881 value for hard registers if that was written in the same mode.
5883 2018-07-26 Martin Liska <mliska@suse.cz>
5885 PR gcov-profile/86536
5886 * gcov.c (format_gcov): Use printf format %.*f directly
5887 and do not handle special values.
5889 2018-07-25 Claudiu Zissulescu <claziss@synopsys.com>
5891 * common/config/arc/arc-common.c (arc_option_optimization_table):
5892 Update default optimizations for size.
5894 2018-07-25 Claudiu Zissulescu <claziss@synopsys.com>
5896 * config/arc/arc.md (movsf_insn): Add short instruction selection.
5897 * config/arc/constraints.md (CfZ): New constraint.
5898 * config/arc/fpu.md (addssf3_fpu): Use CfZ constraint.
5899 (subsf3_fpu): Likewise.
5900 (cmpsf_fpu): Likewise.
5901 (cmpsf_fpu_uneq): Likewise.
5903 2018-07-25 Claudiu Zissulescu <claziss@synopsys.com>
5905 * config/arc/arc.c (compact_memory_operand_p): Check for uncached
5907 (arc_is_uncached_mem_p): uncached applies to both the variable and
5910 2018-07-25 Claudiu Zissulescu <claziss@synopsys.com>
5912 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Add additional
5915 2018-07-25 David Malcolm <dmalcolm@redhat.com>
5917 * optinfo-emit-json.cc (class optrecord_json_writer): Convert
5918 field "m_scopes" from vec to auto_vec.
5920 2018-07-25 Martin Liska <mliska@suse.cz>
5922 * config/powerpcspe/powerpcspe-protos.h (rs6000_loop_align): Fix
5925 2018-07-25 Richard Biener <rguenther@suse.de>
5928 * dwarf2out.c (dwarf2out_decl): Do not handle nested functions
5929 special wrt context_die late.
5930 (gen_subprogram_die): Re-use DIEs in local scope.
5932 2018-07-25 Richard Sandiford <richard.sandiford@arm.com>
5934 PR tree-optimization/86644
5935 * hwint.c (ceil_log2): Fix comment. Return 0 for 0.
5937 2018-07-25 Martin Liska <mliska@suse.cz>
5940 * dumpfile.c: And excluded values with TDF_ALL_VALUES.
5941 * dumpfile.h (enum dump_flag): Defince TDF_ALL_VALUES.
5943 2018-07-25 Martin Liska <mliska@suse.cz>
5946 * params.def: Explain ASan abbreviation and provide
5947 a documentation link.
5949 2018-07-24 Martin Sebor <msebor@redhat.com>
5951 PR tree-optimization/86622
5952 PR tree-optimization/86532
5953 * builtins.h (string_length): Declare.
5954 * builtins.c (c_strlen): Correct handling of non-constant offsets.
5955 (check_access): Be prepared for non-constant length ranges.
5956 (string_length): Make extern.
5957 * expr.c (string_constant): Only handle the minor non-constant
5958 array index. Use string_constant to compute the length of
5959 a generic string constant.
5961 2018-07-24 Richard Sandiford <richard.sandiford@arm.com>
5963 PR tree-optimization/86618
5964 * tree-vect-stmts.c (vectorizable_call): Don't take the address
5965 of LOOP_VINFO_MASKS (loop_vinfo) when loop_vinfo is null.
5967 2018-07-24 David Malcolm <dmalcolm@redhat.com>
5969 PR tree-optimization/86636
5970 * json.cc (json::object::set): Fix comment. Add assertions.
5971 (json::array::append): Move here from json.h. Add comment and an
5973 (json::string::string): Likewise.
5974 * json.h (json::array::append): Move to json.cc.
5975 (json::string::string): Likewise.
5976 * optinfo-emit-json.cc
5977 (optrecord_json_writer::impl_location_to_json): Assert that we
5978 aren't attempting to write out UNKNOWN_LOCATION, or an ad-hoc
5979 wrapper around it. Expand the location once, rather than three
5981 (optrecord_json_writer::inlining_chain_to_json): Fix the check for
5982 UNKNOWN_LOCATION, to use LOCATION_LOCUS to look through ad-hoc
5984 (optrecord_json_writer::optinfo_to_json): Likewise, in four
5985 places. Fix some overlong lines.
5987 2018-07-24 Matthew Malcomson <matthew.malcomson@arm.com>
5989 * config/aarch64/aarch64-simd.md
5990 (aarch64_<ANY_EXTEND:su><ADDSUB:optab>w<mode>): Split into...
5991 (aarch64_<ANY_EXTEND:su>subw<mode>): ... This...
5992 (aarch64_<ANY_EXTEND:su>addw<mode>): ... And this.
5993 (aarch64_<ANY_EXTEND:su><ADDSUB:optab>w<mode>_internal): Split into...
5994 (aarch64_<ANY_EXTEND:su>subw<mode>_internal): ... This...
5995 (aarch64_<ANY_EXTEND:su>addw<mode>_internal): ... And this.
5996 (aarch64_<ANY_EXTEND:su><ADDSUB:optab>w2<mode>_internal): Split into...
5997 (aarch64_<ANY_EXTEND:su>subw2<mode>_internal): ... This...
5998 (aarch64_<ANY_EXTEND:su>addw2<mode>_internal): ... And this.
6000 2018-07-24 Jakub Jelinek <jakub@redhat.com>
6003 * expmed.c (expand_divmod): Punt if d == HOST_WIDE_INT_MIN
6004 and size > HOST_BITS_PER_WIDE_INT. For size > HOST_BITS_PER_WIDE_INT
6005 and abs_d == d, do the power of two handling if profitable.
6007 2018-07-24 Richard Biener <rguenther@suse.de>
6009 * match.pd: Add BIT_FIELD_REF canonicalizations.
6011 2018-07-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
6014 * genmatch.c (dt_operand::gen_match_op): Avoid folding volatile values.
6016 2018-07-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
6018 * gimple-fold.c (gimple_fold_builtin_printf): Don't create a not NUL
6019 terminated STRING_CST object.
6021 2018-07-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
6023 hsa-dump.c (dump_hsa_symbol): Avoid out of scope access to buf.
6025 2018-07-23 Segher Boessenkool <segher@kernel.crashing.org>
6027 * config/rs6000/rs6000-p8swap.c (rtx_is_swappable_p): Adjust.
6028 * config/rs6000/rs6000-protos.h (rs6000_split_v4si_init): Delete.
6029 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Always force
6030 the elements into a register.
6031 (rs6000_split_v4si_init_di_reg): Delete.
6032 (rs6000_split_v4si_init): Delete.
6033 * config/rs6000/vsx.md (unspec): Delete UNSPEC_VSX_VEC_INIT.
6034 (vsx_init_v4si): Rewrite as a define_expand.
6036 2018-07-23 Segher Boessenkool <segher@kernel.crashing.org>
6038 * config/rs6000/rs6000.md (splitters for rldimi and rlwimi with the
6039 zero_extend argument from memory): New.
6041 2018-07-22 Martin Sebor <msebor@redhat.com>
6044 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Avoid
6045 diagnosing calls with unknown arguments unless -Walloca-larger-than
6046 is restricted to less than PTRDIFF_MAX bytes.
6048 2018-07-22 Gerald Pfeifer <gerald@pfeifer.com>
6050 * doc/gcov.texi (Invoking Gcov): Editorial changes.
6052 2018-07-20 David Malcolm <dmalcolm@redhat.com>
6054 * pretty-print.c (text_info::set_location): Remove redundant
6055 "line_table" parameter from call to rich_location::set_range.
6057 2018-07-20 Martin Sebor <msebor@redhat.com>
6060 * builtins.c (expand_builtin_alloca): Adjust.
6061 * calls.c (alloc_max_size): Simplify.
6062 * cgraphunit.c (cgraph_node::expand): Adjust.
6063 * common.opt (larger_than_size, warn_frame_larger_than): Remove
6065 (frame_larger_than_size): Same.
6066 (-Wframe-larger-than, -Wlarger-than, -Wstack-usage): Change options
6067 to take a HOST_WIDE_INT argument and accept a byte-size suffix.
6069 * doc/invoke.texi (GCC Command Options): Document option arguments.
6070 Explain byte-size arguments and suffixes.
6071 (-Wvla-larger-than, -Wno-alloc-size-larger-than): Update.
6072 (-Wno-alloca-larger-than, -Wno-vla-larger-than): Same.
6073 (-Wframe-larger-than, -Wlarger-than, -Wstack-usage): Same.
6074 * doc/options.texi (UInteger): Expand.
6075 (Host_Wide_Int, ByteSize): Document new properties.
6076 * final.c (final_start_function_1): Include sizes in an error message.
6077 * function.c (frame_offset_overflow): Same.
6078 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Adjust.
6079 (alloca_call_type_by_arg): Change function argument to HOST_WIDE_INT.
6080 Diagnose unbounded alloca calls only for limits of less than
6082 (alloca_call_type): Adjust. Diagnose possibly out-of-bounds alloca
6083 calls and VLA size only for limits of less than PTRDIFF_MAX. Same
6085 (pass_walloca::execute): Adjust. Diagnose alloca calls in loops
6086 only for limits of less than PTRDIFF_MAX.
6087 * langhooks-def.h (lhd_handle_option): Change function argument
6089 * langhooks.c (lhd_handle_option): Same.
6090 * langhooks.h (handle_option): Same.
6091 * opt-functions.awk (switch_bit_fields): Handle Host_Wide_Int and
6093 (var_type, var_type_struct): Same.
6094 (var_set): Handle ByteSize flag.
6095 * optc-gen.awk: Add comments to output to ease debugging. Make
6096 use of HOST_WIDE_INT where appropriate.
6097 * opts-gen-save.awk: Use %lx to format unsigned long.
6098 * opth-gen.awk: Change function argument to HOST_WIDE_INT.
6099 * opts-common.c (integral_argument): Return HOST_WIDE_INT and add
6100 arguments. Parse bytes-size suffixes.
6101 (enum_arg_to_value): Change function argument to HOST_WIDE_INT.
6102 (enum_value_to_arg): Same.
6103 (decode_cmdline_option): Handle cl_host_wide_int. Adjust.
6104 (handle_option): Adjust.
6105 (generate_option): Change function argument to HOST_WIDE_INT.
6106 (cmdline_handle_error): Adjust.
6107 (read_cmdline_option): Change function argument to HOST_WIDE_INT.
6108 (set_option): Change function argument to HOST_WIDE_INT.
6109 (option_enabled): Handle cl_host_wide_int.
6110 (get_option_state): Handle CLVC_SIZE.
6111 (control_warning_option): Same.
6112 * opts.c (common_handle_option): Change function argument to
6113 HOST_WIDE_INT. Remove handling of OPT_Walloca_larger_than_ and
6114 OPT_Wvla_larger_than_.
6115 * opts.h (enum cl_var_type): Add an enumerator.
6116 * stor-layout.c (layout_decl): Print a more meaningful warning.
6117 * toplev.c (output_stack_usage): Adjust.
6119 2018-07-20 Qing Zhao <qing.zhao@oracle.com>
6121 * builtins.c (expand_builtin_memcmp): Delete the last parameter for
6122 call to inline_expand_builtin_string_cmp.
6123 (expand_builtin_strcmp): Likewise.
6124 (expand_builtin_strncmp): Likewise.
6125 (inline_string_cmp): Delete the last parameter, change char_type_node
6126 to unsigned_char_type_node for strcmp/strncmp, add conversions to the
6128 (inline_expand_builtin_string_cmp): Delete the last parameter, give up
6129 the inlining expansion on target where the type of the call has same or
6130 narrower precision than unsigned char.
6132 2018-07-20 David Malcolm <dmalcolm@redhat.com>
6134 * Makefile.in (OBJS): Add json.o and optinfo-emit-json.o.
6135 (CFLAGS-optinfo-emit-json.o): Define TARGET_NAME.
6136 * common.opt (fsave-optimization-record): New option.
6137 * coretypes.h (struct kv_pair): Move here from dumpfile.c.
6138 * doc/invoke.texi (-fsave-optimization-record): New option.
6139 * dumpfile.c: Include "optinfo-emit-json.h".
6140 (struct kv_pair): Move to coretypes.h.
6141 (optgroup_options): Make non-static.
6142 (dump_context::end_scope): Call
6143 optimization_records_maybe_pop_dump_scope.
6144 * dumpfile.h (optgroup_options): New decl.
6145 * json.cc: New file.
6147 * optinfo-emit-json.cc: New file.
6148 * optinfo-emit-json.h: New file.
6149 * optinfo.cc: Include "optinfo-emit-json.h".
6150 (optinfo::emit): Call optimization_records_maybe_record_optinfo.
6151 (optinfo_enabled_p): Check optimization_records_enabled_p.
6152 (optinfo_wants_inlining_info_p): Likewise.
6153 * optinfo.h: Update comment.
6154 * profile-count.c (profile_quality_as_string): New function.
6155 * profile-count.h (profile_quality_as_string): New decl.
6156 (profile_count::quality): New accessor.
6157 * selftest-run-tests.c (selftest::run_tests): Call json_cc_tests
6158 and optinfo_emit_json_cc_tests.
6159 * selftest.h (selftest::json_cc_tests): New decl.
6160 (selftest::optinfo_emit_json_cc_tests): New decl.
6161 * toplev.c: Include "optinfo-emit-json.h".
6162 (compile_file): Call optimization_records_finish.
6163 (do_compile): Call optimization_records_start.
6164 * tree-ssa-live.c: Include optinfo.h.
6165 (remove_unused_scope_block_p): Retain inlining information if
6166 optinfo_wants_inlining_info_p returns true.
6168 2018-07-20 Richard Biener <rguenther@suse.de>
6171 * dwarf2out.c (dwarf2out_die_ref_for_decl): Test in_lto_p
6172 to cover -flto-partition=none.
6174 2018-07-20 Martin Liska <mliska@suse.cz>
6176 * tree.h (DECL_LOCATION_RANGE): Remove unused macro.
6177 (get_decl_source_range): Remove unused function.
6179 2018-07-20 Richard Biener <rguenther@suse.de>
6181 * tree-ssa-sccvn.h (struct vn_nary_op_s): Add next member.
6182 (struct vn_phi_s): Likewise.
6183 (struct vn_reference_s): Likewise.
6184 * tree-ssa-sccvn.c (vn_nary_op_hasher::equal): Add shortcut
6185 for searching the slot of an entry known to be in the hash itself.
6186 (vn_phi_hasher::equal): Likewise.
6187 (vn_reference_hasher::equal): Likewise.
6188 (last_inserted_ref, last_inserted_phi, last_inserted_nary): New
6190 (optimistic_info, current_info): Remove, keeping only valid_info.
6191 (vn_reference_lookup_1): Remove fallback lookup.
6192 (vn_reference_lookup_2): Likewise.
6193 (vn_nary_op_lookup_1): Likewise.
6194 (vn_phi_lookup): Likewise.
6195 (vn_nary_build_or_lookup_1): Make sure to not chain the built
6197 (vn_reference_insert): Adjust, chain the inserted hash element
6198 at last_inserted_ref.
6199 (vn_reference_insert_pieces): Likewise.
6200 (visit_reference_op_call): Likewise.
6201 (vn_nary_op_insert_into): Chain the inserted hash element at
6203 (vn_nary_op_insert_pieces): Adjust.
6204 (vn_nary_op_insert): Likewise.
6205 (vn_nary_op_insert_stmt): Likewise.
6206 (vn_phi_insert): Adjust, chain the inserted hash element at
6208 (process_scc): Remove clearing and copying the optimistic
6209 table. Instead remove elements inserted during an optimistic
6210 iteration from the single table we maintain.
6211 (init_scc_vn): Adjust.
6212 (free_scc_vn): Likewise.
6213 (sccvn_dom_walker::record_cond): Likewise.
6214 (sccvn_dom_walker::after_dom_children): Likewise.
6216 2018-07-19 Martin Sebor <msebor@redhat.com>
6218 PR tree-optimization/84047
6219 PR tree-optimization/83776
6220 * tree-vrp.c (vrp_prop::check_mem_ref): New function.
6221 (check_array_bounds): Call it.
6223 2018-07-19 Martin Sebor <msebor@redhat.com>
6225 * align.h (align_flags): Use member initialization.
6227 2018-07-19 David Malcolm <dmalcolm@redhat.com>
6229 * Makefile.in (OBJS): Add optinfo.o.
6230 * coretypes.h (class symtab_node): New forward decl.
6231 (struct cgraph_node): New forward decl.
6232 (class varpool_node): New forward decl.
6233 * dump-context.h: New file.
6234 * dumpfile.c: Include "optinfo.h", "dump-context.h", "cgraph.h",
6236 (refresh_dumps_are_enabled): Use optinfo_enabled_p.
6237 (set_dump_file): Call dumpfile_ensure_any_optinfo_are_flushed.
6238 (set_alt_dump_file): Likewise.
6239 (dump_context::~dump_context): New dtor.
6240 (dump_gimple_stmt): Move implementation to...
6241 (dump_context::dump_gimple_stmt): ...this new member function.
6242 Add the stmt to any pending optinfo, creating one if need be.
6243 (dump_gimple_stmt_loc): Move implementation to...
6244 (dump_context::dump_gimple_stmt_loc): ...this new member function.
6245 Start a new optinfo and add the stmt to it.
6246 (dump_gimple_expr): Move implementation to...
6247 (dump_context::dump_gimple_expr): ...this new member function.
6248 Add the stmt to any pending optinfo, creating one if need be.
6249 (dump_gimple_expr_loc): Move implementation to...
6250 (dump_context::dump_gimple_expr_loc): ...this new member function.
6251 Start a new optinfo and add the stmt to it.
6252 (dump_generic_expr): Move implementation to...
6253 (dump_context::dump_generic_expr): ...this new member function.
6254 Add the tree to any pending optinfo, creating one if need be.
6255 (dump_generic_expr_loc): Move implementation to...
6256 (dump_context::dump_generic_expr_loc): ...this new member
6257 function. Add the tree to any pending optinfo, creating one if
6259 (dump_printf): Move implementation to...
6260 (dump_context::dump_printf_va): ...this new member function. Add
6261 the text to any pending optinfo, creating one if need be.
6262 (dump_printf_loc): Move implementation to...
6263 (dump_context::dump_printf_loc_va): ...this new member function.
6264 Start a new optinfo and add the stmt to it.
6265 (dump_dec): Move implementation to...
6266 (dump_context::dump_dec): ...this new member function. Add the
6267 value to any pending optinfo, creating one if need be.
6268 (dump_context::dump_symtab_node): New member function.
6269 (dump_context::get_scope_depth): New member function.
6270 (dump_context::begin_scope): New member function.
6271 (dump_context::end_scope): New member function.
6272 (dump_context::ensure_pending_optinfo): New member function.
6273 (dump_context::begin_next_optinfo): New member function.
6274 (dump_context::end_any_optinfo): New member function.
6275 (dump_context::s_current): New global.
6276 (dump_context::s_default): New global.
6277 (dump_scope_depth): Delete global.
6278 (dumpfile_ensure_any_optinfo_are_flushed): New function.
6279 (dump_symtab_node): New function.
6280 (get_dump_scope_depth): Reimplement in terms of dump_context.
6281 (dump_begin_scope): Likewise.
6282 (dump_end_scope): Likewise.
6283 (selftest::temp_dump_context::temp_dump_context): New ctor.
6284 (selftest::temp_dump_context::~temp_dump_context): New dtor.
6285 (selftest::verify_item): New function.
6286 (ASSERT_IS_TEXT): New macro.
6287 (ASSERT_IS_TREE): New macro.
6288 (ASSERT_IS_GIMPLE): New macro.
6289 (selftest::test_capture_of_dump_calls): New test.
6290 (selftest::dumpfile_c_tests): Call it.
6291 * dumpfile.h (dump_printf, dump_printf_loc, dump_basic_block)
6292 (dump_generic_expr_loc, dump_generic_expr, dump_gimple_stmt_loc)
6293 (dump_gimple_stmt, dump_dec): Gather these related decls and add a
6294 descriptive comment.
6295 (dump_function, print_combine_total_stats, enable_rtl_dump_file)
6296 (dump_node, dump_bb): Move these unrelated decls.
6297 (class dump_manager): Add leading comment.
6298 * optinfo.cc: New file.
6299 * optinfo.h: New file.
6301 2018-07-19 Michael Collison <michael.collison@arm.com>
6302 Richard Henderson <rth@redhat.com>
6304 * config/aarch64/aarch64.md (subv<GPI>4, usubv<GPI>4): New patterns.
6305 (subti): Handle op1 zero.
6306 (subvti4, usub4ti4): New.
6307 (*sub<GPI>3_compare1_imm): New.
6308 (sub<GPI>3_carryinCV): New.
6309 (*sub<GPI>3_carryinCV_z1_z2, *sub<GPI>3_carryinCV_z1): New.
6310 (*sub<GPI>3_carryinCV_z2, *sub<GPI>3_carryinCV): New.
6312 2018-07-19 Michael Collison <michael.collison@arm.com>
6313 Richard Henderson <rth@redhat.com>
6315 * config/aarch64/aarch64.md: (addv<GPI>4, uaddv<GPI>4): New.
6316 (addti3): Create simpler code if low part is already known to be 0.
6317 (addvti4, uaddvti4): New.
6318 (*add<GPI>3_compareC_cconly_imm): New.
6319 (*add<GPI>3_compareC_cconly): New.
6320 (*add<GPI>3_compareC_imm): New.
6321 (*add<GPI>3_compareC): Rename from add<GPI>3_compare1; do not
6322 handle constants within this pattern..
6323 (*add<GPI>3_compareV_cconly_imm): New.
6324 (*add<GPI>3_compareV_cconly): New.
6325 (*add<GPI>3_compareV_imm): New.
6326 (add<GPI>3_compareV): New.
6327 (add<GPI>3_carryinC, add<GPI>3_carryinV): New.
6328 (*add<GPI>3_carryinC_zero, *add<GPI>3_carryinV_zero): New.
6329 (*add<GPI>3_carryinC, *add<GPI>3_carryinV): New.
6330 ((*add<GPI>3_compareC_cconly_imm): Replace 'ne' operator
6331 with 'comparison' operator.
6332 (*add<GPI>3_compareV_cconly_imm): Ditto.
6333 (*add<GPI>3_compareV_cconly): Ditto.
6334 (*add<GPI>3_compareV_imm): Ditto.
6335 (add<GPI>3_compareV): Ditto.
6336 (add<mode>3_carryinC): Ditto.
6337 (*add<mode>3_carryinC_zero): Ditto.
6338 (*add<mode>3_carryinC): Ditto.
6339 (add<mode>3_carryinV): Ditto.
6340 (*add<mode>3_carryinV_zero): Ditto.
6341 (*add<mode>3_carryinV): Ditto.
6343 2018-07-19 Michael Collison <michael.collison@arm.com>
6344 Richard Henderson <rth@redhat.com>
6346 * config/aarch64/aarch64-modes.def (CC_V): New.
6347 * config/aarch64/aarch64-protos.h
6348 (aarch64_addti_scratch_regs): Declare
6349 (aarch64_subvti_scratch_regs): Declare.
6350 (aarch64_expand_subvti): Declare.
6351 (aarch64_gen_unlikely_cbranch): Declare
6352 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Test
6353 for signed overflow using CC_Vmode.
6354 (aarch64_get_condition_code_1): Handle CC_Vmode.
6355 (aarch64_gen_unlikely_cbranch): New function.
6356 (aarch64_addti_scratch_regs): New function.
6357 (aarch64_subvti_scratch_regs): New function.
6358 (aarch64_expand_subvti): New function.
6360 2018-07-19 Andre Vieira <andre.simoesdiasvieira@arm.com>
6362 * config/aarch64/aarch64-option-extensions.def: New entry for profile
6364 * config/aarch64/aarch64.h (AARCH64_FL_PROFILE): New.
6365 * doc/invoke.texi (aarch64-feature-modifiers): New entry for profile
6368 2018-07-19 Andre Vieira <andre.simoesdiasvieira@arm.com>
6371 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Make
6372 address check not strict.
6374 2018-07-19 Andre Vieira <andre.simoesdiasvieira@arm.com>
6376 * config/aarch64/aarch64-simd.md (aarch64_simd_mov<VQ:mode>): Replace
6378 (store_pair_lanes<mode>): Likewise.
6379 * config/aarch64/aarch64-protos.h (aarch64_addr_query_type): Add new
6380 enum value 'ADDR_QUERY_LDP_STP_N'.
6381 * config/aarch64/aarch64.c (aarch64_addr_query_type): Likewise.
6382 (aarch64_print_address_internal): Add declaration.
6383 (aarch64_print_ldpstp_address): Remove.
6384 (aarch64_classify_address): Adapt mode for 'ADDR_QUERY_LDP_STP_N'.
6385 (aarch64_print_operand): Change printing of 'y'.
6386 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Use
6387 new enum value 'ADDR_QUERY_LDP_STP_N', don't hardcode mode and use
6388 'true' rather than '1'.
6389 * config/aarch64/constraints.md (Uml): Likewise.
6390 (Uml): Rename to Umn.
6393 2018-07-19 Richard Biener <rguenther@suse.de>
6395 * tree-ssa-sccvn.h (struct vn_phi_s): Make phiargs member
6397 * tree-ssa-sccvn.c: Remove alloc-pool.h use.
6398 (vn_phi_hasher): Derive from nofree_ptr_hash and remove remove method.
6399 (vn_reference_hasher): Likewise.
6400 (struct vn_tables_s): Remove obstack and alloc-pool members.
6401 (vn_tables_obstack, vn_tables_insert_obstack): New global obstacks.
6402 (vn_nary_build_or_lookup_1): Manually build in vn_tables_insert_obstack.
6403 (vn_reference_insert): Allocate from obstack instead of from alloc-pool.
6404 (vn_reference_insert_pieces): Likewise.
6405 (alloc_vn_nary_op_noinit): Adjust.
6406 (vn_nary_op_insert_stmt): Allocate phiargs in-place.
6407 (vn_phi_eq): Adjust.
6408 (shared_lookup_phiargs): Remove.
6409 (vn_phi_lookup): Allocate temporary vn_phi_s on the stack.
6410 (vn_phi_insert): Allocate from obstack instead of from alloc-pool.
6411 (visit_reference_op_call): Likewise.
6412 (copy_nary, copy_phi, copy_reference): Remove.
6413 (process_scc): Rewind the obstack when iterating. Do not
6414 copy the elements to valid_info but just move them from one
6415 hashtable to the other.
6416 (allocate_vn_table): Adjust.
6417 (free_vn_table): Likewise.
6418 (init_scc_vn): Likewise.
6419 (free_scc_vn): Likewise.
6421 2018-07-19 H.J. Lu <hongjiu.lu@intel.com>
6424 * config/i386/i386.c (rest_of_insert_endbranch): Lookup
6425 indirect_return as function type attribute.
6426 (ix86_attribute_table): Change indirect_return to function
6428 * doc/extend.texi: Update indirect_return attribute.
6430 2018-07-19 Aldy Hernandez <aldyh@redhat.com>
6432 * wide-int.h (widest2_int): New.
6433 * gimple-fold.c (arith_overflowed_p): Use it.
6434 * tree.h (widest2_int_cst): New.
6435 * tree-vrp.c (wide_int_binop_overflow): Rename from
6436 vrp_int_const_binop.
6437 Rewrite to work on trees.
6438 (extract_range_from_multiplicative_op_1): Abstract code to...
6439 (wide_int_range_min_max): ...here.
6440 (wide_int_range_cross_product): ...and here.
6441 (extract_range_from_binary_expr_1): Abstract overflow code to...
6442 (wide_int_range_mult_wrapping): ...here.
6443 * tree-vrp.h (wide_int_range_cross_product): New.
6444 (wide_int_range_mult_wrapping): New.
6446 2018-07-19 Andrew Senkevich <andrew.senkevich@intel.com>
6447 Julia Koval <julia.koval@intel.com>
6449 * config/i386/x86-tune-costs.h (skylake_memcpy,
6450 skylake_memset): Replace rep_prefix with unrolling for size 512.
6452 2018-07-18 Kugan Vivekanandarajah <kuganv@linaro.org>
6455 * tree-ssa-phiopt.c (cond_removal_in_popcount_pattern): Handle
6456 comparision with EQ_EXPR in last stmt.
6458 2018-07-18 Kelvin Nilsen <kelvin@gcc.gnu.org>
6460 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Rename
6461 this subsection to "PowerPC AltiVec/VSX Built-in Functions".
6462 (PowerPC AltiVec/VSX Built-in Functions): New name for subsection
6463 previously known as "PowerPC AltiVec Built-in Functions". Move
6464 some material to new subsubsections "PowerPC AltiVec Built-in
6465 Functions on ISA 2.06" and "PowerPC AltiVec Built-in Functions on
6467 (PowerPC Altivec Built-in Functions on ISA 2.05): New subsubsection.
6468 (PowerPC Altivec Built-in Functions on ISA 2.06): Likewise.
6469 (PowerPC Altivec Built-in Functions on ISA 2.07): Likewise.
6470 (PowerPC Altivec Built-in Functions on ISA 3.0): Likewise.
6472 2018-07-18 Richard Biener <rguenther@suse.de>
6474 PR tree-optimization/86557
6475 * tree-vect-patterns.c (vect_recog_divmod_pattern): Also handle
6478 2018-07-18 Ilya Leoshkevich <iii@linux.ibm.com>
6480 * config/s390/s390.c (s390_function_profiler): Generate CFI.
6482 2018-07-17 Jeff Law <law@redhat.com>
6484 * config/arm/arm.c (get_label_padding): Update for recent
6485 changes to label_to_alignment.
6487 PR tree-optimization/86010
6488 * tree-ssa-dse.c (compute_trims): Fix typo/thinko.
6490 * config/mips/mips.c (vr4130_align_insns): Update for recent
6491 changes to label_to_alignment.
6493 * config/frv/frv.c (frv_label_align): Update for recent changes
6494 to label_to_alignment.
6496 * config/nios2/nios2.c (nios2_label_align): Update for recent
6497 changes which dropped ALIGN_LABELS_LOG.
6499 2018-07-17 Andreas Schwab <schwab@linux-m68k.org>
6501 * config/m68k/m68k.md (umulsi3_highpart+1, const_umulsi3_highpart)
6502 (smulsi3_highpart+1, const_smulsi3_highpart): Add CC_STATUS_INIT.
6504 2018-07-17 Claudiu Zissulescu <claziss@synopsys.com>
6506 * config/arc/arc.c (arc_label_align): Use align_labels instead of
6507 deprecated align_labels_log.
6509 2018-07-17 Richard Biener <rguenther@suse.de>
6512 * dwarf2out.c (init_sections_and_labels): Always generate
6513 a debug_line_str_section for early LTO debug.
6514 (dwarf2out_finish): Reset debug_line_str_hash output early.
6515 Bump counter for extra dwarf5 .debug_loc labels to not conflict
6517 (dwarf2out_early_finish): Output debug_line_str.
6519 2018-07-17 Robin Dapp <rdapp@linux.ibm.com>
6521 * config/s390/s390.c (preferred_la_operand_p): Do not use LA with
6522 index register on z196 or later.
6524 2018-07-17 Robin Dapp <rdapp@linux.ibm.com>
6526 * config/s390/s390.c (s390_default_align): Set default function
6528 (s390_override_options_after_change): Call s390_default align.
6529 (s390_option_override_internal): Call s390_default_align.
6530 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define.
6532 2018-07-17 Jakub Jelinek <jakub@redhat.com>
6535 * omp-low.c (create_task_copyfn): Copy over also fields corresponding
6536 to _looptemp_ clauses, other than the first two.
6538 2018-07-17 Martin Liska <mliska@suse.cz>
6540 * opts.c: Do not enable OPT_falign_* for -Os.
6542 2018-07-17 Martin Liska <mliska@suse.cz>
6544 * align.h (MAX_CODE_ALIGN): New.
6545 (MAX_CODE_ALIGN_VALUE): New.
6546 * common/config/i386/i386-common.c (ix86_handle_option):
6547 (MAX_CODE_ALIGN): Moved to align.h.
6548 * final.c (MAX_CODE_ALIGN): Likewise.
6549 * opts.c (parse_and_check_align_values):
6550 (MAX_CODE_ALIGN): Likewise.
6551 (MAX_CODE_ALIGN_VALUE): Likewise.
6553 2018-07-17 Martin Liska <mliska@suse.cz>
6555 * config/i386/att.h (ASM_OUTPUT_ALIGN): Fix spacing
6556 in order to fulfil coding style.
6557 * config/i386/cygming.h (ASM_OUTPUT_ALIGN): Likewise.
6558 * config/i386/gas.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
6559 * config/i386/x86-64.h (ASM_OUTPUT_MAX_SKIP_PAD): Likewise.
6560 * config/iq2000/iq2000.h (ASM_OUTPUT_ALIGN): Likewise.
6561 * config/pa/pa.h (ASM_OUTPUT_ALIGN): Likewise.
6562 * config/sparc/sol2.h (ASM_OUTPUT_ALIGN_WITH_NOP): Likewise.
6563 * config/sparc/sparc.h (ASM_OUTPUT_ALIGN): Likewise.
6564 * config/visium/visium.h (ASM_OUTPUT_ALIGN): Likewise.
6565 (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
6567 2018-07-17 Martin Liska <mliska@suse.cz>
6569 * align.h: New file.
6570 * config/alpha/alpha.c (alpha_align_insns_1): Use align_functions
6572 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Use new return type
6573 align_flags of label_to_alignment.
6574 * config/m32r/m32r.h (LOOP_ALIGN): Wrap returned values into
6576 * config/m68k/m68k.c: Do not use removed align_labels_value and
6578 * config/nds32/nds32.h (JUMP_ALIGN): Wrap result into align_flags class.
6579 (LOOP_ALIGN): Likewise.
6580 (LABEL_ALIGN): Likewise.
6581 * config/powerpcspe/powerpcspe.c (TARGET_ASM_LOOP_ALIGN_MAX_SKIP):
6582 Remove not used macro.
6583 (rs6000_loop_align): Change return type to align_flags.
6584 (rs6000_loop_align_max_skip): Remove.
6585 * config/rs6000/rs6000-protos.h (rs6000_loop_align):
6586 Change return type to align_flags.
6587 * config/rs6000/rs6000.c (TARGET_ASM_LOOP_ALIGN_MAX_SKIP):
6588 Remove not used macro.
6589 (rs6000_loop_align): Change return type to align_flags.
6590 (rs6000_loop_align_max_skip): Remove.
6591 * config/rx/rx.h (JUMP_ALIGN): Wrap integer values
6592 * config/rx/rx-protos.h (rx_align_for_label): Make it
6594 * config/rx/rx.c (rx_align_for_label): Change return type
6596 (rx_max_skip_for_label): Remove TARGET_ASM_*_ALIGN_MAX_SKIP
6598 into align_flags class.
6599 (LABEL_ALIGN): Likewise.
6600 (LOOP_ALIGN): Likewise.
6601 * config/s390/s390.c (s390_label_align): Use align_flags
6603 (s390_asm_output_function_label): Likewise.
6604 * config/sh/sh.c (sh_override_options_after_change):
6605 Use align_flags class directly without macros.
6606 (find_barrier): Likewise.
6607 (barrier_align): Likewise.
6608 (sh_loop_align): Likewise.
6609 * config/spu/spu.c (spu_option_override):
6610 Use align_flags_tuple::get_value instead of removed macros.
6611 (spu_sched_init): Likewise.
6612 * config/spu/spu.h (GTY): Likewise.
6613 * config/visium/visium.c (visium_option_override):
6614 Set "8" as default secondary alignment.
6615 * config/visium/visium.h (SUBALIGN_LOG): Define to 3
6616 in order to guarantee secondary alignment of 8.
6617 * coretypes.h: Include align.h header file.
6618 * doc/tm.texi: Remove TARGET_ASM_JUMP_ALIGN_MAX_SKIP,
6619 TARGET_ASM_LOOP_ALIGN_MAX_SKIP, TARGET_ASM_LABEL_ALIGN_MAX_SKIP
6620 and TARGET_ASM_LABEL_ALIGN_AFTER_BARRIER_MAX_SKIP macros.
6621 * doc/tm.texi.in: Likewise.
6622 * final.c (struct label_alignment): Remove not used structure.
6623 (LABEL_ALIGN): Change type to align_flags.
6624 (LOOP_ALIGN): Likewise.
6625 (JUMP_ALIGN): Likewise.
6626 (default_loop_align_max_skip): Remove.
6627 (default_label_align_max_skip): Likewise.
6628 (default_jump_align_max_skip): Likewise.
6629 (default_label_align_after_barrier_max_skip):
6630 (LABEL_TO_ALIGNMENT): Change to access label_align vector.
6631 (LABEL_TO_MAX_SKIP): Remove.
6632 (label_to_alignment): Return align_flags type instead of integer.
6633 (label_to_max_skip): Remove.
6634 (align_fuzz): Use align_flags type.
6635 (compute_alignments): Use align_flags type and use align_flags::max
6636 to combine multiple alignments.
6637 (grow_label_align): Grow vec instead of C array.
6638 (update_alignments): Assign just LABEL_TO_ALIGNMENT.
6639 (shorten_branches): Use align_flags type and use align_flags::max
6640 to combine multiple alignments.
6641 (final_scan_insn_1): Remove usage of secondary alignment that comes
6642 from label alignment, but instead use proper secondary alignment
6643 which is computed in grow_label_align.
6644 * flags.h (struct align_flags_tuple): Move to align.h.
6645 (struct align_flags): Likewise.
6646 (state_align_loops): Rename to align_loops.
6647 (state_align_jumps): Rename to align_jumps.
6648 (state_align_labels): Rename to align_labels.
6649 (state_align_functions): Rename to align_functions.
6650 (align_loops_log): Remove.
6651 (align_jumps_log): Remove.
6652 (align_labels_log): Remove.
6653 (align_functions_log): Remove.
6654 (align_loops_max_skip): Remove.
6655 (align_jumps_max_skip): Remove.
6656 (align_labels_max_skip): Remove.
6657 (align_functions_max_skip): Remove.
6658 (align_loops_value): Remove.
6659 (align_jumps_value): Remove.
6660 (align_labels_value): Remove.
6661 (align_functions_value): Remove.
6662 * output.h (label_to_alignment): Change return type to align_flags.
6663 (label_to_max_skip): Remove.
6664 * target.def: Remove loop_align_max_skip, label_align_max_skip,
6665 jump_align_max_skip macros.
6666 * targhooks.h (default_loop_align_max_skip): Remove.
6667 (default_label_align_max_skip): Likewise.
6668 (default_jump_align_max_skip): Likewise.
6669 (default_label_align_after_barrier_max_skip): Remove.
6670 * toplev.c (read_log_maxskip): Use ::normalize function.
6671 (parse_N_M): Remove not used argument and also call ::normalize.
6672 (parse_alignment_opts): Do not pass unused arguments.
6673 * varasm.c (assemble_start_function): Use directly align_functions
6674 instead of removed macros.
6675 * system.h: Do not poison removed macros.
6677 2018-07-17 Jakub Jelinek <jakub@redhat.com>
6680 * gimplify.c (gimplify_omp_for): Ensure taskloop firstprivatized init
6681 and cond temporaries don't have reference type if iterator has
6682 pointer type. For init use &for_pre_body instead of pre_p if
6683 for_pre_body is non-empty.
6685 2018-07-16 Segher Boessenkool <segher@kernel.crashing.org>
6687 * config/rs6000/rs6000.md (trunc<mode>sf2): Expand truncates of
6688 double-double modes to SFmode directly directly.
6689 (trunc<mode>sf2_fprs): Delete.
6691 2018-07-16 Segher Boessenkool <segher@kernel.crashing.org>
6693 * config/rs6000/rs6000.c (init_float128_ibm): Use the correct names
6694 for conversions between IFmode and the decimal floating point modes.
6695 (init_float128_ieee): Use the correct names for conversions between
6696 KFmode and the decimal floating point modes.
6698 2018-07-16 Segher Boessenkool <segher@kernel.crashing.org>
6700 * config/rs6000/rs6000.c (init_float128_ibm): Use more correct names
6701 for the conversions between TDmode and IFmode.
6702 (init_float128_ieee): Use more correct names for the conversions
6703 between TDmode and KFmode.
6705 2018-07-16 Jakub Jelinek <jakub@redhat.com>
6707 PR tree-optimization/86526
6708 * builtins.c (expand_builtin_memcmp): Formatting fixes.
6709 (inline_expand_builtin_string_cmp): Likewise.
6710 (inline_string_cmp): Likewise. Use c_readstr instead of
6711 builtin_memcpy_read_str. Add unit_mode temporary.
6713 2018-07-16 Bernd Edlinger <bernd.edlinger@hotmail.de>
6716 * builtins.c (check_access): Bail out if range[0] is no INTEGER_CST.
6717 * expr.c (string_constant): Fix the element size of ARRAY_TYPE.
6719 2018-07-16 Kelvin Nilsen <kelvin@gcc.gnu.org>
6721 * doc/extend.texi (PowerPC AltiVec Built-in Functions):
6722 Alphabetize prototypes of built-in functions, separating out
6723 built-in functions that are listed in this section but should be
6724 described elsewhere.
6726 2018-07-16 Uros Bizjak <ubizjak@gmail.com>
6729 * expmed.c (emit_store_flag): Do not emit setcc followed by a
6730 conditional move when trapping comparison was split to a
6731 non-trapping one (and vice versa).
6733 2018-07-16 Ilya Leoshkevich <iii@linux.ibm.com>
6735 * config/s390/s390.c (s390_function_profiler): Generate nops
6736 instead of profiler call sequences.
6737 * config/s390/s390.opt: Add the new option.
6739 2018-07-16 Ilya Leoshkevich <iii@linux.ibm.com>
6741 * config/s390/s390.c (s390_function_profiler): Generate
6742 __mcount_loc section.
6743 * config/s390/s390.opt: Add the new option.
6745 2018-07-16 Ilya Leoshkevich <iii@linux.ibm.com>
6747 * common.opt: Add the new warning.
6748 * config/s390/s390.c (s390_function_profiler): Emit "brasl
6749 %r0,__fentry__" when -mfentry is specified.
6750 (s390_option_override_internal): Disallow -mfentry for 31-bit
6752 * config/s390/s390.opt: Add the new option.
6754 2018-07-16 Richard Biener <rguenther@suse.de>
6757 * dwarf2out.c (dwarf2out_register_external_die): Assign DIE parents
6758 for function-local FUNCTION_DECL and RESULT_DECL immediately.
6760 2018-07-16 Martin Liska <mliska@suse.cz>
6763 * ipa-pure-const.c (malloc_candidate_p): Revert ::get
6766 2017-07-16 Claudiu Zissulescu <claziss@synopsys.com>
6768 * config/arc/arcHS.md: Update ARCHS scheduling rules.
6770 2017-07-16 Claudiu Zissulescu <claziss@synopsys.com>
6772 * config/arc/arc-arch.h (arc_tune_attr): Add new tune parameters
6774 * config/arc/arc-cpus.def (hs4x): New cpu.
6776 * config/arc/arc-tables.opt: Regenerate.
6777 * config/arc/arc.c (arc_sched_issue_rate): New function.
6778 (TARGET_SCHED_ISSUE_RATE): Define.
6779 (TARGET_SCHED_EXPOSED_PIPELINE): Likewise.
6780 * config/arc/arc.md (attr type): Add fpu_fuse, fpu_sdiv, fpu_ddiv,
6782 (attr tune): Add ARCHS4x tune values.
6783 (attr tune_dspmpy): Define.
6784 (*tst): Correct instruction type.
6785 * config/arc/arcHS.md: Don't use this automaton for ARCHS4x cpus.
6786 * config/arc/arcHS4x.md: New file.
6787 * config/arc/fpu.md: Update instruction type attributes.
6788 * config/arc/t-multilib: Regenerate.
6790 2018-07-16 Tom de Vries <tdevries@suse.de>
6793 * var-tracking.c (vt_initialize): Fix pre_dec handling.
6795 2018-07-16 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
6797 * config/aarch64/atomics.md (aarch64_store_execlusive<mode>): Add
6800 2018-07-16 Eric Botcazou <ebotcazou@adacore.com>
6802 PR tree-optimization/86514
6803 * tree-ssa-reassoc.c (init_range_entry) <CASE_CONVERT>: Return for a
6804 conversion to a boolean type from a type with greater precision.
6806 2018-07-16 Tom de Vries <tdevries@suse.de>
6808 * var-tracking.c (vt_initialize): Print adjusted insn slim if
6809 dump_flags request TDF_SLIM.
6811 2018-07-16 Aldy Hernandez <aldyh@redhat.com>
6813 * fold-const.c (int_const_binop_1): Abstract...
6814 (wide_int_binop): ...wide int code here.
6815 (poly_int_binop): ...poly int code here.
6816 Abstract the rest of int_const_binop_1 into int_const_binop.
6817 * fold-const.h (wide_int_binop): New.
6818 * tree-vrp.c (vrp_int_const_binop): Call wide_int_binop.
6819 Remove useless PLUS/MINUS_EXPR case.
6820 (zero_nonzero_bits_from_vr): Move wide int code...
6821 (zero_nonzero_bits_from_bounds): ...here.
6822 (extract_range_from_binary_expr_1): Move mask optimization code...
6823 (range_easy_mask_min_max): ...here.
6824 * tree-vrp.h (zero_nonzero_bits_from_bounds): New.
6825 (range_easy_mask_min_max): New.
6827 2018-07-15 Jeff Law <law@redhat.com>
6830 * config/sh/sh.c (output_mi_thunk): Remove dead conditional
6833 2018-07-14 Jim Wilson <jimw@sifive.com>
6835 * config/riscv/linux.h (TARGET_ASM_FILE_END): New.
6837 2018-07-14 Paul Koning <ni1d@arrl.net>
6839 * config/pdp11/pdp11.c (pdp11_rtx_costs): Bugfixes.
6841 2018-07-13 Jan Hubicka <hubicka@ucw.cz>
6843 * lto-streamer-out.c (copy_function_or_variable): Dump info about
6846 2018-07-13 Bill Schmidt <wschmidt@linux.ibm.com>
6847 Steve Munroe <munroesj52@gmail.com>
6849 * config/rs6000/emmintrin.h (_mm_and_si128): New function.
6850 (_mm_andnot_si128): Likewise.
6851 (_mm_or_si128): Likewise.
6852 (_mm_xor_si128): Likewise.
6854 2018-07-13 Qing Zhao <qing.zhao@oracle.com>
6857 * builtins.c (expand_builtin_memcmp): Inline the calls first
6858 when result_eq is false.
6859 (expand_builtin_strcmp): Inline the calls first.
6860 (expand_builtin_strncmp): Likewise.
6861 (inline_string_cmp): New routine. Expand a string compare
6862 call by using a sequence of char comparison.
6863 (inline_expand_builtin_string_cmp): New routine. Inline expansion
6864 a call to str(n)cmp/memcmp.
6865 * doc/invoke.texi (--param builtin-string-cmp-inline-length):
6867 * params.def (BUILTIN_STRING_CMP_INLINE_LENGTH): New.
6869 2018-07-13 Richard Earnshaw <rearnsha@arm.com>
6871 * config/arm/driver-arm.c: Include arm-native.h.
6872 (host_detect_local_cpu): Use auto-generated data tables.
6873 (vendors, arm_cpu_table): Delete. Move part information to ...
6874 * config/arm/arm-cpus.in: ... here.
6875 * config/arm/parsecpu.awk (gen_native): New function.
6876 (vendor, part): New CPU fields.
6877 (END): Add support for building the native CPU detection tables.
6878 * config/arm/t-arm (arm-native.h): Add build rule.
6879 (driver-arm.o): Add dependency on arm-native.h.
6881 2018-07-13 Richard Biener <rguenther@suse.de>
6884 * match.pd (addr1 - addr2): Allow either of the operand to
6887 2018-07-13 Tom de Vries <tdevries@suse.de>
6889 * tree-inline.c (remap_ssa_name): Save and reuse debug exprs generated
6892 2018-07-13 Jackson Woodruff <jackson.woodruff@arm.com>
6894 * config/aarch64/aarch64.c (aarch64_operands_adjust_ok_for_ldpstp): Use
6895 arrays instead of numbered variables.
6897 2018-07-13 Eric Botcazou <ebotcazou@adacore.com>
6899 * config/sparc/sparc-protos.h (sparc_compute_frame_size): Delete.
6900 * config/sparc/sparc.c (sparc_compute_frame_size): Make static.
6902 2018-07-13 Richard Biener <rguenther@suse.de>
6905 * dwarf2out.c (gen_type_die_with_usage): Use scope_die_for
6906 instead of get_context_die.
6908 2018-07-13 Kugan Vivekanandarajah <kuganv@linaro.org>
6909 Richard Biener <rguenther@suse.de>
6912 * tree-ssa-loop-niter.c (number_of_iterations_popcount): Check
6913 that the loop latch destination where phi is defined.
6915 2018-07-12 Kito Cheng <kito.cheng@gmail.com>
6917 * config/riscv/riscv.c (enum riscv_privilege_levels): Add UNKNOWN_MODE.
6918 (riscv_expand_epilogue): Add assertion to check interrupt mode.
6919 (riscv_set_current_function): Extract getting interrupt type to new
6921 (riscv_get_interrupt_type): New function.
6922 (riscv_merge_decl_attributes): New function, checking interrupt type is
6924 (TARGET_MERGE_DECL_ATTRIBUTES): Define.
6926 2018-07-12 Paul Koning <ni1d@arrl.net>
6928 * config/pdp11/pdp11.c (pdp11_output_def): Fix typo in .set
6931 2018-07-12 Paul Koning <ni1d@arrl.net>
6933 * doc/rtl.texi (REG_NONNEG): Remove decrement and branch until
6934 zero reference, add doloop_end instead.
6935 * doc/md.texi (decrement_and_branch_until_zero): Remove.
6936 (Looping patterns): Remove decrement_and_branch_until_zero. Add
6937 detail for doloop_end.
6939 2018-07-12 Martin Sebor <msebor@redhat.com>
6942 * attribs.c (decl_attributes): Reject conflicting attributes before
6943 calling attribute handlers.
6945 2018-07-12 Jan Hubicka <hubicka@ucw.cz>
6947 * dumpfile.c (gcc::dump_manager::get_dump_file_name): Add PART
6949 (gcc::dump_manager::get_dump_file_name): likewise.
6950 (dump_begin): Likewise.
6951 * dumpfile.h (dump_begin): Update prototype.
6952 (gcc::dump_manager::get_dump_file_name,
6953 gcc::dump_manager::get_dump_file_name): Update prototype.
6955 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
6957 * internal-fn.h (vectorizable_internal_fn_p): New function.
6958 * tree-vect-slp.c (compatible_calls_p): Likewise.
6959 (vect_build_slp_tree_1): Remove nops argument. Handle calls
6960 to internal functions.
6961 (vect_build_slp_tree_2): Update call to vect_build_slp_tree_1.
6963 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
6965 * fold-const.h (inverse_conditions_p): Declare.
6966 * fold-const.c (inverse_conditions_p): New function.
6967 * match.pd: Use inverse_conditions_p. Add folds of view_converts
6968 that test the inverse condition of a conditional internal function.
6969 * internal-fn.h (vectorized_internal_fn_supported_p): Declare.
6970 * internal-fn.c (internal_fn_mask_index): Handle conditional
6972 (vectorized_internal_fn_supported_p): New function.
6973 * tree-if-conv.c: Include internal-fn.h and fold-const.h.
6974 (any_pred_load_store): Replace with...
6975 (need_to_predicate): ...this new variable.
6976 (redundant_ssa_names): New variable.
6977 (ifcvt_can_use_mask_load_store): Move initial checks to...
6978 (ifcvt_can_predicate): ...this new function. Handle tree codes
6979 for which a conditional internal function exists.
6980 (if_convertible_gimple_assign_stmt_p): Use ifcvt_can_predicate
6981 instead of ifcvt_can_use_mask_load_store. Update after variable
6983 (predicate_load_or_store): New function, split out from
6984 predicate_mem_writes.
6985 (check_redundant_cond_expr): New function.
6986 (value_available_p): Likewise.
6987 (predicate_rhs_code): Likewise.
6988 (predicate_mem_writes): Rename to...
6989 (predicate_statements): ...this. Use predicate_load_or_store
6990 and predicate_rhs_code.
6991 (combine_blocks, tree_if_conversion): Update after above name changes.
6992 (ifcvt_local_dce): Handle redundant_ssa_names.
6993 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern): Handle
6994 general conditional functions.
6995 * tree-vect-stmts.c (vectorizable_call): Likewise.
6997 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
6998 Alan Hayward <alan.hayward@arm.com>
6999 David Sherwood <david.sherwood@arm.com>
7001 * internal-fn.h (can_interpret_as_conditional_op_p): Declare.
7002 * internal-fn.c (can_interpret_as_conditional_op_p): New function.
7003 * tree-ssa-math-opts.c (convert_mult_to_fma_1): Handle conditional
7004 plus and minus and convert them into IFN_COND_FMA-based sequences.
7005 (convert_mult_to_fma): Handle conditional plus and minus.
7007 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
7009 * doc/md.texi (cond_fma, cond_fms, cond_fnma, cond_fnms): Document.
7010 * optabs.def (cond_fma_optab, cond_fms_optab, cond_fnma_optab)
7011 (cond_fnms_optab): New optabs.
7012 * internal-fn.def (COND_FMA, COND_FMS, COND_FNMA, COND_FNMS): New
7014 (FMA): Use DEF_INTERNAL_FLT_FN rather than DEF_INTERNAL_FLT_FLOATN_FN.
7015 * internal-fn.h (get_conditional_internal_fn): Declare.
7016 (get_unconditional_internal_fn): Likewise.
7017 * internal-fn.c (cond_ternary_direct): New macro.
7018 (expand_cond_ternary_optab_fn): Likewise.
7019 (direct_cond_ternary_optab_supported_p): Likewise.
7020 (FOR_EACH_COND_FN_PAIR): Likewise.
7021 (get_conditional_internal_fn): New function.
7022 (get_unconditional_internal_fn): Likewise.
7023 * gimple-match.h (gimple_match_op::MAX_NUM_OPS): Bump to 5.
7024 (gimple_match_op::gimple_match_op): Add a new overload for 5
7026 (gimple_match_op::set_op): Likewise.
7027 (gimple_resimplify5): Declare.
7028 * genmatch.c (decision_tree::gen): Generate simplifications for
7030 * gimple-match-head.c (gimple_simplify): Define an overload for
7031 5 operands. Handle calls with 5 arguments in the top-level overload.
7032 (convert_conditional_op): Handle conversions from unconditional
7033 internal functions to conditional ones.
7034 (gimple_resimplify5): New function.
7035 (build_call_internal): Pass a fifth operand.
7036 (maybe_push_res_to_seq): Likewise.
7037 (try_conditional_simplification): Try converting conditional
7038 internal functions to unconditional internal functions.
7039 Handle 3-operand unconditional forms.
7040 * match.pd (UNCOND_TERNARY, COND_TERNARY): Operator lists.
7041 Define ternary equivalents of the current rules for binary conditional
7043 * config/aarch64/aarch64.c (aarch64_preferred_else_value): Handle
7045 * config/aarch64/iterators.md (UNSPEC_COND_FMLA, UNSPEC_COND_FMLS)
7046 (UNSPEC_COND_FNMLA, UNSPEC_COND_FNMLS): New unspecs.
7047 (optab): Handle them.
7048 (SVE_COND_FP_TERNARY): New int iterator.
7049 (sve_fmla_op, sve_fmad_op): New int attributes.
7050 * config/aarch64/aarch64-sve.md (cond_<optab><mode>)
7051 (*cond_<optab><mode>_2, *cond_<optab><mode_4)
7052 (*cond_<optab><mode>_any): New SVE_COND_FP_TERNARY patterns.
7054 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
7056 * target.def (preferred_else_value): New target hook.
7057 * doc/tm.texi.in (TARGET_PREFERRED_ELSE_VALUE): New hook.
7058 * doc/tm.texi: Regenerate.
7059 * targhooks.h (default_preferred_else_value): Declare.
7060 * targhooks.c (default_preferred_else_value): New function.
7061 * internal-fn.h (conditional_internal_fn_code): Declare.
7062 * internal-fn.c (FOR_EACH_CODE_MAPPING): New macro.
7063 (get_conditional_internal_fn): Use it.
7064 (conditional_internal_fn_code): New function.
7065 * gimple-match.h (gimple_match_cond): New struct.
7066 (gimple_match_op): Add a cond member function.
7067 (gimple_match_op::gimple_match_op): Update all forms to take a
7069 * genmatch.c (expr::gen_transform): Use the same condition as res_op
7070 for the suboperation, but don't specify a particular else_value.
7071 * tree-ssa-sccvn.c (vn_nary_simplify, vn_reference_lookup_3)
7072 (visit_nary_op, visit_reference_op_load): Pass
7073 gimple_match_cond::UNCOND to the gimple_match_op constructor.
7074 * gimple-match-head.c: Include tree-eh.h
7075 (convert_conditional_op): New function.
7076 (maybe_resimplify_conditional_op): Likewise.
7077 (gimple_resimplify1): Call maybe_resimplify_conditional_op.
7078 (gimple_resimplify2): Likewise.
7079 (gimple_resimplify3): Likewise.
7080 (gimple_resimplify4): Likewise.
7081 (maybe_push_res_to_seq): Return null for conditional operations.
7082 (try_conditional_simplification): New function.
7083 (gimple_simplify): Call it. Pass conditions to the gimple_match_op
7085 * match.pd: Fold VEC_COND_EXPRs of an IFN_COND_* call to a new
7087 * config/aarch64/aarch64.c (aarch64_preferred_else_value): New
7089 (TARGET_PREFERRED_ELSE_VALUE): Redefine.
7091 2018-07-12 Jan Hubicka <hubicka@ucw.cz>
7093 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not stream
7095 (hash_tree): Do not hash DECL_FCONTEXT
7096 * tree-streamer-in.c (lto_input_ts_field_decl_tree_pointers):
7097 Do not stream DECL_FCONTEXT.
7098 * tree-streamer-out.c (write_ts_field_decl_tree_pointers): Likewise.
7099 * tree.c (free_lang_data_in_decl): Free DECL_FCONTEXT.
7101 2018-07-12 Richard Biener <rguenther@suse.de>
7104 * dwarf2out.c (gen_block_die): Only output blocks when they have
7105 at least one !DECL_IGNORED_P variable.
7107 2018-07-12 Richard Biener <rguenther@suse.de>
7110 * config/gnu-user.h (GNU_USER_TARGET_NO_PTHREADS_LIB_SPEC):
7111 Remove -mieee-fp handling.
7113 2018-07-12 Richard Biener <rguenther@suse.de>
7115 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Remove bogus
7116 left-over from last patch.
7118 2018-07-12 Jakub Jelinek <jakub@redhat.com>
7120 PR tree-optimization/86492
7121 * gimple-ssa-store-merging.c
7122 (imm_store_chain_info::coalesce_immediate_stores): Call
7123 check_no_overlap even for the merge_overlapping case. Formatting fix.
7125 2018-07-12 Richard Biener <rguenther@suse.de>
7128 * fold-const.c (fold_binary_op_with_conditional_arg): Do not
7129 move possibly trapping operations into the conditional.
7131 2018-07-12 Richard Biener <rguenther@suse.de>
7133 * tree-ssa-sccvn.c (mprts_hook_cnt): Remove.
7134 (vn_lookup_simplify_result): Remove recursion limit applied
7136 (vn_nary_build_or_lookup_1): Adjust.
7137 (try_to_simplify): Likewise.
7138 * gimple-match-head.c (gimple_resimplify1): Instead apply one
7140 (gimple_resimplify2): Likewise.
7141 (gimple_resimplify3): Likewise.
7142 (gimple_resimplify4): Likewise.
7144 2018-07-11 Jakub Jelinek <jakub@redhat.com>
7146 * config/i386/avx512bitalgintrin.h (_mm512_mask_bitshuffle_epi64_mask):
7147 Use __mmask64 type instead of __mmask8 for __M argument.
7148 * config/i386/avx512fintrin.h (_mm512_mask_xor_epi64,
7149 _mm512_maskz_xor_epi64): Use __mmask8 type instead of __mmask16 for
7151 (_mm512_mask_cmpneq_epi64_mask): Use __mmask8 type instead of
7152 __mmask16 for __M argument.
7153 (_mm512_maskz_insertf32x4, _mm512_maskz_inserti32x4,
7154 _mm512_mask_insertf32x4, _mm512_mask_inserti32x4): Cast last argument
7155 to __mmask16 instead of __mmask8.
7156 * config/i386/avx512vlintrin.h (_mm_mask_add_ps, _mm_maskz_add_ps,
7157 _mm256_mask_add_ps, _mm256_maskz_add_ps, _mm_mask_sub_ps,
7158 _mm_maskz_sub_ps, _mm256_mask_sub_ps, _mm256_maskz_sub_ps,
7159 _mm256_maskz_cvtepi32_ps, _mm_maskz_cvtepi32_ps): Use __mmask8 type
7160 instead of __mmask16 for __U argument.
7161 * config/i386/avx512vlbwintrin.h (_mm_mask_cmp_epi8_mask): Use
7162 __mmask16 instead of __mmask8 for __U argument.
7163 (_mm256_mask_cmp_epi8_mask): Use __mmask32 instead of __mmask16 for
7165 (_mm256_cmp_epi8_mask): Use __mmask32 return type instead of
7167 (_mm_mask_cmp_epu8_mask): Use __mmask16 instead of __mmask8 for __U
7169 (_mm256_mask_cmp_epu8_mask): Use __mmask32 instead of __mmask16 for
7171 (_mm256_cmp_epu8_mask): Use __mmask32 return type instead of
7173 (_mm_mask_cmp_epi16_mask): Cast last argument to __mmask8 instead
7175 (_mm256_mask_cvtepi8_epi16): Use __mmask16 instead of __mmask32 for
7177 (_mm_mask_cvtepi8_epi16): Use __mmask8 instead of __mmask32 for
7179 (_mm256_mask_cvtepu8_epi16): Use __mmask16 instead of __mmask32 for
7181 (_mm_mask_cvtepu8_epi16): Use __mmask8 instead of __mmask32 for
7183 (_mm256_mask_cmpneq_epu8_mask, _mm256_mask_cmplt_epu8_mask,
7184 _mm256_mask_cmpge_epu8_mask, _mm256_mask_cmple_epu8_mask): Change
7185 return type as well as __M argument type and all casts from __mmask8
7187 (_mm256_mask_cmpneq_epu16_mask, _mm256_mask_cmplt_epu16_mask,
7188 _mm256_mask_cmpge_epu16_mask, _mm256_mask_cmple_epu16_mask): Change
7189 return type as well as __M argument type and all casts from __mmask8
7191 (_mm256_mask_cmpneq_epi8_mask, _mm256_mask_cmplt_epi8_mask,
7192 _mm256_mask_cmpge_epi8_mask, _mm256_mask_cmple_epi8_mask): Change
7193 return type as well as __M argument type and all casts from __mmask8
7195 (_mm256_mask_cmpneq_epi16_mask, _mm256_mask_cmplt_epi16_mask,
7196 _mm256_mask_cmpge_epi16_mask, _mm256_mask_cmple_epi16_mask): Change
7197 return type as well as __M argument type and all casts from __mmask8
7199 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_shrdi_epi32,
7200 _mm_mask_shldi_epi32): Cast last argument to __mmask8 instead of
7203 2018-07-11 Grazvydas Ignotas <notasas@gmail.com>
7205 * config/i386/avx512bwintrin.h: (_mm512_mask_cmp_epi8_mask,
7206 _mm512_mask_cmp_epu8_mask): Use __mmask64 type instead of __mmask32
7209 2018-07-11 Paul Koning <ni1d@arrl.net>
7211 * doc/md.texi (define_subst): Document how multiple occurrences of
7212 the same argument in the replacement pattern are handled.
7214 2018-07-11 Paul Koning <ni1d@arrl.net>
7216 * doc/extend.texi (Common Variable Attributes): Move "mode" into
7218 (Common Type Attributes): Add "mode" attribute.
7220 2018-07-11 Jan Hubicka <hubicka@ucw.cz>
7222 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not
7223 stream DECL_ORIGINAL_TYPE.
7224 (DFS::DFS_write_tree_body): Drop hack handling local external decls.
7225 (hash_tree): Do not walk DECL_ORIGINAL_TYPE.
7226 * tree-streamer-in.c (lto_input_ts_decl_non_common_tree_pointers):
7227 Do not walk original type.
7228 * tree-streamer-out.c (streamer_write_chain): Drop hack handling
7230 (write_ts_decl_non_common_tree_pointers): Do not stream
7232 * tree.c (free_lang_data_in_decl): Clear DECL_ORIGINAL_TYPE.
7233 (find_decls_types_r): Do not walk DEC_ORIGINAL_TYPE.
7235 2018-07-11 Aldy Hernandez <aldyh@redhat.com>
7237 * tree-ssa-threadupdate.c (thread_through_all_blocks): Do not jump
7238 thread twice from the same starting edge.
7240 2018-07-11 Aldy Hernandez <aldyh@redhat.com>
7242 * vr-values.c (gimple_stmt_nonzero_p): Abstract common code to...
7243 * gimple.c (gimple_call_nonnull_result_p): ...here...
7244 (gimple_call_nonnull_arg): ...and here.
7245 * gimple.h (gimple_call_nonnull_result_p): New.
7246 (gimple_call_nonnull_arg): New.
7248 2018-07-11 Richard Earnshaw <rearnsha@arm.com>
7250 * config/arm/arm-cpus.in: Move information from fpu field of each
7251 cpu definition to the isa field.
7252 * config/arm/parsecpu.awk (fpu): Delete match rule.
7253 (gen_comm_data): Don't add bits from the CPU's FPU entry.
7255 2018-07-11 Richard Biener <rguenther@suse.de>
7258 * dwarf2out.c (init_sections_and_labels): Use
7259 output_asm_line_debug_info consistently.
7260 (dwarf2out_early_finish): Likewise.
7261 (dwarf2out_finish): Remove DW_AT_stmt_list from early generated
7264 2018-07-11 Richard Biener <rguenther@suse.de>
7266 * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely_1):
7267 Rework father_bb setting in a way to avoid propagating constants
7268 multiple times on a loop body.
7270 2018-07-10 Mark Wielaard <mark@klomp.org>
7273 * dwarf2out.c (output_macinfo_op): Fix dwarf_FORM typo in gcc_assert.
7275 2018-07-10 Richard Biener <rguenther@suse.de>
7277 * hash-map.h (hash_map::iterator::operator*): Return
7278 references to key and value.
7280 2018-07-10 Jakub Jelinek <jakub@redhat.com>
7283 * gimplify.c (find_combined_omp_for): Add DATA argument, in addition
7284 to finding the inner OMP_FOR/OMP_SIMD stmt find non-trivial wrappers,
7285 BLOCKs with BLOCK_VARs, OMP_PARALLEL in between, OMP_FOR in between.
7286 (gimplify_omp_for): For composite loops, move outer
7287 OMP_{DISTRIBUTE,TASKLOOP,FOR,PARALLEL} right around innermost
7288 OMP_FOR/OMP_SIMD if there are any non-trivial wrappers. For class
7289 iterators add any needed clauses. Allow OMP_FOR_ORIG_DECLS to contain
7290 TREE_LIST for both the original class iterator and the "last" helper
7291 var. Gimplify OMP_FOR_PRE_BODY before the outermost composite
7292 loop, remember has_decl_expr from outer composite loops for the
7293 innermost OMP_SIMD in TREE_PRIVATE bit on OMP_FOR_INIT.
7295 2018-07-09 Martin Sebor <msebor@redhat.com>
7299 * builtins.c (c_strlen): Avoid out-of-bounds warnings when
7300 accessing implicitly initialized array elements.
7301 * expr.c (string_constant): Handle string initializers of
7302 character arrays within aggregates.
7303 * gimple-fold.c (fold_array_ctor_reference): Add argument.
7304 Store element offset. As a special case, handle zero size.
7305 (fold_nonarray_ctor_reference): Same.
7306 (fold_ctor_reference): Add argument. Store subobject offset.
7307 * gimple-fold.h (fold_ctor_reference): Add argument.
7309 2018-07-09 Paul Koning <ni1d@arrl.net>
7311 * config/pdp11/pdp11.c (pdp11_addr_cost): New function.
7312 (pdp11_insn_cost): New function.
7313 (pdp11_md_asm_adjust): New function.
7314 (TARGET_INVALID_WITHIN_DOLOOP): Define.
7315 (pdp11_rtx_costs): Update to match machine better.
7316 (output_addr_const_pdp11): Correct format mismatch warnings.
7317 * config/pdp11/pdp11.h (SLOW_BYTE_ACCESS): Correct definition.
7318 * config/pdp11/pdp11.md: General change to add base_cost and/or
7319 length attributes for use by new pdp11_insn_cost function.
7320 (MIN_BRANCH): Correct definition.
7322 (doloop_end): Use standard pattern name for looping pattern.
7323 (doloop_end_nocc): New.
7324 (movsf): Add another constraint alternative.
7325 (zero_extendqihi2): Add constraint alternatives for not in place
7327 (zero_extendhisi2): Remove.
7328 (shift patterns): Add CC handling variants.
7332 (define_peephole2): New peephole to recognize mov that sets CC for
7335 2018-07-09 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
7337 * config/sparc/sparc.c (sparc_fold_builtin) <SPARC_BUILTIN_PDIST,
7338 SPARC_BUILTIN_PDISTN>: Adapt for signature change of wi::neg,
7341 2018-07-09 Jakub Jelinek <jakub@redhat.com>
7344 * real.c (real_nextafter): Return true if result is denormal.
7346 2018-07-09 Martin Liska <mliska@suse.cz>
7348 * common.opt: Add back wrongly removed attribute.
7350 2018-07-09 Richard Biener <rguenther@suse.de>
7353 * dwarf2out.c (gen_block_die): For an early generated DIE
7354 always output high/low PC attributes.
7356 2018-07-09 Tom de Vries <tdevries@suse.de>
7358 * cfgexpand.c (expand_debug_source_expr): Handle VAR_DECL.
7359 * tree-inline.c (remap_ssa_name): Handle default def ssa_name mapping
7360 onto VAR_DECL with abstract origin.
7362 2018-07-07 Jim Wilson <jimw@sifive.com>
7364 * config/riscv/riscv.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): New.
7366 2018-07-07 Tom de Vries <tdevries@suse.de>
7368 * tree-dfa.c (dump_enumerated_decls): Handle cfun->cfg == NULL.
7370 2018-07-07 Aldy Hernandez <aldyh@redhat.com>
7372 * tree-vrp.c (vrp_int_const_binop): Change overflow type to
7374 (combine_bound): Use wide-int overflow calculation instead of
7376 * calls.c (maybe_warn_alloc_args_overflow): Change overflow type to
7378 * fold-const.c (int_const_binop_2): Same.
7379 (extract_muldiv_1): Same.
7380 (fold_div_compare): Same.
7381 (fold_abs_const): Same.
7383 * poly-int.h (add): Same.
7387 * predict.c (predict_iv_comparison): Same.
7388 * profile-count.c (slow_safe_scale_64bit): Same.
7389 * simplify-rtx.c (simplify_const_binary_operation): Same.
7390 * tree-chrec.c (tree_fold_binomial): Same.
7391 * tree-data-ref.c (split_constant_offset_1): Same.
7392 * tree-if-conv.c (idx_within_array_bound): Same.
7393 * tree-scalar-evolution.c (iv_can_overflow_p): Same.
7394 * tree-ssa-phiopt.c (minmax_replacement): Same.
7395 * tree-vect-loop.c (is_nonwrapping_integer_induction): Same.
7396 * tree-vect-stmts.c (vect_truncate_gather_scatter_offset): Same.
7397 * vr-values.c (vr_values::adjust_range_with_scev): Same.
7398 * wide-int.cc (wi::add_large): Same.
7399 (wi::mul_internal): Same.
7400 (wi::sub_large): Same.
7401 (wi::divmod_internal): Same.
7402 * wide-int.h: Change overflow type to overflow_type for neg, add,
7403 mul, smul, umul, div_trunc, div_floor, div_ceil, div_round,
7404 mod_trunc, mod_ceil, mod_round, add_large, sub_large,
7405 mul_internal, divmod_internal.
7406 (overflow_type): New enum.
7407 (accumulate_overflow): New.
7409 2018-07-06 Kugan Vivekanandarajah <kugan.vivekanandarajah@linaro.org>
7411 * tree-ssa-phiopt.c (cond_removal_in_popcount_pattern): New.
7412 (tree_ssa_phiopt_worker): Call cond_removal_in_popcount_pattern.
7414 2018-07-06 Kugan Vivekanandarajah <kuganv@linaro.org>
7416 * tree-ssa-loop-niter.c (number_of_iterations_popcount): If popcount
7417 argument is checked for zero before entering loop, avoid checking again.
7419 2018-07-06 Kugan Vivekanandarajah <kuganv@linaro.org>
7421 * gimplify.h (generic_expr_could_trap_p): Set as global function.
7422 * gimplify.h (generic_expr_could_trap_p): Likwise.
7423 * tree-scalar-evolution.c (expression_expensive_p): Handle COND_EXPR.
7425 2018-07-06 Jakub Jelinek <jakub@redhat.com>
7427 PR tree-optimization/86401
7428 * fold-const.c (fold_binary_loc) <case BIT_AND_EXPR>: Move the
7429 ((A & N) + B) & M -> (A + B) & M etc. optimization into ...
7430 (fold_bit_and_mask): ... here. New helper function for match.pd.
7431 * fold-const.h (fold_bit_and_mask): Declare.
7432 * match.pd (((A & N) + B) & M -> (A + B) & M): New optimization.
7434 2018-07-06 Peter Bergner <bergner@linux.ibm.com>
7437 * target.def (translate_mode_attribute): New hook.
7438 * targhooks.h (default_translate_mode_attribute): Declare.
7439 * targhooks.c (default_translate_mode_attribute): New function.
7440 * doc/tm.texi.in (TARGET_TRANSLATE_MODE_ATTRIBUTE): New hook.
7441 * doc/tm.texi: Regenerate.
7442 * config/rs6000/rs6000.c (TARGET_TRANSLATE_MODE_ATTRIBUTE): Define.
7443 (rs6000_translate_mode_attribute): New function.
7445 2018-07-06 Paul Koning <ni1d@arrl.net>
7447 * doc/md.texi (define_split): Document DONE and FAIL.
7448 (define_peephole2): Ditto.
7450 2018-07-05 Jeff Law <law@redhat.com>
7452 PR tree-optimization/86010
7453 * tree-ssa-dse.c (compute_trims): More aggressively trim at
7454 both the head and tail of mem* and str* calls.
7456 2018-07-05 Jim Wilson <jimw@sifive.com>
7458 * config.gcc (riscv*-*-*): When setting xlen, handle riscv-*.
7460 2018-07-05 Indu Bhagat <indu.bhagat@oracle.com>
7462 * config/aarch64/aarch64-simd.md: correct flags text for
7463 MIN_EXPR replacement.
7465 2018-07-05 James Clarke <jrtc27@jrtc27.com>
7467 * configure: Regenerated.
7469 2018-07-05 Carl Love <cel@us.ibm.com>
7471 * config/rs6000/rs6000-c.c: Map ALTIVEC_BUILTIN_VEC_UNPACKH for
7472 float argument to VSX_BUILTIN_DOUBLEH_V4SF.
7473 Map ALTIVEC_BUILTIN_VEC_UNPACKL for float argument to
7474 VSX_BUILTIN_DOUBLEL_V4SF.
7476 2018-07-05 Martin Sebor <msebor@redhat.com>
7479 * tree-ssa-strlen.c (maybe_set_strlen_range): Use type size rather
7480 than its domain to compute its the upper bound of a char array.
7482 2018-07-05 Nathan Sidwell <nathan@acm.org>
7484 Replace NO_IMPLICIT_EXTERN_C with SYSTEM_IMPLICIT_EXTERN_C.
7485 * doc/cpp.texi: Update comment.
7486 * doc/tm.texi: Rebuilt.
7487 * doc/tm.texi.in (NO_IMPLICIT_EXTERN_C): Replace with ...
7488 (SYSTEM_IMPLICIT_EXTERN_C): ... this, opposite sense.
7489 * doc/extend.texi (Backwards Compatibility): Clarify it is system
7490 headers affected by extern "C".
7491 * system.h: Poison NO_IMPLICIT_EXTERN_C.
7492 * config/alpha/alpha.h, config/arm/uclinux-elf.h,
7493 config/bfin/elf.h, config/cris/cris.h, config/darwin.h,
7494 config/dragonfly.h, config/freebsd.h, config/gnu-user.h,
7495 config/i386/cygming.h, config/i386/djgpp.h, config/i386/nto.h,
7496 config/ia64/hpux.h, config/lm32/lm32.h, config/lm32/uclinux-elf.h,
7497 config/lynx.h, config/mips/elf.h, config/mmix/mmix.h,
7498 config/netbsd.h, config/pa/pa-hpux.h, config/powerpcspe/sysv4.h,
7499 config/riscv/elf.h, config/rs6000/sysv4.h, config/rtems.h,
7500 config/s390/tpf.h, config/sh/newlib.h, config/sol2.h,
7501 config/sparc/openbsd64.h, config/sparc/sp-elf.h,
7502 config/sparc/sp64-elf.h, config/spu/spu.h,
7503 config/stormy16/stormy16.h, config/v850/v850.h,
7504 config/visium/visium.h, config/vx-common.h, config/xtensa/elf.h: Don't
7505 define NO_IMPLICIT_EXTERN_C.
7506 * config/rs6000/aix.h: Set SYSTEM_IMPLICIT_EXTERN_C.
7508 2018-07-05 Tamar Christina <tamar.christina@arm.com>
7511 * config/arm/arm.c (arm_can_change_mode_class): Use GET_MODE_UNIT_SIZE
7512 instead of GET_MODE_SIZE when comparing Units.
7514 2018-07-05 Tamar Christina <tamar.christina@arm.com>
7517 * rtlanal.c (set_noop_p): Constrain on mode change,
7518 include hard-reg-set.h
7520 2018-07-05 Tamar Christina <tamar.christina@arm.com>
7522 * config/aarch64/aarch64.c (aarch64_expand_movmem): Fix mode size.
7524 2018-07-05 Jakub Jelinek <jakub@redhat.com>
7527 2018-07-04 Maxim Ostapenko <m.ostapenko@samsung.com>
7530 * config/gnu-user.h (LIBASAN_EARLY_SPEC): Pass -lstdc++ for static
7532 * gcc.c: Do not pass LIBUBSAN_SPEC if ASan is enabled with UBSan.
7534 2018-07-04 Maxim Ostapenko <m.ostapenko@samsung.com>
7537 * config/gnu-user.h (LIBASAN_EARLY_SPEC): Pass -lstdc++ for static
7539 * gcc.c: Do not pass LIBUBSAN_SPEC if ASan is enabled with UBSan.
7541 2018-07-04 Eric Botcazou <ebotcazou@adacore.com>
7544 * expmed.c (choose_multiplier): Fix incorrect comparison with mask.
7546 2018-07-04 Aldy Hernandez <aldyh@redhat.com>
7548 * tree-vrp.c (extract_range_from_binary_expr_1): Initialize
7549 neg_*_op* variables.
7551 2018-07-04 Martin Liska <mliska@suse.cz>
7553 * tree-switch-conversion.c: Define
7554 max_ratio_for_speed and max_ratio_for_size constants.
7556 2018-07-04 Denys Vlasenko <dvlasenk@redhat.com>
7557 Martin Liska <mliska@suse.cz>
7562 * common.opt: Rename align options with 'str_' prefix.
7563 * common/config/i386/i386-common.c (set_malign_value): New
7565 (ix86_handle_option): Use it to set -falign-* options/
7566 * config/aarch64/aarch64-protos.h (struct tune_params): Change
7567 type from int to string.
7568 * config/aarch64/aarch64.c: Update default values from int
7570 * config/alpha/alpha.c (alpha_override_options_after_change):
7572 * config/arm/arm.c (arm_override_options_after_change_1): Likewise.
7573 * config/i386/dragonfly.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
7574 max skip conditionally.
7575 * config/i386/freebsd.h (SUBALIGN_LOG): New.
7576 (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
7577 max skip conditionally.
7578 * config/i386/gas.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
7579 max skip conditionally.
7580 * config/i386/gnu-user.h (SUBALIGN_LOG): New.
7581 (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
7582 max skip conditionally.
7583 * config/i386/i386.c (struct ptt): Change type from int to
7585 (ix86_default_align): Set default values.
7586 * config/i386/i386.h (ASM_OUTPUT_MAX_SKIP_PAD): Print
7587 max skip conditionally.
7588 * config/i386/iamcu.h (SUBALIGN_LOG): New.
7589 (ASM_OUTPUT_MAX_SKIP_ALIGN):
7590 * config/i386/lynx.h (ASM_OUTPUT_MAX_SKIP_ALIGN):
7591 * config/i386/netbsd-elf.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
7592 max skip conditionally.
7593 * config/i386/openbsdelf.h (SUBALIGN_LOG): New.
7594 (ASM_OUTPUT_MAX_SKIP_ALIGN) Print max skip conditionally.:
7595 * config/i386/x86-64.h (SUBALIGN_LOG): New.
7596 (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
7597 max skip conditionally.
7598 (ASM_OUTPUT_MAX_SKIP_PAD): Likewise.
7599 * config/ia64/ia64.c (ia64_option_override): Set default values
7600 for alignment options.
7601 * config/m68k/m68k.c: Handle new str_align_* options.
7602 * config/mips/mips.c (mips_set_compression_mode): Change
7604 (mips_option_override): Set default values for options.
7605 * config/powerpcspe/powerpcspe.c (rs6000_option_override_internal):
7607 * config/rs6000/rs6000.c (rs6000_option_override_internal):
7609 * config/rx/rx.c (rx_option_override): Likewise.
7610 * config/rx/rx.h (JUMP_ALIGN): Use align_jumps_log.
7611 (LABEL_ALIGN): Use align_labels_log.
7612 (LOOP_ALIGN): Use align_loops_align.
7613 * config/s390/s390.c (s390_asm_output_function_label): Use new
7615 * config/sh/sh.c (sh_override_options_after_change):
7616 Change type of constants.
7617 * config/spu/spu.c (spu_sched_init): Likewise.
7618 * config/sparc/sparc.c (sparc_option_override): Set default
7620 * config/visium/visium.c (visium_option_override): Likewise.
7621 * config/visium/visium.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Do not
7622 emit p2align format with last argument if it's not needed.
7623 * doc/invoke.texi: Document extended format of -falign-*.
7624 * final.c: Use align_labels alignment.
7625 * flags.h (struct target_flag_state): Change type to use
7627 (struct align_flags_tuple): New.
7628 (struct align_flags): Likewise.
7629 (align_loops_log): Redefine macro to use new types.
7630 (align_loops_max_skip): Redefine macro to use new types.
7631 (align_jumps_log): Redefine macro to use new types.
7632 (align_jumps_max_skip): Redefine macro to use new types.
7633 (align_labels_log): Redefine macro to use new types.
7634 (align_labels_max_skip): Redefine macro to use new types.
7635 (align_functions_log): Redefine macro to use new types.
7636 (align_loops): Redefine macro to use new types.
7637 (align_jumps): Redefine macro to use new types.
7638 (align_labels): Redefine macro to use new types.
7639 (align_functions): Redefine macro to use new types.
7640 (align_functions_max_skip): Redefine macro to use new types.
7641 (align_loops_value): New macro.
7642 (align_jumps_value): New macro.
7643 (align_labels_value): New macro.
7644 (align_functions_value): New macro.
7645 * function.c (invoke_set_current_function_hook): Propagate
7646 alignment values from flags to global variables default in
7648 * ipa-icf.c (sem_function::equals_wpa): Use
7649 cl_optimization_option_eq instead of memcmp.
7650 * lto-streamer.h (cl_optimization_stream_out): Support streaming
7652 (cl_optimization_stream_in): Likewise.
7653 * optc-save-gen.awk: Support strings in cl_optimization.
7654 * opth-gen.awk: Likewise.
7655 * opts.c (finish_options): Remove error checking of invalid
7657 (MAX_CODE_ALIGN): Remove.
7658 (MAX_CODE_ALIGN_VALUE): Likewise.
7659 (parse_and_check_align_values): New function.
7660 (check_alignment_argument): Likewise.
7661 (common_handle_option): Use check_alignment_argument.
7662 * opts.h (parse_and_check_align_values): Declare.
7663 * toplev.c (init_alignments): Remove.
7664 (read_log_maxskip): New.
7665 (parse_N_M): Likewise.
7666 (parse_alignment_opts): Likewise.
7667 (backend_init_target): Remove usage of init_alignments.
7668 * toplev.h (parse_alignment_opts): Declare.
7669 * tree-streamer-in.c (streamer_read_tree_bitfields): Add new
7671 * tree-streamer-out.c (streamer_write_tree_bitfields): Likewise.
7672 * tree.c (cl_option_hasher::equal): New.
7673 * varasm.c: Use new global macros.
7675 2018-07-04 Denys Vlasenko <dvlasenk@redhat.com>
7677 * config/i386/dragonfly.h: (ASM_OUTPUT_MAX_SKIP_ALIGN):
7678 Use a simpler align directive also if MAXSKIP = ALIGN-1.
7679 * config/i386/gas.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
7680 * config/i386/lynx.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
7681 * config/i386/netbsd-elf.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
7682 * config/i386/i386.h (ASM_OUTPUT_MAX_SKIP_PAD): Likewise.
7683 * config/i386/freebsd.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Remove "If N
7684 is large, do at least 8 byte alignment" code. Add SUBALIGN_LOG
7685 define. Use a simpler align directive also if MAXSKIP = ALIGN-1.
7686 * config/i386/gnu-user.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
7687 * config/i386/iamcu.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
7688 * config/i386/openbsdelf.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
7689 * config/i386/x86-64.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
7691 2018-07-04 Martin Liska <mliska@suse.cz>
7692 Jonathan Wakely <jwakely@redhat.com>
7694 * coverage.c: Use correct type.
7695 * doc/invoke.texi: Language correction.
7697 2018-07-03 H.J. Lu <hongjiu.lu@intel.com>
7700 * config/i386/i386.c (rest_of_insert_endbranch): Also generate
7701 ENDBRANCH for non-tail call which may return via indirect branch.
7702 * doc/extend.texi: Document indirect_return attribute.
7704 2018-07-03 Martin Sebor <msebor@redhat.com>
7706 PR tree-optimization/86274
7707 * gimple-ssa-sprintf.c (fmtresult::type_max_digits): Verify
7709 (format_floating): Correct handling of infinities and NaNs.
7711 2018-07-03 Martin Sebor <msebor@redhat.com>
7713 * print-tree.c (print_real_cst): New function.
7714 (print_node_brief): Call it.
7715 (print_node): Ditto.
7717 2018-07-03 Jeff Law <law@redhat.com>
7719 * config/h8300/h8300.md (logical<mode>3_sn, logical<mode>3): Merge
7720 into a single pattern.
7722 * config/h8300/h8300.md (ors code_iterator): New.
7723 (bsetqi_msx, bnotqi_msx patterns and splitters): Consolidate into
7724 a single pattern and single splitter.
7725 (bsethi_msx, bnothi_msx patterns): Consolidate into a single pattern.
7726 (iorqi3_1, xorqi3_1): Likewise.
7727 (iorqi3, xorqi3 expanders): Similarly.
7729 * config/h8300/h8300.md (movmd_internal_normal): Consolidated with
7730 (movmd_internal) into a single pattern using the P mode iterator.
7731 (movmd splitters): Similarly.
7732 (stpcpy_internal_normal, stpcpy_internal): Similarly for thes patterns.
7733 (movsd splitters): Similarly.
7735 * config/h8300/h8300.c (h8300_insn_length_from_table): Consolidate
7736 ADDB, ADDW and ADDL into a single ADD attribute which selects the
7737 right table based on the size of the operand.
7738 * config/h8300/h8300.md (length_table): Corresponding changes. All
7739 references to "addb", "addw" and "addl" changed to "add".
7740 (btst patterns): Merge two variants into a single pattern.
7741 (tstqi, tsthi): Likewise.
7742 (addhi3_incdec, addsi3_incdec): Likewise.
7743 (subhi3_h8300hs, subsi3_h8300hs): Likewise.
7744 (mulhi3, mulsi3): Likewise.
7745 (udivhi3, udivsi3): Likewise.
7746 (divhi3, divsi3): Likewise.
7747 (andorqi3, andorhi3, andorsi3): Likewise.
7749 2018-07-03 Uros Bizjak <ubizjak@gmail.com>
7752 * config/i386/sse.md (uavg<mode>3_ceil): New expander.
7753 (<sse2_avx2>_uavg<mode>3<mask_name>): Simplify expander.
7755 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
7757 PR tree-optimization/85694
7758 * config/aarch64/iterators.md (HADD, RHADD): New int iterators.
7759 (u): Handle UNSPEC_SHADD, UNSPEC_UHADD, UNSPEC_SRHADD and
7761 * config/aarch64/aarch64-simd.md (<u>avg<mode>3_floor)
7762 (<u>avg<mode>3_ceil): New patterns.
7764 2018-07-03 David Malcolm <dmalcolm@redhat.com>
7766 * gcc.dg/vect/slp-perm-1.c: Remove "note: " prefix from
7767 scan-tree-dump directive.
7768 * gcc.dg/vect/slp-perm-2.c: Likewise.
7769 * gcc.dg/vect/slp-perm-3.c: Likewise.
7770 * gcc.dg/vect/slp-perm-5.c: Likewise.
7771 * gcc.dg/vect/slp-perm-6.c: Likewise.
7772 * gcc.dg/vect/slp-perm-7.c: Likewise.
7773 * gcc.dg/vect/slp-perm-8.c: Likewise.
7775 2018-07-03 Marek Polacek <polacek@redhat.com>
7778 * gimple-fold.c (size_must_be_zero_p): Check the type of the size.
7780 2018-07-03 Richard Biener <rguenther@suse.de>
7783 * tree-ssa-structalias.c (find_func_clobbers): Properly
7784 handle indirect calls.
7786 2018-07-03 Jeff Law <law@redhat.com>
7788 * config/h8300/h8300.md (HSI, QHSI, QHSIF): New mode iterators.
7789 (shifts): New code iterator.
7790 (movqi, movhi, movsi, movsf expanders): Consolidate into a single
7791 expander. Fix HImode handling on H8/SX.
7792 (addqi3, addhi3, addsi3 expanders): Consolidate into a single expander.
7793 (subqi3, subhi3, subsi3 expanders): Likewise.
7794 (andqi3, andhi3, andsi3 expanders): Likewise.
7795 (iorqi3, iorhi3, iorsi3 expanders): Likewise.
7796 (xorqi3, xorhi3, xorsi3 expanders): Likewise.
7797 (negqi2, neghi2, negsi2, negsf2 expanders): Likewise.
7798 (one_cmplqi2, one_cmplhi2, one_cmplsi2): Likewise.
7799 (zero_extendqihi2, zero_extendqisi2): Likewise.
7800 (extendqihi2, extendqisi2): Likewise.
7801 (rotlqi3, rotlhi3, rotlsi3): Likewise.
7802 (neghi2_h8300, negsi2_h8300): Likewise for these patterns.
7803 (rotlqi3_1, rotlhi3_1): Likewise.
7804 (logicalhi3_sn, logicalsi3_sn): Likewise.
7805 (logicalhi3, logicalsi3): Likewise.
7807 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
7809 * tree-vect-patterns.c (vect_recog_rotate_pattern)
7810 (vect_recog_vector_vector_shift_pattern, vect_recog_divmod_pattern)
7811 (vect_recog_mixed_size_cond_pattern, adjust_bool_pattern_cast)
7812 (adjust_bool_pattern, vect_recog_bool_pattern): Pass the vector
7813 type to append_pattern_def_seq instead of creating a stmt_vec_info
7815 (build_mask_conversion): Likewise. Remove vinfo argument.
7816 (vect_add_conversion_to_patterm): Likewise, renaming to...
7817 (vect_add_conversion_to_pattern): ...this.
7818 (vect_recog_mask_conversion_pattern): Update call to
7819 build_mask_conversion. Pass the vector type to
7820 append_pattern_def_seq here too.
7821 (vect_recog_gather_scatter_pattern): Update call to
7822 vect_add_conversion_to_pattern.
7824 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
7826 * tree-vect-patterns.c (new_pattern_def_seq): Delete.
7827 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern)
7828 (vect_recog_widen_op_pattern, vect_recog_over_widening_pattern)
7829 (vect_recog_rotate_pattern, vect_synth_mult_by_constant): Don't set
7830 STMT_VINFO_PATTERN_DEF_SEQ to null here.
7831 (vect_recog_pow_pattern, vect_recog_vector_vector_shift_pattern)
7832 (vect_recog_mixed_size_cond_pattern, vect_recog_bool_pattern): Use
7833 append_pattern_def_seq instead of new_pattern_def_seq.
7834 (vect_recog_divmod_pattern): Do both of the above.
7835 (vect_pattern_recog_1): Assert that STMT_VINO_PATTERN_DEF_SEQ
7838 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
7840 * tree-vect-patterns.c (vect_recog_dot_prod_pattern):
7841 (vect_recog_sad_pattern, vect_recog_widen_op_pattern)
7842 (vect_recog_widen_mult_pattern, vect_recog_pow_pattern):
7843 (vect_recog_widen_sum_pattern, vect_recog_over_widening_pattern)
7844 (vect_recog_average_pattern, vect_recog_cast_forwprop_pattern)
7845 (vect_recog_widen_shift_pattern, vect_recog_rotate_pattern)
7846 (vect_recog_vector_vector_shift_pattern, vect_synth_mult_by_constant)
7847 (vect_recog_mult_pattern, vect_recog_divmod_pattern)
7848 (vect_recog_mixed_size_cond_pattern, vect_recog_bool_pattern)
7849 (vect_recog_mask_conversion_pattern): Replace vec<gimple *>
7850 parameter with a single stmt_vec_info.
7851 (vect_recog_func_ptr): Likewise.
7852 (vect_recog_gather_scatter_pattern): Likewise, folding in...
7853 (vect_try_gather_scatter_pattern): ...this.
7854 (vect_pattern_recog_1): Remove stmts_to_replace and just pass
7855 the stmt_vec_info of the statement to be matched. Don't clear
7856 STMT_VINFO_RELATED_STMT.
7857 (vect_pattern_recog): Update call accordingly.
7859 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
7861 PR tree-optimization/85694
7862 * doc/md.texi (avgM3_floor, uavgM3_floor, avgM3_ceil)
7863 (uavgM3_ceil): Document new optabs.
7864 * doc/sourcebuild.texi (vect_avg_qi): Document new target selector.
7865 * internal-fn.def (IFN_AVG_FLOOR, IFN_AVG_CEIL): New internal
7867 * optabs.def (savg_floor_optab, uavg_floor_optab, savg_ceil_optab)
7868 (savg_ceil_optab): New optabs.
7869 * tree-vect-patterns.c (vect_recog_average_pattern): New function.
7870 (vect_vect_recog_func_ptrs): Add it.
7871 * tree-vect-stmts.c (vectorizable_call): Get the type of the zero
7872 constant directly from the associated lhs.
7874 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
7876 * tree-vect-patterns.c (vect_split_statement): New function.
7877 (vect_convert_input): Use it to try to split an existing cast.
7879 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
7881 * poly-int.h (print_hex): New function.
7882 * dumpfile.h (dump_dec, dump_hex): Declare.
7883 * dumpfile.c (dump_dec, dump_hex): New poly_wide_int functions.
7884 * tree-vectorizer.h (_stmt_vec_info): Add min_output_precision,
7885 min_input_precision, operation_precision and operation_sign.
7886 * tree-vect-patterns.c (vect_get_range_info): New function.
7887 (vect_same_loop_or_bb_p, vect_single_imm_use)
7888 (vect_operation_fits_smaller_type): Delete.
7889 (vect_look_through_possible_promotion): Add an optional
7890 single_use_p parameter.
7891 (vect_recog_over_widening_pattern): Rewrite to use new
7892 stmt_vec_info infomration. Handle one operation at a time.
7893 (vect_recog_cast_forwprop_pattern, vect_narrowable_type_p)
7894 (vect_truncatable_operation_p, vect_set_operation_type)
7895 (vect_set_min_input_precision): New functions.
7896 (vect_determine_min_output_precision_1): Likewise.
7897 (vect_determine_min_output_precision): Likewise.
7898 (vect_determine_precisions_from_range): Likewise.
7899 (vect_determine_precisions_from_users): Likewise.
7900 (vect_determine_stmt_precisions, vect_determine_precisions): Likewise.
7901 (vect_vect_recog_func_ptrs): Put over_widening first.
7903 (vect_pattern_recog): Call vect_determine_precisions.
7905 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
7907 * tree-vect-patterns.c (vect_mark_pattern_stmts): Remove pattern
7908 statements that have been replaced by further pattern statements.
7909 (vect_pattern_recog_1): Clear STMT_VINFO_PATTERN_DEF_SEQ on failure.
7911 2018-07-03 Richard Biener <rguenther@suse.de>
7913 * tree-vect-stmts.c (vect_is_simple_use): Consolidate dumping,
7914 always set *dt. Dump vectype in vectype overload.
7915 * dumpfile.h (dump_gimple_expr): New function.
7916 (dump_gimple_expr_loc): Likewise.
7917 * dumpfile.c (dump_gimple_expr): New function.
7918 (dump_gimple_expr_loc): Likewise.
7920 2018-07-02 Jeff Law <law@redhat.com>
7922 * config/h8300/h8300.md (movqi_h8300, movqi_h8300hs): Consolidate
7923 the H8/300, H8/300H and H8/S variants into a single pattern.
7924 (movhi_h8300, movqi_h8300hs): Similarly.
7925 (pushqi_h8300hs, pushhi_h8300hs): Consolidate into a single pattern.
7926 (QHI mode iterator): New.
7928 * config/h8300/h8300.md: Remove trailing whitespace.
7930 2018-07-02 Jim Wilson <jimw@sifive.com>
7932 * config/riscv/riscv.c (riscv_expand_epilogue): Use emit_jump_insn
7933 instead of emit_insn for interrupt returns.
7934 * config/riscv/riscv.md (riscv_met): Add (return) to rtl.
7935 (riscv_sret, riscv_uret): Likewise.
7937 2018-07-02 David Malcolm <dmalcolm@redhat.com>
7939 * pretty-print.c (selftest::test_pp_format): Move save and restore
7940 of quotes to class auto_fix_quotes, and add an instance.
7941 * selftest.c: Include "intl.h".
7942 (selftest::auto_fix_quotes::auto_fix_quotes): New ctor.
7943 (selftest::auto_fix_quotes::~auto_fix_quotes): New dtor.
7944 * selftest.h (selftest::auto_fix_quotes): New class.
7946 2018-07-02 Richard Henderson <richard.henderson@linaro.org>
7948 * config/aarch64/aarch64-protos.h, config/aarch64/aarch64.c
7949 (aarch64_sve_prepare_conditional_op): Remove.
7950 * config/aarch64/aarch64-sve.md (cond_<SVE_INT_BINARY><SVE_I>):
7951 Allow aarch64_simd_reg_or_zero as select operand; remove
7952 the aarch64_sve_prepare_conditional_op call.
7953 (cond_<SVE_INT_BINARY_SD><SVE_SDI>): Likewise.
7954 (cond_<SVE_COND_FP_BINARY><SVE_F>): Likewise.
7955 (*cond_<SVE_INT_BINARY><SVE_I>_z): New pattern.
7956 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_z): New pattern.
7957 (*cond_<SVE_COND_FP_BINARY><SVE_F>_z): New pattern.
7958 (*cond_<SVE_INT_BINARY><SVE_I>_any): New pattern.
7959 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_any): New pattern.
7960 (*cond_<SVE_COND_FP_BINARY><SVE_F>_any): New pattern
7961 and a splitters to match all of the *_any patterns.
7962 * config/aarch64/predicates.md (aarch64_sve_any_binary_operator): New.
7964 * config/aarch64/iterators.md (SVE_INT_BINARY_REV): Remove.
7965 (SVE_COND_FP_BINARY_REV): Remove.
7966 (sve_int_op_rev, sve_fp_op_rev): New.
7967 * config/aarch64/aarch64-sve.md (*cond_<SVE_INT_BINARY><SVE_I>_0): New.
7968 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_0): New.
7969 (*cond_<SVE_COND_FP_BINARY><SVE_F>_0): New.
7970 (*cond_<SVE_INT_BINARY><SVE_I>_2): Rename, add movprfx alternative.
7971 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_2): Similarly.
7972 (*cond_<SVE_COND_FP_BINARY><SVE_F>_2): Similarly.
7973 (*cond_<SVE_INT_BINARY><SVE_I>_3): Similarly; use sve_int_op_rev.
7974 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_3): Similarly.
7975 (*cond_<SVE_COND_FP_BINARY><SVE_F>_3): Similarly; use sve_fp_op_rev.
7977 * config/aarch64/aarch64-sve.md (cond_<SVE_COND_FP_BINARY><SVE_F>):
7978 Remove match_dup 1 from the inner unspec.
7979 (*cond_<SVE_COND_FP_BINARY><SVE_F>): Likewise.
7981 * config/aarch64/aarch64.md (movprfx): New attr.
7982 (length): Default movprfx to 8.
7983 * config/aarch64/aarch64-sve.md (*mul<SVE_I>3): Add movprfx alt.
7984 (*madd<SVE_I>, *msub<SVE_I): Likewise.
7985 (*<su>mul<SVE_I>3_highpart): Likewise.
7986 (*<SVE_INT_BINARY_SD><SVE_SDI>3): Likewise.
7987 (*v<ASHIFT><SVE_I>3): Likewise.
7988 (*<su><MAXMIN><SVE_I>3): Likewise.
7989 (*<su><MAXMIN><SVE_F>3): Likewise.
7990 (*fma<SVE_F>4, *fnma<SVE_F>4): Likewise.
7991 (*fms<SVE_F>4, *fnms<SVE_F>4): Likewise.
7992 (*div<SVE_F>4): Likewise.
7994 2018-07-02 Richard Sandiford <richard.sandiford@arm.com>
7996 * tree-vect-patterns.c (vect_recog_widen_shift_pattern): Fix typo
7999 2018-07-02 Richard Biener <rguenther@suse.de>
8001 PR tree-optimization/86363
8002 * tree-ssa-sccvn.c (vn_reference_lookup_3): Check the
8003 memset argument refers to a non-variable address.
8005 2018-07-02 Aldy Hernandez <aldyh@redhat.com>
8007 * tree-vrp.c (extract_range_from_binary_expr_1): Abstract a lot of the
8008 {PLUS,MINUS}_EXPR code to...
8009 (adjust_symbolic_bound): ...here,
8010 (combine_bound): ...here,
8011 (set_value_range_with_overflow): ...and here.
8013 2018-07-02 Aldy Hernandez <aldyh@redhat.com>
8015 * tree-vrp.c (extract_range_from_unary_expr): Abstract ABS_EXPR
8017 (extract_range_from_abs_expr): ...here.
8019 2018-07-02 Eric Botcazou <ebotcazou@adacore.com>
8021 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Do not overrule
8022 -fno-omit-frame-pointer when not optimizing.
8024 2018-07-02 Martin Liska <mliska@suse.cz>
8027 * ipa-pure-const.c (malloc_candidate_p): Revert usage of ::get.
8028 (propagate_nothrow): Likewise.
8030 2018-07-02 Martin Liska <mliska@suse.cz>
8033 * ipa-inline.c (early_inliner): Revert wrongly added ::get call.
8035 2018-07-02 David Malcolm <dmalcolm@redhat.com>
8037 * dumpfile.c (dump_generic_expr_loc): Undo removal of this
8038 function in r262149, changing "loc" param from source_location to
8039 const dump_location_t &.
8040 * dumpfile.h (dump_generic_expr_loc): Undo removal of this
8041 declaration, as above.
8043 2018-07-01 Paul Koning <ni1d@arrl.net>
8045 * common/config/pdp11/pdp11-common.c (pdp11_handle_option): Handle
8046 -munit-asm, -mgnu-asm, -mdec-asm.
8047 * config/pdp11/pdp11-protos.h (pdp11_gen_int_label): New.
8048 (pdp11_output_labelref): New.
8049 (pdp11_output_def): New.
8050 (pdp11_output_addr_vec_elt): New.
8051 * config/pdp11/pdp11.c: Use tab between opcode and operands. Use
8052 %# and %@ format codes.
8053 (pdp11_option_override): New.
8054 (TARGET_ASM_FILE_START_FILE_DIRECTIVE): Define.
8055 (pdp11_output_ident): New.
8056 (pdp11_asm_named_section): New.
8057 (pdp11_asm_init_sections): New.
8058 (pdp11_file_start): New.
8059 (pdp11_file_end): New.
8060 (output_ascii): Use .ascii/.asciz for -mdec-asm.
8061 (pdp11_asm_print_operand): Update %# and %$ for -mdec-asm. Add
8062 %o, like %c but octal.
8063 (pdp11_option_override): New.
8064 * config/pdp11/pdp11.h (TEXT_SECTION_ASM_OP): Update for
8066 (DATA_SECTION_ASM_OP): Ditto.
8067 (READONLY_DATA_SECTION_ASM_OP): New.
8068 (IS_ASM_LOGICAL_LINE_SEPARATOR): New.
8069 (ASM_GENERATE_INTERNAL_LABEL): Use new function.
8070 (ASM_OUTPUT_LABELREF): Ditto.
8071 (ASM_OUTPUT_DEF): Ditto.
8072 (ASM_OUTPUT_EXTERNAL): New.
8073 (ASM_OUTPUT_SOURCE_FILENAME): New.
8074 (ASM_OUTPUT_ADDR_VEC_ELT): Use new function.
8075 (ASM_OUTPUT_SKIP): Update for -mdec-asm.
8076 * config/pdp11/pdp11.md: Use tab between opcode and operands. Use
8077 %# and %@ format codes.
8078 * config/pdp11/pdp11.opt (mgnu-asm): New.
8079 (mdec-asm): Conflicts with -mgnu-asm and -munix-asm.
8080 (munix-asm): Conflicts with -mdec-asm and -mgnu-asm.
8081 * doc/invoke.txt (PDP-11 Options): Add -mgnu-asm.
8083 2018-07-01 Aldy Hernandez <aldyh@redhat.com>
8085 * tree-ssa-threadupdate.c (mark_threaded_blocks): Avoid
8086 dereferencing path[] beyond its length.
8088 (debug_all_paths): New.
8089 (rewire_first_differing_edge): New.
8090 (adjust_paths_after_duplication): New.
8091 (duplicate_thread_path): Call adjust_paths_after_duplication.
8093 (thread_through_all_blocks): Add new argument to
8094 duplicate_thread_path.
8096 2018-06-30 Jim Wilson <jimw@sifive.com>
8098 * config/riscv/predicates.md (p2m1_shift_operand): New.
8099 (high_mask_shift_operand): New.
8100 * config/riscv/riscv.md (lshrsi3_zero_extend_3+1): New combiner
8101 pattern using p2m1_shift_operand.
8102 (lshsi3_zero_extend_3+2): New combiner pattern using
8103 high_mask_shift_operand.
8105 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
8107 * tree-vect-patterns.c (vect_get_external_def_edge): New function,
8109 (vect_recog_rotate_pattern): ...here.
8110 (vect_convert_input): Try to insert casts of invariants in the
8112 * tree-vect-loop-manip.c (vect_loop_versioning): Don't require the
8113 preheader to be empty.
8115 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
8117 * tree-vect-patterns.c (append_pattern_def_seq): Take an optional
8118 vector type. If given, install it in the new statement's
8120 (vect_element_precision): New function.
8121 (vect_unpromoted_value): New struct.
8122 (vect_unpromoted_value::vect_unpromoted_value): New function.
8123 (vect_unpromoted_value::set_op): Likewise.
8124 (vect_look_through_possible_promotion): Likewise.
8125 (vect_joust_widened_integer, vect_joust_widened_type): Likewise.
8126 (vect_widened_op_tree, vect_convert_input): Likewise.
8127 (vect_convert_inputs, vect_convert_output): Likewise.
8128 (vect_recog_dot_prod_pattern): Use vect_look_through_possible_promotion
8129 to handle the optional cast of the multiplication result and
8130 vect_widened_op_tree to detect the widened multiplication itself.
8131 Do not require the input and output of promotion casts to have
8132 the same sign, but base the signedness of the operation on the
8133 input rather than the result. If the pattern includes two
8134 promotions, check that those promotions have the same sign.
8135 Do not restrict the MULT_EXPR handling to a double-width result;
8136 handle quadruple-width results and wider. Use vect_convert_inputs
8137 to convert the inputs to the common type.
8138 (vect_recog_sad_pattern): Use vect_look_through_possible_promotion
8139 to handle the optional cast of the ABS result. Also allow a sign
8140 change or a sign extension between the ABS and MINUS.
8141 Use vect_widened_op_tree to detect the widened subtraction and use
8142 vect_convert_inputs to convert the inputs to the common type.
8143 (vect_handle_widen_op_by_const): Delete.
8144 (vect_recog_widen_op_pattern): New function.
8145 (vect_recog_widen_mult_pattern): Use it.
8146 (vect_recog_widen_shift_pattern): Likewise.
8147 (vect_recog_widen_sum_pattern): Use
8148 vect_look_through_possible_promotion to handle the promoted
8151 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
8153 * gimple-iterator.c (gsi_for_stmt): Add a new overload that takes
8154 the containing gimple_seq *.
8155 * gimple-iterator.h (gsi_for_stmt): Declare it.
8156 * tree-vect-patterns.c (vect_recog_dot_prod_pattern)
8157 (vect_recog_sad_pattern, vect_recog_widen_sum_pattern)
8158 (vect_recog_widen_shift_pattern, vect_recog_rotate_pattern)
8159 (vect_recog_vector_vector_shift_pattern, vect_recog_divmod_pattern)
8160 (vect_recog_mask_conversion_pattern): Remove STMT_VINFO_IN_PATTERN_P
8162 (vect_init_pattern_stmt, vect_set_pattern_stmt): New functions,
8164 (vect_mark_pattern_stmts): ...here. Handle cases in which the
8165 statement being replaced is part of an existing pattern
8166 definition sequence, inserting the new pattern statements before
8168 (vect_pattern_recog_1): Don't return a bool. If the statement
8169 is already part of a pattern, instead apply pattern matching
8170 to the pattern definition statements. Don't clear the
8171 STMT_VINFO_RELATED_STMT if is_pattern_stmt_p.
8172 (vect_pattern_recog): Don't break after the first match;
8173 continue processing the pattern definition statements instead.
8174 Don't bail out for STMT_VINFO_IN_PATTERN_P here.
8176 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
8178 * tree-vect-patterns.c (vect_reassociating_reduction_p): New function.
8179 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern)
8180 (vect_recog_widen_sum_pattern): Use it.
8182 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
8184 * tree-vect-loop.c (vectorizable_reduction): Assert that the
8185 phi is not a pattern statement and has not been replaced by
8186 a pattern statement.
8187 * tree-vect-patterns.c (type_conversion_p): Don't check
8188 STMT_VINFO_IN_PATTERN_P.
8189 (vect_recog_vector_vector_shift_pattern): Likewise.
8190 (vect_recog_dot_prod_pattern): Expect vect_is_simple_use to return
8191 the pattern statement rather than the original statement; check
8192 directly for a WIDEN_MULT_EXPR here.
8193 * tree-vect-slp.c (vect_get_and_check_slp_defs): Expect
8194 vect_is_simple_use to return the pattern statement rather
8195 than the original statement; use is_pattern_stmt_p to check
8196 for such a pattern statement.
8197 * tree-vect-stmts.c (process_use): Expect vect_is_simple_use
8198 to return the pattern statement rather than the original statement;
8199 don't do the same transformation here.
8200 (vect_is_simple_use): If the defining statement has been replaced
8201 by a pattern statement, return the pattern statement instead.
8202 Remove the corresponding (local) transformation from the vectype
8205 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
8207 * tree-vectorizer.h (vect_is_simple_use): Move the gimple ** to the
8208 end and default to null.
8209 * tree-vect-loop.c (vect_create_epilog_for_reduction)
8210 (vectorizable_reduction): Update calls accordingly, dropping the
8211 gimple ** argument if the passed-back statement isn't needed.
8212 * tree-vect-patterns.c (vect_get_internal_def, type_conversion_p)
8213 (vect_recog_rotate_pattern): Likewise.
8214 (vect_recog_mask_conversion_pattern): Likewise.
8215 * tree-vect-slp.c (vect_get_and_check_slp_defs): Likewise.
8216 (vect_mask_constant_operand_p): Likewise.
8217 * tree-vect-stmts.c (is_simple_and_all_uses_invariant, process_use):
8218 (vect_model_simple_cost, vect_get_vec_def_for_operand): Likewise.
8219 (get_group_load_store_type, get_load_store_type): Likewise.
8220 (vect_check_load_store_mask, vect_check_store_rhs): Likewise.
8221 (vectorizable_call, vectorizable_simd_clone_call): Likewise.
8222 (vectorizable_conversion, vectorizable_assignment): Likewise.
8223 (vectorizable_shift, vectorizable_operation): Likewise.
8224 (vectorizable_store, vect_is_simple_cond): Likewise.
8225 (vectorizable_condition, vectorizable_comparison): Likewise.
8226 (get_same_sized_vectype, vect_get_mask_type_for_stmt): Likewise.
8227 (vect_is_simple_use): Rename the def_stmt argument to def_stmt_out
8228 and move it to the end. Cope with null def_stmt_outs.
8230 2018-06-30 Bernd Edlinger <bernd.edlinger@hotmail.de>
8232 * Makefile.in (FLAGS_TO_PASS): Add RANLIB_FOR_TARGET.
8234 2018-06-29 Jeff Law <law@redhat.com>
8236 * config/v850/v850.c (v850_legitimate_address_p): Handle large
8237 displacements for TARGET_V850E2V3 and newer.
8238 (TARGET_LRA_P): Remove. Defaults to LRA now.
8239 * config/v850/v850.md (sign23byte_load): Remove.
8240 (unsign23byte_load, sign23hword_load, unsign23hword_load): Likewise.
8241 (23word_load, 23byte_store, 23hword_store, 23word_store): Likewise.
8243 2018-06-29 Martin Liska <mliska@suse.cz>
8246 * coverage.c (coverage_init): Mangle full path name.
8247 * doc/invoke.texi: Document the change.
8248 * gcov-io.c (mangle_path): New.
8249 * gcov-io.h (mangle_path): Likewise.
8250 * gcov.c (mangle_name): Use mangle_path for path mangling.
8252 2018-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8254 * config/arm/arm.c (output_move_double): Don't allow STRD instructions
8255 if starting source register is not even.
8257 2018-06-29 Martin Liska <mliska@suse.cz>
8259 PR tree-optimization/86263
8260 * tree-switch-conversion.c (switch_decision_tree::try_switch_expansion):
8261 Make edge redirection.
8263 2018-06-29 David Malcolm <dmalcolm@redhat.com>
8265 * dumpfile.c (dump_loc): Add indentation based on scope depth.
8266 (dump_scope_depth): New variable.
8267 (get_dump_scope_depth): New function.
8268 (dump_begin_scope): New function.
8269 (dump_end_scope): New function.
8270 * dumpfile.h (get_dump_scope_depth): New declaration.
8271 (dump_begin_scope): New declaration.
8272 (dump_end_scope): New declaration.
8273 (class auto_dump_scope): New class.
8274 (AUTO_DUMP_SCOPE): New macro.
8275 * tree-vectorizer.h (DUMP_VECT_SCOPE): Reimplement in terms of
8278 2018-06-29 Richard Biener <rguenther@suse.de>
8280 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Assert
8281 compute_all_dependences succeeds.
8282 * tree-vect-loop.c (vect_get_datarefs_in_loop): Fail early if we
8283 exceed --param loop-max-datarefs-for-datadeps.
8285 2018-06-29 Jakub Jelinek <jakub@redhat.com>
8287 * config/rs6000/t-rs6000: Append rs6000-modes.h to TM_H.
8289 2018-06-28 Uros Bizjak <ubizjak@gmail.com>
8292 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): Use
8293 alternative 0 in preferred_for_speed attribute.
8295 2018-06-28 Paul Koning <ni1d@arrl.net>
8297 * config/pdp11/pdp11-protos.h (pdp11_shift_length): New function.
8298 * config/pdp11/pdp11.c (pdp11_shift_length): New function.
8299 * config/pdp11/pdp11.h (ADJUST_INSN_LENGTH): Remove.
8300 * config/pdp11/pdp11.md: Correct "length" attribute calculation
8301 for shift insn patterns.
8303 2018-06-28 David Malcolm <dmalcolm@redhat.com>
8305 * cgraph.c (cgraph_node::get_body): Replace assignments to
8306 "dump_file" with calls to set_dump_file.
8307 * dumpfile.c (alt_dump_file): Make static, and group with...
8308 (alt_flags): ...this definition.
8309 (dumps_are_enabled): New variable.
8310 (refresh_dumps_are_enabled): New function.
8311 (set_dump_file): New function.
8312 (set_alt_dump_file): New function.
8313 (gcc::dump_manager::dump_start): Replace assignments to
8314 "dump_file" and "alt_dump_file" with calls to set_dump_file and
8316 (gcc::dump_manager::dump_finish): Likewise.
8317 * dumpfile.h (alt_dump_file): Delete decl.
8318 (dumps_are_enabled): New variable decl.
8319 (set_dump_file): New function decl.
8320 (dump_enabled_p): Rewrite in terms of new "dumps_are_enabled"
8322 * tree-nested.c (lower_nested_functions): Replace assignments to
8323 "dump_file" with calls to set_dump_file.
8325 2018-06-28 Eric Botcazou <ebotcazou@adacore.com>
8327 * tree-cfg.c (verify_gimple_in_cfg): Call verify_location on the
8328 goto_locus of each outgoing edge of each basic block.
8330 2018-06-28 Richard Biener <rguenther@suse.de>
8332 * dwarf2out.c (decl_scope_table): Remove.
8333 (push_decl_scope): Likewise.
8334 (pop_decl_scope): Likewise.
8335 (gen_type_die_for_member): Do not call push/pop_decl_scope.
8336 (gen_struct_or_union_type_die): Likewise.
8337 (gen_tagged_type_die): Likewise.
8338 (dwarf2out_init): Do not initialize decl_scope_table.
8339 (dwarf2out_c_finalize): Do not free it.
8341 2018-06-28 Richard Biener <rguenther@suse.de>
8343 * dwarf2out.c (gen_subprogram_die): Use is_unit_die when
8344 deciding whether to not re-use a DIE.
8346 2018-06-28 Richard Biener <rguenther@suse.de>
8348 * dwarf2out.c (gen_subprogram_die): Always re-use DIEs with an
8349 DW_AT_abstract_origin attribute.
8351 2018-06-28 Martin Liska <mliska@suse.cz>
8353 * tree-switch-conversion.c (jump_table_cluster::can_be_handled):
8354 Use newly introduced constants.
8355 * tree-switch-conversion.h (struct jump_table_cluster):
8356 Define max_ratio_for_size and max_ratio_for_speed.
8358 2018-06-28 Martin Liska <mliska@suse.cz>
8360 * tree-switch-conversion.c (jump_table_cluster::find_jump_tables):
8361 Add new checking assert to catch invalid state.
8362 (jump_table_cluster::can_be_handled): Handle single case
8364 (jump_table_cluster::is_beneficial): Bail out for such case.
8365 (bit_test_cluster::find_bit_tests):
8366 Add new checking assert to catch invalid state.
8367 (bit_test_cluster::can_be_handled): Handle single case
8369 (bit_test_cluster::is_beneficial): Bail out for such case.
8370 (switch_decision_tree::analyze_switch_statement):
8373 2018-06-28 Martin Liska <mliska@suse.cz>
8375 * common.opt: Introduce -completion option.
8376 * gcc.c (driver_handle_option): Handle it.
8377 (driver::main): Print completions if completion
8379 * opt-suggestions.c (option_proposer::get_completions):
8381 (option_proposer::suggest_completion): Likewise.
8382 (option_proposer::find_param_completions): Likewise.
8383 (verify_autocompletions): Likewise.
8384 (test_completion_valid_options): Likewise.
8385 (test_completion_valid_params): Likewise.
8386 (in_completion_p): Likewise.
8387 (empty_completion_p): Likewise.
8388 (test_completion_partial_match): Likewise.
8389 (test_completion_garbage): Likewise.
8390 (opt_proposer_c_tests): Likewise.
8391 * opt-suggestions.h: Declare new functions.
8392 * opts.c (common_handle_option): Handle OPT__completion_.
8393 * selftest-run-tests.c (selftest::run_tests): Add
8394 opt_proposer_c_tests.
8395 * selftest.c (assert_str_startswith): New.
8396 * selftest.h (assert_str_startswith): Likewise.
8397 (opt_proposer_c_tests): New.
8398 (ASSERT_STR_STARTSWITH): Likewise.
8400 2018-06-28 Martin Liska <mliska@suse.cz>
8402 * Makefile.in: Add opt-suggestions.o.
8403 * gcc-main.c: Include opt-suggestions.h.
8404 * gcc.c (driver::driver): Likewise.
8405 (driver::~driver): Remove m_option_suggestions.
8406 (driver::build_option_suggestions): Moved to option_proposer.
8407 (driver::suggest_option): Likewise.
8408 (driver::handle_unrecognized_options): Use option_proposer.
8409 * gcc.h (class driver): Add new memver m_option_proposer.
8410 * opt-suggestions.c: New file.
8411 * opt-suggestions.h: New file.
8413 2018-06-28 Martin Liska <mliska@suse.cz>
8415 * vec.h (class auto_string_vec): New (moved from auto_argvec).
8416 (auto_string_vec::~auto_string_vec): Likewise.
8418 2018-06-28 Eric Botcazou <ebotcazou@adacore.com>
8420 * tree-inline.h (struct copy_body_data): Move remapping_type_depth and
8421 prevent_decl_creation_for_types fields up and add reset_location field.
8422 * tree-inline.c (remap_gimple_stmt): Force input_location on the new
8423 statement if id->reset_location is true.
8424 (copy_edges_for_bb): Do not set goto_locus on the new edges if
8425 id->reset_location is true.
8426 (copy_phis_for_bb): Force input_location on the arguments if
8427 id->reset_location is true.
8428 (expand_call_inline): Set id->reset_location if DECL_IGNORED_P
8429 is set on the function to be inlined.
8431 2018-06-27 Stephan Bergmann <sbergman@redhat.com>
8433 * doc/invoke.texi (Debugging Options): Add -gsplit-dwarf.
8435 2018-06-27 Dimitar Dimitrov <dimitar@dinux.eu>
8437 * lra-eliminations.c (update_reg_eliminate): Mark all spanning hard
8438 registers for Pmode.
8439 * lra-lives.c (check_pseudos_live_through_calls): Mark all spanning
8440 hard registers for the clobbered pseudo.
8442 2018-06-27 Paul Koning <ni1d@arrl.net>
8444 * common/config/pdp11/pdp11-common.c (pdp11_handle_option): Handle
8445 mutually exclusive options.
8446 * config/pdp11/constraints.md (h): New constraint.
8447 (O): Update definition to match shift code generation.
8448 (D): New constraint.
8449 * config/pdp11/pdp11-modes.def (CCNZ): Define mode.
8451 * config/pdp11/pdp11-protos.h (int_no_side_effect_operand): New
8453 (output_jump): Change arguments.
8454 (pdp11_fixed_cc_regs): New function.
8455 (pdp11_cc_mode): Ditto.
8456 (pdp11_expand_shift): Ditto.
8457 (pdp11_assemble_shift): Ditto.
8458 (pdp11_small_shift): Ditto.
8459 (pdp11_branch_cost): Remove.
8460 * config/pdp11/pdp11.c (pdp11_assemble_integer): Remove comments
8462 (pdp11_register_move_cost): Update for CC registers.
8463 (pdp11_rtx_costs): Add case for LSHIFTRT.
8464 (pdp11_output_jump): Add CCNZ mode conditional branches.
8465 (notice_update_cc_on_set): Remove.
8466 (pdp11_cc_mode): New function.
8467 (simple_memory_operand): Correct pre/post decrement case.
