1 2018-09-26 Eric Botcazou <ebotcazou@adacore.com>
3 * config/arm/arm.c (arm_reorg): Skip Thumb reorg pass for thunks.
4 (arm32_output_mi_thunk): Deal with long calls.
6 2018-09-26 Richard Biener <rguenther@suse.de>
10 * tree-inline.c (expand_call_inline): When the location
11 of the call is UNKNOWN_LOCATION use DECL_SOURCE_LOCATION
12 or BUILTINS_LOCATION for the BLOCK_SOURCE_LOCATION of
13 the inserted BLOCK to make inlined_function_outer_scope_p
15 * dwarf2out.c (add_call_src_coords_attributes): Do not add
16 coords for reserved locations.
18 2018-09-25 Segher Boessenkool <segher@kernel.crashing.org>
20 * config/rs6000/rs6000.md (*movcc_internal1): Use set_attr_alternative.
21 (*call_indirect_nonlocal_sysv<mode>): Ditto.
22 (*call_value_indirect_nonlocal_sysv<mode>): Ditto.
23 (*sibcall_nonlocal_sysv<mode>): Ditto.
24 (*sibcall_value_nonlocal_sysv<mode>): Ditto.
26 (<bd>tf_<mode>): Ditto.
28 2018-09-25 Segher Boessenkool <segher@kernel.crashing.org>
30 * config/rs6000/altivec.md (*altivec_mov<mode>): Write the output
31 control string as a list of templates instead of as C code.
32 (*altivec_movti): Ditto.
33 * config/rs6000/darwin.md (movdf_low_di): Ditto.
35 2018-09-25 Jim Wilson <jimw@sifive.com>
37 * config/riscv/riscv.c (riscv_split_symbol): Mark auipc label as weak
38 when target symbol is weak.
40 2018-09-25 Bernd Edlinger <bernd.edlinger@hotmail.de>
43 * builtins.c (unterminated_array): Simplify.
44 * expr.c (string_constant): Handle SSA_NAME. Add more exceptions
45 where pointer arithmetic is safe.
47 2018-09-25 Segher Boessenkool <segher@kernel.crashing.org>
50 * config/rs6000/altivec.md (altivec_vspltb): Use
51 const_0_to_15_operand instead of u5bit_cint_operand.
52 (*altivec_vspltb_internal): Ditto.
53 (altivec_vspltb_direct): Ditto.
54 (altivec_vsplth): Use const_0_to_7_operand instead of
56 (*altivec_vsplth_internal): Ditto.
57 (altivec_vsplth_direct): Ditto.
58 (altivec_vspltw): Use const_0_to_3_operand instead of
60 (*altivec_vspltw_internal): Ditto.
61 (altivec_vspltw_direct): Ditto.
62 (altivec_vspltsf): Ditto.
63 (*altivec_vspltsf_internal): Ditto.
64 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Handle the
65 various splats with the proper size immediate. Reorder the various
66 cases by ascending size of immediate, and put all such together.
68 2018-09-25 Richard Biener <rguenther@suse.de>
71 * dwarf2out.c (add_AT_external_die_ref): Remove now redundant
73 (maybe_create_die_with_external_ref): Do not create
74 DW_TAG_imported_unit here.
75 (add_abstract_origin_attribute): Handle external BLOCK refs.
76 (dwarf2out_abstract_function): Simplify LTO case.
77 (dwarf2out_early_finish): Create DW_TAG_imported_unit explicitely
78 rather than using maybe_create_die_with_external_ref.
80 2018-09-25 Uros Bizjak <ubizjak@gmail.com>
82 * config/i386/i386.md (frndintxf2_mask_pm): Remove.
83 (frndintxf2_mask_pm_i387): Ditto.
84 (nearbyintxf2): Rewrite expander pattern to match rintxf2.
85 Enable for !flag_trapping_math.
86 (nearbyint<mode>2): Enable x87 modes for !flag_trapping_math.
87 Enable SSE modes for TARGET_SSE4_1 and expand them with round insn.
88 Change operand 1 predicate to nonimmediate_operand.
89 (attr "i387_cw"): Remove mask_pm.
90 * config/i386/i386.h (enum ix86_stack_slot): Remove SLOT_CW_MASK_PM.
91 (enum ix86_entity): Remove I387_MASK_PM.
92 * config/i386/i386.c (ix86_i387_mode_needed): Do not
94 (ix86_mode_needed): Ditto.
95 (ix86_mode_after): Ditto.
96 (ix86_mode_entry): Ditto.
97 (ix86_mode_exit): Ditto.
98 (emit_i387_cw_initialization): Do not handle I387_CW_MASK_PM.
100 2018-09-25 Jakub Jelinek <jakub@redhat.com>
102 * vr-values.c (vr_values::vr_values): Initialize to_remove_edges and
103 to_update_switch_stmts to vNULL instead of calling create on them
106 2018-09-25 Richard Biener <rguenther@suse.de>
108 PR tree-optimization/87402
109 * tree-ssa-sccvn.c (SSA_VISITED): Remove unused function.
110 (visit_phi): Re-instantiate handling of supposed to be VARYING
111 but non-VARYING backedge value.
113 2018-09-25 Richard Biener <rguenther@suse.de>
116 * dwarf2out.c (struct sym_off_pair): New.
117 (external_die_map): New global.
118 (lookup_decl_die): When in LTO create DIEs lazily from the
120 (lookup_block_die): New function, create DIEs lazily in LTO.
121 (equate_block_to_die): New function.
122 (dwarf2out_die_ref_for_decl): During WPA get the association
123 from the external DIE map.
124 (dwarf2out_register_external_die): Record mapping into the
126 (maybe_create_die_with_external_ref): New function split out from
127 DIE generation part of old dwarf2out_register_external_die.
128 (add_abstract_origin_attribute): Do not return the DIE. When
129 in LTO reference externals directly.
130 (dwarf2out_abstract_function): When in LTO ignore calls for
131 decls with external DIEs (already present abstract instances).
132 (gen_call_site_die): Adjust.
133 (add_high_low_attributes): Likewise.
134 (gen_lexical_block_die): Likewise.
135 (gen_inlined_subroutine_die): Likewie.
136 (gen_block_die): Likewise.
137 (dwarf2out_inline_entry): Likewise.
138 (dwarf2out_early_finish): In LTRANS phase create DW_TAG_imported_unit
141 2018-09-25 Martin Liska <mliska@suse.cz>
143 * ipa-fnsummary.c (estimate_node_size_and_time): Scale by two
144 integers and not by a float value.
146 2018-09-25 Martin Liska <mliska@suse.cz>
149 * dbgcnt.c (dbg_cnt_process_single_pair): Return false
151 * dumpfile.c (dump_enable_all): Remove extra parenthesis.
152 * gcov-tool.c: Declare the function with ATTRIBUTE_NORETURN.
153 * godump.c (go_format_type): Remove extra parenthesis.
155 2018-09-25 Martin Liska <mliska@suse.cz>
157 * alias.c (set_dest_equal_p): Remove unused function.
158 * config/i386/i386.c (def_builtin_pure2): Likewise.
159 * diagnostic-show-locus.c (class layout): Remove
161 (layout::layout): Likewise here.
162 * dump-context.h (class temp_dump_context): Likewise.
163 * dwarf2out.c (add_AT_fde_ref): Remove unused function.
164 (add_AT_loclistsptr): Likewise.
165 (add_AT_offset): Likewise.
166 (get_AT_hi_pc): Likewise.
167 (is_comdat_die): Likewise.
168 (type_is_enum): Likewise.
170 (add_AT_vms_delta): Likewise.
171 (is_class_die): Likewise.
172 * edit-context.c (class line_event): Remove unused field.
173 * graphite-sese-to-poly.c (tree_int_to_gmp): Remove
175 * ipa-cp.c (ipa_get_vr_lat): Likewise.
176 * lra-constraints.c (ok_for_index_p_nonstrict): Likewise.
177 (ok_for_base_p_nonstrict): Likewise.
178 * tree-chrec.c (is_not_constant_evolution): Likewise.
179 (chrec_fold_poly_cst): Likewise.
180 * tree-if-conv.c (has_pred_critical_p): Likewise.
181 * tree-ssa-coalesce.c (print_exprs): Likewise.
182 * tree-ssa-pre.c (bitmap_set_contains_expr): Likewise.
183 * tree-ssa-uninit.c (is_and_or_or_p): Likewise.
184 * tree-vrp.c (value_ranges_intersect_p): Likewise.
185 (value_range_nonnegative_p): Likewise.
187 2018-09-25 Martin Liska <mliska@suse.cz>
189 * config/powerpcspe/powerpcspe.c (rs6000_output_function_epilogue):
190 Do not handle "GNU Pascal".
191 * config/rs6000/rs6000.c (rs6000_output_function_epilogue):
193 * config/sparc/sparc.c (sparc_pass_by_reference): Remove Pascal
194 from documentation. Likewise.
195 * dbxout.c (dbxout_range_type): Likewise.
196 * doc/cpp.texi: Likewise.
197 * doc/extend.texi: Likewise.
198 * doc/frontends.texi: Likewise.
199 * doc/invoke.texi: Remove Pascal entry.
200 * tree.def (CLEANUP_POINT_EXPR): Likewise.
201 * doc/rtl.texi (MODE_FUNCTION): Remove not used entry.
203 2018-09-25 Martin Liska <mliska@suse.cz>
206 * doc/invoke.texi: Document all parameters and remove default
209 2018-09-25 Ilya Leoshkevich <iii@linux.ibm.com>
212 * rtl.c (rtx_code_size): Take into account that EQ_ATTR_ALT
213 contains HOST_WIDE_INTs when computing its size.
215 2018-09-24 Jim Wilson <jimw@sifive.com>
218 * config/riscv/riscv.h (STACK_BOUNDARY): Test riscv_abi == ABI_ILP32E
220 (ABI_STACK_BOUNDARY, MAX_ARGS_IN_REGISTERS): Likewise.
222 2018-09-24 Andrew Pinski <apinski@marvell.com>
224 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Don't
225 access prev before checking it for NULLness in the
226 AARCH64_FUSE_CMP_BRANCH case.
228 2018-09-24 H.J. Lu <hongjiu.lu@intel.com>
231 * config/i386/i386.c (rest_of_insert_endbranch): Set
232 endbr_queued_at_entrance to true and don't insert ENDBR if
233 x86_function_profiler will be called.
234 (x86_function_profiler): Insert ENDBR if endbr_queued_at_entrance
236 * config/i386/i386.h (machine_function): Add
237 endbr_queued_at_entrance.
239 2018-09-24 Ilya Leoshkevich <iii@linux.ibm.com>
241 * genattrtab.c (mk_attr_alt): Use alternative_mask.
242 (attr_rtx_1): Adjust caching to match the new EQ_ATTR_ALT field
244 (check_attr_test): Use alternative_mask.
245 (get_attr_value): Likewise.
246 (compute_alternative_mask): Use alternative_mask and XWINT.
247 (make_alternative_compare): Use alternative_mask.
248 (attr_alt_subset_p): Use XWINT.
249 (attr_alt_subset_of_compl_p): Likewise.
250 (attr_alt_intersection): Use alternative_mask and XWINT.
251 (attr_alt_union): Likewise.
252 (attr_alt_complement): Use HOST_WIDE_INT and XWINT.
253 (mk_attr_alt): Use alternative_mask and HOST_WIDE_INT.
254 (simplify_test_exp): Use alternative_mask and XWINT.
255 (write_test_expr): Use alternative_mask and XWINT, adjust bit
256 number calculation to support 64 bits. Generate code that
258 (main): Use alternative_mask.
259 * rtl.def (EQ_ATTR_ALT): Change field types from ii to ww.
261 2018-09-24 Ilya Leoshkevich <iii@linux.ibm.com>
264 * config/s390/s390.c (s390_emit_epilogue): Do not use PARALLEL
265 RETURN+USE when returning via %r14.
267 2018-09-24 Martin Liska <mliska@suse.cz>
269 * gcov.c (output_lines): Print colorization legend
270 for both flag_use_colors and flag_use_hotness_colors.
273 2018-09-24 Martin Liska <mliska@suse.cz>
275 * coverage.c (get_coverage_counts): Use warning_at
276 with current_function_decl location. Use %qD in warning
279 2018-09-24 Martin Liska <mliska@suse.cz>
281 * memory-block.h (memory_block_pool::release): Annotate with
282 valgrind that the memory is not accessible.
284 2018-09-24 Martin Liska <mliska@suse.cz>
287 * asan.c: Make asan_handled_variables extern.
289 * cfgexpand.c (expand_stack_vars): Make sure
290 a representative is unpoison if another
291 variable in the partition is handled by
292 use-after-scope sanitization.
294 2018-09-24 Richard Biener <rguenther@suse.de>
296 PR tree-optimization/63155
297 * tree-ssa-propagate.c (add_ssa_edge): Avoid adding PHIs to
298 the worklist when the edge of the respective argument isn't
301 2018-09-23 Uros Bizjak <ubizjak@gmail.com>
303 * config/i386/i386.h (enum reg_class): Rename MASK_REGS to
304 ALL_MASK_REGS and MASK_EVEX_REGS to MASK_REGS.
305 (MASK_CLASS_P): Update for rename.
306 (MAYBE_MASK_CLASS_P): Ditto.
307 (REG_CLASS_NAMES): Update.
308 (REG_CLASS_CONTENT): Update.
309 * config/i386/i386.c (regclass_map): Update for MASK_REG
310 and ALL_MASK_REGS rename.
311 * config/i386/constraints.md (Yk): Update for rename.
314 2018-09-23 Uros Bizjak <ubizjak@gmail.com>
316 * config/i386/i386.h (enum reg_class): Remove
317 EVEX_SSE_REGS and MOD4_SSE_REGS.
318 (REG_CLASS_NAMES): Update.
319 (REG_CLASS_CONTENT): Update.
320 * config/i386/i386.c (regclass_map): Declare AVX-512 SSE
321 registers as ALL_SSE_REGS.
322 (ix86_additional_allocno_class_p): Remove.
323 (TARGET_ADDITIONAL_ALLOCNO_CLASS_P): Remove.
324 (ix86_register_priority): Lower priority of EVEX SSE registers.
325 Use IN_RANGE macro where appropriate.
326 (ix86_hard_regno_mode_ok): Merge AVX-5124FMAPS and
327 AVX-5124VNNIW checks.
328 (ix86_modes_tieable_p): Tie 512-bit SSE modes.
329 * config/i386/sse.md (avx5124fmaddps_4fmaddps)
330 (avx5124fmaddps_4fmaddps_mask, avx5124fmaddps_4fmaddps_maskz)
331 (avx5124fmaddps_4fmaddss, avx5124fmaddps_4fmaddss_mask)
332 (avx5124fmaddps_4fmaddss_maskz, avx5124fmaddps_4fnmaddps)
333 (avx5124fmaddps_4fnmaddps_mask, avx5124fmaddps_4fnmaddps_maskz)
334 (avx5124fmaddps_4fnmaddss, avx5124fmaddps_4fnmaddss_mask)
335 (avx5124fmaddps_4fnmaddss_maskz, avx5124vnniw_vp4dpwssd)
336 (avx5124vnniw_vp4dpwssd_mask, avx5124vnniw_vp4dpwssd_maskz)
337 (avx5124vnniw_vp4dpwssds, avx5124vnniw_vp4dpwssds_mask)
338 (avx5124vnniw_vp4dpwssds_maskz): Use "v" instead of "Yh" constraint.
339 * config/i386/constraints.md (Yh): Remove.
341 2018-09-23 Uros Bizjak <ubizjak@gmail.com>
343 * config/i386/i386.c (regclass_map): Declare integer REX registers
346 2018-09-23 Gerald Pfeifer <gerald@pfeifer.com>
348 * doc/service.texi (Service): Switch the fsf.org link to https.
350 2018-09-22 Chung-Ju Wu <jasonwucj@gmail.com>
353 * config/nds32/nds32.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
354 Define to speculation_safe_value_not_needed.
356 2018-09-21 Florian Weimer <fweimer@redhat.com>
359 * doc/extend.texi (Common Function Attributes): Mention that
360 noreturn suppresses tail call optimization.
362 2018-09-21 Jeff Law <law@redhat.com>
364 * gimple-ssa-evrp.c (evrp_dom_walker::cleanup): Call
365 vr_values::cleanup_edges_and_switches.
366 * tree-vrp.c (to_remove_edges, to_update_switch_stmts): Moved into
368 (identify_jump_threads): Remove EDGE_IGNORE handling.
369 (execute_vrp): Move handling of to_remove_edges and
370 to_update_switch_stmts into vr_values class member functions.
371 * tree-vrp.h (switch_update, to_remove_edges): Remove declarations.
372 (to_update_switch_stmts): Likewise.
373 * vr-values.c: Include cfghooks.h.
374 (vr_values::vr_values): Initialize to_remove_edges and
375 to_update_switch_stmts.
376 (vr_values::~vr_values): Verify to_remove_edges and
377 to_update_switch_stmts are empty.
378 (vr_values::simplify_switch_using_ranges): Set EDGE_IGNORE as needed.
379 (vr_values::cleanup_edges_and_switches): New member function.
380 * vr-values.h (vr_values): Add cleanup_edges_and_switches member
381 function. Add new data members.
383 2018-09-21 David Malcolm <dmalcolm@redhat.com>
385 PR tree-optimization/87309
386 * dumpfile.c (dump_context::begin_scope): Filter the dump_loc
387 calls with pflags and alt_flags.
388 (selftest::test_capture_of_dump_calls): Add test of interaction of
389 MSG_OPTIMIZED_LOCATIONS with AUTO_DUMP_SCOPE.
391 2018-09-21 Olivier Hainque <hainque@adacore.com>
393 * config.gcc: Factorize and comment inclusion of vxworks-dummy.h.
395 2018-09-21 Olivier Hainque <hainque@adacore.com>
397 * config/vxworks.h (CLEAR_INSN_CACHE): #define to 1.
399 2018-09-21 Olivier Hainque <hainque@adacore.com>
401 * config/vxworks.h (VXWORKS_LIBGCC_SPEC): Remove -lc_internal.
402 Merge block comment with the one ahead of VXWORKS_LIBS_RTP. Then:
403 (VXWORKS_LIBS_RTP): Minor reordering.
405 2018-09-21 Olivier Hainque <hainque@adacore.com>
407 * config/vxworks.h (STARTFILE_PREFIX_SPEC): Define.
408 (VXWORKS_LIBS_DIR_RTP): Remove definition and use.
410 2018-09-21 Olivier Hainque <hainque@adacore.com>
412 * config/vxworks.h (SIZE_TYPE): Account for TARGET_VXWORKS64.
413 (PTRDIFF_TYPE): Likewise.
415 2018-09-21 Olivier Hainque <hainque@adacore.com>
417 * config.gcc: Enforce def of TARGET_VXWORKS64 to 1 from
418 triplet, similar to support for VxWorks7.
419 * config/vxworks-dummy.h: Provide a default definition
420 of TARGET_VXWORKS64 to 0.
422 2018-09-21 Olivier Hainque <hainque@adacore.com>
424 * config/vxworks.h (TARGET_VXWORKS7): Move default definition ...
425 * config/vxworks-dummy.h: here.
427 2018-09-21 Olivier Hainque <hainque@adacore.com>
429 * config.gcc: Prepend vxworks-dummy.h to tm_file for powerpc*
431 2018-09-21 Shaokun Zhang <zhangshaokun@hisilicon.com>
432 Bo Zhou <zbo.zhou@hisilicon.com>
434 * config/aarch64/aarch64-cores.def (tsv110): New CPU.
435 * config/aarch64/aarch64-tune.md: Regenerated.
436 * doc/invoke.texi (AArch64 Options/-mtune): Add "tsv110".
437 * config/aarch64/aarch64.c (tsv110_tunings): New tuning table.
438 * config/aarch64/aarch64-cost-tables.h: Add "tsv110" extra costs.
440 2018-09-21 Andrew Stubbs <ams@codesourcery.com>
441 Julian Brown <julian@codesourcery.com>
443 * builtins.c (get_builtin_sync_mem): Handle address spaces.
445 2018-09-21 Eric Botcazou <ebotcazou@adacore.com>
447 * config/rs6000/rs6000.c (rs6000_function_ok_for_sibcall): Return false
448 if the call takes a static chain.
450 2018-09-21 Martin Liska <mliska@suse.cz>
452 * auto-profile.c (autofdo_source_profile::read): Do not
454 (read_profile): Do not add working sets.
455 (read_autofdo_file): Remove sum_all.
456 (afdo_callsite_hot_enough_for_early_inline): Remove const
458 * coverage.c (struct counts_entry): Remove gcov_summary.
459 (read_counts_file): Read new GCOV_TAG_OBJECT_SUMMARY,
460 do not support GCOV_TAG_PROGRAM_SUMMARY.
461 (get_coverage_counts): Remove summary and expected
463 * coverage.h (get_coverage_counts): Likewise.
464 * doc/gcov-dump.texi: Remove -w option.
465 * gcov-dump.c (dump_working_sets): Remove.
466 (main): Do not support '-w' option.
467 (print_usage): Likewise.
468 (tag_summary): Likewise.
469 * gcov-io.c (gcov_write_summary): Do not dump
471 (gcov_read_summary): Likewise.
472 (gcov_histo_index): Remove.
473 (gcov_histogram_merge): Likewise.
474 (compute_working_sets): Likewise.
475 * gcov-io.h (GCOV_TAG_OBJECT_SUMMARY): Mark
477 (GCOV_TAG_PROGRAM_SUMMARY): Mark it obsolete.
478 (GCOV_TAG_SUMMARY_LENGTH): Adjust.
479 (GCOV_HISTOGRAM_SIZE): Remove.
480 (GCOV_HISTOGRAM_BITVECTOR_SIZE): Likewise.
481 (struct gcov_summary): Simplify rapidly just
482 to runs and sum_max fields.
483 (gcov_histo_index): Remove.
484 (NUM_GCOV_WORKING_SETS): Likewise.
485 (compute_working_sets): Likewise.
486 * gcov-tool.c (print_overlap_usage_message): Remove
488 * gcov.c (read_count_file): Read GCOV_TAG_OBJECT_SUMMARY.
489 (output_lines): Remove program related line.
490 * ipa-profile.c (ipa_profile): Do not consider GCOV histogram.
491 * lto-cgraph.c (output_profile_summary): Do not stream GCOV
493 (input_profile_summary): Do not read it.
494 (merge_profile_summaries): And do not merge it.
495 (input_symtab): Do not call removed function.
496 * modulo-sched.c (sms_schedule): Do not print sum_max.
497 * params.def (HOT_BB_COUNT_FRACTION): Reincarnate param that was
498 removed when histogram method was invented.
499 (HOT_BB_COUNT_WS_PERMILLE): Mention that it's used only in LTO
501 * postreload-gcse.c (eliminate_partially_redundant_load): Fix
503 * predict.c (get_hot_bb_threshold): Use HOT_BB_COUNT_FRACTION
504 and dump selected value.
505 * profile.c (add_working_set): Remove.
506 (get_working_sets): Likewise.
507 (find_working_set): Likewise.
508 (get_exec_counts): Do not work with working sets.
509 (read_profile_edge_counts): Do not inform as sum_max is removed.
510 (compute_branch_probabilities): Likewise.
511 (compute_value_histograms): Remove argument for call of
513 * profile.h: Do not make gcov_summary const.
515 2018-09-21 Monk Chiang <sh.chiang04@gmail.com>
517 * config.gcc (nds32*-*-*): Set TARGET_DEFAULT_TLSDESC_TRAMPOLINE=0.
519 2018-09-21 Eric Botcazou <ebotcazou@adacore.com>
521 PR tree-optimization/86990
522 * gimple-ssa-store-merging.c (imm_store_chain_info:coalesce_immediate):
523 Check that the entire merged store group is made of constants only for
526 2018-09-20 Allan Sandfeld Jensen <allan.jensen@qt.io>
528 * gcc.c (LINK_COMMAND_SPEC): Handle -r like -nostdlib.
529 (VTABLE_VERIFICATION_SPEC): Likewise.
530 (SANITIZER_EARLY_SPEC): Likewise.
531 (SANITIZER_SPEC): Likewise.
532 * config/darwin.h (LINK_COMMAND_SPEC): Likewise.
533 * doc/invoke.texi (Link Options): Document -r.
535 2018-09-20 Richard Biener <rguenther@suse.de>
538 * gimplify.c (gimplify_expr): Retain alignment of
539 addressable lvalue in dereference.
541 2018-09-20 Alexandre Oliva <aoliva@redhat.com>
544 * configure.ac: Check for .loc is_stmt support.
545 * configure, config.in: Rebuilt.
546 * dwarf2out.c (dwarf2out_source_line): Skip is_stmt
549 2018-09-20 Segher Boessenkool <segher@kernel.crashing.org>
551 * config/rs6000/rs6000.opt (misel=no, misel=yes): Delete.
552 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -misel=yes and
555 2018-09-20 Segher Boessenkool <segher@kernel.crashing.org>
557 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete
559 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete
562 2018-09-20 Marek Polacek <polacek@redhat.com>
564 * doc/invoke.texi: Add -Wno-init-list-lifetime to C++ Language Options.
566 2018-09-20 Richard Sandiford <richard.sandiford@arm.com>
568 PR tree-optimization/87288
569 * tree-vect-loop.c (vect_analyze_loop_2): Take PEELING_FOR_GAPS
570 into account when determining PEELING_FOR_NITERS.
572 2018-09-20 Richard Sandiford <richard.sandiford@arm.com>
574 PR tree-optimization/86877
575 * tree-vect-loop.c (vect_analyze_loop_2): Call
576 vect_verify_datarefs_alignment.
578 2018-09-19 Marek Polacek <polacek@redhat.com>
580 * doc/invoke.texi: Document -Wclass-conversion.
582 2018-09-19 John David Anglin <danglin@gcc.gnu.org>
584 * config/pa/pa.c (pa_adjust_priority): Delete.
585 (TARGET_SCHED_ADJUST_PRIORITY): Delete define.
587 * config/pa/pa.md (atomic_storeqi): Restore deleted expander.
588 (atomic_storehi): Likewise.
589 (atomic_storesi): Likewise.
590 (atomic_loaddi): Restore compare and swap exchange loop code.
592 2018-09-19 Segher Boessenkool <segher@kernel.crashing.org>
594 PR rtl-optimization/86902
595 * combine.c (try_combine): When changing the CC mode used, don't change
596 an unrelated mode in other_insn to that new CC mode.
598 2018-09-19 David Malcolm <dmalcolm@redhat.com>
600 * tree-data-ref.c (runtime_alias_check_p): Use formatted printing
601 with %T in place of calls to dump_generic_expr.
602 (prune_runtime_alias_test_list): Likewise.
603 (create_runtime_alias_checks): Likewise.
604 * tree-vect-data-refs.c (vect_check_nonzero_value): Likewise.
605 (vect_analyze_data_ref_dependence): Likewise.
606 (vect_slp_analyze_data_ref_dependence): Likewise.
607 (vect_record_base_alignment): Likewise. Use %G in place of call
609 (vect_compute_data_ref_alignment): Likewise.
610 (verify_data_ref_alignment): Likewise.
611 (vect_find_same_alignment_drs): Likewise.
612 (vect_analyze_group_access_1): Likewise.
613 (vect_analyze_data_ref_accesses): Likewise.
614 (dependence_distance_ge_vf): Likewise.
615 (dump_lower_bound): Likewise.
616 (vect_prune_runtime_alias_test_list): Likewise.
617 (vect_find_stmt_data_reference): Likewise.
618 (vect_analyze_data_refs): Likewise.
619 (vect_create_addr_base_for_vector_ref): Likewise.
620 (vect_create_data_ref_ptr): Likewise.
621 * tree-vect-loop-manip.c (vect_set_loop_condition): Likewise.
622 (vect_can_advance_ivs_p): Likewise.
623 (vect_update_ivs_after_vectorizer): Likewise.
624 (vect_gen_prolog_loop_niters): Likewise.
625 (vect_prepare_for_masked_peels): Likewise.
626 * tree-vect-loop.c (vect_determine_vf_for_stmt): Likewise.
627 (vect_determine_vectorization_factor): Likewise.
628 (vect_is_simple_iv_evolution): Likewise.
629 (vect_analyze_scalar_cycles_1): Likewise.
630 (vect_analyze_loop_operations): Likewise.
631 (report_vect_op): Likewise.
632 (vect_is_slp_reduction): Likewise.
633 (check_reduction_path): Likewise.
634 (vect_is_simple_reduction): Likewise.
635 (vect_create_epilog_for_reduction): Likewise.
636 (vect_finalize_reduction:): Likewise.
637 (vectorizable_induction): Likewise.
638 (vect_transform_loop_stmt): Likewise.
639 (vect_transform_loop): Likewise.
640 (optimize_mask_stores): Likewise.
641 * tree-vect-patterns.c (vect_pattern_detected): Likewise.
642 (vect_split_statement): Likewise.
643 (vect_recog_over_widening_pattern): Likewise.
644 (vect_recog_average_pattern): Likewise.
645 (vect_determine_min_output_precision_1): Likewise.
646 (vect_determine_precisions_from_range): Likewise.
647 (vect_determine_precisions_from_users): Likewise.
648 (vect_mark_pattern_stmts): Likewise.
649 (vect_pattern_recog_1): Likewise.
650 * tree-vect-slp.c (vect_get_and_check_slp_defs): Likewise.
651 (vect_record_max_nunits): Likewise.
652 (vect_build_slp_tree_1): Likewise.
653 (vect_build_slp_tree_2): Likewise.
654 (vect_print_slp_tree): Likewise.
655 (vect_analyze_slp_instance): Likewise.
656 (vect_detect_hybrid_slp_stmts): Likewise.
657 (vect_detect_hybrid_slp_1): Likewise.
658 (vect_slp_analyze_operations): Likewise.
659 (vect_slp_analyze_bb_1): Likewise.
660 (vect_transform_slp_perm_load): Likewise.
661 (vect_schedule_slp_instance): Likewise.
662 * tree-vect-stmts.c (vect_mark_relevant): Likewise.
663 (vect_mark_stmts_to_be_vectorized): Likewise.
664 (vect_init_vector_1): Likewise.
665 (vect_get_vec_def_for_operand): Likewise.
666 (vect_finish_stmt_generation_1): Likewise.
667 (vect_check_load_store_mask): Likewise.
668 (vectorizable_call): Likewise.
669 (vectorizable_conversion): Likewise.
670 (vectorizable_operation): Likewise.
671 (vectorizable_load): Likewise.
672 (vect_analyze_stmt): Likewise.
673 (vect_is_simple_use): Likewise.
674 (vect_get_vector_types_for_stmt): Likewise.
675 (vect_get_mask_type_for_stmt): Likewise.
676 * tree-vectorizer.c (increase_alignment): Likewise.
678 2018-09-19 Andrew Stubbs <ams@codesourcery.com>
680 * doc/rtl.texi: Adjust vec_select description.
681 * simplify-rtx.c (simplify_binary_operation_1): Allow VEC_SELECT to use
682 non-constant selectors.
684 2018-09-19 Matthew Malcomson <matthew.malcomson@arm.com>
686 * config/aarch64/aarch64-protos.h
687 (aarch64_offset_9bit_signed_unscaled_p): New declaration.
688 * config/aarch64/aarch64.md (arches): New "rcpc8_4" attribute value.
689 (arch_enabled): Add check for "rcpc8_4" attribute value of "arch".
690 * config/aarch64/aarch64.h (AARCH64_FL_RCPC8_4): New bitfield.
691 (AARCH64_FL_FOR_ARCH8_4): Include AARCH64_FL_RCPC8_4.
692 (AARCH64_FL_PROFILE): Move index so flags are ordered.
693 (AARCH64_ISA_RCPC8_4): New flag.
694 * config/aarch64/aarch64.c (offset_9bit_signed_unscaled_p): Renamed
695 to aarch64_offset_9bit_signed_unscaled_p.
696 * config/aarch64/atomics.md (atomic_store<mode>): Allow offset
698 * config/aarch64/constraints.md (Ust): New constraint.
699 * config/aarch64/predicates.md.
700 (aarch64_9bit_offset_memory_operand): New predicate.
701 (aarch64_rcpc_memory_operand): New predicate.
703 2018-09-19 Eric Botcazou <ebotcazou@adacore.com>
705 PR rtl-optimization/87361
706 * rtlanal.c (nonzero_bits1): Revert accidental change.
708 2018-09-19 Richard Biener <rguenther@suse.de>
710 PR tree-optimization/87349
711 PR tree-optimization/87342
712 * tree-ssa-sccvn.c (do_rpo_vn): Iterate max_rpo computation.
714 2018-09-18 Marek Polacek <polacek@redhat.com>
716 P1064R0 - Allowing Virtual Function Calls in Constant Expressions
717 * gimple-fold.c (gimple_get_virt_method_for_vtable): Adjust assert.
719 2018-09-18 Segher Boessenkool <segher@kernel.crashing.org>
721 * config/rs6000/rs6000.md: Remove old "Cygnus sibcall" comment.
723 2018-09-18 Segher Boessenkool <segher@kernel.crashing.org>
725 PR rtl-optimization/86882
726 * rtlanal.c (reg_overlap_mentioned_p): Handle CLOBBER.
728 2018-09-18 Uros Bizjak <ubizjak@gmail.com>
730 * config/i386/i386.md (*<code>extend<mode>xf2): Macroize insn from
731 *<code>extendsfxf2 and *<code>extenddfxf2 using MODEF mode iterator.
733 2018-09-18 Jonathan Wakely <jwakely@redhat.com>
736 * doc/invoke.texi (Link Options): Fix formatting and grammar.
738 2018-09-18 Richard Biener <rguenther@suse.de>
741 * tree-ssa-coalesce.c (tree_int_map_hasher): Remove.
742 (compute_samebase_partition_bases): Likewise.
743 (coalesce_ssa_name): Always use compute_optimized_partition_bases.
744 (gimple_can_coalesce_p): Simplify.
746 2018-09-18 Hans-Peter Nilsson <hp@bitrange.com>
748 Handle a library implementation of ffs calling __builtin_ffs.
749 * config/mmix/mmix.c (TARGET_INIT_LIBFUNCS): Override with...
750 (mmix_init_libfuncs): New function: make __builtin_ffs expand
753 2018-09-17 David Malcolm <dmalcolm@redhat.com>
755 * diagnostic-show-locus.c (class layout_range): Add field
757 (layout_range::layout_range): Add "original_idx" param and use it
758 to initialize new field.
759 (make_range): Use 0 for original_idx.
760 (layout::layout): Pass in index to calls to
761 maybe_add_location_range.
762 (layout::maybe_add_location_range): Add param "original_idx" and
763 pass it on to layout_range.
764 (layout::print_any_labels): Pass on range->m_original_idx to
766 (gcc_rich_location::add_location_if_nearby): Use 0 for
768 * gcc-rich-location.h (text_range_label::get_text): Update for new
770 (range_label_for_type_mismatch::get_text): Likewise.
772 2018-09-17 Uros Bizjak <ubizjak@gmail.com>
774 * config/i386/i386.c (ix86_emit_i387_log1p): Emit fldln2 earlier.
776 2018-09-17 David Malcolm <dmalcolm@redhat.com>
778 * gimple-ssa-sprintf.c (fmtwarn): Update for introduction of
779 format_string_diagnostic_t.
780 (fmtwarn_n): Likewise.
781 * substring-locations.c
782 (format_string_diagnostic_t::format_string_diagnostic_t) New ctor.
783 (format_warning_n_va): Convert to...
784 (format_string_diagnostic_t::emit_warning_n_va): ...this.
785 (format_warning_va): Convert to...
786 (format_string_diagnostic_t::emit_warning_va): ...this.
787 (format_warning_at_substring): Convert to...
788 (format_string_diagnostic_t::emit_warning): ...this.
789 (format_warning_at_substring_n): Convert to...
790 (format_string_diagnostic_t::emit_warning_n): ...this.
791 * substring-locations.h (class format_string_diagnostic_t): New
793 (format_warning_va): Convert to
794 format_string_diagnostic_t::emit_warning_va.
795 (format_warning_n_va): Convert to
796 format_string_diagnostic_t::emit_warning_n_va.
797 (format_warning_at_substring): Convert to
798 format_string_diagnostic_t::emit_warning.
799 (format_warning_at_substring_n): Convert to
800 format_string_diagnostic_t::emit_warning_n.
802 2018-09-17 Cesar Philippidis <cesar@codesourcery.com>
803 Bernd Schmidt <bernds_cb1@t-online.de>
805 * config/nvptx/nvptx.md (atomic_fetch_<logic><mode>): Enable with
808 2018-09-17 Uros Bizjak <ubizjak@gmail.com>
810 * config/i386/i386.md (truncxf<mode>2_i387_noop_unspec): Change
811 operand 0 predicate to nonimmediate operand.
812 (rint<mode>2_frndint): Remove insn pattern.
813 (rint<mode>2): Change operand 1 predicate to general_operand.
814 Extend operand 1 to XFmode and generate rintxf2 insn.
815 (frndintxf2_<rounding>): Rename from frndint<mode>2_<rounding>.
816 Do not use X87MODEF mode macro.
817 (frndintxf2_<rounding>_i387): Rename from
818 frndint<mode>2_<rounding>_i387. Do not use X87MODEF mode macro.
819 (<rounding_insn><mode>2): For non-SSE modes, extend operand 1
820 to XFmode and generate significandxf3 insn.
822 2018-09-17 Richard Biener <rguenther@suse.de>
824 PR tree-optimization/87328
825 * tree-ssa-sccvn.c (process_bb): Remove assertion about not
826 visiting unexecutable backedges when not iterating.
827 (do_rpo_vn): Mark all edges not executable even when not
830 2018-09-17 Martin Jambor <mjambor@suse.cz>
833 * doc/invoke.texi (Warning Options): Likewise.
835 2018-09-17 Richard Biener <rguenther@suse.de>
837 PR tree-optimization/87301
838 * tree-ssa-sccvn.c (eliminate_dom_walker::eliminate_cleanup): Properly
839 clean EH info from leftover copy assignments.
841 2018-09-17 Martin Liska <mliska@suse.cz>
843 PR gcov-profile/85871
844 * gcov.c (output_intermediate_file): Fix out of bounds
847 2018-09-17 Vineet Gupta <vgupta@synopsys.com>
849 * config/arc/arc.c: Object attributes for core4 not reflected
851 * config/arc/arc.h: Don't restrict DBNZ to core3 (core4 includes
854 2018-09-17 Alexey Brodkin <abrodkin@synopsys.com>
856 * config/arc/linux.h (LINK_EH_SPEC): Add missing space.
858 2018-09-17 Martin Liska <mliska@suse.cz>
860 * doc/gcov.texi: Document new option --use-hotness-colors.
861 * gcov.c (struct source_info): Declare new field.
862 (source_info::source_info): Set default for maximum_count.
863 (print_usage): Add new -q option.
864 (process_args): Process it.
865 (accumulate_line_info): Save src->maximum_count.
866 (output_line_beginning): Make color line number if
867 flag_use_hotness_colors is set.
868 (output_line_details): Pass default argument value.
869 (output_lines): Pass src->maximum_count.
871 2018-09-17 Martin Liska <mliska@suse.cz>
873 * common/config/i386/i386-common.c (ix86_get_valid_option_values):
874 Use processor_names table.
875 * config/i386/i386.c (ix86_default_align): Use
876 processor_cost_table for alignment values.
877 (ix86_option_override_internal): Use processor_names.
878 (ix86_function_specific_print): Likewise.
879 * config/i386/i386.h (struct processor_costs):
880 Add alignment values.
881 (struct ptt): Remove and replace with const char *.
882 * config/i386/x86-tune-costs.h (struct processor_costs):
883 Declare default alignments for all costs.
885 2018-09-17 Aldy Hernandez <aldyh@redhat.com>
887 * tree-vrp.c (extract_range_from_unary_expr): Do not special case
888 symbolics or VR_VARYING ranges for ABS_EXPR.
889 * wide-int-range.cc (wide_int_range_abs): Return positive numbers
890 when range will wrap.
892 2018-09-15 Eric Botcazou <ebotcazou@adacore.com>
895 * cfgexpand.c (expand_gimple_basic_block): Be prepared for a BARRIER
896 before and after a JUMP_TABLE_DATA.
898 2018-09-14 John David Anglin <danglin@gcc.gnu.org>
901 * dojump.c (do_compare_and_jump): Canonicalize function pointers
902 when one operand is a function pointer. Use POINTER_TYPE_P and
903 FUNC_OR_METHOD_TYPE_P.
904 * expr.c (do_store_flag): Use POINTER_TYPE_P and FUNC_OR_METHOD_TYPE_P.
905 * fold-const.c (build_range_check): Likewise.
906 * match.pd (simple_comparison): Likewise.
908 2018-09-14 David Malcolm <dmalcolm@redhat.com>
911 * spellcheck.c (get_edit_distance_cutoff): New function.
912 (selftest::test_edit_distance_unit_test_oneway): Rename to...
913 (selftest::test_get_edit_distance_one_way): ...this.
914 (selftest::test_get_edit_distance_unit): Rename to...
915 (selftest::test_get_edit_distance_both_ways): ...this.
916 (selftest::test_edit_distances): Move tests to this new function,
917 and test some more pairs of strings. Update for above renaming.
918 (selftest::get_old_cutoff): New function.
919 (selftest::test_get_edit_distance_cutoff): New function.
920 (selftest::assert_suggested_for): New function.
921 (ASSERT_SUGGESTED_FOR): New macro.
922 (selftest::assert_not_suggested_for): New function.
923 (ASSERT_NOT_SUGGESTED_FOR): New macro.
924 (selftest::test_suggestions): New function.
925 (selftest::spellcheck_c_tests): Move test_get_edit_distance_unit
926 tests to selftest::test_edit_distances and call it. Add calls to
927 selftest::test_get_edit_distance_cutoff and
928 selftest::test_suggestions.
929 * spellcheck.h (get_edit_distance_cutoff): New function declaration.
930 (best_match::consider): Replace hard-coded cutoff calculation with
932 (best_match::get_cutoff): New declaration.
933 (best_match::get_best_meaningful_candidate): Likewise.
935 2018-09-14 Bernd Edlinger <bernd.edlinger@hotmail.de>
937 * builtins.c (fold_builtin_strlen): Remove TODO comment.
939 2018-09-14 Bernd Edlinger <bernd.edlinger@hotmail.de>
942 2018-07-30 Bernd Edlinger <bernd.edlinger@hotmail.de>
944 * tree-ssa-forwprop.c (simplify_builtin_call): Don't create a not NUL
945 terminated string literal.
947 2018-09-14 Martin Sebor <msebor@redhat.com>
949 * builtins.c (unterminated_array): Handle ARRAY_REF.
950 (expand_builtin_stpcpy_1): Detect unterminated char arrays.
951 * builtins.h (unterminated_array): Declare extern.
952 * gimple-fold.c (gimple_fold_builtin_stpcpy): Detect unterminated
954 (gimple_fold_builtin_sprintf): Propagate NO_WARNING to transformed
957 2018-09-14 Martin Sebor <msebor@redhat.com>
958 Jeff Law <law@redhat.com>
960 * builtins.c (unterminated_array): New.
961 (expand_builtin_strcpy): Adjust.
962 (expand_builtin_strcpy_args): Detect unterminated arrays.
963 * gimple-fold.c (get_maxval_strlen): Add argument. Detect
965 * gimple-fold.h (get_maxval_strlen): Add argument.
966 (gimple_fold_builtin_strcpy): Detec unterminated arrays.
968 * gimple-fold.c (get_range_strlen): Add argument.
969 (get_maxval_strlen): Adjust.
970 * gimple-fold.h (get_range_strlen): Add argument.
972 2018-09-14 Wei Xiao <wei3.xiao@intel.com>
974 * config/i386/movdirintrin.h: Fix copyright year.
976 2018-09-14 Uros Bizjak <ubizjak@gmail.com>
978 * reg-stack.c: Include regs.h.
979 (replace_reg): Assert that mode is MODE_FLOAT or MODE_COMPLEX_FLOAT.
980 (emit_pop_insn): Default pop insn mode to the reg_raw_mode of
981 FIRST_STACK_REG, not DFmode.
982 (emit_swap_insn): Default swap insn mode to the reg_raw_mode of
983 FIRST_STACK_REG, not XFmode. Explicitly construct swap RTX.
984 (change stack): Default register mode to the reg_raw_mode of
985 FIRST_STACK_REG, not DFmode.
986 * config/i386/i386.md (*swap<mode>): Remove insn pattern.
987 (*swapxf): Rename from swapxf.
989 2018-09-14 Carl Love <cel@us.ibm.com>
991 * config/rs6000/emmintrin.h: Add _MM_SHUFFLE2.
992 * config/rs6000/xmmintrin.h: Add _MM_SHUFFLE.
994 2018-09-14 Segher Boessenkool <segher@kernel.crashing.org>
997 * config/rs6000/rs6000.md (*mov<mode>_hardfloat64): Add Z to the Y
1000 2018-09-14 Sam Tebbs <sam.tebbs@arm.com>
1003 * config/aarch64/aarch64.md (*aarch64_bfxilsi_uxtw): Define.
1005 2018-09-14 Jason Merrill <jason@redhat.com>
1007 Fix --enable-gather-detailed-mem-stats.
1008 * hash-table.c (hash_table_usage): Change from variable to function.
1009 * hash-table.h: Adjust.
1010 * Makefile.in: Add missing dependencies on hash-table.h.
1012 2018-09-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1014 PR tree-optimization/87259
1016 (pass_cse_reciprocals::execute): Run optimize_recip_sqrt after
1017 execute_cse_reciprocals_1 has tried transforming.
1019 2018-09-14 Aldy Hernandez <aldyh@redhat.com>
1021 * tree-vrp.c (extract_range_from_binary_expr_1): Normalize
1022 VR_VARYING for PLUS/MINUS_EXPR.
1024 2018-09-14 Ilya Leoshkevich <iii@linux.ibm.com>
1026 * config/s390/s390-passes.def (INSERT_PASS_BEFORE): Improve
1029 2018-09-14 Richard Biener <rguenther@suse.de>
1032 * tree-ssa-coalesce.c (ssa_conflicts_merge): Remove conflict
1033 bits for the merged partition.
1035 2018-09-13 Martin Sebor <msebor@redhat.com>
1036 Bernd Edlinger <bernd.edlinger@hotmail.de>
1038 * builtins.h (c_srlen): Add argument.
1039 * builtins.c (warn_string_no_nul): New function.
1040 (c_strlen): Add argument and use it. Update recursive calls.
1041 Pass DECL argument to string_constant to get info on non
1042 terminated strings. Update *NONSTR as needed.
1043 (fold_builtin_strlen): Add argument to calls to c_strlen.
1044 Warn for unterminated arrays.
1045 (warn_string_no_null): Add prototype.
1046 * expr.c (string_constant): Update arguments. Update recursive
1047 calls appropriately. Detect missing NUL terminator and outermost
1048 declaration its missing in.
1049 Improve checks for arrays with nonzero lower bound or elements
1050 that are not a single byte. Simplify offset computation.
1051 Simplify checks for non-NUL terminated strings.
1052 * gimple-fold.c (get_range_strlen): Add argument to c_strlen call.
1053 * gimple-ssa-sprintf.c (get_string_length): Remove unnecessary code.
1055 2018-09-13 Bernd Edlinger <bernd.edlinger@hotmail.de>
1057 * builtins.c (c_strlen): Handle not zero terminated STRING_CSTs
1059 * fold-const.c (c_getstr): Fix function comment. Remove unused third
1060 argument. Fix range checks.
1061 * fold-const.h (c_getstr): Adjust protoype.
1062 * gimple-fold.c (gimple_fold_builtin_memory_op): Avoid folding when
1063 string is constant but contains no NUL byte.
1065 * expr.c (string_constant): Adjust function comment.
1066 Remove bogus check for zero termination.
1068 * fold-const.c (c_getstr): Clamp STRING_LENGTH to STRING_SIZE.
1070 * varasm.c (compare_constant): Compare type size of STRING_CSTs.
1071 (get_constant_size): Don't make STRING_CSTs larger than they are.
1072 (check_string_literal): New check function for STRING_CSTs.
1073 (output_constant): Use it.
1075 2018-09-13 Eric Botcazou <ebotcazou@adacore.com>
1078 * config/visium/visium.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Define.
1080 2018-09-13 Richard Biener <rguenther@suse.de>
1082 PR tree-optimization/87263
1083 * tree-ssa-sccvn.c (visit_phi): Revert some earlier changes.
1084 (struct unwind_state): Add max_rpo field.
1085 (do_rpo_vn): Allow up-to-date loop state to be used when not iterating.
1086 Compute max_rpo, the max RPO number a block can be backwards reached
1087 from. Re-write non-iterating mode to a RPO ordered worklist approach,
1088 separating it from the iterating mode.
1090 2018-09-13 Vlad Lazar <vlad.lazar@arm.com>
1092 * haifa-sched.c (rank_for_schedule): Schedule by INSN_COST.
1093 (rfs_decision): New scheduling decision.
1095 2018-09-13 Richard Biener <rguenther@suse.de>
1098 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Fix assert.
1099 (vn_nary_op_insert_pieces_predicated): Do not write useless
1100 valid_dominated_by_p entry outside of the allocated storage.
1102 2018-09-13 Omar Sandoval <osandov@osandov.com>
1103 Tom de Vries <tdevries@suse.de>
1106 * dwarf2out.c (is_c): New function.
1107 (add_subscript_info): Add DW_AT_count of 0 for C zero-length arrays.
1109 2018-09-13 Sam Tebbs <sam.tebbs@arm.com>
1112 * config/aarch64/aarch64.md (*aarch64_bfxil):
1114 * config/aarch64/constraints.md (Ulc): Define.
1115 * config/aarch64/aarch64-protos.h (aarch64_high_bits_all_ones_p):
1117 * config/aarch64/aarch64.c (aarch64_high_bits_all_ones_p):
1120 2018-09-13 Vlad Lazar <vlad.lazar@arm.com>
1122 * config/aarch64/aarch64.h (TARGET_COMPUTE_FRAME_LAYOUT): Define.
1123 * config/aarch64/aarch64.c (aarch64_expand_prologue): Remove
1124 aarch64_layout_frame call.
1125 (aarch64_expand_epilogue): Likewise.
1126 (aarch64_initial_elimination_offset): Likewise.
1127 (aarch64_get_separate_components): Likewise.
1128 (aarch64_use_return_insn_p): Likewise.
1129 (aarch64_layout_frame): Remove unneeded check.
1131 2018-09-13 Jakub Jelinek <jakub@redhat.com>
1133 * configure.ac: Only append
1134 " : (reconfigured) $TOPLEVEL_CONFIGURE_ARGUMENTS" to
1135 gcc_config_arguments if it was never reconfigured or last reconfigure
1136 was with different arguments.
1137 * configure: Regenerated.
1139 2018-09-13 Jakub Jelinek <jakub@redhat.com>
1140 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1143 * expr.c (maybe_optimize_pow2p_mod_cmp): New function.
1144 (maybe_optimize_mod_cmp): Use it if integer_pow2p treeop1.
1146 2018-09-13 Jakub Jelinek <jakub@redhat.com>
1148 PR tree-optimization/87287
1149 * fold-const.c (fold_binary_loc) <case EQ_EXPR>: Move signed modulo
1150 X % C == 0 to X % (unsigned) C == 0 optimization to ...
1151 * match.pd (X % C == 0): ... here. New optimization.
1153 2018-09-12 Jakub Jelinek <jakub@redhat.com>
1156 * expr.h (maybe_optimize_mod_cmp): Declare.
1157 * expr.c (mod_inv): New function.
1158 (maybe_optimize_mod_cmp): New function.
1159 (do_store_flag): Use it.
1160 * cfgexpand.c (expand_gimple_cond): Likewise.
1162 2018-09-09 Cesar Philippidis <cesar@codesourcery.com>
1163 Julian Brown <julian@codesourcery.com>
1166 * gimplify.c (gimplify_scan_omp_clauses): Set
1167 target_firstprivatize_array_bases in OpenACC parallel and kernels
1168 region contexts. Remove GOMP_MAP_FIRSTPRIVATE_REFERENCE clauses from
1169 OpenACC data regions.
1171 2018-09-12 Uros Bizjak <ubizjak@gmail.com>
1173 * config/i386/i386.md (sqrt_extend<mode>xf3_i387): Remove.
1174 (sqrt<mode>2): Extend operand 1 to XFmode and generate sqrtxf3 insn.
1176 2018-09-12 Richard Biener <rguenther@suse.de>
1178 PR tree-optimization/87280
1179 * tree-ssa-sccvn.c (process_bb): Handle the case of executable
1180 edge but unreachable target.
1181 (do_rpo_vn): For conservatively handling a PHI only mark
1182 the backedge executable but not the block reachable.
1184 2018-09-12 Richard Biener <rguenther@suse.de>
1186 PR tree-optimization/87266
1187 * tree-ssa-sccvn.c (do_rpo_vn): Always iterate to not yet
1190 2018-09-12 Andreas Krebbel <krebbel@linux.ibm.com>
1192 * config/s390/s390.md (PFPO_RND_MODE_DFP, PFPO_RND_MODE_BFP): New
1194 ("trunc<BFP:mode><DFP_ALL:mode>2")
1195 ("trunc<DFP_ALL:mode><BFP:mode>2")
1196 ("extend<BFP:mode><DFP_ALL:mode>2")
1197 ("extend<DFP_ALL:mode><BFP:mode>2"): Set proper rounding mode
1198 according to the target operand type.
1200 2018-09-12 Jakub Jelinek <jakub@redhat.com>
1201 Andreas Krebbel <krebbel@linux.ibm.com>
1203 PR tree-optimization/86844
1204 * gimple-ssa-store-merging.c
1205 (imm_store_chain_info::coalesce_immediate): For overlapping stores, if
1206 there are any overlapping stores in between them, make sure they are
1207 also coalesced or we give up completely.
1209 2018-09-12 Jakub Jelinek <jakub@redhat.com>
1212 * fold-const.c (fold_ternary_loc) <case COND_EXPR>: Verify also that
1213 BIT_AND_EXPR's second operand is a power of two. Formatting fix.
1215 2018-09-12 Tom de Vries <tdevries@suse.de>
1217 * common.opt (gdescribe-dies): Add option.
1218 * dwarf2out.c (add_name_and_src_coords_attributes): Add description
1219 attribute for artifical and nameless decls.
1220 (dwarf2out_register_external_die): Add description attribute to
1221 external reference die.
1222 (add_desc_attribute): New functions.
1223 (gen_subprogram_die): Add description attribute to
1224 DW_TAG_call_site_parameter.
1225 * tree-pretty-print.c (print_generic_expr_to_str): New function.
1226 * tree-pretty-print.h (print_generic_expr_to_str): Declare.
1227 * doc/invoke.texi (@item Debugging Options): Add -gdescribe-dies and
1229 (@item -gdescribe-dies): Add.
1231 2018-09-12 Aldy Hernandez <aldyh@redhat.com>
1233 * tree-vrp.c (vrp_shift_undefined_p): Remove.
1234 (extract_range_from_binary_expr_1: Call
1235 wide_int_range_shift_undefined_p instead of vrp_shift_undefined_p.
1236 * wide-int-range.h (wide_int_range_shift_undefined_p): Do not
1239 2018-09-12 Aldy Hernandez <aldyh@redhat.com>
1241 * gimple-ssa-warn-alloca.c
1242 (alloca_type_and_limit::alloca_type_and_limit): Initialize limit
1243 field for ALLOCA_BOUND_*_LARGE.
1245 2018-09-11 Nathan Sidwell <nathan@acm.org>
1247 * gcc.c (load_specs, execute, run_attempt): Use %qs not '%s'.
1249 2018-09-11 Uros Bizjak <ubizjak@gmail.com>
1251 * reg-stack.c (subst_asm_stack_regs): Call replace_reg also
1252 for clobbers. Remove obsolete comment.
1254 2018-09-11 Uros Bizjak <ubizjak@gmail.com>
1256 * config/i386/i386.md (define_attr "type"): Remove mpxmov, mpxmk,
1257 mpxchk, mpxld and mpxst types.
1258 (define_attr length_immediate): Remove all processing of mpx types.
1259 (define_attr prefix_0f): Ditto.
1260 (define_attr memory): Ditto.
1262 2018-09-11 Uros Bizjak <ubizjak@gmail.com>
1264 * config/i386/i386.md (fyl2x_extend<mode>xf3_i387): Remove.
1265 (log<mode>2): Change operand 1 predicate to general_operand.
1266 Extend operand 1 to XFmode and generate logxf3 insn.
1267 (log10<mode>2): Change operand 1 predicate to general_operand.
1268 Extend operand 1 to XFmode and generate log10xf3 insn.
1269 (log2<mode>2): Change operand 1 predicate to general_operand.
1270 Extend operand 1 to XFmode and generate log2xf3 insn.
1271 (fyl2xp1_extend<mode>xf3_i387): Remove.
1272 (log1p<mode>2): Change operand 1 predicate to general_operand.
1273 Extend operand 1 to XFmode and generate log1pxf3 insn.
1274 (fxtract_extend<mode>xf3_i387): Remove.
1275 (logb<mode>2): Change operand 1 predicate to general_operand.
1276 Extend operand 1 to XFmode and generate logbxf3 insn.
1277 (ilogb<mode>2): Change operand 1 predicate to general_operand.
1278 Extend operand 1 to XFmode and generate fxtractxf3_i387 insn.
1279 (significand<mode>2): Change operand 1 predicate to general_operand.
1280 Extend operand 1 to XFmode and generate significandxf3 insn.
1282 2018-09-11 Nathan Sidwell <nathan@acm.org>
1284 * gcc.c (perror_with_name, pfatal_with_name): Delete.
1285 (load_specs): Use fatal_error.
1286 (DELETE_IF_ORDINARY, process_command): Use error.
1287 (execute, run_attempt): Use fatal_error.
1289 2018-09-11 Andrew Stubbs <ams@codesourcery.com>
1291 * diagnostic-core.h (sorry_at): New prototype.
1292 * diagnostic.c (sorry_at): New function.
1294 2018-09-11 Aldy Hernandez <aldyh@redhat.com>
1296 * tree-vrp.c (extract_range_from_binary_expr_1): Treat all divisions
1297 by zero as VR_UNDEFINED.
1299 2018-09-10 Uros Bizjak <ubizjak@gmail.com>
1301 * config/i386/i386.md (<sincos>xf2): Rename from *<sincos>xf2_i387.
1302 (*<sincos>_extend<mode>xf2_i387): Remove insn pattern.
1303 (<sincos>mode2): New expander.
1304 (sincos_extend<mode>xf3_i387): Remove insn pattern.
1305 (sincos -> sin, cos splitters): Remove splitter patterns.
1306 (sincos<mode>3): Change operand 2 predicate to general_operand.
1307 Extend operand 2 to XFmode and generate sincosxf3 insn.
1308 (fptanxf4_i387): Change mode of operands 0 and 3 to SFmode.
1309 Change operand 3 predicate to const1_operand.
1310 (fptan_extend<mode>xf4_i387): Remove insn pattern.
1311 (tanxf2): Update operands in the call to fptanxf4_i387.
1312 (tan<mode>2): Change operand 1 predicate to general_operand.
1313 Extend operand 1 to XFmode and generate tanxf3 insn.
1314 (atan2xf3): Rename from *fpatanxf3_i387.
1315 (fpatan_extend<mode>xf3_i387): Remove insn pattern.
1316 (atan2xf3): Remove expander.
1317 (atan2<mode<3): Change operand 1 and 2 predicates to general_operand.
1318 Extend operands 1 and 2 to XFmode and generate atan2xf3 insn.
1319 (atan<mode>2): Change operand 1 predicate to general_operand.
1320 Extend operand 1 to XFmode and generate atanxf3 insn.
1322 2018-09-10 Uros Bizjak <ubizjak@gmail.com>
1324 * config/i386/i386.md (x87/SSE constant load splitter): Use
1325 memory_operand instead of nonimmediate_operand for input operand
1328 2018-09-09 Uros Bizjak <ubizjak@gmail.com>
1330 * config/i386/i386.md (float partial SSE register stall splitter): Move
1331 splitter near its instruction pattern.
1332 (float_extend partial SSE register stall splitter): Ditto.
1333 (float_truncate partial SSE register stall splitter): Ditto.
1335 2018-09-09 Hans-Peter Nilsson <hp@bitrange.com>
1338 * config/mmix/mmix.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Redefine
1339 to speculation_safe_value_not_needed.