8468 (no_side_effect_operand): New function.
8469 (pdp11_regno_reg_class): Add CC_REGS class.
8470 (pdp11_fixed_cc_regs): New function.
8471 (pdp11_small_shift): New function.
8472 (pdp11_expand_shift): New function to expand shift insns.
8473 (pdp11_assemble_shift): New function to output shifts.
8474 (pdp11_branch_cost): Remove.
8475 (pdp11_modes_tieable_p): Make QI/HI modes tieable.
8476 * config/pdp11/pdp11.h (SIZE_TYPE): Ensure 16-bit type.
8477 (WCHAR_TYPE): Ditto.
8478 (PTRDIFF_TYPE): Ditto.
8479 (ADJUST_INSN_LENGTH): New macro.
8480 (FIXED_REGISTERS): Add CC registers.
8481 (CALL_USED_REGISTERS): Ditto.
8483 (REG_CLASS_NAMES): Ditto.
8484 (REG_CLASS_CONTENTS): Ditto.
8485 (SELECT_CC_MODE): Use new function.
8486 (TARGET_FLAGS_REGNUM): New macro.
8487 (TARGET_FIXED_CONDITION_CODE_REGS): Ditto.
8488 (cc0_reg_rtx): Remove.
8489 (CC_STATUS_MDEP): Remove.
8490 (CC_STATUS_MDEFP_INIT): Remove.
8491 (CC_IN_FPU): Remove.
8492 (NOTICE_UPDATE_CC): Remove.
8493 (REGISTER_NAMES): Add CC registers.
8494 (BRANCH_COST): Change to constant 1.
8495 * config/pdp11/pdp11.md: Rewrite for CCmode condition code
8497 * config/pdp11/pdp11.opt (mbcopy): Remove.
8498 (mbcopy-builtin): Remove.
8499 (mbranch-cheap): Remove.
8500 (mbranch-expensive): Remove.
8501 * config/pdp11/predicates.md (expand_shift_operand): Update to
8502 match shift code generation.
8503 (ccnz_operator): New predicate.
8504 * doc/invoke.texi (PDP-11 Options): Remove deleted options
8505 -mbcopy, -mbcopy-builtin, -mbranch-cheap, -mbranch-expensive.
8506 Remove non-existent option -mabshi, -mno-abshi. Document mutually
8508 * doc/md.texi (PDP-11): Document new D and h constraints. Update
8509 description of O constraint.
8511 2018-06-27 Jeff Law <law@redhat.com>
8512 Austin Law <austinklaw@gmail.com>
8514 * config/v850/v850.md (addsi3_set_flags): New pattern.
8515 (subsi3_set_flags, negsi2_set_flags, andsi3_set_flags): Likewise.
8516 (iorsi3_set_flags, xorsi3_set_flags, one_cmplsi2_set_flags): Likewise.
8517 (zero_extendhisi2_v850_set_flags): Likewise.
8518 (zero_extendqisi2_v850_set_flags): Likewise.
8519 (ashlsi3_set_flags, ashlsi3_v850e2_set_flags): Likewise.
8520 (lshrsi3_set_flags, lshrsi3_v850e2_set_flags): Likewise.
8521 (ashrsi3_set_flags, ashrsi3_v850e2_set_flags): Likewise.
8523 * config/v850/v850-protos.h (notice_update_cc): Remove.
8524 * config/v850/v850.c (v850_compare_op0, v850_compare_op1): Remove.
8525 (v850_print_operand): Handle 'D' and "d".
8526 (v850_select_cc_mode): Remove ATTRIBUTE_UNUSED for last argument.
8527 Add handling of arithmetic/logical operations compared against zero.
8528 (v850_gen_float_compare): Remove ATTRIBUTE_UNUSED for last argument.
8529 Do not look at v850_compare_op, instead get mode from last argument.
8530 (v850_gen_compare): Remove
8531 (increment_stack): Use addsi3_clobber_flags to avoid splitting failure
8532 after reload for prologue insns.
8533 (expand_prologue): Account for CLOBBER of CC_REGNUM in various
8535 (construct_save_jarl): Likewise.
8536 (TARGET_FLAGS_REGNUM): Define.
8537 * config/v850/v850.h (v850_compare_op0, v850_compare_op1): Remove.
8538 (NOTICE_UPDATE_CC): Remove.
8539 * config/v850/v850.md (v850_tst1): Use (reg:CCZ CC_REGNUM) rather
8540 than cc0. Conditionalize on reload_completed.
8541 (cmpsi_insn, setfcc_insn): Likewise.
8542 (tst1 splitter): Turn into define_and_split which sets the flags
8544 (cstoresi4, cbranchsf4, cbranchdf4, cbranchsi4_insn): Likewise.
8545 (cbranchsi4, branch_normal, branch_invert): Do not expose cc0 here.
8546 (cstoresf4, cstoredf4): Clobber the flags.
8547 (cmpsi, cmpsf, cmpdf): Remove expanders.
8548 (setf_insn): Remove pattern.
8549 (addsi3): Turn into define_and_split which clobbers the flags after
8550 reload and a suitable pattern (addsi3_clobber_flags) for use after
8552 (subsi3, negsi2, andsi3, iorsi3, xorsi3, one_cmplsi2) Likewise.
8553 (ashlsi3, ashlsi3_v850e2, lshrsi3, lsh4si3_v850e2): Likewise.
8554 (ashrsi3, ashrsi3_v850e2): Likewise.
8555 (bins): Clobber the flags.
8556 (movsicc_normal_cc, movsicc_normal, movsicc_tst1): Likewise.
8557 (movsicc_tst1_revesed, sasf, swap and rotate patterns): Likewise.
8558 (fix_loop_counter, call_internal_short, call_internal_long): Likewise.
8559 (call_value_internal_short, call_value_internal_long): Likewise.
8560 (callt_save_interrupt, callt_return_interrupt): Likewise.
8561 (save_interrupt, return_interrupt): Likewise.
8562 (callt_save_all_interrupt, save_all_interrupt): Likewise.
8563 (_save_all_interrupt, callt_restore_all_interrupt): Likewise.
8564 (restore_all_interrupt, _restore_all_interrupt): Likewise.
8565 (All FP comparisons): Only allow after reload has completed.
8567 (divh, divhu): Tweak output template.
8568 (branch_z_normal, branch_z_invert): Remove
8569 (branch_nz_normal, branch_nz_invert): Likewise.
8570 (extendhisi_insn, extendqisi_insn): Do not clobber flags.
8572 * config/v850/v850-modes.def (CCZ, CCNZ): Add new modes.
8573 * config/v850/v850.c (notice_update_cc): Remove.
8574 * config/v850/v850.h (CC_OVERFLOW_UNUSABLE): Remove
8575 (CC_NO_CARRY): Likewise.
8576 (NOTICE_UPDATE_CC): Define to nothing.
8577 * config/v850/v850.md: Remove block comment on cc0 handling
8578 Remove "cc" attribute from all patterns. Remove cc_status handling
8579 from all patterns. Minor formatting fixes.
8581 2018-06-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8583 * config/aarch64/aarch64-cores.def (cortex-a76): New entry.
8584 (cortex-a76.cortex-a55): Likewise.
8585 * config/aarch64/aarch64-tune.md: Regenerate.
8586 * doc/invoke.texi (AArch64 Options): Document cortex-a76 and
8587 cortex-a76.cortex-a55.
8589 2018-06-27 Jeff Law <law@redhat.com>
8591 * config/v850/t-v850 (MULTILIB_OPTIONS): Remove 8byte-align.
8592 (MULTILIB_DIRNAMES): Similarly.
8594 2018-06-27 Eric Botcazou <ebotcazou@adacore.com>
8596 * gimple.h (gimple_return_retbnd): Delete.
8597 (gimple_return_set_retbnd): Likewise.
8598 * cgraphunit.c (cgraph_node::expand_thunk): Remove call to
8599 gimple_return_set_retbnd.
8600 * gimple-pretty-print.c (dump_gimple_return): Remove call to
8601 gimple_return_retbnd and adjust.
8602 * tree-inline.h (struct copy_body_data): Remove retbnd field.
8603 * tree-inline.c (remap_gimple_stmt): Remove handling of retbnd.
8604 Explicitly return NULL in a couple more cases. Move assertion
8605 on debug statements and remove unreachable code.
8606 (reset_debug_binding): Do not test id->retbnd.
8607 (expand_call_inline): Do not set it.
8609 2018-06-27 Rasmus Villemoes <rv@rasmusvillemoes.dk>
8611 * configure.ac: Add --disable-gcov option.
8612 * configure: Regenerate.
8613 * Makefile.in: Honour @enable_gcov@.
8614 * doc/install.texi: Document --disable-gcov.
8616 2018-06-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8618 * config/arm/arm-cpus.in (cortex-a76): New entry.
8619 (cortex-a76.cortex-a55): Likewise.
8620 * config/arm/arm-tables.opt: Regenerate.
8621 * config/arm/arm-tune.md: Likewise.
8622 * config/arm/driver-arm.c (arm_cpu_table): Add Cortex-A76 entry.
8623 * doc/invoke.texi (ARM Options): Document cortex-a76 and
8624 cortex-a76.cortex-a55.
8626 2018-06-27 Tamar Christina <tamar.christina@arm.com>
8629 * config/aarch64/aarch64.md (*movhf_aarch64): Add dup v0.4h pattern.
8631 2018-06-27 Siddhesh Poyarekar <siddhesh@sourceware.org>
8633 * config/aarch64/aarch64.h (CALL_USE_REGISTERS): Fix obsolete
8635 (EPILOGUE_USES): Likewise.
8637 2018-06-26 Eric Botcazou <ebotcazou@adacore.com>
8639 * tree-inline.c (remap_location): New function extracted from...
8640 (copy_edges_for_bb): Add ID parameter. Remap goto_locus.
8641 (copy_phis_for_bb): ...here. Call remap_location.
8642 (copy_cfg_body): Adjust call to copy_edges_for_bb.
8644 2018-06-26 Aaron Sawdey <acsawdey@linux.ibm.com>
8646 * config/rs6000/rs6000-string.c (expand_block_clear): Don't use
8647 unaligned vsx for 16B memset.
8649 2018-06-26 Segher Boessenkool <segher@kernel.crashing.org>
8652 * config/rs6000/rs6000.c (rs6000_init_builtins): Do not set
8653 ieee128_float_type_node to long_double_type_node unless
8654 TARGET_LONG_DOUBLE_128 is set.
8656 2018-06-26 David Malcolm <dmalcolm@redhat.com>
8658 * cfgloop.c (get_loop_location): Convert return type from
8659 location_t to dump_user_location_t, replacing INSN_LOCATION lookups
8660 by implicit construction from rtx_insn *, and using
8661 dump_user_location_t::from_function_decl for the fallback case.
8662 * cfgloop.h (get_loop_location): Convert return type from
8663 location_t to dump_user_location_t.
8664 * cgraphunit.c (walk_polymorphic_call_targets): Update call to
8665 dump_printf_loc to pass in a dump_location_t rather than a
8666 location_t, via the gimple stmt.
8667 * coverage.c (get_coverage_counts): Update calls to
8668 dump_printf_loc to pass in dump_location_t rather than a
8670 * doc/optinfo.texi (Dump types): Convert example of
8671 dump_printf_loc from taking "locus" to taking "insn". Update
8672 description of the "_loc" calls to cover dump_location_t.
8673 * dumpfile.c: Include "backend.h", "gimple.h", "rtl.h", and
8675 (dump_user_location_t::dump_user_location_t): New constructors,
8676 from gimple *stmt and rtx_insn *.
8677 (dump_user_location_t::from_function_decl): New function.
8678 (dump_loc): Make static.
8679 (dump_gimple_stmt_loc): Convert param "loc" from location_t to
8680 const dump_location_t &.
8681 (dump_generic_expr_loc): Delete.
8682 (dump_printf_loc): Convert param "loc" from location_t to
8683 const dump_location_t &.
8684 (selftest::test_impl_location): New function.
8685 (selftest::dumpfile_c_tests): New function.
8686 * dumpfile.h: Include "profile-count.h".
8687 (class dump_user_location_t): New class.
8688 (struct dump_impl_location_t): New struct.
8689 (class dump_location_t): New class.
8690 (dump_printf_loc): Convert 2nd param from source_location to
8691 const dump_location_t &.
8692 (dump_generic_expr_loc): Delete.
8693 (dump_gimple_stmt_loc): Convert 2nd param from source_location to
8694 const dump_location_t &.
8695 * gimple-fold.c (fold_gimple_assign): Update call to
8696 dump_printf_loc to pass in a dump_location_t rather than a
8697 location_t, via the gimple stmt.
8698 (gimple_fold_call): Likewise.
8699 * gimple-loop-interchange.cc
8700 (loop_cand::analyze_iloop_reduction_var): Update for change to
8701 check_reduction_path.
8702 (tree_loop_interchange::interchange): Update for change to
8704 * graphite-isl-ast-to-gimple.c (scop_to_isl_ast): Update for
8705 change in return-type of find_loop_location.
8706 (graphite_regenerate_ast_isl): Likewise.
8707 * graphite-optimize-isl.c (optimize_isl): Likewise.
8708 * graphite.c (graphite_transform_loops): Likewise.
8709 * ipa-devirt.c (ipa_devirt): Update call to dump_printf_loc to
8710 pass in a dump_location_t rather than a location_t, via the
8712 * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
8713 * ipa.c (walk_polymorphic_call_targets): Likewise.
8714 * loop-unroll.c (report_unroll): Convert "locus" param from
8715 location_t to dump_location_t.
8716 (decide_unrolling): Update for change to get_loop_location's
8718 * omp-grid.c (struct grid_prop): Convert field "target_loc" from
8719 location_t to dump_user_location_t.
8720 (grid_find_single_omp_among_assignments_1): Updates calls to
8721 dump_printf_loc to pass in a dump_location_t rather than a
8722 location_t, via the gimple stmt.
8723 (grid_parallel_clauses_gridifiable): Convert "tloc" from
8724 location_t to dump_location_t. Updates calls to dump_printf_loc
8725 to pass in a dump_location_t rather than a location_t, via the
8727 (grid_inner_loop_gridifiable_p): Likewise.
8728 (grid_dist_follows_simple_pattern): Likewise.
8729 (grid_gfor_follows_tiling_pattern): Likewise.
8730 (grid_target_follows_gridifiable_pattern): Likewise.
8731 (grid_attempt_target_gridification): Convert initialization
8732 of local "grid" from memset to zero-initialization; FIXME: does
8733 this require C++11? Update call to dump_printf_loc to pass in a
8734 optinfo_location rather than a location_t, via the gimple stmt.
8735 * profile.c (read_profile_edge_counts): Updates call to
8736 dump_printf_loc to pass in a dump_location_t rather than a
8738 (compute_branch_probabilities): Likewise.
8739 * selftest-run-tests.c (selftest::run_tests): Call
8741 * selftest.h (dumpfile_c_tests): New decl.
8742 * tree-loop-distribution.c (pass_loop_distribution::execute):
8743 Update for change in return type of find_loop_location.
8744 * tree-parloops.c (parallelize_loops): Likewise.
8745 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Convert
8746 "locus" from location_t to dump_user_location_t.
8747 (canonicalize_loop_induction_variables): Likewise.
8748 * tree-ssa-loop-ivopts.c (tree_ssa_iv_optimize_loop): Update
8749 for change in return type of find_loop_location.
8750 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update call
8751 to dump_printf_loc to pass in a dump_location_t rather than a
8752 location_t, via the stmt.
8753 * tree-ssa-sccvn.c (eliminate_dom_walker::before_dom_children):
8755 * tree-vect-loop-manip.c (find_loop_location): Convert return
8756 type from source_location to dump_user_location_t.
8757 (vect_do_peeling): Update for above change.
8758 (vect_loop_versioning): Update for change in type of
8760 * tree-vect-loop.c (check_reduction_path): Convert "loc" param
8761 from location_t to dump_user_location_t.
8762 (vect_estimate_min_profitable_iters): Update for change in type
8764 * tree-vect-slp.c (vect_print_slp_tree): Convert param "loc" from
8765 location_t to dump_location_t.
8766 (vect_slp_bb): Update for change in type of vect_location.
8767 * tree-vectorizer.c (vect_location): Convert from source_location
8768 to dump_user_location_t.
8769 (try_vectorize_loop_1): Update for change in vect_location's type.
8770 (vectorize_loops): Likewise.
8771 (increase_alignment): Likewise.
8772 * tree-vectorizer.h (vect_location): Convert from source_location
8773 to dump_user_location_t.
8774 (find_loop_location): Convert return type from source_location to
8775 dump_user_location_t.
8776 (check_reduction_path): Convert 1st param from location_t to
8777 dump_user_location_t.
8778 * value-prof.c (check_counter): Update call to dump_printf_loc to
8779 pass in a dump_user_location_t rather than a location_t; update
8780 call to error_at for change in type of "locus".
8781 (check_ic_target): Update call to dump_printf_loc to
8782 pass in a dump_user_location_t rather than a location_t, via the
8785 2018-06-26 Robin Dapp <rdapp@linux.vnet.ibm.com>
8787 * config/s390/s390.h (enum processor_flags): Do not use
8788 default tune parameter when -march was specified.
8790 2018-06-26 Jakub Jelinek <jakub@redhat.com>
8793 * config/i386/i386.md (setcc + movzbl to xor + setcc peephole2s):
8794 Check reg_overlap_mentioned_p in addition to reg_set_p with the same
8797 2018-06-26 Richard Biener <rguenther@suse.de>
8799 PR tree-optimization/86287
8801 * tree-vect-loop.c (vect_transform_loop_stmt): Fix read-after-free.
8802 (vect_analyze_loop): Initialize n_stmts.
8804 2018-06-26 Richard Biener <rguenther@suse.de>
8807 * fold-const.c (fold_convertible_p): Pointer extension
8810 2018-06-26 Alexandre Oliva <aoliva@redhat.com>
8813 * dwarf2out.c (loc_list_has_views): Adjust comments.
8814 (dw_loc_list): Split single cross-partition range with
8817 2018-06-25 Jeff Law <law@redhat.com>
8819 * common/config/v850/v850-common.c (TARGET_DEFAULT_TARGET_FLAGS): Turn
8820 on -mbig-switch by default.
8822 * config/v850/predicates.md (const_float_1_operand): Fix match_code
8824 (const_float_0_operand): Remove unused predicate.
8825 * config/v850/v850.md (define_constants): Remove UNSPEC_LOOP.
8826 (define_c_enum unspec): Add LOOP, RCP and RSQRT constants.
8827 (recipsf2): New expander. Original pattern now called
8829 (recipdf2, recipdf2_insn): Similarly.
8830 (rsqrtsf2, rsqrtsf2_insn): Similarly
8831 (rsqrtdf2, rsqrtdf2_insn): Similarly
8833 2018-06-26 Gerald Pfeifer <gerald@pfeifer.com>
8835 * ginclude/stddef.h: Remove an obsolete comment on FreeBSD 5.
8836 Simplify logic for FreeBSD (twice).
8838 2018-06-25 Martin Sebor <msebor@redhat.com>
8840 PR tree-optimization/86204
8841 * tree-ssa-strlen.c (handle_builtin_strlen): Avoid storing
8842 a strnlen result if it's less than the length of the string.
8844 2018-06-25 Martin Sebor <msebor@redhat.com>
8846 PR tree-optimization/85700
8847 * gimple-fold.c (gimple_fold_builtin_strncat): Adjust comment.
8848 * tree-ssa-strlen.c (is_strlen_related_p): Handle integer subtraction.
8849 (maybe_diag_stxncpy_trunc): Distinguish strncat from strncpy.
8851 2018-06-25 Martin Sebor <msebor@redhat.com>
8853 * doc/extend.texi (Zero-length arrays): Update and clarify.
8855 2018-06-25 Michael Meissner <meissner@linux.ibm.com>
8857 * config.gcc (powerpc64le*): Revert January 16th, 2018 patch that
8858 added IEEE/IBM long double multilib support on PowerPC little
8859 endian Linux systems.
8860 * config/rs6000/linux64.h (MULTILIB_DEFAULTS_IEEE): Likewise.
8861 (MULTILIB_DEFAULTS): Likewise.
8862 * config/rs6000/rs6000.c (rs6000_option_override_internal):
8864 * config/rs6000/rs6000.h (TARGET_IEEEQUAD_MULTILIB): Likewise.
8865 * config/rs6000/t-ldouble-linux64le-ibm: Delete, no longer used.
8866 * config/rs6000/t-ldouble-linux64le-ieee: Delete, no longer used.
8868 2018-06-25 Alexander Monakov <amonakov@ispras.ru>
8871 * sort.cc (REORDER_23): Avoid memcpy with same destination and source.
8872 (REORDER_45): Likewise.
8874 2018-06-25 Jeff Law <law@redhat.com>
8876 * config/v850/v850.md (divmodhi4): Make sure to sign extend the
8877 dividend to 32 bits. Adjust length.
8878 (udivmodhi4): Cleanup output template. Fix length.
8880 2018-06-25 Carl Love <cel@us.ibm.com>
8882 * config/rs6000/vsx.md: Change word selector to prefered location.
8884 2018-06-25 Richard Biener <rguenther@suse.de>
8886 PR tree-optimization/86304
8887 * tree-vectorizer.c (vectorize_loops): Walk over new possibly
8888 epilogue-if-converted loops as well.
8890 2018-06-25 Jan Hubicka <hubicka@ucw.cz>
8892 * lto-section-out.c (lto_begin_section): Do not print section
8893 name for noaddr and unnumbered dumps.
8895 2018-06-25 Richard Biener <rguenther@suse.de>
8897 * tree-vectorizer.h (struct vec_info_shared): New structure
8898 with parts split out from struct vec_info and loop_nest from
8899 struct _loop_vec_info.
8900 (struct vec_info): Adjust accordingly.
8901 (struct _loop_vec_info): Likewise.
8902 (LOOP_VINFO_LOOP_NEST): Adjust.
8903 (LOOP_VINFO_DATAREFS): Likewise.
8904 (LOOP_VINFO_DDRS): Likewise.
8905 (struct _bb_vec_info): Likewise.
8906 (BB_VINFO_DATAREFS): Likewise.
8907 (BB_VINFO_DDRS): Likewise.
8908 (struct _stmt_vec_info): Add dr_aux member.
8909 (DR_VECT_AUX): Adjust to refer to member of DR_STMTs vinfo.
8910 (DR_MISALIGNMENT_UNINITIALIZED): New.
8911 (set_dr_misalignment): Adjust.
8912 (dr_misalignment): Assert misalign isn't DR_MISALIGNMENT_UNINITIALIZED.
8913 (vect_analyze_loop): Adjust prototype.
8914 (vect_analyze_loop_form): Likewise.
8915 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences):
8916 Compute dependences lazily.
8917 (vect_record_base_alignments): Use shared datarefs/ddrs.
8918 (vect_verify_datarefs_alignment): Likewise.
8919 (vect_analyze_data_refs_alignment): Likewise.
8920 (vect_analyze_data_ref_accesses): Likewise.
8921 (vect_analyze_data_refs): Likewise.
8922 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Add
8923 constructor parameter for shared part.
8924 (vect_analyze_loop_form): Pass in shared part and adjust.
8925 (vect_analyze_loop_2): Pass in storage for the number of
8926 stmts. Move loop nest finding to the caller. Compute
8928 (vect_analyze_loop): Pass in shared part.
8929 (vect_transform_loop): Verify shared datarefs are unchanged.
8930 * tree-vect-slp.c (_bb_vec_info::_bb_vec_info): Add
8931 constructor parameter for shared part.
8932 (vect_slp_analyze_bb_1): Pass in shared part and adjust.
8933 (vect_slp_bb): Verify shared datarefs are unchanged before
8935 * tree-vect-stmts.c (ensure_base_align): Adjust for DR_AUX
8937 (new_stmt_vec_info): Initialize DR_AUX misalignment to
8938 DR_MISALIGNMENT_UNINITIALIZED.
8939 * tree-vectorizer.c (vec_info::vec_info): Add constructor
8940 parameter for shared part.
8941 (vec_info::~vec_info): Adjust.
8942 (vec_info_shared::vec_info_shared): New.
8943 (vec_info_shared::~vec_info_shared): Likewise.
8944 (vec_info_shared::save_datarefs): Likewise.
8945 (vec_info_shared::check_datarefs): Likewise.
8946 (try_vectorize_loop_1): Construct shared part live for analyses
8947 of a single loop for multiple vector sizes.
8948 * tree-parloops.c (gather_scalar_reductions): Adjust.
8950 2018-06-25 Richard Biener <rguenther@suse.de>
8952 * tree-vect-data-refs.c (vect_find_stmt_data_reference): Modify
8953 DR for SIMD lane accesses here and mark DR with (void *)-1 aux.
8954 (vect_analyze_data_refs): Remove similar code from here and
8955 simplify accordingly.
8957 2018-06-25 Richard Biener <rguenther@suse.de>
8959 * tree-vect-data-refs.c (vect_check_gather_scatter): Fail
8960 for reverse storage order accesses rather than asserting
8961 they cannot happen here.
8963 2018-06-25 Tom de Vries <tdevries@suse.de>
8966 * config/i386/i386.md (define_insn "*tls_global_dynamic_64_<mode>"):
8967 Use data16 instead of .byte for insn prefix.
8969 2018-06-25 Andreas Krebbel <krebbel@linux.ibm.com>
8972 * parser.c (make_char_string_pack): Pass this literal chars
8973 through cpp_interpret_string.
8974 (cp_parser_userdef_numeric_literal): Check the result of
8975 make_char_string_pack.
8977 2018-06-24 Maya Rashish <coypu@sdf.org>
8979 * ginclude/stddef.h: Simplify conditions around avoiding
8980 re-definition of __size_t.
8982 2018-06-22 Jan Hubicka <hubicka@ucw.cz>
8984 * lto-streamer-out.c (tree_is_indexable): Make LABEL_DECL nonindexable
8985 unless it is forced or nonlocal; assert that we stream no IMPORTED_DECL.
8987 2018-06-22 Maya Rashish <coypu@sdf.org>
8989 * doc/invoke.texi (mno-fancy-math-387): Update for changes
8990 made to OpenBSD and NetBSD through the years.
8992 2018-06-22 Kelvin Nilsen <kelvin@gcc.gnu.org>
8994 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Change
8995 behavior of vec_pack (vector double, vector double) to match
8996 behavior of vec_float2 (vector double, vector double).
8998 2018-06-22 Olivier Hainque <hainque@adacore.com>
9000 * gimplify.c (gimplify_function_tree): Prevent creation
9001 of a trampoline for the address of the current function
9002 passed to entry/exit instrumentation hooks.
9004 2018-06-22 Aaron Sawdey <acsawdey@linux.ibm.com>
9007 * config/rs6000/rs6000-string.c (expand_strn_compare): Handle -m32
9010 2018-06-22 Martin Liska <mliska@suse.cz>
9012 PR tree-optimization/86263
9013 * tree-switch-conversion.c (jump_table_cluster::find_jump_tables):
9014 Bail out if is_enabled is false.
9015 * tree-switch-conversion.h (jump_table_cluster::is_enabled):
9017 (jump_table_cluster::is_enabled): New function.
9019 2018-06-22 Jan Hubicka <hubicka@ucw.cz>
9021 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not stream
9022 BINFO_BASE_ACCESSES and BINFO_VPTR_FIELD.
9023 * tree-streamer-in.c (streamer_read_tree_bitfields): Likewise.
9024 (lto_input_ts_binfo_tree_pointers): Likewise.
9025 * tree-streamer-out.c (streamer_write_tree_bitfields,
9026 write_ts_binfo_tree_pointers): Likewise.
9027 * tree.c (free_lang_data_in_binfo): Clear BINFO_VPTR_FIELD.
9029 2018-06-22 Jan Hubicka <hubicka@ucw.cz>
9031 * tree.c (free_lang_data_in_type): Free all TYPE_VFIELDs.
9033 2018-06-22 Martin Liska <mliska@suse.cz>
9035 * symbol-summary.h (get): Make it pure and inline move
9036 functionality from ::get function.
9037 (get): Remove and inline into ::get and ::get_create.
9038 (get_create): Move code from ::get function.
9040 2018-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9043 * config/i386/sol2.h (CPP_SPEC): Don't pass -P for
9044 -x assembler-with-cpp.
9046 2018-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9048 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define
9049 _FILE_OFFSET_BITS=64 for C++.
9051 2018-06-21 Michael Meissner <meissner@linux.ibm.com>
9053 * config/rs6000/rs6000.md (extendtfif2): Add missing 128-bit
9054 conversion insn that shows up when pr85657-3.c is compiled using
9055 IEEE 128-bit long double.
9056 (neg<mode>2_internal): Use the correct mode to check whether the
9057 mode is IBM extended.
9058 * config/rs6000/rs6000.c (init_float128_ieee): Prevent complex
9059 multiply and divide external functions from being created more
9062 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
9064 * cfgrtl.c (fixup_reorder_chain): Do not emit NOPs in DECL_IGNORED_P
9066 (rtl_merge_blocks): Likewise. Do not emit a NOP if the location of
9067 the edge can be forwarded.
9068 (cfg_layout_merge_blocks): Likewise.
9070 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
9072 * except.c (finish_eh_generation): Commit edge insertions only after
9073 the EH edges have been redirected from post-landing to landing pads.
9075 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
9077 * tree-nested.c (get_frame_type): Use create_tmp_var_raw instead of
9078 create_tmp_var_for to create the FRAME decl.
9079 (finalize_nesting_tree_1): Do not unchain the FRAME decl.
9081 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
9083 * tree-inline.c (copy_edges_for_bb): Minor tweak.
9084 (maybe_move_debug_stmts_to_successors): Also reset the locus of the
9085 debug statement when resetting its value.
9086 (expand_call_inline): Copy the locus of the call onto the assignment
9087 of the return value, if any. Use local variable in more cases.
9089 2018-06-21 Martin Liska <mliska@suse.cz>
9091 * ipa-pure-const.c (propagate_nothrow): Use
9092 funct_state_summaries->get.
9093 (dump_malloc_lattice): Likewise.
9094 (propagate_malloc): Likewise.
9096 2018-06-21 Richard Biener <rguenther@suse.de>
9098 * lto-streamer-out.c (DFS::DFS_write_tree_body): Update outdated
9099 comment. Follow BLOCK_ABSTRACT_ORIGIN unconditionally.
9100 * tree-streamer-in.c (lto_input_ts_block_tree_pointers): Update
9102 * tree-streamer-out.c (write_ts_block_tree_pointers): Stream
9103 BLOCK_ABSTRACT_ORIGIN unconditionally.
9105 2018-06-21 David Malcolm <dmalcolm@redhat.com>
9107 * ipa-cp.c (ipcp_driver): Set edge_clone_summaries to NULL after
9109 * ipa-reference.c (ipa_reference_c_finalize): Delete
9110 ipa_ref_opt_sum_summaries and set it to NULL.
9112 2018-06-21 Tom de Vries <tdevries@suse.de>
9114 PR tree-optimization/85859
9115 * tree-ssa-tail-merge.c (stmt_local_def): Copy gimple_is_call
9116 test with comment from bb_no_side_effects_p.
9118 2018-06-21 Richard Biener <rguenther@suse.de>
9120 PR tree-optimization/86232
9121 * tree-ssa-loop-niter.c (number_of_iterations_popcount): Adjust
9122 max for constant niter.
9124 2018-06-21 Andre Vieira <andre.simoesdiasvieira@arm.com>
9126 * config/aarch64/aarch64-simd.md
9127 (*aarch64_crypto_aes<aes_op>v16qi_xor_combine): New.
9129 2018-06-21 Andre Vieira <andre.simoesdiasvieira@arm.com>
9131 * config/aarch64/aarch64-simd.md (aarch64_crypto_aes<aes_op>v16qi):
9132 Make opernads of the unspec commutative.
9134 2018-06-21 Richard Biener <rguenther@suse.de>
9136 * tree-data-ref.c (dr_step_indicator): Handle NULL DR_STEP.
9137 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
9138 Avoid calling vect_mark_for_runtime_alias_test with gathers or scatters.
9139 (vect_analyze_data_ref_dependence): Re-order checks to deal with
9141 (vect_record_base_alignments): Do not record base alignment
9142 for gathers or scatters.
9143 (vect_compute_data_ref_alignment): Drop return value that is always
9144 true. Bail out early for gathers or scatters.
9145 (vect_enhance_data_refs_alignment): Bail out early for gathers
9147 (vect_find_same_alignment_drs): Likewise.
9148 (vect_analyze_data_refs_alignment): Remove dead code.
9149 (vect_slp_analyze_and_verify_node_alignment): Likewise.
9150 (vect_analyze_data_refs): For possible gathers or scatters do
9151 not create an alternate DR, just check their possible validity
9152 and mark them. Adjust DECL_NONALIASED handling to not rely
9154 * tree-vect-loop-manip.c (vect_update_inits_of_drs): Do not
9155 update inits of gathers or scatters.
9156 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern):
9157 Also copy gather/scatter flag to pattern vinfo.
9159 2018-06-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
9161 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Change
9162 behavior of vec_packsu (vector unsigned long long, vector unsigned
9163 long long) to match behavior of vec_packs with same signature.
9165 2018-06-20 Chung-Lin Tang <cltang@codesourcery.com>
9166 Thomas Schwinge <thomas@codesourcery.com>
9167 Cesar Philippidis <cesar@codesourcery.com>
9169 * gimplify.c (gimplify_scan_omp_clauses): Add support for
9170 OMP_CLAUSE_{IF_PRESENT,FINALIZE}.
9171 (gimplify_adjust_omp_clauses): Likewise.
9172 (gimplify_oacc_declare_1): Add support for GOMP_MAP_RELEASE, remove
9173 support for GOMP_MAP_FORCE_{ALLOC,TO,FROM,TOFROM}.
9174 (gimplify_omp_target_update): Update handling of acc update and
9176 * omp-low.c (install_var_field): Remove unused parameter
9177 base_pointers_restrict.
9178 (scan_sharing_clauses): Remove base_pointers_restrict parameter.
9179 Update call to install_var_field. Handle OMP_CLAUSE_{IF_PRESENT,
9181 (omp_target_base_pointers_restrict_p): Delete.
9182 (scan_omp_target): Update call to scan_sharing_clauses.
9183 * tree-core.h (enum omp_clause_code): Add OMP_CLAUSE_{IF_PRESENT,
9185 * tree-nested.c (convert_nonlocal_omp_clauses): Handle
9186 OMP_CLAUSE_{IF_PRESENT,FINALIZE}.
9187 (convert_local_omp_clauses): Likewise.
9188 * tree-pretty-print.c (dump_omp_clause): Likewise.
9189 * tree.c (omp_clause_num_ops): Add entries for OMP_CLAUSE_{IF_PRESENT,
9191 (omp_clause_code_name): Likewise.
9193 2018-06-20 Jakub Jelinek <jakub@redhat.com>
9196 * var-tracking.c (use_narrower_mode_test): Check if shift amount can
9199 PR tree-optimization/86231
9200 * tree-vrp.c (union_ranges): For ( [ ) ] or ( )[ ] range and
9201 anti-range don't overwrite *vr0min before using it to compute *vr0max.
9203 2018-06-20 Tom de Vries <tdevries@suse.de>
9205 PR tree-optimization/86097
9206 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Also convert *nit to
9207 iv type if signedness of iv type is not the same as that of *nit.
9209 2018-06-20 Jakub Jelinek <jakub@redhat.com>
9211 * cfgrtl.c (rtl_verify_edges): Formatting fix. If bb->preds has any
9212 EDGE_EH edges, verify they are all EDGE_EH.
9214 2018-06-20 Maya Rashish <coypu@sdf.org>
9216 * ginclude/stddef.h: Limit #include <machine/ansi.h> to NetBSD.
9218 2018-06-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
9220 * config/aarch64/aarch64-tuning-flags.def (no_ldp_stp_qregs): New.
9221 * config/aarch64/aarch64.c (xgene1_tunings): Add
9222 AARCH64_EXTRA_TUNE_NO_LDP_STP_QREGS to tune_flags.
9223 (aarch64_mode_valid_for_sched_fusion_p):
9224 Allow 16-byte modes.
9225 (aarch64_classify_address): Allow 16-byte modes for load_store_pair_p.
9226 * config/aarch64/aarch64-ldpstp.md: Add peepholes for LDP STP of
9228 * config/aarch64/aarch64-simd.md (load_pair<VQ:mode><VQ2:mode>):
9230 (vec_store_pair<VQ:mode><VQ2:mode>): Likewise.
9231 * config/aarch64/iterators.md (VQ2): New mode iterator.
9233 2018-06-20 Martin Liska <mliska@suse.cz>
9235 * tree-switch-conversion.c (jump_table_cluster::can_be_handled):
9236 Change default ratio from 10 to 8.
9238 2018-06-20 Martin Liska <mliska@suse.cz>
9240 * tree-switch-conversion.c (jump_table_cluster::find_jump_tables):
9242 (bit_test_cluster::find_bit_tests): Likewise.
9243 (switch_decision_tree::analyze_switch_statement): Find clusters.
9244 * tree-switch-conversion.h (struct jump_table_cluster): Document
9247 2018-06-20 Martin Liska <mliska@suse.cz>
9249 * tree-switch-conversion.c (switch_conversion::collect):
9250 Record m_uniq property.
9251 (switch_conversion::expand): Bail out for special conditions.
9252 (group_cluster::~group_cluster): New.
9253 (group_cluster::group_cluster): Likewise.
9254 (group_cluster::dump): Likewise.
9255 (jump_table_cluster::emit): New.
9256 (switch_decision_tree::fix_phi_operands_for_edges): New.
9257 (struct case_node): Remove struct.
9258 (jump_table_cluster::can_be_handled): New.
9259 (case_values_threshold): Moved to header.
9260 (reset_out_edges_aux): Likewise.
9261 (jump_table_cluster::is_beneficial): New.
9262 (bit_test_cluster::can_be_handled): Likewise.
9263 (add_case_node): Remove.
9264 (bit_test_cluster::is_beneficial): New.
9265 (case_bit_test::cmp): New.
9266 (bit_test_cluster::emit): New.
9267 (expand_switch_as_decision_tree_p): Remove.
9268 (bit_test_cluster::hoist_edge_and_branch_if_true): New.
9269 (fix_phi_operands_for_edge): Likewise.
9270 (switch_decision_tree::analyze_switch_statement): New.
9271 (compute_cases_per_edge): Move ...
9272 (switch_decision_tree::compute_cases_per_edge): ... here.
9273 (try_switch_expansion): Likewise.
9274 (switch_decision_tree::try_switch_expansion): Likewise.
9275 (record_phi_operand_mapping): Likewise.
9276 (switch_decision_tree::record_phi_operand_mapping): Likewise.
9277 (emit_case_decision_tree): Likewise.
9278 (switch_decision_tree::emit): Likewise.
9279 (balance_case_nodes): Likewise.
9280 (switch_decision_tree::balance_case_nodes): Likewise.
9281 (dump_case_nodes): Likewise.
9282 (switch_decision_tree::dump_case_nodes): Likewise.
9283 (emit_jump): Likewise.
9284 (switch_decision_tree::emit_jump): Likewise.
9285 (emit_cmp_and_jump_insns): Likewise.
9286 (switch_decision_tree::emit_cmp_and_jump_insns): Likewise.
9287 (emit_case_nodes): Likewise.
9288 (switch_decision_tree::emit_case_nodes): Likewise.
9289 (conditional_probability): Remove.
9290 * tree-switch-conversion.h (enum cluster_type): New.
9292 (struct cluster): Likewise.
9293 (cluster::cluster): Likewise.
9294 (struct simple_cluster): Likewise.
9295 (simple_cluster::simple_cluster): Likewise.
9296 (struct group_cluster): Likewise.
9297 (struct jump_table_cluster): Likewise.
9298 (struct bit_test_cluster): Likewise.
9299 (struct min_cluster_item): Likewise.
9300 (struct case_tree_node): Likewise.
9301 (case_tree_node::case_tree_node): Likewise.
9302 (jump_table_cluster::case_values_threshold): Likewise.
9303 (struct case_bit_test): Likewise.
9304 (struct switch_decision_tree): Likewise.
9305 (struct switch_conversion): Likewise.
9306 (switch_decision_tree::reset_out_edges_aux): Likewise.
9308 2018-06-20 Martin Liska <mliska@suse.cz>
9310 * tree-switch-conversion.c (MAX_CASE_BIT_TESTS): Remove.
9311 (hoist_edge_and_branch_if_true): Likewise.
9312 (expand_switch_using_bit_tests_p): Likewise.
9313 (struct case_bit_test): Likewise.
9314 (case_bit_test_cmp): Likewise.
9315 (emit_case_bit_tests): Likewise.
9316 (switch_conversion::switch_conversion): New class.
9317 (struct switch_conv_info): Remove old struct.
9318 (collect_switch_conv_info): More to ...
9319 (switch_conversion::collect): ... this.
9320 (check_range): Likewise.
9321 (switch_conversion::check_range): Likewise.
9322 (check_all_empty_except_final): Likewise.
9323 (switch_conversion::check_all_empty_except_final): Likewise.
9324 (check_final_bb): Likewise.
9325 (switch_conversion::check_final_bb): Likewise.
9326 (create_temp_arrays): Likewise.
9327 (switch_conversion::create_temp_arrays): Likewise.
9328 (free_temp_arrays): Likewise.
9329 (gather_default_values): Likewise.
9330 (switch_conversion::gather_default_values): Likewise.
9331 (build_constructors): Likewise.
9332 (switch_conversion::build_constructors): Likewise.
9333 (constructor_contains_same_values_p): Likewise.
9334 (switch_conversion::contains_same_values_p): Likewise.
9335 (array_value_type): Likewise.
9336 (switch_conversion::array_value_type): Likewise.
9337 (build_one_array): Likewise.
9338 (switch_conversion::build_one_array): Likewise.
9339 (build_arrays): Likewise.
9340 (switch_conversion::build_arrays): Likewise.
9341 (gen_def_assigns): Likewise.
9342 (switch_conversion::gen_def_assigns): Likewise.
9343 (prune_bbs): Likewise.
9344 (switch_conversion::prune_bbs): Likewise.
9345 (fix_phi_nodes): Likewise.
9346 (switch_conversion::fix_phi_nodes): Likewise.
9347 (gen_inbound_check): Likewise.
9348 (switch_conversion::gen_inbound_check): Likewise.
9349 (process_switch): Use the newly created class.
9350 (switch_conversion::expand): New.
9351 (switch_conversion::~switch_conversion): New.
9352 * tree-switch-conversion.h: New file.
9354 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
9356 * tree-vectorizer.h (NUM_PATTERNS, vect_recog_func_ptr): Move to
9357 tree-vect-patterns.c.
9358 * tree-vect-patterns.c (vect_supportable_direct_optab_p): New function.
9359 (vect_recog_dot_prod_pattern): Use it. Remove the type_in argument.
9360 (vect_recog_sad_pattern): Likewise.
9361 (vect_recog_widen_sum_pattern): Likewise.
9362 (vect_recog_pow_pattern): Likewise. Check for a null vectype.
9363 (vect_recog_widen_shift_pattern): Remove the type_in argument.
9364 (vect_recog_rotate_pattern): Likewise.
9365 (vect_recog_mult_pattern): Likewise.
9366 (vect_recog_vector_vector_shift_pattern): Likewise.
9367 (vect_recog_divmod_pattern): Likewise.
9368 (vect_recog_mixed_size_cond_pattern): Likewise.
9369 (vect_recog_bool_pattern): Likewise.
9370 (vect_recog_mask_conversion_pattern): Likewise.
9371 (vect_try_gather_scatter_pattern): Likewise.
9372 (vect_recog_widen_mult_pattern): Likewise. Check for a null vectype.
9373 (vect_recog_over_widening_pattern): Likewise.
9374 (vect_recog_gather_scatter_pattern): Likewise.
9375 (vect_recog_func_ptr): Move from tree-vectorizer.h
9376 (vect_vect_recog_func_ptrs): Move further down the file.
9377 (vect_recog_func): Likewise. Remove the third argument.
9378 (NUM_PATTERNS): Define based on vect_vect_recog_func_ptrs.
9379 (vect_pattern_recog_1): Expect the pattern function to do any
9380 necessary target tests. Also expect it to provide a vector type.
9381 Remove the type_in handling.
9383 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
9385 * tree-vect-patterns.c (vect_pattern_detected): New function.
9386 (vect_recog_dot_prod_patternm, vect_recog_sad_pattern)
9387 (vect_recog_widen_mult_pattern, vect_recog_widen_sum_pattern)
9388 (vect_recog_over_widening_pattern, vect_recog_widen_shift_pattern
9389 (vect_recog_rotate_pattern, vect_recog_vector_vector_shift_pattern)
9390 (vect_recog_mult_pattern, vect_recog_divmod_pattern)
9391 (vect_recog_mixed_size_cond_pattern, vect_recog_bool_pattern)
9392 (vect_recog_mask_conversion_pattern)
9393 (vect_try_gather_scatter_pattern): Likewise.
9395 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
9397 * tree-vect-patterns.c (vect_get_internal_def): New function.
9398 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern)
9399 (vect_recog_vector_vector_shift_pattern, check_bool_pattern)
9400 (search_type_for_mask_1): Use it.
9402 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
9404 * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Remove
9405 redundant WIDEN_SUM_EXPR handling.
9406 (vect_recog_sad_pattern): Likewise.
9408 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
9410 * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Remove
9411 redundant check that the types of a PLUS_EXPR or MULT_EXPR agree.
9412 (vect_recog_sad_pattern): Likewise PLUS_EXPR, ABS_EXPR and MINUS_EXPR.
9413 (vect_recog_widen_mult_pattern): Likewise MULT_EXPR.
9414 (vect_recog_widen_sum_pattern): Likewise PLUS_EXPR.
9416 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
9418 * tree-vect-stmts.c (vectorizable_call): Make sure that we
9419 use the stmt_vec_info of the original bb statement for the
9420 new zero assignment, even if the call is part of a pattern.
9422 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
9424 * tree-vectorizer.h (_stmt_vec_info): Note above pattern_def_seq
9425 that the sequence is attached to the original statement rather
9426 than the pattern statement.
9427 * tree-vect-loop.c (vect_determine_vf_for_stmt): Take the
9428 PATTERN_DEF_SEQ from the original statement rather than
9429 the main pattern statement.
9430 * tree-vect-stmts.c (free_stmt_vec_info): Likewise.
9431 * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Likewise.
9432 (vect_mark_pattern_stmts): Don't copy the PATTERN_DEF_SEQ.
9434 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
9436 * tree-vect-stmts.c (vect_analyze_stmt): Move the handling of pattern
9437 definition statements before the early exit for statements that aren't
9439 * tree-vect-loop.c (vect_transform_loop_stmt): New function,
9441 (vect_transform_loop): ...here. Process pattern definition
9442 statements without first checking whether the main pattern
9443 statement is live or relevant.
9445 2018-06-19 Eric Botcazou <ebotcazou@adacore.com>
9447 * tree-cfgcleanup.c (tree_forwarder_block_p): Do not return false at
9448 -O0 if the locus represent UNKNOWN_LOCATION but have different values.
9450 2018-06-19 Aaron Sawdey <acsawdey@linux.ibm.com>
9452 * config/rs6000/rs6000-string.c (select_block_compare_mode): Check
9453 TARGET_EFFICIENT_OVERLAPPING_UNALIGNED here instead of in caller.
9454 (do_and3, do_and3_mask, do_compb3, do_rotl3): New functions.
9455 (expand_block_compare): Change select_block_compare_mode call.
9456 (expand_strncmp_align_check): Use new functions, fix comment.
9457 (emit_final_str_compare_gpr): New function.
9458 (expand_strn_compare): Refactor and clean up code.
9459 * config/rs6000/vsx.md (vsx_mov<mode>_64bit): Remove *.
9461 2018-06-19 Tony Reix <tony.reix@atos.com>
9462 Damien Bergamini <damien.bergamini@atos.com>
9463 David Edelsohn <dje.gcc@gmail.com>
9465 * collect2.c (static_obj): New variable.
9466 (static_libs): New variable.
9467 (is_in_list): Uncomment declaration.
9468 (main): Track AIX libraries linked statically.
9469 (is_in_list): Uncomment definition.
9470 (scan_prog_file): Don't add AIX shared libraries initializer
9471 to constructor list if linking statically.
9473 2018-06-19 Max Filippov <jcmvbkbc@gmail.com>
9475 * config/xtensa/xtensa.md (UNSPEC_FRAME_BLOCKAGE): New unspec
9477 (allocate_stack, frame_blockage, *frame_blockage): New patterns.
9479 2018-06-19 Jan Hubicka <hubicka@ucw.cz>
9481 * tree.c (find_decls_types_r): Remove all non-VAR_DECLs from
9484 2018-06-19 Martin Liska <mliska@suse.cz>
9486 * config/i386/i386.c (ix86_can_inline_p): Do not use
9487 ipa_fn_summaries::get_create.
9488 * ipa-cp.c (ipcp_cloning_candidate_p): Replace get_create with
9490 (devirtualization_time_bonus): Likewise.
9491 (ipcp_propagate_stage): Likewise.
9492 * ipa-fnsummary.c (redirect_to_unreachable): Likewise.
9493 (edge_set_predicate): Likewise.
9494 (evaluate_conditions_for_known_args): Likewise.
9495 (evaluate_properties_for_edge): Likewise.
9496 (ipa_call_summary::reset): Tranform to ...
9497 (ipa_call_summary::~ipa_call_summary): ... this.
9498 (ipa_fn_summary::reset): Transform to ...
9499 (ipa_fn_summary::~ipa_fn_summary): ... this.
9500 (ipa_fn_summary_t::remove): Rename to ...
9501 (ipa_fn_summary_t::remove_callees): ... this.
9502 (ipa_fn_summary_t::duplicate): Use placement new
9503 instead of memory copy.
9504 (ipa_call_summary_t::duplicate): Likewise.
9505 (ipa_call_summary_t::remove): Remove.
9506 (dump_ipa_call_summary): Change get_create to get.
9507 (ipa_dump_fn_summary): Dump only when summary exists.
9508 (analyze_function_body): Use symbol_summary::get instead
9510 (compute_fn_summary): Likewise.
9511 (estimate_edge_devirt_benefit): Likewise.
9512 (estimate_edge_size_and_time): Likewise.
9513 (inline_update_callee_summaries): Likewise.
9514 (remap_edge_change_prob): Likewise.
9515 (remap_edge_summaries): Likewise.
9516 (ipa_merge_fn_summary_after_inlining): Likewise.
9517 (write_ipa_call_summary): Likewise.
9518 (ipa_fn_summary_write): Likewise.
9519 (ipa_free_fn_summary): Likewise.
9520 * ipa-fnsummary.h (struct GTY): Add new ctor and copy ctor.
9521 (struct ipa_call_summary): Likewise.
9522 * ipa-icf.c (sem_function::merge): Use symbol_summary::get instead
9524 * ipa-inline-analysis.c (do_estimate_edge_time): Likewise.
9525 (estimate_size_after_inlining): Likewise.
9526 (estimate_growth): Likewise.
9527 (growth_likely_positive): Likewise.
9528 * ipa-inline-transform.c (clone_inlined_nodes): Likewise.
9529 (inline_call): Likewise.
9530 * ipa-inline.c (caller_growth_limits): Likewise.
9531 (can_inline_edge_p): Likewise.
9532 (can_inline_edge_by_limits_p): Likewise.
9533 (compute_uninlined_call_time): Likewise.
9534 (compute_inlined_call_time): Likewise.
9535 (want_inline_small_function_p): Likewise.
9536 (edge_badness): Likewise.
9537 (update_caller_keys): Likewise.
9538 (update_callee_keys): Likewise.
9539 (inline_small_functions): Likewise.
9540 (inline_to_all_callers_1): Likewise.
9541 (dump_overall_stats): Likewise.
9542 (early_inline_small_functions): Likewise.
9543 (early_inliner): Likewise.
9544 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
9545 * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
9546 * ipa-pure-const.c (malloc_candidate_p): Likewise.
9547 * ipa-split.c (execute_split_functions): Likewise.
9548 * symbol-summary.h: Likewise.
9549 * tree-sra.c (ipa_sra_preliminary_function_checks): Likewise.
9551 2018-06-19 Richard Biener <rguenther@suse.de>
9553 * tree-vectorizer.c (try_vectorize_loop_1): Split out of ...
9554 (vectorize_loops): ... here. Fix dbgcnt handling.
9555 (try_vectorize_loop): Wrap try_vectorize_loop_1.
9557 2018-06-19 Segher Boessenkool <segher@kernel.crashing.org>
9560 * config/rs6000/rs6000.md (rs6000_discover_homogeneous_aggregate): An
9561 ieee128 argument takes up only one (vector) register, not two (floating
9564 2018-06-19 Eric Botcazou <ebotcazou@adacore.com>
9566 * gimplify.c (gimplify_init_constructor): Really never clear for an
9567 incomplete constructor if CONSTRUCTOR_NO_CLEARING is set.
9569 2018-06-19 Richard Biener <rguenther@suse.de>
9571 PR tree-optimization/86179
9572 * tree-vect-patterns.c (vect_pattern_recog_1): Clean up
9573 after failed recognition.
9575 2018-06-18 Martin Sebor <msebor@redhat.com>
9578 * calls.c (maybe_warn_nonstring_arg): Handle strncat.
9579 * tree-ssa-strlen.c (is_strlen_related_p): Make extern.
9580 Handle integer subtraction.
9581 (maybe_diag_stxncpy_trunc): Handle nonstring source arguments.
9582 * tree-ssa-strlen.h (is_strlen_related_p): Declare.
9584 2018-06-18 David Malcolm <dmalcolm@redhat.com>
9586 * config/frv/frv-protos.h (frv_ifcvt_modify_insn): Strengthen 3rd
9587 param from rtx to rtx_insn *.
9588 * config/frv/frv.c (frv_ifcvt_add_insn): Likewise for "insn"
9590 (frv_ifcvt_modify_insn): Likwise.
9591 (frv_ifcvt_modify_final): Likwise for local "existing_insn",
9592 adding an as_a <rtx_insn *> cast. Likewise for local "insn".
9593 * config/mips/mips.c (r10k_insert_cache_barriers): Add an
9594 as_a <rtx_insn *> cast to local "unprotected_region" once
9595 it's been established that it's not NULL or pc_rtx.
9596 * config/nds32/nds32-relax-opt.c (nds32_group_insns): Strengthen
9597 param "sethi" from rtx to rtx_insn *.
9598 (nds32_group_float_insns): Likewise for param "insn".
9599 * config/vax/vax-protos.h (vax_output_int_add): Likewise for 1st
9601 (vax_output_int_subtract): Likewise.
9602 * config/vax/vax.c (vax_output_int_add): Likewise for param
9604 (vax_output_int_subtract): Likewise.
9605 * emit-rtl.c (set_insn_deleted): Likewise, removing cast.
9606 (emit_pattern_after): Likewise for param "after".
9607 (emit_insn_after): Likewise.
9608 (emit_jump_insn_after): Likewise.
9609 (emit_call_insn_after): Likewise.
9610 (emit_debug_insn_after): Likewise.
9611 (emit_pattern_before): Likewise for param "before".
9612 (emit_insn_before): Likewise.
9613 (emit_jump_insn_before): Likewise.
9614 * final.c (get_insn_template): Likewise for param "insn", removing
9616 * output.h (get_insn_template): Likewise for 2nd param.
9617 * rtl.h (emit_insn_before): Likewise.
9618 (emit_jump_insn_before): Likewise.
9619 (emit_debug_insn_before_noloc): Likewise.
9620 (emit_insn_after): Likewise.
9621 (emit_jump_insn_after): Likewise.
9622 (emit_call_insn_after): Likewise.
9623 (emit_debug_insn_after): Likewise.
9624 (set_insn_deleted): Likewise for param.
9626 2018-06-18 Michael Meissner <meissner@linux.ibm.com>
9629 * config/rs6000/rs6000-modes.def (toplevel): Rework the 128-bit
9630 floating point modes, so that IFmode is numerically greater than
9631 TFmode, which is greater than KFmode using FRACTIONAL_FLOAT_MODE
9632 to declare the ordering. This prevents IFmode from being
9633 converted to TFmode when long double is IEEE 128-bit on an ISA 3.0
9634 machine. Include rs6000-modes.h to share the fractional values
9635 between genmodes* and the rest of the compiler.
9639 * config/rs6000/rs6000-modes.h: New file.
9640 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Change the
9641 meaning of rs6000_long_double_size so that 126..128 selects an
9642 appropriate 128-bit floating point type.
9643 (rs6000_option_override_internal): Likewise.
9644 * config/rs6000/rs6000.h (toplevel): Include rs6000-modes.h.
9645 (TARGET_LONG_DOUBLE_128): Change the meaning of
9646 rs6000_long_double_size so that 126..128 selects an appropriate
9647 128-bit floating point type.
9648 (LONG_DOUBLE_TYPE_SIZE): Update comment.
9649 * config/rs6000/rs6000.md (trunciftf2): Correct the modes of the
9650 source and destination to match the standard usage.
9651 (truncifkf2): Likewise.
9652 (copysign<mode>3, IEEE iterator): Rework copysign of float128 on
9653 ISA 2.07 to use an explicit clobber, instead of passing in a
9655 (copysign<mode>3_soft): Likewise.
9657 2018-06-18 David Malcolm <dmalcolm@redhat.com>
9659 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences):
9660 Replace dump_printf_loc call with DUMP_VECT_SCOPE.
9661 (vect_slp_analyze_instance_dependence): Likewise.
9662 (vect_enhance_data_refs_alignment): Likewise.
9663 (vect_analyze_data_refs_alignment): Likewise.
9664 (vect_slp_analyze_and_verify_instance_alignment
9665 (vect_analyze_data_ref_accesses): Likewise.
9666 (vect_prune_runtime_alias_test_list): Likewise.
9667 (vect_analyze_data_refs): Likewise.
9668 * tree-vect-loop-manip.c (vect_update_inits_of_drs): Likewise.
9669 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
9670 (vect_analyze_scalar_cycles_1): Likewise.
9671 (vect_get_loop_niters): Likewise.
9672 (vect_analyze_loop_form_1): Likewise.
9673 (vect_update_vf_for_slp): Likewise.
9674 (vect_analyze_loop_operations): Likewise.
9675 (vect_analyze_loop): Likewise.
9676 (vectorizable_induction): Likewise.
9677 (vect_transform_loop): Likewise.
9678 * tree-vect-patterns.c (vect_pattern_recog): Likewise.
9679 * tree-vect-slp.c (vect_analyze_slp): Likewise.
9680 (vect_make_slp_decision): Likewise.
9681 (vect_detect_hybrid_slp): Likewise.
9682 (vect_slp_analyze_operations): Likewise.
9683 (vect_slp_bb): Likewise.
9684 * tree-vect-stmts.c (vect_mark_stmts_to_be_vectorized): Likewise.
9685 (vectorizable_bswap): Likewise.
9686 (vectorizable_call): Likewise.
9687 (vectorizable_simd_clone_call): Likewise.
9688 (vectorizable_conversion): Likewise.
9689 (vectorizable_assignment): Likewise.
9690 (vectorizable_shift): Likewise.
9691 (vectorizable_operation): Likewise.
9692 * tree-vectorizer.h (DUMP_VECT_SCOPE): New macro.
9694 2018-06-18 Martin Sebor <msebor@redhat.com>
9696 PR tree-optimization/81384
9697 * builtin-types.def (BT_FN_SIZE_CONST_STRING_SIZE): New.
9698 * builtins.c (expand_builtin_strnlen): New function.
9699 (expand_builtin): Call it.
9700 (fold_builtin_n): Avoid setting TREE_NO_WARNING.
9701 * builtins.def (BUILT_IN_STRNLEN): New.
9702 * calls.c (maybe_warn_nonstring_arg): Handle BUILT_IN_STRNLEN.
9703 Warn for bounds in excess of maximum object size.
9704 * tree-ssa-strlen.c (maybe_set_strlen_range): Return tree representing
9705 single-value ranges. Handle strnlen.
9706 (handle_builtin_strlen): Handle strnlen.
9707 (strlen_check_and_optimize_stmt): Same.
9708 * doc/extend.texi (Other Builtins): Document strnlen.
9710 2018-06-18 Maya Rashish <coypu@sdf.org>
9712 * config/alpha/openbsd.h (TARGET_DEFAULT): Define.
9713 (LINK_SPEC, STARTFILE_SPEC, ENDFILE_SPEC): Likewise.
9714 (INTMAX_TYPE, UINTMAX_TYPE, WINT_TYPE): Likewise.
9716 * config/alpha/elf.h (STARTFILE_SPEC, ENDFILE_SPEC): Move from
9718 * config/alpha/linux.h (STARTFILE_SPEC, ENDFILE_SPEC): Here.
9720 2018-06-18 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
9722 * tree.c (escaped_string::escape): Replace cast to char * by
9723 const_cast<char *> (unescaped).
9725 2018-06-18 Nick Clifton <nickc@redhat.com>
9728 * tree.c (escaped_string): New class. Converts an unescaped
9729 string into its escaped equivalent.
9730 (warn_deprecated_use): Use the new class to convert the
9731 deprecation message, if present.
9732 (test_escaped_strings): New self test.
9733 (test_c_tests): Add test_escaped_strings.
9734 * doc/extend.texi (deprecated): Add a note that the
9735 deprecation message is affected by the -fmessage-length
9736 option, and that control characters will be escaped.
9737 (#pragma GCC error): Document this pragma.
9738 (#pragma GCC warning): Likewise.
9739 * doc/invoke.texi (-fmessage-length): Document this option's
9740 effect on the #warning and #error preprocessor directives and
9741 the deprecated attribute.
9743 2018-06-18 Eric Botcazou <ebotcazou@adacore.com>
9745 * tree.c (decl_value_expr_lookup): Revert latest change.
9746 (decl_value_expr_insert): Likewise.
9748 2018-06-17 Eric Botcazou <ebotcazou@adacore.com>
9750 * gimplify.c (nonlocal_vlas): Delete.
9751 (nonlocal_vla_vars): Likewise.
9752 (gimplify_var_or_parm_decl): Do not add debug VAR_DECLs for non-local
9754 (gimplify_body): Do not create and destroy nonlocal_vlas.
9755 * tree-nested.c: Include diagnostic.h.
9756 (use_pointer_in_frame): Tweak.
9757 (lookup_field_for_decl): Add assertion and declare the transformation.
9758 (convert_nonlocal_reference_op) <PARM_DECL>: Rework and issue an
9759 internal error when the reference is in a wrong context. Do not
9760 create a debug decl by default.
9761 (note_nonlocal_block_vlas): Delete.
9762 (convert_nonlocal_reference_stmt) <GIMPLE_BIND>: Do not call it.
9763 (convert_local_reference_op) <PARM_DECL>: Skip the frame decl. Do not
9764 create a debug decl by default.
9765 (convert_gimple_call) <GIMPLE_CALL>: Issue an internal error when the
9766 call is in a wrong context.
9767 (fixup_vla_decls): New function.
9768 (finalize_nesting_tree_1): Adjust comment. Call fixup_vla_decls if no
9769 debug variables were created.
9770 * tree.c (decl_value_expr_lookup): Add checking assertion.
9771 (decl_value_expr_insert): Likewise.
9773 2018-06-16 Kugan Vivekanandarajah <kuganv@linaro.org>
9776 * ipa-fnsummary.c (will_be_nonconstant_expr_predicate): Handle CALL_EXPR.
9777 * tree-scalar-evolution.c (interpret_expr): Likewise.
9778 (expression_expensive_p): Likewise.
9779 * tree-ssa-loop-ivopts.c (contains_abnormal_ssa_name_p): Likewise.
9780 * tree-ssa-loop-niter.c (number_of_iterations_popcount): New.
9781 (number_of_iterations_exit_assumptions): Use number_of_iterations_popcount.
9782 (ssa_defined_by_minus_one_stmt_p): New.
9784 2018-06-16 Kugan Vivekanandarajah <kuganv@linaro.org>
9787 * cfgexpand.c (expand_debug_expr): Hande ABSU_EXPR.
9788 * config/i386/i386.c (ix86_add_stmt_cost): Likewise.
9789 * dojump.c (do_jump): Likewise.
9790 * expr.c (expand_expr_real_2): Check operand type's sign.
9791 * fold-const.c (const_unop): Handle ABSU_EXPR.
9792 (fold_abs_const): Likewise.
9793 * gimple-pretty-print.c (dump_unary_rhs): Likewise.
9794 * gimple-ssa-backprop.c (backprop::process_assign_use): Likesie.
9795 (strip_sign_op_1): Likesise.
9796 * match.pd: Add new pattern to generate ABSU_EXPR.
9797 * optabs-tree.c (optab_for_tree_code): Handle ABSU_EXPR.
9798 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
9799 * tree-eh.c (operation_could_trap_helper_p): Likewise.
9800 * tree-inline.c (estimate_operator_cost): Likewise.
9801 * tree-pretty-print.c (dump_generic_node): Likewise.
9802 * tree-vect-patterns.c (vect_recog_sad_pattern): Likewise.
9803 * tree.def (ABSU_EXPR): New.
9805 2018-06-16 Jakub Jelinek <jakub@redhat.com>
9808 * common.opt (Wunsafe-loop-optimizations): Add Ignore, remove Var,
9809 documented as preserved for backward compatibility only.
9810 * doc/invoke.texi: Remove -Wunsafe-loop-optimizations documentation.
9812 PR rtl-optimization/86108
9813 * bb-reorder.c (create_forwarder_block): Renamed to ...
9814 (create_eh_forwarder_block): ... this. Split OLD_BB after labels and
9815 jump from new landing pad to the second part.
9816 (sjlj_fix_up_crossing_landing_pad, dw2_fix_up_crossing_landing_pad):
9819 2018-06-15 Jakub Jelinek <jakub@redhat.com>
9822 * expr.c (expand_assignment): Remove now redundant COMPLEX_MODE_P
9823 check from first store_expr, use to_mode instead of GET_MODE (to_rtx).
9824 Only call store_expr for halves if the mode is the same.
9827 * match.pd ((X / Y) == 0 -> X < Y): Don't transform complex divisions.
9828 Fix up comment formatting.
9830 2018-06-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
9832 * typed-splay-tree.h (typed_splay_tree::remove): New function.
9833 (typed_splay_tree::closure,
9834 typed_splay_tree::inner_foreach_fn, typed_splay_tree::m_inner): Deleted.
9835 (typed_splay_tree::typed_splay_tree,
9836 typed_splay_tree::operator =): Declared private.
9837 (typed_splay_tree::splay_tree_key, typed_splay_tree::splay_tree_value,
9838 typed_splay_tree::splay_tree_node_s, typed_splay_tree::KDEL,
9839 typed_splay_tree::VDEL, typed_splay_tree::splay_tree_delete_helper,
9840 typed_splay_tree::rotate_left, typed_splay_tree::rotate_right,
9841 typed_splay_tree::splay_tree_splay,
9842 typed_splay_tree::splay_tree_foreach_helper,
9843 typed_splay_tree::splay_tree_insert,
9844 typed_splay_tree::splay_tree_remove,
9845 typed_splay_tree::splay_tree_lookup,
9846 typed_splay_tree::splay_tree_predecessor,
9847 typed_splay_tree::splay_tree_successor,
9848 typed_splay_tree::splay_tree_min,
9849 typed_splay_tree::splay_tree_max): Took over from splay-tree.c/.h.
9850 (typed_splay_tree::root, typed_splay_tree::comp,
9851 typed_splay_tree::delete_key,
9852 typed_splay_tree::delete_value): New data members.
9853 * typed-splay-tree.c (selftest::test_str_to_int): Add a test for
9854 typed_splay_tree::remove.
9856 2018-06-15 Matthew Fortune <matthew.fortune@mips.com>
9858 * config/mips/mips.h (ASM_SPEC): Pass through -mcrc, -mno-crc,
9859 -mginv and -mno-ginv to the assembler.
9860 * config/mips/mips.opt (-mcrc): New option.
9862 * doc/invoke.text (-mcrc): Document.
9865 2018-06-15 Nick Clifton <nickc@redhat.com>
9868 * tree.c (escaped_string): New class. Converts an unescaped
9869 string into its escaped equivalent.
9870 (warn_deprecated_use): Use the new class to convert the
9871 deprecation message, if present.
9872 (test_escaped_strings): New self test.
9873 (test_c_tests): Add test_escaped_strings.
9874 * doc/extend.texi (deprecated): Add a note that the
9875 deprecation message is affected by the -fmessage-length
9876 option, and that control characters will be escaped.
9877 (#pragma GCC error): Document this pragma.
9878 (#pragma GCC warning): Likewise.
9879 * doc/invoke.texi (-fmessage-length): Document this option's
9880 effect on the #warning and #error preprocessor directives and
9881 the deprecated attribute.
9883 2018-06-15 Richard Biener <rguenther@suse.de>
9885 * tree-vect-slp.c (vect_slp_bb): Dump MSG_OPTIMIZED_LOCATIONS
9886 here, also noting vector size used.
9887 * tree-vectorizer.c (vectorize_loops): Adjust. Note vector
9888 size used in MSG_OPTIMIZED_LOCATIONS dump.
9889 (pass_slp_vectorize::execute): Adjust.
9891 2018-06-15 Claudiu Zissulescu <claziss@synopsys.com>
9894 * config/arc/arc.c (arc_return_address_register): Fix
9897 2018-06-15 Richard Biener <rguenther@suse.de>
9900 * tree-cfg.c (gimplify_build3): Do not strip sign conversions,
9901 leave useless conversion stripping to force_gimple_operand_gsi.
9902 (gimplify_build2): Likewise.
9903 (gimplify_build1): Likewise.
9905 2018-06-15 Richard Biener <rguenther@suse.de>
9908 * tree-cfg.c (move_stmt_op): unshare invariant addresses
9909 before adjusting their block.
9911 2018-06-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
9913 * config.gcc (riscv*-*-elf* | riscv*-*-rtems*): Use custom
9914 multilibs for *-*-rtems*.
9915 * config/riscv/t-rtems: New file.
9917 2018-06-14 Jakub Jelinek <jakub@redhat.com>
9920 * match.pd ((A +- CST1) +- CST2): Punt if last resort
9921 unsigned_type_for returns NULL.
9924 * lower-subreg.c (find_decomposable_subregs): Don't decompose float
9925 subregs of multi-word pseudos unless the float mode has word size.
9927 2018-06-14 Richard Biener <rguenther@suse.de>
9930 * tree-vect-generic.c (build_word_mode_vector_type): Remove
9931 duplicate and harmful type_hash_canon.
9932 * tree.c (type_hash_canon): Assert we didn't find ourselves.
9934 2018-06-14 Richard Biener <rguenther@suse.de>
9937 * tree-ssa-struct-alias.c (create_variable_info_for): Handle
9940 2018-06-14 Sebastian Huber <sebastian.huber@embedded-brains.de>
9942 * config/rtems.h (STDINT_LONG32): Define.
9944 2018-06-13 Matthew Fortune <matthew.fortune@mips.com>
9945 Prachi Godbole <prachi.godbole@imgtec.com>
9947 * config/mips/mips-cpus.def: Define P6600.
9948 * config/mips/mips-tables.opt: Regenerate.
9949 * config/mips/mips.c (mips_ucbranch_type): New enum.
9950 (mips_rtx_cost_data): Add support for P6600.
9951 (mips_issue_rate): Likewise.
9952 (mips_multipass_dfa_lookahead): Likewise.
9953 (mips_avoid_hazard): Likewise.
9954 (mips_reorg_process_insns): Likewise.
9955 (mips_classify_branch_p6600): New function.
9956 * config/mips/mips.h (TUNE_P6600): New define.
9957 (MIPS_ISA_LEVEL_SPEC): Infer mips64r6 from p6600.
9958 (ENABLE_LD_ST_PAIRS): Enable load/store bonding for p6600.
9959 * config/mips/mips.md: Include p6600.md.
9960 (processor): Add p6600.
9961 * config/mips/p6600.md: New file.
9962 * doc/invoke.texi: Add p6600 to supported architectures.
9964 2018-06-13 Martin Sebor <msebor@redhat.com>
9966 PR tree-optimization/86114
9967 * gimple-fold.c (gimple_fold_builtin_strlen): Only handle LHS
9969 * tree-ssa-strlen.c (maybe_set_strlen_range): Same.
9971 2018-06-13 Richard Biener <rguenther@suse.de>
9973 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern):
9974 Properly set vector type of the intermediate stmt.
9975 * tree-vect-stmts.c (vectorizable_operation): The destination
9976 var always has vectype_out type.
9978 2018-06-13 Jeff Law <law@redhat.com>
9980 * config/rl78/rl78.c (move_elim_pass): Use TDF_NONE rather than
9981 integer 0 for argument to print_rtl_with_bb.
9982 (rl78_reorg): Likewise.
9984 2018-06-13 David Malcolm <dmalcolm@redhat.com>
9986 * config/arc/arc.c (hwloop_optimize): Strengthen local "end_label"
9987 from rtx to rtx_insn *.
9988 * config/bfin/bfin.c (hwloop_optimize): Likewise for local
9990 (add_sched_insns_for_speculation): Likewise for local "target",
9991 converting usage of JUMP_LABEL to JUMP_LABEL_AS_INSN.
9992 * config/c6x/c6x.c (reorg_split_calls): Strengthen param "call_labels"
9993 from rtx_insn ** to rtx_code_label **.
9994 (reorg_emit_nops): Likewise.
9995 (c6x_reorg): Likewise for local "call_labels".
9996 * config/sh/sh-protos.h (get_dest_uid): Strengthen 1st param from
9998 * config/sh/sh.c (dump_table): Strengthen local "lab" from rtx to
9999 rtx_code_label *, adding safe_as_a <rtx_code_label *> casts to
10000 the loops over LABEL_REFS.
10001 (fixup_addr_diff_vecs): Add as_a <rtx_insn *> to usage of
10003 (barrier_align): Convert usage of JUMP_LABEL to JUMP_LABEL_AS_INSN.
10004 (get_dest_uid): Strengthen param "label" from rtx to rtx_insn *.
10005 (split_branches): Strengthen local "olabel" from rtx to
10006 rtx_insn *, adding a safe_as_a cast.
10007 * emit-rtl.c (next_real_insn): Strengthen param from "rtx"
10009 (add_insn_after): Likewise for first two params.
10010 (add_insn_before): Likewise.
10011 (remove_insn): Likewise for param.
10012 (emit_pattern_before_noloc): Likewise for second and third params.
10013 (emit_jump_insn_before_noloc): Convert NULL_RTX to NULL.
10014 (emit_call_insn_before_noloc): Likewise.
10015 (emit_debug_insn_before_noloc): Strengthen "before" param from "rtx"
10017 (emit_barrier_before): Likewise.
10018 (emit_label_before): Strengthen "label" param from "rtx" to
10019 "rtx_code_label *". Strengthen "before" param from "rtx" to
10021 (emit_insn_after_1): Strengthen "after" param from "rtx" to
10023 (emit_pattern_after_noloc): Likewise.
10024 (emit_insn_after_noloc): Likewise.
10025 (emit_jump_insn_after_noloc): Likewise.
10026 (emit_call_insn_after_noloc): Likewise.
10027 (emit_debug_insn_after_noloc): Likewise.
10028 (emit_barrier_after): Likewise.
10029 (emit_label_after): Likewise for both params.
10030 (emit_pattern_after_setloc): Likewise for "after" param. Convert
10031 "loc" param from "int" to "location_t".
10032 (emit_insn_after_setloc): Likewise.
10033 (emit_jump_insn_after_setloc): Likewise.
10034 (emit_call_insn_after_setloc): Likewise.
10035 (emit_debug_insn_after_setloc): Likewise.
10036 (emit_pattern_before_setloc): Likewise for "before" param. Convert
10037 "loc" param from "int" to "location_t".
10038 (emit_pattern_before): Convert NULL_RTX to NULL.
10039 (emit_insn_before_setloc): Convert "loc" param from "int" to
10041 (emit_jump_insn_before_setloc): Likewise.
10042 (emit_call_insn_before_setloc): Likewise.
10043 (emit_debug_insn_before_setloc): Strengthen "before" param from rtx to
10044 rtx_insn *. Convert "loc" param from "int" to "location_t".
10045 * rtl.h (emit_insn_before_setloc, emit_jump_insn_before_setloc,
10046 emit_call_insn_before_setloc, emit_debug_insn_before_setloc):
10047 Convert 3rd param from "int" to "location_t".
10048 (emit_barrier_before, emit_barrier_after, next_real_insn):
10049 Strengthen param from rtx to rtx_insn *.
10050 (emit_label_before): Strengthen 1st param from "rtx" to
10051 "rtx_code_label *". Strengthen 2nd param from "rtx" to
10053 (emit_insn_after_noloc, emit_jump_insn_after_noloc,
10054 emit_call_insn_after_noloc, emit_debug_insn_after_noloc):
10055 Strengthen 2nd param from "rtx" to "rtx_insn *".
10056 (emit_insn_after_setloc, emit_jump_insn_after_setloc)
10057 emit_call_insn_after_setloc, emit_debug_insn_after_setloc):
10058 Likewise. Convert 3rd param from "int" to "location_t".
10059 (emit_label_after): Strengthen 1st param from "rtx" to
10060 "rtx_code_label *".
10061 (next_real_insn, remove_insn): Strengthen param from "rtx" to
10063 (add_insn_before, add_insn_after): Strengthen 1st and 2nd params
10064 from "rtx" to "rtx_insn *".
10066 2018-06-13 Jan Hubicka <hubicka@gcc.gnu.org>
10068 * cgraph.c (cgraph_node::get_untransformed_body): Dump function
10069 bodies streamed in with -Q.
10070 * dumpfile.c (dump_files): Add lto-stream-out dump file.
10071 * dumpfile.h (tree_dump_index): Add lto_stream_out.
10072 * gimple-streamer-out.c: Include gimple-pretty-print.h
10073 (output_bb): Dump stmts streamed.
10074 * lto-section-out.c: Include print-tree.h
10075 (lto_begin_section): Dump sections created.
10076 (lto_output_decl_index): Dump decl encoded.
10077 * lto-streamer-out.c: Include print-tree.h
10078 (create_output_block): Dump output block created.
10079 (DFS::DFS_write_tree_body): Dump DFS SCCs streamed.
10080 (output_function): Dump function output.
10081 (output_constructor): Dump constructor streamed.
10082 (write_global_stream): Output indexes encoded.
10083 (produce_asm_for_decls): Dump streams encoded.
10084 * lto-streamer.c (streamer_dump_file): New global var.
10085 * lto-streamer.h (streamer_dump_file): Declare.
10086 * passes.c (ipa_write_summaries): Initialize streamer dump.
10087 * varpool.c (varpool_node::get_constructor): Dump constructors streamed
10090 2018-06-13 Eric Botcazou <ebotcazou@adacore.com>
10093 * config/i386/winnt.c (i386_pe_seh_cold_init): Do not emit negative
10094 offsets for register save directives. Emit a second batch of save
10095 directives, if need be, when the function accesses prior frames.
10097 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
10099 * config/arc/fpu.md (fmasf4): Force operand to register.
10100 (fnmasf4): Likewise.
10102 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
10104 * config/arc/arc-protos.h (arc_pad_return): Remove.
10105 * config/arc/arc.c (machine_function): Remove force_short_suffix
10107 (arc_print_operand): Adjust printing of '&'.
10108 (arc_verify_short): Remove conditional printing of short suffix.
10109 (arc_final_prescan_insn): Remove reference to size_reason.
10110 (pad_return): New function.
10111 (arc_reorg): Call pad_return.
10112 (arc_pad_return): Remove.
10113 (arc_init_machine_status): Remove reference to force_short_suffix.
10114 * config/arc/arc.md (vunspec): Add VUNSPEC_ARC_BLOCKAGE.
10115 (attr length): When attribute iscompact is true force to 2
10116 regardless; in the case of maybe check if we want to force the
10117 instruction to have 4 bytes length.
10118 (nopv): Change it to generate 4 byte long nop as well.
10119 (blockage): New pattern.
10120 (simple_return): Remove call to arc_pad_return.
10121 (p_return_i): Likewise.
10123 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
10125 * config/arc/elf.h (LINK_GCC_C_SEQUENCE_SPEC): Define.
10127 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
10129 * config/arc/builtins.def (SYNC): SYNC instruction is valid on all
10132 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
10134 * config/arc/arc.c (atomic_exchangesi): EX instruction is default
10135 for ARC700 and ARCv2.
10137 2018-06-13 Chenghua Xu <paul.hua.gm@gmail.com>
10140 * config/mips/loongson.md (vec_setv4hi): Gen_lowpart for
10141 operands[2] instead of operands[1].
10144 2018-06-12 Richard Sandiford <richard.sandiford@linaro.org>
10146 * lra-constraints.c (simplify_operand_subreg): In the paradoxical
10147 case, check whether the outer register overlaps an unallocatable
10148 register, not just whether it fits the required class.
10150 2018-06-12 Richard Sandiford <richard.sandiford@linaro.org>
10152 * poly-int.h (can_div_trunc_p): Add new overload in which all values
10154 * alias.c (get_addr): Extend CONST_INT handling to poly_int_rtx_p.
10155 (memrefs_conflict_p): Likewise.
10156 (init_alias_analysis): Likewise.
10157 * cfgexpand.c (expand_debug_expr): Likewise.
10158 * combine.c (combine_simplify_rtx, force_int_to_mode): Likewise.
10159 * cse.c (fold_rtx): Likewise.
10160 * explow.c (adjust_stack, anti_adjust_stack): Likewise.
10161 * expr.c (emit_block_move_hints): Likewise.
10162 (clear_storage_hints, push_block, emit_push_insn): Likewise.
10163 (store_expr_with_bounds, reduce_to_bit_field_precision): Likewise.
10164 (emit_group_load_1): Use rtx_to_poly_int64 for group offsets.
10165 (emit_group_store): Likewise.
10166 (find_args_size_adjust): Use strip_offset. Use rtx_to_poly_int64
10167 to read the PRE/POST_MODIFY increment.
10168 * calls.c (store_one_arg): Use strip_offset.
10169 * rtlanal.c (rtx_addr_can_trap_p_1): Extend CONST_INT handling to
10171 (set_noop_p): Use rtx_to_poly_int64 for the elements selected
10173 * simplify-rtx.c (avoid_constant_pool_reference): Use strip_offset.
10174 (simplify_binary_operation_1): Extend CONST_INT handling to
10176 * var-tracking.c (compute_cfa_pointer): Take a poly_int64 rather
10177 than a HOST_WIDE_INT.
10178 (hard_frame_pointer_adjustment): Change from HOST_WIDE_INT to
10180 (adjust_mems, add_stores): Update accodingly.
10181 (vt_canonicalize_addr): Track polynomial offsets.
10182 (emit_note_insn_var_location): Likewise.
10183 (vt_add_function_parameter): Likewise.
10184 (vt_initialize): Likewise.
10186 2018-06-12 Jeff Law <law@redhat.com>
10188 * config.gcc (alpha*-*-freebsd*): Remove.
10189 * config/alpha/freebsd.h: Remove.
10191 2018-06-12 David Malcolm <dmalcolm@redhat.com>
10194 * spellcheck-tree.c (levenshtein_distance): Rename to...
10195 (get_edit_distance): ...this, and update for underlying renaming.
10196 * spellcheck-tree.h (levenshtein_distance): Rename to...
10197 (get_edit_distance): ...this.
10198 * spellcheck.c (levenshtein_distance): Rename to...
10199 (get_edit_distance): ...this. Convert from Levenshtein distance
10200 to Damerau-Levenshtein distance by supporting transpositions of
10201 adjacent characters. Rename "v1" to "v_next" and "v0" to
10203 (selftest::levenshtein_distance_unit_test_oneway): Rename to...
10204 (selftest::test_edit_distance_unit_test_oneway): ...this, and
10205 update for underlying renaming.
10206 (selftest::levenshtein_distance_unit_test): Rename to...
10207 (selftest::test_get_edit_distance_unit): ...this, and update for
10208 underlying renaming.
10209 (selftest::test_find_closest_string): Add example from PR 69968
10210 where transposition helps
10211 (selftest::test_metric_conditions): Update for renaming.
10212 (selftest::test_metric_conditions): Likewise.
10213 (selftest::spellcheck_c_tests): Likewise.
10214 * spellcheck.h (levenshtein_distance): Rename both overloads to...
10215 (get_edit_distance): ...this.
10216 (best_match::consider): Update for renaming.
10218 2018-06-12 Martin Sebor <msebor@redhat.com>
10220 PR tree-optimization/85259
10221 * builtins.c (compute_objsize): Handle constant offsets.
10222 * gimple-ssa-warn-restrict.c (maybe_diag_offset_bounds): Return
10223 true iff a warning has been issued.
10224 * gimple.h (gimple_nonartificial_location): New function.
10225 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Call
10226 gimple_nonartificial_location and handle -Wno-system-headers.
10227 (handle_builtin_stxncpy): Same.
10229 2018-06-12 Martin Sebor <msebor@redhat.com>
10232 * fold-const.c (operand_equal_p): Handle SAVE_EXPR.
10234 2018-06-12 Will Schmidt <will_schmidt@vnet.ibm.com>
10236 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
10237 BUILTIN_VEC_XST entries for pointer to double and long long.
10239 2018-06-12 H.J. Lu <hongjiu.lu@intel.com>
10242 * config/i386/gnu-user.h (TARGET_THREAD_SPLIT_STACK_OFFSET):
10244 * config/i386/gnu-user64.h (TARGET_THREAD_SPLIT_STACK_OFFSET):
10247 2018-06-12 Martin Liska <mliska@suse.cz>
10249 * doc/options.texi: Document IntegerRange.
10251 2018-06-12 Martin Liska <mliska@suse.cz>
10253 * config/i386/i386.opt: Make MPX-related options as Deprecated.
10254 * opt-functions.awk: Handle Deprecated flag.
10255 * opts-common.c (decode_cmdline_option): Handle cl_deprecated
10257 (read_cmdline_option): Report warning for a deprecated option.
10258 * opts.h (struct cl_option): Add new field cl_deprecated.
10259 (CL_ERR_DEPRECATED): New.
10261 2018-06-12 Martin Liska <mliska@suse.cz>
10263 * doc/options.texi: Document Deprecated option flag.
10265 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
10267 * config/arc/arc-arch.h (arc_extras): New enum.
10268 (arc_cpu_t):Add field extra.
10269 (arc_cpu_types): Consider the extras.
10270 * config/arc/arc-cpus.def: Add extras info.
10271 * config/arc/arc-opts.h (processor_type): Consider extra field.
10272 * config/arc/arc.c (arc_override_options): Handle extra field.
10274 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
10276 * config/arc/arc-arch.h: Update ARC_OPTX macro.
10277 * config/arc/arc-options.def (ARC_OPTX): Introduce a new doc
10279 * config/arc/arc.c (arc_init): Update pic warning.
10280 (irq_range): Update irq range parsing warnings.
10281 (arc_override_options): Update various warning messages.
10282 (arc_handle_aux_attribute): Likewise.
10284 2018-06-12 Robert Suchanek <robert.suchanek@mips.com>
10286 * config/mips/i6400.md (i6400_fpu_fadd): Remove frint.
10288 2018-06-12 Jozef Lawrynowicz <jozef.l@mittosystems.com>
10290 * doc/sourcebuild.texi: Document usage of line number 0 in verify
10291 compiler messages directives.
10293 2018-06-12 Matthew Fortune <mfortune@gmail.com>
10295 * config/mips/mips-cpus.def: New MIPS_CPU for i6500.
10296 * config/mips/mips-tables.opt: Regenerate.
10297 * config/mips/mips.h (MIPS_ISA_LEVEL_SPEC): Mark i6500 as
10299 * doc/invoke.texi: Document -march=i6500.
10301 2018-06-12 Prachi Godbole <prachi.godbole@imgtec.com>
10303 * config/mips/i6400.md (i6400_gpmuldiv): Remove cpu_unit.
10304 (i6400_gpmul): Add cpu_unit.
10305 (i6400_gpdiv): Likewise.
10306 (i6400_msa_add_d): Update reservations.
10307 (i6400_msa_int_add) Likewise.
10308 (i6400_msa_short_logic3) Likewise.
10309 (i6400_msa_short_logic2) Likewise.
10310 (i6400_msa_short_logic) Likewise.
10311 (i6400_msa_move) Likewise.
10312 (i6400_msa_cmp) Likewise.
10313 (i6400_msa_short_float2) Likewise.
10314 (i6400_msa_div_d) Likewise.
10315 (i6400_msa_long_logic1) Likewise.
10316 (i6400_msa_long_logic2) Likewise.
10317 (i6400_msa_mult) Likewise.
10318 (i6400_msa_long_float2) Likewise.
10319 (i6400_msa_long_float4) Likewise.
10320 (i6400_msa_long_float5) Likewise.
10321 (i6400_msa_long_float8) Likewise.
10322 (i6400_fpu_fadd): Include frint type.
10323 (i6400_fpu_store): New define_insn_reservation.
10324 (i6400_fpu_load): Likewise.
10325 (i6400_fpu_move): Likewise.
10326 (i6400_fpu_fcmp): Likewise.
10327 (i6400_fpu_fmadd): Likewise.
10328 (i6400_int_mult): Include imul3nc type and update reservation.
10329 (i6400_int_div): Include idiv3 type and update reservation.
10330 (i6400_int_load): Update to check type not move_type.
10331 (i6400_int_store): Likewise.
10332 (i6400_int_prefetch): Set zero latency.
10334 2018-06-12 Eric Botcazou <ebotcazou@adacore.com>
10336 * gcc.c: Document new %@{...} sequence.
10337 (LINK_COMMAND_SPEC): Use it for the -L switches.
10338 (cpp_unique_options): Use it for the -I switches.
10339 (at_file_argbuf): New global variable.
10340 (in_at_file): Likewise.
10341 (alloc_args): Create at_file_argbuf.
10342 (clear_args): Truncate at_file_argbuf.
10343 (store_arg): If in_at_file, push the argument onto at_file_argbuf.
10344 (open_at_file): New function.
10345 (close_at_file): Likewise.
10346 (create_at_file): Delete.
10347 (do_spec_1) <'i'>: Use open_at_file/close_at_file.
10350 (validate_switches_from_spec): Deal with %@{...} sequence.
10351 (validate_switches): Likewise.
10352 (driver::finalize): Call clear_args.
10354 2018-06-11 Rasmus Villemoes <rasmus.villemoes@prevas.dk>
10356 * config/vx-common.h (USE_TM_CLONE_REGISTRY): #define to 0.
10358 2018-06-11 Martin Sebor <msebor@redhat.com>
10360 * doc/invoke.texi (-Wall): List -Wc++17-compat.
10361 (Wno-class-memaccess): Add @opindex.
10362 (Wno-templates, Wno-multiple-inheritance): Same.
10363 (Wno-virtual-inheritance, Wno-namespaces): Same.
10364 (Wno-pedantic, Wno-chkp, Wcoverage-mismatch): Same.
10365 (Wno-format-overflow, Wno-format-truncation): Same.
10366 (Wno-unused-local-typedefs, Walloc-size-larger-than): Same.
10367 (Wno-alloc-size-larger-than, Wframe-larger-than): Same
10368 (Wno-stack-usage, Wno-c++-compat, Wno-c++11-compat): Same.
10369 (Wno-c++14-compat, Wno-c++17-compat, Wno-openmp-simd): Same.
10370 (Wno-unsuffixed-float-constants, Wno-addr-space-convert): Same.
10371 (Wno-misspelled-isr): Same.
10373 2018-06-11 Martin Sebor <msebor@redhat.com>
10375 * PR tree-optimization/86083
10376 * tree-ssa-strlen.c (handle_char_store): Use tree_expr_nonzero_p.
10378 2018-06-11 Zhouyi Zhou <zhouzhouyi@gmail.com>
10380 * tree-eh.c (lower_eh_constructs_2): Add a comma to comment.
10382 2018-06-11 Segher Boessenkool <segher@kernel.crashing.org>
10385 * config/rs6000/rs6000.md (*movdi_internal32): Put constraint modifiers
10386 on the correct operand.
10387 (*movdi_internal64): Ditto.
10389 2018-06-11 Martin Liska <mliska@suse.cz>
10391 PR tree-optimization/86089
10392 * tree-ssa-strlen.c (get_string_length): Move back removed hunk.
10394 2018-06-11 Julia Koval <julia.koval@intel.com>
10396 * config/i386/vaesintrin.h (_mm_aesdec_epi128, _mm_aesdeclast_epi128)
10397 _mm_aesenc_epi128, _mm_aesenclast_epi128: Remove.
10398 * config/i386/vpclmulqdqintrin.h (_mm_clmulepi64_epi128): Remove.
10400 2018-06-11 Olivier Hainque <hainque@adacore.com>
10402 * dwarf2out.c (gen_compile_unit_die): Fallout to DW_LANG_Ada83
10403 for Ada with strict dwarf2.
10405 2018-06-08 Peter Bergner <bergner@vnet.ibm.com>
10408 * config/rs6000/rs6000.c (mem_operand_gpr): Enable PRE_INC and PRE_DEC
10411 2018-06-08 Jan Hubicka <hubicka@ucw.cz>
10413 * dumpfile.c (FIRST_ME_AUTO_NUMBERED_DUMP): Bump to 4.
10415 2018-06-08 David Edelsohn <dje.gcc@gmail.com>
10417 * config/rs6000/rs6000.c (rs6000_passes_ieee128): Protect with #if
10420 2018-06-08 Martin Liska <mliska@suse.cz>
10422 * tree-cfg.h (debug_function): Fix argument type to match
10425 2018-06-08 Martin Liska <mliska@suse.cz>
10427 * config/powerpcspe/powerpcspe.c (rs6000_xcoff_visibility):
10428 Remove usage of MPX-related (and removed) fields.
10429 * config/rs6000/rs6000.c (rs6000_xcoff_visibility): Likewise.
10431 2018-06-08 David Malcolm <dmalcolm@redhat.com>
10433 * cfg.c (debug): Use TDF_NONE rather than 0.
10434 * cfghooks.c (debug): Likewise.
10435 * dumpfile.c (DUMP_FILE_INFO): Likewise; also for OPTGROUP.
10436 (struct dump_option_value_info): Convert to...
10437 (struct kv_pair): ...this template type.
10438 (dump_options): Convert to kv_pair<dump_flags_t>; use TDF_NONE
10440 (optinfo_verbosity_options): Likewise.
10441 (optgroup_options): Convert to kv_pair<optgroup_flags_t>; use
10443 (gcc::dump_manager::dump_register): Use optgroup_flags_t rather
10444 than int for "optgroup_flags" param.
10445 (dump_generic_expr_loc): Use dump_flags_t rather than int for
10447 (dump_dec): Likewise.
10448 (dump_finish): Use TDF_NONE rather than 0.
10449 (gcc::dump_manager::opt_info_enable_passes): Use optgroup_flags_t
10450 rather than int for "optgroup_flags" param. Use TDF_NONE rather
10451 than 0. Update for change to option_ptr.
10452 (opt_info_switch_p_1): Convert "optgroup_flags" param from int *
10453 to optgroup_flags_t *. Use TDF_NONE and OPTGROUP_NONE rather than
10454 0. Update for changes to optinfo_verbosity_options and
10456 (opt_info_switch_p): Convert optgroup_flags from int to
10458 (dump_basic_block): Use dump_flags_t rather than int
10459 for "dump_kind" param.
10460 * dumpfile.h (TDF_ADDRESS, TDF_SLIM, TDF_RAW, TDF_DETAILS,
10461 TDF_STATS, TDF_BLOCKS, TDF_VOPS, TDF_LINENO, TDF_UID)
10462 TDF_STMTADDR, TDF_GRAPH, TDF_MEMSYMS, TDF_RHS_ONLY, TDF_ASMNAME,
10463 TDF_EH, TDF_NOUID, TDF_ALIAS, TDF_ENUMERATE_LOCALS, TDF_CSELIB,
10464 TDF_SCEV, TDF_GIMPLE, TDF_FOLDING, MSG_OPTIMIZED_LOCATIONS,
10465 MSG_MISSED_OPTIMIZATION, MSG_NOTE, MSG_ALL, TDF_COMPARE_DEBUG,
10466 TDF_NONE): Convert from macros to...
10467 (enum dump_flag): ...this new enum.
10468 (dump_flags_t): Update to use enum.
10469 (operator|, operator&, operator~, operator|=, operator&=):
10470 Implement for dump_flags_t.
10471 (OPTGROUP_NONE, OPTGROUP_IPA, OPTGROUP_LOOP, OPTGROUP_INLINE,
10472 OPTGROUP_OMP, OPTGROUP_VEC, OPTGROUP_OTHER, OPTGROUP_ALL):
10473 Convert from macros to...
10474 (enum optgroup_flag): ...this new enum.
10475 (optgroup_flags_t): New typedef.
10476 (operator|, operator|=): Implement for optgroup_flags_t.
10477 (struct dump_file_info): Convert field "alt_flags" to
10478 dump_flags_t. Convert field "optgroup_flags" to
10480 (dump_basic_block): Use dump_flags_t rather than int for param.
10481 (dump_generic_expr_loc): Likewise.
10482 (dump_dec): Likewise.
10483 (dump_register): Convert param "optgroup_flags" to
10485 (opt_info_enable_passes): Likewise.
10486 * early-remat.c (early_remat::dump_edge_list): Use TDF_NONE rather
10488 * gimple-pretty-print.c (debug): Likewise.
10489 * gimple-ssa-store-merging.c (bswap_replace): Likewise.
10490 (merged_store_group::apply_stores): Likewise.
10491 * gimple-ssa-strength-reduction.c (insert_initializers): Likewise.
10492 * gimple.c (verify_gimple_pp): Likewise.
10493 * graphite-poly.c (print_pbb_body): Likewise.
10494 * passes.c (pass_manager::register_one_dump_file): Convert
10495 local "optgroup_flags" to optgroup_flags_t.
10496 * print-tree.c (print_node): Use TDF_NONE rather than 0.
10498 (debug_body): Likewise.
10499 * tree-pass.h (struct pass_data): Convert field "optgroup_flags"
10500 to optgroup_flags_t.
10501 * tree-pretty-print.c (print_struct_decl): Use TDF_NONE rather
10503 * tree-ssa-math-opts.c (convert_mult_to_fma_1): Likewise.
10504 (convert_mult_to_fma): Likewise.
10505 * tree-ssa-reassoc.c (undistribute_ops_list): Likewise.
10506 * tree-ssa-sccvn.c (vn_eliminate): Likewise.
10507 * tree-vect-data-refs.c (dump_lower_bound): Convert param
10508 "dump_kind" to dump_flags_t.
10510 2018-06-08 Segher Boessenkool <segher@kernel.crashing.org>
10512 * config/rs6000/rs6000.c (min, max): Delete.
10514 2018-06-08 Segher Boessenkool <segher@kernel.crashing.org>
10516 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete mention of
10517 -mabi=spe and -mabi=no-spe.
10519 2018-06-08 Martin Liska <mliska@suse.cz>
10521 * ipa-pure-const.c (propagate_pure_const): Use ::get at places
10522 where we expect an existing summary.
10524 2018-06-08 Martin Liska <mliska@suse.cz>
10526 * ipa-inline-analysis.c (simple_edge_hints): Use ::get method.
10527 * ipa-inline.h (estimate_edge_growth): Likewise.
10529 2018-06-08 Martin Liska <mliska@suse.cz>
10531 * cgraph.c (function_version_hasher::hash): Use
10532 cgraph_node::get_uid ().
10533 (function_version_hasher::equal):
10534 * cgraph.h (cgraph_node::get_uid): New method.
10535 * ipa-inline.c (update_caller_keys): Use
10536 cgraph_node::get_uid ().
10537 (update_callee_keys): Likewise.
10538 * ipa-utils.c (searchc): Likewise.
10539 (ipa_reduced_postorder): Likewise.
10540 * lto-cgraph.c (input_node): Likewise.
10541 * passes.c (is_pass_explicitly_enabled_or_disabled): Likewise.
10542 * symbol-summary.h (symtab_insertion): Likewise.
10543 (symtab_removal): Likewise.
10544 (symtab_duplication): Likewise.
10545 * tree-pretty-print.c (dump_function_header): Likewise.
10546 * tree-sra.c (convert_callers_for_node): Likewise.
10548 2018-06-08 Martin Liska <mliska@suse.cz>
10550 * cgraph.c (symbol_table::create_edge): Always assign a new
10552 (symbol_table::free_edge): Do not recycle numbers.
10553 * cgraph.h (cgraph_edge::get): New method.
10554 * symbol-summary.h (symtab_removal): Use it.
10555 (symtab_duplication): Likewise.
10556 (call_summary::hashable_uid): Remove.
10558 2018-06-08 Martin Liska <mliska@suse.cz>
10560 * ipa-inline-analysis.c (inline_edge_removal_hook): Remove.
10561 (initialize_growth_caches): Remove.
10562 (free_growth_caches): Likewise.
10563 (do_estimate_edge_time): Use edge_growth_cache.
10564 (do_estimate_edge_size): Likewise.
10565 (do_estimate_edge_hints): Likewise.
10566 * ipa-inline.c (reset_edge_caches): Likewise.
10567 (recursive_inlining): Likewise.
10568 (inline_small_functions): Likewise.
10569 * ipa-inline.h (initialize_growth_caches): Remove.
10570 (estimate_edge_size): Likewise.
10571 (estimate_edge_time): Likewise.
10572 (estimate_edge_hints): Likewise.
10573 (reset_edge_growth_cache): Likewise.
10574 * symbol-summary.h (call_summary::remove): New method.
10576 2018-06-08 Martin Liska <mliska@suse.cz>
10578 * ipa-cp.c (class edge_clone_summary): New summary.
10579 (grow_edge_clone_vectors): Remove.
10580 (ipcp_edge_duplication_hook): Remove.
10581 (class edge_clone_summary_t): New call_summary class.
10582 (ipcp_edge_removal_hook): Remove.
10583 (edge_clone_summary_t::duplicate): New function.
10584 (get_next_cgraph_edge_clone): Use edge_clone_summaries.
10585 (create_specialized_node): Likewise.
10586 (ipcp_driver): Initialize edge_clone_summaries and do not
10589 2018-06-08 Martin Liska <mliska@suse.cz>
10591 * symbol-summary.h (get): New function.
10592 (call_summary::m_initialize_when_cloning): New class member.
10594 2018-06-08 Martin Liska <mliska@suse.cz>
10596 * cgraph.c (cgraph_node::remove): Do not recycle uid.
10597 * cgraph.h (symbol_table::release_symbol): Do not pass uid.
10598 (symbol_table::allocate_cgraph_symbol): Do not set uid.
10599 * passes.c (uid_hash_t): Record removed_nodes by their uids.
10600 (remove_cgraph_node_from_order): Use the removed_nodes set.
10601 (do_per_function_toporder): Likwise.
10602 * symbol-summary.h (symtab_insertion): Use cgraph_node::uid
10603 instead of summary_uid.
10604 (symtab_removal): Likewise.
10605 (symtab_duplication): Likewise.
10607 2018-06-08 Martin Liska <mliska@suse.cz>
10609 * ipa-cp.c (ipcp_store_bits_results): Use
10610 ipcp_transformation_sum.
10611 (ipcp_store_vr_results): Likewise.
10612 * ipa-prop.c (ipcp_grow_transformations_if_necessary): Renamed
10614 (ipcp_transformation_initialize): ... this.
10615 (ipa_set_node_agg_value_chain):
10616 (ipa_node_params_t::duplicate): Use ipcp_transformation_sum.
10617 (write_ipcp_transformation_info): Likewise.
10618 (read_ipcp_transformation_info): Likewise.
10619 (ipcp_update_bits): Likewise.
10620 (ipcp_update_vr): Likewise.
10621 (ipcp_transform_function): Likewise.
10622 * ipa-prop.h: Rename ipcp_transformation_summary to
10623 ipcp_transformation.
10624 (class ipcp_transformation_t): New function summary.
10625 (ipcp_get_transformation_summary): Use ipcp_transformation_sum.
10626 (ipa_get_agg_replacements_for_node): Likewise.
10628 2018-06-08 Martin Liska <mliska@suse.cz>
10630 * ipa-pure-const.c (struct funct_state_d): Do it class instead
10632 (class funct_state_summary_t): New function_summary class.
10633 (has_function_state): Remove.
10634 (get_function_state): Likewise.
10635 (set_function_state): Likewise.
10636 (add_new_function): Likewise.
10637 (funct_state_summary_t::insert): New function.
10638 (duplicate_node_data): Remove.
10639 (remove_node_data): Remove.
10640 (funct_state_summary_t::duplicate): New function.
10641 (register_hooks): Create new funct_state_summaries.
10642 (pure_const_generate_summary): Use it.
10643 (pure_const_write_summary): Likewise.
10644 (pure_const_read_summary): Likewise.
10645 (propagate_pure_const): Likewise.
10646 (propagate_nothrow): Likewise.
10647 (dump_malloc_lattice): Likewise.
10648 (propagate_malloc): Likewise.
10649 (execute): Do not register hooks, just remove summary
10651 (pass_ipa_pure_const::pass_ipa_pure_const): Simplify
10654 2018-06-08 Martin Liska <mliska@suse.cz>
10656 * ipa-reference.c (remove_node_data): Remove.
10657 (duplicate_node_data): Likewise.
10658 (class ipa_ref_var_info_summary_t): New class.
10659 (class ipa_ref_opt_summary_t): Likewise.
10660 (get_reference_vars_info): Use ipa_ref_var_info_summaries.
10661 (get_reference_optimization_summary): Use
10662 ipa_ref_opt_sum_summaries.
10663 (set_reference_vars_info): Remove.
10664 (set_reference_optimization_summary): Likewise.
10665 (ipa_init): Create summaries.
10666 (init_function_info): Use function summary.
10667 (ipa_ref_opt_summary_t::duplicate): New function.
10668 (ipa_ref_opt_summary_t::remove): New function.
10669 (get_read_write_all_from_node): Fix GNU coding style.
10670 (propagate): Use function summary.
10671 (write_node_summary_p): Fix GNU coding style.
10672 (stream_out_bitmap): Likewise.
10673 (ipa_reference_read_optimization_summary): Use function summary.
10674 (ipa_reference_c_finalize): Do not release hooks.
10676 2018-06-08 Martin Liska <mliska@suse.cz>
10678 * ipa-fnsummary.c (dump_ipa_call_summary): Use ::get method.
10679 (analyze_function_body): Extract multiple calls of get_create.
10680 * ipa-inline-analysis.c (simple_edge_hints): Likewise.
10681 * ipa-inline.c (recursive_inlining): Use ::get method.
10682 * ipa-inline.h (estimate_edge_growth): Likewise.
10684 2018-06-08 Martin Liska <mliska@suse.cz>
10686 * hsa-common.h (enum hsa_function_kind): Rename HSA_NONE to
10688 (hsa_function_summary::hsa_function_summary): Use the new enum
10690 (hsa_gpu_implementation_p): Use hsa_summaries::get.
10691 * hsa-gen.c (hsa_get_host_function): Likewise.
10692 (get_brig_function_name): Likewise.
10693 * ipa-hsa.c (process_hsa_functions): Likewise.
10694 (ipa_hsa_write_summary): Likewise.
10695 * symbol-summary.h (symtab_duplication): Use ::get function/
10696 (get): New function.
10698 2018-06-08 Martin Liska <mliska@suse.cz>
10700 * config/i386/i386.c (ix86_can_inline_p): Use get_create instead
10702 * hsa-common.c (hsa_summary_t::link_functions): Likewise.
10703 (hsa_register_kernel): Likewise.
10704 * hsa-common.h (hsa_gpu_implementation_p): Likewise.
10705 * hsa-gen.c (hsa_get_host_function): Likewise.
10706 (get_brig_function_name): Likewise.
10707 (generate_hsa): Likewise.
10708 (pass_gen_hsail::execute): Likewise.
10709 * ipa-cp.c (ipcp_cloning_candidate_p): Likewise.
10710 (devirtualization_time_bonus): Likewise.
10711 (ipcp_propagate_stage): Likewise.
10712 * ipa-fnsummary.c (redirect_to_unreachable): Likewise.
10713 (edge_set_predicate): Likewise.
10714 (evaluate_conditions_for_known_args): Likewise.
10715 (evaluate_properties_for_edge): Likewise.
10716 (ipa_fn_summary::reset): Likewise.
10717 (ipa_fn_summary_t::duplicate): Likewise.
10718 (dump_ipa_call_summary): Likewise.
10719 (ipa_dump_fn_summary): Likewise.
10720 (analyze_function_body): Likewise.
10721 (compute_fn_summary): Likewise.
10722 (estimate_edge_devirt_benefit): Likewise.
10723 (estimate_edge_size_and_time): Likewise.
10724 (estimate_calls_size_and_time): Likewise.
10725 (estimate_node_size_and_time): Likewise.
10726 (inline_update_callee_summaries): Likewise.
10727 (remap_edge_change_prob): Likewise.
10728 (remap_edge_summaries): Likewise.
10729 (ipa_merge_fn_summary_after_inlining): Likewise.
10730 (ipa_update_overall_fn_summary): Likewise.
10731 (read_ipa_call_summary): Likewise.
10732 (inline_read_section): Likewise.
10733 (write_ipa_call_summary): Likewise.
10734 (ipa_fn_summary_write): Likewise.
10735 (ipa_free_fn_summary): Likewise.
10736 * ipa-hsa.c (process_hsa_functions): Likewise.
10737 (ipa_hsa_write_summary): Likewise.
10738 (ipa_hsa_read_section): Likewise.
10739 * ipa-icf.c (sem_function::merge): Likewise.
10740 * ipa-inline-analysis.c (simple_edge_hints): Likewise.
10741 (do_estimate_edge_time): Likewise.
10742 (estimate_size_after_inlining): Likewise.
10743 (estimate_growth): Likewise.
10744 (growth_likely_positive): Likewise.
10745 * ipa-inline-transform.c (clone_inlined_nodes): Likewise.
10746 (inline_call): Likewise.
10747 * ipa-inline.c (caller_growth_limits): Likewise.
10748 (can_inline_edge_p): Likewise.
10749 (can_inline_edge_by_limits_p): Likewise.
10750 (compute_uninlined_call_time): Likewise.
10751 (compute_inlined_call_time): Likewise.
10752 (want_inline_small_function_p): Likewise.
10753 (edge_badness): Likewise.
10754 (update_caller_keys): Likewise.
10755 (update_callee_keys): Likewise.
10756 (recursive_inlining): Likewise.
10757 (inline_small_functions): Likewise.
10758 (inline_to_all_callers_1): Likewise.
10759 (dump_overall_stats): Likewise.
10760 (early_inline_small_functions): Likewise.
10761 (early_inliner): Likewise.
10762 * ipa-inline.h (estimate_edge_growth): Likewise.
10763 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
10764 * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
10765 * ipa-prop.h (IPA_NODE_REF): Likewise.
10766 (IPA_EDGE_REF): Likewise.
10767 * ipa-pure-const.c (malloc_candidate_p): Likewise.
10768 (propagate_malloc): Likewise.
10769 * ipa-split.c (execute_split_functions): Likewise.
10770 * symbol-summary.h: Rename get to get_create.
10772 (get_create): Likewise.
10773 * tree-sra.c (ipa_sra_preliminary_function_checks): Likewise.
10775 2018-06-08 Martin Liska <mliska@suse.cz>
10777 * symbol-summary.h (release): Move definition out of class
10779 (symtab_removal): Likewise.
10780 (symtab_duplication): Likewise.
10782 2018-06-08 Martin Liska <mliska@suse.cz>
10784 * symbol-summary.h (function_summary): Move constructor
10785 implementation out of class declaration.
10786 (release): Likewise.
10787 (symtab_insertion): Likewise.
10788 (symtab_removal): Likewise.
10789 (symtab_duplication): Likewise.
10792 2018-06-08 Martin Liska <mliska@suse.cz>
10794 * Makefile.in: Remove support for MPX (macros, related functions,
10795 fields in cgraph_node, ...).
10796 * builtin-types.def (BT_BND): Likewise.
10797 (BT_FN_BND_CONST_PTR): Likewise.
10798 (BT_FN_CONST_PTR_BND): Likewise.
10799 (BT_FN_VOID_PTR_BND): Likewise.
10800 (BT_FN_BND_CONST_PTR_SIZE): Likewise.
10801 (BT_FN_VOID_CONST_PTR_BND_CONST_PTR): Likewise.
10802 * builtins.c (expand_builtin_memcpy_with_bounds): Likewise.
10803 (expand_builtin_mempcpy_with_bounds): Likewise.
10804 (expand_builtin_memset_with_bounds): Likewise.
10805 (expand_builtin_memset_args): Likewise.
10806 (std_expand_builtin_va_start): Likewise.
10807 (expand_builtin): Likewise.
10808 (expand_builtin_with_bounds): Likewise.
10809 * builtins.def (DEF_BUILTIN_CHKP): Likewise.
10810 (DEF_LIB_BUILTIN_CHKP): Likewise.
10811 (DEF_EXT_LIB_BUILTIN_CHKP): Likewise.
10812 (DEF_CHKP_BUILTIN): Likewise.
10813 (BUILT_IN_MEMCPY): Likewise.
10814 (BUILT_IN_MEMMOVE): Likewise.
10815 (BUILT_IN_MEMPCPY): Likewise.
10816 (BUILT_IN_MEMSET): Likewise.
10817 (BUILT_IN_STPCPY): Likewise.
10818 (BUILT_IN_STRCAT): Likewise.
10819 (BUILT_IN_STRCHR): Likewise.
10820 (BUILT_IN_STRCPY): Likewise.
10821 (BUILT_IN_STRLEN): Likewise.
10822 (BUILT_IN_MEMCPY_CHK): Likewise.
10823 (BUILT_IN_MEMMOVE_CHK): Likewise.
10824 (BUILT_IN_MEMPCPY_CHK): Likewise.
10825 (BUILT_IN_MEMSET_CHK): Likewise.
10826 (BUILT_IN_STPCPY_CHK): Likewise.
10827 (BUILT_IN_STRCAT_CHK): Likewise.
10828 (BUILT_IN_STRCPY_CHK): Likewise.
10829 * calls.c (store_bounds): Likewise.
10830 (emit_call_1): Likewise.
10831 (special_function_p): Likewise.
10832 (maybe_warn_nonstring_arg): Likewise.
10833 (initialize_argument_information): Likewise.
10834 (finalize_must_preallocate): Likewise.
10835 (compute_argument_addresses): Likewise.
10836 (expand_call): Likewise.
10837 * cfgexpand.c (expand_call_stmt): Likewise.
10838 (expand_return): Likewise.
10839 (expand_gimple_stmt_1): Likewise.
10840 (pass_expand::execute): Likewise.
10841 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
10842 (cgraph_node::remove): Likewise.
10843 (cgraph_node::dump): Likewise.
10844 (cgraph_node::verify_node): Likewise.
10845 * cgraph.h (chkp_function_instrumented_p): Likewise.
10846 (symtab_node::get_alias_target): Likewise.
10847 (cgraph_node::can_remove_if_no_direct_calls_and_refs_p): Likewise.
10848 (cgraph_local_p): Likewise.
10849 * cgraphbuild.c (cgraph_edge::rebuild_edges): Likewise.
10850 (cgraph_edge::rebuild_references): Likewise.
10851 * cgraphunit.c (varpool_node::finalize_decl): Likewise.
10852 (walk_polymorphic_call_targets): Likewise.
10853 (cgraph_node::expand_thunk): Likewise.
10854 (symbol_table::output_weakrefs): Likewise.
10855 * common/config/i386/i386-common.c (OPTION_MASK_ISA2_GENERAL_REGS_ONLY_UNSET): Likewise.
10856 (ix86_handle_option): Likewise.
10857 * config/i386/constraints.md: Likewise.
10858 * config/i386/i386-builtin-types.def (BND): Likewise.
10862 * config/i386/i386-builtin.def (BDESC_END): Likewise.
10863 (BDESC_FIRST): Likewise.
10865 * config/i386/i386-c.c (ix86_target_macros_internal): Likewise.
10866 * config/i386/i386-protos.h (ix86_bnd_prefixed_insn_p): Likewise.
10867 * config/i386/i386.c (enum reg_class): Likewise.
10868 (ix86_target_string): Likewise.
10869 (ix86_option_override_internal): Likewise.
10870 (ix86_conditional_register_usage): Likewise.
10871 (ix86_valid_target_attribute_inner_p): Likewise.
10872 (ix86_set_indirect_branch_type): Likewise.
10873 (ix86_set_current_function): Likewise.
10874 (ix86_function_arg_regno_p): Likewise.
10875 (init_cumulative_args): Likewise.
10876 (ix86_function_arg_advance): Likewise.
10877 (ix86_function_arg): Likewise.
10878 (ix86_pass_by_reference): Likewise.
10879 (ix86_function_value_regno_p): Likewise.
10880 (ix86_function_value_1): Likewise.
10881 (ix86_function_value_bounds): Likewise.
10882 (ix86_return_in_memory): Likewise.
10883 (ix86_setup_incoming_vararg_bounds): Likewise.
10884 (ix86_va_start): Likewise.
10885 (indirect_thunk_need_prefix): Likewise.
10886 (print_reg): Likewise.
10887 (ix86_print_operand): Likewise.
10888 (ix86_expand_call): Likewise.
10889 (ix86_output_function_return): Likewise.
10890 (reg_encoded_number): Likewise.
10891 (BDESC_VERIFYS): Likewise.
10892 (ix86_init_mpx_builtins): Likewise.
10893 (ix86_init_builtins): Likewise.
10894 (ix86_emit_cmove): Likewise.
10895 (ix86_emit_move_max): Likewise.
10896 (ix86_expand_builtin): Likewise.
10897 (ix86_builtin_mpx_function): Likewise.
10898 (ix86_get_arg_address_for_bt): Likewise.
10899 (ix86_load_bounds): Likewise.
10900 (ix86_store_bounds): Likewise.
10901 (ix86_load_returned_bounds): Likewise.
10902 (ix86_store_returned_bounds): Likewise.
10903 (ix86_class_likely_spilled_p): Likewise.
10904 (ix86_hard_regno_mode_ok): Likewise.
10905 (x86_order_regs_for_local_alloc): Likewise.
10906 (ix86_mitigate_rop): Likewise.
10907 (ix86_bnd_prefixed_insn_p): Likewise.
10908 (ix86_mpx_bound_mode): Likewise.
10909 (ix86_make_bounds_constant): Likewise.
10910 (ix86_initialize_bounds): Likewise.
10911 (TARGET_LOAD_BOUNDS_FOR_ARG): Likewise.
10912 (TARGET_STORE_BOUNDS_FOR_ARG): Likewise.
10913 (TARGET_LOAD_RETURNED_BOUNDS): Likewise.
10914 (TARGET_STORE_RETURNED_BOUNDS): Likewise.
10915 (TARGET_CHKP_BOUND_MODE): Likewise.
10916 (TARGET_BUILTIN_CHKP_FUNCTION): Likewise.
10917 (TARGET_CHKP_FUNCTION_VALUE_BOUNDS): Likewise.
10918 (TARGET_CHKP_MAKE_BOUNDS_CONSTANT): Likewise.
10919 (TARGET_CHKP_INITIALIZE_BOUNDS): Likewise.
10920 * config/i386/i386.h (TARGET_MPX): Likewise.
10921 (TARGET_MPX_P): Likewise.
10922 (VALID_BND_REG_MODE): Likewise.
10923 (FIRST_BND_REG): Likewise.
10924 (LAST_BND_REG): Likewise.
10925 (enum reg_class): Likewise.
10926 (BND_REG_P): Likewise.
10927 (BND_REGNO_P): Likewise.
10928 (BNDmode): Likewise.
10929 (ADJUST_INSN_LENGTH): Likewise.
10930 * config/i386/i386.md: Likewise.
10931 * config/i386/i386.opt: Likewise.
10932 * config/i386/linux-common.h (LIBMPX_LIBS): Likewise.
10933 (defined): Likewise.
10934 (LINK_MPX): Likewise.
10935 (MPX_SPEC): Likewise.
10936 (LIBMPX_SPEC): Likewise.
10937 (LIBMPXWRAPPERS_SPEC): Likewise.
10938 (CHKP_SPEC): Likewise.
10939 * config/i386/predicates.md: Likewise.
10940 * dbxout.c (dbxout_type): Likewise.
10941 * doc/extend.texi: Likewise.
10942 * doc/invoke.texi: Likewise.
10943 * doc/md.texi: Likewise.
10944 * doc/tm.texi: Likewise.
10945 * doc/tm.texi.in: Likewise.
10946 * dwarf2out.c (is_base_type): Likewise.
10947 (gen_formal_types_die): Likewise.
10948 (gen_subprogram_die): Likewise.
10949 (gen_type_die_with_usage): Likewise.
10950 (gen_decl_die): Likewise.
10951 (dwarf2out_late_global_decl): Likewise.
10952 * expr.c (expand_assignment): Likewise.
10953 (emit_storent_insn): Likewise.
10954 (store_expr_with_bounds): Likewise.
10955 (store_expr): Likewise.
10956 (expand_expr_real_1): Likewise.
10957 * expr.h (store_expr_with_bounds): Likewise.
10958 * function.c (use_register_for_decl): Likewise.
10959 (struct bounds_parm_data): Likewise.
10960 (assign_parms_augmented_arg_list): Likewise.
10961 (assign_parm_find_entry_rtl): Likewise.
10962 (assign_parm_is_stack_parm): Likewise.
10963 (assign_parm_load_bounds): Likewise.
10964 (assign_bounds): Likewise.
10965 (assign_parms): Likewise.
10966 (expand_function_start): Likewise.
10967 * gcc.c (CHKP_SPEC): Likewise.
10968 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
10969 * gimple-ssa-warn-restrict.c (builtin_access::builtin_access): Likewise.
10970 (wrestrict_dom_walker::check_call): Likewise.
10971 * gimple.c (gimple_build_call_from_tree): Likewise.
10972 * gimple.h (enum gf_mask): Likewise.
10973 (gimple_call_with_bounds_p): Likewise.
10974 (gimple_call_set_with_bounds): Likewise.
10975 * gimplify.c (gimplify_init_constructor): Likewise.
10976 * ipa-cp.c (initialize_node_lattices): Likewise.
10977 (propagate_constants_across_call): Likewise.
10978 (find_more_scalar_values_for_callers_subset): Likewise.
10979 * ipa-hsa.c (process_hsa_functions): Likewise.
10980 * ipa-icf-gimple.c (func_checker::compare_gimple_call): Likewise.
10981 * ipa-icf.c (sem_function::merge): Likewise.
10982 * ipa-inline.c (early_inliner): Likewise.
10983 * ipa-pure-const.c (warn_function_noreturn): Likewise.
10984 (warn_function_cold): Likewise.
10985 (propagate_pure_const): Likewise.
10986 * ipa-ref.h (enum GTY): Likewise.
10987 * ipa-split.c (find_retbnd): Likewise.
10988 (consider_split): Likewise.
10989 (split_function): Likewise.
10990 * ipa-visibility.c (cgraph_externally_visible_p): Likewise.
10991 * ipa.c (walk_polymorphic_call_targets): Likewise.
10992 (symbol_table::remove_unreachable_nodes): Likewise.
10993 (process_references): Likewise.
10994 (cgraph_build_static_cdtor_1): Likewise.
10995 * lto-cgraph.c (lto_output_node): Likewise.
10996 (output_refs): Likewise.
10997 (compute_ltrans_boundary): Likewise.
10998 (input_overwrite_node): Likewise.
10999 (input_node): Likewise.
11000 (input_cgraph_1): Likewise.
11001 * params.def (PARAM_CHKP_MAX_CTOR_SIZE): Likewise.
11002 * passes.c (pass_manager::execute_early_local_passes): Likewise.
11003 (class pass_chkp_instrumentation_passes): Likewise.
11004 (make_pass_chkp_instrumentation_passes): Likewise.
11005 * passes.def: Likewise.
11006 * rtl.h (struct GTY): Likewise.
11007 (CALL_EXPR_WITH_BOUNDS_P): Likewise.
11008 * stor-layout.c (layout_type): Likewise.
11009 * symtab.c: Likewise.
11010 * target.def: Likewise.
11011 * targhooks.c (default_chkp_bound_type): Likewise.