1342 * config/mmix/mmix.c (mmix_assemble_integer): Handle byte-size
1343 non-CONST_INT rtx:es using assemble_integer_with_op ".byte".
1344 (MMIX_CFUN_NEEDS_SAVED_EH_RETURN_ADDRESS): Don't call
1345 leaf_function_p, instead use has_hard_reg_initial_val.
1347 2018-09-09 Nathan Sidwell <nathan@acm.org>
1349 * gcc.h (pfatal_with_name): Don't declare here.
1350 * gcc.c (pfatal_with_name): Make static.
1352 2018-09-09 Xianmiao Qu <xianmiao_qu@c-sky.com>
1354 * config/csky/csky.md (*cskyv2_adddi3, *cskyv2_subdi3): Add
1357 2018-09-08 John David Anglin <danglin@gcc.gnu.org>
1359 PR rtl-optimization/85458
1360 * sel-sched.c (sel_target_adjust_priority): Allow backend adjust
1361 priority hook to reduce the priority of EXPR.
1363 2018-09-07 Uros Bizjak <ubizjak@gmail.com>
1365 * config/i386/i386.md (float<SWI48x:mode><MODEF:mode>2) Enable
1366 DImode for x87 on 32bit targets. Conditionally disable x87 modes
1367 with X87_ENABLE_FLOAT. Remove preparation code.
1368 (*float<SWI48:mode><MODEF:mode>2): Rename from
1369 *float<SWI48:mode><MODEF:mode>2_mixed. Handle x87, SSE and mixed
1370 math using "enabled" attribute.
1371 (*floatdi<MODEF:mode>2_i387): Rename from
1372 *float<SWI48x:mode><MODEF:mode>2_i387. Handle only DImode and
1373 enable for 32bit targets only.
1374 (floatdi<X87MODEF:mode>2_i387_with_xmm pre-reload splitter): New
1376 (floatdi<X87MODEF:mode>2_i387_with_xmm): Use register_operand
1377 as operand 1 predicate. Rewrite as define_insn_and_split.
1378 (floatdi<X87MODEF:mode>2_i387_with_xmm memory input splitter): Remove.
1380 2018-09-06 Uros Bizjak <ubizjak@gmail.com>
1382 * reg-stack.c (get_true_reg) <case FLOAT_TRUNCATE>: Reorder
1383 to fallthru to FLOAT case.
1385 2018-09-06 Will Schmidt <will_schmidt@vnet.ibm.com>
1388 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Update logic
1389 around folding of vec_sl to handle out of range shift values.
1391 2018-09-06 Uros Bizjak <ubizjak@gmail.com>
1393 * config/i386/i386.md (fix_trunc<mode>_fisttp_i387_1): Remove.
1394 Update callers to gen_fix_trunc<mode>_i387_fisttp
1395 (fix_trunc<mode>_i387_fisttp): Change operand 0 predicate to
1396 nonimmediate_operand.
1397 (fix_trunc<mode>_i387_fisttp_with_temp): Remove insn pattern
1398 and corresponding splitters.
1399 (*fix_trunc<mode>_i387_1): Always expand with fix_trunc<mode>_i387.
1400 (fix_truncdi_i387): Change operand 0 predicate to nonimmediate_operand.
1401 (fix_truncdi_i387_with_temp): Remove insn pattern
1402 and corresponding splitters.
1403 (fix_trunc<mode>_i387): Change operand 0 predicate to
1404 nonimmediate_operand.
1405 (fix_trunc<mode>_i387_with_temp): Remove insn pattern
1406 and corresponding splitters.
1407 (*fistdi2_1): Remove.
1409 (fistdi2_with_temp): Remove insn pattern and corresponding splitters.
1410 (lrintxfdi2): Remove expander. Reimplement as define_insn.
1411 (*fist<mode>2_1): Remove.
1412 (fist<mode>2): Ditto.
1413 (fist<mode>2_with_temp): Remove insn pattern and corresponding
1415 (lrintxf<mode>2): Remove expander. Reimplement as define_insn.
1416 (*fist<mode>2_<rounding>_1): Always expand with fist<mode>2_<rounding>.
1417 (fistdi2_<rounding>): Change operand 0 predicate to
1418 nonimmediate_operand.
1419 (fistdi2_<rounding>_with_temp): Remove insn pattern
1420 and corresponding splitters.
1421 (fist<mode>2_<rounding>): Change operand 0 predicate to
1422 nonimmediate_operand.
1423 (fist<mode>2_<rounding>_with_temp): Remove insn pattern
1424 and corresponding splitters.
1426 (*fixuns_trunc<mode>si2_avx512f_zext): Depend on TARGET_SSE_MATH.
1428 2018-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
1430 * varasm.c (output_constructor_regular_field): Check TYPE_SIZE_UNIT of
1433 2018-09-06 Will Schmidt <will_schmidt@vnet.ibm.com>
1435 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
1436 early gimple folding of vec_splat().
1437 * tree-vect-generic.c: Remove static from tree_vec_extract() definition.
1438 * gimple-fold.h: Add an extern define for tree_vec_extract().
1440 2018-09-06 Will Schmidt <will_schmidt@vnet.ibm.com>
1442 * config/rs6000/rs6000.c (fold_mergehl_helper): Add types_compatible_p
1443 wrappers around TREE_TYPE comparisons.
1445 2018-09-06 Ilya Leoshkevich <iii@linux.ibm.com>
1448 * config/s390/predicates.md: Add nonsym_memory_operand.
1449 * config/s390/s390.c (s390_legitimize_cs_operand): If operand
1450 contains a SYMBOL_REF, load it into an intermediate pseudo.
1451 (s390_emit_compare_and_swap): Legitimize operand.
1452 * config/s390/s390.md: Use the new nonsym_memory_operand
1453 with UNSPECV_CAS patterns.
1455 2018-09-06 Ilya Leoshkevich <iii@linux.ibm.com>
1458 * config/s390/s390-passes.def: New file.
1459 * config/s390/s390-protos.h (class rtl_opt_pass): Add forward
1461 (make_pass_s390_early_mach): Add declaration.
1462 * config/s390/s390.c (make_pass_s390_early_mach):
1463 (s390_option_override): Remove dynamic registration.
1464 * config/s390/t-s390: Add s390-passes.def.
1466 2018-09-06 Ilya Leoshkevich <iii@linux.ibm.com>
1468 * config/s390/s390.c (s390_decompose_constant_pool_ref):
1469 Remove UNSPEC_LTREL_BASE check.
1470 (annotate_constant_pool_refs): Likewise.
1471 (find_constant_pool_ref): Likewise.
1472 (find_ltrel_base): Removed.
1473 (replace_ltrel_base): Removed.
1474 (s390_mainpool_finish): Remove replace_ltrel_base call.
1475 (s390_chunkify_start): Remove pending LTREL_BASE logic.
1476 (s390_chunkify_finish): Remove replace_ltrel_base call.
1477 * config/s390/s390.md: Remove UNSPEC_LTREL_BASE.
1479 2018-09-06 Hans-Peter Nilsson <hp@axis.com>
1482 * config/cris/cris.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Redefine
1483 to speculation_safe_value_not_needed.
1485 2018-09-05 Cesar Philippidis <cesar@codesourcery.com>
1486 Bernd Schmidt <bernds_cb1@t-online.de>
1488 * config/nvptx/nvptx-opts.h: New file.
1489 * config/nvptx/nvptx.c (nvptx_file_start): Print the correct .target.
1490 * config/nvptx/nvptx.h: Include "nvptx-opts.h".
1492 (TARGET_SM35): New macro.
1493 * config/nvptx/nvptx.md (atomic_fetch_<logic><mode>): Enable with the
1495 * config/nvptx/nvptx.opt (ptx_isa, sm_30, sm_35): New enum and its
1497 (misa=): New option.
1498 * doc/invoke.texi (Nvidia PTX Options): Document -misa.
1500 2018-09-05 Uros Bizjak <ubizjak@gmail.com>
1502 * config/i386/i386.md (truncdfsf2): Remove expander.
1503 (truncdfsf2_with_temp): Ditto.
1504 (truncxf<mode>2): Ditto.
1505 (*truncdfsf_fast_mixed): Remove insn pattern.
1506 (*truncdfsf_fast_i387): Ditto.
1507 (*truncdfsf_mixed): Ditto.
1508 (*truncdfsf_i387): Ditto.
1509 (*truncdfsf2_i387_1): Ditto.
1510 (*truncxfsf2_mixed): Ditto.
1511 (*truncxfdf2_mixed): Ditto.
1512 (*truncxf<mode>2_i387_noop): Ditto. Update callers
1513 to call gen_truncxf<mode>2 instead.
1514 (*truncxf<mode>2_i387): Remove.
1515 (reg->reg splitters): Remove splitter pattern.
1516 (reg->mem splitters): Ditto.
1518 (truncdfsf2): New insn pattern.
1519 (truncxf<mode>2): Ditto.
1521 2018-09-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1523 * tree-ssa-math-opts.c (is_mult_by): New function.
1524 (is_square_of): Use the above.
1525 (optimize_recip_sqrt): New function.
1526 (pass_cse_reciprocals::execute): Use the above.
1528 2018-09-05 Richard Biener <rguenther@suse.de>
1531 * tree-ssa-sccvn.c (rpo_elim::eliminate_push_avail): Make sure
1532 to zero-init the emplaced vec.
1534 2018-09-05 Martin Liska <mliska@suse.cz>
1536 PR tree-optimization/87205
1537 * tree-switch-conversion.c (pass_lower_switch::execute):
1538 Group cases for switch statements.
1540 2018-09-05 Richard Biener <rguenther@suse.de>
1542 PR tree-optimization/87217
1543 * tree-ssa-sccvn.c (vuse_valueize): New.
1544 (vn_reference_lookup_pieces): Use it.
1545 (vn_reference_lookup): Likewise.
1547 2018-09-05 Nathan Sidwell <nathan@acm.org>
1550 * stor-layout.c (place_field): Scan forwards to check last
1551 bitfield when ms_bitfield_placement is in effect.
1553 2018-09-05 Richard Biener <rguenther@suse.de>
1556 * tree-vect-stmts.c (vectorizable_simd_clone_call): Fix bogus
1559 2018-09-05 Siddhesh Poyarekar <siddhesh@sourceware.org>
1560 Kugan Vivekanandarajah <kugan.vivekanandarajah@linaro.org>
1562 * config/aarch64/falkor-tag-collision-avoidance.c: New file.
1563 * config.gcc (extra_objs): Build it.
1564 * config/aarch64/t-aarch64 (falkor-tag-collision-avoidance.o):
1566 * config/aarch64/aarch64-passes.def
1567 (pass_tag_collision_avoidance): New pass.
1568 * config/aarch64/aarch64.c (qdf24xx_tunings): Add
1569 AARCH64_EXTRA_TUNE_RENAME_LOAD_REGS to tuning_flags.
1570 (aarch64_classify_address): Remove static qualifier.
1571 (aarch64_address_info, aarch64_address_type): Move to...
1572 * config/aarch64/aarch64-protos.h: ... here.
1573 (make_pass_tag_collision_avoidance): New function.
1574 * config/aarch64/aarch64-tuning-flags.def (rename_load_regs):
1577 2018-09-05 Martin Liska <mliska@suse.cz>
1579 * doc/gcov.texi: Update documentation of humar
1581 * gcov.c (format_count): Print one decimal place, it provides
1582 more fine number of situations like '1G' vs. '1.4G'.
1584 2018-09-05 Martin Liska <mliska@suse.cz>
1587 * config/rs6000/rs6000.opt: Mark the option as Deprecated.
1588 * optc-gen.awk: Allow 'Var' for Deprecated options in order
1589 to generate a MASK value.
1591 2018-09-04 H.J. Lu <hongjiu.lu@intel.com>
1594 * dwarf2out.c (based_loc_descr): Allow hard frame pointer even
1595 if frame pointer isn't used.
1596 (compute_frame_pointer_to_fb_displacement): Likewise.
1598 2018-09-04 Jakub Jelinek <jakub@redhat.com>
1601 * common/config/i386/i386-common.c (OPTION_MASK_ISA_XSAVEOPT_SET,
1602 OPTION_MASK_ISA_XSAVES_SET, OPTION_MASK_ISA_XSAVEC_SET): Use
1603 OPTION_MASK_ISA_XSAVE_SET instead of OPTION_MASK_ISA_XSAVE.
1604 (OPTION_MASK_ISA_XSAVE_UNSET): Add OPTION_MASK_ISA_XSAVES_UNSET
1605 and OPTION_MASK_ISA_XSAVEC_UNSET.
1607 2018-09-04 Max Filippov <jcmvbkbc@gmail.com>
1609 * config/xtensa/xtensa.c (xtensa_expand_atomic): Reorder AND and
1610 XOR operations in NAND case.
1612 2018-09-04 Aldy Hernandez <aldyh@redhat.com>
1614 * wide-int-range.cc (wide_int_range_convert): New.
1615 * wide-int-range.h (wide_int_range_convert): New.
1616 * tree-vrp.c (extract_range_from_unary_expr): Abstract wide int
1617 code into wide_int_range_convert.
1618 (extract_range_into_wide_ints): Do not munge anti range constants
1619 into the entire domain. Just return the range back.
1621 2018-09-04 Martin Liska <mliska@suse.cz>
1623 * genmatch.c (output_line_directive): Add new argument
1625 (dt_simplify::gen_1): Encapsulate dump within __builtin_expect.
1627 2018-09-04 Jonathan Wakely <jwakely@redhat.com>
1629 * doc/invoke.texi (Option Summary): Add whitespace.
1631 * doc/invoke.texi (Option Summary): Add -Waligned-new.
1633 2018-09-04 Richard Biener <rguenther@suse.de>
1635 PR tree-optimization/87211
1636 * tree-ssa-sccvn.c (visit_phi): When value-numbering to a
1637 backedge value we're supposed to treat as VARYING also number
1638 the PHI to VARYING in case it got a different value-number already.
1640 2018-09-04 Aldy Hernandez <aldyh@redhat.com>
1642 * tree-vrp.c (vrp_can_optimize_bit_op): Remove.
1643 (extract_range_from_binary_expr_1): Do not call
1644 vrp_can_optimize_bit_op.
1645 * wide-int-range.cc (wide_int_range_can_optimize_bit_op): Make
1647 (wide_int_range_get_mask_and_bounds): New.
1648 (wide_int_range_optimize_bit_op): New.
1649 (wide_int_range_bit_ior): Call wide_int_range_optimize_bit_op.
1650 (wide_int_range_bit_and): Same.
1651 * wide-int-range.h (wide_int_range_can_optimize_bit_op): Remove.
1652 (wide_int_range_optimize_bit_op): New.
1653 (wide_int_range_get_mask_and_bounds): New.
1655 2018-09-04 Richard Biener <rguenther@suse.de>
1657 PR tree-optimization/87176
1658 * tree-ssa-sccvn.c (visit_phi): Remove redundant allsame
1659 variable. When value-numbering a virtual PHI node make sure
1660 to not value-number to the backedge value.
1662 2018-09-04 Jonathan Wakely <jwakely@redhat.com>
1664 * doc/extend.texi (Long Long, Hex Floats): Document support for
1665 long long and hex floats in more recent versions of ISO C++.
1667 2018-09-03 Richard Biener <rguenther@suse.de>
1669 PR tree-optimization/87177
1670 * tree-ssa-sccvn.c (vuse_ssa_val): Revert previous change, keep
1673 2018-09-03 Alexander Monakov <amonakov@ispras.ru>
1675 * bb-reorder.c (edge_order): Convert to C-qsort-style
1676 tri-state comparator.
1677 (reorder_basic_blocks_simple): Change std::stable_sort to
1680 2018-09-03 Alexander Monakov <amonakov@ispras.ru>
1682 * tree-loop-distribution.c (offset_cmp): Convert to C-qsort-style
1683 tri-state comparator.
1684 (fuse_memset_builtins): Change std::stable_sort to gcc_stablesort.
1686 2018-09-03 Alexander Monakov <amonakov@ispras.ru>
1688 * sort.cc (struct sort_ctx): New field 'nlim'. Use it...
1689 (mergesort): ... here as maximum count for using netsort.
1690 (gcc_qsort): Set nlim to 3 if stable sort is requested.
1691 (gcc_stablesort): New.
1692 * system.h (gcc_stablesort): Declare.
1694 2018-09-03 Alexander Monakov <amonakov@ispras.ru>
1696 * sort.cc (gcc_qsort) [CHECKING_P]: Call qsort_chk.
1697 * system.h (qsort): Always redirect to gcc_qsort. Update comment.
1698 * vec.c (qsort_chk): Do not call gcc_qsort. Update comment.
1700 2018-09-03 Segher Boessenkool <segher@kernel.crashing.org>
1702 * config/rs6000/rs6000.md (*mov<mode>_hardfloat32): Remove %U from the
1703 lxsdx and stxsdx alternatives.
1704 (*mov<mode>_hardfloat64): Ditto.
1705 * config/rs6000/vsx.md (*vsx_extract_<mode>_store): Ditto.
1707 2018-09-03 Richard Biener <rguenther@suse.de>
1709 PR tree-optimization/87200
1710 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Valueize a
1713 2018-09-03 Martin Liska <mliska@suse.cz>
1715 PR tree-optimization/87201
1716 * tree-switch-conversion.c (switch_decision_tree::balance_case_nodes):
1717 Fix parenthesis in an expression.
1719 2018-09-03 Richard Biener <rguenther@suse.de>
1721 PR tree-optimization/87197
1722 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Mark the new def
1723 visited. CSE the VN_INFO hashtable lookup.
1725 PR tree-optimization/87169
1726 * tree-ssa-sccvn.c (do_rpo_vn): When marking loops for not
1727 iterating make sure there's no extra backedges from irreducible
1728 regions feeding the header. Mark the destination block
1731 2018-09-03 Martin Liska <mliska@suse.cz>
1734 * common/common-target.def: Add TARGET_GET_VALID_OPTION_VALUES.
1735 * common/common-targhooks.c (default_get_valid_option_values):
1737 * common/common-targhooks.h (default_get_valid_option_values):
1739 * common/config/i386/i386-common.c: Move processor_target_table
1741 (ix86_get_valid_option_values): New function.
1742 (TARGET_GET_VALID_OPTION_VALUES): New macro.
1743 * config/i386/i386.c (struct ptt): Move to i386-common.c.
1744 (PTA_*): Move all defined masks into i386-common.c.
1745 (ix86_function_specific_restore): Use new processor_cost_table.
1746 * config/i386/i386.h (struct ptt): Moved from i386.c.
1747 (struct pta): Likewise.
1748 * doc/tm.texi: Document new TARGET_GET_VALID_OPTION_VALUES.
1749 * doc/tm.texi.in: Likewise.
1750 * opt-suggestions.c (option_proposer::suggest_option):
1751 Pass prefix to build_option_suggestions.
1752 (option_proposer::get_completions): Likewise.
1753 (option_proposer::build_option_suggestions): Use the new target
1755 * opts.c (struct option_help_tuple): New struct.
1756 (print_filtered_help): Use the new target hook.
1758 2018-09-03 Martin Liska <mliska@suse.cz>
1761 * predict.c (set_even_probabilities): Add likely_edges
1762 argument and handle cases where we have precisely one
1764 (combine_predictions_for_bb): Catch also likely_edges.
1765 (tree_predict_by_opcode): Handle gswitch statements.
1766 * tree-cfg.h (find_case_label_for_value): New declaration.
1767 (find_taken_edge_switch_expr): Likewise.
1768 * tree-switch-conversion.c (switch_decision_tree::balance_case_nodes):
1769 Find pivot in decision tree based on probabily, not by number of
1772 2018-09-02 Gerald Pfeifer <gerald@pfeifer.com>
1774 * doc/standards.texi (Standards): Update Objective-C reference.
1776 2018-09-01 Gerald Pfeifer <gerald@pfeifer.com>
1778 * doc/install.texi (Prerequisites): Update link for MPC.
1780 2018-09-01 Michael Matz <matz@suse.de>
1782 PR tree-optimization/87074
1783 * gimple-loop-jam.c (unroll_jam_possible_p): Check loop exit
1784 PHIs for outer-loop uses.
1786 2018-09-01 Gerald Pfeifer <gerald@pfeifer.com>
1788 * doc/generic.texi (OpenMP): Adjust link to openmp.org.
1789 * doc/invoke.texi (C Dialect Options): Ditto.
1791 2018-09-01 Gerald Pfeifer <gerald@pfeifer.com>
1793 * doc/install.texi (Prerequisites): Adjust link mpfr.org.
1795 2018-08-31 Richard Biener <rguenther@suse.de>
1797 PR tree-optimization/87168
1798 * tree-ssa-sccvn.c (SSA_VAL): Add visited output parameter.
1799 (rpo_elim::eliminate_avail): When OP was not visited it must
1802 2018-08-31 David Malcolm <dmalcolm@redhat.com>
1804 * tree-vrp.c (copy_value_range): Convert param "from" from
1805 "value_range *" to "const value_range *".
1806 (range_is_null): Likewise for param "vr".
1807 (range_int_cst_p): Likewise.
1808 (range_int_cst_singleton_p): Likewise.
1809 (symbolic_range_p): Likewise.
1810 (value_ranges_intersect_p): Likewise for both params.
1811 (value_range_nonnegative_p): Likewise for param "vr".
1812 (value_range_constant_singleton): Likewise.
1813 (vrp_set_zero_nonzero_bits): Likewise for param "ar".
1814 (extract_range_into_wide_ints): Likewise for param "vr".
1815 (extract_range_from_multiplicative_op): Likewise for params "vr0"
1817 (vrp_can_optimize_bit_op): Likewise.
1818 (extract_range_from_binary_expr_1): Likewise for params "vr0_" and
1820 (extract_range_from_unary_expr): Likewise.
1821 (debug_value_range): Likewise for param "vr".
1822 (value_range::dump): Add "const" qualifier.
1823 (vrp_prop::check_array_ref): Convert local "vr" from
1824 "value_range *" to "const value_range *".
1825 (vrp_prop::check_mem_ref): Likewise.
1826 (vrp_prop::visit_stmt): Likewise for local "old_vr".
1827 (vrp_intersect_ranges_1): Likewise for param "vr_1".
1828 (vrp_intersect_ranges): Likewise.
1829 (simplify_stmt_for_jump_threading): Likewise for local "vr".
1830 (vrp_prop::vrp_finalize): Likewise.
1831 * tree-vrp.h (value_range::dump): Add "const" qualifier.
1832 (vrp_intersect_ranges): Add "const" qualifier to params as above.
1833 (extract_range_from_unary_expr): Likewise.
1834 (value_range_constant_singleton): Likewise.
1835 (symbolic_range_p): Likewise.
1836 (copy_value_range): Likewise.
1837 (extract_range_from_binary_expr_1): Likewise.
1838 (range_int_cst_p): Likewise.
1839 (vrp_set_zero_nonzero_bits): Likewise.
1840 (range_int_cst_singleton_p): Likewise.
1842 2018-08-31 Vlad Lazar <vlad.lazar@arm.com>
1844 * config/aarch64/arm_neon.h (vabsd_s64): New.
1845 (vnegd_s64): Likewise.
1847 2018-08-31 Martin Jambor <mjambor@suse.cz>
1849 * ipa-cp.c (estimate_local_effects): Replace wrong MAX with MIN.
1851 2018-08-31 Martin Liska <mliska@suse.cz>
1853 * ipa-icf.c (sem_item::add_type): Use
1854 sem_item::m_type_hash_cache.
1855 * ipa-icf.h: Move the cache from sem_item_optimizer
1858 2018-08-31 Nathan Sidwell <nathan@acm.org>
1860 * doc/extend.texi (Backwards Compatibility): Remove implicit
1861 extern C leeway of () being (...).
1863 2018-08-31 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1865 * ipa-inline.c (can_inline_edge_by_limits_p): Fix typos in comment.
1867 2018-08-31 Segher Boessenkool <segher@kernel.crashing.org>
1871 * config/rs6000/rs6000.md (lround<mode>di2): Gate on TARGET_FPRND.
1873 2018-08-31 Jakub Jelinek <jakub@redhat.com>
1876 * expmed.c (expand_mult_const): Use immed_wide_int_const instead of
1877 gen_int_mode. Formatting fixes.
1879 2018-08-30 Sandra Loosemore <sandra@codesourcery.com>
1881 * target.def (custom_function_descriptors): Improve documentation.
1882 * doc/tm.texi.in (Trampolines): Expand discussion of function
1883 descriptors and move TARGET_CUSTOM_FUNCTION_DESCRIPTORS to the
1884 beginning of the section.
1885 * doc/tm.texi: Regenerated.
1887 2018-08-30 Jose E. Marchesi <jose.marchesi@oracle.com>
1889 * cfg.h (class auto_edge_flag): Spell out the template-id of the
1890 base class in the initializer list. This is a workaround for
1891 building with older GCC.
1892 (class auto_bb_flag): Likewise.
1894 2018-08-30 Aaron Sawdey <acsawdey@linux.ibm.com>
1896 * config/rs6000/altivec.md (altivec_eq<mode>): Remove star.
1897 (altivec_vcmpequ<VI_char>_p): Remove star.
1898 * config/rs6000/rs6000-string.c (do_load_for_compare): Support
1900 (expand_strncmp_vec_sequence): New function.
1901 (emit_final_str_compare_vec): New function.
1902 (expand_strn_compare): Add support for vector strncmp.
1903 * config/rs6000/rs6000.opt (-mstring-compare-inline-limit): Change
1904 length specification to bytes.
1905 * config/rs6000/vsx.md (vsx_ld_elemrev_v16qi_internal): Remove star.
1906 (vcmpnezb_p): New pattern.
1907 * doc/invoke.texi (RS/6000 and PowerPC Options): Update documentation
1908 for option -mstring-compare-inline-limit.
1910 2018-08-30 Thiago Macieira <thiago.macieira@intel.com>
1912 * config/i386/i386.c (PTA_WESTMERE): Remove PTA_AES.
1913 (PTA_SKYLAKE): Add PTA_AES.
1914 (PTA_GOLDMONT): Likewise.
1916 2018-08-29 Jan Hubicka <jh@suse.cz>
1919 * lto-opts.c (lto_write_options): Always stream PIC/PIE mode.
1920 * lto-wrapper.c (merge_and_complain): Fix merging of PIC/PIE.
1922 2018-08-29 Jan Hubicka <jh@suse.cz>
1924 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not follow
1926 (hash_tree): Do not visit TYPE_STUB_DECL.
1927 * tree-streamer-out.c (write_ts_type_common_tree_pointers): Do not
1928 stream TYPE_STUB_DECL.
1929 * tree-streamer-in.c (lto_input_ts_type_common_tree_pointers): Likewise.
1930 * ipa-utils.h (type_with_linkage_p): Do not rely on TYPE_STUB_DECL
1931 after free_lang_data.
1932 (type_in_anonymous_namespace_p): Likewise.
1934 2018-08-29 Jan Hubicka <jh@suse.cz>
1936 * sreal.h (SREAL_PART_BITS): Change to 31; remove seemingly unnecessary
1937 comment that it has to be even number.
1938 (class sreal): Change m_sig type to int32_t.
1939 * sreal.c (sreal::dump, sreal::to_int, opreator+, operator-): Use
1940 int64_t for temporary calculations.
1941 (sreal_verify_basics): Drop one bit from minimum and maximum.
1943 2018-08-30 Richard Biener <rguenther@suse.de>
1945 PR tree-optimization/87147
1946 * tree-ssa-sccvn.c (SSA_VISITED): New function.
1947 (visit_phi): When the degenerate result is from the backedge and
1948 we didn't visit its definition yet drop to VARYING.
1949 (do_rpo_vn): Properly mark blocks with incoming backedges as executable.
1951 2018-08-29 Jan Hubicka <jh@suse.cz>
1953 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not walk
1955 (hash_tree): Likewise.
1957 2018-08-29 Jan Hubicka <jh@suse.cz>
1959 * tree.c (find_decls_types_r): Walk also TYPE_NEXT_PTR_TO
1960 and TYPE_NEXT_REF_TO.
1962 2018-08-29 Jan Hubicka <jh@suse.cz>
1964 * sreal.h (SREAL_PART_BITS): Change to 31; remove seemingly unnecessary
1965 comment that it has to be even number.
1966 (class sreal): Change m_sig type to int32_t.
1967 * sreal.c (sreal::dump, sreal::to_int, opreator+, operator-): Use
1968 int64_t for temporary calculations.
1969 (sreal_verify_basics): Drop one bit from minimum and maximum.
1971 2018-08-30 Tamar Christina <tamar.christina@arm.com>
1973 * config/aarch64/aarch64.c (aarch64_expand_movmem): Set TImode max.
1975 2018-08-30 Vlad Lazar <vlad.lazar@arm.com>
1978 * expmed.c (canonicalize_comparison): Use wi::sub instead of wi::add
1979 if to_add is negative.
1981 2018-08-29 Bernd Edlinger <bernd.edlinger@hotmail.de>
1984 * builtins.c (c_strlen): Improve range checks.
1986 2018-08-29 Martin Sebor <msebor@redhat.com>
1987 Jeff Law <law@redhat.com>
1989 PR tree-optimization/86714
1990 PR tree-optimization/86711
1991 * builtins.c (c_strlen): Add arguments to call to string_constant.
1992 * expr.c (string_constant): Add argument. Detect missing nul
1993 terminator and outermost declaration it's missing in.
1994 * expr.h (string_constant): Add argument.
1995 * fold-const.c (read_from_constant_string): Add arguments to call to
1997 (c_getstr): Likewise.
1998 * tree-ssa-forwprop.c (simplify_builtin_call): Likewise.
2000 * tree-ssa-strlen.c (get_stridx): Likewise.
2002 2018-08-29 Jan Hubicka <jh@suse.cz>
2004 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers):
2005 Do not stream DECL_VINDEX.
2006 * tree-streamer-out.c (write_ts_function_decl_tree_pointers): Likewise.
2007 * tree.c (free_lang_data_in_decl): Clear DECL_VINDEX.
2008 (decl_function_context): Use DECL_VIRTUAL_P rather than DECL_VINDEX.
2010 2018-08-29 Richard Biener <rguenther@suse.de>
2012 * tree-ssa-sccvn.c (vuse_ssa_val): Return NULL for unvisited
2013 virtual operands that are not default defs to honor region
2015 (rpo_vn_valueize): Remove ineffective code here.
2017 2018-08-29 Richard Biener <rguenther@suse.de>
2019 PR tree-optimization/87132
2020 * tree-ssa-alias.c (get_continuation_for_phi): Do not translate
2021 when skipping defs reachable over backedges.
2023 2018-08-29 Richard Biener <rguenther@suse.de>
2025 * tree-core.h: Document use of deprecated_flag in SSA_NAME.
2026 * tree.h (SSA_NAME_POINTS_TO_READONLY_MEMORY): Define.
2027 * tree-into-ssa.c (pass_build_ssa::execute): Initialize
2028 function parameters SSA_NAME_POINTS_TO_READONLY_MEMORY from fnspec.
2029 * tree-ssa-sccvn.c (const_parms, init_const_parms): Remove.
2030 (vn_reference_lookup_3): Remove use of const_parms.
2031 (free_rpo_vn): Do not free const_parms.
2032 (do_rpo_vn): Do not call init_const_parms.
2033 * tree-ssa-alias.c (refs_may_alias_p_1): Honor
2034 SSA_NAME_POINTS_TO_READONLY_MEMORY.
2035 (call_may_clobber_ref_p_1): Likewise.
2037 2018-08-29 Alexander Monakov <amonakov@ispras.ru>
2040 * invoke.texi (Optimization Options): List -ftree-scev-cprop.
2042 (-ftree-scev-cprop): Document.
2044 2018-08-29 Jan Hubicka <jh@suse.cz>
2046 * sreal.h (normalize, normalize_up, normalize_down): Add new_sig/new_exp
2048 (sreal constructor): Update.
2049 * sreal.c (sreal:operator+, sreal:operator-, sreal:operator*,
2050 sreal:operator/): Update.
2052 2018-08-29 Martin Liska <mliska@suse.cz>
2054 * tree-switch-conversion.c (switch_conversion::expand):
2055 Strenghten assumption about gswitch statements.
2057 2018-08-29 Richard Biener <rguenther@suse.de>
2059 PR tree-optimization/87117
2060 * tree-ssa-sccvn.c (eliminate_dom_walker::eliminate_stmt): Only
2061 re-value-number released SSA VDEFs.
2063 2018-08-29 Richard Biener <rguenther@suse.de>
2065 PR tree-optimization/87126
2066 * tree-ssa-sccvn.c (vn_reference_insert): Remove assert.
2068 2018-08-28 Jim Wilson <jimw@sifive.com>
2070 * config/riscv/pic.md: Rewrite.
2071 * config/riscv/riscv.c (riscv_address_insns): Return cost of 3 for
2073 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Delete.
2074 (SOFTF, default_load, softload, softstore): New.
2076 2018-08-28 Jeff Law <law@redhat.com>
2078 * fold-const.c (fold_binary_loc): Remove recently added assert.
2080 2018-08-28 Joern Rennecke <joern.rennecke@riscy-ip.com>
2082 * genpreds.c (write_predicate_subfunction): Also add ATTRIBUTE_UNUSED
2083 to OP parmeter of generated function.
2085 2018-08-28 MCC CS <deswurstes@users.noreply.github.com>
2087 PR tree-optimization/87009
2088 * match.pd: Add boolean optimizations.
2090 2018-08-28 Martin Sebor <msebor@redhat.com>
2093 * calls.c (alloc_max_size): Treat HOST_WIDE_INT special.
2094 * gimple-ssa-warn-alloca.c (adjusted_warn_limit): New function.
2095 (pass_walloca::gate): Use it.
2096 (alloca_call_type): Same.
2097 (pass_walloca::execute): Same.
2098 * stor-layout.c (layout_decl): Treat HOST_WIDE_INT special.
2100 2018-08-28 David Malcolm <dmalcolm@redhat.com>
2102 * dumpfile.h (ATTRIBUTE_GCC_DUMP_PRINTF): Change version check on
2103 GCC_VERSION for usage of "__gcc_dump_printf__" format from
2106 2018-08-28 Richard Biener <rguenther@suse.de>
2108 PR tree-optimization/87124
2109 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Guard against
2110 constants before looking up avail.
2112 2018-08-28 Jakub Jelinek <jakub@redhat.com>
2115 * calls.c (maybe_warn_nonstring_arg): Punt early if
2116 warn_stringop_overflow is zero. Don't call get_range_strlen
2117 on 3rd argument, keep iterating until lenrng[1] is INTEGER_CST.
2118 Swap comparison operands to have constants on rhs. Only use
2119 lenrng[1] if non-NULL and INTEGER_CST. Don't uselessly
2120 increment lenrng[0].
2122 2018-08-28 Richard Sandiford <richard.sandiford@arm.com>
2124 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Fix unguarded
2125 use of tree_to_shwi. Remove duplicated test for the size being
2126 a whole number of bytes.
2128 2018-08-28 Richard Biener <rguenther@suse.de>
2130 PR tree-optimization/87117
2131 * tree-ssa-sccvn.c (eliminate_dom_walker::eliminate_cleanup):
2132 Handle removed stmt without LHS (GIMPLE_NOP).
2134 2018-08-28 Richard Biener <rguenther@suse.de>
2136 PR tree-optimization/87117
2137 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Exclude
2138 void which is is_gimple_reg_type by checking for COMPLETE_TYPE_P.
2140 2018-08-28 Richard Biener <rguenther@suse.de>
2142 PR tree-optimization/87117
2143 * tree-ssa-pre.c (compute_avail): Do not make expressions
2144 with predicated values available.
2145 (get_expr_value_id): Assert we do not run into predicated value
2148 2018-08-28 Richard Biener <rguenther@suse.de>
2150 PR tree-optimization/87117
2151 * tree-ssa-operands.c (add_stmt_operand): STRING_CST may
2152 get virtual operands.
2153 (get_expr_operands): Handle STRING_CST like other decls.
2155 2018-08-28 Martin Liska <mliska@suse.cz>
2157 * tree.h: Update documentation of fndecl_built_in_p
2161 2018-08-27 Jeff Law <law@redhat.com>
2162 PR tree-optimization/87110
2163 * tree-ssa-dse.c (compute_trims): Handle non-constant
2166 2018-08-27 Martin Sebor <msebor@redhat.com>
2168 PR tree-optimization/86914
2169 * tree-ssa-strlen.c (maybe_set_strlen_range): Avoid MEM_REF.
2171 2018-08-27 Martin Sebor <msebor@redhat.com>
2173 PR tree-optimization/87112
2174 * builtins.c (expand_builtin_strnlen): Convert c_strlen result to
2175 the type of the bound argument.
2177 2018-08-27 Jeff Law <law@redhat.com>
2179 * tree-ssa-dse.c (compute_trims): Handle case where the reference's
2180 type does not have a TYPE_SIZE_UNIT.
2182 2018-08-27 Steve Ellcey <sellcey@cavium.com>
2184 * config/aarch64/aarch64-speculation.cc: Replace include of cfg.h
2185 with include of backend.h.
2187 2018-08-27 Richard Biener <rguenther@suse.de>
2189 PR tree-optimization/86927
2190 * tree-vect-loop.c (vect_create_epilog_for_reduction): Properly
2191 use const cond reduction code.
2193 2018-08-27 Alexander Monakov <amonakov@ispras.ru>
2195 PR tree-optimization/85758
2196 * match.pd ((X & Y) ^ Y): Add :s qualifier to inner expression.
2198 2018-08-27 David Malcolm <dmalcolm@redhat.com>
2201 * diagnostic-show-locus.c (class layout_range): Update for
2202 conversion of show_caret_p to a tri-state.
2203 (layout_range::layout_range): Likewise.
2204 (make_range): Likewise.
2205 (layout::maybe_add_location_range): Likewise.
2206 (layout::should_print_annotation_line_p): Don't show annotation
2207 lines for ranges that are SHOW_LINES_WITHOUT_RANGE.
2208 (layout::get_state_at_point): Update for conversion of
2209 show_caret_p to a tri-state. Bail out early for
2210 SHOW_LINES_WITHOUT_RANGE, so that such ranges don't affect
2211 underlining or source colorization.
2212 (gcc_rich_location::add_location_if_nearby): Update for conversion
2213 of show_caret_p to a tri-state.
2214 (selftest::test_one_liner_multiple_carets_and_ranges): Likewise.
2215 (selftest::test_one_liner_fixit_replace_equal_secondary_range):
2217 (selftest::test_one_liner_labels): Likewise.
2218 * gcc-rich-location.c (gcc_rich_location::add_expr): Update for
2219 conversion of show_caret_p to a tri-state.
2220 * pretty-print.c (text_info::set_location): Likewise.
2221 * pretty-print.h (text_info::set_location): Likewise.
2222 * substring-locations.c (format_warning_n_va): Likewise.
2223 * tree-diagnostic.c (default_tree_printer): Likewise.
2224 * tree-pretty-print.c (newline_and_indent): Likewise.
2226 2018-08-27 David Malcolm <dmalcolm@redhat.com>
2229 * diagnostic-show-locus.c (get_line_span_for_fixit_hint): Show the
2230 line above for line-insertion fix-it hints.
2231 (selftest::test_fixit_insert_containing_newline): Update the
2232 expected results, and add a test with line-numbering enabled.
2234 2018-08-27 Martin Liska <mliska@suse.cz>
2237 * sanopt.c (sanitize_rewrite_addressable_params): Ignore
2238 params with DECL_HAS_VALUE_EXPR_P.
2240 2018-08-27 Martin Liska <mliska@suse.cz>
2242 * config/i386/i386.c (ix86_expand_set_or_movmem): Dump
2243 selected expansion strategy.
2245 2018-08-27 Martin Liska <mliska@suse.cz>
2247 * builtins.h (is_builtin_fn): Remove and fndecl_built_in_p.
2248 * builtins.c (is_builtin_fn): Likewise.
2249 * attribs.c (diag_attr_exclusions): Use new function
2250 fndecl_built_in_p and remove check for FUNCTION_DECL if
2252 (builtin_mathfn_code): Likewise.
2253 (fold_builtin_expect): Likewise.
2254 (fold_call_expr): Likewise.
2255 (fold_builtin_call_array): Likewise.
2256 (fold_call_stmt): Likewise.
2257 (set_builtin_user_assembler_name): Likewise.
2258 (is_simple_builtin): Likewise.
2259 * calls.c (gimple_alloca_call_p): Likewise.
2260 (maybe_warn_nonstring_arg): Likewise.
2261 * cfgexpand.c (expand_call_stmt): Likewise.
2262 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Likewise.
2263 (cgraph_edge::verify_corresponds_to_fndecl): Likewise.
2264 (cgraph_node::verify_node): Likewise.
2265 * cgraphclones.c (build_function_decl_skip_args): Likewise.
2266 (cgraph_node::create_clone): Likewise.
2267 * config/arm/arm.c (arm_insert_attributes): Likewise.
2268 * config/i386/i386.c (ix86_gimple_fold_builtin): Likewise.
2269 * dse.c (scan_insn): Likewise.
2270 * expr.c (expand_expr_real_1): Likewise.
2271 * fold-const.c (operand_equal_p): Likewise.
2272 (fold_binary_loc): Likewise.
2273 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
2274 * gimple-low.c (lower_stmt): Likewise.
2275 * gimple-pretty-print.c (dump_gimple_call): Likewise.
2276 * gimple-ssa-warn-restrict.c (wrestrict_dom_walker::check_call):
2278 * gimple.c (gimple_build_call_from_tree): Likewise.
2279 (gimple_call_builtin_p): Likewise.
2280 (gimple_call_combined_fn): Likewise.
2281 * gimplify.c (gimplify_call_expr): Likewise.
2282 (gimple_boolify): Likewise.
2283 (gimplify_modify_expr): Likewise.
2284 (gimplify_addr_expr): Likewise.
2285 * hsa-gen.c (gen_hsa_insns_for_call): Likewise.
2286 * ipa-cp.c (determine_versionability): Likewise.
2287 * ipa-fnsummary.c (compute_fn_summary): Likewise.
2288 * ipa-param-manipulation.c (ipa_modify_formal_parameters): Likewise.
2289 * ipa-split.c (visit_bb): Likewise.
2290 (split_function): Likewise.
2291 * ipa-visibility.c (cgraph_externally_visible_p): Likewise.
2292 * lto-cgraph.c (input_node): Likewise.
2293 * lto-streamer-out.c (write_symbol): Likewise.
2294 * omp-low.c (setjmp_or_longjmp_p): Likewise.
2295 (lower_omp_1): Likewise.
2296 * predict.c (strip_predict_hints): Likewise.
2297 * print-tree.c (print_node): Likewise.
2298 * symtab.c (symtab_node::output_to_lto_symbol_table_p): Likewise.
2299 * trans-mem.c (is_tm_irrevocable): Likewise.
2300 (is_tm_load): Likewise.
2301 (is_tm_simple_load): Likewise.
2302 (is_tm_store): Likewise.
2303 (is_tm_simple_store): Likewise.
2304 (is_tm_abort): Likewise.
2305 (tm_region_init_1): Likewise.
2306 * tree-call-cdce.c (gen_shrink_wrap_conditions): Likewise.
2307 * tree-cfg.c (verify_gimple_call): Likewise.
2308 (move_stmt_r): Likewise.
2309 (stmt_can_terminate_bb_p): Likewise.
2310 * tree-eh.c (lower_eh_constructs_2): Likewise.
2311 * tree-if-conv.c (if_convertible_stmt_p): Likewise.
2312 * tree-inline.c (remap_gimple_stmt): Likewise.
2313 (copy_bb): Likewise.
2314 (estimate_num_insns): Likewise.
2315 (fold_marked_statements): Likewise.
2316 * tree-sra.c (scan_function): Likewise.
2317 * tree-ssa-ccp.c (surely_varying_stmt_p): Likewise.
2318 (optimize_stack_restore): Likewise.
2319 (pass_fold_builtins::execute): Likewise.
2320 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
2321 (mark_all_reaching_defs_necessary_1): Likewise.
2322 * tree-ssa-dom.c (dom_opt_dom_walker::optimize_stmt): Likewise.
2323 * tree-ssa-forwprop.c (simplify_builtin_call): Likewise.
2324 (pass_forwprop::execute): Likewise.
2325 * tree-ssa-loop-im.c (stmt_cost): Likewise.
2326 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
2327 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Likewise.
2328 * tree-ssa-strlen.c (get_string_length): Likewise.
2329 * tree-ssa-structalias.c (handle_lhs_call): Likewise.
2330 (find_func_aliases_for_call): Likewise.
2331 * tree-ssa-ter.c (find_replaceable_in_bb): Likewise.
2332 * tree-stdarg.c (optimize_va_list_gpr_fpr_size): Likewise.
2333 * tree-tailcall.c (find_tail_calls): Likewise.
2334 * tree.c (need_assembler_name_p): Likewise.
2335 (free_lang_data_in_decl): Likewise.
2336 (get_call_combined_fn): Likewise.
2337 * ubsan.c (is_ubsan_builtin_p): Likewise.
2338 * varasm.c (incorporeal_function_p): Likewise.
2339 * tree.h (DECL_BUILT_IN): Remove and replace with
2341 (DECL_BUILT_IN_P): Transfort to fndecl_built_in_p.
2342 (fndecl_built_in_p): New.
2344 2018-08-27 Martin Liska <mliska@suse.cz>
2346 PR tree-optimization/86847
2347 * tree-switch-conversion.c (switch_decision_tree::dump_case_nodes):
2348 Dump also subtree probability.
2349 (switch_decision_tree::do_jump_if_equal): New function.
2350 (switch_decision_tree::emit_case_nodes): Handle special
2351 situations in balanced tree that can be emitted much simpler.
2352 Fix calculation of probabilities that happen in tree expansion.
2353 * tree-switch-conversion.h (struct cluster): Add
2355 (struct simple_cluster): Likewise.
2356 (struct case_tree_node): Add new function has_child.
2357 (do_jump_if_equal): New.
2359 2018-08-27 Martin Liska <mliska@suse.cz>
2361 * tree-switch-conversion.c (bit_test_cluster::find_bit_tests):
2362 Add new argument to bit_test_cluster constructor.
2363 (bit_test_cluster::emit): Set bits really number of values
2365 (bit_test_cluster::hoist_edge_and_branch_if_true): Add
2366 probability argument.
2367 * tree-switch-conversion.h (struct bit_test_cluster):
2368 Add m_handles_entire_switch.
2370 2018-08-27 Martin Liska <mliska@suse.cz>
2372 PR tree-optimization/86702
2373 * tree-switch-conversion.c (jump_table_cluster::emit):
2374 Make probabilities even for values in jump table
2375 according to number of cases handled.
2376 (switch_decision_tree::compute_cases_per_edge): Pass
2377 argument to reset_out_edges_aux function.
2378 (switch_decision_tree::analyze_switch_statement): Likewise.
2379 * tree-switch-conversion.h (switch_decision_tree::reset_out_edges_aux):
2382 2018-08-27 Martin Liska <mliska@suse.cz>
2384 * cfgexpand.c (expand_asm_stmt): Use label_to_block and pass
2385 cfun argument explicitly.
2386 * gimple-pretty-print.c (dump_gimple_switch): Likewise.
2387 * hsa-gen.c (gen_hsa_insns_for_switch_stmt): Use new
2388 function gimple_switch_default_bb.
2389 (convert_switch_statements):
2391 * ipa-fnsummary.c (set_switch_stmt_execution_predicate):
2392 * stmt.c (label_to_block_fn): Use label_to_block and pass
2393 cfun argument explicitly and use gimple_switch_label_bb.
2394 (expand_case): Likewise.
2395 * tree-cfg.c (lower_phi_internal_fn): Use label_to_block and pass
2396 cfun argument explicitly. Likewise.
2397 (make_edges_bb): Likewise.
2398 (make_cond_expr_edges): Likewise.
2399 (get_cases_for_edge): Likewise.
2400 (make_gimple_switch_edges): Likewise.
2401 (label_to_block_fn): Likewise.
2402 (label_to_block): Likewise.
2403 (make_goto_expr_edges): Likewise.
2404 (make_gimple_asm_edges): Likewise.
2405 (main_block_label): Likewise.
2406 (group_case_labels_stmt): Likewise.
2407 (find_taken_edge_computed_goto): Likewise.
2408 (find_taken_edge_switch_expr): Likewise.
2409 (gimple_verify_flow_info): Likewise.
2410 (gimple_redirect_edge_and_branch): Likewise.
2411 (gimple_switch_label_bb): New function.
2412 (gimple_switch_default_bb): Likewise.
2413 (gimple_switch_edge): Likewise.
2414 (gimple_switch_default_edge): Likewise.
2415 * tree-cfg.h (label_to_block_fn): Remove and replace ...
2416 (label_to_block): ... with this.
2417 (gimple_switch_label_bb): New.
2418 (gimple_switch_default_bb): Likewise.
2419 (gimple_switch_edge): Likewise.
2420 (gimple_switch_default_edge): Likewise.
2421 * tree-cfgcleanup.c (convert_single_case_switch): Use
2422 new gimple functions and pass new argument to label_to_block.
2423 (cleanup_control_flow_bb):
2424 * tree-eh.c (make_eh_dispatch_edges): Use label_to_block and pass
2425 cfun argument explicitly.
2426 (make_eh_edges): Likewise.
2427 (redirect_eh_dispatch_edge): Likewise.
2428 (lower_resx): Likewise.
2429 (lower_eh_dispatch): Likewise.
2430 (maybe_remove_unreachable_handlers): Likewise.
2431 (unsplit_eh): Likewise.
2432 (cleanup_empty_eh): Likewise.
2433 (verify_eh_edges): Likewise.
2434 (verify_eh_dispatch_edge): Likewise.
2435 * tree-ssa-dom.c (record_edge_info): Likewise.
2436 * tree-ssa-forwprop.c (simplify_gimple_switch_label_vec): Likewise.
2437 * tree-ssa-threadedge.c (thread_around_empty_blocks): Likewise.
2438 (thread_through_normal_block): Likewise.
2439 * tree-ssa-uncprop.c (associate_equivalences_with_edges): Likewise.
2440 * tree-ssa-uninit.c (convert_control_dep_chain_into_preds):
2441 * tree-switch-conversion.c (switch_conversion::collect): Use new
2443 (switch_conversion::check_final_bb): Likewise.
2444 (switch_conversion::gather_default_values): Pass new argument
2446 (switch_conversion::build_constructors): Likewise.
2447 (switch_decision_tree::compute_cases_per_edge): Use new
2448 gimple_switch_edge function.
2449 (switch_decision_tree::analyze_switch_statement): Pass new argument
2451 (switch_decision_tree::try_switch_expansion): Use
2452 gimple_switch_default_edge.
2453 * tree-vrp.c (find_switch_asserts): Pass new argument
2455 * vr-values.c (vr_values::vrp_visit_switch_stmt): Likewise.
2456 (vr_values::simplify_switch_using_ranges): Likewise.
2458 2018-08-27 Richard Biener <rguenther@suse.de>
2460 * cfganal.h (rev_post_order_and_mark_dfs_back_seme): Declare.
2461 * cfganal.c (rev_post_order_and_mark_dfs_back_seme): New function.
2463 * tree-ssa-sccvn.h (struct vn_pval): New structure.
2464 (struct vn_nary_op_s): Add unwind_to member. Add
2465 predicated_values flag and put result into a union together
2466 with a linked list of vn_pval.
2467 (struct vn_ssa_aux): Add name member to make maintaining
2468 a map of SSA name to vn_ssa_aux possible. Remove no longer
2469 needed info, dfsnum, low, visited, on_sccstack, use_processed
2470 and range_info_anti_range_p members.
2471 (run_scc_vn, vn_eliminate, free_scc_vn, vn_valueize): Remove.
2472 (do_rpo_vn, run_rpo_vn, eliminate_with_rpo_vn, free_rpo_vn):
2474 (vn_valueize): New global.
2475 (vn_context_bb): Likewise.
2476 (VN_INFO_RANGE_INFO, VN_INFO_ANTI_RANGE_P, VN_INFO_RANGE_TYPE,
2477 VN_INFO_PTR_INFO): Remove.
2478 * tree-ssa-sccvn.c: ... (rewrite)
2479 (pass_fre::execute): For -O2+ initialize loops and run
2480 RPO VN in optimistic mode (iterating). For -O1 and -Og
2481 run RPO VN in non-optimistic mode.
2482 * params.def (PARAM_SCCVN_MAX_SCC_SIZE): Remove.
2483 (PARAM_RPO_VN_MAX_LOOP_DEPTH): Add.
2484 * doc/invoke.texi (sccvn-max-scc-size): Remove.
2485 (rpo-vn-max-loop-depth): Document.
2486 * tree-ssa-alias.c (walk_non_aliased_vuses): Stop walking
2487 when valuezing the VUSE signals we walked out of the region.
2488 * tree-ssa-pre.c (phi_translate_1): Ignore predicated values.
2489 (phi_translate): Set VN context block to use for availability
2491 (compute_avail): Likewise.
2492 (pre_valueize): New function.
2493 (pass_pre::execute): Adjust to the RPO VN API.
2495 * tree-ssa-loop-ivcanon.c: Include tree-ssa-sccvn.h.
2496 (propagate_constants_for_unrolling): Remove.
2497 (tree_unroll_loops_completely): Perform value-numbering
2498 on the unrolled bodies loop parent.
2500 2018-08-27 Richard Biener <rguenther@suse.de>
2502 * tree-ssa-pre.c (compute_antic): Re-use inverted postorder
2503 for partial antic compute.
2505 2018-08-27 Jakub Jelinek <jakub@redhat.com>
2507 PR rtl-optimization/87065
2508 * combine.c (simplify_if_then_else): Formatting fix.
2509 (if_then_else_cond): Guard MULT optimization with SCALAR_INT_MODE_P
2511 (known_cond): Don't return const_true_rtx for vector modes. Use
2512 CONST0_RTX instead of const0_rtx. Formatting fixes.
2514 2018-08-27 Martin Liska <mliska@suse.cz>
2516 PR gcov-profile/87069
2517 * gcov.c (process_file): Record files already processed
2518 and warn about a file being processed multiple times.
2520 2018-08-27 Martin Liska <mliska@suse.cz>
2523 * config/aarch64/aarch64.c (aarch64_override_options_internal):
2524 Set default values for x_aarch64_*_string strings.
2525 * config/aarch64/aarch64.opt: Remove --{march,mcpu,mtune}==
2526 prefix. For -mabi do not print '=ABI' in help and use
2527 <option_value> format for -msve-vector-bits and -moverride
2530 2018-08-26 Jeff Law <law@redhat.com>
2532 * config/mips/frame-header-opt.c: Include "backend.h" rather than
2535 2018-08-26 Marek Polacek <polacek@redhat.com>
2537 PR c++/87029, Implement -Wredundant-move.
2538 * doc/invoke.texi: Document -Wredundant-move.
2540 2018-08-25 Martin Sebor <msebor@redhat.com>
2542 PR tree-optimization/87059
2543 * builtins.c (expand_builtin_strncmp): Convert MIN_EXPR operand
2544 to the same type as the other.
2545 * fold-const.c (fold_binary_loc): Assert expectation.
2547 2018-08-25 Iain Sandoe <iain@sandoe.co.uk>
2549 * config/darwin.c (machopic_legitimize_pic_address): Clean up
2550 extraneous parentheses, dead code section and formatting.
2552 2018-08-24 David Malcolm <dmalcolm@redhat.com>
2555 * diagnostic-show-locus.c (layout::layout): Ensure the margin is
2556 wide enough for jumps in the line-numbering to be visible.
2557 (layout::print_gap_in_line_numbering): New member function.
2558 (layout::calculate_line_spans): When using line numbering, merge
2559 line spans that are only 1 line apart.
2560 (diagnostic_show_locus): When printing line numbers, show gaps in
2561 line numbering directly, rather than printing headers.
2562 (selftest::test_diagnostic_show_locus_fixit_lines): Add test of
2563 line-numbering with multiple line spans.
2564 (selftest::test_fixit_insert_containing_newline_2): Add test of
2565 line-numbering, in which the spans are close enough to be merged.
2567 2018-08-24 Aldy Hernandez <aldyh@redhat.com>
2569 * gimple-ssa-evrp-analyze.c (set_ssa_range_info): Pass value_range
2570 to range_includes_zero_p. Do not special case VR_ANTI_RANGE.
2571 * tree-vrp.c (range_is_nonnull): Remove.
2572 (range_includes_zero_p): Accept value_range instead of min/max.
2573 (extract_range_from_binary_expr_1): Do not early bail on
2575 Use range_includes_zero_p instead of range_is_nonnull.
2576 (extract_range_from_unary_expr): Use range_includes_zero_p instead
2577 of range_is_nonnull.
2578 (vrp_meet_1): Pass value_range to range_includes_zero_p. Do not
2579 special case VR_ANTI_RANGE.
2580 (vrp_finalize): Same.
2581 * tree-vrp.h (range_includes_zero_p): Pass value_range as argument
2583 (range_is_nonnull): Remove.
2584 * vr-values.c (vrp_stmt_computes_nonzero): Use
2585 range_includes_zero_p instead of range_is_nonnull.
2586 (extract_range_basic): Pass value_range to range_includes_zero_p
2587 instead of range_is_nonnull.
2589 2018-08-24 Uros Bizjak <ubizjak@gmail.com>
2591 * emit-rtl.c (init_emit_once): Do not emit MODE_POINTER_BOUNDS RTXes.
2592 * emit-rtl.h (rtl_data): Remove return_bnd.
2593 * explow.c (trunc_int_for_mode): Do not handle POINTER_BOUNDS_MODE_P.
2594 * function.c (diddle_return_value): Do not handle crtl->return_bnd.
2595 * genmodes.c (complete_mode): Do not handle MODE_POINTER_BOUNDS.
2596 (POINTER_BOUNDS_MODE): Remove definition.
2597 (make_pointer_bounds_mode): Remove.
2598 (get_mode_class): Do not handle MODE_POINTER_BOUNDS.
2599 * machmode.h (POINTER_BOUNDS_MODE_P): Remove definition.
2600 (scalare_mode::includes_p): Do not handle MODE_POINTER_BOUNDS.
2601 * mode-classes.def: Do not define MODE_POINTER_BOUNDS.
2602 * stor-layout.c (int_mode_for_mode): Do not handle MODE_POINTER_BOUNDS.
2603 * tree-core.h (enum tree_index): Remove TI_POINTER_BOUNDS_TYPE.
2604 * varasm.c (output_constant_pool_2): Do not handle MODE_POINTER_BOUNDS.
2606 * config/i386/i386-modes.def (BND32, BND64): Remove.
2607 * config/i386/i386.c (dbx_register_map): Remove bound registers.
2608 (dbx64_register_map): Ditto.
2609 (svr4_dbx_register_map): Ditto.
2610 (indirect_thunk_bnd_needed): Remove.
2611 (indirect_thunks_bnd_used): Ditto.
2612 (indirect_return_bnd_needed): Ditto.
2613 (indirect_return_via_cx_bnd): Ditto.
2614 (enum indirect_thunk_prefix): Remove indirect_thunk_prefix_bnd.
2615 (indirect_thunk_name): Remove handling of indirect_thunk_prefix_bnd.
2616 (output_indirect_thunk): Ditto. Remove need_prefix argument.
2617 (output_indirect_thunk_function): Remove handling of
2618 indirect_return_bnd_needed, indirect_return_via_cx_bnd,
2619 indirect_thunk_bnd_needed and indirect_thunks_bnd_used variables.
2620 (ix86_save_reg): Remove handling of crtl->return_bnd.
2621 (ix86_legitimate_constant_p): Remove handling of POINTER_BOUNDS_MODE_P.
2622 (ix86_print_operand_address_as): Remove handling of UNSPEC_BNDMK_ADDR
2623 and UNSPEC_BNDLX_ADDR.
2624 (ix86_output_indirect_branch_via_reg): Remove handling of
2625 indirect_thunk_prefix_bnd.
2626 (ix86_output_indirect_branch_via_push): Ditto.
2627 (ix86_output_function_return): Ditto.
2628 (ix86_output_indirect_function_return): Ditto.
2629 (avoid_func_arg_motion): Do not handle UNSPEC_BNDSTX.
2630 * config/i386/i386.h (FIXED_REGISTERS): Remove bound registers.
2631 (CALL_USED_REGISTERS): Ditto.
2632 (REG_ALLOC_ORDER): Update for removal of bound registers.
2633 (HI_REGISTER_NAMES): Ditto.