11012 (default_chkp_bound_mode): Likewise.
11013 (default_builtin_chkp_function): Likewise.
11014 (default_chkp_function_value_bounds): Likewise.
11015 (default_chkp_make_bounds_constant): Likewise.
11016 (default_chkp_initialize_bounds): Likewise.
11017 * targhooks.h (default_chkp_bound_type): Likewise.
11018 (default_chkp_bound_mode): Likewise.
11019 (default_builtin_chkp_function): Likewise.
11020 (default_chkp_function_value_bounds): Likewise.
11021 (default_chkp_make_bounds_constant): Likewise.
11022 (default_chkp_initialize_bounds): Likewise.
11023 * toplev.c (compile_file): Likewise.
11024 (process_options): Likewise.
11025 * tree-core.h (DEF_BUILTIN): Likewise.
11026 (DEF_BUILTIN_CHKP): Likewise.
11027 * tree-inline.c (declare_return_variable): Likewise.
11028 (remap_gimple_stmt): Likewise.
11029 (copy_bb): Likewise.
11030 (initialize_inlined_parameters): Likewise.
11031 (expand_call_inline): Likewise.
11032 * tree-pass.h (make_pass_ipa_chkp_versioning): Likewise.
11033 (make_pass_ipa_chkp_early_produce_thunks): Likewise.
11034 (make_pass_ipa_chkp_produce_thunks): Likewise.
11035 (make_pass_chkp): Likewise.
11036 (make_pass_chkp_opt): Likewise.
11037 (make_pass_chkp_instrumentation_passes): Likewise.
11038 * tree-pretty-print.c (dump_generic_node): Likewise.
11039 * tree-ssa-ccp.c (insert_clobber_before_stack_restore): Likewise.
11040 * tree-ssa-dce.c (propagate_necessity): Likewise.
11041 (eliminate_unnecessary_stmts): Likewise.
11042 * tree-ssa-pre.c (create_expression_by_pieces): Likewise.
11043 * tree-ssa-sccvn.c (copy_reference_ops_from_call): Likewise.
11044 * tree-ssa-sccvn.h: Likewise.
11045 * tree-ssa-strlen.c (get_string_length): Likewise.
11046 (valid_builtin_call): Likewise.
11047 (adjust_last_stmt): Likewise.
11048 (handle_builtin_strchr): Likewise.
11049 (handle_builtin_strcpy): Likewise.
11050 (handle_builtin_stxncpy): Likewise.
11051 (handle_builtin_memcpy): Likewise.
11052 (handle_builtin_strcat): Likewise.
11053 (strlen_check_and_optimize_stmt): Likewise.
11054 * tree-stdarg.c (expand_ifn_va_arg_1): Likewise.
11055 * tree-streamer-in.c: Likewise.
11056 * tree-streamer.c (record_common_node): Likewise.
11057 * tree.c (tree_code_size): Likewise.
11058 (wide_int_to_tree_1): Likewise.
11059 (type_contains_placeholder_1): Likewise.
11060 (build_common_tree_nodes): Likewise.
11061 * tree.def (POINTER_BOUNDS_TYPE): Likewise.
11062 * tree.h (POINTER_BOUNDS_TYPE_P): Likewise.
11063 (POINTER_BOUNDS_P): Likewise.
11064 (BOUNDED_TYPE_P): Likewise.
11065 (BOUNDED_P): Likewise.
11066 (CALL_WITH_BOUNDS_P): Likewise.
11067 (pointer_bounds_type_node): Likewise.
11068 * value-prof.c (gimple_ic): Likewise.
11069 * var-tracking.c (vt_add_function_parameters): Likewise.
11070 * varasm.c (make_decl_rtl): Likewise.
11071 (assemble_start_function): Likewise.
11072 (output_constant): Likewise.
11073 (maybe_assemble_visibility): Likewise.
11074 * varpool.c (ctor_for_folding): Likewise.
11075 * chkp-builtins.def: Remove.
11076 * ipa-chkp.c: Remove.
11077 * ipa-chkp.h: Remove.
11078 * rtl-chkp.c: Remove.
11079 * rtl-chkp.h: Remove.
11080 * tree-chkp-opt.c: Remove.
11081 * tree-chkp.c: Remove.
11082 * tree-chkp.h: Remove.
11084 2018-06-07 Carl Love <cel@us.ibm.com>
11086 * config/rs6000/vsx.md (vextract_fp_from_shorth,
11087 vextract_fp_from_shortl): Add BE support.
11089 2018-06-07 Paul Koning <ni1d@arrl.net>
11091 * compare-elim.c (try_merge_compare): Don't merge compare if
11092 address contains a side effect.
11093 (try_eliminate_compare): Likewise.
11095 2018-06-07 Olga Makhotina <olga.makhotina@intel.com>
11097 * config.gcc: Support "tremont".
11098 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "tremont".
11099 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
11101 * config/i386/i386.c (m_TREMONT): Define.
11102 (processor_target_table): Add "tremont".
11103 (PTA_TREMONT): Define.
11104 (ix86_lea_outperforms): Add TARGET_TREMONT.
11105 (get_builtin_code_for_version): Handle PROCESSOR_TREMONT.
11106 (fold_builtin_cpu): Add M_INTEL_TREMONT, replace M_INTEL_GOLDMONT
11107 and M_INTEL_GOLDMONT_PLUS.
11108 (fold_builtin_cpu): Add "tremont".
11109 (ix86_add_stmt_cost): Add TARGET_TREMONT.
11110 (ix86_option_override_internal): Add "tremont".
11111 * config/i386/i386.h (processor_costs): Define TARGET_TREMONT.
11112 (processor_type): Add PROCESSOR_TREMONT.
11113 * config/i386/x86-tune.def: Add m_TREMONT.
11114 * doc/invoke.texi: Add tremont as x86 -march=/-mtune= CPU type.
11116 2018-06-07 Jozef Lawrynowicz <jozef.l@mittosystems.com>
11118 * config/msp430/msp430.c (msp430_mcu_name): Set the "i" in the
11119 symbol defined for msp430i* devices to be lower case.
11121 2018-06-07 Richard Biener <rguenther@suse.de>
11123 * graphite-sese-to-poly.c (extract_affine): Avoid unneded
11124 wrapping. Properly wrap the result of a BIT_NOT_EXPR.
11125 Properly wrap signed arithmetic if overflow wraps.
11127 2018-06-07 Jakub Jelinek <jakub@redhat.com>
11129 PR tree-optimization/69615
11130 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If rhs2 is lhs
11131 of a cast from a same precision integral SSA_NAME in a bb dominated
11132 by first_bb, retry with rhs2 set to the rhs1 of the cast. Don't emit
11133 cast to utype if rhs2 has already a compatible type.
11135 2018-06-07 Richard Biener <rguenther@suse.de>
11137 PR tree-optimization/85935
11138 * graphite-scop-detection.c (find_params_in_bb): Analyze
11139 condition operands with respect to the correct loop. Assert
11140 the analysis doesn't fail.
11142 2018-06-04 Carl Love <cel@us.ibm.com>
11144 * config/rs6000/vsx.md (first_match_index_<mode>): Calculate index
11145 using natural element order. Use gen_lshrsi3 instead of gen_ashrsi3
11146 as it is slightly cheaper.
11147 (first_match_or_eos_index_<mode>):
11148 Calculate index using natural element order.
11149 (first_match_index_<mode>):
11150 Calculate index using natural element order.
11151 (first_match_or_eos_index_<mode>):
11152 Calculate index using natural order.
11153 (define_insn vclzlsbb): Change to define_insn vclzlsbb_<mode>.
11154 for BE and LE modes.
11155 * config/rs6000/rs6000-c.c: Rename P9V_BUILTIN_VCLZLSBB,
11156 P9V_BUILTIN_VCLZLSBB_V16QI.
11157 * config/rs6000/rs6000-builtin.def: Make VCLZLSBB mode
11160 2018-06-06 Kelvin Nilsen <kelvin@gcc.gnu.org>
11162 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
11163 indentation and line wrap for many prototypes. Add missing
11164 @smallexample directives around block of prototypes for vec_xl and
11167 2018-06-05 Michael Meissner <meissner@linux.ibm.com>
11169 * config/rs6000/rs6000.c (rs6000_passes_ieee128): New boolean to
11170 track if we pass or return IEEE 128-bit floating point.
11171 (ieee128_mangling_gcc_8_1): New boolean to say whether to generate
11172 C++ mangling that is compatible with GCC 8.1.
11173 (TARGET_ASM_GLOBALIZE_DECL_NAME): Override target hook.
11174 (init_cumulative_args): Note if we pass or return IEEE 128-bit
11175 floating point types.
11176 (rs6000_function_arg_advance_1): Likewise.
11177 (rs6000_mangle_type): Optionally generate mangled names that match
11178 what GCC 8.1 generated for IEEE 128-bit floating point types.
11179 (rs6000_globalize_decl_name): If we have an external function that
11180 passes or returns IEEE 128-bit types, generate a weak reference
11181 from the mangled name used in GCC 8.1 to the current mangled
11183 (rs6000_init_builtins): Make __ibm128 use the long double type if
11184 long double is IBM extended double. Make __float128 use the long
11185 double type if long double is IEEE 128-bit.
11188 * config/rs6000/rs6000-builtin.def (BU_IBM128_2): New helper
11189 macro for __ibm128 built-in functions.
11190 (PACK_IF): Add __ibm128 pack/unpack functions.
11191 (UNPACK_IF): Likewise.
11192 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Do not
11193 enable long double built-in functions if long double is IEEE
11194 128-bit floating point.
11195 (rs6000_invalid_builtin): Update long double built-in function
11197 (rs6000_expand_builtin): For PACK_IF and UNPACK_IF built-in
11198 functions, adjust the built-in function to use the long double
11199 built-in function if __ibm128 and long double are the same type.
11200 * doc/extend.texi (PowerPC builtins): Update documention for
11201 __builtin_{,un}pack_longdouble. Add documentation for
11202 __builtin_{,un}pack_ibm128.
11204 2018-06-06 Jim Wilson <jimw@sifive.com>
11206 * config/riscv/riscv.c (enum riscv_privilege_levels): New.
11207 (struct machine_function): New field interrupt_mode.
11208 (riscv_handle_type_attribute): New function. Add forward declaration.
11209 (riscv_attribute_table) <interrupt>: Use riscv_handle_type_attribute.
11210 (riscv_expand_epilogue): Check interrupt_mode field.
11211 (riscv_set_current_function): Check interrupt attribute args and
11212 set interrupt_mode field.
11213 * config/riscv/riscv.md (UNSPECV_SRET, UNSPECV_URET): New.
11214 (riscv_sret, riscv_uret): New.
11215 * doc/extend.texi (RISC-V Function Attributes) <interrupt>: Document
11216 new arguments to interrupt attribute.
11218 2018-06-06 Peter Bergner <bergner@vnet.ibm.com>
11221 * /config/rs6000/rs6000.h (ASM_CPU_SPEC): Add support for -mpower9.
11222 Don't handle -mcpu=power8 if -mpower9-vector is also used.
11224 2018-06-06 Kelvin Nilsen <kelvin@gcc.gnu.org>
11226 * config/rs6000/rs6000-builtin.def (VSX_BUILTIN_VEC_LD,
11227 VSX_BUILTIN_VEC_ST): Add comment to explain non-traditional uses.
11228 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
11229 several redundant entries.
11231 2018-06-06 David Malcolm <dmalcolm@redhat.com>
11233 * config/i386/i386-protos.h (ix86_expand_call): Strengthen return
11234 type from "rtx" to "rtx_insn *".
11235 * config/i386/i386.c (ix86_expand_split_stack_prologue): Likewise
11236 for local "call_insn", removing cast.
11237 (ix86_expand_call): Likewise, introducing a "call_insn" local.
11239 2018-06-06 Eric Botcazou <ebotcazou@adacore.com>
11241 PR tree-optimization/86066
11242 * gimple-ssa-store-merging.c (process_store): Do not bypass BIT_NOT_EXPR
11243 for BIT_INSERT_EXPR stores.
11245 2018-06-06 Richard Biener <rguenther@suse.de>
11247 PR tree-optimization/86062
11248 * tree-ssa-sccvn.c (vn_reference_lookup_3): Handle arbitrary
11249 component refs ontop
11250 of to be offsetted base.
11252 2018-06-06 Jozef Lawrynowicz <jozef.l@mittosystems.com>
11254 * config/msp430/msp430.c (msp430_attr): Allow interrupt handlers
11255 to be static and remove check on interrupt attribute name.
11257 2018-06-05 Kelvin Nilsen <kelvin@gcc.gnu.org>
11259 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Remove
11260 volatile qualifier from vec_lvsl and vec_lvsr argument prototypes.
11262 2018-06-05 Steve Ellcey <sellcey@cavium.com>
11265 * config/aarch64/aarch64-protos.h (aarch64_err_no_fpadvsimd): Remove
11267 * config/aarch64/aarch64-protos..c (aarch64_err_no_fpadvsimd):
11268 Remove second argument, change how error is called.
11269 (aarch64_layout_arg): Remove second argument from
11270 aarch64_err_no_fpadvsimd call.
11271 (aarch64_init_cumulative_args): Ditto.
11272 (aarch64_gimplify_va_arg_expr): Ditto.
11273 * config/aarch64/aarch64.md (mov<mode>): Ditto.
11275 2018-06-05 Uros Bizjak <ubizjak@gmail.com>
11277 * config/i386/i386.md (simple_return_indirect_internal): New expander.
11278 (*simple_return_indirect_internal<mode>): Rename from
11279 simple_return_indirect_internal. Use W mode iterator.
11280 (rstorssp): New expander.
11281 (*rstorssp<mode>): Rename from rstorssp. Use P mode iterator.
11282 (clrssbsy): New expander.
11283 (*clrssbsy<mode>): Rename from clrssbsy. Use P mode iterator.
11285 2018-06-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
11287 * config/arm/arm_cmse.h (cmse_nsfptr_create): Change typeof to
11289 (cmse_check_pointed_object): Likewise.
11291 2018-06-05 Martin Liska <mliska@suse.cz>
11293 PR gcov-profile/47618
11294 * doc/invoke.texi: Document how -fprofile-dir format
11297 2018-06-05 Richard Biener <rguenther@suse.de>
11299 * tree-cfgcleanup.c (cleanup_control_flow_pre): For edge
11300 removal pretend DOM info isn't available so we do not update
11301 it and only remove edges, not dominated blocks. Actually free
11302 DOM info in case we removed something. Remove unreachable blocks.
11303 (mfb_keep_latches): Work with either DOM info or marked backedges.
11304 (cleanup_tree_cfg_noloop): Do not remove unreachable blocks
11305 first. Mark backedges if DOM info isn't available.
11306 (Re-)compute DOM info after cleanup_control_flow_pre.
11308 2018-06-05 Richard Biener <rguenther@suse.de>
11310 * tree-cfg.c (struct locus_discrim_map): Store line, not location.
11311 (locus_discrim_hasher::hash): Adjust.
11312 (locus_discrim_hasher::equal): Likewise.
11313 (next_discriminator_for_locus): Work on line directly.
11314 (same_line_p): Pass in expanded locus1 as well.
11315 (assign_discriminators): Avoid redundant location expansions.
11317 2018-06-05 Richard Biener <rguenther@suse.de>
11319 PR tree-optimization/86046
11320 * tree-ssa.c (maybe_optimize_var): Clear DECL_GIMPLE_REG_P
11321 if required after clearing TREE_ADDRESSABLE.
11323 2018-06-05 Richard Biener <rguenther@suse.de>
11325 PR tree-optimization/86047
11326 * tree-ssa-loop.c (for_each_index): Glob handling of all
11327 decls and constants and really handle all of them.
11329 2018-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11332 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
11333 qualifier_void_pointer and qualifier_const_void_pointer.
11334 (arm_ldc_qualifiers, arm_stc_qualifiers): Use the above.
11335 (arm_init_builtins): Handle the above.
11336 * config/arm/arm_acle.h (__arm_cdp, __arm_ldc, __arm_ldcl, __arm_stc,
11337 __arm_stcl, __arm_mcr, __arm_cdp2, __arm_ldc2, __arm_ldcl2, __arm_stc2,
11338 __arm_stcl2,__arm_mcr2, __arm_mcrr, __arm_mcrr2): Remove return for
11341 2018-06-05 Martin Liska <mliska@suse.cz>
11343 * auto-profile.c (read_autofdo_file): Do not use
11344 gcov_ctr_summary struct.
11345 (afdo_callsite_hot_enough_for_early_inline): Likewise.
11346 * coverage.c (struct counts_entry): Likewise.
11347 (read_counts_file): Read just single summary entry.
11348 (get_coverage_counts): Use gcov_summary struct.
11349 * coverage.h (get_coverage_counts): Likewise.
11350 * gcov-dump.c (dump_working_sets): Likewise.
11351 (tag_summary): Dump just single summary.
11352 * gcov-io.c (gcov_write_summary): Write just histogram
11354 (gcov_read_summary): Read just single summary.
11355 (compute_working_sets): Use gcov_summary struct.
11356 * gcov-io.h (GCOV_TAG_SUMMARY_LENGTH): Remove usage
11357 of GCOV_COUNTERS_SUMMABLE.
11358 (GCOV_COUNTERS_SUMMABLE): Remove.
11359 (GCOV_FIRST_VALUE_COUNTER): Replace with
11360 GCOV_COUNTER_V_INTERVAL.
11361 (struct gcov_ctr_summary): Remove.
11362 (struct gcov_summary): Directly use fields of former
11364 (compute_working_sets): Use gcov_summary struct.
11365 * gcov.c (read_count_file): Do not use ctrs fields.
11366 * lto-cgraph.c (merge_profile_summaries): Use gcov_summary
11368 * lto-streamer.h (struct GTY): Make profile_info gcov_summary
11370 * profile.c: Likewise.
11371 * profile.h: Likewise.
11373 2018-06-05 Martin Liska <mliska@suse.cz>
11375 PR gcov-profile/84846
11376 * gcov.c (output_lines): Print working directory only
11377 in intermediate format.
11379 2018-06-05 Andreas Krebbel <krebbel@linux.ibm.com>
11381 * config/s390/s390-builtin-types.def: Add void function type.
11382 * config/s390/s390-builtins.def: Use the function type for the
11385 2018-06-04 Jim Wilson <jimw@sifive.com>
11387 * config/riscv/riscv-protos.h (riscv_expand_epilogue): Change bool arg
11389 * config/riscv/riscv.c (riscv_for_each_saved_reg): New args epilogue
11390 and maybe_eh_return. Change regno to unsigned int. Use new args to
11391 handle EH_RETURN_DATA_REGNO registers properly.
11392 (riscv_expand_prologue): Pass new args to riscv_for_each_saved_reg.
11393 (riscv_expand_epilogue): Update comment. Change argument name and
11394 type. Update code to use new name and type. Pass new args to
11395 riscv_for_each_saved_reg. Only use EH_RETURN_STACKADJ_RTX when
11397 * config/riscv/riscv.md (NORMAL_RETURN): New.
11398 (SIBCALL_RETURN, EXCEPTION_RETURN): New.
11399 (epilogue, sibcall_epilogue): Update riscv_expand_epilogue arg.
11400 (eh_return): Call gen_eh_return_internal and emit barrier.
11401 (eh_return_internal): Call riscv_expand_epilogue.
11403 2018-06-04 Eric Botcazou <ebotcazou@adacore.com>
11405 * gimple-ssa-store-merging.c (struct merged_store_group): Move up
11406 bit_insertion field and declare can_be_merged_into method.
11407 (merged_store_group::can_be_merged_into): New method.
11408 (imm_store_chain_info::coalesce_immediate): Call it to decide whether
11409 consecutive non-overlapping stores can be merged. Turn MEM_REF stores
11410 into BIT_INSERT_EXPR stores if the group contains a non-MEM_REF store.
11412 2018-06-04 Richard Biener <rguenther@suse.de>
11414 PR tree-optimization/85955
11415 * builtins.c (fold_builtin_sincos): Convert pointers to
11416 destination to appropriate type before dereferencing.
11418 2018-06-04 Segher Boessenkool <segher@kernel.crashing.org>
11420 * config/rs6000/rs6000.md (abs<mode>2 for FLOAT128): Handle IFmode.
11422 2018-06-04 Richard Sandiford <richard.sandiford@linaro.org>
11424 * expr.c (expand_expr_real_1): Force the operand into memory if
11425 its TYPE_MODE is BLKmode and if there is no integer mode for
11426 the number of bits being extracted.
11428 2018-06-04 Jakub Jelinek <jakub@redhat.com>
11432 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>_1):
11433 Use vptestnm rather than vptestm in (=Yc,v,C) variant.
11435 2018-06-04 Richard Biener <rguenther@suse.de>
11437 * tree-cfgcleanup.c (cleanup_tree_cfg_1): Fold into...
11438 (cleanup_tree_cfg_noloop): ... single caller. Do
11439 start_recording_case_labels later.
11441 2018-06-04 Sebastian Peryt <sebastian.peryt@intel.com>
11443 * config/i386/cldemoteintrin.h: Change define from _X86INTRIN_H_INCLUDED
11444 to _IMMINTRIN_H_INCLUDED.
11445 * config/i386/pconfigintrin.h: Ditto.
11446 * config/i386/waitpkgintrin.h: Ditto.
11447 * config/i386/immintrin.h: Add includes for sgxintrin.h,
11448 pconfigintrin.h, waitpkgintrin.h and cldemoteintrin.h.
11449 * config/i386/x86intrin.h: Remove includes for mintrin.h, xmmintrin.h,
11450 emmintrin.h, pmmintrin.h, tmmintrin.h, smmintrin.h, wmmintrin.h,
11451 bmiintrin.h, bmi2intrin.h, lzcntintrin.h, sgxintrin.h, pconfigintrin.h,
11452 waitpkgintrin.h and cldemoteintrin.h.
11454 2018-06-04 Richard Biener <rguenther@suse.de>
11456 PR tree-optimization/86038
11457 * tracer.c (find_best_successor): Check probability for
11458 being initialized, bail out if not.
11460 2018-06-04 Richard Earnshaw <rearnsha@arm.com>
11463 * config/arm/arm-cpus.in (ALL_QUIRKS): Add xscale feature to the list
11464 of bits to ignore when comparing architectures.
11466 2018-06-04 Jakub Jelinek <jakub@redhat.com>
11468 PR tree-optimization/69615
11469 * fold-const.c (merge_ranges): If range1 is - [x, x] and x is the
11470 maximum or minimum of the type, try to merge it also as if
11471 range1 is + [-, x - 1] or + [x + 1, -].
11474 * tree.c (inchash::add_expr): Handle IDENTIFIER_NODE.
11476 2018-06-03 Eric Botcazou <ebotcazou@adacore.com>
11478 PR tree-optimization/86034
11479 * gimple-ssa-store-merging.c (output_merged_store): Convert the RHS to
11480 the unsigned bitfield type in a bit insertion sequence if it does not
11481 have a larger precision than the bitfield size.
11482 (process_store): Also bypass widening conversions for BIT_INSERT_EXPR.
11484 2018-06-03 Kito Cheng <kito.cheng@gmail.com>
11486 * config/nds32/nds32-peephole2.md: Add new patterns for code size.
11488 2018-06-03 Chung-Ju Wu <jasonwucj@gmail.com>
11490 * config/nds32/nds32-opts.h (nds32_arch_type): Add ARCH_V3J.
11491 * config/nds32/nds32.c (nds32_option_override): Consider ARCH_V3J.
11492 * config/nds32/nds32.h (TARGET_ISA_V3): Add ARCH_V3J checking.
11493 * config/nds32/nds32.opt (march): Add enum value Value(ARCH_V3J).
11495 2018-06-03 Chung-Ju Wu <jasonwucj@gmail.com>
11497 * common/config/nds32/nds32-common.c (nds32_option_optimization_table):
11498 Disable -fdelete-null-pointer-checks for ELF toolchain.
11500 2018-06-02 Chung-Ju Wu <jasonwucj@gmail.com>
11501 Kito Cheng <kito.cheng@gmail.com>
11503 * config.gcc (nds32*): Use nds32-linux.opt and nds32-elf.opt.
11504 (nds32le-*-*, nds32be-*-*): Integrate checking process.
11505 (nds32*-*-*): Add glibc and uclibc conditions.
11506 * common/config/nds32/nds32-common.c (nds32_except_unwind_info): New.
11507 (TARGET_EXCEPT_UNWIND_INFO): Define.
11508 * config/nds32/elf.h: New file.
11509 * config/nds32/linux.h: New file.
11510 * config/nds32/nds32-elf.opt: New file.
11511 * config/nds32/nds32-linux.opt: New file.
11512 * config/nds32/nds32-fp-as-gp.c
11513 (pass_nds32_fp_as_gp::gate): Consider TARGET_LINUX_ABI.
11514 * config/nds32/nds32.c (nds32_conditional_register_usage): Consider
11516 (nds32_asm_file_end): Ditto.
11517 (nds32_print_operand): Ditto.
11518 (nds32_insert_attributes): Ditto.
11519 (nds32_init_libfuncs): New function.
11520 (TARGET_HAVE_TLS): Define.
11521 (TARGET_INIT_LIBFUNCS): Define.
11522 * config/nds32/nds32.h (TARGET_DEFAULT_RELAX): Apply different relax
11524 (TARGET_ELF): Apply different mcmodel setting.
11525 (LINK_SPEC, LIB_SPEC, STARTFILE_SPEC, ENDFILE_SPEC): The content has
11526 been migrated into elf.h and linux.h files.
11527 * config/nds32/nds32.md (add_pc): Consider TARGET_LINUX_ABI.
11528 * config/nds32/nds32.opt (mvh): Consider TARGET_LINUX_ABI.
11529 (mcmodel): The content has been migrated into nds32-elf.opt and
11530 nds32-linux.opt files.
11531 * config/nds32/t-elf: New file.
11532 * config/nds32/t-linux: New file.
11534 2018-06-02 Chung-Ju Wu <jasonwucj@gmail.com>
11535 Shiva Chen <shiva0217@gmail.com>
11537 * config/nds32/constants.md (unspec_volatile_element): Add
11538 UNSPEC_VOLATILE_OMIT_FP_BEGIN and UNSPEC_VOLATILE_OMIT_FP_END.
11539 * config/nds32/nds32-fp-as-gp.c: New implementation of fp_as_gp
11541 * config/nds32/nds32-protos.h (nds32_naked_function_p): Declare.
11542 (make_pass_nds32_fp_as_gp): Declare.
11543 * config/nds32/nds32.c (nds32_register_passes): Add fp_as_gp as one
11545 (nds32_asm_function_end_prologue): Remove unused asm output.
11546 (nds32_asm_function_begin_epilogue): Remove unused asm output.
11547 (nds32_asm_file_start): Output necessary fp_as_gp information.
11548 (nds32_option_override): Adjust register usage.
11549 (nds32_expand_prologue): Consider fp_as_gp situation.
11550 (nds32_expand_prologue_v3push): Consider fp_as_gp situation.
11551 * config/nds32/nds32.md (prologue): Check fp_as_gp_p and naked_p.
11554 (simple_return): Ditto.
11555 (omit_fp_begin): Output special directive for fp_as_gp.
11556 (omit_fp_end): Output special directive for fp_as_gp.
11557 * config/nds32/nds32.opt (mfp-as-gp, mno-fp-as-gp, mforce-fp-as-gp,
11558 mforbid-fp-as-gp): New options.
11560 2018-06-01 Mark Wielaard <mark@klomp.org>
11562 * dwarf2out.c (dwarf2out_finish): Remove generation of
11563 DW_AT_loclists_base.
11565 2018-06-01 Eric Botcazou <ebotcazou@adacore.com>
11567 * gimple-ssa-store-merging.c: Include gimple-fold.h.
11568 (struct store_immediate_info): Document BIT_INSERT_EXPR stores.
11569 (struct merged_store_group): Add bit_insertion field.
11570 (dump_char_array): Use standard hexadecimal format.
11571 (merged_store_group::merged_store_group): Set bit_insertion to false.
11572 (merged_store_group::apply_stores): Use optimal buffer size. Deal
11573 with BIT_INSERT_EXPR stores. Move up code updating the mask and
11574 also print the mask in the dump file.
11575 (pass_store_merging::gate): Minor tweak.
11576 (imm_store_chain_info::coalesce_immediate): Fix wrong association
11577 of stores with groups in dump. Allow coalescing of BIT_INSERT_EXPR
11578 stores with INTEGER_CST stores.
11579 (count_multiple_uses) <BIT_INSERT_EXPR>: New case.
11580 (imm_store_chain_info::output_merged_store): Add try_bitpos variable
11581 and use it throughout. Generate bit insertion sequences if need be.
11582 (pass_store_merging::process_store): Remove redundant condition.
11583 Record stores from a SSA name to a bit-field with BIT_INSERT_EXPR.
11585 2018-06-01 Segher Boessenkool <segher@kernel.crashing.org>
11587 * config/rs6000/rs6000.c (rs6000_mangle_type): Change the mangling of
11588 the 128-bit floating point types. Fix function comment.
11590 2018-06-01 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11592 * config/aarch64/aarch64-simd.md
11593 (aarch64_simd_vec_unpack<su>_lo_<mode>): Use UXTL and SXTL assembler
11595 (aarch64_simd_vec_unpack<su>_hi_<mode>): Use UXTL2 and SXTL2 assembler
11598 2018-06-01 Richard Sandiford <richard.sandiford@linaro.org>
11600 PR tree-optimization/85989
11601 * gimple-ssa-backprop.c (backprop::m_visited_phis): New member
11603 (backprop::intersect_uses): Check it when deciding whether this
11604 is a backedge reference.
11605 (backprop::process_block): Add each phi to m_visited_phis
11606 after visiting it, then clear it at the end.
11608 2018-06-01 Richard Biener <rguenther@suse.de>
11610 * tree-vectorizer.h (vect_dr_stmt): New function.
11611 (vect_get_load_cost): Adjust.
11612 (vect_get_store_cost): Likewise.
11613 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence):
11614 Use vect_dr_stmt instead of DR_SMTT.
11615 (vect_record_base_alignments): Likewise.
11616 (vect_calculate_target_alignment): Likewise.
11617 (vect_compute_data_ref_alignment): Likewise and make static.
11618 (vect_update_misalignment_for_peel): Likewise.
11619 (vect_verify_datarefs_alignment): Likewise.
11620 (vector_alignment_reachable_p): Likewise.
11621 (vect_get_data_access_cost): Likewise. Pass down
11622 vinfo to vect_get_load_cost/vect_get_store_cost instead of DR.
11623 (vect_get_peeling_costs_all_drs): Likewise.
11624 (vect_peeling_hash_get_lowest_cost): Likewise.
11625 (vect_enhance_data_refs_alignment): Likewise.
11626 (vect_find_same_alignment_drs): Likewise.
11627 (vect_analyze_data_refs_alignment): Likewise.
11628 (vect_analyze_group_access_1): Likewise.
11629 (vect_analyze_group_access): Likewise.
11630 (vect_analyze_data_ref_access): Likewise.
11631 (vect_analyze_data_ref_accesses): Likewise.
11632 (vect_vfa_segment_size): Likewise.
11633 (vect_small_gap_p): Likewise.
11634 (vectorizable_with_step_bound_p): Likewise.
11635 (vect_prune_runtime_alias_test_list): Likewise.
11636 (vect_analyze_data_refs): Likewise.
11637 (vect_supportable_dr_alignment): Likewise.
11638 * tree-vect-loop-manip.c (get_misalign_in_elems): Likewise.
11639 (vect_gen_prolog_loop_niters): Likewise.
11640 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
11641 * tree-vect-patterns.c (vect_recog_bool_pattern): Do not
11643 (vect_recog_mask_conversion_pattern): Likewise.
11644 (vect_try_gather_scatter_pattern): Likewise.
11645 * tree-vect-stmts.c (vect_model_store_cost): Pass stmt_info
11646 to vect_get_store_cost.
11647 (vect_get_store_cost): Get stmt_info instead of DR.
11648 (vect_model_load_cost): Pass stmt_info to vect_get_load_cost.
11649 (vect_get_load_cost): Get stmt_info instead of DR.
11651 2018-06-01 Richard Biener <rguenther@suse.de>
11653 PR middle-end/86017
11654 * gimple-fold.c (var_decl_component_p): Also allow offsetted
11655 vars wrapped in MEM_REFs.
11657 2018-06-01 Richard Sandiford <richard.sandiford@linaro.org>
11659 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
11660 Fix subreg tests so that we only return a choice between
11661 GENERAL_REGS and FP_REGS if the original classes included both.
11663 2018-06-01 Richard Biener <rguenther@suse.de>
11666 * tree-ssa-structalias.c (get_function_part_constraint):
11667 Handle NULL fi->decl.
11668 (find_func_aliases_for_call): Properly handle indirect
11669 fi from direct call.
11670 (find_func_clobbers): Likewise.
11671 (ipa_pta_execute): Likewise.
11672 (create_variable_info_for): For functions that are ifunc_resolver
11673 resolve to a varinfo that contains the result of the resolver call.
11674 (associate_varinfo_to_alias): Do not treat ifunc resolvers as
11677 2018-05-31 Michael Collison <michael.collison@arm.com>
11679 * config/aarch64/aarch64.md:
11680 (*fix_to_zero_extenddfdi2): New pattern.
11681 * gcc.target/aarch64/fix_extend1.c: New testcase.
11683 2018-05-31 Qing Zhao <qing.zhao@oracle.com>
11685 PR middle-end/78809
11686 PR middle-end/83026
11687 * builtins.c (expand_builtin): Add the handling of BUILT_IN_STRCMP_EQ
11688 and BUILT_IN_STRNCMP_EQ.
11689 * builtins.def: Add new builtins BUILT_IN_STRCMP_EQ and
11690 BUILT_IN_STRNCMP_EQ.
11691 * gimple-fold.c (gimple_fold_builtin_string_compare): Add the
11692 handling of BUILTIN_IN_STRCMP_EQ and BUILT_IN_STRNCMP_EQ.
11693 (gimple_fold_builtin): Likewise.
11694 * tree-ssa-strlen.c (compute_string_length): New function.
11695 (determine_min_obsize): New function.
11696 (handle_builtin_string_cmp): New function to handle calls to
11697 string compare functions.
11698 (strlen_optimize_stmt): Add handling to builtin string compare
11700 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
11701 Add the handling of BUILT_IN_STRCMP_EQ and BUILT_IN_STRNCMP_EQ.
11702 * tree.c (build_common_builtin_nodes): Add new defines of
11703 BUILT_IN_STRNCMP_EQ and BUILT_IN_STRCMP_EQ.
11705 2018-05-31 Jakub Jelinek <jakub@redhat.com>
11708 * bb-reorder.c (pass_partition_blocks::gate): Return false for
11709 functions with naked attribute.
11711 2018-05-31 Uros Bizjak <ubizjak@gmail.com>
11713 * config/i386/sse.md (avx_vec_concat<mode>):
11714 Substitute concat_tg_mode mode attribute with xtg_mode.
11715 (<mask_codefor>avx512dq_broadcast<mode><mask_name>_1): Ditto.
11716 (concat_tg_mode): Remove mode attribute.
11718 2018-05-31 Martin Sebor <msebor@redhat.com>
11721 * calls.c (alloc_max_size): Correct a logic error/typo.
11722 Treat excessive arguments as infinite. Warn for invalid arguments.
11723 * doc/invoke.texi (-Walloc-size-larger-than): Update.
11725 2018-05-31 H.J. Lu <hongjiu.lu@intel.com>
11728 * config/i386/x86-tune.def: Re-enable partial_reg_dependency
11729 and movx for Haswell.
11731 2018-05-31 Chung-Lin Tang <cltang@codesourcery.com>
11732 Cesar Philippidis <cesar@codesourcery.com>
11734 PR middle-end/85879
11735 * gimplify.c (gimplify_adjust_omp_clauses): Add 'remove = true'
11736 when emitting error on private/firstprivate reductions.
11737 * omp-low.c (lower_omp_target): Avoid reference-type processing
11738 on pointers for firstprivate clause.
11740 2018-05-31 Sameera Deshpande <sameera.deshpande@linaro.org>
11742 * config/aarch64/aarch64-simd-builtins.def (ld1x3): New.
11745 * config/aarch64/aarch64-simd.md
11746 (aarch64_ld1x3<VALLDIF:mode>): New pattern.
11747 (aarch64_ld1_x3_<mode>): Likewise
11748 (aarch64_st1x2<VALLDIF:mode>): Likewise
11749 (aarch64_st1_x2_<mode>): Likewise
11750 (aarch64_st1x3<VALLDIF:mode>): Likewise
11751 (aarch64_st1_x3_<mode>): Likewise
11752 * config/aarch64/arm_neon.h (vld1_u8_x3): New function.
11753 (vld1_s8_x3): Likewise.
11754 (vld1_u16_x3): Likewise.
11755 (vld1_s16_x3): Likewise.
11756 (vld1_u32_x3): Likewise.
11757 (vld1_s32_x3): Likewise.
11758 (vld1_u64_x3): Likewise.
11759 (vld1_s64_x3): Likewise.
11760 (vld1_f16_x3): Likewise.
11761 (vld1_f32_x3): Likewise.
11762 (vld1_f64_x3): Likewise.
11763 (vld1_p8_x3): Likewise.
11764 (vld1_p16_x3): Likewise.
11765 (vld1_p64_x3): Likewise.
11766 (vld1q_u8_x3): Likewise.
11767 (vld1q_s8_x3): Likewise.
11768 (vld1q_u16_x3): Likewise.
11769 (vld1q_s16_x3): Likewise.
11770 (vld1q_u32_x3): Likewise.
11771 (vld1q_s32_x3): Likewise.
11772 (vld1q_u64_x3): Likewise.
11773 (vld1q_s64_x3): Likewise.
11774 (vld1q_f16_x3): Likewise.
11775 (vld1q_f32_x3): Likewise.
11776 (vld1q_f64_x3): Likewise.
11777 (vld1q_p8_x3): Likewise.
11778 (vld1q_p16_x3): Likewise.
11779 (vld1q_p64_x3): Likewise.
11780 (vst1_s64_x2): Likewise.
11781 (vst1_u64_x2): Likewise.
11782 (vst1_f64_x2): Likewise.
11783 (vst1_s8_x2): Likewise.
11784 (vst1_p8_x2): Likewise.
11785 (vst1_s16_x2): Likewise.
11786 (vst1_p16_x2): Likewise.
11787 (vst1_s32_x2): Likewise.
11788 (vst1_u8_x2): Likewise.
11789 (vst1_u16_x2): Likewise.
11790 (vst1_u32_x2): Likewise.
11791 (vst1_f16_x2): Likewise.
11792 (vst1_f32_x2): Likewise.
11793 (vst1_p64_x2): Likewise.
11794 (vst1q_s8_x2): Likewise.
11795 (vst1q_p8_x2): Likewise.
11796 (vst1q_s16_x2): Likewise.
11797 (vst1q_p16_x2): Likewise.
11798 (vst1q_s32_x2): Likewise.
11799 (vst1q_s64_x2): Likewise.
11800 (vst1q_u8_x2): Likewise.
11801 (vst1q_u16_x2): Likewise.
11802 (vst1q_u32_x2): Likewise.
11803 (vst1q_u64_x2): Likewise.
11804 (vst1q_f16_x2): Likewise.
11805 (vst1q_f32_x2): Likewise.
11806 (vst1q_f64_x2): Likewise.
11807 (vst1q_p64_x2): Likewise.
11808 (vst1_s64_x3): Likewise.
11809 (vst1_u64_x3): Likewise.
11810 (vst1_f64_x3): Likewise.
11811 (vst1_s8_x3): Likewise.
11812 (vst1_p8_x3): Likewise.
11813 (vst1_s16_x3): Likewise.
11814 (vst1_p16_x3): Likewise.
11815 (vst1_s32_x3): Likewise.
11816 (vst1_u8_x3): Likewise.
11817 (vst1_u16_x3): Likewise.
11818 (vst1_u32_x3): Likewise.
11819 (vst1_f16_x3): Likewise.
11820 (vst1_f32_x3): Likewise.
11821 (vst1_p64_x3): Likewise.
11822 (vst1q_s8_x3): Likewise.
11823 (vst1q_p8_x3): Likewise.
11824 (vst1q_s16_x3): Likewise.
11825 (vst1q_p16_x3): Likewise.
11826 (vst1q_s32_x3): Likewise.
11827 (vst1q_s64_x3): Likewise.
11828 (vst1q_u8_x3): Likewise.
11829 (vst1q_u16_x3): Likewise.
11830 (vst1q_u32_x3): Likewise.
11831 (vst1q_u64_x3): Likewise.
11832 (vst1q_f16_x3): Likewise.
11833 (vst1q_f32_x3): Likewise.
11834 (vst1q_f64_x3): Likewise.
11835 (vst1q_p64_x3): Likewise.
11837 2018-05-30 Jozef Lawrynowicz <jozef.l@mittosystems.com>
11839 * config/msp430/msp430.c (msp430_output_labelref): Prepend
11840 user_label_prefix to name.
11842 * tree-core.h: Update comment about the format of NAME string
11843 passed to handler in attribute_spec.
11845 * config/msp430/msp430.md: Remove erroneous subreg expression from
11846 zero_extendqisi2 insn pattern. Remove msp430x ISA restriction on
11847 zero_extend{q,h}isi2.
11849 2018-05-30 Borislav Petkov <bp@suse.de>
11851 * doc/extend.texi: Document some architecture specific
11852 constraints and sort entries.
11854 2018-05-30 Martin Sebor <msebor@redhat.com>
11856 PR middle-end/85369
11857 * builtins.c (expand_builtin_stpcpy_1): New function.
11858 (expand_builtin_stpcpy): Call it, and call maybe_warn_nonstring_arg
11859 only if the former succeeds.
11861 2018-05-31 Sameera Deshpande <sameera.deshpande@linaro.org>
11863 * config/aarch64/aarch64-cores.def (saphira) : Add support for ARMv8.4
11866 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
11868 * doc/invoke.texi (-flinker-output): Document
11870 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
11872 * passes.c (ipa_write_summaries): Only modify statements if body
11874 * cgraphunit.c (ipa_passes): Also produce intermeidate code when
11875 incrementally linking.
11876 (ipa_passes): Likewise.
11877 * lto-cgraph.c (lto_output_node): When incrementally linking do not
11878 pass down resolution info.
11879 * common.opt (flag_incremental_link): Update info.
11880 * gcc.c (plugin specs): Turn flinker-output=* to
11881 -plugin-opt=-linker-output-known
11882 * toplev.c (compile_file): Also cut compilation when doing incremental
11884 * flag-types. (enum lto_partition_model): Add
11885 LTO_LINKER_OUTPUT_NOLTOREL.
11886 (invoke.texi): Add -flinker-output docs.
11887 * ipa.c (symbol_table::remove_unreachable_nodes): Handle LTO incremental
11888 link same way as WPA; do not stream in dead initializers.
11890 * dwarf2out.c (dwarf2out_die_ref_for_decl,
11891 darf2out_register_external_decl): Support incremental link.
11893 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
11895 * lto-opts.c (lto_write_options): Skip OPT_dumpdir, OPT_fresolution_.
11897 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
11899 * lto-wrapper.c (debug_objcopy): Add rename parameter; pass
11900 it down to simple_object_copy_lto_debug_sections.
11901 (run_gcc): Determine incremental LTO link time and configure
11902 lto1 into non-wpa mode, disable renaming of debug sections.
11904 2018-05-30 Kelvin Nilsen <kelvin@gcc.gnu.org>
11906 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Remove
11907 descriptions of various incorrectly documented functions.
11909 2018-05-30 Andre Vieira <andre.simoesdiasvieira@arm.com>
11912 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Make
11913 address check not strict.
11915 2018-05-30 Richard Biener <rguenther@suse.de>
11917 PR tree-optimization/85964
11918 * tracer.c (better_p): Drop initialized count check, we only
11919 call the function with initialized counts now.
11920 (find_best_successor): Do find a best edge if one
11921 has uninitialized count.
11922 (find_best_predecessor): Likewise. Do BB frequency check only
11923 if count is initialized.
11925 2017-05-30 Jackson Woodruff <jackson.woodruff@arm.com>
11927 * config/aarch64/aarch64.c (aarch64_host_wide_int_compare): New.
11928 (aarch64_ldrstr_offset_compare): New.
11929 (aarch64_operands_adjust_ok_for_ldpstp): Update to consider all
11930 load/store orderings.
11931 (aarch64_gen_adjusted_ldpstp): Likewise.
11933 2018-05-30 Wilco Dijkstra <wdijkstr@arm.com>
11935 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
11936 Check for subset of GENERAL_REGS and FP_REGS.
11937 * config/aarch64/aarch64-simd.md (aarch64_get_lane): Increase cost of
11940 2018-05-30 Richard Sandiford <richard.sandiford@linaro.org>
11942 * alias.c (adjust_offset_for_component_ref): Use poly_int_tree_p
11943 and wi::to_poly_offset. Add the current offset and then check
11944 whether the sum fits, rather than using an unchecked addition of
11945 a checked term. Check for a shwi rather than a uhwi.
11946 * expr.c (get_bit_range): Use tree_to_poly_uint64.
11947 (store_constructor): Use poly_int_tree_p.
11948 (expand_expr_real_1): Likewise.
11949 * function.c (assign_temp): Likewise.
11950 * fold-const.c (const_binop): Use poly_int_tree_p and
11951 wi::to_poly_offset.
11952 (fold_indirect_ref_1): Likewise. Use multiple_p to attempt an exact
11954 * ipa-icf-gimple.c (func_checker::compare_operand): Use
11955 to_poly_offset for MEM offsets.
11956 * ipa-icf.c (sem_variable::equals): Likewise.
11957 * stor-layout.c (compute_record_mode): Use poly_int_tree_p.
11958 * tree-ssa-sccvn.c (ao_ref_init_from_vn_reference): Use
11959 wi::to_poly_offset for BIT_FIELD_REF offsets.
11960 (vn_reference_maybe_forwprop_address): Use poly_int_tree_p and
11961 wi::to_poly_offset.
11962 * var-tracking.c (emit_note_insn_var_location): Use
11963 tree_to_poly_uint64.
11965 2018-05-29 Jim Wilson <jimw@sifive.com>
11967 * config/riscv/riscv.c (riscv_interrupt_type): Fix comment typo.
11969 2018-05-29 Uros Bizjak <ubizjak@gmail.com>
11972 * config/i386/i386.md (l<rounding_insn><MODEF:mode><SWI48:mode>2):
11973 Enable for TARGET_SSE4_1 and generate rounds{s,d} and cvtts{s,d}2si{,q}
11975 (sse4_1_round<mode>2): Use nonimmediate_operand
11976 for operand 1 predicate.
11978 2018-05-29 Martin Sebor <msebor@redhat.com>
11979 Richard Biener <rguenther@suse.de>
11982 * calls.c (get_size_range): Call determine_value_range instead
11983 of get_value_range..
11984 * tree-vrp.h (determine_value_range): Declared new function.
11985 * tree-vrp.c (determine_value_range_1, determine_value_range): New.
11987 2018-05-29 Richard Biener <rguenther@suse.de>
11989 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Make
11990 sure to use non-pattern stmts for get_earlier_stmt arguments.
11991 * tree-vectorizer.h (get_earlier_stmt): Assert we do not get
11992 called on pattern stmts.
11993 (get_later_stmt): Likewise.
11995 2018-05-29 Martin Liska <mliska@suse.cz>
11997 PR gcov-profile/85759
11998 * doc/gcov.texi: Document GCOV_ERROR_FILE and GCOV_EXIT_AT_ERROR
12001 2018-05-29 Jakub Jelinek <jakub@redhat.com>
12003 * tree-cfg.c (verify_gimple_assign_unary): Add checking for
12005 (verify_gimple_assign_binary): Check TYPE_VECTOR_SUBPARTS for
12009 * tree.def (VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR,
12010 VEC_PACK_FLOAT_EXPR): New tree codes.
12011 * tree-pretty-print.c (op_code_prio): Handle
12012 VEC_UNPACK_FIX_TRUNC_HI_EXPR and VEC_UNPACK_FIX_TRUNC_LO_EXPR.
12013 (dump_generic_node): Handle VEC_UNPACK_FIX_TRUNC_HI_EXPR,
12014 VEC_UNPACK_FIX_TRUNC_LO_EXPR and VEC_PACK_FLOAT_EXPR.
12015 * tree-inline.c (estimate_operator_cost): Likewise.
12016 * gimple-pretty-print.c (dump_binary_rhs): Handle VEC_PACK_FLOAT_EXPR.
12017 * fold-const.c (const_binop): Likewise.
12018 (const_unop): Handle VEC_UNPACK_FIX_TRUNC_HI_EXPR and
12019 VEC_UNPACK_FIX_TRUNC_LO_EXPR.
12020 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
12021 (verify_gimple_assign_binary): Handle VEC_PACK_FLOAT_EXPR.
12022 * cfgexpand.c (expand_debug_expr): Handle VEC_UNPACK_FIX_TRUNC_HI_EXPR,
12023 VEC_UNPACK_FIX_TRUNC_LO_EXPR and VEC_PACK_FLOAT_EXPR.
12024 * expr.c (expand_expr_real_2): Likewise.
12025 * optabs.def (vec_packs_float_optab, vec_packu_float_optab,
12026 vec_unpack_sfix_trunc_hi_optab, vec_unpack_sfix_trunc_lo_optab,
12027 vec_unpack_ufix_trunc_hi_optab, vec_unpack_ufix_trunc_lo_optab): New
12029 * optabs.c (expand_widen_pattern_expr): For
12030 VEC_UNPACK_FIX_TRUNC_HI_EXPR and VEC_UNPACK_FIX_TRUNC_LO_EXPR use
12031 sign from result type rather than operand's type.
12032 (expand_binop_directly): For vec_packu_float_optab and
12033 vec_packs_float_optab allow result type to be different from operand's
12035 * optabs-tree.c (optab_for_tree_code): Handle
12036 VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR and
12037 VEC_PACK_FLOAT_EXPR. Formatting fixes.
12038 * tree-vect-generic.c (expand_vector_operations_1): Handle
12039 VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR and
12040 VEC_PACK_FLOAT_EXPR.
12041 * tree-vect-stmts.c (supportable_widening_operation): Handle
12043 (supportable_narrowing_operation): Handle FLOAT_EXPR.
12044 * config/i386/i386.md (fixprefix, floatprefix): New code attributes.
12045 * config/i386/sse.md (*float<floatunssuffix>v2div2sf2): Rename to ...
12046 (float<floatunssuffix>v2div2sf2): ... this. Formatting fix.
12047 (vpckfloat_concat_mode, vpckfloat_temp_mode, vpckfloat_op_mode): New
12049 (vec_pack<floatprefix>_float_<mode>): New expander.
12050 (vunpckfixt_mode, vunpckfixt_model, vunpckfixt_extract_mode): New mode
12052 (vec_unpack_<fixprefix>fix_trunc_lo_<mode>,
12053 vec_unpack_<fixprefix>fix_trunc_hi_<mode>): New expanders.
12054 * doc/md.texi (vec_packs_float_@var{m}, vec_packu_float_@var{m},
12055 vec_unpack_sfix_trunc_hi_@var{m}, vec_unpack_sfix_trunc_lo_@var{m},
12056 vec_unpack_ufix_trunc_hi_@var{m}, vec_unpack_ufix_trunc_lo_@var{m}):
12058 * doc/generic.texi (VEC_UNPACK_FLOAT_HI_EXPR,
12059 VEC_UNPACK_FLOAT_LO_EXPR): Fix pasto in description.
12060 (VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR,
12061 VEC_PACK_FLOAT_EXPR): Document.
12063 2018-05-29 Richard Biener <rguenther@suse.de>
12065 * tree-vectorizer.h (struct vec_info): Add stmt_vec_infos
12067 (stmt_vec_info_vec): Make pointer.
12068 (init_stmt_vec_info_vec): Remove.
12069 (free_stmt_vec_info_vec): Likewise.
12070 (set_stmt_vec_info_vec): New function.
12071 (free_stmt_vec_infos): Likewise.
12072 (vinfo_for_stmt): Adjust for stmt_vec_info_vec indirection.
12073 (set_vinfo_for_stmt): Likewise.
12074 (get_earlier_stmt): Likewise.
12075 (get_later_stmt): Likewise.
12076 * tree-vectorizer.c (stmt_vec_info_vec): Make pointer.
12077 (vec_info::vec_info): Allocate stmt_vec_infos and set the global.
12078 (vec_info::~vec_info): Free stmt_vec_infos.
12079 (vectorize_loops): Set the global stmt_vec_info_vec to NULL.
12080 Remove old init_stmt_vec_info_vec/free_stmt_vec_info_vec calls.
12081 (pass_slp_vectorize::execute): Likewise.
12082 * tree-vect-stmts.c (init_stmt_vec_info_vec): Remove.
12083 (free_stmt_vec_info_vec): Likewise.
12084 (set_stmt_vec_info_vec): New function.
12085 (free_stmt_vec_infos): Likewise.
12086 * tree-vect-loop.c (_loop_vec_info::~_loop_vec_info): Set
12087 the global stmt_vec_info_vec.
12088 * tree-parloops.c (gather_scalar_reductions): Use
12089 set_stmt_vec_info_vec/free_stmt_vec_infos and maintain a local
12092 2018-05-29 Richard Biener <rguenther@suse.de>
12094 * dominance.c (iterate_fix_dominators): Push/pop TV_DOMINANCE.
12096 2018-05-29 Martin Liska <mliska@suse.cz>
12097 David Malcolm <dmalcolm@redhat.com>
12099 * vec.c (test_reverse): New.
12100 (vec_c_tests): Add new test.
12101 * vec.h (vl_ptr>::reverse): New function.
12103 2018-05-29 Gerald Pfeifer <gerald@pfeifer.com>
12105 * config.gcc: Identify FreeBSD 3.x and 4.x as unsupported.
12107 * config/freebsd-spec.h (FBSD_LIB_SPEC): Only consider FreeBSD 5
12110 2018-05-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
12112 * tree-dump.c (dump_node): Use splay_tree_delete_pointers.
12114 2018-05-28 Richard Biener <rguenther@suse.de>
12116 PR tree-optimization/85933
12117 * tree-vect-data-refs.c (vect_record_base_alignments): Only
12118 look at stmts marked as vectorizable.
12120 2018-05-28 Richard Biener <rguenther@suse.de>
12122 PR tree-optimization/85934
12123 * tree-vect-generic.c (expand_vector_operations_1): Hoist
12124 vector boolean check before scalar optimization.
12126 2018-05-28 Jakub Jelinek <jakub@redhat.com>
12128 * doc/invoke.texi (ARM Options): Use @item instead of @itemx
12131 2018-05-28 Mark Wielaard <mark@klomp.org>
12133 * dwarf2asm.c (dw2_asm_output_delta_uleb128): Add brackets around lab2
12134 if it is an expression containing a minus sign.
12136 2018-05-27 John David Anglin <danglin@gcc.gnu.org>
12138 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Define to 0.
12140 2018-05-27 Paul Koning <ni1d@arrl.net>
12142 * config/pdp11/pdp11.md (truncsihi2): Remove.
12144 2018-05-27 Monk Chiang <sh.chiang04@gmail.com>
12145 Chung-Ju Wu <jasonwucj@gmail.com>
12147 * config/nds32/nds32-intrinsic.md (unaligned_storedi): Modify patterns
12149 (unaligned_store_dw): Ditto.
12150 * config/nds32/nds32-memory-manipulation.c
12151 (nds32_expand_movmemsi_loop_known_size): Refactoring implementation.
12152 (nds32_gen_dup_4_byte_to_word_value): Rename to ...
12153 (nds32_gen_dup_4_byte_to_word_value_aux): ... this.
12154 (emit_setmem_word_loop): Rename to ...
12155 (emit_setmem_doubleword_loop): ... this.
12156 (nds32_gen_dup_4_byte_to_word_value): New function.
12157 (nds32_gen_dup_8_byte_to_double_word_value): New function.
12158 (nds32_expand_setmem_loop): Refine implementation.
12159 (nds32_expand_setmem_loop_v3m): Ditto.
12160 * config/nds32/nds32-multiple.md (unaligned_store_update_base_dw): New
12163 2018-05-27 Chung-Ju Wu <jasonwucj@gmail.com>
12165 * config/nds32/nds32.md (bswapsi2, bswaphi2): New patterns.
12167 2018-05-27 Chung-Ju Wu <jasonwucj@gmail.com>
12169 * config/nds32/nds32.c (nds32_attribute_table): Add "no_prologue".
12170 (nds32_init_machine_status): Initialize machine->attr_naked_p and
12171 machine->attr_no_prologue_p.
12172 (nds32_compute_stack_frame): Check "naked" and "no_prologue" attributes.
12173 (nds32_naked_function_p): Handle "naked" and "no_prologue" attributes.
12174 (nds32_expand_epilogue): Consider attr_naked_p.
12175 (nds32_expand_epilogue_v3pop): Likewise.
12176 (nds32_can_use_return_insn): Likewise.
12177 * config/nds32/nds32.h (machine_function): Add attr_naked_p and
12178 attr_no_prologue_p fields.
12179 * config/nds32/nds32.opt (mret-in-naked-func): New option.
12181 2018-05-27 Jakub Jelinek <jakub@redhat.com>
12184 * config/i386/i386.md (fixunssuffix, floatunssuffix): New code
12186 * config/i386/sse.md
12187 (<floatsuffix>float<sseintvecmodelower><mode>2<mask_name><round_name>):
12189 (float<floatunssuffix><sseintvecmodelower><mode>2<mask_name><round_name>):
12191 (<floatsuffix>float<sselongvecmodelower><mode>2<mask_name><round_name>):
12193 (float<floatunssuffix><sselongvecmodelower><mode>2<mask_name><round_name>):
12195 (*<floatsuffix>floatv2div2sf2): Rename to ...
12196 (*float<floatunssuffix>v2div2sf2): ... this.
12197 (<floatsuffix>floatv2div2sf2_mask): Rename to ...
12198 (float<floatunssuffix>v2div2sf2_mask): ... this.
12199 (*<floatsuffix>floatv2div2sf2_mask_1): Rename to ...
12200 (*float<floatunssuffix>v2div2sf2_mask_1): ... this.
12201 (<fixsuffix>fix_truncv8dfv8si2<mask_name><round_saeonly_name>): Rename
12203 (fix<fixunssuffix>_truncv8dfv8si2<mask_name><round_saeonly_name>):
12205 (<fixsuffix>fix_trunc<mode><sseintvecmodelower>2<mask_name><round_saeonly_name>):
12207 (fix<fixunssuffix>_trunc<mode><sseintvecmodelower>2<mask_name><round_saeonly_name>):
12209 (<fixsuffix>fix_trunc<mode><sselongvecmodelower>2<mask_name><round_saeonly_name>):
12211 (fix<fixunssuffix>_trunc<mode><sselongvecmodelower>2<mask_name><round_saeonly_name>):
12213 (<fixsuffix>fix_truncv2sfv2di2<mask_name>): Rename to ...
12214 (fix<fixunssuffix>_truncv2sfv2di2<mask_name>): ... this.
12215 (vec_pack_ufix_trunc_<mode>): Use gen_fixuns_truncv8dfv8si2 instead of
12216 gen_ufix_truncv8dfv8si2.
12217 * config/i386/i386-builtin.def (__builtin_ia32_cvttpd2uqq256_mask,
12218 __builtin_ia32_cvttpd2uqq128_mask, __builtin_ia32_cvttps2uqq256_mask,
12219 __builtin_ia32_cvttps2uqq128_mask, __builtin_ia32_cvtuqq2ps256_mask,
12220 __builtin_ia32_cvtuqq2ps128_mask, __builtin_ia32_cvtuqq2pd256_mask,
12221 __builtin_ia32_cvtuqq2pd128_mask, __builtin_ia32_cvttpd2udq512_mask,
12222 __builtin_ia32_cvtuqq2ps512_mask, __builtin_ia32_cvtuqq2pd512_mask,
12223 __builtin_ia32_cvttps2uqq512_mask, __builtin_ia32_cvttpd2uqq512_mask):
12224 Use fixuns instead ufix or floatuns instead ufloat in CODE_FOR_ names.
12226 2018-05-24 H.J. Lu <hongjiu.lu@intel.com>
12230 * varasm.c (assemble_alias): Lookup ifunc attribute on error.
12232 2018-05-25 Jim Wilson <jimw@sifive.com>
12234 * config/riscv/riscv-protos.h (riscv_epilogue_uses): New.
12235 * config/riscv/riscv.c (struct machine_function): Add
12236 interrupt_handler_p and attribute_checked_p fields.
12237 (riscv_attribute_table): Add interrupt.
12238 (riscv_interrupt_type_p): New.
12239 (riscv_save_reg_p): Save extra regs for interrupt handler.
12240 (riscv_use_save_libcall): Return false for interrupt handler.
12241 (riscv_first_stack_step): Add forward declaration.
12242 (riscv_compute_frame_info): New local interrupt_save_t1. Set it
12243 for interrupt handler with large frame. Use it for saved reg list.
12244 (riscv_expand_prologue): Move flag_stack_usage_info support to
12245 eliminate duplication.
12246 (riscv_expand_epilogue): Generate mret for interrupt handler.
12247 (riscv_epilogue_uses): New.
12248 (riscv_can_use_return_insn): Return false for interrupt handler.
12249 (riscv_function_ok_for_sibcall): Likewise.
12250 (riscv_set_current_function): Add interrupt handler support.
12251 * config/riscv/riscv.h (EPILOGUE_USES): Call riscv_epilogue_uses.
12252 * config/riscv/riscv.md (UNSPECV_MRET): New.
12254 (riscv_frflags, riscv_fsflags): Use tab after opcode.
12256 * doc/extend.texi (RISC-V Function Attributes) <interrupt>: New.
12258 2018-05-25 Bill Schmidt <wschmidt@linux.ibm.com>
12260 PR tree-optimization/85712
12261 * gimple-ssa-strength-reduction.c (replace_one_candidate): Skip if
12262 this candidate has already been replaced in-situ by a copy.
12264 2018-05-25 Jason Merrill <jason@redhat.com>
12266 PR c++/80485 - inline function non-zero address.
12267 * symtab.c (nonzero_address): Check DECL_COMDAT.
12269 2018-05-25 Uros Bizjak <ubizjak@gmail.com>
12272 * config/alpha/alpha.md (ashlsi3): New insn pattern.
12273 (*ashlsi_se): Rename from *ashldi_se. Define as sign
12274 extension of SImode operation. Use const123_operand predicate.
12275 (*saddsi_1): Remove.
12276 (*saddl_se_1): Ditto.
12277 (*ssubsi_1): Ditto.
12278 (*ssubl_se_1): Ditto.
12279 * config/alpha/predicates.md (const123_operand): New predicate.
12280 * config/alpha/constraints.md (P): Use IN_RANGE.
12282 2018-05-25 Richard Biener <rguenther@suse.de>
12284 * tree-ssa-alias.h (refs_may_alias_p): Add tbaa_p bool parameter,
12286 (ref_maybe_used_by_stmt_p): Likewise.
12287 (stmt_may_clobber_ref_p): Likewise.
12288 (stmt_may_clobber_ref_p_1): Likewise.
12289 * tree-ssa-alias.c (refs_may_alias_p): Add tbaa_p bool parameter
12291 (ref_maybe_used_by_stmt_p): Likewise.
12292 (stmt_may_clobber_ref_p): Likewise.
12293 (stmt_may_clobber_ref_p_1): Likewise.
12294 * tree-vect-data-refs.c (vect_slp_analyze_node_dependences): Use
12295 the alias oracle to disambiguate DRs with stmts DR analysis
12297 (vect_analyze_data_refs): Do not give up on not analyzable
12298 DRs for BB vectorization. Remove code truncating the dataref
12301 2018-05-25 Jakub Jelinek <jakub@redhat.com>
12304 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>_1):
12305 Add (=Yk,v,C) variant using vptestm insn. Use TARGET_AVX512BW
12306 in test instead of TARGET_AVX512F for VI12_AVX512VL iterator.
12308 2018-05-25 Richard Biener <rguenther@suse.de>
12310 * tree-vect-data-refs.c (vect_find_stmt_data_reference): New
12311 function, combining stmt data ref gathering and fatal analysis
12313 (vect_analyze_data_refs): Remove now redudnant code and simplify.
12314 * tree-vect-loop.c (vect_get_datarefs_in_loop): Factor out from
12315 vect_analyze_loop_2 and use vect_find_stmt_data_reference.
12316 * tree-vect-slp.c (vect_slp_bb): Use vect_find_stmt_data_reference.
12317 * tree-vectorizer.h (vect_find_stmt_data_reference): Declare.
12319 2018-05-25 Bin Cheng <bin.cheng@arm.com>
12321 PR tree-optimization/85720
12322 * tree-loop-distribution.c (break_alias_scc_partitions): Don't merge
12323 SCC if all partitions are builtins.
12324 (version_loop_by_alias_check): New parameter. Generate cancelable
12325 runtime alias check if all partitions are builtins.
12326 (distribute_loop): Update call to above function.
12328 2018-05-25 Bin Cheng <bin.cheng@arm.com>
12330 * tree-outof-ssa.c (tree-ssa.h, tree-dfa.h): Include header files.
12331 (create_default_def, for_all_parms): Moved from tree-ssa-coalesce.c.
12332 (parm_default_def_partition_arg): Ditto.
12333 (set_parm_default_def_partition): Ditto.
12334 (get_parm_default_def_partitions): Ditto and make it static.
12335 (get_undefined_value_partitions): Ditto and make it static.
12336 (remove_ssa_form): Refactor call to init_var_map here.
12337 * tree-ssa-coalesce.c (build_ssa_conflict_graph): Support live range
12338 computation for loop region.
12339 (coalesce_partitions, compute_optimized_partition_bases): Ditto.
12340 (register_default_def): Delete.
12341 (for_all_parms, create_default_def): Move to tree-outof-ssa.c.
12342 (parm_default_def_partition_arg): Ditto.
12343 (set_parm_default_def_partition): Ditto.
12344 (get_parm_default_def_partitions): Ditto and make it static.
12345 (get_undefined_value_partitions): Ditto and make it static.
12346 (coalesce_with_default, coalesce_with_default): Update comment.
12347 (create_coalesce_list_for_region): New func factored out from
12348 create_outofssa_var_map.
12349 (populate_coalesce_list_for_outofssa): New func factored out from
12350 create_outofssa_var_map and coalesce_ssa_name.
12351 (create_outofssa_var_map): Delete.
12352 (coalesce_ssa_name): Refactor to support live range computation.
12353 * tree-ssa-coalesce.h (coalesce_ssa_name): Change decl.
12354 (get_parm_default_def_partitions): Delete.
12355 (get_undefined_value_partitions): Ditto.
12356 * tree-ssa-live.c (init_var_map, delete_var_map): Support live range
12357 computation for loop region.
12358 (new_tree_live_info, loe_visit_block): Ditto.
12359 (live_worklist, set_var_live_on_entry): Ditto.
12360 (calculate_live_on_exit, verify_live_on_entry): Ditto.
12361 * tree-ssa-live.h (struct _var_map): New fields.
12362 (init_var_map): Change decl.
12363 (region_contains_p): New.
12365 2018-05-25 Bin Cheng <bin.cheng@arm.com>
12367 * tree-ssa-live.h (live_merge_and_clear): Delete.
12369 2018-05-25 Richard Biener <rguenther@suse.de>
12372 * tree-dump.c (dequeue_and_dump): Remove access to removed
12373 operand 2 of a SWITCH_EXPR.
12375 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
12377 * doc/sourcebuild.texi (vect_double_cond_arith): Include
12378 multiplication and division.
12379 * doc/md.texi (cond_mul@var{m}, cond_div@var{m}, cond_mod@var{m})
12380 (cond_udiv@var{m}, cond_umod@var{m}): Document.
12381 * optabs.def (cond_smul_optab, cond_sdiv_optab, cond_smod_optab)
12382 (cond_udiv_optab, cond_umod_optab): New optabs.
12383 * internal-fn.def (IFN_COND_MUL, IFN_COND_DIV, IFN_COND_MOD)
12384 (IFN_COND_RDIV): New internal functions.
12385 * internal-fn.c (get_conditional_internal_fn): Handle TRUNC_DIV_EXPR,
12386 TRUNC_MOD_EXPR and RDIV_EXPR.
12387 * match.pd (UNCOND_BINARY, COND_BINARY): Handle them.
12388 * config/aarch64/iterators.md (UNSPEC_COND_MUL, UNSPEC_COND_DIV):
12390 (SVE_INT_BINARY): Include mult.
12391 (SVE_COND_FP_BINARY): Include UNSPEC_MUL and UNSPEC_DIV.
12392 (optab, sve_int_op): Handle mult.
12393 (optab, sve_fp_op, commutative): Handle UNSPEC_COND_MUL and
12395 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): New pattern
12396 for SVE_INT_BINARY_SD.
12398 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
12400 * config/aarch64/iterators.md (SVE_INT_BINARY_SD): New code iterator.
12401 (optab, sve_int_op): Handle div and udiv.
12402 * config/aarch64/aarch64-sve.md (<optab><mode>3): New expander
12403 for SVE_INT_BINARY_SD.
12404 (*<optab><mode>3): New insn for the same.
12406 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
12408 * tree-vect-patterns.c: Include predict.h.
12409 (vect_recog_divmod_pattern): Restrict check for division support
12410 to when optimizing for size.
12412 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
12414 * doc/sourcebuild.texi (vect_double_cond_arith: Document.
12415 * gimple-match.h (gimple_match_op::MAX_NUM_OPS): Bump to 4.
12416 (gimple_match_op::gimple_match_op): Add an overload for 4 operands.
12417 (gimple_match_op::set_op): Likewise.
12418 (gimple_resimplify4): Declare.
12419 * genmatch.c (get_operand_type): Handle CFN_COND_* functions.
12420 (expr::gen_transform): Likewise.
12421 (decision_tree::gen): Generate a simplification routine for 4 operands.
12422 * gimple-match-head.c (gimple_simplify): Add an overload for
12423 4 operands. In the top-level function, handle up to 4 call
12424 arguments and call gimple_resimplify4.
12425 (gimple_resimplify4): New function.
12426 (build_call_internal): Pass a fourth operand.
12427 (maybe_push_to_seq): Likewise.
12428 * match.pd (UNCOND_BINARY, COND_BINARY): New operator lists.
12429 Fold VEC_COND_EXPRs of an operation and a default value into
12430 an IFN_COND_* function if possible.
12431 * config/aarch64/iterators.md (UNSPEC_COND_MAX, UNSPEC_COND_MIN):
12433 (SVE_COND_FP_BINARY): Include them.
12434 (optab, sve_fp_op): Handle them.
12435 (SVE_INT_BINARY_REV): New code iterator.
12436 (SVE_COND_FP_BINARY_REV): New int iterator.
12437 (commutative): New int attribute.
12438 * config/aarch64/aarch64-protos.h (aarch64_sve_prepare_conditional_op):
12440 * config/aarch64/aarch64.c (aarch64_sve_prepare_conditional_op): New
12442 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): Use it.
12443 (*cond_<optab><mode>): New patterns for reversed operands.
12445 2018-05-25 Richard Biener <rguenther@suse.de>
12447 * tree-vectorizer.h (STMT_VINFO_GROUP_*, GROUP_*): Remove.
12448 (DR_GROUP_*): New, assert we have non-NULL ->data_ref_info.
12449 (REDUC_GROUP_*): New, assert we have NULL ->data_ref_info.
12450 (STMT_VINFO_GROUPED_ACCESS): Adjust.
12451 * tree-vect-data-refs.c (everywhere): Adjust users.
12452 * tree-vect-loop.c (everywhere): Likewise.
12453 * tree-vect-slp.c (everywhere): Likewise.
12454 * tree-vect-stmts.c (everywhere): Likewise.
12455 * tree-vect-patterns.c (vect_reassociating_reduction_p): Likewise.
12457 2018-05-25 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
12459 * configure.ac (gcc_cv_as_section_has_e): Move to common section.
12461 (gcc_cv_as_section_exclude): ... this.
12462 Try Solaris as #exclude syntax.
12463 * configure: Regenerate.
12464 * config.in: Regenerate.
12465 * config/i386/i386.c (i386_solaris_elf_named_section): Handle
12467 * config/sparc/sparc.c (sparc_solaris_elf_asm_named_section)
12468 [HAVE_GAS_SECTION_EXCLUDE]: Handle SECTION_EXCLUDE.
12470 * varasm.c (default_elf_asm_named_section): Don't check if
12471 HAVE_GAS_SECTION_EXCLUDE is defined.
12473 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
12475 * doc/md.texi: Update the documentation of the cond_* optabs
12476 to mention the new final operand. Fix GET_MODE_NUNITS call.
12477 Describe the scalar case too.
12478 * internal-fn.def (IFN_EXTRACT_LAST): Change type to fold_left.
12479 * internal-fn.c (expand_cond_unary_optab_fn): Expect 3 operands
12481 (expand_cond_binary_optab_fn): Expect 4 operands instead of 3.
12482 (get_conditional_internal_fn): Update comment.
12483 * tree-vect-loop.c (vectorizable_reduction): Pass the original
12484 accumulator value as a final argument to conditional functions.
12485 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): Turn into
12486 a define_expand and add an "else" operand. Assert for now that
12487 the else operand is equal to operand 2. Use SVE_INT_BINARY and
12488 SVE_COND_FP_BINARY instead of SVE_COND_INT_OP and SVE_COND_FP_OP.
12489 (*cond_<optab><mode>): New patterns.
12490 * config/aarch64/iterators.md (UNSPEC_COND_SMAX, UNSPEC_COND_UMAX)
12491 (UNSPEC_COND_SMIN, UNSPEC_COND_UMIN, UNSPEC_COND_AND, UNSPEC_COND_ORR)
12492 (UNSPEC_COND_EOR): Delete.
12493 (optab): Remove associated mappings.
12494 (SVE_INT_BINARY): New code iterator.
12495 (sve_int_op): Remove int attribute and add "minus" to the code
12497 (SVE_COND_INT_OP): Delete.
12498 (SVE_COND_FP_OP): Rename to...
12499 (SVE_COND_FP_BINARY): ...this.
12501 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
12503 * optabs.c (can_reuse_operands_p): New function.
12504 (maybe_legitimize_operands): Try to reuse the results for
12507 2018-05-24 Uros Bizjak <ubizjak@gmail.com>
12509 * config/i386/sse.md (cvtusi2<ssescalarmodesuffix>64<round_name>):
12510 Add {q} suffix to insn mnemonic.
12512 2018-05-23 Jozef Lawrynowicz <jozef.l@mittosystems.com>
12514 * config/msp430/msp430.c (TARGET_WARN_FUNC_RETURN): Define.
12515 (msp430_warn_func_return): New.
12517 2018-05-24 Roger Sayle <roger@nextmovesoftware.com>
12519 * fold-const.c (tree_nonzero_bits): New function.
12520 * fold-const.h (tree_nonzero_bits): Likewise.
12521 * match.pd (POPCOUNT): New patterns to fold BUILTIN_POPCOUNT and
12522 friends. POPCOUNT(x&1) => x&1, POPCOUNT(x)==0 => x==0, etc.
12524 2018-05-24 H.J. Lu <hongjiu.lu@intel.com>
12528 * varasm.c (assemble_alias): Check ifunc_resolver only on
12531 2018-05-24 Uros Bizjak <ubizjak@gmail.com>
12534 * config/i386/sse.md (movdi_to_sse): Do not generate pseudo
12535 when memory input operand is handled.
12537 2018-05-24 Luis Machado <luis.machado@linaro.org>
12539 * config/aarch64/aarch64.c (qdf24xx_addrcost_table): New static
12541 (qdf24xx_tunings) <addr_costs>: Set to qdf24xx_addrcost_table.
12543 2018-05-24 Richard Sandiford <richard.sandiford@linaro.org>
12545 * match.pd: Delay FMA folds until after vectorization.
12547 2018-05-24 Andre Vieira <andre.simoesdiasvieira@arm.com>
12550 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Make
12551 address check not strict.
12553 2018-05-24 Richard Sandiford <richard.sandiford@linaro.org>
12555 * gimple-match.h (gimple_match_op): New class.
12556 (mprts_hook): Replace parameters with a gimple_match_op *.
12557 (maybe_build_generic_op): Likewise.
12558 (gimple_simplified_result_is_gimple_val): Replace parameters with
12559 a const gimple_match_op *.
12560 (gimple_simplify): Replace code_helper * and tree * parameters with
12561 a gimple_match_op * parameter.
12562 (gimple_resimplify1): Replace code_helper *, tree and tree *
12563 parameters with a gimple_match_op * parameter.
12564 (gimple_resimplify2): Likewise.
12565 (gimple_resimplify3): Likewise.
12566 (maybe_push_res_to_seq): Replace code_helper, tree and tree *
12567 parameters with a gimple_match_op * parameter.
12568 * gimple-match-head.c (gimple_simplify): Change prototypes of
12569 auto-generated functions to take a gimple_match_op * instead of
12570 separate code_helper * and tree * parameters. Make the same
12571 change in the top-level overload and update calls to the
12572 gimple_resimplify routines. Update calls to the auto-generated
12573 functions and to maybe_push_res_to_seq in the publicly-facing
12574 operation-specific gimple_simplify overloads.
12575 (gimple_match_op::MAX_NUM_OPS): Define.
12576 (gimple_resimplify1): Replace rcode and ops with a single res_op
12577 parameter. Update call to gimple_simplify.
12578 (gimple_resimplify2): Likewise.
12579 (gimple_resimplify3): Likewise.
12580 (mprts_hook): Replace parameters with a gimple_match_op *.
12581 (maybe_build_generic_op): Likewise.
12582 (build_call_internal): Replace type, nargs and ops with
12583 a gimple_match_op *.
12584 (maybe_push_res_to_seq): Replace res_code, type and ops parameters
12585 with a single gimple_match_op *. Update calls to mprts_hook,
12586 build_call_internal and gimple_simplified_result_is_gimple_val.
12587 Factor out code that is common to the tree_code and combined_fn cases.
12588 * genmatch.c (expr::gen_transform): Replace tem_code and
12589 tem_ops with a gimple_match_op called tem_op. Update calls
12590 to the gimple_resimplify functions and maybe_push_res_to_seq.
12591 (dt_simplify::gen_1): Manipulate res_op instead of res_code and
12592 res_ops. Update call to the gimple_resimplify functions.
12593 (dt_simplify::gen): Pass res_op instead of res_code and res_ops.
12594 (decision_tree::gen): Make the functions take a gimple_match_op *
12595 called res_op instead of separate res_code and res_ops parameters.
12596 Update call accordingly.
12597 * gimple-fold.c (replace_stmt_with_simplification): Replace rcode
12598 and ops with a single res_op parameter. Update calls to
12599 maybe_build_generic_op and maybe_push_res_to_seq.
12600 (fold_stmt_1): Update calls to gimple_simplify and
12601 replace_stmt_with_simplification.
12602 (gimple_fold_stmt_to_constant_1): Update calls to gimple_simplify
12603 and gimple_simplified_result_is_gimple_val.
12604 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update call to
12606 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Replace parameters
12607 with a gimple_match_op *.
12608 (vn_nary_build_or_lookup): Likewise. Update call to
12609 vn_nary_build_or_lookup_1.
12610 (vn_nary_build_or_lookup_1): Replace rcode, type and ops with a
12611 gimple_match_op *. Update calls to the gimple_resimplify routines
12612 and to gimple_simplified_result_is_gimple_val.
12613 (vn_nary_simplify): Update call to vn_nary_build_or_lookup_1.
12614 Use gimple_match_op::MAX_NUM_OPS instead of a hard-coded 3.
12615 (vn_reference_lookup_3): Update call to vn_nary_build_or_lookup.
12616 (visit_nary_op): Likewise.
12617 (visit_reference_op_load): Likewise.
12619 2018-05-23 Luis Machado <luis.machado@linaro.org>
12621 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Use correct type
12622 modifier for printing the step amount.
12624 2018-05-23 Jozef Lawrynowicz <jozef.l@somniumtech.com>
12627 * tree.c (build_common_tree_nodes): Dont set TYPE_SIZE for __intN
12630 2018-05-23 Segher Boessenkool <segher@kernel.crashing.org>
12632 * doc/sourcebuild.texi (Endianness): New subsubsection.
12634 2018-05-23 Luis Machado <luis.machado@linaro.org>
12636 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
12637 <prefetch_dynamic_strides>: New const bool field.
12638 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
12639 prefetch_dynamic_strides.
12640 (exynosm1_prefetch_tune): Likewise.
12641 (thunderxt88_prefetch_tune): Likewise.
12642 (thunderx_prefetch_tune): Likewise.
12643 (thunderx2t99_prefetch_tune): Likewise.
12644 (qdf24xx_prefetch_tune): Likewise. Set prefetch_dynamic_strides to
12646 (aarch64_override_options_internal): Update to set
12647 PARAM_PREFETCH_DYNAMIC_STRIDES.
12648 * doc/invoke.texi (prefetch-dynamic-strides): Document new option.
12649 * params.def (PARAM_PREFETCH_DYNAMIC_STRIDES): New.
12650 * params.h (PARAM_PREFETCH_DYNAMIC_STRIDES): Define.
12651 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Account for
12652 prefetch-dynamic-strides setting.
12654 2018-05-23 Luis Machado <luis.machado@linaro.org>
12656 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
12657 <minimum_stride>: New const int field.
12658 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
12659 minimum_stride field defaulting to -1.
12660 (exynosm1_prefetch_tune): Likewise.
12661 (thunderxt88_prefetch_tune): Likewise.
12662 (thunderx_prefetch_tune): Likewise.
12663 (thunderx2t99_prefetch_tune): Likewise.
12664 (qdf24xx_prefetch_tune) <minimum_stride>: Set to 2048.
12665 <default_opt_level>: Set to 3.
12666 (aarch64_override_options_internal): Update to set
12667 PARAM_PREFETCH_MINIMUM_STRIDE.
12668 * doc/invoke.texi (prefetch-minimum-stride): Document new option.
12669 * params.def (PARAM_PREFETCH_MINIMUM_STRIDE): New.
12670 * params.h (PARAM_PREFETCH_MINIMUM_STRIDE): Define.
12671 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Return false if
12672 stride is constant and is below the minimum stride threshold.
12674 2018-05-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12676 * config/arm/arm-cpus.in (mode26): Delete.
12677 (armv4): Delete mode26 reference.
12678 * config/arm/arm.c (arm_configure_build_target): Delete use of
12681 2018-05-23 Uros Bizjak <ubizjak@gmail.com>
12683 * config/i386/i386.md (*floatuns<SWI48:mode><MODEF:mode>2_avx512):
12685 (floatunssi<mode>2): Also enable for AVX512F and TARGET_SSE_MATH.
12686 Rewrite expander pattern. Emit gen_floatunssi<mode>2_i387_with_xmm
12688 (floatunsdisf2): Rewrite expander pattern. Hanlde TARGET_AVX512F.
12689 (floatunsdidf2): Ditto.
12691 2018-05-23 Uros Bizjak <ubizjak@gmail.com>
12693 * config/i386/i386.md (fixuns_trunc<mode>di2): New insn pattern.
12694 (fixuns_trunc<mode>si2_avx512f): Ditto.
12695 (*fixuns_trunc<mode>si2_avx512f_zext): Ditto.
12696 (fixuns_trunc<mode>si2): Also enable for AVX512F and TARGET_SSE_MATH.
12697 Emit fixuns_trunc<mode>si2_avx512f for AVX512F targets.
12699 2018-05-23 Alexander Monakov <amonakov@ispras.ru>
12701 PR rtl-optimization/79985
12702 * df-scan.c (df_insn_refs_collect): Remove special case for
12703 global registers and asm statements.
12705 2018-05-23 Alexander Monakov <amonakov@ispras.ru>
12707 * extend.texi (Global Register Variables): Rewrite the bullet list.
12708 Note that the register is available for allocation. Note that access
12709 via inline asm must use constraints. Add note about async-signal
12710 handlers. Remove paragraph about automagic register selection.
12712 2018-05-23 Richard Biener <rguenther@suse.de>
12714 * tree-ssa-sccvn.c (vn_reference_lookup_3): Remove restriction
12715 of fixed offset from memset VN.
12717 2018-05-23 Bill Schmidt <wschmidt@linux.ibm.com>
12719 * gimple-ssa-strength-reduction.c (struct slsr_cand_d): Add
12720 first_interp field.
12721 (alloc_cand_and_find_basis): Initialize first_interp field.
12722 (slsr_process_mul): Modify first_interp field.
12723 (slsr_process_add): Likewise.
12724 (slsr_process_cast): Modify first_interp field for each new
12726 (slsr_process_copy): Likewise.
12727 (dump_candidate): Dump first_interp field.
12728 (replace_mult_candidate): Process all interpretations, not just
12730 (replace_rhs_if_not_dup): Likewise.
12731 (replace_one_candidate): Likewise.
12733 2018-05-23 Wilco Dijkstra <wdijkstr@arm.com>
12735 * config/aarch64/aarch64.c (aarch64_use_frame_pointer):
12737 (aarch64_needs_frame_chain): New function.
12738 (aarch64_parse_override_string): Set aarch64_use_frame_pointer.
12740 2018-05-23 Sudakshina Das <sudi.das@arm.com>
12743 * common/config/aarch64/aarch64-common.c (aarch64_handle_option):
12744 Check val before adding MASK_STRICT_ALIGN to opts->x_target_flags.
12745 * config/aarch64/aarch64.opt (mstrict-align): Remove RejectNegative.
12746 * config/aarch64/aarch64.c (aarch64_attributes): Mark allow_neg
12747 as true for strict-align.
12748 (aarch64_can_inline_p): Perform checks even when callee has no
12749 attributes to check for strict alignment.
12750 * doc/extend.texi (AArch64 Function Attributes): Document
12752 * doc/invoke.texi: (AArch64 Options): Likewise.
12754 2018-05-23 Richard Sandiford <richard.sandiford@linaro.org>
12756 PR tree-optimization/85853
12757 * tree-vect-slp.c (vect_slp_analyze_node_operations): Split out
12758 the handling of the root of the node to...
12759 (vect_slp_analyze_node_operations_1): ...this new function,
12760 and run the whole thing with the child nodes' def types
12761 set according to their SLP node's def type.
12763 2018-05-23 Richard Biener <rguenther@suse.de>
12765 PR middle-end/85874
12766 * tree-data-ref.c (create_runtime_alias_checks): Defer
12767 and ignore overflow warnings.
12769 2018-05-23 Yury Gribov <tetra2005@gmail.com>
12771 PR tree-optimization/85822
12772 * tree-vrp.c (is_masked_range_test): Fix handling of negative
12775 2018-05-23 Richard Biener <rguenther@suse.de>
12777 * tree-ssa-sccvn.c (vn_reference_lookup_3): Handle arbitrary
12778 memset constants via native_interpret_expr.
12780 2018-05-22 H.J. Lu <hongjiu.lu@intel.com>
12783 * cgraph.h (cgraph_node::create): Set ifunc_resolver for ifunc
12785 (cgraph_node::create_alias): Likewise.
12786 (cgraph_node::get_availability): Check ifunc_resolver instead
12787 of looking up ifunc attribute.
12788 * cgraphunit.c (maybe_diag_incompatible_alias): Likewise.
12789 * varasm.c (do_assemble_alias): Likewise.
12790 (assemble_alias): Likewise.
12791 (default_binds_local_p_3): Likewise.
12792 * cgraph.h (cgraph_node): Add ifunc_resolver.
12793 (cgraph_node::only_called_directly_or_aliased_p): Return false
12794 for IFUNC resolver.
12795 * lto-cgraph.c (input_node): Set ifunc_resolver for ifunc
12797 * symtab.c (symtab_node::verify_base): Verify that ifunc_resolver
12798 is equivalent to lookup_attribute ("ifunc", DECL_ATTRIBUTES (decl)).
12799 (symtab_node::binds_to_current_def_p): Check ifunc_resolver
12800 instead of looking up ifunc attribute.
12802 2018-05-22 Luis Machado <luis.machado@linaro.org>
12804 * config/aarch64/aarch64.md (*ashift<mode>_extv_bfiz): New pattern.
12806 2018-05-22 Martin Sebor <msebor@redhat.com>
12808 PR middle-end/85359
12809 * builtins.c (expand_builtin_strcpy): Call maybe_warn_nonstring_arg
12810 only when expasion succeeds.
12811 (expand_builtin_strcmp): Same.
12812 (expand_builtin_strncmp): Same.
12814 2018-05-22 Martin Sebor <msebor@redhat.com>
12816 * calls.c (maybe_warn_nonstring_arg): Fix a typo in a comment.
12818 2018-05-22 Jackson Woodruff <jackson.woodruff@arm.com>
12819 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12821 * config/aarch64/aarch64-ldpstp.md: Replace uses of
12822 aarch64_mem_pair_operand with memory_operand and delete operand swapping
12824 * config/aarch64/aarch64.c (aarch64_operands_ok_for_ldpstp):
12825 Add check for legitimate_address.
12826 (aarch64_gen_adjusted_ldpstp): Swap operands where appropriate.
12827 (aarch64_swap_ldrstr_operands): New.
12828 * config/aarch64/aarch64-protos.h (aarch64_swap_ldrstr_operands):
12831 2018-05-22 Jackson Woodruff <jackson.woodruff@arm.com>
12832 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12834 * config/aarch64/aarch64.md: New patterns to generate stp
12836 (store_pair_sw, store_pair_dw): New patterns to generate stp for
12837 single words and double words.
12838 (load_pair_sw, load_pair_dw): Likewise.
12839 (store_pair_sf, store_pair_df, store_pair_si, store_pair_di):
12841 (load_pair_sf, load_pair_df, load_pair_si, load_pair_di):
12843 * config/aarch64/aarch64-ldpstp.md: Modify peephole
12844 for different mode ldpstp and add peephole for merged zero stores.
12845 Likewise for loads.
12846 * config/aarch64/aarch64.c (aarch64_operands_ok_for_ldpstp):
12848 (aarch64_gen_store_pair): Rename calls to match new patterns.
12849 (aarch64_gen_load_pair): Rename calls to match new patterns.
12850 * config/aarch64/aarch64-simd.md (load_pair<mode>): Rename to...
12851 (load_pair<DREG:mode><DREG2:mode>): ... This.
12852 (store_pair<mode>): Rename to...
12853 (vec_store_pair<DREG:mode><DREG2:mode>): ... This.
12854 * config/aarch64/iterators.md (DREG, DREG2, DX2, SX, SX2, DSX):
12855 New mode iterators.
12856 (V_INT_EQUIV): Handle SImode.
12857 * config/aarch64/predicates.md (aarch64_reg_zero_or_fp_zero):
12860 2018-05-22 Martin Sebor <msebor@redhat.com>
12863 * calls.c (maybe_warn_nonstring_arg): Use string length to set
12864 or ajust the presumed bound on an operation to avoid unnecessary
12867 2018-05-22 Martin Sebor <msebor@redhat.com>
12869 PR tree-optimization/85826
12870 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Avoid
12871 assuming that a DECL necesarily has a constant size.
12873 2018-05-22 Richard Sandiford <richard.sandiford@linaro.org>
12875 PR middle-end/85862
12876 * internal-fn.c (expand_direct_optab_fn): Cope with a null lhs.
12878 2018-05-22 Richard Biener <rguenther@suse.de>
12880 PR tree-optimization/85834
12881 * tree-ssa-sccvn.c (vn_reference_lookup_3): Properly handle
12882 non-constant and non-zero memset arguments.
12884 2018-05-22 Martin Liska <mliska@suse.cz>
12887 * ipa-icf.c (sem_item::add_type): Do not ICE for incomplete types.
12889 2018-05-22 Richard Biener <rguenther@suse.de>
12891 PR tree-optimization/85863
12892 * tree-vect-stmts.c (vect_is_simple_cond): Only widen invariant
12893 comparisons when vectype is specified.
12894 (vectorizable_condition): Do not specify vectype for
12895 vect_is_simple_cond when SLP vectorizing.
12897 2018-05-21 Michael Meissner <meissner@linux.ibm.com>
12900 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Do not
12901 define __ibm128 as long double.
12902 * config/rs6000/rs6000.c (rs6000_init_builtins): Create __ibm128
12903 as a distinct type when IEEE 128-bit support is enabled.
12904 (init_float128_ieee): Fix up conversions between IFmode and IEEE
12905 128-bit types to use the correct functions.
12906 (rs6000_expand_float128_convert): Use explicit FLOAT_EXTEND to
12907 convert between 128-bit floating point types that have different
12908 modes but the same representation, instead of using gen_lowpart to
12910 * config/rs6000/rs6000.md (IFKF): New iterator for IFmode and
12912 (IFKF_reg): New attributes to give the register constraints for
12914 (extend<mode>tf2_internal): New insns to mark an explicit
12915 conversion between 128-bit floating point types that have a
12916 different mode but share the same representation.
12918 2018-05-21 Richard Sandiford <richard.sandiford@linaro.org>
12920 PR tree-optimization/85814
12921 * tree-ssa-strlen.c (get_stridx_plus_constant): Cope with
12922 a null return from get_strinfo when unsharing the next
12923 strinfo in the chain.
12925 2018-05-21 Vladimir Mezentsev <vladimir.mezentsev@oracle.com>
12928 * varasm.c (weak_finish): Clean up weak_decls.
12930 2018-05-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12932 * config/aarch64/aarch64.md ("unspec"): Define UNSPEC_SABAL,
12933 UNSPEC_SABDL2, UNSPEC_SADALP, UNSPEC_UABAL, UNSPEC_UABDL2,
12934 UNSPEC_UADALP values.
12935 * config/aarch64/iterators.md (ABAL): New int iterator.
12938 (sur): Add mappings for the above.
12939 * config/aarch64/aarch64-simd.md (aarch64_<sur>abdl2<mode>_3):
12941 (aarch64_<sur>abal<mode>_4): Likewise.
12942 (aarch64_<sur>adalp<mode>_3): Likewise.
12943 (<sur>sadv16qi): New define_expand.
12945 2018-05-21 Alexander Nesterovskiy <alexander.nesterovskiy@intel.com>
12947 * config/i386/i386.md (*movsf_internal): AVX falsedep fix.
12948 (*movdf_internal): Ditto.
12949 (*rcpsf2_sse): Ditto.
12950 (*rsqrtsf2_sse): Ditto.
12951 (*sqrt<mode>2_sse): Ditto.
12953 2018-05-21 Tamar Christina <tamar.christina@arm.com>
12955 * config/aarch64/aarch64-simd.md (aarch64_eor3qv8hi): Change to
12957 (aarch64_bcaxqv8hi): Change to bcaxq<mode>4.
12958 * config/aarch64/aarch64-simd-builtins.def (veor3q_u8, veor3q_u32,
12959 veor3q_u64, veor3q_s8, veor3q_s16, veor3q_s32, veor3q_s64, vbcaxq_u8,
12960 vbcaxq_u32, vbcaxq_u64, vbcaxq_s8, vbcaxq_s16, vbcaxq_s32,
12962 * config/aarch64/arm_neon.h: Likewise.
12963 * config/aarch64/iterators.md (VQ_I): New.
12965 2018-05-21 Alexey Brodkin <abrodkin@synopsys.com>
12967 * config.gcc: Add arc/t-multilib-linux to tmake_file for
12969 * config/arc/t-multilib-linux: Specify MULTILIB_OPTIONS and
12972 2018-05-20 Chung-Ju Wu <jasonwucj@gmail.com>
12974 * config/nds32/constraints.md (S): New constraint.
12975 * config/nds32/nds32.md (call_internal): Use constraint S.
12976 (call_value_internal): Likewise.
12977 (sibcall_internal): Likewise.
12978 (sibcall_value_internal): Likewise.
12980 2018-05-20 Kito Cheng <kito.cheng@gmail.com>
12981 Chung-Ju Wu <jasonwucj@gmail.com>
12983 * config/nds32/nds32.c (nds32_register_move_cost): Take garywolf cpu
12984 into consideration.
12986 2018-05-20 Kito Cheng <kito.cheng@gmail.com>
12987 Chung-Ju Wu <jasonwucj@gmail.com>
12989 * config/nds32/nds32-cost.c (rtx_cost_model_t): New structure.
12990 (insn_size_16bit, insn_size_32bit): New variables for cost evaluation.
12991 (nds32_rtx_costs_impl): Simplify.
12992 (nds32_address_cost_impl): Simplify.
12993 (nds32_init_rtx_costs): New function.
12994 (nds32_rtx_costs_speed_prefer): Likewise.
12995 (nds32_rtx_costs_size_prefer): Likewise.
12996 (nds32_address_cost_speed_prefer): Likewise.
12997 (nds32_address_cost_speed_fwprop): Likewise.
12998 (nds32_address_cost_size_prefer): Likewise.
12999 * config/nds32/nds32-protos.h (nds32_init_rtx_costs): Declare.
13000 * config/nds32/nds32.c (nds32_option_override): Use
13001 nds32_init_rtx_costs function.
13003 2018-05-20 Chung-Ju Wu <jasonwucj@gmail.com>
13005 * config/nds32/nds32.c (nds32_asm_file_start): Output pipeline model.
13006 * config/nds32/nds32.h (TARGET_PIPELINE_N7): Define.
13007 (TARGET_PIPELINE_N8): Likewise.
13008 (TARGET_PIPELINE_N10): Likewise.
13009 (TARGET_PIPELINE_N13): Likewise.
13010 (TARGET_PIPELINE_GRAYWOLF): Likewise.
13012 2018-05-19 Monk Chiang <sh.chiang04@gmail.com>
13014 * config/nds32/nds32-fpu.md: Update copyright year.
13016 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
13018 * config/nds32/nds32.h (ASM_SPEC): Adjust spec rule.
13020 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
13022 * config/nds32/nds32.c
13023 (nds32_md_asm_adjust): Consider flag_inline_asm_r15 variable.
13024 * config/nds32/nds32.opt (minline-asm-r15): New option.
13026 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
13028 * common/config/nds32/nds32-common.c (TARGET_DEFAULT_TARGET_FLAGS): Add
13030 * config/nds32/nds32.md (abssi2): New pattern.
13032 2018-05-19 Uros Bizjak <ubizjak@gmail.com>
13034 * config/i386/i386.md (rex64namesuffix): New mode attribute.
13035 * config/i386/sse.md (sse_cvtsi2ss<rex64namesuffix><round_name>):
13036 Merge insn pattern from sse_cvtsi2ss<round_name> and
13037 sse_cvtsi2ssq<round_name> using SWI48 mode iterator.
13038 (sse_cvtss2si<rex64namesuffix><round_name>): Merge insn pattern
13039 from sse_cvtss2si<round_name> and sse_cvtss2siq<round_name>
13040 using SWI48 mode iterator.
13041 (sse_cvtss2si<rex64namesuffix>_2): Merge insn pattern from
13042 sse_cvtss2si_2 and sse_cvtss2siq_2 using SWI48 mode iterator.
13043 (sse_cvttss2si<rex64namesuffix><round_saeonly_name>): Merge insn
13044 pattern from sse_cvttss2si<round_saeonly_name>
13045 and sse_cvttss2siq<round_saeonly_name> using SWI48 mode iterator.
13046 (avx512f_vcvtss2usi<rex64namesuffix><round_name>): Merge insn pattern
13047 from avx512f_vcvtss2usi<round_name> and avx512f_vcvtss2usiq<round_name>
13048 using SWI48 mode iterator.
13049 (avx512f_vcvttss2usi<rex64namesuffix><round_saeonly_name>): Merge
13050 insn pattern from avx512f_vcvttss2usi<round_saeonly_name> and
13051 avx512f_vcvttss2usiq<round_saeonly_name> using SWI48 mode iterator.
13052 (avx512f_vcvtsd2usi<rex64namesuffix><round_name>): Merge insn pattern
13053 from avx512f_vcvtsd2usi<round_name> and avx512f_vcvtsd2usiq<round_name>
13054 using SWI48 mode iterator.
13055 (avx512f_vcvttsd2usi<rex64namesuffix><round_saeonly_name>): Merge
13056 insn pattern from avx512f_vcvttsd2usi<round_saeonly_name> and
13057 avx512f_vcvttsd2usiq<round_saeonly_name> using SWI48 mode iterator.
13058 (sse2_cvtsd2si<rex64namesuffix><round_name>): Merge insn pattern from
13059 sse2_cvtsd2si<round_name> and sse2_cvtsd2siq<round_name> using
13060 SWI48 mode iterator.
13061 (sse2_cvtsd2si<rex64namesuffix>_2): Merge insn pattern from
13062 sse2_cvtsd2si_2 and sse2_cvtsd2siq_2 using SWI48 mode iterator.
13063 (sse_cvttsd2si<rex64namesuffix><round_saeonly_name>): Merge insn
13064 pattern from sse_cvttsd2si<round_saeonly_name>
13065 and sse_cvttsd2siq<round_saeonly_name> using SWI48 mode iterator.
13067 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
13069 * config/nds32/nds32-md-auxiliary.c
13070 (nds32_valid_smw_lwm_base_p): Refine.
13071 (nds32_output_smw_single_word): Refine.
13072 (nds32_output_smw_double_word): New.
13073 * config/nds32/nds32-protos.h (nds32_output_smw_double_word): New.
13075 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
13077 * config/nds32/nds32-md-auxiliary.c (nds32_output_stack_push): Refine.
13078 (nds32_output_stack_pop): Refine.
13079 (nds32_expand_unaligned_load): Refine.
13080 (nds32_expand_unaligned_store): Refine.
13082 2018-05-19 Kuan-Lin Chen <kuanlinchentw@gmail.com>
13083 Chung-Ju Wu <jasonwucj@gmail.com>
13085 * config/nds32/constants.md: Add TP_REGNUM constant.
13086 (unspec_element): Add UNSPEC_GOTINIT, UNSPEC_GOT, UNSPEC_GOTOFF,
13087 UNSPEC_PLT, UNSPEC_TLSGD, UNSPEC_TLSLD, UNSPEC_TLSIE, UNSPEC_TLSLE and
13089 * config/nds32/nds32-doubleword.md: Consider flag_pic.
13090 * config/nds32/nds32-dspext.md (mov<mode>): Expand TLS and PIC cases.
13091 * config/nds32/nds32-predicates.c (nds32_const_unspec_p): New.
13092 * config/nds32/nds32-md-auxiliary.c: Implementation that support TLS
13093 and PIC code generation.
13094 * config/nds32/nds32-protos.h: Declarations that support TLS and PIC
13096 * config/nds32/nds32-relax-opt.c: Consider TLS and PIC for relax
13098 * config/nds32/nds32.md: Support TLS and PIC.
13099 * config/nds32/nds32.c: Support TLS and PIC.
13100 * config/nds32/nds32.h (nds32_relax_insn_type): New enum type.
13101 * config/nds32/predicates.md (nds32_nonunspec_symbolic_operand): New
13104 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
13106 * config/nds32/nds32-predicates.c (const_vector_to_hwint): Use machine
13107 mode with E_ prefix.
13109 2018-05-19 Kuan-Lin Chen <kuanlinchentw@gmail.com>
13110 Chung-Ju Wu <jasonwucj@gmail.com>
13112 * config/nds32/constants.md (unspec_element): Add UNSPEC_ICT.
13113 * config/nds32/nds32-md-auxiliary.c
13114 (symbolic_reference_mentioned_p): New.
13115 (nds32_legitimize_ict_address): New.
13116 (nds32_expand_ict_move): New.
13117 (nds32_indirect_call_referenced_p): New.
13118 (nds32_symbol_binds_local_p): Delete.
13119 (nds32_long_call_p): Modify.
13120 * config/nds32/nds32-opts.h (nds32_ict_model_type): New enum type.
13121 * config/nds32/nds32-protos.h
13122 (symbolic_reference_mentioned_p): Declare.
13123 (nds32_legitimize_ict_address): Declare.
13124 (nds32_expand_ict_move): Declare.
13125 (nds32_indirect_call_referenced_p): Declare.
13126 * config/nds32/nds32-relax-opt.c (nds32_ict_const_p): New.
13127 (nds32_relax_group): Use nds32_ict_const_p as condition.
13128 * config/nds32/nds32.c (nds32_attribute_table): Add "indirect_call".
13129 (nds32_asm_file_start): Output ict_model directive in asm code.
13130 (nds32_legitimate_address_p): Consider indirect call.
13131 (nds32_print_operand): Consider indirect call.
13132 (nds32_print_operand_address): Consider indirect call.
13133 (nds32_insert_attributes): Handle "indirect_call" attribute.
13134 (TARGET_LEGITIMATE_ADDRESS_P): Define.
13135 (TARGET_LEGITIMATE_CONSTANT_P): Define.
13136 (TARGET_CANNOT_FORCE_CONST_MEM): Define.
13137 (TARGET_DELEGITIMIZE_ADDRESS): Define.
13138 (TARGET_ASM_OUTPUT_ADDR_CONST_EXTRA): Define.
13139 * config/nds32/nds32.h (SYMBOLIC_CONST_P): Define.
13140 (TARGET_ICT_MODEL_SMALL): Define.
13141 (TARGET_ICT_MODEL_LARGE): Define.
13142 * config/nds32/nds32.md (movsi): Consider ict model.
13143 (call, call_value): Consider ict model.
13144 (sibcall, sibcall_value): Consider ict model.
13145 * config/nds32/nds32.opt (mict-model): New option.
13146 * config/nds32/predicates.md (nds32_symbolic_operand): Consider ict
13149 2018-05-18 Kito Cheng <kito.cheng@gmail.com>
13150 Monk Chiang <sh.chiang04@gmail.com>
13151 Jim Wilson <jimw@sifive.com>
13153 * common/config/riscv/riscv-common.c (riscv_parse_arch_string):
13154 Add support to parse rv32e*. Clear MASK_RVE for rv32i and rv64i.
13155 * config.gcc (riscv*-*-*): Add support for rv32e* and ilp32e.
13156 * config/riscv/riscv-c.c (riscv_cpu_cpp_builtins): Define
13157 __riscv_32e when TARGET_RVE. Handle ABI_ILP32E as soft-float ABI.
13158 * config/riscv/riscv-opts.h (riscv_abi_type): Add ABI_ILP32E.
13159 * config/riscv/riscv.c (riscv_compute_frame_info): When TARGET_RVE,
13160 compute save_libcall_adjustment properly.
13161 (riscv_option_override): Call error if TARGET_RVE and not ABI_ILP32E.
13162 (riscv_conditional_register_usage): Handle TARGET_RVE and ABI_ILP32E.
13163 * config/riscv/riscv.h (UNITS_PER_FP_ARG): Handle ABI_ILP32E.
13164 (STACK_BOUNDARY, ABI_STACK_BOUNDARY): Handle TARGET_RVE.
13165 (GP_REG_LAST, MAX_ARGS_IN_REGISTERS): Likewise.
13166 (ABI_SPEC): Handle mabi=ilp32e.
13167 * config/riscv/riscv.opt (abi_type): Add ABI_ILP32E.
13168 (RVE): Add RVE mask.
13169 * doc/invoke.texi (RISC-V options) <-mabi>: Add ilp32e info.
13170 <-march>: Add rv32e as an example.
13172 2018-05-18 Marc Glisse <marc.glisse@inria.fr>
13175 * tree-ssa-structalias.c (create_variable_info_for_1): Extra argument.
13176 (intra_create_variable_infos): Handle C++ constructors.
13178 2018-05-18 Martin Liska <mliska@suse.cz>
13180 * passes.def: Remove a redundant pass.
13182 2018-05-18 Eric Botcazou <ebotcazou@adacore.com>
13185 * config/sparc/sparc.c (sparc_expand_builtin): Always initialize op[0].
13187 2018-05-18 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13189 * config/arm/arm-cpus.in (armv3m, mode32): Delete features.
13191 (ARMv2, ARMv3, ARMv3m): Delete fgroups.
13193 (armv2, armv2a, armv3, armv3m): Delete architectures.
13194 (arm2, arm250, arm3, arm6, arm60, arm600, arm610, arm620,
13195 arm7, arm7d, arm7di, arm70, arm700, arm700i, arm710, arm720,
13196 arm710c, arm7100, arm7500, arm7500fe, arm7m, arm7dm, arm7dmi):
13198 * config/arm/arm.md (maddsidi4): Remove check for arm_arch3m.
13199 (*mulsidi3adddi): Likewise.
13200 (mulsidi3): Likewise.
13201 (*mulsidi3_nov6): Likewise.
13202 (umulsidi3): Likewise.
13203 (umulsidi3_nov6): Likewise.
13204 (umaddsidi4): Likewise.
13205 (*umulsidi3adddi): Likewise.
13206 (smulsi3_highpart): Likewise.
13207 (*smulsi3_highpart_nov6): Likewise.
13208 (umulsi3_highpart): Likewise.
13209 (*umulsi3_highpart_nov6): Likewise.
13210 * config/arm/arm.h (arm_arch3m): Delete.
13211 * config/arm/arm.c (arm_arch3m): Delete.
13212 (arm_option_override_internal): Update armv3-related comment.
13213 (arm_configure_build_target): Delete use of isa_bit_mode32.
13214 (arm_option_reconfigure_globals): Delete set of arm_ach3m.
13215 (arm_rtx_costs_internal): Delete check of arm_arch3m.
13216 * config/arm/arm-fixed.md (mulsq3): Delete check for arm_arch3m.
13217 (mulsa3): Likewise.
13218 (mulusa3): Likewise.
13219 * config/arm/arm-protos.h (arm_arch3m): Delete.
13220 * config/arm/arm-tables.opt: Regenerate.
13221 * config/arm/arm-tune.md: Likewise.
13222 * config/arm/t-arm-elf (all_early_nofp): Delete mentions of
13223 deleted architectures.
13225 2018-05-18 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13227 * config/arm/arm-cpus.in (armv5, armv5e): Delete features.
13228 (armv5t, armv5te): New features.
13229 (ARMv5, ARMv5e): Delete fgroups.
13230 (ARMv5t, ARMv5te): Adjust for above changes.
13231 (ARMv6m): Likewise.
13232 (armv5, armv5e): Delete arches.
13233 * config/arm/arm.md (*call_reg_armv5): Use arm_arch5t instead of
13235 (*call_reg_arm): Likewise.
13236 (*call_value_reg_armv5): Likewise.
13237 (*call_value_reg_arm): Likewise.
13238 (*call_symbol): Likewise.
13239 (*call_value_symbol): Likewise.
13240 (*sibcall_insn): Likewise.
13241 (*sibcall_value_insn): Likewise.
13242 (clzsi2): Likewise.
13243 (prefetch): Likewise.
13244 (define_split and define_peephole2 dependent on arm_arch5):
13246 * config/arm/arm.h (TARGET_LDRD): Use arm_arch5te instead of
13248 (TARGET_ARM_QBIT): Likewise.
13249 (TARGET_DSP_MULTIPLY): Likewise.
13250 (enum base_architecture): Delete BASE_ARCH_5, BASE_ARCH_5E.
13251 (arm_arch5, arm_arch5e): Delete.
13252 (arm_arch5t, arm_arch5te): Declare.
13253 * config/arm/arm.c (arm_arch5, arm_arch5e): Delete.
13254 (arm_arch5t): Declare.
13255 (arm_option_reconfigure_globals): Update for the above.
13256 (arm_options_perform_arch_sanity_checks): Update comment, replace
13257 use of arm_arch5 with arm_arch5t.
13258 (use_return_insn): Likewise.
13259 (arm_emit_call_insn): Likewise.
13260 (output_return_instruction): Likewise.
13261 (arm_final_prescan_insn): Likewise.
13262 (arm_coproc_builtin_available): Likewise.
13263 * config/arm/arm-c.c (arm_cpu_builtins): Replace arm_arch5 and
13264 arm_arch5e with arm_arch5t and arm_arch5te.
13265 * config/arm/arm-protos.h (arm_arch5, arm_arch5e): Delete.
13266 (arm_arch5t, arm_arch5te): Declare.
13267 * config/arm/arm-tables.opt: Regenerate.
13268 * config/arm/t-arm-elf: Remove references to armv5, armv5e.
13269 * config/arm/t-multilib: Likewise.
13270 * config/arm/thumb1.md (*call_reg_thumb1_v5): Check arm_arch5t
13271 instead of arm_arch5.
13272 (*call_reg_thumb1): Likewise.
13273 (*call_value_reg_thumb1_v5): Likewise.
13274 (*call_value_reg_thumb1): Likewise.
13275 * config/arm/vxworks.h (TARGET_OS_CPP_BUILTINS): Remove now
13277 * doc/invoke.texi (ARM Options): Remove references to armv5, armv5e.
13279 2018-05-18 Martin Liska <mliska@suse.cz>
13281 PR gcov-profile/84846
13282 * doc/gcov.texi: Document -t option of gcov tool.
13284 2018-05-18 Martin Liska <mliska@suse.cz>
13286 PR gcov-profile/84846
13287 * gcov.c (print_usage): Add new -t option.
13288 (process_args): Handle the option.
13289 (generate_results): Use stdout as output when requested by
13292 2018-05-18 Martin Liska <mliska@suse.cz>
13294 PR gcov-profile/84846
13295 * coverage.c (coverage_init): Write PWD to .gcno file.
13296 * doc/gcov.texi: Document how working directory is printed.
13297 * gcov-dump.c (dump_gcov_file): Print PWD.
13298 * gcov.c (output_intermediate_file): Likewise.
13299 (read_graph_file): Read PWD string.
13300 (output_lines): Print PWD.
13302 2018-05-18 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
13304 PR middle-end/85817
13305 * ipa-pure-const.c (malloc_candidate_p): Remove the check integer_zerop
13306 for retval and return false if all args to phi are zero.
13308 2018-05-18 Richard Biener <rguenther@suse.de>
13310 * gimple-ssa-evrp.c (class evrp_folder): Add simplify_stmt_using_ranges
13312 (evrp_dom_walker::before_dom_children): Call it.
13314 2018-05-18 Richard Biener <rguenther@suse.de>
13316 * tree-dfa.c (get_ref_base_and_extent): Use range-info to refine
13317 results when processing array refs with variable index.
13319 2018-05-18 Toon Moene <toon@moene.org>
13321 * doc/invoke.texi: Move -floop-unroll-and-jam documentation
13322 directly after that of -floop-interchange. Indicate that both
13323 options are enabled by default when specifying -O3.
13325 2018-05-18 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13327 * config/aarch64/aarch64-simd.md (vec_set<mode>): Use VALL_F16 mode
13328 iterator. Delete separate integer-mode vec_set<mode> expander.
13329 (aarch64_simd_vec_setv2di): Delete.
13330 (vec_setv2di): Delete.
13331 (aarch64_simd_vec_set<mode>): Delete all other patterns with that name.
13332 Use VALL_F16 mode iterator. Add LD1 alternative and use vwcore for
13333 the "w, r" alternative.
13335 2018-05-18 Martin Liska <mliska@suse.cz>
13337 * passes.def: Add pass_lower_switch and pass_lower_switch_O0.
13338 * tree-pass.h (make_pass_lower_switch_O0): New function.
13339 * tree-switch-conversion.c (node_has_low_bound): Remove.
13340 (node_has_high_bound): Likewise.
13341 (node_is_bounded): Likewise.
13342 (class pass_lower_switch): Make it a template type and create
13344 (pass_lower_switch::execute): Add template argument.
13345 (make_pass_lower_switch): New function.
13346 (make_pass_lower_switch_O0): New function.
13347 (do_jump_if_equal): Remove.
13348 (emit_case_nodes): Simplify to just handle all 3 cases and leave
13349 all the hard work to tree optimization passes.
13351 2018-05-18 Martin Liska <mliska@suse.cz>
13353 * dbgcnt.c (limit_low): Renamed from limit.
13354 (limit_high): New variable.
13355 (dbg_cnt_is_enabled): Check for upper limit.
13356 (dbg_cnt): Adjust dumping.
13357 (dbg_cnt_set_limit_by_index): Add new argument for high
13359 (dbg_cnt_set_limit_by_name): Likewise.
13360 (dbg_cnt_process_single_pair): Parse new format.
13361 (dbg_cnt_process_opt): Use strtok.
13362 (dbg_cnt_list_all_counters): Remove 'value' and add
13364 * doc/invoke.texi: Document changes.
13366 2018-05-18 Richard Sandiford <richard.sandiford@linaro.org>
13368 * doc/sourcebuild.texi (scalar_all_fma): Document.
13369 * tree.def (FMA_EXPR): Delete.
13370 * internal-fn.def (FMA, FMS, FNMA, FNMS): New internal functions.
13371 * internal-fn.c (ternary_direct): New macro.
13372 (expand_ternary_optab_fn): Likewise.
13373 (direct_ternary_optab_supported_p): Likewise.
13374 * Makefile.in (build/genmatch.o): Depend on case-fn-macros.h.
13375 * builtins.c (fold_builtin_fma): Delete.
13376 (fold_builtin_3): Don't call it.
13377 * cfgexpand.c (expand_debug_expr): Remove FMA_EXPR handling.
13378 * expr.c (expand_expr_real_2): Likewise.
13379 * fold-const.c (operand_equal_p): Likewise.
13380 (fold_ternary_loc): Likewise.
13381 * gimple-pretty-print.c (dump_ternary_rhs): Likewise.
13382 * gimple.c (DEFTREECODE): Likewise.
13383 * gimplify.c (gimplify_expr): Likewise.
13384 * optabs-tree.c (optab_for_tree_code): Likewise.
13385 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
13386 * tree-eh.c (operation_could_trap_p): Likewise.
13387 (stmt_could_throw_1_p): Likewise.
13388 * tree-inline.c (estimate_operator_cost): Likewise.
13389 * tree-pretty-print.c (dump_generic_node): Likewise.
13390 (op_code_prio): Likewise.
13391 * tree-ssa-loop-im.c (stmt_cost): Likewise.
13392 * tree-ssa-operands.c (get_expr_operands): Likewise.
13393 * tree.c (commutative_ternary_tree_code, add_expr): Likewise.
13394 * fold-const-call.h (fold_fma): Delete.
13395 * fold-const-call.c (fold_const_call_ssss): Handle CFN_FMS,
13396 CFN_FNMA and CFN_FNMS.
13397 (fold_fma): Delete.
13398 * genmatch.c (combined_fn): New enum.
13399 (commutative_ternary_tree_code): Remove FMA_EXPR handling.
13400 (commutative_op): New function.
13401 (commutate): Use it. Handle more than 2 operands.
13402 (dt_operand::gen_gimple_expr): Use commutative_op.
13403 (parser::parse_expr): Allow :c to be used with non-binary
13404 operators if the commutative operand is known.
13405 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Handle
13406 CFN_FMS, CFN_FNMA and CFN_FNMS.
13407 (backprop::process_assign_use): Remove FMA_EXPR handling.
13408 * hsa-gen.c (gen_hsa_insns_for_operation_assignment): Likewise.
13409 (gen_hsa_fma): New function.
13410 (gen_hsa_insn_for_internal_fn_call): Use it for IFN_FMA, IFN_FMS,
13411 IFN_FNMA and IFN_FNMS.
13412 * match.pd: Add folds for IFN_FMS, IFN_FNMA and IFN_FNMS.
13413 * gimple-fold.h (follow_all_ssa_edges): Declare.
13414 * gimple-fold.c (follow_all_ssa_edges): New function.
13415 * tree-ssa-math-opts.c (convert_mult_to_fma_1): Use the
13416 gimple_build interface and use follow_all_ssa_edges to fold the result.
13417 (convert_mult_to_fma): Use direct_internal_fn_suppoerted_p
13418 instead of checking for optabs directly.
13419 * config/i386/i386.c (ix86_add_stmt_cost): Recognize FMAs as calls
13420 rather than FMA_EXPRs.
13421 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Create a
13422 call to IFN_FMA instead of an FMA_EXPR.
13424 2018-05-17 Jim Wilson <jimw@sifive.com>
13426 * expr.c (do_tablejump): When converting index to Pmode, if we have a
13427 sign extended promoted subreg, and the range does not have the sign bit
13428 set, then do a sign extend.
13430 * config/riscv/riscv.c (riscv_extend_comparands): In unsigned QImode
13431 test, check for sign extended subreg and/or constant operands, and
13432 do a sign extend in that case.
13434 2018-05-17 Steve Ellcey <sellcey@cavium.com>
13436 * config/aarch64/thunderx2t99.md (thunderx2t99_ls_both): Delete.
13437 (thunderx2t99_multiple): Delete psuedo-units from used cpus.
13439 (thunderx2t99_alu_shift): Remove alu_shift_reg, alus_shift_reg.
13440 Change logics_shift_reg to logics_shift_imm.
13441 (thunderx2t99_fp_loadpair_basic): Delete.
13442 (thunderx2t99_fp_storepair_basic): Delete.
13443 (thunderx2t99_asimd_int): Add neon_sub and neon_sub_q types.
13444 (thunderx2t99_asimd_polynomial): Delete.
13445 (thunderx2t99_asimd_fp_simple): Add neon_fp_mul_s_scalar_q
13446 and neon_fp_mul_d_scalar_q.
13447 (thunderx2t99_asimd_fp_conv): Add *int_to_fp* types.
13448 (thunderx2t99_asimd_misc): Delete neon_dup and neon_dup_q.
13449 (thunderx2t99_asimd_recip_step): Add missing *sqrt* types.
13450 (thunderx2t99_asimd_lut): Add missing tbl types.
13451 (thunderx2t99_asimd_ext): Delete.
13452 (thunderx2t99_asimd_load1_1_mult): Delete.
13453 (thunderx2t99_asimd_load1_2_mult): Delete.
13454 (thunderx2t99_asimd_load1_ldp): New.
13455 (thunderx2t99_asimd_load1): New.
13456 (thunderx2t99_asimd_load2): Add missing *load2* types.
13457 (thunderx2t99_asimd_load3): New.
13458 (thunderx2t99_asimd_load4): New.
13459 (thunderx2t99_asimd_store1_1_mult): Delete.
13460 (thunderx2t99_asimd_store1_2_mult): Delete.
13461 (thunderx2t99_asimd_store2_mult): Delete.
13462 (thunderx2t99_asimd_store2_onelane): Delete.
13463 (thunderx2t99_asimd_store_stp): New.
13464 (thunderx2t99_asimd_store1): New.
13465 (thunderx2t99_asimd_store2): New.
13466 (thunderx2t99_asimd_store3): New.
13467 (thunderx2t99_asimd_store4): New.
13469 2018-05-17 Jerome Lambourg <lambourg@adacore.com>
13471 * config/arm/arm_cmse.h (cmse_nsfptr_create, cmse_is_nsfptr): Remove
13472 #include <stdint.h>. Replace intptr_t with __INTPTR_TYPE__.
13474 2018-05-17 Pat Haugen <pthaugen@us.ibm.com>
13475 Segher Boessenkool <segher@kernel.crashing.org>
13478 * config/rs6000/rs6000.c (rs6000_output_move_128bit): Check dest
13481 2018-05-17 Richard Biener <rguenther@suse.de>
13483 * tree-ssa-dse.c (dse_classify_store): Fix iterator increment
13484 for pruning loop and prune defs feeding only already visited PHIs.
13486 2018-05-17 Richard Biener <rguenther@suse.de>
13488 * tree-ssa-sccvn.c (vn_reference_lookup_3): Improve memset handling.
13490 2018-05-17 Bin Cheng <bin.cheng@arm.com>
13491 Richard Biener <rguenther@suse.de>
13493 PR tree-optimization/85793
13494 * tree-vect-stmts.c (vectorizable_load): Handle 1 element-wise load
13495 for VMAT_ELEMENTWISE.
13497 2018-05-17 Richard Sandiford <richard.sandiford@linaro.org>
13499 * internal-fn.h (lookup_internal_fn): Declare
13500 * internal-fn.c (lookup_internal_fn): New function.
13501 * gimple.c (gimple_build_call_from_tree): Handle calls to
13502 internal functions.
13503 * gimple-pretty-print.c (dump_gimple_call): Print "." before
13504 internal function names.
13505 * tree-pretty-print.c (dump_generic_node): Likewise.
13506 * tree-ssa-scopedtables.c (expr_hash_elt::print): Likewise.
13508 2018-05-17 Richard Sandiford <richard.sandiford@linaro.org>
13510 * gimple-fold.h (gimple_build): Make the function forms take
13511 combined_fn rather than built_in_function.
13512 (gimple_simplify): Likewise.
13513 * gimple-match-head.c (gimple_simplify): Likewise.
13514 * gimple-fold.c (gimple_build): Likewise.
13515 * tree-vect-loop.c (get_initial_def_for_reduction): Use gimple_build
13516 rather than gimple_build_call_internal.
13517 (get_initial_defs_for_reduction): Likewise.
13518 (vect_create_epilog_for_reduction): Likewise.
13519 (vectorizable_live_operation): Likewise.
13521 2018-05-17 Martin Liska <mliska@suse.cz>
13523 * gimple-ssa-sprintf.c (format_directive): Do not use
13524 space in between 'G_' and '('.
13526 2018-05-17 Jakub Jelinek <jakub@redhat.com>
13529 * config/i386/i386.c (ix86_fold_builtin): Handle masked shifts
13530 even if the mask is not all ones.
13533 * config/i386/i386.c (ix86_fold_builtin): Fold shift builtins by
13535 (ix86_gimple_fold_builtin): Likewise.
13538 * config/i386/i386.c: Include tree-vector-builder.h.
13539 (ix86_vector_shift_count): New function.
13540 (ix86_fold_builtin): Fold shift builtins by scalar count.
13541 (ix86_gimple_fold_builtin): Likewise.
13543 * config/i386/avx512fintrin.h (_mm512_set_epi16, _mm512_set_epi8,
13544 _mm512_setzero): New intrinsics.
13546 2018-05-17 James Greenhalgh <james.greenhalgh@arm.com>
13547 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13549 * config/aarch64/aarch64.c (aarch64_expand_vector_init): Modify
13550 code generation for cases where splatting a value is not useful.
13551 * simplify-rtx.c (simplify_ternary_operation): Simplify
13552 vec_merge across a vec_duplicate and a paradoxical subreg forming
13553 a vector mode to a vec_concat.
13555 2018-05-17 Olga Makhotina <olga.makhotina@intel.com>
13557 * config.gcc: Support "goldmont-plus".
13558 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
13560 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
13561 PROCESSOR_GOLDMONT_PLUS.
13562 * config/i386/i386.c (m_GOLDMONT_PLUS): Define.
13563 (processor_target_table): Add "goldmont-plus".
13564 (PTA_GOLDMONT_PLUS): Define.
13565 (ix86_lea_outperforms): Add TARGET_GOLDMONT_PLUS.
13566 (get_builtin_code_for_version): Handle PROCESSOR_GOLDMONT_PLUS.
13567 (fold_builtin_cpu): Add M_INTEL_GOLDMONT_PLUS.
13568 (fold_builtin_cpu): Add "goldmont-plus".
13569 (ix86_add_stmt_cost): Add TARGET_GOLDMONT_PLUS.
13570 (ix86_option_override_internal): Add "goldmont-plus".
13571 * config/i386/i386.h (processor_costs): Define TARGET_GOLDMONT_PLUS.
13572 (processor_type): Add PROCESSOR_GOLDMONT_PLUS.
13573 * config/i386/x86-tune.def: Add m_GOLDMONT_PLUS.
13574 * doc/invoke.texi: Add goldmont-plus as x86 -march=/-mtune= CPU type.
13576 2018-05-17 Richard Biener <rguenther@suse.de>
13578 PR tree-optimization/85757
13579 * tree-ssa-dse.c (dse_classify_store): Record a PHI def and
13580 remove defs that only feed that PHI from further processing.
13582 2018-05-16 Jim Wilson <jimw@sifive.com>
13584 * config/riscv/riscv.md (<optab>si3_mask, <optab>si3_mask_1): Prepend
13586 (<optab>di3_mask, <optab>di3_mask_1): Likewise.
13588 2018-05-16 Mark Wielaard <mark@klomp.org>
13590 * dwarf2out.c (count_index_strings): New function.
13591 (output_indirect_strings): Call count_index_strings and generate
13592 header for dwarf_version >= 5.
13594 2018-05-16 Mark Wielaard <mark@klomp.org>
13596 * dwarf2out.c (dwarf_FORM): New function.
13597 (set_indirect_string): Use dwarf_FORM.
13598 (reset_indirect_string): Likewise.
13599 (size_of_die): Likewise.
13600 (value_format): Likewise.
13601 (output_die): Likewise.
13602 (add_skeleton_AT_string): Likewise.
13603 (output_macinfo_op): Likewise.
13604 (index_string): Likewise.
13605 (output_index_string_offset): Likewise.
13606 (output_index_string): Likewise.
13607 (count_index_strings): Likewise.
13609 2018-05-16 Carl Love <cel@us.ibm.com>
13611 * config/rs6000/rs6000.md (prefetch): Generate ISA 2.06 instructions
13612 dcbt and dcbtstt with TH=16 if operands[2] is 0 and Power 8 or newer.
13614 2018-05-16 Martin Jambor <mjambor@suse.cz>
13616 * ipa-prop.c (ipa_free_all_edge_args): Remove.
13617 * ipa-prop.h (ipa_free_all_edge_args): Likewise.
13619 2018-05-16 Wilco Dijkstra <wdijkstr@arm.com>
13621 * config/aarch64/aarch64.md (fma<mode>4): Change into expand pattern.
13622 (fnma<mode>4): Likewise.
13623 (fms<mode>4): Likewise.
13624 (fnms<mode>4): Likewise.
13625 (aarch64_fma<mode>4): Rename insn, reorder accumulator operand.
13626 (aarch64_fnma<mode>4): Likewise.
13627 (aarch64_fms<mode>4): Likewise.
13628 (aarch64_fnms<mode>4): Likewise.
13629 (aarch64_fnmadd<mode>4): Likewise.
13631 2018-05-16 Jason Merrill <jason@redhat.com>
13633 * tree.c (warn_deprecated_use): Return bool. Simplify logic.
13635 2018-05-16 Richard Biener <rguenther@suse.de>
13637 * tree-vectorizer.h (struct stmt_info_for_cost): Add where member.
13638 (dump_stmt_cost): Declare.
13639 (add_stmt_cost): Dump cost we add.
13640 (add_stmt_costs): New function.
13641 (vect_model_simple_cost, vect_model_store_cost, vect_model_load_cost):
13642 No longer exported.
13643 (vect_analyze_stmt): Adjust prototype.
13644 (vectorizable_condition): Likewise.
13645 (vectorizable_live_operation): Likewise.
13646 (vectorizable_reduction): Likewise.
13647 (vectorizable_induction): Likewise.
13648 * tree-vect-loop.c (vect_analyze_loop_operations): Create local
13649 cost vector to pass to vectorizable_ and record afterwards.
13650 (vect_model_reduction_cost): Take cost vector argument and adjust.
13651 (vect_model_induction_cost): Likewise.
13652 (vectorizable_reduction): Likewise.
13653 (vectorizable_induction): Likewise.
13654 (vectorizable_live_operation): Likewise.
13655 * tree-vect-slp.c (vect_create_new_slp_node): Initialize
13656 SLP_TREE_NUMBER_OF_VEC_STMTS.
13657 (vect_analyze_slp_cost_1): Remove.
13658 (vect_analyze_slp_cost): Likewise.
13659 (vect_slp_analyze_node_operations): Take visited args and
13660 a target cost vector. Avoid processing already visited stmt sets.
13661 (vect_slp_analyze_operations): Use a local cost vector to gather
13662 costs and register those of non-discarded instances.
13663 (vect_bb_vectorization_profitable_p): Use add_stmt_costs.
13664 (vect_schedule_slp_instance): Remove copying of
13665 SLP_TREE_NUMBER_OF_VEC_STMTS. Instead assert that it is not
13667 * tree-vect-stmts.c (record_stmt_cost): Remove path directly
13668 adding cost. Record cost entry location.
13669 (vect_prologue_cost_for_slp_op): Function to compute cost of
13670 a constant or invariant generated for SLP vect in the prologue,
13671 split out from vect_analyze_slp_cost_1.
13672 (vect_model_simple_cost): Make static. Adjust for SLP costing.
13673 (vect_model_promotion_demotion_cost): Likewise.
13674 (vect_model_store_cost): Likewise, make static.
13675 (vect_model_load_cost): Likewise.
13676 (vectorizable_bswap): Add cost vector arg and adjust.
13677 (vectorizable_call): Likewise.
13678 (vectorizable_simd_clone_call): Likewise.
13679 (vectorizable_conversion): Likewise.
13680 (vectorizable_assignment): Likewise.
13681 (vectorizable_shift): Likewise.
13682 (vectorizable_operation): Likewise.
13683 (vectorizable_store): Likewise.
13684 (vectorizable_load): Likewise.
13685 (vectorizable_condition): Likewise.
13686 (vectorizable_comparison): Likewise.
13687 (can_vectorize_live_stmts): Likewise.
13688 (vect_analyze_stmt): Likewise.
13689 (vect_transform_stmt): Adjust calls to vectorizable_*.
13690 * tree-vectorizer.c: Include gimple-pretty-print.h.
13691 (dump_stmt_cost): New function.
13693 2018-05-16 Richard Biener <rguenther@suse.de>
13695 * params.def (PARAM_DSE_MAX_ALIAS_QUERIES_PER_STORE): New param.
13696 * doc/invoke.texi (dse-max-alias-queries-per-store): Document.
13697 * tree-ssa-dse.c: Include tree-ssa-loop.h.
13698 (check_name): New callback.
13699 (dse_classify_store): Track cycles via a visited bitmap of PHI
13700 defs and simplify handling of in-loop and across loop dead stores
13701 and properly fail for loop-variant refs. Handle byte-tracking with
13702 multiple defs. Use PARAM_DSE_MAX_ALIAS_QUERIES_PER_STORE for
13705 2018-05-16 Richard Sandiford <richard.sandiford@linaro.org>
13707 * tree-vectorizer.h (vect_get_vector_types_for_stmt): Declare.
13708 (vect_get_mask_type_for_stmt): Likewise.
13709 * tree-vect-slp.c (vect_two_operations_perm_ok_p): New function,
13711 (vect_build_slp_tree_1): ...here. Use vect_get_vector_types_for_stmt
13712 to determine the statement's vector type and the vector type that
13713 should be used for calculating nunits. Deal with cases in which
13714 the type has to be deferred.
13715 (vect_slp_analyze_node_operations): Use vect_get_vector_types_for_stmt
13716 and vect_get_mask_type_for_stmt to calculate STMT_VINFO_VECTYPE.
13717 * tree-vect-loop.c (vect_determine_vf_for_stmt_1)
13718 (vect_determine_vf_for_stmt): New functions, split out from...
13719 (vect_determine_vectorization_factor): ...here.
13720 * tree-vect-stmts.c (vect_get_vector_types_for_stmt)
13721 (vect_get_mask_type_for_stmt): New functions, split out from
13722 vect_determine_vectorization_factor.
13724 2018-05-16 Richard Biener <rguenther@suse.de>
13726 * tree-cfg.c (verify_gimple_assign_ternary): Properly
13727 verify the [VEC_]COND_EXPR embedded comparison.
13729 2018-05-15 Martin Sebor <msebor@redhat.com>
13731 PR tree-optimization/85753
13732 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Handle
13733 RECORD_TYPE in addition to ARRAY_TYPE.
13735 2018-05-15 Martin Sebor <msebor@redhat.com>
13737 PR middle-end/85643
13738 * calls.c (get_attr_nonstring_decl): Handle MEM_REF.
13740 2018-05-15 Richard Biener <rguenther@suse.de>
13742 * tree-ssa-dse.c (dse_classify_store): Remove use_stmt parameter,
13743 add by_clobber_p one. Change algorithm to collect all defs
13744 representing uses we need to walk and try reducing them to
13745 a single one before failing.
13746 (dse_dom_walker::dse_optimize_stmt): Adjust.
13748 2018-05-13 Mark Wielaard <mark@klomp.org>
13750 * dwarf2out.c (dwarf_OP): Handle DW_OP_addrx and DW_OP_constx.
13751 (size_of_loc_descr): Likewise.
13752 (output_loc_operands): Likewise.
13753 (output_loc_operands_raw): Likewise.
13754 (dw_addr_op): Use dwarf_OP () for DW_OP_constx and DW_OP_addrx.
13755 (resolve_addr_in_expr): Handle DW_OP_addrx and DW_OP_constx.
13756 (hash_loc_operands): Likewise.
13757 (compare_loc_operands): Likewise.
13759 2018-05-14 Mark Wielaard <mark@klomp.org>
13761 * dwarf2out.c (count_index_addrs): New function.
13762 (dwarf2out_finish): Use count_index_addrs to calculate addrs_length.
13764 2018-05-15 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
13766 PR tree-optimization/83648
13767 * ipa-pure-const.c (malloc_candidate_p): Allow function with NULL
13768 return value as malloc candidate.
13770 2018-05-15 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
13773 * ipa-pure-const.c (warn_function_malloc): Pass value of known_finite
13774 param as true in call to suggest_attribute.
13776 2018-05-14 Segher Boessenkool <segher@kernel.crashing.org>
13778 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete @itemx for
13779 -mreadonly-in-sdata.
13781 2018-05-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13783 * config/aarch64/aarch64-simd.md (*aarch64_crypto_aese_fused):
13785 (aarch64_crypto_aesd_fused): Likewise.
13787 2018-05-14 Wilco Dijkstra <wdijkstr@arm.com>
13789 * config/aarch64/aarch64.md (mov<mode>): Remove '*' in alternatives.
13790 (movsi_aarch64): Likewise.
13791 (load_pairsi): Likewise.
13792 (load_pairdi): Likewise.
13793 (store_pairsi): Likewise.
13794 (store_pairdi): Likewise.
13795 (load_pairsf): Likewise.
13796 (load_pairdf): Likewise.
13797 (store_pairsf): Likewise.
13798 (store_pairdf): Likewise.
13799 (zero_extend): Likewise.
13800 (trunc): Swap alternatives.
13801 (fcvt_target): Add '?' to prefer w over r.
13803 2018-05-14 Jakub Jelinek <jakub@redhat.com>
13806 * config/i386/i386.md: Disallow non-commutative arithmetics in
13807 last twpeephole for mem {+,-,&,|,^}= x; mem != 0 after cmpelim
13808 optimization. Use COMMUTATIVE_ARITH_P test rather than != MINUS
13809 in the peephole2 before it.
13811 2018-05-14 Sebastian Peryt <sebastian.peryt@intel.com>
13813 * common/config/i386/i386-common.c (OPTION_MASK_ISA_CLDEMOTE_SET,
13814 OPTION_MASK_ISA_CLDEMOTE_UNSET): New defines.
13815 (ix86_handle_option): Handle -mcldemote.
13816 * config.gcc: New header.
13817 * config/i386/cldemoteintrin.h: New file.
13818 * config/i386/cpuid.h (bit_CLDEMOTE): New bit.
13819 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
13821 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
13822 OPTION_MASK_ISA_CLDEMOTE.
13823 * config/i386/i386.c (ix86_target_string): Add -mcldemote.
13824 (ix86_valid_target_attribute_inner_p): Ditto.
13825 (enum ix86_builtins): Add IX86_BUILTIN_CLDEMOTE.
13826 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_cldemote.
13827 (ix86_expand_builtin): Expand IX86_BUILTIN_CLDEMOTE.
13828 * config/i386/i386.h (TARGET_CLDEMOTE, TARGET_CLDEMOTE_P): New.
13829 * config/i386/i386.md (UNSPECV_CLDEMOTE): New.
13831 * config/i386/i386.opt: Add -mcldemote.
13832 * config/i386/x86intrin.h: New header.
13833 * doc/invoke.texi: Add -mcldemote.
13835 2018-05-14 Richard Biener <rguenther@suse.de>
13837 * doc/match-and-simplify.texi: Adjust :s documentation.
13839 2018-05-14 Alexander Monakov <amonakov@ispras.ru>
13841 * sort.cc (REORDER_23): Pass the type for the temporaries instead of
13842 intended memcpy size.
13843 (REORDER_45): Likewise.
13845 2018-05-13 Alexander Monakov <amonakov@ispras.ru>
13847 * sort.cc: New file.
13848 * system.h [!CHECKING_P] (qsort): Redirect to gcc_qsort.
13849 * vec.c (qsort_chk): Use gcc_qsort.
13850 * Makefile.in (OBJS-libcommon): Add sort.o.
13851 (build/sort.o): New target. Use it...
13852 (BUILD_RTL): ... here, and...
13853 (build/gencfn-macros): ... here, and...
13854 (build/genmatch): ... here.
13856 2018-05-13 Kito Cheng <kito.cheng@gmail.com>
13857 Chung-Ju Wu <jasonwucj@gmail.com>
13859 * config.gcc (nds32*-*-*): Check that n15 is valid to --with-cpu.
13860 * config/nds32/nds32-graywolf.md: New file.
13861 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_GRAYWOLF.
13862 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n15
13864 * config/nds32/nds32-protos.h: More declarations for n15 pipeline.
13865 * config/nds32/nds32-utils.c: More implementations for n15 pipeline.
13866 * config/nds32/nds32.md (pipeline_model): Add graywolf.
13867 * config/nds32/nds32.opt (mcpu): Support n15 pipeline cpus.
13868 * config/nds32/pipelines.md: Include n15 settings.
13870 2018-05-13 Kito Cheng <kito.cheng@gmail.com>
13871 Chung-Ju Wu <jasonwucj@gmail.com>
13873 * config.gcc (nds32*-*-*): Check that n12/n13 are valid to --with-cpu.
13874 * config/nds32/nds32-n13.md: New file.
13875 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N12 and CPU_N13.
13876 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n12/n13
13878 * config/nds32/nds32-protos.h: More declarations for n12/n13 pipeline.
13879 * config/nds32/nds32.md (pipeline_model): Add n13.
13880 * config/nds32/nds32.opt (mcpu): Support n12/n13 pipeline cpus.
13881 * config/nds32/pipelines.md: Include n13 settings.
13883 2018-05-13 Kito Cheng <kito.cheng@gmail.com>
13884 Chung-Ju Wu <jasonwucj@gmail.com>
13886 * config.gcc (nds32*-*-*): Check that n10/d10 are valid to --with-cpu.
13887 * config/nds32/nds32-n10.md: New file.
13888 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N10.
13889 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n10
13891 * config/nds32/nds32-protos.h: More declarations for n10 pipeline.
13892 * config/nds32/nds32-utils.c: More implementations for n10 pipeline.
13893 * config/nds32/nds32.md (pipeline_model): Add n10.
13894 * config/nds32/nds32.opt (mcpu): Support n10 pipeline cpus.
13895 * config/nds32/pipelines.md: Include n10 settings.
13897 2018-05-13 Monk Chiang <sh.chiang04@gmail.com>
13898 Kito Cheng <kito.cheng@gmail.com>
13899 Chung-Ju Wu <jasonwucj@gmail.com>
13901 * config.gcc (nds32be-*-*): Handle --with-ext-dsp.
13902 * config/nds32/constants.md (unspec_element, unspec_volatile_element):
13903 Add enum values for DSP extension instructions.
13904 * config/nds32/constraints.md (Iu06, IU06, CVp5, CVs5, CVs2, CVhi):
13906 * config/nds32/iterators.md (shifts, shiftrt, sat_plus, all_plus,
13907 sat_minus, all_minus, plus_minus, extend, sumax, sumin, sumin_max):
13908 New code iterators.
13909 (su, zs, uk, opcode, add_rsub, add_sub): New code attributes.
13910 * config/nds32/nds32-dspext.md: New file for DSP implementation.
13911 * config/nds32/nds32-intrinsic.c: Implementation of DSP extension.
13912 * config/nds32/nds32-intrinsic.md: Likewise.
13913 * config/nds32/nds32_intrinsic.h: Likewise.
13914 * config/nds32/nds32-md-auxiliary.c: Likewise.
13915 * config/nds32/nds32-memory-manipulation.c: Consider DSP extension.
13916 * config/nds32/nds32-predicates.c (const_vector_to_hwint): New.
13917 (nds32_valid_CVp5_p, nds32_valid_CVs5_p): New.
13918 (nds32_valid_CVs2_p, nds32_valid_CVhi_p): New.
13919 * config/nds32/nds32-protos.h: New declarations for DSP extension.
13920 * config/nds32/nds32-utils.c (extract_mac_non_acc_rtx): New case
13921 TYPE_DMAC in switch statement.
13922 * config/nds32/nds32.c: New checking and implementation for DSP
13923 extension instructions.
13924 * config/nds32/nds32.h: Likewise.
13925 * config/nds32/nds32.md: Likewise.
13926 * config/nds32/nds32.opt (mhw-abs, mext-dsp): New options.
13927 * config/nds32/predicates.md: Implement new predicates for DSP
13930 2018-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
13932 * config/rs6000/rs6000.md (mov<mode>_softfloat, FMOVE32):
13933 Reformat alternatives and attributes so it is easier to identify
13934 which constraints/attributes go with which instruction.
13935 (mov<mode>_hardfloat32, FMOVE64): Likewise.
13936 (mov<mode>_softfloat32, FMOVE64): Likewise.
13937 (mov<mode>_hardfloat64, FMOVE64): Likewise.
13938 (mov<mode>_softfloat64, FMOVE64): Likewise.
13940 2018-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
13942 * doc/extend.texi (PowerPC Built-in Functions): Rename this
13944 (Basic PowerPC Built-in Functions): The new name of the
13945 subsection previously known as "PowerPC Built-in Functions".
13946 (Basic PowerPC Built-in Functions Available on all Configurations):
13948 (Basic PowerPC Built-in Functions Available on ISA 2.05): Likewise.
13949 (Basic PowerPC Built-in Functions Available on ISA 2.06): Likewise.
13950 (Basic PowerPC Built-in Functions Available on ISA 2.07): Likewise.
13951 (Basic PowerPC Built-in Functions Available on ISA 3.0): Likewise.
13953 2018-05-11 Martin Jambor <mjambor@suse.cz>
13956 * ipa-cp.c (intersect_with_plats): Check that the lattice contains
13959 2018-05-11 Richard Earnshaw <rearnsha@arm.com>
13962 * config/arm/arm-cpus.in (fgroup ARMv6m): Add be8 feature.
13964 2018-05-11 Sebastian Peryt <sebastian.peryt@intel.com>
13966 * common/config/i386/i386-common.c (OPTION_MASK_ISA_WAITPKG_SET,
13967 OPTION_MASK_ISA_WAITPKG_UNSET): New defines.
13968 (ix86_handle_option): Handle -mwaitpkg.
13969 * config.gcc: New header.
13970 * config/i386/cpuid.h (bit_WAITPKG): New bit.
13971 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mwaitpkg.
13972 * config/i386/i386-builtin-types.def ((UINT8, UNSIGNED, UINT64)): New
13974 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
13975 OPTION_MASK_ISA_WAITPKG.
13976 * config/i386/i386.c (ix86_target_string): Add -mwaitpkg.
13977 (ix86_option_override_internal): Add PTA_WAITPKG.
13978 (ix86_valid_target_attribute_inner_p): Add -mwaitpkg.
13979 (enum ix86_builtins): Add IX86_BUILTIN_UMONITOR, IX86_BUILTIN_UMWAIT,
13980 IX86_BUILTIN_TPAUSE.
13981 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_umonitor,
13982 __builtin_ia32_umwait and __builtin_ia32_tpause.
13983 (ix86_expand_builtin): Expand IX86_BUILTIN_UMONITOR,
13984 IX86_BUILTIN_UMWAIT, IX86_BUILTIN_TPAUSE.
13985 * config/i386/i386.h (TARGET_WAITPKG, TARGET_WAITPKG_P): New.
13986 * config/i386/i386.md (UNSPECV_UMWAIT, UNSPECV_UMONITOR,
13987 UNSPECV_TPAUSE): New.
13988 (umwait, umwait_rex64, umonitor_<mode>, tpause, tpause_rex64): New.
13989 * config/i386/i386.opt: Add -mwaitpkg.
13990 * config/i386/waitpkgintrin.h: New file.
13991 * config/i386/x86intrin.h: New header.
13992 * doc/invoke.texi: Add -mwaitpkg.
13994 2018-05-11 Richard Earnshaw <rearnsha@arm.com>
13997 * config/arm/arm-cpus.in: Add comment that ARMv6-m and ARMv6S-m are now
13999 (cortex-m0): Use armv6s-m isa.
14000 (cortex-m0plus): Likewise.
14001 (cortex-m1): Likewise.
14002 (cortex-m0.small-multiply): Likewise.
14003 (cortex-m0plus.small-multiply): Likewise.
14004 (cortex-m1.small-multiply): Likewise.
14006 2018-05-11 Allan Sandfeld Jensen <allan.jensen@qt.io>
14007 Jakub Jelinek <jakub@redhat.com>
14009 PR tree-optimization/85692
14010 * tree-ssa-forwprop.c (simplify_vector_constructor): Try two
14011 source permute as well.
14013 2018-05-11 Martin Liska <mliska@suse.cz>
14016 * doc/extend.texi: Document LLVM style format for no_sanitize
14019 2018-05-10 Michael Meissner <meissner@linux.vnet.ibm.com>
14021 * config/rs6000/rs6000.c (mode_supports_dq_form): Rename
14022 mode_supports_vsx_dform_quad to mode_supports_dq_form.
14023 (mode_supports_vsx_dform_quad): Likewise.
14024 (mode_supports_vmx_dform): Move these functions to be next to the
14025 other mode_supports functions.
14026 (mode_supports_dq_form): Likewise.
14027 (quad_address_p): Change calls of mode_supports_vsx_dform_quad to
14028 mode_supports_dq_form.
14029 (reg_offset_addressing_ok_p): Likewise.
14030 (offsettable_ok_by_alignment): Likewise.
14031 (rs6000_legitimate_offset_address_p): Likewise.
14032 (legitimate_lo_sum_address_p): Likewise.
14033 (rs6000_legitimize_address): Likewise.
14034 (rs6000_legitimize_reload_address): Likewise.
14035 (rs6000_secondary_reload_inner): Likewise.
14036 (rs6000_preferred_reload_class): Likewise.
14037 (rs6000_output_move_128bit): Likewise.
14039 2018-05-10 Uros Bizjak <ubizjak@gmail.com>
14041 * config/i386/i386.c (ix86_expand_builtin) <case IX86_BUILTIN_RDPID>:
14042 Generate SImode target register for null target.
14043 <case IX86_BUILTIN_XGETBV>: Ditto.
14044 <case IX86_BUILTIN_XSETBV>: Optimize LSHIFTRT generation.
14045 * config/i386/xsaveintrin.h (_xgetbv): Add missing return.
14047 2018-05-10 Carl Love <cel@us.ibm.com>
14049 * config/rs6000/rs6000.md (prefetch): Generate ISA 2.06 instructions
14050 dcbtt and dcbtstt if operands[2] is 0.
14052 2018-05-10 Uros Bizjak <ubizjak@gmail.com>
14055 * config/i386/sse.md (usadv64qi): New expander.
14057 2018-05-10 Segher Boessenkool <segher@kernel.crashing.org>
14059 * config/rs6000/altivec.md (altivec_vmrghb, altivec_vmrghh,
14060 altivec_vmrghw, altivec_vmrglb, altivec_vmrglh, altivec_vmrglw): Remove
14061 -maltivec=be support.
14062 (vec_widen_umult_even_v16qi, vec_widen_smult_even_v16qi,
14063 vec_widen_umult_even_v8hi, vec_widen_smult_even_v8hi,
14064 vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
14065 vec_widen_umult_odd_v16qi, vec_widen_smult_odd_v16qi,
14066 vec_widen_umult_odd_v8hi, vec_widen_smult_odd_v8hi,
14067 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si, altivec_vpkpx,
14068 altivec_vpks<VI_char>ss, altivec_vpks<VI_char>us,
14069 altivec_vpku<VI_char>us, altivec_vpku<VI_char>um, altivec_vsum2sws,
14070 altivec_vsumsws): Adjust.
14071 (altivec_vspltb *altivec_vspltb_internal, altivec_vsplth,
14072 *altivec_vsplth_internal, altivec_vspltw, *altivec_vspltw_internal,
14073 altivec_vspltsf, *altivec_vspltsf_internal): Remove -maltivec=be
14075 (altivec_vperm_<mode>, altivec_vperm_<mode>_uns,
14076 altivec_vupkhs<VU_char>, altivec_vupkls<VU_char>, altivec_vupkhpx,
14077 altivec_vupklpx, altivec_lvsl, altivec_lvsr): Adjust.
14078 (altivec_lve<VI_char>x): Delete expand.
14079 (*altivec_lve<VI_char>x_internal): Rename to...
14080 (altivec_lve<VI_char>x): ... this.
14081 (altivec_lvxl_<mode>): Delete expand.
14082 (*altivec_lvxl_<mode>_internal): Rename to ...
14083 (altivec_lvxl_<mode>): ... this.
14084 (altivec_stvxl_<mode>): Delete expand.
14085 (*altivec_stvxl_<mode>_internal): Rename to ...
14086 (altivec_stvxl_<mode>): ... this.
14087 (altivec_stve<VI_char>x): Delete expand.
14088 (*altivec_stve<VI_char>x_internal): Rename to ...
14089 (altivec_stve<VI_char>x): ... this.
14090 (doublee<mode>2, unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2,
14091 doubleh<mode>2, unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2,
14092 reduc_plus_scal_<mode>): Adjust.
14093 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Adjust
14095 (rs6000_cpu_cpp_builtins): Adjust.
14096 (altivec_resolve_overloaded_builtin): Remove -maltivec=be support.
14097 * config/rs6000/rs6000-protos.h (altivec_expand_lvx_be,
14098 altivec_expand_stvx_be, altivec_expand_stvex_be): Delete.
14099 * config/rs6000/rs6000.c (rs6000_option_override_internal): Remove
14100 -maltivec=be support.
14101 (rs6000_split_vec_extract_var): Adjust.
14102 (rs6000_split_v4si_init): Adjust.
14103 (swap_selector_for_mode): Delete.
14104 (altivec_expand_lvx_be, altivec_expand_stvx_be,
14105 altivec_expand_stvex_be): Delete.
14106 (altivec_expand_lv_builtin, altivec_expand_stv_builtin): Remove
14107 -maltivec=be support.
14108 (rs6000_gimple_fold_builtin): Ditto.
14109 (rs6000_generate_float2_double_code, rs6000_generate_float2_code):
14111 * config/rs6000/rs6000.h (VECTOR_ELT_ORDER_BIG): Delete.
14112 (TARGET_DIRECT_MOVE_64BIT): Adjust.
14113 * config/rs6000/rs6000.md (split for extendsidi2 for vectors): Adjust.
14114 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Delete.
14115 * config/rs6000/vsx.md (floate<mode>, unsfloatev2di, floato<mode>,
14116 unsfloatov2di, vsignedo_v2df, vsignede_v2df, vunsignedo_v2df,
14117 vunsignede_v2df, vsx_extract_<mode>_p9, *vsx_extract_si,
14118 *vsx_extract_<mode>_p8, *vsx_extract_si_<uns>float_df,
14119 *vsx_extract_si_<uns>float_<mode>, vsx_set_<mode>_p9, vsx_set_v4sf_p9,
14120 *vsx_insert_extract_v4sf_p9, *vsx_insert_extract_v4sf_p9_2, and an
14121 anonymous split): Adjust.
14122 (vsx_mergel_<mode>, vsx_mergeh_<mode>): Remove -maltivec=be support.
14123 (vsx_xxspltd_<mode>, extract4b, insert4b): Adjust.
14125 2018-05-10 Eric Botcazou <ebotcazou@adacore.com>
14127 * configure.ac (gcc_gxx_include_dir_add_sysroot): Set it to 1 only
14128 when --with-gxx-include-dir is also specified.
14129 * configure: Regenerate.
14131 2018-05-09 Jim Wilson <jimw@sifive.com>
14134 * config.gcc (riscv*-*-*): Handle --with-multilib-list.
14135 * config/riscv/t-withmultilib: New.
14136 * config/riscv/withmultilib.h: New.
14137 * doc/install.texi: Document RISC-V --with-multilib-list support.
14139 2018-05-09 Richard Biener <rguenther@suse.de>
14141 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Fill a cost
14143 (vect_bb_vectorization_profitable_p): Adjust. Compute
14144 actual scalar cost using the cost vector and the add_stmt_cost
14147 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
14149 PR rtl-optimization/85645
14150 * config/rs6000/rs6000.c (rs6000_emit_prologue_components): Put a SET
14151 in the REG_CFA_REGISTER note for LR, don't leave it empty.
14153 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
14155 PR rtl-optimization/85645
14156 * shrink-wrap.c (spread_components): Return a boolean saying if
14157 anything was changed.
14158 (try_shrink_wrapping_separate): Iterate spread_components until
14159 nothing changes anymore.
14161 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
14163 PR rtl-optimization/85645
14164 * regrename.c (build_def_use): Also kill the chains that include the
14165 destination of a REG_CFA_REGISTER note.
14167 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
14169 PR rtl-optimization/85645
14170 * regcprop.c (copyprop_hardreg_forward_1): Don't propagate into an
14171 insn that has a REG_CFA_REGISTER note.
14173 2018-05-09 Richard Sandiford <richard.sandiford@linaro.org>
14175 * cfgexpand.c (expand_clobber): New function.
14176 (expand_gimple_stmt_1): Use it.
14177 * tree-vect-stmts.c (vect_clobber_variable): New function,
14179 (vectorizable_simd_clone_call): ...here.
14180 (vectorizable_store): Emit a clobber either side of an
14181 IFN_STORE_LANES sequence.
14182 (vectorizable_load): Emit a clobber after an IFN_LOAD_LANES sequence.
14184 2018-05-09 Tom de Vries <tom@codesourcery.com>
14187 * config/nvptx/nvptx.md (define_insn "trap", define_insn "trap_if_true")
14188 (define_insn "trap_if_false"): Add exit after trap.
14190 2018-05-09 Eric Botcazou <ebotcazou@adacore.com>
14192 PR rtl-optimization/85638
14193 * bb-reorder.c: Include common/common-target.h.
14194 (create_forwarder_block): New function extracted from...
14195 (fix_up_crossing_landing_pad): ...here. Rename into...
14196 (dw2_fix_up_crossing_landing_pad): ...this.
14197 (sjlj_fix_up_crossing_landing_pad): New function.
14198 (find_rarely_executed_basic_blocks_and_crossing_edges): In SJLJ mode,
14199 call sjlj_fix_up_crossing_landing_pad if there are incoming EH edges
14200 from both partitions and exit the loop after one iteration.
14202 2018-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
14205 * doc/extend.texi (PowerPC Built-in Functions): Rename this
14207 (Basic PowerPC Built-in Functions): The new name of the
14208 subsection previously known as "PowerPC Built-in Functions".
14209 (Basic PowerPC Built-in Functions Available on all Configurations):
14211 (Basic PowerPC Built-in Functions Available on ISA 2.05): New
14213 (Basic PowerPC Built-in Functions Available on ISA 2.06): New
14215 (Basic PowerPC Built-in Functions Available on ISA 2.07): New
14217 (Basic PowerPC Built-in Functions Available on ISA 3.0): New
14220 2018-05-08 Jim Wilson <jimw@sifive.com>
14222 * config/riscv/linux.h (MUSL_ABI_SUFFIX): Delete unnecessary backslash.
14223 (LD_EMUL_SUFFIX): New.
14224 (LINK_SPEC): Use it.
14226 2018-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
14228 * doc/extend.texi (PowerPC Built-in Functions): Rename this
14230 (Basic PowerPC Built-in Functions): The new name of the
14231 subsection previously known as "PowerPC Built-in Functions".
14232 (Basic PowerPC Built-in Functions Available on all Configurations):
14234 (Basic PowerPC Built-in Functions Available on ISA 2.05): New
14236 (Basic PowerPC Built-in Functions Available on ISA 2.06): New
14238 (Basic PowerPC Built-in Functions Available on ISA 2.07): New
14240 (Basic PowerPC Built-in Functions Available on ISA 3.0): New
14243 2018-05-08 Jakub Jelinek <jakub@redhat.com>
14246 * config/i386/i386.md: Add peepholes for mem {+,-,&,|,^}= x; mem != 0
14247 after cmpelim optimization.
14249 2018-05-08 Olga Makhotina <olga.makhotina@intel.com>
14251 * config.gcc: Support "goldmont".
14252 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "goldmont".
14253 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
14254 PROCESSOR_GOLDMONT.
14255 * config/i386/i386.c (m_GOLDMONT): Define.
14256 (processor_target_table): Add "goldmont".
14257 (PTA_GOLDMONT): Define.
14258 (ix86_lea_outperforms): Add TARGET_GOLDMONT.
14259 (get_builtin_code_for_version): Handle PROCESSOR_GOLDMONT.
14260 (fold_builtin_cpu): Add M_INTEL_GOLDMONT.
14261 (fold_builtin_cpu): Add "goldmont".
14262 (ix86_add_stmt_cost): Add TARGET_GOLDMONT.
14263 (ix86_option_override_internal): Add "goldmont".
14264 * config/i386/i386.h (processor_costs): Define TARGET_GOLDMONT.
14265 (processor_type): Add PROCESSOR_GOLDMONT.
14266 * config/i386/i386.md: Add CPU "glm".
14267 * config/i386/glm.md: New file.
14268 * config/i386/x86-tune.def: Add m_GOLDMONT.
14269 * doc/invoke.texi: Add goldmont as x86 -march=/-mtune= CPU type.
14271 2018-05-08 Jakub Jelinek <jakub@redhat.com>
14274 * config/i386/i386.c (ix86_expand_sse2_abs): Handle E_V2DImode and
14276 * config/i386/sse.md (abs<mode>2): Use VI_AVX2 iterator instead of
14277 VI1248_AVX512VL_AVX512BW. Handle V2DImode and V4DImode if not
14278 TARGET_AVX512VL using ix86_expand_sse2_abs. Formatting fixes.
14281 * config/i386/i386.c (ix86_fold_builtin): Handle
14282 IX86_BUILTIN_{,P}MOVMSK{PS,PD,B}{,128,256}.
14285 * config/i386/sse.md (ssequaterinsnmode): New mode attribute.
14286 (*<extract_type>_vinsert<shuffletype><extract_suf>_0): New pattern.
14288 2018-05-08 Richard Earnshaw <rearnsha@arm.com>
14291 * config/arm/parsecpu.awk (check_cpu): Fix operator precedence.
14292 (check_arch): Likewise.
14293 (check_fpu): Return the result rather than printing it.
14294 (end arch): Fix operator precedence.
14295 (end cpu): Likewise.
14296 (END): Print the result from check_fpu.
14298 2018-05-08 Richard Sandiford <richard.sandiford@linaro.org>
14299 Alan Hayward <alan.hayward@arm.com>
14300 David Sherwood <david.sherwood@arm.com>
14302 * config/aarch64/aarch64-sve.md (*pred_cmp<cmp_op><mode>_combine)
14303 (*pred_cmp<cmp_op><mode>, *fcm<cmp_op><mode>_and_combine)
14304 (*fcmuo<mode>_and_combine, *fcm<cmp_op><mode>_and)
14305 (*fcmuo<mode>_and): New patterns.
14307 2018-05-08 Richard Sandiford <richard.sandiford@linaro.org>
14309 * config/aarch64/iterators.md (UNSPEC_COND_LO, UNSPEC_COND_LS)
14310 (UNSPEC_COND_HI, UNSPEC_COND_HS, UNSPEC_COND_UO): Delete.
14311 (SVE_INT_CMP, SVE_FP_CMP): New code iterators.
14312 (cmp_op, sve_imm_con): New code attributes.
14313 (SVE_COND_INT_CMP, imm_con): Delete.
14314 (cmp_op): Remove above unspecs from int attribute.
14315 * config/aarch64/aarch64-sve.md (*vec_cmp<cmp_op>_<mode>): Rename
14317 (*cmp<cmp_op><mode>): ...this. Use UNSPEC_MERGE_PTRUE instead of
14318 comparison-specific unspecs.
14319 (*vec_cmp<cmp_op>_<mode>_ptest): Rename to...
14320 (*cmp<cmp_op><mode>_ptest): ...this and adjust likewise.
14321 (*vec_cmp<cmp_op>_<mode>_cc): Rename to...
14322 (*cmp<cmp_op><mode>_cc): ...this and adjust likewise.
14323 (*vec_fcm<cmp_op><mode>): Rename to...
14324 (*fcm<cmp_op><mode>): ...this and adjust likewise.
14325 (*vec_fcmuo<mode>): Rename to...
14326 (*fcmuo<mode>): ...this and adjust likewise.
14327 (*pred_fcm<cmp_op><mode>): New pattern.
14328 * config/aarch64/aarch64.c (aarch64_emit_unop, aarch64_emit_binop)
14329 (aarch64_emit_sve_ptrue_op, aarch64_emit_sve_ptrue_op_cc): New
14331 (aarch64_unspec_cond_code): Remove handling of LTU, GTU, LEU, GEU
14333 (aarch64_gen_unspec_cond, aarch64_emit_unspec_cond): Delete.
14334 (aarch64_emit_sve_predicated_cond): New function.
14335 (aarch64_expand_sve_vec_cmp_int): Use aarch64_emit_sve_ptrue_op_cc.
14336 (aarch64_emit_unspec_cond_or): Replace with...
14337 (aarch64_emit_sve_or_conds): ...this new function. Use
14338 aarch64_emit_sve_ptrue_op for the individual comparisons and
14339 aarch64_emit_binop to OR them together.
14340 (aarch64_emit_inverted_unspec_cond): Replace with...
14341 (aarch64_emit_sve_inverted_cond): ...this new function. Use
14342 aarch64_emit_sve_ptrue_op for the comparison and
14343 aarch64_emit_unop to invert the result.
14344 (aarch64_expand_sve_vec_cmp_float): Update after the above
14345 changes. Use aarch64_emit_sve_ptrue_op for native comparisons.
14347 2018-05-07 Nathan Sidwell <nathan@acm.org>
14349 * doc/invoke.texi (C++ Dialect Options): Remove -ffor-scope.
14350 * doc/extend.texi (Deprecated Features): Remove -fno-for-scope
14351 (Backwards Compatibility): Likewise.
14353 2018-05-07 Luis Machado <luis.machado@linaro.org>
14357 2018-05-07 Luis Machado <luis.machado@linaro.org>
14359 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
14360 <prefetch_dynamic_strides>: New const bool field.
14361 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
14362 prefetch_dynamic_strides.
14363 (exynosm1_prefetch_tune): Likewise.
14364 (thunderxt88_prefetch_tune): Likewise.
14365 (thunderx_prefetch_tune): Likewise.
14366 (thunderx2t99_prefetch_tune): Likewise.
14367 (qdf24xx_prefetch_tune): Likewise. Set prefetch_dynamic_strides
14369 (aarch64_override_options_internal): Update to set
14370 PARAM_PREFETCH_DYNAMIC_STRIDES.
14371 * doc/invoke.texi (prefetch-dynamic-strides): Document new option.
14372 * params.def (PARAM_PREFETCH_DYNAMIC_STRIDES): New.
14373 * params.h (PARAM_PREFETCH_DYNAMIC_STRIDES): Define.
14374 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Account for
14375 prefetch-dynamic-strides setting.
14377 2018-05-07 Luis Machado <luis.machado@linaro.org>
14379 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
14380 <minimum_stride>: New const int field.
14381 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
14382 minimum_stride field.
14383 (exynosm1_prefetch_tune): Likewise.
14384 (thunderxt88_prefetch_tune): Likewise.
14385 (thunderx_prefetch_tune): Likewise.
14386 (thunderx2t99_prefetch_tune): Likewise.
14387 (qdf24xx_prefetch_tune): Likewise. Set minimum_stride to 2048.
14388 (aarch64_override_options_internal): Update to set
14389 PARAM_PREFETCH_MINIMUM_STRIDE.
14390 * doc/invoke.texi (prefetch-minimum-stride): Document new option.
14391 * params.def (PARAM_PREFETCH_MINIMUM_STRIDE): New.
14392 * params.h (PARAM_PREFETCH_MINIMUM_STRIDE): Define.
14393 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Return false if
14394 stride is constant and is below the minimum stride threshold.
14396 2018-05-07 Luis Machado <luis.machado@linaro.org>
14398 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune) <l2_cache_size>: Set
14401 2018-05-07 Luis Machado <luis.machado@linaro.org>
14403 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
14404 <prefetch_dynamic_strides>: New const bool field.
14405 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
14406 prefetch_dynamic_strides.
14407 (exynosm1_prefetch_tune): Likewise.
14408 (thunderxt88_prefetch_tune): Likewise.
14409 (thunderx_prefetch_tune): Likewise.
14410 (thunderx2t99_prefetch_tune): Likewise.
14411 (qdf24xx_prefetch_tune): Likewise. Set prefetch_dynamic_strides
14413 (aarch64_override_options_internal): Update to set
14414 PARAM_PREFETCH_DYNAMIC_STRIDES.
14415 * doc/invoke.texi (prefetch-dynamic-strides): Document new option.
14416 * params.def (PARAM_PREFETCH_DYNAMIC_STRIDES): New.
14417 * params.h (PARAM_PREFETCH_DYNAMIC_STRIDES): Define.
14418 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Account for
14419 prefetch-dynamic-strides setting.
14421 2018-05-07 Luis Machado <luis.machado@linaro.org>
14423 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
14424 <minimum_stride>: New const int field.
14425 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
14426 minimum_stride field.
14427 (exynosm1_prefetch_tune): Likewise.
14428 (thunderxt88_prefetch_tune): Likewise.
14429 (thunderx_prefetch_tune): Likewise.
14430 (thunderx2t99_prefetch_tune): Likewise.
14431 (qdf24xx_prefetch_tune): Likewise. Set minimum_stride to 2048.
14432 (aarch64_override_options_internal): Update to set
14433 PARAM_PREFETCH_MINIMUM_STRIDE.
14434 * doc/invoke.texi (prefetch-minimum-stride): Document new option.
14435 * params.def (PARAM_PREFETCH_MINIMUM_STRIDE): New.
14436 * params.h (PARAM_PREFETCH_MINIMUM_STRIDE): Define.
14437 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Return false if
14438 stride is constant and is below the minimum stride threshold.
14440 2018-05-06 Jakub Jelinek <jakub@redhat.com>
14443 * cfgexpand.c (expand_asm_stmt): Don't create a temporary if
14444 the type is addressable. Don't force op into register if it has
14447 2018-05-05 Roland McGrath <mcgrathr@google.com>
14450 * varasm.c (default_section_type_flags): Set SECTION_NOTYPE for
14451 any section for which we don't know a specific type it should have,
14452 regardless of name. Previously this was done only for the exact
14453 names ".init_array", ".fini_array", and ".preinit_array".
14454 (default_elf_asm_named_section): Add comment about
14455 relationship with default_section_type_flags and SECTION_NOTYPE.
14456 (get_section): Don't consider it a type conflict if one side has
14457 SECTION_NOTYPE and the other doesn't, as long as neither has the
14458 SECTION_BSS et al used in the default_section_type_flags logic.
14460 2018-05-05 Tom de Vries <tom@codesourcery.com>
14463 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG_3): Define.
14464 (workaround_barsyncs): New function.
14465 (nvptx_reorg): Use workaround_barsyncs.
14466 * config/nvptx/nvptx.md (define_c_enum "unspecv"): Add UNSPECV_MEMBAR.
14467 (define_expand "nvptx_membar_cta"): New define_expand.
14468 (define_insn "*nvptx_membar_cta"): New insn.
14470 2018-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
14472 * brig-builtins.def: Add consts to ptrs etc. in BRIG builtin defs.
14473 To improve optimization opportunities.
14474 * builtin-types.def: The new needed builtin types for the above.
14476 2018-05-04 Richard Biener <rguenther@suse.de>
14478 * bb-reorder.c (sanitize_hot_paths): Release hot_bbs_to_check.
14479 * gimple-ssa-store-merging.c
14480 (imm_store_chain_info::output_merged_store): Remove redundant create,
14481 release split_store vector contents on failure.
14482 * tree-vect-slp.c (vect_schedule_slp_instance): Avoid leaking
14483 scalar stmt vector on cache hit.
14485 2018-05-04 Segher Boessenkool <segher@kernel.crashing.org>
14487 * common/config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
14489 * config.gcc (powerpc-xilinx-eabi*): Remove.
14490 * config/rs6000/predicates.md (easy_fp_constant): Remove Xilinx FP
14492 (fusion_addis_mem_combo_load): Ditto.
14493 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove Xilinx
14495 (rs6000_cpu_cpp_builtins): Ditto.
14496 * config/rs6000/rs6000-linux.c
14497 (rs6000_linux_float_exceptions_rounding_supported_p): Ditto.
14498 * config/rs6000/rs6000-opts.h (enum fpu_type_t): Delete.
14499 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove Xilinx FP
14501 (rs6000_setup_reg_addr_masks): Ditto.
14502 (rs6000_init_hard_regno_mode_ok): Ditto.
14503 (rs6000_option_override_internal): Ditto.
14504 (legitimate_lo_sum_address_p): Ditto.
14505 (rs6000_legitimize_address): Ditto.
14506 (rs6000_legitimize_reload_address): Ditto.
14507 (rs6000_legitimate_address_p): Ditto.
14508 (abi_v4_pass_in_fpr): Ditto.
14509 (setup_incoming_varargs): Ditto.
14510 (rs6000_gimplify_va_arg): Ditto.
14511 (rs6000_split_multireg_move): Ditto.
14512 (rs6000_savres_strategy): Ditto.
14513 (rs6000_emit_prologue_components): Ditto.
14514 (rs6000_emit_epilogue_components): Ditto.
14515 (rs6000_emit_prologue): Ditto.
14516 (rs6000_emit_epilogue): Ditto.
14517 (rs6000_elf_file_end): Ditto.
14518 (rs6000_function_value): Ditto.
14519 (rs6000_libcall_value): Ditto.
14520 * config/rs6000/rs6000.h: Ditto.
14521 (TARGET_MINMAX_SF, TARGET_MINMAX_DF): Delete, merge to ...
14522 (TARGET_MINMAX): ... this. New.
14523 (TARGET_SF_FPR, TARGET_DF_FPR, TARGET_SF_INSN, TARGET_DF_INSN): Delete.
14524 * config/rs6000/rs6000.md: Remove Xilinx FP support.
14525 (*movsi_internal1_single): Delete.
14526 * config/rs6000/rs6000.opt (msingle-float, mdouble-float, msimple-fpu,
14527 mfpu=, mxilinx-fpu): Delete.
14528 * config/rs6000/singlefp.h: Delete.
14529 * config/rs6000/sysv4.h: Remove Xilinx FP support.
14530 * config/rs6000/t-rs6000: Ditto.
14531 * config/rs6000/t-xilinx: Delete.
14532 * config/rs6000/titan.md: Adjust for fp_type removal.
14533 * config/rs6000/vsx.md: Remove Xilinx FP support.
14534 (VStype_simple): Delete.
14535 (VSfptype_simple, VSfptype_mul, VSfptype_div, VSfptype_sqrt): Delete.
14536 * config/rs6000/xfpu.h: Delete.
14537 * config/rs6000/xfpu.md: Delete.
14538 * config/rs6000/xilinx.h: Delete.
14539 * config/rs6000/xilinx.opt: Delete.
14540 * doc/invoke.texi (RS/6000 and PowerPC Options): Remove
14541 -msingle-float, -mdouble-float, -msimple-fpu, -mfpu=, and -mxilinx-fpu.
14543 2018-05-04 Tom de Vries <tom@codesourcery.com>
14546 * builtins.c (expand_builtin_goacc_parlevel_id_size): Handle null target
14549 2018-05-04 Richard Biener <rguenther@suse.de>
14551 PR middle-end/85627
14552 * tree-complex.c (update_complex_assignment): We are always in SSA form.
14553 (expand_complex_div_wide): Likewise.
14554 (expand_complex_operations_1): Likewise.
14555 (expand_complex_libcall): Preserve EH info of the original stmt.
14556 (tree_lower_complex): Handle removed blocks.
14557 * tree.c (build_common_builtin_nodes): Do not set ECF_NOTRHOW
14558 on complex multiplication and division libcall builtins.
14560 2018-05-04 Richard Biener <rguenther@suse.de>
14562 PR middle-end/85574
14563 * fold-const.c (negate_expr_p): Restrict negation of operand
14564 zero of a division to when we know that can happen without
14566 (fold_negate_expr_1): Likewise.
14568 2018-05-04 Jakub Jelinek <jakub@redhat.com>
14571 * real.h (real_nextafter): Declare.
14572 * real.c (real_nextafter): New function.
14573 * fold-const-call.c (fold_const_nextafter): New function.
14574 (fold_const_call_sss): Call it for CASE_CFN_NEXTAFTER and
14575 CASE_CFN_NEXTTOWARD.
14576 (fold_const_call_1): For CASE_CFN_NEXTTOWARD call fold_const_call_sss
14577 even when arg1_mode is different from arg0_mode.
14579 2018-05-03 Nathan Sidwell <nathan@acm.org>
14581 * doc/extend.texi (Deprecated Features): Remove
14582 -ffriend-injection.
14583 (Backwards Compatibility): Likewise.
14584 * doc/invoke.texi (C++ Language Options): Likewise.
14585 (C++ Dialect Options): Likewise.
14587 2018-05-03 Jakub Jelinek <jakub@redhat.com>
14590 * config/i386/avx512fintrin.h (_mm512_mullox_epi64,
14591 _mm512_mask_mullox_epi64): New intrinsics.
14593 2018-05-03 Tom de Vries <tom@codesourcery.com>
14596 * doc/sourcebuild.texi (Commands for use in dg-final, Scan optimization
14597 dump files): Add offload-tree.
14599 2018-05-03 Richard Biener <rguenther@suse.de>
14601 PR tree-optimization/85615
14602 * tree-ssa-threadupdate.c (thread_block_1): Only allow exits
14603 to loops not nested in BBs loop father to avoid creating multi-entry
14606 2018-05-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14608 PR tree-optimization/70291
14609 * tree-complex.c (expand_complex_libcall): Add type, inplace_p
14610 arguments. Change return type to tree. Emit libcall as a new
14611 statement rather than replacing existing one when inplace_p is true.
14612 (expand_complex_multiplication_components): New function.
14613 (expand_complex_multiplication): Expand floating-point complex
14614 multiplication using the above.
14615 (expand_complex_division): Rename inner_type parameter to type.
14616 Update expand_complex_libcall call-site.
14617 (expand_complex_operations_1): Update expand_complex_multiplication
14618 and expand_complex_division call-sites.
14620 2018-05-02 Jakub Jelinek <jakub@redhat.com>
14623 * config/i386/i386.md (*ashl<dwi>3_doubleword_mask,
14624 *ashl<dwi>3_doubleword_mask_1, *<shift_insn><dwi>3_doubleword_mask,
14625 *<shift_insn><dwi>3_doubleword_mask_1): In condition require that
14626 the highest significant bit of the shift count mask is clear. In
14627 check whether and[sq]i3 is needed verify that all significant bits
14628 of the shift count other than the highest are set.
14630 2018-05-02 Tom de Vries <tom@codesourcery.com>
14633 * builtins.def (DEF_GOACC_BUILTIN_ONLY): Define.
14634 * omp-builtins.def (BUILT_IN_GOACC_PARLEVEL_ID)
14635 (BUILT_IN_GOACC_PARLEVEL_SIZE): New builtin.
14636 * builtins.c (expand_builtin_goacc_parlevel_id_size): New function.
14637 (expand_builtin): Call expand_builtin_goacc_parlevel_id_size.
14638 * doc/extend.texi (Other Builtins): Add __builtin_goacc_parlevel_id and
14639 __builtin_goacc_parlevel_size.
14641 2018-05-02 Richard Biener <rguenther@suse.de>
14643 PR tree-optimization/85597
14644 * tree-vect-stmts.c (vectorizable_operation): For ternary SLP
14645 do not use split vect_get_vec_defs call but call vect_get_slp_defs
14648 2018-05-02 Tom de Vries <tom@codesourcery.com>
14651 * doc/sourcebuild.texi (Commands for use in dg-final, Scan optimization
14652 dump files): Add ltrans-tree.
14654 2018-05-02 Tom de Vries <tom@codesourcery.com>
14657 * doc/sourcebuild.texi (Commands for use in dg-final, Scan optimization
14658 dump files): Add wpa-ipa.
14660 2018-05-02 Segher Boessenkool <segher@kernel.crashing.org>
14662 * config.gcc (powerpc*-*-*): Remove paired.h. Unsupport the
14663 powerpc*-*-linux*paired* target.
14664 * config/rs6000/750cl.h: Delete.
14665 * config/rs6000/paired.h: Delete.
14666 * config/rs6000/paired.md: Delete.
14667 * config/rs6000/predicates.md (easy_vector_constant): Remove paired
14669 * config/rs6000/rs6000-builtin.def: Remove paired float support.
14670 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
14671 comment. Remove paired float support.
14672 * config/rs6000/rs6000-modes.def: Remove V2SF and V2SI.
14673 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete
14675 * config/rs6000/rs6000-protos.h (paired_expand_vector_init,
14676 paired_emit_vector_cond_expr, paired_expand_vector_move): Delete
14678 * config/rs6000/rs6000.c: Remove paired float support.
14679 (paired_expand_vector_init, paired_expand_vector_move,
14680 paired_emit_vector_compare, paired_emit_vector_cond_expr,
14681 (paired_expand_lv_builtin, paired_expand_stv_builtin,
14682 paired_expand_builtin, paired_expand_predicate_builtin,
14683 paired_init_builtins): Delete.
14684 * config/rs6000/rs6000.h: Remove paired float support.
14685 * config/rs6000/rs6000.md: Remove paired float support.
14686 (move_from_CR_ov_bit): Delete.
14687 * config/rs6000/rs6000.opt (mpaired): Delete.
14688 * config/rs6000/t-rs6000: Remove paired.md from MD_INCLUDES.
14689 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mpaired.
14691 2018-05-02 Richard Biener <rguenther@suse.de>
14693 PR middle-end/85567
14694 * gimplify.c (gimplify_save_expr): When in SSA form allow
14695 SAVE_EXPRs to compute to SSA vars.
14697 2018-05-02 Jakub Jelinek <jakub@redhat.com>
14700 * config/i386/i386.md (*ashl<dwi>3_doubleword_mask,
14701 *ashl<dwi>3_doubleword_mask_1, *<shift_insn><dwi>3_doubleword_mask,
14702 *<shift_insn><dwi>3_doubleword_mask_1): If and[sq]i3 is needed, don't
14703 clobber operands[2], instead use a new pseudo. Formatting fixes.
14705 2018-05-02 Richard Sandiford <richard.sandiford@linaro.org>
14707 PR tree-optimization/85586
14708 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Only
14709 exit early for statements in the same group if the accesses are
14712 2018-05-02 Tom de Vries <tom@codesourcery.com>
14715 * lto-wrapper.c (compile_offload_image): Add "could not find mkoffload"
14718 2018-05-01 Marc Glisse <marc.glisse@inria.fr>
14720 PR tree-optimization/85143
14721 * match.pd (A<B&A<C): Extend to BIT_IOR_EXPR.
14723 2018-05-01 Tom de Vries <tom@codesourcery.com>
14726 * config/nvptx/mkoffload.c (main): Suggest using -B in "offload compiler
14727 not found" error message.
14729 2018-05-01 Tom de Vries <tom@codesourcery.com>
14732 * vec.h (VEC_ORDERED_REMOVE_IF, VEC_ORDERED_REMOVE_IF_FROM_TO): Define.
14733 * vec.c (test_ordered_remove_if): New function.
14734 (vec_c_tests): Call test_ordered_remove_if.
14735 * dwarf2cfi.c (connect_traces): Use VEC_ORDERED_REMOVE_IF_FROM_TO.
14736 * lto-streamer-out.c (prune_offload_funcs): Use VEC_ORDERED_REMOVE_IF.
14737 * tree-vect-patterns.c (vect_pattern_recog_1): Use
14738 VEC_ORDERED_REMOVE_IF.
14740 2018-05-01 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
14742 PR tree-optimization/82665
14743 * vr-values.c (vr_values::extract_range_from_binary_expr): Handle
14744 pointer subtraction where arguments come from a memchr call.
14746 2018-05-01 Jakub Jelinek <jakub@redhat.com>
14748 * configure.ac (LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION): Use
14749 --push-state --as-needed and --pop-state instead of --as-needed and
14750 --no-as-needed if ld supports it.
14751 * configure: Regenerated.
14754 * doc/install.texi2html: Replace _002d with - and _002a with * in
14755 generated html files using sed.
14757 2018-04-30 David Malcolm <dmalcolm@redhat.com>
14760 * gcc-rich-location.c (blank_line_before_p): New function.
14761 (use_new_line): New function.
14762 (gcc_rich_location::add_fixit_insert_formatted): New function.
14763 * gcc-rich-location.h
14764 (gcc_rich_location::add_fixit_insert_formatted): New function.
14766 2018-04-30 David Malcolm <dmalcolm@redhat.com>
14768 * selftest.c (assert_streq): Rename "expected" and "actual" to
14769 "val1" and "val2". Extend NULL-handling to cover both inputs
14770 symmetrically, while still requiring both to be non-NULL for a pass.
14771 * selftest.h (assert_streq): Rename "expected" and "actual" to
14773 (ASSERT_EQ): Likewise.
14774 (ASSERT_EQ_AT): Likewise.
14775 (ASSERT_KNOWN_EQ): Likewise.
14776 (ASSERT_KNOWN_EQ_AT): Likewise.
14777 (ASSERT_NE): Likewise.
14778 (ASSERT_MAYBE_NE): Likewise.
14779 (ASSERT_MAYBE_NE_AT): Likewise.
14780 (ASSERT_STREQ): Likewise. Clarify that both must be non-NULL for
14781 the assertion to pass.
14782 (ASSERT_STREQ_AT): Likewise.
14784 2018-04-30 Jonathan Wakely <jwakely@redhat.com>
14786 * doc/invoke.texi (-fpie, -fPIE): Fix grammar and clarify
14787 interaction with -pie.
14789 2018-04-30 David Malcolm <dmalcolm@redhat.com>
14791 * selftest.h: Fix alphabetization of per-source-file selftest
14794 2018-04-30 Jason Merrill <jason@redhat.com>
14796 PR c++/61982 - dead stores to destroyed objects.
14797 * gimplify.c (gimplify_modify_expr): Simplify complex lvalue on LHS
14800 2018-04-30 Jason Merrill <jason@redhat.com>
14802 * tree.c (build_clobber): New.
14803 * tree.h: Declare it.
14804 * gimplify.c (gimplify_bind_expr, gimplify_target_expr): Use it.
14806 2018-04-30 David Malcolm <dmalcolm@redhat.com>
14808 * diagnostic-show-locus.c (layout::layout): Update for
14809 location_get_source_line returning a char_span.
14810 (struct char_span): Move to input.h.
14811 (struct correction): Update for fields in char_span becoming
14813 (struct source_line): Update for location_get_source_line
14814 returning a char_span.
14815 (layout::print_line): Likewise.
14816 * edit-context.c (edited_file::print_content): Likewise.
14817 (edited_file::print_diff_hunk): Likewise.
14818 (edited_file::print_run_of_changed_lines): Likewise.
14819 (edited_file::get_num_lines): Likewise.
14820 (edited_line::edited_line): Likewise.
14821 * final.c (asm_show_source): Likewise.
14822 * input.c (location_get_source_line): Convert return type
14823 from const char * to char_span, losing the final "line_len"
14825 (dump_location_info): Update for the above.
14826 (get_substring_ranges_for_loc): Likewise. Use a char_span
14827 when handling the literal within the line.
14828 (test_reading_source_line): Update for location_get_source_line
14829 returning a char_span.
14830 * input.h (class char_span): Move here from
14831 diagnostic-show-locus.c, converting from a struct to a class.
14832 Make data members private.
14833 (char_span::operator bool): New.
14834 (char_span::length): New.
14835 (char_span::get_buffer): New.
14836 (char_span::operator[]): New.
14837 (char_span::subspan): Make const.
14838 (char_span::xstrdup): New.
14839 (location_get_source_line): Convert return type from const char *
14840 to char_span, losing the final "line_size" param.
14842 2018-04-30 Jan Hubicka <jh@suse.cz>
14844 * lto-wrapper.c (ltrans_priorities): New static var.
14845 (cmp_priority): New.
14846 (run_gcc): Read priorities and if doing parallel build order
14847 the Makefile by them.
14849 2018-04-30 David Malcolm <dmalcolm@redhat.com>
14851 * input.h (builtins_location_check): Convert to a STATIC_ASSERT.
14853 2018-04-30 Richard Biener <rguenther@suse.de>
14855 * tree-cfg.c (verify_address): Remove base argument, add
14856 flag whether to check TREE_ADDRESSABLE and do that.
14857 (verify_expr): Remove.
14858 (verify_types_in_gimple_reference): Add pieces from verify_expr.
14859 (verify_gimple_assign_single): Likewise.
14860 (verify_gimple_switch): Likewise.
14861 (verify_expr_location_1): Dereference tp once. Add (disabled)
14862 piece from verify_expr.
14863 (verify_gimple_in_cfg): Do not call verify_expr on all ops.
14865 2018-04-30 Claudiu Zissulescu <claziss@synopsys.com>
14867 * config/arc/linux.h (CLEAR_INSN_CACHE): Define.
14869 2018-04-30 Claudiu Zissulescu <claziss@synopsys.com>
14871 * config/arc/arc-protos.h (prepare_extend_operands): Remove.
14872 (small_data_pattern): Likewise.
14873 (arc_rewrite_small_data): Likewise.
14874 * config/arc/arc.c (LEGITIMATE_SMALL_DATA_OFFSET_P): Remove.
14875 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Likewise.
14876 (get_symbol_alignment): New function.
14877 (legitimate_small_data_address_p): Likewise.
14878 (legitimate_scaled_address): Update, call
14879 legitimate_small_data_address_p.
14880 (output_sdata): New static variable.
14881 (arc_print_operand): Update how we handle small data operands.
14882 (arc_print_operand_address): Likewise.
14883 (arc_legitimate_address_p): Update, use
14884 legitimate_small_data_address_p.
14885 (arc_rewrite_small_data_p): Remove.
14886 (arc_rewrite_small_data_1): Likewise.
14887 (arc_rewrite_small_data): Likewise.
14888 (small_data_pattern): Likewise.
14889 (compact_sda_memory_operand): Update to use
14890 legitimate_small_data_address_p and get_symbol_alignment.
14891 (prepare_move_operands): Don't rewite sdata pattern.
14892 (prepare_extend_operands): Remove.
14893 * config/arc/arc.md (zero_extendqihi2): Don't rewrite sdata
14895 (zero_extendqisi2): Likewise.
14896 (zero_extendhisi2): Likewise.
14897 (extendqihi2): Likewise.
14898 (extendqisi2): Likewise.
14899 (extendhisi2): Likewise.
14900 (addsi3): Likewise.
14901 (subsi3): Likewise.
14902 (andsi3): Likewise.
14903 * config/arc/constraints.md (Usd): Change it to memory constraint.
14905 2018-04-30 Claudiu Zissulescu <claziss@synopsys.com>
14907 * config/arc/arc.c (arc_split_move): Allow signed 6-bit constants
14908 as source of std instructions.
14909 * config/arc/arc.md (movsi_insn): Update pattern predicate to
14910 allow 6-bit constants as source for store instructions.
14911 (movdi_insn): Update instruction pattern to allow 6-bit constants
14912 as source for store instructions.
14914 2018-04-30 Jonathan Wakely <jwakely@redhat.com>
14916 * doc/invoke.texi (-fdebug-types-section): Fix grammar.
14918 2018-04-30 Nathan Sidwell <nathan@acm.org>
14919 Sandra Loosemore <sandra@codesourcery.com>
14921 * dumpfile.c (dump_open): Allow '-' for stdout.
14922 * doc/invoke.texi (Developer Options): Document dump filename
14923 determination early. Document stdin/stdout selection.
14925 2018-04-30 Andrew Sadek <andrew.sadek.se@gmail.com>
14927 Microblaze Target: PIC data text relative
14929 * config/microblaze/microblaze.opt: add new option -mpic-data-text-rel.
14930 * config/microblaze/microblaze-protos.h (microblaze_constant_address_p):
14932 * config/microblaze/microblaze.h (microblaze_constant_address_p):
14933 CONSTANT_ADDRESS_P definition to microblaze_constant_address_p.
14934 * config/microblaze/microblaze.c (TARGET_PIC_DATA_TEXT_REL):
14935 New addressing mode for data-text relative position indepenedent code.
14936 (microblaze_classify_unspec): add 'UNSPEC_TEXT' case ->
14937 'ADDRESS_SYMBOLIC_TXT_REL'.
14938 (microblaze_classify_address): Add handling for UNSPEC + CONST_INT.
14939 (microblaze_legitimate_pic_operand): Exclude function calls from
14940 pic operands in case of TARGET_PIC_DATA_TEXT_REL option.
14941 (microblaze_legitimize_address): Generate 'UNSPEC_TEXT' for all possible
14943 (microblaze_address_insns): Add 'ADDRESS_SYMBOLIC_TXT_REL' case.
14944 (print_operand): Add 'ADDRESS_SYMBOLIC_TXT_REL' case.
14945 (print_operand_address): Add 'ADDRESS_SYMBOLIC_TXT_REL' case + handling
14946 for 'address + offset'.
14947 (microblaze_expand_prologue): Add new function prologue call for
14949 (microblaze_asm_generate_pic_addr_dif_vec): Override new target hook
14950 'TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC' to disable address diff vector
14951 table in case of TARGET_PIC_DATA_TEXT_REL.
14952 (expand_pic_symbol_ref): Add handling for 'UNSPEC_TEXT'.
14953 * config/microblaze/microblaze.md (TARGET_PIC_DATA_TEXT_REL):
14954 Add new macros 'UNSPEC_TEXT',
14955 'UNSPEC_SET_TEXT' + add rule for setting r20 in function prologue
14956 + exclude function calls from 'UNSPEC_PLT' in case of data text
14958 * doc/tm.texi.in (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC): Add
14959 new target hook for generating address diff vector tables in case of
14961 * doc/tm.texi : Regenerate.
14962 * stmt.c (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC): Append new condition
14963 'targetm.asm_out.generate_pic_addr_diff_vec' to flag_pic in case
14964 of addr diff vector generation.
14965 * target.def (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC): Add
14966 target hook definition.
14967 * targhooks.h, targhooks.c (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC):
14968 Add default function for generate_pic_addr_diff_vec -> flag_pic.
14969 * doc/invoke.texi (Add new pic option): Add new microblaze pic
14970 option for data text relative.
14972 2018-04-30 Richard Biener <rguenther@suse.de>
14974 * tree-chrec.h (evolution_function_is_constant_p): Remove
14976 * tree-cfg.c (tree_node_can_be_shared): Re-order checks.
14978 2018-04-30 Richard Biener <rguenther@suse.de>
14981 * dwarf2out.c (gen_producer_string): Ignore -fchecking[=].
14983 2018-04-30 Richard Biener <rguenther@suse.de>
14985 PR tree-optimization/28364
14986 PR tree-optimization/85275
14987 * tree-ssa-loop-ch.c (ch_base::copy_headers): Stop after
14988 copying first exit test.
14990 2018-04-28 Mark Wielaard <mark@klomp.org>
14992 * dwarf2out.c (dwarf2out_finish): Add .debug_addr table header for
14993 dwarf_version >= 5.
14994 (dwarf_AT): Handle DW_AT_addr_base.
14995 (add_top_level_skeleton_die_attrs): Use dwarf_AT for DW_AT_addr_base.
14997 2018-04-28 Uros Bizjak <ubizjak@gmail.com>
15000 * config/i386/i386.md (*ashl<dwi>3_doubleword_mask): New pattern.
15001 (*ashl<dwi>3_doubleword_mask_1): Ditto.
15002 (*<shift_insn><dwi>3_doubleword_mask): Ditto.
15003 (*<shift_insn><dwi>3_doubleword_mask_1): Ditto.
15005 2018-04-28 Richard Biener <rguenther@suse.de>
15007 * tree-cfg.c (verify_gimple_phi): Take a gphi * argument.
15008 (verify_gimple_in_cfg): Rename visited_stmts to visited_throwing_stmts
15009 to reflect use. Only add interesting stmts.
15011 2018-04-27 Martin Jambor <mjambor@suse.cz>
15014 * ipa-cp.c (find_aggregate_values_for_callers_subset): Make sure
15015 the jump function allows for passing through aggregate values.
15017 2018-04-27 David Malcolm <dmalcolm@redhat.com>
15019 * input.h (in_system_header_at): Convert from macro to inline
15021 (from_macro_expansion_at): Likewise.
15022 (from_macro_definition_at): Likewise.
15024 2018-04-27 Jeff Law <law@redhat.com>
15026 * config.gcc: Mark tile* targets as deprecated/obsolete.
15028 2018-04-27 Richard Biener <rguenther@suse.de>
15030 * config/aarch64/aarch64.c: Simplify ap.__stack advance and
15033 2018-04-27 Richard Biener <rguenther@suse.de>
15035 * tree-cfg.c (verify_expr): Make dead code hit gcc_unreachable.
15037 2018-04-27 Uros Bizjak <ubizjak@gmail.com>
15039 * config/i386/i386.md (*movti_internal): Substitute Ye constraint
15040 with Yd constraint. Set "preferred_for_speed" attribute from
15041 TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC for alternatives
15042 with Yd constraint.
15043 (*movdi_internal): Ditto.
15044 (movti_interunit splitters): Remove
15045 TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC from insn condition.
15046 (movdi_interunit splitters): Ditto.
15047 * config/i386/constraints.md (Ye): Remove.
15048 (Yd): Do not depend on TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC.
15050 2018-04-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15053 * config/aarch64/constraints.md (Usg): Limit to 31.
15054 (Usj): Limit to 63.
15056 2018-04-27 Jakub Jelinek <jakub@redhat.com>
15058 PR tree-optimization/85529
15059 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Add FIRST_BB
15060 argument. Don't call get_nonzero_bits if opcode is ERROR_MARK_NODE,
15061 rhs2 def stmt's bb is dominated by first_bb and it isn't an obvious
15062 zero extension or masking of the MSB bit.
15063 (optimize_range_tests): Add FIRST_BB argument, pass it through
15064 to optimize_range_tests_var_bound.
15065 (maybe_optimize_range_tests, reassociate_bb): Adjust
15066 optimize_range_tests callers.
15068 2018-04-26 Richard Biener <rguenther@suse.de>
15069 Jakub Jelinek <jakub@redhat.com>
15071 * cgraph.h (symbol_table): Just declare debug method here.
15072 * symtab.c (symbol_table::debug): Define.
15074 2018-04-26 Eric Botcazou <ebotcazou@adacore.com>
15076 * loop-invariant.c (may_assign_reg_p): Return false for frame pointer.
15078 2018-04-26 Uros Bizjak <ubizjak@gmail.com>
15080 * config/i386/i386.md ("isa" attribute): Add x64_sse2.
15081 ("enabled" attribute): Handle x64_sse2 "isa" attribute.
15082 (*movdi_internal): Substitute Yi and Yj constraint with x
15083 and Ym and Yn constraint with y constraint. Update "isa"
15084 attribute and set "preferred_for_speed" attribute from
15085 TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC for updated alternatives.
15086 (*movsi_internal): Ditto.
15087 (*movdf_internal): Ditto.
15088 (*movsf_internal): Ditto.
15089 (*zero_extendsidi2): Ditto.
15090 * config/i386/sse.md (vec_set<mode>_0): Ditto.
15091 (sse2_loadld): Ditto.
15092 (*vec_extract<ssevecmodelower>_0): Ditto.
15093 (*vec_extractv4si_0_zext_sse4): Ditto.
15094 (vec_concatv2di): Ditto.
15095 (*vec_dup<mode>): Ditto.
15096 * config/i386/mmx.md (*mov<mode>_internal): Ditto.
15097 * config/i386/constraints.md (Yi): Remove.
15102 2018-04-26 Nathan Sidwell <nathan@acm.org>
15104 * dumpfile.c (dump_open): New.
15105 (dump_open_alternate_stream, dump_start, dump_begin): Call it.
15106 (dump_finish): Detect stdio/stderr by value not name.
15108 2018-04-26 Jonathan Wakely <jwakely@redhat.com>
15110 * doc/invoke.texi (-Wreturn-type): Document default status for C++.
15112 2018-04-26 Tom de Vries <tom@codesourcery.com>
15115 * config/nvptx/nvptx.c (verify_neutering_jumps)
15116 (verify_neutering_labels): New function
15117 (nvptx_single): Use verify_neutering_jumps and verify_neutering_labels.
15119 2018-04-26 Tom de Vries <tom@codesourcery.com>
15122 * config/nvptx/nvptx.c (needs_neutering_p): New function.
15123 (nvptx_single): Use needs_neutering_p to skip over insns that do not
15126 2018-04-26 Richard Biener <rguenther@suse.de>
15127 Tom de Vries <tom@codesourcery.com>
15130 * lto-streamer-out.c (output_function): Fixup loops if required to match
15131 discovery done in the reader.
15133 2018-04-26 Richard Biener <rguenther@suse.de>
15135 PR tree-optimization/85116
15136 * tree-ssa-loop-ch.c (do_while_loop_p): A do-while loop should
15137 have a loop exit from the single latch predecessor. Remove
15138 case of header with just condition.
15139 (ch_base::copy_headers): Exclude infinite loops from any
15141 (pass_ch::execute): Record exits.
15143 2018-04-26 Richard Biener <rguenther@suse.de>
15145 * tree-vect-data-refs.c (vect_get_data_access_cost): Get
15146 prologue cost vector and pass it to vect_get_load_cost.
15147 (vect_get_peeling_costs_all_drs): Likewise.
15148 (vect_peeling_hash_get_lowest_cost): Likewise.
15149 (vect_enhance_data_refs_alignment): Likewise.
15151 2018-04-26 Richard Biener <rguenther@suse.de>
15153 PR middle-end/85450
15154 * tree-cfg.c (verify_gimple_assign_unary): Restore proper
15155 checking of integer<->pointer conversions.
15156 * omp-expand.c (expand_omp_for_static_nochunk): Avoid
15157 sign-/zero-extending pointer types.
15158 (expand_omp_for_static_chunk): Likewise.
15160 2018-03-22 Hans-Peter Nilsson <hp@axis.com>
15161 Jean Lee <xiaoyur347@gmail.com>
15163 * config/mips/mips.c (mips_asan_shadow_offset): New function.
15164 (TARGET_ASAN_SHADOW_OFFSET): Define.
15165 * config/mips/mips.h (FRAME_GROWS_DOWNWARD): Augment to also be
15166 true for -fsanitize=address.
15168 2018-04-25 Mark Wielaard <mark@klomp.org>
15170 * dwarf2out.c (file_info_cmp): Sort longer dir prefixes before
15173 2018-04-25 Jakub Jelinek <jakub@redhat.com>
15175 * config/i386/i386.md (*x86_mov<mode>cc_0_m1): Use type "alu1" rather
15176 than "alu", remove explicit "memory" and "imm_disp" attributes.
15177 (*x86_mov<mode>cc_0_m1_se, *x86_mov<mode>cc_0_m1_neg): Likewise.
15179 PR middle-end/85414
15180 * simplify-rtx.c (simplify_unary_operation_1) <case SIGN_EXTEND,
15181 case ZERO_EXTEND>: Pass SUBREG_REG (op) rather than op to
15182 gen_lowpart_no_emit.
15184 2018-04-25 Sebastian Peryt <sebastian.peryt@intel.com>
15187 * config/i386/i386.c (ix86_expand_builtin): Change memory
15188 operand to XI, extend p0 to Pmode.
15189 * config/i386/i386.md: Change unspec volatile and operand
15190 1 mode to XI, change operand 0 mode to P.
15192 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
15194 * config/nds32/nds32-predicates.c (nds32_can_use_bclr_p): Mask with
15195 GET_MODE_MASK before any checking.
15196 (nds32_can_use_bset_p): Likewise.
15197 (nds32_can_use_btgl_p): Likewise.
15199 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
15201 * config/nds32/nds32-doubleword.md: New define_split pattern for
15202 illegal register number.
15204 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
15206 * config/nds32/nds32.c (nds32_print_operand): Set op_value ealier.
15208 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
15210 * config/nds32/nds32.h (ASM_APP_ON): Add missing newline character.
15212 2018-04-25 Richard Biener <rguenther@suse.de>
15214 * lto-streamer.h (LTO_major_version): Bump to 8.
15216 2018-04-25 Jakub Jelinek <jakub@redhat.com>
15218 * BASE-VER: Set to 9.0.0.
15220 2018-04-24 Segher Boessenkool <segher@kernel.crashing.org>
15222 * config/rs6000/rs6000.c (init_float128_ieee): Fix spelling mistakes
15223 in __abskf2 and __powikf2.
15225 2018-04-24 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15228 * config/aarch64/constraints.md (Usg, Usj): New constraints.
15229 * config/aarch64/iterators.md (cmode_simd): New mode attribute.
15230 * config/aarch64/aarch64.md (*aarch64_ashr_sisd_or_int_<mode>3):
15231 Use the above on operand 2. Reindent.
15232 (*aarch64_lshr_sisd_or_int_<mode>3): Likewise.
15234 2018-04-24 H.J. Lu <hongjiu.lu@intel.com>
15237 * common/config/i386/i386-common.c (ix86_handle_option): Don't
15239 * config/i386/i386.opt (mcet): Removed.
15240 * doc/install.texi: Remove -mcet documentation.
15241 * doc/invoke.texi: Likewise.
15243 2018-04-24 H.J. Lu <hongjiu.lu@intel.com>
15246 * doc/install.texi: Remove -mcet from bootstrap-cet.
15248 2018-04-24 Jakub Jelinek <jakub@redhat.com>
15251 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Don't define
15252 __builtin_ia32_readeflags_u32 and __builtin_ia32_writeeflags_u32
15256 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Punt if
15257 const_vector is not CONST_VECTOR or SYMBOL_REF for a constant pool
15258 containing a CONST_VECTOR.
15260 2018-04-24 Cesar Philippidis <cesar@codesourcery.com>
15262 * doc/install.texi: Update newlib dependency for nvptx.
15264 2018-04-24 Jakub Jelinek <jakub@redhat.com>
15267 * config/i386/i386.c (ix86_expand_vector_init_one_var): Use UINTVAL
15268 instead of INTVAL when shifting x left.
15270 2018-04-24 Andreas Krebbel <krebbel@linux.ibm.com>
15272 PR tree-optimization/85478
15273 * tree-vect-loop.c (vect_analyze_loop_2): Do not call
15274 vect_grouped_store_supported for single element vectors.
15276 2018-04-24 Richard Biener <rguenther@suse.de>
15279 * config/i386/i386.c (ix86_add_stmt_cost): Restrict strided
15280 load cost increase to the case of non-constant step.
15282 2018-04-24 Jakub Jelinek <jakub@redhat.com>
15285 * reg-stack.c (move_for_stack_reg): Don't fail assertion about dead
15286 destination if any_malformed_asm.
15288 2018-04-23 Eric Botcazou <ebotcazou@adacore.com>
15290 PR middle-end/85496
15291 * expr.c (store_field): In the bitfield case, if the value comes from
15292 a function call and is returned in registers by means of a PARALLEL,
15293 do not change the mode of the temporary unless BLKmode and VOIDmode.
15295 2018-04-23 Andrey Belevantsev <abel@ispras.ru>
15297 PR rtl-optimization/85423
15298 * sel-sched-ir.c (has_dependence_note_mem_dep): Only discard
15299 dependencies to debug insns when the previous insn is non-debug.
15301 2018-04-23 Claudiu Zissulescu <claziss@synopsys.com>
15303 * config/arc/arc.md ("vunspec"): Delete it, unify all the unspec
15304 enums into a single definition.
15305 (fls): Fix predicates and printing.
15308 2018-04-23 Claudiu Zissulescu <claziss@synopsys.com>
15310 * config/arc/arc-protos.h (check_if_valid_sleep_operand): Remove.
15311 * config/arc/arc.c (arc_expand_builtin): Sleep accepts registers
15312 and short u6 immediate.
15313 (check_if_valid_sleep_operand): Remove.
15314 * config/arc/arc.md (Sleep): Accepts registers and u6 immediates.
15316 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
15318 * config/nds32/nds32.c (nds32_compute_stack_frame): Consider
15319 flag_always_save_lp condition.
15320 * config/nds32/nds32.opt (malways-save-lp): New option.
15322 2018-04-22 Shiva Chen <shiva0217@gmail.com>
15324 * config/nds32/nds32-protos.h (nds32_use_load_post_increment): Declare.
15325 * config/nds32/nds32.c (nds32_use_load_post_increment): New.
15326 * config/nds32/nds32.h
15327 (USE_LOAD_POST_INCREMENT, USE_LOAD_POST_DECREMENT): Define.
15328 (USE_STORE_POST_INCREMENT, USE_STORE_POST_DECREMENT): Define.
15330 2018-04-22 Shiva Chen <shiva0217@gmail.com>
15332 * config/nds32/nds32-protos.h (nds32_ls_333_p): Remove.
15333 * config/nds32/nds32.c (nds32_ls_333_p): Remove.
15335 2018-04-22 Shiva Chen <shiva0217@gmail.com>
15336 Chung-Ju Wu <jasonwucj@gmail.com>
15338 * config/nds32/nds32-protos.h (nds32_case_vector_shorten_mode):
15340 * config/nds32/nds32.c (nds32_case_vector_shorten_mode): New function.
15341 * config/nds32/nds32.h (CASE_VECTOR_SHORTEN_MODE): Modify.
15343 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
15345 * config/nds32/nds32.c (nds32_compute_stack_frame): Fix wrong value.
15347 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
15349 * config/nds32/nds32-protos.h (nds32_data_alignment,
15350 nds32_local_alignment): Declare.
15351 * config/nds32/nds32.c (nds32_data_alignment, nds32_constant_alignment,
15352 nds32_local_alignment): New functions.
15353 (TARGET_CONSTANT_ALIGNMENT): Define.
15354 * config/nds32/nds32.h (DATA_ALIGNMENT, LOCAL_ALIGNMENT): Define.
15356 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
15358 * config/nds32/nds32.c
15359 (TARGET_HARD_REGNO_MODE_OK): Move to the bottom of file.
15360 (TARGET_MODES_TIEABLE_P): Likewise.
15362 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
15364 * config/nds32/nds32.c (nds32_asm_file_start): Display optimization
15365 level Ofast and Og.
15367 2018-04-22 Monk Chiang <sh.chiang04@gmail.com>
15368 Chung-Ju Wu <jasonwucj@gmail.com>
15370 * config/nds32/constants.md (unspec_volatile_element): Add enum values
15371 for unaligned access.
15372 * config/nds32/nds32-intrinsic.c: Implementation of expanding
15374 * config/nds32/nds32-intrinsic.md: Likewise.
15375 * config/nds32/nds32_intrinsic.h: Likewise.
15376 * config/nds32/nds32.h (nds32_builtins): Likewise.
15377 * config/nds32/nds32.opt (munaligned-access): New option.
15378 * config/nds32/nds32.c (nds32_asm_file_start): Display
15379 flag_unaligned_access status.
15381 2018-04-20 Kito Cheng <kito.cheng@gmail.com>
15383 * config/riscv/elf.h (LINK_SPEC): Pass --no-relax if
15384 -mno-relax is present.
15385 * config/riscv/linux.h (LINK_SPEC): Ditto.
15387 2018-04-20 Martin Sebor <msebor@redhat.com>
15390 * gimple-fold.c (gimple_fold_builtin_strcpy): Suppress -Wrestrict
15392 (gimple_fold_builtin_stxcpy_chk): Same.
15393 * gimple-ssa-warn-restrict.c (check_bounds_or_overlap): Same.
15395 2018-04-20 Michael Meissner <meissner@linux.ibm.com>
15398 * config/rs6000/rs6000.c (init_float128_ieee): Add support to call
15399 __powikf2 when long double is IEEE 128-bit.
15401 2018-04-20 Kito Cheng <kito.cheng@gmail.com>
15403 * config/riscv/riscv.c (riscv_first_stack_step): Round up min
15404 step to make sure stack always aligned.
15406 2018-04-20 Carl Love <cel@us.ibm.com>
15409 * config/rs6000/rs6000-c.c (rs6000_gimple_fold_builtin): Add
15410 size check for arg0.
15412 2018-04-20 Nathan Sidwell <nathan@codesourcery.com>
15413 Tom de Vries <tom@codesourcery.com>
15416 * config/nvptx/nvptx.c (nvptx_emit_forking, nvptx_emit_joining):
15417 Emit insns for calls too.
15418 (nvptx_find_par): Always look for worker-level predecessor insn.
15419 (nvptx_propagate): Add is_call parm, return bool. Copy frame for
15421 (nvptx_vpropagate, nvptx_wpropagate): Adjust.
15422 (nvptx_process_pars): Propagate frames for calls.
15424 2018-04-20 H.J. Lu <hongjiu.lu@intel.com>
15427 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET):
15429 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
15430 (ix86_handle_option): Don't handle OPT_mibt.
15431 * config/i386/cet.h: Check __CET__ instead of __IBT__ and
15433 * config/i386/driver-i386.c (host_detect_local_cpu): Remove
15435 * config/i386/i386-c.c (ix86_target_macros_internal): Don't
15436 check OPTION_MASK_ISA_IBT nor flag_cf_protection.
15437 (ix86_target_macros): Define __CET__ with flag_cf_protection
15438 for -fcf-protection.
15439 * config/i386/i386.c (isa2_opts): Remove -mibt.
15440 * config/i386/i386.h (TARGET_IBT): Removed.
15441 (TARGET_IBT_P): Likewise.
15442 (ix86_valid_target_attribute_inner_p): Don't check OPT_mibt.
15443 * config/i386/i386.md (nop_endbr): Don't check TARGET_IBT.
15444 * config/i386/i386.opt (mcet): Update help message.
15445 (mshstk): Likewise.
15447 * doc/invoke.texi: Remove -mibt. Document __CET__. Document
15448 -mcet as an alias for -mshstk.
15450 2018-04-20 Richard Biener <rguenther@suse.de>
15452 PR middle-end/85475
15453 * match.pd ((X * CST) * Y -> (X * Y) * CST): Avoid exponential
15454 complexity by forcing a single use of the multiply operand.
15456 2018-04-20 Martin Jambor <mjambor@suse.cz>
15459 * ipa-cp.c (cgraph_edge_brings_value_p): Move check for self-feeding
15460 recursion dependency to only apply to non-clones.
15462 2018-04-20 Martin Jambor <mjambor@suse.cz>
15465 * ipa-cp.c (create_specialized_node): Check that clones of
15466 self-recursive edges exist during IPA-CP.
15468 2018-04-19 Toon Moene <toon@moene.org>
15470 * doc/invoke.texi: Add -floop-unroll-and-jam to options enabled
15473 2018-04-19 Jakub Jelinek <jakub@redhat.com>
15475 PR tree-optimization/85467
15476 * fold-const.c (fold_ternary_loc) <case BIT_FIELD_REF>: Use
15477 VECTOR_TYPE_P macro. If type is vector type, VIEW_CONVERT_EXPR the
15478 VECTOR_CST element to type.
15480 2018-04-19 H.J. Lu <hongjiu.lu@intel.com>
15483 * config/i386/i386.h (STACK_SAVEAREA_MODE): New.
15484 * config/i386/i386.md (builtin_setjmp_setup): Removed.
15485 (builtin_longjmp): Likewise.
15486 (save_stack_nonlocal): New pattern.
15487 (restore_stack_nonlocal): Likewise.
15489 2018-04-19 H.J. Lu <hongjiu.lu@intel.com>
15492 * config/i386/cet.c (file_end_indicate_exec_stack_and_cet):
15493 Replace ASM_OUTPUT_LABEL with fprintf.
15495 2018-04-19 H.J. Lu <hongjiu.lu@intel.com>
15498 * config/i386/cet.c (file_end_indicate_exec_stack_and_cet):
15499 Check flag_cf_protection instead of TARGET_IBT and TARGET_SHSTK.
15500 * config/i386/i386-c.c (ix86_target_macros_internal): Also
15501 define __IBT__ and __SHSTK__ for -fcf-protection.
15502 * config/i386/i386.c (pass_insert_endbranch::gate): Don't check
15504 (ix86_trampoline_init): Likewise.
15505 (x86_output_mi_thunk): Likewise.
15506 (ix86_notrack_prefixed_insn_p): Likewise.
15507 (ix86_option_override_internal): Don't disallow -fcf-protection.
15508 * config/i386/i386.md (rdssp<mode>): Also enable for
15510 (incssp<mode>): Likewise.
15511 (nop_endbr): Likewise.
15512 * config/i386/i386.opt (mcet): Change help message to built-in
15515 (mshstk): Likewise.
15516 * doc/invoke.texi: Remove -mcet, -mibt and -mshstk condition
15517 on -fcf-protection. Change -mcet, -mibt and -mshstk to only
15518 enable CET built-in functions.
15520 2018-04-19 Sebastian Peryt <sebastian.peryt@intel.com>
15522 * common/config/i386/i386-common.c
15523 (OPTION_MASK_ISA_MOVDIRI_SET, OPTION_MASK_ISA_MOVDIR64B_SET,
15524 OPTION_MASK_ISA_MOVDIRI_UNSET,
15525 OPTION_MASK_ISA_MOVDIR64B_UNSET): New defines.
15526 (ix86_handle_option): Handle -mmovdiri and -mmovdir64b.
15527 * config.gcc (movdirintrin.h): New header.
15528 * config/i386/cpuid.h (bit_MOVDIRI,
15529 bit_MOVDIR64B): New bits.
15530 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mmovdiri
15532 * config/i386/i386-builtin-types.def ((VOID, PUNSIGNED, UNSIGNED),
15533 (VOID, PVOID, PCVOID)): New function types.
15534 * config/i386/i386-builtin.def (__builtin_ia32_directstoreu_u32,
15535 __builtin_ia32_directstoreu_u64,
15536 __builtin_ia32_movdir64b): New builtins.
15537 * config/i386/i386-c.c (__MOVDIRI__, __MOVDIR64B__): New.
15538 * config/i386/i386.c (ix86_target_string): Added -mmovdir64b
15540 (ix86_valid_target_attribute_inner_p): Ditto.
15541 (ix86_expand_special_args_builtin): Added VOID_FTYPE_PUNSIGNED_UNSIGNED
15542 and VOID_FTYPE_PUNSIGNED_UNSIGNED.
15543 (ix86_expand_builtin): Expand IX86_BUILTIN_MOVDIR64B.
15544 * config/i386/i386.h (TARGET_MOVDIRI, TARGET_MOVDIRI_P,
15545 TARGET_MOVDIR64B, TARGET_MOVDIR64B_P): New.
15546 * config/i386/i386.md (UNSPECV_MOVDIRI, UNSPECV_MOVDIR64B): New.
15547 (movdiri<mode>, movdir64b_<mode>): New.
15548 * config/i386/i386.opt: Add -mmovdiri and -mmovdir64b.
15549 * config/i386/immintrin.h: Include movdirintrin.h.
15550 * config/i386/movdirintrin.h: New file.
15551 * doc/invoke.texi: Added -mmovdiri and -mmovdir64b.
15553 2018-04-19 Richard Biener <rguenther@suse.de>
15555 PR middle-end/85455
15556 * cfg.c (clear_bb_flags): When loop state says we have
15557 marked irreducible regions also preserve BB_IRREDUCIBLE_LOOP.
15559 2018-04-19 Richard Biener <rguenther@suse.de>
15561 PR tree-optimization/84737
15562 * tree-vect-data-refs.c (vect_copy_ref_info): New function
15563 copying restrict info.
15564 (vect_setup_realignment): Use it.
15565 * tree-vectorizer.h (vect_copy_ref_info): Declare.
15566 * tree-vect-stmts.c (vectorizable_store): Copy ref info from
15567 the first DR to all generated stores.
15568 (vectorizable_load): Likewise for loads.
15570 2018-04-19 Jakub Jelinek <jakub@redhat.com>
15572 PR tree-optimization/85446
15573 * match.pd ((intptr_t) x eq/ne CST to x eq/ne (typeof x) cst): Require
15574 the integral and pointer types to have the same precision.
15576 * doc/install.texi: Document --disable-cet being the default and
15579 2018-04-18 Martin Liska <mliska@suse.cz>
15581 * ipa-devirt.c (odr_subtypes_equivalent_p): Fix GNU coding
15584 2018-04-18 Martin Liska <mliska@suse.cz>
15587 2018-03-02 Eric Botcazou <ebotcazou@adacore.com>
15590 * ipa-devirt.c (odr_subtypes_equivalent_p): Get the ODR type of both
15591 arguments if they are comparable.
15593 2018-04-18 Martin Liska <mliska@suse.cz>
15596 2018-03-13 Eric Botcazou <ebotcazou@adacore.com>
15599 * ipa-devirt.c (odr_subtypes_equivalent_p): Do not get the ODR type of
15602 2018-04-18 H.J. Lu <hongjiu.lu@intel.com>
15605 * config/i386/i386.c (ix86_expand_split_stack_prologue): Insert
15606 ENDBR after calling __morestack.
15608 2018-04-18 David Malcolm <dmalcolm@redhat.com>
15611 * configure.ac (gcc-driver-name.h): Honor --with-gcc-major-version
15612 by using gcc_base_ver to generate a gcc_driver_version, and use
15613 it when generating GCC_DRIVER_NAME.
15614 * configure: Regenerate.
15616 2018-04-18 Jakub Jelinek <jakub@redhat.com>
15619 * config.gcc: Obsolete powerpc*-*-*spe*.
15621 2018-04-17 Jakub Jelinek <jakub@redhat.com>
15624 * dbxout.c (dbxout_int): Perform negation in unsigned int type.
15625 (stabstr_D): Change type of unum from unsigned int to
15626 unsigned HOST_WIDE_INT. Perform negation in unsigned HOST_WIDE_INT
15629 2018-04-17 Jim Wilson <jimw@sifive.com>
15632 * config/riscv/riscv.c (riscv_compute_frame_info): Add calls to
15633 RISCV_STACK_ALIGN when using outgoing_args_size and pretend_args_size.
15634 Set arg_pointer_offset after using pretend_args_size.
15636 2018-04-17 Jakub Jelinek <jakub@redhat.com>
15638 PR rtl-optimization/85431
15639 * dse.c (record_store): Ignore zero width stores.
15642 * asan.c (handle_builtin_stack_restore): Adjust comment. Emit
15643 __asan_allocas_unpoison call and last_alloca_addr = new_sp before
15644 __builtin_stack_restore rather than after it.
15645 * builtins.c (expand_asan_emit_allocas_unpoison): Pass
15646 arg1 + (virtual_dynamic_stack_rtx - stack_pointer_rtx) as second
15647 argument instead of virtual_dynamic_stack_rtx.
15649 2018-04-17 Kelvin Nilsen <kelvin@gcc.gnu.org>
15651 * config/rs6000/rs6000-protos.h (rs6000_builtin_is_supported_p):
15653 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
15654 Add note to error message to explain internal mapping of overloaded
15655 built-in function name to non-overloaded built-in function name.
15656 * config/rs6000/rs6000.c (rs6000_builtin_is_supported_p): New
15659 2018-04-17 Michael Meissner <meissner@linux.vnet.ibm.com>
15662 * config/rs6000/rs6000.md (pack<mode>): Do not try handle a pack
15663 where the inputs overlap with the output.
15665 2018-04-17 Jakub Jelinek <jakub@redhat.com>
15667 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): Add
15668 (=v, v) alternative and explicit "memory" attribute.
15669 (vec_extract_lo_<mode><mask_name>): Likewise. Also add
15670 "type", "prefix", "prefix_extra", "length_immediate" and "mode"
15672 (vec_extract_lo_<mode><mask_name>): Add (=v, v) alternative and use
15673 "sselog1" type instead of "sselog".
15674 (vec_extract_hi_<mode><mask_name>): Use "sselog1" type instead of
15675 "sselog". Remove explicit "memory" attribute.
15676 (vec_extract_lo_v32hi): Add (=v, v) alternative and explicit "memory",
15677 "type", "prefix", "prefix_extra", "length_immediate" and "mode"
15679 (vec_extract_hi_v32hi): Merge all alternatives into one, use
15680 "sselog1" type instead of "sselog". Remove explicit "memory"
15682 (vec_extract_hi_v16hi): Merge each pair of alternatives into one,
15683 use "sselog1" type instead of "sselog". Remove explicit "memory"
15685 (vec_extract_lo_v64qi): Add (=v, v) alternative and explicit "memory",
15686 "type", "prefix", "prefix_extra", "length_immediate" and "mode"
15688 (vec_extract_hi_v64qi): Merge all alternatives into one, use
15689 "sselog1" type instead of "sselog". Remove explicit "memory"
15691 (vec_extract_hi_v32qi): Merge each pair of alternatives into one,
15692 use "sselog1" type instead of "sselog". Remove explicit "memory"
15696 * config/i386/i386.md (*ashlqi3_1_slp): Use alu1 type instead of alu.
15698 PR middle-end/85414
15699 * rtlhooks.c (gen_lowpart_if_possible): Don't call gen_lowpart_SUBREG
15702 2018-04-17 Martin Jambor <mjambor@suse.cz>
15705 * ipa-cp.c (create_specialized_node): Call
15706 expand_all_artificial_thunks if necessary.
15708 2018-04-17 Martin Liska <mliska@suse.cz>
15711 * ipa-devirt.c (odr_types_equivalent_p): Remove trailing
15712 in message, remote space in between '_G' and '('.
15714 2018-04-17 Jakub Jelinek <jakub@redhat.com>
15717 * config/i386/sse.md (reduces<mode><mask_scalar_name>,
15718 avx512f_vmcmp<mode>3<round_saeonly_name>,
15719 avx512f_vmcmp<mode>3_mask<round_saeonly_name>,
15720 avx512f_sgetexp<mode><mask_scalar_name><round_saeonly_scalar_name>,
15721 avx512f_rndscale<mode><round_saeonly_name>,
15722 avx512dq_ranges<mode><mask_scalar_name><round_saeonly_scalar_name>,
15723 avx512f_vgetmant<mode><mask_scalar_name><round_saeonly_scalar_name>):
15724 Use %<iptr>2 instead of %2 for -masm=intel.
15725 (avx512f_vcvtss2usi<round_name>, avx512f_vcvtss2usiq<round_name>,
15726 avx512f_vcvttss2usi<round_saeonly_name>,
15727 avx512f_vcvttss2usiq<round_saeonly_name>): Use %k1 instead of %1 for
15729 (avx512f_vcvtsd2usi<round_name>, avx512f_vcvtsd2usiq<round_name>,
15730 avx512f_vcvttsd2usi<round_saeonly_name>,
15731 avx512f_vcvttsd2usiq<round_saeonly_name>, ufloatv2siv2df2<mask_name>):
15732 Use %q1 instead of %1 for -masm=intel.
15733 (avx512f_sfixupimm<mode><sd_maskz_name><round_saeonly_name>,
15734 avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Use %<iptr>3 instead
15735 of %3 for -masm=intel.
15736 (sse2_shufpd_v2df_mask): Fix a typo, change %{6%} to %{%6%} for
15738 (*avx512vl_<code>v2div2qi2_store): Use %w0 instead of %0 for
15740 (*avx512vl_<code><mode>v4qi2_store): Use %k0 instead of %0 for
15742 (avx512vl_<code><mode>v4qi2_mask_store): Use a single pattern with
15743 %k0 and %1 for -masm=intel rather than two patterns, one with %0 and
15745 (*avx512vl_<code><mode>v8qi2_store): Use %q0 instead of %0 for
15747 (avx512vl_<code><mode>v8qi2_mask_store): Use a single pattern with
15748 %q0 and %1 for -masm=intel rather than two patterns, one with %0 and
15749 %g1 and one with %0 and %1.
15750 (avx512er_vmrcp28<mode><round_saeonly_name>,
15751 avx512er_vmrsqrt28<mode><round_saeonly_name>): Use %<iptr>1 instead of
15752 %1 for -masm=intel.
15753 (avx5124fmaddps_4fmaddps_mask, avx5124fmaddps_4fmaddss_mask,
15754 avx5124fmaddps_4fnmaddps_mask, avx5124fmaddps_4fnmaddss_mask,
15755 avx5124vnniw_vp4dpwssd_mask, avx5124vnniw_vp4dpwssds_mask): Swap order
15756 of %0 and %{%4%} for -masm=intel.
15757 (avx5124fmaddps_4fmaddps_maskz, avx5124fmaddps_4fmaddss_maskz,
15758 avx5124fmaddps_4fnmaddps_maskz, avx5124fmaddps_4fnmaddss_maskz,
15759 avx5124vnniw_vp4dpwssd_maskz, avx5124vnniw_vp4dpwssds_maskz): Swap
15760 order of %0 and %{%5%}%{z%} for -masm=intel.
15762 2018-04-17 Jan Hubicka <jh@suse.cz>
15765 * ipa-devirt.c (odr_types_equivalent_p): Handle bit fields.
15767 2018-04-17 Martin Liska <mliska@suse.cz>
15770 * multiple_target.c (create_dispatcher_calls): Set apostrophes
15771 for target_clone error message. Make default implementation
15772 clone to be a local declaration.
15773 (separate_attrs): Add new argument and check for an empty
15775 (expand_target_clones): Handle it.
15776 (ipa_target_clone): Make redirection just for target_clones
15779 2018-04-16 Cesar Philippidis <cesar@codesourcery.com>
15780 Tom de Vries <tom@codesourcery.com>
15782 PR middle-end/84955
15783 * omp-expand.c (expand_oacc_for): Add dummy false branch for
15784 tiled basic blocks without omp continue statements.
15786 2018-04-16 Aaron Sawdey <acsawdey@linux.ibm.com>
15789 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Mark
15790 vec_extract expression as having side effects to make sure it gets
15793 2018-04-16 H.J. Lu <hongjiu.lu@intel.com>
15796 * config/i386/i386.c (get_builtin_code_for_version): Check
15799 2018-04-16 Olga Makhotina <olga.makhotina@intel.com>
15802 * config.gcc: Support "skylake".
15803 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
15805 * config/i386/i386.c (m_SKYLAKE): Define.
15806 (processor_target_table): Add "skylake".
15807 (ix86_option_override_internal): Add "skylake".
15808 (get_builtin_code_for_version): Handle PROCESSOR_SKYLAKE,
15809 PROCESSOR_CANNONLAKE.
15810 (get_builtin_code_for_version): Fix priority for
15811 PROCESSOR_ICELAKE_CLIENT, PROCESSOR_ICELAKE_SERVER,
15812 PROCESSOR_SKYLAKE-AVX512.
15813 * config/i386/i386.h (processor_costs): Define TARGET_SKYLAKE.
15814 (processor_type): Add PROCESSOR_SKYLAKE.
15816 2018-04-16 Paolo Carlini <paolo.carlini@oracle.com>
15817 Jason Merrill <jason@redhat.com>
15820 * convert.c (convert_to_integer_1): Use direct recursion for
15821 enumeral types and types with a precision less than the number
15822 of bits in their mode.
15824 2018-04-16 Julia Koval <julia.koval@intel.com>
15827 * config/i386/x86-tune.def (X86_TUNE_SSE_UNALIGNED_LOAD_OPTIMAL,
15828 X86_TUNE_SSE_UNALIGNED_STORE_OPTIMAL): Add m_SKYLAKE_AVX512
15830 2018-04-14 Segher Boessenkool <segher@kernel.crashing.org>
15833 * config/rs6000/rs6000.opt (mdirect-move): Make deprecated.
15834 * doc/invoke.texi (RS/6000 and PowerPC Options): Remove -mdirect-move
15835 and -mno-direct-move.
15837 2018-04-13 Paul A. Clarke <pc@us.ibm.com>
15840 * config/rs6000/emmintrin.h (_mm_slli_epi{16,32,64}):
15841 Ensure that vec_splat_s32 is only called with 0 <= shift < 16.
15842 Ensure negative shifts result in {0}.
15844 2018-04-13 Vladimir Makarov <vmakarov@redhat.com>
15846 PR rtl-optimization/79916
15847 * config/rs6000/rs6000.c (rs6000_emit_move): Use assigned hard
15848 regs (if any) to define how to gnerate SD moves when LRA is in
15851 2018-04-13 Jakub Jelinek <jakub@redhat.com>
15853 PR rtl-optimization/85393
15854 * except.h (expand_dw2_landing_pad_for_region): Remove declaration.
15855 * except.c (expand_dw2_landing_pad_for_region): Make static.
15856 * bb-reorder.c (fix_up_crossing_landing_pad): In new_bb emit just
15857 a label and unconditional jump to old_bb, rather than
15858 expand_dw2_landing_pad_for_region insn(s) and jump to single_succ
15861 PR rtl-optimization/85376
15862 * simplify-rtx.c (simplify_const_unary_operation): For CLZ and CTZ and
15863 zero op0, if C?Z_DEFINED_VALUE_AT_ZERO is false, return NULL_RTX
15864 instead of a specific value.
15866 2018-04-13 Jan Hubicka <hubicka@ucw.cz>
15867 Bin Cheng <bin.cheng@arm.com>
15869 PR tree-optimization/82965
15870 PR tree-optimization/83991
15871 * cfgloopanal.c (expected_loop_iterations_unbounded): Add
15872 by_profile_only parameter.
15873 * cfgloopmanip.c (scale_loop_profile): Further scale loop's profile
15874 information if the loop was predicted to iterate too many times.
15875 * cfgloop.h (expected_loop_iterations_unbounded): Update prototype
15877 2018-04-13 Jan Hubicka <hubicka@ucw.cz>
15880 * config/i386/i386.c (ix86_can_inline_p): Allow safe transitions for
15883 2018-04-13 Martin Liska <mliska@suse.cz>
15884 Jakub Jelinek <jakub@redhat.com>
15886 PR middle-end/81657
15887 * expr.h (enum block_op_methods): Add BLOCK_OP_NO_LIBCALL_RET.
15888 * expr.c (emit_block_move_hints): Handle BLOCK_OP_NO_LIBCALL_RET.
15889 * builtins.c (expand_builtin_memory_copy_args): Use
15890 BLOCK_OP_NO_LIBCALL_RET method for mempcpy with non-ignored target,
15891 handle dest_addr == pc_rtx.
15893 2018-04-12 Segher Boessenkool <segher@kernel.crashing.org>
15896 * config/rs6000/rs6000.md (fix_trunc<mode>si2): Use legacy code if
15897 asked to not generate direct moves.
15898 (fix_trunc<mode>si2_stfiwx): Similar.
15899 (fix_trunc<mode>si2_internal): Similar.
15901 2018-04-12 Jakub Jelinek <jakub@redhat.com>
15904 * var-tracking.c (add_stores): Handle STRICT_LOW_PART SET_DEST.
15905 * cselib.c (cselib_record_sets): For STRICT_LOW_PART dest,
15906 lookup if dest in some wider mode is known to be const0_rtx and
15907 if so, record permanent equivalence for it to be ZERO_EXTEND of
15908 the narrower mode destination.
15910 2018-04-12 Cesar Philippidis <cesar@codesourcery.com>
15912 * lto-streamer-out.c (output_function): Revert 259346.
15913 * omp-expand.c (expand_oacc_for): Likewise.
15915 2018-04-12 Alexander Monakov <amonakov@ispras.ru>
15917 PR rtl-optimization/85354
15918 * sel-sched-ir.c (sel_init_pipelining): Move cfg_cleanup call...
15919 * sel-sched.c (sel_global_init): ... here.
15921 2018-04-12 Eric Botcazou <ebotcazou@adacore.com>
15924 * lto-wrapper.c (debug_objcopy): Open the files in binary mode.
15925 * dwarf2out.c (dwarf2out_early_finish): Do not generate assembly in LTO
15926 mode for PE-COFF targets.
15927 * config/i386/i386-protos.h (i386_pe_asm_lto_start): Declare.
15928 (i386_pe_asm_lto_end): Likewise.
15929 * config/i386/cygming.h (TARGET_ASM_LTO_START): Define.
15930 (TARGET_ASM_LTO_END): Likewise.
15931 * config/i386/winnt.c (saved_debug_info_level): New static variable.
15932 (i386_pe_asm_lto_start): New function.
15933 (i386_pe_asm_lto_end): Likewise.
15935 2018-04-12 Cesar Philippidis <cesar@codesourcery.com>
15936 Richard Biener <rguenther@suse.de>
15938 PR middle-end/84955
15939 * lto-streamer-out.c (output_function): Fix CFG loop state before
15941 * omp-expand.c (expand_oacc_for): Handle calls to internal
15942 functions like regular functions.
15944 2018-04-12 Richard Biener <rguenther@suse.de>
15947 * dwarf2out.c (init_sections_and_labels): Use debug_line_section[_label]
15948 for the early LTO debug to properly generate references to it
15949 during DIE emission. Do not re-use that for the skeleton for
15951 (dwarf2out_early_finish): Likewise.
15953 2018-04-12 Jakub Jelinek <jakub@redhat.com>
15956 * config/i386/sse.md
15957 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name> split,
15958 <mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name> split,
15959 vec_extract_lo_<mode><mask_name> split, vec_extract_lo_v32hi,
15960 vec_extract_lo_v64qi): For non-AVX512VL if input is xmm16+ reg
15961 and output is a reg, avoid creating invalid lowpart subreg, but
15962 instead split into a 512-bit move. Don't split if not AVX512VL,
15963 input is xmm16+ reg and output is a mem.
15964 (vec_extract_lo_<mode><mask_name>, vec_extract_lo_v32hi,
15965 vec_extract_lo_v64qi): Don't require split if not AVX512VL, input is
15966 xmm16+ reg and output is a mem.
15968 2018-04-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15970 * config/s390/s390.c (s390_output_indirect_thunk_function): Check
15971 also for flag_dwarf2_cfi_asm.
15973 2018-04-12 Jakub Jelinek <jakub@redhat.com>
15975 PR rtl-optimization/85342
15976 * regcprop.c (copyprop_hardreg_forward_1): Remove replaced array, use
15977 a bool scalar var inside of the loop instead. Don't try to update
15978 recog_data.operand after failed apply_change_group.
15980 2018-04-12 Tom de Vries <tom@codesourcery.com>
15983 * config/nvptx/nvptx.c (flexible_array_member_type_p): New function.
15984 (nvptx_assemble_decl_begin): Add undefined param. Declare undefined
15985 array with flexible array member as array without given dimension.
15986 (nvptx_assemble_undefined_decl): Set nvptx_assemble_decl_begin call
15987 argument for undefined param to true.
15989 2018-04-11 Aaron Sawdey <acsawdey@linux.ibm.com>
15992 * doc/invoke.texi (RS/6000 and PowerPC Options): Document options
15993 -mcall- and -mtraceback=. Remove options -mabi=spe and -mabi=no-spe
15994 from PowerPC section.
15995 * config/rs6000/sysv4.opt (mcall-): Improve help text.
15996 * config/rs6000/rs6000.opt (mblock-compare-inline-limit=): Trim
15997 help text that is too long.
15998 * config/rs6000/rs6000.opt (mblock-compare-inline-loop-limit=): Trim
15999 help text that is too long.
16000 * config/rs6000/rs6000.opt (mstring-compare-inline-limit=): Trim
16001 help text that is too long.
16003 2018-04-11 Uros Bizjak <ubizjak@gmail.com>
16005 * config/alpha/alpha.md (stack_probe_internal): Rename
16006 from "probe_stack". Update all callers.
16008 2018-04-11 Alexander Monakov <amonakov@ispras.ru>
16010 PR rtl-optimization/84566
16011 * sched-deps.c (sched_analyze_insn): Check deps->readonly when invoking
16012 sched_macro_fuse_insns.
16014 2018-04-11 Alexander Monakov <amonakov@ispras.ru>
16017 * sched-rgn.c (add_branch_dependences): Move sel_sched_p check here...
16018 (compute_block_dependences): ... from here.
16020 2018-04-11 Jakub Jelinek <jakub@redhat.com>
16022 PR tree-optimization/85331
16023 * vec-perm-indices.h (vec_perm_indices::clamp): Change input type
16024 from int to HOST_WIDE_INT.
16026 2018-04-11 Martin Jambor <mjambor@suse.cz>
16029 * ipa-cp.c (propagate_vals_across_pass_through): Expand comment.
16030 (cgraph_edge_brings_value_p): New parameter dest_val, check if it is
16031 not the same as the source val.
16032 (cgraph_edge_brings_value_p): New parameter.
16033 (gather_edges_for_value): Pass destination value to
16034 cgraph_edge_brings_value_p.
16035 (perhaps_add_new_callers): Likewise.
16036 (get_info_about_necessary_edges): Likewise and exclude values brought
16037 only by self-recursive edges.
16038 (create_specialized_node): Redirect only clones of self-calling edges.
16039 (+self_recursive_pass_through_p): New function.
16040 (find_more_scalar_values_for_callers_subset): Use it.
16041 (find_aggregate_values_for_callers_subset): Likewise.
16042 (known_aggs_to_agg_replacement_list): Removed.
16043 (decide_whether_version_node): Re-calculate known constants for all
16044 remaining context clones.
16046 2018-04-11 Richard Biener <rguenther@suse.de>
16049 * dwarf2out.c (dwarf2out_finish): Remove DW_AT_stmt_list attribute
16050 from early DWARF output.
16051 (dwarf2out_early_finish): Output line info unconditionally into
16052 early DWARF and add reference to it.
16054 2018-04-11 Jakub Jelinek <jakub@redhat.com>
16057 * config/i386/sse.md (iptr): Add V16SFmode and V8DFmode cases.
16058 (<avx512>_vec_dup<mode><mask_name>): Use a single pattern for modes
16059 other than V2DFmode using iptr mode attribute.
16060 (<avx512>_vec_dup<mode><mask_name>): Use iptr mode attribute.
16062 2018-04-11 Alexander Monakov <amonakov@ispras.ru>
16064 PR rtl-optimization/84659
16065 * sel-sched-ir.c (sel_init_pipelining): Invoke cleanup_cfg.
16067 2018-04-11 Jakub Jelinek <jakub@redhat.com>
16070 * dwarf2out.c (skip_loc_list_entry): Don't call size_of_locs if
16072 (output_loc_list): Pass address of a dummy size variable even in the
16073 locview handling loop.
16074 (index_location_lists): Add comment on why skip_loc_list_entry can't
16077 2018-04-11 Thomas Preud'homme <thomas.preudhomme@arm.com>
16080 * config/arm/arm-builtins.c (arm_expand_builtin): Force input operand
16083 2018-04-10 Aaron Sawdey <acsawdey@linux.ibm.com>
16086 * doc/invoke.texi (RS/6000 and PowerPC Options): Document options
16087 -mblock-compare-inline-limit, -mblock-compare-inline-loop-limit,
16088 and -mstring-compare-inline-limit.
16090 2018-04-10 Segher Boessenkool <segher@kernel.crashing.org>
16093 * config/rs6000/rs6000.md (allocate_stack): Put the residual size
16094 for stack clash protection in a register whenever we need it to be in
16097 2018-04-10 Segher Boessenkool <segher@kernel.crashing.org>
16099 * common/config/rs6000/rs6000-common.c (rs6000_option_init_struct):
16100 Enable -fasynchronous-unwind-tables by default if OBJECT_FORMAT_ELF.
16102 2018-04-10 Segher Boessenkool <segher@kernel.crashing.org>
16105 * config/rs6000/rs6000.opt (mtraceback=): Show the allowed values in
16107 (mlong-double-): Ditto.
16108 * config/rs6000/sysv4.opt (msdata=): Ditto.
16109 (mtls-size=): Ditto.
16111 2018-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
16113 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
16114 erroneous entries for
16115 "vector int vec_ldl (int, long int *)", and
16116 "vector unsigned int vec_ldl (int, unsigned long int *)".
16117 Add comments and entries for
16118 "vector bool char vec_ldl (int, bool char *)",
16119 "vector bool short vec_ldl (int, bool short *)",
16120 "vector bool int vec_ldl (int, bool int *)",
16121 "vector bool long long vec_ldl (int, bool long long *)",
16122 "vector pixel vec_ldl (int, pixel *)",
16123 "vector long long vec_ldl (int, long long *)",
16124 "vector unsigned long long vec_ldl (int, unsigned long long *)".
16125 * config/rs6000/rs6000.c (rs6000_init_builtins): Initialize new
16126 type tree bool_long_long_type_node and correct definition of
16127 bool_V2DI_type_node to make reference to this new type tree.
16128 (rs6000_mangle_type): Replace erroneous reference to
16129 bool_long_type_node with bool_long_long_type_node.
16130 * config/rs6000/rs6000.h (enum rs6000_builtin_type_index): Add
16131 comments to emphasize sign distinctions for char and int types and
16132 replace RS6000_BTI_bool_long constant with
16133 RS6000_BTI_bool_long_long constant. Also add comment to restrict
16134 use of RS6000_BTI_pixel.
16135 (bool_long_type_node): Remove this macro definition.
16136 (bool_long_long_type_node): New macro definition
16138 2018-04-10 Jakub Jelinek <jakub@redhat.com>
16140 PR rtl-optimization/85300
16141 * combine.c (subst): Handle subst of CONST_SCALAR_INT_P new_rtx also
16142 into FLOAT and UNSIGNED_FLOAT like ZERO_EXTEND, return a CLOBBER if
16143 simplify_unary_operation fails.
16145 2018-04-10 Martin Liska <mliska@suse.cz>
16147 * gdbhooks.py: Add pretty-printers for varpool_node, symtab_node,
16148 cgraph_edge and ipa_ref.
16150 2018-04-10 Jakub Jelinek <jakub@redhat.com>
16154 * config/i386/sse.md
16155 (<extract_type>_vinsert<shuffletype><extract_suf>_mask): Fix
16156 computation of the VEC_MERGE selector from mask.
16157 (<extract_type>_vinsert<shuffletype><extract_suf>_1<mask_name>):
16158 Fix decoding of the VEC_MERGE selector into mask.
16160 2018-04-10 Richard Sandiford <richard.sandiford@linaro.org>
16162 PR tree-optimization/85286
16163 * tree-vect-data-refs.c (vect_get_smallest_scalar_type):
16165 2018-04-10 Richard Sandiford <richard.sandiford@linaro.org>
16167 * final.c (final_1): Set insn_last_address as well as
16168 insn_current_address.
16170 2018-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16173 * explow.c (emit_stack_probe): Call validize_mem on memory location
16174 before passing it to gen_probe_stack. Create address operand and
16175 legitimize it for the probe_stack_address case.
16177 2018-04-09 Jan Hubicka <jh@suse.cz>
16180 * ipa-devirt.c (rebuild_type_inheritance-hash): New.
16181 * ipa-utils.h (rebuild_type_inheritance-hash): Declare.
16182 * tree.c (free_lang_data_in_type): Fix handling of binfos;
16184 (free_lang_data): Rebuild type inheritance graph.
16186 2018-04-09 Martin Sebor <msebor@redhat.com>
16188 * invoke.texi (-finline-small-functions): Mention other optimization
16190 (-findirect-inlining, -fpartial-inlining): Same.
16191 (-finline-functions-called-once): Same.
16192 (-freorder-blocks-and-partition): Same.
16194 2018-04-09 Jan Hubicka <jh@suse.cz>
16197 * cfgcleanup.c (try_forward_edges): Do not give up on crossing
16198 jumps; choose last target that matches the criteria (i.e.
16199 no partition changes for non-crossing jumps).
16200 * cfgrtl.c (cfg_layout_redirect_edge_and_branch): Add basic
16201 support for redirecting crossing jumps to non-crossing.
16203 2018-04-09 Alexey Brodkin <abrodkin@synopsys.com>
16205 * config/arc/arc.c (arc_expand_prologue): Set stack usage info
16206 also for naked functions.
16208 2018-04-09 Claudiu Zissulescu <claziss@synopsys.com>
16210 * config/arc/arc.md (add_shift): New pattern.
16211 (add_shift2): Likewise.
16212 (sub_shift): Likewise.
16213 (sub_shift_cmp0_noout): Likewise.
16214 (compare_si_ashiftsi): Likewise.
16215 (xbfu_cmp0_noout): New combine pattern.
16216 (xbfu_cmp0"): Likewise.
16217 (movsi_set_cc_insn): Place the predicable variant first.
16218 (commutative_binary_cmp0_noout): Remove clobber.
16219 (commutative_binary_cmp0): New pattern.
16220 (noncommutative_binary_cmp0): Likewise.