2634 * config/i386/i386.md (UNSPEC_BNDMK, UNSPEC_BNDMK_ADDR, UNSPEC_BNDSTX)
2635 (UNSPEC_BNDLDX, UNSPEC_BNDLDX_ADDR, UNSPEC_BNDCL, UNSPEC_BNDCU)
2636 (UNSPEC_BNDCN, UNSPEC_MPX_FENCE): Remove.
2637 (BND0_REG, BND1_REG, BND2_REG, BND3_REG): Remove
2638 (FIRST_PSEUDO_REG): Update.
2639 (BND): Remove mode iterator.
2640 * config/i386/predicates.md (bnd_mem_operator): Remove.
2642 2018-08-24 Richard Sandiford <richard.sandiford@arm.com>
2644 * tree-vect-stmts.c (vectorizable_bswap): Handle variable-length
2647 2018-08-24 Richard Sandiford <richard.sandiford@arm.com>
2649 * tree-vect-slp.c (vect_transform_slp_perm_load): Separate out
2650 the case in which the permute needs only a single element and
2651 repeats for every vector of the result. Extend that case to
2652 handle variable-length vectors.
2653 * tree-vect-stmts.c (vectorizable_load): Update accordingly.
2655 2018-08-24 H.J. Lu <hongjiu.lu@intel.com>
2658 * dwarf2out.c (save_macinfo_strings): Call set_indirect_string
2659 on DW_MACINFO_start_file for -gsplit-dwarf -g3.
2661 2018-08-24 Richard Biener <rguenther@suse.de>
2663 * cfg.h (struct control_flow_graph): Add edge_flags_allocated and
2664 bb_flags_allocated members.
2665 (auto_flag): New RAII class for allocating flags.
2666 (auto_edge_flag): New RAII class for allocating edge flags.
2667 (auto_bb_flag): New RAII class for allocating bb flags.
2668 * cfgloop.c (verify_loop_structure): Allocate temporary edge
2670 * cfganal.c (dfs_enumerate_from): Remove use of visited sbitmap
2671 in favor of temporarily allocated BB flag.
2672 * hsa-brig.c: Re-order includes.
2673 * hsa-dump.c: Likewise.
2674 * hsa-regalloc.c: Likewise.
2675 * print-rtl.c: Likewise.
2676 * profile-count.c: Likewise.
2678 2018-08-24 Segher Boessenkool <segher@kernel.crashing.org>
2681 * config/rs6000/rs6000.c (toc_relative_expr_p): Check that the base is
2684 2018-08-24 Aldy Hernandez <aldyh@redhat.com>
2687 * wide-int-range.cc (wide_int_range_div): Do not ignore result
2688 from wide_int_range_multiplicative_op.
2690 2018-08-23 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
2692 * tree-vect-data-refs.c (vect_grouped_store_supported): Fix typo
2695 2018-08-23 Giuliano Belinassi <giuliano.belinassi@usp.br>
2697 * genmatch.c (parser::parse_operation): Fix typo 'exapnded'
2700 2018-08-23 Alexander Monakov <amonakov@ispras.ru>
2702 * tree-scalar-evolution.c (final_value_replacement_loop): Dump
2703 full GENERIC expression used for replacement.
2705 2018-08-23 Aldy Hernandez <aldyh@redhat.com>
2707 * tree-vrp.c (abs_extent_range): Remove.
2708 (extract_range_into_wide_ints): Pass wide ints by reference.
2709 (extract_range_from_binary_expr_1): Rewrite the *DIV_EXPR code.
2710 Pass wide ints by reference in all calls to
2711 extract_range_into_wide_ints.
2712 * wide-int-range.cc (wide_int_range_div): New.
2713 * wide-int-range.h (wide_int_range_div): New.
2714 (wide_int_range_includes_zero_p): New.
2715 (wide_int_range_zero_p): New.
2717 2018-08-23 Matthew Malcomson <matthew.malcomson@arm.com>
2719 * config/aarch64/aarch64.md (arches): New enum.
2720 (arch): New enum attr.
2721 (arch_enabled): New attr.
2722 (enabled): Now uses arch_enabled only.
2723 (simd, sve, fp16): Removed attribute.
2724 (fp): Attr now defined in terms of 'arch'.
2725 (*mov<mode>_aarch64, *movsi_aarch64, *movdi_aarch64, *movti_aarch64,
2726 *movhf_aarch64, <optab><fcvt_target><GPF:mode>2,
2727 <FCVT_F2FIXED:fcvt_fixed_insn><GPF:mode>3,
2728 <FCVT_FIXED2F:fcvt_fixed_insn><GPI:mode>3): Merge 'fp' and 'simd'
2729 attributes into 'arch'.
2730 (*movsf_aarch64, *movdf_aarch64, *movtf_aarch64, *add<mode>3_aarch64,
2731 subdi3, neg<mode>2, <optab><mode>3, one_cmpl<mode>2,
2732 *<NLOGICAL:optab>_one_cmpl<mode>3, *xor_one_cmpl<mode>3,
2733 *aarch64_ashl_sisd_or_int_<mode>3, *aarch64_lshr_sisd_or_int_<mode>3,
2734 *aarch64_ashr_sisd_or_int_<mode>3, *aarch64_sisd_ushl): Convert use of
2735 'simd' attribute into 'arch'.
2736 (load_pair_sw_<SX:mode><SX2:mode>, load_pair_dw_<DX:mode><DX2:mode>,
2737 store_pair_sw_<SX:mode><SX2:mode>, store_pair_dw_<DX:mode><DX2:mode>):
2738 Convert use of 'fp' attribute to 'arch'.
2739 * config/aarch64/aarch64-simd.md (move_lo_quad_internal_<mode>,
2740 move_lo_quad_internal_<mode>): (different modes) Merge 'fp' and 'simd'
2742 (move_lo_quad_internal_be_<mode>, move_lo_quad_internal_be_<mode>):
2743 (different modes) Merge 'fp' and 'simd' into 'arch'.
2744 (*aarch64_combinez<mode>, *aarch64_combinez_be<mode>): Merge 'fp' and
2747 2018-08-23 Segher Boessenkool <segher@kernel.crashing.org>
2749 PR rtl-optimization/87026
2750 * expmed.c (canonicalize_comparison): If we can no longer create
2751 pseudoregisters, don't.
2753 2018-08-23 Richard Earnshaw <rearnsha@arm.com>
2756 * config/arm/arm-protos.h (arm_emit_speculation_barrier): New
2758 * config/arm/arm.c (speculation_barrier_libfunc): New static
2760 (arm_init_libfuncs): Initialize it.
2761 (arm_emit_speculation_barrier): New function.
2762 * config/arm/arm.md (speculation_barrier): Call
2763 arm_emit_speculation_barrier for architectures that do not have
2765 (speculation_barrier_insn): Only match on Armv7 or later.
2767 2018-08-23 Richard Biener <rguenther@suse.de>
2770 * tree-inline.c (copy_bb): Drop unused __builtin_va_arg_pack_len
2773 2018-08-23 Richard Sandiford <richard.sandiford@arm.com>
2775 * config/aarch64/aarch64.c (aarch64_evpc_sve_tbl): Fix handling
2776 of single-vector TBLs.
2777 (aarch64_vectorize_vec_perm_const): Set one_vector_p when only
2780 2018-08-23 Richard Sandiford <richard.sandiford@arm.com>
2783 * config/aarch64/aarch64.c (aarch64_expand_vec_perm_const_1): Fix
2784 aarch64_evpc_tbl guard.
2786 2018-08-22 Bernd Edlinger <bernd.edlinger@hotmail.de>
2788 * tree-ssa-dse.c (compute_trims): Avoid folding away undefined
2791 2018-08-22 Martin Sebor <msebor@redhat.com>
2794 * tree-pretty-print.c (pretty_print_string): Add argument.
2795 (dump_generic_node): Call to pretty_print_string with string size.
2797 2018-08-22 Segher Boessenkool <segher@kernel.crashing.org>
2799 PR rtl-optimization/86771
2800 * combine.c (try_combine): Do not allow splitting a resulting PARALLEL
2801 of two SETs into those two SETs, one to be placed at i2, if that SETs
2802 destination is modified between i2 and i3.
2804 2018-08-22 Richard Sandiford <richard.sandiford@arm.com>
2806 PR tree-optimization/86725
2807 * tree-vect-loop.c (vect_inner_phi_in_double_reduction_p): New
2809 (vect_analyze_scalar_cycles_1): Check it.
2811 2018-08-22 Richard Sandiford <richard.sandiford@arm.com>
2813 PR tree-optimization/86725
2814 * tree-vect-loop.c (vect_is_simple_reduction): When treating
2815 an outer loop phi as a double reduction, make sure that the
2816 single user of the phi result is an inner loop phi.
2818 2018-08-22 Richard Sandiford <richard.sandiford@arm.com>
2820 * tree-vect-data-refs.c (vect_analyze_group_access_1): Convert
2821 grouped stores with gaps to a strided group.
2823 2018-08-22 Richard Sandiford <richard.sandiford@arm.com>
2825 * tree-vect-stmts.c (get_group_load_store_type)
2826 (get_load_store_type): Only test STMT_VINFO_STRIDED_P for the
2827 first statement in a group.
2829 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
2831 * config/darwin.h (LINK_COMMAND_SPEC_A): Sync LTO options with
2832 the sequence used in gcc/gcc.c.
2834 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
2837 * gcc-ar.c (main): Don’t try to invoke the plug-in if we’re not
2840 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
2842 * config/darwin10.h (LINK_GCC_C_SEQUENCE_SPEC): Adjust to use the
2843 Darwin10-specific unwinder-shim.
2844 * config/darwin12.h (LINK_GCC_C_SEQUENCE_SPEC): Remove.
2845 * config/rs6000/darwin.h (DARWIN_CRT1_SPEC, DARWIN_DYLIB1_SPEC):
2846 New to cater for Darwin10 Rosetta.
2848 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
2850 * config/i386/i386.c (ix86_output_addr_diff_elt): Move the MACH-O
2851 specific test before the one for HAVE_AS_GOTOFF_IN_DATA.
2853 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
2858 * gcc/dwarf2out.c (FUNC_SECOND_SECT_LABEL): New.
2859 (dwarf2out_switch_text_section): Generate a local label for the
2860 second function sub-section and apply it as the second FDE start
2862 * gcc/final.c (final_scan_insn_1): Emit second FDE label after the
2863 second sub-section start.
2865 2018-08-22 Richard Biener <rguenther@suse.de>
2867 PR tree-optimization/86988
2868 * tree-vrp.c (vrp_prop::check_mem_ref): Bail out on VLAs.
2870 2018-08-22 Richard Biener <rguenther@suse.de>
2872 PR tree-optimization/86945
2873 * tree-cfg.c (generate_range_test): Use unsigned arithmetic.
2875 2018-08-22 Alexandre Oliva <oliva@adacore.com>
2877 * config/rs6000/rs6000.c (SMALL_DATA_RELOC, SMALL_DATA_REG): Add
2878 a comment about how uses of r2 for .sdata2 come about.
2880 2018-08-22 Alexandre Oliva <aoliva@redhat.com>
2882 * tree-ssa-reassoc.c (is_reassociable_op): Fix cut&pasto.
2884 2018-08-21 Marek Polacek <polacek@redhat.com>
2886 PR c++/86981, Implement -Wpessimizing-move.
2887 * doc/invoke.texi: Document -Wpessimizing-move.
2889 2018-08-21 Jan Hubicka <jh@suse.cz>
2891 * tree.c (find_decls_types_r): Do not check for redundant typedefs.
2892 * tree.h (is_redundant_typedef): Remove.
2893 * dwarf2out.c (is_redundant_typedef): Turn into static function.
2895 2018-08-21 Jan Hubicka <jh@suse.cz>
2897 * tree.c (free_lang_data_in_decl): Remove types from DECL_CONTEXT
2900 2018-08-21 Tamar Christina <tamar.christina@arm.com>
2902 * expmed.c (extract_low_bits): Reject invalid subregs early.
2904 2018-08-21 Bernd Edlinger <bernd.edlinger@hotmail.de>
2907 * tree-ssa-strlen.c (adjust_last_stmt): Avoid folding away undefined
2910 2018-08-21 Rasmus Villemoes <rv@rasmusvillemoes.dk>
2912 * config/vxworks.h: Guard vxworks_asm_out_constructor and
2913 vxworks_asm_out_destructor by !HAVE_INITFINI_ARRAY_SUPPORT
2914 * config/vxworks.c: Likewise.
2916 2018-08-21 Rasmus Villemoes <rv@rasmusvillemoes.dk>
2918 * config/vxworks.c: Set targetm.have_ctors_dtors
2919 if HAVE_INITFINI_ARRAY_SUPPORT.
2920 * config/vxworks.h: Set SUPPORTS_INIT_PRIORITY
2921 if HAVE_INITFINI_ARRAY_SUPPORT.
2923 2018-08-21 Rasmus Villemoes <rv@rasmusvillemoes.dk>
2925 * config/vxworks.h: Add $(WIND_BASE)/target/h/wrn/coreip to
2926 default search path for VxWorks < 7.
2928 2018-08-21 Bernd Edlinger <bernd.edlinger@hotmail.de>
2930 * gimple-ssa-sprintf.c (decl_constant_value): Remove.
2931 (get_format_string): Refer to c_getstr.
2933 2018-08-21 Tom de Vries <tdevries@suse.de>
2935 * cgraph.h (debuginfo_early_init, debuginfo_init, debuginfo_fini)
2936 (debuginfo_start, debuginfo_stop, debuginfo_early_start)
2937 (debuginfo_early_stop): Declare.
2938 * cgraphunit.c (debuginfo_early_init, debuginfo_init, debuginfo_fini)
2939 (debuginfo_start, debuginfo_stop, debuginfo_early_start)
2940 (debuginfo_early_stop): New function.
2941 (symbol_table::finalize_compilation_unit): Call debuginfo_early_start
2942 and debuginfo_early_stop.
2943 * dwarf2out.c (dwarf2out_finish, dwarf2out_early_finish): Dump dwarf.
2944 * toplev.c (compile_file): Call debuginfo_start and debuginfo_stop.
2945 (general_init): Call debuginfo_early_init.
2946 (finalize): Call debuginfo_fini.
2947 (do_compile): Call debuginfo_init.
2948 * doc/invoke.texi (@gccoptlist): Add -fdump-debug and
2950 (@item -fdump-debug, @item -fdump-earlydebug): Add.
2952 2018-08-21 Tom de Vries <tdevries@suse.de>
2954 * dwarf2out.c (print_dw_val, print_loc_descr, print_die): Handle
2955 flag_dump_noaddr and flag_dump_unnumbered.
2957 2018-08-21 Aldy Hernandez <aldyh@redhat.com>
2959 * wide-int-range.cc (wide_int_range_abs): New.
2960 (wide_int_range_order_set): Rename from wide_int_range_min_max.
2961 * wide-int-range.h (wide_int_range_abs): New.
2962 (wide_int_range_min_max): New.
2963 * tree-vrp.c (extract_range_from_unary_expr): Rewrite ABS_EXPR
2964 case to call wide_int_range_abs.
2965 Rewrite MIN/MAX_EXPR to call wide_int_range_min_max.
2966 (extract_range_from_abs_expr): Delete.
2968 2018-08-20 Michael Meissner <meissner@linux.ibm.com>
2971 * config/rs6000/rs6000.md (extendsi<mode>2): Change constraints
2972 from 'Y' to 'YZ' to enable the LWAX instruction to be generated
2975 2018-08-20 Nathan Sidwell <nathan@acm.org>
2976 Jeff Law <law@redhat.com>
2978 * config/s390/s390-c (s390_macro_to_expand): Use cpp_macro_p.
2979 * config/spu/spu-c.c (spu_macro_to_expand): Likewise.
2981 2018-08-20 David Malcolm <dmalcolm@redhat.com>
2984 * attribs.c (diag_attr_exclusions): Add auto_diagnostic_group instance.
2985 (decl_attributes): Likewise.
2986 * calls.c (maybe_warn_nonstring_arg): Add auto_diagnostic_group
2988 * cgraphunit.c (maybe_diag_incompatible_alias): Likewise.
2989 * diagnostic-core.h (class auto_diagnostic_group): New class.
2990 * diagnostic.c (diagnostic_initialize): Initialize the new fields.
2991 (diagnostic_report_diagnostic): Handle the first diagnostics within
2993 (emit_diagnostic): Add auto_diagnostic_group instance.
2995 (inform_n): Likewise.
2996 (warning): Likewise.
2997 (warning_at): Likewise.
2998 (warning_n): Likewise.
2999 (pedwarn): Likewise.
3000 (permerror): Likewise.
3002 (error_n): Likewise.
3003 (error_at): Likewise.
3005 (fatal_error): Likewise.
3006 (internal_error): Likewise.
3007 (internal_error_no_backtrace): Likewise.
3008 (auto_diagnostic_group::auto_diagnostic_group): New ctor.
3009 (auto_diagnostic_group::~auto_diagnostic_group): New dtor.
3010 * diagnostic.h (struct diagnostic_context): Add fields
3011 "diagnostic_group_nesting_depth",
3012 "diagnostic_group_emission_count", "begin_group_cb",
3014 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior):
3015 Add auto_diagnostic_group instance(s).
3016 (find_explicit_erroneous_behavior): Likewise.
3017 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Likewise.
3018 * gimple-ssa-warn-restrict.c (maybe_diag_offset_bounds): Likewise.
3019 * gimplify.c (warn_implicit_fallthrough_r): Likewise.
3020 (gimplify_va_arg_expr): Likewise.
3021 * hsa-gen.c (HSA_SORRY_ATV): Likewise.
3022 (HSA_SORRY_AT): Likewise.
3023 * ipa-devirt.c (compare_virtual_tables): Likewise.
3024 (warn_odr): Likewise.
3025 * multiple_target.c (expand_target_clones): Likewise.
3026 * opts-common.c (cmdline_handle_error): Likewise.
3027 * reginfo.c (globalize_reg): Likewise.
3028 * substring-locations.c (format_warning_n_va): Likewise.
3029 * tree-inline.c (expand_call_inline): Likewise.
3030 * tree-ssa-ccp.c (pass_post_ipa_warn::execute): Likewise.
3031 * tree-ssa-loop-niter.c
3032 (do_warn_aggressive_loop_optimizations): Likewise.
3033 * tree-ssa-uninit.c (warn_uninit): Likewise.
3034 * tree.c (warn_deprecated_use): Likewise.
3036 2018-08-20 H.J. Lu <hongjiu.lu@intel.com>
3039 * config/i386/i386.md (eh_return): Always update EH return
3040 address in word_mode.
3042 2018-08-20 Chung-Lin Tang <cltang@codesourcery.com>
3044 * targhooks.c (std_gimplify_va_arg_expr): Properly handle case of when
3045 TARGET_SPLIT_COMPLEX_ARG is defined.
3047 2018-08-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
3049 * expr.c (store_field): Change gcc_assert to gcc_checking_assert.
3051 2018-08-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
3054 * expr.c (expand_assignment): Assert that bitpos is positive.
3055 (store_field): Likewise
3056 (expand_expr_real_1): Make sure that bitpos is positive.
3057 * config/alpha/alpha.h (CONSTANT_ADDRESS_P): Avoid signed
3060 2018-08-20 Nathan Sidwell <nathan@acm.org>
3062 * Makefile.in (CPP_ID_DATA_H): Delete.
3063 (CPP_INTERNAL_H): Don't add it.
3064 (GTFILES): Replace CPP_ID_DATA_H with CPPLIB_H.
3065 * gengtype.c (open_base_files): Replace cpp-id-data.h with cpplib.h
3067 2018-08-20 Richard Biener <rguenther@suse.de>
3069 PR tree-optimization/78655
3070 * tree-vrp.c (extract_range_from_binary_expr_1): Make
3071 pointer + offset nonnull if either operand is nonnull work.
3073 2018-08-20 Tom de Vries <tdevries@suse.de>
3075 * dwarf2out.c (add_scalar_info): Don't add reference to existing die
3076 unless the referenced die describes the added property using
3077 DW_AT_location or DW_AT_const_value. Fall back to exprloc case.
3078 Otherwise, add a DW_AT_location to the referenced die.
3080 2018-08-19 Uros Bizjak <ubizjak@gmail.com>
3083 * config/i386/i386.c (ix86_rtx_costs) [case SET]: Check source for
3084 register_operand when calling ix86_set_reg_reg_cost.
3085 [case CONST_INT, case CONST, case LABEL_REF, case SYMBOL_REF]:
3086 Set *total to 0 for operands that satisfy x86_64_immediate_operand
3087 predicate and to 1 otherwise.
3089 2018-08-18 Iain Sandoe <iain@sandoe.co.uk>
3091 * config/darwin.c (darwin_override_options): If -gsplit-dwarf is set,
3092 emit a diagnostic that it is not supported and reset the option.
3093 * config/darwin.h (DRIVER_SELF_SPECS): Note that gsplit-dwarf is not
3094 supported and consume the option. (ASM_FINAL_SPEC): New.
3096 2018-08-17 Segher Boessenkool <segher@kernel.crashing.org>
3098 * doc/md.texi (Patterns): Use @ref instead of @xref in the middle of
3101 2018-08-17 Sandra Loosemore <sandra@codesourcery.com>
3103 C-SKY port: Documentation
3105 * doc/extend.texi (C-SKY Function Attributes): New section.
3106 * doc/invoke.texi (Option Summary): Add C-SKY options.
3107 (C-SKY Options): New section.
3108 * doc/md.texi (Machine Constraints): Document C-SKY constraints.
3110 2018-08-17 Jojo <jijie_rong@c-sky.com>
3111 Huibin Wang <huibin_wang@c-sky.com>
3112 Sandra Loosemore <sandra@codesourcery.com>
3113 Chung-Lin Tang <cltang@codesourcery.com>
3115 C-SKY port: Backend implementation
3117 * config/csky/*: New.
3118 * common/config/csky/*: New.
3120 2018-08-17 Jojo <jijie_rong@c-sky.com>
3121 Huibin Wang <huibin_wang@c-sky.com>
3122 Sandra Loosemore <sandra@codesourcery.com>
3123 Chung-Lin Tang <cltang@codesourcery.com>
3124 Andrew Jenner <andrew@codesourcery.com>
3126 C-SKY port: Configury
3128 * config.gcc (csky-*-*): New.
3129 * configure.ac: Add csky to targets for dwarf2 debug_line support.
3130 * configure: Regenerated.
3132 2018-08-17 David Malcolm <dmalcolm@redhat.com>
3134 * dump-context.h: Include "dumpfile.h".
3135 (dump_context::dump_printf_va): Convert final param from va_list
3136 to va_list *. Convert from ATTRIBUTE_PRINTF to
3137 ATTRIBUTE_GCC_DUMP_PRINTF.
3138 (dump_context::dump_printf_loc_va): Likewise.
3139 * dumpfile.c: Include "stringpool.h".
3140 (make_item_for_dump_printf_va): Delete.
3141 (make_item_for_dump_printf): Delete.
3142 (class dump_pretty_printer): New class.
3143 (dump_pretty_printer::dump_pretty_printer): New ctor.
3144 (dump_pretty_printer::emit_items): New member function.
3145 (dump_pretty_printer::emit_any_pending_textual_chunks): New member
3147 (dump_pretty_printer::emit_item): New member function.
3148 (dump_pretty_printer::stash_item): New member function.
3149 (dump_pretty_printer::format_decoder_cb): New member function.
3150 (dump_pretty_printer::decode_format): New member function.
3151 (dump_context::dump_printf_va): Reimplement in terms of
3152 dump_pretty_printer.
3153 (dump_context::dump_printf_loc_va): Convert final param from va_list
3155 (dump_context::begin_scope): Reimplement call to
3156 make_item_for_dump_printf.
3157 (dump_printf): Update for change to dump_printf_va.
3158 (dump_printf_loc): Likewise.
3159 (selftest::test_capture_of_dump_calls): Convert "stmt" from
3160 greturn * to gimple *. Add a test_decl. Add tests of dump_printf
3161 with %T, %E, and %G.
3162 * dumpfile.h (ATTRIBUTE_GCC_DUMP_PRINTF): New macro.
3163 (dump_printf): Replace ATTRIBUTE_PRINTF_2 with
3164 ATTRIBUTE_GCC_DUMP_PRINTF (2, 3).
3165 (dump_printf_loc): Replace ATTRIBUTE_PRINTF_3 with
3166 ATTRIBUTE_GCC_DUMP_PRINTF (3, 0).
3167 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Convert
3168 use of HOST_WIDE_INT_PRINT_DEC on unsigned HOST_WIDE_INT "count"
3169 within a dump_printf_loc call to "%wu".
3170 (vector_alignment_reachable_p): Merge two dump_printf[_loc] calls,
3171 converting a use of HOST_WIDE_INT_PRINT_DEC to "%wd". Add a
3172 missing space after "=".
3173 * tree-vect-loop.c (vect_analyze_loop_2) Within a dump_printf
3174 call, convert use of HOST_WIDE_INT_PRINT_DEC to "%wd".
3175 * tree-vect-slp.c (vect_slp_bb): Within a dump_printf_loc call,
3176 convert use of HOST_WIDE_INT_PRINT_UNSIGNED to "%wu".
3177 * tree-vectorizer.c (try_vectorize_loop_1): Likewise. Remove
3178 duplicate "vectorized" from message.
3180 2018-08-17 Szabolcs Nagy <szabolcs.nagy@arm.com>
3182 * config/arm/arm-builtins.c (arm_init_simd_builtin_types): Clear
3183 polyNxK_t element's TYPE_STRING_FLAG.
3185 2018-08-17 Segher Boessenkool <segher@kernel.crashing.org>
3187 * config/rs6000/rs6000.md (*cbranch, *creturn): Name these patterns
3188 (they were unnamed before). Fix comments.
3190 2018-08-17 Nathan Sidwell <nathan@acm.org>
3192 * cppbuiltin.c: Include "cpplib.h", not "cpp-id-data.h".
3194 2018-08-17 Richard Biener <rguenther@suse.de>
3196 PR tree-optimization/86841
3197 * wide-int-range.cc (wide_int_range_lshift): Use to_uhwi.
3199 2018-08-17 Martin Liska <mliska@suse.cz>
3201 * common.opt: Remove Warn, Init and Report for options with
3202 Ignore/Deprecated flag. Warning is done automatically for
3204 * config/i386/i386.opt: Likewise.
3205 * config/ia64/ia64.opt: Likewise.
3206 * config/rs6000/rs6000.opt: Likewise.
3207 * cppbuiltin.c (define_builtin_macros_for_compilation_flags):
3208 Remove usage of flag_check_pointer_bounds.
3209 * lto-wrapper.c (merge_and_complain): Do not handle
3210 OPT_fcheck_pointer_bounds.
3211 (append_compiler_options): Likewise.
3212 * opt-functions.awk: Do not handle Deprecated.
3213 * optc-gen.awk: Check that Var, Report and Init are not
3214 used for an option with Ignore/Deprecated flag.
3215 * opts-common.c (decode_cmdline_option): Do not report
3217 (read_cmdline_option): Report warning for OPT_SPECIAL_deprecated
3219 * opts.h (struct cl_option): Remove cl_deprecated flag.
3220 (CL_ERR_DEPRECATED): Remove error enum value.
3222 2018-08-17 Richard Biener <rguenther@suse.de>
3225 * tree-inline.c (copy_bb): When inlining __builtin_va_arg_pack_len ()
3226 across a va-arg-pack using call adjust its return value accordingly.
3228 2018-08-16 Martin Sebor <msebor@redhat.com>
3230 PR tree-optimization/86853
3231 * gimple-ssa-sprintf.c (struct format_result): Rename member.
3232 (struct fmtresult): Add member and initialize it in ctors.
3233 (format_character): Handle %C. Extend range to NUL. Set MAYFAIL.
3234 (format_string): Handle %S the same as %ls. Set MAYFAIL.
3235 (format_directive): Set POSUNDER4K when MAYFAIL is set.
3236 (parse_directive): Handle %C same as %c.
3237 (sprintf_dom_walker::compute_format_length): Adjust.
3238 (is_call_safe): Adjust.
3240 2018-08-16 Bernd Edlinger <bernd.edlinger@hotmail.de>
3242 * builtins.c (c_strlen): Add new parameter eltsize. Use it
3243 for determining how to count the elements.
3244 * builtins.h (c_strlen): Adjust prototype.
3245 * expr.c (string_constant): Add new parameter mem_size.
3246 Set *mem_size appropriately.
3247 * expr.h (string_constant): Adjust protoype.
3248 * gimple-fold.c (get_range_strlen): Add new parameter eltsize.
3249 * gimple-fold.h (get_range_strlen): Adjust prototype.
3250 * gimple-ssa-sprintf.c (get_string_length): Add new parameter eltsize.
3251 (format_string): Call get_string_length with eltsize.
3253 2018-08-16 David Malcolm <dmalcolm@redhat.com>
3255 * diagnostic.c (default_diagnostic_start_span_fn): Call pp_string
3256 to emit the span, rather than setting it as the prefix.
3258 2018-08-16 David Malcolm <dmalcolm@redhat.com>
3260 * diagnostic-show-locus.c (layout::start_annotation_line): Add
3261 "margin_char" parameter, defaulting to space. Use it in place
3262 of pp_space for the initial part of the margin.
3263 (layout::print_leading_fixits): Use '+' when filling the margin
3264 of line-insertion fix-it hints.
3266 2018-08-16 Segher Boessenkool <segher@kernel.crashing.org>
3268 * config/rs6000/rs6000.md (two unnamed define_insn and define_split):
3271 2018-08-16 Segher Boessenkool <segher@kernel.crashing.org>
3273 * config/rs6000/altivec.md: Don't set length attribute to the default
3275 * config/rs6000/darwin.md: Ditto.
3276 * config/rs6000/dfp.md: Ditto.
3277 * config/rs6000/htm.md: Ditto.
3278 * config/rs6000/rs6000.md: Ditto.
3279 * config/rs6000/sync.md: Ditto.
3280 * config/rs6000/vsx.md: Ditto.
3282 2018-08-16 Segher Boessenkool <segher@kernel.crashing.org>
3284 * config/rs6000/altivec.md: Don't set length attribute to the default
3285 value, for branch instructions.
3286 * config/rs6000/darwin.md: Ditto.
3287 * config/rs6000/rs6000.md: Ditto.
3289 2018-08-16 Segher Boessenkool <segher@kernel.crashing.org>
3291 * config/rs6000/rs6000.md (length): Always define as const_int 4.
3292 (unnamed conditional branch define_insn): Set length to 4 or 8
3293 depending on offset.
3294 (<bd>_<mode>): Similar, for alternative 0.
3295 (<bd>tf_<mode>): Ditto.
3297 2018-08-16 Tamar Christina <tamar.christina@arm.com>
3299 * expr.c (copy_blkmode_to_reg): Perform larger copies when safe.
3301 2018-08-16 Matthew Malcomson <matthew.malcomson@arm.com>
3303 * doc/rtl.texi: Replace old RTX class names with new names.
3306 2018-08-16 Vlad Lazar <vlad.lazar@arm.com>
3308 * expmed.h (canonicalize_comparison): New declaration.
3309 * expmed.c (canonicalize_comparison, equivalent_cmp_code): New function.
3310 * expmed.c (emit_store_flag_1): Add call to canonicalize_comparison.
3311 * optabs.c (prepare_cmp_insn): Likewise.
3312 * rtl.h (unsigned_condition_p): New function which checks if a
3313 comparison operator is unsigned.
3315 2018-08-16 Nathan Sidwell <nathan@acm.org>
3317 * config/rs6000/rs6000-c.c (rs6000_macro_to_expend): Use cpp_macro_p.
3318 * config/powerpcspc/powerpcspe-c.c (rs6000_macro_to_expend): Likewise.
3320 2018-08-16 Tamar Christina <tamar.christina@arm.com>
3323 * config/arm/arm.c (arm_can_change_mode_class): Disallow subreg.
3324 * config/arm/neon.md (movv4hf, movv8hf): Refactored to..
3325 (mov<mov>): ..this and enable unconditionally.
3327 2018-08-16 Tamar Christina <tamar.christina@arm.com>
3329 * config/arm/neon.md (*neon_mov<mode>): Remove reg-to-reg alternative.
3331 2018-08-16 Sam Tebbs <sam.tebbs@arm.com>
3333 * config/aarch64/aarch64.opt (mlow-precision-recip-sqrt)
3334 (mlow-precision-sqrt, mlow-precision-div, mverbose-cost-dump): Replace
3335 "Common" with "Target".
3337 2018-08-15 Uros Bizjak <ubizjak@gmail.com>
3339 * config/i386/i386.opt (mmitigate-rop): Mark as deprecated.
3340 * doc/invoke.texi (mmitigate-rop): Remove.
3341 * config/i386/i386.c: Do not include "regrename.h".
3342 (ix86_rop_should_change_byte_p, reg_encoded_number)
3343 (ix86_get_modrm_for_rop, set_rop_modrm_reg_bits, ix86_mitigate_rop):
3345 (ix86_reorg): Remove call to ix86_mitigate_rop.
3346 * config/i386/i386.md (attr "modrm_class"): Remove.
3347 (cmp<mode>_ccno_1, mov<mode>_xor, movstrict<mode>_xor)
3348 (x86_mov<mode>cc_0_m1. x86_mov<mode>cc_0_m1_se)
3349 (x86_mov<mode>cc_0_m1_neg): Remove modrm_class attribute override.
3351 2018-08-15 Will Schmidt <will_schmidt@vnet.ibm.com>
3353 * config/rs6000/rs600.c (rs6000_gimple_fold_builtin): Add entries to
3354 allow folding of mergeh() and mergel() for the float and double types.
3355 (fold_mergehl_helper): Rework to handle building a permute tree
3358 2018-08-15 Uros Bizjak <ubizjak@gmail.com>
3360 * config/i386/i386.c (expand_vec_perm_movs): Enable V4SFmode
3363 2018-08-15 David Malcolm <dmalcolm@redhat.com>
3365 * common.opt (fdiagnostics-show-labels): New option.
3366 * diagnostic-show-locus.c (class layout_range): Add field
3368 (class layout): Add field "m_show_labels_p".
3369 (layout_range::layout_range): Add param "label" and use it to
3371 (make_range): Pass in NULL for new "label" param of layout_range's
3373 (layout::layout): Initialize m_show_labels_p.
3374 (layout::maybe_add_location_range): Pass in loc_range->m_label
3375 when constructing layout_range instances.
3376 (struct line_label): New struct.
3377 (layout::print_any_labels): New member function.
3378 (layout::print_line): Call it if label-printing is enabled.
3379 (selftest::test_one_liner_labels): New test.
3380 (selftest::test_diagnostic_show_locus_one_liner): Call it.
3381 * diagnostic.c (diagnostic_initialize): Initialize
3382 context->show_labels_p.
3383 * diagnostic.h (struct diagnostic_context): Add field
3385 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
3386 -fno-diagnostics-show-labels.
3387 * dwarf2out.c (gen_producer_string): Add
3388 OPT_fdiagnostics_show_labels to the ignored options.
3389 * gcc-rich-location.c (gcc_rich_location::add_expr): Add "label"
3391 (gcc_rich_location::maybe_add_expr): Likewise.
3392 * gcc-rich-location.h (gcc_rich_location::gcc_rich_location): Add
3393 label" param, defaulting to NULL.
3394 (gcc_rich_location::add_expr): Add "label" param.
3395 (gcc_rich_location::maybe_add_expr): Likewise.
3396 (class text_range_label): New class.
3397 (class range_label_for_type_mismatch): New class.
3398 * gimple-ssa-sprintf.c (fmtwarn): Pass NULL for new label params
3399 of format_warning_va.
3400 (fmtwarn_n): Likewise for new params of format_warning_n_va.
3401 * lto-wrapper.c (merge_and_complain): Add
3402 OPT_fdiagnostics_show_labels to the "pick one setting" options.
3403 (append_compiler_options): Likewise to the dropped options.
3404 (append_diag_options): Likewise to the passed-on options.
3405 * opts.c (common_handle_option): Handle the new option.
3406 * selftest-diagnostic.c
3407 (test_diagnostic_context::test_diagnostic_context): Enable
3409 * substring-locations.c: Include "gcc-rich-location.h".
3410 (format_warning_n_va): Add "fmt_label" and "param_label" params
3411 and use them as appropriate.
3412 (format_warning_va): Add "fmt_label" and "param_label" params,
3413 passing them on to format_warning_n_va.
3414 (format_warning_at_substring): Likewise.
3415 (format_warning_at_substring_n): Likewise.
3416 * substring-locations.h (format_warning_va): Add "fmt_label" and
3417 "param_label" params.
3418 (format_warning_n_va): Likewise.
3419 (format_warning_at_substring): Likewise.
3420 (format_warning_at_substring_n): Likewise.
3421 * toplev.c (general_init): Initialize global_dc->show_labels_p.
3423 2018-08-15 Qing Zhao <qing.zhao@oracle.com>
3426 * builtins.c (expand_builtin_memcmp): Do not expand the call
3427 when overflow is detected.
3429 2018-08-15 Martin Sebor <msebor@redhat.com>
3431 PR tree-optimization/71625
3432 * config/aarch64/aarch64-builtins.c
3433 (aarch64_init_simd_builtin_types): Clear Poly8_t's TYPE_STRING_FLAG.
3435 2018-08-15 Ilya Leoshkevich <iii@linux.ibm.com>
3437 * config/s390/s390.c (s390_reorg): Remove loop.
3439 2018-08-15 Iain Sandoe <iain@sandoe.co.uk>
3442 (darwin_function_switched_text_sections): Delete.
3443 * gcc/config/darwin.h
3444 (TARGET_ASM_FUNCTION_SWITCHED_TEXT_SECTIONS): Likewise.
3446 2018-08-15 Iain Sandoe <iain@sandoe.co.uk>
3449 * config/darwin.h: (DEBUG_STR_OFFSETS_SECTION, DEBUG_LOCLISTS_SECTION,
3450 DEBUG_RNGLISTS_SECTION) new macros. (DEBUG_PUBNAMES_SECTION,
3451 DEBUG_PUBTYPES_SECTION) update to include GNU variant.
3453 2018-08-15 Martin Liska <mliska@suse.cz>
3455 PR tree-optimization/86925
3456 * predict.c (expr_expected_value_1): When taking
3457 later predictor, assign also probability.
3458 Use fold_build2_initializer_loc in order to fold
3459 the expression in -frounding-math.
3461 2018-08-14 Allan Sandfeld Jensen <allan.jensen@qt.io>
3463 * config/i386/i386.c (expand_vec_perm_movs): New method matching movs
3465 (expand_vec_perm_1): Try the new method.
3467 2018-08-14 Ilya Leoshkevich <iii@linux.ibm.com>
3470 * lra-lives.c (remove_some_program_points_and_update_live_ranges):
3471 Check whether lra_live_max_point is 0 before dividing.
3473 2018-08-14 Martin Sebor <msebor@redhat.com>
3475 PR tree-optimization/86650
3476 * tree-vrp.c (vrp_prop::check_array_ref): Print an inform message.
3477 (vrp_prop::check_mem_ref): Same.
3479 2018-08-13 Liu Hao <lh_mouse@126.com>
3481 * pretty-print.c (eat_esc_sequence): Swap the foreground and
3482 background colors if the COMMON_LVB_REVERSE_VIDEO flag is set,
3483 and clear it thereafter, as it only works for DBCS.
3485 2018-08-13 Liu Hao <lh_mouse@126.com>
3487 * pretty-print.c (mingw_ansi_fputs): Do not call _close() on the
3488 handle returned by _get_osf_handle().
3490 2018-08-13 Will Schmidt <will_schmidt@vnet.ibm.com>
3492 * gcc/config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support
3493 for folding vec_perm.
3495 2018-08-13 Will Schmidt <will_schmidt@vnet.ibm.com>
3497 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin):
3498 Add support for gimple-folding of vec_pack() and vec_unpack()
3501 2018-08-13 Will Schmidt <will_schmidt@vnet.ibm.com>
3503 * config/rs6000/rs6000.c (rs6000_builtin_valid_without_lhs): Add
3504 vec_xst variants to the list.
3505 (rs6000_gimple_fold_builtin): Add support for folding unaligned
3506 vector loads and stores.
3508 2018-08-13 David Edelsohn <dje.gcc@gmail.com>
3510 * config.gcc (rs6000-ibm-aix4.x): Delete.
3511 (rs6000-ibm-aix5.1): Delete.
3512 (rs6000-ibm-aix5.2): Delete.
3513 (rs6000-ibm-aix5.3): Delete.
3514 * config/rs6000/aix43.h: Delete.
3515 * config/rs6000/aix51.h: Delete.
3516 * config/rs6000/aix52.h: Delete.
3517 * config/rs6000/t-aix43: Delete.
3519 2018-08-13 Ilya Leoshkevich <iii@linux.ibm.com>
3521 * config/s390/s390.c (s390_decompose_constant_pool_ref):
3523 (s390_decompose_address): Factor out constant pool ref
3526 2018-08-12 Chung-Ju Wu <jasonwucj@gmail.com>
3528 * config/nds32/nds32-predicates.c
3529 (nds32_can_use_bclr_p): Change return type as bool.
3530 (nds32_can_use_bset_p): Ditto.
3531 (nds32_can_use_btgl_p): Ditto.
3532 (nds32_can_use_bitci_p): Ditto.
3533 * config/nds32/nds32-protos.h
3534 (nds32_can_use_bclr_p): Change declaration.
3535 (nds32_can_use_bset_p): Ditto.
3536 (nds32_can_use_btgl_p): Ditto.
3537 (nds32_can_use_bitci_p): Ditto.
3539 2018-08-12 Chung-Ju Wu <jasonwucj@gmail.com>
3541 * config/nds32/nds32.c (nds32_expand_prologue, nds32_expand_epilogue):
3542 Support -msched-prolog-epilog option.
3543 * config/nds32/nds32.opt (msched-prolog-epilog): New option.
3545 2018-08-12 Chung-Ju Wu <jasonwucj@gmail.com>
3547 * common/config/nds32/nds32-common.c
3548 (nds32_option_optimization_table): Enalbe -malways-align.
3550 2018-08-12 Chung-Ju Wu <jasonwucj@gmail.com>
3552 * config.gcc (nds32*): Add nds32_isr.h and nds32_init.inc in
3554 * common/config/nds32/nds32-common.c (nds32_handle_option): Handle
3555 OPT_misr_secure_ case.
3556 * config/nds32/nds32-isr.c: Implementation of backward compatibility.
3557 * config/nds32/nds32-protos.h (nds32_isr_function_critical_p): New.
3558 * config/nds32/nds32.c (nds32_attribute_table): Add critical and
3560 * config/nds32/nds32.h (nds32_isr_nested_type): Add NDS32_CRITICAL.
3561 (nds32_isr_info): New field security_level.
3562 (TARGET_ISR_VECTOR_SIZE_4_BYTE): New macro.
3563 * config/nds32/nds32.md (return_internal): Consider critical attribute.
3564 * config/nds32/nds32.opt (misr-secure): New option.
3565 * config/nds32/nds32_init.inc: New file.
3566 * config/nds32/nds32_isr.h: New file.
3568 2018-08-11 John David Anglin <danglin@gcc.gnu.org>
3570 * config/pa/pa.md (UNSPEC_MEMORY_BARRIER): New unspec enum.
3571 Update comment for atomic instructions.
3572 (atomic_storeqi, atomic_storehi, atomic_storesi, atomic_storesf,
3573 atomic_loaddf, atomic_loaddf_1, atomic_storedf, atomic_storedf_1):
3575 (atomic_loaddi): Revise fence expansion to only emit fence prior to
3576 load for __ATOMIC_SEQ_CST model.
3577 (atomic_loaddi_1): Remove float register target.
3578 (atomic_storedi): Handle CONST_INT values.
3579 (atomic_storedi_1): Remove float register source. Add special case
3581 (memory_barrier): New expander and insn.
3583 2018-08-11 Jakub Jelinek <jakub@redhat.com>
3585 PR tree-optimization/86835
3586 * tree-ssa-math-opts.c (insert_reciprocals): Even when inserting
3587 new_stmt after def_gsi, make sure to insert new_square_stmt after
3588 that stmt, not 2 stmts before it.
3590 2018-08-10 Alexander Monakov <amonakov@ispras.ru>
3593 * config/i386/i386.md (<s>mul<mode>3_highpart): Use DWIH mode iterator
3596 2018-08-10 Martin Liska <mliska@suse.cz>
3599 * builtin-types.def (BT_FN_LONG_LONG_LONG_DOUBLE): Add new
3601 * builtins.c (expand_builtin_expect_with_probability):
3603 (expand_builtin_expect_with_probability): New function.
3604 (build_builtin_expect_predicate): Add new argumnet probability
3605 for BUILT_IN_EXPECT_WITH_PROBABILITY.
3606 (fold_builtin_expect):
3609 * builtins.def (BUILT_IN_EXPECT_WITH_PROBABILITY):
3610 * builtins.h (fold_builtin_expect): Set new argument.
3611 * doc/extend.texi: Document __builtin_expect_with_probability.
3612 * doc/invoke.texi: Likewise.
3613 * gimple-fold.c (gimple_fold_call): Pass new argument.
3614 * ipa-fnsummary.c (find_foldable_builtin_expect): Handle
3615 also BUILT_IN_EXPECT_WITH_PROBABILITY.
3616 * predict.c (get_predictor_value): New function.
3617 (expr_expected_value): Add new argument probability. Assume
3618 that predictor and probability are always non-null.
3619 (expr_expected_value_1): Likewise. For __builtin_expect and
3620 __builtin_expect_with_probability set probability. Handle
3621 combination in binary expressions.
3622 (tree_predict_by_opcode): Simplify code by simply calling
3623 get_predictor_value.
3624 (pass_strip_predict_hints::execute): Add handling of
3625 BUILT_IN_EXPECT_WITH_PROBABILITY.
3626 * predict.def (PRED_BUILTIN_EXPECT_WITH_PROBABILITY): Add
3628 * tree.h (DECL_BUILT_IN_P): New function.
3630 2018-08-10 Martin Liska <mliska@suse.cz>
3632 PR tree-optimization/85799
3633 * passes.def: Add argument for pass_strip_predict_hints.
3634 * predict.c (class pass_strip_predict_hints): Add new argument
3636 (strip_predictor_early): New function.
3637 (pass_strip_predict_hints::execute): Call the function to
3639 (strip_predict_hints): New function.
3640 * predict.def: Fix comment.
3642 2018-08-10 Thomas Preud'homme <thomas.preudhomme@linaro.org>
3644 * Makefile.in: Clarify which tm.texi to copy over to assert the
3645 right to grant a GFDL license for all.
3647 2018-08-09 Jeff Law <law@redhat.com>
3649 * config/m68k/m68k.c (m68k_adjust_decorated_operand): Remove
3652 2018-08-09 Andreas Schwab <schwab@linux-m68k.org>
3654 * config/m68k/m68k-protos.h (m68k_final_prescan_insn): Remove
3657 2018-08-09 Richard Sandiford <richard.sandiford@arm.com>
3659 * tree-vect-loop.c (vectorizable_reduction): Allow inner-loop
3660 reductions for variable-length vectors.
3662 2018-08-09 David Malcolm <dmalcolm@redhat.com>
3665 * common.opt (fdiagnostics-show-line-numbers): New option.
3666 * diagnostic-show-locus.c (class layout): Add fields
3667 "m_show_line_numbers_p" and "m_linenum_width";
3668 (num_digits): New function.
3669 (test_num_digits): New function.
3670 (layout::layout): Initialize new fields. Update m_x_offset
3671 logic to handle any left margin.
3672 (layout::print_source_line): Print line number when requested.
3673 (layout::start_annotation_line): New member function.
3674 (layout::print_annotation_line): Call it.
3675 (layout::print_leading_fixits): Likewise.
3676 (layout::print_trailing_fixits): Likewise. Update calls to
3677 move_to_column for new parameter.
3678 (layout::get_x_bound_for_row): Add "add_left_margin" param and use
3679 it to potentially call start_annotation_line.
3680 (layout::show_ruler): Call start_annotation_line.
3681 (selftest::test_line_numbers_multiline_range): New selftest.
3682 (selftest::diagnostic_show_locus_c_tests): Call test_num_digits
3683 and selftest::test_line_numbers_multiline_range.
3684 * diagnostic.c (diagnostic_initialize): Initialize
3685 show_line_numbers_p.
3686 * diagnostic.h (struct diagnostic_context): Add field
3687 "show_line_numbers_p".
3688 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
3689 -fno-diagnostics-show-line-numbers.
3690 * dwarf2out.c (gen_producer_string): Add
3691 OPT_fdiagnostics_show_line_numbers to the ignored options.
3692 * lto-wrapper.c (merge_and_complain): Likewise to the "pick
3693 one setting" options.
3694 (append_compiler_options): Likewise to the dropped options.
3695 (append_diag_options): Likewise to the passed-on options.
3696 * opts.c (common_handle_option): Handle the new option.
3697 * toplev.c (general_init): Set up global_dc->show_line_numbers_p.
3699 2018-08-09 Kelvin Nilsen <kelvin@gcc.gnu.org>
3701 * doc/extend.texi (PowerPC AltiVec Built-in Functions Available on
3702 ISA 2.07): Correct spelling of bcdsub to be __builtin_bcdsub. Add
3703 third argument of type "const signed char" to descriptions of
3704 __builtin_bcdadd, __builtin_bcdadd_lt, __builtin_bcdadd_eq,
3705 __builtin_bcdadd_gt, __builtin_bcdadd_ov, __builtin_bcdsub,
3706 __builtin_bcdsub_lt, __builtin_bcdsub_eq, __builtin_bcdsub_gt,
3707 __builtin_bcdsub_ov functions.
3709 2018-08-09 Richard Sandiford <richard.sandiford@arm.com>
3711 PR tree-optimization/86858
3712 * tree-vect-loop.c (vect_is_simple_reduction): Restore
3713 flow_bb_inside_loop_p calls.
3715 2018-08-09 Richard Sandiford <richard.sandiford@arm.com>
3717 PR tree-optimization/86871
3718 * tree-vect-stmts.c (vect_transform_stmt): Use gimple_get_lhs
3719 instead of gimple_assign_lhs.
3721 2018-08-09 Richard Earnshaw <rearnsha@arm.com>
3724 * config/aarch64/aarch64.md (add<mode>3_carryinC_zero): Add missing
3725 register constraint to operand 0.
3726 (add<mode>3_carryinC): Likewise.
3727 (add<mode>3_carryinV_zero, add<mode>3_carryinV): Likewise.
3729 2018-08-09 Martin Liska <mliska@suse.cz>
3732 * common.opt: Remove extra line.
3734 2018-08-09 Martin Liska <mliska@suse.cz>
3736 * params.def (PARAM_ALIGN_LOOP_ITERATIONS): Remove double dots
3737 at the end of a line, make first letter capital and end up
3738 a sentence with a dot.
3739 (PARAM_LOOP_INTERCHANGE_STRIDE_RATIO): Likewise.
3740 (PARAM_LOOP_BLOCK_TILE_SIZE): Likewise.
3741 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Likewise.
3742 (PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP): Likewise.
3743 (PARAM_MAX_ISL_OPERATIONS): Likewise.
3744 (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): Likewise.
3745 (PARAM_PROFILE_FUNC_INTERNAL_ID): Likewise.
3746 (PARAM_INDIR_CALL_TOPN_PROFILE): Likewise.
3747 (PARAM_SLP_MAX_INSNS_IN_BB): Likewise.
3748 (PARAM_IPA_CP_EVAL_THRESHOLD): Likewise.
3749 (PARAM_IPA_CP_RECURSION_PENALTY): Likewise.
3750 (PARAM_IPA_CP_SINGLE_CALL_PENALTY): Likewise.
3751 (PARAM_IPA_CP_LOOP_HINT_BONUS): Likewise.
3752 (PARAM_IPA_CP_ARRAY_INDEX_HINT_BONUS): Likewise.
3753 (PARAM_TREE_REASSOC_WIDTH): Likewise.
3754 (PARAM_HSA_GEN_DEBUG_STORES): Likewise.
3755 (PARAM_MAX_SPECULATIVE_DEVIRT_MAYDEFS): Likewise.
3756 (PARAM_MAX_VRP_SWITCH_ASSERTIONS): Likewise.
3758 2018-08-09 Andreas Krebbel <krebbel@linux.ibm.com>
3761 * config/s390/s390.c (s390_option_override_internal): Reduce the
3762 stack-clash-protection-probe-interval param if it would be too big
3765 2018-08-08 Andreas Schwab <schwab@linux-m68k.org>
3768 * config/m68k/m68k.h (FINAL_PRESCAN_INSN): Don't define.
3769 * config/m68k/m68k.c (handle_move_double): Don't call
3770 m68k_final_prescan_insn.
3771 (m68k_adjust_decorated_operand): Renamed from
3772 m68k_final_prescan_insn, remove first and third operand and
3774 (print_operand): Call it.
3775 (print_operand_address): Call it.
3777 2018-08-08 Nathan Sidwell <nathan@acm.org>
3779 * diagnostic.c (diagnostic_report_current_module): Use
3780 linemap_included_from & linemap_included_from_linemap.
3782 2018-08-08 Hongbo Zhang <hongbo.zhang@linaro.org>
3784 * config/aarch64/aarch64-cores.def: Add phecda core.
3785 * config/aarch64/aarch64-tune.md: Regenerate.
3786 * doc/invoke.texi: Add phecda core.
3788 2018-08-08 Andreas Krebbel <krebbel@linux.ibm.com>
3791 * config/s390/constraints.md ("NxHD0", "NxSD0"): New constraint
3793 * config/s390/s390.md ("movti"): Add more alternatives for
3794 constant to GPR copies.
3796 2018-08-08 Andreas Krebbel <krebbel@linux.ibm.com>
3798 * config/s390/s390.c: Fix whitespace damage throughout the file.
3799 * config/s390/s390.h: Likewise.
3800 * config/s390/tpf.h: Likewise.
3802 2018-08-08 Ilya Leoshkevich <iii@linux.ibm.com>
3804 * config/s390/s390.c (s390_loadrelative_operand_p):
3805 Remove TARGET_CPU_ZARCH usages.
3806 (s390_rtx_costs): Likewise.
3807 (s390_legitimate_constant_p): Likewise.
3808 (s390_cannot_force_const_mem): Likewise.
3809 (legitimate_reload_constant_p): Likewise.
3810 (s390_preferred_reload_class): Likewise.
3811 (legitimize_pic_address): Likewise.
3812 (legitimize_tls_address): Likewise.
3813 (s390_split_branches): Removed.
3814 (s390_add_execute): Removed.
3815 (s390_dump_pool): Remove TARGET_CPU_ZARCH usages.
3816 (s390_mainpool_start): Likewise.
3817 (s390_mainpool_finish): Likewise.
3818 (s390_mainpool_cancel): Removed.
3819 (s390_chunkify_start): Remove TARGET_CPU_ZARCH usages.
3820 (s390_chunkify_cancel): Likewise.
3821 (s390_return_addr_rtx): Likewise.
3822 (s390_register_info): Remove split_branches_pending_p uages.
3823 (s390_optimize_register_info): Likewise.
3824 (s390_init_frame_layout): Remove TARGET_CPU_ZARCH and
3825 split_branches_pending_p usages.
3826 (s390_can_eliminate): Remove TARGET_CPU_ZARCH usages.
3827 (s390_load_got): Likewise.
3828 (s390_expand_split_stack_prologue): Likewise.
3829 (output_asm_nops): Likewise.
3830 (s390_function_profiler): Likewise.
3831 (s390_emit_call): Likewise.
3832 (s390_conditional_register_usage): Likewise.
3833 (s390_optimize_prologue): Likewise.
3834 (s390_reorg): Remove TARGET_CPU_ZARCH and
3835 split_branches_pending_p usages.
3836 (s390_option_override_internal): Remove TARGET_CPU_ZARCH
3838 (s390_output_indirect_thunk_function): Likewise.
3839 * config/s390/s390.h (TARGET_CPU_ZARCH): Removed.
3840 (TARGET_CPU_ZARCH_P): Removed.
3841 (struct machine_function): Remove split_branches_pending_p.
3842 * config/s390/s390.md: Remove TARGET_CPU_ZARCH usages.
3844 2018-08-08 Ilya Leoshkevich <iii@linux.ibm.com>
3846 * common/config/s390/s390-common.c (processor_flags_table):
3848 * config.gcc: Remove with_arch/with_tune support.
3849 * config/s390/2064.md: Remove cpu attribute comparisons.
3850 * config/s390/driver-native.c (s390_host_detect_local_cpu):
3852 * config/s390/linux.h (ASM_SPEC):
3853 Remove -march support.
3854 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal):
3855 Use a table to get an arch level.
3856 * config/s390/s390-opts.h (enum processor_type):
3858 * config/s390/s390.c
3859 (processor_table): Remove entries, add arch_level values.
3860 (s390_issue_rate): Remove cases.
3861 (s390_option_override): Adjust
3862 s390_option_override_internal() call.
3863 (s390_option_override_internal): Remove deprecation warning.
3864 (s390_valid_target_attribute_tree): Adjust
3865 s390_option_override_internal() call.
3866 * config/s390/s390.h (struct s390_processor):
3867 Share with s390-c.c, add arch_level field.
3868 * config/s390/s390.md:
3869 Remove occurrences in cpu attribute.
3870 * config/s390/s390.opt: Remove -march/-mtune support.
3871 * config/s390/tpf.h (ASM_SPEC): Remove -march support.
3872 * doc/invoke.texi: Remove deprecation warning.
3874 2018-08-08 Luis Machado <luis.machado@linaro.org>
3876 * config/aarch64/aarch64.c (qdf24xx_vector_cost): New static
3878 (qdf24xx_tunings): Set vector cost structure to
3879 qdf24xx_vector_cost.
3881 * config/aarch64/aarch64.c (qdf24xx_addrcost_table)
3882 <register_sextend>: Set to 3.
3884 2018-08-07 Richard Sandiford <richard.sandiford@arm.com>
3887 * config/aarch64/iterators.md (FRECP, frecp_suffix): Delete.
3888 * config/aarch64/aarch64-simd.md
3889 (aarch64_frecp<FRECP:frecp_suffix><mode>): Fold FRECPE into...
3890 (@aarch64_frecpe<mode>): ...here and the move FRECPX to...
3891 (aarch64_frecpx<mode>): ...this new pattern.
3892 * config/aarch64/aarch64-simd-builtins.def: Remove comment
3893 about aarch64_frecp<FRECP:frecp_suffix><mode>.
3895 2018-08-07 Martin Liska <mliska@suse.cz>
3898 * predict.c (expr_expected_value_1): Handle DECL_IS_MALLOC,
3899 BUILT_IN_REALLOC and DECL_IS_OPERATOR_NEW.
3900 * predict.def (PRED_MALLOC_NONNULL): New predictor.
3901 * doc/extend.texi: Document that malloc attribute adds
3904 2018-08-06 John David Anglin <danglin@gcc.gnu.org>
3907 * config/pa/pa.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
3908 Define to speculation_safe_value_not_needed.
3910 2018-08-06 Jeff Law <law@redhat.com>
3912 * tree-ssa-dom.c (dom_opt_dom_walker::optimize_stmt): Pass down
3913 the vr_values instance to cprop_into_stmt.
3914 (cprop_into_stmt): Pass vr_values instance down to cprop_operand.
3915 (cprop_operand): Also query EVRP to determine if OP is a constant.
3917 2018-08-06 Nathan Sidwell <nathan@acm.org>
3919 * diagnostic.c (diagnostic_report_current_module): Reroll
3920 included-at loop. Translate text.
3922 2018-08-06 David Malcolm <dmalcolm@redhat.com>
3924 * function-tests.c (selftest::test_expansion_to_rtl): Call
3925 free_after_compilation.
3927 2018-08-06 Alan Hayward <alan.hayward@arm.com>
3929 * config/aarch64/aarch64.md: Add clobber highs to tls_desc.
3931 2018-08-06 Andreas Krebbel <krebbel@linux.ibm.com>
3933 * config/s390/s390.c (s390_loop_unroll_adjust): Prevent small
3934 loops with memory block operations from getting unrolled.
3936 2018-08-06 Ulrich Weigand <uweigand@de.ibm.com>
3939 * config/spu/spu.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
3940 Define to speculation_safe_value_not_needed.
3942 2018-08-06 Jeff Law <law@redhat.com>
3944 * reload1.c (forget_old_reloads_1): Adjust CLOBBER_HIGH
3947 2018-08-06 Jozef Lawrynowicz <jozef.l@mittosystems.com>
3950 * gcc/tree.c (build_common_tree_nodes): Initialize integer_types array
3951 with all enabled __intN types.
3953 * gcc/testsuite/gcc.target/msp430/pr86662.c: New test.
3955 2018-08-06 Alan Hayward <alan.hayward@arm.com>
3957 * alias.c (record_set): Check for clobber high.
3958 * cfgexpand.c (expand_gimple_stmt): Likewise.
3959 * combine-stack-adj.c (single_set_for_csa): Likewise.
3960 * combine.c (find_single_use_1): Likewise.
3961 (set_nonzero_bits_and_sign_copies): Likewise.
3962 (get_combine_src_dest): Likewise.
3963 (is_parallel_of_n_reg_sets): Likewise.
3964 (try_combine): Likewise.
3965 (record_dead_and_set_regs_1): Likewise.
3966 (reg_dead_at_p_1): Likewise.
3967 (reg_dead_at_p): Likewise.
3968 * dce.c (deletable_insn_p): Likewise.
3969 (mark_nonreg_stores_1): Likewise.
3970 (mark_nonreg_stores_2): Likewise.
3971 * df-scan.c (df_find_hard_reg_defs): Likewise.
3972 (df_uses_record): Likewise.
3973 (df_get_call_refs): Likewise.
3974 * dwarf2out.c (mem_loc_descriptor): Likewise.
3975 * haifa-sched.c (haifa_classify_rtx): Likewise.
3976 * ira-build.c (create_insn_allocnos): Likewise.
3977 * ira-costs.c (scan_one_insn): Likewise.
3978 * ira.c (equiv_init_movable_p): Likewise.
3979 (rtx_moveable_p): Likewise.
3980 (interesting_dest_for_shprep): Likewise.
3981 * jump.c (mark_jump_label_1): Likewise.
3982 * postreload-gcse.c (record_opr_changes): Likewise.
3983 * postreload.c (reload_cse_simplify): Likewise.
3984 (struct reg_use): Add source expr.
3985 (reload_combine): Check for clobber high.
3986 (reload_combine_note_use): Likewise.
3987 (reload_cse_move2add): Likewise.
3988 (move2add_note_store): Likewise.
3989 * print-rtl.c (print_pattern): Likewise.
3990 * recog.c (decode_asm_operands): Likewise.
3991 (store_data_bypass_p): Likewise.
3992 (if_test_bypass_p): Likewise.
3993 * regcprop.c (kill_clobbered_value): Likewise.
3994 (kill_set_value): Likewise.
3995 * reginfo.c (reg_scan_mark_refs): Likewise.
3996 * reload1.c (maybe_fix_stack_asms): Likewise.
3997 (eliminate_regs_1): Likewise.
3998 (elimination_effects): Likewise.
3999 (mark_not_eliminable): Likewise.
4000 (scan_paradoxical_subregs): Likewise.
4001 (forget_old_reloads_1): Likewise.
4002 * reorg.c (find_end_label): Likewise.
4003 (try_merge_delay_insns): Likewise.
4004 (redundant_insn): Likewise.
4005 (own_thread_p): Likewise.
4006 (fill_simple_delay_slots): Likewise.
4007 (fill_slots_from_thread): Likewise.
4008 (dbr_schedule): Likewise.
4009 * resource.c (update_live_status): Likewise.
4010 (mark_referenced_resources): Likewise.
4011 (mark_set_resources): Likewise.
4012 * rtl.c (copy_rtx): Likewise.
4013 * rtlanal.c (reg_referenced_p): Likewise.
4014 (single_set_2): Likewise.
4015 (noop_move_p): Likewise.
4016 (note_stores): Likewise.
4017 * sched-deps.c (sched_analyze_reg): Likewise.
4018 (sched_analyze_insn): Likewise.
4020 2018-08-06 Alan Hayward <alan.hayward@arm.com>
4022 * cse.c (invalidate_reg): New function extracted from...
4023 (invalidate): ...here.
4024 (canonicalize_insn): Check for clobber high.
4025 (invalidate_from_clobbers): invalidate clobber highs.
4026 (invalidate_from_sets_and_clobbers): Likewise.
4027 (count_reg_usage): Check for clobber high.
4028 (insn_live_p): Likewise.
4029 * cselib.c (cselib_expand_value_rtx_1):Likewise.
4030 (cselib_invalidate_regno): Check for clobber in setter.
4031 (cselib_invalidate_rtx): Pass through setter.
4032 (cselib_invalidate_rtx_note_stores):
4033 (cselib_process_insn): Check for clobber high.
4034 * cselib.h (cselib_invalidate_rtx): Add operand.
4036 2018-08-06 Alan Hayward <alan.hayward@arm.com>
4038 * lra-eliminations.c (lra_eliminate_regs_1): Check for clobber high.
4039 (mark_not_eliminable): Likewise.
4040 * lra-int.h (struct lra_insn_reg): Add clobber high marker.
4041 * lra-lives.c (process_bb_lives): Check for clobber high.
4042 * lra.c (new_insn_reg): Remember clobber highs.
4043 (collect_non_operand_hard_regs): Check for clobber high.
4044 (lra_set_insn_recog_data): Likewise.
4045 (add_regs_to_insn_regno_info): Likewise.
4046 (lra_update_insn_regno_info): Likewise.
4048 2018-08-06 Alan Hayward <alan.hayward@arm.com>
4050 * rtl.h (reg_is_clobbered_by_clobber_high): Add declarations.
4051 * rtlanal.c (reg_is_clobbered_by_clobber_high): Add function.
4053 2018-08-06 Alan Hayward <alan.hayward@arm.com>
4055 * emit-rtl.c (verify_rtx_sharing): Check for CLOBBER_HIGH.
4056 (copy_insn_1): Likewise.
4057 (gen_hard_reg_clobber_high): New gen function.
4058 * genconfig.c (walk_insn_part): Check for CLOBBER_HIGH.
4059 * genemit.c (gen_exp): Likewise.
4060 (gen_emit_seq): Pass through info.
4061 (gen_insn): Check for CLOBBER_HIGH.
4062 (gen_expand): Pass through info.
4063 (gen_split): Likewise.
4064 (output_add_clobbers): Likewise.
4065 * genrecog.c (validate_pattern): Check for CLOBBER_HIGH.
4066 (remove_clobbers): Likewise.
4067 * rtl.h (gen_hard_reg_clobber_high): New declaration.
4069 2018-08-06 Alan Hayward <alan.hayward@arm.com>
4071 * doc/rtl.texi (clobber_high): Add.
4072 (parallel): Add in clobber high
4073 * rtl.c (rtl_check_failed_code3): Add function.
4074 * rtl.def (CLOBBER_HIGH): Add expression.
4075 * rtl.h (RTL_CHECKC3): Add macro.
4076 (rtl_check_failed_code3): Add declaration.
4077 (XC3EXP): Add macro.
4079 2018-08-05 H.J. Lu <hongjiu.lu@intel.com>
4082 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Set
4083 cfun->machine->max_used_stack_alignment if needed.
4085 2018-08-04 Martin Sebor <msebor@redhat.com>
4087 PR tree-optimization/86571
4088 * gimple-ssa-sprintf.c (format_floating): Extend upper bound of
4091 2018-08-03 Sandra Loosemore <sandra@codesourcery.com>
4094 * config/nios2/nios2.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
4097 2018-08-03 Jeff Law <law@redhat.com>
4100 * config/mn10300/mn10300.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
4101 Define to speculation_safe_value_not_needed.
4103 2018-08-03 David Malcolm <dmalcolm@redhat.com>
4105 * doc/gcov.texi (-x): Remove duplicate "to".
4106 * doc/invoke.texi (-Wnoexcept-type): Remove duplicate "calls".
4107 (-Wif-not-aligned): Remove duplicate "is".
4108 (-flto): Remove duplicate "the".
4109 (MicroBlaze Options): In examples of "-mcpu=cpu-type", remove
4110 duplicate "v5.00.b".
4111 (MSP430 Options): Remove duplicate "and" from the description
4112 of "-mgprel-sec=regexp".
4113 (x86 Options): Remove duplicate copies of "vmldLog102" and
4114 vmlsLog104 from description of "-mveclibabi=type".
4116 2018-08-03 Richard Sandiford <richard.sandiford@arm.com>
4118 * internal-fn.h (first_commutative_argument): Declare.
4119 * internal-fn.c (first_commutative_argument): New function.
4120 * tree-vect-slp.c (vect_get_and_check_slp_defs): Remove extra
4121 restrictions for pattern statements. Use first_commutative_argument
4122 to look for commutative operands in calls to internal functions.
4124 2018-08-03 Aldy Hernandez <aldyh@redhat.com>
4126 * Makefile.in (wide-int-range.o): New.
4127 * tree-vrp.c: Move all the wide_int_* functions to...
4128 * wide-int-range.cc: ...here.
4129 * tree-vrp.h: Move all the wide_int_* prototypes to...
4130 * wide-int-range.h: ...here.
4132 2018-08-03 Tom de Vries <tdevries@suse.de>
4134 * common/config/nvptx/nvptx-common.c (nvptx_except_unwind_info): Return
4136 * config/nvptx/nvptx.c (TARGET_ASM_BYTE_OP): Remove define.
4137 * except.c (output_function_exception_table): Do early exit if
4138 targetm_common.except_unwind_info (&global_options) == UI_NONE.
4140 2018-08-03 Martin Liska <mliska@suse.cz>
4142 * predict.c (dump_prediction): Change to 2 digits
4145 2018-08-03 Siddhesh Poyarekar <siddhesh@sourceware.org>
4147 * config/aarch64/falkor.md (falkor_am_1_vxvy_vxvy): Move
4149 (falkor_am_1_gtov_gtov): ... a new insn reservation.
4151 2018-07-19 Ilya Leoshkevich <iii@linux.ibm.com>
4153 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Replace > with >=.
4154 * df-problems.c (df_remove_dead_eq_notes): Replace > with >=.
4155 * dwarf2out.c (mem_loc_descriptor): Replace > with >=.
4156 * lra-constraints.c (spill_hard_reg_in_range): Replace <= with <.
4157 * lra-remat.c (call_used_input_regno_present_p): Replace <= with <.
4159 2018-08-02 David Malcolm <dmalcolm@redhat.com>
4161 * diagnostic-show-locus.c (diagnostic_show_locus): Use
4162 pp_take_prefix when saving the existing prefix.
4163 * diagnostic.c (diagnostic_append_note): Likewise.
4164 * langhooks.c (lhd_print_error_function): Likewise.
4165 * pretty-print.c (pp_set_prefix): Drop the "const" from "prefix"
4166 param's type. Free the existing prefix.
4167 (pp_take_prefix): New function.
4168 (pretty_printer::pretty_printer): Drop the prefix parameter.
4169 Rename the length parameter to match the comment.
4170 (pretty_printer::~pretty_printer): Free the prefix.
4171 * pretty-print.h (pretty_printer::pretty_printer): Drop the prefix
4173 (struct pretty_printer): Drop the "const" from "prefix" field's
4174 type and clarify memory management.
4175 (pp_set_prefix): Drop the "const" from the 2nd param.
4176 (pp_take_prefix): New decl.
4178 2018-08-02 Aaron Sawdey <acsawdey@linux.ibm.com>
4180 * config/rs6000/rs6000-string.c (select_block_compare_mode): Move test
4181 for word_mode_ok here instead of passing as argument.
4182 (expand_block_compare): Change select_block_compare_mode() call.
4183 (expand_strncmp_gpr_sequence): New function.
4184 (expand_strn_compare): Make use of expand_strncmp_gpr_sequence.
4186 2018-08-02 Jeff Law <law@redhat.com>
4189 * config/m68k/m68k.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
4190 Define to speculation_safe_value_not_needed.
4193 * config/h8300/h8300.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
4194 Define to speculation_safe_value_not_needed.
4196 2018-08-02 Tom de Vries <tdevries@suse.de>
4199 * common/config/nvptx/nvptx-common.c (nvptx_except_unwind_info): New
4200 function. Return UI_TARGET unconditionally.
4201 (TARGET_EXCEPT_UNWIND_INFO): Redefine to nvptx_except_unwind_info.
4202 * config/nvptx/nvptx.c (TARGET_ASM_BYTE_OP): Emit commented out '.byte'.
4204 2018-08-02 Richard Sandiford <richard.sandiford@arm.com>
4206 * genemit.c (print_overload_test): Fix typo.
4208 2018-08-02 Richard Biener <rguenther@suse.de>
4210 PR tree-optimization/86816
4211 * tree-ssa-tail-merge.c (tail_merge_valueize): New function
4212 which checks for value availability before querying it.
4213 (gvn_uses_equal): Use it.
4214 (same_succ_hash): Likewise.
4215 (gimple_equal_p): Likewise.
4217 2018-08-02 Nick Clifton <nickc@redhat.com>
4220 * config/stormy16/stormy16.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
4221 Define to speculation_safe_value_not_needed.
4224 * config/v850/v850.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
4225 Define to speculation_safe_value_not_needed.
4228 * config/v850/v850.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
4229 Define to speculation_safe_value_not_needed.
4232 * config/rx/rx.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
4233 Define to speculation_safe_value_not_needed.
4236 * config/msp430/msp430.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
4237 Define to speculation_safe_value_not_needed.
4240 * config/mcore/mcore.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
4241 Define to speculation_safe_value_not_needed.
4244 * config/m32r/m32r.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
4245 Define to speculation_safe_value_not_needed.
4248 * config/iq2000/iq2000.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
4249 Define to speculation_safe_value_not_needed.
4252 * config/frv/frv.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Define to
4253 speculation_safe_value_not_needed.
4256 * config/fr30/fr30.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Define
4257 to speculation_safe_value_not_needed.
4259 2018-08-02 Richard Sandiford <richard.sandiford@arm.com>
4261 * doc/md.texi: Expand the documentation of instruction names
4262 to mention port-local uses. Document '@' in pattern names.
4263 * read-md.h (overloaded_instance, overloaded_name): New structs.
4265 (md_reader::handle_overloaded_name): New member function.
4266 (md_reader::get_overloads): Likewise.
4267 (md_reader::m_first_overload): New member variable.
4268 (md_reader::m_next_overload_ptr): Likewise.
4269 (md_reader::m_overloads_htab): Likewise.
4270 * read-md.c (md_reader::md_reader): Initialize m_first_overload,
4271 m_next_overload_ptr and m_overloads_htab.
4272 * read-rtl.c (iterator_group): Add "type" and "get_c_token" fields.
4273 (get_mode_token, get_code_token, get_int_token): New functions.
4274 (map_attr_string): Add an optional argument that passes back
4275 the associated iterator.
4276 (overloaded_name_hash, overloaded_name_eq_p, named_rtx_p):
4277 (md_reader::handle_overloaded_name, add_overload_instance): New
4279 (apply_iterators): Handle '@' names. Report an error if '@'
4280 is used without iterators.
4281 (initialize_iterators): Initialize the new iterator_group fields.
4282 * genopinit.c (handle_overloaded_code_for)
4283 (handle_overloaded_gen): New functions.
4284 (main): Use them to print declarations of maybe_code_for_* and
4285 maybe_gen_* functions, and inline definitions of code_for_* and gen_*.
4286 * genemit.c (print_overload_arguments, print_overload_test)
4287 (handle_overloaded_code_for, handle_overloaded_gen): New functions.
4288 (main): Use it to print definitions of maybe_code_for_* and
4289 maybe_gen_* functions.
4290 * config/aarch64/aarch64.c (aarch64_split_128bit_move): Use
4291 gen_aarch64_mov{low,high}_di and gen_aarch64_movdi_{low,high}
4292 instead of explicit mode checks.
4293 (aarch64_split_simd_combine): Likewise gen_aarch64_simd_combine.
4294 (aarch64_split_simd_move): Likewise gen_aarch64_split_simd_mov.
4295 (aarch64_emit_load_exclusive): Likewise gen_aarch64_load_exclusive.
4296 (aarch64_emit_store_exclusive): Likewise gen_aarch64_store_exclusive.
4297 (aarch64_expand_compare_and_swap): Likewise
4298 gen_aarch64_compare_and_swap and gen_aarch64_compare_and_swap_lse
4299 (aarch64_gen_atomic_cas): Likewise gen_aarch64_atomic_cas.
4300 (aarch64_emit_atomic_swap): Likewise gen_aarch64_atomic_swp.
4301 (aarch64_constant_pool_reload_icode): Delete.
4302 (aarch64_secondary_reload): Use code_for_aarch64_reload_movcp
4303 instead of aarch64_constant_pool_reload_icode. Use
4304 code_for_aarch64_reload_mov instead of explicit mode checks.
4305 (rsqrte_type, get_rsqrte_type, rsqrts_type, get_rsqrts_type): Delete.
4306 (aarch64_emit_approx_sqrt): Use gen_aarch64_rsqrte instead of
4307 get_rsqrte_type and gen_aarch64_rsqrts instead of gen_rqrts_type.
4308 (recpe_type, get_recpe_type, recps_type, get_recps_type): Delete.
4309 (aarch64_emit_approx_div): Use gen_aarch64_frecpe instead of
4310 get_recpe_type and gen_aarch64_frecps instead of get_recps_type.
4311 (aarch64_atomic_load_op_code): Delete.
4312 (aarch64_emit_atomic_load_op): Likewise.
4313 (aarch64_gen_atomic_ldop): Use UNSPECV_ATOMIC_* instead of
4314 aarch64_atomic_load_op_code. Use gen_aarch64_atomic_load
4315 instead of aarch64_emit_atomic_load_op.
4316 * config/aarch64/aarch64.md (aarch64_reload_movcp<GPF_TF:mode><P:mode>)
4317 (aarch64_reload_movcp<VALL:mode><P:mode>, aarch64_reload_mov<mode>)
4318 (aarch64_movdi_<mode>low, aarch64_movdi_<mode>high)
4319 (aarch64_mov<mode>high_di, aarch64_mov<mode>low_di): Add a '@'
4320 character before the pattern name.
4321 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
4322 (aarch64_rsqrte<mode>, aarch64_rsqrts<mode>)
4323 (aarch64_simd_combine<mode>, aarch64_frecpe<mode>)
4324 (aarch64_frecps<mode>): Likewise.
4325 * config/aarch64/atomics.md (atomic_compare_and_swap<mode>)
4326 (aarch64_compare_and_swap<mode>, aarch64_compare_and_swap<mode>_lse)
4327 (aarch64_load_exclusive<mode>, aarch64_store_exclusive<mode>)
4328 (aarch64_atomic_swp<mode>, aarch64_atomic_cas<mode>)
4329 (aarch64_atomic_load<atomic_ldop><mode>): Likewise.
4331 2018-08-02 Richard Sandiford <richard.sandiford@arm.com>
4333 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
4334 Allow HFmode constants if TARGET_FP_F16INST.
4336 2018-08-02 Jackson Woodruff <jackson.woodruff@arm.com>
4339 * config/aarch64/aarch64.c (aarch64_operands_adjust_ok_for_ldpstp):
4340 No longer check last store for clobber of address register.
4342 2018-08-02 Martin Liska <mliska@suse.cz>
4344 PR gcov-profile/86817
4345 * gcov.c (process_all_functions): New function.
4347 (process_file): Move functions processing to
4348 process_all_functions.
4350 2018-08-02 David Malcolm <dmalcolm@redhat.com>
4352 * dumpfile.c (dump_user_location_t::dump_user_location_t): Add
4353 "const" to the "gimple *" and "rtx_insn *" parameters.
4354 * dumpfile.h (dump_user_location_t::dump_user_location_t):
4356 (dump_location_t::dump_location_t): Likewise.
4358 2018-08-01 Martin Sebor <msebor@redhat.com>
4360 PR tree-optimization/86650
4361 * gimple-pretty-print.c (percent_G_format): Accept a "gimple *"
4362 rather than a "gcall *". Directly pass the data of interest
4363 to percent_K_format, rather than building a temporary CALL_EXPR
4365 * gimple-fold.c (gimple_fold_builtin_strncpy): Adjust.
4366 (gimple_fold_builtin_strncat): Adjust.
4367 * gimple-ssa-warn-restrict.h (check_bounds_or_overlap): Replace
4368 gcall* argument with gimple*.
4369 * gimple-ssa-warn-restrict.c (check_call): Same.
4370 (wrestrict_dom_walker::before_dom_children): Same.
4371 (builtin_access::builtin_access): Same.
4372 (check_bounds_or_overlap): Same
4373 (maybe_diag_overlap): Same.
4374 (maybe_diag_offset_bounds): Same.
4375 * tree-diagnostic.c (default_tree_printer): Move usage of
4376 EXPR_LOCATION (t) and TREE_BLOCK (t) from within percent_K_format
4378 * tree-pretty-print.c (percent_K_format): Add argument.
4379 * tree-pretty-print.h: Add argument.
4380 * tree-ssa-ccp.c (pass_post_ipa_warn::execute): Adjust.
4381 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Adjust.
4382 (maybe_diag_stxncpy_trunc): Same.
4383 (handle_builtin_stxncpy): Same.
4384 (handle_builtin_strcat): Same.
4386 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
4388 * match.pd: Optimise pointer range checks.
4390 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
4392 PR tree-optimization/86758
4393 * tree-vect-stmts.c (vectorizable_simd_clone_call): Don't try
4394 to remove pattern statements.
4396 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
4398 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Use the
4399 result of dfs_enumerate_from when constructing stmt_vec_infos,
4400 instead of additionally calling get_loop_body.
4402 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
4404 * tree-vectorizer.h (vect_create_data_ref_ptr): Remove inv_p
4406 * tree-vect-data-refs.c (vect_create_data_ref_ptr): Likewise.
4407 When creating an iv, assert that the step is not known to be zero.
4408 (vect_setup_realignment): Update call accordingly.
4409 * tree-vect-stmts.c (vectorizable_store): Likewise.
4410 (vectorizable_load): Likewise. Handle VMAT_INVARIANT separately.
4412 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
4414 * tree-vectorizer.h (vect_stmt_to_vectorize): New function.
4415 * tree-vect-loop.c (vect_update_vf_for_slp): Use it.
4416 (vectorizable_reduction): Likewise.
4417 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
4418 (vect_detect_hybrid_slp_stmts): Likewise.
4419 * tree-vect-stmts.c (vect_is_simple_use): Likewise.
4421 2018-08-01 Aldy Hernandez <aldyh@redhat.com>
4423 * tree-vrp (zero_nonzero_bits_from_bounds): Rename to...
4424 (wide_int_set_zero_nonzero_bits): ...this.
4425 (zero_nonzero_bits_from_vr): Rename to...
4426 (vrp_set_zero_nonzero_bits): ...this.
4427 (extract_range_from_multiplicative_op_1): Abstract wide int
4429 (wide_int_range_multiplicative_op): ...here.
4430 (extract_range_from_binary_expr_1): Extract wide int binary
4431 operations into their own functions.
4432 (wide_int_range_lshift): New.
4433 (wide_int_range_can_optimize_bit_op): New.
4434 (wide_int_range_shift_undefined_p): New.
4435 (wide_int_range_bit_xor): New.
4436 (wide_int_range_bit_ior): New.
4437 (wide_int_range_bit_and): New.
4438 (wide_int_range_trunc_mod): New.
4439 (extract_range_into_wide_ints): New.
4440 (vrp_shift_undefined_p): New.
4441 (extract_range_from_multiplicative_op): New.
4442 (vrp_can_optimize_bit_op): New.
4443 * tree-vrp.h (value_range::dump): New.
4444 (wide_int_range_multiplicative_op): New.
4445 (wide_int_range_lshift):New.
4446 (wide_int_range_shift_undefined_p): New.
4447 (wide_int_range_bit_xor): New.
4448 (wide_int_range_bit_ior): New.
4449 (wide_int_range_bit_and): New.
4450 (wide_int_range_trunc_mod): New.
4451 (zero_nonzero_bits_from_bounds): Rename to...
4452 (wide_int_set_zero_nonzero_bits): ...this.
4453 (zero_nonzero_bits_from_vr): Rename to...
4454 (vrp_set_zero_nonzero_bits): ...this.
4455 (range_easy_mask_min_max): Rename to...
4456 (wide_int_range_can_optimize_bit_op): this.
4457 * vr-values.c (simplify_bit_ops_using_ranges): Rename
4458 zero_nonzero_bits_from_vr into vrp_set_zero_nonzero_bits.
4460 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
4462 * tree-vectorizer.h (vect_orig_stmt): New function.
4463 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Use it.
4464 * tree-vect-loop.c (vect_model_reduction_cost): Likewise.
4465 (vect_create_epilog_for_reduction): Likewise.
4466 (vectorizable_live_operation): Likewise.
4467 * tree-vect-slp.c (vect_find_last_scalar_stmt_in_slp): Likewise.
4468 (vect_detect_hybrid_slp_stmts, vect_schedule_slp): Likewise.
4469 * tree-vect-stmts.c (vectorizable_call): Likewise.
4470 (vectorizable_simd_clone_call, vect_remove_stores): Likewise.
4472 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
4474 * tree-vectorizer.h (vect_transform_stmt): Remove grouped_store
4476 * tree-vect-stmts.c (vect_transform_stmt): Likewise.
4477 * tree-vect-loop.c (vect_transform_loop_stmt): Update call accordingly.
4478 (vect_transform_loop): Likewise.
4479 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
4481 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
4483 * tree-vectorizer.h (vect_schedule_slp): Return void.
4484 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
4485 (vect_schedule_slp): Likewise.
4487 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
4489 * tree-vect-loop.c (vect_transform_loop_stmt): Remove slp_scheduled
4491 (vect_transform_loop): Update calls accordingly. Schedule SLP
4492 instances before the main loop, if any exist.
4494 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
4496 PR tree-optimization/86749
4497 * tree-vect-patterns.c (vect_determine_min_output_precision_1):
4498 If the lhs is used in a COND_EXPR, check that it is being used
4499 as the "then" or "else" value.
4501 2018-08-01 Tom de Vries <tdevries@suse.de>
4504 * config/nvptx/nvptx.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Define to
4505 speculation_safe_value_not_needed.
4507 2018-08-01 Richard Biener <rguenther@suse.de>
4509 * tree-ssa-sccvn.c (visit_phi): Compare invariant addresses
4512 2018-08-01 Martin Liska <mliska@suse.cz>
4514 * value-prof.c (gimple_divmod_fixed_value_transform): Unify
4515 format how successful transformation is dumped.
4516 (gimple_mod_pow2_value_transform): Likewise.
4517 (gimple_mod_subtract_transform): Likewise.
4518 (gimple_stringops_transform): Likewise.
4520 2018-08-01 Martin Liska <mliska@suse.cz>
4523 * value-prof.c (interesting_stringop_to_profile_p):
4524 Simplify the code and add BUILT_IN_MEMMOVE.
4525 (gimple_stringops_transform): Likewise.
4527 2018-08-01 Sam Tebbs <sam.tebbs@arm.com>
4529 * config/aarch64/aarch64-simd.md
4530 (*aarch64_get_lane_zero_extendsi<mode>): Rename to...
4531 (*aarch64_get_lane_zero_extend<GPI:mode><VDQQH:mode>): ... This and
4532 use GPI iterator instead of SI mode.
4534 2018-08-01 Richard Earnshaw <rearnsha@arm.com>
4536 * config/rs6000/rs6000.md (speculation_barrier): Renamed from
4537 rs6000_speculation_barrier.
4538 * config/rs6000/rs6000.c (rs6000_expand_builtin): Adjust for
4539 new barrier pattern name.
4541 2018-08-01 Richard Earnshaw <rearnsha@arm.com>
4543 * config/i386/i386.md (unspecv): Add UNSPECV_SPECULATION_BARRIER.
4544 (speculation_barrier): New insn.
4546 2018-08-01 Richard Biener <rguenther@suse.de>
4549 * graphite.h: Include isl/id.h and isl/space.h to allow build
4552 2018-08-01 Jan Willem Jagersma <jwjagersma@gmail.com>
4555 * dwarf2out.c (dwarf2out_early_finish): Do not generate assembly in LTO
4556 mode for COFF targets.
4557 * defaults.h (TARGET_COFF): Define.
4558 * config/i386/djgpp.h (TARGET_ASM_LTO_START, TARGET_ASM_LTO_END,
4559 TARGET_COFF): Define.
4560 (i386_djgpp_asm_lto_start, i386_djgpp_asm_lto_end): Declare.
4561 * config/i386/djgpp.c (saved_debug_info_level): New static variable.
4562 (i386_djgpp_asm_lto_start, i386_djgpp_asm_lto_end): New functions.
4564 2018-07-31 Alexandre Oliva <oliva@adacore.com>
4566 * gimple-streamer-in.c (input_bb): Restore BB discriminator.
4567 * gimple-streamer-out.c (output_bb): Save it.
4568 * lto-streamer-in.c (input_struct_function_base): Restore
4569 instance discriminator if available. Create map on demand.
4570 * lto-streamer-out.c (output_struct_function_base): Save it if
4572 * final.c (decl_to_instance_map): Document LTO strategy.
4574 2018-07-31 Alexandre Oliva <oliva@adacore.com>
4575 Olivier Hainque <hainque@adacore.com>
4577 * debug.h (decl_to_instance_map_t): New type.
4578 (decl_to_instance_map): Declare.
4579 (maybe_create_decl_to_instance_map): New inline function.
4580 * final.c (bb_discriminator, last_bb_discriminator): New statics,
4581 to track basic block discriminators.
4582 (final_start_function_1): Initialize them.
4583 (final_scan_insn_1): On NOTE_INSN_BASIC_BLOCK, track
4585 (decl_to_instance_map): New variable.
4586 (map_decl_to_instance, maybe_set_discriminator): New functions.
4587 (notice_source_line): Set discriminator.
4589 2018-07-31 Ian Lance Taylor <iant@golang.org>
4591 * targhooks.c (default_have_speculation_safe_value): Add
4594 2018-07-31 David Malcolm <dmalcolm@redhat.com>
4596 * dump-context.h: Include "pretty-print.h".
4597 (dump_context::refresh_dumps_are_enabled): New decl.
4598 (dump_context::emit_item): New decl.
4599 (class dump_context): Add fields "m_test_pp" and
4601 (temp_dump_context::temp_dump_context): Add param "test_pp_flags".
4602 (temp_dump_context::get_dumped_text): New decl.
4603 (class temp_dump_context): Add field "m_pp".
4604 * dumpfile.c (refresh_dumps_are_enabled): Convert to...
4605 (dump_context::refresh_dumps_are_enabled): ...and add a test for
4607 (set_dump_file): Update for above change.
4608 (set_alt_dump_file): Likewise.
4609 (dump_loc): New overload, taking a pretty_printer *.
4610 (dump_context::dump_loc): Call end_any_optinfo. Dump the location
4611 to any test pretty-printer.
4612 (make_item_for_dump_gimple_stmt): New function, adapted from
4613 optinfo::add_gimple_stmt.
4614 (dump_context::dump_gimple_stmt): Call it, and use the result,
4615 eliminating the direct usage of dump_file and alt_dump_file in
4616 favor of indirectly using them via emit_item.
4617 (make_item_for_dump_gimple_expr): New function, adapted from
4618 optinfo::add_gimple_expr.
4619 (dump_context::dump_gimple_expr): Call it, and use the result,
4620 eliminating the direct usage of dump_file and alt_dump_file in
4621 favor of indirectly using them via emit_item.
4622 (make_item_for_dump_generic_expr): New function, adapted from
4624 (dump_context::dump_generic_expr): Call it, and use the result,
4625 eliminating the direct usage of dump_file and alt_dump_file in
4626 favor of indirectly using them via emit_item.
4627 (make_item_for_dump_printf_va): New function, adapted from
4628 optinfo::add_printf_va.
4629 (make_item_for_dump_printf): New function.
4630 (dump_context::dump_printf_va): Call make_item_for_dump_printf_va,
4631 and use the result, eliminating the direct usage of dump_file and
4632 alt_dump_file in favor of indirectly using them via emit_item.
4633 (make_item_for_dump_dec): New function.
4634 (dump_context::dump_dec): Call it, and use the result,
4635 eliminating the direct usage of dump_file and alt_dump_file in
4636 favor of indirectly using them via emit_item.
4637 (make_item_for_dump_symtab_node): New function, adapted from
4638 optinfo::add_symtab_node.
4639 (dump_context::dump_symtab_node): Call it, and use the result,
4640 eliminating the direct usage of dump_file and alt_dump_file in
4641 favor of indirectly using them via emit_item.
4642 (dump_context::begin_scope): Reimplement, avoiding direct usage
4643 of dump_file and alt_dump_file in favor of indirectly using them
4645 (dump_context::emit_item): New member function.
4646 (temp_dump_context::temp_dump_context): Add param "test_pp_flags".
4647 Set up test pretty-printer on the underlying context. Call
4648 refresh_dumps_are_enabled.
4649 (temp_dump_context::~temp_dump_context): Call
4650 refresh_dumps_are_enabled.
4651 (temp_dump_context::get_dumped_text): New member function.
4652 (selftest::verify_dumped_text): New function.
4653 (ASSERT_DUMPED_TEXT_EQ): New macro.
4654 (selftest::test_capture_of_dump_calls): Run all tests twice, with
4655 and then without optinfo enabled. Add uses of
4656 ASSERT_DUMPED_TEXT_EQ to all tests. Add test of nested scopes.
4657 * dumpfile.h: Update comment for the dump_* API.
4658 * optinfo-emit-json.cc
4659 (selftest::test_building_json_from_dump_calls): Update for new
4660 param for temp_dump_context ctor.
4661 * optinfo.cc (optinfo_item::optinfo_item): Remove "owned" param
4662 and "m_owned" field.
4663 (optinfo_item::~optinfo_item): Likewise.
4664 (optinfo::add_item): New member function.
4665 (optinfo::emit): Update comment.
4666 (optinfo::add_string): Delete.
4667 (optinfo::add_printf): Delete.
4668 (optinfo::add_printf_va): Delete.
4669 (optinfo::add_gimple_stmt): Delete.
4670 (optinfo::add_gimple_expr): Delete.
4671 (optinfo::add_tree): Delete.
4672 (optinfo::add_symtab_node): Delete.
4673 (optinfo::add_dec): Delete.
4674 * optinfo.h (class dump_context): New forward decl.
4675 (optinfo::add_item): New decl.
4676 (optinfo::add_string): Delete.
4677 (optinfo::add_printf): Delete.
4678 (optinfo::add_printf_va): Delete.
4679 (optinfo::add_gimple_stmt): Delete.
4680 (optinfo::add_gimple_expr): Delete.
4681 (optinfo::add_tree): Delete.
4682 (optinfo::add_symtab_node): Delete.
4683 (optinfo::add_dec): Delete.
4684 (optinfo::add_poly_int): Delete.
4685 (optinfo_item::optinfo_item): Remove "owned" param.
4686 (class optinfo_item): Remove field "m_owned".
4688 2018-07-31 Jozef Lawrynowicz <jozef.l@mittosystems.com>
4691 * gcc/cfgexpand.c (set_parm_rtl): Use the alignment of Pmode when
4692 MAX_SUPPORTED_STACK_ALIGNMENT would otherwise be exceeded by the
4693 requested variable alignment.
4694 (expand_one_ssa_partition): Likewise.
4695 (expand_one_var): Likewise.
4697 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
4699 * config/pdp11/pdp11.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Redefine
4700 to speculation_safe_value_not_needed.
4702 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
4704 * targhooks.h (speculation_safe_value_not_needed): New prototype.
4705 * targhooks.c (speculation_safe_value_not_needed): New function.
4706 * target.def (have_speculation_safe_value): Update documentation.
4707 * doc/tm.texi: Regenerated.
4709 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
4711 * config/aarch64/iterators.md (ALLI_TI): New iterator.
4712 * config/aarch64/aarch64.md (despeculate_copy<ALLI_TI:mode>): New
4714 (despeculate_copy<ALLI:mode>_insn): New insn.
4715 (despeculate_copyti_insn): New insn.
4716 (despeculate_simple<ALLI:mode>): New insn
4717 (despeculate_simpleti): New insn.
4718 * config/aarch64/aarch64.c (aarch64_speculation_safe_value): New
4720 (TARGET_SPECULATION_SAFE_VALUE): Redefine to
4721 aarch64_speculation_safe_value.
4722 (aarch64_print_operand): Handle const0_rtx in modifier 'H'.
4724 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
4726 * config/aarch64/aarch64-speculation.cc: New file.
4727 * config/aarch64/aarch64-passes.def (pass_track_speculation): Add
4728 before pass_reorder_blocks.
4729 * config/aarch64/aarch64-protos.h (make_pass_track_speculation): Add
4731 * config/aarch64/aarch64.c (aarch64_conditional_register_usage): Fix
4732 X14 and X15 when tracking speculation.
4733 * config/aarch64/aarch64.md (register name constants): Add
4734 SPECULATION_TRACKER_REGNUM and SPECULATION_SCRATCH_REGNUM.
4735 (unspec): Add UNSPEC_SPECULATION_TRACKER.
4736 (speculation_barrier): New insn attribute.
4737 (cmp<mode>): Allow SP in comparisons.
4738 (speculation_tracker): New insn.
4739 (speculation_barrier): Add speculation_barrier attribute.
4740 * config/aarch64/t-aarch64: Add make rule for aarch64-speculation.o.
4741 * config.gcc (aarch64*-*-*): Add aarch64-speculation.o to extra_objs.
4742 * doc/invoke.texi (AArch64 Options): Document -mtrack-speculation.
4744 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
4746 * config/aarch64/aarch64.md (cb<optab><mode>1): Disable when
4747 aarch64_track_speculation is true.
4748 (tb<optab><mode>1): Likewise.
4749 * config/aarch64/aarch64.c (aarch64_split_compare_regs): Do not
4750 generate CB[N]Z when tracking speculation.
4751 (aarch64_split_compare_and_swap): Likewise.
4752 (aarch64_split_atomic_op): Likewise.
4754 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
4756 * config/aarch64/aarch64.opt (mtrack-speculation): New target option.
4758 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
4760 * config/aarch64.md (unspecv): Add UNSPECV_SPECULAION_BARRIER.
4761 (speculation_barrier): New insn.
4763 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
4765 * config/arm/unspecs.md (unspecv): Add VUNSPEC_SPECULATION_BARRIER.
4766 * config/arm/arm.md (speculation_barrier): New expand.
4767 (speculation_barrier_insn): New pattern.
4769 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
4771 * builtin-types.def (BT_FN_PTR_PTR_VAR): New function type.
4772 (BT_FN_I1_I1_VAR, BT_FN_I2_I2_VAR, BT_FN_I4_I4_VAR): Likewise.
4773 (BT_FN_I8_I8_VAR, BT_FN_I16_I16_VAR): Likewise.
4774 * builtin-attrs.def (ATTR_NOVOPS_NOTHROW_LEAF_LIST): New attribute
4776 * builtins.def (BUILT_IN_SPECULATION_SAFE_VALUE_N): New builtin.
4777 (BUILT_IN_SPECULATION_SAFE_VALUE_PTR): New internal builtin.
4778 (BUILT_IN_SPECULATION_SAFE_VALUE_1): Likewise.
4779 (BUILT_IN_SPECULATION_SAFE_VALUE_2): Likewise.
4780 (BUILT_IN_SPECULATION_SAFE_VALUE_4): Likewise.
4781 (BUILT_IN_SPECULATION_SAFE_VALUE_8): Likewise.
4782 (BUILT_IN_SPECULATION_SAFE_VALUE_16): Likewise.
4783 * builtins.c (expand_speculation_safe_value): New function.
4784 (expand_builtin): Call it.
4785 * doc/cpp.texi: Document predefine __HAVE_SPECULATION_SAFE_VALUE.
4786 * doc/extend.texi: Document __builtin_speculation_safe_value.
4787 * doc/md.texi: Document "speculation_barrier" pattern.
4788 * doc/tm.texi.in: Pull in TARGET_SPECULATION_SAFE_VALUE and
4789 TARGET_HAVE_SPECULATION_SAFE_VALUE.
4790 * doc/tm.texi: Regenerated.
4791 * target.def (have_speculation_safe_value, speculation_safe_value): New
4793 * targhooks.c (default_have_speculation_safe_value): New function.
4794 (default_speculation_safe_value): New function.
4795 * targhooks.h (default_have_speculation_safe_value): Add prototype.
4796 (default_speculation_safe_value): Add prototype.
4798 2018-07-31 David Malcolm <dmalcolm@redhat.com>
4800 * dump-context.h (dump_context::dump_loc): New decl.
4801 * dumpfile.c (dump_context::dump_loc): New member function.
4802 (dump_context::dump_gimple_stmt_loc): Reimplement using dump_loc
4803 and dump_gimple_stmt.
4804 (dump_context::dump_gimple_expr_loc): Likewise, using
4806 (dump_context::dump_generic_expr_loc): Likewise, using
4808 (dump_context::dump_printf_loc_va): Likewise, using
4810 (dump_context::begin_scope): Explicitly using the global function
4811 "dump_loc", rather than the member function.
4813 2018-07-31 Martin Sebor <msebor@redhat.com>
4815 PR tree-optimization/86741
4816 * tree-vrp.c (vrp_prop::check_mem_ref): Avoid incomplete types.
4818 2018-07-31 Andreas Krebbel <krebbel@linux.ibm.com>
4820 * config/s390/s390.c (s390_expand_setmem): Make the unrolling to
4821 depend on whether prefetch instructions will be emitted or not.
4822 Use TARGET_SETMEM_PFD for checking whether prefetch instructions
4823 will be emitted or not.
4824 * config/s390/s390.h (TARGET_SETMEM_PREFETCH_DISTANCE)
4825 (TARGET_SETMEM_PFD): New macros.
4827 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4829 * tree-vectorizer.h (stmt_vec_info): Turn back into a typedef.
4830 (NULL_STMT_VEC_INFO): Delete.
4831 (stmt_vec_info::operator*): Likewise.
4832 (stmt_vec_info::operator gimple *): Likewise.
4833 * tree-vect-loop.c (vectorizable_reduction): Use NULL instead
4834 of NULL_STMT_VEC_INFO.
4835 * tree-vect-patterns.c (vect_init_pattern_stmt): Likewise.
4836 (vect_reassociating_reduction_p): Likewise.
4837 * tree-vect-stmts.c (vect_build_gather_load_calls): Likewise.
4838 (vectorizable_store): Likewise.
4839 * tree-vectorizer.c (vec_info::set_vinfo_for_stmt): Likewise.
4840 (vec_info::free_stmt_vec_infos): Likewise.
4842 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4844 * tree-vectorizer.h (vect_stmt_in_region_p): Delete.
4845 * tree-vectorizer.c (vect_stmt_in_region_p): Likewise.
4847 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4849 * tree-vectorizer.h (vec_info::new_vinfo_for_stmt)
4850 (vec_info::set_vinfo_for_stmt, vec_info::free_stmt_vec_infos)
4851 (vec_info::free_stmt_vec_info): New private member functions.
4852 (set_stmt_vec_info_vec, free_stmt_vec_infos, vinfo_for_stmt)
4853 (set_vinfo_for_stmt, new_stmt_vec_info, free_stmt_vec_info): Delete.
4854 * tree-parloops.c (gather_scalar_reductions): Remove calls to
4855 set_stmt_vec_info_vec and free_stmt_vec_infos.
4856 * tree-vect-loop.c (_loop_vec_info): Remove call to
4857 set_stmt_vec_info_vec.
4858 * tree-vect-stmts.c (new_stmt_vec_info, set_stmt_vec_info_vec)
4859 (free_stmt_vec_infos, free_stmt_vec_info): Delete in favor of...
4860 * tree-vectorizer.c (vec_info::new_stmt_vec_info)
4861 (vec_info::set_vinfo_for_stmt, vec_info::free_stmt_vec_infos)
4862 (vec_info::free_stmt_vec_info): ...these new functions. Remove
4863 assignments in {vec_info::,}new_stmt_vec_info that are redundant
4864 with the clearing in the xcalloc.
4865 (stmt_vec_info_vec): Delete.
4866 (vec_info::vec_info): Don't call set_stmt_vec_info_vec.
4867 (vectorize_loops): Likewise.
4868 (vec_info::~vec_info): Remove argument from call to
4869 free_stmt_vec_infos.
4870 (vec_info::add_stmt): Remove vinfo argument from call to
4873 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4875 * tree-vectorizer.h (free_stmt_vec_info): Take a stmt_vec_info
4876 rather than a gimple stmt.
4877 * tree-vect-stmts.c (free_stmt_vec_info): Likewise. Don't free
4878 information for pattern statements when passed the original
4879 statement; instead wait to be passed the pattern statement itself.
4880 Don't call set_vinfo_for_stmt here.
4881 (free_stmt_vec_infos): Update call to free_stmt_vec_info.
4882 * tree-vect-loop.c (_loop_vec_info::~loop_vec_info): Don't free
4883 stmt_vec_infos here.
4884 * tree-vect-slp.c (_bb_vec_info::~bb_vec_info): Likewise.
4885 * tree-vectorizer.c (vec_info::remove_stmt): Nullify the statement's
4886 stmt_vec_infos entry.
4888 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4890 * tree-vectorizer.h (vec_info::replace_stmt): Declare.
4891 * tree-vectorizer.c (vec_info::replace_stmt): New function.
4892 * tree-vect-slp.c (vect_remove_slp_scalar_calls): Use it.
4893 * tree-vect-stmts.c (vectorizable_call): Likewise.
4894 (vectorizable_simd_clone_call): Likewise.
4896 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4898 * tree-vectorizer.h (vec_info::remove_stmt): Declare.
4899 * tree-vectorizer.c (vec_info::remove_stmt): New function.
4900 * tree-vect-loop-manip.c (vect_set_loop_condition): Use it.
4901 * tree-vect-loop.c (vect_transform_loop): Likewise.
4902 * tree-vect-slp.c (vect_schedule_slp): Likewise.
4903 * tree-vect-stmts.c (vect_remove_stores): Likewise.
4905 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4907 * tree-vectorizer.h (vec_info::lookup_dr): New member function.
4908 (vect_dr_stmt): Delete.
4909 * tree-vectorizer.c (vec_info::lookup_dr): New function.
4910 * tree-vect-loop-manip.c (vect_update_inits_of_drs): Use it instead
4912 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr)
4913 (vect_analyze_data_ref_dependence, vect_record_base_alignments)
4914 (vect_verify_datarefs_alignment, vect_peeling_supportable)
4915 (vect_analyze_data_ref_accesses, vect_prune_runtime_alias_test_list)
4916 (vect_analyze_data_refs): Likewise.
4917 (vect_slp_analyze_data_ref_dependence): Likewise. Take a vec_info
4919 (vect_find_same_alignment_drs): Likewise.
4920 (vect_slp_analyze_node_dependences): Update calls accordingly.
4921 (vect_analyze_data_refs_alignment): Likewise. Use vec_info::lookup_dr
4922 instead of DR_VECT_AUX.
4923 (vect_get_peeling_costs_all_drs): Take a loop_vec_info instead
4924 of a vector data references. Use vec_info::lookup_dr instead of
4926 (vect_peeling_hash_get_lowest_cost): Update calls accordingly.
4927 (vect_enhance_data_refs_alignment): Likewise. Use vec_info::lookup_dr
4928 instead of DR_VECT_AUX.
4930 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4932 * tree-vectorizer.h (_loop_vec_info::unaligned_dr): Change to
4934 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Update
4936 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
4937 * tree-vect-loop-manip.c (get_misalign_in_elems): Likewise.
4938 (vect_gen_prolog_loop_niters): Likewise.
4940 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4942 * tree-vectorizer.h (set_dr_misalignment, dr_misalignment)
4943 (DR_TARGET_ALIGNMENT, aligned_access_p, known_alignment_for_access_p)
4944 (vect_known_alignment_in_bytes, vect_dr_behavior)
4945 (vect_get_scalar_dr_size): Take references as dr_vec_infos
4946 instead of data_references. Update calls to other routines for
4947 which the same change has been made.
4948 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Take
4949 dr_vec_infos instead of stmt_vec_infos.
4950 (vect_analyze_data_ref_dependence): Update call accordingly.
4951 (vect_slp_analyze_data_ref_dependence)
4952 (vect_record_base_alignments): Use DR_VECT_AUX.
4953 (vect_calculate_target_alignment, vect_compute_data_ref_alignment)
4954 (vect_update_misalignment_for_peel, verify_data_ref_alignment)
4955 (vector_alignment_reachable_p, vect_get_data_access_cost)
4956 (vect_peeling_supportable, vect_analyze_group_access_1)
4957 (vect_analyze_group_access, vect_analyze_data_ref_access)
4958 (vect_vfa_segment_size, vect_vfa_access_size, vect_vfa_align)
4959 (vect_compile_time_alias, vect_small_gap_p)
4960 (vectorizable_with_step_bound_p, vect_duplicate_ssa_name_ptr_info):
4961 (vect_supportable_dr_alignment): Take references as dr_vec_infos
4962 instead of data_references. Update calls to other routines for
4963 which the same change has been made.
4964 (vect_verify_datarefs_alignment, vect_get_peeling_costs_all_drs)
4965 (vect_find_same_alignment_drs, vect_analyze_data_refs_alignment)
4966 (vect_slp_analyze_and_verify_node_alignment)
4967 (vect_analyze_data_ref_accesses, vect_prune_runtime_alias_test_list)
4968 (vect_create_addr_base_for_vector_ref, vect_create_data_ref_ptr)
4969 (vect_setup_realignment): Use dr_vec_infos. Update calls after
4971 (_vect_peel_info::dr): Replace with...
4972 (_vect_peel_info::dr_info): ...this new field.
4973 (vect_peeling_hash_get_most_frequent)
4974 (vect_peeling_hash_choose_best_peeling): Update accordingly.
4975 (vect_peeling_hash_get_lowest_cost):
4976 (vect_enhance_data_refs_alignment): Likewise. Update calls to other
4977 routines for which the same change has been made.
4978 (vect_peeling_hash_insert): Likewise. Take a dr_vec_info instead of a
4980 * tree-vect-loop-manip.c (get_misalign_in_elems)
4981 (vect_gen_prolog_loop_niters): Use dr_vec_infos. Update calls after
4983 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
4984 * tree-vect-stmts.c (vect_get_store_cost, vect_get_load_cost)
4985 (vect_truncate_gather_scatter_offset, compare_step_with_zero)
4986 (get_group_load_store_type, get_negative_load_store_type)
4987 (vect_get_data_ptr_increment, vectorizable_store)
4988 (vectorizable_load): Likewise.
4989 (ensure_base_align): Take a dr_vec_info instead of a data_reference.
4990 Update calls to other routines for which the same change has been made.
4992 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4994 * tree-vectorizer.h (vec_info::move_dr): New member function.
4995 (dataref_aux): Rename to...
4996 (dr_vec_info): ...this and add "dr" and "stmt" fields.
4997 (_stmt_vec_info::dr_aux): Update accordingly.
4998 (_stmt_vec_info::data_ref_info): Delete.
4999 (STMT_VINFO_GROUPED_ACCESS, DR_GROUP_FIRST_ELEMENT)
5000 (DR_GROUP_NEXT_ELEMENT, DR_GROUP_SIZE, DR_GROUP_STORE_COUNT)
5001 (DR_GROUP_GAP, DR_GROUP_SAME_DR_STMT, REDUC_GROUP_FIRST_ELEMENT):
5002 (REDUC_GROUP_NEXT_ELEMENT, REDUC_GROUP_SIZE): Use dr_aux.dr instead
5004 (STMT_VINFO_DATA_REF): Likewise. Turn into an lvalue.
5005 (STMT_VINFO_DR_INFO): New macro.
5006 (DR_VECT_AUX): Use STMT_VINFO_DR_INKFO and vect_dr_stmt.
5007 (set_dr_misalignment): Update after rename of dataref_aux.
5008 (vect_dr_stmt): Move earlier in file. Return dr_aux.stmt.
5009 * tree-vect-stmts.c (new_stmt_vec_info): Remove redundant
5010 initialization of STMT_VINFO_DATA_REF.
5011 * tree-vectorizer.c (vec_info::move_dr): New function.
5012 * tree-vect-patterns.c (vect_recog_bool_pattern)
5013 (vect_recog_mask_conversion_pattern)
5014 (vect_recog_gather_scatter_pattern): Use it.
5015 * tree-vect-data-refs.c (vect_analyze_data_refs): Initialize
5016 the "dr" and "stmt" fields of dr_vec_info instead of
5017 STMT_VINFO_DATA_REF.
5019 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
5021 * tree-vectorizer.h (_stmt_vec_info::pattern_stmt_p): New field.
5022 (is_pattern_stmt_p): Use it.
5023 * tree-vect-patterns.c (vect_init_pattern_stmt): Set pattern_stmt_p
5024 on pattern statements.
5026 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
5028 * tree-vect-patterns.c (vect_mark_pattern_stmts): Take the
5029 original stmt as a stmt_vec_info rather than a gimple stmt.
5030 (vect_pattern_recog_1): Take the statement directly as a
5031 stmt_vec_info, rather than via a gimple_stmt_iterator.
5032 Update call to vect_mark_pattern_stmts.
5033 (vect_pattern_recog): Update calls accordingly.
5035 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
5037 * tree-vectorizer.h (vect_get_vec_defs_for_stmt_copy)
5038 (vect_get_vec_def_for_stmt_copy): Take a vec_info rather than
5039 a vect_def_type for the first argument.
5040 * tree-vect-stmts.c (vect_get_vec_defs_for_stmt_copy): Likewise.
5041 (vect_get_vec_def_for_stmt_copy): Likewise. Return the original
5042 operand if it isn't defined by a vectorized statement.
5043 (vect_build_gather_load_calls): Remove the mask_dt argument and
5044 update calls to vect_get_vec_def_for_stmt_copy.
5045 (vectorizable_bswap): Likewise the dt argument.
5046 (vectorizable_call): Update calls to vectorizable_bswap and
5047 vect_get_vec_def_for_stmt_copy.
5048 (vectorizable_simd_clone_call, vectorizable_assignment)
5049 (vectorizable_shift, vectorizable_operation, vectorizable_condition)
5050 (vectorizable_comparison): Update calls to
5051 vect_get_vec_def_for_stmt_copy.
5052 (vectorizable_store): Likewise. Remove now-unnecessary calls to
5054 (vect_get_loop_based_defs): Remove dt argument and update call
5055 to vect_get_vec_def_for_stmt_copy.
5056 (vectorizable_conversion): Update calls to vect_get_loop_based_defs
5057 and vect_get_vec_def_for_stmt_copy.
5058 (vectorizable_load): Update calls to vect_build_gather_load_calls
5059 and vect_get_vec_def_for_stmt_copy.
5060 * tree-vect-loop.c (vect_create_epilog_for_reduction)
5061 (vectorizable_reduction, vectorizable_live_operation): Update calls
5062 to vect_get_vec_def_for_stmt_copy.
5064 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
5066 * tree-vect-data-refs.c (vect_record_base_alignment): Replace vec_info
5067 and gimple stmt arguments with a stmt_vec_info.
5068 (vect_record_base_alignments): Update calls accordingly.
5069 * tree-vect-slp.c (vect_record_max_nunits): Replace vec_info
5070 and gimple stmt arguments with a stmt_vec_info.
5071 (vect_build_slp_tree_1): Remove vinfo argument and update call
5072 to vect_record_max_nunits.
5073 (vect_build_slp_tree_2): Update calls to vect_build_slp_tree_1
5074 and vect_record_max_nunits.
5076 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
5078 * tree-vectorizer.h (nested_in_vect_loop_p): Move further down
5079 file and take a stmt_vec_info instead of a gimple stmt.
5080 (supportable_widening_operation, vect_finish_replace_stmt)
5081 (vect_finish_stmt_generation, vect_get_store_rhs)
5082 (vect_get_vec_def_for_operand_1, vect_get_vec_def_for_operand)
5083 (vect_get_vec_defs, vect_init_vector, vect_transform_stmt)
5084 (vect_remove_stores, vect_analyze_stmt, vectorizable_condition)
5085 (vect_get_smallest_scalar_type, vect_check_gather_scatter)
5086 (vect_create_data_ref_ptr, bump_vector_ptr)
5087 (vect_permute_store_chain, vect_setup_realignment)
5088 (vect_transform_grouped_load, vect_record_grouped_load_vectors)
5089 (vect_create_addr_base_for_vector_ref, vectorizable_live_operation)
5090 (vectorizable_reduction, vectorizable_induction)
5091 (get_initial_def_for_reduction, is_simple_and_all_uses_invariant)
5092 (vect_get_place_in_interleaving_chain): Take stmt_vec_infos rather
5093 than gimple stmts as arguments.
5094 * tree-vect-data-refs.c (vect_get_smallest_scalar_type)
5095 (vect_preserves_scalar_order_p, vect_slp_analyze_node_dependences)
5096 (can_group_stmts_p, vect_check_gather_scatter)
5097 (vect_create_addr_base_for_vector_ref, vect_create_data_ref_ptr)
5098 (bump_vector_ptr, vect_permute_store_chain, vect_setup_realignment)
5099 (vect_permute_load_chain, vect_shift_permute_load_chain)
5100 (vect_transform_grouped_load)
5101 (vect_record_grouped_load_vectors): Likewise.
5102 * tree-vect-loop.c (vect_fixup_reduc_chain)
5103 (get_initial_def_for_reduction, vect_create_epilog_for_reduction)
5104 (vectorize_fold_left_reduction, is_nonwrapping_integer_induction)
5105 (vectorizable_reduction, vectorizable_induction)
5106 (vectorizable_live_operation, vect_loop_kill_debug_uses): Likewise.
5107 * tree-vect-patterns.c (type_conversion_p, adjust_bool_stmts)
5108 (vect_get_load_store_mask): Likewise.
5109 * tree-vect-slp.c (vect_get_place_in_interleaving_chain)
5110 (vect_analyze_slp_instance, vect_mask_constant_operand_p): Likewise.
5111 * tree-vect-stmts.c (vect_mark_relevant)
5112 (is_simple_and_all_uses_invariant)
5113 (exist_non_indexing_operands_for_use_p, process_use)
5114 (vect_init_vector_1, vect_init_vector, vect_get_vec_def_for_operand_1)
5115 (vect_get_vec_def_for_operand, vect_get_vec_defs)
5116 (vect_finish_stmt_generation_1, vect_finish_replace_stmt)
5117 (vect_finish_stmt_generation, vect_truncate_gather_scatter_offset)
5118 (compare_step_with_zero, vect_get_store_rhs, get_group_load_store_type)
5119 (get_negative_load_store_type, get_load_store_type)
5120 (vect_check_load_store_mask, vect_check_store_rhs)
5121 (vect_build_gather_load_calls, vect_get_strided_load_store_ops)
5122 (vectorizable_bswap, vectorizable_call, vectorizable_simd_clone_call)
5123 (vect_create_vectorized_demotion_stmts, vectorizable_conversion)
5124 (vectorizable_assignment, vectorizable_shift, vectorizable_operation)
5125 (get_group_alias_ptr_type, vectorizable_store, hoist_defs_of_uses)
5126 (vectorizable_load, vectorizable_condition, vectorizable_comparison)
5127 (vect_analyze_stmt, vect_transform_stmt, vect_remove_stores)
5128 (supportable_widening_operation): Likewise.
5130 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
5132 * tree-vect-data-refs.c (vect_describe_gather_scatter_call): Take
5133 a stmt_vec_info instead of a gcall.
5134 (vect_check_gather_scatter): Update call accordingly.
5135 * tree-vect-loop-manip.c (iv_phi_p): Take a stmt_vec_info instead
5137 (vect_can_advance_ivs_p, vect_update_ivs_after_vectorizer)
5138 (slpeel_update_phi_nodes_for_loops):): Update calls accordingly.
5139 * tree-vect-loop.c (vect_transform_loop_stmt): Take a stmt_vec_info
5140 instead of a gimple stmt.
5141 (vect_transform_loop): Update calls accordingly.
5142 * tree-vect-slp.c (vect_split_slp_store_group): Take and return
5143 stmt_vec_infos instead of gimple stmts.
5144 (vect_analyze_slp_instance): Update use accordingly.
5145 * tree-vect-stmts.c (read_vector_array, write_vector_array)
5146 (vect_clobber_variable, vect_stmt_relevant_p, permute_vec_elements)
5147 (vect_use_strided_gather_scatters_p, vect_build_all_ones_mask)
5148 (vect_build_zero_merge_argument, vect_get_gather_scatter_ops)
5149 (vect_gen_widened_results_half, vect_get_loop_based_defs)
5150 (vect_create_vectorized_promotion_stmts, can_vectorize_live_stmts):
5151 Take a stmt_vec_info instead of a gimple stmt and pass stmt_vec_infos
5152 down to subroutines.