16221 (noncommutative_binary_cmp0_noout): Likewise.
16222 (noncommutative_binary_comparison_result_used): Removed.
16223 (rsub_cmp0): New pattern.
16224 (rsub_cmp0_noout): Likewise.
16225 (extzvsi): Changed, keep only meaningful variants.
16226 (SQH, SEZ): New iterators.
16227 (SQH_postfix): New mode attribute.
16228 (SEZ_prefix): New code attribute.
16229 (<SEZ_prefix>xt<SQH_postfix>_cmp0_noout): New instruction pattern.
16230 (<SEZ_prefix>xt<SQH_postfix>_cmp0): Likewise.
16231 * config/arc/predicates.md (cc_set_register): Use CC_REG instead
16232 of numerical value.
16233 (noncommutative_operator): Check the availability of barrel
16236 2018-04-09 Richard Biener <rguenther@suse.de>
16238 PR tree-optimization/85284
16239 * tree-ssa-loop-niter.c (number_of_iterations_exit_assumptions):
16240 Only use the niter constraining form of simple_iv when the exit
16241 is always executed.
16243 2018-04-09 Tom de Vries <tom@codesourcery.com>
16246 * config/nvptx/nvptx.md (define_c_enum "unspecv"): Add UNSPECV_MEMBAR.
16247 (define_expand "*memory_barrier"): New define_expand.
16248 (define_insn "memory_barrier"): New insn.
16250 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
16252 PR rtl-optimization/80463
16253 PR rtl-optimization/83972
16254 PR rtl-optimization/83480
16256 * sel-sched-ir.c (has_dependence_note_mem_dep): Take into account the
16257 correct producer for the insn.
16258 (tidy_control_flow): Fixup seqnos in case of debug insns.
16260 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
16262 PR rtl-optimization/83913
16264 * sel-sched-ir.c (merge_expr_data): Choose the middle between two
16265 different sched-times when merging exprs.
16267 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
16269 PR rtl-optimization/83962
16271 * sel-sched-ir.c (tidy_control_flow): Correct the order in which we call
16272 tidy_fallthru_edge and tidy_control_flow.
16274 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
16276 PR rtl-optimization/83530
16278 * sel-sched.c (force_next_insn): New global variable.
16279 (remove_insn_for_debug): When force_next_insn is true, also leave only
16280 next insn in the ready list.
16281 (sel_sched_region): When the region wasn't scheduled, make another pass
16282 over it with force_next_insn set to 1.
16284 2018-04-08 Monk Chiang <sh.chiang04@gmail.com>
16286 * config.gcc (nds32le-*-*, nds32be-*-*): Add nds32/nds32_intrinsic.h
16288 * config/nds32/constants.md (unspec_volatile_element): Add enum values
16289 for interrupt control.
16290 * config/nds32/nds32-intrinsic.c: Implementation of intrinsic
16291 functions for interrupt control.
16292 * config/nds32/nds32-intrinsic.md: Likewise.
16293 * config/nds32/nds32_intrinsic.h: Likewise.
16294 * config/nds32/nds32.h (nds32_builtins): Likewise.
16296 2018-04-08 Chung-Ju Wu <jasonwucj@gmail.com>
16298 * config/nds32/nds32.c (nds32_init_machine_status,
16299 nds32_legitimate_index_p, nds32_legitimate_address_p): Consider
16300 strict_aligned_p field.
16301 (nds32_expand_to_rtl_hook): New function.
16302 (TARGET_EXPAND_TO_RTL_HOOK): Define.
16303 * config/nds32/nds32.h (machine_function): Add strict_aligned_p field.
16305 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
16306 Chung-Ju Wu <jasonwucj@gmail.com>
16308 * config.gcc (nds32*-*-*): Check that n7 is valid to --with-cpu.
16309 * config/nds32/nds32-n7.md: New file.
16310 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N7.
16311 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n7
16313 * config/nds32/nds32-protos.h: More declarations for n7 pipeline.
16314 * config/nds32/nds32.md (pipeline_model): Add n7.
16315 * config/nds32/nds32.opt (mcpu): Support n7 pipeline cpus.
16316 * config/nds32/pipelines.md: Include n7 settings.
16318 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
16319 Chung-Ju Wu <jasonwucj@gmail.com>
16321 * config.gcc (nds32*-*-*): Check that e8 is valid to --with-cpu.
16322 * config/nds32/nds32-e8.md: New file.
16323 * config/nds32/nds32-opts.h (nds32-cpu_type): Add CPU_E8.
16324 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for e8
16326 * config/nds32/nds32-protos.h: More declarations for e8 pipeline.
16327 * config/nds32/nds32.md (pipeline_model): Add e8.
16328 * config/nds32/nds32.opt (mcpu): Support e8 pipeline cpus.
16329 * config/nds32/pipelines.md: Include e8 settings.
16331 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
16332 Chung-Ju Wu <jasonwucj@gmail.com>
16334 * config.gcc (nds32*-*-*): Check that n6/n8/s8 are valid to --with-cpu.
16335 * config/nds32/nds32-n8.md: New file.
16336 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N6 and CPU_N8.
16337 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n8
16339 * config/nds32/nds32-protos.h: More declarations for n8 pipeline.
16340 * config/nds32/nds32-utils.c: More implementations for n8 pipeline.
16341 * config/nds32/nds32.md (pipeline_model): Add n8.
16342 * config/nds32/nds32.opt (mcpu): Support n8 pipeline cpus.
16343 * config/nds32/pipelines.md: Include n8 settings.
16345 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
16346 Chung-Ju Wu <jasonwucj@gmail.com>
16348 * config.gcc (nds32*): Add nds32-utils.o into extra_objs.
16349 * config/nds32/nds32-n9-2r1w.md: New file.
16350 * config/nds32/nds32-n9-3r2w.md: New file.
16351 * config/nds32/nds32-opts.h (nds32_cpu_type, nds32_mul_type,
16352 nds32_register_ports): New or modify for cpu n9.
16353 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n9
16355 * config/nds32/nds32-protos.h: More declarations for n9 pipeline.
16356 * config/nds32/nds32-utils.c: New file.
16357 * config/nds32/nds32.h (TARGET_PIPELINE_N9, TARGET_PIPELINE_SIMPLE,
16358 TARGET_MUL_SLOW): Define.
16359 * config/nds32/nds32.md (pipeline_model): New attribute.
16360 * config/nds32/nds32.opt (mcpu, mconfig-mul, mconfig-register-ports):
16361 New options that support cpu n9.
16362 * config/nds32/pipelines.md: Include n9 settings.
16363 * config/nds32/t-nds32 (nds32-utils.o): Add dependency.
16365 2018-04-08 Chung-Ju Wu <jasonwucj@gmail.com>
16367 * config/nds32/nds32-md-auxiliary.c (output_cond_branch): Output align
16368 information if necessary.
16369 (output_cond_branch_compare_zero): Likewise.
16370 * config/nds32/nds32.c (nds32_adjust_insn_length): Consider align case.
16371 (nds32_target_alignment): Refine for alignment.
16372 * config/nds32/nds32.h (NDS32_ALIGN_P): Define.
16373 (FUNCTION_BOUNDARY): Modify.
16374 * config/nds32/nds32.md (call_internal, call_value_internal): Consider
16376 * config/nds32/nds32.opt (malways-align, malign-functions): New.
16378 2018-04-08 Monk Chiang <sh.chiang04@gmail.com>
16380 * config/nds32/constants.md (unspec_volatile_element): Add values for
16381 TLB operation and data prefetch.
16382 * config/nds32/nds32-intrinsic.c: Implementation of intrinsic
16383 functions for TLB operation and data prefetch.
16384 * config/nds32/nds32-intrinsic.md: Likewise.
16385 * config/nds32/nds32_intrinsic.h: Likewise.
16386 * config/nds32/nds32.c (nds32_dpref_names): Likewise.
16387 (nds32_print_operand): Likewise.
16388 * config/nds32/nds32.h (nds32_builtins): Likewise.
16390 2018-04-07 Thomas Koenig <tkoenig@gcc.gnu.org>
16391 Andrew Pinski <pinsika@gcc.gnu.org>
16393 PR middle-end/82976
16394 * match.pd: Use constant_boolean_node of correct type instead of
16395 boolean_true_node or boolean_false_node for simplifying
16396 pointer comparisons to zero.
16398 2018-04-07 Jakub Jelinek <jakub@redhat.com>
16400 PR tree-optimization/80021
16401 * tree.c (verify_type_variant): Make error call in verify_variant_match
16402 translatable and remove final full stop.
16404 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
16406 * config/nds32/constants.md (unspec_volatile_element): Add
16407 UNSPEC_VOLATILE_EH_RETURN.
16408 * config/nds32/nds32-md-auxiliary.c (nds32_output_stack_push,
16409 nds32_output_stack_pop): Support dwarf exception handling process.
16410 * config/nds32/nds32-protos.h (nds32_dynamic_chain_address): Declare.
16411 * config/nds32/nds32.c (nds32_init_machine_status): Support dwarf
16412 exception handling process.
16413 (nds32_compute_stack_frame): Likewise.
16414 (nds32_return_addr_rtx): Likewise.
16415 (nds32_initial_elimination_offset): Likewise.
16416 (nds32_expand_prologue): Likewise.
16417 (nds32_expand_epilogue): Likewise.
16418 (nds32_dynamic_chain_address): New function.
16419 * config/nds32/nds32.h (machine_function): Add fields for dwarf
16420 exception handling.
16421 (DYNAMIC_CHAIN_ADDRESS): Define.
16422 (EH_RETURN_DATA_REGNO): Define.
16423 (EH_RETURN_STACKADJ_RTX): Define.
16424 * config/nds32/nds32.md (eh_return, nds32_eh_return): Implement
16425 patterns for dwarf exception handling.
16427 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
16429 * config/nds32/nds32.h: Clean up obsolete macros.
16431 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
16433 * config/nds32/constants.md (unspec_element, unspec_volatile_element):
16434 Add enum values for particular instructions.
16435 * config/nds32/nds32-intrinsic.c: Implementation of expanding
16436 particular intrinsic functions.
16437 * config/nds32/nds32-intrinsic.md: Likewise.
16438 * config/nds32/nds32_intrinsic.h: Likewise.
16439 * config/nds32/nds32.h (nds32_builtins): Likewise.
16440 * config/nds32/nds32.md (type): Add pbsad and pbsada.
16441 (btst, ave): New patterns for particular instructions.
16443 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
16445 * config/nds32/constants.md (unspec_element, unspec_volatile_element):
16446 Add enum values for atomic load/store and memory sync.
16447 * config/nds32/nds32-intrinsic.c: Implementation for atomic load/store
16449 * config/nds32/nds32-intrinsic.md: Likewise.
16450 * config/nds32/nds32_intrinsic.h: Likewise.
16451 * config/nds32/nds32.h (nds32_builtins): Likewise.
16453 2018-04-07 Jakub Jelinek <jakub@redhat.com>
16455 PR tree-optimization/85257
16456 * fold-const.c (native_encode_vector): If not all elts could fit
16457 and off is -1, return 0 rather than offset.
16458 * tree-ssa-sccvn.c (vn_reference_lookup_3): Pass
16459 (offseti - offset2) / BITS_PER_UNIT as 4th argument to
16460 native_encode_expr. Verify len * BITS_PER_UNIT >= maxsizei. Don't
16461 adjust buffer in native_interpret_expr call.
16463 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
16465 * config/nds32/constants.md (unspec_volatile_element): Add cache
16466 control enum values.
16467 * config/nds32/nds32-intrinsic.c: Add cache control expand functions.
16468 * config/nds32/nds32-intrinsic.md: Add cache control patterns.
16469 * config/nds32/nds32.c (nds32_cctl_names): New.
16470 (nds32_print_operand): Handle cache control register names.
16471 * config/nds32/nds32.h (nds32_builtins): New enum values.
16472 * config/nds32/nds32_intrinsic.h: Add cache control enum types and
16474 * config/nds32/nds32.md (type): Add mmu.
16475 * config/nds32/pipelines.md (simple_insn): Add mmu.
16477 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
16479 * config/nds32/nds32.md (type): Remove call.
16480 * config/nds32/pipelines.md (simple_insn): Likewise.
16482 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
16484 * config/nds32/constants.md (unspec_volatile_element): Add
16485 UNSPEC_VOLATILE_FMFCSR, UNSPEC_VOLATILE_FMTCSR and
16486 UNSPEC_VOLATILE_FMFCFG.
16487 * config/nds32/nds32-intrinsic.c (bdesc_noarg): New builtin
16488 description for fmfcfg and fmfcsr.
16489 (bdesc_1arg): Add fmtcsr.
16490 (bdesc_2arg): Add fcpynss, fcpyss, fcpynsd and fcpysd.
16491 (nds32_expand_builtin_impl): Deal with FPU intrinsic functions.
16492 * config/nds32/nds32-intrinsic.md (unspec_fcpynsd, unspec_fcpysd,
16493 unspec_fcpynss, unspec_fcpysd, unspec_fcpyss, unspec_fmfcsr,
16494 unspec_fmfcfg): New patterns.
16495 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_FMFCFG,
16496 NDS32_BUILTIN_FMFCSR, NDS32_BUILTIN_FMTCSR, NDS32_BUILTIN_FCPYNSS,
16497 NDS32_BUILTIN_FCPYSS,NDS32_BUILTIN_FCPYNSD and NDS32_BUILTIN_FCPYSD.
16498 * config/nds32/nds32_intrinsic.h (__nds32__fcpynsd, __nds32__fcpynss,
16499 __nds32__fcpysd, __nds32__fcpyss, __nds32__fmfcsr, __nds32__fmtcsr,
16500 __nds32__fmfcfg): Define.
16502 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
16504 * config/nds32/nds32.c (nds32_intrinsic_register_names): Add more
16505 intrinsic register names.
16506 * config/nds32/nds32_intrinsic.h (nds32_intrinsic_registers): Add more
16507 intrinsic register enum values and macros.
16509 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
16511 * config/nds32/nds32.c (nds32_legitimate_index_p): Modify condition
16512 for load/store addressing form.
16513 (nds32_print_operand_address): Likewise.
16515 2018-04-06 Eric Botcazou <ebotcazou@adacore.com>
16518 * config/sparc/sparc.c (sparc_expand_move): Deal with symbolic operands
16519 based on LABEL_REF. Remove useless assertion.
16520 (pic_address_needs_scratch): Fix formatting.
16521 (sparc_legitimize_pic_address): Minor tweaks.
16522 (sparc_delegitimize_address): Adjust assertion accordingly.
16523 * config/sparc/sparc.md (movsi_pic_label_ref): Change label_ref_operand
16524 into symbolic_operand.
16525 (movsi_high_pic_label_ref): Likewise.
16526 (movsi_lo_sum_pic_label_ref): Likewise.
16527 (movdi_pic_label_ref): Likewise.
16528 (movdi_high_pic_label_ref): Likewise.
16529 (movdi_lo_sum_pic_label_ref): Likewise.
16531 2018-04-06 Amaan Cheval <amaan.cheval@gmail.com>
16533 * config.gcc (x86_64-*-rtems*): Add rtems.h to tm_file for
16534 custom LIB_SPEC setup.
16536 2018-04-06 Ruslan Bukin <br@bsdpad.com>
16537 Kito Cheng <kito.cheng@gmail.com>
16539 * config.gcc (riscv*-*-freebsd*): Add RISC-V FreeBSD support.
16540 * config/riscv/freebsd.h: New.
16542 2018-04-06 Chung-Ju Wu <jasonwucj@gmail.com>
16544 * config/nds32/nds32.c (nds32_adjust_insn_length): Refine.
16545 * config/nds32/nds32.h (ADJUST_INSN_LENGTH): Change the location in
16548 2018-04-06 Chung-Ju Wu <jasonwucj@gmail.com>
16549 Kito Cheng <kito.cheng@gmail.com>
16551 * config/nds32/nds32-md-auxiliary.c (nds32_output_return,
16552 nds32_output_call, nds32_symbol_binds_local_p): New functions.
16553 * config/nds32/nds32-protos.h (nds32_output_call,
16554 nds32_output_return): Declare.
16555 * config/nds32/nds32.md: Refine all the call and return patterns.
16557 2018-04-06 Jakub Jelinek <jakub@redhat.com>
16560 * dwarf2out.c (rtl_for_decl_init): For STRING_CST initializer only
16561 build CONST_STRING if TYPE_MAX_VALUE is non-NULL and is INTEGER_CST.
16563 PR rtl-optimization/84872
16564 * cfgloopmanip.c (create_preheader): Use make_forwarder_block even if
16565 nentry == 1 when CP_FALLTHRU_PREHEADERS and single_entry is
16566 EDGE_CROSSING edge.
16568 2018-04-06 Tamar Christina <tamar.christina@arm.com>
16570 * expr.c (copy_blkmode_to_reg): Revert 254862.
16571 * doc/sourcebuild.texi (word_mode_no_slow_unalign): Likewise.
16573 2018-04-06 Richard Biener <rguenther@suse.de>
16575 PR middle-end/85244
16576 * tree-dfa.c (get_ref_base_and_extent): Reset seen_variable_array_ref
16577 after seeing a component reference with an adjacent field. Treat
16578 refs to arrays at struct end of external decls similar to
16579 refs to unconstrained commons.
16581 2018-04-06 Jakub Jelinek <jakub@redhat.com>
16584 * fold-const.c (twoval_comparison_p): Remove SAVE_P argument and don't
16585 look through SAVE_EXPRs with non-side-effects argument. Adjust
16587 (fold_comparison): Adjust twoval_comparison_p caller, don't handle
16590 2018-04-06 Richard Biener <rguenther@suse.de>
16592 PR middle-end/85180
16593 * alias.c (find_base_term): New wrapper around find_base_term
16594 unwinding CSELIB_VAL_PTR changes.
16595 (find_base_term): Do not restore CSELIB_VAL_PTR during the
16598 2018-04-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16600 * config/s390/s390.c (s390_z10_optimize_cmp): Expand dedicated NOP
16602 * config/s390/s390.md (UNSPECV_NOP_LR_0, UNSPECV_NOP_LR_1): New
16603 constant definitions.
16604 ("nop"): lr 0,0 -> nopr r0
16605 ("nop_lr0", "nop_lr1"): New insn definitions.
16607 2018-04-06 Chung-Ju Wu <jasonwucj@gmail.com>
16609 * config/nds32/nds32.md (*stack_push, *stack_pop): Use
16610 NDS32_V3PUSH_AVAILABLE_P macro.
16612 2018-04-06 Monk Chiang <sh.chiang04@gmail.com>
16613 Chung-Ju Wu <jasonwucj@gmail.com>
16615 * config.gcc (nds32*-*-*): Add v2j v3f v3s checking.
16616 (nds32*-*-*): Add float and fpu_config into supported_defaults.
16617 * common/config/nds32/nds32-common.c (TARGET_DEFAULT_TARGET_FLAGS):
16618 Include TARGET_DEFAULT_FPU_ISA and TARGET_DEFAULT_FPU_FMA.
16619 * config/nds32/constants.md (unspec_element): Add UNSPEC_COPYSIGN,
16620 UNSPEC_FCPYNSD, UNSPEC_FCPYNSS, UNSPEC_FCPYSD and UNSPEC_FCPYSS.
16621 * config/nds32/constraints.md: New constraints and checking for hard
16622 float configuration.
16623 * config/nds32/iterators.md: New mode iterator and attribute for hard
16624 float configuration.
16625 * config/nds32/nds32-doubleword.md: Use hard float alternatives and
16627 * config/nds32/nds32-fpu.md: New file.
16628 * config/nds32/nds32-md-auxiliary.c: New functions and checkings to
16629 deal with hard float code generation.
16630 * config/nds32/nds32-opts.h (nds32_arch_type): Add ARCH_V3F and
16632 (abi_type, float_reg_number): New enum type.
16633 * config/nds32/nds32-predicates.c: New predicates for hard float.
16634 * config/nds32/nds32-protos.h: Declare functions for hard float.
16635 * config/nds32/nds32.c: Implementation for hard float configuration.
16636 * config/nds32/nds32.h: Definitions for hard float configuration.
16637 * config/nds32/nds32.md: Include hard float machine description and
16638 modify patterns for hard float configuration.
16639 * config/nds32/nds32.opt: New options for hard float configuration.
16640 * config/nds32/predicates.md: New predicates for hard float
16643 2018-04-06 Kuan-Lin Chen <kuanlinchentw@gmail.com>
16645 * common/config/nds32/nds32-common.c
16646 (nds32_option_optimization_table): Enable -mreleax-hint by default.
16648 2018-04-05 Jakub Jelinek <jakub@redhat.com>
16650 PR middle-end/85195
16651 * match.pd (BIT_FIELD_REF CONSTRUCTOR@0 @1 @2): Use view_convert around
16652 CONSTRUCTOR_ELT (ctor, ...)->value.
16654 2018-04-05 Uros Bizjak <ubizjak@gmail.com>
16657 * config/i386/i386.md (define_attr "memory"): Handle rotate1 type.
16659 2018-04-05 Tom de Vries <tom@codesourcery.com>
16662 * config/nvptx/nvptx.c (nvptx_single): Fix neutering of bb with only
16665 2018-04-05 Shiva Chen <shiva0217@gmail.com>
16666 Kito Cheng <kito.cheng@gmail.com>
16668 * config/nds32/constraints.md (U33): Fine-tune checking condition.
16669 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Ditto.
16670 * config/nds32/nds32.h (nds32_16bit_address_type): Add
16671 ADDRESS_POST_MODIFY_LO_REG_IMM3U.
16673 2018-04-05 Shiva Chen <shiva0217@gmail.com>
16674 Kito Cheng <kito.cheng@gmail.com>
16676 * config/nds32/constraints.md (Ufe): New memory constraint.
16677 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format,
16678 nds32_output_16bit_load): Consider r8 register for lwi45.fe format.
16679 * config/nds32/nds32.c (nds32_print_operand): Output lwi45.fe
16681 * config/nds32/nds32.h (nds32_16bit_address_type): Add ADDRESS_R8_IMM7U.
16682 * config/nds32/nds32.md (*mov<mode>): Adjust pattern.
16684 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
16686 * config/nds32/nds32.md: Use optimize_size in the condition for
16687 alu-shift instructions.
16689 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
16691 * config/nds32/nds32.md (divsi4, udivsi4): New patterns.
16693 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
16695 * config/nds32/nds32.md (negsi2): Refine pattern.
16697 2018-04-05 Kito Cheng <kito.cheng@gmail.com>
16698 Chung-Ju Wu <jasonwucj@gmail.com>
16700 * config/nds32/iterators.md (shift_rotate): New code iterator.
16701 (shift): New code attribute.
16702 * config/nds32/nds32-md-auxiliary.c (nds32_expand_constant): New.
16703 * config/nds32/nds32-protos.h (nds32_expand_constant): Declare.
16704 * config/nds32/nds32.c (nds32_print_operand): Deal with more cases.
16705 * config/nds32/nds32.md (addsi3, *add_srli): Refine implementation for
16706 bit-wise operations.
16707 (andsi3, *andsi3): Ditto.
16708 (iorsi3, *iorsi3, *or_slli, *or_srli): Ditto.
16709 (xorsi3, *xorsi3, *xor_slli, *xor_srli): Ditto.
16710 (<shift>si3, *ashlsi3, *ashrsi3, *lshrsi3, *rotrsi3): Ditto.
16711 * config/nds32/predicates.md (nds32_rimm5u_operand, nds32_and_operand,
16712 nds32_ior_operand, nds32_xor_operand): New predicates.
16714 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
16716 * config/nds32/nds32.md (add<mode>3, sub<mode>3): Rename to ...
16717 (addsi3, subsi3): ... this.
16719 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
16721 * config/nds32/nds32.md (*sub_srli, *and_slli): Fine-tune predicator.
16723 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
16725 * config/nds32/nds32.md: Adjust indention.
16727 2018-04-05 Kito Cheng <kito.cheng@gmail.com>
16729 * config/nds32/nds32.md (feature): New attribute.
16731 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
16733 * config/nds32/nds32.md (subtype): New attribute.
16735 2018-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
16738 * config/arm/arm-builtins.c (arm_expand_builtin): Change
16739 expansion to perform a bitwise AND of the argument followed by a
16740 boolean negation of the result.
16742 2018-04-04 Peter Bergner <bergner@vnet.ibm.com>
16744 PR rtl-optimization/84878
16745 * ddg.c (add_cross_iteration_register_deps): Use DF_REF_BB to determine
16746 the basic block. Assert the use reference is not artificial and that
16747 it has an associated insn.
16749 2018-04-04 Michael Matz <matz@suse.de>
16751 * builtins.c (compute_objsize): Pass correct operand
16752 to array_at_struct_end_p.
16754 2018-04-04 Richard Biener <rguenther@suse.de>
16757 * dwarf2out.c (dwarf2out_register_external_die): Peel namespaces
16758 from contexts for DINFO_LEVEL_TERSE and below.
16760 2018-04-04 Kito Cheng <kito.cheng@gmail.com>
16762 * config/nds32/nds32-doubleword.md (move_<mode>): Require
16763 resiter_operand condition.
16764 * config/nds32/nds32.md (*move<mode>): Ditto.
16766 2018-04-04 Kito Cheng <kito.cheng@gmail.com>
16767 Monk Chiang <sh.chiang04@gmail.com>
16769 * config/nds32/nds32.md (movmisalign<mode>): New pattern.
16771 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
16773 * config/nds32/nds32.md (movqi, movhi): Merge into mov<mode>.
16775 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
16776 Kito Cheng <kito.cheng@gmail.com>
16778 * config/nds32/nds32-md-auxiliary.c (nds32_inverse_cond_code,
16779 nds32_cond_code_str, output_cond_branch,
16780 output_cond_branch_compare_zero, nds32_expand_cbranch,
16781 nds32_expand_cstore, nds32_expand_movcc,
16782 nds32_output_cbranchsi4_equality_zero,
16783 nds32_output_cbranchsi4_equality_reg,
16784 nds32_output_cbranchsi4_equality_reg_or_const_int,
16785 nds32_output_cbranchsi4_greater_less_zero: New functions.
16786 * config/nds32/nds32-protos.h (nds32_expand_cbranch,
16787 nds32_expand_cstore, nds32_expand_movcc,
16788 nds32_output_cbranchsi4_equality_zero,
16789 nds32_output_cbranchsi4_equality_reg,
16790 nds32_output_cbranchsi4_equality_reg_or_const_int,
16791 nds32_output_cbranchsi4_greater_less_zero): Declare.
16792 * config/nds32/predicates.md (nds32_movecc_comparison_operator,
16793 nds32_rimm11s_operand): New predicates.
16794 * config/nds32/nds32.h (nds32_expand_result_type): New enum type.
16795 * config/nds32/nds32.md: Rewrite all the branch and conditional move
16798 2018-04-04 Kito Cheng <kito.cheng@gmail.com>
16800 * config/nds32/nds32-doubleword.md: Refine all the instruction type.
16801 * config/nds32/nds32.md: Ditto.
16802 * config/nds32/pipelines.md: Ditto.
16804 2018-04-04 Richard Biener <rguenther@suse.de>
16806 PR tree-optimization/85168
16807 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address): Avoid
16808 propagating abnormals.
16810 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
16812 * config/nds32/nds32.md (enabled): Use yes/no for this attribute.
16814 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
16815 Kito Cheng <kito.cheng@gmail.com>
16817 * config/nds32/nds32-md-auxiliary.c (nds32_long_call_p): New function.
16818 * config/nds32/nds32-protos.h (nds32_long_call_p): Declare.
16819 * config/nds32/nds32.c (nds32_function_ok_for_sibcall): New function.
16820 (TARGET_FUNCTION_OK_FOR_SIBCALL): Define.
16821 * config/nds32/nds32.md (sibcall_internal): New.
16822 (sibcall_register): Remove.
16823 (sibcall_immediate): Remove.
16824 (sibcall_value_internal): New.
16825 (sibcall_value_register): Remove.
16826 (sibcall_value_immediate): Remove.
16827 * config/nds32/predicates.md (nds32_general_register_operand): New.
16828 (nds32_call_address_operand): New.
16830 2018-04-03 Jakub Jelinek <jakub@redhat.com>
16832 PR rtl-optimization/85167
16833 * shrink-wrap.c (move_insn_for_shrink_wrap): Don't set bb_uses and
16834 bb_defs if *split_p, instead preinitialize it to NULL.
16836 PR tree-optimization/85156
16837 * builtins.c (fold_builtin_expect): Use save_expr on arg1 to avoid
16838 evaluating the argument multiple times.
16840 2018-04-03 Bill Schmidt <wschmidt@linux.ibm.com>
16842 * config/rs6000/emmintrin.h (_mm_cvtpd_epi32): Use __vector rather
16844 (_mm_cvtpd_ps): Likewise.
16845 (_mm_cvttpd_epi32): Likewise.
16846 * config/rs6000/mmintrin.h (_mm_unpacklo_pi8): Likewise.
16847 * config/rs6000/xmmintrin.h: For strict-ANSI C++ or C11, undefine
16848 vector, pixel, and bool following altivec.h include.
16850 2018-04-03 Martin Sebor <msebor@redhat.com>
16852 * doc/extend.texi (Common Function Attributes): Clarify.
16853 (const attribute): Likewise.
16854 (pure attribute): Likewise.
16856 2018-04-03 Jakub Jelinek <jakub@redhat.com>
16859 * config/i386/i386.c (ix86_expand_vector_set): Use
16860 HOST_WIDE_INT_1U << elt instead of 1 << elt. Formatting fix.
16862 2018-04-03 Uros Bizjak <ubizjak@gmail.com>
16864 * config/i386/i386.c (emit_i387_cw_initialization): Always use logic
16865 instructions when changing rounding bits to preserve precision bits
16866 in the x87 control word.
16868 2018-04-03 Martin Liska <mliska@suse.cz>
16870 PR tree-optimization/82491
16871 * rtl.h (strip_offset_and_add): Replace += suboffset with
16872 poly_uint64 () + suboffset.
16874 2018-03-29 Martin Liska <mliska@suse.cz>
16875 Martin Jambor <mjambor@suse.cz>
16878 * ipa-cp.c (propagate_bits_across_jump_function): Bail out if
16879 param_type is not an integral or pointer type.
16881 2018-04-03 Richard Biener <rguenther@suse.de>
16883 * sese.h (recompute_all_dominators): Remove.
16885 2018-04-02 Martin Sebor <msebor@redhat.com>
16887 * doc/invoke.texi (-Wrestrict): Fix typos.
16889 2018-04-02 Jim Wilson <jimw@sifive.com>
16891 * config/riscv/riscv.h (SHIFT_COUNT_TRUNCATED): Set to zero.
16892 * config/riscv/riscv.md (<optab>si3): Use QImode shift count.
16893 (<optab>di3, <optab>si3_extend): Likewise.
16894 (<optab>si3_mask, <optab>si3_mask_1): New.
16895 (<optab>di3_mask, <optab>di3_mask_1): New.
16896 (<optab>si3_extend_mask, <optab>si3_extend_mask_1): New.
16897 (lshrsi3_zero_extend_1): Use VOIDmode shift count.
16898 * config/riscv/sync.md (atomic_test_and_set): Emit QImode shift count.
16900 2018-04-02 Gerald Pfeifer <gerald@pfeifer.com>
16902 * doc/cpp.texi (Variadic Macros): Fix line continuation in an
16905 2018-04-02 Chung-Ju Wu <jasonwucj@gmail.com>
16907 * config/nds32/nds32.c (TARGET_CANONICALIZE_COMPARISON): Define.
16908 (nds32_canonicalize_comparison): New function.
16910 2018-04-01 Chung-Ju Wu <jasonwucj@gmail.com>
16911 Kito Cheng <kito.cheng@gmail.com>
16912 Kuan-Lin Chen <kuanlinchentw@gmail.com>
16914 * config.gcc (nds32): Add nds32-relax-opt.o into extra_objs.
16915 * config/nds32/constants.md (unspec_volatile_element): Add
16916 UNSPEC_VOLATILE_RELAX_GROUP.
16917 * config/nds32/nds32-relax-opt.c: New file.
16918 * config/nds32/nds32-predicates.c
16919 (nds32_symbol_load_store_p): New function.
16920 * config/nds32/nds32-protos.h
16921 (nds32_symbol_load_store_p): Declare function.
16922 (make_pass_nds32_relax_opt): Declare new rtl pass function.
16923 * config/nds32/nds32.c
16924 (nds32_register_pass): New function to register pass.
16925 (nds32_register_passes): New function to register passes.
16926 * config/nds32/nds32.md (relax_group): New pattern.
16927 * config/nds32/nds32.opt (mrelax-hint): New option.
16928 * config/nds32/t-nds32 (nds32-relax-opt.o): New dependency.
16930 2018-04-01 Kito Cheng <kito.cheng@gmail.com>
16932 * config/nds32/t-nds32: Modify files dependency.
16934 2018-04-01 Chung-Ju Wu <jasonwucj@gmail.com>
16936 * config/nds32/nds32.h (FUNCTION_PROFILER): Output newline character.
16937 (PROFILE_HOOK): Define its implementation.
16939 2018-04-01 Chung-Ju Wu <jasonwucj@gmail.com>
16941 * config/nds32/nds32.h (WCHAR_TYPE, WCHAR_TYPE_SIZE): Use unsigned int
16942 type and 32-bit size.
16944 2018-04-01 Jakub Jelinek <jakub@redhat.com>
16946 PR middle-end/85090
16947 * config/i386/sse.md (V): Add V64QI and V32HI for TARGET_AVX512F.
16948 (V_128_256): New mode iterator.
16949 (*avx512dq_vextract<shuffletype>64x2_1 splitter): New define_split.
16950 (*avx512f_vextract<shuffletype>32x4_1 splitter): Likewise.
16951 (xop_pcmov_<mode><avxsizesuffix>): Use V_128_256 mode iterator instead
16953 * config/i386/i386.c (ix86_expand_vector_set): Improve V32HImode and
16954 V64QImode expansion for !TARGET_AVX512BW && TARGET_AVX512F.
16956 2018-03-31 Segher Boessenkool <segher@kernel.crashing.org>
16959 * config/rs6000/xmmintrin.h (_mm_set_ps, _mm_max_ps): Handle (quiet)
16960 NaN inputs correctly.
16962 2018-03-30 Peter Bergner <bergner@vnet.ibm.com>
16965 * config/rs6000/vsx.md (??r): New mode attribute.
16966 (*vsx_mov<mode>_64bit): Use it.
16967 (*vsx_mov<mode>_32bit): Likewise.
16969 2018-03-30 Martin Sebor <msebor@redhat.com>
16971 PR tree-optimization/84818
16972 * builtins.c (check_access): Use warning_n.
16974 2018-03-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
16977 * config/rs6000/rs6000-string.c (expand_compare_loop): Fix redundant
16979 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Fix redundant
16982 2018-03-30 Julia Koval <julia.koval@intel.com>
16985 * x86-tune.def (movx, partial_reg_dependency): Enable for
16988 2018-03-29 Vladimir Makarov <vmakarov@redhat.com>
16990 PR inline-asm/84985
16991 * lra-constraints.c (process_alt_operands): Move setting
16992 this_alternative_matches below.
16994 2018-03-29 Martin Liska <mliska@suse.cz>
16997 * doc/invoke.texi: Document how LTO works with debug info.
16998 Describe auto-load support of binutils. Mention 'x86-64'
16999 as valid option value of -march option.
17001 2018-03-29 Jakub Jelinek <jakub@redhat.com>
17003 * config/i386/sse.md (<avx512>_blendm<mode>): Use <sseintprefix>.
17006 * fold-const.c (operand_equal_p): Handle DEBUG_BEGIN_STMT.
17007 For STATEMENT_LIST, pass down OEP_LEXICOGRAPHIC and maybe
17008 OEP_NO_HASH_CHECK for recursive call, to avoid exponential
17011 2018-03-28 Peter Bergner <bergner@vnet.ibm.com>
17014 * config/rs6000/rs6000.h (RS6000_BTM_POWERPC64): New define.
17015 (RS6000_BTM_COMMON): Add RS6000_BTM_POWERPC64.
17016 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add support
17017 for RS6000_BTM_POWERPC64.
17018 (rs6000_invalid_builtin): Add handling for RS6000_BTM_POWERPC64
17019 (rs6000_builtin_mask_names): Add RS6000_BTM_POWERPC64.
17020 * config/rs6000/rs6000-builtin.def (BU_P7_POWERPC64_MISC_2): New macro
17023 (DIVDEU): Likewise.
17025 2018-03-28 Carl Love <cel@us.ibm.com>
17028 2017-09-27 Carl Love <cel@us.ibm.com>
17030 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
17031 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
17032 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
17035 2018-03-28 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17037 * config/rs6000/xmmintrin.h (_mm_max_pi16): Use __vector __bool
17038 instead of __vector bool.
17039 (_mm_max_pu8): Likewise.
17040 (_mm_min_pi16): Likewise.
17042 2018-03-28 Peter Bergner <bergner@vnet.ibm.com>
17045 * config/rs6000/rs6000-builtin.def (DIVWEO): Delete macro expansion.
17046 (DIVWEUO): Likewise.
17047 (DIVDEO): Likewise.
17048 (DIVDEUO): Likewise.
17049 * config/rs6000/rs6000.c (builtin_function_type): Remove support for
17050 DIVWEUO and DIVDEUO.
17051 * config/rs6000/rs6000.md (UNSPEC_DIVEO, UNSPEC_DIVEUO): Delete unspecs.
17052 (UNSPEC_DIV_EXTEND): Remove deleted unspecs.
17053 (div_extend): Likewise.
17054 * doc/extend.texi (__builtin_divweo): Remove documentation for deleted
17056 (__builtin_divweuo): Likewise.
17057 (__builtin_divdeo): Likewise.
17058 (__builtin_divdeuo): Likewise.
17060 2018-03-28 Jakub Jelinek <jakub@redhat.com>
17063 * config/i386/i386.md (*add<mode>3_carry_0, *addsi3_carry_zext_0,
17064 *sub<mode>3_carry_0, *subsi3_carry_zext_0): New patterns.
17066 PR tree-optimization/82004
17067 * gimple-match-head.c (optimize_pow_to_exp): New function.
17068 * match.pd (pow(C,x) -> exp(log(C)*x)): Wrap with #if GIMPLE.
17069 Don't fold to exp if optimize_pow_to_exp is false.
17071 2018-03-28 Martin Liska <mliska@suse.cz>
17074 * calls.c (initialize_argument_information): Fix trailing space.
17075 * common.opt: Fix typo and provide better explanation for
17076 -fsanitize-coverage option.
17077 * config/i386/i386.opt: Fix typo.
17079 2018-03-28 Jakub Jelinek <jakub@redhat.com>
17080 Martin Liska <mliska@suse.cz>
17083 * gimplify.c (asan_poison_variable): Don't do the check for
17084 gimplify_omp_ctxp here.
17085 (gimplify_decl_expr): Do it here.
17086 (gimplify_target_expr): Likewise.
17088 2018-03-28 Martin Liska <mliska@suse.cz>
17091 * config/i386/i386.c (ix86_function_arg_advance): Do not call
17092 chkp_type_bounds_count if MPX is not enabled.
17094 2018-03-27 Chung-Ju Wu <jasonwucj@gmail.com>
17096 * config/nds32/nds32.h (BRANCH_COST): Adjust cost.
17098 2018-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
17101 * config/rs6000/rs6000.c (create_complex_muldiv): New helper
17102 function to create the function decl for complex long double
17103 multiply and divide for -mabi=ieeelongdouble.
17104 (init_float128_ieee): Call it.
17106 2018-03-27 H.J. Lu <hongjiu.lu@intel.com>
17109 * config/i386/i386.c (ix86_trampoline_init): Insert ENDBR for
17110 -fcf-protection=branch -mibt.
17111 * config/i386/i386.h (TRAMPOLINE_SIZE): Increased by 4 bytes.
17113 2018-03-27 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
17116 * config/arm/arm.c (arm_valid_symbolic_address): Handle
17117 arm_word_relocations.
17119 2018-03-27 Cesar Philippidis <cesar@codesourcery.com>
17122 * config/nvptx/nvptx.c (nvptx_assemble_decl_begin): Add '[]' to
17123 extern array declarations.
17125 2018-03-27 Richard Biener <rguenther@suse.de>
17127 PR middle-end/84067
17128 * match.pd ((A * C) +- (B * C) -> (A+-B) * C): Guard with
17129 explicit single_use checks.
17131 2018-03-27 Richard Biener <rguenther@suse.de>
17133 PR tree-optimization/85082
17134 * tree-ssa-sccvn.c (vn_reference_lookup_or_insert_for_pieces):
17137 2018-03-27 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
17139 * config.gcc (aarch64*-*-linux*): New TARGET_DEFAULT_ASYNC_UNWIND_TABLES
17140 * common/config/aarch64/aarch64-common.c (aarch64_optimization_table[]):
17141 Turn on fasynchronous-unwind-tables and funwind-tables.
17143 2018-03-26 Uros Bizjak <ubizjak@gmail.com>
17146 * config/i386/i386.md (*bmi_blsr_<mode>_cmp): New insn pattern.
17147 (*bmi_blsr_<mode>_ccz): Ditto.
17149 2018-03-26 Tom de Vries <tom@codesourcery.com>
17151 PR tree-optimization/85063
17152 * omp-general.c (offloading_function_p): New function. Factor out
17154 * omp-offload.c (pass_omp_target_link::gate): ... here.
17155 * omp-general.h (offloading_function_p): Declare.
17156 * tree-switch-conversion.c (build_one_array): Mark CSWTCH.x variable
17157 with attribute omp declare target for offloading functions.
17159 2018-03-24 Richard Sandiford <richard.sandiford@linaro.org>
17161 PR tree-optimization/84005
17162 * tree-data-ref.h (get_base_for_alignment): Declare.
17163 * tree-data-ref.c (get_base_for_alignment_1): New function.
17164 (get_base_for_alignment): Likewise.
17165 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
17166 get_base_for_alignment to find a suitable base object, instead
17167 of always using drb->base_address.
17169 2018-03-23 Jakub Jelinek <jakub@redhat.com>
17171 PR inline-asm/85022
17172 * emit-rtl.c (init_emit_regs): Indicate that VOIDmode MEMs don't have
17173 known size by default.
17175 2018-03-23 Vladimir Makarov <vmakarov@redhat.com>
17177 PR inline-asm/85030
17178 * lra-constraints.c (process_alt_operands): Don't match BLKmode
17179 and non BLKmode operands.
17181 2018-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17184 * config/arm/arm.md (unaligned_loadhis): Remove first alternative.
17185 Clean up attributes.
17187 2018-03-23 Richard Biener <rguenther@suse.de>
17190 * dwarf2out.c (rtl_for_decl_location): Do not generate RTL early when
17191 we are going to emit early debug for LTO.
17193 2018-03-23 Jakub Jelinek <jakub@redhat.com>
17195 PR inline-asm/85034
17196 * function.c (match_asm_constraints_1): Don't optimize if input
17197 doesn't satisfy general_operand predicate for output's mode.
17199 PR inline-asm/85022
17200 * alias.c (write_dependence_p): Don't require for x_canonicalized
17201 non-VOIDmode if x has VOIDmode.
17204 * sanopt.c (maybe_optimize_ubsan_ptr_ifn): If DECL_REGISTER (base),
17205 just don't try to optimize it rather than assert it never happens.
17207 2018-03-22 Kelvin Nilsen <kelvin@gcc.gnu.org>
17209 * config/rs6000/rs6000-builtin.def: Remove various BU_ALTIVEC_X
17210 macro expansions for definition of ST_INTERNAL_<mode> and
17211 LD_INTERNAL_<mode> builtins.
17212 * config/rs6000/rs6000-protos.h (rs6000_address_for_altivec):
17214 * config/rs6000/rs6000.c (altivec_expand_ld_builtin): Delete this
17216 (altivec_expand_st_builtin): Likewise.
17217 (altivec_expand_builtin): Remove calls to deleted functions.
17218 (rs6000_address_for_altivec): Delete this function.
17219 * config/rs6000/vector.md: Remove expands for
17220 vector_altivec_load_<mode> and vector_altivec_store_<mode>.
17222 2018-03-22 Sudakshina Das <sudi.das@arm.com>
17225 * config/arm/arm.h (machine_function): Add static_chain_stack_bytes.
17226 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Avoid
17227 re-computing once computed.
17228 (arm_expand_prologue): Compute machine->static_chain_stack_bytes.
17229 (arm_init_machine_status): Initialize
17230 machine->static_chain_stack_bytes.
17232 2018-03-22 Kelvin Nilsen <kelvin@gcc.gnu.org>
17235 * doc/extend.texi: Add four new prototypes for vec_ld.
17236 * config/rs6000/rs6000-builtin.def (LVX_V1TI): Reorder symbol
17237 definitions for more logical presentation.
17238 * config/rs6000/rs6000-c.c: (altivec_overloaded_builtins): Add
17239 entries for V1TI variants of __builtin_altivec_ld builtin.
17240 * config/rs6000/rs6000.c: (altivec_expand_lv_builtin): Add test and
17241 handling of V1TI variant of LVX icode pattern.
17242 (altivec_expand_builtin): Add case for ALTIVEC_BUILTIN_LVX_V1TI.
17243 (rs6000_gimple_fold_builtin): Likewise.
17244 (altivec_init_builtins): Add code to define
17245 __builtin_altivec_lvx_v1ti function.
17247 2018-03-22 Jakub Jelinek <jakub@redhat.com>
17249 PR inline-asm/84941
17250 * function.c (match_asm_constraints_1): Don't do the optimization
17251 if input isn't a REG, SUBREG, MEM or constant.
17253 2018-03-22 Tom de Vries <tom@codesourcery.com>
17255 PR tree-optimization/84956
17256 * tree-ssa-tail-merge.c (find_clusters_1): Skip bbs with
17257 bb_has_abnormal_pred.
17259 2018-03-22 Jakub Jelinek <jakub@redhat.com>
17262 * dwarf2asm.c (dw2_output_indirect_constant_1): Set
17263 DECL_INITIAL (decl) to decl at the end.
17264 * varasm.c (use_blocks_for_decl_p): Revert the 2018-03-20 change,
17265 adjust the comment.
17267 2018-03-21 Joseph Myers <joseph@codesourcery.com>
17269 * doc/extend.texi (__builtin_tgmath): Document when complex
17270 integer types are treated as _Complex _Float64.
17272 2018-03-21 Tom de Vries <tom@codesourcery.com>
17274 * doc/extend.texi (__builtin_extend_pointer): Remove pasto.
17276 2018-03-21 Jakub Jelinek <jakub@redhat.com>
17278 PR tree-optimization/84960
17279 * tree-cfg.c (remove_bb): Don't move forced labels into bb->prev_bb
17280 if it is ENTRY block, move them into single succ of ENTRY in that case.
17282 2018-03-21 Richard Sandiford <richard.sandiford@linaro.org>
17284 PR tree-optimization/84811
17285 * poly-int.h (poly_span_traits): Remove the T3 parameter and
17286 promote HOST_WIDE_INT T2 - T1 results to unsigned HOST_WIDE_INT.
17287 (maybe_in_range_p, known_in_range_p, ranges_known_overlap_p):
17288 (known_subrange_p): Update accordingly. Cast each value involved
17289 in the size comparison, rather than casting the result of the
17292 2018-03-21 Jakub Jelinek <jakub@redhat.com>
17294 PR tree-optimization/84982
17295 * gimple-ssa-store-merging.c (invert_op): Handle boolean inversion
17296 by flipping the least significant bit rather than all bits from
17297 bitpos to bitpos + bitsize - 1.
17299 2018-03-21 Nathan Sidwell <nathan@acm.org>
17301 * doc/extend.texi (Deprecated Features): Remove mention of
17302 long-deleted deprecations.
17304 2018-03-21 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
17307 * configure.ac (gcc_cv_ld_soname) <*-*-solaris2*>: Set.
17308 * configure: Regenerate.
17310 2018-03-21 Tom de Vries <tom@codesourcery.com>
17312 PR tree-optimization/83126
17313 * tree-parloops.c (num_phis): New function.
17314 (gen_parallel_loop): Detect and handle canonicalize_loop_ivs failure.
17316 2018-03-21 Nathan Sidwell <nathan@acm.org>
17318 * doc/extend.texi (Deprecated Features): Update deprecated flags,
17319 mention anon-struct/union members and trailing attributes.
17321 2018-03-21 Bin Cheng <bin.cheng@arm.com>
17323 PR tree-optimization/84969
17324 * tree-loop-distribution.c (fuse_memset_builtins): Don't reorder
17325 builtin memset partitions if they set different rhs values.
17327 2018-03-21 Jakub Jelinek <jakub@redhat.com>
17329 PR rtl-optimization/84989
17330 * simplify-rtx.c (simplify_unary_operation_1): Don't try to simplify
17331 VEC_DUPLICATE with scalar result mode.
17333 2018-03-21 Martin Liska <mliska@suse.cz>
17336 * ipa-icf.c (sem_item_optimizer::fixup_points_to_sets): Remove
17337 not intended return statement.
17339 2018-03-21 Martin Liska <mliska@suse.cz>
17342 * tree-chkp.c (CHKP_ARRAY_MAX_CHECK_STEPS): Define a new macro.
17343 (chkp_find_bound_slots_1): Limit number of iterations.
17345 2018-03-20 David H. Gutteridge <dhgutteridge@sympatico.ca>
17348 * Minor grammar fixes for x86 options.
17350 2018-03-20 Jakub Jelinek <jakub@redhat.com>
17353 * dce.c (delete_unmarked_insns): Don't remove frame related noop moves
17354 holding REG_CFA_RESTORE notes, instead turn them into a USE.
17356 2018-03-20 Peter Bergner <bergner@vnet.ibm.com>
17359 * config/rs6000/altivec.md (altivec_lvx_<mode>_2op): Delete define_insn.
17360 (altivec_lvx_<mode>_1op): Likewise.
17361 (altivec_stvx_<mode>_2op): Likewise.
17362 (altivec_stvx_<mode>_1op): Likewise.
17363 (altivec_lvx_<VM2:mode>): New define_expand.
17364 (altivec_stvx_<VM2:mode>): Likewise.
17365 (altivec_lvx_<VM2:mode>_2op_<P:mptrsize>): New define_insn.
17366 (altivec_lvx_<VM2:mode>_1op_<P:mptrsize>): Likewise.
17367 (altivec_stvx_<VM2:mode>_2op_<P:mptrsize>): Likewise.
17368 (altivec_stvx_<VM2:mode>_1op_<P:mptrsize>): Likewise.
17369 * config/rs6000/rs6000-p8swap.c (rs6000_gen_stvx): Use new expanders.
17370 (rs6000_gen_lvx): Likewise.
17371 * config/rs6000/rs6000.c (altivec_expand_lv_builtin): Likewise.
17372 (altivec_expand_stv_builtin): Likewise.
17373 (altivec_expand_builtin): Likewise.
17374 * config/rs6000/vector.md: Likewise.
17376 2018-03-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17379 * config/arm/arm.c (arm_array_mode_supported_p): Return false for
17382 2018-03-20 Richard Biener <rguenther@suse.de>
17385 * config/i386/i386.c (ix86_add_stmt_cost): Only cost
17386 sign-conversions as zero, fall back to standard scalar_stmt
17389 2018-03-20 Martin Liska <mliska@suse.cz>
17392 * predict.c (rebuild_frequencies): Handle case when we have
17393 PROFILE_ABSENT, but flag_guess_branch_prob is false.
17395 2018-03-20 Jakub Jelinek <jakub@redhat.com>
17398 * dwarf2asm.c (dw2_output_indirect_constant_1): Temporarily turn off
17399 flag_section_anchors.
17400 * varasm.c (use_blocks_for_decl_p): Remove hack for
17401 dw2_force_const_mem.
17404 * config/aarch64/aarch64.md (*aarch64_reg_<mode>3_neg_mask2): Rename
17406 (*aarch64_<optab>_reg_<mode>3_neg_mask2): ... this. If pseudos can't
17407 be created, use lowpart_subreg of operands[0] rather than operands[0]
17409 (*aarch64_reg_<mode>3_minus_mask): Rename to ...
17410 (*aarch64_ashl_reg_<mode>3_minus_mask): ... this.
17411 (*aarch64_<optab>_reg_di3_mask2): Use const_int_operand predicate
17412 and n constraint instead of aarch64_shift_imm_di and Usd.
17413 (*aarch64_reg_<optab>_minus<mode>3): Rename to ...
17414 (*aarch64_<optab>_reg_minus<mode>3): ... this.
17416 2018-03-20 Sudakshina Das <sudi.das@arm.com>
17419 * config/arm/neon.md (ashldi3_neon): Update ?s for constraints
17420 to favor GPR over NEON registers.
17421 (<shift>di3_neon): Likewise.
17423 2018-03-20 Tom de Vries <tom@codesourcery.com>
17426 * config/nvptx/nvptx.c (nvptx_single): Don't neuter bar.sync.
17427 (nvptx_process_pars): Emit bar.sync asap and alap.
17429 2018-03-20 Tom de Vries <tom@codesourcery.com>
17432 * config/nvptx/nvptx.c (prevent_branch_around_nothing): Also update
17433 seen_label if seen_label is already set.
17435 2018-03-20 Jakub Jelinek <jakub@redhat.com>
17438 * config/i386/i386.c (fold_builtin_cpu): For features above 31
17439 use __cpu_features2 variable instead of __cpu_model.__cpu_features[0].
17440 Use 1U instead of 1. Formatting fixes.
17443 * builtins.c (fold_builtin_strpbrk): For strpbrk(x, "") use type
17444 instead of TREE_TYPE (s1) for the return value.
17446 2018-03-19 Jakub Jelinek <jakub@redhat.com>
17448 PR tree-optimization/84946
17449 * gimple-ssa-store-merging.c (mem_valid_for_store_merging): Compute
17450 bitsize + bitsize in poly_uint64 rather than poly_int64.
17453 * dwarf2asm.c: Include fold-const.c.
17454 (dw2_output_indirect_constant_1): Set DECL_INITIAL (decl) to ADDR_EXPR
17455 of decl rather than decl itself.
17457 PR rtl-optimization/84643
17458 * memmodel.h (enum memmodel): Add MEMMODEL_MAX enumerator.
17460 2018-03-19 Maxim Ostapenko <m.ostapenko@samsung.com>
17463 * dwarf2asm.c (dw2_output_indirect_constant_1): Disable ASan before
17464 calling assemble_variable.
17466 2018-03-19 Sudakshina Das <sudi.das@arm.com>
17469 * config/aarch64/aarch64-simd.md (vec_cmp<mode><v_int_equiv>): Modify
17470 instructions for UNLT, UNLE, UNGT, UNGE, UNEQ, UNORDERED and ORDERED.
17472 2018-03-19 Jim Wilson <jimw@sifive.com>
17475 * config/riscv/riscv.c (riscv_function_arg_boundary): Use
17476 PREFERRED_STACK_BOUNDARY instead of STACK_BOUNDARY.
17477 (riscv_first_stack_step): Likewise.
17478 (riscv_option_override): Use STACK_BOUNDARY instead of
17479 MIN_STACK_BOUNDARY.
17480 * config/riscv/riscv.h (STACK_BOUNDARY): Renamed from
17481 MIN_STACK_BOUNDARY.
17482 (BIGGEST_ALIGNMENT): Set to 128.
17483 (PREFERRED_STACK_BOUNDARY): Renamed from STACK_BOUNDARY.
17484 (RISCV_STACK_ALIGN): Use PREFERRED_STACK_BOUNDARY instead of
17487 2018-03-19 Richard Biener <rguenther@suse.de>
17489 PR tree-optimization/84933
17490 * tree-vrp.c (set_and_canonicalize_value_range): Treat out-of-bound
17491 values as -INF/INF when canonicalizing an ANTI_RANGE to a RANGE.
17493 2018-03-19 Richard Biener <rguenther@suse.de>
17495 PR tree-optimization/84859
17496 * tree-ssa-phiopt.c (single_trailing_store_in_bb): New function.
17497 (cond_if_else_store_replacement): Perform sinking operation on
17498 single-store BBs regardless of MAX_STORES_TO_SINK setting.
17499 Generalize what a BB with a single eligible store is.
17501 2018-03-19 Richard Biener <rguenther@suse.de>
17503 PR tree-optimization/84929
17504 * tree-data-ref.c (analyze_siv_subscript_cst_affine): Guard
17505 chrec_is_positive against non-chrec arg.
17507 2018-03-19 Tamar Christina <tamar.christina@arm.com>
17510 * config/arm/arm.c (arm_can_change_mode_class): revert r258554.
17512 2018-03-18 Martin Liska <mliska@suse.cz>
17514 PR rtl-optimization/84635
17515 * regrename.c (build_def_use): Use matches_mode only when
17518 2018-03-18 Richard Sandiford <richard.sandiford@linaro.org>
17520 PR tree-optimization/84913
17521 * tree-vect-loop.c (vectorizable_reduction): Don't try to
17522 vectorize chains of COND_EXPRs.
17524 2018-03-18 Chung-Ju Wu <jasonwucj@gmail.com>
17526 * config/nds32/nds32.h (MAX_REGS_PER_ADDRESS): Fix the value.
17528 2018-03-18 Chung-Ju Wu <jasonwucj@gmail.com>
17530 * config/nds32/nds32.h (LOGICAL_OP_NON_SHORT_CIRCUIT): Define.
17532 2018-03-18 Chung-Ju Wu <jasonwucj@gmail.com>
17534 * config/nds32/nds32.h (CLZ_DEFINED_VALUE_AT_ZERO): Define.
17536 2018-03-17 Chung-Ju Wu <jasonwucj@gmail.com>
17537 Kito Cheng <kito.cheng@gmail.com>
17539 * config/nds32/nds32-protos.h (nds32_adjust_reg_alloc_order): Declare.
17540 * config/nds32/nds32.c (nds32_reg_alloc_order_for_speed): New array.
17541 (nds32_adjust_reg_alloc_order): New function.
17542 * config/nds32/nds32.h (ADJUST_REG_ALLOC_ORDER): Define.
17544 2018-03-17 Kito Cheng <kito.cheng@gmail.com>
17546 * config/nds32/nds32.c (nds32_asm_output_mi_thunk,
17547 nds32_print_operand, nds32_print_operand_address): Use
17548 HOST_WIDE_INT_PRINT_DEC instead.
17550 2018-03-17 Chung-Ju Wu <jasonwucj@gmail.com>
17552 * config/nds32/nds32.c (nds32_register_priority): Modify cost.
17554 2018-03-17 Jakub Jelinek <jakub@redhat.com>
17557 * config/i386/i386.c (initial_ix86_tune_features,
17558 initial_ix86_arch_features): Use unsigned HOST_WIDE_INT rather than
17559 unsigned long long.
17560 (set_ix86_tune_features): Change ix86_tune_mask from unsigned int
17561 to unsigned HOST_WIDE_INT, initialize to HOST_WIDE_INT_1U << ix86_tune
17562 rather than 1u << ix86_tune. Formatting fix.
17563 (ix86_option_override_internal): Change ix86_arch_mask from
17564 unsigned int to unsigned HOST_WIDE_INT, initialize to
17565 HOST_WIDE_INT_1U << ix86_arch rather than 1u << ix86_arch.
17566 (ix86_function_specific_restore): Likewise.
17568 2018-03-16 Jakub Jelinek <jakub@redhat.com>
17571 * postreload.c (reload_combine_recognize_pattern): Perform
17572 INTVAL addition in unsigned HOST_WIDE_INT type to avoid UB and
17573 truncate_int_for_mode the result for the destination's mode.
17576 * hsa-gen.c (mem_type_for_type): Fix comment typo.
17577 * tree-vect-loop-manip.c (vect_create_cond_for_niters_checks):
17579 * gimple-ssa-warn-restrict.c (builtin_memref::set_base_and_offset):
17582 2018-03-16 Vladimir Makarov <vmakarov@redhat.com>
17585 * lra-assigns.c (lra_split_hard_reg_for): Don't use
17586 regno_allocno_class_array and sorted_pseudos.
17587 * lra-constraints.c (spill_hard_reg_in_range): Ignore hard regs in
17588 insns where regno is used.
17590 2018-03-16 Martin Liska <mliska@suse.cz>
17593 * multiple_target.c (create_dispatcher_calls): Redirect
17594 reference in the symbol table.
17596 2018-03-16 Martin Liska <mliska@suse.cz>
17599 * multiple_target.c (create_dispatcher_calls): Redirect also
17602 2018-03-16 Jakub Jelinek <jakub@redhat.com>
17606 * tree.h (TARGET_EXPR_NO_ELIDE): Define.
17607 * gimplify.c (gimplify_modify_expr_rhs): Don't elide TARGET_EXPRs with
17608 TARGET_EXPR_NO_ELIDE flag set unless *expr_p is INIT_EXPR.
17610 2018-03-16 Julia Koval <julia.koval@intel.com>
17612 * doc/invoke.texi (Skylake Server): Add CLWB.
17613 Cannonlake): Remove CLWB.
17615 2018-03-16 Jakub Jelinek <jakub@redhat.com>
17617 PR tree-optimization/84841
17618 * tree-ssa-reassoc.c (INTEGER_CONST_TYPE): Change to 1 << 4 from
17620 (FLOAT_ONE_CONST_TYPE): Define.
17621 (constant_type): Return FLOAT_ONE_CONST_TYPE for -1.0 and 1.0.
17622 (sort_by_operand_rank): Put entries with higher constant_type last
17623 rather than first to match comments.
17625 2018-03-15 Sandra Loosemore <sandra@codesourcery.com>
17627 * config/nios2/nios2.md (movsi_internal): Fix thinko in
17630 2018-03-15 Jakub Jelinek <jakub@redhat.com>
17633 * calls.c (expand_call): For TREE_ADDRESSABLE rettype ignore alignment
17634 check and use address of target always.
17636 2018-03-15 H.J. Lu <hongjiu.lu@intel.com>
17639 * config/i386/i386.c (indirect_thunk_needed): Update comments.
17640 (indirect_thunk_bnd_needed): Likewise.
17641 (indirect_thunks_used): Likewise.
17642 (indirect_thunks_bnd_used): Likewise.
17643 (indirect_return_needed): New.
17644 (indirect_return_bnd_needed): Likewise.
17645 (output_indirect_thunk_function): Add a bool argument for
17647 (output_indirect_thunk_function): Don't generate alias for
17648 function return thunk.
17649 (ix86_code_end): Call output_indirect_thunk_function to generate
17650 function return thunks.
17651 (ix86_output_function_return): Set indirect_return_bnd_needed
17652 and indirect_return_needed instead of indirect_thunk_bnd_needed
17653 and indirect_thunk_needed.
17655 2018-03-15 Olga Makhotina <olga.makhotina@intel.com>
17657 * config/i386/sgxintrin.h (_enclv_u32): New intrinsic.
17658 (__enclv_bc, __enclv_cd, __enclv_generic): New definitions.
17659 (ERDINFO, ETRACKC, ELDBC, ELDUC): New leaves.
17661 2018-03-15 David Malcolm <dmalcolm@redhat.com>
17662 Paul Hua <paul.hua.gm@gmail.com>
17665 * gcc.dg/fixits-pr84852-1.c: Fix filename in dg-regexp.
17667 2018-03-15 Segher Boessenkool <segher@kernel.crashing.org>
17669 * config/rs6000/rs6000.c (abi_v4_pass_in_fpr): Add back the
17670 TARGET_DOUBLE_FLOAT and TARGET_SINGLE_FLOAT conditions on the DFmode
17671 resp. SFmode cases.
17673 2018-03-15 Tamar Christina <tamar.christina@arm.com>
17676 * config/arm/arm.c (arm_can_change_mode_class): Use GET_MODE_UNIT_SIZE
17677 instead of GET_MODE_SIZE when comparing Units.
17679 2018-03-15 Vladimir Mezentsev <vladimir.mezentsev@oracle.com>
17682 * varasm.c (hash_section): Return an unchangeble hash value
17683 * config/aarch64/aarch64.c (aarch64_use_blocks_for_constant_p):
17684 Return !aarch64_can_use_per_function_literal_pools_p ().
17686 2018-03-15 Jakub Jelinek <jakub@redhat.com>
17689 * optabs.c (emit_conditional_move): Pass address of cmode's copy
17690 rather than address of cmode as last argument to prepare_cmp_insn.
17692 2018-03-15 Julia Koval <julia.koval@intel.com>
17694 * config/i386/i386.c (F_AVX512VBMI2, F_GFNI, F_VPCLMULQDQ,
17695 F_AVX512VNNI, F_AVX512BITALG): New.
17697 2018-03-14 John David Anglin <danglin@gcc.gnu.org>
17700 * config/pa/pa.c (pa_emit_move_sequence): Always emit secondary reload
17701 insn for floating-point loads and stores.
17703 2018-03-14 Carl Love <cel@us.ibm.com>
17705 * config/rs6000/rs6000-c.c: Add macro definitions for
17706 ALTIVEC_BUILTIN_VEC_PERMXOR.
17707 * config/rs6000/rs6000.h: Add #define for vec_permxor builtin.
17708 * config/rs6000/rs6000-builtin.def: Add macro expansions for VPERMXOR.
17709 * config/rs6000/altivec.md (altivec_vpermxor): New define expand.
17710 * config/rs6000/rs6000-p8swap.c (rtx_is_swappable_p): Add case
17712 * config/doc/extend.texi: Add prototypes for vec_permxor.
17714 2018-03-14 David Malcolm <dmalcolm@redhat.com>
17717 * diagnostic-show-locus.c (class layout_point): Convert m_line
17718 from int to linenum_type.
17719 (line_span::comparator): Use linenum "compare" function when
17720 comparing line numbers.
17721 (test_line_span): New function.
17722 (layout_range::contains_point): Convert param "row" from int to
17724 (layout_range::intersects_line_p): Likewise.
17725 (layout::will_show_line_p): Likewise.
17726 (layout::print_source_line): Likewise.
17727 (layout::should_print_annotation_line_p): Likewise.
17728 (layout::print_annotation_line): Likewise.
17729 (layout::print_leading_fixits): Likewise.
17730 (layout::annotation_line_showed_range_p): Likewise.
17731 (struct line_corrections): Likewise for field m_row.
17732 (line_corrections::line_corrections): Likewise for param "row".
17733 (layout::print_trailing_fixits): Likewise.
17734 (layout::get_state_at_point): Likewise.
17735 (layout::get_x_bound_for_row): Likewise.
17736 (layout::print_line): Likewise.
17737 (diagnostic_show_locus): Likewise for locals "last_line" and "row".
17738 (selftest::diagnostic_show_locus_c_tests): Call test_line_span.
17739 * input.c (selftest::test_linenum_comparisons): New function.
17740 (selftest::input_c_tests): Call it.
17741 * selftest.c (selftest::test_assertions): Test ASSERT_GT,
17742 ASSERT_GT_AT, ASSERT_LT, and ASSERT_LT_AT.
17743 * selftest.h (ASSERT_GT): New macro.
17744 (ASSERT_GT_AT): New macro.
17745 (ASSERT_LT): New macro.
17746 (ASSERT_LT_AT): New macro.
17748 2018-03-14 Segher Boessenkool <segher@kernel.crashing.org>
17750 PR rtl-optimization/84780
17751 * combine.c (distribute_links): Don't make a link based on pc_rtx.
17753 2018-03-14 Martin Liska <mliska@suse.cz>
17755 * tree.c (record_node_allocation_statistics): Use
17756 get_stats_node_kind.
17757 (get_stats_node_kind): New function extracted from
17758 record_node_allocation_statistics.
17759 (free_node): Use get_stats_node_kind.
17761 2018-03-14 Richard Biener <rguenther@suse.de>
17763 * tree-ssa-pre.c (compute_antic_aux): Remove code that asserts
17764 that the value-set of ANTIC_IN doesn't grow.
17767 * tree-ssa-pre.c (struct bb_bitmap_sets): Add visited_with_visited_succs
17769 (BB_VISITED_WITH_VISITED_SUCCS): New define.
17770 (compute_antic): Initialize BB_VISITED_WITH_VISITED_SUCCS.
17772 2018-03-14 Julia Koval <julia.koval@intel.com>
17774 * config.gcc (icelake-client, icelake-server): New.
17776 * config/i386/i386.c (initial_ix86_tune_features): Extend to 64 bit.
17777 (initial_ix86_arch_features): Ditto.
17778 (PTA_SKYLAKE): Add SGX.
17779 (PTA_ICELAKE): Remove.
17780 (PTA_ICELAKE_CLIENT): New.
17781 (PTA_ICELAKE_SERVER): New.
17782 (ix86_option_override_internal): Split up icelake on icelake client and
17784 (get_builtin_code_for_version): Ditto.
17785 (fold_builtin_cpu): Ditto.
17786 * config/i386/driver-i386.c (config/i386/driver-i386.c): Ditto.
17787 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto
17788 * config/i386/i386.h (processor_type): Ditto.
17789 * doc/invoke.texi: Ditto.
17791 2018-03-14 Jakub Jelinek <jakub@redhat.com>
17794 * sanopt.c (maybe_optimize_ubsan_ptr_ifn): Handle also
17795 INTEGER_CST offset, add it together with bitpos / 8 and
17796 sign extend based on POINTER_SIZE.
17800 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
17803 * config/i386/constraints.md (Yc): New register constraint.
17804 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
17805 Use Yc constraint for alternative 2 of operand 0. Remove
17806 preferred_for_speed attribute.
17808 2018-03-14 Richard Biener <rguenther@suse.de>
17810 PR tree-optimization/84830
17811 * tree-ssa-pre.c (compute_antic_aux): Intersect the new ANTIC_IN
17812 with the old one to avoid oscillations.
17814 2018-03-13 Vladimir Makarov <vmakarov@redhat.com>
17817 * lra-assigns.c (find_all_spills_for): Ignore uninteresting
17819 (assign_by_spills): Return a flag of reload assignment failure.
17820 Do not process the reload assignment failures. Do not spill other
17821 reload pseudos if they has the same reg class. Update n if
17823 (lra_assign): Add a return arg. Set up from the result of
17824 assign_by_spills call.
17825 (find_reload_regno_insns, lra_split_hard_reg_for): New functions.
17826 * lra-constraints.c (split_reg): Add a new arg. Use it instead of
17827 usage_insns if it is not NULL.
17828 (spill_hard_reg_in_range): New function.
17829 (split_if_necessary, inherit_in_ebb): Pass a new arg to split_reg.
17830 * lra-int.h (spill_hard_reg_in_range, lra_split_hard_reg_for): New
17831 function prototypes.
17832 (lra_assign): Change prototype.
17833 * lra.c (lra): Add code to deal with fails by splitting hard reg
17836 2018-03-01 Palmer Dabbelt <palmer@sifive.com>
17838 * config/riscv/riscv.opt (mrelax): New option.
17839 * config/riscv/riscv.c (riscv_file_start): Emit ".option
17840 "norelax" when riscv_mrelax is disabled.
17841 * doc/invoke.texi (RISC-V): Document "-mrelax" and "-mno-relax".
17843 2018-03-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
17846 * config/rs6000/rs6000.c (rs6000_reassociation_width): Disable parallel
17847 reassociation for int modes.
17849 2018-03-13 Richard Sandiford <richard.sandiford@linaro.org>
17851 * tree-vect-loop-manip.c (vect_maybe_permute_loop_masks):
17852 Reverse the choice between VEC_UNPACK_LO_EXPR and VEC_UNPACK_HI_EXPR
17854 * config/aarch64/iterators.md (hi_lanes_optab): New int attribute.
17855 * config/aarch64/aarch64-sve.md
17856 (*aarch64_sve_<perm_insn><perm_hilo><mode>): Rename to...
17857 (aarch64_sve_<perm_insn><perm_hilo><mode>): ...this.
17858 (*extend<mode><Vwide>2): Rename to...
17859 (aarch64_sve_extend<mode><Vwide>2): ...this.
17860 (vec_unpack<su>_<perm_hilo>_<mode>): Turn into a define_expand,
17861 renaming the old pattern to...
17862 (aarch64_sve_punpk<perm_hilo>_<mode>): ...this. Only define
17864 (vec_unpack<su>_<perm_hilo>_<SVE_BHSI:mode>): Turn into a
17865 define_expand, renaming the old pattern to...
17866 (aarch64_sve_<su>unpk<perm_hilo>_<SVE_BHSI:mode>): ...this.
17867 (*vec_unpacku_<perm_hilo>_<mode>_no_convert): Delete.
17868 (vec_unpacks_<perm_hilo>_<mode>): Take BYTES_BIG_ENDIAN into
17869 account when deciding which SVE instruction the optab should use.
17870 (vec_unpack<su_optab>_float_<perm_hilo>_vnx4si): Likewise.
17872 2018-03-13 Richard Sandiford <richard.sandiford@linaro.org>
17874 * config/aarch64/aarch64.md (V4_REGNUM, V8_REGNUM, V12_REGNUM)
17875 (V20_REGNUM, V24_REGNUM, V28_REGNUM, P1_REGNUM, P2_REGNUM, P3_REGNUM)
17876 (P4_REGNUM, P5_REGNUM, P6_REGNUM, P8_REGNUM, P9_REGNUM, P10_REGNUM)
17877 (P11_REGNUM, P12_REGNUM, P13_REGNUM, P14_REGNUM): New define_constants.
17878 (tlsdesc_small_<mode>): Turn a define_expand and use
17879 tlsdesc_small_sve_<mode> for SVE. Rename original define_insn to...
17880 (tlsdesc_small_advsimd_<mode>): ...this.
17881 (tlsdesc_small_sve_<mode>): New pattern.
17883 2018-03-13 Richard Sandiford <richard.sandiford@linaro.org>
17885 * config/aarch64/iterators.md (UNSPEC_SMUL_HIGHPART)
17886 (UNSPEC_UMUL_HIGHPART): New constants.
17887 (MUL_HIGHPART): New int iteraor.
17888 (su): Handle UNSPEC_SMUL_HIGHPART and UNSPEC_UMUL_HIGHPART.
17889 * config/aarch64/aarch64-sve.md (<su>mul<mode>3_highpart): New
17891 (*<su>mul<mode>3_highpart): New define_insn.
17893 2018-03-13 Eric Botcazou <ebotcazou@adacore.com>
17896 * ipa-devirt.c (odr_subtypes_equivalent_p): Do not get the ODR type of
17899 2018-03-13 Martin Liska <mliska@suse.cz>
17902 * (sem_item_optimizer::sem_item_optimizer): Initialize new
17904 (sem_item_optimizer::~sem_item_optimizer): Release it.
17905 (sem_item_optimizer::merge_classes): Register variable aliases.
17906 (sem_item_optimizer::fixup_pt_set): New function.
17907 (sem_item_optimizer::fixup_points_to_sets): Likewise.
17908 * ipa-icf.h: Declare new variables and functions.
17910 2018-03-13 Jakub Jelinek <jakub@redhat.com>
17912 PR middle-end/84834
17913 * match.pd ((A & C) != 0 ? D : 0): Use INTEGER_CST@2 instead of
17914 integer_pow2p@2 and test integer_pow2p in condition.
17915 (A < 0 ? C : 0): Similarly for @1.
17917 PR middle-end/84831
17918 * stmt.c (parse_output_constraint): If the CONSTRAINT_LEN (*p, p)
17919 characters starting at p contain '\0' character, don't look beyond
17923 * config/i386/i386.md (round<mode>2): For 387 fancy math, disable
17924 pattern if -ftrapping-math -fno-fp-int-builtin-inexact.
17927 * reg-stack.c (change_stack): Change update_end var from int to
17928 rtx_insn *, if non-NULL don't update just BB_END (current_block), but
17929 also call set_block_for_insn on the newly added insns and rescan.
17932 * config/i386/sse.md (sse2_loadhpd): Use Yv constraint rather than v
17933 on the last operand.
17936 * tree.c (stabilize_reference_1): Return save_expr (e) for
17937 STATEMENT_LIST even if it doesn't have side-effects.
17939 2018-03-12 Jonathan Wakely <jwakely@redhat.com>
17941 * doc/invoke.texi (-mclflushopt): Fix spelling of option.
17943 2018-03-12 Renlin Li <renlin.li@arm.com>
17945 * config/aarch64/aarch64.md (movhf_aarch64): Fix mode argument to
17946 aarch64_output_scalar_simd_mov_immediate.
17948 2018-03-12 Martin Sebor <msebor@redhat.com>
17950 PR tree-optimization/83456
17951 * gimple-fold.c (gimple_fold_builtin_memory_op): Avoid warning
17952 for perfectly overlapping calls to memcpy.
17953 (gimple_fold_builtin_memory_chk): Same.
17954 (gimple_fold_builtin_strcpy): Handle no-warning.
17955 (gimple_fold_builtin_stxcpy_chk): Same.
17956 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Handle no-warning.
17958 2018-03-12 Segher Boessenkool <segher@kernel.crashing.org>
17960 * config/rs6000/rs6000.c (abi_v4_pass_in_fpr): Add bool "named"
17961 parameter. Use it for SFmode.
17962 (rs6000_function_arg_advance_1): Adjust.
17963 (rs6000_function_arg): Adjust.
17964 (rs6000_gimplify_va_arg): Pass false for that new parameter.
17966 2018-03-12 Segher Boessenkool <segher@kernel.crashing.org>
17968 PR rtl-optimization/84169
17969 PR rtl-optimization/84780
17970 * combine.c (can_combine_p): Check for a 2-insn combination whether
17971 the destination register is used between the two insns, too.
17973 2018-03-12 Richard Biener <rguenther@suse.de>
17975 PR tree-optimization/84803
17976 * tree-if-conv.c (ifcvt_memrefs_wont_trap): Don't do anything
17977 for refs DR analysis didn't process.
17979 2018-03-12 Richard Biener <rguenther@suse.de>
17981 PR tree-optimization/84777
17982 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): For
17983 force-vectorize loops ignore whether we are optimizing for size.
17985 2018-03-12 Chung-Ju Wu <jasonwucj@gmail.com>
17987 * config/nds32/nds32.c (nds32_md_asm_adjust): New function.
17988 (TARGET_MD_ASM_ADJUST): Define.
17990 2018-03-12 Monk Chiang <sh.chiang04@gmail.com>
17991 Kito Cheng <kito.cheng@gmail.com>
17992 Chung-Ju Wu <jasonwucj@gmail.com>
17994 * config/nds32/nds32.c (nds32_compute_stack_frame,
17995 nds32_emit_stack_push_multiple, nds32_emit_stack_pop_multiple,
17996 nds32_emit_stack_v3push, nds32_emit_stack_v3pop,
17997 nds32_emit_adjust_frame, nds32_expand_prologue, nds32_expand_epilogue,
17998 nds32_expand_prologue_v3push, nds32_expand_epilogue_v3pop): Refine.
17999 * config/nds32/nds32.h (NDS32_FIRST_CALLEE_SAVE_GPR_REGNUM,
18000 NDS32_LAST_CALLEE_SAVE_GPR_REGNUM, NDS32_V3PUSH_AVAILABLE_P): New.
18001 * config/nds32/nds32.md (prologue, epilogue): Use macro
18002 NDS32_V3PUSH_AVAILABLE_P to do checking.
18004 2018-03-11 Jakub Jelinek <jakub@redhat.com>
18007 * dwarf2out.c (gen_enumeration_type_die): Don't guard adding
18008 DW_AT_declaration for ENUM_IS_OPAQUE on -gdwarf-4 or -gno-strict-dwarf,
18009 but on TYPE_SIZE. Don't do anything for ENUM_IS_OPAQUE if not creating
18010 a new die. Don't set TREE_ASM_WRITTEN if ENUM_IS_OPAQUE. Guard
18011 addition of most attributes on !orig_type_die or the attribute not
18012 being present already. Assert TYPE_VALUES is NULL for ENUM_IS_OPAQUE.
18014 2018-03-11 Kito Cheng <kito.cheng@gmail.com>
18015 Chung-Ju Wu <jasonwucj@gmail.com>
18017 * config/nds32/nds32.c (nds32_cpu_cpp_builtins): Modify to define
18018 __NDS32_VH__ macro.
18019 * config/nds32/nds32.opt (mvh): New option.
18021 2018-03-11 Kito Cheng <kito.cheng@gmail.com>
18022 Chung-Ju Wu <jasonwucj@gmail.com>
18024 * config/nds32/nds32-protos.h (nds32_cpu_cpp_builtins): Declare
18026 * config/nds32/nds32.c (nds32_cpu_cpp_builtins): New function.
18027 * config/nds32/nds32.h (TARGET_CPU_CPP_BUILTINS): Modify its
18030 2018-03-11 Kito Cheng <kito.cheng@gmail.com>
18031 Chung-Ju Wu <jasonwucj@gmail.com>
18033 * config/nds32/nds32-memory-manipulation.c (nds32_expand_strlen): New
18035 * config/nds32/nds32-multiple.md (strlensi): New pattern.
18036 * config/nds32/nds32-protos.h (nds32_expand_strlen): Declare function.
18038 2018-03-11 Monk Chiang <sh.chiang04@gmail.com>
18039 Kito Cheng <kito.cheng@gmail.com>
18040 Chung-Ju Wu <jasonwucj@gmail.com>
18042 * config/nds32/constants.md (unspec_element): Add UNSPEC_FFB,
18043 UNSPEC_FFMISM and UNSPEC_FLMISM.
18044 * config/nds32/nds32-intrinsic.c (bdesc_2arg): Add builtin description
18045 for ffb, ffmism and flmism.
18046 * config/nds32/nds32-intrinsic.md (unspec_ffb): Define new pattern.
18047 (unspec_ffmism): Ditto.
18048 (unspec_flmism): Ditto.
18049 (nds32_expand_builtin_impl): Check if string extension is available.
18050 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_FFB,
18051 NDS32_BUILTIN_FFMISM and NDS32_BUILTIN_FLMISM.
18053 2018-03-10 Vladimir Makarov <vmakarov@redhat.com>
18056 2018-03-09 Vladimir Makarov <vmakarov@redhat.com>
18059 * lra-assigns.c (assign_by_spills): Return a flag of reload
18060 assignment failure. Do not process the reload assignment
18061 failures. Do not spill other reload pseudos if they has the same
18063 (lra_assign): Add a return arg. Set up from the result of
18064 assign_by_spills call.
18065 (find_reload_regno_insns, lra_split_hard_reg_for): New functions.
18066 * lra-constraints.c (split_reg): Add a new arg. Use it instead of
18067 usage_insns if it is not NULL.
18068 (spill_hard_reg_in_range): New function.
18069 (split_if_necessary, inherit_in_ebb): Pass a new arg to split_reg.
18070 * lra-int.h (spill_hard_reg_in_range, lra_split_hard_reg_for): New
18071 function prototypes.
18072 (lra_assign): Change prototype.
18073 * lra.c (lra): Add code to deal with fails by splitting hard reg
18076 2018-03-10 H.J. Lu <hongjiu.lu@intel.com>
18079 * config/i386/i386.opt: Replace Enforcment with Enforcement.
18081 2018-03-10 Alexandre Oliva <aoliva@redhat.com>
18084 * dwarf2out.h (dw_val_class): Add dw_val_class_symview.
18085 (dw_val_node): Add val_symbolic_view.
18086 * dwarf2out.c (dw_line_info_table): Add symviews_since_reset.
18087 (symview_upper_bound): New.
18088 (new_line_info_table): Initialize symviews_since_reset.
18089 (dwarf2out_source_line): Count symviews_since_reset and set
18090 symview_upper_bound.
18091 (dw_val_equal_p): Handle symview.
18092 (add_AT_symview): New.
18093 (print_dw_val): Handle symview.
18094 (attr_checksum, attr_checksum_ordered): Likewise.
18095 (same_dw_val_p, size_of_die): Likewise.
18096 (value_format, output_die): Likewise.
18097 (add_high_low_attributes): Use add_AT_symview for entry_view.
18098 (dwarf2out_finish): Reset symview_upper_bound, clear
18101 2018-03-09 Peter Bergner <bergner@vnet.ibm.com>
18104 * config/rs6000/rs6000.c (rs6000_offsettable_memref_p): New prototype.
18105 Add strict argument and use it.
18106 (rs6000_split_multireg_move): Update for new strict argument.
18107 (mem_operand_gpr): Disallow all non-offsettable addresses.
18108 * config/rs6000/rs6000.md (*movdi_internal64): Use YZ constraint.
18110 2018-03-09 Jakub Jelinek <jakub@redhat.com>
18113 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Mark va_arg_tmp
18114 temporary TREE_ADDRESSABLE before gimplification of BUILT_IN_MEMCPY.
18115 * config/powerpcspe/powerpcspe.c (rs6000_gimplify_va_arg): Likewise.
18118 * tree-inline.c (copy_tree_body_r): For INDIRECT_REF of a remapped
18119 decl, use remap_type if we want to use the type.
18121 2018-03-09 Martin Sebor <msebor@redhat.com>
18123 PR tree-optimization/84526
18124 * gimple-ssa-warn-restrict.c (builtin_memref::set_base_and_offset):
18126 (builtin_access::generic_overlap): Be prepared to handle non-array
18129 2018-03-09 Alexandre Oliva <aoliva@redhat.com>
18131 PR rtl-optimization/84682
18132 * lra-constraints.c (process_address_1): Check is_address flag
18133 for address constraints.
18134 (process_alt_operands): Likewise.
18135 * lra.c (lra_set_insn_recog_data): Pass asm operand locs to
18136 preprocess_constraints.
18137 * recog.h (preprocess_constraints): Add oploc parameter.
18139 * recog.c (preprocess_constraints): Test address_operand for
18140 CT_ADDRESS constraints.
18142 2018-03-09 Vladimir Makarov <vmakarov@redhat.com>
18145 * lra-assigns.c (assign_by_spills): Return a flag of reload
18146 assignment failure. Do not process the reload assignment
18147 failures. Do not spill other reload pseudos if they has the same
18149 (lra_assign): Add a return arg. Set up from the result of
18150 assign_by_spills call.
18151 (find_reload_regno_insns, lra_split_hard_reg_for): New functions.
18152 * lra-constraints.c (split_reg): Add a new arg. Use it instead of
18153 usage_insns if it is not NULL.
18154 (spill_hard_reg_in_range): New function.
18155 (split_if_necessary, inherit_in_ebb): Pass a new arg to split_reg.
18156 * lra-int.h (spill_hard_reg_in_range, lra_split_hard_reg_for): New
18157 function prototypes.
18158 (lra_assign): Change prototype.
18159 * lra.c (lra): Add code to deal with fails by splitting hard reg
18162 2018-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18165 * common/config/arm/arm-common.c (arm_parse_arch_option_name):
18166 Accept complain bool parameter. Only emit errors if it is true.
18167 (arm_parse_cpu_option_name): Likewise.
18168 (arm_target_thumb_only): Adjust callers of the above.
18169 * config/arm/arm-protos.h (arm_parse_cpu_option_name): Adjust
18170 prototype to take a default true bool parameter.
18171 (arm_parse_arch_option_name): Likewise.
18173 2018-03-09 David Malcolm <dmalcolm@redhat.com>
18174 Francois-Xavier Coudert <fxcoudert@gcc.gnu.org>
18178 * Makefile.in (LD_VERSION_SCRIPT_OPTION, LD_SONAME_OPTION): New.
18179 * configure: Regenerate.
18180 * configure.ac ("linker --version-script option"): New.
18181 ("linker soname option"): New.
18183 2018-03-09 Richard Biener <rguenther@suse.de>
18185 PR tree-optimization/84775
18186 * tree-if-conv.c (add_bb_predicate_gimplified_stmts): Delink
18187 immediate uses of predicate stmts and mark them modified.
18190 PR tree-optimization/84178
18191 * tree-if-conv.c (combine_blocks): Move insert_gimplified_predicates
18193 (version_loop_for_if_conversion): Delay update_ssa call.
18194 (tree_if_conversion): Delay update_ssa until after predicate
18197 2018-03-09 Eric Botcazou <ebotcazou@adacore.com>
18200 * config/i386/winnt.c (i386_pe_seh_cold_init): Use small pre-allocation
18201 when the function accesses prior frames.
18203 2018-03-08 Jakub Jelinek <jakub@redhat.com>
18206 * dwarf2out.c (dw_loc_list): If list && loc_list->first->next, call
18207 gen_llsym, otherwise call maybe_gen_llsym.
18209 PR inline-asm/84742
18210 * recog.c (asm_operand_ok): Return 0 if multi-character constraint
18211 has ',' character inside of it.
18213 2018-03-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18216 * config/aarch64/aarch64.md (*compare_cstore<mode>_insn): Mark pattern
18217 as clobbering CC_REGNUM.
18219 2018-03-08 Richard Biener <rguenther@suse.de>
18221 PR middle-end/84552
18222 * tree-scalar-evolution.c: Include tree-into-ssa.h.
18223 (follow_copies_to_constant): Do not follow SSA names registered
18226 2018-03-08 Richard Biener <rguenther@suse.de>
18228 PR tree-optimization/84178
18229 * tree-if-conv.c (combine_blocks): Move insert_gimplified_predicates
18231 (version_loop_for_if_conversion): Delay update_ssa call.
18232 (tree_if_conversion): Delay update_ssa until after predicate
18235 2018-03-08 David Malcolm <dmalcolm@redhat.com>
18237 PR tree-optimization/84178
18238 * tree-if-conv.c (release_bb_predicate): Remove the
18239 the assertion that the stmts have NULL use_ops.
18240 Discard the statements, asserting that they haven't
18241 yet been added to a BB.
18243 2018-03-08 Richard Biener <rguenther@suse.de>
18245 PR tree-optimization/84746
18246 * tree-ssa-pre.c (find_leader_in_sets): Deal with SET1 being NULL.
18247 (phi_translate): Pass in destination ANTIC_OUT set.
18248 (phi_translate_1): Likewise. For a simplified result lookup
18249 a leader in ANTIC_OUT and AVAIL_OUT, not the ANTIC_IN sets.
18250 (phi_translate_set): Adjust.
18251 (do_pre_regular_insertion): Likewise.
18252 (do_pre_partial_partial_insertion): Likewise.
18254 2018-03-08 Martin Liska <mliska@suse.cz>
18256 PR gcov-profile/84735
18257 * doc/gcov.texi: Document usage of profile files.
18258 * gcov-io.h: Document changes in the format.
18260 2018-03-08 Alexandre Oliva <aoliva@redhat.com>
18264 * dwarf2out.c (struct dw_line_info_table): Update comments for
18266 (FORCE_RESET_NEXT_VIEW): New.
18267 (FORCE_RESETTING_VIEW_P): New.
18268 (RESETTING_VIEW_P): Check for -1 too.
18269 (ZERO_VIEW_P): Likewise.
18270 (new_line_info_table): Force-reset next view.
18271 (dwarf2out_begin_function): Likewise.
18272 (dwarf2out_source_line): Simplify zero_view_p initialization.
18273 Test FORCE_RESETTING_VIEW_P and RESETTING_VIEW_P instead of
18274 view directly. Omit view when omitting .loc at line 0.
18276 2018-03-08 Jakub Jelinek <jakub@redhat.com>
18278 PR tree-optimization/84740
18279 * tree-switch-conversion.c (process_switch): Call build_constructors
18280 only if info.phi_count is non-zero.
18282 PR tree-optimization/84739
18283 * tree-tailcall.c (find_tail_calls): Check call arguments against
18284 DECL_ARGUMENTS (current_function_decl) rather than
18285 DECL_ARGUMENTS (func) when checking for tail recursion.
18287 2018-03-07 Jakub Jelinek <jakub@redhat.com>
18289 * doc/contrib.texi: Add entries for Martin Liska, David Malcolm,
18290 Marek Polacek, extend Vladimir Makarov's, Jonathan Wakely's and
18291 Volker Reichelt's entry and add entries for people that perform
18292 GCC fuzzy testing and report numerous bugs.
18294 2018-03-07 Segher Boessenkool <segher@kernel.crashing.org>
18297 * config/rs6000/rs6000.c (rs6000_elf_in_small_data_p): Don't put
18298 readonly data in sdata, if that is disabled.
18299 * config/rs6000/sysv4.opt (mreadonly-in-sdata): New option.
18300 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
18301 -mreadonly-in-sdata option.
18303 2018-03-07 Martin Sebor <msebor@redhat.com>
18305 PR tree-optimization/84468
18306 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Consider successor
18307 basic block when looking for nul assignment.
18309 2018-03-07 Eric Botcazou <ebotcazou@adacore.com>
18312 * except.h (output_function_exception_table): Adjust prototype.
18313 * except.c (output_function_exception_table): Remove FNNAME parameter
18314 and add SECTION parameter. Ouput one part of the table at a time.
18315 * final.c (final_scan_insn_1) <NOTE_INSN_SWITCH_TEXT_SECTIONS>: Output
18316 the first part of the exception table and emit unwind directives.
18317 * config/i386/i386-protos.h (i386_pe_end_cold_function): Declare.
18318 (i386_pe_seh_cold_init): Likewise.
18319 * config/i386/cygming.h (ASM_DECLARE_COLD_FUNCTION_NAME): New macro.
18320 (ASM_DECLARE_COLD_FUNCTION_SIZE): Likewise.
18321 * config/i386/i386.c (x86_expand_epilogue): Fix wording in comment.
18322 (ix86_output_call_insn): Emit a nop in one more case for SEH.
18323 * config/i386/winnt.c: Include except.h.
18324 (struct seh_frame_state): Add reg_offset, after_prologue and
18325 in_cold_section fields.
18326 (i386_pe_seh_end_prologue): Set seh->after_prologue.
18327 (i386_pe_seh_cold_init): New function.
18328 (i386_pe_seh_fini): Add COLD parameter and bail out if it is not equal
18329 to seh->in_cold_section.
18330 (seh_emit_push): Record the offset of the push.
18331 (seh_emit_save): Record the offet of the save.
18332 (i386_pe_seh_unwind_emit): Deal with NOTE_INSN_SWITCH_TEXT_SECTIONS.
18333 Test seh->after_prologue to disregard the epilogue.
18334 (i386_pe_end_function): Pass FALSE to i386_pe_seh_fini.
18335 (i386_pe_end_cold_function): New function.
18337 2018-03-07 Jakub Jelinek <jakub@redhat.com>
18340 * config/aarch64/predicates.md (aarch64_simd_reg_or_zero): Use
18341 aarch64_simd_or_scalar_imm_zero rather than aarch64_simd_imm_zero.
18344 * gimple-expr.c (create_tmp_var_raw): Set DECL_NAMELESS flag
18346 * tree-pretty-print.c (dump_decl_name): For TDF_COMPARE_DEBUG,
18347 don't print names of DECL_NAMELESS DECL_IGNORED_P decls.
18349 PR middle-end/84723
18350 * multiple_target.c: Include tree-inline.h and intl.h.
18351 (expand_target_clones): Diagnose and fail if node->definition and
18352 !tree_versionable_function_p (node->decl).
18354 2018-03-06 John David Anglin <danglin@gcc.gnu.org>
18356 * config/pa/pa.h (ASM_GENERATE_INTERNAL_LABEL): Revise to use
18358 (ASM_OUTPUT_ADDR_VEC_ELT): Revise for above change.
18359 (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
18360 * config/pa/pa64-hpux.h (ASM_GENERATE_INTERNAL_LABEL): Revise as above.
18362 2018-03-06 Jakub Jelinek <jakub@redhat.com>
18365 * combine.c (try_combine): Use reg_or_subregno instead of handling
18366 just paradoxical SUBREGs and REGs.
18368 2018-03-06 Claudiu Zissulescu <claziss@synopsys.com>
18370 * config/arc/arc.c (arc_finalize_pic): Remove function.
18371 (arc_must_save_register): We use single base PIC register, remove
18372 checks to save/restore the PIC register.
18373 (arc_expand_prologue): Likewise.
18374 * config/arc/arc-protos.h (arc_set_default_type_attributes):
18376 (arc_verify_short): Likewise.
18377 (arc_attr_type): Likewise.
18378 * config/arc/arc.c (arc_set_default_type_attributes): Remove.
18379 (walk_stores): Likewise.
18380 (arc_address_cost): Make it static.
18381 (arc_verify_short): Likewise.
18382 (branch_dest): Likewise.
18383 (arc_attr_type): Likewise.
18384 * config/arc/arc.c (TARGET_ADJUST_INSN_LENGTH): Remove.
18385 (TARGET_INSN_LENGTH_PARAMETERS): Likewise.
18386 (arc_final_prescan_insn): Remove inserting the nops due to
18387 hardware hazards. It is done in reorg step.
18388 (insn_length_variant_t): Remove.
18389 (insn_length_parameters_t): Likewise.
18390 (arc_insn_length_parameters): Likewise.
18391 (arc_get_insn_variants): Likewise.
18392 * config/arc/arc.h (TARGET_UPSIZE_DBR): Remove.
18394 2018-03-06 Jakub Jelinek <jakub@redhat.com>
18396 PR inline-asm/84683
18397 * reg-stack.c (move_for_stack_reg): If any_malformed_asm, avoid
18400 PR tree-optimization/84687
18401 * omp-simd-clone.c (simd_clone_create): Clear DECL_BUILT_IN_CLASS
18403 * match.pd (pow(C,x)*expN(y) -> expN(logN(C)*x+y)): New optimization.
18405 2018-03-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18407 * config/rs6000/rs6000-builtin.def (rs6000_speculation_barrier):
18408 Rename to ppc_speculation_barrier.
18409 * config/rs6000/rs6000.c (rs6000_init_builtins): Rename builtin to
18410 __builtin_ppc_speculation_barrier.
18412 2018-03-05 Jakub Jelinek <jakub@redhat.com>
18415 * combine.c (combine_simplify_rtx): Don't try to simplify if
18416 if_then_else_cond returned non-NULL, but either true_rtx or false_rtx
18419 2018-03-05 Segher Boessenkool <segher@kernel.crashing.org>
18421 * config/rs6000/rs6000.c (rs6000_loop_align): Don't align tiny loops
18422 to 32 bytes when compiling for POWER9.
18424 2018-03-05 Jakub Jelinek <jakub@redhat.com>
18427 * config/i386/i386.c (ix86_function_ok_for_sibcall): Check for
18428 regparm >= 3 with no arg reg available also for calls with
18429 flag_force_indirect_call. Pass decl to ix86_function_regparm.
18432 * config/i386/sse.md (*<code><mode>3): Replace <mask_prefix3> with
18434 (*<plusminus_insn><mode>3): Likewise. Remove <mask_operand3> uses.
18436 2018-03-05 Peter Bergner <bergner@vnet.ibm.com>
18439 * config/rs6000/vector.md (mov<mode>): Disallow altivec memory operands.
18441 2018-03-05 Richard Biener <rguenther@suse.de>
18443 PR tree-optimization/84486
18444 * tree-ssa-pre.c (create_expression_by_pieces): Remove dead code.
18445 When inserting a __builtin_assume_aligned call set the LHS
18446 SSA name alignment info accordingly.
18448 2018-03-05 Wilco Dijkstra <wdijkstr@arm.com>
18450 PR tree-optimization/84114
18451 * config/aarch64/aarch64.c (aarch64_reassociation_width)
18452 Avoid reassociation of FLOAT_MODE addition.
18454 2018-03-05 Olga Makhotina <olga.makhotina@intel.com>
18456 * common/config/i386/i386-common.c (OPTION_MASK_ISA_PCONFIG_SET,
18457 OPTION_MASK_ISA_PCONFIG_UNSET, OPTION_MASK_ISA_WBNOINVD_SET,
18458 OPTION_MASK_ISA_WBNOINVD_UNSET): New definitions.
18459 (ix86_handle_option): Handle -mpconfig and -mwbnoinvd.
18460 * config.gcc (pconfigintrin.h, wbnoinvdintrin.h) : Add headers.
18461 * config/i386/cpuid.h (bit_PCONFIG, bit_WBNOINVD): New.
18462 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mpconfig
18464 * config/i386/i386-builtin.def (__builtin_ia32_wbnoinvd,
18465 __builtin_ia32_wbinvd): New builtins.
18466 (SPECIAL_ARGS2): New.
18467 * config/i386/i386-c.c (__WBNOINVD__, __PCONFIG__): New.
18468 (SPECIAL_ARGS2): New.
18469 * config/i386/i386.c (ix86_target_string): Add -mpconfig and -mwbnoinvd.
18470 (ix86_valid_target_attribute_inner_p): Ditto.
18471 (ix86_init_mmx_sse_builtins): Add special_args2.
18472 * config/i386/i386.h (TARGET_PCONFIG, TARGET_PCONFIG_P, TARGET_WBNOINVD,
18473 TARGET_WBNOINVD_P): New.
18474 * config/i386/i386.md (UNSPECV_WBINVD, UNSPECV_WBNOINVD): New.
18475 (define_insn "wbinvd", define_insn "wbnoinvd"): New.
18476 * config/i386/i386.opt: Add -mpconfig and -mwbnoinvd.
18477 * config/i386/immintrin.h (_wbinvd): New intrinsic.
18478 * config/i386/pconfigintrin.h: New file.
18479 * config/i386/wbnoinvdintrin.h: Ditto.
18480 * config/i386/x86intrin.h: Add headers pconfigintrin.h and
18482 * doc/invoke.texi (-mpconfig, -mwbnoinvd): New.
18484 2018-03-05 Richard Biener <rguenther@suse.de>
18486 PR tree-optimization/84670
18487 * tree-ssa-pre.c (struct bb_bitmap_sets): Add visited_with_visited_succs
18489 (BB_VISITED_WITH_VISITED_SUCCS): New define.
18490 (compute_antic): Initialize BB_VISITED_WITH_VISITED_SUCCS.
18491 (compute_antic_aux): Only assert the number of values in ANTIC_IN
18492 doesn't grow if all successors (recursively) were visited at least
18495 2018-03-05 Richard Biener <rguenther@suse.de>
18497 PR tree-optimization/84650
18498 * tree-ssa-loop-im.c (pass_lim::execute): Reset the SCEV cache
18499 if executed in the loop pipeline.
18501 2018-03-05 Sandra Loosemore <sandra@codesourcery.com>
18503 * doc/configfiles.texi (Configuration Files): Move info about
18504 conditionalizing $target-protos.h to...
18505 * doc/sourcebuild.texi (Back End): Here. Explain how $target.h
18506 differs from $target-protos.h.
18508 2018-03-05 Kito Cheng <kito.cheng@gmail.com>
18509 Chung-Ju Wu <jasonwucj@gmail.com>
18511 * config/nds32/nds32-protos.h (nds32_expand_setmem): Declare.
18512 * config/nds32/nds32-multiple.md (setmemsi): Define.
18513 * config/nds32/nds32-memory-manipulation.c
18514 (nds32_gen_dup_4_byte_to_word_value): New.
18515 (emit_setmem_word_loop): New.
18516 (emit_setmem_byte_loop): New.
18517 (nds32_expand_setmem_loop): New.
18518 (nds32_expand_setmem_loop_v3m): New.
18519 (nds32_expand_setmem_unroll): New.
18520 (nds32_expand_setmem): New.
18522 2018-03-04 Kito Cheng <kito.cheng@gmail.com>
18523 Chung-Ju Wu <jasonwucj@gmail.com>
18525 * config/nds32/nds32-memory-manipulation.c
18526 (nds32_emit_load_store): New.
18527 (nds32_emit_post_inc_load_store): New.
18528 (nds32_emit_mem_move): New.
18529 (nds32_emit_mem_move_block): New.
18530 (nds32_expand_movmemsi_loop_unknown_size): New.
18531 (nds32_expand_movmemsi_loop_known_size): New.
18532 (nds32_expand_movmemsi_loop): New.
18533 (nds32_expand_movmemsi_unroll): New.
18534 (nds32_expand_movmemqi): Rename ...
18535 (nds32_expand_movmemsi): ... to this.
18536 * config/nds32/nds32-multiple.md (movmemqi): Rename ...
18537 (movmemsi): ... to this.
18538 * config/nds32/nds32-protos.h (nds32_expand_movmemqi): Rename ...
18539 (nds32_expand_movmemsi): ... to this.
18541 2018-03-04 Kito Cheng <kito.cheng@gmail.com>
18542 Monk Chiang <sh.chiang04@gmail.com>
18543 Chung-Ju Wu <jasonwucj@gmail.com>
18545 * config/nds32/nds32-protos.h
18546 (nds32_expand_load_multiple): New arguments.
18547 (nds32_expand_store_multiple): Ditto.
18548 (nds32_valid_multiple_load_store): Rename ...
18549 (nds32_valid_multiple_load_store_p): ... to this.
18550 * config/nds32/nds32-memory-manipulation.c
18551 (nds32_expand_load_multiple): Refine implementation.
18552 (nds32_expand_store_multiple): Ditto.
18553 * config/nds32/nds32-multiple.md
18554 (load_multiple): Update nds32_expand_load_multiple interface.
18555 (store_multiple): Update nds32_expand_store_multiple interface.
18556 * config/nds32/nds32-predicates.c
18557 (nds32_valid_multiple_load_store): Rename ...
18558 (nds32_valid_multiple_load_store_p): ... to this and refine
18560 * config/nds32/predicates.md
18561 (nds32_load_multiple_and_update_address_operation): New predicate.
18562 (nds32_store_multiple_and_update_address_operation): New predicate.
18564 2018-03-04 Kito Cheng <kito.cheng@gmail.com>
18565 Chung-Ju Wu <jasonwucj@gmail.com>
18567 * config/nds32/nds32.md (type): Add load_multiple and store_multiple.
18568 (combo): New attribute.
18569 * config/nds32/nds32-multiple.md: Refine patterns with new attributes.
18571 2018-03-03 Chung-Ju Wu <jasonwucj@gmail.com>
18573 * config/nds32/nds32.opt: Change -mcmodel= default value.
18575 2018-03-03 Kito Cheng <kito.cheng@gmail.com>
18576 Monk Chiang <sh.chiang04@gmail.com>
18577 Chung-Ju Wu <jasonwucj@gmail.com>
18579 * config/nds32/constants.md (unspec_element): New enum.
18580 * config/nds32/constraints.md (Umw): New constraint.
18581 * config/nds32/nds32-intrinsic.c: Add more builtin functions.
18582 * config/nds32/nds32-intrinsic.md: Likewise.
18583 * config/nds32/nds32-md-auxiliary.c (nds32_regno_to_enable4): New.
18584 (nds32_valid_smw_lwm_base_p): New.
18585 (nds32_output_smw_single_word): New.
18586 (nds32_output_lmw_single_word): New.
18587 (nds32_expand_unaligned_load): New.
18588 (nds32_expand_unaligned_store): New.
18589 * config/nds32/nds32-protos.h (nds32_valid_smw_lwm_base_p): Declare.
18590 (nds32_output_smw_single_word): Declare.
18591 (nds32_output_lmw_single_word): Declare.
18592 (nds32_expand_unaligned_load): Declare.
18593 (nds32_expand_unaligned_store): Declare.
18594 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_UALOAD_HW,
18595 NDS32_BUILTIN_UALOAD_W, NDS32_BUILTIN_UALOAD_DW,
18596 NDS32_BUILTIN_UASTORE_HW, NDS32_BUILTIN_UASTORE_W,
18597 NDS32_BUILTIN_UASTORE_DW.
18598 * config/nds32/predicates.md (nds32_lmw_smw_base_operand): New
18601 2018-03-03 Monk Chiang <sh.chiang04@gmail.com>
18602 Kito Cheng <kito.cheng@gmail.com>
18603 Chung-Ju Wu <jasonwucj@gmail.com>
18605 * config/nds32/nds32-intrinsic.c
18606 (nds32_expand_builtin_null_ftype_reg): Delete.
18607 (nds32_expand_builtin_reg_ftype_imm): Ditto.
18608 (nds32_expand_builtin_null_ftype_reg_imm): Ditto.
18609 (nds32_read_argument): New.
18610 (nds32_legitimize_target): Ditto.
18611 (nds32_legitimize_argument): Ditto.
18612 (nds32_check_constant_argument): Ditto.
18613 (nds32_expand_unop_builtin): Ditto.
18614 (nds32_expand_unopimm_builtin): Ditto.
18615 (nds32_expand_binop_builtin): Ditto.
18616 (nds32_builtin_decl_impl): Ditto.
18617 (builtin_description): Ditto.
18618 (nds32_expand_builtin_impl): Rewrite with new infrastructure.
18619 (nds32_init_builtins_impl): Ditto.
18620 * config/nds32/nds32.c (TARGET_BUILTIN_DECL): Define.
18621 (nds32_builtin_decl): New.
18622 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_COUNT.
18623 * config/nds32/nds32-protos.h (nds32_builtin_decl_impl): Declare.
18625 2018-03-02 Jeff Law <law@redhat.com>
18627 * reorg.c (stop_search_p): Handle DEBUG_INSN.
18628 (redundant_insn, fill_simple_delay_slots): Likewise.
18629 (fill_slots_from_thread): Likewise.
18630 * resource.c (mark_referenced_resources): Likewise.
18631 (mark_set_resources, find_dead_or_set_registers): Likewise.
18633 2018-03-02 Jakub Jelinek <jakub@redhat.com>
18635 * substring-locations.h (format_warning_va): Formatting fix for
18636 ATTRIBUTE_GCC_DIAG.
18637 (format_warning_at_substring): Fix up ATTRIBUTE_GCC_DIAG second
18639 (format_warning_n_va, format_warning_at_substring_n): New prototypes.
18640 * substring-locations.c: Include intl.h.
18641 (format_warning_va): Turned into small wrapper around
18642 format_warning_n_va, renamed to ...
18643 (format_warning_n_va): ... this, add N and PLURAL_GMSGID arguments,
18644 rename GMSGID to SINGULAR_GMSGID, if SINGULAR_GMSGID != PLURAL_GMSGID,
18646 (format_warning_at_substring_n): New function.
18647 * gimple-ssa-sprintf.c: Remove GCC diagnostic ignored pragma.
18648 (fmtwarn): Add ATTRIBUTE_GCC_DIAG. Turn into a copy of
18649 format_warning_at_substring with just a shorter name instead of
18650 const function pointer.
18651 (fmtwarn_n): New function.
18652 (maybe_warn, format_directive, parse_directive): Use fmtwarn_n where
18653 appropriate, get rid of all the fmtstr temporaries, move conditionals
18654 with G_() wrapped string literals directly into fmtwarn arguments,
18655 cast dir.len to (int), formatting fixes.
18657 2018-03-02 Thomas Schwinge <thomas@codesourcery.com>
18659 * doc/invoke.texi: Remove "Cilk Plus" references.
18661 2018-03-02 Jakub Jelinek <jakub@redhat.com>
18662 Richard Biener <rguenther@suse.de>
18665 * expr.c (expand_expr_real_1) <case CALL_EXPR>: Don't emit diagnostics
18666 for error or warning attributes if CALL_FROM_THUNK_P is set.
18669 2018-03-02 Jakub Jelinek <jakub@redhat.com>
18672 * config/pa/pa.h (TARGET_CPU_CPP_BUILTINS): Predefine
18673 __SIZEOF_128__ macro if HPUX_LONG_DOUBLE_LIBRARY.
18676 * config/ia64/ia64.h (TARGET_CPU_CPP_BUILTINS): Predefine
18677 __SIZEOF_{FPREG,FLOAT{80,128}}__ macros.
18679 * predict.c (test_prediction_value_range): Use PROB_UNINITIALIZED
18680 instead of -1U in last predictors element's probability member.
18682 2018-03-02 Eric Botcazou <ebotcazou@adacore.com>
18685 * ipa-devirt.c (odr_subtypes_equivalent_p): Get the ODR type of both
18686 arguments if they are comparable.
18688 2018-03-02 Richard Sandiford <richard.sandiford@linaro.org>
18690 PR tree-optimization/84634
18691 * tree-vect-stmts.c (vectorizable_store, vectorizable_load): Replace
18692 masks and masked_loop_p with a single loop_masks, making sure it's
18693 null for bb vectorization.
18695 2018-03-02 Richard Sandiford <richard.sandiford@linaro.org>
18697 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence)
18698 (vect_analyze_data_ref_access): Use loop->safe_len rather than
18699 loop->force_vectorize to check whether there is no alias.
18701 2018-03-02 Jakub Jelinek <jakub@redhat.com>
18704 * rtl.h (prev_real_nondebug_insn, next_real_nondebug_insn): New
18706 * emit-rtl.c (next_real_insn, prev_real_insn): Fix up function
18708 (next_real_nondebug_insn, prev_real_nondebug_insn): New functions.
18709 * cfgcleanup.c (try_head_merge_bb): Use prev_real_nondebug_insn
18710 instead of a loop around prev_real_insn.
18711 * combine.c (move_deaths): Use prev_real_nondebug_insn instead of
18714 PR inline-asm/84625
18715 * config/i386/i386.c (ix86_print_operand): Use conditional
18716 output_operand_lossage instead of gcc_assert if CONST_VECTOR is not
18719 2018-03-02 Richard Biener <rguenther@suse.de>
18721 PR tree-optimization/84427
18722 * tree-ssa-pre.c (bitmap_remove_expr_from_set): Remove.
18723 (bitmap_set_subtract_values): Rewrite to handle multiple
18726 (prune_clobbered_mems): Likewise.
18727 (phi_translate): Take edge instead of pred/phiblock.
18728 (phi_translate_1): Likewise.
18729 (phi_translate_set): Likewise. Insert all translated
18730 exprs for a value into the set, keeping possibly multiple
18731 expressions per value.
18732 (compute_antic_aux): Adjust for phi_translate changes.
18733 When intersecting union the expressions and prune those
18734 not in the final value set, keeping possibly multiple
18735 expressions per value. Do not use value-insertion
18736 for unioning ANTIC_OUT U EXP_GEN - TMP_GEN but merge
18737 all expressions. Add verification that the value-sets
18738 only shrink during iteration.
18739 (compute_partial_antic_aux): Adjust for the phi_translate changes.
18740 (do_pre_regular_insertion): Likewise.
18741 (do_pre_partial_partial_insertion): Likewise.
18743 2018-03-02 Richard Biener <rguenther@suse.de>
18746 * config/darwin.c (saved_debug_info_level): New static global.
18747 (darwin_asm_lto_start): Disable debug info generation for LTO out.
18748 (darwin_asm_lto_end): Restore debug info generation settings.
18750 2018-03-01 Martin Liska <mliska@suse.cz>
18753 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
18754 volatile arguments.
18756 2018-03-01 Richard Biener <rguenther@suse.de>
18759 * dwarf2out.c (gen_variable_die): Properly handle late VLA
18760 type annotation with LTO when debug was disabled at compile-time.
18762 2018-03-01 Matthew Fortune <mfortune@gmail.com>
18764 * config/mips/mips.c (mips_final_prescan_insn): Fix incorrect
18766 (mips_final_postscan_insn): Likewise.
18768 2018-03-01 Richard Sandiford <richard.sandiford@linaro.org>
18770 PR rtl-optimization/84528
18771 * alias.c (init_alias_target): Add commentary.
18772 (init_alias_analysis): Only give HARD_FRAME_POINTER_REGNUM
18773 a unique base value if the frame pointer is not eliminated
18774 to the stack pointer.
18776 2018-03-01 Tom de Vries <tom@codesourcery.com>
18778 PR rtl-optimization/83327
18779 * lra-int.h (hard_regs_spilled_into): Declare.
18780 * lra.c (hard_regs_spilled_into): Define.
18781 (init_reg_info): Init hard_regs_spilled_into.
18782 * lra-spills.c (assign_spill_hard_regs): Update hard_regs_spilled_into.
18783 * lra-lives.c (make_hard_regno_born, make_hard_regno_dead)
18784 (process_bb_lives): Handle hard_regs_spilled_into.
18785 (lra_create_live_ranges_1): Before doing liveness propagation, clear
18786 regs in all_hard_regs_bitmap if set in hard_regs_spilled_into.
18788 2018-02-28 David Edelsohn <dje.gcc@gmail.com>
18790 * config.gcc (powerpc-ibm-aix7.1.*): New stanza.
18791 (powerpc-ibm-aix[789]*): Default to AIX 7.2.
18792 * config/rs6000/aix71.h (TARGET_DEFAULT): Revert to Power4 ISA.
18793 * config/rs6000/aix72.h: New file.
18795 2018-02-28 Jakub Jelinek <jakub@redhat.com>
18797 * gimple-ssa-warn-restrict.c (maybe_diag_overlap): Use warning_n
18798 instead of warning_at with conditional singular and plural messages
18802 * stor-layout.c (update_alignment_for_field): For
18803 targetm.ms_bitfield_layout_p (rli->t), if !is_bitfield
18804 && !DECL_PACKED (field), do the alignment update, just use
18805 only desired_align instead of MAX (type_align, desired_align)
18807 (place_field): Don't do known_align < desired_align handling
18808 early if targetm.ms_bitfield_layout_p (rli->t) and rli->prev_field
18809 is non-NULL, instead do it after rli->prev_field handling and
18810 only if not within a bitfield word. For DECL_PACKED (field)
18811 use type_align of BITS_PER_UNIT.
18813 2018-02-28 Eric Botcazou <ebotcazou@adacore.com>
18815 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Remove
18816 superfluous parentheses and trailing spaces.
18818 2018-02-28 Richard Biener <rguenther@suse.de>
18820 PR tree-optimization/84584
18821 * graphite-scop-detection.c (scop_detection::add_scop): Discard
18822 SCoPs with fake exit edge.
18824 2018-02-28 Martin Liska <mliska@suse.cz>
18827 * timevar.c (timer::print): Fix format to properly print 100%
18830 2018-02-28 Richard Biener <rguenther@suse.de>
18832 PR middle-end/84607
18833 * genmatch.c (capture_info::walk_match): Do not mark
18834 captured expressions without operands as expr_p given
18835 they act more like predicates and should be subject to
18836 "lost tail" side-effect preserving.
18838 2018-02-28 Alexandre Oliva <aoliva@redhat.com>
18840 PR rtl-optimization/81611
18841 * auto-inc-dec.c (attempt_change): Move dead note from
18842 mem_insn if it's the next use of regno
18843 (find_address): Take address use of reg holding
18844 non-incremented value. Add parm to limit search to the named
18846 (merge_in_block): Attempt to use a mem insn that is the next
18847 use of the original regno.
18849 2018-02-27 Martin Sebor <msebor@redhat.com>
18852 * doc/invoke.texi (-Wmissing-attributes): New option.
18853 * print-tree.c (print_node): Handle DECL_UNINLINABLE.
18855 2018-02-27 Martin Sebor <msebor@redhat.com>
18857 PR translation/84207
18858 * diagnostic-core.h (warning_n, error_n, inform_n): Change
18859 n argument to unsigned HOST_WIDE_INT.
18860 * diagnostic.c (warning_n, error_n, inform_n): Ditto.
18861 (diagnostic_n_impl): Ditto. Handle arguments in excess of LONG_MAX.
18862 * gimple-ssa-sprintf.c (format_directive): Simplify inform_n call.
18863 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Use warning_n.
18865 2018-02-27 Richard Biener <rguenther@suse.de>
18867 PR tree-optimization/84512
18868 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost):
18869 Do not use the estimate returned from record_stmt_cost for
18870 the scalar iteration cost but sum properly using add_stmt_cost.
18872 2018-02-27 Richard Biener <rguenther@suse.de>
18874 PR tree-optimization/84466
18875 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
18876 Adjust last change to less strictly validate use operands.
18878 2018-02-27 Martin Liska <mliska@suse.cz>
18880 PR gcov-profile/84548
18881 * gcov.c (process_file): Allow partial overlap and consider it
18882 also as group functions.
18883 (output_lines): Properly calculate range of lines for a group.
18885 2018-02-27 Martin Liska <mliska@suse.cz>
18887 * timevar.c (timer::print_row): Remove 'usr', 'sys', 'wall' and
18888 'ggc' suffixes. Change first column width.
18889 (timer::print): Fix formatting of the column.
18891 2018-02-27 Alexandre Oliva <aoliva@redhat.com>
18893 * tree-ssa-live.c (remove_unused_scope_block_p): Do not
18894 preserve inline entry blocks for the sake of debug inline
18895 entry point markers alone.
18896 (remove_unused_locals): Suggest in comments a better place to
18897 force the preservation of inline entry blocks that are
18898 otherwise unused, but do not preserve them.
18900 2018-02-26 H.J. Lu <hongjiu.lu@intel.com>
18902 * config/i386/i386.c (ix86_output_indirect_jmp): Update comments.
18904 2018-02-26 H.J. Lu <hongjiu.lu@intel.com>
18907 * config/i386/constraints.md (Bs): Replace
18908 ix86_indirect_branch_register with
18909 TARGET_INDIRECT_BRANCH_REGISTER.
18911 * config/i386/i386.md (indirect_jump): Likewise.
18912 (tablejump): Likewise.
18913 (*sibcall_memory): Likewise.
18914 (*sibcall_value_memory): Likewise.
18915 Peepholes of indirect call and jump via memory: Likewise.
18916 (*sibcall_GOT_32): Disallowed for TARGET_INDIRECT_BRANCH_REGISTER.
18917 (*sibcall_value_GOT_32): Likewise.
18918 * config/i386/predicates.md (indirect_branch_operand): Likewise.
18919 (GOT_memory_operand): Likewise.
18920 (call_insn_operand): Likewise.
18921 (sibcall_insn_operand): Likewise.
18922 (GOT32_symbol_operand): Likewise.
18923 * config/i386/i386.h (TARGET_INDIRECT_BRANCH_REGISTER): New.
18925 2018-02-26 Eric Botcazou <ebotcazou@adacore.com>
18927 PR rtl-optimization/83496
18928 * reorg.c (steal_delay_list_from_target): Change REDUNDANT array from
18929 booleans to RTXes. Call fix_reg_dead_note on every non-null element.
18930 (steal_delay_list_from_fallthrough): Call fix_reg_dead_note on a
18931 redundant insn, if any.
18932 (relax_delay_slots): Likewise.
18933 (update_reg_unused_notes): Rename REDUNDANT_INSN to OTHER_INSN.
18935 2018-02-26 Richard Sandiford <richard.sandiford@linaro.org>
18937 PR tree-optimization/83965
18938 * tree-vect-patterns.c (vect_reassociating_reduction_p): Assume
18939 that grouped statements are part of a reduction chain. Return
18940 true if the statement is not marked as a reduction itself but
18941 is part of a group.
18942 (vect_recog_dot_prod_pattern): Don't check whether the statement
18943 is part of a group here.
18944 (vect_recog_sad_pattern): Likewise.
18945 (vect_recog_widen_sum_pattern): Likewise.
18947 2018-02-26 Eric Botcazou <ebotcazou@adacore.com>
18950 * final.c (rest_of_clean_state): Also look for calls inside sequences.
18952 2018-02-26 H.J. Lu <hongjiu.lu@intel.com>
18955 * config/i386/i386-protos.h (ix86_output_indirect_jmp): Remove
18957 (ix86_output_indirect_function_return): New prototype.
18958 (ix86_split_simple_return_pop_internal): Likewise.
18959 * config/i386/i386.c (indirect_return_via_cx): New.
18960 (indirect_return_via_cx_bnd): Likewise.
18961 (indirect_thunk_name): Handle return va CX_REG.
18962 (output_indirect_thunk_function): Create alias for
18963 __x86_return_thunk_[re]cx and __x86_return_thunk_[re]cx_bnd.
18964 (ix86_output_indirect_jmp): Remove the bool argument.
18965 (ix86_output_indirect_function_return): New function.
18966 (ix86_split_simple_return_pop_internal): Likewise.
18967 * config/i386/i386.md (*indirect_jump): Don't pass false
18968 to ix86_output_indirect_jmp.
18969 (*tablejump_1): Likewise.
18970 (simple_return_pop_internal): Change it to define_insn_and_split.
18971 Call ix86_split_simple_return_pop_internal to split it for
18972 -mfunction-return=.
18973 (simple_return_indirect_internal): Call
18974 ix86_output_indirect_function_return instead of
18975 ix86_output_indirect_jmp.
18977 2018-02-26 Jakub Jelinek <jakub@redhat.com>
18980 * vec.h (vec_default_construct): For BROKEN_VALUE_INITIALIZATION use
18981 memset and value initialization afterwards.
18983 2018-02-26 Christophe Lyon <christophe.lyon@linaro.org>
18985 * Makefile.in (lto-wrapper): Use ALL_LINKERFLAGS.
18987 2018-02-26 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
18990 * common/config/aarch64/aarch64-common.c
18991 (aarch_option_optimization_table[]): Switch
18992 off fomit-frame-pointer
18994 2018-02-26 Kito Cheng <kito.cheng@gmail.com>
18995 Chung-Ju Wu <jasonwucj@gmail.com>
18997 * config/nds32/nds32-multiple.md (load_multiple): Disallow
18999 (store_multiple): Ditto.
19001 2018-02-26 Kito Cheng <kito.cheng@gmail.com>
19003 * config.gcc: Add --with-cpu support for nds32 target.
19004 * config/nds32/nds32-opts.h (nds32_cpu_type): New.
19005 * config/nds32/nds32.opt: Add -mcpu= option.
19007 2018-02-25 Segher Boessenkool <segher@kernel.crashing.org>
19009 * config/rs6000/rs6000.opt (mvrsave=no, mvrsave=yes, isel=no,
19010 isel=yes): Warn for these deprecated options.
19012 2018-02-23 David Edelsohn <dje.gcc@gmail.com>
19014 * config/rs6000/aix71.h (TARGET_DEFAULT): Change to
19015 ISA_2_5_MASKS_EMBEDDED.
19017 2018-02-23 Jakub Jelinek <jakub@redhat.com>
19019 * ipa-prop.c (ipa_vr_ggc_hash_traits::hash): Hash p->min and
19020 p->max as pointers rather than using iterative_hash_expr.
19022 2018-02-23 Carl Love <cel@us.ibm.com>
19024 * config/rs6000/rs6000-builtin.def: Change VSIGNED2 and VUNSIGNED2
19025 macro expansions from BU_VSX_2 to BU_P8V_VSX_2 and BU_VSX_OVERLOAD_2 to
19027 * config/rs6000/rs6000-c.c: Change VSX_BUILTIN_VEC_VSIGNED2 to
19028 P8V_BUILTIN_VEC_VSIGNED2. Change VSX_BUILTIN_VEC_VUNSIGNED2 to
19029 P8V_BUILTIN_VEC_VUNSIGNED2.
19031 2018-02-22 Vladimir Makarov <vmakarov@redhat.com>
19034 * lra-int.h (LRA_UNKNOWN_ALT, LRA_NON_CLOBBERED_ALT): New macros.
19035 * lra.c (lra_set_insn_recog_data, lra_update_insn_recog_data): Use
19037 * lra-constraints.c (curr_insn_transform): Set up
19038 LRA_NON_CLOBBERED_ALT for moves processed on the fast path. Use
19040 (remove_inheritance_pseudos): Use LRA_UNKNOWN_ALT.
19041 * lra-eliminations.c (spill_pseudos): Ditto.
19042 (process_insn_for_elimination): Ditto.
19043 * lra-lives.c (reg_early_clobber_p): Use the new macros.
19044 * lra-spills.c (spill_pseudos): Use LRA_UNKNOWN_ALT and
19045 LRA_NON_CLOBBERED_ALT.
19047 2018-02-22 Martin Sebor <msebor@redhat.com>
19049 PR tree-optimization/84480
19050 * gimple-fold.c (gimple_fold_builtin_strcpy): Move warnings
19051 to maybe_diag_stxncpy_trunc. Call it.
19052 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Integrate warnings
19053 from gimple_fold_builtin_strcpy. Print inlining stack.
19054 (handle_builtin_stxncpy): Print inlining stack.
19055 * tree-ssa-strlen.h (maybe_diag_stxncpy_trunc): Declare.
19057 2018-02-22 H.J. Lu <hongjiu.lu@intel.com>
19060 * config/i386/i386.c (ix86_set_indirect_branch_type): Issue an
19061 error when -mindirect-branch=thunk-extern, -fcf-protection=branch
19062 and -fcheck-pointer-bounds are used together.
19063 (indirect_thunk_prefix): New enum.
19064 (indirect_thunk_need_prefix): New function.
19065 (indirect_thunk_name): Replace need_bnd_p with need_prefix. Use
19066 "_nt" instead of "_bnd" for NOTRACK prefix.
19067 (output_indirect_thunk): Replace need_bnd_p with need_prefix.
19068 (output_indirect_thunk_function): Likewise.
19070 (ix86_code_end): Update output_indirect_thunk_function calls.
19071 (ix86_output_indirect_branch_via_reg): Replace
19072 ix86_bnd_prefixed_insn_p with indirect_thunk_need_prefix.
19073 (ix86_output_indirect_branch_via_push): Likewise.
19074 (ix86_output_function_return): Likewise.
19075 * doc/invoke.texi: Document -mindirect-branch=thunk-extern is
19076 incompatible with -fcf-protection=branch and
19077 -fcheck-pointer-bounds.
19079 2018-02-22 Steve Ellcey <sellcey@cavium.com>
19082 * config/aarch64/aarch64.c (aarch64_print_address_internal):
19083 Change gcc_assert call to output_operand_lossage.
19085 2018-02-22 Steve Ellcey <sellcey@cavium.com>
19087 * doc/extend.texi (__builtin_extend_pointer): Document builtin.
19089 2018-02-22 DJ Delorie <dj@redhat.com>
19090 Sebastian Perta <sebastian.perta@renesas.com>
19091 Oleg Endo <olegendo@gcc.gnu.org>
19093 * config/rx/rx.c (rx_rtx_costs): New function.
19094 (TARGET_RTX_COSTS): Override to use rx_rtx_costs.
19096 2018-02-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
19098 * config/arm/t-multilib: Map Armv8-R to Armv7 multilibs.
19100 2018-02-22 Martin Liska <mliska@suse.cz>
19103 * common/config/arm/arm-common.c (arm_print_hint_for_cpu_option):
19104 Add "native" as a possible value.
19106 2018-02-22 Martin Liska <mliska@suse.cz>
19109 * config/i386/i386.c (ix86_option_override_internal):
19110 Add "native" as a possible value for -march and -mtune.
19112 2018-02-22 Jakub Jelinek <jakub@redhat.com>
19115 * stor-layout.c (finalize_type_size): Propagate TYPE_EMPTY_P flag
19116 to all type variants.
19118 PR tree-optimization/84503
19119 * gimple-ssa-store-merging.c (merged_store_group::merge_into): Compute
19120 width as info->bitpos + info->bitsize - start.
19121 (merged_store_group::merge_overlapping): Simplify width computation.
19122 (check_no_overlap): New function.
19123 (imm_store_chain_info::try_coalesce_bswap): Compute expected
19124 start + width and last_order of the group, fail if check_no_overlap
19126 (imm_store_chain_info::coalesce_immediate_stores): Don't merge info
19127 to group if check_no_overlap fails.
19129 2018-02-21 Segher Boessenkool <segher@kernel.crashing.org>
19131 * config/rs6000/altivec.md: Delete contraint arguments to
19132 define_expand, define_split, and define_peephole2, and in
19133 define_insn_and_split if always unused.
19134 * config/rs6000/darwin.md: Ditto.
19135 * config/rs6000/dfp.md: Ditto.
19136 * config/rs6000/rs6000.md: Ditto.
19137 * config/rs6000/sync.md: Ditto.
19138 * config/rs6000/vector.md: Ditto.
19139 * config/rs6000/vsx.md: Ditto.
19141 2018-02-21 Segher Boessenkool <segher@kernel.crashing.org>
19143 * config/rs6000/altivec.md: Write output control strings as braced
19144 blocks instead of double-quoted strings.
19145 * config/rs6000/darwin.md: Ditto.
19146 * config/rs6000/rs6000.md: Ditto.
19147 * config/rs6000/vector.md: Ditto.
19148 * config/rs6000/vsx.md: Ditto.
19150 2018-02-21 Jason Merrill <jason@redhat.com>
19152 PR c++/84314 - ICE with templates and fastcall attribute.
19153 * attribs.c (build_type_attribute_qual_variant): Remove assert.
19155 2018-02-21 Jan Hubicka <hubicka@ucw.cz>
19157 * ipa-cp.c (determine_versionability): Fix comment typos.
19159 2018-02-21 Jan Hubicka <hubicka@ucw.cz>
19162 * ipa-cp.c (determine_versionability): Do not version functions caling
19165 2018-02-21 Martin Liska <mliska@suse.cz>
19168 * config/aarch64/aarch64.c (aarch64_print_hint_for_core_or_arch):
19169 Add "native" as a possible value.
19170 * config/aarch64/aarch64.h (HAVE_LOCAL_CPU_DETECT): Define
19171 the macro when native cpu detection is available.
19173 2018-02-21 Martin Liska <mliska@suse.cz>
19176 * common/config/arm/arm-common.c (arm_print_hint_for_arch_option):
19177 Add "native" as a possible value.
19178 * config/arm/arm.h (HAVE_LOCAL_CPU_DETECT): Define the macro
19179 when native cpu detection is available.
19181 2018-02-21 Jakub Jelinek <jakub@redhat.com>
19182 Martin Sebor <msebor@redhat.com>
19184 PR tree-optimization/84478
19185 * gimple-fold.h (get_range_strlen): Add a bool argument defaulted to
19187 * gimple-fold.c (get_range_strlen): Make minlen const and assume it
19188 can't be NULL. Change FUZZY from bool to int, for 1 add PHI/COND_EXPR
19189 support which is conservatively correct, for 2 only stay conservative
19190 for maxlen. Formatting and comment capitalization fixes. Add STRICT
19191 argument to the 2 argument get_range_strlen, adjust 6 arg
19192 get_range_strlen caller and clear minmaxlen[0] and [1] if it returned
19194 (get_maxval_strlen): Adjust 6 arg get_range_strlen caller.
19195 (gimple_fold_builtin_strlen): Pass true as last argument to
19198 2018-02-20 Martin Sebor <msebor@redhat.com>
19200 PR middle-end/84095
19201 * gimple-ssa-warn-restrict.c (builtin_memref::extend_offset_range): New.
19202 (builtin_memref::set_base_and_offset): Same. Handle inner references.
19203 (builtin_memref::builtin_memref): Factor out parts into
19204 set_base_and_offset and call it.
19206 2018-02-20 Richard Sandiford <richard.sandiford@linaro.org>
19208 PR middle-end/84406
19209 * optabs-query.c (find_widening_optab_handler_and_mode): If from_mode
19210 is a scalar_int_mode, assert that to_mode is a scalar_int_mode with
19211 greater precision. If to_mode is a MODE_PARTIAL_INT, stop the
19212 search at the associated MODE_INT.
19214 2018-02-20 Jeff Law <law@redhat.com>
19216 PR middle-end/82123
19217 PR tree-optimization/81592
19218 PR middle-end/79257
19219 * gimple-ssa-sprintf.c (format_integer): Query EVRP range analyzer
19220 for range data rather than using global data.
19221 * gimple-ssa-sprintf.c (get_int_range): Query EVRP range analyzer for
19222 range data rather than using global data.
19223 * gimple-ssa-sprintf.c (get_int_range): Accept vr_values parameter
19224 pass it to children as needed.
19225 (struct directive::fmtresult): Similarly.
19226 (struct directive::set_width): Similarly.
19227 (struct directive::set_precision): Similarly.
19228 (format_integer, format_directive, parse_directive): Similarly.
19229 (format_none): Accept unnamed vr_values parameter.
19230 (format_percent, format_floating, format_character): Similarly.
19231 (format_string, format_plain): Similarly.
19232 * gimple-ssa-sprintf.c (sprintf_dom_walker::handle_gimple_call): Query
19233 the EVRP range analyzer for range data rather than using global data.
19234 * gimple-ssa-sprintf.c: Include alloc-pool.h, vr-values.h and
19235 gimple-ssa-evrp-analyze.h
19236 (class sprintf_dom_walker): Add after_dom_children member function.
19237 Add evrp_range_analyzer member.
19238 (sprintf_dom_walker::before_dom_children): Call into the EVRP
19239 range analyzer as needed.
19240 (sprintf_dom_walker::after_dom_children): New member function.
19241 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::enter): Do nothing
19243 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
19244 (evrp_range_analyzer::pop_to_marker): Likewise.
19246 2018-02-20 Richard Sandiford <richard.sandiford@linaro.org>
19248 PR tree-optimization/84419
19249 * internal-fn.c (expand_call_mem_ref): Create a TARGET_MEM_REF
19250 with the required type if its current type is compatible but
19253 2018-02-20 Jakub Jelinek <jakub@redhat.com>
19255 PR middle-end/82004
19256 * match.pd (pow(C,x) -> exp(log(C)*x)): Delay all folding until
19257 after vectorization.
19259 2018-02-20 Martin Liska <mliska@suse.cz>
19262 * config/aarch64/aarch64.c (aarch64_print_hint_for_core_or_arch): Print
19263 possible values if we don't have a hint.
19265 2018-02-20 Martin Liska <mliska@suse.cz>
19269 * final.c (shorten_branches): Build align_tab array with one
19271 * opts.c (finish_options): Add alignment option limit check.
19272 (MAX_CODE_ALIGN): Likewise.
19273 (MAX_CODE_ALIGN_VALUE): Likewise.
19274 * doc/invoke.texi: Document maximum allowed option value for
19275 all -falign-* options.
19277 2018-02-19 Jakub Jelinek <jakub@redhat.com>
19280 * reg-notes.def (REG_CALL_ARG_LOCATION): New reg note.
19281 * insn-notes.def (NOTE_INSN_CALL_ARG_LOCATION): Remove.
19282 * var-tracking.c (emit_note_insn_var_location): Remove all references
19283 to NOTE_INSN_CALL_ARG_LOCATION.
19284 (emit_notes_in_bb): Emit arguments as REG_CALL_ARG_LOCATION note on
19285 the CALL_INSN rather than separate NOTE_INSN_CALL_ARG_LOCATION note.
19286 Use copy_rtx_if_shared.
19287 * dwarf2out.c (gen_subprogram_die): Use XEXP with 0 instead of
19288 NOTE_VAR_LOCATION on ca_loc->call_arg_loc_note.
19289 (dwarf2out_var_location): Remove handling of
19290 NOTE_INSN_CALL_ARG_LOCATION, instead handle REG_CALL_ARG_LOCATION note
19292 * final.c (final_scan_insn): Remove all references to
19293 NOTE_INSN_CALL_ARG_LOCATION.
19294 (rest_of_clean_state): Likewise. Remove REG_CALL_ARG_LOCATION notes
19295 before dumping final insns.
19296 * except.c (emit_note_eh_region_end): Remove all references to
19297 NOTE_INSN_CALL_ARG_LOCATION.
19298 * config/alpha/alpha.c (alpha_pad_function_end): Likewise.
19299 * config/c6x/c6x.c (c6x_gen_bundles): Likewise.
19300 * config/arc/arc.c (hwloop_optimize): Likewise.
19301 * config/arm/arm.c (create_fix_barrier): Likewise.
19302 * config/s390/s390.c (s390_chunkify_start): Likewise.
19303 * config/sh/sh.c (find_barrier): Likewise.
19304 * config/i386/i386.c (rest_of_insert_endbranch,
19305 ix86_seh_fixup_eh_fallthru): Likewise.
19306 * config/xtensa/xtensa.c (hwloop_optimize): Likewise.
19307 * config/iq2000/iq2000.c (final_prescan_insn): Likewise.
19308 * config/frv/frv.c (frv_function_prologue): Likewise.
19309 * emit-rtl.c (try_split): Likewise. Copy over REG_CALL_ARG_LOCATION
19311 (note_outside_basic_block_p): Remove all references to
19312 NOTE_INSN_CALL_ARG_LOCATION.
19313 * gengtype.c (adjust_field_rtx_def): Likewise.
19314 * print-rtl.c (rtx_writer::print_rtx_operand_code_0, print_insn):
19316 * jump.c (cleanup_barriers, delete_related_insns): Likewise.
19317 * cfgrtl.c (force_nonfallthru_and_redirect): Likewise.
19320 * builtins.c (builtin_mathfn_code): Don't check if CALL_EXPR_FN (t)
19323 PR tree-optimization/84452
19324 * tree-vect-patterns.c (vect_recog_pow_pattern): Don't call
19325 expand_simd_clones if targetm.simd_clone.compute_vecsize_and_simdlen
19328 2018-02-19 Martin Liska <mliska@suse.cz>
19331 * passes.def: Put pass_sancov_O0 before pass_lower_switch with -O0.
19333 2018-02-19 Martin Liska <mliska@suse.cz>
19334 Richard Sandiford <richard.sandiford@linaro.org>
19336 PR tree-optimization/82491
19337 * gimple-fold.c (get_base_constructor): Make earlier bail out
19340 2018-02-19 Carl Love <cel@us.ibm.com>
19342 * config/rs6000/rs6000-builtin.def: Change NEG macro expansions from
19343 BU_ALTIVEC_A to BU_P8V_AV_1 and BU_ALTIVEC_OVERLOAD_1 to
19345 * config/rs6000/rs6000-c.c: Change ALTIVEC_BUILTIN_VEC_NEG to
19346 P8V_BUILTIN_VEC_NEG.
19348 2018-02-19 Sebastian Perta <sebastian.perta@renesas.com>
19350 * config/rl78/rl78.md (movdf): New define expand.
19352 2018-02-19 Martin Liska <mliska@suse.cz>
19355 * doc/invoke.texi: Fix typo.
19356 * params.def (PARAM_MAX_LOOP_HEADER_INSNS): Likewise.
19358 2018-02-18 Segher Boessenkool <segher@kernel.crashing.org>
19360 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
19361 handle rs6000_single_float and rs6000_double_float specially for
19364 2018-02-16 Jeff Law <law@redhat.com>
19366 * config/rx/rx.c (add_pop_cfi_notes): New function.;
19367 (pop_regs): Use it.
19369 2018-02-16 Jakub Jelinek <jakub@redhat.com>
19372 * ipa-inline.c (inline_small_functions): Fix a typo.
19374 2018-02-16 Nathan Sidwell <nathan@acm.org>
19376 * doc/extend.texi (Backwards Compatibility): Americanize 'behaviour'.
19378 2018-02-16 Carl Love <cel@us.ibm.com>
19380 * config/rs6000/rs6000-builtin.def: Add BU_P8V_VSX_2 macro definition.
19381 Change FLOAT2 expansions from BU_VSX_2 to BU_P8V_VSX_2 and
19382 from BU_VSX_OVERLOAD_2 to BU_P8V_OVERLOAD_2.
19383 * config/rs6000/rs6000-c.c: Changed macro VSX_BUILTIN_VEC_FLOAT2
19384 expansion to P8V_BUILTIN_VEC_FLOAT2.
19386 2018-02-16 Vladimir Makarov <vmakarov@redhat.com>
19388 PR rtl-optimization/70023
19389 * lra-constraints.c (inherit_in_ebb): Take hard reg mode of
19390 src_regno into account.
19392 2018-02-16 Carl Love <cel@us.ibm.com>
19394 * config/rs6000/altivec.h: Remove vec_vextract4b and vec_vinsert4b.
19395 * config/rs6000/rs6000-builtin.def: Remove macro expansion for
19396 VEXTRACT4B, VINSERT4B, VINSERT4B_DI and VEXTRACT4B.
19397 * config/rs6000/rs6000.c: Remove case statements for
19398 P9V_BUILTIN_VEXTRACT4B, P9V_BUILTIN_VEC_VEXTRACT4B,
19399 P9V_BUILTIN_VINSERT4B, P9V_BUILTIN_VINSERT4B_DI,
19400 and P9V_BUILTIN_VEC_VINSERT4B.
19401 * config/rs6000/rs6000-c.c (altivec_expand_builtin): Remove entries for
19402 P9V_BUILTIN_VEC_VEXTRACT4B and P9V_BUILTIN_VEC_VINSERT4B.
19403 * config/rs6000/vsx.md:
19404 * doc/extend.texi: Remove vec_vextract4b, non ABI definitions for
19407 2018-02-16 Carl Love <cel@us.ibm.com>
19409 * config/rs6000/altivec.h: Add builtin names vec_extract4b
19411 * config/rs6000/rs6000-builtin.def: Add INSERT4B and EXTRACT4B
19413 * config/rs6000/rs6000-c.c: Add the definitions for
19414 P9V_BUILTIN_VEC_EXTRACT4B and P9V_BUILTIN_VEC_INSERT4B.
19415 * config/rs6000/rs6000.c (altivec_expand_builtin): Add
19416 P9V_BUILTIN_EXTRACT4B and P9V_BUILTIN_INSERT4B case statements.
19417 * config/rs6000/vsx.md: Add define_insn extract4b. Add define_expand
19418 definition for insert4b and define insn *insert3b_internal.
19419 * doc/extend.texi: Add documentation for vec_extract4b.
19421 2018-02-16 Nathan Sidwell <nathan@acm.org>
19423 * doc/extend.texi (Backwards Compatibility): Mention friend
19424 injection. Note for-scope is deprecated.
19425 * doc/invoke.texi (-ffriend-injection): Deprecate.
19427 2018-02-16 Segher Boessenkool <segher@kernel.crashing.org>
19429 * combine.c (try_combine): When adjusting LOG_LINKS for the destination
19430 that moved to I2, also allow destinations that are a paradoxical
19431 subreg (instead of a normal reg).
19433 2018-02-16 Oleg Endo <olegendo@gcc.gnu.org>
19436 * config/rx/rx.c (rx_fuse_in_memory_bitop): Convert shift operand
19439 2018-02-16 Richard Biener <rguenther@suse.de>
19441 PR tree-optimization/84037
19442 PR tree-optimization/84016
19444 * config/i386/i386.c (ix86_builtin_vectorization_cost):
19445 Adjust vec_construct for the fact we need additional higher latency
19446 128bit inserts for AVX256 and AVX512 vector builds.
19447 (ix86_add_stmt_cost): Scale vector construction cost for
19450 2018-02-16 Richard Biener <rguenther@suse.de>
19452 PR tree-optimization/84417
19453 * tree-ssa.c (non_rewritable_mem_ref_base): Properly constrain
19454 the MEM_REF offset when conversion to BIT_FIELD_REF is desired.
19455 (non_rewritable_lvalue_p): Likewise, use poly-ints.
19457 2018-02-16 Martin Liska <mliska@suse.cz>
19460 * internal-fn.def (ASAN_CHECK): Set proper flags.
19461 (ASAN_MARK): Likewise.
19463 2018-02-16 Julia Koval <julia.koval@intel.com>
19465 * config/i386/i386.c (ix86_option_override_internal): Remove PTA_CLWB
19466 from PTA_CANNONLAKE.
19468 2018-02-16 Jakub Jelinek <jakub@redhat.com>
19471 * config/aarch64/cortex-a57-fma-steering.c (fma_forest::merge_forest):
19472 Use ++iter rather than iter++ for std::list iterators.
19473 (func_fma_steering::dfs): Likewise. Don't delete nodes right away,
19474 defer deleting them until all nodes in the forest are processed. Do
19475 free even leaf nodes. Change to_process into auto_vec.
19478 * system.h (BROKEN_VALUE_INITIALIZATION): Define for GCC < 4.3.
19479 * vec.h (vec_default_construct): Use memset instead of placement new
19480 if BROKEN_VALUE_INITIALIZATION is defined.
19481 * hash-table.h (hash_table<Descriptor, Allocator>::empty_slow): Use
19482 memset instead of value initialization if BROKEN_VALUE_INITIALIZATION
19485 PR rtl-optimization/83723
19486 * lra-int.h (lra_substitute_pseudo): Add DEBUG_P argument.
19487 * lra.c (lra_substitute_pseudo): Likewise. If true, use
19488 gen_rtx_raw_SUBREG instead of gen_rtx_SUBREG. Pass DEBUG_P to
19490 (lra_substitute_pseudo_within_insn): Adjust lra_substitute_pseudo
19492 * lra-constraints.c (inherit_reload_reg, split_reg): Likewise.
19494 2018-02-16 Eric Botcazou <ebotcazou@adacore.com>
19496 PR rtl-optimization/81443
19497 * rtlanal.c (num_sign_bit_copies1) <SUBREG>: Do not propagate results
19498 from inner REGs to paradoxical SUBREGs.
19500 2018-02-16 Richard Biener <rguenther@suse.de>
19502 PR tree-optimization/84399
19503 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
19504 For operands we can analyze at their definition make sure we can
19505 analyze them at each use as well.
19507 2018-02-16 Richard Biener <rguenther@suse.de>
19509 PR tree-optimization/84190
19510 * tree-ssa.c (non_rewritable_mem_ref_base): Do not touch
19511 volatile accesses if the decl isn't volatile.
19513 2018-02-15 Jason Merrill <jason@redhat.com>
19515 PR c++/84314 - ICE with templates and fastcall attribute.
19516 * attribs.c (build_type_attribute_qual_variant): Don't clobber
19517 TYPE_CANONICAL on an existing type.
19519 2018-02-15 Jakub Jelinek <jakub@redhat.com>
19521 PR tree-optimization/84383
19522 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Don't look at
19523 dstoff nor call operand_equal_p if dstbase is NULL.
19525 PR tree-optimization/84334
19526 * match.pd ((A +- CST1) +- CST2 -> A + CST3): If A is
19527 also a CONSTANT_CLASS_P, punt.
19529 2018-02-14 Jim Wilson <jimw@sifive.com>
19531 * config/riscv/riscv.c (riscv_first_stack_step): Move locals after
19532 first SMALL_OPERAND check. New local min_second_step. Move assert
19533 to where locals are set. Add TARGET_RVC support.
19534 * config/riscv/riscv.h (C_SxSP_BITS, SWSP_REACH, SDSP_REACH): New.
19536 2018-02-14 Indu Bhagat <indu.bhagat@oracle.com>
19538 * doc/invoke.texi: Correct -Wformat-overflow code sample.
19540 2018-02-14 Martin Sebor <msebor@redhat.com>
19542 PR tree-optimization/83698
19543 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): For
19544 arrays constrain the offset range to their bounds.
19545 (builtin_access::strcat_overlap): Adjust the bounds of overlap offset.
19546 (builtin_access::overlap): Avoid setting the size of overlap if it's
19548 (maybe_diag_overlap): Also consider arrays when deciding what values
19549 of offsets to include in diagnostics.
19551 2018-02-14 Martin Sebor <msebor@redhat.com>
19554 * attribs.c (diag_attr_exclusions): Consider the exclusion(s)
19555 that correspond to the kind of a declaration.
19557 2018-02-14 John David Anglin <danglin@gcc.gnu.org>
19560 * config/pa/pa.md: Load address of PIC label using the linkage table
19561 if the label is nonlocal.
19563 2018-02-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
19565 * config/rs6000/rs6000.c (rs6000_option_override_internal): Issue
19566 warning message if user requests -maltivec=be.
19567 * doc/invoke.texi: Document deprecation of -maltivec=be.
19569 2018-02-14 Will Schmidt <will_schmidt@vnet.ibm.com>
19572 * config/rs6000/rs6000-c.c: Update definitions for
19573 ALTIVEC_BUILTIN_VEC_SLD, ALTIVEC_BUILTIN_VEC_SLDW,
19574 VEC_XXSLDWI and ALTIVEC_BUILTIN_VEC_XXPERMDI builtins.
19576 2018-02-14 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
19579 * config/i386/cetintrin.h: Remove _rdssp[d|q] and
19580 add _get_ssp intrinsics. Remove argument from
19581 __builtin_ia32_rdssp[d|q].
19582 * config/i386/i386-builtin-types.def: Add UINT_FTYPE_VOID.
19583 * config/i386/i386-builtin.def: Remove argument from
19584 __builtin_ia32_rdssp[d|q].
19585 * config/i386/i386.c: Use UINT_FTYPE_VOID. Use
19586 ix86_expand_special_args_builtin for _rdssp[d|q].
19587 * config/i386/i386.md: Remove argument from rdssp[si|di] insn.
19588 Clear register before usage.
19589 * doc/extend.texi: Remove argument from __builtin_ia32_rdssp[d|q].
19590 Add documentation for new _get_ssp and _inc_ssp intrinsics.
19592 2018-02-14 Richard Sandiford <richard.sandiford@linaro.org>
19594 PR tree-optimization/84357
19595 * tree-data-ref.c (object_address_invariant_in_loop_p): Check
19596 operand 1 of an ARRAY_REF too.
19598 2018-02-14 Oleg Endo <olegendo@gcc.gnu.org>
19601 * config/rx/rx-protos.h (rx_reg_dead_or_unused_after_insn,
19602 rx_copy_reg_dead_or_unused_notes, rx_fuse_in_memory_bitop): New
19604 (set_of_reg): New struct.
19605 (rx_find_set_of_reg, rx_find_use_of_reg): New functions.
19606 * config/rx/rx.c (rx_reg_dead_or_unused_after_insn,
19607 rx_copy_reg_dead_or_unused_notes, rx_fuse_in_memory_bitop): New
19609 * config/rx/rx.md (andsi3, iorsi3, xorsi3): Convert to insn_and_split.
19610 Split into bitclr, bitset, bitinvert patterns if appropriate.
19611 (*bitset, *bitinvert, *bitclr): Convert to named insn_and_split and
19612 use rx_fuse_in_memory_bitop.
19613 (*bitset_in_memory, *bitinvert_in_memory, *bitclr_in_memory): Convert
19614 to named insn, correct maximum insn length.
19616 2018-02-14 Jozef Lawrynowicz <jozefl.gcc@gmail.com>
19619 * machmode.def: Define a complex mode for PARTIAL_INT.
19620 * genmodes.c (complex_class): Return MODE_COMPLEX_INT for
19622 * doc/rtl.texi: Document CSPImode.
19623 * config/msp430/msp430.c (msp430_hard_regno_nregs): Add CPSImode
19625 (msp430_hard_regno_nregs_with_padding): Likewise.
19627 2018-02-13 Peter Bergner <bergner@vnet.ibm.com>
19630 * config/rs6000/rs6000.c (mem_operand_gpr): Disallow altivec addresses.
19632 2018-02-13 Segher Boessenkool <segher@kernel.crashing.org>
19634 PR rtl-optimization/84169
19635 * combine.c (try_combine): New variable split_i2i3. Set it to true if
19636 we generated a parallel as new i3 and we split that to new i2 and i3
19637 instructions. Handle split_i2i3 similar to swap_i2i3: scan the
19638 LOG_LINKs of i3 to see which of those need to link to i2 now. Link
19639 those to i2, not i1. Partially rewrite this scan code.
19641 2018-02-13 Jakub Jelinek <jakub@redhat.com>
19644 * stor-layout.c (place_field): For variable length fields, adjust
19645 offset_align afterwards not just based on the field's alignment,
19646 but also on the size.
19648 PR middle-end/84309
19649 * match.pd (pow(C,x) -> exp(log(C)*x)): Use exp2s and log2s instead
19650 of exps and logs in the use_exp2 case.
19652 2018-02-13 Jeff Law <law@redhat.com>
19654 * config/rl/rl78.c (rl78_attribute_table): Fix terminator and
19655 entry for "vector".
19657 * config/rl78/rl78.c (rl78_handle_func_attribute): Mark
19660 2018-02-13 Alexandre Oliva <aoliva@redhat.com>
19664 * common.opt (gas-loc-support, gas-locview-support): New.
19665 (ginline-points, ginternal-reset-location-views): New.
19666 * doc/invoke.texi: Document them. Use @itemx where intended.
19667 (gvariable-location-views): Adjust.
19668 * target.def (reset_location_view): New.
19669 * doc/tm.texi.in (DWARF2_ASM_VIEW_DEBUG_INFO): New.
19670 (TARGET_RESET_LOCATION_VIEW): New.
19671 * doc/tm.texi: Rebuilt.
19672 * dwarf2out.c (dwarf2out_default_as_loc_support): New.
19673 (dwarf2out_default_as_locview_support): New.
19674 (output_asm_line_debug_info): Use option variables.
19675 (dwarf2out_maybe_output_loclist_view_pair): Likewise.
19676 (output_loc_list): Likewise.
19677 (add_high_low_attributes): Check option variables.
19678 Don't output entry view attribute in strict mode.
19679 (gen_inlined_subroutine_die): Check option variables.
19680 (dwarf2out_inline_entry): Likewise.
19681 (init_sections_and_labels): Likewise.
19682 (dwarf2out_early_finish): Likewise.
19683 (maybe_reset_location_view): New, from...
19684 (dwarf2out_var_location): ... here. Call it.
19685 * debug.h (dwarf2out_default_as_loc_support): Declare.
19686 (dwarf2out_default_as_locview_support): Declare.
19687 * hooks.c (hook_int_rtx_insn_0): New.
19688 * hooks.h (hook_int_rtx_insn_0): Declare.
19689 * toplev.c (process_options): Take -gas-loc-support and
19690 -gas-locview-support from dwarf2out. Enable
19691 -gvariable-location-views by default only with locview
19692 assembler support. Enable -ginternal-reset-location-views by
19693 default only if the target defines the corresponding hook.
19694 Enable -ginline-points by default if location views are
19695 enabled; force it disabled if statement frontiers are
19697 * tree-inline.c (expand_call_inline): Check option variables.
19698 * tree-ssa-live.c (remove_unused_scope_block_p): Likewise.
19700 2018-02-13 Richard Sandiford <richard.sandiford@linaro.org>
19702 PR tree-optimization/84321
19703 * tree-vrp.c (intersect_range_with_nonzero_bits): Fix VR_ANTI_RANGE
19704 handling. Also check whether the anti-range contains any values
19705 that satisfy the mask; switch to a VR_RANGE if not.
19707 2018-02-13 Paolo Bonzini <bonzini@gnu.org>
19710 * internal-fn.def (ASAN_CHECK, ASAN_MARK): Revert changes to fnspec.
19712 2018-02-13 Martin Jambor <mjambor@suse.cz>
19715 * ipa-param-manipulation.c (ipa_modify_call_arguments): Use location
19716 of call statements, also set location of a load to a temporary.
19718 2018-02-13 Sebastian Perta <sebastian.perta@renesas.com>
19720 * config/rl78/rl78.c (add_vector_labels): New function.
19721 * config/rl78/rl78.c (rl78_handle_vector_attribute): New function.
19722 * config/rl78/rl78.c (rl78_start_function): Call add_vector_labels.
19723 * config/rl78/rl78.c (rl78_handle_func_attribute): Removed the assert
19724 which checks that no arguments are passed.
19725 * config/rl78/rl78.c (rl78_attribute_table): Add "vector" attribute.
19726 * doc/extend.texi: Documentation for the new attribute.
19728 2018-02-13 Andreas Schwab <schwab@suse.de>
19730 * config/riscv/linux.h (CPP_SPEC): Define.
19732 2018-02-13 Jakub Jelinek <jakub@redhat.com>
19735 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Pass
19736 OPTION_MASK_ISA_AES | OPTION_MASK_ISA_SSE2 instead of
19737 OPTION_MASK_ISA_AES as first argument to def_builtin_const
19738 for AES builtins. Pass OPTION_MASK_ISA_PCLMUL | OPTION_MASK_ISA_SSE2
19739 instead of OPTION_MASK_ISA_PCLMUL as first argument to
19740 def_builtin_const for __builtin_ia32_pclmulqdq128 builtin.
19741 * config/i386/wmmintrin.h: If __SSE2__ is not defined, enable it
19742 temporarily for AES and PCLMUL builtins.
19744 PR tree-optimization/84339
19745 * gimple-fold.c (get_range_strlen): Set *FLEXP to true when handling
19746 ARRAY_REF where first operand is array_at_struct_end_p COMPONENT_REF.
19749 PR middle-end/84309
19750 * match.pd (pow(C,x) -> exp(log(C)*x)): Optimize instead into
19751 exp2(log2(C)*x) if C is a power of 2 and c99 runtime is available.
19752 * generic-match-head.c (canonicalize_math_after_vectorization_p): New
19754 * gimple-match-head.c (canonicalize_math_after_vectorization_p): New
19756 * omp-simd-clone.h: New file.
19757 * omp-simd-clone.c: Include omp-simd-clone.h.
19758 (expand_simd_clones): No longer static.
19759 * tree-vect-patterns.c: Include fold-const-call.h, attribs.h,
19760 cgraph.h and omp-simd-clone.h.
19761 (vect_recog_pow_pattern): Optimize pow(C,x) to exp(log(C)*x).
19762 (vect_recog_widen_shift_pattern): Formatting fix.
19763 (vect_pattern_recog_1): Don't check optab for calls.
19766 * config/i386/sse.md (<avx512>_vpermi2var<mode>3_mask): Force
19767 operands[2] into a REG before using gen_lowpart on it.
19769 2018-02-12 Jeff Law <law@redhat.com>
19772 * config/sh/sh.c (find_barrier): Consider a sibling call
19775 * cse.c (try_back_substitute_reg): Move any REG_ARGS_SIZE note when
19776 successfully back substituting a reg.
19778 2018-02-12 Richard Biener <rguenther@suse.de>
19780 PR tree-optimization/84037
19781 * tree-vect-slp.c (vect_analyze_slp_cost): Add visited
19782 parameter, move visited init to caller.
19783 (vect_slp_analyze_operations): Separate cost from validity
19784 check, initialize visited once for all instances.
19785 (vect_schedule_slp): Analyze map to CSE vectorized nodes once
19787 * tree-vect-stmts.c (vect_model_simple_cost): Make early
19789 (vect_model_promotion_demotion_cost): Likewise.
19790 (vectorizable_bswap): Guard cost modeling with !slp_node
19791 instead of !PURE_SLP_STMT to avoid double-counting on hybrid
19793 (vectorizable_call): Likewise.
19794 (vectorizable_conversion): Likewise.
19795 (vectorizable_assignment): Likewise.
19796 (vectorizable_shift): Likewise.
19797 (vectorizable_operation): Likewise.
19798 (vectorizable_store): Likewise.
19799 (vectorizable_load): Likewise.
19800 (vectorizable_condition): Likewise.
19801 (vectorizable_comparison): Likewise.
19803 2018-02-12 Paolo Bonzini <bonzini@gnu.org>
19806 * internal-fn.def (ASAN_CHECK): Fix fnspec to account for return value.
19807 (ASAN_MARK): Fix fnspec to account for return value, change pointer
19808 argument from 'R' to 'W' so that the pointed-to datum is clobbered.
19810 2018-02-08 Jan Hubicka <hubicka@ucw.cz>
19812 PR middle-end/83665
19813 * params.def (inline-min-speedup): Increase from 8 to 15.
19814 (max-inline-insns-auto): Decrease from 40 to 30.
19815 * ipa-split.c (consider_split): Add some buffer for function to
19816 be considered inlining candidate.
19817 * invoke.texi (max-inline-insns-auto, inline-min-speedup): UPdate
19820 2018-02-12 Richard Biener <rguenther@suse.de>
19822 PR tree-optimization/84037
19823 * tree-vect-slp.c (vect_build_slp_tree_2): Try swapping the
19824 matched stmts if we cannot swap the non-matched ones.
19826 2018-02-12 Olga Makhotina <olga.makhotina@intel.com>
19828 * config/i386/avx512fintrin.h (_mm_mask_scalef_round_sd,
19829 _mm_maskz_scalef_round_sd, _mm_mask_scalef_round_ss,
19830 _mm_maskz_scalef_round_ss): New intrinsics.
19831 (__builtin_ia32_scalefsd_round, __builtin_ia32_scalefss_round): Fix.
19832 * config/i386/i386-builtin.def (__builtin_ia32_scalefsd_round,
19833 __builtin_ia32_scalefss_round): Remove.
19834 (__builtin_ia32_scalefsd_mask_round,
19835 __builtin_ia32_scalefss_mask_round): New intrinsics.
19836 * config/i386/sse.md (vmscalef<mode><round_name>): Renamed to ...
19837 (vmscalef<mode><mask_scalar_name><round_scalar_name>): ... this.
19838 ((match_operand:VF_128 2 "<round_nimm_predicate>"
19839 "<round_constraint>")): Changed to ...
19840 ((match_operand:VF_128 2 "<round_scalar_nimm_predicate>"
19841 "<round_scalar_constraint>")): ... this.
19842 ("vscalef<ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|
19843 %0, %1, %2<round_op3>}"): Changed to ...
19844 ("vscalef<ssescalarmodesuffix>\t{<round_scalar_mask_op3>%2, %1,
19845 %0<mask_scalar_operand3>|%0<mask_scalar_operand3>, %1,
19846 %2<round_scalar_mask_op3>}"): ... this.
19847 * config/i386/subst.md (round_scalar_nimm_predicate): New.
19849 2018-02-12 Olga Makhotina <olga.makhotina@intel.com>
19851 * config/i386/avx512fintrin.h (_mm_mask_sqrt_round_sd)
19852 (_mm_maskz_sqrt_round_sd, _mm_mask_sqrt_round_ss)
19853 (_mm_maskz_sqrt_round_ss): New intrinsics.
19854 (__builtin_ia32_sqrtsd_round, __builtin_ia32_sqrtss_round): Remove.
19855 (__builtin_ia32_sqrtsd_mask_round)
19856 (__builtin_ia32_sqrtss_mask_round): New builtins.
19857 * config/i386/i386-builtin.def (__builtin_ia32_sqrtsd_round)
19858 (__builtin_ia32_sqrtss_round): Remove.
19859 (__builtin_ia32_sqrtsd_mask_round)
19860 (__builtin_ia32_sqrtss_mask_round): New builtins.
19861 * config/i386/sse.md (vmsqrt<mode>2<round_name>): Renamed to ...
19862 (vmsqrt<mode>2<mask_scalar_name><round_scalar_name>): ... this.
19863 ((match_operand:VF_128 1 "vector_operand"
19864 "xBm,<round_constraint>")): Changed to ...
19865 ((match_operand:VF_128 1 "vector_operand"
19866 "xBm,<round_scalar_constraint>")): ... this.
19867 (vsqrt<ssescalarmodesuffix>\t{<round_op3>%1, %2, %0|
19868 %0, %2, %<iptr>1<round_op3>}): Changed to ...
19869 (vsqrt<ssescalarmodesuffix>\t{<round_scalar_mask_op3>%1, %2,
19870 %0<mask_scalar_operand3>|%0<mask_scalar_operand3>, %2,
19871 %<iptr>1<round_scalar_mask_op3>}): ... this.
19872 ((set_attr "prefix" "<round_prefix>")): Changed to ...
19873 ((set_attr "prefix" "<round_scalar_prefix>")): ... this.
19875 2018-02-11 Steven Munroe <munroesj@gcc.gnu.org>
19878 * config/rs6000/mmintrin.h (_mm_cmpeq_pi32 [_ARCH_PWR9]):
19879 Cast vec_cmpeq result to correct type.
19880 * config/rs6000/mmintrin.h (_mm_cmpgt_pi32 [_ARCH_PWR9]):
19881 Cast vec_cmpgt result to correct type.
19883 2018-02-11 Alexandre Oliva <aoliva@redhat.com>
19885 * final.c (final_scan_insn_1): Renamed from...
19886 (final_scan_insn): ... this. New wrapper, to recover
19887 seen from the outermost call in recursive ones.
19888 * config/sparc/sparc.c (output_return): Drop seen from call.
19889 (output_sibcall): Likewise.
19890 * config/visium/visium.c (output_branch): Likewise.
19892 2018-02-10 John David Anglin <danglin@gcc.gnu.org>
19894 * config/pa/pa.c (hppa_profile_hook): Mark SYMBOL_REF for _mcount as
19897 2018-02-10 Alan Modra <amodra@gmail.com>
19900 * config/rs6000/rs6000.md (split_stack_return): Remove (use ..).
19901 Specify LR as an input.
19903 2018-02-10 Jakub Jelinek <jakub@redhat.com>
19906 * omp-low.c (maybe_remove_omp_member_access_dummy_vars,
19907 remove_member_access_dummy_vars): New functions.
19908 (lower_omp_for, lower_omp_taskreg, lower_omp_target,
19909 lower_omp_1, execute_lower_omp): Use them.
19911 PR rtl-optimization/84308
19912 * shrink-wrap.c (spread_components): Release todo vector.
19914 2018-02-09 Vladimir Makarov <vmakarov@redhat.com>
19916 PR rtl-optimization/57193
19917 * ira-color.c (struct allocno_color_data): Add member
19918 conflict_allocno_hard_prefs.
19919 (update_conflict_allocno_hard_prefs): New.
19920 (bucket_allocno_compare_func): Add a preference based on
19921 conflict_allocno_hard_prefs.
19922 (push_allocno_to_stack): Update conflict_allocno_hard_prefs.
19923 (color_allocnos): Remove a dead code. Initiate
19924 conflict_allocno_hard_prefs. Call update_costs_from_prefs.
19926 2018-02-09 Jakub Jelinek <jakub@redhat.com>
19929 * config/rs6000/vsx.md (p9_xxbrq_v16qi): Change input operand
19930 constraint from =wa to wa. Avoid a subreg on the output operand,
19931 instead use a pseudo and subreg it in a move.
19932 (p9_xxbrd_<mode>): Changed to ...
19933 (p9_xxbrd_v2di): ... this insn, without VSX_D iterator.
19934 (p9_xxbrd_v2df): New expander.
19935 (p9_xxbrw_<mode>): Changed to ...
19936 (p9_xxbrw_v4si): ... this insn, without VSX_W iterator.
19937 (p9_xxbrw_v4sf): New expander.
19939 2018-02-09 Sebastian Perta <sebastian.perta@renesas.com>
19941 * config/rx/rx.md (movsicc): Update expander to be matched by GCC.
19943 2018-02-09 Peter Bergner <bergner@vnet.ibm.com>
19946 * config/rs6000/vsx.md (vsx_mul_v2di): Handle generating a 64-bit
19947 multiply in 32-bit mode.
19948 (vsx_div_v2di): Handle generating a 64-bit signed divide in 32-bit mode.
19949 (vsx_udiv_v2di): Handle generating a 64-bit unsigned divide in 32-bit
19952 2018-02-09 Sebastian Perta <sebastian.perta@renesas.com>
19954 * config/rx/constraints.md (CALL_OP_SYMBOL_REF): Added new constraint
19955 to allow or block "symbol_ref" depending on the value of TARGET_JSR.
19956 * config/rx/rx.md (call_internal): Use CALL_OP_SYMBOL_REF.
19957 * config/rx/rx.md (call_value_internal): Use CALL_OP_SYMBOL_REF.
19959 2018-02-09 Pierre-Marie de Rodat <derodat@adacore.com>
19962 * dwarf2out.c (is_trivial_indirect_ref): New function.
19963 (dwarf2out_late_global_decl): Do not generate a location
19964 attribute for variables that have a non-trivial DECL_VALUE_EXPR
19965 and that are not defined in the current unit.
19967 2018-02-09 Eric Botcazou <ebotcazou@adacore.com>
19969 * optabs.c (prepare_cmp_insn): Try harder to emit a direct comparison
19970 instead of a libcall for UNORDERED.
19972 2018-02-09 Tamar Christina <tamar.christina@arm.com>
19975 * config/arm/arm-c.c (arm_cpu_builtins): Un-define __ARM_FEATURE_LDREX,
19976 __ARM_ARCH_PROFILE, __ARM_ARCH_ISA_THUMB, __ARM_FP and __ARM_NEON_FP.
19978 2018-02-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19981 * config/s390/s390.c (s390_set_current_function): Invoke
19982 s390_indirect_branch_settings also if fndecl didn't change.
19984 2018-02-09 Alexandre Oliva <aoliva@redhat.com>
19986 * config/rs6000/rs6000.md (blockage): Set length to zero.
19988 2018-02-09 Eric Botcazou <ebotcazou@adacore.com>
19990 * expr.c (optimize_bitfield_assignment_op): Remove obsolete assertion.
19992 2018-02-09 Jakub Jelinek <jakub@redhat.com>
19995 * gcc.c (STATIC_LIBASAN_LIBS, STATIC_LIBTSAN_LIBS,
19996 STATIC_LIBLSAN_LIBS, STATIC_LIBUBSAN_LIBS): Handle -static like
20000 * var-tracking.c (vt_add_function_parameter): Punt for non-onepart
20001 PARALLEL incoming that failed vt_get_decl_and_offset check.
20003 PR middle-end/84237
20004 * output.h (bss_initializer_p): Add NAMED argument, defaulted to false.
20005 * varasm.c (bss_initializer_p): Add NAMED argument, if true, ignore
20007 (get_variable_section): For decls in named .bss* sections pass true as
20008 second argument to bss_initializer_p.
20010 2018-02-09 Marek Polacek <polacek@redhat.com>
20011 Jakub Jelinek <jakub@redhat.com>
20014 * fold-const.c (fold_indirect_ref_1): Use VECTOR_TYPE_P macro.
20015 Formatting fixes. Verify first that tree_fits_poly_int64_p (op01).
20016 Sync some changes from cxx_fold_indirect_ref.
20018 2018-02-09 Alexandre Oliva <aoliva@redhat.com>
20020 * cfgexpand.c (expand_gimple_basic_block): Handle inline entry
20022 * dwarf2out.c (dwarf2_debug_hooks): Enable inline_entry hook.
20023 (BLOCK_INLINE_ENTRY_LABEL): New.
20024 (dwarf2out_var_location): Disregard inline entry markers.
20025 (inline_entry_data): New struct.
20026 (inline_entry_data_hasher): New hashtable type.
20027 (inline_entry_data_hasher::hash): New.
20028 (inline_entry_data_hasher::equal): New.
20029 (inline_entry_data_table): New variable.
20030 (add_high_low_attributes): Add DW_AT_entry_pc and
20031 DW_AT_GNU_entry_view attributes if a pending entry is found
20032 in inline_entry_data_table. Add old entry_pc attribute only
20033 if debug nonbinding markers are disabled.
20034 (gen_inlined_subroutine_die): Set BLOCK_DIE if nonbinding
20035 markers are enabled.
20036 (block_within_block_p, dwarf2out_inline_entry): New.
20037 (dwarf2out_finish): Check that no entries remained in
20038 inline_entry_data_table.
20039 * final.c (reemit_insn_block_notes): Handle inline entry notes.
20040 (final_scan_insn, notice_source_line): Likewise.
20041 (rest_of_clean_state): Skip inline entry markers.
20042 * gimple-pretty-print.c (dump_gimple_debug): Handle inline entry
20044 * gimple.c (gimple_build_debug_inline_entry): New.
20045 * gimple.h (enum gimple_debug_subcode): Add
20046 GIMPLE_DEBUG_INLINE_ENTRY.
20047 (gimple_build_debug_inline_entry): Declare.
20048 (gimple_debug_inline_entry_p): New.
20049 (gimple_debug_nonbind_marker_p): Adjust.
20050 * insn-notes.def (INLINE_ENTRY): New.
20051 * print-rtl.c (rtx_writer::print_rtx_operand_code_0): Handle
20052 inline entry marker notes.
20053 (print_insn): Likewise.
20054 * rtl.h (NOTE_MARKER_P): Add INLINE_ENTRY support.
20055 (INSN_DEBUG_MARKER_KIND): Likewise.
20056 (GEN_RTX_DEBUG_MARKER_INLINE_ENTRY_PAT): New.
20057 * tree-inline.c (expand_call_inline): Build and insert
20058 debug_inline_entry stmt.
20059 * tree-ssa-live.c (remove_unused_scope_block_p): Preserve
20060 inline entry blocks early, if nonbind markers are enabled.
20061 (dump_scope_block): Dump fragment info.
20062 * var-tracking.c (reemit_marker_as_note): Handle inline entry note.
20063 * doc/gimple.texi (gimple_debug_inline_entry_p): New.
20064 (gimple_build_debug_inline_entry): New.
20065 * doc/invoke.texi (gstatement-frontiers, gno-statement-frontiers):
20066 Enable/disable inline entry points too.
20067 * doc/rtl.texi (NOTE_INSN_INLINE_ENTRY): New.
20068 (DEBUG_INSN): Describe inline entry markers.
20070 * common.opt (gvariable-location-views): New.
20071 (gvariable-location-views=incompat5): New.
20072 * config.in: Rebuilt.
20073 * configure: Rebuilt.
20074 * configure.ac: Test assembler for view support.
20075 * dwarf2asm.c (dw2_asm_output_symname_uleb128): New.
20076 * dwarf2asm.h (dw2_asm_output_symname_uleb128): Declare.
20077 * dwarf2out.c (var_loc_view): New typedef.
20078 (struct dw_loc_list_struct): Add vl_symbol, vbegin, vend.
20079 (dwarf2out_locviews_in_attribute): New.
20080 (dwarf2out_locviews_in_loclist): New.
20081 (dw_val_equal_p): Compare val_view_list of dw_val_class_view_lists.
20082 (enum dw_line_info_opcode): Add LI_adv_address.
20083 (struct dw_line_info_table): Add view.
20084 (RESET_NEXT_VIEW, RESETTING_VIEW_P): New macros.
20085 (DWARF2_ASM_VIEW_DEBUG_INFO): Define default.
20086 (zero_view_p): New variable.
20087 (ZERO_VIEW_P): New macro.
20088 (output_asm_line_debug_info): New.
20089 (struct var_loc_node): Add view.
20090 (add_AT_view_list, AT_loc_list): New.
20091 (add_var_loc_to_decl): Add view param. Test it against last.
20092 (new_loc_list): Add view params. Record them.
20093 (AT_loc_list_ptr): Handle loc and view lists.
20094 (view_list_to_loc_list_val_node): New.
20095 (print_dw_val): Handle dw_val_class_view_list.
20096 (size_of_die): Likewise.
20097 (value_format): Likewise.
20098 (loc_list_has_views): New.
20099 (gen_llsym): Set vl_symbol too.
20100 (maybe_gen_llsym, skip_loc_list_entry): New.
20101 (dwarf2out_maybe_output_loclist_view_pair): New.
20102 (output_loc_list): Output view list or entries too.
20103 (output_view_list_offset): New.
20104 (output_die): Handle dw_val_class_view_list.
20105 (output_dwarf_version): New.
20106 (output_compilation_unit_header): Use it.
20107 (output_skeleton_debug_sections): Likewise.
20108 (output_rnglists, output_line_info): Likewise.
20109 (output_pubnames, output_aranges): Update version comments.
20110 (output_one_line_info_table): Output view numbers in asm comments.
20111 (dw_loc_list): Determine current endview, pass it to new_loc_list.
20112 Call maybe_gen_llsym.
20113 (loc_list_from_tree_1): Adjust.
20114 (add_AT_location_description): Create view list attribute if
20115 needed, check it's absent otherwise.
20116 (convert_cfa_to_fb_loc_list): Adjust.
20117 (maybe_emit_file): Call output_asm_line_debug_info for test.
20118 (dwarf2out_var_location): Reset views as needed. Precompute
20119 add_var_loc_to_decl args. Call get_attr_min_length only if we have the
20120 attribute. Set view.
20121 (new_line_info_table): Reset next view.
20122 (set_cur_line_info_table): Call output_asm_line_debug_info for test.
20123 (dwarf2out_source_line): Likewise. Output view resets and labels to
20124 the assembler, or select appropriate line info opcodes.
20125 (prune_unused_types_walk_attribs): Handle dw_val_class_view_list.
20126 (optimize_string_length): Catch it. Adjust.
20127 (resolve_addr): Copy vl_symbol along with ll_symbol. Handle
20128 dw_val_class_view_list, and remove it if no longer needed.
20129 (hash_loc_list): Hash view numbers.
20130 (loc_list_hasher::equal): Compare them.
20131 (optimize_location_lists): Check whether a view list symbol is
20132 needed, and whether the locview attribute is present, and
20133 whether they match. Remove the locview attribute if no longer
20135 (index_location_lists): Call skip_loc_list_entry for test.
20136 (dwarf2out_finish): Call output_asm_line_debug_info for test.
20137 Use output_dwarf_version.
20138 * dwarf2out.h (enum dw_val_class): Add dw_val_class_view_list.
20139 (struct dw_val_node): Add val_view_list.
20140 * final.c (SEEN_NEXT_VIEW): New.
20141 (set_next_view_needed): New.
20142 (clear_next_view_needed): New.
20143 (maybe_output_next_view): New.
20144 (final_start_function): Rename to...
20145 (final_start_function_1): ... this. Take pointer to FIRST,
20146 add SEEN parameter. Emit param bindings in the initial view.
20147 (final_start_function): Reintroduce SEEN-less interface.
20148 (final): Rename to...
20149 (final_1): ... this. Take SEEN parameter. Output final pending
20150 next view at the end.
20151 (final): Reintroduce seen-less interface.
20152 (final_scan_insn): Output pending next view before switching
20153 sections or ending a block. Mark the next view as needed when
20154 outputting variable locations. Notify debug backend of section
20155 changes, and of location view changes.
20156 (rest_of_handle_final): Adjust.
20157 * toplev.c (process_options): Autodetect value for debug variable
20158 location views option. Warn on incompat5 without -gdwarf-5.
20159 * doc/invoke.texi (gvariable-location-views): New.
20160 (gvariable-location-views=incompat5): New.
20161 (gno-variable-location-views): New.
20163 2018-02-08 David Malcolm <dmalcolm@redhat.com>
20165 PR tree-optimization/84136
20166 * tree-cfg.c (find_taken_edge_computed_goto): Remove assertion
20167 that the result of find_edge is non-NULL.
20169 2018-02-08 Sergey Shalnov <sergey.shalnov@intel.com>
20172 * config/i386/x86-tune-costs.h (skylake_cost): Fix cost of
20173 storing integer register in SImode. Fix cost of 256 and 512
20174 byte aligned SSE register store.
20176 2018-02-08 Sergey Shalnov <sergey.shalnov@intel.com>
20178 * config/i386/i386.c (ix86_multiplication_cost): Fix
20179 multiplication cost for TARGET_AVX512DQ.
20181 2018-02-08 Marek Polacek <polacek@redhat.com>
20183 PR tree-optimization/84238
20184 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Verify the result of
20187 2018-02-08 Richard Sandiford <richard.sandiford@linaro.org>
20189 PR tree-optimization/84265
20190 * tree-vect-stmts.c (vectorizable_store): Don't treat
20191 VMAT_CONTIGUOUS accesses as grouped.
20192 (vectorizable_load): Likewise.
20194 2018-02-08 Richard Sandiford <richard.sandiford@linaro.org>
20196 PR tree-optimization/81635
20197 * wide-int.h (wi::round_down_for_mask, wi::round_up_for_mask): Declare.
20198 * wide-int.cc (wi::round_down_for_mask, wi::round_up_for_mask)
20199 (test_round_for_mask): New functions.
20200 (wide_int_cc_tests): Call test_round_for_mask.
20201 * tree-vrp.h (intersect_range_with_nonzero_bits): Declare.
20202 * tree-vrp.c (intersect_range_with_nonzero_bits): New function.
20203 * tree-data-ref.c (split_constant_offset_1): Use it to refine the
20204 range returned by get_range_info.
20206 2018-02-08 Jan Hubicka <hubicka@ucw.cz>
20209 * cgraph.h (symtab_node::output_to_lto_symbol_table_p): Declare
20210 * symtab.c: Include builtins.h
20211 (symtab_node::output_to_lto_symbol_table_p): Move here
20212 from lto-streamer-out.c:output_symbol_p.
20213 * lto-streamer-out.c (write_symbol): Turn early exit to assert.
20214 (output_symbol_p): Move all logic to symtab.c
20215 (produce_symtab): Update.
20217 2018-02-08 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20219 * config/s390/s390-opts.h (enum indirect_branch): Define.
20220 * config/s390/s390-protos.h (s390_return_addr_from_memory)
20221 (s390_indirect_branch_via_thunk)
20222 (s390_indirect_branch_via_inline_thunk): Add function prototypes.
20223 (enum s390_indirect_branch_type): Define.
20224 * config/s390/s390.c (struct s390_frame_layout, struct
20225 machine_function): Remove.
20226 (indirect_branch_prez10thunk_mask, indirect_branch_z10thunk_mask)
20227 (indirect_branch_table_label_no, indirect_branch_table_name):
20229 (INDIRECT_BRANCH_NUM_OPTIONS): Define macro.
20230 (enum s390_indirect_branch_option): Define.
20231 (s390_return_addr_from_memory): New function.
20232 (s390_handle_string_attribute): New function.
20233 (s390_attribute_table): Add new attribute handler.
20234 (s390_execute_label): Handle UNSPEC_EXECUTE_JUMP patterns.
20235 (s390_indirect_branch_via_thunk): New function.
20236 (s390_indirect_branch_via_inline_thunk): New function.
20237 (s390_function_ok_for_sibcall): When jumping via thunk disallow
20238 sibling call optimization for non z10 compiles.
20239 (s390_emit_call): Force indirect branch target to be a single
20240 register. Add r1 clobber for non-z10 compiles.
20241 (s390_emit_epilogue): Emit return jump via return_use expander.
20242 (s390_reorg): Handle JUMP_INSNs as execute targets.
20243 (s390_option_override_internal): Perform validity checks for the
20244 new command line options.
20245 (s390_indirect_branch_attrvalue): New function.
20246 (s390_indirect_branch_settings): New function.
20247 (s390_set_current_function): Invoke s390_indirect_branch_settings.
20248 (s390_output_indirect_thunk_function): New function.
20249 (s390_code_end): Implement target hook.
20250 (s390_case_values_threshold): Implement target hook.
20251 (TARGET_ASM_CODE_END, TARGET_CASE_VALUES_THRESHOLD): Define target
20253 * config/s390/s390.h (struct s390_frame_layout)
20254 (struct machine_function): Move here from s390.c.
20255 (TARGET_INDIRECT_BRANCH_NOBP_RET)
20256 (TARGET_INDIRECT_BRANCH_NOBP_JUMP)
20257 (TARGET_INDIRECT_BRANCH_NOBP_JUMP_THUNK)
20258 (TARGET_INDIRECT_BRANCH_NOBP_JUMP_INLINE_THUNK)
20259 (TARGET_INDIRECT_BRANCH_NOBP_CALL)
20260 (TARGET_DEFAULT_INDIRECT_BRANCH_TABLE)
20261 (TARGET_INDIRECT_BRANCH_THUNK_NAME_EXRL)
20262 (TARGET_INDIRECT_BRANCH_THUNK_NAME_EX)
20263 (TARGET_INDIRECT_BRANCH_TABLE): Define macros.
20264 * config/s390/s390.md (UNSPEC_EXECUTE_JUMP)
20265 (INDIRECT_BRANCH_THUNK_REGNUM): Define constants.
20266 (mnemonic attribute): Add values which aren't recognized
20268 ("*cjump_long", "*icjump_long", "*basr", "*basr_r"): Disable
20269 pattern for branch conversion. Fix mnemonic attribute.
20270 ("*c<code>", "*sibcall_br", "*sibcall_value_br", "*return"): Emit
20271 indirect branch via thunk if requested.
20272 ("indirect_jump", "<code>"): Expand patterns for branch conversion.
20273 ("*indirect_jump"): Disable for branch conversion using out of
20275 ("indirect_jump_via_thunk<mode>_z10")
20276 ("indirect_jump_via_thunk<mode>")
20277 ("indirect_jump_via_inlinethunk<mode>_z10")
20278 ("indirect_jump_via_inlinethunk<mode>", "*casesi_jump")
20279 ("casesi_jump_via_thunk<mode>_z10", "casesi_jump_via_thunk<mode>")
20280 ("casesi_jump_via_inlinethunk<mode>_z10")
20281 ("casesi_jump_via_inlinethunk<mode>", "*basr_via_thunk<mode>_z10")
20282 ("*basr_via_thunk<mode>", "*basr_r_via_thunk_z10")
20283 ("*basr_r_via_thunk", "return<mode>_prez10"): New pattern.
20284 ("*indirect2_jump"): Disable for branch conversion.
20285 ("casesi_jump"): Turn into expander and expand patterns for branch
20287 ("return_use"): New expander.
20288 ("*return"): Emit return via thunk and rename it to ...
20289 ("*return<mode>"): ... this one.
20290 * config/s390/s390.opt: Add new options and and enum for the
20293 2018-02-08 Richard Sandiford <richard.sandiford@linaro.org>
20295 * lra-constraints.c (match_reload): Unconditionally use
20296 gen_lowpart_SUBREG, rather than selecting between that
20297 and equivalent gen_rtx_SUBREG code.
20299 2018-02-08 Richard Biener <rguenther@suse.de>
20301 PR tree-optimization/84233
20302 * tree-ssa-phiprop.c (propagate_with_phi): Use separate
20303 changed flag instead of boguously re-using phi_inserted.
20305 2018-02-08 Martin Jambor <mjambor@suse.cz>
20307 * hsa-gen.c (get_symbol_for_decl): Set program allocation for
20308 static local variables.
20310 2018-02-08 Richard Biener <rguenther@suse.de>
20312 PR tree-optimization/84278
20313 * tree-vect-stmts.c (vectorizable_store): When looking for
20314 smaller vector types to perform grouped strided loads/stores
20315 make sure the mode is supported by the target.
20316 (vectorizable_load): Likewise.
20318 2018-02-08 Wilco Dijkstra <wdijkstr@arm.com>
20320 * config/aarch64/aarch64.c (aarch64_components_for_bb):
20321 Increase LDP/STP opportunities by adding adjacent callee-saves.
20323 2018-02-08 Wilco Dijkstra <wdijkstr@arm.com>
20325 PR rtl-optimization/84068
20326 PR rtl-optimization/83459
20327 * haifa-sched.c (rank_for_schedule): Fix SCHED_PRESSURE_MODEL sorting.
20329 2018-02-08 Aldy Hernandez <aldyh@redhat.com>
20331 PR tree-optimization/84224
20332 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Remove assert.
20333 * calls.c (gimple_alloca_call_p): Only return TRUE when we have
20334 non-zero arguments.
20336 2018-02-07 Iain Sandoe <iain@codesourcery.com>
20339 * config/rs6000/altivec.md (*restore_world): Remove LR use.
20340 * config/rs6000/predicates.md (restore_world_operation): Adjust op
20341 count, remove one USE.
20343 2018-02-07 Michael Meissner <meissner@linux.vnet.ibm.com>
20345 * doc/install.texi (Configuration): Document the
20346 --with-long-double-format={ibm,ieee} PowerPC configuration
20350 * config/rs6000/rs6000.md (fix_trunc<SFDF:mode><QHI:mode>2):
20351 Convert from define_expand to be define_insn_and_split. Rework
20352 float/double/_Float128 conversions to QI/HI/SImode to work with
20353 both ISA 2.07 (power8) or ISA 3.0 (power9). Fix regression where
20354 conversions to QI/HImode types did a store and then a load to
20355 truncate the value. For conversions to VSX registers, don't split
20356 the insn, instead emit the code directly. Use the code iterator
20357 any_fix to combine signed and unsigned conversions.
20358 (fix<uns>_trunc<SFDF:mode>si2_p8): Likewise.
20359 (fixuns_trunc<SFDF:mode><QHI:mode>2): Likewise.
20360 (fix_trunc<IEEE128:mode><QHI:mode>2): Likewise.
20361 (fix<uns>_trunc<SFDF:mode><QHI:mode>2): Likewise.
20362 (fix_<mode>di2_hw): Likewise.
20363 (fixuns_<mode>di2_hw): Likewise.
20364 (fix_<mode>si2_hw): Likewise.
20365 (fixuns_<mode>si2_hw): Likewise.
20366 (fix<uns>_<IEEE128:mode><SDI:mode>2_hw): Likewise.
20367 (fix<uns>_trunc<IEEE128:mode><QHI:mode>2): Likewise.
20368 (fctiw<u>z_<mode>_smallint): Rename fctiw<u>z_<mode>_smallint to
20369 fix<uns>_trunc<SFDF:mode>si2_p8.
20370 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Delete, no longer used.
20371 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
20372 (fix<uns>_<mode>_mem): Likewise.
20373 (fctiw<u>z_<mode>_mem): Likewise.
20374 (fix<uns>_<mode>_mem): Likewise.
20375 (fix<uns>_trunc<SFDF:mode><QHSI:mode>2_mem): On ISA 3.0, prevent
20376 the register allocator from doing a direct move to the GPRs to do
20377 a store, and instead use the ISA 3.0 store byte/half-word from
20378 vector register instruction. For IEEE 128-bit floating point,
20379 also optimize stores of 32-bit ints.
20380 (fix<uns>_trunc<IEEE128:mode><QHSI:mode>2_mem): Likewise.
20382 2018-02-07 Alan Hayward <alan.hayward@arm.com>
20384 * genextract.c (push_pathstr_operand): New function to support
20386 (walk_rtx): Call push_pathstr_operand.
20387 (print_path): Support [a-zA-Z].
20389 2018-02-07 Richard Biener <rguenther@suse.de>
20391 PR tree-optimization/84037
20392 * tree-vectorizer.h (struct _loop_vec_info): Add ivexpr_map member.
20393 (cse_and_gimplify_to_preheader): Declare.
20394 (vect_get_place_in_interleaving_chain): Likewise.
20395 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
20397 (_loop_vec_info::~_loop_vec_info): Delete it.
20398 (cse_and_gimplify_to_preheader): New function.
20399 * tree-vect-slp.c (vect_get_place_in_interleaving_chain): Export.
20400 * tree-vect-stmts.c (vectorizable_store): CSE base and steps.
20401 (vectorizable_load): Likewise. For grouped stores always base
20402 the IV on the first element.
20403 * tree-vect-loop-manip.c (vect_loop_versioning): Unshare versioning
20404 condition before gimplifying.
20406 2018-02-07 Jakub Jelinek <jakub@redhat.com>
20408 * tree-eh.c (operation_could_trap_helper_p): Ignore honor_trapv for
20409 *DIV_EXPR and *MOD_EXPR.
20411 2018-02-07 H.J. Lu <hongjiu.lu@intel.com>
20414 * config/i386/i386.c (ix86_option_override_internal): Mask out
20415 the CF_SET bit when checking -fcf-protection.
20417 2018-02-07 Tom de Vries <tom@codesourcery.com>
20420 * omp-expand.c (expand_oacc_collapse_init): Ensure diff_type is large
20423 2018-02-07 Richard Biener <rguenther@suse.de>
20425 PR tree-optimization/84204
20426 * tree-chrec.c (chrec_fold_plus_1): Remove size limiting in
20429 PR tree-optimization/84205
20430 * graphite-isl-ast-to-gimple.c (binary_op_to_tree): Also
20431 special-case isl_ast_op_zdiv_r.
20433 PR tree-optimization/84223
20434 * graphite-scop-detection.c (gather_bbs::before_dom_children):
20435 Only add conditions from within the region.
20436 (gather_bbs::after_dom_children): Adjust.
20438 2018-02-07 Georg-Johann Lay <avr@gjlay.de>
20441 * config/avr/avr.h (GENERAL_REGNO_P, GENERAL_REG_P): New macros.
20442 * config/avr/avr.md: Only post-reload split REG-REG moves if
20443 either register is GENERAL_REG_P.
20445 2018-02-07 Jakub Jelinek <jakub@redhat.com>
20447 PR tree-optimization/84235
20448 * tree-ssa-scopedtables.c
20449 (avail_exprs_stack::simplify_binary_operation): Fir MINUS_EXPR, punt
20450 if the subtraction is performed in floating point type where NaNs are
20451 honored. For *DIV_EXPR, punt for ALL_FRACT_MODE_Ps where we can't
20452 build 1. Formatting fix.
20454 2018-02-06 Jakub Jelinek <jakub@redhat.com>
20457 * config/i386/i386.c (rest_of_insert_endbranch): Only skip
20458 NOTE_INSN_CALL_ARG_LOCATION after a call, not anything else,
20459 and skip it regardless of bb boundaries. Use CALL_P macro,
20460 don't test INSN_P (insn) together with CALL_P or JUMP_P check
20461 unnecessarily, formatting fix.
20463 2018-02-06 Michael Collison <michael.collison@arm.com>
20465 * config/arm/thumb2.md:
20466 (*thumb2_mov_negscc): Split only if TARGET_THUMB2 && !arm_restrict_it.
20467 (*thumb_mov_notscc): Ditto.
20469 2018-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
20472 * config/rs6000/rs6000.md (su code attribute): Use "u" for
20473 unsigned_fix, not "s".
20475 2018-02-06 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
20477 * configure.ac (gcc_fn_eh_frame_ro): New function.
20478 (gcc_cv_as_cfi_directive): Check both 32 and 64-bit assembler for
20479 correct .eh_frame permissions.
20480 * configure: Regenerate.
20482 2018-02-06 Andrew Jenner <andrew@codeourcery.com>
20484 * doc/invoke.texi: Add section for the PowerPC SPE backend. Remove
20485 irrelevant options.
20487 2018-02-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20489 * config/rs6000/rs6000.c (rs6000_option_override_internal):
20490 Display warning message for -mno-speculate-indirect-jumps.
20492 2018-02-06 Andrew Jenner <andrew@codesourcery.com>
20494 * config/powerpcspe/powerpcspe.opt: (msimple-fpu, mfpu) Add
20496 * config/powerpcspe/sysv4.opt (mbit-align): Likewise.
20498 2018-02-06 Aldy Hernandez <aldyh@redhat.com>
20500 PR tree-optimization/84225
20501 * tree-eh.c (find_trapping_overflow): Only call
20502 operation_no_trapping_overflow when ANY_INTEGRAL_TYPE_P.
20504 2018-02-06 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
20507 * config/i386/i386.c: Reimplement the check of possible options
20508 -mibt/-mshstk conbination. Change error messages.
20509 * doc/invoke.texi: Fix a typo: remove extra '='.
20511 2018-02-06 Marek Polacek <polacek@redhat.com>
20513 PR tree-optimization/84228
20514 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Skip debug statements.
20516 2018-02-06 Tamar Christina <tamar.christina@arm.com>
20519 * config/arm/arm.c (arm_print_asm_arch_directives): Record already
20520 emitted arch directives.
20521 * config/arm/arm-c.c (arm_cpu_builtins): Undefine __ARM_ARCH and
20522 __ARM_FEATURE_COPROC before changing architectures.
20524 2018-02-06 Richard Biener <rguenther@suse.de>
20526 * config/i386/i386.c (print_reg): Fix typo.
20527 (ix86_loop_unroll_adjust): Do not unroll beyond the original nunroll.
20529 2018-02-06 Eric Botcazou <ebotcazou@adacore.com>
20531 * configure: Regenerate.
20533 2018-02-05 Martin Sebor <msebor@redhat.com>
20535 PR tree-optimization/83369
20536 * tree-ssa-ccp.c (pass_post_ipa_warn::execute): Use %G to print
20539 2018-02-05 Martin Liska <mliska@suse.cz>
20541 * doc/invoke.texi: Cherry-pick upstream r323995.
20543 2018-02-05 Richard Sandiford <richard.sandiford@linaro.org>
20545 * ira.c (ira_init_register_move_cost): Adjust comment.
20547 2018-02-05 Martin Liska <mliska@suse.cz>
20549 PR gcov-profile/84137
20550 * doc/gcov.texi: Fix typo in documentation.
20552 2018-02-05 Martin Liska <mliska@suse.cz>
20554 PR gcov-profile/83879
20555 * doc/gcov.texi: Document necessity of --dynamic-list-data when
20556 using dlopen functionality.
20558 2018-02-05 Olga Makhotina <olga.makhotina@intel.com>
20560 * config/i386/avx512dqintrin.h (_mm_mask_range_sd, _mm_maskz_range_sd,
20561 _mm_mask_range_round_sd, _mm_maskz_range_round_sd, _mm_mask_range_ss,
20562 _mm_maskz_range_ss, _mm_mask_range_round_ss,
20563 _mm_maskz_range_round_ss): New intrinsics.
20564 (__builtin_ia32_rangesd128_round)
20565 (__builtin_ia32_rangess128_round): Remove.
20566 (__builtin_ia32_rangesd128_mask_round,
20567 __builtin_ia32_rangess128_mask_round): New builtins.
20568 * config/i386/i386-builtin.def (__builtin_ia32_rangesd128_round,
20569 __builtin_ia32_rangess128_round): Remove.
20570 (__builtin_ia32_rangesd128_mask_round,
20571 __builtin_ia32_rangess128_mask_round): New builtins.
20572 * config/i386/sse.md (ranges<mode><round_saeonly_name>): Renamed to ...
20573 (ranges<mode><mask_scalar_name><round_saeonly_scalar_name>): ... this.
20574 ((match_operand:VF_128 2 "<round_saeonly_nimm_predicate>"
20575 "<round_saeonly_constraint>")): Changed to ...
20576 ((match_operand:VF_128 2 "<round_saeonly_scalar_nimm_predicate>"
20577 "<round_saeonly_scalar_constraint>")): ... this.
20578 ("vrange<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
20579 %0, %1, %2<round_saeonly_op4>, %3}"): Changed to ...
20580 ("vrange<ssescalarmodesuffix>\t{%3, <round_saeonly_scalar_mask_op4>%2,
20581 %1, %0<mask_scalar_operand4>|%0<mask_scalar_operand4>, %1,
20582 %2<round_saeonly_scalar_mask_op4>, %3}"): ... this.
20584 2018-02-02 Andrew Jenner <andrew@codesourcery.com>
20586 * config/powerpcspe/powerpcspe.opt: Add Undocumented to irrelevant
20588 * config/powerpcspe/powerpcspe-tables.opt (rs6000_cpu_opt_value):
20589 Remove all values except native, 8540 and 8548.
20591 2018-02-02 H.J. Lu <hongjiu.lu@intel.com>
20593 * config/i386/i386.c (ix86_output_function_return): Pass
20594 INVALID_REGNUM, instead of -1, as invalid register number to
20595 indirect_thunk_name and output_indirect_thunk.
20597 2018-02-02 Julia Koval <julia.koval@intel.com>
20599 * config.gcc: Add -march=icelake.
20600 * config/i386/driver-i386.c (host_detect_local_cpu): Detect icelake.
20601 * config/i386/i386-c.c (ix86_target_macros_internal): Handle icelake.
20602 * config/i386/i386.c (processor_costs): Add m_ICELAKE.
20603 (PTA_ICELAKE, PTA_AVX512VNNI, PTA_GFNI, PTA_VAES, PTA_AVX512VBMI2,
20604 PTA_VPCLMULQDQ, PTA_RDPID, PTA_AVX512BITALG): New.
20605 (processor_target_table): Add icelake.
20606 (ix86_option_override_internal): Handle new PTAs.
20607 (get_builtin_code_for_version): Handle icelake.
20608 (M_INTEL_COREI7_ICELAKE): New.
20609 (fold_builtin_cpu): Handle icelake.
20610 * config/i386/i386.h (TARGET_ICELAKE, PROCESSOR_ICELAKE): New.
20611 * doc/invoke.texi: Add -march=icelake.
20613 2018-02-02 Julia Koval <julia.koval@intel.com>
20615 * config/i386/i386.c (ix86_option_override_internal): Change flags type
20616 to wide_int_bitmask.
20617 * wide-int-bitmask.h: New.
20619 2018-02-02 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
20622 * config/i386/i386.md: Replace Pmode with word_mode in
20623 builtin_setjmp_setup and builtin_longjmp to support x32.
20625 2018-02-01 Peter Bergner <bergner@vnet.ibm.com>
20629 * config/rs6000/driver-rs6000.c: #include "diagnostic.h".
20631 (rs6000_supported_cpu_names): New static variable.
20632 (linux_cpu_translation_table): Likewise.
20633 (elf_platform) <cpu>: Define new static variable and use it.
20634 Translate kernel AT_PLATFORM name to canonical name if needed.
20635 Error if platform name is unknown.
20637 2018-02-01 Aldy Hernandez <aldyh@redhat.com>
20640 * config/pa/predicates.md (base14_operand): Handle E_VOIDmode.
20642 2018-02-01 Jeff Law <law@redhat.com>
20645 * config/i386/i386.c (release_scratch_register_on_entry): Add new
20646 OFFSET and RELEASE_VIA_POP arguments. Use SP+OFFSET to restore
20647 the scratch if RELEASE_VIA_POP is false.
20648 (ix86_adjust_stack_and_probe_stack_clash): Un-constify SIZE.
20649 If we have to save a temporary register, decrement SIZE appropriately.
20650 Pass new arguments to release_scratch_register_on_entry.
20651 (ix86_adjust_stack_and_probe): Likewise.
20652 (ix86_emit_probe_stack_range): Pass new arguments to
20653 release_scratch_register_on_entry.
20655 2018-02-01 Uros Bizjak <ubizjak@gmail.com>
20657 PR rtl-optimization/84157
20658 * combine.c (change_zero_ext): Use REG_P predicate in
20659 front of HARD_REGISTER_P predicate.
20661 2018-02-01 Georg-Johann Lay <avr@gjlay.de>
20663 * config/avr/avr.c (avr_option_override): Move disabling of
20664 -fdelete-null-pointer-checks to...
20665 * common/config/avr/avr-common.c (avr_option_optimization_table):
20668 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
20670 PR tree-optimization/81635
20671 * tree-data-ref.c (split_constant_offset_1): For types that
20672 wrap on overflow, try to use range info to prove that wrapping
20675 2018-02-01 Renlin Li <renlin.li@arm.com>
20678 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
20679 TAILCALL_ADDR_REGS.
20680 (aarch64_register_move_cost): Likewise.
20681 * config/aarch64/aarch64.h (reg_class): Rename CALLER_SAVE_REGS to
20682 TAILCALL_ADDR_REGS.
20683 (REG_CLASS_NAMES): Likewise.
20684 (REG_CLASS_CONTENTS): Rename CALLER_SAVE_REGS to
20685 TAILCALL_ADDR_REGS. Remove IP registers.
20686 * config/aarch64/aarch64.md (Ucs): Update register constraint.
20688 2018-02-01 Richard Biener <rguenther@suse.de>
20690 * domwalk.h (dom_walker::dom_walker): Add additional constructor
20691 for specifying RPO order and allow NULL for that.
20692 * domwalk.c (dom_walker::dom_walker): Likewise.
20693 (dom_walker::walk): Handle NULL RPO order.
20694 * tree-into-ssa.c (rewrite_dom_walker): Do not walk dom children
20696 (rewrite_update_dom_walker): Likewise.
20697 (mark_def_dom_walker): Likewise.
20699 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
20701 * config/aarch64/aarch64-protos.h (aarch64_split_sve_subreg_move)
20702 (aarch64_maybe_expand_sve_subreg_move): Declare.
20703 * config/aarch64/aarch64.md (UNSPEC_REV_SUBREG): New unspec.
20704 * config/aarch64/predicates.md (aarch64_any_register_operand): New
20706 * config/aarch64/aarch64-sve.md (mov<mode>): Optimize subreg moves
20707 that are semantically a reverse operation.
20708 (*aarch64_sve_mov<mode>_subreg_be): New pattern.
20709 * config/aarch64/aarch64.c (aarch64_maybe_expand_sve_subreg_move):
20710 (aarch64_replace_reg_mode, aarch64_split_sve_subreg_move): New
20712 (aarch64_can_change_mode_class): For big-endian, forbid changes
20713 between two SVE modes if they have different element sizes.
20715 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
20717 * config/aarch64/aarch64.c (aarch64_expand_sve_const_vector): Prefer
20718 the TImode handling for big-endian targets.
20720 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
20722 * config/aarch64/aarch64-sve.md (sve_ld1rq): Replace with...
20723 (*sve_ld1rq<Vesize>): ... this new pattern. Handle all element sizes,
20725 * config/aarch64/aarch64.c (aarch64_expand_sve_widened_duplicate):
20726 Remove BSWAP handing for big-endian targets and use the form of
20727 LD1RQ appropariate for the mode.
20729 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
20731 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Handle
20732 all CONST_VECTOR_DUPLICATE_P vectors, not just those with a single
20733 duplicated element.
20735 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
20738 * config/aarch64/aarch64.c (aarch64_secondary_reload): Tighten
20739 check for operands that need to go through aarch64_sve_reload_be.
20741 2018-02-01 Jakub Jelinek <jakub@redhat.com>
20743 PR tree-optimization/81661
20744 PR tree-optimization/84117
20745 * tree-eh.h (rewrite_to_non_trapping_overflow): Declare.
20746 * tree-eh.c: Include gimplify.h.
20747 (find_trapping_overflow, replace_trapping_overflow,
20748 rewrite_to_non_trapping_overflow): New functions.
20749 * tree-vect-loop.c: Include tree-eh.h.
20750 (vect_get_loop_niters): Use rewrite_to_non_trapping_overflow.
20751 * tree-data-ref.c: Include tree-eh.h.
20752 (get_segment_min_max): Use rewrite_to_non_trapping_overflow.
20754 2018-01-31 Uros Bizjak <ubizjak@gmail.com>
20756 PR rtl-optimization/84123
20757 * combine.c (change_zero_ext): Check if hard register satisfies
20758 can_change_dest_mode before calling gen_lowpart_SUBREG.
20760 2018-01-31 Vladimir Makarov <vmakarov@redhat.com>
20763 * ira.c (ira_init_register_move_cost): Remove assert.
20765 2018-01-31 Eric Botcazou <ebotcazou@adacore.com>
20767 PR rtl-optimization/84071
20768 * doc/tm.texi.in (WORD_REGISTER_OPERATIONS): Add explicit case.
20769 * doc/tm.texi: Regenerate.
20771 2018-01-31 Richard Biener <rguenther@suse.de>
20773 PR tree-optimization/84132
20774 * tree-data-ref.c (analyze_miv_subscript): Properly
20775 check whether evolution_function_is_affine_multivariate_p
20776 before calling gcd_of_steps_may_divide_p.
20778 2018-01-31 Julia Koval <julia.koval@intel.com>
20781 * config/i386/i386.c (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
20782 * config/i386/i386.md (rdpid_rex64) New.
20783 (rdpid): Make 32bit only.
20785 2018-01-29 Aldy Hernandez <aldyh@redhat.com>
20788 * tree-pretty-print.c (dump_generic_node): Handle a TYPE_NAME with
20789 an IDENTIFIER_NODE for FUNCTION_TYPE's.
20791 2018-01-31 Eric Botcazou <ebotcazou@adacore.com>
20794 2018-01-12 Eric Botcazou <ebotcazou@adacore.com>
20796 * config/sparc/sparc.md (vxworks_load_got): Set the GOT register.
20798 2018-01-31 Eric Botcazou <ebotcazou@adacore.com>
20800 PR rtl-optimization/84071
20801 * combine.c (record_dead_and_set_regs_1): Record the source unmodified
20802 for a paradoxical SUBREG on a WORD_REGISTER_OPERATIONS target.
20804 2018-01-31 Claudiu Zissulescu <claziss@synopsys.com>
20806 * config/arc/arc.c (arc_handle_aux_attribute): New function.
20807 (arc_attribute_table): Add 'aux' attribute.
20808 (arc_in_small_data_p): Consider aux like variables.
20809 (arc_is_aux_reg_p): New function.
20810 (arc_asm_output_aligned_decl_local): Ignore 'aux' like variables.
20811 (arc_get_aux_arg): New function.
20812 (prepare_move_operands): Handle aux-register access.
20813 (arc_handle_aux_attribute): New function.
20814 * doc/extend.texi (ARC Variable attributes): Add subsection.
20816 2018-01-31 Claudiu Zissulescu <claziss@synopsys.com>
20818 * config/arc/arc-protos.h (arc_is_uncached_mem_p): Function proto.
20819 * config/arc/arc.c (arc_handle_uncached_attribute): New function.
20820 (arc_attribute_table): Add 'uncached' attribute.
20821 (arc_print_operand): Print '.di' flag for uncached memory
20823 (arc_in_small_data_p): Do not consider for small data the uncached
20825 (arc_is_uncached_mem_p): New function.
20826 * config/arc/predicates.md (compact_store_memory_operand): Check
20827 for uncached memory accesses.
20828 (nonvol_nonimm_operand): Likewise.
20829 * doc/extend.texi (ARC Type Attribute): New subsection.
20831 2018-01-31 Jakub Jelinek <jakub@redhat.com>
20834 * common.opt (falign-functions=, falign-jumps=, falign-labels=,
20835 falign-loops=): Add Optimization flag.
20837 2018-01-30 Jeff Law <law@redhat.com>
20840 * i386.c (ix86_adjust_stack_and_probe_stack_clash): New argument
20841 INT_REGISTERS_SAVED. Check it prior to calling
20842 get_scratch_register_on_entry.
20843 (ix86_adjust_stack_and_probe): Similarly.
20844 (ix86_emit_probe_stack_range): Similarly.
20845 (ix86_expand_prologue): Corresponding changes.
20847 2018-01-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
20850 * config/sol2.h (STARTFILE_ARCH_SPEC): Use -std=c*,
20851 -std=iso9899:199409 instead of -pedantic to select values-Xc.o.
20853 2018-01-30 Vladimir Makarov <vmakarov@redhat.com>
20856 * lra-constraints.c (curr_insn_transform): Process AND in the
20859 2018-01-30 Jakub Jelinek <jakub@redhat.com>
20861 PR rtl-optimization/83986
20862 * sched-deps.c (sched_analyze_insn): For frame related insns, add anti
20863 dependence against last_pending_memory_flush in addition to
20864 pending_jump_insns.
20866 2018-01-30 Alexandre Oliva <aoliva@redhat.com>
20868 PR tree-optimization/81611
20869 * tree-ssa-dom.c (simple_iv_increment_p): Skip intervening
20872 2018-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
20875 * config/rs6000/rs6000.c (rs6000_internal_arg_pointer): Only return
20878 2018-01-30 Richard Biener <rguenther@suse.de>
20879 Jakub Jelinek <jakub@redhat.com>
20881 PR tree-optimization/84111
20882 * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely_1): Skip
20883 inner loops added during recursion, as they don't have up-to-date
20886 2018-01-30 Jan Hubicka <hubicka@ucw.cz>
20889 * ipa-inline.c (can_inline_edge_p): Break out late tests to...
20890 (can_inline_edge_by_limits_p): ... here.
20891 (can_early_inline_edge_p, check_callers,
20892 update_caller_keys, update_callee_keys, recursive_inlining,
20893 add_new_edges_to_heap, speculation_useful_p,
20894 inline_small_functions,
20895 inline_small_functions, flatten_function,
20896 inline_to_all_callers_1): Update.
20898 2018-01-30 Jan Hubicka <hubicka@ucw.cz>
20900 * profile-count.c (profile_count::combine_with_ipa_count): Handle
20903 2018-01-30 Richard Biener <rguenther@suse.de>
20905 PR tree-optimization/83008
20906 * tree-vect-slp.c (vect_analyze_slp_cost_1): Properly cost
20907 invariant and constant vector uses in stmts when they need
20908 more than one stmt.
20910 2018-01-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
20913 * configure.ac (gcc_cv_as_shf_merge): Disable on Solaris 10/x86.
20914 * configure: Regenerate.
20916 2018-01-30 Richard Sandiford <richard.sandiford@linaro.org>
20918 * config/aarch64/aarch64-sve.md (*vec_extract<mode><Vel>_0): New
20920 (*vec_extract<mode><Vel>_v128): Require a nonzero lane number.
20921 Use gen_rtx_REG rather than gen_lowpart.
20923 2018-01-30 Richard Sandiford <richard.sandiford@linaro.org>
20925 * lra-constraints.c (match_reload): Use subreg_lowpart_offset
20926 rather than 0 when creating partial subregs.
20928 2018-01-30 Richard Sandiford <richard.sandiford@linaro.org>
20930 * vec-perm-indices.c (vec_perm_indices::series_p): Give examples
20933 2018-01-29 Michael Meissner <meissner@linux.vnet.ibm.com>
20936 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): If DFmode
20937 and SFmode can go in Altivec registers (-mcpu=power7 for DFmode,
20938 -mcpu=power8 for SFmode) don't set the PRE_INCDEC or PRE_MODIFY
20939 flags. This restores the settings used before the 2017-07-24.
20940 Turning off pre increment/decrement/modify allows IVOPTS to
20941 optimize DF/SF loops where the index is an int.
20943 2018-01-29 Richard Biener <rguenther@suse.de>
20944 Kelvin Nilsen <kelvin@gcc.gnu.org>
20947 * tree-vect-stmts.c (vectorizable_call): Don't call
20948 targetm.vectorize_builtin_md_vectorized_function if callee is
20951 2018-01-22 Carl Love <cel@us.ibm.com>
20953 * doc/extend.tex: Fix typo in second arg in
20954 __builtin_bcdadd_{lt|eq|gt|ov} and __builtin_bcdsub_{lt|eq|gt|ov}.
20956 2018-01-29 Richard Biener <rguenther@suse.de>
20958 PR tree-optimization/84086
20959 * tree-ssanames.c: Include cfgloop.h and tree-scalar-evolution.h.
20960 (flush_ssaname_freelist): When SSA names were released reset
20961 the SCEV hash table.
20963 2018-01-29 Richard Biener <rguenther@suse.de>
20965 PR tree-optimization/84057
20966 * tree-ssa-loop-ivcanon.c (unloop_loops): Deal with already
20967 removed paths when removing edges.
20969 2018-01-27 H.J. Lu <hongjiu.lu@intel.com>
20971 * doc/invoke.texi: Replace -mfunction-return==@var{choice} with
20972 -mfunction-return=@var{choice}.
20974 2018-01-27 Bernd Edlinger <bernd.edlinger@hotmail.de>
20976 PR diagnostic/84034
20977 * diagnostic-show-locus.c (get_line_width_without_trailing_whitespace):
20978 Handle CR like TAB.
20979 (layout::print_source_line): Likewise.
20980 (test_get_line_width_without_trailing_whitespace): Add test cases.
20982 2018-01-27 Jakub Jelinek <jakub@redhat.com>
20984 PR middle-end/84040
20985 * sched-deps.c (sched_macro_fuse_insns): Return immediately for
20988 2018-01-26 Jim Wilson <jimw@sifive.com>
20990 * config/riscv/riscv.h (MAX_FIXED_MODE_SIZE): New.
20992 * config/riscv/elf.h (LIB_SPEC): Don't include -lgloss when nosys.specs
20995 2018-01-26 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20997 * config/aarch64/aarch64.md: Add peepholes for CMP + SUB -> SUBS
20998 and CMP + SUB-immediate -> SUBS.
21000 2018-01-26 Martin Sebor <msebor@redhat.com>
21002 PR tree-optimization/83896
21003 * tree-ssa-strlen.c (get_string_len): Rename...
21004 (get_string_cst_length): ...to this. Return HOST_WIDE_INT.
21005 Avoid assuming length is constant.
21006 (handle_char_store): Use HOST_WIDE_INT for string length.
21008 2018-01-26 Uros Bizjak <ubizjak@gmail.com>
21011 * config/i386/i386.md (*andndi3_doubleword): Add earlyclobber
21012 to (=&r,r,rm) alternative. Add (=r,0,rm) and (=r,r,0) alternatives.
21014 2018-01-26 Richard Biener <rguenther@suse.de>
21016 PR rtl-optimization/84003
21017 * dse.c (record_store): Only record redundant stores when
21018 the earlier store aliases at least all accesses the later one does.
21020 2018-01-26 Jakub Jelinek <jakub@redhat.com>
21022 PR rtl-optimization/83985
21023 * dce.c (deletable_insn_p): Return false for separate shrink wrapping
21024 REG_CFA_RESTORE insns.
21025 (delete_unmarked_insns): Don't ignore separate shrink wrapping
21026 REG_CFA_RESTORE insns here.
21029 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Don't
21030 use SSA_NAME_VAR as base for SSA_NAMEs with non-NULL SSA_NAME_VAR.
21032 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
21034 * config/arc/arc-arch.h (arc_tune_attr): Add ARC_TUNE_CORE_3.
21035 * config/arc/arc.c (arc_sched_issue_rate): Use ARC_TUNE_... .
21036 (arc_init): Likewise.
21037 (arc_override_options): Likewise.
21038 (arc_file_start): Choose Tag_ARC_CPU_variation based on arc_tune
21040 (hwloop_fail): Use TARGET_DBNZ when we want to check for dbnz insn
21042 * config/arc/arc.h (TARGET_DBNZ): Define.
21043 * config/arc/arc.md (attr tune): Add core_3, use ARC_TUNE_... to
21044 properly set the tune attribute.
21045 (dbnz): Use TARGET_DBNZ guard.
21046 * config/arc/arc.opt (mtune): Add core3 option.
21048 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
21050 * config/arc/arc.c (arc_delegitimize_address_0): Refactored to
21051 recognize new pic like addresses.
21052 (arc_delegitimize_address): Clean up.
21054 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
21056 * config/arc/arc-arches.def: Option mrf16 valid for all
21058 * config/arc/arc-c.def (__ARC_RF16__): New predefined macro.
21059 * config/arc/arc-cpus.def (em_mini): New cpu with rf16 on.
21060 * config/arc/arc-options.def (FL_RF16): Add mrf16 option.
21061 * config/arc/arc-tables.opt: Regenerate.
21062 * config/arc/arc.c (arc_conditional_register_usage): Handle
21063 reduced register file case.
21064 (arc_file_start): Set must have build attributes.
21065 * config/arc/arc.h (MAX_ARC_PARM_REGS): Conditional define using
21066 mrf16 option value.
21067 * config/arc/arc.opt (mrf16): Add new option.
21068 * config/arc/elf.h (ATTRIBUTE_PCS): Define.
21069 * config/arc/genmultilib.awk: Handle new mrf16 option.
21070 * config/arc/linux.h (ATTRIBUTE_PCS): Define.
21071 * config/arc/t-multilib: Regenerate.
21072 * doc/invoke.texi (ARC Options): Document mrf16 option.
21074 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
21076 * config/arc/arc-protos.h: Add arc_is_secure_call_p proto.
21077 * config/arc/arc.c (arc_handle_secure_attribute): New function.
21078 (arc_attribute_table): Add 'secure_call' attribute.
21079 (arc_print_operand): Print secure call operand.
21080 (arc_function_ok_for_sibcall): Don't optimize tail calls when
21082 (arc_is_secure_call_p): New function. * config/arc/arc.md
21083 (call_i): Add support for sjli instruction.
21084 (call_value_i): Likewise.
21085 * config/arc/constraints.md (Csc): New constraint.
21087 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
21088 John Eric Martin <John.Martin@emmicro-us.com>
21090 * config/arc/arc-protos.h: Add arc_is_jli_call_p proto.
21091 * config/arc/arc.c (_arc_jli_section): New struct.
21092 (arc_jli_section): New type.
21093 (rc_jli_sections): New static variable.
21094 (arc_handle_jli_attribute): New function.
21095 (arc_attribute_table): Add jli_always and jli_fixed attribute.
21096 (arc_file_end): New function.
21097 (TARGET_ASM_FILE_END): Define.
21098 (arc_print_operand): Reuse 'S' letter for JLI output instruction.
21099 (arc_add_jli_section): New function.
21100 (jli_call_scan): Likewise.
21101 (arc_reorg): Call jli_call_scan.
21102 (arc_output_addsi): Remove 'S' from printing asm operand.
21103 (arc_is_jli_call_p): New function.
21104 * config/arc/arc.md (movqi_insn): Remove 'S' from printing asm
21106 (movhi_insn): Likewise.
21107 (movsi_insn): Likewise.
21108 (movsi_set_cc_insn): Likewise.
21109 (loadqi_update): Likewise.
21110 (load_zeroextendqisi_update): Likewise.
21111 (load_signextendqisi_update): Likewise.
21112 (loadhi_update): Likewise.
21113 (load_zeroextendhisi_update): Likewise.
21114 (load_signextendhisi_update): Likewise.
21115 (loadsi_update): Likewise.
21116 (loadsf_update): Likewise.
21117 (movsicc_insn): Likewise.
21118 (bset_insn): Likewise.
21119 (bxor_insn): Likewise.
21120 (bclr_insn): Likewise.
21121 (bmsk_insn): Likewise.
21122 (bicsi3_insn): Likewise.
21123 (cmpsi_cc_c_insn): Likewise.
21124 (movsi_ne): Likewise.
21125 (movsi_cond_exec): Likewise.
21126 (clrsbsi2): Likewise.
21127 (norm_f): Likewise.
21137 (call_i): Remove 'S' asm letter, add jli instruction.
21138 (call_value_i): Likewise.
21139 * config/arc/arc.op (mjli-always): New option.
21140 * config/arc/constraints.md (Cji): New constraint.
21141 * config/arc/fpx.md (addsf3_fpx): Remove 'S' from printing asm
21143 (subsf3_fpx): Likewise.
21144 (mulsf3_fpx): Likewise.
21145 * config/arc/simdext.md (vendrec_insn): Remove 'S' from printing
21147 * doc/extend.texi (ARC): Document 'jli-always' and 'jli-fixed'
21148 function attrbutes.
21149 * doc/invoke.texi (ARC): Document mjli-always option.
21151 2018-01-26 Sebastian Perta <sebastian.perta@renesas.com>
21153 * config/rl78/rl78.c (rl78_addsi3_internal): If operand 2 is const
21154 avoid addition with 0 and use incw and decw where possible.
21156 2018-01-26 Richard Biener <rguenther@suse.de>
21158 PR tree-optimization/81082
21159 * fold-const.c (fold_plusminus_mult_expr): Do not perform the
21160 association if it requires casting to unsigned.
21161 * match.pd ((A * C) +- (B * C) -> (A+-B)): New patterns derived
21162 from fold_plusminus_mult_expr to catch important cases late when
21163 range info is available.
21165 2018-01-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
21167 * config/i386/sol2.h (USE_HIDDEN_LINKONCE): Remove.
21168 * configure.ac (hidden_linkonce): New test.
21169 * configure: Regenerate.
21170 * config.in: Regenerate.
21172 2018-01-26 Julia Koval <julia.koval@intel.com>
21174 * config/i386/avx512bitalgintrin.h (_mm512_bitshuffle_epi64_mask,
21175 _mm512_mask_bitshuffle_epi64_mask, _mm256_bitshuffle_epi64_mask,
21176 _mm256_mask_bitshuffle_epi64_mask, _mm_bitshuffle_epi64_mask,
21177 _mm_mask_bitshuffle_epi64_mask): Fix type.
21178 * config/i386/i386-builtin-types.def (UHI_FTYPE_V2DI_V2DI_UHI,
21179 USI_FTYPE_V4DI_V4DI_USI): Remove.
21180 * config/i386/i386-builtin.def (__builtin_ia32_vpshufbitqmb512_mask,
21181 __builtin_ia32_vpshufbitqmb256_mask,
21182 __builtin_ia32_vpshufbitqmb128_mask): Fix types.
21183 * config/i386/i386.c (ix86_expand_args_builtin): Remove old types.
21184 * config/i386/sse.md (VI1_AVX512VLBW): Change types.
21186 2018-01-26 Alan Modra <amodra@gmail.com>
21189 * config/rs6000/rs6000-p8swap.c (rtx_is_swappable_p): Exclude
21190 UNSPEC_VBPERMQ. Sort other unspecs.
21192 2018-01-25 David Edelsohn <dje.gcc@gmail.com>
21194 * doc/invoke.texi (PowerPC Options): Document 'native' cpu type.
21196 2018-01-25 Jan Hubicka <hubicka@ucw.cz>
21198 PR middle-end/83055
21199 * predict.c (drop_profile): Do not push/pop cfun; update also
21201 (handle_missing_profiles): Fix logic looking for zero profiles.
21203 2018-01-25 Jakub Jelinek <jakub@redhat.com>
21205 PR middle-end/83977
21206 * ipa-fnsummary.c (compute_fn_summary): Clear can_change_signature
21207 on functions with #pragma omp declare simd or functions with simd
21209 * omp-simd-clone.c (expand_simd_clones): Revert 2018-01-24 change.
21210 * config/i386/i386.c (ix86_simd_clone_compute_vecsize_and_simdlen):
21211 Remove trailing \n from warning_at calls.
21213 2018-01-25 Tom de Vries <tom@codesourcery.com>
21216 * config/nvptx/nvptx.c (nvptx_single): Add exit insn after noreturn call
21217 for neutered workers.
21219 2018-01-24 Joseph Myers <joseph@codesourcery.com>
21222 * config/m68k/m68k.c (m68k_promote_function_mode): New function.
21223 (TARGET_PROMOTE_FUNCTION_MODE): New macro.
21225 2018-01-24 Jeff Law <law@redhat.com>
21228 * i386.c (get_probe_interval): Move to earlier point.
21229 (ix86_compute_frame_layout): If -fstack-clash-protection and
21230 the frame is larger than the probe interval, then use pushes
21231 to save registers rather than reg->mem moves.
21232 (ix86_expand_prologue): Remove conditional for int_registers_saved
21235 2018-01-24 Vladimir Makarov <vmakarov@redhat.com>
21238 * ira-build.c (setup_min_max_allocno_live_range_point): Set up
21239 min/max for never referenced object.
21241 2018-01-24 Jakub Jelinek <jakub@redhat.com>
21243 PR middle-end/83977
21244 * tree.c (free_lang_data_in_decl): Don't clear DECL_ABSTRACT_ORIGIN
21246 * omp-low.c (create_omp_child_function): Remove "omp declare simd"
21247 attributes from DECL_ATTRIBUTES (decl) without affecting
21248 DECL_ATTRIBUTES (current_function_decl).
21249 * omp-simd-clone.c (expand_simd_clones): Ignore DECL_ARTIFICIAL
21250 functions with non-NULL DECL_ABSTRACT_ORIGIN.
21252 2018-01-24 Richard Sandiford <richard.sandiford@linaro.org>
21254 PR tree-optimization/83979
21255 * fold-const.c (fold_comparison): Use constant_boolean_node
21256 instead of boolean_{true,false}_node.
21258 2018-01-24 Jan Hubicka <hubicka@ucw.cz>
21260 * ipa-profile.c (ipa_propagate_frequency_1): Fix logic skipping calls
21263 2018-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21265 * config/rs6000/rs6000.md (*call_indirect_nonlocal_sysv<mode>):
21266 Simplify the clause that sets the length attribute.
21267 (*call_value_indirect_nonlocal_sysv<mode>): Likewise.
21268 (*sibcall_nonlocal_sysv<mode>): Clean up code block; simplify the
21269 clause that sets the length attribute.
21270 (*sibcall_value_nonlocal_sysv<mode>): Likewise.
21272 2018-01-24 Tom de Vries <tom@codesourcery.com>
21275 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG_2): Define to 1.
21276 (nvptx_pc_set, nvptx_condjump_label): New function. Copy from jump.c.
21277 Add strict parameter.
21278 (prevent_branch_around_nothing): Insert dummy insn between branch to
21279 label and label with no ptx insn inbetween.
21280 * config/nvptx/nvptx.md (define_insn "fake_nop"): New insn.
21282 2018-01-24 Tom de Vries <tom@codesourcery.com>
21285 * config/nvptx/nvptx.c (nvptx_single): Add exit insn after noreturn call
21286 for neutered threads in warp.
21287 * config/nvptx/nvptx.md (define_insn "exit"): New insn.
21289 2018-01-24 Richard Biener <rguenther@suse.de>
21291 PR tree-optimization/83176
21292 * tree-chrec.c (chrec_fold_plus_1): Handle (signed T){(T) .. }
21295 2018-01-24 Richard Biener <rguenther@suse.de>
21297 PR tree-optimization/82819
21298 * graphite-isl-ast-to-gimple.c (binary_op_to_tree): Avoid
21299 code generating pluses that are no-ops in the target precision.
21301 2018-01-24 Richard Biener <rguenther@suse.de>
21303 PR middle-end/84000
21304 * tree-cfg.c (replace_loop_annotate): Handle annot_expr_parallel_kind.
21306 2018-01-23 Jan Hubicka <hubicka@ucw.cz>
21308 * cfgcleanup.c (try_crossjump_to_edge): Use combine_with_count
21309 to merge probabilities.
21310 * predict.c (probably_never_executed): Also mark as cold functions
21311 with global 0 profile and guessed local profile.
21312 * profile-count.c (profile_probability::combine_with_count): New
21314 * profile-count.h (profile_probability::operator*,
21315 profile_probability::operator*=, profile_probability::operator/,
21316 profile_probability::operator/=): Reduce precision to adjusted
21317 and set value to guessed on contradictory divisions.
21318 (profile_probability::combine_with_freq): Remove.
21319 (profile_probability::combine_wiht_count): Declare.
21320 (profile_count::force_nonzero):: Set to adjusted.
21321 (profile_count::probability_in):: Set quality to adjusted.
21322 * tree-ssa-tail-merge.c (replace_block_by): Use
21323 combine_with_count.
21325 2018-01-23 Andrew Waterman <andrew@sifive.com>
21326 Jim Wilson <jimw@sifive.com>
21328 * config/riscv/riscv.c (riscv_stack_boundary): New.
21329 (riscv_option_override): Set riscv_stack_boundary. Handle
21330 riscv_preferred_stack_boundary_arg.
21331 * config/riscv/riscv.h (MIN_STACK_BOUNDARY, ABI_STACK_BOUNDARY): New.
21332 (BIGGEST_ALIGNMENT): Set to STACK_BOUNDARY.
21333 (STACK_BOUNDARY): Set to riscv_stack_boundary.
21334 (RISCV_STACK_ALIGN): Use STACK_BOUNDARY.
21335 * config/riscv/riscv.opt (mpreferred-stack-boundary): New.
21336 * doc/invoke.tex (RISC-V Options): Add -mpreferred-stack-boundary.
21338 2018-01-23 H.J. Lu <hongjiu.lu@intel.com>
21341 * config/i386/i386.c (ix86_expand_prologue): Use cost reference
21342 of struct ix86_frame.
21343 (ix86_expand_epilogue): Likewise. Add a local variable for
21344 the reg_save_offset field in struct ix86_frame.
21346 2018-01-23 Bin Cheng <bin.cheng@arm.com>
21348 PR tree-optimization/82604
21349 * tree-loop-distribution.c (enum partition_kind): New enum item
21350 PKIND_PARTIAL_MEMSET.
21351 (partition_builtin_p): Support above new enum item.
21352 (generate_code_for_partition): Ditto.
21353 (compute_access_range): Differentiate cases that equality can be
21354 proven at all loops, the innermost loops or no loops.
21355 (classify_builtin_st, classify_builtin_ldst): Adjust call to above
21356 function. Set PKIND_PARTIAL_MEMSET for partition appropriately.
21357 (finalize_partitions, distribute_loop): Don't fuse partition of
21358 PKIND_PARTIAL_MEMSET kind when distributing 3-level loop nest.
21359 (prepare_perfect_loop_nest): Distribute 3-level loop nest only if
21360 parloop is enabled.
21362 2018-01-23 Martin Liska <mliska@suse.cz>
21364 * predict.def (PRED_INDIR_CALL): Set probability to PROB_EVEN in
21365 order to ignore the predictor.
21366 (PRED_POLYMORPHIC_CALL): Likewise.
21367 (PRED_RECURSIVE_CALL): Likewise.
21369 2018-01-23 Martin Liska <mliska@suse.cz>
21371 * tree-profile.c (tree_profiling): Print function header to
21372 aware reader which function we are working on.
21373 * value-prof.c (gimple_find_values_to_profile): Do not print
21374 not interesting value histograms.
21376 2018-01-23 Martin Liska <mliska@suse.cz>
21378 * profile-count.h (enum profile_quality): Add
21379 profile_uninitialized as the first value. Do not number values
21380 as they are zero based.
21381 (profile_count::verify): Update sanity check.
21382 (profile_probability::verify): Likewise.
21384 2018-01-23 Nathan Sidwell <nathan@acm.org>
21386 * doc/invoke.texi (ffor-scope): Deprecate.
21388 2018-01-23 David Malcolm <dmalcolm@redhat.com>
21390 PR tree-optimization/83510
21391 * domwalk.c (set_all_edges_as_executable): New function.
21392 (dom_walker::dom_walker): Convert bool param
21393 "skip_unreachable_blocks" to enum reachability. Move setup of
21394 edge flags to set_all_edges_as_executable and only do it when
21395 reachability is REACHABLE_BLOCKS.
21396 * domwalk.h (enum dom_walker::reachability): New enum.
21397 (dom_walker::dom_walker): Convert bool param
21398 "skip_unreachable_blocks" to enum reachability.
21399 (set_all_edges_as_executable): New decl.
21400 * graphite-scop-detection.c (gather_bbs::gather_bbs): Convert
21401 from false for "skip_unreachable_blocks" to ALL_BLOCKS for
21403 * tree-ssa-dom.c (dom_opt_dom_walker::dom_opt_dom_walker): Likewise,
21404 but converting true to REACHABLE_BLOCKS.
21405 * tree-ssa-sccvn.c (sccvn_dom_walker::sccvn_dom_walker): Likewise.
21407 (check_array_bounds_dom_walker::check_array_bounds_dom_walker):
21408 Likewise, but converting it to REACHABLE_BLOCKS_PRESERVING_FLAGS.
21409 (vrp_dom_walker::vrp_dom_walker): Likewise, but converting it to
21411 (vrp_prop::vrp_finalize): Call set_all_edges_as_executable
21412 if check_all_array_refs will be called.
21414 2018-01-23 David Malcolm <dmalcolm@redhat.com>
21416 * tree.c (selftest::test_location_wrappers): Add more test
21419 2018-01-23 David Malcolm <dmalcolm@redhat.com>
21421 * sbitmap.c (selftest::test_set_range): Fix memory leaks.
21422 (selftest::test_bit_in_range): Likewise.
21424 2018-01-23 Richard Sandiford <richard.sandiford@linaro.org>
21427 * doc/sourcebuild.texi (vect_float): Say that the selector
21428 only describes the situation when -funsafe-math-optimizations is on.
21429 (vect_float_strict): Document.
21431 2018-01-23 Richard Sandiford <richard.sandiford@linaro.org>
21433 PR tree-optimization/83965
21434 * tree-vect-patterns.c (vect_reassociating_reduction_p): New function.
21435 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern): Use it
21436 instead of checking only for a reduction.
21437 (vect_recog_widen_sum_pattern): Likewise.
21439 2018-01-23 Jan Hubicka <hubicka@ucw.cz>
21441 * predict.c (probably_never_executed): Only use precise profile info.
21442 (compute_function_frequency): Skip after inlining hack since we now
21443 have quality checking.
21445 2018-01-23 Jan Hubicka <hubicka@ucw.cz>
21447 * profile-count.h (profile_probability::very_unlikely,
21448 profile_probability::unlikely, profile_probability::even): Set
21449 precision to guessed.
21451 2018-01-23 Richard Biener <rguenther@suse.de>
21453 PR tree-optimization/83963
21454 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
21455 Properly terminate dominator walk when crossing the exit edge not
21456 when visiting its source block.
21458 2018-01-23 Jakub Jelinek <jakub@redhat.com>
21461 * tree.c (maybe_wrap_with_location): Use NON_LVALUE_EXPR rather than
21462 VIEW_CONVERT_EXPR to wrap CONST_DECLs.
21464 2018-01-22 Jakub Jelinek <jakub@redhat.com>
21466 PR tree-optimization/83957
21467 * omp-expand.c (expand_omp_for_generic): Ignore virtual PHIs. Remove
21468 semicolon after for body surrounded by braces.
21470 PR tree-optimization/83081
21471 * profile-count.h (profile_probability::split): New method.
21472 * dojump.c (do_jump_1) <case TRUTH_ANDIF_EXPR, case TRUTH_ORIF_EXPR>:
21473 Use profile_probability::split.
21474 (do_compare_rtx_and_jump): Fix adjustment of probabilities
21475 when splitting a single conditional jump into 2.
21477 2018-01-22 David Malcolm <dmalcolm@redhat.com>
21479 PR tree-optimization/69452
21480 * tree-ssa-loop-im.c (class move_computations_dom_walker): Remove
21483 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
21485 * config/rl78/rl78-expand.md (bswaphi2): New define_expand.
21486 * config/rl78/rl78-virt.md (*bswaphi2_virt): New define_insn.
21487 * config/rl78/rl78-real.md (*bswaphi2_real): New define_insn.
21489 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
21491 * config/rl78/rl78-protos.h (rl78_split_movdi): New function
21493 * config/rl78/rl78.md (movdi): New define_expand.
21494 * config/rl78/rl78.c (rl78_split_movdi): New function.
21496 2018-01-22 Michael Meissner <meissner@linux.vnet.ibm.com>
21499 * config/rs6000/rs6000-protos.h (rs6000_split_signbit): Delete,
21501 * config/rs6000/rs6000.c (rs6000_split_signbit): Likewise.
21502 * config/rs6000/rs6000.md (signbit<mode>2): Change code for IEEE
21503 128-bit to produce an UNSPEC move to get the double word with the
21504 signbit and then a shift directly to do signbit.
21505 (signbit<mode>2_dm): Replace old IEEE 128-bit signbit
21506 implementation with a new version that just does either a direct
21507 move or a regular move. Move memory interface to separate insns.
21508 Move insns so they are next to the expander.
21509 (signbit<mode>2_dm_mem_be): New combiner insns to combine load
21510 with signbit move. Split big and little endian case.
21511 (signbit<mode>2_dm_mem_le): Likewise.
21512 (signbit<mode>2_dm_<su>ext): Delete, no longer used.
21513 (signbit<mode>2_dm2): Likewise.
21515 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
21517 * config/rl78/rl78.md (anddi3): New define_expand.
21519 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
21521 * config/rl78/rl78.md (umindi3): New define_expand.
21523 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
21525 * config/rl78/rl78.md (smindi3): New define_expand.
21527 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
21529 * config/rl78/rl78.md (smaxdi3): New define_expand.
21531 2018-01-22 Carl Love <cel@us.ibm.com>
21533 * config/rs6000/rs6000-builtin.def (ST_ELEMREV_V1TI, LD_ELEMREV_V1TI,
21534 LVX_V1TI): Add macro expansion.
21535 * config/rs6000/rs6000-c.c (altivec_builtin_types): Add argument
21536 definitions for VSX_BUILTIN_VEC_XST_BE, VSX_BUILTIN_VEC_ST,
21537 VSX_BUILTIN_VEC_XL, LD_ELEMREV_V1TI builtins.
21538 * config/rs6000/rs6000-p8swap.c (insn_is_swappable_p);
21539 Change check to determine if the instruction is a byte reversing
21540 entry. Fix typo in comment.
21541 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case entry
21542 for VSX_BUILTIN_ST_ELEMREV_V1TI and VSX_BUILTIN_LD_ELEMREV_V1TI.
21543 Add def_builtin calls for new builtins.
21544 * config/rs6000/vsx.md (vsx_st_elemrev_v1ti, vsx_ld_elemrev_v1ti):
21545 Add define_insn expansion.
21547 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
21549 * config/rl78/rl78.md (umaxdi3): New define_expand.
21551 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
21553 * config/rl78/rl78.c (rl78_note_reg_set): Fixed dead reg check
21554 for non-QImode registers.
21556 2018-01-22 Richard Biener <rguenther@suse.de>
21558 PR tree-optimization/83963
21559 * graphite-scop-detection.c (scop_detection::get_sese): Delay
21560 including the loop exit block.
21561 (scop_detection::merge_sese): Likewise.
21562 (scop_detection::add_scop): Do it here instead.
21564 2018-01-22 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21566 * doc/sourcebuild.texi (arm_softfloat): Document.
21568 2018-01-21 John David Anglin <danglin@gcc.gnu.org>
21571 * config/pa/pa.c (pa_function_ok_for_sibcall): Use
21572 targetm.binds_local_p instead of TREE_PUBLIC to check local binding.
21573 Move TARGET_PORTABLE_RUNTIME check after TARGET_64BIT check.
21575 2018-01-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21576 David Edelsohn <dje.gcc@gmail.com>
21579 * config/rs6000/rs6000.md (*call_indirect_nonlocal_sysv<mode>):
21580 Change "crset eq" to "crset 2".
21581 (*call_value_indirect_nonlocal_sysv<mode>): Likewise.
21582 (*call_indirect_aix<mode>_nospec): Likewise.
21583 (*call_value_indirect_aix<mode>_nospec): Likewise.
21584 (*call_indirect_elfv2<mode>_nospec): Likewise.
21585 (*call_value_indirect_elfv2<mode>_nospec): Likewise.
21586 (*sibcall_nonlocal_sysv<mode>): Change "crset eq" to "crset 2";
21587 change assembly output from . to $.
21588 (*sibcall_value_nonlocal_sysv<mode>): Likewise.
21589 (indirect_jump<mode>_nospec): Change assembly output from . to $.
21590 (*tablejump<mode>_internal1_nospec): Likewise.
21592 2018-01-21 Oleg Endo <olegendo@gcc.gnu.org>
21595 * config/sh/sh_optimize_sett_clrt.cc:
21596 Use INCLUDE_ALGORITHM and INCLUDE_VECTOR instead of direct includes.
21598 2018-01-20 Richard Sandiford <richard.sandiford@linaro.org>
21600 PR tree-optimization/83940
21601 * tree-vect-stmts.c (vect_truncate_gather_scatter_offset): Set
21602 offset_dt to vect_constant_def rather than vect_unknown_def_type.
21603 (vect_check_load_store_mask): Add a mask_dt_out parameter and
21604 use it to pass back the definition type.
21605 (vect_check_store_rhs): Likewise rhs_dt_out.
21606 (vect_build_gather_load_calls): Add a mask_dt argument and use
21607 it instead of a call to vect_is_simple_use.
21608 (vectorizable_store): Update calls to vect_check_load_store_mask
21609 and vect_check_store_rhs. Use the dt returned by the latter instead
21610 of scatter_src_dt. Use the cached mask_dt and gs_info.offset_dt
21611 instead of calls to vect_is_simple_use. Pass the scalar rather
21612 than the vector operand to vect_is_simple_use when handling
21613 second and subsequent copies of an rhs value.
21614 (vectorizable_load): Update calls to vect_check_load_store_mask
21615 and vect_build_gather_load_calls. Use the cached mask_dt and
21616 gs_info.offset_dt instead of calls to vect_is_simple_use.
21618 2018-01-20 Jakub Jelinek <jakub@redhat.com>
21620 PR middle-end/83945
21621 * tree-emutls.c: Include gimplify.h.
21622 (lower_emutls_2): New function.
21623 (lower_emutls_1): If ADDR_EXPR is a gimple invariant and walk_tree
21624 with lower_emutls_2 callback finds some TLS decl in it, unshare_expr
21625 it before further processing.
21628 * simplify-rtx.c (simplify_binary_operation_1) <case UMOD>: Use
21629 UINTVAL (trueop1) instead of INTVAL (op1).
21631 2018-01-19 Jakub Jelinek <jakub@redhat.com>
21635 * dwarf2cfi.c (DEFAULT_INCOMING_FRAME_SP_OFFSET): Define to
21636 INCOMING_FRAME_SP_OFFSET if not defined.
21637 (scan_trace): Add ENTRY argument. If true and
21638 DEFAULT_INCOMING_FRAME_SP_OFFSET != INCOMING_FRAME_SP_OFFSET,
21639 emit a note to adjust the CFA offset.
21640 (create_cfi_notes): Adjust scan_trace callers.
21641 (create_cie_data): Use DEFAULT_INCOMING_FRAME_SP_OFFSET rather than
21642 INCOMING_FRAME_SP_OFFSET in the CIE.
21643 * config/i386/i386.h (DEFAULT_INCOMING_FRAME_SP_OFFSET): Define.
21644 * config/stormy16/stormy16.h (DEFAULT_INCOMING_FRAME_SP_OFFSET):
21646 * doc/tm.texi.in (DEFAULT_INCOMING_FRAME_SP_OFFSET): Document.
21647 * doc/tm.texi: Regenerated.
21649 2018-01-19 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21651 PR rtl-optimization/83147
21652 * lra-constraints.c (remove_inheritance_pseudos): Use
21653 lra_substitute_pseudo_within_insn.
21655 2018-01-19 Tom de Vries <tom@codesourcery.com>
21656 Cesar Philippidis <cesar@codesourcery.com>
21659 * config/nvptx/nvptx.c (nvptx_single): Fix jit workaround.
21661 2018-01-19 Cesar Philippidis <cesar@codesourcery.com>
21664 * config/nvptx/nvptx.c (output_init_frag): Don't use generic address
21665 spaces for function labels.
21667 2018-01-19 Martin Liska <mliska@suse.cz>
21669 * predict.def (PRED_LOOP_EXIT): Change from 85 to 89.
21670 (PRED_LOOP_EXIT_WITH_RECURSION): Change from 72 to 78.
21671 (PRED_LOOP_EXTRA_EXIT): Change from 83 to 67.
21672 (PRED_OPCODE_POSITIVE): Change from 64 to 59.
21673 (PRED_TREE_OPCODE_POSITIVE): Change from 64 to 59.
21674 (PRED_CONST_RETURN): Change from 69 to 65.
21675 (PRED_NULL_RETURN): Change from 91 to 71.
21676 (PRED_LOOP_IV_COMPARE_GUESS): Change from 98 to 64.
21677 (PRED_LOOP_GUARD): Change from 66 to 73.
21679 2018-01-19 Martin Liska <mliska@suse.cz>
21681 * predict.c (predict_insn_def): Add new assert.