5154 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
5156 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Change the type
5157 of the worklist from a vector of gimple stmts to a vector of
5159 * tree-vect-stmts.c (vect_mark_relevant, process_use)
5160 (vect_mark_stmts_to_be_vectorized): Likewise
5162 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
5164 * tree-vect-loop.c (vect_analyze_loop_operations): Look up the
5165 statement before passing it to vect_analyze_stmt.
5166 (vect_create_epilog_for_reduction): Use a stmt_vec_info to walk
5167 the chain of phi vector definitions. Track the exit phi via its
5169 (vectorizable_reduction): Set cond_stmt_vinfo directly from the
5170 STMT_VINFO_REDUC_DEF.
5171 * tree-vect-slp.c (vect_get_place_in_interleaving_chain): Use
5172 stmt_vec_infos to handle the statement chains.
5173 (vect_get_slp_defs): Record the first statement in the node
5174 using a stmt_vec_info.
5175 * tree-vect-stmts.c (vect_mark_stmts_to_be_vectorized): Look up
5176 statements here and pass their stmt_vec_info down to subroutines.
5177 (vect_init_vector_1): Hoist call to vinfo_for_stmt and pass it
5178 down to vect_finish_stmt_generation.
5179 (vect_init_vector, vect_get_vec_defs, vect_finish_replace_stmt)
5180 (vect_finish_stmt_generation): Call vinfo_for_stmt and pass
5181 stmt_vec_infos to subroutines.
5182 (vect_remove_stores): Use stmt_vec_infos to handle the statement
5185 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
5187 * tree-vect-data-refs.c (vect_slp_analyze_node_dependences):
5188 (vect_check_gather_scatter, vect_create_data_ref_ptr, bump_vector_ptr)
5189 (vect_permute_store_chain, vect_setup_realignment)
5190 (vect_permute_load_chain, vect_shift_permute_load_chain)
5191 (vect_transform_grouped_load): Use stmt_vec_info rather than gimple
5192 stmts internally, and when passing values to other vectorizer routines.
5193 * tree-vect-loop-manip.c (vect_can_advance_ivs_p): Likewise.
5194 * tree-vect-loop.c (vect_analyze_scalar_cycles_1)
5195 (vect_analyze_loop_operations, get_initial_def_for_reduction)
5196 (vect_create_epilog_for_reduction, vectorize_fold_left_reduction)
5197 (vectorizable_reduction, vectorizable_induction)
5198 (vectorizable_live_operation, vect_transform_loop_stmt)
5199 (vect_transform_loop): Likewise.
5200 * tree-vect-patterns.c (vect_reassociating_reduction_p)
5201 (vect_recog_widen_op_pattern, vect_recog_mixed_size_cond_pattern)
5202 (vect_recog_bool_pattern, vect_recog_gather_scatter_pattern): Likewise.
5203 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
5204 (vect_slp_analyze_node_operations_1): Likewise.
5205 * tree-vect-stmts.c (vect_mark_relevant, process_use)
5206 (exist_non_indexing_operands_for_use_p, vect_init_vector_1)
5207 (vect_mark_stmts_to_be_vectorized, vect_get_vec_def_for_operand)
5208 (vect_finish_stmt_generation_1, get_group_load_store_type)
5209 (get_load_store_type, vect_build_gather_load_calls)
5210 (vectorizable_bswap, vectorizable_call, vectorizable_simd_clone_call)
5211 (vect_create_vectorized_demotion_stmts, vectorizable_conversion)
5212 (vectorizable_assignment, vectorizable_shift, vectorizable_operation)
5213 (vectorizable_store, vectorizable_load, vectorizable_condition)
5214 (vectorizable_comparison, vect_analyze_stmt, vect_transform_stmt)
5215 (supportable_widening_operation): Likewise.
5216 (vect_get_vector_types_for_stmt): Likewise.
5217 * tree-vectorizer.h (vect_dr_behavior): Likewise.
5219 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
5221 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence)
5222 (vect_slp_analyze_node_dependences, vect_analyze_data_ref_accesses)
5223 (vect_permute_store_chain, vect_permute_load_chain)
5224 (vect_shift_permute_load_chain, vect_transform_grouped_load): Avoid
5225 repeated stmt_vec_info lookups.
5226 * tree-vect-loop-manip.c (vect_can_advance_ivs_p): Likewise.
5227 (vect_update_ivs_after_vectorizer): Likewise.
5228 * tree-vect-loop.c (vect_is_simple_reduction): Likewise.
5229 (vect_create_epilog_for_reduction, vectorizable_reduction): Likewise.
5230 * tree-vect-patterns.c (adjust_bool_stmts): Likewise.
5231 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
5232 (vect_bb_slp_scalar_cost): Likewise.
5233 * tree-vect-stmts.c (get_group_alias_ptr_type): Likewise.
5235 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
5237 * tree-vect-data-refs.c (vect_check_gather_scatter): Pass the
5238 gcall rather than the generic gimple stmt to gimple_call_internal_fn.
5239 (vect_get_smallest_scalar_type, can_group_stmts_p): Use dyn_cast
5240 to get gassigns and gcalls, rather than operating on generc gimple
5242 * tree-vect-stmts.c (exist_non_indexing_operands_for_use_p)
5243 (vect_mark_stmts_to_be_vectorized, vectorizable_store)
5244 (vectorizable_load, vect_analyze_stmt): Likewise.
5245 * tree-vect-loop.c (vectorizable_reduction): Likewise gphi.
5247 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
5249 * tree-vectorizer.h (get_earlier_stmt, get_later_stmt): Take and
5250 return stmt_vec_infos rather than gimple stmts. Do not accept
5252 (vect_find_last_scalar_stmt_in_slp): Return a stmt_vec_info instead
5254 * tree-vect-slp.c (vect_find_last_scalar_stmt_in_slp): Likewise.
5255 Update use of get_later_stmt.
5256 (vect_get_constant_vectors): Update call accordingly.
5257 (vect_schedule_slp_instance): Likewise
5258 * tree-vect-data-refs.c (vect_slp_analyze_node_dependences): Likewise.
5259 (vect_slp_analyze_instance_dependence): Likewise.
5260 (vect_preserves_scalar_order_p): Update use of get_earlier_stmt.
5262 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
5264 * tree-vectorizer.h (stmt_info_for_cost::stmt): Replace with...
5265 (stmt_info_for_cost::stmt_info): ...this new field.
5266 (add_stmt_costs): Update accordingly.
5267 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost)
5268 (vect_get_known_peeling_cost): Likewise.
5269 (vect_estimate_min_profitable_iters): Likewise.
5270 * tree-vect-stmts.c (record_stmt_cost): Likewise.
5272 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
5274 * tree-vectorizer.h (_loop_vec_info::may_misalign_stmts): Change
5275 from an auto_vec<gimple *> to an auto_vec<stmt_vec_info>.
5276 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Update
5278 * tree-vect-loop-manip.c (vect_create_cond_for_align_checks): Likewise.
5280 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
5282 * tree-vectorizer.h (_stmt_vec_info::same_dr_stmt): Change from
5283 a gimple stmt to a stmt_vec_info.
5284 * tree-vect-stmts.c (vectorizable_load): Update accordingly.
5286 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
5288 * tree-vectorizer.h (vec_info::grouped_stores): Change from
5289 an auto_vec<gimple *> to an auto_vec<stmt_vec_info>.
5290 (_loop_vec_info::reduction_chains): Likewise.
5291 * tree-vect-loop.c (vect_fixup_scalar_cycles_with_patterns): Update
5293 * tree-vect-slp.c (vect_analyze_slp): Likewise.
5295 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
5297 * tree-vectorizer.h (_stmt_vec_info::first_element): Change from
5298 a gimple stmt to a stmt_vec_info.
5299 (_stmt_vec_info::next_element): Likewise.
5300 * tree-vect-data-refs.c (vect_update_misalignment_for_peel)
5301 (vect_slp_analyze_and_verify_node_alignment)
5302 (vect_analyze_group_access_1, vect_analyze_group_access)
5303 (vect_small_gap_p, vect_prune_runtime_alias_test_list)
5304 (vect_create_data_ref_ptr, vect_record_grouped_load_vectors)
5305 (vect_supportable_dr_alignment): Update accordingly.
5306 * tree-vect-loop.c (vect_fixup_reduc_chain): Likewise.
5307 (vect_fixup_scalar_cycles_with_patterns, vect_is_slp_reduction)
5308 (vect_is_simple_reduction, vectorizable_reduction): Likewise.
5309 * tree-vect-patterns.c (vect_reassociating_reduction_p): Likewise.
5310 * tree-vect-slp.c (vect_build_slp_tree_1)
5311 (vect_attempt_slp_rearrange_stmts, vect_supported_load_permutation_p)
5312 (vect_split_slp_store_group, vect_analyze_slp_instance)
5313 (vect_analyze_slp, vect_transform_slp_perm_load): Likewise.
5314 * tree-vect-stmts.c (vect_model_store_cost, vect_model_load_cost)
5315 (get_group_load_store_type, get_load_store_type)
5316 (get_group_alias_ptr_type, vectorizable_store, vectorizable_load)
5317 (vect_transform_stmt, vect_remove_stores): Likewise.
5319 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
5321 * tree-vectorizer.h (vect_dr_stmt): Return a stmt_vec_info rather
5323 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence)
5324 (vect_slp_analyze_data_ref_dependence, vect_record_base_alignments)
5325 (vect_calculate_target_alignmentm, vect_compute_data_ref_alignment)
5326 (vect_update_misalignment_for_peel, vect_verify_datarefs_alignment)
5327 (vector_alignment_reachable_p, vect_get_data_access_cost)
5328 (vect_get_peeling_costs_all_drs, vect_peeling_hash_get_lowest_cost)
5329 (vect_peeling_supportable, vect_enhance_data_refs_alignment)
5330 (vect_find_same_alignment_drs, vect_analyze_data_refs_alignment)
5331 (vect_analyze_group_access_1, vect_analyze_group_access)
5332 (vect_analyze_data_ref_access, vect_analyze_data_ref_accesses)
5333 (vect_vfa_access_size, vect_small_gap_p, vect_analyze_data_refs)
5334 (vect_supportable_dr_alignment): Remove vinfo_for_stmt from the
5335 result of vect_dr_stmt and use the stmt_vec_info instead of
5336 the associated gimple stmt.
5337 * tree-vect-loop-manip.c (get_misalign_in_elems): Likewise.
5338 (vect_gen_prolog_loop_niters): Likewise.
5339 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
5341 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
5343 * tree-vectorizer.h (_slp_tree::stmts): Change from a vec<gimple *>
5344 to a vec<stmt_vec_info>.
5345 * tree-vect-slp.c (vect_free_slp_tree): Update accordingly.
5346 (vect_create_new_slp_node): Take a vec<gimple *> instead of a
5348 (_slp_oprnd_info::def_stmts): Change from a vec<gimple *>
5349 to a vec<stmt_vec_info>.
5350 (bst_traits::value_type, bst_traits::value_type): Likewise.
5351 (bst_traits::hash): Update accordingly.
5352 (vect_get_and_check_slp_defs): Change the stmts parameter from
5353 a vec<gimple *> to a vec<stmt_vec_info>.
5354 (vect_two_operations_perm_ok_p, vect_build_slp_tree_1): Likewise.
5355 (vect_build_slp_tree): Likewise.
5356 (vect_build_slp_tree_2): Likewise. Update uses of
5357 SLP_TREE_SCALAR_STMTS.
5358 (vect_print_slp_tree): Update uses of SLP_TREE_SCALAR_STMTS.
5359 (vect_mark_slp_stmts, vect_mark_slp_stmts_relevant)
5360 (vect_slp_rearrange_stmts, vect_attempt_slp_rearrange_stmts)
5361 (vect_supported_load_permutation_p, vect_find_last_scalar_stmt_in_slp)
5362 (vect_detect_hybrid_slp_stmts, vect_slp_analyze_node_operations_1)
5363 (vect_slp_analyze_node_operations, vect_slp_analyze_operations)
5364 (vect_bb_slp_scalar_cost, vect_slp_analyze_bb_1)
5365 (vect_get_constant_vectors, vect_get_slp_defs)
5366 (vect_transform_slp_perm_load, vect_schedule_slp_instance)
5367 (vect_remove_slp_scalar_calls, vect_schedule_slp): Likewise.
5368 (vect_analyze_slp_instance): Build up a vec of stmt_vec_infos
5369 instead of gimple stmts.
5370 * tree-vect-data-refs.c (vect_slp_analyze_node_dependences): Change
5371 the stores parameter for a vec<gimple *> to a vec<stmt_vec_info>.
5372 (vect_slp_analyze_instance_dependence): Update uses of
5373 SLP_TREE_SCALAR_STMTS.
5374 (vect_slp_analyze_and_verify_node_alignment): Likewise.
5375 (vect_slp_analyze_and_verify_instance_alignment): Likewise.
5376 * tree-vect-loop.c (neutral_op_for_slp_reduction): Likewise.
5377 (get_initial_defs_for_reduction): Likewise.
5378 (vect_create_epilog_for_reduction): Likewise.
5379 (vectorize_fold_left_reduction): Likewise.
5380 * tree-vect-stmts.c (vect_prologue_cost_for_slp_op): Likewise.
5381 (vect_model_simple_cost, vectorizable_shift, vectorizable_load)
5382 (can_vectorize_live_stmts): Likewise.
5384 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
5386 * tree-vectorizer.h (_loop_vec_info::reductions): Change from an
5387 auto_vec<gimple *> to an auto_vec<stmt_vec_info>.
5388 (vect_force_simple_reduction): Take and return stmt_vec_infos rather
5390 * tree-parloops.c (valid_reduction_p): Take a stmt_vec_info instead
5392 (gather_scalar_reductions): Update after above interface changes.
5393 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Likewise.
5394 (vect_is_simple_reduction): Take and return stmt_vec_infos rather
5396 (vect_force_simple_reduction): Likewise.
5397 * tree-vect-patterns.c (vect_pattern_recog_1): Update use of
5398 LOOP_VINFO_REDUCTIONS.
5399 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
5401 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
5403 * tree-vectorizer.h (_stmt_vec_info::reduc_def): Change from
5404 a gimple stmt to a stmt_vec_info.
5405 * tree-vect-loop.c (vect_active_double_reduction_p)
5406 (vect_force_simple_reduction, vectorizable_reduction): Update
5409 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
5411 * tree-vectorizer.h (_slp_tree::vec_stmts): Change from a
5412 vec<gimple *> to a vec<stmt_vec_info>.
5413 * tree-vect-loop.c (vect_create_epilog_for_reduction): Change
5414 the reduction_phis argument from a vec<gimple *> to a
5416 (vectorizable_reduction): Likewise the phis local variable that
5417 is passed to vect_create_epilog_for_reduction. Update for new type
5418 of SLP_TREE_VEC_STMTS.
5419 (vectorizable_induction): Update for new type of SLP_TREE_VEC_STMTS.
5420 (vectorizable_live_operation): Likewise.
5421 * tree-vect-slp.c (vect_get_slp_vect_defs): Likewise.
5422 (vect_transform_slp_perm_load, vect_schedule_slp_instance): Likewise.
5424 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
5426 * tree-vectorizer.h (_stmt_vec_info::vectorized_stmt): Change from
5427 a gimple stmt to a stmt_vec_info.
5428 (vectorizable_condition, vectorizable_live_operation)
5429 (vectorizable_reduction, vectorizable_induction): Pass back the
5430 vectorized statement as a stmt_vec_info.
5431 * tree-vect-data-refs.c (vect_record_grouped_load_vectors): Update
5432 use of STMT_VINFO_VEC_STMT.
5433 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise,
5434 accumulating the inner phis that feed the STMT_VINFO_VEC_STMT
5435 as stmt_vec_infos rather than gimple stmts.
5436 (vectorize_fold_left_reduction): Change vec_stmt from a gimple stmt
5438 (vectorizable_live_operation): Likewise.
5439 (vectorizable_reduction, vectorizable_induction): Likewise,
5440 updating use of STMT_VINFO_VEC_STMT.
5441 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Update use
5442 of STMT_VINFO_VEC_STMT.
5443 (vect_build_gather_load_calls, vectorizable_bswap, vectorizable_call)
5444 (vectorizable_simd_clone_call, vectorizable_conversion)
5445 (vectorizable_assignment, vectorizable_shift, vectorizable_operation)
5446 (vectorizable_store, vectorizable_load, vectorizable_condition)
5447 (vectorizable_comparison, can_vectorize_live_stmts): Change vec_stmt
5448 from a gimple stmt to a stmt_vec_info.
5449 (vect_transform_stmt): Update use of STMT_VINFO_VEC_STMT. Pass a
5450 pointer to a stmt_vec_info to the vectorizable_* routines.
5452 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
5454 * tree-vectorizer.h (_stmt_vec_info::related_stmt): Change from
5455 a gimple stmt to a stmt_vec_info.
5456 (is_pattern_stmt_p): Update accordingly.
5457 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Likewise.
5458 (vect_record_grouped_load_vectors): Likewise.
5459 * tree-vect-loop.c (vect_determine_vf_for_stmt): Likewise.
5460 (vect_fixup_reduc_chain, vect_update_vf_for_slp): Likewise.
5461 (vect_model_reduction_cost): Likewise.
5462 (vect_create_epilog_for_reduction): Likewise.
5463 (vectorizable_reduction, vectorizable_induction): Likewise.
5464 * tree-vect-patterns.c (vect_init_pattern_stmt): Likewise.
5465 Return the stmt_vec_info for the pattern statement.
5466 (vect_set_pattern_stmt): Update use of STMT_VINFO_RELATED_STMT.
5467 (vect_split_statement, vect_mark_pattern_stmts): Likewise.
5468 * tree-vect-slp.c (vect_detect_hybrid_slp_stmts): Likewise.
5469 (vect_detect_hybrid_slp, vect_get_slp_defs): Likewise.
5470 * tree-vect-stmts.c (vect_mark_relevant): Likewise.
5471 (vect_get_vec_def_for_operand_1, vectorizable_call): Likewise.
5472 (vectorizable_simd_clone_call, vect_analyze_stmt, new_stmt_vec_info)
5473 (free_stmt_vec_info, vect_is_simple_use): Likewise.
5475 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
5477 * tree-vectorizer.h (vect_finish_replace_stmt): Return a stmt_vec_info
5478 (vect_finish_stmt_generation): Likewise.
5479 * tree-vect-stmts.c (vect_finish_stmt_generation_1): Likewise.
5480 (vect_finish_replace_stmt, vect_finish_stmt_generation): Likewise.
5481 (vect_build_gather_load_calls): Use the return value of the above
5482 functions instead of a separate call to vinfo_for_stmt. Use narrow
5483 scopes for the input gimple stmt and wider scopes for the associated
5484 stmt_vec_info. Use vec_info::lookup_def when setting these
5485 stmt_vec_infos from an SSA_NAME definition.
5486 (vectorizable_bswap, vectorizable_call, vectorizable_simd_clone_call)
5487 (vect_create_vectorized_demotion_stmts, vectorizable_conversion)
5488 (vectorizable_assignment, vectorizable_shift, vectorizable_operation)
5489 (vectorizable_store, vectorizable_load, vectorizable_condition)
5490 (vectorizable_comparison): Likewise.
5491 * tree-vect-loop.c (vectorize_fold_left_reduction): Likewise.
5492 (vectorizable_reduction): Likewise.
5494 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
5496 * tree-vectorizer.h (vect_is_simple_use): Add an optional
5497 stmt_vec_info * parameter before the optional gimple **.
5498 * tree-vect-stmts.c (vect_is_simple_use): Likewise.
5499 (process_use, vect_get_vec_def_for_operand_1): Update callers.
5500 (vect_get_vec_def_for_operand, vectorizable_shift): Likewise.
5501 * tree-vect-loop.c (vectorizable_reduction): Likewise.
5502 (vectorizable_live_operation): Likewise.
5503 * tree-vect-patterns.c (type_conversion_p): Likewise.
5504 (vect_look_through_possible_promotion): Likewise.
5505 (vect_recog_rotate_pattern): Likewise.
5506 * tree-vect-slp.c (vect_get_and_check_slp_defs): Likewise.
5508 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
5510 * tree-vectorizer.h (stmt_vec_info): Temporarily change from
5511 a typedef to a wrapper class.
5512 (NULL_STMT_VEC_INFO): New macro.
5513 (vec_info::stmt_infos): Change to vec<stmt_vec_info>.
5514 (stmt_vec_info::operator*): New function.
5515 (stmt_vec_info::operator gimple *): Likewise.
5516 (set_vinfo_for_stmt): Use NULL_STMT_VEC_INFO.
5517 (add_stmt_costs): Likewise.
5518 * tree-vect-loop-manip.c (iv_phi_p): Likewise.
5519 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost)
5520 (vect_get_known_peeling_cost): Likewise.
5521 (vect_estimate_min_profitable_iters): Likewise.
5522 * tree-vect-patterns.c (vect_init_pattern_stmt): Likewise.
5523 * tree-vect-slp.c (vect_remove_slp_scalar_calls): Likewise.
5524 * tree-vect-stmts.c (vect_build_gather_load_calls): Likewise.
5525 (vectorizable_store, free_stmt_vec_infos): Likewise.
5526 (new_stmt_vec_info): Change return type of xcalloc to
5529 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
5531 * tree-vectorizer.h (vec_info::lookup_single_use): Declare.
5532 * tree-vectorizer.c (vec_info::lookup_single_use): New function.
5533 * tree-vect-loop.c (vectorizable_reduction): Use it instead of
5534 a single_imm_use-based sequence.
5535 * tree-vect-stmts.c (supportable_widening_operation): Likewise.
5537 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
5539 * tree-vectorizer.h (vec_info::lookup_def): Declare.
5540 * tree-vectorizer.c (vec_info::lookup_def): New function.
5541 * tree-vect-patterns.c (vect_get_internal_def): Use it.
5542 (vect_widened_op_tree): Likewise.
5543 * tree-vect-stmts.c (vect_is_simple_use): Likewise.
5544 * tree-vect-loop.c (vect_analyze_loop_operations): Likewise.
5545 (vectorizable_reduction): Likewise.
5546 (vect_valid_reduction_input_p): Take a stmt_vec_info instead
5548 (vect_is_slp_reduction): Update calls accordingly. Use
5549 vec_info::lookup_def.
5550 (vect_is_simple_reduction): Likewise
5551 * tree-vect-slp.c (vect_detect_hybrid_slp_1): Use vec_info::lookup_def.
5553 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
5555 * tree-vectorizer.h (vec_info::lookup_stmt): Declare.
5556 * tree-vectorizer.c (vec_info::lookup_stmt): New function.
5557 * tree-vect-loop.c (vect_determine_vf_for_stmt): Use it instead
5559 (vect_determine_vectorization_factor, vect_analyze_scalar_cycles_1)
5560 (vect_compute_single_scalar_iteration_cost, vect_analyze_loop_form)
5561 (vect_update_vf_for_slp, vect_analyze_loop_operations)
5562 (vect_is_slp_reduction, vectorizable_induction)
5563 (vect_transform_loop_stmt, vect_transform_loop): Likewise.
5564 * tree-vect-patterns.c (vect_init_pattern_stmt):
5565 (vect_determine_min_output_precision_1, vect_determine_precisions)
5566 (vect_pattern_recog): Likewise.
5567 * tree-vect-stmts.c (vect_analyze_stmt, vect_transform_stmt): Likewise.
5568 * config/powerpcspe/powerpcspe.c (rs6000_density_test): Likewise.
5569 * config/rs6000/rs6000.c (rs6000_density_test): Likewise.
5570 * tree-vect-slp.c (vect_detect_hybrid_slp_stmts): Likewise.
5571 (vect_detect_hybrid_slp_1, vect_detect_hybrid_slp_2)
5572 (vect_detect_hybrid_slp): Likewise. Change the walk_stmt_info
5573 info field from a loop to a loop_vec_info.
5575 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
5577 * tree-vectorizer.h (stmt_vec_info): Move typedef earlier in file.
5578 (vec_info::add_stmt): Declare.
5579 * tree-vectorizer.c (vec_info::add_stmt): New function.
5580 * tree-vect-data-refs.c (vect_create_data_ref_ptr): Use it.
5581 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Likewise.
5582 (vect_create_epilog_for_reduction, vectorizable_reduction): Likewise.
5583 (vectorizable_induction): Likewise.
5584 * tree-vect-slp.c (_bb_vec_info::_bb_vec_info): Likewise.
5585 * tree-vect-stmts.c (vect_finish_stmt_generation_1): Likewise.
5586 (vectorizable_simd_clone_call, vectorizable_store): Likewise.
5587 (vectorizable_load): Likewise.
5588 * tree-vect-patterns.c (vect_init_pattern_stmt): Likewise.
5589 (vect_recog_bool_pattern, vect_recog_mask_conversion_pattern)
5590 (vect_recog_gather_scatter_pattern): Likewise.
5591 (append_pattern_def_seq): Likewise. Remove a check that is
5592 performed by add_stmt itself.
5594 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
5596 * tree-vect-loop.c (vectorizable_reduction): Fix an instance in
5597 which make_ssa_name was called with new_stmt before new_stmt
5600 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
5602 * tree-vect-loop.c (vect_valid_reduction_input_p): New function,
5604 (vect_is_slp_reduction): ...here...
5605 (vect_is_simple_reduction): ...and here. Remove repetition of tests
5606 that are already known to be false.
5608 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
5610 * tree-vectorizer.h (vect_free_slp_instance): Add a final_p parameter.
5611 * tree-vect-slp.c (vect_free_slp_tree): Likewise. Don't update
5612 STMT_VINFO_NUM_SLP_USES when it's true.
5613 (vect_free_slp_instance): Add a final_p parameter and pass it to
5615 (vect_build_slp_tree_2): Update call to vect_free_slp_instance.
5616 (vect_analyze_slp_instance): Likewise.
5617 (vect_slp_analyze_operations): Likewise.
5618 (vect_slp_analyze_bb_1): Likewise.
5619 * tree-vectorizer.c (vec_info): Likewise.
5620 * tree-vect-loop.c (vect_transform_loop): Likewise.
5622 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
5624 * tree-vect-loop.c (vectorizable_reduction): Assert that the
5625 function is not called for second and subsequent members of
5628 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
5630 * tree-vect-loop.c (get_initial_def_for_reduction): Move special
5631 cases for nested loops from here to ...
5632 (vect_create_epilog_for_reduction): ...here. Only call
5633 vect_is_simple_use for inner-loop reductions.
5635 2018-07-31 Martin Liska <mliska@suse.cz>
5637 PR gcov-profile/85338
5638 PR gcov-profile/85350
5639 PR gcov-profile/85372
5640 * profile.c (struct location_triplet): New.
5641 (struct location_triplet_hash): Likewise.
5642 (output_location): Do not output a BB that
5643 is already recorded for a line.
5644 (branch_prob): Use streamed_locations.
5646 2018-07-31 Martin Liska <mliska@suse.cz>
5648 PR gcov-profile/85370
5649 * coverage.c (coverage_begin_function): Do not mark target
5650 clones as artificial functions.
5652 2018-07-31 Martin Liska <mliska@suse.cz>
5654 PR gcov-profile/83813
5655 PR gcov-profile/84758
5656 PR gcov-profile/85217
5657 PR gcov-profile/85332
5658 * profile.c (branch_prob): Do not record GOTO expressions
5659 for GIMPLE statements which locations are already streamed.
5661 2018-07-31 Olivier Hainque <hainque@adacore.com>
5663 * gcc.c (handle_spec_function): Accept a soft_matched_part
5664 argument, as do_spec_1. Pass it down to ...
5665 (eval_spec_function): Accept a soft_matched_part argument,
5666 and pass it down to ...
5667 (do_spec_2): Accept a soft_matched_part argument, and pass
5668 it down to do_spec_1.
5669 (do_spec_1): Pass soft_matched_part to handle_spec_function.
5670 (handle_braces): Update call to handle_spec_function.
5671 (driver::set_up_specs): Update calls to do_spec_2.
5672 (compare_debug_dump_opt_spec_function): Likewise.
5673 (compare_debug_self_opt_spec_function): Likewise.
5675 2018-07-31 Olivier Hainque <hainque@adacore.com>
5677 * common.opt (nolibc): New option.
5678 * doc/invoke.texi (Link Options): Document it.
5679 * gcc.c (LINK_GCC_C_SEQUENCE_SPEC): Honor nolibc.
5680 * config/alpha/linux.h: Likewise.
5681 * config/arc/elf.h: Likewise.
5682 * config/arm/uclinux-elf.h: Likewise.
5683 * config/arm/unknown-elf.h: Likewise.
5684 * config/avr/avrlibc.h: Likewise.
5685 * config/bfin/bfin.h: Likewise.
5686 * config/bfin/linux.h: Likewise.
5687 * config/bfin/uclinux.h: Likewise.
5688 * config/darwin.h: Likewise.
5689 * config/darwin10.h: Likewise.
5690 * config/darwin12.h: Likewise.
5691 * config/gnu-user.h: Likewise.
5692 * config/lm32/uclinux-elf.h: Likewise.
5693 * config/pa/pa-hpux11.h: Likewise.
5694 * config/pa/pa64-hpux.h: Likewise.
5695 * config/sparc/sparc.h: Likewise.
5697 2018-07-31 Olivier Hainque <hainque@adacore.com>
5699 * gcc.c (getenv_spec_function): Prepend '/' to value for allowed
5700 undefined variables.
5702 2018-07-30 Segher Boessenkool <segher@kernel.crashing.org>
5705 * config/arm/arm.c (arm_block_set_aligned_vect): Use gen_int_mode
5708 2018-07-30 Bernd Edlinger <bernd.edlinger@hotmail.de>
5710 * tree-ssa-forwprop.c (simplify_builtin_call): Don't create a not NUL
5711 terminated string literal.
5713 2018-07-30 Segher Boessenkool <segher@kernel.crashing.org>
5715 PR rtl-optimization/85160
5716 * combine.c (is_just_move): New function.
5717 (try_combine): Allow combining two instructions into two if neither of
5718 the original instructions was a move.
5720 2018-07-30 Alexander Monakov <amonakov@ispras.ru>
5723 * doc/extend.texi (Global Register Variables): Discourage use of type
5725 (Local Register Variables): Likewise.
5727 2018-07-30 Richard Sandiford <richard.sandiford@arm.com>
5729 PR tree-optimization/86506
5730 * hwint.h (ceil_log2): Resync with hwint.c implementation.
5732 2018-07-30 Ilya Leoshkevich <iii@linux.ibm.com>
5735 * lra-constraints.c (spill_hard_reg_in_range): When selecting the
5736 hard_regno, make sure no insn between `from` and `to` clobbers it.
5738 2018-07-30 Cesar Philippidis <cesar@codesourcery.com>
5739 Tom de Vries <tdevries@suse.de>
5741 * config/nvptx/nvptx.c (PTX_GANG_DEFAULT): Rename to ...
5742 (PTX_DEFAULT_RUNTIME_DIM): ... this.
5743 (nvptx_goacc_validate_dims): Set default worker and gang dims to
5744 PTX_DEFAULT_RUNTIME_DIM.
5745 (nvptx_dim_limit): Ignore GOMP_DIM_WORKER.
5747 2018-07-29 John David Anglin <danglin@gcc.gnu.org>
5749 * config/pa/pa.c (pa_output_addr_vec): Align address table.
5750 * config/pa/pa.h (JUMP_TABLES_IN_TEXT_SECTION): Revise comment.
5751 * config/pa/pa32-linux.h (JUMP_TABLES_IN_TEXT_SECTION): Define.
5753 2018-07-27 Michael Meissner <meissner@linux.ibm.com>
5755 * config/rs6000/constraints.md (wG constraint): Delete, no longer
5757 * config/rs6000/predicates.md (p9_fusion_reg_operand): Rename
5758 predicate to reflect toc fusion has been deleted.
5759 (toc_fusion_mem_raw): Delete, no longer used.
5760 (toc_fusion_mem_wrapped): Likewise.
5761 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete toc
5763 * config/rs6000/rs6000-protos.h (fusion_wrap_memory_address):
5764 Delete, no longer used.
5765 * config/rs6000/rs6000.c (struct rs6000_reg_addr): Delete fields
5766 meant to be used for toc fusion.
5767 (rs6000_debug_print_mode): Delete toc fusion debugging.
5768 (rs6000_debug_reg_global): Likewise.
5769 (rs6000_init_hard_regno_mode_ok): Delete setting up fields for toc
5770 fusion and secondary reload support that were never used.
5771 (rs6000_option_override_internal): Delete TOC fusion, that was only
5772 partially defined, and it did not work unless you also used the
5774 (rs6000_legitimate_address_p): Delete TOC fusion support.
5775 (rs6000_opt_masks): Likewise.
5776 (fusion_wrap_memory_address): Delete function, no longer used.
5777 (fusion_split_address); Delete TOC fusion support.
5778 * config/rs6000/rs6000.h (TARGET_TOC_FUSION_INT): Delete, no
5779 longer used with toc fusion being deleted.
5780 (TARGET_TOC_FUSION_FP): Likewise.
5781 * config/rs6000/rs6000.md (UNSPEC_FUSION_ADDIS): Delete TOC fusion
5783 (toc fusion spliter): Delete TOC fusion support.
5784 (toc_fusionload_<mode>): Likewise.
5785 (toc_fusionload_di): Likewise.
5786 (fusion_gpr_load_<mode>): Delete generator function, this insn no
5787 longer needs to be named. Rename predicate to delete TOC fusion.
5788 (fusion_gpr_<P:mode>_<GPR_FUSION:mode>_load): Likewise.
5789 (fusion_gpr_<P:mode>_<GPR_FUSION:mode>_store): Likewise.
5790 (fusion_vsx_<P:mode>_<GPR_FUSION:mode>_load): Likewise.
5791 (fusion_vsx_<P:mode>_<GPR_FUSION:mode>_store): Likewise.
5792 (p9 fusion peephole2s): Rename predicate to delete TOC fusion.
5794 2018-07-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
5796 * doc/extend.texi (Basic PowerPC Built-in Functions Available on
5797 ISA 2.05): Replace __uint128_t with __uint128 and __int128_t with
5798 __int128 in built-in function prototypes.
5799 (PowerPC AltiVec Built-in Functions on ISA 2.07): Likewise.
5800 (PowerPC AltiVec Built-in Functions on ISA 3.0): Likewise.
5802 2018-07-27 Martin Sebor <msebor@redhat.com>
5804 PR tree-optimization/86696
5805 * tree-ssa-strlen.c (get_min_string_length): Handle all integer
5806 types, including enums.
5807 (handle_char_store): Be prepared for the above function to fail.
5809 2018-07-26 Qing Zhao <qing.zhao@oracle.com>
5811 * builtins.c (inline_expand_builtin_string_cmp): Disable inlining
5812 when optimization level is lower than 2 or optimize for size.
5814 2018-07-26 Martin Sebor <msebor@redhat.com>
5816 PR tree-optimization/86043
5817 PR tree-optimization/86042
5818 * tree-ssa-strlen.c (handle_builtin_memcpy): Handle strict overlaps.
5819 (get_string_cst_length): Rename...
5820 (get_min_string_length): ...to this. Add argument.
5821 (handle_char_store): Extend to handle multi-character stores by
5823 * tree.c (initializer_zerop): Use new argument. Handle MEM_REF.
5824 * tree.h (initializer_zerop): Add argument.
5826 2018-07-26 Jakub Jelinek <jakub@redhat.com>
5829 * omp-low.c (scan_sharing_clauses): Don't ignore map clauses for
5830 declare target to variables if they have always,{to,from,tofrom} map
5833 2018-07-26 Martin Liska <mliska@suse.cz>
5836 * lto-wrapper.c: Add linker_output as prefix
5837 for ltrans_output_file.
5839 2018-07-26 Segher Boessenkool <segher@kernel.crashing.org>
5841 PR rtl-optimization/85805
5842 * combine.c (reg_nonzero_bits_for_combine): Only use the last set
5843 value for hard registers if that was written in the same mode.
5845 2018-07-26 Martin Liska <mliska@suse.cz>
5847 PR gcov-profile/86536
5848 * gcov.c (format_gcov): Use printf format %.*f directly
5849 and do not handle special values.
5851 2018-07-25 Claudiu Zissulescu <claziss@synopsys.com>
5853 * common/config/arc/arc-common.c (arc_option_optimization_table):
5854 Update default optimizations for size.
5856 2018-07-25 Claudiu Zissulescu <claziss@synopsys.com>
5858 * config/arc/arc.md (movsf_insn): Add short instruction selection.
5859 * config/arc/constraints.md (CfZ): New constraint.
5860 * config/arc/fpu.md (addssf3_fpu): Use CfZ constraint.
5861 (subsf3_fpu): Likewise.
5862 (cmpsf_fpu): Likewise.
5863 (cmpsf_fpu_uneq): Likewise.
5865 2018-07-25 Claudiu Zissulescu <claziss@synopsys.com>
5867 * config/arc/arc.c (compact_memory_operand_p): Check for uncached
5869 (arc_is_uncached_mem_p): uncached applies to both the variable and
5872 2018-07-25 Claudiu Zissulescu <claziss@synopsys.com>
5874 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Add additional
5877 2018-07-25 David Malcolm <dmalcolm@redhat.com>
5879 * optinfo-emit-json.cc (class optrecord_json_writer): Convert
5880 field "m_scopes" from vec to auto_vec.
5882 2018-07-25 Martin Liska <mliska@suse.cz>
5884 * config/powerpcspe/powerpcspe-protos.h (rs6000_loop_align): Fix
5887 2018-07-25 Richard Biener <rguenther@suse.de>
5890 * dwarf2out.c (dwarf2out_decl): Do not handle nested functions
5891 special wrt context_die late.
5892 (gen_subprogram_die): Re-use DIEs in local scope.
5894 2018-07-25 Richard Sandiford <richard.sandiford@arm.com>
5896 PR tree-optimization/86644
5897 * hwint.c (ceil_log2): Fix comment. Return 0 for 0.
5899 2018-07-25 Martin Liska <mliska@suse.cz>
5902 * dumpfile.c: And excluded values with TDF_ALL_VALUES.
5903 * dumpfile.h (enum dump_flag): Defince TDF_ALL_VALUES.
5905 2018-07-25 Martin Liska <mliska@suse.cz>
5908 * params.def: Explain ASan abbreviation and provide
5909 a documentation link.
5911 2018-07-24 Martin Sebor <msebor@redhat.com>
5913 PR tree-optimization/86622
5914 PR tree-optimization/86532
5915 * builtins.h (string_length): Declare.
5916 * builtins.c (c_strlen): Correct handling of non-constant offsets.
5917 (check_access): Be prepared for non-constant length ranges.
5918 (string_length): Make extern.
5919 * expr.c (string_constant): Only handle the minor non-constant
5920 array index. Use string_constant to compute the length of
5921 a generic string constant.
5923 2018-07-24 Richard Sandiford <richard.sandiford@arm.com>
5925 PR tree-optimization/86618
5926 * tree-vect-stmts.c (vectorizable_call): Don't take the address
5927 of LOOP_VINFO_MASKS (loop_vinfo) when loop_vinfo is null.
5929 2018-07-24 David Malcolm <dmalcolm@redhat.com>
5931 PR tree-optimization/86636
5932 * json.cc (json::object::set): Fix comment. Add assertions.
5933 (json::array::append): Move here from json.h. Add comment and an
5935 (json::string::string): Likewise.
5936 * json.h (json::array::append): Move to json.cc.
5937 (json::string::string): Likewise.
5938 * optinfo-emit-json.cc
5939 (optrecord_json_writer::impl_location_to_json): Assert that we
5940 aren't attempting to write out UNKNOWN_LOCATION, or an ad-hoc
5941 wrapper around it. Expand the location once, rather than three
5943 (optrecord_json_writer::inlining_chain_to_json): Fix the check for
5944 UNKNOWN_LOCATION, to use LOCATION_LOCUS to look through ad-hoc
5946 (optrecord_json_writer::optinfo_to_json): Likewise, in four
5947 places. Fix some overlong lines.
5949 2018-07-24 Matthew Malcomson <matthew.malcomson@arm.com>
5951 * config/aarch64/aarch64-simd.md
5952 (aarch64_<ANY_EXTEND:su><ADDSUB:optab>w<mode>): Split into...
5953 (aarch64_<ANY_EXTEND:su>subw<mode>): ... This...
5954 (aarch64_<ANY_EXTEND:su>addw<mode>): ... And this.
5955 (aarch64_<ANY_EXTEND:su><ADDSUB:optab>w<mode>_internal): Split into...
5956 (aarch64_<ANY_EXTEND:su>subw<mode>_internal): ... This...
5957 (aarch64_<ANY_EXTEND:su>addw<mode>_internal): ... And this.
5958 (aarch64_<ANY_EXTEND:su><ADDSUB:optab>w2<mode>_internal): Split into...
5959 (aarch64_<ANY_EXTEND:su>subw2<mode>_internal): ... This...
5960 (aarch64_<ANY_EXTEND:su>addw2<mode>_internal): ... And this.
5962 2018-07-24 Jakub Jelinek <jakub@redhat.com>
5965 * expmed.c (expand_divmod): Punt if d == HOST_WIDE_INT_MIN
5966 and size > HOST_BITS_PER_WIDE_INT. For size > HOST_BITS_PER_WIDE_INT
5967 and abs_d == d, do the power of two handling if profitable.
5969 2018-07-24 Richard Biener <rguenther@suse.de>
5971 * match.pd: Add BIT_FIELD_REF canonicalizations.
5973 2018-07-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
5976 * genmatch.c (dt_operand::gen_match_op): Avoid folding volatile values.
5978 2018-07-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
5980 * gimple-fold.c (gimple_fold_builtin_printf): Don't create a not NUL
5981 terminated STRING_CST object.
5983 2018-07-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
5985 hsa-dump.c (dump_hsa_symbol): Avoid out of scope access to buf.
5987 2018-07-23 Segher Boessenkool <segher@kernel.crashing.org>
5989 * config/rs6000/rs6000-p8swap.c (rtx_is_swappable_p): Adjust.
5990 * config/rs6000/rs6000-protos.h (rs6000_split_v4si_init): Delete.
5991 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Always force
5992 the elements into a register.
5993 (rs6000_split_v4si_init_di_reg): Delete.
5994 (rs6000_split_v4si_init): Delete.
5995 * config/rs6000/vsx.md (unspec): Delete UNSPEC_VSX_VEC_INIT.
5996 (vsx_init_v4si): Rewrite as a define_expand.
5998 2018-07-23 Segher Boessenkool <segher@kernel.crashing.org>
6000 * config/rs6000/rs6000.md (splitters for rldimi and rlwimi with the
6001 zero_extend argument from memory): New.
6003 2018-07-22 Martin Sebor <msebor@redhat.com>
6006 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Avoid
6007 diagnosing calls with unknown arguments unless -Walloca-larger-than
6008 is restricted to less than PTRDIFF_MAX bytes.
6010 2018-07-22 Gerald Pfeifer <gerald@pfeifer.com>
6012 * doc/gcov.texi (Invoking Gcov): Editorial changes.
6014 2018-07-20 David Malcolm <dmalcolm@redhat.com>
6016 * pretty-print.c (text_info::set_location): Remove redundant
6017 "line_table" parameter from call to rich_location::set_range.
6019 2018-07-20 Martin Sebor <msebor@redhat.com>
6022 * builtins.c (expand_builtin_alloca): Adjust.
6023 * calls.c (alloc_max_size): Simplify.
6024 * cgraphunit.c (cgraph_node::expand): Adjust.
6025 * common.opt (larger_than_size, warn_frame_larger_than): Remove
6027 (frame_larger_than_size): Same.
6028 (-Wframe-larger-than, -Wlarger-than, -Wstack-usage): Change options
6029 to take a HOST_WIDE_INT argument and accept a byte-size suffix.
6031 * doc/invoke.texi (GCC Command Options): Document option arguments.
6032 Explain byte-size arguments and suffixes.
6033 (-Wvla-larger-than, -Wno-alloc-size-larger-than): Update.
6034 (-Wno-alloca-larger-than, -Wno-vla-larger-than): Same.
6035 (-Wframe-larger-than, -Wlarger-than, -Wstack-usage): Same.
6036 * doc/options.texi (UInteger): Expand.
6037 (Host_Wide_Int, ByteSize): Document new properties.
6038 * final.c (final_start_function_1): Include sizes in an error message.
6039 * function.c (frame_offset_overflow): Same.
6040 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Adjust.
6041 (alloca_call_type_by_arg): Change function argument to HOST_WIDE_INT.
6042 Diagnose unbounded alloca calls only for limits of less than
6044 (alloca_call_type): Adjust. Diagnose possibly out-of-bounds alloca
6045 calls and VLA size only for limits of less than PTRDIFF_MAX. Same
6047 (pass_walloca::execute): Adjust. Diagnose alloca calls in loops
6048 only for limits of less than PTRDIFF_MAX.
6049 * langhooks-def.h (lhd_handle_option): Change function argument
6051 * langhooks.c (lhd_handle_option): Same.
6052 * langhooks.h (handle_option): Same.
6053 * opt-functions.awk (switch_bit_fields): Handle Host_Wide_Int and
6055 (var_type, var_type_struct): Same.
6056 (var_set): Handle ByteSize flag.
6057 * optc-gen.awk: Add comments to output to ease debugging. Make
6058 use of HOST_WIDE_INT where appropriate.
6059 * opts-gen-save.awk: Use %lx to format unsigned long.
6060 * opth-gen.awk: Change function argument to HOST_WIDE_INT.
6061 * opts-common.c (integral_argument): Return HOST_WIDE_INT and add
6062 arguments. Parse bytes-size suffixes.
6063 (enum_arg_to_value): Change function argument to HOST_WIDE_INT.
6064 (enum_value_to_arg): Same.
6065 (decode_cmdline_option): Handle cl_host_wide_int. Adjust.
6066 (handle_option): Adjust.
6067 (generate_option): Change function argument to HOST_WIDE_INT.
6068 (cmdline_handle_error): Adjust.
6069 (read_cmdline_option): Change function argument to HOST_WIDE_INT.
6070 (set_option): Change function argument to HOST_WIDE_INT.
6071 (option_enabled): Handle cl_host_wide_int.
6072 (get_option_state): Handle CLVC_SIZE.
6073 (control_warning_option): Same.
6074 * opts.c (common_handle_option): Change function argument to
6075 HOST_WIDE_INT. Remove handling of OPT_Walloca_larger_than_ and
6076 OPT_Wvla_larger_than_.
6077 * opts.h (enum cl_var_type): Add an enumerator.
6078 * stor-layout.c (layout_decl): Print a more meaningful warning.
6079 * toplev.c (output_stack_usage): Adjust.
6081 2018-07-20 Qing Zhao <qing.zhao@oracle.com>
6083 * builtins.c (expand_builtin_memcmp): Delete the last parameter for
6084 call to inline_expand_builtin_string_cmp.
6085 (expand_builtin_strcmp): Likewise.
6086 (expand_builtin_strncmp): Likewise.
6087 (inline_string_cmp): Delete the last parameter, change char_type_node
6088 to unsigned_char_type_node for strcmp/strncmp, add conversions to the
6090 (inline_expand_builtin_string_cmp): Delete the last parameter, give up
6091 the inlining expansion on target where the type of the call has same or
6092 narrower precision than unsigned char.
6094 2018-07-20 David Malcolm <dmalcolm@redhat.com>
6096 * Makefile.in (OBJS): Add json.o and optinfo-emit-json.o.
6097 (CFLAGS-optinfo-emit-json.o): Define TARGET_NAME.
6098 * common.opt (fsave-optimization-record): New option.
6099 * coretypes.h (struct kv_pair): Move here from dumpfile.c.
6100 * doc/invoke.texi (-fsave-optimization-record): New option.
6101 * dumpfile.c: Include "optinfo-emit-json.h".
6102 (struct kv_pair): Move to coretypes.h.
6103 (optgroup_options): Make non-static.
6104 (dump_context::end_scope): Call
6105 optimization_records_maybe_pop_dump_scope.
6106 * dumpfile.h (optgroup_options): New decl.
6107 * json.cc: New file.
6109 * optinfo-emit-json.cc: New file.
6110 * optinfo-emit-json.h: New file.
6111 * optinfo.cc: Include "optinfo-emit-json.h".
6112 (optinfo::emit): Call optimization_records_maybe_record_optinfo.
6113 (optinfo_enabled_p): Check optimization_records_enabled_p.
6114 (optinfo_wants_inlining_info_p): Likewise.
6115 * optinfo.h: Update comment.
6116 * profile-count.c (profile_quality_as_string): New function.
6117 * profile-count.h (profile_quality_as_string): New decl.
6118 (profile_count::quality): New accessor.
6119 * selftest-run-tests.c (selftest::run_tests): Call json_cc_tests
6120 and optinfo_emit_json_cc_tests.
6121 * selftest.h (selftest::json_cc_tests): New decl.
6122 (selftest::optinfo_emit_json_cc_tests): New decl.
6123 * toplev.c: Include "optinfo-emit-json.h".
6124 (compile_file): Call optimization_records_finish.
6125 (do_compile): Call optimization_records_start.
6126 * tree-ssa-live.c: Include optinfo.h.
6127 (remove_unused_scope_block_p): Retain inlining information if
6128 optinfo_wants_inlining_info_p returns true.
6130 2018-07-20 Richard Biener <rguenther@suse.de>
6133 * dwarf2out.c (dwarf2out_die_ref_for_decl): Test in_lto_p
6134 to cover -flto-partition=none.
6136 2018-07-20 Martin Liska <mliska@suse.cz>
6138 * tree.h (DECL_LOCATION_RANGE): Remove unused macro.
6139 (get_decl_source_range): Remove unused function.
6141 2018-07-20 Richard Biener <rguenther@suse.de>
6143 * tree-ssa-sccvn.h (struct vn_nary_op_s): Add next member.
6144 (struct vn_phi_s): Likewise.
6145 (struct vn_reference_s): Likewise.
6146 * tree-ssa-sccvn.c (vn_nary_op_hasher::equal): Add shortcut
6147 for searching the slot of an entry known to be in the hash itself.
6148 (vn_phi_hasher::equal): Likewise.
6149 (vn_reference_hasher::equal): Likewise.
6150 (last_inserted_ref, last_inserted_phi, last_inserted_nary): New
6152 (optimistic_info, current_info): Remove, keeping only valid_info.
6153 (vn_reference_lookup_1): Remove fallback lookup.
6154 (vn_reference_lookup_2): Likewise.
6155 (vn_nary_op_lookup_1): Likewise.
6156 (vn_phi_lookup): Likewise.
6157 (vn_nary_build_or_lookup_1): Make sure to not chain the built
6159 (vn_reference_insert): Adjust, chain the inserted hash element
6160 at last_inserted_ref.
6161 (vn_reference_insert_pieces): Likewise.
6162 (visit_reference_op_call): Likewise.
6163 (vn_nary_op_insert_into): Chain the inserted hash element at
6165 (vn_nary_op_insert_pieces): Adjust.
6166 (vn_nary_op_insert): Likewise.
6167 (vn_nary_op_insert_stmt): Likewise.
6168 (vn_phi_insert): Adjust, chain the inserted hash element at
6170 (process_scc): Remove clearing and copying the optimistic
6171 table. Instead remove elements inserted during an optimistic
6172 iteration from the single table we maintain.
6173 (init_scc_vn): Adjust.
6174 (free_scc_vn): Likewise.
6175 (sccvn_dom_walker::record_cond): Likewise.
6176 (sccvn_dom_walker::after_dom_children): Likewise.
6178 2018-07-19 Martin Sebor <msebor@redhat.com>
6180 PR tree-optimization/84047
6181 PR tree-optimization/83776
6182 * tree-vrp.c (vrp_prop::check_mem_ref): New function.
6183 (check_array_bounds): Call it.
6185 2018-07-19 Martin Sebor <msebor@redhat.com>
6187 * align.h (align_flags): Use member initialization.
6189 2018-07-19 David Malcolm <dmalcolm@redhat.com>
6191 * Makefile.in (OBJS): Add optinfo.o.
6192 * coretypes.h (class symtab_node): New forward decl.
6193 (struct cgraph_node): New forward decl.
6194 (class varpool_node): New forward decl.
6195 * dump-context.h: New file.
6196 * dumpfile.c: Include "optinfo.h", "dump-context.h", "cgraph.h",
6198 (refresh_dumps_are_enabled): Use optinfo_enabled_p.
6199 (set_dump_file): Call dumpfile_ensure_any_optinfo_are_flushed.
6200 (set_alt_dump_file): Likewise.
6201 (dump_context::~dump_context): New dtor.
6202 (dump_gimple_stmt): Move implementation to...
6203 (dump_context::dump_gimple_stmt): ...this new member function.
6204 Add the stmt to any pending optinfo, creating one if need be.
6205 (dump_gimple_stmt_loc): Move implementation to...
6206 (dump_context::dump_gimple_stmt_loc): ...this new member function.
6207 Start a new optinfo and add the stmt to it.
6208 (dump_gimple_expr): Move implementation to...
6209 (dump_context::dump_gimple_expr): ...this new member function.
6210 Add the stmt to any pending optinfo, creating one if need be.
6211 (dump_gimple_expr_loc): Move implementation to...
6212 (dump_context::dump_gimple_expr_loc): ...this new member function.
6213 Start a new optinfo and add the stmt to it.
6214 (dump_generic_expr): Move implementation to...
6215 (dump_context::dump_generic_expr): ...this new member function.
6216 Add the tree to any pending optinfo, creating one if need be.
6217 (dump_generic_expr_loc): Move implementation to...
6218 (dump_context::dump_generic_expr_loc): ...this new member
6219 function. Add the tree to any pending optinfo, creating one if
6221 (dump_printf): Move implementation to...
6222 (dump_context::dump_printf_va): ...this new member function. Add
6223 the text to any pending optinfo, creating one if need be.
6224 (dump_printf_loc): Move implementation to...
6225 (dump_context::dump_printf_loc_va): ...this new member function.
6226 Start a new optinfo and add the stmt to it.
6227 (dump_dec): Move implementation to...
6228 (dump_context::dump_dec): ...this new member function. Add the
6229 value to any pending optinfo, creating one if need be.
6230 (dump_context::dump_symtab_node): New member function.
6231 (dump_context::get_scope_depth): New member function.
6232 (dump_context::begin_scope): New member function.
6233 (dump_context::end_scope): New member function.
6234 (dump_context::ensure_pending_optinfo): New member function.
6235 (dump_context::begin_next_optinfo): New member function.
6236 (dump_context::end_any_optinfo): New member function.
6237 (dump_context::s_current): New global.
6238 (dump_context::s_default): New global.
6239 (dump_scope_depth): Delete global.
6240 (dumpfile_ensure_any_optinfo_are_flushed): New function.
6241 (dump_symtab_node): New function.
6242 (get_dump_scope_depth): Reimplement in terms of dump_context.
6243 (dump_begin_scope): Likewise.
6244 (dump_end_scope): Likewise.
6245 (selftest::temp_dump_context::temp_dump_context): New ctor.
6246 (selftest::temp_dump_context::~temp_dump_context): New dtor.
6247 (selftest::verify_item): New function.
6248 (ASSERT_IS_TEXT): New macro.
6249 (ASSERT_IS_TREE): New macro.
6250 (ASSERT_IS_GIMPLE): New macro.
6251 (selftest::test_capture_of_dump_calls): New test.
6252 (selftest::dumpfile_c_tests): Call it.
6253 * dumpfile.h (dump_printf, dump_printf_loc, dump_basic_block)
6254 (dump_generic_expr_loc, dump_generic_expr, dump_gimple_stmt_loc)
6255 (dump_gimple_stmt, dump_dec): Gather these related decls and add a
6256 descriptive comment.
6257 (dump_function, print_combine_total_stats, enable_rtl_dump_file)
6258 (dump_node, dump_bb): Move these unrelated decls.
6259 (class dump_manager): Add leading comment.
6260 * optinfo.cc: New file.
6261 * optinfo.h: New file.
6263 2018-07-19 Michael Collison <michael.collison@arm.com>
6264 Richard Henderson <rth@redhat.com>
6266 * config/aarch64/aarch64.md (subv<GPI>4, usubv<GPI>4): New patterns.
6267 (subti): Handle op1 zero.
6268 (subvti4, usub4ti4): New.
6269 (*sub<GPI>3_compare1_imm): New.
6270 (sub<GPI>3_carryinCV): New.
6271 (*sub<GPI>3_carryinCV_z1_z2, *sub<GPI>3_carryinCV_z1): New.
6272 (*sub<GPI>3_carryinCV_z2, *sub<GPI>3_carryinCV): New.
6274 2018-07-19 Michael Collison <michael.collison@arm.com>
6275 Richard Henderson <rth@redhat.com>
6277 * config/aarch64/aarch64.md: (addv<GPI>4, uaddv<GPI>4): New.
6278 (addti3): Create simpler code if low part is already known to be 0.
6279 (addvti4, uaddvti4): New.
6280 (*add<GPI>3_compareC_cconly_imm): New.
6281 (*add<GPI>3_compareC_cconly): New.
6282 (*add<GPI>3_compareC_imm): New.
6283 (*add<GPI>3_compareC): Rename from add<GPI>3_compare1; do not
6284 handle constants within this pattern..
6285 (*add<GPI>3_compareV_cconly_imm): New.
6286 (*add<GPI>3_compareV_cconly): New.
6287 (*add<GPI>3_compareV_imm): New.
6288 (add<GPI>3_compareV): New.
6289 (add<GPI>3_carryinC, add<GPI>3_carryinV): New.
6290 (*add<GPI>3_carryinC_zero, *add<GPI>3_carryinV_zero): New.
6291 (*add<GPI>3_carryinC, *add<GPI>3_carryinV): New.
6292 ((*add<GPI>3_compareC_cconly_imm): Replace 'ne' operator
6293 with 'comparison' operator.
6294 (*add<GPI>3_compareV_cconly_imm): Ditto.
6295 (*add<GPI>3_compareV_cconly): Ditto.
6296 (*add<GPI>3_compareV_imm): Ditto.
6297 (add<GPI>3_compareV): Ditto.
6298 (add<mode>3_carryinC): Ditto.
6299 (*add<mode>3_carryinC_zero): Ditto.
6300 (*add<mode>3_carryinC): Ditto.
6301 (add<mode>3_carryinV): Ditto.
6302 (*add<mode>3_carryinV_zero): Ditto.
6303 (*add<mode>3_carryinV): Ditto.
6305 2018-07-19 Michael Collison <michael.collison@arm.com>
6306 Richard Henderson <rth@redhat.com>
6308 * config/aarch64/aarch64-modes.def (CC_V): New.
6309 * config/aarch64/aarch64-protos.h
6310 (aarch64_addti_scratch_regs): Declare
6311 (aarch64_subvti_scratch_regs): Declare.
6312 (aarch64_expand_subvti): Declare.
6313 (aarch64_gen_unlikely_cbranch): Declare
6314 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Test
6315 for signed overflow using CC_Vmode.
6316 (aarch64_get_condition_code_1): Handle CC_Vmode.
6317 (aarch64_gen_unlikely_cbranch): New function.
6318 (aarch64_addti_scratch_regs): New function.
6319 (aarch64_subvti_scratch_regs): New function.
6320 (aarch64_expand_subvti): New function.
6322 2018-07-19 Andre Vieira <andre.simoesdiasvieira@arm.com>
6324 * config/aarch64/aarch64-option-extensions.def: New entry for profile
6326 * config/aarch64/aarch64.h (AARCH64_FL_PROFILE): New.
6327 * doc/invoke.texi (aarch64-feature-modifiers): New entry for profile
6330 2018-07-19 Andre Vieira <andre.simoesdiasvieira@arm.com>
6333 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Make
6334 address check not strict.
6336 2018-07-19 Andre Vieira <andre.simoesdiasvieira@arm.com>
6338 * config/aarch64/aarch64-simd.md (aarch64_simd_mov<VQ:mode>): Replace
6340 (store_pair_lanes<mode>): Likewise.
6341 * config/aarch64/aarch64-protos.h (aarch64_addr_query_type): Add new
6342 enum value 'ADDR_QUERY_LDP_STP_N'.
6343 * config/aarch64/aarch64.c (aarch64_addr_query_type): Likewise.
6344 (aarch64_print_address_internal): Add declaration.
6345 (aarch64_print_ldpstp_address): Remove.
6346 (aarch64_classify_address): Adapt mode for 'ADDR_QUERY_LDP_STP_N'.
6347 (aarch64_print_operand): Change printing of 'y'.
6348 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Use
6349 new enum value 'ADDR_QUERY_LDP_STP_N', don't hardcode mode and use
6350 'true' rather than '1'.
6351 * config/aarch64/constraints.md (Uml): Likewise.
6352 (Uml): Rename to Umn.
6355 2018-07-19 Richard Biener <rguenther@suse.de>
6357 * tree-ssa-sccvn.h (struct vn_phi_s): Make phiargs member
6359 * tree-ssa-sccvn.c: Remove alloc-pool.h use.
6360 (vn_phi_hasher): Derive from nofree_ptr_hash and remove remove method.
6361 (vn_reference_hasher): Likewise.
6362 (struct vn_tables_s): Remove obstack and alloc-pool members.
6363 (vn_tables_obstack, vn_tables_insert_obstack): New global obstacks.
6364 (vn_nary_build_or_lookup_1): Manually build in vn_tables_insert_obstack.
6365 (vn_reference_insert): Allocate from obstack instead of from alloc-pool.
6366 (vn_reference_insert_pieces): Likewise.
6367 (alloc_vn_nary_op_noinit): Adjust.
6368 (vn_nary_op_insert_stmt): Allocate phiargs in-place.
6369 (vn_phi_eq): Adjust.
6370 (shared_lookup_phiargs): Remove.
6371 (vn_phi_lookup): Allocate temporary vn_phi_s on the stack.
6372 (vn_phi_insert): Allocate from obstack instead of from alloc-pool.
6373 (visit_reference_op_call): Likewise.
6374 (copy_nary, copy_phi, copy_reference): Remove.
6375 (process_scc): Rewind the obstack when iterating. Do not
6376 copy the elements to valid_info but just move them from one
6377 hashtable to the other.
6378 (allocate_vn_table): Adjust.
6379 (free_vn_table): Likewise.
6380 (init_scc_vn): Likewise.
6381 (free_scc_vn): Likewise.
6383 2018-07-19 H.J. Lu <hongjiu.lu@intel.com>
6386 * config/i386/i386.c (rest_of_insert_endbranch): Lookup
6387 indirect_return as function type attribute.
6388 (ix86_attribute_table): Change indirect_return to function
6390 * doc/extend.texi: Update indirect_return attribute.
6392 2018-07-19 Aldy Hernandez <aldyh@redhat.com>
6394 * wide-int.h (widest2_int): New.
6395 * gimple-fold.c (arith_overflowed_p): Use it.
6396 * tree.h (widest2_int_cst): New.
6397 * tree-vrp.c (wide_int_binop_overflow): Rename from
6398 vrp_int_const_binop.
6399 Rewrite to work on trees.
6400 (extract_range_from_multiplicative_op_1): Abstract code to...
6401 (wide_int_range_min_max): ...here.
6402 (wide_int_range_cross_product): ...and here.
6403 (extract_range_from_binary_expr_1): Abstract overflow code to...
6404 (wide_int_range_mult_wrapping): ...here.
6405 * tree-vrp.h (wide_int_range_cross_product): New.
6406 (wide_int_range_mult_wrapping): New.
6408 2018-07-19 Andrew Senkevich <andrew.senkevich@intel.com>
6409 Julia Koval <julia.koval@intel.com>
6411 * config/i386/x86-tune-costs.h (skylake_memcpy,
6412 skylake_memset): Replace rep_prefix with unrolling for size 512.
6414 2018-07-18 Kugan Vivekanandarajah <kuganv@linaro.org>
6417 * tree-ssa-phiopt.c (cond_removal_in_popcount_pattern): Handle
6418 comparision with EQ_EXPR in last stmt.
6420 2018-07-18 Kelvin Nilsen <kelvin@gcc.gnu.org>
6422 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Rename
6423 this subsection to "PowerPC AltiVec/VSX Built-in Functions".
6424 (PowerPC AltiVec/VSX Built-in Functions): New name for subsection
6425 previously known as "PowerPC AltiVec Built-in Functions". Move
6426 some material to new subsubsections "PowerPC AltiVec Built-in
6427 Functions on ISA 2.06" and "PowerPC AltiVec Built-in Functions on
6429 (PowerPC Altivec Built-in Functions on ISA 2.05): New subsubsection.
6430 (PowerPC Altivec Built-in Functions on ISA 2.06): Likewise.
6431 (PowerPC Altivec Built-in Functions on ISA 2.07): Likewise.
6432 (PowerPC Altivec Built-in Functions on ISA 3.0): Likewise.
6434 2018-07-18 Richard Biener <rguenther@suse.de>
6436 PR tree-optimization/86557
6437 * tree-vect-patterns.c (vect_recog_divmod_pattern): Also handle
6440 2018-07-18 Ilya Leoshkevich <iii@linux.ibm.com>
6442 * config/s390/s390.c (s390_function_profiler): Generate CFI.
6444 2018-07-17 Jeff Law <law@redhat.com>
6446 * config/arm/arm.c (get_label_padding): Update for recent
6447 changes to label_to_alignment.
6449 PR tree-optimization/86010
6450 * tree-ssa-dse.c (compute_trims): Fix typo/thinko.
6452 * config/mips/mips.c (vr4130_align_insns): Update for recent
6453 changes to label_to_alignment.
6455 * config/frv/frv.c (frv_label_align): Update for recent changes
6456 to label_to_alignment.
6458 * config/nios2/nios2.c (nios2_label_align): Update for recent
6459 changes which dropped ALIGN_LABELS_LOG.
6461 2018-07-17 Andreas Schwab <schwab@linux-m68k.org>
6463 * config/m68k/m68k.md (umulsi3_highpart+1, const_umulsi3_highpart)
6464 (smulsi3_highpart+1, const_smulsi3_highpart): Add CC_STATUS_INIT.
6466 2018-07-17 Claudiu Zissulescu <claziss@synopsys.com>
6468 * config/arc/arc.c (arc_label_align): Use align_labels instead of
6469 deprecated align_labels_log.
6471 2018-07-17 Richard Biener <rguenther@suse.de>
6474 * dwarf2out.c (init_sections_and_labels): Always generate
6475 a debug_line_str_section for early LTO debug.
6476 (dwarf2out_finish): Reset debug_line_str_hash output early.
6477 Bump counter for extra dwarf5 .debug_loc labels to not conflict
6479 (dwarf2out_early_finish): Output debug_line_str.
6481 2018-07-17 Robin Dapp <rdapp@linux.ibm.com>
6483 * config/s390/s390.c (preferred_la_operand_p): Do not use LA with
6484 index register on z196 or later.
6486 2018-07-17 Robin Dapp <rdapp@linux.ibm.com>
6488 * config/s390/s390.c (s390_default_align): Set default function
6490 (s390_override_options_after_change): Call s390_default align.
6491 (s390_option_override_internal): Call s390_default_align.
6492 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define.
6494 2018-07-17 Jakub Jelinek <jakub@redhat.com>
6497 * omp-low.c (create_task_copyfn): Copy over also fields corresponding
6498 to _looptemp_ clauses, other than the first two.
6500 2018-07-17 Martin Liska <mliska@suse.cz>
6502 * opts.c: Do not enable OPT_falign_* for -Os.
6504 2018-07-17 Martin Liska <mliska@suse.cz>
6506 * align.h (MAX_CODE_ALIGN): New.
6507 (MAX_CODE_ALIGN_VALUE): New.
6508 * common/config/i386/i386-common.c (ix86_handle_option):
6509 (MAX_CODE_ALIGN): Moved to align.h.
6510 * final.c (MAX_CODE_ALIGN): Likewise.
6511 * opts.c (parse_and_check_align_values):
6512 (MAX_CODE_ALIGN): Likewise.
6513 (MAX_CODE_ALIGN_VALUE): Likewise.
6515 2018-07-17 Martin Liska <mliska@suse.cz>
6517 * config/i386/att.h (ASM_OUTPUT_ALIGN): Fix spacing
6518 in order to fulfil coding style.
6519 * config/i386/cygming.h (ASM_OUTPUT_ALIGN): Likewise.
6520 * config/i386/gas.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
6521 * config/i386/x86-64.h (ASM_OUTPUT_MAX_SKIP_PAD): Likewise.
6522 * config/iq2000/iq2000.h (ASM_OUTPUT_ALIGN): Likewise.
6523 * config/pa/pa.h (ASM_OUTPUT_ALIGN): Likewise.
6524 * config/sparc/sol2.h (ASM_OUTPUT_ALIGN_WITH_NOP): Likewise.
6525 * config/sparc/sparc.h (ASM_OUTPUT_ALIGN): Likewise.
6526 * config/visium/visium.h (ASM_OUTPUT_ALIGN): Likewise.
6527 (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
6529 2018-07-17 Martin Liska <mliska@suse.cz>
6531 * align.h: New file.
6532 * config/alpha/alpha.c (alpha_align_insns_1): Use align_functions
6534 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Use new return type
6535 align_flags of label_to_alignment.
6536 * config/m32r/m32r.h (LOOP_ALIGN): Wrap returned values into
6538 * config/m68k/m68k.c: Do not use removed align_labels_value and
6540 * config/nds32/nds32.h (JUMP_ALIGN): Wrap result into align_flags class.
6541 (LOOP_ALIGN): Likewise.
6542 (LABEL_ALIGN): Likewise.
6543 * config/powerpcspe/powerpcspe.c (TARGET_ASM_LOOP_ALIGN_MAX_SKIP):
6544 Remove not used macro.
6545 (rs6000_loop_align): Change return type to align_flags.
6546 (rs6000_loop_align_max_skip): Remove.
6547 * config/rs6000/rs6000-protos.h (rs6000_loop_align):
6548 Change return type to align_flags.
6549 * config/rs6000/rs6000.c (TARGET_ASM_LOOP_ALIGN_MAX_SKIP):
6550 Remove not used macro.
6551 (rs6000_loop_align): Change return type to align_flags.
6552 (rs6000_loop_align_max_skip): Remove.
6553 * config/rx/rx.h (JUMP_ALIGN): Wrap integer values
6554 * config/rx/rx-protos.h (rx_align_for_label): Make it
6556 * config/rx/rx.c (rx_align_for_label): Change return type
6558 (rx_max_skip_for_label): Remove TARGET_ASM_*_ALIGN_MAX_SKIP
6560 into align_flags class.
6561 (LABEL_ALIGN): Likewise.
6562 (LOOP_ALIGN): Likewise.
6563 * config/s390/s390.c (s390_label_align): Use align_flags
6565 (s390_asm_output_function_label): Likewise.
6566 * config/sh/sh.c (sh_override_options_after_change):
6567 Use align_flags class directly without macros.
6568 (find_barrier): Likewise.
6569 (barrier_align): Likewise.
6570 (sh_loop_align): Likewise.
6571 * config/spu/spu.c (spu_option_override):
6572 Use align_flags_tuple::get_value instead of removed macros.
6573 (spu_sched_init): Likewise.
6574 * config/spu/spu.h (GTY): Likewise.
6575 * config/visium/visium.c (visium_option_override):
6576 Set "8" as default secondary alignment.
6577 * config/visium/visium.h (SUBALIGN_LOG): Define to 3
6578 in order to guarantee secondary alignment of 8.
6579 * coretypes.h: Include align.h header file.
6580 * doc/tm.texi: Remove TARGET_ASM_JUMP_ALIGN_MAX_SKIP,
6581 TARGET_ASM_LOOP_ALIGN_MAX_SKIP, TARGET_ASM_LABEL_ALIGN_MAX_SKIP
6582 and TARGET_ASM_LABEL_ALIGN_AFTER_BARRIER_MAX_SKIP macros.
6583 * doc/tm.texi.in: Likewise.
6584 * final.c (struct label_alignment): Remove not used structure.
6585 (LABEL_ALIGN): Change type to align_flags.
6586 (LOOP_ALIGN): Likewise.
6587 (JUMP_ALIGN): Likewise.
6588 (default_loop_align_max_skip): Remove.
6589 (default_label_align_max_skip): Likewise.
6590 (default_jump_align_max_skip): Likewise.
6591 (default_label_align_after_barrier_max_skip):
6592 (LABEL_TO_ALIGNMENT): Change to access label_align vector.
6593 (LABEL_TO_MAX_SKIP): Remove.
6594 (label_to_alignment): Return align_flags type instead of integer.
6595 (label_to_max_skip): Remove.
6596 (align_fuzz): Use align_flags type.
6597 (compute_alignments): Use align_flags type and use align_flags::max
6598 to combine multiple alignments.
6599 (grow_label_align): Grow vec instead of C array.
6600 (update_alignments): Assign just LABEL_TO_ALIGNMENT.
6601 (shorten_branches): Use align_flags type and use align_flags::max
6602 to combine multiple alignments.
6603 (final_scan_insn_1): Remove usage of secondary alignment that comes
6604 from label alignment, but instead use proper secondary alignment
6605 which is computed in grow_label_align.
6606 * flags.h (struct align_flags_tuple): Move to align.h.
6607 (struct align_flags): Likewise.
6608 (state_align_loops): Rename to align_loops.
6609 (state_align_jumps): Rename to align_jumps.
6610 (state_align_labels): Rename to align_labels.
6611 (state_align_functions): Rename to align_functions.
6612 (align_loops_log): Remove.
6613 (align_jumps_log): Remove.
6614 (align_labels_log): Remove.
6615 (align_functions_log): Remove.
6616 (align_loops_max_skip): Remove.
6617 (align_jumps_max_skip): Remove.
6618 (align_labels_max_skip): Remove.
6619 (align_functions_max_skip): Remove.
6620 (align_loops_value): Remove.
6621 (align_jumps_value): Remove.
6622 (align_labels_value): Remove.
6623 (align_functions_value): Remove.
6624 * output.h (label_to_alignment): Change return type to align_flags.
6625 (label_to_max_skip): Remove.
6626 * target.def: Remove loop_align_max_skip, label_align_max_skip,
6627 jump_align_max_skip macros.
6628 * targhooks.h (default_loop_align_max_skip): Remove.
6629 (default_label_align_max_skip): Likewise.
6630 (default_jump_align_max_skip): Likewise.
6631 (default_label_align_after_barrier_max_skip): Remove.
6632 * toplev.c (read_log_maxskip): Use ::normalize function.
6633 (parse_N_M): Remove not used argument and also call ::normalize.
6634 (parse_alignment_opts): Do not pass unused arguments.
6635 * varasm.c (assemble_start_function): Use directly align_functions
6636 instead of removed macros.
6637 * system.h: Do not poison removed macros.
6639 2018-07-17 Jakub Jelinek <jakub@redhat.com>
6642 * gimplify.c (gimplify_omp_for): Ensure taskloop firstprivatized init
6643 and cond temporaries don't have reference type if iterator has
6644 pointer type. For init use &for_pre_body instead of pre_p if
6645 for_pre_body is non-empty.
6647 2018-07-16 Segher Boessenkool <segher@kernel.crashing.org>
6649 * config/rs6000/rs6000.md (trunc<mode>sf2): Expand truncates of
6650 double-double modes to SFmode directly directly.
6651 (trunc<mode>sf2_fprs): Delete.
6653 2018-07-16 Segher Boessenkool <segher@kernel.crashing.org>
6655 * config/rs6000/rs6000.c (init_float128_ibm): Use the correct names
6656 for conversions between IFmode and the decimal floating point modes.
6657 (init_float128_ieee): Use the correct names for conversions between
6658 KFmode and the decimal floating point modes.
6660 2018-07-16 Segher Boessenkool <segher@kernel.crashing.org>
6662 * config/rs6000/rs6000.c (init_float128_ibm): Use more correct names
6663 for the conversions between TDmode and IFmode.
6664 (init_float128_ieee): Use more correct names for the conversions
6665 between TDmode and KFmode.
6667 2018-07-16 Jakub Jelinek <jakub@redhat.com>
6669 PR tree-optimization/86526
6670 * builtins.c (expand_builtin_memcmp): Formatting fixes.
6671 (inline_expand_builtin_string_cmp): Likewise.
6672 (inline_string_cmp): Likewise. Use c_readstr instead of
6673 builtin_memcpy_read_str. Add unit_mode temporary.
6675 2018-07-16 Bernd Edlinger <bernd.edlinger@hotmail.de>
6678 * builtins.c (check_access): Bail out if range[0] is no INTEGER_CST.
6679 * expr.c (string_constant): Fix the element size of ARRAY_TYPE.
6681 2018-07-16 Kelvin Nilsen <kelvin@gcc.gnu.org>
6683 * doc/extend.texi (PowerPC AltiVec Built-in Functions):
6684 Alphabetize prototypes of built-in functions, separating out
6685 built-in functions that are listed in this section but should be
6686 described elsewhere.
6688 2018-07-16 Uros Bizjak <ubizjak@gmail.com>
6691 * expmed.c (emit_store_flag): Do not emit setcc followed by a
6692 conditional move when trapping comparison was split to a
6693 non-trapping one (and vice versa).
6695 2018-07-16 Ilya Leoshkevich <iii@linux.ibm.com>
6697 * config/s390/s390.c (s390_function_profiler): Generate nops
6698 instead of profiler call sequences.
6699 * config/s390/s390.opt: Add the new option.
6701 2018-07-16 Ilya Leoshkevich <iii@linux.ibm.com>
6703 * config/s390/s390.c (s390_function_profiler): Generate
6704 __mcount_loc section.
6705 * config/s390/s390.opt: Add the new option.
6707 2018-07-16 Ilya Leoshkevich <iii@linux.ibm.com>
6709 * common.opt: Add the new warning.
6710 * config/s390/s390.c (s390_function_profiler): Emit "brasl
6711 %r0,__fentry__" when -mfentry is specified.
6712 (s390_option_override_internal): Disallow -mfentry for 31-bit
6714 * config/s390/s390.opt: Add the new option.
6716 2018-07-16 Richard Biener <rguenther@suse.de>
6719 * dwarf2out.c (dwarf2out_register_external_die): Assign DIE parents
6720 for function-local FUNCTION_DECL and RESULT_DECL immediately.
6722 2018-07-16 Martin Liska <mliska@suse.cz>
6725 * ipa-pure-const.c (malloc_candidate_p): Revert ::get
6728 2017-07-16 Claudiu Zissulescu <claziss@synopsys.com>
6730 * config/arc/arcHS.md: Update ARCHS scheduling rules.
6732 2017-07-16 Claudiu Zissulescu <claziss@synopsys.com>
6734 * config/arc/arc-arch.h (arc_tune_attr): Add new tune parameters
6736 * config/arc/arc-cpus.def (hs4x): New cpu.
6738 * config/arc/arc-tables.opt: Regenerate.
6739 * config/arc/arc.c (arc_sched_issue_rate): New function.
6740 (TARGET_SCHED_ISSUE_RATE): Define.
6741 (TARGET_SCHED_EXPOSED_PIPELINE): Likewise.
6742 * config/arc/arc.md (attr type): Add fpu_fuse, fpu_sdiv, fpu_ddiv,
6744 (attr tune): Add ARCHS4x tune values.
6745 (attr tune_dspmpy): Define.
6746 (*tst): Correct instruction type.
6747 * config/arc/arcHS.md: Don't use this automaton for ARCHS4x cpus.
6748 * config/arc/arcHS4x.md: New file.
6749 * config/arc/fpu.md: Update instruction type attributes.
6750 * config/arc/t-multilib: Regenerate.
6752 2018-07-16 Tom de Vries <tdevries@suse.de>
6755 * var-tracking.c (vt_initialize): Fix pre_dec handling.
6757 2018-07-16 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
6759 * config/aarch64/atomics.md (aarch64_store_execlusive<mode>): Add
6762 2018-07-16 Eric Botcazou <ebotcazou@adacore.com>
6764 PR tree-optimization/86514
6765 * tree-ssa-reassoc.c (init_range_entry) <CASE_CONVERT>: Return for a
6766 conversion to a boolean type from a type with greater precision.
6768 2018-07-16 Tom de Vries <tdevries@suse.de>
6770 * var-tracking.c (vt_initialize): Print adjusted insn slim if
6771 dump_flags request TDF_SLIM.
6773 2018-07-16 Aldy Hernandez <aldyh@redhat.com>
6775 * fold-const.c (int_const_binop_1): Abstract...
6776 (wide_int_binop): ...wide int code here.
6777 (poly_int_binop): ...poly int code here.
6778 Abstract the rest of int_const_binop_1 into int_const_binop.
6779 * fold-const.h (wide_int_binop): New.
6780 * tree-vrp.c (vrp_int_const_binop): Call wide_int_binop.
6781 Remove useless PLUS/MINUS_EXPR case.
6782 (zero_nonzero_bits_from_vr): Move wide int code...
6783 (zero_nonzero_bits_from_bounds): ...here.
6784 (extract_range_from_binary_expr_1): Move mask optimization code...
6785 (range_easy_mask_min_max): ...here.
6786 * tree-vrp.h (zero_nonzero_bits_from_bounds): New.
6787 (range_easy_mask_min_max): New.
6789 2018-07-15 Jeff Law <law@redhat.com>
6792 * config/sh/sh.c (output_mi_thunk): Remove dead conditional
6795 2018-07-14 Jim Wilson <jimw@sifive.com>
6797 * config/riscv/linux.h (TARGET_ASM_FILE_END): New.
6799 2018-07-14 Paul Koning <ni1d@arrl.net>
6801 * config/pdp11/pdp11.c (pdp11_rtx_costs): Bugfixes.
6803 2018-07-13 Jan Hubicka <hubicka@ucw.cz>
6805 * lto-streamer-out.c (copy_function_or_variable): Dump info about
6808 2018-07-13 Bill Schmidt <wschmidt@linux.ibm.com>
6809 Steve Munroe <munroesj52@gmail.com>
6811 * config/rs6000/emmintrin.h (_mm_and_si128): New function.
6812 (_mm_andnot_si128): Likewise.
6813 (_mm_or_si128): Likewise.
6814 (_mm_xor_si128): Likewise.
6816 2018-07-13 Qing Zhao <qing.zhao@oracle.com>
6819 * builtins.c (expand_builtin_memcmp): Inline the calls first
6820 when result_eq is false.
6821 (expand_builtin_strcmp): Inline the calls first.
6822 (expand_builtin_strncmp): Likewise.
6823 (inline_string_cmp): New routine. Expand a string compare
6824 call by using a sequence of char comparison.
6825 (inline_expand_builtin_string_cmp): New routine. Inline expansion
6826 a call to str(n)cmp/memcmp.
6827 * doc/invoke.texi (--param builtin-string-cmp-inline-length):
6829 * params.def (BUILTIN_STRING_CMP_INLINE_LENGTH): New.
6831 2018-07-13 Richard Earnshaw <rearnsha@arm.com>
6833 * config/arm/driver-arm.c: Include arm-native.h.
6834 (host_detect_local_cpu): Use auto-generated data tables.
6835 (vendors, arm_cpu_table): Delete. Move part information to ...
6836 * config/arm/arm-cpus.in: ... here.
6837 * config/arm/parsecpu.awk (gen_native): New function.
6838 (vendor, part): New CPU fields.
6839 (END): Add support for building the native CPU detection tables.
6840 * config/arm/t-arm (arm-native.h): Add build rule.
6841 (driver-arm.o): Add dependency on arm-native.h.
6843 2018-07-13 Richard Biener <rguenther@suse.de>
6846 * match.pd (addr1 - addr2): Allow either of the operand to
6849 2018-07-13 Tom de Vries <tdevries@suse.de>
6851 * tree-inline.c (remap_ssa_name): Save and reuse debug exprs generated
6854 2018-07-13 Jackson Woodruff <jackson.woodruff@arm.com>
6856 * config/aarch64/aarch64.c (aarch64_operands_adjust_ok_for_ldpstp): Use
6857 arrays instead of numbered variables.
6859 2018-07-13 Eric Botcazou <ebotcazou@adacore.com>
6861 * config/sparc/sparc-protos.h (sparc_compute_frame_size): Delete.
6862 * config/sparc/sparc.c (sparc_compute_frame_size): Make static.
6864 2018-07-13 Richard Biener <rguenther@suse.de>
6867 * dwarf2out.c (gen_type_die_with_usage): Use scope_die_for
6868 instead of get_context_die.
6870 2018-07-13 Kugan Vivekanandarajah <kuganv@linaro.org>
6871 Richard Biener <rguenther@suse.de>
6874 * tree-ssa-loop-niter.c (number_of_iterations_popcount): Check
6875 that the loop latch destination where phi is defined.
6877 2018-07-12 Kito Cheng <kito.cheng@gmail.com>
6879 * config/riscv/riscv.c (enum riscv_privilege_levels): Add UNKNOWN_MODE.
6880 (riscv_expand_epilogue): Add assertion to check interrupt mode.
6881 (riscv_set_current_function): Extract getting interrupt type to new
6883 (riscv_get_interrupt_type): New function.
6884 (riscv_merge_decl_attributes): New function, checking interrupt type is
6886 (TARGET_MERGE_DECL_ATTRIBUTES): Define.
6888 2018-07-12 Paul Koning <ni1d@arrl.net>
6890 * config/pdp11/pdp11.c (pdp11_output_def): Fix typo in .set
6893 2018-07-12 Paul Koning <ni1d@arrl.net>
6895 * doc/rtl.texi (REG_NONNEG): Remove decrement and branch until
6896 zero reference, add doloop_end instead.
6897 * doc/md.texi (decrement_and_branch_until_zero): Remove.
6898 (Looping patterns): Remove decrement_and_branch_until_zero. Add
6899 detail for doloop_end.
6901 2018-07-12 Martin Sebor <msebor@redhat.com>
6904 * attribs.c (decl_attributes): Reject conflicting attributes before
6905 calling attribute handlers.
6907 2018-07-12 Jan Hubicka <hubicka@ucw.cz>
6909 * dumpfile.c (gcc::dump_manager::get_dump_file_name): Add PART
6911 (gcc::dump_manager::get_dump_file_name): likewise.
6912 (dump_begin): Likewise.
6913 * dumpfile.h (dump_begin): Update prototype.
6914 (gcc::dump_manager::get_dump_file_name,
6915 gcc::dump_manager::get_dump_file_name): Update prototype.
6917 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
6919 * internal-fn.h (vectorizable_internal_fn_p): New function.
6920 * tree-vect-slp.c (compatible_calls_p): Likewise.
6921 (vect_build_slp_tree_1): Remove nops argument. Handle calls
6922 to internal functions.
6923 (vect_build_slp_tree_2): Update call to vect_build_slp_tree_1.
6925 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
6927 * fold-const.h (inverse_conditions_p): Declare.
6928 * fold-const.c (inverse_conditions_p): New function.
6929 * match.pd: Use inverse_conditions_p. Add folds of view_converts
6930 that test the inverse condition of a conditional internal function.
6931 * internal-fn.h (vectorized_internal_fn_supported_p): Declare.
6932 * internal-fn.c (internal_fn_mask_index): Handle conditional
6934 (vectorized_internal_fn_supported_p): New function.
6935 * tree-if-conv.c: Include internal-fn.h and fold-const.h.
6936 (any_pred_load_store): Replace with...
6937 (need_to_predicate): ...this new variable.
6938 (redundant_ssa_names): New variable.
6939 (ifcvt_can_use_mask_load_store): Move initial checks to...
6940 (ifcvt_can_predicate): ...this new function. Handle tree codes
6941 for which a conditional internal function exists.
6942 (if_convertible_gimple_assign_stmt_p): Use ifcvt_can_predicate
6943 instead of ifcvt_can_use_mask_load_store. Update after variable
6945 (predicate_load_or_store): New function, split out from
6946 predicate_mem_writes.
6947 (check_redundant_cond_expr): New function.
6948 (value_available_p): Likewise.
6949 (predicate_rhs_code): Likewise.
6950 (predicate_mem_writes): Rename to...
6951 (predicate_statements): ...this. Use predicate_load_or_store
6952 and predicate_rhs_code.
6953 (combine_blocks, tree_if_conversion): Update after above name changes.
6954 (ifcvt_local_dce): Handle redundant_ssa_names.
6955 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern): Handle
6956 general conditional functions.
6957 * tree-vect-stmts.c (vectorizable_call): Likewise.
6959 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
6960 Alan Hayward <alan.hayward@arm.com>
6961 David Sherwood <david.sherwood@arm.com>
6963 * internal-fn.h (can_interpret_as_conditional_op_p): Declare.
6964 * internal-fn.c (can_interpret_as_conditional_op_p): New function.
6965 * tree-ssa-math-opts.c (convert_mult_to_fma_1): Handle conditional
6966 plus and minus and convert them into IFN_COND_FMA-based sequences.
6967 (convert_mult_to_fma): Handle conditional plus and minus.
6969 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
6971 * doc/md.texi (cond_fma, cond_fms, cond_fnma, cond_fnms): Document.
6972 * optabs.def (cond_fma_optab, cond_fms_optab, cond_fnma_optab)
6973 (cond_fnms_optab): New optabs.
6974 * internal-fn.def (COND_FMA, COND_FMS, COND_FNMA, COND_FNMS): New
6976 (FMA): Use DEF_INTERNAL_FLT_FN rather than DEF_INTERNAL_FLT_FLOATN_FN.
6977 * internal-fn.h (get_conditional_internal_fn): Declare.
6978 (get_unconditional_internal_fn): Likewise.
6979 * internal-fn.c (cond_ternary_direct): New macro.
6980 (expand_cond_ternary_optab_fn): Likewise.
6981 (direct_cond_ternary_optab_supported_p): Likewise.
6982 (FOR_EACH_COND_FN_PAIR): Likewise.
6983 (get_conditional_internal_fn): New function.
6984 (get_unconditional_internal_fn): Likewise.
6985 * gimple-match.h (gimple_match_op::MAX_NUM_OPS): Bump to 5.
6986 (gimple_match_op::gimple_match_op): Add a new overload for 5
6988 (gimple_match_op::set_op): Likewise.
6989 (gimple_resimplify5): Declare.
6990 * genmatch.c (decision_tree::gen): Generate simplifications for
6992 * gimple-match-head.c (gimple_simplify): Define an overload for
6993 5 operands. Handle calls with 5 arguments in the top-level overload.
6994 (convert_conditional_op): Handle conversions from unconditional
6995 internal functions to conditional ones.
6996 (gimple_resimplify5): New function.
6997 (build_call_internal): Pass a fifth operand.
6998 (maybe_push_res_to_seq): Likewise.
6999 (try_conditional_simplification): Try converting conditional
7000 internal functions to unconditional internal functions.
7001 Handle 3-operand unconditional forms.
7002 * match.pd (UNCOND_TERNARY, COND_TERNARY): Operator lists.
7003 Define ternary equivalents of the current rules for binary conditional
7005 * config/aarch64/aarch64.c (aarch64_preferred_else_value): Handle
7007 * config/aarch64/iterators.md (UNSPEC_COND_FMLA, UNSPEC_COND_FMLS)
7008 (UNSPEC_COND_FNMLA, UNSPEC_COND_FNMLS): New unspecs.
7009 (optab): Handle them.
7010 (SVE_COND_FP_TERNARY): New int iterator.
7011 (sve_fmla_op, sve_fmad_op): New int attributes.
7012 * config/aarch64/aarch64-sve.md (cond_<optab><mode>)
7013 (*cond_<optab><mode>_2, *cond_<optab><mode_4)
7014 (*cond_<optab><mode>_any): New SVE_COND_FP_TERNARY patterns.
7016 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
7018 * target.def (preferred_else_value): New target hook.
7019 * doc/tm.texi.in (TARGET_PREFERRED_ELSE_VALUE): New hook.
7020 * doc/tm.texi: Regenerate.
7021 * targhooks.h (default_preferred_else_value): Declare.
7022 * targhooks.c (default_preferred_else_value): New function.
7023 * internal-fn.h (conditional_internal_fn_code): Declare.
7024 * internal-fn.c (FOR_EACH_CODE_MAPPING): New macro.
7025 (get_conditional_internal_fn): Use it.
7026 (conditional_internal_fn_code): New function.
7027 * gimple-match.h (gimple_match_cond): New struct.
7028 (gimple_match_op): Add a cond member function.
7029 (gimple_match_op::gimple_match_op): Update all forms to take a
7031 * genmatch.c (expr::gen_transform): Use the same condition as res_op
7032 for the suboperation, but don't specify a particular else_value.
7033 * tree-ssa-sccvn.c (vn_nary_simplify, vn_reference_lookup_3)
7034 (visit_nary_op, visit_reference_op_load): Pass
7035 gimple_match_cond::UNCOND to the gimple_match_op constructor.
7036 * gimple-match-head.c: Include tree-eh.h
7037 (convert_conditional_op): New function.
7038 (maybe_resimplify_conditional_op): Likewise.
7039 (gimple_resimplify1): Call maybe_resimplify_conditional_op.
7040 (gimple_resimplify2): Likewise.
7041 (gimple_resimplify3): Likewise.
7042 (gimple_resimplify4): Likewise.
7043 (maybe_push_res_to_seq): Return null for conditional operations.
7044 (try_conditional_simplification): New function.
7045 (gimple_simplify): Call it. Pass conditions to the gimple_match_op
7047 * match.pd: Fold VEC_COND_EXPRs of an IFN_COND_* call to a new
7049 * config/aarch64/aarch64.c (aarch64_preferred_else_value): New
7051 (TARGET_PREFERRED_ELSE_VALUE): Redefine.
7053 2018-07-12 Jan Hubicka <hubicka@ucw.cz>
7055 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not stream
7057 (hash_tree): Do not hash DECL_FCONTEXT
7058 * tree-streamer-in.c (lto_input_ts_field_decl_tree_pointers):
7059 Do not stream DECL_FCONTEXT.
7060 * tree-streamer-out.c (write_ts_field_decl_tree_pointers): Likewise.
7061 * tree.c (free_lang_data_in_decl): Free DECL_FCONTEXT.
7063 2018-07-12 Richard Biener <rguenther@suse.de>
7066 * dwarf2out.c (gen_block_die): Only output blocks when they have
7067 at least one !DECL_IGNORED_P variable.
7069 2018-07-12 Richard Biener <rguenther@suse.de>
7072 * config/gnu-user.h (GNU_USER_TARGET_NO_PTHREADS_LIB_SPEC):
7073 Remove -mieee-fp handling.
7075 2018-07-12 Richard Biener <rguenther@suse.de>
7077 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Remove bogus
7078 left-over from last patch.
7080 2018-07-12 Jakub Jelinek <jakub@redhat.com>
7082 PR tree-optimization/86492
7083 * gimple-ssa-store-merging.c
7084 (imm_store_chain_info::coalesce_immediate_stores): Call
7085 check_no_overlap even for the merge_overlapping case. Formatting fix.
7087 2018-07-12 Richard Biener <rguenther@suse.de>
7090 * fold-const.c (fold_binary_op_with_conditional_arg): Do not
7091 move possibly trapping operations into the conditional.
7093 2018-07-12 Richard Biener <rguenther@suse.de>
7095 * tree-ssa-sccvn.c (mprts_hook_cnt): Remove.
7096 (vn_lookup_simplify_result): Remove recursion limit applied
7098 (vn_nary_build_or_lookup_1): Adjust.
7099 (try_to_simplify): Likewise.
7100 * gimple-match-head.c (gimple_resimplify1): Instead apply one
7102 (gimple_resimplify2): Likewise.
7103 (gimple_resimplify3): Likewise.
7104 (gimple_resimplify4): Likewise.
7106 2018-07-11 Jakub Jelinek <jakub@redhat.com>
7108 * config/i386/avx512bitalgintrin.h (_mm512_mask_bitshuffle_epi64_mask):
7109 Use __mmask64 type instead of __mmask8 for __M argument.
7110 * config/i386/avx512fintrin.h (_mm512_mask_xor_epi64,
7111 _mm512_maskz_xor_epi64): Use __mmask8 type instead of __mmask16 for
7113 (_mm512_mask_cmpneq_epi64_mask): Use __mmask8 type instead of
7114 __mmask16 for __M argument.
7115 (_mm512_maskz_insertf32x4, _mm512_maskz_inserti32x4,
7116 _mm512_mask_insertf32x4, _mm512_mask_inserti32x4): Cast last argument
7117 to __mmask16 instead of __mmask8.
7118 * config/i386/avx512vlintrin.h (_mm_mask_add_ps, _mm_maskz_add_ps,
7119 _mm256_mask_add_ps, _mm256_maskz_add_ps, _mm_mask_sub_ps,
7120 _mm_maskz_sub_ps, _mm256_mask_sub_ps, _mm256_maskz_sub_ps,
7121 _mm256_maskz_cvtepi32_ps, _mm_maskz_cvtepi32_ps): Use __mmask8 type
7122 instead of __mmask16 for __U argument.
7123 * config/i386/avx512vlbwintrin.h (_mm_mask_cmp_epi8_mask): Use
7124 __mmask16 instead of __mmask8 for __U argument.
7125 (_mm256_mask_cmp_epi8_mask): Use __mmask32 instead of __mmask16 for
7127 (_mm256_cmp_epi8_mask): Use __mmask32 return type instead of
7129 (_mm_mask_cmp_epu8_mask): Use __mmask16 instead of __mmask8 for __U
7131 (_mm256_mask_cmp_epu8_mask): Use __mmask32 instead of __mmask16 for
7133 (_mm256_cmp_epu8_mask): Use __mmask32 return type instead of
7135 (_mm_mask_cmp_epi16_mask): Cast last argument to __mmask8 instead
7137 (_mm256_mask_cvtepi8_epi16): Use __mmask16 instead of __mmask32 for
7139 (_mm_mask_cvtepi8_epi16): Use __mmask8 instead of __mmask32 for
7141 (_mm256_mask_cvtepu8_epi16): Use __mmask16 instead of __mmask32 for
7143 (_mm_mask_cvtepu8_epi16): Use __mmask8 instead of __mmask32 for
7145 (_mm256_mask_cmpneq_epu8_mask, _mm256_mask_cmplt_epu8_mask,
7146 _mm256_mask_cmpge_epu8_mask, _mm256_mask_cmple_epu8_mask): Change
7147 return type as well as __M argument type and all casts from __mmask8
7149 (_mm256_mask_cmpneq_epu16_mask, _mm256_mask_cmplt_epu16_mask,
7150 _mm256_mask_cmpge_epu16_mask, _mm256_mask_cmple_epu16_mask): Change
7151 return type as well as __M argument type and all casts from __mmask8
7153 (_mm256_mask_cmpneq_epi8_mask, _mm256_mask_cmplt_epi8_mask,
7154 _mm256_mask_cmpge_epi8_mask, _mm256_mask_cmple_epi8_mask): Change
7155 return type as well as __M argument type and all casts from __mmask8
7157 (_mm256_mask_cmpneq_epi16_mask, _mm256_mask_cmplt_epi16_mask,
7158 _mm256_mask_cmpge_epi16_mask, _mm256_mask_cmple_epi16_mask): Change
7159 return type as well as __M argument type and all casts from __mmask8
7161 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_shrdi_epi32,
7162 _mm_mask_shldi_epi32): Cast last argument to __mmask8 instead of
7165 2018-07-11 Grazvydas Ignotas <notasas@gmail.com>
7167 * config/i386/avx512bwintrin.h: (_mm512_mask_cmp_epi8_mask,
7168 _mm512_mask_cmp_epu8_mask): Use __mmask64 type instead of __mmask32
7171 2018-07-11 Paul Koning <ni1d@arrl.net>
7173 * doc/md.texi (define_subst): Document how multiple occurrences of
7174 the same argument in the replacement pattern are handled.
7176 2018-07-11 Paul Koning <ni1d@arrl.net>
7178 * doc/extend.texi (Common Variable Attributes): Move "mode" into
7180 (Common Type Attributes): Add "mode" attribute.
7182 2018-07-11 Jan Hubicka <hubicka@ucw.cz>
7184 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not
7185 stream DECL_ORIGINAL_TYPE.
7186 (DFS::DFS_write_tree_body): Drop hack handling local external decls.
7187 (hash_tree): Do not walk DECL_ORIGINAL_TYPE.
7188 * tree-streamer-in.c (lto_input_ts_decl_non_common_tree_pointers):
7189 Do not walk original type.
7190 * tree-streamer-out.c (streamer_write_chain): Drop hack handling
7192 (write_ts_decl_non_common_tree_pointers): Do not stream
7194 * tree.c (free_lang_data_in_decl): Clear DECL_ORIGINAL_TYPE.
7195 (find_decls_types_r): Do not walk DEC_ORIGINAL_TYPE.
7197 2018-07-11 Aldy Hernandez <aldyh@redhat.com>
7199 * tree-ssa-threadupdate.c (thread_through_all_blocks): Do not jump
7200 thread twice from the same starting edge.
7202 2018-07-11 Aldy Hernandez <aldyh@redhat.com>
7204 * vr-values.c (gimple_stmt_nonzero_p): Abstract common code to...
7205 * gimple.c (gimple_call_nonnull_result_p): ...here...
7206 (gimple_call_nonnull_arg): ...and here.
7207 * gimple.h (gimple_call_nonnull_result_p): New.
7208 (gimple_call_nonnull_arg): New.
7210 2018-07-11 Richard Earnshaw <rearnsha@arm.com>
7212 * config/arm/arm-cpus.in: Move information from fpu field of each
7213 cpu definition to the isa field.
7214 * config/arm/parsecpu.awk (fpu): Delete match rule.
7215 (gen_comm_data): Don't add bits from the CPU's FPU entry.
7217 2018-07-11 Richard Biener <rguenther@suse.de>
7220 * dwarf2out.c (init_sections_and_labels): Use
7221 output_asm_line_debug_info consistently.
7222 (dwarf2out_early_finish): Likewise.
7223 (dwarf2out_finish): Remove DW_AT_stmt_list from early generated
7226 2018-07-11 Richard Biener <rguenther@suse.de>
7228 * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely_1):
7229 Rework father_bb setting in a way to avoid propagating constants
7230 multiple times on a loop body.
7232 2018-07-10 Mark Wielaard <mark@klomp.org>
7235 * dwarf2out.c (output_macinfo_op): Fix dwarf_FORM typo in gcc_assert.
7237 2018-07-10 Richard Biener <rguenther@suse.de>
7239 * hash-map.h (hash_map::iterator::operator*): Return
7240 references to key and value.
7242 2018-07-10 Jakub Jelinek <jakub@redhat.com>
7245 * gimplify.c (find_combined_omp_for): Add DATA argument, in addition
7246 to finding the inner OMP_FOR/OMP_SIMD stmt find non-trivial wrappers,
7247 BLOCKs with BLOCK_VARs, OMP_PARALLEL in between, OMP_FOR in between.
7248 (gimplify_omp_for): For composite loops, move outer
7249 OMP_{DISTRIBUTE,TASKLOOP,FOR,PARALLEL} right around innermost
7250 OMP_FOR/OMP_SIMD if there are any non-trivial wrappers. For class
7251 iterators add any needed clauses. Allow OMP_FOR_ORIG_DECLS to contain
7252 TREE_LIST for both the original class iterator and the "last" helper
7253 var. Gimplify OMP_FOR_PRE_BODY before the outermost composite
7254 loop, remember has_decl_expr from outer composite loops for the
7255 innermost OMP_SIMD in TREE_PRIVATE bit on OMP_FOR_INIT.
7257 2018-07-09 Martin Sebor <msebor@redhat.com>
7261 * builtins.c (c_strlen): Avoid out-of-bounds warnings when
7262 accessing implicitly initialized array elements.
7263 * expr.c (string_constant): Handle string initializers of
7264 character arrays within aggregates.
7265 * gimple-fold.c (fold_array_ctor_reference): Add argument.
7266 Store element offset. As a special case, handle zero size.
7267 (fold_nonarray_ctor_reference): Same.
7268 (fold_ctor_reference): Add argument. Store subobject offset.
7269 * gimple-fold.h (fold_ctor_reference): Add argument.
7271 2018-07-09 Paul Koning <ni1d@arrl.net>
7273 * config/pdp11/pdp11.c (pdp11_addr_cost): New function.
7274 (pdp11_insn_cost): New function.
7275 (pdp11_md_asm_adjust): New function.
7276 (TARGET_INVALID_WITHIN_DOLOOP): Define.
7277 (pdp11_rtx_costs): Update to match machine better.
7278 (output_addr_const_pdp11): Correct format mismatch warnings.
7279 * config/pdp11/pdp11.h (SLOW_BYTE_ACCESS): Correct definition.
7280 * config/pdp11/pdp11.md: General change to add base_cost and/or
7281 length attributes for use by new pdp11_insn_cost function.
7282 (MIN_BRANCH): Correct definition.
7284 (doloop_end): Use standard pattern name for looping pattern.
7285 (doloop_end_nocc): New.
7286 (movsf): Add another constraint alternative.
7287 (zero_extendqihi2): Add constraint alternatives for not in place
7289 (zero_extendhisi2): Remove.
7290 (shift patterns): Add CC handling variants.
7294 (define_peephole2): New peephole to recognize mov that sets CC for
7297 2018-07-09 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
7299 * config/sparc/sparc.c (sparc_fold_builtin) <SPARC_BUILTIN_PDIST,
7300 SPARC_BUILTIN_PDISTN>: Adapt for signature change of wi::neg,
7303 2018-07-09 Jakub Jelinek <jakub@redhat.com>
7306 * real.c (real_nextafter): Return true if result is denormal.
7308 2018-07-09 Martin Liska <mliska@suse.cz>
7310 * common.opt: Add back wrongly removed attribute.
7312 2018-07-09 Richard Biener <rguenther@suse.de>
7315 * dwarf2out.c (gen_block_die): For an early generated DIE
7316 always output high/low PC attributes.
7318 2018-07-09 Tom de Vries <tdevries@suse.de>
7320 * cfgexpand.c (expand_debug_source_expr): Handle VAR_DECL.
7321 * tree-inline.c (remap_ssa_name): Handle default def ssa_name mapping
7322 onto VAR_DECL with abstract origin.
7324 2018-07-07 Jim Wilson <jimw@sifive.com>
7326 * config/riscv/riscv.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): New.
7328 2018-07-07 Tom de Vries <tdevries@suse.de>
7330 * tree-dfa.c (dump_enumerated_decls): Handle cfun->cfg == NULL.
7332 2018-07-07 Aldy Hernandez <aldyh@redhat.com>
7334 * tree-vrp.c (vrp_int_const_binop): Change overflow type to
7336 (combine_bound): Use wide-int overflow calculation instead of
7338 * calls.c (maybe_warn_alloc_args_overflow): Change overflow type to
7340 * fold-const.c (int_const_binop_2): Same.
7341 (extract_muldiv_1): Same.
7342 (fold_div_compare): Same.
7343 (fold_abs_const): Same.
7345 * poly-int.h (add): Same.
7349 * predict.c (predict_iv_comparison): Same.
7350 * profile-count.c (slow_safe_scale_64bit): Same.
7351 * simplify-rtx.c (simplify_const_binary_operation): Same.
7352 * tree-chrec.c (tree_fold_binomial): Same.
7353 * tree-data-ref.c (split_constant_offset_1): Same.
7354 * tree-if-conv.c (idx_within_array_bound): Same.
7355 * tree-scalar-evolution.c (iv_can_overflow_p): Same.
7356 * tree-ssa-phiopt.c (minmax_replacement): Same.
7357 * tree-vect-loop.c (is_nonwrapping_integer_induction): Same.
7358 * tree-vect-stmts.c (vect_truncate_gather_scatter_offset): Same.
7359 * vr-values.c (vr_values::adjust_range_with_scev): Same.
7360 * wide-int.cc (wi::add_large): Same.
7361 (wi::mul_internal): Same.
7362 (wi::sub_large): Same.
7363 (wi::divmod_internal): Same.
7364 * wide-int.h: Change overflow type to overflow_type for neg, add,
7365 mul, smul, umul, div_trunc, div_floor, div_ceil, div_round,
7366 mod_trunc, mod_ceil, mod_round, add_large, sub_large,
7367 mul_internal, divmod_internal.
7368 (overflow_type): New enum.
7369 (accumulate_overflow): New.
7371 2018-07-06 Kugan Vivekanandarajah <kugan.vivekanandarajah@linaro.org>
7373 * tree-ssa-phiopt.c (cond_removal_in_popcount_pattern): New.
7374 (tree_ssa_phiopt_worker): Call cond_removal_in_popcount_pattern.
7376 2018-07-06 Kugan Vivekanandarajah <kuganv@linaro.org>
7378 * tree-ssa-loop-niter.c (number_of_iterations_popcount): If popcount
7379 argument is checked for zero before entering loop, avoid checking again.
7381 2018-07-06 Kugan Vivekanandarajah <kuganv@linaro.org>
7383 * gimplify.h (generic_expr_could_trap_p): Set as global function.
7384 * gimplify.h (generic_expr_could_trap_p): Likwise.
7385 * tree-scalar-evolution.c (expression_expensive_p): Handle COND_EXPR.
7387 2018-07-06 Jakub Jelinek <jakub@redhat.com>
7389 PR tree-optimization/86401
7390 * fold-const.c (fold_binary_loc) <case BIT_AND_EXPR>: Move the
7391 ((A & N) + B) & M -> (A + B) & M etc. optimization into ...
7392 (fold_bit_and_mask): ... here. New helper function for match.pd.
7393 * fold-const.h (fold_bit_and_mask): Declare.
7394 * match.pd (((A & N) + B) & M -> (A + B) & M): New optimization.
7396 2018-07-06 Peter Bergner <bergner@linux.ibm.com>
7399 * target.def (translate_mode_attribute): New hook.
7400 * targhooks.h (default_translate_mode_attribute): Declare.
7401 * targhooks.c (default_translate_mode_attribute): New function.
7402 * doc/tm.texi.in (TARGET_TRANSLATE_MODE_ATTRIBUTE): New hook.
7403 * doc/tm.texi: Regenerate.
7404 * config/rs6000/rs6000.c (TARGET_TRANSLATE_MODE_ATTRIBUTE): Define.
7405 (rs6000_translate_mode_attribute): New function.
7407 2018-07-06 Paul Koning <ni1d@arrl.net>
7409 * doc/md.texi (define_split): Document DONE and FAIL.
7410 (define_peephole2): Ditto.
7412 2018-07-05 Jeff Law <law@redhat.com>
7414 PR tree-optimization/86010
7415 * tree-ssa-dse.c (compute_trims): More aggressively trim at
7416 both the head and tail of mem* and str* calls.
7418 2018-07-05 Jim Wilson <jimw@sifive.com>
7420 * config.gcc (riscv*-*-*): When setting xlen, handle riscv-*.
7422 2018-07-05 Indu Bhagat <indu.bhagat@oracle.com>
7424 * config/aarch64/aarch64-simd.md: correct flags text for
7425 MIN_EXPR replacement.
7427 2018-07-05 James Clarke <jrtc27@jrtc27.com>
7429 * configure: Regenerated.
7431 2018-07-05 Carl Love <cel@us.ibm.com>
7433 * config/rs6000/rs6000-c.c: Map ALTIVEC_BUILTIN_VEC_UNPACKH for
7434 float argument to VSX_BUILTIN_DOUBLEH_V4SF.
7435 Map ALTIVEC_BUILTIN_VEC_UNPACKL for float argument to
7436 VSX_BUILTIN_DOUBLEL_V4SF.
7438 2018-07-05 Martin Sebor <msebor@redhat.com>
7441 * tree-ssa-strlen.c (maybe_set_strlen_range): Use type size rather
7442 than its domain to compute its the upper bound of a char array.
7444 2018-07-05 Nathan Sidwell <nathan@acm.org>
7446 Replace NO_IMPLICIT_EXTERN_C with SYSTEM_IMPLICIT_EXTERN_C.
7447 * doc/cpp.texi: Update comment.
7448 * doc/tm.texi: Rebuilt.
7449 * doc/tm.texi.in (NO_IMPLICIT_EXTERN_C): Replace with ...
7450 (SYSTEM_IMPLICIT_EXTERN_C): ... this, opposite sense.
7451 * doc/extend.texi (Backwards Compatibility): Clarify it is system
7452 headers affected by extern "C".
7453 * system.h: Poison NO_IMPLICIT_EXTERN_C.
7454 * config/alpha/alpha.h, config/arm/uclinux-elf.h,
7455 config/bfin/elf.h, config/cris/cris.h, config/darwin.h,
7456 config/dragonfly.h, config/freebsd.h, config/gnu-user.h,
7457 config/i386/cygming.h, config/i386/djgpp.h, config/i386/nto.h,
7458 config/ia64/hpux.h, config/lm32/lm32.h, config/lm32/uclinux-elf.h,
7459 config/lynx.h, config/mips/elf.h, config/mmix/mmix.h,
7460 config/netbsd.h, config/pa/pa-hpux.h, config/powerpcspe/sysv4.h,
7461 config/riscv/elf.h, config/rs6000/sysv4.h, config/rtems.h,
7462 config/s390/tpf.h, config/sh/newlib.h, config/sol2.h,
7463 config/sparc/openbsd64.h, config/sparc/sp-elf.h,
7464 config/sparc/sp64-elf.h, config/spu/spu.h,
7465 config/stormy16/stormy16.h, config/v850/v850.h,
7466 config/visium/visium.h, config/vx-common.h, config/xtensa/elf.h: Don't
7467 define NO_IMPLICIT_EXTERN_C.
7468 * config/rs6000/aix.h: Set SYSTEM_IMPLICIT_EXTERN_C.
7470 2018-07-05 Tamar Christina <tamar.christina@arm.com>
7473 * config/arm/arm.c (arm_can_change_mode_class): Use GET_MODE_UNIT_SIZE
7474 instead of GET_MODE_SIZE when comparing Units.
7476 2018-07-05 Tamar Christina <tamar.christina@arm.com>
7479 * rtlanal.c (set_noop_p): Constrain on mode change,
7480 include hard-reg-set.h
7482 2018-07-05 Tamar Christina <tamar.christina@arm.com>
7484 * config/aarch64/aarch64.c (aarch64_expand_movmem): Fix mode size.
7486 2018-07-05 Jakub Jelinek <jakub@redhat.com>
7489 2018-07-04 Maxim Ostapenko <m.ostapenko@samsung.com>
7492 * config/gnu-user.h (LIBASAN_EARLY_SPEC): Pass -lstdc++ for static
7494 * gcc.c: Do not pass LIBUBSAN_SPEC if ASan is enabled with UBSan.
7496 2018-07-04 Maxim Ostapenko <m.ostapenko@samsung.com>
7499 * config/gnu-user.h (LIBASAN_EARLY_SPEC): Pass -lstdc++ for static
7501 * gcc.c: Do not pass LIBUBSAN_SPEC if ASan is enabled with UBSan.
7503 2018-07-04 Eric Botcazou <ebotcazou@adacore.com>
7506 * expmed.c (choose_multiplier): Fix incorrect comparison with mask.
7508 2018-07-04 Aldy Hernandez <aldyh@redhat.com>
7510 * tree-vrp.c (extract_range_from_binary_expr_1): Initialize
7511 neg_*_op* variables.
7513 2018-07-04 Martin Liska <mliska@suse.cz>
7515 * tree-switch-conversion.c: Define
7516 max_ratio_for_speed and max_ratio_for_size constants.
7518 2018-07-04 Denys Vlasenko <dvlasenk@redhat.com>
7519 Martin Liska <mliska@suse.cz>
7524 * common.opt: Rename align options with 'str_' prefix.
7525 * common/config/i386/i386-common.c (set_malign_value): New
7527 (ix86_handle_option): Use it to set -falign-* options/
7528 * config/aarch64/aarch64-protos.h (struct tune_params): Change
7529 type from int to string.
7530 * config/aarch64/aarch64.c: Update default values from int
7532 * config/alpha/alpha.c (alpha_override_options_after_change):
7534 * config/arm/arm.c (arm_override_options_after_change_1): Likewise.
7535 * config/i386/dragonfly.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
7536 max skip conditionally.
7537 * config/i386/freebsd.h (SUBALIGN_LOG): New.
7538 (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
7539 max skip conditionally.
7540 * config/i386/gas.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
7541 max skip conditionally.
7542 * config/i386/gnu-user.h (SUBALIGN_LOG): New.
7543 (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
7544 max skip conditionally.
7545 * config/i386/i386.c (struct ptt): Change type from int to
7547 (ix86_default_align): Set default values.
7548 * config/i386/i386.h (ASM_OUTPUT_MAX_SKIP_PAD): Print
7549 max skip conditionally.
7550 * config/i386/iamcu.h (SUBALIGN_LOG): New.
7551 (ASM_OUTPUT_MAX_SKIP_ALIGN):
7552 * config/i386/lynx.h (ASM_OUTPUT_MAX_SKIP_ALIGN):
7553 * config/i386/netbsd-elf.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
7554 max skip conditionally.
7555 * config/i386/openbsdelf.h (SUBALIGN_LOG): New.
7556 (ASM_OUTPUT_MAX_SKIP_ALIGN) Print max skip conditionally.:
7557 * config/i386/x86-64.h (SUBALIGN_LOG): New.
7558 (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
7559 max skip conditionally.
7560 (ASM_OUTPUT_MAX_SKIP_PAD): Likewise.
7561 * config/ia64/ia64.c (ia64_option_override): Set default values
7562 for alignment options.
7563 * config/m68k/m68k.c: Handle new str_align_* options.
7564 * config/mips/mips.c (mips_set_compression_mode): Change
7566 (mips_option_override): Set default values for options.
7567 * config/powerpcspe/powerpcspe.c (rs6000_option_override_internal):
7569 * config/rs6000/rs6000.c (rs6000_option_override_internal):
7571 * config/rx/rx.c (rx_option_override): Likewise.
7572 * config/rx/rx.h (JUMP_ALIGN): Use align_jumps_log.
7573 (LABEL_ALIGN): Use align_labels_log.
7574 (LOOP_ALIGN): Use align_loops_align.
7575 * config/s390/s390.c (s390_asm_output_function_label): Use new
7577 * config/sh/sh.c (sh_override_options_after_change):
7578 Change type of constants.
7579 * config/spu/spu.c (spu_sched_init): Likewise.
7580 * config/sparc/sparc.c (sparc_option_override): Set default
7582 * config/visium/visium.c (visium_option_override): Likewise.
7583 * config/visium/visium.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Do not
7584 emit p2align format with last argument if it's not needed.
7585 * doc/invoke.texi: Document extended format of -falign-*.
7586 * final.c: Use align_labels alignment.
7587 * flags.h (struct target_flag_state): Change type to use
7589 (struct align_flags_tuple): New.
7590 (struct align_flags): Likewise.
7591 (align_loops_log): Redefine macro to use new types.
7592 (align_loops_max_skip): Redefine macro to use new types.
7593 (align_jumps_log): Redefine macro to use new types.
7594 (align_jumps_max_skip): Redefine macro to use new types.
7595 (align_labels_log): Redefine macro to use new types.
7596 (align_labels_max_skip): Redefine macro to use new types.
7597 (align_functions_log): Redefine macro to use new types.
7598 (align_loops): Redefine macro to use new types.
7599 (align_jumps): Redefine macro to use new types.
7600 (align_labels): Redefine macro to use new types.
7601 (align_functions): Redefine macro to use new types.
7602 (align_functions_max_skip): Redefine macro to use new types.
7603 (align_loops_value): New macro.
7604 (align_jumps_value): New macro.
7605 (align_labels_value): New macro.
7606 (align_functions_value): New macro.
7607 * function.c (invoke_set_current_function_hook): Propagate
7608 alignment values from flags to global variables default in
7610 * ipa-icf.c (sem_function::equals_wpa): Use
7611 cl_optimization_option_eq instead of memcmp.
7612 * lto-streamer.h (cl_optimization_stream_out): Support streaming
7614 (cl_optimization_stream_in): Likewise.
7615 * optc-save-gen.awk: Support strings in cl_optimization.
7616 * opth-gen.awk: Likewise.
7617 * opts.c (finish_options): Remove error checking of invalid
7619 (MAX_CODE_ALIGN): Remove.
7620 (MAX_CODE_ALIGN_VALUE): Likewise.
7621 (parse_and_check_align_values): New function.
7622 (check_alignment_argument): Likewise.
7623 (common_handle_option): Use check_alignment_argument.
7624 * opts.h (parse_and_check_align_values): Declare.
7625 * toplev.c (init_alignments): Remove.
7626 (read_log_maxskip): New.
7627 (parse_N_M): Likewise.
7628 (parse_alignment_opts): Likewise.
7629 (backend_init_target): Remove usage of init_alignments.
7630 * toplev.h (parse_alignment_opts): Declare.