21682 (struct branch_predictor): Change type to signed integer.
21683 (test_prediction_value_range): Amend test to cover
21684 PROB_UNINITIALIZED.
21685 * predict.def (PRED_LOOP_ITERATIONS): Use the new constant.
21686 (PRED_LOOP_ITERATIONS_GUESSED): Likewise.
21687 (PRED_LOOP_ITERATIONS_MAX): Likewise.
21688 (PRED_LOOP_IV_COMPARE): Likewise.
21689 * predict.h (PROB_UNINITIALIZED): Define new constant.
21691 2018-01-19 Martin Liska <mliska@suse.cz>
21693 * predict.c (dump_prediction): Add new format for
21694 analyze_brprob.py script which is enabled with -details
21696 * profile-count.h (precise_p): New function.
21698 2018-01-19 Richard Sandiford <richard.sandiford@linaro.org>
21700 PR tree-optimization/83922
21701 * tree-vect-loop.c (vect_verify_full_masking): Return false if
21702 there are no statements that need masking.
21703 (vect_active_double_reduction_p): New function.
21704 (vect_analyze_loop_operations): Use it when handling phis that
21705 are not in the loop header.
21707 2018-01-19 Richard Sandiford <richard.sandiford@linaro.org>
21709 PR tree-optimization/83914
21710 * tree-vect-loop.c (vectorizable_induction): Don't convert
21711 init_expr or apply the peeling adjustment for inductions
21712 that are nested within the vectorized loop.
21714 2018-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21716 * config/arm/thumb2.md (*thumb2_negsi2_short): Use RSB mnemonic
21719 2018-01-18 Jakub Jelinek <jakub@redhat.com>
21723 * function.h (gimplify_parameters): Add gimple_seq * argument.
21724 * function.c: Include gimple.h and options.h.
21725 (gimplify_parameters): Add cleanup argument, add CLOBBER stmts
21726 for the added local temporaries if needed.
21727 * gimplify.c (gimplify_body): Adjust gimplify_parameters caller,
21728 if there are any parameter cleanups, wrap whole body into a
21729 try/finally with the cleanups.
21731 2018-01-18 Wilco Dijkstra <wdijkstr@arm.com>
21734 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
21735 Use GET_MODE_CLASS for scalar floating point.
21737 2018-01-18 Jan Hubicka <hubicka@ucw.cz>
21741 * cgraphclones.c (cgraph_node::create_version_clone_with_body):
21742 Fix call of call_cgraph_insertion_hooks.
21744 2018-01-18 Martin Sebor <msebor@redhat.com>
21746 * doc/invoke.texi (-Wclass-memaccess): Tweak text.
21748 2018-01-18 Jan Hubicka <hubicka@ucw.cz>
21751 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Update edge
21754 2018-01-18 Boris Kolpackov <boris@codesynthesis.com>
21757 * common.opt: (-ffile-prefix-map): New option.
21758 * opts.c (common_handle_option): Defer it.
21759 * opts-global.c (handle_common_deferred_options): Handle it.
21760 * debug.h (remap_debug_filename, add_debug_prefix_map): Move to...
21761 * file-prefix-map.h: New file.
21762 (remap_debug_filename, add_debug_prefix_map): ...here.
21763 (add_macro_prefix_map, add_file_prefix_map, remap_macro_filename): New.
21764 * final.c (debug_prefix_map, add_debug_prefix_map
21765 remap_debug_filename): Move to...
21766 * file-prefix-map.c: New file.
21767 (file_prefix_map, add_prefix_map, remap_filename) ...here and rename,
21768 generalize, get rid of alloca(), use strrchr() instead of strchr().
21769 (add_macro_prefix_map, add_debug_prefix_map, add_file_prefix_map):
21770 Implement in terms of add_prefix_map().
21771 (remap_macro_filename, remap_debug_filename): Implement in term of
21773 * Makefile.in (OBJS, PLUGIN_HEADERS): Add new files.
21774 * builtins.c (fold_builtin_FILE): Call remap_macro_filename().
21775 * dbxout.c: Include file-prefix-map.h.
21776 * varasm.c: Likewise.
21777 * vmsdbgout.c: Likewise.
21778 * xcoffout.c: Likewise.
21779 * dwarf2out.c: Likewise plus omit new options from DW_AT_producer.
21780 * doc/cppopts.texi (-fmacro-prefix-map): Document.
21781 * doc/invoke.texi (-ffile-prefix-map): Document.
21782 (-fdebug-prefix-map): Update description.
21784 2018-01-18 Martin Liska <mliska@suse.cz>
21786 * config/i386/i386.c (indirect_thunk_name): Document that also
21788 (output_indirect_thunk): Document why both instructions
21789 (pause and lfence) are generated.
21791 2018-01-18 Richard Biener <rguenther@suse.de>
21793 PR tree-optimization/83887
21794 * graphite-scop-detection.c
21795 (scop_detection::get_nearest_dom_with_single_entry): Remove.
21796 (scop_detection::get_nearest_pdom_with_single_exit): Likewise.
21797 (scop_detection::merge_sese): Re-implement with a flood-fill
21798 algorithm that properly finds a SESE region if it exists.
21800 2018-01-18 Jakub Jelinek <jakub@redhat.com>
21803 * match.pd ((P + A) - P, P - (P + A), (P + A) - (P + B)): For
21804 pointer_diff optimizations use view_convert instead of convert.
21806 2018-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21808 * config/rs6000/rs6000.md (*call_indirect_nonlocal_sysv<mode>):
21809 Generate different code for -mno-speculate-indirect-jumps.
21810 (*call_value_indirect_nonlocal_sysv<mode>): Likewise.
21811 (*call_indirect_aix<mode>): Disable for
21812 -mno-speculate-indirect-jumps.
21813 (*call_indirect_aix<mode>_nospec): New define_insn.
21814 (*call_value_indirect_aix<mode>): Disable for
21815 -mno-speculate-indirect-jumps.
21816 (*call_value_indirect_aix<mode>_nospec): New define_insn.
21817 (*sibcall_nonlocal_sysv<mode>): Generate different code for
21818 -mno-speculate-indirect-jumps.
21819 (*sibcall_value_nonlocal_sysv<mode>): Likewise.
21821 2018-01-17 Michael Meissner <meissner@linux.vnet.ibm.com>
21823 * config/rs6000/rs6000.c (rs6000_emit_move): If we load or store a
21824 long double type, set the flags for noting the default long double
21825 type, even if we don't pass or return a long double type.
21827 2018-01-17 Jan Hubicka <hubicka@ucw.cz>
21830 * ipa-inline.c (flatten_function): Do not overwrite final inlining
21833 2018-01-17 Will Schmidt <will_schmidt@vnet.ibm.com>
21835 * config/rs6000/rs6000.c (rs6000_gimple_builtin): Add gimple folding
21836 support for merge[hl].
21837 (fold_mergehl_helper): New helper function.
21838 (tree-vector-builder.h): New #include for tree_vector_builder usage.
21839 * config/rs6000/altivec.md (altivec_vmrghw_direct): Add xxmrghw insn.
21840 (altivec_vmrglw_direct): Add xxmrglw insn.
21842 2018-01-17 Andrew Waterman <andrew@sifive.com>
21844 * config/riscv/riscv.c (riscv_conditional_register_usage): If
21845 UNITS_PER_FP_ARG is 0, set call_used_regs to 1 for all FP regs.
21847 2018-01-17 David Malcolm <dmalcolm@redhat.com>
21850 * ipa-devirt.c (add_type_duplicate): When comparing memory layout,
21851 call the lto_location_cache before reading the
21852 DECL_SOURCE_LOCATION of the types.
21854 2018-01-17 Wilco Dijkstra <wdijkstr@arm.com>
21855 Richard Sandiford <richard.sandiford@linaro.org>
21857 * config/aarch64/aarch64.md (movti_aarch64): Use Uti constraint.
21858 * config/aarch64/aarch64.c (aarch64_mov128_immediate): New function.
21859 (aarch64_legitimate_constant_p): Just support CONST_DOUBLE
21860 SF/DF/TF mode to avoid creating illegal CONST_WIDE_INT immediates.
21861 * config/aarch64/aarch64-protos.h (aarch64_mov128_immediate):
21863 * config/aarch64/constraints.md (aarch64_movti_operand):
21865 * config/aarch64/predicates.md (Uti): Add new constraint.
21867 2018-01-17 Carl Love <cel@us.ibm.com>
21869 * config/rs6000/vsx.md (define_expand xl_len_r,
21870 define_expand stxvl, define_expand *stxvl): Add match_dup argument.
21871 (define_insn): Add, match_dup 1 argument to define_insn stxvll and
21873 (define_expand, define_insn): Move the shift left from the
21874 define_insn to the define_expand for lxvl and stxvl instructions.
21875 * config/rs6000/rs6000-builtin.def (BU_P9V_64BIT_VSX_2): Change LXVL
21876 and XL_LEN_R definitions to PURE.
21878 2018-01-17 Uros Bizjak <ubizjak@gmail.com>
21880 * config/i386/i386.c (indirect_thunk_name): Declare regno
21881 as unsigned int. Compare regno with INVALID_REGNUM.
21882 (output_indirect_thunk): Ditto.
21883 (output_indirect_thunk_function): Ditto.
21884 (ix86_code_end): Declare regno as unsigned int. Use INVALID_REGNUM
21885 in the call to output_indirect_thunk_function.
21887 2018-01-17 Richard Sandiford <richard.sandiford@linaro.org>
21889 PR middle-end/83884
21890 * expr.c (expand_expr_real_1): Use the size of GET_MODE (op0)
21891 rather than the size of inner_type to determine the stack slot size
21892 when handling VIEW_CONVERT_EXPRs on strict-alignment targets.
21894 2018-01-16 Sebastian Peryt <sebastian.peryt@intel.com>
21897 * config/i386/i386.c (ix86_option_override_internal): Add PTA_RDRND
21900 2018-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
21902 * config.gcc (powerpc*-linux*-*): Add support for 64-bit little
21903 endian Linux systems to optionally enable multilibs for selecting
21904 the long double type if the user configured an explicit type.
21905 * config/rs6000/rs6000.h (TARGET_IEEEQUAD_MULTILIB): Indicate we
21906 have no long double multilibs if not defined.
21907 * config/rs6000/rs6000.c (rs6000_option_override_internal): Do not
21908 warn if the user used -mabi={ieee,ibm}longdouble and we built
21909 multilibs for long double.
21910 * config/rs6000/linux64.h (MULTILIB_DEFAULTS_IEEE): Define as the
21911 appropriate multilib option.
21912 (MULTILIB_DEFAULTS): Add MULTILIB_DEFAULTS_IEEE to the default
21914 * config/rs6000/t-ldouble-linux64le-ibm: New configuration files
21915 for building long double multilibs.
21916 * config/rs6000/t-ldouble-linux64le-ieee: Likewise.
21918 2018-01-16 John David Anglin <danglin@gcc.gnu.org>
21920 * config.gcc (hppa*-*-linux*): Change callee copies ABI to caller
21923 * config/pa.h (MALLOC_ABI_ALIGNMENT): Set 32-bit alignment default to
21925 * config/pa/pa32-linux.h (MALLOC_ABI_ALIGNMENT): Set alignment to
21928 * config/pa/som.h (ASM_DECLARE_FUNCTION_NAME): Cleanup type and mode
21931 * config/pa/pa.c (pa_function_arg_size): Apply CEIL to GET_MODE_SIZE
21934 2018-01-16 Eric Botcazou <ebotcazou@adacore.com>
21936 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): For an
21937 ADDR_EXPR, do not count the offset of a COMPONENT_REF twice.
21939 2018-01-16 Kelvin Nilsen <kelvin@gcc.gnu.org>
21941 * config/rs6000/rs6000-p8swap.c (rs6000_gen_stvx): Generate
21942 different rtl trees depending on TARGET_64BIT.
21943 (rs6000_gen_lvx): Likewise.
21945 2018-01-16 Eric Botcazou <ebotcazou@adacore.com>
21947 * config/visium/visium.md (nop): Tweak comment.
21948 (hazard_nop): Likewise.
21950 2018-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21952 * config/rs6000/rs6000.c (rs6000_opt_vars): Add entry for
21953 -mspeculate-indirect-jumps.
21954 * config/rs6000/rs6000.md (*call_indirect_elfv2<mode>): Disable
21955 for -mno-speculate-indirect-jumps.
21956 (*call_indirect_elfv2<mode>_nospec): New define_insn.
21957 (*call_value_indirect_elfv2<mode>): Disable for
21958 -mno-speculate-indirect-jumps.
21959 (*call_value_indirect_elfv2<mode>_nospec): New define_insn.
21960 (indirect_jump): Emit different RTL for
21961 -mno-speculate-indirect-jumps.
21962 (*indirect_jump<mode>): Disable for
21963 -mno-speculate-indirect-jumps.
21964 (*indirect_jump<mode>_nospec): New define_insn.
21965 (tablejump): Emit different RTL for
21966 -mno-speculate-indirect-jumps.
21967 (tablejumpsi): Disable for -mno-speculate-indirect-jumps.
21968 (tablejumpsi_nospec): New define_expand.
21969 (tablejumpdi): Disable for -mno-speculate-indirect-jumps.
21970 (tablejumpdi_nospec): New define_expand.
21971 (*tablejump<mode>_internal1): Disable for
21972 -mno-speculate-indirect-jumps.
21973 (*tablejump<mode>_internal1_nospec): New define_insn.
21974 * config/rs6000/rs6000.opt (mspeculate-indirect-jumps): New
21977 2018-01-16 Artyom Skrobov tyomitch@gmail.com
21979 * caller-save.c (insert_save): Drop unnecessary parameter. All
21982 2018-01-16 Jakub Jelinek <jakub@redhat.com>
21983 Richard Biener <rguenth@suse.de>
21986 * gimplify.c (gimplify_one_sizepos): For is_gimple_constant (expr)
21987 return early, inline manually is_gimple_sizepos. Make sure if we
21988 call gimplify_expr we don't end up with a gimple constant.
21989 * tree.c (variably_modified_type_p): Don't return true for
21990 is_gimple_constant (_t). Inline manually is_gimple_sizepos.
21991 * gimplify.h (is_gimple_sizepos): Remove.
21993 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
21995 PR tree-optimization/83857
21996 * tree-vect-loop.c (vect_analyze_loop_operations): Don't call
21997 vectorizable_live_operation for pure SLP statements.
21998 (vectorizable_live_operation): Handle PHIs.
22000 2018-01-16 Richard Biener <rguenther@suse.de>
22002 PR tree-optimization/83867
22003 * tree-vect-stmts.c (vect_transform_stmt): Precompute
22004 nested_in_vect_loop_p since the scalar stmt may get invalidated.
22006 2018-01-16 Jakub Jelinek <jakub@redhat.com>
22009 * stor-layout.c (handle_warn_if_not_align): Use byte_position and
22010 multiple_of_p instead of unchecked tree_to_uhwi and UHWI check.
22011 If off is not INTEGER_CST, issue a may not be aligned warning
22012 rather than isn't aligned. Use isn%'t rather than isn't.
22013 * fold-const.c (multiple_of_p) <case BIT_AND_EXPR>: Don't fall through
22015 <case MULT_EXPR>: Improve the case when bottom and one of the
22016 MULT_EXPR operands are INTEGER_CSTs and bottom is multiple of that
22017 operand, in that case check if the other operand is multiple of
22018 bottom divided by the INTEGER_CST operand.
22020 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
22023 * config/pa/pa.h (FUNCTION_ARG_SIZE): Delete.
22024 * config/pa/pa-protos.h (pa_function_arg_size): Declare.
22025 * config/pa/som.h (ASM_DECLARE_FUNCTION_NAME): Use
22026 pa_function_arg_size instead of FUNCTION_ARG_SIZE.
22027 * config/pa/pa.c (pa_function_arg_advance): Likewise.
22028 (pa_function_arg, pa_arg_partial_bytes): Likewise.
22029 (pa_function_arg_size): New function.
22031 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
22033 * fold-const.c (fold_ternary_loc): Construct the vec_perm_indices
22034 in a separate statement.
22036 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
22038 PR tree-optimization/83847
22039 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Don't
22040 group gathers and scatters.
22042 2018-01-16 Jakub Jelinek <jakub@redhat.com>
22044 PR rtl-optimization/86620
22045 * params.def (max-sched-ready-insns): Bump minimum value to 1.
22047 PR rtl-optimization/83213
22048 * recog.c (peep2_attempt): Copy over CROSSING_JUMP_P from peepinsn
22049 to last if both are JUMP_INSNs.
22051 PR tree-optimization/83843
22052 * gimple-ssa-store-merging.c
22053 (imm_store_chain_info::output_merged_store): Handle bit_not_p on
22054 store_immediate_info for bswap/nop orig_stores.
22056 2018-01-15 Andrew Waterman <andrew@sifive.com>
22058 * config/riscv/riscv.c (riscv_rtx_costs) <MULT>: Increase cost if
22060 <UDIV>: Increase cost if !TARGET_DIV.
22062 2018-01-15 Segher Boessenkool <segher@kernel.crashing.org>
22064 * config/rs6000/rs6000.md (define_attr "type"): Remove delayed_cr.
22065 (define_attr "cr_logical_3op"): New.
22066 (cceq_ior_compare): Adjust.
22067 (cceq_ior_compare_complement): Adjust.
22068 (*cceq_rev_compare): Adjust.
22069 * config/rs6000/rs6000.c (rs6000_adjust_cost): Adjust.
22070 (is_cracked_insn): Adjust.
22071 (insn_must_be_first_in_group): Adjust.
22072 * config/rs6000/40x.md: Adjust.
22073 * config/rs6000/440.md: Adjust.
22074 * config/rs6000/476.md: Adjust.
22075 * config/rs6000/601.md: Adjust.
22076 * config/rs6000/603.md: Adjust.
22077 * config/rs6000/6xx.md: Adjust.
22078 * config/rs6000/7450.md: Adjust.
22079 * config/rs6000/7xx.md: Adjust.
22080 * config/rs6000/8540.md: Adjust.
22081 * config/rs6000/cell.md: Adjust.
22082 * config/rs6000/e300c2c3.md: Adjust.
22083 * config/rs6000/e500mc.md: Adjust.
22084 * config/rs6000/e500mc64.md: Adjust.
22085 * config/rs6000/e5500.md: Adjust.
22086 * config/rs6000/e6500.md: Adjust.
22087 * config/rs6000/mpc.md: Adjust.
22088 * config/rs6000/power4.md: Adjust.
22089 * config/rs6000/power5.md: Adjust.
22090 * config/rs6000/power6.md: Adjust.
22091 * config/rs6000/power7.md: Adjust.
22092 * config/rs6000/power8.md: Adjust.
22093 * config/rs6000/power9.md: Adjust.
22094 * config/rs6000/rs64.md: Adjust.
22095 * config/rs6000/titan.md: Adjust.
22097 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
22099 * config/i386/predicates.md (indirect_branch_operand): Rewrite
22100 ix86_indirect_branch_register logic.
22102 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
22104 * config/i386/constraints.md (Bs): Update
22105 ix86_indirect_branch_register check. Don't check
22106 ix86_indirect_branch_register with GOT_memory_operand.
22108 * config/i386/predicates.md (GOT_memory_operand): Don't check
22109 ix86_indirect_branch_register here.
22110 (GOT32_symbol_operand): Likewise.
22112 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
22114 * config/i386/predicates.md (constant_call_address_operand):
22115 Rewrite ix86_indirect_branch_register logic.
22116 (sibcall_insn_operand): Likewise.
22118 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
22120 * config/i386/constraints.md (Bs): Replace
22121 ix86_indirect_branch_thunk_register with
22122 ix86_indirect_branch_register.
22124 * config/i386/i386.md (indirect_jump): Likewise.
22125 (tablejump): Likewise.
22126 (*sibcall_memory): Likewise.
22127 (*sibcall_value_memory): Likewise.
22128 Peepholes of indirect call and jump via memory: Likewise.
22129 * config/i386/i386.opt: Likewise.
22130 * config/i386/predicates.md (indirect_branch_operand): Likewise.
22131 (GOT_memory_operand): Likewise.
22132 (call_insn_operand): Likewise.
22133 (sibcall_insn_operand): Likewise.
22134 (GOT32_symbol_operand): Likewise.
22136 2018-01-15 Jakub Jelinek <jakub@redhat.com>
22138 PR middle-end/83837
22139 * omp-expand.c (expand_omp_atomic_pipeline): Use loaded_val
22140 type rather than type addr's type points to.
22141 (expand_omp_atomic_mutex): Likewise.
22142 (expand_omp_atomic): Likewise.
22144 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
22147 * config/i386/i386.c (output_indirect_thunk_function): Use
22148 ASM_OUTPUT_LABEL, instead of ASM_OUTPUT_DEF, for TARGET_MACHO
22149 for __x86_return_thunk.
22151 2018-01-15 Richard Biener <rguenther@suse.de>
22153 PR middle-end/83850
22154 * expmed.c (extract_bit_field_1): Fix typo.
22156 2018-01-15 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22159 * config/arm/iterators.md (VF): New mode iterator.
22160 * config/arm/neon.md (neon_vabd<mode>_2): Use the above.
22161 Remove integer-related logic from pattern.
22162 (neon_vabd<mode>_3): Likewise.
22164 2018-01-15 Jakub Jelinek <jakub@redhat.com>
22166 PR middle-end/82694
22167 * common.opt (fstrict-overflow): No longer an alias.
22168 (fwrapv-pointer): New option.
22169 * tree.h (TYPE_OVERFLOW_WRAPS, TYPE_OVERFLOW_UNDEFINED): Define
22170 also for pointer types based on flag_wrapv_pointer.
22171 * opts.c (common_handle_option) <case OPT_fstrict_overflow>: Set
22172 opts->x_flag_wrap[pv] to !value, clear opts->x_flag_trapv if
22173 opts->x_flag_wrapv got set.
22174 * fold-const.c (fold_comparison, fold_binary_loc): Revert 2017-08-01
22175 changes, just use TYPE_OVERFLOW_UNDEFINED on pointer type instead of
22176 POINTER_TYPE_OVERFLOW_UNDEFINED.
22177 * match.pd: Likewise in address comparison pattern.
22178 * doc/invoke.texi: Document -fwrapv and -fstrict-overflow.
22180 2018-01-15 Richard Biener <rguenther@suse.de>
22183 * tree.c (free_lang_data_in_type): Always unlink TYPE_DECLs
22184 from TYPE_FIELDS. Free TYPE_BINFO if not used by devirtualization.
22185 Reset type names to their identifier if their TYPE_DECL doesn't
22186 have linkage (and thus is used for ODR and devirt).
22187 (save_debug_info_for_decl): Remove.
22188 (save_debug_info_for_type): Likewise.
22189 (add_tree_to_fld_list): Adjust.
22190 * tree-pretty-print.c (dump_generic_node): Make dumping of
22191 type names more robust.
22193 2018-01-15 Richard Biener <rguenther@suse.de>
22195 * BASE-VER: Bump to 8.0.1.
22197 2018-01-14 Martin Sebor <msebor@redhat.com>
22200 * builtins.c (check_access): Avoid warning when the no-warning bit
22203 2018-01-14 Cory Fields <cory-nospam-@coryfields.com>
22205 * tree-ssa-loop-im.c (sort_bbs_in_loop_postorder_cmp): Stabilize sort.
22206 * ira-color (allocno_hard_regs_compare): Likewise.
22208 2018-01-14 Nathan Rossi <nathan@nathanrossi.com>
22211 * config/microblaze/microblaze.c (microblaze_asm_output_ident):
22212 Use .pushsection/.popsection.
22214 2018-01-14 Martin Sebor <msebor@redhat.com>
22217 * doc/invoke.texi (-Wlass-memaccess): Document suppression by casting.
22219 2018-01-14 Jakub Jelinek <jakub@redhat.com>
22221 * config.gcc (i[34567]86-*-*): Remove one duplicate gfniintrin.h
22222 entry from extra_headers.
22223 (x86_64-*-*): Remove two duplicate gfniintrin.h entries from
22224 extra_headers, make the list bitwise identical to the i?86-*-* one.
22226 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
22228 * config/i386/i386.c (ix86_set_indirect_branch_type): Disallow
22229 -mcmodel=large with -mindirect-branch=thunk,
22230 -mindirect-branch=thunk-extern, -mfunction-return=thunk and
22231 -mfunction-return=thunk-extern.
22232 * doc/invoke.texi: Document -mcmodel=large is incompatible with
22233 -mindirect-branch=thunk, -mindirect-branch=thunk-extern,
22234 -mfunction-return=thunk and -mfunction-return=thunk-extern.
22236 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
22238 * config/i386/i386.c (print_reg): Print the name of the full
22239 integer register without '%'.
22240 (ix86_print_operand): Handle 'V'.
22241 * doc/extend.texi: Document 'V' modifier.
22243 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
22245 * config/i386/constraints.md (Bs): Disallow memory operand for
22246 -mindirect-branch-register.
22248 * config/i386/predicates.md (indirect_branch_operand): Likewise.
22249 (GOT_memory_operand): Likewise.
22250 (call_insn_operand): Likewise.
22251 (sibcall_insn_operand): Likewise.
22252 (GOT32_symbol_operand): Likewise.
22253 * config/i386/i386.md (indirect_jump): Call convert_memory_address
22254 for -mindirect-branch-register.
22255 (tablejump): Likewise.
22256 (*sibcall_memory): Likewise.
22257 (*sibcall_value_memory): Likewise.
22258 Disallow peepholes of indirect call and jump via memory for
22259 -mindirect-branch-register.
22260 (*call_pop): Replace m with Bw.
22261 (*call_value_pop): Likewise.
22262 (*sibcall_pop_memory): Replace m with Bs.
22263 * config/i386/i386.opt (mindirect-branch-register): New option.
22264 * doc/invoke.texi: Document -mindirect-branch-register option.
22266 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
22268 * config/i386/i386-protos.h (ix86_output_function_return): New.
22269 * config/i386/i386.c (ix86_set_indirect_branch_type): Also
22270 set function_return_type.
22271 (indirect_thunk_name): Add ret_p to indicate thunk for function
22273 (output_indirect_thunk_function): Pass false to
22274 indirect_thunk_name.
22275 (ix86_output_indirect_branch_via_reg): Likewise.
22276 (ix86_output_indirect_branch_via_push): Likewise.
22277 (output_indirect_thunk_function): Create alias for function
22278 return thunk if regno < 0.
22279 (ix86_output_function_return): New function.
22280 (ix86_handle_fndecl_attribute): Handle function_return.
22281 (ix86_attribute_table): Add function_return.
22282 * config/i386/i386.h (machine_function): Add
22283 function_return_type.
22284 * config/i386/i386.md (simple_return_internal): Use
22285 ix86_output_function_return.
22286 (simple_return_internal_long): Likewise.
22287 * config/i386/i386.opt (mfunction-return=): New option.
22288 (indirect_branch): Mention -mfunction-return=.
22289 * doc/extend.texi: Document function_return function attribute.
22290 * doc/invoke.texi: Document -mfunction-return= option.
22292 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
22294 * config/i386/i386-opts.h (indirect_branch): New.
22295 * config/i386/i386-protos.h (ix86_output_indirect_jmp): Likewise.
22296 * config/i386/i386.c (ix86_using_red_zone): Disallow red-zone
22297 with local indirect jump when converting indirect call and jump.
22298 (ix86_set_indirect_branch_type): New.
22299 (ix86_set_current_function): Call ix86_set_indirect_branch_type.
22300 (indirectlabelno): New.
22301 (indirect_thunk_needed): Likewise.
22302 (indirect_thunk_bnd_needed): Likewise.
22303 (indirect_thunks_used): Likewise.
22304 (indirect_thunks_bnd_used): Likewise.
22305 (INDIRECT_LABEL): Likewise.
22306 (indirect_thunk_name): Likewise.
22307 (output_indirect_thunk): Likewise.
22308 (output_indirect_thunk_function): Likewise.
22309 (ix86_output_indirect_branch_via_reg): Likewise.
22310 (ix86_output_indirect_branch_via_push): Likewise.
22311 (ix86_output_indirect_branch): Likewise.
22312 (ix86_output_indirect_jmp): Likewise.
22313 (ix86_code_end): Call output_indirect_thunk_function if needed.
22314 (ix86_output_call_insn): Call ix86_output_indirect_branch if
22316 (ix86_handle_fndecl_attribute): Handle indirect_branch.
22317 (ix86_attribute_table): Add indirect_branch.
22318 * config/i386/i386.h (machine_function): Add indirect_branch_type
22319 and has_local_indirect_jump.
22320 * config/i386/i386.md (indirect_jump): Set has_local_indirect_jump
22322 (tablejump): Likewise.
22323 (*indirect_jump): Use ix86_output_indirect_jmp.
22324 (*tablejump_1): Likewise.
22325 (simple_return_indirect_internal): Likewise.
22326 * config/i386/i386.opt (mindirect-branch=): New option.
22327 (indirect_branch): New.
22330 (thunk-inline): Likewise.
22331 (thunk-extern): Likewise.
22332 * doc/extend.texi: Document indirect_branch function attribute.
22333 * doc/invoke.texi: Document -mindirect-branch= option.
22335 2018-01-14 Jan Hubicka <hubicka@ucw.cz>
22338 * ipa-inline.c (edge_badness): Tolerate roundoff errors.
22340 2018-01-14 Richard Sandiford <richard.sandiford@linaro.org>
22342 * ipa-inline.c (want_inline_small_function_p): Return false if
22343 inlining has already failed with CIF_FINAL_ERROR.
22344 (update_caller_keys): Call want_inline_small_function_p before
22346 (update_callee_keys): Likewise.
22348 2018-01-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
22350 * config/rs6000/rs6000-p8swap.c (rs6000_sum_of_two_registers_p):
22352 (rs6000_quadword_masked_address_p): Likewise.
22353 (quad_aligned_load_p): Likewise.
22354 (quad_aligned_store_p): Likewise.
22355 (const_load_sequence_p): Add comment to describe the outer-most loop.
22356 (mimic_memory_attributes_and_flags): New function.
22357 (rs6000_gen_stvx): Likewise.
22358 (replace_swapped_aligned_store): Likewise.
22359 (rs6000_gen_lvx): Likewise.
22360 (replace_swapped_aligned_load): Likewise.
22361 (replace_swapped_load_constant): Capitalize argument name in
22362 comment describing this function.
22363 (rs6000_analyze_swaps): Add a third pass to search for vector loads
22364 and stores that access quad-word aligned addresses and replace
22365 with stvx or lvx instructions when appropriate.
22366 * config/rs6000/rs6000-protos.h (rs6000_sum_of_two_registers_p):
22367 New function prototype.
22368 (rs6000_quadword_masked_address_p): Likewise.
22369 (rs6000_gen_lvx): Likewise.
22370 (rs6000_gen_stvx): Likewise.
22371 * config/rs6000/vsx.md (*vsx_le_perm_load_<mode>): For modes
22372 VSX_D (V2DF, V2DI), modify this split to select lvx instruction
22373 when memory address is aligned.
22374 (*vsx_le_perm_load_<mode>): For modes VSX_W (V4SF, V4SI), modify
22375 this split to select lvx instruction when memory address is aligned.
22376 (*vsx_le_perm_load_v8hi): Modify this split to select lvx
22377 instruction when memory address is aligned.
22378 (*vsx_le_perm_load_v16qi): Likewise.
22379 (four unnamed splitters): Modify to select the stvx instruction
22380 when memory is aligned.
22382 2018-01-13 Jan Hubicka <hubicka@ucw.cz>
22384 * predict.c (determine_unlikely_bbs): Handle correctly BBs
22385 which appears in the queue multiple times.
22387 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
22388 Alan Hayward <alan.hayward@arm.com>
22389 David Sherwood <david.sherwood@arm.com>
22391 * tree-vectorizer.h (vec_lower_bound): New structure.
22392 (_loop_vec_info): Add check_nonzero and lower_bounds.
22393 (LOOP_VINFO_CHECK_NONZERO): New macro.
22394 (LOOP_VINFO_LOWER_BOUNDS): Likewise.
22395 (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Check lower_bounds too.
22396 * tree-data-ref.h (dr_with_seg_len): Add access_size and align
22397 fields. Make seg_len the distance travelled, not including the
22399 (dr_direction_indicator): Declare.
22400 (dr_zero_step_indicator): Likewise.
22401 (dr_known_forward_stride_p): Likewise.
22402 * tree-data-ref.c: Include stringpool.h, tree-vrp.h and
22404 (runtime_alias_check_p): Allow runtime alias checks with
22406 (operator ==): Compare access_size and align.
22407 (prune_runtime_alias_test_list): Rework for new distinction between
22408 the access_size and seg_len.
22409 (create_intersect_range_checks_index): Likewise. Cope with polynomial
22411 (get_segment_min_max): New function.
22412 (create_intersect_range_checks): Use it.
22413 (dr_step_indicator): New function.
22414 (dr_direction_indicator): Likewise.
22415 (dr_zero_step_indicator): Likewise.
22416 (dr_known_forward_stride_p): Likewise.
22417 * tree-loop-distribution.c (data_ref_segment_size): Return
22418 DR_STEP * (niters - 1).
22419 (compute_alias_check_pairs): Update call to the dr_with_seg_len
22421 * tree-vect-data-refs.c (vect_check_nonzero_value): New function.
22422 (vect_preserves_scalar_order_p): New function, split out from...
22423 (vect_analyze_data_ref_dependence): ...here. Check for zero steps.
22424 (vect_vfa_segment_size): Return DR_STEP * (length_factor - 1).
22425 (vect_vfa_access_size): New function.
22426 (vect_vfa_align): Likewise.
22427 (vect_compile_time_alias): Take access_size_a and access_b arguments.
22428 (dump_lower_bound): New function.
22429 (vect_check_lower_bound): Likewise.
22430 (vect_small_gap_p): Likewise.
22431 (vectorizable_with_step_bound_p): Likewise.
22432 (vect_prune_runtime_alias_test_list): Ignore cross-iteration
22433 depencies if the vectorization factor is 1. Convert the checks
22434 for nonzero steps into checks on the bounds of DR_STEP. Try using
22435 a bunds check for variable steps if the minimum required step is
22436 relatively small. Update calls to the dr_with_seg_len
22437 constructor and to vect_compile_time_alias.
22438 * tree-vect-loop-manip.c (vect_create_cond_for_lower_bounds): New
22440 (vect_loop_versioning): Call it.
22441 * tree-vect-loop.c (vect_analyze_loop_2): Clear LOOP_VINFO_LOWER_BOUNDS
22443 (vect_estimate_min_profitable_iters): Account for any bounds checks.
22445 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
22446 Alan Hayward <alan.hayward@arm.com>
22447 David Sherwood <david.sherwood@arm.com>
22449 * doc/sourcebuild.texi (vect_scatter_store): Document.
22450 * optabs.def (scatter_store_optab, mask_scatter_store_optab): New
22452 * doc/md.texi (scatter_store@var{m}, mask_scatter_store@var{m}):
22454 * genopinit.c (main): Add supports_vec_scatter_store and
22455 supports_vec_scatter_store_cached to target_optabs.
22456 * gimple.h (gimple_expr_type): Handle IFN_SCATTER_STORE and
22457 IFN_MASK_SCATTER_STORE.
22458 * internal-fn.def (SCATTER_STORE, MASK_SCATTER_STORE): New internal
22460 * internal-fn.h (internal_store_fn_p): Declare.
22461 (internal_fn_stored_value_index): Likewise.
22462 * internal-fn.c (scatter_store_direct): New macro.
22463 (expand_scatter_store_optab_fn): New function.
22464 (direct_scatter_store_optab_supported_p): New macro.
22465 (internal_store_fn_p): New function.
22466 (internal_gather_scatter_fn_p): Handle IFN_SCATTER_STORE and
22467 IFN_MASK_SCATTER_STORE.
22468 (internal_fn_mask_index): Likewise.
22469 (internal_fn_stored_value_index): New function.
22470 (internal_gather_scatter_fn_supported_p): Adjust operand numbers
22471 for scatter stores.
22472 * optabs-query.h (supports_vec_scatter_store_p): Declare.
22473 * optabs-query.c (supports_vec_scatter_store_p): New function.
22474 * tree-vectorizer.h (vect_get_store_rhs): Declare.
22475 * tree-vect-data-refs.c (vect_analyze_data_ref_access): Return
22476 true for scatter stores.
22477 (vect_gather_scatter_fn_p): Handle scatter stores too.
22478 (vect_check_gather_scatter): Consider using scatter stores if
22479 supports_vec_scatter_store_p.
22480 * tree-vect-patterns.c (vect_try_gather_scatter_pattern): Handle
22481 scatter stores too.
22482 * tree-vect-stmts.c (exist_non_indexing_operands_for_use_p): Use
22483 internal_fn_stored_value_index.
22484 (check_load_store_masking): Handle scatter stores too.
22485 (vect_get_store_rhs): Make public.
22486 (vectorizable_call): Use internal_store_fn_p.
22487 (vectorizable_store): Handle scatter store internal functions.
22488 (vect_transform_stmt): Compare GROUP_STORE_COUNT with GROUP_SIZE
22489 when deciding whether the end of the group has been reached.
22490 * config/aarch64/aarch64.md (UNSPEC_ST1_SCATTER): New unspec.
22491 * config/aarch64/aarch64-sve.md (scatter_store<mode>): New expander.
22492 (mask_scatter_store<mode>): New insns.
22494 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
22495 Alan Hayward <alan.hayward@arm.com>
22496 David Sherwood <david.sherwood@arm.com>
22498 * tree-vectorizer.h (vect_gather_scatter_fn_p): Declare.
22499 * tree-vect-data-refs.c (vect_gather_scatter_fn_p): Make public.
22500 * tree-vect-stmts.c (vect_truncate_gather_scatter_offset): New
22502 (vect_use_strided_gather_scatters_p): Take a masked_p argument.
22503 Use vect_truncate_gather_scatter_offset if we can't treat the
22504 operation as a normal gather load or scatter store.
22505 (get_group_load_store_type): Take the gather_scatter_info
22506 as argument. Try using a gather load or scatter store for
22507 single-element groups.
22508 (get_load_store_type): Update calls to get_group_load_store_type
22509 and vect_use_strided_gather_scatters_p.
22511 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
22512 Alan Hayward <alan.hayward@arm.com>
22513 David Sherwood <david.sherwood@arm.com>
22515 * tree-vectorizer.h (vect_create_data_ref_ptr): Take an extra
22516 optional tree argument.
22517 * tree-vect-data-refs.c (vect_check_gather_scatter): Check for
22519 (vect_create_data_ref_ptr): Take the iv_step as an optional argument,
22520 but continue to use the current value as a fallback.
22521 (bump_vector_ptr): Use operand_equal_p rather than tree_int_cst_compare
22522 to compare the updates.
22523 * tree-vect-stmts.c (vect_use_strided_gather_scatters_p): New function.
22524 (get_load_store_type): Use it when handling a strided access.
22525 (vect_get_strided_load_store_ops): New function.
22526 (vect_get_data_ptr_increment): Likewise.
22527 (vectorizable_load): Handle strided gather loads. Always pass
22528 a step to vect_create_data_ref_ptr and bump_vector_ptr.
22530 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
22531 Alan Hayward <alan.hayward@arm.com>
22532 David Sherwood <david.sherwood@arm.com>
22534 * doc/md.texi (gather_load@var{m}): Document.
22535 (mask_gather_load@var{m}): Likewise.
22536 * genopinit.c (main): Add supports_vec_gather_load and
22537 supports_vec_gather_load_cached to target_optabs.
22538 * optabs-tree.c (init_tree_optimization_optabs): Use
22539 ggc_cleared_alloc to allocate target_optabs.
22540 * optabs.def (gather_load_optab, mask_gather_laod_optab): New optabs.
22541 * internal-fn.def (GATHER_LOAD, MASK_GATHER_LOAD): New internal
22543 * internal-fn.h (internal_load_fn_p): Declare.
22544 (internal_gather_scatter_fn_p): Likewise.
22545 (internal_fn_mask_index): Likewise.
22546 (internal_gather_scatter_fn_supported_p): Likewise.
22547 * internal-fn.c (gather_load_direct): New macro.
22548 (expand_gather_load_optab_fn): New function.
22549 (direct_gather_load_optab_supported_p): New macro.
22550 (direct_internal_fn_optab): New function.
22551 (internal_load_fn_p): Likewise.
22552 (internal_gather_scatter_fn_p): Likewise.
22553 (internal_fn_mask_index): Likewise.
22554 (internal_gather_scatter_fn_supported_p): Likewise.
22555 * optabs-query.c (supports_at_least_one_mode_p): New function.
22556 (supports_vec_gather_load_p): Likewise.
22557 * optabs-query.h (supports_vec_gather_load_p): Declare.
22558 * tree-vectorizer.h (gather_scatter_info): Add ifn, element_type
22559 and memory_type field.
22560 (NUM_PATTERNS): Bump to 15.
22561 * tree-vect-data-refs.c: Include internal-fn.h.
22562 (vect_gather_scatter_fn_p): New function.
22563 (vect_describe_gather_scatter_call): Likewise.
22564 (vect_check_gather_scatter): Try using internal functions for
22565 gather loads. Recognize existing calls to a gather load function.
22566 (vect_analyze_data_refs): Consider using gather loads if
22567 supports_vec_gather_load_p.
22568 * tree-vect-patterns.c (vect_get_load_store_mask): New function.
22569 (vect_get_gather_scatter_offset_type): Likewise.
22570 (vect_convert_mask_for_vectype): Likewise.
22571 (vect_add_conversion_to_patterm): Likewise.
22572 (vect_try_gather_scatter_pattern): Likewise.
22573 (vect_recog_gather_scatter_pattern): New pattern recognizer.
22574 (vect_vect_recog_func_ptrs): Add it.
22575 * tree-vect-stmts.c (exist_non_indexing_operands_for_use_p): Use
22576 internal_fn_mask_index and internal_gather_scatter_fn_p.
22577 (check_load_store_masking): Take the gather_scatter_info as an
22578 argument and handle gather loads.
22579 (vect_get_gather_scatter_ops): New function.
22580 (vectorizable_call): Check internal_load_fn_p.
22581 (vectorizable_load): Likewise. Handle gather load internal
22583 (vectorizable_store): Update call to check_load_store_masking.
22584 * config/aarch64/aarch64.md (UNSPEC_LD1_GATHER): New unspec.
22585 * config/aarch64/iterators.md (SVE_S, SVE_D): New mode iterators.
22586 * config/aarch64/predicates.md (aarch64_gather_scale_operand_w)
22587 (aarch64_gather_scale_operand_d): New predicates.
22588 * config/aarch64/aarch64-sve.md (gather_load<mode>): New expander.
22589 (mask_gather_load<mode>): New insns.
22591 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
22592 Alan Hayward <alan.hayward@arm.com>
22593 David Sherwood <david.sherwood@arm.com>
22595 * optabs.def (fold_left_plus_optab): New optab.
22596 * doc/md.texi (fold_left_plus_@var{m}): Document.
22597 * internal-fn.def (IFN_FOLD_LEFT_PLUS): New internal function.
22598 * internal-fn.c (fold_left_direct): Define.
22599 (expand_fold_left_optab_fn): Likewise.
22600 (direct_fold_left_optab_supported_p): Likewise.
22601 * fold-const-call.c (fold_const_fold_left): New function.
22602 (fold_const_call): Use it to fold CFN_FOLD_LEFT_PLUS.
22603 * tree-parloops.c (valid_reduction_p): New function.
22604 (gather_scalar_reductions): Use it.
22605 * tree-vectorizer.h (FOLD_LEFT_REDUCTION): New vect_reduction_type.
22606 (vect_finish_replace_stmt): Declare.
22607 * tree-vect-loop.c (fold_left_reduction_fn): New function.
22608 (needs_fold_left_reduction_p): New function, split out from...
22609 (vect_is_simple_reduction): ...here. Accept reductions that
22610 forbid reassociation, but give them type FOLD_LEFT_REDUCTION.
22611 (vect_force_simple_reduction): Also store the reduction type in
22612 the assignment's STMT_VINFO_REDUC_TYPE.
22613 (vect_model_reduction_cost): Handle FOLD_LEFT_REDUCTION.
22614 (merge_with_identity): New function.
22615 (vect_expand_fold_left): Likewise.
22616 (vectorize_fold_left_reduction): Likewise.
22617 (vectorizable_reduction): Handle FOLD_LEFT_REDUCTION. Leave the
22618 scalar phi in place for it. Check for target support and reject
22619 cases that would reassociate the operation. Defer the transform
22620 phase to vectorize_fold_left_reduction.
22621 * config/aarch64/aarch64.md (UNSPEC_FADDA): New unspec.
22622 * config/aarch64/aarch64-sve.md (fold_left_plus_<mode>): New expander.
22623 (*fold_left_plus_<mode>, *pred_fold_left_plus_<mode>): New insns.
22625 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
22627 * tree-if-conv.c (predicate_mem_writes): Remove redundant
22628 call to ifc_temp_var.
22630 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
22631 Alan Hayward <alan.hayward@arm.com>
22632 David Sherwood <david.sherwood@arm.com>
22634 * target.def (legitimize_address_displacement): Take the original
22635 offset as a poly_int.
22636 * targhooks.h (default_legitimize_address_displacement): Update
22638 * targhooks.c (default_legitimize_address_displacement): Likewise.
22639 * doc/tm.texi: Regenerate.
22640 * lra-constraints.c (base_plus_disp_to_reg): Take the displacement
22641 as an argument, moving assert of ad->disp == ad->disp_term to...
22642 (process_address_1): ...here. Update calls to base_plus_disp_to_reg.
22643 Try calling targetm.legitimize_address_displacement before expanding
22644 the address rather than afterwards, and adjust for the new interface.
22645 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
22646 Match the new hook interface. Handle SVE addresses.
22647 * config/sh/sh.c (sh_legitimize_address_displacement): Make the
22648 new hook interface.
22650 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
22652 * Makefile.in (OBJS): Add early-remat.o.
22653 * target.def (select_early_remat_modes): New hook.
22654 * doc/tm.texi.in (TARGET_SELECT_EARLY_REMAT_MODES): New hook.
22655 * doc/tm.texi: Regenerate.
22656 * targhooks.h (default_select_early_remat_modes): Declare.
22657 * targhooks.c (default_select_early_remat_modes): New function.
22658 * timevar.def (TV_EARLY_REMAT): New timevar.
22659 * passes.def (pass_early_remat): New pass.
22660 * tree-pass.h (make_pass_early_remat): Declare.
22661 * early-remat.c: New file.
22662 * config/aarch64/aarch64.c (aarch64_select_early_remat_modes): New
22664 (TARGET_SELECT_EARLY_REMAT_MODES): Define.
22666 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
22667 Alan Hayward <alan.hayward@arm.com>
22668 David Sherwood <david.sherwood@arm.com>
22670 * tree-vect-loop-manip.c (vect_gen_scalar_loop_niters): Replace
22671 vfm1 with a bound_epilog parameter.
22672 (vect_do_peeling): Update calls accordingly, and move the prologue
22673 call earlier in the function. Treat the base bound_epilog as 0 for
22674 fully-masked loops and retain vf - 1 for other loops. Add 1 to
22675 this base when peeling for gaps.
22676 * tree-vect-loop.c (vect_analyze_loop_2): Allow peeling for gaps
22677 with fully-masked loops.
22678 (vect_estimate_min_profitable_iters): Handle the single peeled
22679 iteration in that case.
22681 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
22682 Alan Hayward <alan.hayward@arm.com>
22683 David Sherwood <david.sherwood@arm.com>
22685 * tree-vect-data-refs.c (vect_analyze_group_access_1): Allow
22686 single-element interleaving even if the size is not a power of 2.
22687 * tree-vect-stmts.c (get_load_store_type): Disallow elementwise
22688 accesses for single-element interleaving if the group size is
22691 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
22692 Alan Hayward <alan.hayward@arm.com>
22693 David Sherwood <david.sherwood@arm.com>
22695 * doc/md.texi (fold_extract_last_@var{m}): Document.
22696 * doc/sourcebuild.texi (vect_fold_extract_last): Likewise.
22697 * optabs.def (fold_extract_last_optab): New optab.
22698 * internal-fn.def (FOLD_EXTRACT_LAST): New internal function.
22699 * internal-fn.c (fold_extract_direct): New macro.
22700 (expand_fold_extract_optab_fn): Likewise.
22701 (direct_fold_extract_optab_supported_p): Likewise.
22702 * tree-vectorizer.h (EXTRACT_LAST_REDUCTION): New vect_reduction_type.
22703 * tree-vect-loop.c (vect_model_reduction_cost): Handle
22704 EXTRACT_LAST_REDUCTION.
22705 (get_initial_def_for_reduction): Do not create an initial vector
22706 for EXTRACT_LAST_REDUCTION reductions.
22707 (vectorizable_reduction): Leave the scalar phi in place for
22708 EXTRACT_LAST_REDUCTIONs. Try using EXTRACT_LAST_REDUCTION
22709 ahead of INTEGER_INDUC_COND_REDUCTION. Do not check for an
22710 epilogue code for EXTRACT_LAST_REDUCTION and defer the
22711 transform phase to vectorizable_condition.
22712 * tree-vect-stmts.c (vect_finish_stmt_generation_1): New function,
22714 (vect_finish_stmt_generation): ...here.
22715 (vect_finish_replace_stmt): New function.
22716 (vectorizable_condition): Handle EXTRACT_LAST_REDUCTION.
22717 * config/aarch64/aarch64-sve.md (fold_extract_last_<mode>): New
22719 * config/aarch64/aarch64.md (UNSPEC_CLASTB): New unspec.
22721 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
22722 Alan Hayward <alan.hayward@arm.com>
22723 David Sherwood <david.sherwood@arm.com>
22725 * doc/md.texi (extract_last_@var{m}): Document.
22726 * optabs.def (extract_last_optab): New optab.
22727 * internal-fn.def (EXTRACT_LAST): New internal function.
22728 * internal-fn.c (cond_unary_direct): New macro.
22729 (expand_cond_unary_optab_fn): Likewise.
22730 (direct_cond_unary_optab_supported_p): Likewise.
22731 * tree-vect-loop.c (vectorizable_live_operation): Allow fully-masked
22732 loops using EXTRACT_LAST.
22733 * config/aarch64/aarch64-sve.md (aarch64_sve_lastb<mode>): Rename to...
22734 (extract_last_<mode>): ...this optab.
22735 (vec_extract<mode><Vel>): Update accordingly.
22737 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
22738 Alan Hayward <alan.hayward@arm.com>
22739 David Sherwood <david.sherwood@arm.com>
22741 * target.def (empty_mask_is_expensive): New hook.
22742 * doc/tm.texi.in (TARGET_VECTORIZE_EMPTY_MASK_IS_EXPENSIVE): New hook.
22743 * doc/tm.texi: Regenerate.
22744 * targhooks.h (default_empty_mask_is_expensive): Declare.
22745 * targhooks.c (default_empty_mask_is_expensive): New function.
22746 * tree-vectorizer.c (vectorize_loops): Only call optimize_mask_stores
22747 if the target says that empty masks are expensive.
22748 * config/aarch64/aarch64.c (aarch64_empty_mask_is_expensive):
22750 (TARGET_VECTORIZE_EMPTY_MASK_IS_EXPENSIVE): Redefine.
22752 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
22753 Alan Hayward <alan.hayward@arm.com>
22754 David Sherwood <david.sherwood@arm.com>
22756 * tree-vectorizer.h (_loop_vec_info::mask_skip_niters): New field.
22757 (LOOP_VINFO_MASK_SKIP_NITERS): New macro.
22758 (vect_use_loop_mask_for_alignment_p): New function.
22759 (vect_prepare_for_masked_peels, vect_gen_while_not): Declare.
22760 * tree-vect-loop-manip.c (vect_set_loop_masks_directly): Add an
22761 niters_skip argument. Make sure that the first niters_skip elements
22762 of the first iteration are inactive.
22763 (vect_set_loop_condition_masked): Handle LOOP_VINFO_MASK_SKIP_NITERS.
22764 Update call to vect_set_loop_masks_directly.
22765 (get_misalign_in_elems): New function, split out from...
22766 (vect_gen_prolog_loop_niters): ...here.
22767 (vect_update_init_of_dr): Take a code argument that specifies whether
22768 the adjustment should be added or subtracted.
22769 (vect_update_init_of_drs): Likewise.
22770 (vect_prepare_for_masked_peels): New function.
22771 (vect_do_peeling): Skip prologue peeling if we're using a mask
22772 instead. Update call to vect_update_inits_of_drs.
22773 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
22775 (vect_analyze_loop_2): Allow fully-masked loops with peeling for
22776 alignment. Do not include the number of peeled iterations in
22777 the minimum threshold in that case.
22778 (vectorizable_induction): Adjust the start value down by
22779 LOOP_VINFO_MASK_SKIP_NITERS iterations.
22780 (vect_transform_loop): Call vect_prepare_for_masked_peels.
22781 Take the number of skipped iterations into account when calculating
22783 * tree-vect-stmts.c (vect_gen_while_not): New function.
22785 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
22786 Alan Hayward <alan.hayward@arm.com>
22787 David Sherwood <david.sherwood@arm.com>
22789 * doc/sourcebuild.texi (vect_fully_masked): Document.
22790 * params.def (PARAM_MIN_VECT_LOOP_BOUND): Change minimum and
22791 default value to 0.
22792 * tree-vect-loop.c (vect_analyze_loop_costing): New function,
22794 (vect_analyze_loop_2): ...here. Don't check the vectorization
22795 factor against the number of loop iterations if the loop is
22798 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
22799 Alan Hayward <alan.hayward@arm.com>
22800 David Sherwood <david.sherwood@arm.com>
22802 * tree-ssa-loop-ivopts.c (USE_ADDRESS): Split into...
22803 (USE_REF_ADDRESS, USE_PTR_ADDRESS): ...these new use types.
22804 (dump_groups): Update accordingly.
22805 (iv_use::mem_type): New member variable.
22806 (address_p): New function.
22807 (record_use): Add a mem_type argument and initialize the new
22809 (record_group_use): Add a mem_type argument. Use address_p.
22810 Remove obsolete null checks of base_object. Update call to record_use.
22811 (find_interesting_uses_op): Update call to record_group_use.
22812 (find_interesting_uses_cond): Likewise.
22813 (find_interesting_uses_address): Likewise.
22814 (get_mem_type_for_internal_fn): New function.
22815 (find_address_like_use): Likewise.
22816 (find_interesting_uses_stmt): Try find_address_like_use before
22817 calling find_interesting_uses_op.
22818 (addr_offset_valid_p): Use the iv mem_type field as the type
22819 of the addressed memory.
22820 (add_autoinc_candidates): Likewise.
22821 (get_address_cost): Likewise.
22822 (split_small_address_groups_p): Use address_p.
22823 (split_address_groups): Likewise.
22824 (add_iv_candidate_for_use): Likewise.
22825 (autoinc_possible_for_pair): Likewise.
22826 (rewrite_groups): Likewise.
22827 (get_use_type): Check for USE_REF_ADDRESS instead of USE_ADDRESS.
22828 (determine_group_iv_cost): Update after split of USE_ADDRESS.
22829 (get_alias_ptr_type_for_ptr_address): New function.
22830 (rewrite_use_address): Rewrite address uses in calls that were
22831 identified by find_address_like_use.
22833 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
22834 Alan Hayward <alan.hayward@arm.com>
22835 David Sherwood <david.sherwood@arm.com>
22837 * expr.c (expand_expr_addr_expr_1): Handle ADDR_EXPRs of
22839 * gimple-expr.h (is_gimple_addressable: Likewise.
22840 * gimple-expr.c (is_gimple_address): Likewise.
22841 * internal-fn.c (expand_call_mem_ref): New function.
22842 (expand_mask_load_optab_fn): Use it.
22843 (expand_mask_store_optab_fn): Likewise.
22845 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
22846 Alan Hayward <alan.hayward@arm.com>
22847 David Sherwood <david.sherwood@arm.com>
22849 * doc/md.texi (cond_add@var{mode}, cond_sub@var{mode})
22850 (cond_and@var{mode}, cond_ior@var{mode}, cond_xor@var{mode})
22851 (cond_smin@var{mode}, cond_smax@var{mode}, cond_umin@var{mode})
22852 (cond_umax@var{mode}): Document.
22853 * optabs.def (cond_add_optab, cond_sub_optab, cond_and_optab)
22854 (cond_ior_optab, cond_xor_optab, cond_smin_optab, cond_smax_optab)
22855 (cond_umin_optab, cond_umax_optab): New optabs.
22856 * internal-fn.def (COND_ADD, COND_SUB, COND_MIN, COND_MAX, COND_AND)
22857 (COND_IOR, COND_XOR): New internal functions.
22858 * internal-fn.h (get_conditional_internal_fn): Declare.
22859 * internal-fn.c (cond_binary_direct): New macro.
22860 (expand_cond_binary_optab_fn): Likewise.
22861 (direct_cond_binary_optab_supported_p): Likewise.
22862 (get_conditional_internal_fn): New function.
22863 * tree-vect-loop.c (vectorizable_reduction): Handle fully-masked loops.
22864 Cope with reduction statements that are vectorized as calls rather
22866 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): New insns.
22867 * config/aarch64/iterators.md (UNSPEC_COND_ADD, UNSPEC_COND_SUB)
22868 (UNSPEC_COND_SMAX, UNSPEC_COND_UMAX, UNSPEC_COND_SMIN)
22869 (UNSPEC_COND_UMIN, UNSPEC_COND_AND, UNSPEC_COND_ORR)
22870 (UNSPEC_COND_EOR): New unspecs.
22871 (optab): Add mappings for them.
22872 (SVE_COND_INT_OP, SVE_COND_FP_OP): New int iterators.
22873 (sve_int_op, sve_fp_op): New int attributes.
22875 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
22876 Alan Hayward <alan.hayward@arm.com>
22877 David Sherwood <david.sherwood@arm.com>
22879 * optabs.def (while_ult_optab): New optab.
22880 * doc/md.texi (while_ult@var{m}@var{n}): Document.
22881 * internal-fn.def (WHILE_ULT): New internal function.
22882 * internal-fn.h (direct_internal_fn_supported_p): New override
22883 that takes two types as argument.
22884 * internal-fn.c (while_direct): New macro.
22885 (expand_while_optab_fn): New function.
22886 (convert_optab_supported_p): Likewise.
22887 (direct_while_optab_supported_p): New macro.
22888 * wide-int.h (wi::udiv_ceil): New function.
22889 * tree-vectorizer.h (rgroup_masks): New structure.
22890 (vec_loop_masks): New typedef.
22891 (_loop_vec_info): Add masks, mask_compare_type, can_fully_mask_p
22892 and fully_masked_p.
22893 (LOOP_VINFO_CAN_FULLY_MASK_P, LOOP_VINFO_FULLY_MASKED_P)
22894 (LOOP_VINFO_MASKS, LOOP_VINFO_MASK_COMPARE_TYPE): New macros.
22895 (vect_max_vf): New function.
22896 (slpeel_make_loop_iterate_ntimes): Delete.
22897 (vect_set_loop_condition, vect_get_loop_mask_type, vect_gen_while)
22898 (vect_halve_mask_nunits, vect_double_mask_nunits): Declare.
22899 (vect_record_loop_mask, vect_get_loop_mask): Likewise.
22900 * tree-vect-loop-manip.c: Include tree-ssa-loop-niter.h,
22901 internal-fn.h, stor-layout.h and optabs-query.h.
22902 (vect_set_loop_mask): New function.
22903 (add_preheader_seq): Likewise.
22904 (add_header_seq): Likewise.
22905 (interleave_supported_p): Likewise.
22906 (vect_maybe_permute_loop_masks): Likewise.
22907 (vect_set_loop_masks_directly): Likewise.
22908 (vect_set_loop_condition_masked): Likewise.
22909 (vect_set_loop_condition_unmasked): New function, split out from
22910 slpeel_make_loop_iterate_ntimes.
22911 (slpeel_make_loop_iterate_ntimes): Rename to..
22912 (vect_set_loop_condition): ...this. Use vect_set_loop_condition_masked
22913 for fully-masked loops and vect_set_loop_condition_unmasked otherwise.
22914 (vect_do_peeling): Update call accordingly.
22915 (vect_gen_vector_loop_niters): Use VF as the step for fully-masked
22917 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
22918 mask_compare_type, can_fully_mask_p and fully_masked_p.
22919 (release_vec_loop_masks): New function.
22920 (_loop_vec_info): Use it to free the loop masks.
22921 (can_produce_all_loop_masks_p): New function.
22922 (vect_get_max_nscalars_per_iter): Likewise.
22923 (vect_verify_full_masking): Likewise.
22924 (vect_analyze_loop_2): Save LOOP_VINFO_CAN_FULLY_MASK_P around
22925 retries, and free the mask rgroups before retrying. Check loop-wide
22926 reasons for disallowing fully-masked loops. Make the final decision
22927 about whether use a fully-masked loop or not.
22928 (vect_estimate_min_profitable_iters): Do not assume that peeling
22929 for the number of iterations will be needed for fully-masked loops.
22930 (vectorizable_reduction): Disable fully-masked loops.
22931 (vectorizable_live_operation): Likewise.
22932 (vect_halve_mask_nunits): New function.
22933 (vect_double_mask_nunits): Likewise.
22934 (vect_record_loop_mask): Likewise.
22935 (vect_get_loop_mask): Likewise.
22936 (vect_transform_loop): Handle the case in which the final loop
22937 iteration might handle a partial vector. Call vect_set_loop_condition
22938 instead of slpeel_make_loop_iterate_ntimes.
22939 * tree-vect-stmts.c: Include tree-ssa-loop-niter.h and gimple-fold.h.
22940 (check_load_store_masking): New function.
22941 (prepare_load_store_mask): Likewise.
22942 (vectorizable_store): Handle fully-masked loops.
22943 (vectorizable_load): Likewise.
22944 (supportable_widening_operation): Use vect_halve_mask_nunits for
22946 (supportable_narrowing_operation): Likewise vect_double_mask_nunits.
22947 (vect_gen_while): New function.
22948 * config/aarch64/aarch64.md (umax<mode>3): New expander.
22949 (aarch64_uqdec<mode>): New insn.
22951 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
22952 Alan Hayward <alan.hayward@arm.com>
22953 David Sherwood <david.sherwood@arm.com>
22955 * optabs.def (reduc_and_scal_optab, reduc_ior_scal_optab)
22956 (reduc_xor_scal_optab): New optabs.
22957 * doc/md.texi (reduc_and_scal_@var{m}, reduc_ior_scal_@var{m})
22958 (reduc_xor_scal_@var{m}): Document.
22959 * doc/sourcebuild.texi (vect_logical_reduc): Likewise.
22960 * internal-fn.def (IFN_REDUC_AND, IFN_REDUC_IOR, IFN_REDUC_XOR): New
22961 internal functions.
22962 * fold-const-call.c (fold_const_call): Handle them.
22963 * tree-vect-loop.c (reduction_fn_for_scalar_code): Return the new
22964 internal functions for BIT_AND_EXPR, BIT_IOR_EXPR and BIT_XOR_EXPR.
22965 * config/aarch64/aarch64-sve.md (reduc_<bit_reduc>_scal_<mode>):
22966 (*reduc_<bit_reduc>_scal_<mode>): New patterns.
22967 * config/aarch64/iterators.md (UNSPEC_ANDV, UNSPEC_ORV)
22968 (UNSPEC_XORV): New unspecs.
22969 (optab): Add entries for them.
22970 (BITWISEV): New int iterator.
22971 (bit_reduc_op): New int attributes.
22973 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
22974 Alan Hayward <alan.hayward@arm.com>
22975 David Sherwood <david.sherwood@arm.com>
22977 * doc/md.texi (vec_shl_insert_@var{m}): New optab.
22978 * internal-fn.def (VEC_SHL_INSERT): New internal function.
22979 * optabs.def (vec_shl_insert_optab): New optab.
22980 * tree-vectorizer.h (can_duplicate_and_interleave_p): Declare.
22981 (duplicate_and_interleave): Likewise.
22982 * tree-vect-loop.c: Include internal-fn.h.
22983 (neutral_op_for_slp_reduction): New function, split out from
22984 get_initial_defs_for_reduction.
22985 (get_initial_def_for_reduction): Handle option 2 for variable-length
22986 vectors by loading the neutral value into a vector and then shifting
22987 the initial value into element 0.
22988 (get_initial_defs_for_reduction): Replace the code argument with
22989 the neutral value calculated by neutral_op_for_slp_reduction.
22990 Use gimple_build_vector for constant-length vectors.
22991 Use IFN_VEC_SHL_INSERT for variable-length vectors if all
22992 but the first group_size elements have a neutral value.
22993 Use duplicate_and_interleave otherwise.
22994 (vect_create_epilog_for_reduction): Take a neutral_op parameter.
22995 Update call to get_initial_defs_for_reduction. Handle SLP
22996 reductions for variable-length vectors by creating one vector
22997 result for each scalar result, with the elements associated
22998 with other scalar results stubbed out with the neutral value.
22999 (vectorizable_reduction): Call neutral_op_for_slp_reduction.
23000 Require IFN_VEC_SHL_INSERT for double reductions on
23001 variable-length vectors, or SLP reductions that have
23002 a neutral value. Require can_duplicate_and_interleave_p
23003 support for variable-length unchained SLP reductions if there
23004 is no neutral value, such as for MIN/MAX reductions. Also require
23005 the number of vector elements to be a multiple of the number of
23006 SLP statements when doing variable-length unchained SLP reductions.
23007 Update call to vect_create_epilog_for_reduction.
23008 * tree-vect-slp.c (can_duplicate_and_interleave_p): Make public
23009 and remove initial values.
23010 (duplicate_and_interleave): Make public.
23011 * config/aarch64/aarch64.md (UNSPEC_INSR): New unspec.
23012 * config/aarch64/aarch64-sve.md (vec_shl_insert_<mode>): New insn.
23014 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
23015 Alan Hayward <alan.hayward@arm.com>
23016 David Sherwood <david.sherwood@arm.com>
23018 * tree-vect-slp.c: Include gimple-fold.h and internal-fn.h
23019 (can_duplicate_and_interleave_p): New function.
23020 (vect_get_and_check_slp_defs): Take the vector of statements
23021 rather than just the current one. Remove excess parentheses.
23022 Restriction rejectinon of vect_constant_def and vect_external_def
23023 for variable-length vectors to boolean types, or types for which
23024 can_duplicate_and_interleave_p is false.
23025 (vect_build_slp_tree_2): Update call to vect_get_and_check_slp_defs.
23026 (duplicate_and_interleave): New function.
23027 (vect_get_constant_vectors): Use gimple_build_vector for
23028 constant-length vectors and suitable variable-length constant
23029 vectors. Use duplicate_and_interleave for other variable-length
23030 vectors. Don't defer the update when inserting new statements.
23032 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
23033 Alan Hayward <alan.hayward@arm.com>
23034 David Sherwood <david.sherwood@arm.com>
23036 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Make sure
23037 min_profitable_iters doesn't go negative.
23039 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
23040 Alan Hayward <alan.hayward@arm.com>
23041 David Sherwood <david.sherwood@arm.com>
23043 * doc/md.texi (vec_mask_load_lanes@var{m}@var{n}): Document.
23044 (vec_mask_store_lanes@var{m}@var{n}): Likewise.
23045 * optabs.def (vec_mask_load_lanes_optab): New optab.
23046 (vec_mask_store_lanes_optab): Likewise.
23047 * internal-fn.def (MASK_LOAD_LANES): New internal function.
23048 (MASK_STORE_LANES): Likewise.
23049 * internal-fn.c (mask_load_lanes_direct): New macro.
23050 (mask_store_lanes_direct): Likewise.
23051 (expand_mask_load_optab_fn): Handle masked operations.
23052 (expand_mask_load_lanes_optab_fn): New macro.
23053 (expand_mask_store_optab_fn): Handle masked operations.
23054 (expand_mask_store_lanes_optab_fn): New macro.
23055 (direct_mask_load_lanes_optab_supported_p): Likewise.
23056 (direct_mask_store_lanes_optab_supported_p): Likewise.
23057 * tree-vectorizer.h (vect_store_lanes_supported): Take a masked_p
23059 (vect_load_lanes_supported): Likewise.
23060 * tree-vect-data-refs.c (strip_conversion): New function.
23061 (can_group_stmts_p): Likewise.
23062 (vect_analyze_data_ref_accesses): Use it instead of checking
23063 for a pair of assignments.
23064 (vect_store_lanes_supported): Take a masked_p parameter.
23065 (vect_load_lanes_supported): Likewise.
23066 * tree-vect-loop.c (vect_analyze_loop_2): Update calls to
23067 vect_store_lanes_supported and vect_load_lanes_supported.
23068 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
23069 * tree-vect-stmts.c (get_group_load_store_type): Take a masked_p
23070 parameter. Don't allow gaps for masked accesses.
23071 Use vect_get_store_rhs. Update calls to vect_store_lanes_supported
23072 and vect_load_lanes_supported.
23073 (get_load_store_type): Take a masked_p parameter and update
23074 call to get_group_load_store_type.
23075 (vectorizable_store): Update call to get_load_store_type.
23076 Handle IFN_MASK_STORE_LANES.
23077 (vectorizable_load): Update call to get_load_store_type.
23078 Handle IFN_MASK_LOAD_LANES.
23080 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
23081 Alan Hayward <alan.hayward@arm.com>
23082 David Sherwood <david.sherwood@arm.com>
23084 * config/aarch64/aarch64-modes.def: Define x2, x3 and x4 vector
23086 * config/aarch64/aarch64-protos.h
23087 (aarch64_sve_struct_memory_operand_p): Declare.
23088 * config/aarch64/iterators.md (SVE_STRUCT): New mode iterator.
23089 (vector_count, insn_length, VSINGLE, vsingle): New mode attributes.
23090 (VPRED, vpred): Handle SVE structure modes.
23091 * config/aarch64/constraints.md (Utx): New constraint.
23092 * config/aarch64/predicates.md (aarch64_sve_struct_memory_operand)
23093 (aarch64_sve_struct_nonimmediate_operand): New predicates.
23094 * config/aarch64/aarch64.md (UNSPEC_LDN, UNSPEC_STN): New unspecs.
23095 * config/aarch64/aarch64-sve.md (mov<mode>, *aarch64_sve_mov<mode>_le)
23096 (*aarch64_sve_mov<mode>_be, pred_mov<mode>): New patterns for
23097 structure modes. Split into pieces after RA.
23098 (vec_load_lanes<mode><vsingle>, vec_mask_load_lanes<mode><vsingle>)
23099 (vec_store_lanes<mode><vsingle>, vec_mask_store_lanes<mode><vsingle>):
23101 * config/aarch64/aarch64.c (aarch64_classify_vector_mode): Handle
23102 SVE structure modes.
23103 (aarch64_classify_address): Likewise.
23104 (sizetochar): Move earlier in file.
23105 (aarch64_print_operand): Handle SVE register lists.
23106 (aarch64_array_mode): New function.
23107 (aarch64_sve_struct_memory_operand_p): Likewise.
23108 (TARGET_ARRAY_MODE): Redefine.
23110 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
23111 Alan Hayward <alan.hayward@arm.com>
23112 David Sherwood <david.sherwood@arm.com>
23114 * target.def (array_mode): New target hook.
23115 * doc/tm.texi.in (TARGET_ARRAY_MODE): New hook.
23116 * doc/tm.texi: Regenerate.
23117 * hooks.h (hook_optmode_mode_uhwi_none): Declare.
23118 * hooks.c (hook_optmode_mode_uhwi_none): New function.
23119 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Use
23120 targetm.array_mode.
23121 * stor-layout.c (mode_for_array): Likewise. Support polynomial
23124 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
23125 Alan Hayward <alan.hayward@arm.com>
23126 David Sherwood <david.sherwood@arm.com>
23128 * fold-const.c (fold_binary_loc): Check the argument types
23129 rather than the result type when testing for a vector operation.
23131 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
23133 * doc/tm.texi.in (DWARF_LAZY_REGISTER_VALUE): Document.
23134 * doc/tm.texi: Regenerate.
23136 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
23137 Alan Hayward <alan.hayward@arm.com>
23138 David Sherwood <david.sherwood@arm.com>
23140 * doc/invoke.texi (-msve-vector-bits=): Document new option.
23141 (sve): Document new AArch64 extension.
23142 * doc/md.texi (w): Extend the description of the AArch64
23143 constraint to include SVE vectors.
23144 (Upl, Upa): Document new AArch64 predicate constraints.
23145 * config/aarch64/aarch64-opts.h (aarch64_sve_vector_bits_enum): New
23147 * config/aarch64/aarch64.opt (sve_vector_bits): New enum.
23148 (msve-vector-bits=): New option.
23149 * config/aarch64/aarch64-option-extensions.def (fp, simd): Disable
23150 SVE when these are disabled.
23151 (sve): New extension.
23152 * config/aarch64/aarch64-modes.def: Define SVE vector and predicate
23153 modes. Adjust their number of units based on aarch64_sve_vg.
23154 (MAX_BITSIZE_MODE_ANY_MODE): Define.
23155 * config/aarch64/aarch64-protos.h (ADDR_QUERY_ANY): New
23156 aarch64_addr_query_type.
23157 (aarch64_const_vec_all_same_in_range_p, aarch64_sve_pred_mode)
23158 (aarch64_sve_cnt_immediate_p, aarch64_sve_addvl_addpl_immediate_p)
23159 (aarch64_sve_inc_dec_immediate_p, aarch64_add_offset_temporaries)
23160 (aarch64_split_add_offset, aarch64_output_sve_cnt_immediate)
23161 (aarch64_output_sve_addvl_addpl, aarch64_output_sve_inc_dec_immediate)
23162 (aarch64_output_sve_mov_immediate, aarch64_output_ptrue): Declare.
23163 (aarch64_simd_imm_zero_p): Delete.
23164 (aarch64_check_zero_based_sve_index_immediate): Declare.
23165 (aarch64_sve_index_immediate_p, aarch64_sve_arith_immediate_p)
23166 (aarch64_sve_bitmask_immediate_p, aarch64_sve_dup_immediate_p)
23167 (aarch64_sve_cmp_immediate_p, aarch64_sve_float_arith_immediate_p)
23168 (aarch64_sve_float_mul_immediate_p): Likewise.
23169 (aarch64_classify_symbol): Take the offset as a HOST_WIDE_INT
23170 rather than an rtx.
23171 (aarch64_sve_ld1r_operand_p, aarch64_sve_ldr_operand_p): Declare.
23172 (aarch64_expand_mov_immediate): Take a gen_vec_duplicate callback.
23173 (aarch64_emit_sve_pred_move, aarch64_expand_sve_mem_move): Declare.
23174 (aarch64_expand_sve_vec_cmp_int, aarch64_expand_sve_vec_cmp_float)
23175 (aarch64_expand_sve_vcond, aarch64_expand_sve_vec_perm): Declare.
23176 (aarch64_regmode_natural_size): Likewise.
23177 * config/aarch64/aarch64.h (AARCH64_FL_SVE): New macro.
23178 (AARCH64_FL_V8_3, AARCH64_FL_RCPC, AARCH64_FL_DOTPROD): Shift
23180 (AARCH64_ISA_SVE, TARGET_SVE): New macros.
23181 (FIXED_REGISTERS, CALL_USED_REGISTERS, REGISTER_NAMES): Add entries
23182 for VG and the SVE predicate registers.
23183 (V_ALIASES): Add a "z"-prefixed alias.
23184 (FIRST_PSEUDO_REGISTER): Change to P15_REGNUM + 1.
23185 (AARCH64_DWARF_VG, AARCH64_DWARF_P0): New macros.
23186 (PR_REGNUM_P, PR_LO_REGNUM_P): Likewise.
23187 (PR_LO_REGS, PR_HI_REGS, PR_REGS): New reg_classes.
23188 (REG_CLASS_NAMES): Add entries for them.
23189 (REG_CLASS_CONTENTS): Likewise. Update ALL_REGS to include VG
23190 and the predicate registers.
23191 (aarch64_sve_vg): Declare.
23192 (BITS_PER_SVE_VECTOR, BYTES_PER_SVE_VECTOR, BYTES_PER_SVE_PRED)
23193 (SVE_BYTE_MODE, MAX_COMPILE_TIME_VEC_BYTES): New macros.
23194 (REGMODE_NATURAL_SIZE): Define.
23195 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins): Handle
23197 * config/aarch64/aarch64.c: Include cfgrtl.h.
23198 (simd_immediate_info): Add a constructor for series vectors,
23199 and an associated step field.
23200 (aarch64_sve_vg): New variable.
23201 (aarch64_dbx_register_number): Handle VG and the predicate registers.
23202 (aarch64_vect_struct_mode_p, aarch64_vector_mode_p): Delete.
23203 (VEC_ADVSIMD, VEC_SVE_DATA, VEC_SVE_PRED, VEC_STRUCT, VEC_ANY_SVE)
23204 (VEC_ANY_DATA, VEC_STRUCT): New constants.
23205 (aarch64_advsimd_struct_mode_p, aarch64_sve_pred_mode_p)
23206 (aarch64_classify_vector_mode, aarch64_vector_data_mode_p)
23207 (aarch64_sve_data_mode_p, aarch64_sve_pred_mode)
23208 (aarch64_get_mask_mode): New functions.
23209 (aarch64_hard_regno_nregs): Handle SVE data modes for FP_REGS
23210 and FP_LO_REGS. Handle PR_REGS, PR_LO_REGS and PR_HI_REGS.
23211 (aarch64_hard_regno_mode_ok): Handle VG. Also handle the SVE
23212 predicate modes and predicate registers. Explicitly restrict
23213 GPRs to modes of 16 bytes or smaller. Only allow FP registers
23214 to store a vector mode if it is recognized by
23215 aarch64_classify_vector_mode.
23216 (aarch64_regmode_natural_size): New function.
23217 (aarch64_hard_regno_caller_save_mode): Return the original mode
23219 (aarch64_sve_cnt_immediate_p, aarch64_output_sve_cnt_immediate)
23220 (aarch64_sve_addvl_addpl_immediate_p, aarch64_output_sve_addvl_addpl)
23221 (aarch64_sve_inc_dec_immediate_p, aarch64_output_sve_inc_dec_immediate)
23222 (aarch64_add_offset_1_temporaries, aarch64_offset_temporaries): New
23224 (aarch64_add_offset): Add a temp2 parameter. Assert that temp1
23225 does not overlap dest if the function is frame-related. Handle
23227 (aarch64_split_add_offset): New function.
23228 (aarch64_add_sp, aarch64_sub_sp): Add temp2 parameters and pass
23229 them aarch64_add_offset.
23230 (aarch64_allocate_and_probe_stack_space): Add a temp2 parameter
23231 and update call to aarch64_sub_sp.
23232 (aarch64_add_cfa_expression): New function.
23233 (aarch64_expand_prologue): Pass extra temporary registers to the
23234 functions above. Handle the case in which we need to emit new
23235 DW_CFA_expressions for registers that were originally saved
23236 relative to the stack pointer, but now have to be expressed
23237 relative to the frame pointer.
23238 (aarch64_output_mi_thunk): Pass extra temporary registers to the
23240 (aarch64_expand_epilogue): Likewise. Prevent inheritance of
23241 IP0 and IP1 values for SVE frames.
23242 (aarch64_expand_vec_series): New function.
23243 (aarch64_expand_sve_widened_duplicate): Likewise.
23244 (aarch64_expand_sve_const_vector): Likewise.
23245 (aarch64_expand_mov_immediate): Add a gen_vec_duplicate parameter.
23246 Handle SVE constants. Use emit_move_insn to move a force_const_mem
23247 into the register, rather than emitting a SET directly.
23248 (aarch64_emit_sve_pred_move, aarch64_expand_sve_mem_move)
23249 (aarch64_get_reg_raw_mode, offset_4bit_signed_scaled_p)
23250 (offset_6bit_unsigned_scaled_p, aarch64_offset_7bit_signed_scaled_p)
23251 (offset_9bit_signed_scaled_p): New functions.
23252 (aarch64_replicate_bitmask_imm): New function.
23253 (aarch64_bitmask_imm): Use it.
23254 (aarch64_cannot_force_const_mem): Reject expressions involving
23255 a CONST_POLY_INT. Update call to aarch64_classify_symbol.
23256 (aarch64_classify_index): Handle SVE indices, by requiring
23257 a plain register index with a scale that matches the element size.
23258 (aarch64_classify_address): Handle SVE addresses. Assert that
23259 the mode of the address is VOIDmode or an integer mode.
23260 Update call to aarch64_classify_symbol.
23261 (aarch64_classify_symbolic_expression): Update call to
23262 aarch64_classify_symbol.
23263 (aarch64_const_vec_all_in_range_p): New function.
23264 (aarch64_print_vector_float_operand): Likewise.
23265 (aarch64_print_operand): Handle 'N' and 'C'. Use "zN" rather than
23266 "vN" for FP registers with SVE modes. Handle (const ...) vectors
23267 and the FP immediates 1.0 and 0.5.
23268 (aarch64_print_address_internal): Handle SVE addresses.
23269 (aarch64_print_operand_address): Use ADDR_QUERY_ANY.
23270 (aarch64_regno_regclass): Handle predicate registers.
23271 (aarch64_secondary_reload): Handle big-endian reloads of SVE
23273 (aarch64_class_max_nregs): Handle SVE modes and predicate registers.
23274 (aarch64_rtx_costs): Check for ADDVL and ADDPL instructions.
23275 (aarch64_convert_sve_vector_bits): New function.
23276 (aarch64_override_options): Use it to handle -msve-vector-bits=.
23277 (aarch64_classify_symbol): Take the offset as a HOST_WIDE_INT
23278 rather than an rtx.
23279 (aarch64_legitimate_constant_p): Use aarch64_classify_vector_mode.
23280 Handle SVE vector and predicate modes. Accept VL-based constants
23281 that need only one temporary register, and VL offsets that require
23282 no temporary registers.
23283 (aarch64_conditional_register_usage): Mark the predicate registers
23284 as fixed if SVE isn't available.
23285 (aarch64_vector_mode_supported_p): Use aarch64_classify_vector_mode.
23286 Return true for SVE vector and predicate modes.
23287 (aarch64_simd_container_mode): Take the number of bits as a poly_int64
23288 rather than an unsigned int. Handle SVE modes.
23289 (aarch64_preferred_simd_mode): Update call accordingly. Handle
23291 (aarch64_autovectorize_vector_sizes): Add BYTES_PER_SVE_VECTOR
23293 (aarch64_sve_index_immediate_p, aarch64_sve_arith_immediate_p)
23294 (aarch64_sve_bitmask_immediate_p, aarch64_sve_dup_immediate_p)
23295 (aarch64_sve_cmp_immediate_p, aarch64_sve_float_arith_immediate_p)
23296 (aarch64_sve_float_mul_immediate_p): New functions.
23297 (aarch64_sve_valid_immediate): New function.
23298 (aarch64_simd_valid_immediate): Use it as the fallback for SVE vectors.
23299 Explicitly reject structure modes. Check for INDEX constants.
23300 Handle PTRUE and PFALSE constants.
23301 (aarch64_check_zero_based_sve_index_immediate): New function.
23302 (aarch64_simd_imm_zero_p): Delete.
23303 (aarch64_mov_operand_p): Use aarch64_simd_valid_immediate for
23304 vector modes. Accept constants in the range of CNT[BHWD].
23305 (aarch64_simd_scalar_immediate_valid_for_move): Explicitly
23306 ask for an Advanced SIMD mode.
23307 (aarch64_sve_ld1r_operand_p, aarch64_sve_ldr_operand_p): New functions.
23308 (aarch64_simd_vector_alignment): Handle SVE predicates.
23309 (aarch64_vectorize_preferred_vector_alignment): New function.
23310 (aarch64_simd_vector_alignment_reachable): Use it instead of
23312 (aarch64_shift_truncation_mask): Use aarch64_vector_data_mode_p.
23313 (aarch64_output_sve_mov_immediate, aarch64_output_ptrue): New
23315 (MAX_VECT_LEN): Delete.
23316 (expand_vec_perm_d): Add a vec_flags field.
23317 (emit_unspec2, aarch64_expand_sve_vec_perm): New functions.
23318 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
23319 (aarch64_evpc_ext): Don't apply a big-endian lane correction
23321 (aarch64_evpc_rev): Rename to...
23322 (aarch64_evpc_rev_local): ...this. Use a predicated operation for SVE.
23323 (aarch64_evpc_rev_global): New function.
23324 (aarch64_evpc_dup): Enforce a 64-byte range for SVE DUP.
23325 (aarch64_evpc_tbl): Use MAX_COMPILE_TIME_VEC_BYTES instead of
23327 (aarch64_evpc_sve_tbl): New function.
23328 (aarch64_expand_vec_perm_const_1): Update after rename of
23329 aarch64_evpc_rev. Handle SVE permutes too, trying
23330 aarch64_evpc_rev_global and using aarch64_evpc_sve_tbl rather
23331 than aarch64_evpc_tbl.
23332 (aarch64_vectorize_vec_perm_const): Initialize vec_flags.
23333 (aarch64_sve_cmp_operand_p, aarch64_unspec_cond_code)
23334 (aarch64_gen_unspec_cond, aarch64_expand_sve_vec_cmp_int)
23335 (aarch64_emit_unspec_cond, aarch64_emit_unspec_cond_or)
23336 (aarch64_emit_inverted_unspec_cond, aarch64_expand_sve_vec_cmp_float)
23337 (aarch64_expand_sve_vcond): New functions.
23338 (aarch64_modes_tieable_p): Use aarch64_vector_data_mode_p instead
23339 of aarch64_vector_mode_p.
23340 (aarch64_dwarf_poly_indeterminate_value): New function.
23341 (aarch64_compute_pressure_classes): Likewise.
23342 (aarch64_can_change_mode_class): Likewise.
23343 (TARGET_GET_RAW_RESULT_MODE, TARGET_GET_RAW_ARG_MODE): Redefine.
23344 (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): Likewise.
23345 (TARGET_VECTORIZE_GET_MASK_MODE): Likewise.
23346 (TARGET_DWARF_POLY_INDETERMINATE_VALUE): Likewise.
23347 (TARGET_COMPUTE_PRESSURE_CLASSES): Likewise.
23348 (TARGET_CAN_CHANGE_MODE_CLASS): Likewise.
23349 * config/aarch64/constraints.md (Upa, Upl, Uav, Uat, Usv, Usi, Utr)
23350 (Uty, Dm, vsa, vsc, vsd, vsi, vsn, vsl, vsm, vsA, vsM, vsN): New
23352 (Dn, Dl, Dr): Accept const as well as const_vector.
23353 (Dz): Likewise. Compare against CONST0_RTX.
23354 * config/aarch64/iterators.md: Refer to "Advanced SIMD" instead
23355 of "vector" where appropriate.
23356 (SVE_ALL, SVE_BH, SVE_BHS, SVE_BHSI, SVE_HSDI, SVE_HSF, SVE_SD)
23357 (SVE_SDI, SVE_I, SVE_F, PRED_ALL, PRED_BHS): New mode iterators.
23358 (UNSPEC_SEL, UNSPEC_ANDF, UNSPEC_IORF, UNSPEC_XORF, UNSPEC_COND_LT)
23359 (UNSPEC_COND_LE, UNSPEC_COND_EQ, UNSPEC_COND_NE, UNSPEC_COND_GE)
23360 (UNSPEC_COND_GT, UNSPEC_COND_LO, UNSPEC_COND_LS, UNSPEC_COND_HS)
23361 (UNSPEC_COND_HI, UNSPEC_COND_UO): New unspecs.
23362 (Vetype, VEL, Vel, VWIDE, Vwide, vw, vwcore, V_INT_EQUIV)
23363 (v_int_equiv): Extend to SVE modes.
23364 (Vesize, V128, v128, Vewtype, V_FP_EQUIV, v_fp_equiv, VPRED): New
23366 (LOGICAL_OR, SVE_INT_UNARY, SVE_FP_UNARY): New code iterators.
23367 (optab): Handle popcount, smin, smax, umin, umax, abs and sqrt.
23368 (logical_nn, lr, sve_int_op, sve_fp_op): New code attributs.
23369 (LOGICALF, OPTAB_PERMUTE, UNPACK, UNPACK_UNSIGNED, SVE_COND_INT_CMP)
23370 (SVE_COND_FP_CMP): New int iterators.
23371 (perm_hilo): Handle the new unpack unspecs.
23372 (optab, logicalf_op, su, perm_optab, cmp_op, imm_con): New int
23374 * config/aarch64/predicates.md (aarch64_sve_cnt_immediate)
23375 (aarch64_sve_addvl_addpl_immediate, aarch64_split_add_offset_immediate)
23376 (aarch64_pluslong_or_poly_operand, aarch64_nonmemory_operand)
23377 (aarch64_equality_operator, aarch64_constant_vector_operand)
23378 (aarch64_sve_ld1r_operand, aarch64_sve_ldr_operand): New predicates.
23379 (aarch64_sve_nonimmediate_operand): Likewise.
23380 (aarch64_sve_general_operand): Likewise.
23381 (aarch64_sve_dup_operand, aarch64_sve_arith_immediate): Likewise.
23382 (aarch64_sve_sub_arith_immediate, aarch64_sve_inc_dec_immediate)
23383 (aarch64_sve_logical_immediate, aarch64_sve_mul_immediate): Likewise.
23384 (aarch64_sve_dup_immediate, aarch64_sve_cmp_vsc_immediate): Likewise.
23385 (aarch64_sve_cmp_vsd_immediate, aarch64_sve_index_immediate): Likewise.
23386 (aarch64_sve_float_arith_immediate): Likewise.
23387 (aarch64_sve_float_arith_with_sub_immediate): Likewise.
23388 (aarch64_sve_float_mul_immediate, aarch64_sve_arith_operand): Likewise.
23389 (aarch64_sve_add_operand, aarch64_sve_logical_operand): Likewise.
23390 (aarch64_sve_lshift_operand, aarch64_sve_rshift_operand): Likewise.
23391 (aarch64_sve_mul_operand, aarch64_sve_cmp_vsc_operand): Likewise.
23392 (aarch64_sve_cmp_vsd_operand, aarch64_sve_index_operand): Likewise.
23393 (aarch64_sve_float_arith_operand): Likewise.
23394 (aarch64_sve_float_arith_with_sub_operand): Likewise.
23395 (aarch64_sve_float_mul_operand): Likewise.
23396 (aarch64_sve_vec_perm_operand): Likewise.
23397 (aarch64_pluslong_operand): Include aarch64_sve_addvl_addpl_immediate.
23398 (aarch64_mov_operand): Accept const_poly_int and const_vector.
23399 (aarch64_simd_lshift_imm, aarch64_simd_rshift_imm): Accept const
23400 as well as const_vector.
23401 (aarch64_simd_imm_zero, aarch64_simd_imm_minus_one): Move earlier
23402 in file. Use CONST0_RTX and CONSTM1_RTX.
23403 (aarch64_simd_or_scalar_imm_zero): Likewise. Add match_codes.
23404 (aarch64_simd_reg_or_zero): Accept const as well as const_vector.
23405 Use aarch64_simd_imm_zero.
23406 * config/aarch64/aarch64-sve.md: New file.
23407 * config/aarch64/aarch64.md: Include it.
23408 (VG_REGNUM, P0_REGNUM, P7_REGNUM, P15_REGNUM): New register numbers.
23409 (UNSPEC_REV, UNSPEC_LD1_SVE, UNSPEC_ST1_SVE, UNSPEC_MERGE_PTRUE)
23410 (UNSPEC_PTEST_PTRUE, UNSPEC_UNPACKSHI, UNSPEC_UNPACKUHI)
23411 (UNSPEC_UNPACKSLO, UNSPEC_UNPACKULO, UNSPEC_PACK)
23412 (UNSPEC_FLOAT_CONVERT, UNSPEC_WHILE_LO): New unspec constants.
23413 (sve): New attribute.
23414 (enabled): Disable instructions with the sve attribute unless
23416 (movqi, movhi): Pass CONST_POLY_INT operaneds through
23417 aarch64_expand_mov_immediate.
23418 (*mov<mode>_aarch64, *movsi_aarch64, *movdi_aarch64): Handle
23419 CNT[BHSD] immediates.
23420 (movti): Split CONST_POLY_INT moves into two halves.
23421 (add<mode>3): Accept aarch64_pluslong_or_poly_operand.
23422 Split additions that need a temporary here if the destination
23423 is the stack pointer.
23424 (*add<mode>3_aarch64): Handle ADDVL and ADDPL immediates.
23425 (*add<mode>3_poly_1): New instruction.
23426 (set_clobber_cc): New expander.
23428 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
23430 * simplify-rtx.c (simplify_immed_subreg): Add an inner_bytes
23431 parameter and use it instead of GET_MODE_SIZE (innermode). Use
23432 inner_bytes * BITS_PER_UNIT instead of GET_MODE_BITSIZE (innermode).
23433 Use CEIL (inner_bytes, GET_MODE_UNIT_SIZE (innermode)) instead of
23434 GET_MODE_NUNITS (innermode). Also add a first_elem parameter.
23435 Change innermode from fixed_mode_size to machine_mode.
23436 (simplify_subreg): Update call accordingly. Handle a constant-sized
23437 subreg of a variable-length CONST_VECTOR.
23439 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
23440 Alan Hayward <alan.hayward@arm.com>
23441 David Sherwood <david.sherwood@arm.com>
23443 * tree-ssa-address.c (mem_ref_valid_without_offset_p): New function.
23444 (add_offset_to_base): New function, split out from...
23445 (create_mem_ref): ...here. When handling a scale other than 1,
23446 check first whether the address is valid without the offset.
23447 Add it into the base if so, leaving the index and scale as-is.
23449 2018-01-12 Jakub Jelinek <jakub@redhat.com>
23452 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Call
23453 fold_for_warn before checking if arg2 is INTEGER_CST.
23455 2018-01-12 Segher Boessenkool <segher@kernel.crashing.org>
23457 * config/rs6000/predicates.md (load_multiple_operation): Delete.
23458 (store_multiple_operation): Delete.
23459 * config/rs6000/rs6000-cpus.def (601): Remove MASK_STRING.
23460 * config/rs6000/rs6000-protos.h (rs6000_output_load_multiple): Delete.
23461 * config/rs6000/rs6000-string.c (expand_block_move): Delete everything
23462 guarded by TARGET_STRING.
23463 (rs6000_output_load_multiple): Delete.
23464 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
23465 OPTION_MASK_STRING / TARGET_STRING handling.
23466 (print_operand) <'N', 'O'>: Add comment that these are unused now.
23467 (const rs6000_opt_masks) <"string">: Change mask to 0.
23468 * config/rs6000/rs6000.h (TARGET_DEFAULT): Remove MASK_STRING.
23469 (MASK_STRING): Delete.
23470 * config/rs6000/rs6000.md (*mov<mode>_string): Delete TARGET_STRING
23472 (load_multiple): Delete.
23479 (store_multiple): Delete.
23486 (movmemsi_8reg): Delete.
23487 (corresponding unnamed define_insn): Delete.
23488 (movmemsi_6reg): Delete.
23489 (corresponding unnamed define_insn): Delete.
23490 (movmemsi_4reg): Delete.
23491 (corresponding unnamed define_insn): Delete.
23492 (movmemsi_2reg): Delete.
23493 (corresponding unnamed define_insn): Delete.
23494 (movmemsi_1reg): Delete.
23495 (corresponding unnamed define_insn): Delete.
23496 * config/rs6000/rs6000.opt (mno-string): New.
23497 (mstring): Replace by deprecation warning stub.
23498 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mstring.
23500 2018-01-12 Jakub Jelinek <jakub@redhat.com>
23502 * regrename.c (regrename_do_replace): If replacing the same
23503 reg multiple times, try to reuse last created gen_raw_REG.
23506 * bb-reorder.c (pass_partition_blocks::gate): In lto don't partition
23507 main to workaround a bug in GDB.
23509 2018-01-12 Tom de Vries <tom@codesourcery.com>
23512 * config.gcc (nvptx*-*-*): Set use_gcc_stdint=wrap.
23514 2018-01-12 Vladimir Makarov <vmakarov@redhat.com>
23516 PR rtl-optimization/80481
23517 * ira-color.c (get_cap_member): New function.
23518 (allocnos_conflict_by_live_ranges_p): Use it.
23519 (slot_coalesced_allocno_live_ranges_intersect_p): Add assert.
23520 (setup_slot_coalesced_allocno_live_ranges): Ditto.
23522 2018-01-12 Uros Bizjak <ubizjak@gmail.com>
23525 * config/alpha/alpha.md (*saddsi_1): New insn_ans_split pattern.
23526 (*saddl_se_1): Ditto.
23527 (*ssubsi_1): Ditto.
23528 (*ssubl_se_1): Ditto.
23530 2018-01-12 Richard Sandiford <richard.sandiford@linaro.org>
23532 * tree-predcom.c (aff_combination_dr_offset): Use wi::to_poly_widest
23533 rather than wi::to_widest for DR_INITs.
23534 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Use
23535 wi::to_poly_offset rather than wi::to_offset for DR_INIT.
23536 (vect_analyze_data_ref_accesses): Require both DR_INITs to be
23538 (vect_analyze_group_access_1): Note that here.
23540 2018-01-12 Richard Sandiford <richard.sandiford@linaro.org>
23542 * tree-vectorizer.c (get_vec_alignment_for_array_type): Handle
23543 polynomial type sizes.
23545 2018-01-12 Richard Sandiford <richard.sandiford@linaro.org>
23547 * gimplify.c (gimple_add_tmp_var_fn): Allow variables to have a
23548 poly_uint64 size, rather than requiring an unsigned HOST_WIDE_INT size.
23549 (gimple_add_tmp_var): Likewise.
23551 2018-01-12 Martin Liska <mliska@suse.cz>
23553 * gimple.c (gimple_alloc_counts): Use uint64_t instead of int.
23554 (gimple_alloc_sizes): Likewise.
23555 (dump_gimple_statistics): Use PRIu64 in printf format.
23556 * gimple.h: Change uint64_t to int.
23558 2018-01-12 Martin Liska <mliska@suse.cz>
23560 * tree-core.h: Use uint64_t instead of int.
23561 * tree.c (tree_node_counts): Likewise.
23562 (tree_node_sizes): Likewise.
23563 (dump_tree_statistics): Use PRIu64 in printf format.
23565 2018-01-12 Martin Liska <mliska@suse.cz>
23567 * Makefile.in: As qsort_chk is implemented in vec.c, add
23568 vec.o to linkage of gencfn-macros.
23569 * tree.c (build_new_poly_int_cst): Add CXX_MEM_STAT_INFO as it's
23570 passing the info to record_node_allocation_statistics.
23571 (test_vector_cst_patterns): Add CXX_MEM_STAT_INFO to declaration
23573 * ggc-common.c (struct ggc_usage): Add operator== and use
23574 it in operator< and compare function.
23575 * mem-stats.h (struct mem_usage): Likewise.
23576 * vec.c (struct vec_usage): Remove operator< and compare
23577 function. Can be simply inherited.
23579 2018-01-12 Martin Jambor <mjambor@suse.cz>
23582 * params.def: New parameter PARAM_AVOID_FMA_MAX_BITS.
23583 * tree-ssa-math-opts.c: Include domwalk.h.
23584 (convert_mult_to_fma_1): New function.
23585 (fma_transformation_info): New type.
23586 (fma_deferring_state): Likewise.
23587 (cancel_fma_deferring): New function.
23588 (result_of_phi): Likewise.
23589 (last_fma_candidate_feeds_initial_phi): Likewise.
23590 (convert_mult_to_fma): Added deferring logic, split actual
23591 transformation to convert_mult_to_fma_1.
23592 (math_opts_dom_walker): New type.
23593 (math_opts_dom_walker::after_dom_children): New method, body moved
23594 here from pass_optimize_widening_mul::execute, added deferring logic
23596 (pass_optimize_widening_mul::execute): Moved most of code to
23597 math_opts_dom_walker::after_dom_children.
23598 * config/i386/x86-tune.def (X86_TUNE_AVOID_128FMA_CHAINS): New.
23599 * config/i386/i386.c (ix86_option_override_internal): Added
23600 maybe_setting of PARAM_AVOID_FMA_MAX_BITS.
23602 2018-01-12 Richard Biener <rguenther@suse.de>
23605 * dwarf2out.c (gen_variable_die): Do not reset old_die for
23606 inline instance vars.
23608 2018-01-12 Oleg Endo <olegendo@gcc.gnu.org>
23611 * config/rx/rx.c (rx_is_restricted_memory_address):
23612 Handle SUBREG case.
23614 2018-01-12 Richard Biener <rguenther@suse.de>
23616 PR tree-optimization/80846
23617 * target.def (split_reduction): New target hook.
23618 * targhooks.c (default_split_reduction): New function.
23619 * targhooks.h (default_split_reduction): Declare.
23620 * tree-vect-loop.c (vect_create_epilog_for_reduction): If the
23621 target requests first reduce vectors by combining low and high
23623 * tree-vect-stmts.c (vect_gen_perm_mask_any): Adjust.
23624 (get_vectype_for_scalar_type_and_size): Export.
23625 * tree-vectorizer.h (get_vectype_for_scalar_type_and_size): Declare.
23626 * doc/tm.texi.in (TARGET_VECTORIZE_SPLIT_REDUCTION): Document.
23627 * doc/tm.texi: Regenerate.
23628 * config/i386/i386.c (ix86_split_reduction): Implement
23629 TARGET_VECTORIZE_SPLIT_REDUCTION.
23631 2018-01-12 Eric Botcazou <ebotcazou@adacore.com>
23634 * config/sparc/sparc.h (PIC_OFFSET_TABLE_REGNUM): Set to INVALID_REGNUM
23635 in PIC mode except for TARGET_VXWORKS_RTP.
23636 * config/sparc/sparc.c: Include cfgrtl.h.
23637 (TARGET_INIT_PIC_REG): Define.
23638 (TARGET_USE_PSEUDO_PIC_REG): Likewise.
23639 (sparc_pic_register_p): New predicate.
23640 (sparc_legitimate_address_p): Use it.
23641 (sparc_legitimize_pic_address): Likewise.
23642 (sparc_delegitimize_address): Likewise.
23643 (sparc_mode_dependent_address_p): Likewise.
23644 (gen_load_pcrel_sym): Remove 4th parameter.
23645 (load_got_register): Adjust call to above. Remove obsolete stuff.
23646 (sparc_expand_prologue): Do not call load_got_register here.
23647 (sparc_flat_expand_prologue): Likewise.
23648 (sparc_output_mi_thunk): Set the pic_offset_table_rtx object.
23649 (sparc_use_pseudo_pic_reg): New function.
23650 (sparc_init_pic_reg): Likewise.
23651 * config/sparc/sparc.md (vxworks_load_got): Set the GOT register.
23652 (builtin_setjmp_receiver): Enable only for TARGET_VXWORKS_RTP.
23654 2018-01-12 Christophe Lyon <christophe.lyon@linaro.org>
23656 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
23657 Add item for branch_cost.
23659 2018-01-12 Eric Botcazou <ebotcazou@adacore.com>
23661 PR rtl-optimization/83565
23662 * rtlanal.c (nonzero_bits1): On WORD_REGISTER_OPERATIONS machines, do
23663 not extend the result to a larger mode for rotate operations.
23664 (num_sign_bit_copies1): Likewise.
23666 2018-01-12 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
23669 * config/sol2.h (STARTFILE_ARCH_SPEC): Don't use with -shared or
23671 Use values-Xc.o for -pedantic.
23672 Link with values-xpg4.o for C90, values-xpg6.o otherwise.
23674 2018-01-12 Martin Liska <mliska@suse.cz>
23677 * ipa-devirt.c (final_warning_record::grow_type_warnings):
23679 (possible_polymorphic_call_targets): Use it.
23680 (ipa_devirt): Likewise.
23682 2018-01-12 Martin Liska <mliska@suse.cz>
23684 * profile-count.h (enum profile_quality): Use 0 as invalid
23685 enum value of profile_quality.
23687 2018-01-12 Chung-Ju Wu <jasonwucj@gmail.com>
23689 * doc/invoke.texi (NDS32 Options): Add -mext-perf, -mext-perf2 and
23690 -mext-string options.
23692 2018-01-12 Richard Biener <rguenther@suse.de>
23694 * lto-streamer-out.c (DFS::DFS_write_tree_body): Process
23695 DECL_DEBUG_EXPR conditional on DECL_HAS_DEBUG_EXPR_P.
23696 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
23698 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
23700 2018-01-11 Michael Meissner <meissner@linux.vnet.ibm.com>
23702 * configure.ac (--with-long-double-format): Add support for the
23703 configuration option to change the default long double format on
23705 * config.gcc (powerpc*-linux*-*): Likewise.
23706 * configure: Regenerate.
23707 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If long
23708 double is IEEE, define __KC__ and __KF__ to allow floatn.h to be
23709 used without modification.
23711 2018-01-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23713 * config/rs6000/rs6000-builtin.def (BU_P7_MISC_X): New #define.
23714 (SPEC_BARRIER): New instantiation of BU_P7_MISC_X.
23715 * config/rs6000/rs6000.c (rs6000_expand_builtin): Handle
23716 MISC_BUILTIN_SPEC_BARRIER.
23717 (rs6000_init_builtins): Likewise.
23718 * config/rs6000/rs6000.md (UNSPECV_SPEC_BARRIER): New UNSPECV
23720 (speculation_barrier): New define_insn.
23721 * doc/extend.texi: Document __builtin_speculation_barrier.
23723 2018-01-11 Jakub Jelinek <jakub@redhat.com>
23726 * config/i386/i386.c (ix86_expand_vector_init_one_nonzero): If one_var
23727 is 0, for V{8,16}S[IF] and V[48]D[IF]mode use gen_vec_set<mode>_0.
23728 * config/i386/sse.md (VI8_AVX_AVX512F, VI4F_256_512): New mode
23730 (ssescalarmodesuffix): Add 512-bit vectors. Use "d" or "q" for
23731 integral modes instead of "ss" and "sd".
23732 (vec_set<mode>_0): New define_insns for 256-bit and 512-bit
23733 vectors with 32-bit and 64-bit elements.
23734 (vecdupssescalarmodesuffix): New mode attribute.
23735 (vec_dup<mode>): Use it.
23737 2018-01-11 H.J. Lu <hongjiu.lu@intel.com>
23740 * config/i386/i386.c (ix86_compute_frame_layout): Align stack
23741 frame if argument is passed on stack.
23743 2018-01-11 Jakub Jelinek <jakub@redhat.com>
23746 * ree.c (combine_reaching_defs): Optimize also
23747 reg2=exp; reg1=reg2; reg2=any_extend(reg1); into
23748 reg2=any_extend(exp); reg1=reg2;, formatting fix.
23750 2018-01-11 Jan Hubicka <hubicka@ucw.cz>
23752 PR middle-end/83189
23753 * gimple-ssa-isolate-paths.c (isolate_path): Fix profile update.
23755 2018-01-11 Jan Hubicka <hubicka@ucw.cz>
23757 PR middle-end/83718
23758 * tree-inline.c (copy_cfg_body): Adjust num&den for scaling
23759 after they are computed.
23761 2018-01-11 Bin Cheng <bin.cheng@arm.com>
23763 PR tree-optimization/83695
23764 * gimple-loop-linterchange.cc
23765 (tree_loop_interchange::interchange_loops): Call scev_reset_htab to
23766 reset cached scev information after interchange.
23767 (pass_linterchange::execute): Remove call to scev_reset_htab.
23769 2018-01-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23771 * config/arm/arm_neon.h (vfmlal_lane_low_u32, vfmlal_lane_high_u32,
23772 vfmlalq_laneq_low_u32, vfmlalq_lane_low_u32, vfmlal_laneq_low_u32,
23773 vfmlalq_laneq_high_u32, vfmlalq_lane_high_u32, vfmlal_laneq_high_u32,
23774 vfmlsl_lane_low_u32, vfmlsl_lane_high_u32, vfmlslq_laneq_low_u32,
23775 vfmlslq_lane_low_u32, vfmlsl_laneq_low_u32, vfmlslq_laneq_high_u32,
23776 vfmlslq_lane_high_u32, vfmlsl_laneq_high_u32): Define.
23777 * config/arm/arm_neon_builtins.def (vfmal_lane_low,
23778 vfmal_lane_lowv4hf, vfmal_lane_lowv8hf, vfmal_lane_high,
23779 vfmal_lane_highv4hf, vfmal_lane_highv8hf, vfmsl_lane_low,
23780 vfmsl_lane_lowv4hf, vfmsl_lane_lowv8hf, vfmsl_lane_high,
23781 vfmsl_lane_highv4hf, vfmsl_lane_highv8hf): New sets of builtins.
23782 * config/arm/iterators.md (VFMLSEL2, vfmlsel2): New mode attributes.
23783 (V_lane_reg): Likewise.
23784 * config/arm/neon.md (neon_vfm<vfml_op>l_lane_<vfml_half><VCVTF:mode>):
23786 (neon_vfm<vfml_op>l_lane_<vfml_half><vfmlsel2><mode>): Likewise.
23787 (vfmal_lane_low<mode>_intrinsic,
23788 vfmal_lane_low<vfmlsel2><mode>_intrinsic,
23789 vfmal_lane_high<vfmlsel2><mode>_intrinsic,
23790 vfmal_lane_high<mode>_intrinsic, vfmsl_lane_low<mode>_intrinsic,
23791 vfmsl_lane_low<vfmlsel2><mode>_intrinsic,
23792 vfmsl_lane_high<vfmlsel2><mode>_intrinsic,
23793 vfmsl_lane_high<mode>_intrinsic): New define_insns.
23795 2018-01-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23797 * config/arm/arm-cpus.in (fp16fml): New feature.
23798 (ALL_SIMD): Add fp16fml.
23799 (armv8.2-a): Add fp16fml as an option.
23800 (armv8.3-a): Likewise.
23801 (armv8.4-a): Add fp16fml as part of fp16.
23802 * config/arm/arm.h (TARGET_FP16FML): Define.
23803 * config/arm/arm-c.c (arm_cpu_builtins): Define __ARM_FEATURE_FP16_FML
23805 * config/arm/arm-modes.def (V2HF): Define.
23806 * config/arm/arm_neon.h (vfmlal_low_u32, vfmlsl_low_u32,
23807 vfmlal_high_u32, vfmlsl_high_u32, vfmlalq_low_u32,
23808 vfmlslq_low_u32, vfmlalq_high_u32, vfmlslq_high_u32): Define.
23809 * config/arm/arm_neon_builtins.def (vfmal_low, vfmal_high,
23810 vfmsl_low, vfmsl_high): New set of builtins.
23811 * config/arm/iterators.md (PLUSMINUS): New code iterator.
23812 (vfml_op): New code attribute.
23813 (VFMLHALVES): New int iterator.
23814 (VFML, VFMLSEL): New mode attributes.
23815 (V_reg): Define mapping for V2HF.
23816 (V_hi, V_lo): New mode attributes.
23817 (VF_constraint): Likewise.
23818 (vfml_half, vfml_half_selector): New int attributes.
23819 * config/arm/neon.md (neon_vfm<vfml_op>l_<vfml_half><mode>): New
23821 (vfmal_low<mode>_intrinsic, vfmsl_high<mode>_intrinsic,
23822 vfmal_high<mode>_intrinsic, vfmsl_low<mode>_intrinsic):
23824 * config/arm/t-arm-elf (v8_fps): Add fp16fml.
23825 * config/arm/t-multilib (v8_2_a_simd_variants): Add fp16fml.
23826 * config/arm/unspecs.md (UNSPEC_VFML_LO, UNSPEC_VFML_HI): New unspecs.
23827 * doc/invoke.texi (ARM Options): Document fp16fml. Update armv8.4-a
23829 * doc/sourcebuild.texi (arm_fp16fml_neon_ok, arm_fp16fml_neon):
23830 Document new effective target and option set.
23832 2018-01-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23834 * config/arm/arm-cpus.in (armv8_4): New feature.
23835 (ARMv8_4a): New fgroup.
23836 (armv8.4-a): New arch.
23837 * config/arm/arm-tables.opt: Regenerate.
23838 * config/arm/t-aprofile: Add matching rules for -march=armv8.4-a.
23839 * config/arm/t-arm-elf (all_v8_archs): Add armv8.4-a.
23840 * config/arm/t-multilib (v8_4_a_simd_variants): New variable.
23841 Add matching rules for -march=armv8.4-a and extensions.
23842 * doc/invoke.texi (ARM Options): Document -march=armv8.4-a.
23844 2018-01-11 Oleg Endo <olegendo@gcc.gnu.org>
23847 * config/rx/rx.md (BW): New mode attribute.
23848 (sync_lock_test_and_setsi): Add mode suffix to insn output.
23850 2018-01-11 Richard Biener <rguenther@suse.de>
23852 PR tree-optimization/83435
23853 * graphite.c (canonicalize_loop_form): Ignore fake loop exit edges.
23854 * graphite-scop-detection.c (scop_detection::get_sese): Likewise.
23855 * tree-vrp.c (add_assert_info): Drop TREE_OVERFLOW if they appear.
23857 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
23858 Alan Hayward <alan.hayward@arm.com>
23859 David Sherwood <david.sherwood@arm.com>
23861 * config/aarch64/aarch64.c (aarch64_address_info): Add a const_offset
23863 (aarch64_classify_address): Initialize it. Track polynomial offsets.
23864 (aarch64_print_address_internal): Use it to check for a zero offset.
23866 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
23867 Alan Hayward <alan.hayward@arm.com>
23868 David Sherwood <david.sherwood@arm.com>
23870 * config/aarch64/aarch64-modes.def (NUM_POLY_INT_COEFFS): Set to 2.
23871 * config/aarch64/aarch64-protos.h (aarch64_initial_elimination_offset):
23872 Return a poly_int64 rather than a HOST_WIDE_INT.
23873 (aarch64_offset_7bit_signed_scaled_p): Take the offset as a poly_int64
23874 rather than a HOST_WIDE_INT.
23875 * config/aarch64/aarch64.h (aarch64_frame): Protect with
23876 HAVE_POLY_INT_H rather than HOST_WIDE_INT. Change locals_offset,
23877 hard_fp_offset, frame_size, initial_adjust, callee_offset and
23878 final_offset from HOST_WIDE_INT to poly_int64.
23879 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args): Use
23880 to_constant when getting the number of units in an Advanced SIMD
23882 (aarch64_builtin_vectorized_function): Check for a constant number
23884 * config/aarch64/aarch64-simd.md (mov<mode>): Handle polynomial
23886 (aarch64_ld<VSTRUCT:nregs>_lane<VALLDIF:mode>): Use the nunits
23887 attribute instead of GET_MODE_NUNITS.
23888 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs)
23889 (aarch64_class_max_nregs): Use the constant_lowest_bound of the
23890 GET_MODE_SIZE for fixed-size registers.
23891 (aarch64_const_vec_all_same_in_range_p): Use const_vec_duplicate_p.
23892 (aarch64_hard_regno_call_part_clobbered, aarch64_classify_index)
23893 (aarch64_mode_valid_for_sched_fusion_p, aarch64_classify_address)
23894 (aarch64_legitimize_address_displacement, aarch64_secondary_reload)
23895 (aarch64_print_operand, aarch64_print_address_internal)
23896 (aarch64_address_cost, aarch64_rtx_costs, aarch64_register_move_cost)
23897 (aarch64_short_vector_p, aapcs_vfp_sub_candidate)
23898 (aarch64_simd_attr_length_rglist, aarch64_operands_ok_for_ldpstp):
23899 Handle polynomial GET_MODE_SIZE.
23900 (aarch64_hard_regno_caller_save_mode): Likewise. Return modes
23901 wider than SImode without modification.
23902 (tls_symbolic_operand_type): Use strip_offset instead of split_const.
23903 (aarch64_pass_by_reference, aarch64_layout_arg, aarch64_pad_reg_upward)
23904 (aarch64_gimplify_va_arg_expr): Assert that we don't yet handle
23905 passing and returning SVE modes.
23906 (aarch64_function_value, aarch64_layout_arg): Use gen_int_mode
23907 rather than GEN_INT.
23908 (aarch64_emit_probe_stack_range): Take the size as a poly_int64
23909 rather than a HOST_WIDE_INT, but call sorry if it isn't constant.
23910 (aarch64_allocate_and_probe_stack_space): Likewise.
23911 (aarch64_layout_frame): Cope with polynomial offsets.
23912 (aarch64_save_callee_saves, aarch64_restore_callee_saves): Take the
23913 start_offset as a poly_int64 rather than a HOST_WIDE_INT. Track
23914 polynomial offsets.
23915 (offset_9bit_signed_unscaled_p, offset_12bit_unsigned_scaled_p)
23916 (aarch64_offset_7bit_signed_scaled_p): Take the offset as a
23917 poly_int64 rather than a HOST_WIDE_INT.
23918 (aarch64_get_separate_components, aarch64_process_components)
23919 (aarch64_expand_prologue, aarch64_expand_epilogue)
23920 (aarch64_use_return_insn_p): Handle polynomial frame offsets.
23921 (aarch64_anchor_offset): New function, split out from...
23922 (aarch64_legitimize_address): ...here.
23923 (aarch64_builtin_vectorization_cost): Handle polynomial
23924 TYPE_VECTOR_SUBPARTS.
23925 (aarch64_simd_check_vect_par_cnst_half): Handle polynomial
23927 (aarch64_simd_make_constant, aarch64_expand_vector_init): Get the
23928 number of elements from the PARALLEL rather than the mode.
23929 (aarch64_shift_truncation_mask): Use GET_MODE_UNIT_BITSIZE
23930 rather than GET_MODE_BITSIZE.
23931 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_ext)
23932 (aarch64_evpc_rev, aarch64_evpc_dup, aarch64_evpc_zip)
23933 (aarch64_expand_vec_perm_const_1): Handle polynomial
23934 d->perm.length () and d->perm elements.
23935 (aarch64_evpc_tbl): Likewise. Use nelt rather than GET_MODE_NUNITS.
23936 Apply to_constant to d->perm elements.
23937 (aarch64_simd_valid_immediate, aarch64_vec_fpconst_pow_of_2): Handle
23938 polynomial CONST_VECTOR_NUNITS.
23939 (aarch64_move_pointer): Take amount as a poly_int64 rather
23941 (aarch64_progress_pointer): Avoid temporary variable.
23942 * config/aarch64/aarch64.md (aarch64_<crc_variant>): Use
23943 the mode attribute instead of GET_MODE.
23945 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
23946 Alan Hayward <alan.hayward@arm.com>
23947 David Sherwood <david.sherwood@arm.com>
23949 * config/aarch64/aarch64.c (aarch64_force_temporary): Assert that
23950 x exists before using it.
23951 (aarch64_add_constant_internal): Rename to...
23952 (aarch64_add_offset_1): ...this. Replace regnum with separate
23953 src and dest rtxes. Handle the case in which they're different,
23954 including when the offset is zero. Replace scratchreg with an rtx.
23955 Use 2 additions if there is no spare register into which we can
23956 move a 16-bit constant.
23957 (aarch64_add_constant): Delete.
23958 (aarch64_add_offset): Replace reg with separate src and dest
23959 rtxes. Take a poly_int64 offset instead of a HOST_WIDE_INT.
23960 Use aarch64_add_offset_1.
23961 (aarch64_add_sp, aarch64_sub_sp): Take the scratch register as
23962 an rtx rather than an int. Take the delta as a poly_int64
23963 rather than a HOST_WIDE_INT. Use aarch64_add_offset.
23964 (aarch64_expand_mov_immediate): Update uses of aarch64_add_offset.
23965 (aarch64_expand_prologue): Update calls to aarch64_sub_sp,
23966 aarch64_allocate_and_probe_stack_space and aarch64_add_offset.
23967 (aarch64_expand_epilogue): Update calls to aarch64_add_offset
23968 and aarch64_add_sp.
23969 (aarch64_output_mi_thunk): Use aarch64_add_offset rather than
23970 aarch64_add_constant.
23972 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
23974 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
23975 Use scalar_float_mode.
23977 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
23979 * config/aarch64/aarch64-simd.md
23980 (aarch64_fml<f16mac1>l<f16quad>_low<mode>): Avoid GET_MODE_NUNITS.
23981 (aarch64_fml<f16mac1>l<f16quad>_high<mode>): Likewise.
23982 (aarch64_fml<f16mac1>l_lane_lowv2sf): Likewise.
23983 (aarch64_fml<f16mac1>l_lane_highv2sf): Likewise.
23984 (aarch64_fml<f16mac1>lq_laneq_lowv4sf): Likewise.
23985 (aarch64_fml<f16mac1>lq_laneq_highv4sf): Likewise.
23986 (aarch64_fml<f16mac1>l_laneq_lowv2sf): Likewise.
23987 (aarch64_fml<f16mac1>l_laneq_highv2sf): Likewise.
23988 (aarch64_fml<f16mac1>lq_lane_lowv4sf): Likewise.
23989 (aarch64_fml<f16mac1>lq_lane_highv4sf): Likewise.
23991 2018-01-11 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
23994 * config/arm/arm.c (arm_declare_function_name): Set arch_to_print if
23995 targ_options->x_arm_arch_string is non NULL.
23997 2018-01-11 Tamar Christina <tamar.christina@arm.com>
23999 * config/aarch64/aarch64.h
24000 (AARCH64_FL_FOR_ARCH8_4): Add AARCH64_FL_DOTPROD.
24002 2018-01-11 Sudakshina Das <sudi.das@arm.com>
24005 * expmed.c (emit_store_flag_force): Swap if const op0
24006 and change VOIDmode to mode of op0.
24008 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
24010 PR rtl-optimization/83761
24011 * caller-save.c (replace_reg_with_saved_mem): Pass bits rather
24012 than bytes to mode_for_size.
24014 2018-01-10 Jan Hubicka <hubicka@ucw.cz>
24016 PR middle-end/83189
24017 * gfortran.fortran-torture/compile/pr83189.f90: New testcase.
24018 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Handle zero
24021 2018-01-10 Jan Hubicka <hubicka@ucw.cz>
24023 PR middle-end/83575
24024 * cfgrtl.c (rtl_verify_edges): Only verify fixability of partition
24025 when in layout mode.
24026 (cfg_layout_finalize): Do not verify cfg before we are out of layout.
24027 * cfgcleanup.c (try_optimize_cfg): Only verify flow info when doing
24030 2018-01-10 Michael Collison <michael.collison@arm.com>
24032 * config/aarch64/aarch64-modes.def (V2HF): New VECTOR_MODE.
24033 * config/aarch64/aarch64-option-extension.def: Add
24034 AARCH64_OPT_EXTENSION of 'fp16fml'.
24035 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
24036 (__ARM_FEATURE_FP16_FML): Define if TARGET_F16FML is true.
24037 * config/aarch64/predicates.md (aarch64_lane_imm3): New predicate.
24038 * config/aarch64/constraints.md (Ui7): New constraint.
24039 * config/aarch64/iterators.md (VFMLA_W): New mode iterator.
24040 (VFMLA_SEL_W): Ditto.
24043 (VFMLA16_LOW): New int iterator.
24044 (VFMLA16_HIGH): Ditto.
24045 (UNSPEC_FMLAL): New unspec.
24046 (UNSPEC_FMLSL): Ditto.
24047 (UNSPEC_FMLAL2): Ditto.
24048 (UNSPEC_FMLSL2): Ditto.
24049 (f16mac): New code attribute.
24050 * config/aarch64/aarch64-simd-builtins.def
24051 (aarch64_fmlal_lowv2sf): Ditto.
24052 (aarch64_fmlsl_lowv2sf): Ditto.
24053 (aarch64_fmlalq_lowv4sf): Ditto.
24054 (aarch64_fmlslq_lowv4sf): Ditto.
24055 (aarch64_fmlal_highv2sf): Ditto.
24056 (aarch64_fmlsl_highv2sf): Ditto.
24057 (aarch64_fmlalq_highv4sf): Ditto.
24058 (aarch64_fmlslq_highv4sf): Ditto.
24059 (aarch64_fmlal_lane_lowv2sf): Ditto.
24060 (aarch64_fmlsl_lane_lowv2sf): Ditto.
24061 (aarch64_fmlal_laneq_lowv2sf): Ditto.
24062 (aarch64_fmlsl_laneq_lowv2sf): Ditto.
24063 (aarch64_fmlalq_lane_lowv4sf): Ditto.