7631 * tree-streamer-in.c (streamer_read_tree_bitfields): Add new
7633 * tree-streamer-out.c (streamer_write_tree_bitfields): Likewise.
7634 * tree.c (cl_option_hasher::equal): New.
7635 * varasm.c: Use new global macros.
7637 2018-07-04 Denys Vlasenko <dvlasenk@redhat.com>
7639 * config/i386/dragonfly.h: (ASM_OUTPUT_MAX_SKIP_ALIGN):
7640 Use a simpler align directive also if MAXSKIP = ALIGN-1.
7641 * config/i386/gas.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
7642 * config/i386/lynx.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
7643 * config/i386/netbsd-elf.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
7644 * config/i386/i386.h (ASM_OUTPUT_MAX_SKIP_PAD): Likewise.
7645 * config/i386/freebsd.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Remove "If N
7646 is large, do at least 8 byte alignment" code. Add SUBALIGN_LOG
7647 define. Use a simpler align directive also if MAXSKIP = ALIGN-1.
7648 * config/i386/gnu-user.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
7649 * config/i386/iamcu.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
7650 * config/i386/openbsdelf.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
7651 * config/i386/x86-64.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
7653 2018-07-04 Martin Liska <mliska@suse.cz>
7654 Jonathan Wakely <jwakely@redhat.com>
7656 * coverage.c: Use correct type.
7657 * doc/invoke.texi: Language correction.
7659 2018-07-03 H.J. Lu <hongjiu.lu@intel.com>
7662 * config/i386/i386.c (rest_of_insert_endbranch): Also generate
7663 ENDBRANCH for non-tail call which may return via indirect branch.
7664 * doc/extend.texi: Document indirect_return attribute.
7666 2018-07-03 Martin Sebor <msebor@redhat.com>
7668 PR tree-optimization/86274
7669 * gimple-ssa-sprintf.c (fmtresult::type_max_digits): Verify
7671 (format_floating): Correct handling of infinities and NaNs.
7673 2018-07-03 Martin Sebor <msebor@redhat.com>
7675 * print-tree.c (print_real_cst): New function.
7676 (print_node_brief): Call it.
7677 (print_node): Ditto.
7679 2018-07-03 Jeff Law <law@redhat.com>
7681 * config/h8300/h8300.md (logical<mode>3_sn, logical<mode>3): Merge
7682 into a single pattern.
7684 * config/h8300/h8300.md (ors code_iterator): New.
7685 (bsetqi_msx, bnotqi_msx patterns and splitters): Consolidate into
7686 a single pattern and single splitter.
7687 (bsethi_msx, bnothi_msx patterns): Consolidate into a single pattern.
7688 (iorqi3_1, xorqi3_1): Likewise.
7689 (iorqi3, xorqi3 expanders): Similarly.
7691 * config/h8300/h8300.md (movmd_internal_normal): Consolidated with
7692 (movmd_internal) into a single pattern using the P mode iterator.
7693 (movmd splitters): Similarly.
7694 (stpcpy_internal_normal, stpcpy_internal): Similarly for thes patterns.
7695 (movsd splitters): Similarly.
7697 * config/h8300/h8300.c (h8300_insn_length_from_table): Consolidate
7698 ADDB, ADDW and ADDL into a single ADD attribute which selects the
7699 right table based on the size of the operand.
7700 * config/h8300/h8300.md (length_table): Corresponding changes. All
7701 references to "addb", "addw" and "addl" changed to "add".
7702 (btst patterns): Merge two variants into a single pattern.
7703 (tstqi, tsthi): Likewise.
7704 (addhi3_incdec, addsi3_incdec): Likewise.
7705 (subhi3_h8300hs, subsi3_h8300hs): Likewise.
7706 (mulhi3, mulsi3): Likewise.
7707 (udivhi3, udivsi3): Likewise.
7708 (divhi3, divsi3): Likewise.
7709 (andorqi3, andorhi3, andorsi3): Likewise.
7711 2018-07-03 Uros Bizjak <ubizjak@gmail.com>
7714 * config/i386/sse.md (uavg<mode>3_ceil): New expander.
7715 (<sse2_avx2>_uavg<mode>3<mask_name>): Simplify expander.
7717 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
7719 PR tree-optimization/85694
7720 * config/aarch64/iterators.md (HADD, RHADD): New int iterators.
7721 (u): Handle UNSPEC_SHADD, UNSPEC_UHADD, UNSPEC_SRHADD and
7723 * config/aarch64/aarch64-simd.md (<u>avg<mode>3_floor)
7724 (<u>avg<mode>3_ceil): New patterns.
7726 2018-07-03 David Malcolm <dmalcolm@redhat.com>
7728 * gcc.dg/vect/slp-perm-1.c: Remove "note: " prefix from
7729 scan-tree-dump directive.
7730 * gcc.dg/vect/slp-perm-2.c: Likewise.
7731 * gcc.dg/vect/slp-perm-3.c: Likewise.
7732 * gcc.dg/vect/slp-perm-5.c: Likewise.
7733 * gcc.dg/vect/slp-perm-6.c: Likewise.
7734 * gcc.dg/vect/slp-perm-7.c: Likewise.
7735 * gcc.dg/vect/slp-perm-8.c: Likewise.
7737 2018-07-03 Marek Polacek <polacek@redhat.com>
7740 * gimple-fold.c (size_must_be_zero_p): Check the type of the size.
7742 2018-07-03 Richard Biener <rguenther@suse.de>
7745 * tree-ssa-structalias.c (find_func_clobbers): Properly
7746 handle indirect calls.
7748 2018-07-03 Jeff Law <law@redhat.com>
7750 * config/h8300/h8300.md (HSI, QHSI, QHSIF): New mode iterators.
7751 (shifts): New code iterator.
7752 (movqi, movhi, movsi, movsf expanders): Consolidate into a single
7753 expander. Fix HImode handling on H8/SX.
7754 (addqi3, addhi3, addsi3 expanders): Consolidate into a single expander.
7755 (subqi3, subhi3, subsi3 expanders): Likewise.
7756 (andqi3, andhi3, andsi3 expanders): Likewise.
7757 (iorqi3, iorhi3, iorsi3 expanders): Likewise.
7758 (xorqi3, xorhi3, xorsi3 expanders): Likewise.
7759 (negqi2, neghi2, negsi2, negsf2 expanders): Likewise.
7760 (one_cmplqi2, one_cmplhi2, one_cmplsi2): Likewise.
7761 (zero_extendqihi2, zero_extendqisi2): Likewise.
7762 (extendqihi2, extendqisi2): Likewise.
7763 (rotlqi3, rotlhi3, rotlsi3): Likewise.
7764 (neghi2_h8300, negsi2_h8300): Likewise for these patterns.
7765 (rotlqi3_1, rotlhi3_1): Likewise.
7766 (logicalhi3_sn, logicalsi3_sn): Likewise.
7767 (logicalhi3, logicalsi3): Likewise.
7769 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
7771 * tree-vect-patterns.c (vect_recog_rotate_pattern)
7772 (vect_recog_vector_vector_shift_pattern, vect_recog_divmod_pattern)
7773 (vect_recog_mixed_size_cond_pattern, adjust_bool_pattern_cast)
7774 (adjust_bool_pattern, vect_recog_bool_pattern): Pass the vector
7775 type to append_pattern_def_seq instead of creating a stmt_vec_info
7777 (build_mask_conversion): Likewise. Remove vinfo argument.
7778 (vect_add_conversion_to_patterm): Likewise, renaming to...
7779 (vect_add_conversion_to_pattern): ...this.
7780 (vect_recog_mask_conversion_pattern): Update call to
7781 build_mask_conversion. Pass the vector type to
7782 append_pattern_def_seq here too.
7783 (vect_recog_gather_scatter_pattern): Update call to
7784 vect_add_conversion_to_pattern.
7786 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
7788 * tree-vect-patterns.c (new_pattern_def_seq): Delete.
7789 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern)
7790 (vect_recog_widen_op_pattern, vect_recog_over_widening_pattern)
7791 (vect_recog_rotate_pattern, vect_synth_mult_by_constant): Don't set
7792 STMT_VINFO_PATTERN_DEF_SEQ to null here.
7793 (vect_recog_pow_pattern, vect_recog_vector_vector_shift_pattern)
7794 (vect_recog_mixed_size_cond_pattern, vect_recog_bool_pattern): Use
7795 append_pattern_def_seq instead of new_pattern_def_seq.
7796 (vect_recog_divmod_pattern): Do both of the above.
7797 (vect_pattern_recog_1): Assert that STMT_VINO_PATTERN_DEF_SEQ
7800 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
7802 * tree-vect-patterns.c (vect_recog_dot_prod_pattern):
7803 (vect_recog_sad_pattern, vect_recog_widen_op_pattern)
7804 (vect_recog_widen_mult_pattern, vect_recog_pow_pattern):
7805 (vect_recog_widen_sum_pattern, vect_recog_over_widening_pattern)
7806 (vect_recog_average_pattern, vect_recog_cast_forwprop_pattern)
7807 (vect_recog_widen_shift_pattern, vect_recog_rotate_pattern)
7808 (vect_recog_vector_vector_shift_pattern, vect_synth_mult_by_constant)
7809 (vect_recog_mult_pattern, vect_recog_divmod_pattern)
7810 (vect_recog_mixed_size_cond_pattern, vect_recog_bool_pattern)
7811 (vect_recog_mask_conversion_pattern): Replace vec<gimple *>
7812 parameter with a single stmt_vec_info.
7813 (vect_recog_func_ptr): Likewise.
7814 (vect_recog_gather_scatter_pattern): Likewise, folding in...
7815 (vect_try_gather_scatter_pattern): ...this.
7816 (vect_pattern_recog_1): Remove stmts_to_replace and just pass
7817 the stmt_vec_info of the statement to be matched. Don't clear
7818 STMT_VINFO_RELATED_STMT.
7819 (vect_pattern_recog): Update call accordingly.
7821 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
7823 PR tree-optimization/85694
7824 * doc/md.texi (avgM3_floor, uavgM3_floor, avgM3_ceil)
7825 (uavgM3_ceil): Document new optabs.
7826 * doc/sourcebuild.texi (vect_avg_qi): Document new target selector.
7827 * internal-fn.def (IFN_AVG_FLOOR, IFN_AVG_CEIL): New internal
7829 * optabs.def (savg_floor_optab, uavg_floor_optab, savg_ceil_optab)
7830 (savg_ceil_optab): New optabs.
7831 * tree-vect-patterns.c (vect_recog_average_pattern): New function.
7832 (vect_vect_recog_func_ptrs): Add it.
7833 * tree-vect-stmts.c (vectorizable_call): Get the type of the zero
7834 constant directly from the associated lhs.
7836 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
7838 * tree-vect-patterns.c (vect_split_statement): New function.
7839 (vect_convert_input): Use it to try to split an existing cast.
7841 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
7843 * poly-int.h (print_hex): New function.
7844 * dumpfile.h (dump_dec, dump_hex): Declare.
7845 * dumpfile.c (dump_dec, dump_hex): New poly_wide_int functions.
7846 * tree-vectorizer.h (_stmt_vec_info): Add min_output_precision,
7847 min_input_precision, operation_precision and operation_sign.
7848 * tree-vect-patterns.c (vect_get_range_info): New function.
7849 (vect_same_loop_or_bb_p, vect_single_imm_use)
7850 (vect_operation_fits_smaller_type): Delete.
7851 (vect_look_through_possible_promotion): Add an optional
7852 single_use_p parameter.
7853 (vect_recog_over_widening_pattern): Rewrite to use new
7854 stmt_vec_info infomration. Handle one operation at a time.
7855 (vect_recog_cast_forwprop_pattern, vect_narrowable_type_p)
7856 (vect_truncatable_operation_p, vect_set_operation_type)
7857 (vect_set_min_input_precision): New functions.
7858 (vect_determine_min_output_precision_1): Likewise.
7859 (vect_determine_min_output_precision): Likewise.
7860 (vect_determine_precisions_from_range): Likewise.
7861 (vect_determine_precisions_from_users): Likewise.
7862 (vect_determine_stmt_precisions, vect_determine_precisions): Likewise.
7863 (vect_vect_recog_func_ptrs): Put over_widening first.
7865 (vect_pattern_recog): Call vect_determine_precisions.
7867 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
7869 * tree-vect-patterns.c (vect_mark_pattern_stmts): Remove pattern
7870 statements that have been replaced by further pattern statements.
7871 (vect_pattern_recog_1): Clear STMT_VINFO_PATTERN_DEF_SEQ on failure.
7873 2018-07-03 Richard Biener <rguenther@suse.de>
7875 * tree-vect-stmts.c (vect_is_simple_use): Consolidate dumping,
7876 always set *dt. Dump vectype in vectype overload.
7877 * dumpfile.h (dump_gimple_expr): New function.
7878 (dump_gimple_expr_loc): Likewise.
7879 * dumpfile.c (dump_gimple_expr): New function.
7880 (dump_gimple_expr_loc): Likewise.
7882 2018-07-02 Jeff Law <law@redhat.com>
7884 * config/h8300/h8300.md (movqi_h8300, movqi_h8300hs): Consolidate
7885 the H8/300, H8/300H and H8/S variants into a single pattern.
7886 (movhi_h8300, movqi_h8300hs): Similarly.
7887 (pushqi_h8300hs, pushhi_h8300hs): Consolidate into a single pattern.
7888 (QHI mode iterator): New.
7890 * config/h8300/h8300.md: Remove trailing whitespace.
7892 2018-07-02 Jim Wilson <jimw@sifive.com>
7894 * config/riscv/riscv.c (riscv_expand_epilogue): Use emit_jump_insn
7895 instead of emit_insn for interrupt returns.
7896 * config/riscv/riscv.md (riscv_met): Add (return) to rtl.
7897 (riscv_sret, riscv_uret): Likewise.
7899 2018-07-02 David Malcolm <dmalcolm@redhat.com>
7901 * pretty-print.c (selftest::test_pp_format): Move save and restore
7902 of quotes to class auto_fix_quotes, and add an instance.
7903 * selftest.c: Include "intl.h".
7904 (selftest::auto_fix_quotes::auto_fix_quotes): New ctor.
7905 (selftest::auto_fix_quotes::~auto_fix_quotes): New dtor.
7906 * selftest.h (selftest::auto_fix_quotes): New class.
7908 2018-07-02 Richard Henderson <richard.henderson@linaro.org>
7910 * config/aarch64/aarch64-protos.h, config/aarch64/aarch64.c
7911 (aarch64_sve_prepare_conditional_op): Remove.
7912 * config/aarch64/aarch64-sve.md (cond_<SVE_INT_BINARY><SVE_I>):
7913 Allow aarch64_simd_reg_or_zero as select operand; remove
7914 the aarch64_sve_prepare_conditional_op call.
7915 (cond_<SVE_INT_BINARY_SD><SVE_SDI>): Likewise.
7916 (cond_<SVE_COND_FP_BINARY><SVE_F>): Likewise.
7917 (*cond_<SVE_INT_BINARY><SVE_I>_z): New pattern.
7918 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_z): New pattern.
7919 (*cond_<SVE_COND_FP_BINARY><SVE_F>_z): New pattern.
7920 (*cond_<SVE_INT_BINARY><SVE_I>_any): New pattern.
7921 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_any): New pattern.
7922 (*cond_<SVE_COND_FP_BINARY><SVE_F>_any): New pattern
7923 and a splitters to match all of the *_any patterns.
7924 * config/aarch64/predicates.md (aarch64_sve_any_binary_operator): New.
7926 * config/aarch64/iterators.md (SVE_INT_BINARY_REV): Remove.
7927 (SVE_COND_FP_BINARY_REV): Remove.
7928 (sve_int_op_rev, sve_fp_op_rev): New.
7929 * config/aarch64/aarch64-sve.md (*cond_<SVE_INT_BINARY><SVE_I>_0): New.
7930 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_0): New.
7931 (*cond_<SVE_COND_FP_BINARY><SVE_F>_0): New.
7932 (*cond_<SVE_INT_BINARY><SVE_I>_2): Rename, add movprfx alternative.
7933 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_2): Similarly.
7934 (*cond_<SVE_COND_FP_BINARY><SVE_F>_2): Similarly.
7935 (*cond_<SVE_INT_BINARY><SVE_I>_3): Similarly; use sve_int_op_rev.
7936 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_3): Similarly.
7937 (*cond_<SVE_COND_FP_BINARY><SVE_F>_3): Similarly; use sve_fp_op_rev.
7939 * config/aarch64/aarch64-sve.md (cond_<SVE_COND_FP_BINARY><SVE_F>):
7940 Remove match_dup 1 from the inner unspec.
7941 (*cond_<SVE_COND_FP_BINARY><SVE_F>): Likewise.
7943 * config/aarch64/aarch64.md (movprfx): New attr.
7944 (length): Default movprfx to 8.
7945 * config/aarch64/aarch64-sve.md (*mul<SVE_I>3): Add movprfx alt.
7946 (*madd<SVE_I>, *msub<SVE_I): Likewise.
7947 (*<su>mul<SVE_I>3_highpart): Likewise.
7948 (*<SVE_INT_BINARY_SD><SVE_SDI>3): Likewise.
7949 (*v<ASHIFT><SVE_I>3): Likewise.
7950 (*<su><MAXMIN><SVE_I>3): Likewise.
7951 (*<su><MAXMIN><SVE_F>3): Likewise.
7952 (*fma<SVE_F>4, *fnma<SVE_F>4): Likewise.
7953 (*fms<SVE_F>4, *fnms<SVE_F>4): Likewise.
7954 (*div<SVE_F>4): Likewise.
7956 2018-07-02 Richard Sandiford <richard.sandiford@arm.com>
7958 * tree-vect-patterns.c (vect_recog_widen_shift_pattern): Fix typo
7961 2018-07-02 Richard Biener <rguenther@suse.de>
7963 PR tree-optimization/86363
7964 * tree-ssa-sccvn.c (vn_reference_lookup_3): Check the
7965 memset argument refers to a non-variable address.
7967 2018-07-02 Aldy Hernandez <aldyh@redhat.com>
7969 * tree-vrp.c (extract_range_from_binary_expr_1): Abstract a lot of the
7970 {PLUS,MINUS}_EXPR code to...
7971 (adjust_symbolic_bound): ...here,
7972 (combine_bound): ...here,
7973 (set_value_range_with_overflow): ...and here.
7975 2018-07-02 Aldy Hernandez <aldyh@redhat.com>
7977 * tree-vrp.c (extract_range_from_unary_expr): Abstract ABS_EXPR
7979 (extract_range_from_abs_expr): ...here.
7981 2018-07-02 Eric Botcazou <ebotcazou@adacore.com>
7983 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Do not overrule
7984 -fno-omit-frame-pointer when not optimizing.
7986 2018-07-02 Martin Liska <mliska@suse.cz>
7989 * ipa-pure-const.c (malloc_candidate_p): Revert usage of ::get.
7990 (propagate_nothrow): Likewise.
7992 2018-07-02 Martin Liska <mliska@suse.cz>
7995 * ipa-inline.c (early_inliner): Revert wrongly added ::get call.
7997 2018-07-02 David Malcolm <dmalcolm@redhat.com>
7999 * dumpfile.c (dump_generic_expr_loc): Undo removal of this
8000 function in r262149, changing "loc" param from source_location to
8001 const dump_location_t &.
8002 * dumpfile.h (dump_generic_expr_loc): Undo removal of this
8003 declaration, as above.
8005 2018-07-01 Paul Koning <ni1d@arrl.net>
8007 * common/config/pdp11/pdp11-common.c (pdp11_handle_option): Handle
8008 -munit-asm, -mgnu-asm, -mdec-asm.
8009 * config/pdp11/pdp11-protos.h (pdp11_gen_int_label): New.
8010 (pdp11_output_labelref): New.
8011 (pdp11_output_def): New.
8012 (pdp11_output_addr_vec_elt): New.
8013 * config/pdp11/pdp11.c: Use tab between opcode and operands. Use
8014 %# and %@ format codes.
8015 (pdp11_option_override): New.
8016 (TARGET_ASM_FILE_START_FILE_DIRECTIVE): Define.
8017 (pdp11_output_ident): New.
8018 (pdp11_asm_named_section): New.
8019 (pdp11_asm_init_sections): New.
8020 (pdp11_file_start): New.
8021 (pdp11_file_end): New.
8022 (output_ascii): Use .ascii/.asciz for -mdec-asm.
8023 (pdp11_asm_print_operand): Update %# and %$ for -mdec-asm. Add
8024 %o, like %c but octal.
8025 (pdp11_option_override): New.
8026 * config/pdp11/pdp11.h (TEXT_SECTION_ASM_OP): Update for
8028 (DATA_SECTION_ASM_OP): Ditto.
8029 (READONLY_DATA_SECTION_ASM_OP): New.
8030 (IS_ASM_LOGICAL_LINE_SEPARATOR): New.
8031 (ASM_GENERATE_INTERNAL_LABEL): Use new function.
8032 (ASM_OUTPUT_LABELREF): Ditto.
8033 (ASM_OUTPUT_DEF): Ditto.
8034 (ASM_OUTPUT_EXTERNAL): New.
8035 (ASM_OUTPUT_SOURCE_FILENAME): New.
8036 (ASM_OUTPUT_ADDR_VEC_ELT): Use new function.
8037 (ASM_OUTPUT_SKIP): Update for -mdec-asm.
8038 * config/pdp11/pdp11.md: Use tab between opcode and operands. Use
8039 %# and %@ format codes.
8040 * config/pdp11/pdp11.opt (mgnu-asm): New.
8041 (mdec-asm): Conflicts with -mgnu-asm and -munix-asm.
8042 (munix-asm): Conflicts with -mdec-asm and -mgnu-asm.
8043 * doc/invoke.txt (PDP-11 Options): Add -mgnu-asm.
8045 2018-07-01 Aldy Hernandez <aldyh@redhat.com>
8047 * tree-ssa-threadupdate.c (mark_threaded_blocks): Avoid
8048 dereferencing path[] beyond its length.
8050 (debug_all_paths): New.
8051 (rewire_first_differing_edge): New.
8052 (adjust_paths_after_duplication): New.
8053 (duplicate_thread_path): Call adjust_paths_after_duplication.
8055 (thread_through_all_blocks): Add new argument to
8056 duplicate_thread_path.
8058 2018-06-30 Jim Wilson <jimw@sifive.com>
8060 * config/riscv/predicates.md (p2m1_shift_operand): New.
8061 (high_mask_shift_operand): New.
8062 * config/riscv/riscv.md (lshrsi3_zero_extend_3+1): New combiner
8063 pattern using p2m1_shift_operand.
8064 (lshsi3_zero_extend_3+2): New combiner pattern using
8065 high_mask_shift_operand.
8067 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
8069 * tree-vect-patterns.c (vect_get_external_def_edge): New function,
8071 (vect_recog_rotate_pattern): ...here.
8072 (vect_convert_input): Try to insert casts of invariants in the
8074 * tree-vect-loop-manip.c (vect_loop_versioning): Don't require the
8075 preheader to be empty.
8077 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
8079 * tree-vect-patterns.c (append_pattern_def_seq): Take an optional
8080 vector type. If given, install it in the new statement's
8082 (vect_element_precision): New function.
8083 (vect_unpromoted_value): New struct.
8084 (vect_unpromoted_value::vect_unpromoted_value): New function.
8085 (vect_unpromoted_value::set_op): Likewise.
8086 (vect_look_through_possible_promotion): Likewise.
8087 (vect_joust_widened_integer, vect_joust_widened_type): Likewise.
8088 (vect_widened_op_tree, vect_convert_input): Likewise.
8089 (vect_convert_inputs, vect_convert_output): Likewise.
8090 (vect_recog_dot_prod_pattern): Use vect_look_through_possible_promotion
8091 to handle the optional cast of the multiplication result and
8092 vect_widened_op_tree to detect the widened multiplication itself.
8093 Do not require the input and output of promotion casts to have
8094 the same sign, but base the signedness of the operation on the
8095 input rather than the result. If the pattern includes two
8096 promotions, check that those promotions have the same sign.
8097 Do not restrict the MULT_EXPR handling to a double-width result;
8098 handle quadruple-width results and wider. Use vect_convert_inputs
8099 to convert the inputs to the common type.
8100 (vect_recog_sad_pattern): Use vect_look_through_possible_promotion
8101 to handle the optional cast of the ABS result. Also allow a sign
8102 change or a sign extension between the ABS and MINUS.
8103 Use vect_widened_op_tree to detect the widened subtraction and use
8104 vect_convert_inputs to convert the inputs to the common type.
8105 (vect_handle_widen_op_by_const): Delete.
8106 (vect_recog_widen_op_pattern): New function.
8107 (vect_recog_widen_mult_pattern): Use it.
8108 (vect_recog_widen_shift_pattern): Likewise.
8109 (vect_recog_widen_sum_pattern): Use
8110 vect_look_through_possible_promotion to handle the promoted
8113 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
8115 * gimple-iterator.c (gsi_for_stmt): Add a new overload that takes
8116 the containing gimple_seq *.
8117 * gimple-iterator.h (gsi_for_stmt): Declare it.
8118 * tree-vect-patterns.c (vect_recog_dot_prod_pattern)
8119 (vect_recog_sad_pattern, vect_recog_widen_sum_pattern)
8120 (vect_recog_widen_shift_pattern, vect_recog_rotate_pattern)
8121 (vect_recog_vector_vector_shift_pattern, vect_recog_divmod_pattern)
8122 (vect_recog_mask_conversion_pattern): Remove STMT_VINFO_IN_PATTERN_P
8124 (vect_init_pattern_stmt, vect_set_pattern_stmt): New functions,
8126 (vect_mark_pattern_stmts): ...here. Handle cases in which the
8127 statement being replaced is part of an existing pattern
8128 definition sequence, inserting the new pattern statements before
8130 (vect_pattern_recog_1): Don't return a bool. If the statement
8131 is already part of a pattern, instead apply pattern matching
8132 to the pattern definition statements. Don't clear the
8133 STMT_VINFO_RELATED_STMT if is_pattern_stmt_p.
8134 (vect_pattern_recog): Don't break after the first match;
8135 continue processing the pattern definition statements instead.
8136 Don't bail out for STMT_VINFO_IN_PATTERN_P here.
8138 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
8140 * tree-vect-patterns.c (vect_reassociating_reduction_p): New function.
8141 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern)
8142 (vect_recog_widen_sum_pattern): Use it.
8144 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
8146 * tree-vect-loop.c (vectorizable_reduction): Assert that the
8147 phi is not a pattern statement and has not been replaced by
8148 a pattern statement.
8149 * tree-vect-patterns.c (type_conversion_p): Don't check
8150 STMT_VINFO_IN_PATTERN_P.
8151 (vect_recog_vector_vector_shift_pattern): Likewise.
8152 (vect_recog_dot_prod_pattern): Expect vect_is_simple_use to return
8153 the pattern statement rather than the original statement; check
8154 directly for a WIDEN_MULT_EXPR here.
8155 * tree-vect-slp.c (vect_get_and_check_slp_defs): Expect
8156 vect_is_simple_use to return the pattern statement rather
8157 than the original statement; use is_pattern_stmt_p to check
8158 for such a pattern statement.
8159 * tree-vect-stmts.c (process_use): Expect vect_is_simple_use
8160 to return the pattern statement rather than the original statement;
8161 don't do the same transformation here.
8162 (vect_is_simple_use): If the defining statement has been replaced
8163 by a pattern statement, return the pattern statement instead.
8164 Remove the corresponding (local) transformation from the vectype
8167 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
8169 * tree-vectorizer.h (vect_is_simple_use): Move the gimple ** to the
8170 end and default to null.
8171 * tree-vect-loop.c (vect_create_epilog_for_reduction)
8172 (vectorizable_reduction): Update calls accordingly, dropping the
8173 gimple ** argument if the passed-back statement isn't needed.
8174 * tree-vect-patterns.c (vect_get_internal_def, type_conversion_p)
8175 (vect_recog_rotate_pattern): Likewise.
8176 (vect_recog_mask_conversion_pattern): Likewise.
8177 * tree-vect-slp.c (vect_get_and_check_slp_defs): Likewise.
8178 (vect_mask_constant_operand_p): Likewise.
8179 * tree-vect-stmts.c (is_simple_and_all_uses_invariant, process_use):
8180 (vect_model_simple_cost, vect_get_vec_def_for_operand): Likewise.
8181 (get_group_load_store_type, get_load_store_type): Likewise.
8182 (vect_check_load_store_mask, vect_check_store_rhs): Likewise.
8183 (vectorizable_call, vectorizable_simd_clone_call): Likewise.
8184 (vectorizable_conversion, vectorizable_assignment): Likewise.
8185 (vectorizable_shift, vectorizable_operation): Likewise.
8186 (vectorizable_store, vect_is_simple_cond): Likewise.
8187 (vectorizable_condition, vectorizable_comparison): Likewise.
8188 (get_same_sized_vectype, vect_get_mask_type_for_stmt): Likewise.
8189 (vect_is_simple_use): Rename the def_stmt argument to def_stmt_out
8190 and move it to the end. Cope with null def_stmt_outs.
8192 2018-06-30 Bernd Edlinger <bernd.edlinger@hotmail.de>
8194 * Makefile.in (FLAGS_TO_PASS): Add RANLIB_FOR_TARGET.
8196 2018-06-29 Jeff Law <law@redhat.com>
8198 * config/v850/v850.c (v850_legitimate_address_p): Handle large
8199 displacements for TARGET_V850E2V3 and newer.
8200 (TARGET_LRA_P): Remove. Defaults to LRA now.
8201 * config/v850/v850.md (sign23byte_load): Remove.
8202 (unsign23byte_load, sign23hword_load, unsign23hword_load): Likewise.
8203 (23word_load, 23byte_store, 23hword_store, 23word_store): Likewise.
8205 2018-06-29 Martin Liska <mliska@suse.cz>
8208 * coverage.c (coverage_init): Mangle full path name.
8209 * doc/invoke.texi: Document the change.
8210 * gcov-io.c (mangle_path): New.
8211 * gcov-io.h (mangle_path): Likewise.
8212 * gcov.c (mangle_name): Use mangle_path for path mangling.
8214 2018-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8216 * config/arm/arm.c (output_move_double): Don't allow STRD instructions
8217 if starting source register is not even.
8219 2018-06-29 Martin Liska <mliska@suse.cz>
8221 PR tree-optimization/86263
8222 * tree-switch-conversion.c (switch_decision_tree::try_switch_expansion):
8223 Make edge redirection.
8225 2018-06-29 David Malcolm <dmalcolm@redhat.com>
8227 * dumpfile.c (dump_loc): Add indentation based on scope depth.
8228 (dump_scope_depth): New variable.
8229 (get_dump_scope_depth): New function.
8230 (dump_begin_scope): New function.
8231 (dump_end_scope): New function.
8232 * dumpfile.h (get_dump_scope_depth): New declaration.
8233 (dump_begin_scope): New declaration.
8234 (dump_end_scope): New declaration.
8235 (class auto_dump_scope): New class.
8236 (AUTO_DUMP_SCOPE): New macro.
8237 * tree-vectorizer.h (DUMP_VECT_SCOPE): Reimplement in terms of
8240 2018-06-29 Richard Biener <rguenther@suse.de>
8242 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Assert
8243 compute_all_dependences succeeds.
8244 * tree-vect-loop.c (vect_get_datarefs_in_loop): Fail early if we
8245 exceed --param loop-max-datarefs-for-datadeps.
8247 2018-06-29 Jakub Jelinek <jakub@redhat.com>
8249 * config/rs6000/t-rs6000: Append rs6000-modes.h to TM_H.
8251 2018-06-28 Uros Bizjak <ubizjak@gmail.com>
8254 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): Use
8255 alternative 0 in preferred_for_speed attribute.
8257 2018-06-28 Paul Koning <ni1d@arrl.net>
8259 * config/pdp11/pdp11-protos.h (pdp11_shift_length): New function.
8260 * config/pdp11/pdp11.c (pdp11_shift_length): New function.
8261 * config/pdp11/pdp11.h (ADJUST_INSN_LENGTH): Remove.
8262 * config/pdp11/pdp11.md: Correct "length" attribute calculation
8263 for shift insn patterns.
8265 2018-06-28 David Malcolm <dmalcolm@redhat.com>
8267 * cgraph.c (cgraph_node::get_body): Replace assignments to
8268 "dump_file" with calls to set_dump_file.
8269 * dumpfile.c (alt_dump_file): Make static, and group with...
8270 (alt_flags): ...this definition.
8271 (dumps_are_enabled): New variable.
8272 (refresh_dumps_are_enabled): New function.
8273 (set_dump_file): New function.
8274 (set_alt_dump_file): New function.
8275 (gcc::dump_manager::dump_start): Replace assignments to
8276 "dump_file" and "alt_dump_file" with calls to set_dump_file and
8278 (gcc::dump_manager::dump_finish): Likewise.
8279 * dumpfile.h (alt_dump_file): Delete decl.
8280 (dumps_are_enabled): New variable decl.
8281 (set_dump_file): New function decl.
8282 (dump_enabled_p): Rewrite in terms of new "dumps_are_enabled"
8284 * tree-nested.c (lower_nested_functions): Replace assignments to
8285 "dump_file" with calls to set_dump_file.
8287 2018-06-28 Eric Botcazou <ebotcazou@adacore.com>
8289 * tree-cfg.c (verify_gimple_in_cfg): Call verify_location on the
8290 goto_locus of each outgoing edge of each basic block.
8292 2018-06-28 Richard Biener <rguenther@suse.de>
8294 * dwarf2out.c (decl_scope_table): Remove.
8295 (push_decl_scope): Likewise.
8296 (pop_decl_scope): Likewise.
8297 (gen_type_die_for_member): Do not call push/pop_decl_scope.
8298 (gen_struct_or_union_type_die): Likewise.
8299 (gen_tagged_type_die): Likewise.
8300 (dwarf2out_init): Do not initialize decl_scope_table.
8301 (dwarf2out_c_finalize): Do not free it.
8303 2018-06-28 Richard Biener <rguenther@suse.de>
8305 * dwarf2out.c (gen_subprogram_die): Use is_unit_die when
8306 deciding whether to not re-use a DIE.
8308 2018-06-28 Richard Biener <rguenther@suse.de>
8310 * dwarf2out.c (gen_subprogram_die): Always re-use DIEs with an
8311 DW_AT_abstract_origin attribute.
8313 2018-06-28 Martin Liska <mliska@suse.cz>
8315 * tree-switch-conversion.c (jump_table_cluster::can_be_handled):
8316 Use newly introduced constants.
8317 * tree-switch-conversion.h (struct jump_table_cluster):
8318 Define max_ratio_for_size and max_ratio_for_speed.
8320 2018-06-28 Martin Liska <mliska@suse.cz>
8322 * tree-switch-conversion.c (jump_table_cluster::find_jump_tables):
8323 Add new checking assert to catch invalid state.
8324 (jump_table_cluster::can_be_handled): Handle single case
8326 (jump_table_cluster::is_beneficial): Bail out for such case.
8327 (bit_test_cluster::find_bit_tests):
8328 Add new checking assert to catch invalid state.
8329 (bit_test_cluster::can_be_handled): Handle single case
8331 (bit_test_cluster::is_beneficial): Bail out for such case.
8332 (switch_decision_tree::analyze_switch_statement):
8335 2018-06-28 Martin Liska <mliska@suse.cz>
8337 * common.opt: Introduce -completion option.
8338 * gcc.c (driver_handle_option): Handle it.
8339 (driver::main): Print completions if completion
8341 * opt-suggestions.c (option_proposer::get_completions):
8343 (option_proposer::suggest_completion): Likewise.
8344 (option_proposer::find_param_completions): Likewise.
8345 (verify_autocompletions): Likewise.
8346 (test_completion_valid_options): Likewise.
8347 (test_completion_valid_params): Likewise.
8348 (in_completion_p): Likewise.
8349 (empty_completion_p): Likewise.
8350 (test_completion_partial_match): Likewise.
8351 (test_completion_garbage): Likewise.
8352 (opt_proposer_c_tests): Likewise.
8353 * opt-suggestions.h: Declare new functions.
8354 * opts.c (common_handle_option): Handle OPT__completion_.
8355 * selftest-run-tests.c (selftest::run_tests): Add
8356 opt_proposer_c_tests.
8357 * selftest.c (assert_str_startswith): New.
8358 * selftest.h (assert_str_startswith): Likewise.
8359 (opt_proposer_c_tests): New.
8360 (ASSERT_STR_STARTSWITH): Likewise.
8362 2018-06-28 Martin Liska <mliska@suse.cz>
8364 * Makefile.in: Add opt-suggestions.o.
8365 * gcc-main.c: Include opt-suggestions.h.
8366 * gcc.c (driver::driver): Likewise.
8367 (driver::~driver): Remove m_option_suggestions.
8368 (driver::build_option_suggestions): Moved to option_proposer.
8369 (driver::suggest_option): Likewise.
8370 (driver::handle_unrecognized_options): Use option_proposer.
8371 * gcc.h (class driver): Add new memver m_option_proposer.
8372 * opt-suggestions.c: New file.
8373 * opt-suggestions.h: New file.
8375 2018-06-28 Martin Liska <mliska@suse.cz>
8377 * vec.h (class auto_string_vec): New (moved from auto_argvec).
8378 (auto_string_vec::~auto_string_vec): Likewise.
8380 2018-06-28 Eric Botcazou <ebotcazou@adacore.com>
8382 * tree-inline.h (struct copy_body_data): Move remapping_type_depth and
8383 prevent_decl_creation_for_types fields up and add reset_location field.
8384 * tree-inline.c (remap_gimple_stmt): Force input_location on the new
8385 statement if id->reset_location is true.
8386 (copy_edges_for_bb): Do not set goto_locus on the new edges if
8387 id->reset_location is true.
8388 (copy_phis_for_bb): Force input_location on the arguments if
8389 id->reset_location is true.
8390 (expand_call_inline): Set id->reset_location if DECL_IGNORED_P
8391 is set on the function to be inlined.
8393 2018-06-27 Stephan Bergmann <sbergman@redhat.com>
8395 * doc/invoke.texi (Debugging Options): Add -gsplit-dwarf.
8397 2018-06-27 Dimitar Dimitrov <dimitar@dinux.eu>
8399 * lra-eliminations.c (update_reg_eliminate): Mark all spanning hard
8400 registers for Pmode.
8401 * lra-lives.c (check_pseudos_live_through_calls): Mark all spanning
8402 hard registers for the clobbered pseudo.
8404 2018-06-27 Paul Koning <ni1d@arrl.net>
8406 * common/config/pdp11/pdp11-common.c (pdp11_handle_option): Handle
8407 mutually exclusive options.
8408 * config/pdp11/constraints.md (h): New constraint.
8409 (O): Update definition to match shift code generation.
8410 (D): New constraint.
8411 * config/pdp11/pdp11-modes.def (CCNZ): Define mode.
8413 * config/pdp11/pdp11-protos.h (int_no_side_effect_operand): New
8415 (output_jump): Change arguments.
8416 (pdp11_fixed_cc_regs): New function.
8417 (pdp11_cc_mode): Ditto.
8418 (pdp11_expand_shift): Ditto.
8419 (pdp11_assemble_shift): Ditto.
8420 (pdp11_small_shift): Ditto.
8421 (pdp11_branch_cost): Remove.
8422 * config/pdp11/pdp11.c (pdp11_assemble_integer): Remove comments
8424 (pdp11_register_move_cost): Update for CC registers.
8425 (pdp11_rtx_costs): Add case for LSHIFTRT.
8426 (pdp11_output_jump): Add CCNZ mode conditional branches.
8427 (notice_update_cc_on_set): Remove.
8428 (pdp11_cc_mode): New function.
8429 (simple_memory_operand): Correct pre/post decrement case.
8430 (no_side_effect_operand): New function.
8431 (pdp11_regno_reg_class): Add CC_REGS class.
8432 (pdp11_fixed_cc_regs): New function.
8433 (pdp11_small_shift): New function.
8434 (pdp11_expand_shift): New function to expand shift insns.
8435 (pdp11_assemble_shift): New function to output shifts.
8436 (pdp11_branch_cost): Remove.
8437 (pdp11_modes_tieable_p): Make QI/HI modes tieable.
8438 * config/pdp11/pdp11.h (SIZE_TYPE): Ensure 16-bit type.
8439 (WCHAR_TYPE): Ditto.
8440 (PTRDIFF_TYPE): Ditto.
8441 (ADJUST_INSN_LENGTH): New macro.
8442 (FIXED_REGISTERS): Add CC registers.
8443 (CALL_USED_REGISTERS): Ditto.
8445 (REG_CLASS_NAMES): Ditto.
8446 (REG_CLASS_CONTENTS): Ditto.
8447 (SELECT_CC_MODE): Use new function.
8448 (TARGET_FLAGS_REGNUM): New macro.
8449 (TARGET_FIXED_CONDITION_CODE_REGS): Ditto.
8450 (cc0_reg_rtx): Remove.
8451 (CC_STATUS_MDEP): Remove.
8452 (CC_STATUS_MDEFP_INIT): Remove.
8453 (CC_IN_FPU): Remove.
8454 (NOTICE_UPDATE_CC): Remove.
8455 (REGISTER_NAMES): Add CC registers.
8456 (BRANCH_COST): Change to constant 1.
8457 * config/pdp11/pdp11.md: Rewrite for CCmode condition code
8459 * config/pdp11/pdp11.opt (mbcopy): Remove.
8460 (mbcopy-builtin): Remove.
8461 (mbranch-cheap): Remove.
8462 (mbranch-expensive): Remove.
8463 * config/pdp11/predicates.md (expand_shift_operand): Update to
8464 match shift code generation.
8465 (ccnz_operator): New predicate.
8466 * doc/invoke.texi (PDP-11 Options): Remove deleted options
8467 -mbcopy, -mbcopy-builtin, -mbranch-cheap, -mbranch-expensive.
8468 Remove non-existent option -mabshi, -mno-abshi. Document mutually
8470 * doc/md.texi (PDP-11): Document new D and h constraints. Update
8471 description of O constraint.
8473 2018-06-27 Jeff Law <law@redhat.com>
8474 Austin Law <austinklaw@gmail.com>
8476 * config/v850/v850.md (addsi3_set_flags): New pattern.
8477 (subsi3_set_flags, negsi2_set_flags, andsi3_set_flags): Likewise.
8478 (iorsi3_set_flags, xorsi3_set_flags, one_cmplsi2_set_flags): Likewise.
8479 (zero_extendhisi2_v850_set_flags): Likewise.
8480 (zero_extendqisi2_v850_set_flags): Likewise.
8481 (ashlsi3_set_flags, ashlsi3_v850e2_set_flags): Likewise.
8482 (lshrsi3_set_flags, lshrsi3_v850e2_set_flags): Likewise.
8483 (ashrsi3_set_flags, ashrsi3_v850e2_set_flags): Likewise.
8485 * config/v850/v850-protos.h (notice_update_cc): Remove.
8486 * config/v850/v850.c (v850_compare_op0, v850_compare_op1): Remove.
8487 (v850_print_operand): Handle 'D' and "d".
8488 (v850_select_cc_mode): Remove ATTRIBUTE_UNUSED for last argument.
8489 Add handling of arithmetic/logical operations compared against zero.
8490 (v850_gen_float_compare): Remove ATTRIBUTE_UNUSED for last argument.
8491 Do not look at v850_compare_op, instead get mode from last argument.
8492 (v850_gen_compare): Remove
8493 (increment_stack): Use addsi3_clobber_flags to avoid splitting failure
8494 after reload for prologue insns.
8495 (expand_prologue): Account for CLOBBER of CC_REGNUM in various
8497 (construct_save_jarl): Likewise.
8498 (TARGET_FLAGS_REGNUM): Define.
8499 * config/v850/v850.h (v850_compare_op0, v850_compare_op1): Remove.
8500 (NOTICE_UPDATE_CC): Remove.
8501 * config/v850/v850.md (v850_tst1): Use (reg:CCZ CC_REGNUM) rather
8502 than cc0. Conditionalize on reload_completed.
8503 (cmpsi_insn, setfcc_insn): Likewise.
8504 (tst1 splitter): Turn into define_and_split which sets the flags
8506 (cstoresi4, cbranchsf4, cbranchdf4, cbranchsi4_insn): Likewise.
8507 (cbranchsi4, branch_normal, branch_invert): Do not expose cc0 here.
8508 (cstoresf4, cstoredf4): Clobber the flags.
8509 (cmpsi, cmpsf, cmpdf): Remove expanders.
8510 (setf_insn): Remove pattern.
8511 (addsi3): Turn into define_and_split which clobbers the flags after
8512 reload and a suitable pattern (addsi3_clobber_flags) for use after
8514 (subsi3, negsi2, andsi3, iorsi3, xorsi3, one_cmplsi2) Likewise.
8515 (ashlsi3, ashlsi3_v850e2, lshrsi3, lsh4si3_v850e2): Likewise.
8516 (ashrsi3, ashrsi3_v850e2): Likewise.
8517 (bins): Clobber the flags.
8518 (movsicc_normal_cc, movsicc_normal, movsicc_tst1): Likewise.
8519 (movsicc_tst1_revesed, sasf, swap and rotate patterns): Likewise.
8520 (fix_loop_counter, call_internal_short, call_internal_long): Likewise.
8521 (call_value_internal_short, call_value_internal_long): Likewise.
8522 (callt_save_interrupt, callt_return_interrupt): Likewise.
8523 (save_interrupt, return_interrupt): Likewise.
8524 (callt_save_all_interrupt, save_all_interrupt): Likewise.
8525 (_save_all_interrupt, callt_restore_all_interrupt): Likewise.
8526 (restore_all_interrupt, _restore_all_interrupt): Likewise.
8527 (All FP comparisons): Only allow after reload has completed.
8529 (divh, divhu): Tweak output template.
8530 (branch_z_normal, branch_z_invert): Remove
8531 (branch_nz_normal, branch_nz_invert): Likewise.
8532 (extendhisi_insn, extendqisi_insn): Do not clobber flags.
8534 * config/v850/v850-modes.def (CCZ, CCNZ): Add new modes.
8535 * config/v850/v850.c (notice_update_cc): Remove.
8536 * config/v850/v850.h (CC_OVERFLOW_UNUSABLE): Remove
8537 (CC_NO_CARRY): Likewise.
8538 (NOTICE_UPDATE_CC): Define to nothing.
8539 * config/v850/v850.md: Remove block comment on cc0 handling
8540 Remove "cc" attribute from all patterns. Remove cc_status handling
8541 from all patterns. Minor formatting fixes.
8543 2018-06-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8545 * config/aarch64/aarch64-cores.def (cortex-a76): New entry.
8546 (cortex-a76.cortex-a55): Likewise.
8547 * config/aarch64/aarch64-tune.md: Regenerate.
8548 * doc/invoke.texi (AArch64 Options): Document cortex-a76 and
8549 cortex-a76.cortex-a55.
8551 2018-06-27 Jeff Law <law@redhat.com>
8553 * config/v850/t-v850 (MULTILIB_OPTIONS): Remove 8byte-align.
8554 (MULTILIB_DIRNAMES): Similarly.
8556 2018-06-27 Eric Botcazou <ebotcazou@adacore.com>
8558 * gimple.h (gimple_return_retbnd): Delete.
8559 (gimple_return_set_retbnd): Likewise.
8560 * cgraphunit.c (cgraph_node::expand_thunk): Remove call to
8561 gimple_return_set_retbnd.
8562 * gimple-pretty-print.c (dump_gimple_return): Remove call to
8563 gimple_return_retbnd and adjust.
8564 * tree-inline.h (struct copy_body_data): Remove retbnd field.
8565 * tree-inline.c (remap_gimple_stmt): Remove handling of retbnd.
8566 Explicitly return NULL in a couple more cases. Move assertion
8567 on debug statements and remove unreachable code.
8568 (reset_debug_binding): Do not test id->retbnd.
8569 (expand_call_inline): Do not set it.
8571 2018-06-27 Rasmus Villemoes <rv@rasmusvillemoes.dk>
8573 * configure.ac: Add --disable-gcov option.
8574 * configure: Regenerate.
8575 * Makefile.in: Honour @enable_gcov@.
8576 * doc/install.texi: Document --disable-gcov.
8578 2018-06-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8580 * config/arm/arm-cpus.in (cortex-a76): New entry.
8581 (cortex-a76.cortex-a55): Likewise.
8582 * config/arm/arm-tables.opt: Regenerate.
8583 * config/arm/arm-tune.md: Likewise.
8584 * config/arm/driver-arm.c (arm_cpu_table): Add Cortex-A76 entry.
8585 * doc/invoke.texi (ARM Options): Document cortex-a76 and
8586 cortex-a76.cortex-a55.
8588 2018-06-27 Tamar Christina <tamar.christina@arm.com>
8591 * config/aarch64/aarch64.md (*movhf_aarch64): Add dup v0.4h pattern.
8593 2018-06-27 Siddhesh Poyarekar <siddhesh@sourceware.org>
8595 * config/aarch64/aarch64.h (CALL_USE_REGISTERS): Fix obsolete
8597 (EPILOGUE_USES): Likewise.
8599 2018-06-26 Eric Botcazou <ebotcazou@adacore.com>
8601 * tree-inline.c (remap_location): New function extracted from...
8602 (copy_edges_for_bb): Add ID parameter. Remap goto_locus.
8603 (copy_phis_for_bb): ...here. Call remap_location.
8604 (copy_cfg_body): Adjust call to copy_edges_for_bb.
8606 2018-06-26 Aaron Sawdey <acsawdey@linux.ibm.com>
8608 * config/rs6000/rs6000-string.c (expand_block_clear): Don't use
8609 unaligned vsx for 16B memset.
8611 2018-06-26 Segher Boessenkool <segher@kernel.crashing.org>
8614 * config/rs6000/rs6000.c (rs6000_init_builtins): Do not set
8615 ieee128_float_type_node to long_double_type_node unless
8616 TARGET_LONG_DOUBLE_128 is set.
8618 2018-06-26 David Malcolm <dmalcolm@redhat.com>
8620 * cfgloop.c (get_loop_location): Convert return type from
8621 location_t to dump_user_location_t, replacing INSN_LOCATION lookups
8622 by implicit construction from rtx_insn *, and using
8623 dump_user_location_t::from_function_decl for the fallback case.
8624 * cfgloop.h (get_loop_location): Convert return type from
8625 location_t to dump_user_location_t.
8626 * cgraphunit.c (walk_polymorphic_call_targets): Update call to
8627 dump_printf_loc to pass in a dump_location_t rather than a
8628 location_t, via the gimple stmt.
8629 * coverage.c (get_coverage_counts): Update calls to
8630 dump_printf_loc to pass in dump_location_t rather than a
8632 * doc/optinfo.texi (Dump types): Convert example of
8633 dump_printf_loc from taking "locus" to taking "insn". Update
8634 description of the "_loc" calls to cover dump_location_t.
8635 * dumpfile.c: Include "backend.h", "gimple.h", "rtl.h", and
8637 (dump_user_location_t::dump_user_location_t): New constructors,
8638 from gimple *stmt and rtx_insn *.
8639 (dump_user_location_t::from_function_decl): New function.
8640 (dump_loc): Make static.
8641 (dump_gimple_stmt_loc): Convert param "loc" from location_t to
8642 const dump_location_t &.
8643 (dump_generic_expr_loc): Delete.
8644 (dump_printf_loc): Convert param "loc" from location_t to
8645 const dump_location_t &.
8646 (selftest::test_impl_location): New function.
8647 (selftest::dumpfile_c_tests): New function.
8648 * dumpfile.h: Include "profile-count.h".
8649 (class dump_user_location_t): New class.
8650 (struct dump_impl_location_t): New struct.
8651 (class dump_location_t): New class.
8652 (dump_printf_loc): Convert 2nd param from source_location to
8653 const dump_location_t &.
8654 (dump_generic_expr_loc): Delete.
8655 (dump_gimple_stmt_loc): Convert 2nd param from source_location to
8656 const dump_location_t &.
8657 * gimple-fold.c (fold_gimple_assign): Update call to
8658 dump_printf_loc to pass in a dump_location_t rather than a
8659 location_t, via the gimple stmt.
8660 (gimple_fold_call): Likewise.
8661 * gimple-loop-interchange.cc
8662 (loop_cand::analyze_iloop_reduction_var): Update for change to
8663 check_reduction_path.
8664 (tree_loop_interchange::interchange): Update for change to
8666 * graphite-isl-ast-to-gimple.c (scop_to_isl_ast): Update for
8667 change in return-type of find_loop_location.
8668 (graphite_regenerate_ast_isl): Likewise.
8669 * graphite-optimize-isl.c (optimize_isl): Likewise.
8670 * graphite.c (graphite_transform_loops): Likewise.
8671 * ipa-devirt.c (ipa_devirt): Update call to dump_printf_loc to
8672 pass in a dump_location_t rather than a location_t, via the
8674 * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
8675 * ipa.c (walk_polymorphic_call_targets): Likewise.
8676 * loop-unroll.c (report_unroll): Convert "locus" param from
8677 location_t to dump_location_t.
8678 (decide_unrolling): Update for change to get_loop_location's
8680 * omp-grid.c (struct grid_prop): Convert field "target_loc" from
8681 location_t to dump_user_location_t.
8682 (grid_find_single_omp_among_assignments_1): Updates calls to
8683 dump_printf_loc to pass in a dump_location_t rather than a
8684 location_t, via the gimple stmt.
8685 (grid_parallel_clauses_gridifiable): Convert "tloc" from
8686 location_t to dump_location_t. Updates calls to dump_printf_loc
8687 to pass in a dump_location_t rather than a location_t, via the
8689 (grid_inner_loop_gridifiable_p): Likewise.
8690 (grid_dist_follows_simple_pattern): Likewise.
8691 (grid_gfor_follows_tiling_pattern): Likewise.
8692 (grid_target_follows_gridifiable_pattern): Likewise.
8693 (grid_attempt_target_gridification): Convert initialization
8694 of local "grid" from memset to zero-initialization; FIXME: does
8695 this require C++11? Update call to dump_printf_loc to pass in a
8696 optinfo_location rather than a location_t, via the gimple stmt.
8697 * profile.c (read_profile_edge_counts): Updates call to
8698 dump_printf_loc to pass in a dump_location_t rather than a
8700 (compute_branch_probabilities): Likewise.
8701 * selftest-run-tests.c (selftest::run_tests): Call
8703 * selftest.h (dumpfile_c_tests): New decl.
8704 * tree-loop-distribution.c (pass_loop_distribution::execute):
8705 Update for change in return type of find_loop_location.
8706 * tree-parloops.c (parallelize_loops): Likewise.
8707 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Convert
8708 "locus" from location_t to dump_user_location_t.
8709 (canonicalize_loop_induction_variables): Likewise.
8710 * tree-ssa-loop-ivopts.c (tree_ssa_iv_optimize_loop): Update
8711 for change in return type of find_loop_location.
8712 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update call
8713 to dump_printf_loc to pass in a dump_location_t rather than a
8714 location_t, via the stmt.
8715 * tree-ssa-sccvn.c (eliminate_dom_walker::before_dom_children):
8717 * tree-vect-loop-manip.c (find_loop_location): Convert return
8718 type from source_location to dump_user_location_t.
8719 (vect_do_peeling): Update for above change.
8720 (vect_loop_versioning): Update for change in type of
8722 * tree-vect-loop.c (check_reduction_path): Convert "loc" param
8723 from location_t to dump_user_location_t.
8724 (vect_estimate_min_profitable_iters): Update for change in type
8726 * tree-vect-slp.c (vect_print_slp_tree): Convert param "loc" from
8727 location_t to dump_location_t.
8728 (vect_slp_bb): Update for change in type of vect_location.
8729 * tree-vectorizer.c (vect_location): Convert from source_location
8730 to dump_user_location_t.
8731 (try_vectorize_loop_1): Update for change in vect_location's type.
8732 (vectorize_loops): Likewise.
8733 (increase_alignment): Likewise.
8734 * tree-vectorizer.h (vect_location): Convert from source_location
8735 to dump_user_location_t.
8736 (find_loop_location): Convert return type from source_location to
8737 dump_user_location_t.
8738 (check_reduction_path): Convert 1st param from location_t to
8739 dump_user_location_t.
8740 * value-prof.c (check_counter): Update call to dump_printf_loc to
8741 pass in a dump_user_location_t rather than a location_t; update
8742 call to error_at for change in type of "locus".
8743 (check_ic_target): Update call to dump_printf_loc to
8744 pass in a dump_user_location_t rather than a location_t, via the
8747 2018-06-26 Robin Dapp <rdapp@linux.vnet.ibm.com>
8749 * config/s390/s390.h (enum processor_flags): Do not use
8750 default tune parameter when -march was specified.
8752 2018-06-26 Jakub Jelinek <jakub@redhat.com>
8755 * config/i386/i386.md (setcc + movzbl to xor + setcc peephole2s):
8756 Check reg_overlap_mentioned_p in addition to reg_set_p with the same
8759 2018-06-26 Richard Biener <rguenther@suse.de>
8761 PR tree-optimization/86287
8763 * tree-vect-loop.c (vect_transform_loop_stmt): Fix read-after-free.
8764 (vect_analyze_loop): Initialize n_stmts.
8766 2018-06-26 Richard Biener <rguenther@suse.de>
8769 * fold-const.c (fold_convertible_p): Pointer extension
8772 2018-06-26 Alexandre Oliva <aoliva@redhat.com>
8775 * dwarf2out.c (loc_list_has_views): Adjust comments.
8776 (dw_loc_list): Split single cross-partition range with
8779 2018-06-25 Jeff Law <law@redhat.com>
8781 * common/config/v850/v850-common.c (TARGET_DEFAULT_TARGET_FLAGS): Turn
8782 on -mbig-switch by default.
8784 * config/v850/predicates.md (const_float_1_operand): Fix match_code
8786 (const_float_0_operand): Remove unused predicate.
8787 * config/v850/v850.md (define_constants): Remove UNSPEC_LOOP.
8788 (define_c_enum unspec): Add LOOP, RCP and RSQRT constants.
8789 (recipsf2): New expander. Original pattern now called
8791 (recipdf2, recipdf2_insn): Similarly.
8792 (rsqrtsf2, rsqrtsf2_insn): Similarly
8793 (rsqrtdf2, rsqrtdf2_insn): Similarly
8795 2018-06-26 Gerald Pfeifer <gerald@pfeifer.com>
8797 * ginclude/stddef.h: Remove an obsolete comment on FreeBSD 5.
8798 Simplify logic for FreeBSD (twice).
8800 2018-06-25 Martin Sebor <msebor@redhat.com>
8802 PR tree-optimization/86204
8803 * tree-ssa-strlen.c (handle_builtin_strlen): Avoid storing
8804 a strnlen result if it's less than the length of the string.
8806 2018-06-25 Martin Sebor <msebor@redhat.com>
8808 PR tree-optimization/85700
8809 * gimple-fold.c (gimple_fold_builtin_strncat): Adjust comment.
8810 * tree-ssa-strlen.c (is_strlen_related_p): Handle integer subtraction.
8811 (maybe_diag_stxncpy_trunc): Distinguish strncat from strncpy.
8813 2018-06-25 Martin Sebor <msebor@redhat.com>
8815 * doc/extend.texi (Zero-length arrays): Update and clarify.
8817 2018-06-25 Michael Meissner <meissner@linux.ibm.com>
8819 * config.gcc (powerpc64le*): Revert January 16th, 2018 patch that
8820 added IEEE/IBM long double multilib support on PowerPC little
8821 endian Linux systems.
8822 * config/rs6000/linux64.h (MULTILIB_DEFAULTS_IEEE): Likewise.
8823 (MULTILIB_DEFAULTS): Likewise.
8824 * config/rs6000/rs6000.c (rs6000_option_override_internal):
8826 * config/rs6000/rs6000.h (TARGET_IEEEQUAD_MULTILIB): Likewise.
8827 * config/rs6000/t-ldouble-linux64le-ibm: Delete, no longer used.
8828 * config/rs6000/t-ldouble-linux64le-ieee: Delete, no longer used.
8830 2018-06-25 Alexander Monakov <amonakov@ispras.ru>
8833 * sort.cc (REORDER_23): Avoid memcpy with same destination and source.
8834 (REORDER_45): Likewise.
8836 2018-06-25 Jeff Law <law@redhat.com>
8838 * config/v850/v850.md (divmodhi4): Make sure to sign extend the
8839 dividend to 32 bits. Adjust length.
8840 (udivmodhi4): Cleanup output template. Fix length.