24064 (aarch64_fmlsl_lane_lowv4sf): Ditto.
24065 (aarch64_fmlalq_laneq_lowv4sf): Ditto.
24066 (aarch64_fmlsl_laneq_lowv4sf): Ditto.
24067 (aarch64_fmlal_lane_highv2sf): Ditto.
24068 (aarch64_fmlsl_lane_highv2sf): Ditto.
24069 (aarch64_fmlal_laneq_highv2sf): Ditto.
24070 (aarch64_fmlsl_laneq_highv2sf): Ditto.
24071 (aarch64_fmlalq_lane_highv4sf): Ditto.
24072 (aarch64_fmlsl_lane_highv4sf): Ditto.
24073 (aarch64_fmlalq_laneq_highv4sf): Ditto.
24074 (aarch64_fmlsl_laneq_highv4sf): Ditto.
24075 * config/aarch64/aarch64-simd.md:
24076 (aarch64_fml<f16mac1>l<f16quad>_low<mode>): New pattern.
24077 (aarch64_fml<f16mac1>l<f16quad>_high<mode>): Ditto.
24078 (aarch64_simd_fml<f16mac1>l<f16quad>_low<mode>): Ditto.
24079 (aarch64_simd_fml<f16mac1>l<f16quad>_high<mode>): Ditto.
24080 (aarch64_fml<f16mac1>l_lane_lowv2sf): Ditto.
24081 (aarch64_fml<f16mac1>l_lane_highv2sf): Ditto.
24082 (aarch64_simd_fml<f16mac>l_lane_lowv2sf): Ditto.
24083 (aarch64_simd_fml<f16mac>l_lane_highv2sf): Ditto.
24084 (aarch64_fml<f16mac1>lq_laneq_lowv4sf): Ditto.
24085 (aarch64_fml<f16mac1>lq_laneq_highv4sf): Ditto.
24086 (aarch64_simd_fml<f16mac>lq_laneq_lowv4sf): Ditto.
24087 (aarch64_simd_fml<f16mac>lq_laneq_highv4sf): Ditto.
24088 (aarch64_fml<f16mac1>l_laneq_lowv2sf): Ditto.
24089 (aarch64_fml<f16mac1>l_laneq_highv2sf): Ditto.
24090 (aarch64_simd_fml<f16mac>l_laneq_lowv2sf): Ditto.
24091 (aarch64_simd_fml<f16mac>l_laneq_highv2sf): Ditto.
24092 (aarch64_fml<f16mac1>lq_lane_lowv4sf): Ditto.
24093 (aarch64_fml<f16mac1>lq_lane_highv4sf): Ditto.
24094 (aarch64_simd_fml<f16mac>lq_lane_lowv4sf): Ditto.
24095 (aarch64_simd_fml<f16mac>lq_lane_highv4sf): Ditto.
24096 * config/aarch64/arm_neon.h (vfmlal_low_u32): New intrinsic.
24097 (vfmlsl_low_u32): Ditto.
24098 (vfmlalq_low_u32): Ditto.
24099 (vfmlslq_low_u32): Ditto.
24100 (vfmlal_high_u32): Ditto.
24101 (vfmlsl_high_u32): Ditto.
24102 (vfmlalq_high_u32): Ditto.
24103 (vfmlslq_high_u32): Ditto.
24104 (vfmlal_lane_low_u32): Ditto.
24105 (vfmlsl_lane_low_u32): Ditto.
24106 (vfmlal_laneq_low_u32): Ditto.
24107 (vfmlsl_laneq_low_u32): Ditto.
24108 (vfmlalq_lane_low_u32): Ditto.
24109 (vfmlslq_lane_low_u32): Ditto.
24110 (vfmlalq_laneq_low_u32): Ditto.
24111 (vfmlslq_laneq_low_u32): Ditto.
24112 (vfmlal_lane_high_u32): Ditto.
24113 (vfmlsl_lane_high_u32): Ditto.
24114 (vfmlal_laneq_high_u32): Ditto.
24115 (vfmlsl_laneq_high_u32): Ditto.
24116 (vfmlalq_lane_high_u32): Ditto.
24117 (vfmlslq_lane_high_u32): Ditto.
24118 (vfmlalq_laneq_high_u32): Ditto.
24119 (vfmlslq_laneq_high_u32): Ditto.
24120 * config/aarch64/aarch64.h (AARCH64_FL_F16SML): New flag.
24121 (AARCH64_FL_FOR_ARCH8_4): New.
24122 (AARCH64_ISA_F16FML): New ISA flag.
24123 (TARGET_F16FML): New feature flag for fp16fml.
24124 (doc/invoke.texi): Document new fp16fml option.
24126 2018-01-10 Michael Collison <michael.collison@arm.com>
24128 * config/aarch64/aarch64-builtins.c:
24129 (aarch64_types_ternopu_imm_qualifiers, TYPES_TERNOPUI): New.
24130 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
24131 (__ARM_FEATURE_SHA3): Define if TARGET_SHA3 is true.
24132 * config/aarch64/aarch64.h (AARCH64_FL_SHA3): New flags.
24133 (AARCH64_ISA_SHA3): New ISA flag.
24134 (TARGET_SHA3): New feature flag for sha3.
24135 * config/aarch64/iterators.md (sha512_op): New int attribute.
24136 (CRYPTO_SHA512): New int iterator.
24137 (UNSPEC_SHA512H): New unspec.
24138 (UNSPEC_SHA512H2): Ditto.
24139 (UNSPEC_SHA512SU0): Ditto.
24140 (UNSPEC_SHA512SU1): Ditto.
24141 * config/aarch64/aarch64-simd-builtins.def
24142 (aarch64_crypto_sha512hqv2di): New builtin.
24143 (aarch64_crypto_sha512h2qv2di): Ditto.
24144 (aarch64_crypto_sha512su0qv2di): Ditto.
24145 (aarch64_crypto_sha512su1qv2di): Ditto.
24146 (aarch64_eor3qv8hi): Ditto.
24147 (aarch64_rax1qv2di): Ditto.
24148 (aarch64_xarqv2di): Ditto.
24149 (aarch64_bcaxqv8hi): Ditto.
24150 * config/aarch64/aarch64-simd.md:
24151 (aarch64_crypto_sha512h<sha512_op>qv2di): New pattern.
24152 (aarch64_crypto_sha512su0qv2di): Ditto.
24153 (aarch64_crypto_sha512su1qv2di): Ditto.
24154 (aarch64_eor3qv8hi): Ditto.
24155 (aarch64_rax1qv2di): Ditto.
24156 (aarch64_xarqv2di): Ditto.
24157 (aarch64_bcaxqv8hi): Ditto.
24158 * config/aarch64/arm_neon.h (vsha512hq_u64): New intrinsic.
24159 (vsha512h2q_u64): Ditto.
24160 (vsha512su0q_u64): Ditto.
24161 (vsha512su1q_u64): Ditto.
24162 (veor3q_u16): Ditto.
24163 (vrax1q_u64): Ditto.
24164 (vxarq_u64): Ditto.
24165 (vbcaxq_u16): Ditto.
24166 * config/arm/types.md (crypto_sha512): New type attribute.
24167 (crypto_sha3): Ditto.
24168 (doc/invoke.texi): Document new sha3 option.
24170 2018-01-10 Michael Collison <michael.collison@arm.com>
24172 * config/aarch64/aarch64-builtins.c:
24173 (aarch64_types_quadopu_imm_qualifiers, TYPES_QUADOPUI): New.
24174 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
24175 (__ARM_FEATURE_SM3): Define if TARGET_SM4 is true.
24176 (__ARM_FEATURE_SM4): Define if TARGET_SM4 is true.
24177 * config/aarch64/aarch64.h (AARCH64_FL_SM4): New flags.
24178 (AARCH64_ISA_SM4): New ISA flag.
24179 (TARGET_SM4): New feature flag for sm4.
24180 * config/aarch64/aarch64-simd-builtins.def
24181 (aarch64_sm3ss1qv4si): Ditto.
24182 (aarch64_sm3tt1aq4si): Ditto.
24183 (aarch64_sm3tt1bq4si): Ditto.
24184 (aarch64_sm3tt2aq4si): Ditto.
24185 (aarch64_sm3tt2bq4si): Ditto.
24186 (aarch64_sm3partw1qv4si): Ditto.
24187 (aarch64_sm3partw2qv4si): Ditto.
24188 (aarch64_sm4eqv4si): Ditto.
24189 (aarch64_sm4ekeyqv4si): Ditto.
24190 * config/aarch64/aarch64-simd.md:
24191 (aarch64_sm3ss1qv4si): Ditto.
24192 (aarch64_sm3tt<sm3tt_op>qv4si): Ditto.
24193 (aarch64_sm3partw<sm3part_op>qv4si): Ditto.
24194 (aarch64_sm4eqv4si): Ditto.
24195 (aarch64_sm4ekeyqv4si): Ditto.
24196 * config/aarch64/iterators.md (sm3tt_op): New int iterator.
24197 (sm3part_op): Ditto.
24198 (CRYPTO_SM3TT): Ditto.
24199 (CRYPTO_SM3PART): Ditto.
24200 (UNSPEC_SM3SS1): New unspec.
24201 (UNSPEC_SM3TT1A): Ditto.
24202 (UNSPEC_SM3TT1B): Ditto.
24203 (UNSPEC_SM3TT2A): Ditto.
24204 (UNSPEC_SM3TT2B): Ditto.
24205 (UNSPEC_SM3PARTW1): Ditto.
24206 (UNSPEC_SM3PARTW2): Ditto.
24207 (UNSPEC_SM4E): Ditto.
24208 (UNSPEC_SM4EKEY): Ditto.
24209 * config/aarch64/constraints.md (Ui2): New constraint.
24210 * config/aarch64/predicates.md (aarch64_imm2): New predicate.
24211 * config/arm/types.md (crypto_sm3): New type attribute.
24212 (crypto_sm4): Ditto.
24213 * config/aarch64/arm_neon.h (vsm3ss1q_u32): New intrinsic.
24214 (vsm3tt1aq_u32): Ditto.
24215 (vsm3tt1bq_u32): Ditto.
24216 (vsm3tt2aq_u32): Ditto.
24217 (vsm3tt2bq_u32): Ditto.
24218 (vsm3partw1q_u32): Ditto.
24219 (vsm3partw2q_u32): Ditto.
24220 (vsm4eq_u32): Ditto.
24221 (vsm4ekeyq_u32): Ditto.
24222 (doc/invoke.texi): Document new sm4 option.
24224 2018-01-10 Michael Collison <michael.collison@arm.com>
24226 * config/aarch64/aarch64-arches.def (armv8.4-a): New architecture.
24227 * config/aarch64/aarch64.h (AARCH64_ISA_V8_4): New ISA flag.
24228 (AARCH64_FL_FOR_ARCH8_4): New.
24229 (AARCH64_FL_V8_4): New flag.
24230 (doc/invoke.texi): Document new armv8.4-a option.
24232 2018-01-10 Michael Collison <michael.collison@arm.com>
24234 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
24235 (__ARM_FEATURE_AES): Define if TARGET_AES is true.
24236 (__ARM_FEATURE_SHA2): Define if TARGET_SHA2 is true.
24237 * config/aarch64/aarch64-option-extension.def: Add
24238 AARCH64_OPT_EXTENSION of 'sha2'.
24239 (aes): Add AARCH64_OPT_EXTENSION of 'aes'.
24240 (crypto): Disable sha2 and aes if crypto disabled.
24241 (crypto): Enable aes and sha2 if enabled.
24242 (simd): Disable sha2 and aes if simd disabled.
24243 * config/aarch64/aarch64.h (AARCH64_FL_AES, AARCH64_FL_SHA2):
24245 (AARCH64_ISA_AES, AARCH64_ISA_SHA2): New ISA flags.
24246 (TARGET_SHA2): New feature flag for sha2.
24247 (TARGET_AES): New feature flag for aes.
24248 * config/aarch64/aarch64-simd.md:
24249 (aarch64_crypto_aes<aes_op>v16qi): Make pattern
24250 conditional on TARGET_AES.
24251 (aarch64_crypto_aes<aesmc_op>v16qi): Ditto.
24252 (aarch64_crypto_sha1hsi): Make pattern conditional
24254 (aarch64_crypto_sha1hv4si): Ditto.
24255 (aarch64_be_crypto_sha1hv4si): Ditto.
24256 (aarch64_crypto_sha1su1v4si): Ditto.
24257 (aarch64_crypto_sha1<sha1_op>v4si): Ditto.
24258 (aarch64_crypto_sha1su0v4si): Ditto.
24259 (aarch64_crypto_sha256h<sha256_op>v4si): Ditto.
24260 (aarch64_crypto_sha256su0v4si): Ditto.
24261 (aarch64_crypto_sha256su1v4si): Ditto.
24262 (doc/invoke.texi): Document new aes and sha2 options.
24264 2018-01-10 Martin Sebor <msebor@redhat.com>
24266 PR tree-optimization/83781
24267 * gimple-fold.c (get_range_strlen): Avoid treating arrays of pointers
24270 2018-01-11 Martin Sebor <msebor@gmail.com>
24271 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
24273 PR tree-optimization/83501
24274 PR tree-optimization/81703
24276 * tree-ssa-strlen.c (get_string_cst): Rename...
24277 (get_string_len): ...to this. Handle global constants.
24278 (handle_char_store): Adjust.
24280 2018-01-10 Kito Cheng <kito.cheng@gmail.com>
24281 Jim Wilson <jimw@sifive.com>
24283 * config/riscv/riscv-protos.h (riscv_output_return): New.
24284 * config/riscv/riscv.c (struct machine_function): New naked_p field.
24285 (riscv_attribute_table, riscv_output_return),
24286 (riscv_handle_fndecl_attribute, riscv_naked_function_p),
24287 (riscv_allocate_stack_slots_for_args, riscv_warn_func_return): New.
24288 (riscv_compute_frame_info): Only compute frame->mask if not a naked
24290 (riscv_expand_prologue): Add early return for naked function.
24291 (riscv_expand_epilogue): Likewise.
24292 (riscv_function_ok_for_sibcall): Return false for naked function.
24293 (riscv_set_current_function): New.
24294 (TARGET_SET_CURRENT_FUNCTION, TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS),
24295 (TARGET_ATTRIBUTE_TABLE, TARGET_WARN_FUNC_RETURN): New.
24296 * config/riscv/riscv.md (simple_return): Call riscv_output_return.
24297 * doc/extend.texi (RISC-V Function Attributes): New.
24299 2018-01-10 Michael Meissner <meissner@linux.vnet.ibm.com>
24301 * config/rs6000/rs6000.c (is_complex_IBM_long_double): Explicitly
24302 check for 128-bit long double before checking TCmode.
24303 * config/rs6000/rs6000.h (FLOAT128_IEEE_P): Explicitly check for
24304 128-bit long doubles before checking TFmode or TCmode.
24305 (FLOAT128_IBM_P): Likewise.
24307 2018-01-10 Martin Sebor <msebor@redhat.com>
24309 PR tree-optimization/83671
24310 * builtins.c (c_strlen): Unconditionally return zero for the empty
24312 Use -Warray-bounds for warnings.
24313 * gimple-fold.c (get_range_strlen): Handle non-constant lengths
24314 for non-constant array indices with COMPONENT_REF, arrays of
24315 arrays, and pointers to arrays.
24316 (gimple_fold_builtin_strlen): Determine and set length range for
24317 non-constant character arrays.
24319 2018-01-10 Aldy Hernandez <aldyh@redhat.com>
24321 PR middle-end/81897
24322 * tree-ssa-uninit.c (convert_control_dep_chain_into_preds): Skip
24325 2018-01-10 Eric Botcazou <ebotcazou@adacore.com>
24327 * dwarf2out.c (dwarf2out_var_location): Do not pass NULL to fprintf.
24329 2018-01-10 Peter Bergner <bergner@vnet.ibm.com>
24332 * config/rs6000/rs6000.c (print_operand) <'y'>: Use
24333 VECTOR_MEM_ALTIVEC_OR_VSX_P.
24334 * config/rs6000/vsx.md (*vsx_le_perm_load_<mode> for VSX_D): Use
24335 indexed_or_indirect_operand predicate.
24336 (*vsx_le_perm_load_<mode> for VSX_W): Likewise.
24337 (*vsx_le_perm_load_v8hi): Likewise.
24338 (*vsx_le_perm_load_v16qi): Likewise.
24339 (*vsx_le_perm_store_<mode> for VSX_D): Likewise.
24340 (*vsx_le_perm_store_<mode> for VSX_W): Likewise.
24341 (*vsx_le_perm_store_v8hi): Likewise.
24342 (*vsx_le_perm_store_v16qi): Likewise.
24343 (eight unnamed splitters): Likewise.
24345 2018-01-10 Peter Bergner <bergner@vnet.ibm.com>
24347 * config/rs6000/x86intrin.h: Change #warning to #error. Update message.
24348 * config/rs6000/emmintrin.h: Likewise.
24349 * config/rs6000/mmintrin.h: Likewise.
24350 * config/rs6000/xmmintrin.h: Likewise.
24352 2018-01-10 David Malcolm <dmalcolm@redhat.com>
24355 * tree-core.h: Document EXPR_LOCATION_WRAPPER_P's usage of
24357 * tree.c (tree_nop_conversion): Return true for location wrapper
24359 (maybe_wrap_with_location): New function.
24360 (selftest::check_strip_nops): New function.
24361 (selftest::test_location_wrappers): New function.
24362 (selftest::tree_c_tests): Call it.
24363 * tree.h (STRIP_ANY_LOCATION_WRAPPER): New macro.
24364 (maybe_wrap_with_location): New decl.
24365 (EXPR_LOCATION_WRAPPER_P): New macro.
24366 (location_wrapper_p): New inline function.
24367 (tree_strip_any_location_wrapper): New inline function.
24369 2018-01-10 H.J. Lu <hongjiu.lu@intel.com>
24372 * config/i386/i386.c (ix86_compute_frame_layout): Always adjust
24373 stack_realign_offset for the largest alignment of stack slot
24375 (ix86_find_max_used_stack_alignment): New function.
24376 (ix86_finalize_stack_frame_flags): Use it. Set
24377 max_used_stack_alignment if we don't realign stack.
24378 * config/i386/i386.h (machine_function): Add
24379 max_used_stack_alignment.
24381 2018-01-10 Christophe Lyon <christophe.lyon@linaro.org>
24383 * config/arm/arm.opt (-mbranch-cost): New option.
24384 * config/arm/arm.h (BRANCH_COST): Take arm_branch_cost into
24387 2018-01-10 Segher Boessenkool <segher@kernel.crashing.org>
24390 * config/rs6000/rs6000.md (load_toc_v4_PIC_2, load_toc_v4_PIC_3b,
24391 load_toc_v4_PIC_3c): Wrap const term in CONST RTL.
24393 2018-01-10 Richard Biener <rguenther@suse.de>
24396 * dwarf2out.c (gen_subprogram_die): Hoist old_die && declaration
24397 early out so it also covers the case where we have a non-NULL
24400 2018-01-10 Richard Sandiford <richard.sandiford@linaro.org>
24402 PR tree-optimization/83753
24403 * tree-vect-stmts.c (get_group_load_store_type): Use VMAT_CONTIGUOUS
24404 for non-strided grouped accesses if the number of elements is 1.
24406 2018-01-10 Jan Hubicka <hubicka@ucw.cz>
24409 * i386.c (ix86_vectorize_builtin_gather): Check TARGET_USE_GATHER.
24410 * i386.h (TARGET_USE_GATHER): Define.
24411 * x86-tune.def (X86_TUNE_USE_GATHER): New.
24413 2018-01-10 Martin Liska <mliska@suse.cz>
24416 * basic-block.h (CLEANUP_NO_PARTITIONING): New define.
24417 * bb-reorder.c (pass_reorder_blocks::execute): Do not clean up
24419 * cfgcleanup.c (try_optimize_cfg): Fix up partitioning if
24420 CLEANUP_NO_PARTITIONING is not set.
24422 2018-01-10 Richard Sandiford <richard.sandiford@linaro.org>
24424 * doc/rtl.texi: Remove documentation of (const ...) wrappers
24425 for vectors, as a partial revert of r254296.
24426 * rtl.h (const_vec_p): Delete.
24427 (const_vec_duplicate_p): Don't test for vector CONSTs.
24428 (unwrap_const_vec_duplicate, const_vec_series_p): Likewise.
24429 * expmed.c (make_tree): Likewise.
24432 * common.md (E, F): Use CONSTANT_P instead of checking for
24434 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
24435 checking for CONST_VECTOR.
24437 2018-01-09 Jan Hubicka <hubicka@ucw.cz>
24439 PR middle-end/83575
24440 * predict.c (force_edge_cold): Handle in more sane way edges
24441 with no prediction.
24443 2018-01-09 Carl Love <cel@us.ibm.com>
24445 * config/rs6002/altivec.md (p8_vmrgow): Add support for V2DI, V2DF,
24447 (p8_vmrgew): Add support for V2DI, V2DF, V4SF types.
24448 * config/rs6000/rs6000-builtin.def: Add definitions for FLOAT2_V2DF,
24449 VMRGEW_V2DI, VMRGEW_V2DF, VMRGEW_V4SF, VMRGOW_V4SI, VMRGOW_V4SF,
24450 VMRGOW_V2DI, VMRGOW_V2DF. Remove definition for VMRGOW.
24451 * config/rs6000/rs6000-c.c (VSX_BUILTIN_VEC_FLOAT2,
24452 P8V_BUILTIN_VEC_VMRGEW, P8V_BUILTIN_VEC_VMRGOW): Add definitions.
24453 * config/rs6000/rs6000-protos.h: Add extern defition for
24454 rs6000_generate_float2_double_code.
24455 * config/rs6000/rs6000.c (rs6000_generate_float2_double_code): Add
24457 * config/rs6000/vsx.md (vsx_xvcdpsp): Add define_insn.
24458 (float2_v2df): Add define_expand.
24460 2018-01-09 Uros Bizjak <ubizjak@gmail.com>
24463 * combine.c (force_int_to_mode) <case ASHIFT>: Use mode instead of
24464 op_mode in the force_to_mode call.
24466 2018-01-09 Richard Sandiford <richard.sandiford@linaro.org>
24468 * config/aarch64/aarch64.c (aarch64_evpc_trn): Use d.perm.series_p
24469 instead of checking each element individually.
24470 (aarch64_evpc_uzp): Likewise.
24471 (aarch64_evpc_zip): Likewise.
24472 (aarch64_evpc_ext): Likewise.
24473 (aarch64_evpc_rev): Likewise.
24474 (aarch64_evpc_dup): Test the encoding for a single duplicated element,
24475 instead of checking each element individually. Return true without
24477 (aarch64_vectorize_vec_perm_const): Use all_from_input_p to test
24478 whether all selected elements come from the same input, instead of
24479 checking each element individually. Remove calls to gen_rtx_REG,
24480 start_sequence and end_sequence and instead assert that no rtl is
24483 2018-01-09 Richard Sandiford <richard.sandiford@linaro.org>
24485 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p): Fix
24486 order of HIGH and CONST checks.
24488 2018-01-09 Richard Sandiford <richard.sandiford@linaro.org>
24490 * tree-vect-stmts.c (permute_vec_elements): Create a fresh variable
24491 if the destination isn't an SSA_NAME.
24493 2018-01-09 Richard Biener <rguenther@suse.de>
24495 PR tree-optimization/83668
24496 * graphite.c (canonicalize_loop_closed_ssa): Add edge argument,
24498 (canonicalize_loop_form): ... here, renamed from ...
24499 (canonicalize_loop_closed_ssa_form): ... this and amended to
24500 swap successor edges for loop exit blocks to make us use
24501 the RPO order we need for initial schedule generation.
24503 2018-01-09 Joseph Myers <joseph@codesourcery.com>
24505 PR tree-optimization/64811
24506 * match.pd: When optimizing comparisons with Inf, avoid
24507 introducing or losing exceptions from comparisons with NaN.
24509 2018-01-09 Martin Liska <mliska@suse.cz>
24512 * asan.c (shadow_mem_size): Add gcc_assert.
24514 2018-01-09 Georg-Johann Lay <avr@gjlay.de>
24516 Don't save registers in main().
24519 * doc/invoke.texi (AVR Options) [-mmain-is-OS_task]: Document it.
24520 * config/avr/avr.opt (-mmain-is-OS_task): New target option.
24521 * config/avr/avr.c (avr_set_current_function): Don't error if
24522 naked, OS_task or OS_main are specified at the same time.
24523 (avr_function_ok_for_sibcall): Don't disable sibcalls for OS_task,
24525 (avr_insert_attributes) [-mmain-is-OS_task] <main>: Add OS_task
24527 * common/config/avr/avr-common.c (avr_option_optimization_table):
24528 Switch on -mmain-is-OS_task for optimizing compilations.
24530 2018-01-09 Richard Biener <rguenther@suse.de>
24532 PR tree-optimization/83572
24533 * graphite.c: Include cfganal.h.
24534 (graphite_transform_loops): Connect infinite loops to exit
24535 and remove fake edges at the end.
24537 2018-01-09 Jan Hubicka <hubicka@ucw.cz>
24539 * ipa-inline.c (edge_badness): Revert accidental checkin.
24541 2018-01-09 Jan Hubicka <hubicka@ucw.cz>
24544 * ipa-comdats.c (set_comdat_group): Only set comdat group of real
24545 symbols; not inline clones.
24547 2018-01-09 Jakub Jelinek <jakub@redhat.com>
24550 * modulo-sched.c (schedule_reg_moves): Punt if we'd need to move
24551 hard registers. Formatting fixes.
24553 PR preprocessor/83722
24554 * gcc.c (try_generate_repro): Pass
24555 &temp_stderr_files[RETRY_ICE_ATTEMPTS - 1] rather than
24556 &temp_stdout_files[RETRY_ICE_ATTEMPTS - 1] as last argument to
24559 2018-01-08 Monk Chiang <sh.chiang04@gmail.com>
24560 Kito Cheng <kito.cheng@gmail.com>
24562 * config/riscv/riscv.c (machine_function::is_leaf): Remove field.
24563 (riscv_leaf_function_p): Delete.
24564 (riscv_function_ok_for_sibcall): Return false when TARGET_SAVE_RESTORE.
24566 2018-01-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
24568 * config/rs6000/rs6000-string.c (do_load_for_compare_from_addr): New
24570 (do_ifelse): New function.
24571 (do_isel): New function.
24572 (do_sub3): New function.
24573 (do_add3): New function.
24574 (do_load_mask_compare): New function.
24575 (do_overlap_load_compare): New function.
24576 (expand_compare_loop): New function.
24577 (expand_block_compare): Call expand_compare_loop() when appropriate.
24578 * config/rs6000/rs6000.opt (-mblock-compare-inline-limit): Change
24579 option description.
24580 (-mblock-compare-inline-loop-limit): New option.
24582 2018-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24585 * config/rs6000/altivec.md (*altivec_vpermr_<mode>_internal):
24586 Reverse order of second and third operands in first alternative.
24587 * config/rs6000/rs6000.c (rs6000_expand_vector_set): Reverse order
24588 of first and second elements in UNSPEC_VPERMR vector.
24589 (altivec_expand_vec_perm_le): Likewise.
24591 2018-01-08 Jeff Law <law@redhat.com>
24593 PR rtl-optimizatin/81308
24594 * tree-switch-conversion.c (cfg_altered): New file scoped static.
24595 (process_switch): If group_case_labels makes a change, then set
24597 (pass_convert_switch::execute): If a switch is converted, then
24598 set cfg_altered. Return TODO_cfg_cleanup if cfg_altered is true.
24600 PR rtl-optimization/81308
24601 * recog.c (split_all_insns): Conditionally cleanup the CFG after
24604 2018-01-08 Vidya Praveen <vidyapraveen@arm.com>
24606 PR target/83663 - Revert r255946
24607 * config/aarch64/aarch64.c (aarch64_expand_vector_init): Modify code
24608 generation for cases where splatting a value is not useful.
24609 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
24610 across a vec_duplicate and a paradoxical subreg forming a vector
24611 mode to a vec_concat.
24613 2018-01-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24615 * config/arm/t-aprofile (MULTILIB_MATCHES): Add mapping rules for
24616 -march=armv8.3-a variants.
24617 * config/arm/t-multilib: Likewise.
24618 * config/arm/t-arm-elf: Likewise. Handle dotprod extension.
24620 2018-01-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
24622 * config/rs6000/rs6000.md (cceq_ior_compare): Remove * so I can use it
24624 (cceq_ior_compare_complement): Give it a name so I can use it, and
24625 change boolean_or_operator predicate to boolean_operator so it can
24626 be used to generate a crand.
24627 (eqne): New code iterator.
24628 (bd/bd_neg): New code_attrs.
24629 (<bd>_<mode>): New name for ctr<mode>_internal[12] now combined into
24630 a single define_insn.
24631 (<bd>tf_<mode>): A new insn pattern for the conditional form branch
24632 decrement (bdnzt/bdnzf/bdzt/bdzf).
24633 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Updated
24634 with the new names of the branch decrement patterns, and added the
24635 names of the branch decrement conditional patterns.
24637 2018-01-08 Richard Biener <rguenther@suse.de>
24639 PR tree-optimization/83563
24640 * graphite.c (canonicalize_loop_closed_ssa_form): Reset the SCEV
24643 2018-01-08 Richard Biener <rguenther@suse.de>
24645 PR middle-end/83713
24646 * convert.c (do_narrow): Properly guard TYPE_OVERFLOW_WRAPS checks.
24648 2018-01-08 Richard Biener <rguenther@suse.de>
24650 PR tree-optimization/83685
24651 * tree-ssa-pre.c (create_expression_by_pieces): Do not insert
24652 references to abnormals.
24654 2018-01-08 Richard Biener <rguenther@suse.de>
24657 * dwarf2out.c (output_indirect_strings): Handle empty
24658 skeleton_debug_str_hash.
24659 (dwarf2out_early_finish): Index strings for -gsplit-dwarf.
24661 2018-01-08 Claudiu Zissulescu <claziss@synopsys.com>
24663 * config/arc/arc.c (TARGET_TRAMPOLINE_ADJUST_ADDRESS): Delete.
24664 (emit_store_direct): Likewise.
24665 (arc_trampoline_adjust_address): Likewise.
24666 (arc_asm_trampoline_template): New function.
24667 (arc_initialize_trampoline): Use asm_trampoline_template.
24668 (TARGET_ASM_TRAMPOLINE_TEMPLATE): Define.
24669 * config/arc/arc.h (TRAMPOLINE_SIZE): Adjust to 16.
24670 * config/arc/arc.md (flush_icache): Delete pattern.
24672 2018-01-08 Claudiu Zissulescu <claziss@synopsys.com>
24674 * config/arc/arc-c.def (__ARC_UNALIGNED__): New define.
24675 * config/arc/arc.h (STRICT_ALIGNMENT): Control this macro using
24678 2018-01-08 Sebastian Huber <sebastian.huber@embedded-brains.de>
24681 * config/epiphany/epiphany.h (make_pass_mode_switch_use): Guard
24682 by not USED_FOR_TARGET.
24683 (make_pass_resolve_sw_modes): Likewise.
24685 2018-01-08 Sebastian Huber <sebastian.huber@embedded-brains.de>
24687 * config/nios2/nios2.h (nios2_section_threshold): Guard by not
24690 2018-01-08 Richard Biener <rguenther@suse.de>
24692 PR middle-end/83580
24693 * tree-data-ref.c (split_constant_offset): Remove STRIP_NOPS.
24695 2018-01-08 Richard Biener <rguenther@suse.de>
24697 PR middle-end/83517
24698 * match.pd ((t * 2) / 2) -> t): Add missing :c.
24700 2018-01-06 Aldy Hernandez <aldyh@redhat.com>
24702 PR middle-end/81897
24703 * tree-ssa-uninit.c (compute_control_dep_chain): Do not bail on
24704 basic blocks with a small number of successors.
24705 (convert_control_dep_chain_into_preds): Improve handling of
24707 (dump_predicates): Split apart into...
24708 (dump_pred_chain): ...here...
24709 (dump_pred_info): ...and here.
24710 (can_one_predicate_be_invalidated_p): Add debugging printfs.
24711 (can_chain_union_be_invalidated_p): Improve check for invalidation
24713 (uninit_uses_cannot_happen): Avoid unnecessary if
24714 convert_control_dep_chain_into_preds yielded nothing.
24716 2018-01-06 Martin Sebor <msebor@redhat.com>
24718 PR tree-optimization/83640
24719 * gimple-ssa-warn-restrict.c (builtin_access::builtin_access): Avoid
24720 subtracting negative offset from size.
24721 (builtin_access::overlap): Adjust offset bounds of the access to fall
24722 within the size of the object if possible.
24724 2018-01-06 Richard Sandiford <richard.sandiford@linaro.org>
24726 PR rtl-optimization/83699
24727 * expmed.c (extract_bit_field_1): Restrict the vector usage of
24728 extract_bit_field_as_subreg to cases in which the extracted
24729 value is also a vector.
24731 * lra-constraints.c (process_alt_operands): Test for the equivalence
24732 substitutions when detecting a possible reload cycle.
24734 2018-01-06 Jakub Jelinek <jakub@redhat.com>
24737 * toplev.c (process_options): Don't enable debug_nonbind_markers_p
24738 by default if flag_selective_schedling{,2}. Formatting fixes.
24740 PR rtl-optimization/83682
24741 * rtl.h (const_vec_duplicate_p): Only return true for VEC_DUPLICATE
24742 if it has non-VECTOR_MODE element mode.
24743 (vec_duplicate_p): Likewise.
24745 PR middle-end/83694
24746 * cfgexpand.c (expand_debug_expr): Punt if mode1 is VOIDmode
24747 and bitsize might be greater than MAX_BITSIZE_MODE_ANY_INT.
24749 2018-01-05 Jakub Jelinek <jakub@redhat.com>
24752 * config/i386/i386-builtin.def
24753 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
24754 __builtin_ia32_vgf2p8affineqb_v64qi, __builtin_ia32_vgf2p8mulb_v64qi):
24755 Require also OPTION_MASK_ISA_AVX512F in addition to
24756 OPTION_MASK_ISA_GFNI.
24757 (__builtin_ia32_vgf2p8affineinvqb_v16qi_mask,
24758 __builtin_ia32_vgf2p8affineqb_v16qi_mask): Require
24759 OPTION_MASK_ISA_AVX512VL instead of OPTION_MASK_ISA_SSE in addition
24760 to OPTION_MASK_ISA_GFNI.
24761 (__builtin_ia32_vgf2p8mulb_v32qi_mask): Require
24762 OPTION_MASK_ISA_AVX512VL in addition to OPTION_MASK_ISA_GFNI and
24763 OPTION_MASK_ISA_AVX512BW.
24764 (__builtin_ia32_vgf2p8mulb_v16qi_mask): Require
24765 OPTION_MASK_ISA_AVX512VL instead of OPTION_MASK_ISA_AVX512BW in
24766 addition to OPTION_MASK_ISA_GFNI.
24767 (__builtin_ia32_vgf2p8affineinvqb_v16qi,
24768 __builtin_ia32_vgf2p8affineqb_v16qi, __builtin_ia32_vgf2p8mulb_v16qi):
24769 Require OPTION_MASK_ISA_SSE2 instead of OPTION_MASK_ISA_SSE in addition
24770 to OPTION_MASK_ISA_GFNI.
24771 * config/i386/i386.c (def_builtin): Change to builtin isa/isa2 being
24772 a requirement for all ISAs rather than any of them with a few
24774 (ix86_add_new_builtins): Clear OPTION_MASK_ISA_64BIT from isa before
24776 (ix86_expand_builtin): Require all ISAs from builtin's isa and isa2
24777 bitmasks to be enabled with 3 exceptions, instead of requiring any
24778 enabled ISA with lots of exceptions.
24779 * config/i386/sse.md (vgf2p8affineinvqb_<mode><mask_name>,
24780 vgf2p8affineqb_<mode><mask_name>, vgf2p8mulb_<mode><mask_name>):
24781 Change avx512bw in isa attribute to avx512f.
24782 * config/i386/sgxintrin.h: Add license boilerplate.
24783 * config/i386/vaesintrin.h: Likewise. Fix macro spelling __AVX512F
24784 to __AVX512F__ and __AVX512VL to __AVX512VL__.
24785 (_mm256_aesdec_epi128, _mm256_aesdeclast_epi128, _mm256_aesenc_epi128,
24786 _mm256_aesenclast_epi128): Enable temporarily avx if __AVX__ is not
24788 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8,
24789 _mm_gf2p8affineinv_epi64_epi8, _mm_gf2p8affine_epi64_epi8): Enable
24790 temporarily sse2 rather than sse if not enabled already.
24793 * config/i386/sse.md (VI248_VLBW): Rename to ...
24794 (VI248_AVX512VL): ... this. Don't guard V32HI with TARGET_AVX512BW.
24795 (vpshrd_<mode><mask_name>, vpshld_<mode><mask_name>,
24796 vpshrdv_<mode>, vpshrdv_<mode>_mask, vpshrdv_<mode>_maskz,
24797 vpshrdv_<mode>_maskz_1, vpshldv_<mode>, vpshldv_<mode>_mask,
24798 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1): Use VI248_AVX512VL
24799 mode iterator instead of VI248_VLBW.
24801 2018-01-05 Jan Hubicka <hubicka@ucw.cz>
24803 * ipa-fnsummary.c (record_modified_bb_info): Add OP.
24804 (record_modified): Skip clobbers; add debug output.
24805 (param_change_prob): Use sreal frequencies.
24807 2018-01-05 Richard Sandiford <richard.sandiford@linaro.org>
24809 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
24810 punt for user-aligned variables.
24812 2018-01-05 Richard Sandiford <richard.sandiford@linaro.org>
24814 * tree-chrec.c (chrec_contains_symbols): Return true for
24817 2018-01-05 Sudakshina Das <sudi.das@arm.com>
24820 * simplify-rtx.c (simplify_relational_operation_1): Add simplifications
24821 of (x|y) == x for BICS pattern.
24823 2018-01-05 Jakub Jelinek <jakub@redhat.com>
24825 PR tree-optimization/83605
24826 * gimple-ssa-strength-reduction.c: Include tree-eh.h.
24827 (find_candidates_dom_walker::before_dom_children): Ignore stmts that
24830 2018-01-05 Sebastian Huber <sebastian.huber@embedded-brains.de>
24832 * config.gcc (epiphany-*-elf*): Add (epiphany-*-rtems*) configuration.
24833 * config/epiphany/rtems.h: New file.
24835 2018-01-04 Jakub Jelinek <jakub@redhat.com>
24836 Uros Bizjak <ubizjak@gmail.com>
24839 * config/i386/i386.md (*<rotate_insn>hi3_1 splitter): Use
24840 QIreg_operand instead of register_operand predicate.
24841 * config/i386/i386.c (ix86_rop_should_change_byte_p,
24842 set_rop_modrm_reg_bits, ix86_mitigate_rop): Use -mmitigate-rop in
24843 comments instead of -fmitigate[-_]rop.
24845 2018-01-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
24848 * cgraphunit.c (symbol_table::compile): Switch to text_section
24849 before calling assembly_start debug hook.
24850 * run-rtl-passes.c (run_rtl_passes): Likewise.
24853 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
24855 * tree-vrp.c (extract_range_from_binary_expr_1): Check
24856 range_int_cst_p rather than !symbolic_range_p before calling
24857 extract_range_from_multiplicative_op_1.
24859 2018-01-04 Jeff Law <law@redhat.com>
24861 * tree-ssa-math-opts.c (execute_cse_reciprocals_1): Remove
24862 redundant test in assertion.
24864 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
24866 * doc/rtl.texi: Document machine_mode wrapper classes.
24868 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
24870 * fold-const.c (fold_ternary_loc): Check tree_fits_uhwi_p before
24871 using tree_to_uhwi.
24873 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
24875 * tree-ssa-forwprop.c (is_combined_permutation_identity): Allow
24876 the VEC_PERM_EXPR fold to fail.
24878 2018-01-04 Jakub Jelinek <jakub@redhat.com>
24881 * bb-reorder.c (insert_section_boundary_note): Set has_bb_partition
24882 to switched_sections.
24884 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
24887 * config/arm/arm.c (arm_vectorize_vec_perm_const): Fix inverted
24888 test for d.testing.
24890 2018-01-04 Peter Bergner <bergner@vnet.ibm.com>
24893 * config/rs6000/rs6000.c (rs6000_discover_homogeneous_aggregate): Do not
24894 allow arguments in FP registers if TARGET_HARD_FLOAT is false.
24896 2018-01-04 Jakub Jelinek <jakub@redhat.com>
24899 * cfgexpand.c (expand_debug_expr) <case BIT_FIELD_REF>: Punt if mode
24900 is BLKmode and bitpos not zero or mode change is needed.
24902 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
24905 * config/sparc/sparc.c (sparc_vectorize_vec_perm_const): Require
24908 2018-01-04 Uros Bizjak <ubizjak@gmail.com>
24911 * config/alpha/alpha.md (*sadd<modesuffix>): Use ASHIFT
24912 instead of MULT rtx. Update all corresponding splitters.
24913 (*saddl_se): Ditto.
24914 (*ssub<modesuffix>): Ditto.
24915 (*ssubl_se): Ditto.
24916 (*cmp_sadd_di): Update split patterns.
24917 (*cmp_sadd_si): Ditto.
24918 (*cmp_sadd_sidi): Ditto.
24919 (*cmp_ssub_di): Ditto.
24920 (*cmp_ssub_si): Ditto.
24921 (*cmp_ssub_sidi): Ditto.
24922 * config/alpha/predicates.md (const23_operand): New predicate.
24923 * config/alpha/alpha.c (alpha_rtx_costs) [PLUS, MINUS]:
24924 Look for ASHIFT, not MULT inner operand.
24925 (alpha_split_conditional_move): Update for *sadd<modesuffix> change.
24927 2018-01-04 Martin Liska <mliska@suse.cz>
24929 PR gcov-profile/83669
24930 * gcov.c (output_intermediate_file): Add version to intermediate
24932 * doc/gcov.texi: Document new field 'version' in intermediate
24933 file format. Fix location of '-k' option of gcov command.
24935 2018-01-04 Martin Liska <mliska@suse.cz>
24938 * ipa-icf.c (sem_function::merge): Do not cross comdat boundary.
24940 2018-01-04 Jakub Jelinek <jakub@redhat.com>
24942 * gimple-ssa-sprintf.c (parse_directive): Cast second dir.len to uhwi.
24944 2018-01-03 Martin Sebor <msebor@redhat.com>
24946 PR tree-optimization/83655
24947 * gimple-ssa-warn-restrict.c (wrestrict_dom_walker::check_call): Avoid
24948 checking calls with invalid arguments.
24950 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24952 * tree-vect-stmts.c (vect_get_store_rhs): New function.
24953 (vectorizable_mask_load_store): Delete.
24954 (vectorizable_call): Return false for masked loads and stores.
24955 (vectorizable_store): Handle IFN_MASK_STORE. Use vect_get_store_rhs
24956 instead of gimple_assign_rhs1.
24957 (vectorizable_load): Handle IFN_MASK_LOAD.
24958 (vect_transform_stmt): Don't set is_store for call_vec_info_type.
24960 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24962 * tree-vect-stmts.c (vect_build_gather_load_calls): New function,
24964 (vectorizable_mask_load_store): ...here.
24965 (vectorizable_load): ...and here.
24967 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24969 * tree-vect-stmts.c (vect_build_all_ones_mask)
24970 (vect_build_zero_merge_argument): New functions, split out from...
24971 (vectorizable_load): ...here.
24973 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24975 * tree-vect-stmts.c (vect_check_store_rhs): New function,
24977 (vectorizable_mask_load_store): ...here.
24978 (vectorizable_store): ...and here.
24980 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24982 * tree-vect-stmts.c (vect_check_load_store_mask): New function,
24984 (vectorizable_mask_load_store): ...here.
24986 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24988 * tree-vectorizer.h (vec_load_store_type): Moved from tree-vec-stmts.c
24989 (vect_model_store_cost): Take a vec_load_store_type instead of a
24991 * tree-vect-stmts.c (vec_load_store_type): Move to tree-vectorizer.h.
24992 (vect_model_store_cost): Take a vec_load_store_type instead of a
24994 (vectorizable_mask_load_store): Update accordingly.
24995 (vectorizable_store): Likewise.
24996 * tree-vect-slp.c (vect_analyze_slp_cost_1): Update accordingly.
24998 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25000 * tree-vect-loop.c (vect_transform_loop): Stub out scalar
25001 IFN_MASK_LOAD calls here rather than...
25002 * tree-vect-stmts.c (vectorizable_mask_load_store): ...here.
25004 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25005 Alan Hayward <alan.hayward@arm.com>
25006 David Sherwood <david.sherwood@arm.com>
25008 * expmed.c (extract_bit_field_1): For vector extracts,
25009 fall back to extract_bit_field_as_subreg if vec_extract
25012 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25013 Alan Hayward <alan.hayward@arm.com>
25014 David Sherwood <david.sherwood@arm.com>
25016 * lra-spills.c (pseudo_reg_slot_compare): Sort slots by whether
25017 they are variable or constant sized.
25018 (assign_stack_slot_num_and_sort_pseudos): Don't reuse variable-sized
25019 slots for constant-sized data.
25021 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25022 Alan Hayward <alan.hayward@arm.com>
25023 David Sherwood <david.sherwood@arm.com>
25025 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern): When
25026 handling COND_EXPRs with boolean comparisons, try to find a better
25027 basis for the mask type than the boolean itself.
25029 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25031 * doc/rtl.texi (MAX_BITSIZE_MODE_ANY_MODE): Describe how the default
25032 is calculated and how it can be overridden.
25033 * genmodes.c (max_bitsize_mode_any_mode): New variable.
25034 (create_modes): Initialize it from MAX_BITSIZE_MODE_ANY_MODE,
25036 (emit_max_int): Use it to set the output MAX_BITSIZE_MODE_ANY_MODE,
25039 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25040 Alan Hayward <alan.hayward@arm.com>
25041 David Sherwood <david.sherwood@arm.com>
25043 * config/aarch64/aarch64-protos.h (aarch64_output_simd_mov_immediate):
25044 Remove the mode argument.
25045 (aarch64_simd_valid_immediate): Remove the mode and inverse
25047 * config/aarch64/iterators.md (bitsize): New iterator.
25048 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>, and<mode>3)
25049 (ior<mode>3): Update calls to aarch64_output_simd_mov_immediate.
25050 * config/aarch64/constraints.md (Do, Db, Dn): Update calls to
25051 aarch64_simd_valid_immediate.
25052 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Likewise.
25053 (aarch64_reg_or_bic_imm): Likewise.
25054 * config/aarch64/aarch64.c (simd_immediate_info): Replace mvn
25055 with an insn_type enum and msl with a modifier_type enum.
25056 Replace element_width with a scalar_mode. Change the shift
25057 to unsigned int. Add constructors for scalar_float_mode and
25058 scalar_int_mode elements.
25059 (aarch64_vect_float_const_representable_p): Delete.
25060 (aarch64_can_const_movi_rtx_p)
25061 (aarch64_simd_scalar_immediate_valid_for_move)
25062 (aarch64_simd_make_constant): Update call to
25063 aarch64_simd_valid_immediate.
25064 (aarch64_advsimd_valid_immediate_hs): New function.
25065 (aarch64_advsimd_valid_immediate): Likewise.
25066 (aarch64_simd_valid_immediate): Remove mode and inverse
25067 arguments. Rewrite to use the above. Use const_vec_duplicate_p
25068 to detect duplicated constants and use aarch64_float_const_zero_rtx_p
25069 and aarch64_float_const_representable_p on the result.
25070 (aarch64_output_simd_mov_immediate): Remove mode argument.
25071 Update call to aarch64_simd_valid_immediate and use of
25072 simd_immediate_info.
25073 (aarch64_output_scalar_simd_mov_immediate): Update call
25076 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25077 Alan Hayward <alan.hayward@arm.com>
25078 David Sherwood <david.sherwood@arm.com>
25080 * machmode.h (mode_precision): Prefix with CONST_MODE_PRECISION.
25081 (mode_nunits): Likewise CONST_MODE_NUNITS.
25082 * machmode.def (ADJUST_NUNITS): Document.
25083 * genmodes.c (mode_data::need_nunits_adj): New field.
25084 (blank_mode): Update accordingly.
25085 (adj_nunits): New variable.
25086 (print_maybe_const_decl): Replace CATEGORY with a NEEDS_ADJ
25088 (emit_mode_size_inline): Set need_bytesize_adj for all modes
25089 listed in adj_nunits.
25090 (emit_mode_nunits_inline): Set need_nunits_adj for all modes
25091 listed in adj_nunits. Don't emit case statements for such modes.
25092 (emit_insn_modes_h): Emit definitions of CONST_MODE_NUNITS
25093 and CONST_MODE_PRECISION. Make CONST_MODE_SIZE expand to
25094 nothing if adj_nunits is nonnull.
25095 (emit_mode_precision, emit_mode_nunits): Use print_maybe_const_decl.
25096 (emit_mode_unit_size, emit_mode_base_align, emit_mode_ibit)
25097 (emit_mode_fbit): Update use of print_maybe_const_decl.
25098 (emit_move_size): Likewise. Treat the array as non-const
25100 (emit_mode_adjustments): Handle adj_nunits.
25102 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25104 * machmode.def (VECTOR_MODES_WITH_PREFIX): Document.
25105 * genmodes.c (VECTOR_MODES_WITH_PREFIX): New macro.
25106 (VECTOR_MODES): Use it.
25107 (make_vector_modes): Take the prefix as an argument.
25109 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25110 Alan Hayward <alan.hayward@arm.com>
25111 David Sherwood <david.sherwood@arm.com>
25113 * mode-classes.def (MODE_VECTOR_BOOL): New mode class.
25114 * machmode.h (INTEGRAL_MODE_P, VECTOR_MODE_P): Return true
25115 for MODE_VECTOR_BOOL.
25116 * machmode.def (VECTOR_BOOL_MODE): Document.
25117 * genmodes.c (VECTOR_BOOL_MODE): New macro.
25118 (make_vector_bool_mode): New function.
25119 (complete_mode, emit_mode_wider, emit_mode_adjustments): Handle
25121 * lto-streamer-in.c (lto_input_mode_table): Likewise.
25122 * rtx-vector-builder.c (rtx_vector_builder::find_cached_value):
25124 * stor-layout.c (int_mode_for_mode): Likewise.
25125 * tree.c (build_vector_type_for_mode): Likewise.
25126 * varasm.c (output_constant_pool_2): Likewise.
25127 * emit-rtl.c (init_emit_once): Make sure that CONST1_RTX (BImode) and
25128 CONSTM1_RTX (BImode) are the same thing. Initialize const_tiny_rtx
25129 for MODE_VECTOR_BOOL.
25130 * expr.c (expand_expr_real_1): Use VECTOR_MODE_P instead of a list
25131 of mode class checks.
25132 * tree-vect-generic.c (expand_vector_operation): Use VECTOR_MODE_P
25133 instead of a list of mode class checks.
25134 (expand_vector_scalar_condition): Likewise.
25135 (type_for_widest_vector_mode): Handle BImode as an inner mode.
25137 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25138 Alan Hayward <alan.hayward@arm.com>
25139 David Sherwood <david.sherwood@arm.com>
25141 * machmode.h (mode_size): Change from unsigned short to
25143 (mode_to_bytes): Return a poly_uint16 rather than an unsigned short.
25144 (GET_MODE_SIZE): Return a constant if ONLY_FIXED_SIZE_MODES,
25145 or if measurement_type is not polynomial.
25146 (fixed_size_mode::includes_p): Check for constant-sized modes.
25147 * genmodes.c (emit_mode_size_inline): Make mode_size_inline
25148 return a poly_uint16 rather than an unsigned short.
25149 (emit_mode_size): Change the type of mode_size from unsigned short
25150 to poly_uint16_pod. Use ZERO_COEFFS for the initializer.
25151 (emit_mode_adjustments): Cope with polynomial vector sizes.
25152 * lto-streamer-in.c (lto_input_mode_table): Use bp_unpack_poly_value
25154 * lto-streamer-out.c (lto_write_mode_table): Use bp_pack_poly_value
25156 * auto-inc-dec.c (try_merge): Treat GET_MODE_SIZE as polynomial.
25157 * builtins.c (expand_ifn_atomic_compare_exchange_into_call): Likewise.
25158 * caller-save.c (setup_save_areas): Likewise.
25159 (replace_reg_with_saved_mem): Likewise.
25160 * calls.c (emit_library_call_value_1): Likewise.
25161 * combine-stack-adj.c (combine_stack_adjustments_for_block): Likewise.
25162 * combine.c (simplify_set, make_extraction, simplify_shift_const_1)
25163 (gen_lowpart_for_combine): Likewise.
25164 * convert.c (convert_to_integer_1): Likewise.
25165 * cse.c (equiv_constant, cse_insn): Likewise.
25166 * cselib.c (autoinc_split, cselib_hash_rtx): Likewise.
25167 (cselib_subst_to_values): Likewise.
25168 * dce.c (word_dce_process_block): Likewise.
25169 * df-problems.c (df_word_lr_mark_ref): Likewise.
25170 * dwarf2cfi.c (init_one_dwarf_reg_size): Likewise.
25171 * dwarf2out.c (multiple_reg_loc_descriptor, mem_loc_descriptor)
25172 (concat_loc_descriptor, concatn_loc_descriptor, loc_descriptor)
25173 (rtl_for_decl_location): Likewise.
25174 * emit-rtl.c (gen_highpart, widen_memory_access): Likewise.
25175 * expmed.c (extract_bit_field_1, extract_integral_bit_field): Likewise.
25176 * expr.c (emit_group_load_1, clear_storage_hints): Likewise.
25177 (emit_move_complex, emit_move_multi_word, emit_push_insn): Likewise.
25178 (expand_expr_real_1): Likewise.
25179 * function.c (assign_parm_setup_block_p, assign_parm_setup_block)
25180 (pad_below): Likewise.
25181 * gimple-fold.c (optimize_atomic_compare_exchange_p): Likewise.
25182 * gimple-ssa-store-merging.c (rhs_valid_for_store_merging_p): Likewise.
25183 * ira.c (get_subreg_tracking_sizes): Likewise.
25184 * ira-build.c (ira_create_allocno_objects): Likewise.
25185 * ira-color.c (coalesced_pseudo_reg_slot_compare): Likewise.
25186 (ira_sort_regnos_for_alter_reg): Likewise.
25187 * ira-costs.c (record_operand_costs): Likewise.
25188 * lower-subreg.c (interesting_mode_p, simplify_gen_subreg_concatn)
25189 (resolve_simple_move): Likewise.
25190 * lra-constraints.c (get_reload_reg, operands_match_p): Likewise.
25191 (process_addr_reg, simplify_operand_subreg, curr_insn_transform)
25192 (lra_constraints): Likewise.
25193 (CONST_POOL_OK_P): Reject variable-sized modes.
25194 * lra-spills.c (slot, assign_mem_slot, pseudo_reg_slot_compare)
25195 (add_pseudo_to_slot, lra_spill): Likewise.
25196 * omp-low.c (omp_clause_aligned_alignment): Likewise.
25197 * optabs-query.c (get_best_extraction_insn): Likewise.
25198 * optabs-tree.c (expand_vec_cond_expr_p): Likewise.
25199 * optabs.c (expand_vec_perm_var, expand_vec_cond_expr): Likewise.
25200 (expand_mult_highpart, valid_multiword_target_p): Likewise.
25201 * recog.c (offsettable_address_addr_space_p): Likewise.
25202 * regcprop.c (maybe_mode_change): Likewise.
25203 * reginfo.c (choose_hard_reg_mode, record_subregs_of_mode): Likewise.
25204 * regrename.c (build_def_use): Likewise.
25205 * regstat.c (dump_reg_info): Likewise.
25206 * reload.c (complex_word_subreg_p, push_reload, find_dummy_reload)
25207 (find_reloads, find_reloads_subreg_address): Likewise.
25208 * reload1.c (eliminate_regs_1): Likewise.
25209 * rtlanal.c (for_each_inc_dec_find_inc_dec, rtx_cost): Likewise.
25210 * simplify-rtx.c (avoid_constant_pool_reference): Likewise.
25211 (simplify_binary_operation_1, simplify_subreg): Likewise.
25212 * targhooks.c (default_function_arg_padding): Likewise.
25213 (default_hard_regno_nregs, default_class_max_nregs): Likewise.
25214 * tree-cfg.c (verify_gimple_assign_binary): Likewise.
25215 (verify_gimple_assign_ternary): Likewise.
25216 * tree-inline.c (estimate_move_cost): Likewise.
25217 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
25218 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Likewise.
25219 (get_address_cost_ainc): Likewise.
25220 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
25221 (vect_supportable_dr_alignment): Likewise.
25222 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
25223 (vectorizable_reduction): Likewise.
25224 * tree-vect-stmts.c (vectorizable_assignment, vectorizable_shift)
25225 (vectorizable_operation, vectorizable_load): Likewise.
25226 * tree.c (build_same_sized_truth_vector_type): Likewise.
25227 * valtrack.c (cleanup_auto_inc_dec): Likewise.
25228 * var-tracking.c (emit_note_insn_var_location): Likewise.
25229 * config/arc/arc.h (ASM_OUTPUT_CASE_END): Use as_a <scalar_int_mode>.
25230 (ADDR_VEC_ALIGN): Likewise.
25232 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25233 Alan Hayward <alan.hayward@arm.com>
25234 David Sherwood <david.sherwood@arm.com>
25236 * machmode.h (mode_to_bits): Return a poly_uint16 rather than an
25238 (GET_MODE_BITSIZE): Return a constant if ONLY_FIXED_SIZE_MODES,
25239 or if measurement_type is polynomial.
25240 * calls.c (shift_return_value): Treat GET_MODE_BITSIZE as polynomial.
25241 * combine.c (make_extraction): Likewise.
25242 * dse.c (find_shift_sequence): Likewise.
25243 * dwarf2out.c (mem_loc_descriptor): Likewise.
25244 * expmed.c (store_integral_bit_field, extract_bit_field_1): Likewise.
25245 (extract_bit_field, extract_low_bits): Likewise.
25246 * expr.c (convert_move, convert_modes, emit_move_insn_1): Likewise.
25247 (optimize_bitfield_assignment_op, expand_assignment): Likewise.
25248 (store_expr_with_bounds, store_field, expand_expr_real_1): Likewise.
25249 * fold-const.c (optimize_bit_field_compare, merge_ranges): Likewise.
25250 * gimple-fold.c (optimize_atomic_compare_exchange_p): Likewise.
25251 * reload.c (find_reloads): Likewise.
25252 * reload1.c (alter_reg): Likewise.
25253 * stor-layout.c (bitwise_mode_for_mode, compute_record_mode): Likewise.
25254 * targhooks.c (default_secondary_memory_needed_mode): Likewise.
25255 * tree-if-conv.c (predicate_mem_writes): Likewise.
25256 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
25257 * tree-vect-patterns.c (adjust_bool_pattern): Likewise.
25258 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
25259 * valtrack.c (dead_debug_insert_temp): Likewise.
25260 * varasm.c (mergeable_constant_section): Likewise.
25261 * config/sh/sh.h (LOCAL_ALIGNMENT): Use as_a <fixed_size_mode>.
25263 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25264 Alan Hayward <alan.hayward@arm.com>
25265 David Sherwood <david.sherwood@arm.com>
25267 * expr.c (expand_assignment): Cope with polynomial mode sizes
25268 when assigning to a CONCAT.
25270 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25271 Alan Hayward <alan.hayward@arm.com>
25272 David Sherwood <david.sherwood@arm.com>
25274 * machmode.h (mode_precision): Change from unsigned short to
25276 (mode_to_precision): Return a poly_uint16 rather than an unsigned
25278 (GET_MODE_PRECISION): Return a constant if ONLY_FIXED_SIZE_MODES,
25279 or if measurement_type is not polynomial.
25280 (HWI_COMPUTABLE_MODE_P): Turn into a function. Optimize the case
25281 in which the mode is already known to be a scalar_int_mode.
25282 * genmodes.c (emit_mode_precision): Change the type of mode_precision
25283 from unsigned short to poly_uint16_pod. Use ZERO_COEFFS for the
25285 * lto-streamer-in.c (lto_input_mode_table): Use bp_unpack_poly_value
25286 for GET_MODE_PRECISION.
25287 * lto-streamer-out.c (lto_write_mode_table): Use bp_pack_poly_value
25288 for GET_MODE_PRECISION.
25289 * combine.c (update_rsp_from_reg_equal): Treat GET_MODE_PRECISION
25291 (try_combine, find_split_point, combine_simplify_rtx): Likewise.
25292 (expand_field_assignment, make_extraction): Likewise.
25293 (make_compound_operation_int, record_dead_and_set_regs_1): Likewise.
25294 (get_last_value): Likewise.
25295 * convert.c (convert_to_integer_1): Likewise.
25296 * cse.c (cse_insn): Likewise.
25297 * expr.c (expand_expr_real_1): Likewise.
25298 * lra-constraints.c (simplify_operand_subreg): Likewise.
25299 * optabs-query.c (can_atomic_load_p): Likewise.
25300 * optabs.c (expand_atomic_load): Likewise.
25301 (expand_atomic_store): Likewise.
25302 * ree.c (combine_reaching_defs): Likewise.
25303 * rtl.h (partial_subreg_p, paradoxical_subreg_p): Likewise.
25304 * rtlanal.c (nonzero_bits1, lsb_bitfield_op_p): Likewise.
25305 * tree.h (type_has_mode_precision_p): Likewise.
25306 * ubsan.c (instrument_si_overflow): Likewise.
25308 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25309 Alan Hayward <alan.hayward@arm.com>
25310 David Sherwood <david.sherwood@arm.com>
25312 * tree.h (TYPE_VECTOR_SUBPARTS): Turn into a function and handle
25313 polynomial numbers of units.
25314 (SET_TYPE_VECTOR_SUBPARTS): Likewise.
25315 (valid_vector_subparts_p): New function.
25316 (build_vector_type): Remove temporary shim and take the number
25317 of units as a poly_uint64 rather than an int.
25318 (build_opaque_vector_type): Take the number of units as a
25319 poly_uint64 rather than an int.
25320 * tree.c (build_vector_from_ctor): Handle polynomial
25321 TYPE_VECTOR_SUBPARTS.
25322 (type_hash_canon_hash, type_cache_hasher::equal): Likewise.
25323 (uniform_vector_p, vector_type_mode, build_vector): Likewise.
25324 (build_vector_from_val): If the number of units is variable,
25325 use build_vec_duplicate_cst for constant operands and
25326 VEC_DUPLICATE_EXPR otherwise.
25327 (make_vector_type): Remove temporary is_constant ().
25328 (build_vector_type, build_opaque_vector_type): Take the number of
25329 units as a poly_uint64 rather than an int.
25330 (check_vector_cst): Handle polynomial TYPE_VECTOR_SUBPARTS and
25332 * cfgexpand.c (expand_debug_expr): Likewise.
25333 * expr.c (count_type_elements, categorize_ctor_elements_1): Likewise.
25334 (store_constructor, expand_expr_real_1): Likewise.
25335 (const_scalar_mask_from_tree): Likewise.
25336 * fold-const-call.c (fold_const_reduction): Likewise.
25337 * fold-const.c (const_binop, const_unop, fold_convert_const): Likewise.
25338 (operand_equal_p, fold_vec_perm, fold_ternary_loc): Likewise.
25339 (native_encode_vector, vec_cst_ctor_to_array): Likewise.
25340 (fold_relational_const): Likewise.
25341 (native_interpret_vector): Likewise. Change the size from an
25342 int to an unsigned int.
25343 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Handle polynomial
25344 TYPE_VECTOR_SUBPARTS.
25345 (gimple_fold_indirect_ref, gimple_build_vector): Likewise.
25346 (gimple_build_vector_from_val): Use VEC_DUPLICATE_EXPR when
25347 duplicating a non-constant operand into a variable-length vector.
25348 * hsa-brig.c (hsa_op_immed::emit_to_buffer): Handle polynomial
25349 TYPE_VECTOR_SUBPARTS and VECTOR_CST_NELTS.
25350 * ipa-icf.c (sem_variable::equals): Likewise.
25351 * match.pd: Likewise.
25352 * omp-simd-clone.c (simd_clone_subparts): Likewise.
25353 * print-tree.c (print_node): Likewise.
25354 * stor-layout.c (layout_type): Likewise.
25355 * targhooks.c (default_builtin_vectorization_cost): Likewise.
25356 * tree-cfg.c (verify_gimple_comparison): Likewise.
25357 (verify_gimple_assign_binary): Likewise.
25358 (verify_gimple_assign_ternary): Likewise.
25359 (verify_gimple_assign_single): Likewise.
25360 * tree-pretty-print.c (dump_generic_node): Likewise.
25361 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
25362 (simplify_bitfield_ref, is_combined_permutation_identity): Likewise.
25363 * tree-vect-data-refs.c (vect_permute_store_chain): Likewise.
25364 (vect_grouped_load_supported, vect_permute_load_chain): Likewise.
25365 (vect_shift_permute_load_chain): Likewise.
25366 * tree-vect-generic.c (nunits_for_known_piecewise_op): Likewise.
25367 (expand_vector_condition, optimize_vector_constructor): Likewise.
25368 (lower_vec_perm, get_compute_type): Likewise.
25369 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
25370 (get_initial_defs_for_reduction, vect_transform_loop): Likewise.
25371 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
25372 (vect_recog_mask_conversion_pattern): Likewise.
25373 * tree-vect-slp.c (vect_supported_load_permutation_p): Likewise.
25374 (vect_get_constant_vectors, vect_transform_slp_perm_load): Likewise.
25375 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
25376 (get_group_load_store_type, vectorizable_mask_load_store): Likewise.
25377 (vectorizable_bswap, simd_clone_subparts, vectorizable_assignment)
25378 (vectorizable_shift, vectorizable_operation, vectorizable_store)
25379 (vectorizable_load, vect_is_simple_cond, vectorizable_comparison)
25380 (supportable_widening_operation): Likewise.
25381 (supportable_narrowing_operation): Likewise.
25382 * tree-vector-builder.c (tree_vector_builder::binary_encoded_nelts):
25384 * varasm.c (output_constant): Likewise.
25386 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25387 Alan Hayward <alan.hayward@arm.com>
25388 David Sherwood <david.sherwood@arm.com>
25390 * tree-vect-data-refs.c (vect_permute_store_chain): Reorganize
25391 so that both the length == 3 and length != 3 cases set up their
25392 own permute vectors. Add comments explaining why we know the
25393 number of elements is constant.
25394 (vect_permute_load_chain): Likewise.
25396 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25397 Alan Hayward <alan.hayward@arm.com>
25398 David Sherwood <david.sherwood@arm.com>
25400 * machmode.h (mode_nunits): Change from unsigned char to
25402 (ONLY_FIXED_SIZE_MODES): New macro.
25403 (pod_mode::measurement_type, scalar_int_mode::measurement_type)
25404 (scalar_float_mode::measurement_type, scalar_mode::measurement_type)
25405 (complex_mode::measurement_type, fixed_size_mode::measurement_type):
25407 (mode_to_nunits): Return a poly_uint16 rather than an unsigned short.
25408 (GET_MODE_NUNITS): Return a constant if ONLY_FIXED_SIZE_MODES,
25409 or if measurement_type is not polynomial.
25410 * genmodes.c (ZERO_COEFFS): New macro.
25411 (emit_mode_nunits_inline): Make mode_nunits_inline return a
25413 (emit_mode_nunits): Change the type of mode_nunits to poly_uint16_pod.
25414 Use ZERO_COEFFS when emitting initializers.
25415 * data-streamer.h (bp_pack_poly_value): New function.
25416 (bp_unpack_poly_value): Likewise.
25417 * lto-streamer-in.c (lto_input_mode_table): Use bp_unpack_poly_value
25418 for GET_MODE_NUNITS.
25419 * lto-streamer-out.c (lto_write_mode_table): Use bp_pack_poly_value
25420 for GET_MODE_NUNITS.
25421 * tree.c (make_vector_type): Remove temporary shim and make
25422 the real function take the number of units as a poly_uint64
25423 rather than an int.
25424 (build_vector_type_for_mode): Handle polynomial nunits.
25425 * dwarf2out.c (loc_descriptor, add_const_value_attribute): Likewise.
25426 * emit-rtl.c (const_vec_series_p_1): Likewise.
25427 (gen_rtx_CONST_VECTOR): Likewise.
25428 * fold-const.c (test_vec_duplicate_folding): Likewise.
25429 * genrecog.c (validate_pattern): Likewise.
25430 * optabs-query.c (can_vec_perm_var_p, can_mult_highpart_p): Likewise.
25431 * optabs-tree.c (expand_vec_cond_expr_p): Likewise.
25432 * optabs.c (expand_vector_broadcast, expand_binop_directly): Likewise.
25433 (shift_amt_for_vec_perm_mask, expand_vec_perm_var): Likewise.
25434 (expand_vec_cond_expr, expand_mult_highpart): Likewise.
25435 * rtlanal.c (subreg_get_info): Likewise.
25436 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
25437 (vect_grouped_load_supported): Likewise.
25438 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
25439 * tree-vect-loop.c (have_whole_vector_shift): Likewise.
25440 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
25441 (simplify_const_unary_operation, simplify_binary_operation_1)
25442 (simplify_const_binary_operation, simplify_ternary_operation)
25443 (test_vector_ops_duplicate, test_vector_ops): Likewise.
25444 (simplify_immed_subreg): Use GET_MODE_NUNITS on a fixed_size_mode
25445 instead of CONST_VECTOR_NUNITS.
25446 * varasm.c (output_constant_pool_2): Likewise.
25447 * rtx-vector-builder.c (rtx_vector_builder::build): Only include the
25448 explicit-encoded elements in the XVEC for variable-length vectors.
25450 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25452 * lra-constraints.c (curr_insn_transform): Use partial_subreg_p.
25454 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25455 Alan Hayward <alan.hayward@arm.com>
25456 David Sherwood <david.sherwood@arm.com>
25458 * coretypes.h (fixed_size_mode): Declare.
25459 (fixed_size_mode_pod): New typedef.
25460 * builtins.h (target_builtins::x_apply_args_mode)
25461 (target_builtins::x_apply_result_mode): Change type to
25462 fixed_size_mode_pod.
25463 * builtins.c (apply_args_size, apply_result_size, result_vector)
25464 (expand_builtin_apply_args_1, expand_builtin_apply)
25465 (expand_builtin_return): Update accordingly.
25467 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25469 * cse.c (hash_rtx_cb): Hash only the encoded elements.
25470 * cselib.c (cselib_hash_rtx): Likewise.
25471 * expmed.c (make_tree): Build VECTOR_CSTs directly from the
25472 CONST_VECTOR encoding.
25474 2018-01-03 Jakub Jelinek <jakub@redhat.com>
25475 Jeff Law <law@redhat.com>
25478 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): For
25479 noreturn probe, use gen_pop instead of ix86_emit_restore_reg_using_pop,
25480 only set RTX_FRAME_RELATED_P on both the push and pop if cfa_reg is sp
25481 and add REG_CFA_ADJUST_CFA notes in that case to both insns.
25484 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Do not
25485 explicitly probe *sp in a noreturn function if there were any callee
25486 register saves or frame pointer is needed.
25488 2018-01-03 Jakub Jelinek <jakub@redhat.com>
25491 * cfgexpand.c (expand_debug_expr): Return NULL if mode is
25492 BLKmode for ternary, binary or unary expressions.
25495 * var-tracking.c (delete_vta_debug_insn): New inline function.
25496 (delete_vta_debug_insns): Add USE_CFG argument, if true, walk just
25497 insns from get_insns () to NULL instead of each bb separately.
25498 Use delete_vta_debug_insn. No longer static.
25499 (vt_debug_insns_local, variable_tracking_main_1): Adjust
25500 delete_vta_debug_insns callers.
25501 * rtl.h (delete_vta_debug_insns): Declare.
25502 * final.c (rest_of_handle_final): Call delete_vta_debug_insns
25503 instead of variable_tracking_main.
25505 2018-01-03 Martin Sebor <msebor@redhat.com>
25507 PR tree-optimization/83603
25508 * calls.c (maybe_warn_nonstring_arg): Avoid accessing function
25509 arguments past the endof the argument list in functions declared
25510 without a prototype.
25511 * gimple-ssa-warn-restrict.c (wrestrict_dom_walker::check_call):
25512 Avoid checking when arguments are null.
25514 2018-01-03 Martin Sebor <msebor@redhat.com>
25517 * doc/extend.texi (attribute const): Fix a typo.
25518 * ipa-pure-const.c ((warn_function_const, warn_function_pure): Avoid
25519 issuing -Wsuggest-attribute for void functions.
25521 2018-01-03 Martin Sebor <msebor@redhat.com>
25523 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Use
25524 offset_int::from instead of wide_int::to_shwi.
25525 (maybe_diag_overlap): Remove assertion.
25526 Use HOST_WIDE_INT_PRINT_DEC instead of %lli.
25527 * gimple-ssa-sprintf.c (format_directive): Same.
25528 (parse_directive): Same.
25529 (sprintf_dom_walker::compute_format_length): Same.
25530 (try_substitute_return_value): Same.
25532 2018-01-03 Jeff Law <law@redhat.com>
25534 PR middle-end/83654
25535 * explow.c (anti_adjust_stack_and_probe_stack_clash): Test a
25536 non-constant residual for zero at runtime and avoid probing in
25537 that case. Reorganize code for trailing problem to mirror handling
25540 2018-01-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
25542 PR tree-optimization/83501
25543 * tree-ssa-strlen.c (get_string_cst): New.
25544 (handle_char_store): Call get_string_cst.
25546 2018-01-03 Martin Liska <mliska@suse.cz>
25548 PR tree-optimization/83593
25549 * tree-ssa-strlen.c: Include tree-cfg.h.
25550 (strlen_check_and_optimize_stmt): Add new argument cleanup_eh.
25551 (strlen_dom_walker): Add new member variable m_cleanup_cfg.
25552 (strlen_dom_walker::strlen_dom_walker): Initialize m_cleanup_cfg
25554 (strlen_dom_walker::before_dom_children): Call
25555 gimple_purge_dead_eh_edges. Dump tranformation with details
25557 (strlen_dom_walker::before_dom_children): Update call by adding
25558 new argument cleanup_eh.
25559 (pass_strlen::execute): Return TODO_cleanup_cfg if needed.
25561 2018-01-03 Martin Liska <mliska@suse.cz>
25564 * cif-code.def (VARIADIC_THUNK): New enum value.
25565 * ipa-fnsummary.c (compute_fn_summary): Do not inline variadic
25568 2018-01-03 Jan Beulich <jbeulich@suse.com>
25570 * sse.md (mov<mode>_internal): Tighten condition for when to use
25571 vmovdqu<ssescalarsize> for TI and OI modes.
25573 2018-01-03 Jakub Jelinek <jakub@redhat.com>
25575 Update copyright years.
25577 2018-01-03 Martin Liska <mliska@suse.cz>
25580 * ipa-visibility.c (function_and_variable_visibility): Skip
25581 functions with noipa attribure.
25583 2018-01-03 Jakub Jelinek <jakub@redhat.com>
25585 * gcc.c (process_command): Update copyright notice dates.
25586 * gcov-dump.c (print_version): Ditto.
25587 * gcov.c (print_version): Ditto.
25588 * gcov-tool.c (print_version): Ditto.
25589 * gengtype.c (create_file): Ditto.
25590 * doc/cpp.texi: Bump @copying's copyright year.
25591 * doc/cppinternals.texi: Ditto.
25592 * doc/gcc.texi: Ditto.
25593 * doc/gccint.texi: Ditto.
25594 * doc/gcov.texi: Ditto.
25595 * doc/install.texi: Ditto.
25596 * doc/invoke.texi: Ditto.
25598 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25600 * vector-builder.h (vector_builder::m_full_nelts): Change from
25601 unsigned int to poly_uint64.
25602 (vector_builder::full_nelts): Update prototype accordingly.
25603 (vector_builder::new_vector): Likewise.
25604 (vector_builder::encoded_full_vector_p): Handle polynomial full_nelts.
25605 (vector_builder::operator ==): Likewise.
25606 (vector_builder::finalize): Likewise.
25607 * int-vector-builder.h (int_vector_builder::int_vector_builder):
25608 Take the number of elements as a poly_uint64 rather than an
25610 * vec-perm-indices.h (vec_perm_indices::m_nelts_per_input): Change
25611 from unsigned int to poly_uint64.
25612 (vec_perm_indices::vec_perm_indices): Update prototype accordingly.
25613 (vec_perm_indices::new_vector): Likewise.
25614 (vec_perm_indices::length): Likewise.
25615 (vec_perm_indices::nelts_per_input): Likewise.
25616 (vec_perm_indices::input_nelts): Likewise.
25617 * vec-perm-indices.c (vec_perm_indices::new_vector): Take the
25618 number of elements per input as a poly_uint64 rather than an
25619 unsigned int. Use the original encoding for variable-length
25620 vectors, rather than clamping each individual element.
25621 For the second and subsequent elements in each pattern,
25622 clamp the step and base before clamping their sum.
25623 (vec_perm_indices::series_p): Handle polynomial element counts.
25624 (vec_perm_indices::all_in_range_p): Likewise.
25625 (vec_perm_indices_to_tree): Likewise.
25626 (vec_perm_indices_to_rtx): Likewise.
25627 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
25628 * tree-vector-builder.c (tree_vector_builder::new_unary_operation)
25629 (tree_vector_builder::new_binary_operation): Handle polynomial
25630 element counts. Return false if we need to know the number
25631 of elements at compile time.
25632 * fold-const.c (fold_vec_perm): Punt if the number of elements
25633 isn't known at compile time.
25635 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25637 * vec-perm-indices.h (vec_perm_builder): Change element type
25638 from HOST_WIDE_INT to poly_int64.
25639 (vec_perm_indices::element_type): Update accordingly.
25640 (vec_perm_indices::clamp): Handle polynomial element_types.
25641 * vec-perm-indices.c (vec_perm_indices::series_p): Likewise.
25642 (vec_perm_indices::all_in_range_p): Likewise.
25643 (tree_to_vec_perm_builder): Check for poly_int64 trees rather
25645 * vector-builder.h (vector_builder::stepped_sequence_p): Handle
25646 polynomial vec_perm_indices element types.
25647 * int-vector-builder.h (int_vector_builder::equal_p): Likewise.
25648 * fold-const.c (fold_vec_perm): Likewise.
25649 * optabs.c (shift_amt_for_vec_perm_mask): Likewise.
25650 * tree-vect-generic.c (lower_vec_perm): Likewise.
25651 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
25652 * config/aarch64/aarch64.c (aarch64_evpc_tbl): Cast d->perm
25653 element type to HOST_WIDE_INT.
25655 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25656 Alan Hayward <alan.hayward@arm.com>
25657 David Sherwood <david.sherwood@arm.com>
25659 * alias.c (addr_side_effect_eval): Take the size as a poly_int64
25660 rather than an int. Use plus_constant.
25661 (memrefs_conflict_p): Take the sizes as poly_int64s rather than ints.
25662 Take the offset "c" as a poly_int64 rather than a HOST_WIDE_INT.
25664 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25665 Alan Hayward <alan.hayward@arm.com>
25666 David Sherwood <david.sherwood@arm.com>
25668 * calls.c (emit_call_1, expand_call): Change struct_value_size from
25669 a HOST_WIDE_INT to a poly_int64.
25671 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25672 Alan Hayward <alan.hayward@arm.com>
25673 David Sherwood <david.sherwood@arm.com>
25675 * calls.c (load_register_parameters): Cope with polynomial
25676 mode sizes. Require a constant size for BLKmode parameters
25677 that aren't described by a PARALLEL. If BLOCK_REG_PADDING
25678 forces a parameter to be padded at the lsb end in order to
25679 fill a complete number of words, require the parameter size
25680 to be ordered wrt UNITS_PER_WORD.
25682 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25683 Alan Hayward <alan.hayward@arm.com>
25684 David Sherwood <david.sherwood@arm.com>
25686 * reload1.c (spill_stack_slot_width): Change element type
25687 from unsigned int to poly_uint64_pod.
25688 (alter_reg): Treat mode sizes as polynomial.
25690 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25691 Alan Hayward <alan.hayward@arm.com>
25692 David Sherwood <david.sherwood@arm.com>
25694 * reload.c (complex_word_subreg_p): New function.
25695 (reload_inner_reg_of_subreg, push_reload): Use it.
25697 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25698 Alan Hayward <alan.hayward@arm.com>
25699 David Sherwood <david.sherwood@arm.com>
25701 * lra-constraints.c (process_alt_operands): Reject matched
25702 operands whose sizes aren't ordered.
25703 (match_reload): Refer to this check here.
25705 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25706 Alan Hayward <alan.hayward@arm.com>
25707 David Sherwood <david.sherwood@arm.com>
25709 * builtins.c (expand_ifn_atomic_compare_exchange_into_call): Assert
25710 that the mode size is in the set {1, 2, 4, 8, 16}.
25712 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25713 Alan Hayward <alan.hayward@arm.com>
25714 David Sherwood <david.sherwood@arm.com>
25716 * var-tracking.c (adjust_mems): Treat mode sizes as polynomial.
25717 Use plus_constant instead of gen_rtx_PLUS.
25719 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25720 Alan Hayward <alan.hayward@arm.com>
25721 David Sherwood <david.sherwood@arm.com>
25723 * config/cr16/cr16-protos.h (cr16_push_rounding): Declare.
25724 * config/cr16/cr16.h (PUSH_ROUNDING): Move implementation to...
25725 * config/cr16/cr16.c (cr16_push_rounding): ...this new function.
25726 * config/h8300/h8300-protos.h (h8300_push_rounding): Declare.
25727 * config/h8300/h8300.h (PUSH_ROUNDING): Move implementation to...
25728 * config/h8300/h8300.c (h8300_push_rounding): ...this new function.
25729 * config/i386/i386-protos.h (ix86_push_rounding): Declare.
25730 * config/i386/i386.h (PUSH_ROUNDING): Move implementation to...
25731 * config/i386/i386.c (ix86_push_rounding): ...this new function.
25732 * config/m32c/m32c-protos.h (m32c_push_rounding): Take and return
25734 * config/m32c/m32c.c (m32c_push_rounding): Likewise.
25735 * config/m68k/m68k-protos.h (m68k_push_rounding): Declare.
25736 * config/m68k/m68k.h (PUSH_ROUNDING): Move implementation to...
25737 * config/m68k/m68k.c (m68k_push_rounding): ...this new function.
25738 * config/pdp11/pdp11-protos.h (pdp11_push_rounding): Declare.
25739 * config/pdp11/pdp11.h (PUSH_ROUNDING): Move implementation to...
25740 * config/pdp11/pdp11.c (pdp11_push_rounding): ...this new function.
25741 * config/stormy16/stormy16-protos.h (xstormy16_push_rounding): Declare.
25742 * config/stormy16/stormy16.h (PUSH_ROUNDING): Move implementation to...
25743 * config/stormy16/stormy16.c (xstormy16_push_rounding): ...this new
25745 * expr.c (emit_move_resolve_push): Treat the input and result
25746 of PUSH_ROUNDING as a poly_int64.
25747 (emit_move_complex_push, emit_single_push_insn_1): Likewise.
25748 (emit_push_insn): Likewise.
25749 * lra-eliminations.c (mark_not_eliminable): Likewise.
25750 * recog.c (push_operand): Likewise.
25751 * reload1.c (elimination_effects): Likewise.
25752 * rtlanal.c (nonzero_bits1): Likewise.
25753 * calls.c (store_one_arg): Likewise. Require the padding to be
25754 known at compile time.
25756 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25757 Alan Hayward <alan.hayward@arm.com>
25758 David Sherwood <david.sherwood@arm.com>
25760 * expr.c (emit_single_push_insn_1): Treat mode sizes as polynomial.
25761 Use plus_constant instead of gen_rtx_PLUS.
25763 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25764 Alan Hayward <alan.hayward@arm.com>
25765 David Sherwood <david.sherwood@arm.com>
25767 * auto-inc-dec.c (set_inc_state): Take the mode size as a poly_int64
25768 rather than an int.
25770 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25771 Alan Hayward <alan.hayward@arm.com>
25772 David Sherwood <david.sherwood@arm.com>
25774 * expr.c (expand_expr_real_1): Use tree_to_poly_uint64
25775 instead of int_size_in_bytes when handling VIEW_CONVERT_EXPRs
25776 via stack temporaries. Treat the mode size as polynomial too.
25778 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25779 Alan Hayward <alan.hayward@arm.com>
25780 David Sherwood <david.sherwood@arm.com>
25782 * expr.c (expand_expr_real_2): When handling conversions involving
25783 unions, apply tree_to_poly_uint64 to the TYPE_SIZE rather than
25784 multiplying int_size_in_bytes by BITS_PER_UNIT. Treat GET_MODE_BISIZE
25785 as a poly_uint64 too.
25787 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25788 Alan Hayward <alan.hayward@arm.com>
25789 David Sherwood <david.sherwood@arm.com>
25791 * rtlanal.c (subreg_get_info): Handle polynomial mode sizes.
25793 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25794 Alan Hayward <alan.hayward@arm.com>
25795 David Sherwood <david.sherwood@arm.com>
25797 * combine.c (can_change_dest_mode): Handle polynomial
25798 REGMODE_NATURAL_SIZE.
25799 * expmed.c (store_bit_field_1): Likewise.
25800 * expr.c (store_constructor): Likewise.
25801 * emit-rtl.c (validate_subreg): Operate on polynomial mode sizes
25802 and polynomial REGMODE_NATURAL_SIZE.
25803 (gen_lowpart_common): Likewise.
25804 * reginfo.c (record_subregs_of_mode): Likewise.
25805 * rtlanal.c (read_modify_subreg_p): Likewise.
25807 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25808 Alan Hayward <alan.hayward@arm.com>
25809 David Sherwood <david.sherwood@arm.com>
25811 * internal-fn.c (expand_vector_ubsan_overflow): Handle polynomial
25812 numbers of elements.
25814 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25815 Alan Hayward <alan.hayward@arm.com>
25816 David Sherwood <david.sherwood@arm.com>
25818 * match.pd: Cope with polynomial numbers of vector elements.
25820 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25821 Alan Hayward <alan.hayward@arm.com>
25822 David Sherwood <david.sherwood@arm.com>
25824 * fold-const.c (fold_indirect_ref_1): Handle polynomial offsets
25825 in a POINTER_PLUS_EXPR.
25827 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25828 Alan Hayward <alan.hayward@arm.com>
25829 David Sherwood <david.sherwood@arm.com>
25831 * omp-simd-clone.c (simd_clone_subparts): New function.
25832 (simd_clone_init_simd_arrays): Use it instead of TYPE_VECTOR_SUBPARTS.
25833 (ipa_simd_modify_function_body): Likewise.
25835 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25836 Alan Hayward <alan.hayward@arm.com>
25837 David Sherwood <david.sherwood@arm.com>
25839 * tree-vect-generic.c (nunits_for_known_piecewise_op): New function.
25840 (expand_vector_piecewise): Use it instead of TYPE_VECTOR_SUBPARTS.
25841 (expand_vector_addition, add_rshift, expand_vector_divmod): Likewise.
25842 (expand_vector_condition, vector_element): Likewise.
25843 (subparts_gt): New function.
25844 (get_compute_type): Use subparts_gt.
25845 (count_type_subparts): Delete.
25846 (expand_vector_operations_1): Use subparts_gt instead of
25847 count_type_subparts.
25849 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25850 Alan Hayward <alan.hayward@arm.com>
25851 David Sherwood <david.sherwood@arm.com>
25853 * tree-vect-data-refs.c (vect_no_alias_p): Replace with...
25854 (vect_compile_time_alias): ...this new function. Do the calculation
25855 on poly_ints rather than trees.
25856 (vect_prune_runtime_alias_test_list): Update call accordingly.
25858 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25859 Alan Hayward <alan.hayward@arm.com>
25860 David Sherwood <david.sherwood@arm.com>
25862 * tree-vect-slp.c (vect_build_slp_tree_1): Handle polynomial
25864 (vect_schedule_slp_instance): Likewise.
25866 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25867 Alan Hayward <alan.hayward@arm.com>
25868 David Sherwood <david.sherwood@arm.com>
25870 * tree-vect-slp.c (vect_get_and_check_slp_defs): Reject
25871 constant and extern definitions for variable-length vectors.
25872 (vect_get_constant_vectors): Note that the number of units
25873 is known to be constant.
25875 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25876 Alan Hayward <alan.hayward@arm.com>
25877 David Sherwood <david.sherwood@arm.com>
25879 * tree-vect-stmts.c (vectorizable_conversion): Treat the number
25880 of units as polynomial. Choose between WIDE and NARROW based
25883 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25884 Alan Hayward <alan.hayward@arm.com>
25885 David Sherwood <david.sherwood@arm.com>
25887 * tree-vect-stmts.c (simd_clone_subparts): New function.
25888 (vectorizable_simd_clone_call): Use it instead of TYPE_VECTOR_SUBPARTS.
25890 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25891 Alan Hayward <alan.hayward@arm.com>
25892 David Sherwood <david.sherwood@arm.com>
25894 * tree-vect-stmts.c (vectorizable_call): Treat the number of
25895 vectors as polynomial. Use build_index_vector for
25896 IFN_GOMP_SIMD_LANE.
25898 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25899 Alan Hayward <alan.hayward@arm.com>
25900 David Sherwood <david.sherwood@arm.com>
25902 * tree-vect-stmts.c (get_load_store_type): Treat the number of
25903 units as polynomial. Reject VMAT_ELEMENTWISE and VMAT_STRIDED_SLP
25904 for variable-length vectors.
25905 (vectorizable_mask_load_store): Treat the number of units as
25906 polynomial, asserting that it is constant if the condition has
25907 already been enforced.
25908 (vectorizable_store, vectorizable_load): Likewise.
25910 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25911 Alan Hayward <alan.hayward@arm.com>
25912 David Sherwood <david.sherwood@arm.com>
25914 * tree-vect-loop.c (vectorizable_live_operation): Treat the number
25915 of units as polynomial. Punt if we can't tell at compile time
25916 which vector contains the final result.
25918 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25919 Alan Hayward <alan.hayward@arm.com>
25920 David Sherwood <david.sherwood@arm.com>
25922 * tree-vect-loop.c (vectorizable_induction): Treat the number
25923 of units as polynomial. Punt on SLP inductions. Use an integer
25924 VEC_SERIES_EXPR for variable-length integer reductions. Use a
25925 cast of such a series for variable-length floating-point
25928 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25929 Alan Hayward <alan.hayward@arm.com>
25930 David Sherwood <david.sherwood@arm.com>
25932 * tree.h (build_index_vector): Declare.
25933 * tree.c (build_index_vector): New function.
25934 * tree-vect-loop.c (get_initial_defs_for_reduction): Treat the number
25935 of units as polynomial, forcibly converting it to a constant if
25936 vectorizable_reduction has already enforced the condition.
25937 (vect_create_epilog_for_reduction): Likewise. Use build_index_vector
25938 to create a {1,2,3,...} vector.
25939 (vectorizable_reduction): Treat the number of units as polynomial.
25940 Choose vectype_in based on the largest scalar element size rather
25941 than the smallest number of units. Enforce the restrictions
25944 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25945 Alan Hayward <alan.hayward@arm.com>
25946 David Sherwood <david.sherwood@arm.com>
25948 * tree-vect-data-refs.c (vector_alignment_reachable_p): Treat the
25949 number of units as polynomial.
25951 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25952 Alan Hayward <alan.hayward@arm.com>
25953 David Sherwood <david.sherwood@arm.com>
25955 * target.h (vector_sizes, auto_vector_sizes): New typedefs.
25956 * target.def (autovectorize_vector_sizes): Return the vector sizes
25957 by pointer, using vector_sizes rather than a bitmask.
25958 * targhooks.h (default_autovectorize_vector_sizes): Update accordingly.
25959 * targhooks.c (default_autovectorize_vector_sizes): Likewise.
25960 * config/aarch64/aarch64.c (aarch64_autovectorize_vector_sizes):
25962 * config/arc/arc.c (arc_autovectorize_vector_sizes): Likewise.
25963 * config/arm/arm.c (arm_autovectorize_vector_sizes): Likewise.
25964 * config/i386/i386.c (ix86_autovectorize_vector_sizes): Likewise.
25965 * config/mips/mips.c (mips_autovectorize_vector_sizes): Likewise.
25966 * omp-general.c (omp_max_vf): Likewise.
25967 * omp-low.c (omp_clause_aligned_alignment): Likewise.
25968 * optabs-query.c (can_vec_mask_load_store_p): Likewise.
25969 * tree-vect-loop.c (vect_analyze_loop): Likewise.
25970 * tree-vect-slp.c (vect_slp_bb): Likewise.
25971 * doc/tm.texi: Regenerate.
25972 * tree-vectorizer.h (current_vector_size): Change from an unsigned int
25974 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Take
25975 the vector size as a poly_uint64 rather than an unsigned int.
25976 (current_vector_size): Change from an unsigned int to a poly_uint64.
25977 (get_vectype_for_scalar_type): Update accordingly.
25978 * tree.h (build_truth_vector_type): Take the size and number of
25979 units as a poly_uint64 rather than an unsigned int.
25980 (build_vector_type): Add a temporary overload that takes
25981 the number of units as a poly_uint64 rather than an unsigned int.
25982 * tree.c (make_vector_type): Likewise.
25983 (build_truth_vector_type): Take the number of units as a poly_uint64
25984 rather than an unsigned int.
25986 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25987 Alan Hayward <alan.hayward@arm.com>
25988 David Sherwood <david.sherwood@arm.com>
25990 * target.def (get_mask_mode): Take the number of units and length
25991 as poly_uint64s rather than unsigned ints.
25992 * targhooks.h (default_get_mask_mode): Update accordingly.
25993 * targhooks.c (default_get_mask_mode): Likewise.
25994 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
25995 * doc/tm.texi: Regenerate.
25997 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25998 Alan Hayward <alan.hayward@arm.com>
25999 David Sherwood <david.sherwood@arm.com>
26001 * omp-general.h (omp_max_vf): Return a poly_uint64 instead of an int.
26002 * omp-general.c (omp_max_vf): Likewise.
26003 * omp-expand.c (omp_adjust_chunk_size): Update call to omp_max_vf.
26004 (expand_omp_simd): Handle polynomial safelen.
26005 * omp-low.c (omplow_simd_context): Add a default constructor.
26006 (omplow_simd_context::max_vf): Change from int to poly_uint64.
26007 (lower_rec_simd_input_clauses): Update accordingly.
26008 (lower_rec_input_clauses): Likewise.
26010 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
26011 Alan Hayward <alan.hayward@arm.com>
26012 David Sherwood <david.sherwood@arm.com>
26014 * tree-vectorizer.h (vect_nunits_for_cost): New function.
26015 * tree-vect-loop.c (vect_model_reduction_cost): Use it.
26016 * tree-vect-slp.c (vect_analyze_slp_cost_1): Likewise.
26017 (vect_analyze_slp_cost): Likewise.
26018 * tree-vect-stmts.c (vect_model_store_cost): Likewise.
26019 (vect_model_load_cost): Likewise.
26021 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
26022 Alan Hayward <alan.hayward@arm.com>
26023 David Sherwood <david.sherwood@arm.com>
26025 * tree-vect-slp.c (vect_record_max_nunits, vect_build_slp_tree_1)
26026 (vect_build_slp_tree_2, vect_build_slp_tree): Change max_nunits
26027 from an unsigned int * to a poly_uint64_pod *.
26028 (calculate_unrolling_factor): New function.
26029 (vect_analyze_slp_instance): Use it. Track polynomial max_nunits.
26031 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
26032 Alan Hayward <alan.hayward@arm.com>
26033 David Sherwood <david.sherwood@arm.com>
26035 * tree-vectorizer.h (_slp_instance::unrolling_factor): Change
26036 from an unsigned int to a poly_uint64.
26037 (_loop_vec_info::slp_unrolling_factor): Likewise.
26038 (_loop_vec_info::vectorization_factor): Change from an int
26040 (MAX_VECTORIZATION_FACTOR): Bump from 64 to INT_MAX.
26041 (vect_get_num_vectors): New function.
26042 (vect_update_max_nunits, vect_vf_for_cost): Likewise.
26043 (vect_get_num_copies): Use vect_get_num_vectors.
26044 (vect_analyze_data_ref_dependences): Change max_vf from an int *
26045 to an unsigned int *.
26046 (vect_analyze_data_refs): Change min_vf from an int * to a
26048 (vect_transform_slp_perm_load): Take the vf as a poly_uint64 rather
26049 than an unsigned HOST_WIDE_INT.
26050 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr)
26051 (vect_analyze_data_ref_dependence): Change max_vf from an int *
26052 to an unsigned int *.
26053 (vect_analyze_data_ref_dependences): Likewise.
26054 (vect_compute_data_ref_alignment): Handle polynomial vf.
26055 (vect_enhance_data_refs_alignment): Likewise.
26056 (vect_prune_runtime_alias_test_list): Likewise.
26057 (vect_shift_permute_load_chain): Likewise.
26058 (vect_supportable_dr_alignment): Likewise.
26059 (dependence_distance_ge_vf): Take the vectorization factor as a
26060 poly_uint64 rather than an unsigned HOST_WIDE_INT.
26061 (vect_analyze_data_refs): Change min_vf from an int * to a
26063 * tree-vect-loop-manip.c (vect_gen_scalar_loop_niters): Take
26064 vfm1 as a poly_uint64 rather than an int. Make the same change
26065 for the returned bound_scalar.
26066 (vect_gen_vector_loop_niters): Handle polynomial vf.
26067 (vect_do_peeling): Likewise. Update call to
26068 vect_gen_scalar_loop_niters and handle polynomial bound_scalars.
26069 (vect_gen_vector_loop_niters_mult_vf): Assert that the vf must
26071 * tree-vect-loop.c (vect_determine_vectorization_factor)
26072 (vect_update_vf_for_slp, vect_analyze_loop_2): Handle polynomial vf.
26073 (vect_get_known_peeling_cost): Likewise.
26074 (vect_estimate_min_profitable_iters, vectorizable_reduction): Likewise.
26075 (vect_worthwhile_without_simd_p, vectorizable_induction): Likewise.
26076 (vect_transform_loop): Likewise. Use the lowest possible VF when
26077 updating the upper bounds of the loop.
26078 (vect_min_worthwhile_factor): Make static. Return an unsigned int
26079 rather than an int.
26080 * tree-vect-slp.c (vect_attempt_slp_rearrange_stmts): Cope with
26081 polynomial unroll factors.
26082 (vect_analyze_slp_cost_1, vect_analyze_slp_instance): Likewise.
26083 (vect_make_slp_decision): Likewise.
26084 (vect_supported_load_permutation_p): Likewise, and polynomial
26086 (vect_analyze_slp_cost): Handle polynomial vf.
26087 (vect_slp_analyze_node_operations): Likewise.
26088 (vect_slp_analyze_bb_1): Likewise.
26089 (vect_transform_slp_perm_load): Take the vf as a poly_uint64 rather
26090 than an unsigned HOST_WIDE_INT.
26091 * tree-vect-stmts.c (vectorizable_simd_clone_call, vectorizable_store)
26092 (vectorizable_load): Handle polynomial vf.
26093 * tree-vectorizer.c (simduid_to_vf::vf): Change from an int to
26095 (adjust_simduid_builtins, shrink_simd_arrays): Update accordingly.
26097 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
26098 Alan Hayward <alan.hayward@arm.com>
26099 David Sherwood <david.sherwood@arm.com>
26101 * match.pd: Handle bit operations involving three constants
26102 and try to fold one pair.
26104 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
26106 * tree-vect-loop-manip.c: Include gimple-fold.h.
26107 (slpeel_make_loop_iterate_ntimes): Add step, final_iv and
26108 niters_maybe_zero parameters. Handle other cases besides a step of 1.
26109 (vect_gen_vector_loop_niters): Add a step_vector_ptr parameter.
26110 Add a path that uses a step of VF instead of 1, but disable it
26112 (vect_do_peeling): Add step_vector, niters_vector_mult_vf_var
26113 and niters_no_overflow parameters. Update calls to
26114 slpeel_make_loop_iterate_ntimes and vect_gen_vector_loop_niters.
26115 Create a new SSA name if the latter choses to use a ste other
26116 than zero, and return it via niters_vector_mult_vf_var.
26117 * tree-vect-loop.c (vect_transform_loop): Update calls to
26118 vect_do_peeling, vect_gen_vector_loop_niters and
26119 slpeel_make_loop_iterate_ntimes.
26120 * tree-vectorizer.h (slpeel_make_loop_iterate_ntimes, vect_do_peeling)
26121 (vect_gen_vector_loop_niters): Update declarations after above changes.
26123 2018-01-02 Michael Meissner <meissner@linux.vnet.ibm.com>
26125 * config/rs6000/rs6000.md (floor<mode>2): Add support for IEEE
26126 128-bit round to integer instructions.
26127 (ceil<mode>2): Likewise.
26128 (btrunc<mode>2): Likewise.
26129 (round<mode>2): Likewise.
26131 2018-01-02 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
26133 * config/rs6000/rs6000-string.c (expand_block_move): Allow the use of
26134 unaligned VSX load/store on P8/P9.
26135 (expand_block_clear): Allow the use of unaligned VSX
26136 load/store on P8/P9.
26138 2018-01-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26140 * config/rs6000/rs6000-p8swap.c (swap_feeds_both_load_and_store):
26142 (rs6000_analyze_swaps): Mark a web unoptimizable if it contains a
26143 swap associated with both a load and a store.
26145 2018-01-02 Andrew Waterman <andrew@sifive.com>
26147 * config/riscv/linux.h (ICACHE_FLUSH_FUNC): New.
26148 * config/riscv/riscv.md (clear_cache): Use it.
26150 2018-01-02 Artyom Skrobov <tyomitch@gmail.com>
26152 * web.c: Remove out-of-date comment.
26154 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
26156 * expr.c (fixup_args_size_notes): Check that any existing
26157 REG_ARGS_SIZE notes are correct, and don't try to re-add them.
26158 (emit_single_push_insn_1): Move stack_pointer_delta adjustment to...
26159 (emit_single_push_insn): ...here.
26161 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
26163 * rtl.h (CONST_VECTOR_ELT): Redefine to const_vector_elt.
26164 (const_vector_encoded_nelts): New function.
26165 (CONST_VECTOR_NUNITS): Redefine to use GET_MODE_NUNITS.
26166 (const_vector_int_elt, const_vector_elt): Declare.
26167 * emit-rtl.c (const_vector_int_elt_1): New function.
26168 (const_vector_elt): Likewise.
26169 * simplify-rtx.c (simplify_immed_subreg): Avoid taking the address
26170 of CONST_VECTOR_ELT.
26172 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
26174 * expr.c: Include rtx-vector-builder.h.
26175 (const_vector_mask_from_tree): Use rtx_vector_builder and operate
26176 directly on the tree encoding.
26177 (const_vector_from_tree): Likewise.
26178 * optabs.c: Include rtx-vector-builder.h.
26179 (expand_vec_perm_var): Use rtx_vector_builder and create a repeating
26180 sequence of "u" values.
26181 * vec-perm-indices.c: Include rtx-vector-builder.h.
26182 (vec_perm_indices_to_rtx): Use rtx_vector_builder and operate
26183 directly on the vec_perm_indices encoding.
26185 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
26187 * doc/rtl.texi (const_vector): Describe new encoding scheme.
26188 * Makefile.in (OBJS): Add rtx-vector-builder.o.
26189 * rtx-vector-builder.h: New file.
26190 * rtx-vector-builder.c: Likewise.
26191 * rtl.h (rtx_def::u2): Add a const_vector field.
26192 (CONST_VECTOR_NPATTERNS): New macro.
26193 (CONST_VECTOR_NELTS_PER_PATTERN): Likewise.
26194 (CONST_VECTOR_DUPLICATE_P): Likewise.
26195 (CONST_VECTOR_STEPPED_P): Likewise.
26196 (CONST_VECTOR_ENCODED_ELT): Likewise.
26197 (const_vec_duplicate_p): Check for a duplicated vector encoding.
26198 (unwrap_const_vec_duplicate): Likewise.
26199 (const_vec_series_p): Check for a non-duplicated vector encoding.
26200 Say that the function only returns true for integer vectors.
26201 * emit-rtl.c: Include rtx-vector-builder.h.
26202 (gen_const_vec_duplicate_1): Delete.
26203 (gen_const_vector): Call gen_const_vec_duplicate instead of
26204 gen_const_vec_duplicate_1.
26205 (const_vec_series_p_1): Operate directly on the CONST_VECTOR encoding.
26206 (gen_const_vec_duplicate): Use rtx_vector_builder.
26207 (gen_const_vec_series): Likewise.
26208 (gen_rtx_CONST_VECTOR): Likewise.
26209 * config/powerpcspe/powerpcspe.c: Include rtx-vector-builder.h.
26210 (swap_const_vector_halves): Take an rtx pointer rather than rtx.
26211 Build a new vector rather than modifying a CONST_VECTOR in-place.
26212 (handle_special_swappables): Update call accordingly.
26213 * config/rs6000/rs6000-p8swap.c: Include rtx-vector-builder.h.
26214 (swap_const_vector_halves): Take an rtx pointer rather than rtx.
26215 Build a new vector rather than modifying a CONST_VECTOR in-place.
26216 (handle_special_swappables): Update call accordingly.
26218 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
26220 * simplify-rtx.c (simplify_const_binary_operation): Use
26221 CONST_VECTOR_ELT instead of XVECEXP.
26223 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
26225 * tree-cfg.c (verify_gimple_assign_ternary): Allow the size of
26226 the selector elements to be different from the data elements
26227 if the selector is a VECTOR_CST.
26228 * tree-vect-stmts.c (vect_gen_perm_mask_any): Use a vector of
26229 ssizetype for the selector.
26231 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
26233 * optabs.c (shift_amt_for_vec_perm_mask): Try using series_p
26234 before testing each element individually.
26235 * tree-vect-generic.c (lower_vec_perm): Likewise.
26237 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
26239 * selftest.h (selftest::vec_perm_indices_c_tests): Declare.
26240 * selftest-run-tests.c (selftest::run_tests): Call it.
26241 * vector-builder.h (vector_builder::operator ==): New function.
26242 (vector_builder::operator !=): Likewise.
26243 * vec-perm-indices.h (vec_perm_indices::series_p): Declare.
26244 (vec_perm_indices::all_from_input_p): New function.
26245 * vec-perm-indices.c (vec_perm_indices::series_p): Likewise.
26246 (test_vec_perm_12, selftest::vec_perm_indices_c_tests): Likewise.
26247 * fold-const.c (fold_ternary_loc): Use tree_to_vec_perm_builder
26248 instead of reading the VECTOR_CST directly. Detect whether both
26249 vector inputs are the same before constructing the vec_perm_indices,
26250 and update the number of inputs argument accordingly. Use the
26251 utility functions added above. Only construct sel2 if we need to.
26253 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
26255 * optabs.c (expand_vec_perm_var): Use an explicit encoding for
26256 the broadcast of the low byte.
26257 (expand_mult_highpart): Use an explicit encoding for the permutes.
26258 * optabs-query.c (can_mult_highpart_p): Likewise.
26259 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
26260 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
26261 (vectorizable_bswap): Likewise.
26262 * tree-vect-data-refs.c (vect_grouped_store_supported): Use an
26263 explicit encoding for the power-of-2 permutes.
26264 (vect_permute_store_chain): Likewise.
26265 (vect_grouped_load_supported): Likewise.
26266 (vect_permute_load_chain): Likewise.
26268 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
26270 * vec-perm-indices.h (vec_perm_indices_to_tree): Declare.
26271 * vec-perm-indices.c (vec_perm_indices_to_tree): New function.
26272 * tree-ssa-forwprop.c (simplify_vector_constructor): Use it.
26273 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
26274 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
26275 (vect_gen_perm_mask_any): Likewise.
26277 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
26279 * int-vector-builder.h: New file.
26280 * vec-perm-indices.h: Include int-vector-builder.h.
26281 (vec_perm_indices): Redefine as an int_vector_builder.
26282 (auto_vec_perm_indices): Delete.
26283 (vec_perm_builder): Redefine as a stand-alone class.
26284 (vec_perm_indices::vec_perm_indices): New function.
26285 (vec_perm_indices::clamp): Likewise.
26286 * vec-perm-indices.c: Include fold-const.h and tree-vector-builder.h.
26287 (vec_perm_indices::new_vector): New function.
26288 (vec_perm_indices::new_expanded_vector): Update for new
26289 vec_perm_indices class.
26290 (vec_perm_indices::rotate_inputs): New function.
26291 (vec_perm_indices::all_in_range_p): Operate directly on the
26292 encoded form, without computing elided elements.
26293 (tree_to_vec_perm_builder): Operate directly on the VECTOR_CST
26294 encoding. Update for new vec_perm_indices class.
26295 * optabs.c (expand_vec_perm_const): Create a vec_perm_indices for
26296 the given vec_perm_builder.
26297 (expand_vec_perm_var): Update vec_perm_builder constructor.
26298 (expand_mult_highpart): Use vec_perm_builder instead of
26299 auto_vec_perm_indices.
26300 * optabs-query.c (can_mult_highpart_p): Use vec_perm_builder and
26301 vec_perm_indices instead of auto_vec_perm_indices. Use a single
26302 or double series encoding as appropriate.
26303 * fold-const.c (fold_ternary_loc): Use vec_perm_builder and
26304 vec_perm_indices instead of auto_vec_perm_indices.
26305 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
26306 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
26307 (vect_permute_store_chain): Likewise.
26308 (vect_grouped_load_supported): Likewise.
26309 (vect_permute_load_chain): Likewise.
26310 (vect_shift_permute_load_chain): Likewise.
26311 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
26312 (vect_transform_slp_perm_load): Likewise.
26313 (vect_schedule_slp_instance): Likewise.
26314 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
26315 (vectorizable_mask_load_store): Likewise.
26316 (vectorizable_bswap): Likewise.
26317 (vectorizable_store): Likewise.
26318 (vectorizable_load): Likewise.
26319 * tree-vect-generic.c (lower_vec_perm): Use vec_perm_builder and
26320 vec_perm_indices instead of auto_vec_perm_indices. Use
26321 tree_to_vec_perm_builder to read the vector from a tree.
26322 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Take a
26323 vec_perm_builder instead of a vec_perm_indices.
26324 (have_whole_vector_shift): Use vec_perm_builder and
26325 vec_perm_indices instead of auto_vec_perm_indices. Leave the
26326 truncation to calc_vec_perm_mask_for_shift.
26327 (vect_create_epilog_for_reduction): Likewise.
26328 * config/aarch64/aarch64.c (expand_vec_perm_d::perm): Change
26329 from auto_vec_perm_indices to vec_perm_indices.
26330 (aarch64_expand_vec_perm_const_1): Use rotate_inputs on d.perm
26331 instead of changing individual elements.
26332 (aarch64_vectorize_vec_perm_const): Use new_vector to install
26333 the vector in d.perm.
26334 * config/arm/arm.c (expand_vec_perm_d::perm): Change
26335 from auto_vec_perm_indices to vec_perm_indices.
26336 (arm_expand_vec_perm_const_1): Use rotate_inputs on d.perm
26337 instead of changing individual elements.
26338 (arm_vectorize_vec_perm_const): Use new_vector to install
26339 the vector in d.perm.
26340 * config/powerpcspe/powerpcspe.c (rs6000_expand_extract_even):
26341 Update vec_perm_builder constructor.
26342 (rs6000_expand_interleave): Likewise.
26343 * config/rs6000/rs6000.c (rs6000_expand_extract_even): Likewise.
26344 (rs6000_expand_interleave): Likewise.
26346 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
26348 * optabs-query.c (can_vec_perm_var_p): Check whether lowering
26349 to qimode could truncate the indices.
26350 * optabs.c (expand_vec_perm_var): Likewise.
26352 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
26354 * Makefile.in (OBJS): Add vec-perm-indices.o.
26355 * vec-perm-indices.h: New file.
26356 * vec-perm-indices.c: Likewise.
26357 * target.h (vec_perm_indices): Replace with a forward class
26359 (auto_vec_perm_indices): Move to vec-perm-indices.h.
26360 * optabs.h: Include vec-perm-indices.h.
26361 (expand_vec_perm): Delete.
26362 (selector_fits_mode_p, expand_vec_perm_var): Declare.
26363 (expand_vec_perm_const): Declare.
26364 * target.def (vec_perm_const_ok): Replace with...
26365 (vec_perm_const): ...this new hook.
26366 * doc/tm.texi.in (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Replace with...
26367 (TARGET_VECTORIZE_VEC_PERM_CONST): ...this new hook.
26368 * doc/tm.texi: Regenerate.
26369 * optabs.def (vec_perm_const): Delete.
26370 * doc/md.texi (vec_perm_const): Likewise.
26371 (vec_perm): Refer to TARGET_VECTORIZE_VEC_PERM_CONST.
26372 * expr.c (expand_expr_real_2): Use expand_vec_perm_const rather than
26373 expand_vec_perm for constant permutation vectors. Assert that
26374 the mode of variable permutation vectors is the integer equivalent
26375 of the mode that is being permuted.
26376 * optabs-query.h (selector_fits_mode_p): Declare.
26377 * optabs-query.c: Include vec-perm-indices.h.
26378 (selector_fits_mode_p): New function.
26379 (can_vec_perm_const_p): Check whether targetm.vectorize.vec_perm_const
26380 is defined, instead of checking whether the vec_perm_const_optab
26381 exists. Use targetm.vectorize.vec_perm_const instead of
26382 targetm.vectorize.vec_perm_const_ok. Check whether the indices
26383 fit in the vector mode before using a variable permute.
26384 * optabs.c (shift_amt_for_vec_perm_mask): Take a mode and a
26385 vec_perm_indices instead of an rtx.
26386 (expand_vec_perm): Replace with...
26387 (expand_vec_perm_const): ...this new function. Take the selector
26388 as a vec_perm_indices rather than an rtx. Also take the mode of
26389 the selector. Update call to shift_amt_for_vec_perm_mask.
26390 Use targetm.vectorize.vec_perm_const instead of vec_perm_const_optab.
26391 Use vec_perm_indices::new_expanded_vector to expand the original
26392 selector into bytes. Check whether the indices fit in the vector
26393 mode before using a variable permute.
26394 (expand_vec_perm_var): Make global.
26395 (expand_mult_highpart): Use expand_vec_perm_const.
26396 * fold-const.c: Includes vec-perm-indices.h.
26397 * tree-ssa-forwprop.c: Likewise.
26398 * tree-vect-data-refs.c: Likewise.
26399 * tree-vect-generic.c: Likewise.
26400 * tree-vect-loop.c: Likewise.
26401 * tree-vect-slp.c: Likewise.
26402 * tree-vect-stmts.c: Likewise.
26403 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm_const):
26405 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>): Delete.
26406 * config/aarch64/aarch64.c (aarch64_expand_vec_perm_const)
26407 (aarch64_vectorize_vec_perm_const_ok): Fuse into...
26408 (aarch64_vectorize_vec_perm_const): ...this new function.
26409 (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
26410 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
26411 * config/arm/arm-protos.h (arm_expand_vec_perm_const): Delete.
26412 * config/arm/vec-common.md (vec_perm_const<mode>): Delete.
26413 * config/arm/arm.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
26414 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
26415 (arm_expand_vec_perm_const, arm_vectorize_vec_perm_const_ok): Merge
26417 (arm_vectorize_vec_perm_const): ...this new function. Explicitly
26418 check for NEON modes.
26419 * config/i386/i386-protos.h (ix86_expand_vec_perm_const): Delete.
26420 * config/i386/sse.md (VEC_PERM_CONST, vec_perm_const<mode>): Delete.
26421 * config/i386/i386.c (ix86_expand_vec_perm_const_1): Update comment.
26422 (ix86_expand_vec_perm_const, ix86_vectorize_vec_perm_const_ok): Merge
26424 (ix86_vectorize_vec_perm_const): ...this new function. Incorporate
26425 the old VEC_PERM_CONST conditions.
26426 * config/ia64/ia64-protos.h (ia64_expand_vec_perm_const): Delete.
26427 * config/ia64/vect.md (vec_perm_const<mode>): Delete.
26428 * config/ia64/ia64.c (ia64_expand_vec_perm_const)
26429 (ia64_vectorize_vec_perm_const_ok): Merge into...
26430 (ia64_vectorize_vec_perm_const): ...this new function.
26431 * config/mips/loongson.md (vec_perm_const<mode>): Delete.
26432 * config/mips/mips-msa.md (vec_perm_const<mode>): Delete.
26433 * config/mips/mips-ps-3d.md (vec_perm_constv2sf): Delete.
26434 * config/mips/mips-protos.h (mips_expand_vec_perm_const): Delete.
26435 * config/mips/mips.c (mips_expand_vec_perm_const)
26436 (mips_vectorize_vec_perm_const_ok): Merge into...
26437 (mips_vectorize_vec_perm_const): ...this new function.
26438 * config/powerpcspe/altivec.md (vec_perm_constv16qi): Delete.
26439 * config/powerpcspe/paired.md (vec_perm_constv2sf): Delete.
26440 * config/powerpcspe/spe.md (vec_perm_constv2si): Delete.
26441 * config/powerpcspe/vsx.md (vec_perm_const<mode>): Delete.
26442 * config/powerpcspe/powerpcspe-protos.h (altivec_expand_vec_perm_const)
26443 (rs6000_expand_vec_perm_const): Delete.
26444 * config/powerpcspe/powerpcspe.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK):
26446 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
26447 (altivec_expand_vec_perm_const_le): Take each operand individually.
26448 Operate on constant selectors rather than rtxes.
26449 (altivec_expand_vec_perm_const): Likewise. Update call to
26450 altivec_expand_vec_perm_const_le.
26451 (rs6000_expand_vec_perm_const): Delete.
26452 (rs6000_vectorize_vec_perm_const_ok): Delete.
26453 (rs6000_vectorize_vec_perm_const): New function.
26454 (rs6000_do_expand_vec_perm): Take a vec_perm_builder instead of
26455 an element count and rtx array.
26456 (rs6000_expand_extract_even): Update call accordingly.
26457 (rs6000_expand_interleave): Likewise.
26458 * config/rs6000/altivec.md (vec_perm_constv16qi): Delete.
26459 * config/rs6000/paired.md (vec_perm_constv2sf): Delete.
26460 * config/rs6000/vsx.md (vec_perm_const<mode>): Delete.
26461 * config/rs6000/rs6000-protos.h (altivec_expand_vec_perm_const)
26462 (rs6000_expand_vec_perm_const): Delete.
26463 * config/rs6000/rs6000.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
26464 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
26465 (altivec_expand_vec_perm_const_le): Take each operand individually.
26466 Operate on constant selectors rather than rtxes.
26467 (altivec_expand_vec_perm_const): Likewise. Update call to
26468 altivec_expand_vec_perm_const_le.
26469 (rs6000_expand_vec_perm_const): Delete.
26470 (rs6000_vectorize_vec_perm_const_ok): Delete.
26471 (rs6000_vectorize_vec_perm_const): New function. Remove stray
26472 reference to the SPE evmerge intructions.
26473 (rs6000_do_expand_vec_perm): Take a vec_perm_builder instead of
26474 an element count and rtx array.
26475 (rs6000_expand_extract_even): Update call accordingly.
26476 (rs6000_expand_interleave): Likewise.
26477 * config/sparc/sparc.md (vec_perm_constv8qi): Delete in favor of...
26478 * config/sparc/sparc.c (sparc_vectorize_vec_perm_const): ...this
26480 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
26482 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
26484 * optabs.c (expand_vec_perm_1): Assert that SEL has an integer
26485 vector mode and that that mode matches the mode of the data
26487 (expand_vec_perm): Split handling of non-CONST_VECTOR selectors
26488 out into expand_vec_perm_var. Do all CONST_VECTOR handling here,
26489 directly using expand_vec_perm_1 when forcing selectors into
26491 (expand_vec_perm_var): New function, split out from expand_vec_perm.
26493 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
26495 * optabs-query.h (can_vec_perm_p): Delete.
26496 (can_vec_perm_var_p, can_vec_perm_const_p): Declare.
26497 * optabs-query.c (can_vec_perm_p): Split into...
26498 (can_vec_perm_var_p, can_vec_perm_const_p): ...these two functions.
26499 (can_mult_highpart_p): Use can_vec_perm_const_p to test whether a
26500 particular selector is valid.
26501 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
26502 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
26503 (vect_grouped_load_supported): Likewise.
26504 (vect_shift_permute_load_chain): Likewise.
26505 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
26506 (vect_transform_slp_perm_load): Likewise.
26507 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
26508 (vectorizable_bswap): Likewise.
26509 (vect_gen_perm_mask_checked): Likewise.
26510 * fold-const.c (fold_ternary_loc): Likewise. Don't take
26511 implementations of variable permutation vectors into account
26512 when deciding which selector to use.
26513 * tree-vect-loop.c (have_whole_vector_shift): Don't check whether
26514 vec_perm_const_optab is supported; instead use can_vec_perm_const_p
26515 with a false third argument.
26516 * tree-vect-generic.c (lower_vec_perm): Use can_vec_perm_const_p
26517 to test whether the constant selector is valid and can_vec_perm_var_p
26518 to test whether a variable selector is valid.
26520 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
26522 * optabs-query.h (can_vec_perm_p): Take a const vec_perm_indices *.
26523 * optabs-query.c (can_vec_perm_p): Likewise.
26524 * fold-const.c (fold_vec_perm): Take a const vec_perm_indices &
26525 instead of vec_perm_indices.
26526 * tree-vectorizer.h (vect_gen_perm_mask_any): Likewise,
26527 (vect_gen_perm_mask_checked): Likewise,
26528 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise,
26529 (vect_gen_perm_mask_checked): Likewise,
26531 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
26533 * optabs-query.h (qimode_for_vec_perm): Declare.
26534 * optabs-query.c (can_vec_perm_p): Split out qimode search to...
26535 (qimode_for_vec_perm): ...this new function.
26536 * optabs.c (expand_vec_perm): Use qimode_for_vec_perm.
26538 2018-01-02 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
26540 * rtlanal.c (canonicalize_condition): Return 0 if final rtx
26541 does not have a conditional at the top.
26543 2018-01-02 Richard Biener <rguenther@suse.de>
26545 * ipa-inline.c (big_speedup_p): Fix expression.
26547 2018-01-02 Jan Hubicka <hubicka@ucw.cz>
26550 * config/i386/x86-tune-costs.h: Increase cost of integer load costs
26553 2018-01-02 Jan Hubicka <hubicka@ucw.cz>
26557 * x86-tune-costs.h (generic_cost): Reduce cost of FDIV 20->17,
26558 cost of sqrt 20->14, DIVSS 18->13, DIVSD 32->17, SQRtSS 30->14
26559 and SQRTsD 58->18, cond_not_taken_branch_cost. 2->1. Increase
26560 cond_taken_branch_cost 3->4.
26562 2018-01-01 Jakub Jelinek <jakub@redhat.com>
26564 PR tree-optimization/83581
26565 * tree-loop-distribution.c (pass_loop_distribution::execute): Return
26566 TODO_cleanup_cfg if any changes have been made.
26568 PR middle-end/83608
26569 * expr.c (store_expr_with_bounds): Use simplify_gen_subreg instead of
26570 convert_modes if target mode has the right side, but different mode
26573 PR middle-end/83609
26574 * expr.c (expand_assignment): Fix up a typo in simplify_gen_subreg
26575 last argument when extracting from CONCAT. If either from_real or
26576 from_imag is NULL, use expansion through memory. If result is not
26577 a CONCAT and simplify_gen_subreg fails, try to simplify_gen_subreg
26578 the parts directly to inner mode, if even that fails, use expansion
26581 PR middle-end/83623
26582 * expmed.c (expand_shift_1): For 2-byte rotates by BITS_PER_UNIT,
26583 check for bswap in mode rather than HImode and use that in expand_unop
26586 Copyright (C) 2018 Free Software Foundation, Inc.
26588 Copying and distribution of this file, with or without modification,
26589 are permitted in any medium without royalty provided the copyright
26590 notice and this notice are preserved.