8842 2018-06-25 Carl Love <cel@us.ibm.com>
8844 * config/rs6000/vsx.md: Change word selector to prefered location.
8846 2018-06-25 Richard Biener <rguenther@suse.de>
8848 PR tree-optimization/86304
8849 * tree-vectorizer.c (vectorize_loops): Walk over new possibly
8850 epilogue-if-converted loops as well.
8852 2018-06-25 Jan Hubicka <hubicka@ucw.cz>
8854 * lto-section-out.c (lto_begin_section): Do not print section
8855 name for noaddr and unnumbered dumps.
8857 2018-06-25 Richard Biener <rguenther@suse.de>
8859 * tree-vectorizer.h (struct vec_info_shared): New structure
8860 with parts split out from struct vec_info and loop_nest from
8861 struct _loop_vec_info.
8862 (struct vec_info): Adjust accordingly.
8863 (struct _loop_vec_info): Likewise.
8864 (LOOP_VINFO_LOOP_NEST): Adjust.
8865 (LOOP_VINFO_DATAREFS): Likewise.
8866 (LOOP_VINFO_DDRS): Likewise.
8867 (struct _bb_vec_info): Likewise.
8868 (BB_VINFO_DATAREFS): Likewise.
8869 (BB_VINFO_DDRS): Likewise.
8870 (struct _stmt_vec_info): Add dr_aux member.
8871 (DR_VECT_AUX): Adjust to refer to member of DR_STMTs vinfo.
8872 (DR_MISALIGNMENT_UNINITIALIZED): New.
8873 (set_dr_misalignment): Adjust.
8874 (dr_misalignment): Assert misalign isn't DR_MISALIGNMENT_UNINITIALIZED.
8875 (vect_analyze_loop): Adjust prototype.
8876 (vect_analyze_loop_form): Likewise.
8877 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences):
8878 Compute dependences lazily.
8879 (vect_record_base_alignments): Use shared datarefs/ddrs.
8880 (vect_verify_datarefs_alignment): Likewise.
8881 (vect_analyze_data_refs_alignment): Likewise.
8882 (vect_analyze_data_ref_accesses): Likewise.
8883 (vect_analyze_data_refs): Likewise.
8884 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Add
8885 constructor parameter for shared part.
8886 (vect_analyze_loop_form): Pass in shared part and adjust.
8887 (vect_analyze_loop_2): Pass in storage for the number of
8888 stmts. Move loop nest finding to the caller. Compute
8890 (vect_analyze_loop): Pass in shared part.
8891 (vect_transform_loop): Verify shared datarefs are unchanged.
8892 * tree-vect-slp.c (_bb_vec_info::_bb_vec_info): Add
8893 constructor parameter for shared part.
8894 (vect_slp_analyze_bb_1): Pass in shared part and adjust.
8895 (vect_slp_bb): Verify shared datarefs are unchanged before
8897 * tree-vect-stmts.c (ensure_base_align): Adjust for DR_AUX
8899 (new_stmt_vec_info): Initialize DR_AUX misalignment to
8900 DR_MISALIGNMENT_UNINITIALIZED.
8901 * tree-vectorizer.c (vec_info::vec_info): Add constructor
8902 parameter for shared part.
8903 (vec_info::~vec_info): Adjust.
8904 (vec_info_shared::vec_info_shared): New.
8905 (vec_info_shared::~vec_info_shared): Likewise.
8906 (vec_info_shared::save_datarefs): Likewise.
8907 (vec_info_shared::check_datarefs): Likewise.
8908 (try_vectorize_loop_1): Construct shared part live for analyses
8909 of a single loop for multiple vector sizes.
8910 * tree-parloops.c (gather_scalar_reductions): Adjust.
8912 2018-06-25 Richard Biener <rguenther@suse.de>
8914 * tree-vect-data-refs.c (vect_find_stmt_data_reference): Modify
8915 DR for SIMD lane accesses here and mark DR with (void *)-1 aux.
8916 (vect_analyze_data_refs): Remove similar code from here and
8917 simplify accordingly.
8919 2018-06-25 Richard Biener <rguenther@suse.de>
8921 * tree-vect-data-refs.c (vect_check_gather_scatter): Fail
8922 for reverse storage order accesses rather than asserting
8923 they cannot happen here.
8925 2018-06-25 Tom de Vries <tdevries@suse.de>
8928 * config/i386/i386.md (define_insn "*tls_global_dynamic_64_<mode>"):
8929 Use data16 instead of .byte for insn prefix.
8931 2018-06-25 Andreas Krebbel <krebbel@linux.ibm.com>
8934 * parser.c (make_char_string_pack): Pass this literal chars
8935 through cpp_interpret_string.
8936 (cp_parser_userdef_numeric_literal): Check the result of
8937 make_char_string_pack.
8939 2018-06-24 Maya Rashish <coypu@sdf.org>
8941 * ginclude/stddef.h: Simplify conditions around avoiding
8942 re-definition of __size_t.
8944 2018-06-22 Jan Hubicka <hubicka@ucw.cz>
8946 * lto-streamer-out.c (tree_is_indexable): Make LABEL_DECL nonindexable
8947 unless it is forced or nonlocal; assert that we stream no IMPORTED_DECL.
8949 2018-06-22 Maya Rashish <coypu@sdf.org>
8951 * doc/invoke.texi (mno-fancy-math-387): Update for changes
8952 made to OpenBSD and NetBSD through the years.
8954 2018-06-22 Kelvin Nilsen <kelvin@gcc.gnu.org>
8956 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Change
8957 behavior of vec_pack (vector double, vector double) to match
8958 behavior of vec_float2 (vector double, vector double).
8960 2018-06-22 Olivier Hainque <hainque@adacore.com>
8962 * gimplify.c (gimplify_function_tree): Prevent creation
8963 of a trampoline for the address of the current function
8964 passed to entry/exit instrumentation hooks.
8966 2018-06-22 Aaron Sawdey <acsawdey@linux.ibm.com>
8969 * config/rs6000/rs6000-string.c (expand_strn_compare): Handle -m32
8972 2018-06-22 Martin Liska <mliska@suse.cz>
8974 PR tree-optimization/86263
8975 * tree-switch-conversion.c (jump_table_cluster::find_jump_tables):
8976 Bail out if is_enabled is false.
8977 * tree-switch-conversion.h (jump_table_cluster::is_enabled):
8979 (jump_table_cluster::is_enabled): New function.
8981 2018-06-22 Jan Hubicka <hubicka@ucw.cz>
8983 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not stream
8984 BINFO_BASE_ACCESSES and BINFO_VPTR_FIELD.
8985 * tree-streamer-in.c (streamer_read_tree_bitfields): Likewise.
8986 (lto_input_ts_binfo_tree_pointers): Likewise.
8987 * tree-streamer-out.c (streamer_write_tree_bitfields,
8988 write_ts_binfo_tree_pointers): Likewise.
8989 * tree.c (free_lang_data_in_binfo): Clear BINFO_VPTR_FIELD.
8991 2018-06-22 Jan Hubicka <hubicka@ucw.cz>
8993 * tree.c (free_lang_data_in_type): Free all TYPE_VFIELDs.
8995 2018-06-22 Martin Liska <mliska@suse.cz>
8997 * symbol-summary.h (get): Make it pure and inline move
8998 functionality from ::get function.
8999 (get): Remove and inline into ::get and ::get_create.
9000 (get_create): Move code from ::get function.
9002 2018-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9005 * config/i386/sol2.h (CPP_SPEC): Don't pass -P for
9006 -x assembler-with-cpp.
9008 2018-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9010 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define
9011 _FILE_OFFSET_BITS=64 for C++.
9013 2018-06-21 Michael Meissner <meissner@linux.ibm.com>
9015 * config/rs6000/rs6000.md (extendtfif2): Add missing 128-bit
9016 conversion insn that shows up when pr85657-3.c is compiled using
9017 IEEE 128-bit long double.
9018 (neg<mode>2_internal): Use the correct mode to check whether the
9019 mode is IBM extended.
9020 * config/rs6000/rs6000.c (init_float128_ieee): Prevent complex
9021 multiply and divide external functions from being created more
9024 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
9026 * cfgrtl.c (fixup_reorder_chain): Do not emit NOPs in DECL_IGNORED_P
9028 (rtl_merge_blocks): Likewise. Do not emit a NOP if the location of
9029 the edge can be forwarded.
9030 (cfg_layout_merge_blocks): Likewise.
9032 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
9034 * except.c (finish_eh_generation): Commit edge insertions only after
9035 the EH edges have been redirected from post-landing to landing pads.
9037 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
9039 * tree-nested.c (get_frame_type): Use create_tmp_var_raw instead of
9040 create_tmp_var_for to create the FRAME decl.
9041 (finalize_nesting_tree_1): Do not unchain the FRAME decl.
9043 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
9045 * tree-inline.c (copy_edges_for_bb): Minor tweak.
9046 (maybe_move_debug_stmts_to_successors): Also reset the locus of the
9047 debug statement when resetting its value.
9048 (expand_call_inline): Copy the locus of the call onto the assignment
9049 of the return value, if any. Use local variable in more cases.
9051 2018-06-21 Martin Liska <mliska@suse.cz>
9053 * ipa-pure-const.c (propagate_nothrow): Use
9054 funct_state_summaries->get.
9055 (dump_malloc_lattice): Likewise.
9056 (propagate_malloc): Likewise.
9058 2018-06-21 Richard Biener <rguenther@suse.de>
9060 * lto-streamer-out.c (DFS::DFS_write_tree_body): Update outdated
9061 comment. Follow BLOCK_ABSTRACT_ORIGIN unconditionally.
9062 * tree-streamer-in.c (lto_input_ts_block_tree_pointers): Update
9064 * tree-streamer-out.c (write_ts_block_tree_pointers): Stream
9065 BLOCK_ABSTRACT_ORIGIN unconditionally.
9067 2018-06-21 David Malcolm <dmalcolm@redhat.com>
9069 * ipa-cp.c (ipcp_driver): Set edge_clone_summaries to NULL after
9071 * ipa-reference.c (ipa_reference_c_finalize): Delete
9072 ipa_ref_opt_sum_summaries and set it to NULL.
9074 2018-06-21 Tom de Vries <tdevries@suse.de>
9076 PR tree-optimization/85859
9077 * tree-ssa-tail-merge.c (stmt_local_def): Copy gimple_is_call
9078 test with comment from bb_no_side_effects_p.
9080 2018-06-21 Richard Biener <rguenther@suse.de>
9082 PR tree-optimization/86232
9083 * tree-ssa-loop-niter.c (number_of_iterations_popcount): Adjust
9084 max for constant niter.
9086 2018-06-21 Andre Vieira <andre.simoesdiasvieira@arm.com>
9088 * config/aarch64/aarch64-simd.md
9089 (*aarch64_crypto_aes<aes_op>v16qi_xor_combine): New.
9091 2018-06-21 Andre Vieira <andre.simoesdiasvieira@arm.com>
9093 * config/aarch64/aarch64-simd.md (aarch64_crypto_aes<aes_op>v16qi):
9094 Make opernads of the unspec commutative.
9096 2018-06-21 Richard Biener <rguenther@suse.de>
9098 * tree-data-ref.c (dr_step_indicator): Handle NULL DR_STEP.
9099 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
9100 Avoid calling vect_mark_for_runtime_alias_test with gathers or scatters.
9101 (vect_analyze_data_ref_dependence): Re-order checks to deal with
9103 (vect_record_base_alignments): Do not record base alignment
9104 for gathers or scatters.
9105 (vect_compute_data_ref_alignment): Drop return value that is always
9106 true. Bail out early for gathers or scatters.
9107 (vect_enhance_data_refs_alignment): Bail out early for gathers
9109 (vect_find_same_alignment_drs): Likewise.
9110 (vect_analyze_data_refs_alignment): Remove dead code.
9111 (vect_slp_analyze_and_verify_node_alignment): Likewise.
9112 (vect_analyze_data_refs): For possible gathers or scatters do
9113 not create an alternate DR, just check their possible validity
9114 and mark them. Adjust DECL_NONALIASED handling to not rely
9116 * tree-vect-loop-manip.c (vect_update_inits_of_drs): Do not
9117 update inits of gathers or scatters.
9118 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern):
9119 Also copy gather/scatter flag to pattern vinfo.
9121 2018-06-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
9123 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Change
9124 behavior of vec_packsu (vector unsigned long long, vector unsigned
9125 long long) to match behavior of vec_packs with same signature.
9127 2018-06-20 Chung-Lin Tang <cltang@codesourcery.com>
9128 Thomas Schwinge <thomas@codesourcery.com>
9129 Cesar Philippidis <cesar@codesourcery.com>
9131 * gimplify.c (gimplify_scan_omp_clauses): Add support for
9132 OMP_CLAUSE_{IF_PRESENT,FINALIZE}.
9133 (gimplify_adjust_omp_clauses): Likewise.
9134 (gimplify_oacc_declare_1): Add support for GOMP_MAP_RELEASE, remove
9135 support for GOMP_MAP_FORCE_{ALLOC,TO,FROM,TOFROM}.
9136 (gimplify_omp_target_update): Update handling of acc update and
9138 * omp-low.c (install_var_field): Remove unused parameter
9139 base_pointers_restrict.
9140 (scan_sharing_clauses): Remove base_pointers_restrict parameter.
9141 Update call to install_var_field. Handle OMP_CLAUSE_{IF_PRESENT,
9143 (omp_target_base_pointers_restrict_p): Delete.
9144 (scan_omp_target): Update call to scan_sharing_clauses.
9145 * tree-core.h (enum omp_clause_code): Add OMP_CLAUSE_{IF_PRESENT,
9147 * tree-nested.c (convert_nonlocal_omp_clauses): Handle
9148 OMP_CLAUSE_{IF_PRESENT,FINALIZE}.
9149 (convert_local_omp_clauses): Likewise.
9150 * tree-pretty-print.c (dump_omp_clause): Likewise.
9151 * tree.c (omp_clause_num_ops): Add entries for OMP_CLAUSE_{IF_PRESENT,
9153 (omp_clause_code_name): Likewise.
9155 2018-06-20 Jakub Jelinek <jakub@redhat.com>
9158 * var-tracking.c (use_narrower_mode_test): Check if shift amount can
9161 PR tree-optimization/86231
9162 * tree-vrp.c (union_ranges): For ( [ ) ] or ( )[ ] range and
9163 anti-range don't overwrite *vr0min before using it to compute *vr0max.
9165 2018-06-20 Tom de Vries <tdevries@suse.de>
9167 PR tree-optimization/86097
9168 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Also convert *nit to
9169 iv type if signedness of iv type is not the same as that of *nit.
9171 2018-06-20 Jakub Jelinek <jakub@redhat.com>
9173 * cfgrtl.c (rtl_verify_edges): Formatting fix. If bb->preds has any
9174 EDGE_EH edges, verify they are all EDGE_EH.
9176 2018-06-20 Maya Rashish <coypu@sdf.org>
9178 * ginclude/stddef.h: Limit #include <machine/ansi.h> to NetBSD.
9180 2018-06-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
9182 * config/aarch64/aarch64-tuning-flags.def (no_ldp_stp_qregs): New.
9183 * config/aarch64/aarch64.c (xgene1_tunings): Add
9184 AARCH64_EXTRA_TUNE_NO_LDP_STP_QREGS to tune_flags.
9185 (aarch64_mode_valid_for_sched_fusion_p):
9186 Allow 16-byte modes.
9187 (aarch64_classify_address): Allow 16-byte modes for load_store_pair_p.
9188 * config/aarch64/aarch64-ldpstp.md: Add peepholes for LDP STP of
9190 * config/aarch64/aarch64-simd.md (load_pair<VQ:mode><VQ2:mode>):
9192 (vec_store_pair<VQ:mode><VQ2:mode>): Likewise.
9193 * config/aarch64/iterators.md (VQ2): New mode iterator.
9195 2018-06-20 Martin Liska <mliska@suse.cz>
9197 * tree-switch-conversion.c (jump_table_cluster::can_be_handled):
9198 Change default ratio from 10 to 8.
9200 2018-06-20 Martin Liska <mliska@suse.cz>
9202 * tree-switch-conversion.c (jump_table_cluster::find_jump_tables):
9204 (bit_test_cluster::find_bit_tests): Likewise.
9205 (switch_decision_tree::analyze_switch_statement): Find clusters.
9206 * tree-switch-conversion.h (struct jump_table_cluster): Document
9209 2018-06-20 Martin Liska <mliska@suse.cz>
9211 * tree-switch-conversion.c (switch_conversion::collect):
9212 Record m_uniq property.
9213 (switch_conversion::expand): Bail out for special conditions.
9214 (group_cluster::~group_cluster): New.
9215 (group_cluster::group_cluster): Likewise.
9216 (group_cluster::dump): Likewise.
9217 (jump_table_cluster::emit): New.
9218 (switch_decision_tree::fix_phi_operands_for_edges): New.
9219 (struct case_node): Remove struct.
9220 (jump_table_cluster::can_be_handled): New.
9221 (case_values_threshold): Moved to header.
9222 (reset_out_edges_aux): Likewise.
9223 (jump_table_cluster::is_beneficial): New.
9224 (bit_test_cluster::can_be_handled): Likewise.
9225 (add_case_node): Remove.
9226 (bit_test_cluster::is_beneficial): New.
9227 (case_bit_test::cmp): New.
9228 (bit_test_cluster::emit): New.
9229 (expand_switch_as_decision_tree_p): Remove.
9230 (bit_test_cluster::hoist_edge_and_branch_if_true): New.
9231 (fix_phi_operands_for_edge): Likewise.
9232 (switch_decision_tree::analyze_switch_statement): New.
9233 (compute_cases_per_edge): Move ...
9234 (switch_decision_tree::compute_cases_per_edge): ... here.
9235 (try_switch_expansion): Likewise.
9236 (switch_decision_tree::try_switch_expansion): Likewise.
9237 (record_phi_operand_mapping): Likewise.
9238 (switch_decision_tree::record_phi_operand_mapping): Likewise.
9239 (emit_case_decision_tree): Likewise.
9240 (switch_decision_tree::emit): Likewise.
9241 (balance_case_nodes): Likewise.
9242 (switch_decision_tree::balance_case_nodes): Likewise.
9243 (dump_case_nodes): Likewise.
9244 (switch_decision_tree::dump_case_nodes): Likewise.
9245 (emit_jump): Likewise.
9246 (switch_decision_tree::emit_jump): Likewise.
9247 (emit_cmp_and_jump_insns): Likewise.
9248 (switch_decision_tree::emit_cmp_and_jump_insns): Likewise.
9249 (emit_case_nodes): Likewise.
9250 (switch_decision_tree::emit_case_nodes): Likewise.
9251 (conditional_probability): Remove.
9252 * tree-switch-conversion.h (enum cluster_type): New.
9254 (struct cluster): Likewise.
9255 (cluster::cluster): Likewise.
9256 (struct simple_cluster): Likewise.
9257 (simple_cluster::simple_cluster): Likewise.
9258 (struct group_cluster): Likewise.
9259 (struct jump_table_cluster): Likewise.
9260 (struct bit_test_cluster): Likewise.
9261 (struct min_cluster_item): Likewise.
9262 (struct case_tree_node): Likewise.
9263 (case_tree_node::case_tree_node): Likewise.
9264 (jump_table_cluster::case_values_threshold): Likewise.
9265 (struct case_bit_test): Likewise.
9266 (struct switch_decision_tree): Likewise.
9267 (struct switch_conversion): Likewise.
9268 (switch_decision_tree::reset_out_edges_aux): Likewise.
9270 2018-06-20 Martin Liska <mliska@suse.cz>
9272 * tree-switch-conversion.c (MAX_CASE_BIT_TESTS): Remove.
9273 (hoist_edge_and_branch_if_true): Likewise.
9274 (expand_switch_using_bit_tests_p): Likewise.
9275 (struct case_bit_test): Likewise.
9276 (case_bit_test_cmp): Likewise.
9277 (emit_case_bit_tests): Likewise.
9278 (switch_conversion::switch_conversion): New class.
9279 (struct switch_conv_info): Remove old struct.
9280 (collect_switch_conv_info): More to ...
9281 (switch_conversion::collect): ... this.
9282 (check_range): Likewise.
9283 (switch_conversion::check_range): Likewise.
9284 (check_all_empty_except_final): Likewise.
9285 (switch_conversion::check_all_empty_except_final): Likewise.
9286 (check_final_bb): Likewise.
9287 (switch_conversion::check_final_bb): Likewise.
9288 (create_temp_arrays): Likewise.
9289 (switch_conversion::create_temp_arrays): Likewise.
9290 (free_temp_arrays): Likewise.
9291 (gather_default_values): Likewise.
9292 (switch_conversion::gather_default_values): Likewise.
9293 (build_constructors): Likewise.
9294 (switch_conversion::build_constructors): Likewise.
9295 (constructor_contains_same_values_p): Likewise.
9296 (switch_conversion::contains_same_values_p): Likewise.
9297 (array_value_type): Likewise.
9298 (switch_conversion::array_value_type): Likewise.
9299 (build_one_array): Likewise.
9300 (switch_conversion::build_one_array): Likewise.
9301 (build_arrays): Likewise.
9302 (switch_conversion::build_arrays): Likewise.
9303 (gen_def_assigns): Likewise.
9304 (switch_conversion::gen_def_assigns): Likewise.
9305 (prune_bbs): Likewise.
9306 (switch_conversion::prune_bbs): Likewise.
9307 (fix_phi_nodes): Likewise.
9308 (switch_conversion::fix_phi_nodes): Likewise.
9309 (gen_inbound_check): Likewise.
9310 (switch_conversion::gen_inbound_check): Likewise.
9311 (process_switch): Use the newly created class.
9312 (switch_conversion::expand): New.
9313 (switch_conversion::~switch_conversion): New.
9314 * tree-switch-conversion.h: New file.
9316 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
9318 * tree-vectorizer.h (NUM_PATTERNS, vect_recog_func_ptr): Move to
9319 tree-vect-patterns.c.
9320 * tree-vect-patterns.c (vect_supportable_direct_optab_p): New function.
9321 (vect_recog_dot_prod_pattern): Use it. Remove the type_in argument.
9322 (vect_recog_sad_pattern): Likewise.
9323 (vect_recog_widen_sum_pattern): Likewise.
9324 (vect_recog_pow_pattern): Likewise. Check for a null vectype.
9325 (vect_recog_widen_shift_pattern): Remove the type_in argument.
9326 (vect_recog_rotate_pattern): Likewise.
9327 (vect_recog_mult_pattern): Likewise.
9328 (vect_recog_vector_vector_shift_pattern): Likewise.
9329 (vect_recog_divmod_pattern): Likewise.
9330 (vect_recog_mixed_size_cond_pattern): Likewise.
9331 (vect_recog_bool_pattern): Likewise.
9332 (vect_recog_mask_conversion_pattern): Likewise.
9333 (vect_try_gather_scatter_pattern): Likewise.
9334 (vect_recog_widen_mult_pattern): Likewise. Check for a null vectype.
9335 (vect_recog_over_widening_pattern): Likewise.
9336 (vect_recog_gather_scatter_pattern): Likewise.
9337 (vect_recog_func_ptr): Move from tree-vectorizer.h
9338 (vect_vect_recog_func_ptrs): Move further down the file.
9339 (vect_recog_func): Likewise. Remove the third argument.
9340 (NUM_PATTERNS): Define based on vect_vect_recog_func_ptrs.
9341 (vect_pattern_recog_1): Expect the pattern function to do any
9342 necessary target tests. Also expect it to provide a vector type.
9343 Remove the type_in handling.
9345 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
9347 * tree-vect-patterns.c (vect_pattern_detected): New function.
9348 (vect_recog_dot_prod_patternm, vect_recog_sad_pattern)
9349 (vect_recog_widen_mult_pattern, vect_recog_widen_sum_pattern)
9350 (vect_recog_over_widening_pattern, vect_recog_widen_shift_pattern
9351 (vect_recog_rotate_pattern, vect_recog_vector_vector_shift_pattern)
9352 (vect_recog_mult_pattern, vect_recog_divmod_pattern)
9353 (vect_recog_mixed_size_cond_pattern, vect_recog_bool_pattern)
9354 (vect_recog_mask_conversion_pattern)
9355 (vect_try_gather_scatter_pattern): Likewise.
9357 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
9359 * tree-vect-patterns.c (vect_get_internal_def): New function.
9360 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern)
9361 (vect_recog_vector_vector_shift_pattern, check_bool_pattern)
9362 (search_type_for_mask_1): Use it.
9364 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
9366 * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Remove
9367 redundant WIDEN_SUM_EXPR handling.
9368 (vect_recog_sad_pattern): Likewise.
9370 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
9372 * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Remove
9373 redundant check that the types of a PLUS_EXPR or MULT_EXPR agree.
9374 (vect_recog_sad_pattern): Likewise PLUS_EXPR, ABS_EXPR and MINUS_EXPR.
9375 (vect_recog_widen_mult_pattern): Likewise MULT_EXPR.
9376 (vect_recog_widen_sum_pattern): Likewise PLUS_EXPR.
9378 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
9380 * tree-vect-stmts.c (vectorizable_call): Make sure that we
9381 use the stmt_vec_info of the original bb statement for the
9382 new zero assignment, even if the call is part of a pattern.
9384 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
9386 * tree-vectorizer.h (_stmt_vec_info): Note above pattern_def_seq
9387 that the sequence is attached to the original statement rather
9388 than the pattern statement.
9389 * tree-vect-loop.c (vect_determine_vf_for_stmt): Take the
9390 PATTERN_DEF_SEQ from the original statement rather than
9391 the main pattern statement.
9392 * tree-vect-stmts.c (free_stmt_vec_info): Likewise.
9393 * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Likewise.
9394 (vect_mark_pattern_stmts): Don't copy the PATTERN_DEF_SEQ.
9396 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
9398 * tree-vect-stmts.c (vect_analyze_stmt): Move the handling of pattern
9399 definition statements before the early exit for statements that aren't
9401 * tree-vect-loop.c (vect_transform_loop_stmt): New function,
9403 (vect_transform_loop): ...here. Process pattern definition
9404 statements without first checking whether the main pattern
9405 statement is live or relevant.
9407 2018-06-19 Eric Botcazou <ebotcazou@adacore.com>
9409 * tree-cfgcleanup.c (tree_forwarder_block_p): Do not return false at
9410 -O0 if the locus represent UNKNOWN_LOCATION but have different values.
9412 2018-06-19 Aaron Sawdey <acsawdey@linux.ibm.com>
9414 * config/rs6000/rs6000-string.c (select_block_compare_mode): Check
9415 TARGET_EFFICIENT_OVERLAPPING_UNALIGNED here instead of in caller.
9416 (do_and3, do_and3_mask, do_compb3, do_rotl3): New functions.
9417 (expand_block_compare): Change select_block_compare_mode call.
9418 (expand_strncmp_align_check): Use new functions, fix comment.
9419 (emit_final_str_compare_gpr): New function.
9420 (expand_strn_compare): Refactor and clean up code.
9421 * config/rs6000/vsx.md (vsx_mov<mode>_64bit): Remove *.
9423 2018-06-19 Tony Reix <tony.reix@atos.com>
9424 Damien Bergamini <damien.bergamini@atos.com>
9425 David Edelsohn <dje.gcc@gmail.com>
9427 * collect2.c (static_obj): New variable.
9428 (static_libs): New variable.
9429 (is_in_list): Uncomment declaration.
9430 (main): Track AIX libraries linked statically.
9431 (is_in_list): Uncomment definition.
9432 (scan_prog_file): Don't add AIX shared libraries initializer
9433 to constructor list if linking statically.
9435 2018-06-19 Max Filippov <jcmvbkbc@gmail.com>
9437 * config/xtensa/xtensa.md (UNSPEC_FRAME_BLOCKAGE): New unspec
9439 (allocate_stack, frame_blockage, *frame_blockage): New patterns.
9441 2018-06-19 Jan Hubicka <hubicka@ucw.cz>
9443 * tree.c (find_decls_types_r): Remove all non-VAR_DECLs from
9446 2018-06-19 Martin Liska <mliska@suse.cz>
9448 * config/i386/i386.c (ix86_can_inline_p): Do not use
9449 ipa_fn_summaries::get_create.
9450 * ipa-cp.c (ipcp_cloning_candidate_p): Replace get_create with
9452 (devirtualization_time_bonus): Likewise.
9453 (ipcp_propagate_stage): Likewise.
9454 * ipa-fnsummary.c (redirect_to_unreachable): Likewise.
9455 (edge_set_predicate): Likewise.
9456 (evaluate_conditions_for_known_args): Likewise.
9457 (evaluate_properties_for_edge): Likewise.
9458 (ipa_call_summary::reset): Tranform to ...
9459 (ipa_call_summary::~ipa_call_summary): ... this.
9460 (ipa_fn_summary::reset): Transform to ...
9461 (ipa_fn_summary::~ipa_fn_summary): ... this.
9462 (ipa_fn_summary_t::remove): Rename to ...
9463 (ipa_fn_summary_t::remove_callees): ... this.
9464 (ipa_fn_summary_t::duplicate): Use placement new
9465 instead of memory copy.
9466 (ipa_call_summary_t::duplicate): Likewise.
9467 (ipa_call_summary_t::remove): Remove.
9468 (dump_ipa_call_summary): Change get_create to get.
9469 (ipa_dump_fn_summary): Dump only when summary exists.
9470 (analyze_function_body): Use symbol_summary::get instead
9472 (compute_fn_summary): Likewise.
9473 (estimate_edge_devirt_benefit): Likewise.
9474 (estimate_edge_size_and_time): Likewise.
9475 (inline_update_callee_summaries): Likewise.
9476 (remap_edge_change_prob): Likewise.
9477 (remap_edge_summaries): Likewise.
9478 (ipa_merge_fn_summary_after_inlining): Likewise.
9479 (write_ipa_call_summary): Likewise.
9480 (ipa_fn_summary_write): Likewise.
9481 (ipa_free_fn_summary): Likewise.
9482 * ipa-fnsummary.h (struct GTY): Add new ctor and copy ctor.
9483 (struct ipa_call_summary): Likewise.
9484 * ipa-icf.c (sem_function::merge): Use symbol_summary::get instead
9486 * ipa-inline-analysis.c (do_estimate_edge_time): Likewise.
9487 (estimate_size_after_inlining): Likewise.
9488 (estimate_growth): Likewise.
9489 (growth_likely_positive): Likewise.
9490 * ipa-inline-transform.c (clone_inlined_nodes): Likewise.
9491 (inline_call): Likewise.
9492 * ipa-inline.c (caller_growth_limits): Likewise.
9493 (can_inline_edge_p): Likewise.
9494 (can_inline_edge_by_limits_p): Likewise.
9495 (compute_uninlined_call_time): Likewise.
9496 (compute_inlined_call_time): Likewise.
9497 (want_inline_small_function_p): Likewise.
9498 (edge_badness): Likewise.
9499 (update_caller_keys): Likewise.
9500 (update_callee_keys): Likewise.
9501 (inline_small_functions): Likewise.
9502 (inline_to_all_callers_1): Likewise.
9503 (dump_overall_stats): Likewise.
9504 (early_inline_small_functions): Likewise.
9505 (early_inliner): Likewise.
9506 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
9507 * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
9508 * ipa-pure-const.c (malloc_candidate_p): Likewise.
9509 * ipa-split.c (execute_split_functions): Likewise.
9510 * symbol-summary.h: Likewise.
9511 * tree-sra.c (ipa_sra_preliminary_function_checks): Likewise.
9513 2018-06-19 Richard Biener <rguenther@suse.de>
9515 * tree-vectorizer.c (try_vectorize_loop_1): Split out of ...
9516 (vectorize_loops): ... here. Fix dbgcnt handling.
9517 (try_vectorize_loop): Wrap try_vectorize_loop_1.
9519 2018-06-19 Segher Boessenkool <segher@kernel.crashing.org>
9522 * config/rs6000/rs6000.md (rs6000_discover_homogeneous_aggregate): An
9523 ieee128 argument takes up only one (vector) register, not two (floating
9526 2018-06-19 Eric Botcazou <ebotcazou@adacore.com>
9528 * gimplify.c (gimplify_init_constructor): Really never clear for an
9529 incomplete constructor if CONSTRUCTOR_NO_CLEARING is set.
9531 2018-06-19 Richard Biener <rguenther@suse.de>
9533 PR tree-optimization/86179
9534 * tree-vect-patterns.c (vect_pattern_recog_1): Clean up
9535 after failed recognition.
9537 2018-06-18 Martin Sebor <msebor@redhat.com>
9540 * calls.c (maybe_warn_nonstring_arg): Handle strncat.
9541 * tree-ssa-strlen.c (is_strlen_related_p): Make extern.
9542 Handle integer subtraction.
9543 (maybe_diag_stxncpy_trunc): Handle nonstring source arguments.
9544 * tree-ssa-strlen.h (is_strlen_related_p): Declare.
9546 2018-06-18 David Malcolm <dmalcolm@redhat.com>
9548 * config/frv/frv-protos.h (frv_ifcvt_modify_insn): Strengthen 3rd
9549 param from rtx to rtx_insn *.
9550 * config/frv/frv.c (frv_ifcvt_add_insn): Likewise for "insn"
9552 (frv_ifcvt_modify_insn): Likwise.
9553 (frv_ifcvt_modify_final): Likwise for local "existing_insn",
9554 adding an as_a <rtx_insn *> cast. Likewise for local "insn".
9555 * config/mips/mips.c (r10k_insert_cache_barriers): Add an
9556 as_a <rtx_insn *> cast to local "unprotected_region" once
9557 it's been established that it's not NULL or pc_rtx.
9558 * config/nds32/nds32-relax-opt.c (nds32_group_insns): Strengthen
9559 param "sethi" from rtx to rtx_insn *.
9560 (nds32_group_float_insns): Likewise for param "insn".
9561 * config/vax/vax-protos.h (vax_output_int_add): Likewise for 1st
9563 (vax_output_int_subtract): Likewise.
9564 * config/vax/vax.c (vax_output_int_add): Likewise for param
9566 (vax_output_int_subtract): Likewise.
9567 * emit-rtl.c (set_insn_deleted): Likewise, removing cast.
9568 (emit_pattern_after): Likewise for param "after".
9569 (emit_insn_after): Likewise.
9570 (emit_jump_insn_after): Likewise.
9571 (emit_call_insn_after): Likewise.
9572 (emit_debug_insn_after): Likewise.
9573 (emit_pattern_before): Likewise for param "before".
9574 (emit_insn_before): Likewise.
9575 (emit_jump_insn_before): Likewise.
9576 * final.c (get_insn_template): Likewise for param "insn", removing
9578 * output.h (get_insn_template): Likewise for 2nd param.
9579 * rtl.h (emit_insn_before): Likewise.
9580 (emit_jump_insn_before): Likewise.
9581 (emit_debug_insn_before_noloc): Likewise.
9582 (emit_insn_after): Likewise.
9583 (emit_jump_insn_after): Likewise.
9584 (emit_call_insn_after): Likewise.
9585 (emit_debug_insn_after): Likewise.
9586 (set_insn_deleted): Likewise for param.
9588 2018-06-18 Michael Meissner <meissner@linux.ibm.com>
9591 * config/rs6000/rs6000-modes.def (toplevel): Rework the 128-bit
9592 floating point modes, so that IFmode is numerically greater than
9593 TFmode, which is greater than KFmode using FRACTIONAL_FLOAT_MODE
9594 to declare the ordering. This prevents IFmode from being
9595 converted to TFmode when long double is IEEE 128-bit on an ISA 3.0
9596 machine. Include rs6000-modes.h to share the fractional values
9597 between genmodes* and the rest of the compiler.
9601 * config/rs6000/rs6000-modes.h: New file.
9602 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Change the
9603 meaning of rs6000_long_double_size so that 126..128 selects an
9604 appropriate 128-bit floating point type.
9605 (rs6000_option_override_internal): Likewise.
9606 * config/rs6000/rs6000.h (toplevel): Include rs6000-modes.h.
9607 (TARGET_LONG_DOUBLE_128): Change the meaning of
9608 rs6000_long_double_size so that 126..128 selects an appropriate
9609 128-bit floating point type.
9610 (LONG_DOUBLE_TYPE_SIZE): Update comment.
9611 * config/rs6000/rs6000.md (trunciftf2): Correct the modes of the
9612 source and destination to match the standard usage.
9613 (truncifkf2): Likewise.
9614 (copysign<mode>3, IEEE iterator): Rework copysign of float128 on
9615 ISA 2.07 to use an explicit clobber, instead of passing in a
9617 (copysign<mode>3_soft): Likewise.
9619 2018-06-18 David Malcolm <dmalcolm@redhat.com>
9621 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences):
9622 Replace dump_printf_loc call with DUMP_VECT_SCOPE.
9623 (vect_slp_analyze_instance_dependence): Likewise.
9624 (vect_enhance_data_refs_alignment): Likewise.
9625 (vect_analyze_data_refs_alignment): Likewise.
9626 (vect_slp_analyze_and_verify_instance_alignment
9627 (vect_analyze_data_ref_accesses): Likewise.
9628 (vect_prune_runtime_alias_test_list): Likewise.
9629 (vect_analyze_data_refs): Likewise.
9630 * tree-vect-loop-manip.c (vect_update_inits_of_drs): Likewise.
9631 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
9632 (vect_analyze_scalar_cycles_1): Likewise.
9633 (vect_get_loop_niters): Likewise.
9634 (vect_analyze_loop_form_1): Likewise.
9635 (vect_update_vf_for_slp): Likewise.
9636 (vect_analyze_loop_operations): Likewise.
9637 (vect_analyze_loop): Likewise.
9638 (vectorizable_induction): Likewise.
9639 (vect_transform_loop): Likewise.
9640 * tree-vect-patterns.c (vect_pattern_recog): Likewise.
9641 * tree-vect-slp.c (vect_analyze_slp): Likewise.
9642 (vect_make_slp_decision): Likewise.
9643 (vect_detect_hybrid_slp): Likewise.
9644 (vect_slp_analyze_operations): Likewise.
9645 (vect_slp_bb): Likewise.
9646 * tree-vect-stmts.c (vect_mark_stmts_to_be_vectorized): Likewise.
9647 (vectorizable_bswap): Likewise.
9648 (vectorizable_call): Likewise.
9649 (vectorizable_simd_clone_call): Likewise.
9650 (vectorizable_conversion): Likewise.
9651 (vectorizable_assignment): Likewise.
9652 (vectorizable_shift): Likewise.
9653 (vectorizable_operation): Likewise.
9654 * tree-vectorizer.h (DUMP_VECT_SCOPE): New macro.
9656 2018-06-18 Martin Sebor <msebor@redhat.com>
9658 PR tree-optimization/81384
9659 * builtin-types.def (BT_FN_SIZE_CONST_STRING_SIZE): New.
9660 * builtins.c (expand_builtin_strnlen): New function.
9661 (expand_builtin): Call it.
9662 (fold_builtin_n): Avoid setting TREE_NO_WARNING.
9663 * builtins.def (BUILT_IN_STRNLEN): New.
9664 * calls.c (maybe_warn_nonstring_arg): Handle BUILT_IN_STRNLEN.
9665 Warn for bounds in excess of maximum object size.
9666 * tree-ssa-strlen.c (maybe_set_strlen_range): Return tree representing
9667 single-value ranges. Handle strnlen.
9668 (handle_builtin_strlen): Handle strnlen.
9669 (strlen_check_and_optimize_stmt): Same.
9670 * doc/extend.texi (Other Builtins): Document strnlen.
9672 2018-06-18 Maya Rashish <coypu@sdf.org>
9674 * config/alpha/openbsd.h (TARGET_DEFAULT): Define.
9675 (LINK_SPEC, STARTFILE_SPEC, ENDFILE_SPEC): Likewise.
9676 (INTMAX_TYPE, UINTMAX_TYPE, WINT_TYPE): Likewise.
9678 * config/alpha/elf.h (STARTFILE_SPEC, ENDFILE_SPEC): Move from
9680 * config/alpha/linux.h (STARTFILE_SPEC, ENDFILE_SPEC): Here.
9682 2018-06-18 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
9684 * tree.c (escaped_string::escape): Replace cast to char * by
9685 const_cast<char *> (unescaped).
9687 2018-06-18 Nick Clifton <nickc@redhat.com>
9690 * tree.c (escaped_string): New class. Converts an unescaped
9691 string into its escaped equivalent.
9692 (warn_deprecated_use): Use the new class to convert the
9693 deprecation message, if present.
9694 (test_escaped_strings): New self test.
9695 (test_c_tests): Add test_escaped_strings.
9696 * doc/extend.texi (deprecated): Add a note that the
9697 deprecation message is affected by the -fmessage-length
9698 option, and that control characters will be escaped.
9699 (#pragma GCC error): Document this pragma.
9700 (#pragma GCC warning): Likewise.
9701 * doc/invoke.texi (-fmessage-length): Document this option's
9702 effect on the #warning and #error preprocessor directives and
9703 the deprecated attribute.
9705 2018-06-18 Eric Botcazou <ebotcazou@adacore.com>
9707 * tree.c (decl_value_expr_lookup): Revert latest change.
9708 (decl_value_expr_insert): Likewise.
9710 2018-06-17 Eric Botcazou <ebotcazou@adacore.com>
9712 * gimplify.c (nonlocal_vlas): Delete.
9713 (nonlocal_vla_vars): Likewise.
9714 (gimplify_var_or_parm_decl): Do not add debug VAR_DECLs for non-local
9716 (gimplify_body): Do not create and destroy nonlocal_vlas.
9717 * tree-nested.c: Include diagnostic.h.
9718 (use_pointer_in_frame): Tweak.
9719 (lookup_field_for_decl): Add assertion and declare the transformation.
9720 (convert_nonlocal_reference_op) <PARM_DECL>: Rework and issue an
9721 internal error when the reference is in a wrong context. Do not
9722 create a debug decl by default.
9723 (note_nonlocal_block_vlas): Delete.
9724 (convert_nonlocal_reference_stmt) <GIMPLE_BIND>: Do not call it.
9725 (convert_local_reference_op) <PARM_DECL>: Skip the frame decl. Do not
9726 create a debug decl by default.
9727 (convert_gimple_call) <GIMPLE_CALL>: Issue an internal error when the
9728 call is in a wrong context.
9729 (fixup_vla_decls): New function.
9730 (finalize_nesting_tree_1): Adjust comment. Call fixup_vla_decls if no
9731 debug variables were created.
9732 * tree.c (decl_value_expr_lookup): Add checking assertion.
9733 (decl_value_expr_insert): Likewise.
9735 2018-06-16 Kugan Vivekanandarajah <kuganv@linaro.org>
9738 * ipa-fnsummary.c (will_be_nonconstant_expr_predicate): Handle CALL_EXPR.
9739 * tree-scalar-evolution.c (interpret_expr): Likewise.
9740 (expression_expensive_p): Likewise.
9741 * tree-ssa-loop-ivopts.c (contains_abnormal_ssa_name_p): Likewise.
9742 * tree-ssa-loop-niter.c (number_of_iterations_popcount): New.
9743 (number_of_iterations_exit_assumptions): Use number_of_iterations_popcount.
9744 (ssa_defined_by_minus_one_stmt_p): New.
9746 2018-06-16 Kugan Vivekanandarajah <kuganv@linaro.org>
9749 * cfgexpand.c (expand_debug_expr): Hande ABSU_EXPR.
9750 * config/i386/i386.c (ix86_add_stmt_cost): Likewise.
9751 * dojump.c (do_jump): Likewise.
9752 * expr.c (expand_expr_real_2): Check operand type's sign.
9753 * fold-const.c (const_unop): Handle ABSU_EXPR.
9754 (fold_abs_const): Likewise.
9755 * gimple-pretty-print.c (dump_unary_rhs): Likewise.
9756 * gimple-ssa-backprop.c (backprop::process_assign_use): Likesie.
9757 (strip_sign_op_1): Likesise.
9758 * match.pd: Add new pattern to generate ABSU_EXPR.
9759 * optabs-tree.c (optab_for_tree_code): Handle ABSU_EXPR.
9760 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
9761 * tree-eh.c (operation_could_trap_helper_p): Likewise.
9762 * tree-inline.c (estimate_operator_cost): Likewise.
9763 * tree-pretty-print.c (dump_generic_node): Likewise.
9764 * tree-vect-patterns.c (vect_recog_sad_pattern): Likewise.
9765 * tree.def (ABSU_EXPR): New.
9767 2018-06-16 Jakub Jelinek <jakub@redhat.com>
9770 * common.opt (Wunsafe-loop-optimizations): Add Ignore, remove Var,
9771 documented as preserved for backward compatibility only.
9772 * doc/invoke.texi: Remove -Wunsafe-loop-optimizations documentation.
9774 PR rtl-optimization/86108
9775 * bb-reorder.c (create_forwarder_block): Renamed to ...
9776 (create_eh_forwarder_block): ... this. Split OLD_BB after labels and
9777 jump from new landing pad to the second part.
9778 (sjlj_fix_up_crossing_landing_pad, dw2_fix_up_crossing_landing_pad):
9781 2018-06-15 Jakub Jelinek <jakub@redhat.com>
9784 * expr.c (expand_assignment): Remove now redundant COMPLEX_MODE_P
9785 check from first store_expr, use to_mode instead of GET_MODE (to_rtx).
9786 Only call store_expr for halves if the mode is the same.
9789 * match.pd ((X / Y) == 0 -> X < Y): Don't transform complex divisions.
9790 Fix up comment formatting.
9792 2018-06-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
9794 * typed-splay-tree.h (typed_splay_tree::remove): New function.
9795 (typed_splay_tree::closure,
9796 typed_splay_tree::inner_foreach_fn, typed_splay_tree::m_inner): Deleted.
9797 (typed_splay_tree::typed_splay_tree,
9798 typed_splay_tree::operator =): Declared private.
9799 (typed_splay_tree::splay_tree_key, typed_splay_tree::splay_tree_value,
9800 typed_splay_tree::splay_tree_node_s, typed_splay_tree::KDEL,
9801 typed_splay_tree::VDEL, typed_splay_tree::splay_tree_delete_helper,
9802 typed_splay_tree::rotate_left, typed_splay_tree::rotate_right,
9803 typed_splay_tree::splay_tree_splay,
9804 typed_splay_tree::splay_tree_foreach_helper,
9805 typed_splay_tree::splay_tree_insert,
9806 typed_splay_tree::splay_tree_remove,
9807 typed_splay_tree::splay_tree_lookup,
9808 typed_splay_tree::splay_tree_predecessor,
9809 typed_splay_tree::splay_tree_successor,
9810 typed_splay_tree::splay_tree_min,
9811 typed_splay_tree::splay_tree_max): Took over from splay-tree.c/.h.
9812 (typed_splay_tree::root, typed_splay_tree::comp,
9813 typed_splay_tree::delete_key,
9814 typed_splay_tree::delete_value): New data members.
9815 * typed-splay-tree.c (selftest::test_str_to_int): Add a test for
9816 typed_splay_tree::remove.
9818 2018-06-15 Matthew Fortune <matthew.fortune@mips.com>
9820 * config/mips/mips.h (ASM_SPEC): Pass through -mcrc, -mno-crc,
9821 -mginv and -mno-ginv to the assembler.
9822 * config/mips/mips.opt (-mcrc): New option.
9824 * doc/invoke.text (-mcrc): Document.
9827 2018-06-15 Nick Clifton <nickc@redhat.com>
9830 * tree.c (escaped_string): New class. Converts an unescaped
9831 string into its escaped equivalent.
9832 (warn_deprecated_use): Use the new class to convert the
9833 deprecation message, if present.
9834 (test_escaped_strings): New self test.
9835 (test_c_tests): Add test_escaped_strings.
9836 * doc/extend.texi (deprecated): Add a note that the
9837 deprecation message is affected by the -fmessage-length
9838 option, and that control characters will be escaped.
9839 (#pragma GCC error): Document this pragma.
9840 (#pragma GCC warning): Likewise.
9841 * doc/invoke.texi (-fmessage-length): Document this option's
9842 effect on the #warning and #error preprocessor directives and
9843 the deprecated attribute.
9845 2018-06-15 Richard Biener <rguenther@suse.de>
9847 * tree-vect-slp.c (vect_slp_bb): Dump MSG_OPTIMIZED_LOCATIONS
9848 here, also noting vector size used.
9849 * tree-vectorizer.c (vectorize_loops): Adjust. Note vector
9850 size used in MSG_OPTIMIZED_LOCATIONS dump.
9851 (pass_slp_vectorize::execute): Adjust.
9853 2018-06-15 Claudiu Zissulescu <claziss@synopsys.com>
9856 * config/arc/arc.c (arc_return_address_register): Fix
9859 2018-06-15 Richard Biener <rguenther@suse.de>
9862 * tree-cfg.c (gimplify_build3): Do not strip sign conversions,
9863 leave useless conversion stripping to force_gimple_operand_gsi.
9864 (gimplify_build2): Likewise.
9865 (gimplify_build1): Likewise.
9867 2018-06-15 Richard Biener <rguenther@suse.de>
9870 * tree-cfg.c (move_stmt_op): unshare invariant addresses
9871 before adjusting their block.
9873 2018-06-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
9875 * config.gcc (riscv*-*-elf* | riscv*-*-rtems*): Use custom
9876 multilibs for *-*-rtems*.
9877 * config/riscv/t-rtems: New file.
9879 2018-06-14 Jakub Jelinek <jakub@redhat.com>
9882 * match.pd ((A +- CST1) +- CST2): Punt if last resort
9883 unsigned_type_for returns NULL.
9886 * lower-subreg.c (find_decomposable_subregs): Don't decompose float
9887 subregs of multi-word pseudos unless the float mode has word size.
9889 2018-06-14 Richard Biener <rguenther@suse.de>
9892 * tree-vect-generic.c (build_word_mode_vector_type): Remove
9893 duplicate and harmful type_hash_canon.
9894 * tree.c (type_hash_canon): Assert we didn't find ourselves.
9896 2018-06-14 Richard Biener <rguenther@suse.de>
9899 * tree-ssa-struct-alias.c (create_variable_info_for): Handle
9902 2018-06-14 Sebastian Huber <sebastian.huber@embedded-brains.de>
9904 * config/rtems.h (STDINT_LONG32): Define.
9906 2018-06-13 Matthew Fortune <matthew.fortune@mips.com>
9907 Prachi Godbole <prachi.godbole@imgtec.com>
9909 * config/mips/mips-cpus.def: Define P6600.
9910 * config/mips/mips-tables.opt: Regenerate.
9911 * config/mips/mips.c (mips_ucbranch_type): New enum.
9912 (mips_rtx_cost_data): Add support for P6600.
9913 (mips_issue_rate): Likewise.
9914 (mips_multipass_dfa_lookahead): Likewise.
9915 (mips_avoid_hazard): Likewise.
9916 (mips_reorg_process_insns): Likewise.
9917 (mips_classify_branch_p6600): New function.
9918 * config/mips/mips.h (TUNE_P6600): New define.
9919 (MIPS_ISA_LEVEL_SPEC): Infer mips64r6 from p6600.
9920 (ENABLE_LD_ST_PAIRS): Enable load/store bonding for p6600.
9921 * config/mips/mips.md: Include p6600.md.
9922 (processor): Add p6600.
9923 * config/mips/p6600.md: New file.
9924 * doc/invoke.texi: Add p6600 to supported architectures.
9926 2018-06-13 Martin Sebor <msebor@redhat.com>
9928 PR tree-optimization/86114
9929 * gimple-fold.c (gimple_fold_builtin_strlen): Only handle LHS
9931 * tree-ssa-strlen.c (maybe_set_strlen_range): Same.
9933 2018-06-13 Richard Biener <rguenther@suse.de>
9935 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern):
9936 Properly set vector type of the intermediate stmt.
9937 * tree-vect-stmts.c (vectorizable_operation): The destination
9938 var always has vectype_out type.
9940 2018-06-13 Jeff Law <law@redhat.com>
9942 * config/rl78/rl78.c (move_elim_pass): Use TDF_NONE rather than
9943 integer 0 for argument to print_rtl_with_bb.
9944 (rl78_reorg): Likewise.
9946 2018-06-13 David Malcolm <dmalcolm@redhat.com>
9948 * config/arc/arc.c (hwloop_optimize): Strengthen local "end_label"
9949 from rtx to rtx_insn *.
9950 * config/bfin/bfin.c (hwloop_optimize): Likewise for local
9952 (add_sched_insns_for_speculation): Likewise for local "target",
9953 converting usage of JUMP_LABEL to JUMP_LABEL_AS_INSN.
9954 * config/c6x/c6x.c (reorg_split_calls): Strengthen param "call_labels"
9955 from rtx_insn ** to rtx_code_label **.
9956 (reorg_emit_nops): Likewise.
9957 (c6x_reorg): Likewise for local "call_labels".
9958 * config/sh/sh-protos.h (get_dest_uid): Strengthen 1st param from
9960 * config/sh/sh.c (dump_table): Strengthen local "lab" from rtx to
9961 rtx_code_label *, adding safe_as_a <rtx_code_label *> casts to
9962 the loops over LABEL_REFS.
9963 (fixup_addr_diff_vecs): Add as_a <rtx_insn *> to usage of
9965 (barrier_align): Convert usage of JUMP_LABEL to JUMP_LABEL_AS_INSN.
9966 (get_dest_uid): Strengthen param "label" from rtx to rtx_insn *.
9967 (split_branches): Strengthen local "olabel" from rtx to
9968 rtx_insn *, adding a safe_as_a cast.
9969 * emit-rtl.c (next_real_insn): Strengthen param from "rtx"
9971 (add_insn_after): Likewise for first two params.
9972 (add_insn_before): Likewise.
9973 (remove_insn): Likewise for param.
9974 (emit_pattern_before_noloc): Likewise for second and third params.
9975 (emit_jump_insn_before_noloc): Convert NULL_RTX to NULL.
9976 (emit_call_insn_before_noloc): Likewise.
9977 (emit_debug_insn_before_noloc): Strengthen "before" param from "rtx"
9979 (emit_barrier_before): Likewise.
9980 (emit_label_before): Strengthen "label" param from "rtx" to
9981 "rtx_code_label *". Strengthen "before" param from "rtx" to
9983 (emit_insn_after_1): Strengthen "after" param from "rtx" to
9985 (emit_pattern_after_noloc): Likewise.
9986 (emit_insn_after_noloc): Likewise.
9987 (emit_jump_insn_after_noloc): Likewise.
9988 (emit_call_insn_after_noloc): Likewise.
9989 (emit_debug_insn_after_noloc): Likewise.
9990 (emit_barrier_after): Likewise.
9991 (emit_label_after): Likewise for both params.
9992 (emit_pattern_after_setloc): Likewise for "after" param. Convert
9993 "loc" param from "int" to "location_t".
9994 (emit_insn_after_setloc): Likewise.
9995 (emit_jump_insn_after_setloc): Likewise.
9996 (emit_call_insn_after_setloc): Likewise.
9997 (emit_debug_insn_after_setloc): Likewise.
9998 (emit_pattern_before_setloc): Likewise for "before" param. Convert
9999 "loc" param from "int" to "location_t".
10000 (emit_pattern_before): Convert NULL_RTX to NULL.
10001 (emit_insn_before_setloc): Convert "loc" param from "int" to
10003 (emit_jump_insn_before_setloc): Likewise.
10004 (emit_call_insn_before_setloc): Likewise.
10005 (emit_debug_insn_before_setloc): Strengthen "before" param from rtx to
10006 rtx_insn *. Convert "loc" param from "int" to "location_t".
10007 * rtl.h (emit_insn_before_setloc, emit_jump_insn_before_setloc,
10008 emit_call_insn_before_setloc, emit_debug_insn_before_setloc):
10009 Convert 3rd param from "int" to "location_t".
10010 (emit_barrier_before, emit_barrier_after, next_real_insn):
10011 Strengthen param from rtx to rtx_insn *.
10012 (emit_label_before): Strengthen 1st param from "rtx" to
10013 "rtx_code_label *". Strengthen 2nd param from "rtx" to
10015 (emit_insn_after_noloc, emit_jump_insn_after_noloc,
10016 emit_call_insn_after_noloc, emit_debug_insn_after_noloc):
10017 Strengthen 2nd param from "rtx" to "rtx_insn *".
10018 (emit_insn_after_setloc, emit_jump_insn_after_setloc)
10019 emit_call_insn_after_setloc, emit_debug_insn_after_setloc):
10020 Likewise. Convert 3rd param from "int" to "location_t".
10021 (emit_label_after): Strengthen 1st param from "rtx" to
10022 "rtx_code_label *".
10023 (next_real_insn, remove_insn): Strengthen param from "rtx" to
10025 (add_insn_before, add_insn_after): Strengthen 1st and 2nd params
10026 from "rtx" to "rtx_insn *".
10028 2018-06-13 Jan Hubicka <hubicka@gcc.gnu.org>
10030 * cgraph.c (cgraph_node::get_untransformed_body): Dump function
10031 bodies streamed in with -Q.
10032 * dumpfile.c (dump_files): Add lto-stream-out dump file.
10033 * dumpfile.h (tree_dump_index): Add lto_stream_out.
10034 * gimple-streamer-out.c: Include gimple-pretty-print.h
10035 (output_bb): Dump stmts streamed.
10036 * lto-section-out.c: Include print-tree.h
10037 (lto_begin_section): Dump sections created.
10038 (lto_output_decl_index): Dump decl encoded.
10039 * lto-streamer-out.c: Include print-tree.h
10040 (create_output_block): Dump output block created.
10041 (DFS::DFS_write_tree_body): Dump DFS SCCs streamed.
10042 (output_function): Dump function output.
10043 (output_constructor): Dump constructor streamed.
10044 (write_global_stream): Output indexes encoded.
10045 (produce_asm_for_decls): Dump streams encoded.
10046 * lto-streamer.c (streamer_dump_file): New global var.
10047 * lto-streamer.h (streamer_dump_file): Declare.
10048 * passes.c (ipa_write_summaries): Initialize streamer dump.
10049 * varpool.c (varpool_node::get_constructor): Dump constructors streamed
10052 2018-06-13 Eric Botcazou <ebotcazou@adacore.com>
10055 * config/i386/winnt.c (i386_pe_seh_cold_init): Do not emit negative
10056 offsets for register save directives. Emit a second batch of save
10057 directives, if need be, when the function accesses prior frames.
10059 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
10061 * config/arc/fpu.md (fmasf4): Force operand to register.
10062 (fnmasf4): Likewise.
10064 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
10066 * config/arc/arc-protos.h (arc_pad_return): Remove.
10067 * config/arc/arc.c (machine_function): Remove force_short_suffix
10069 (arc_print_operand): Adjust printing of '&'.
10070 (arc_verify_short): Remove conditional printing of short suffix.
10071 (arc_final_prescan_insn): Remove reference to size_reason.
10072 (pad_return): New function.
10073 (arc_reorg): Call pad_return.
10074 (arc_pad_return): Remove.
10075 (arc_init_machine_status): Remove reference to force_short_suffix.
10076 * config/arc/arc.md (vunspec): Add VUNSPEC_ARC_BLOCKAGE.
10077 (attr length): When attribute iscompact is true force to 2
10078 regardless; in the case of maybe check if we want to force the
10079 instruction to have 4 bytes length.
10080 (nopv): Change it to generate 4 byte long nop as well.
10081 (blockage): New pattern.
10082 (simple_return): Remove call to arc_pad_return.
10083 (p_return_i): Likewise.
10085 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
10087 * config/arc/elf.h (LINK_GCC_C_SEQUENCE_SPEC): Define.
10089 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
10091 * config/arc/builtins.def (SYNC): SYNC instruction is valid on all
10094 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
10096 * config/arc/arc.c (atomic_exchangesi): EX instruction is default
10097 for ARC700 and ARCv2.
10099 2018-06-13 Chenghua Xu <paul.hua.gm@gmail.com>
10102 * config/mips/loongson.md (vec_setv4hi): Gen_lowpart for
10103 operands[2] instead of operands[1].
10106 2018-06-12 Richard Sandiford <richard.sandiford@linaro.org>
10108 * lra-constraints.c (simplify_operand_subreg): In the paradoxical
10109 case, check whether the outer register overlaps an unallocatable
10110 register, not just whether it fits the required class.
10112 2018-06-12 Richard Sandiford <richard.sandiford@linaro.org>
10114 * poly-int.h (can_div_trunc_p): Add new overload in which all values
10116 * alias.c (get_addr): Extend CONST_INT handling to poly_int_rtx_p.
10117 (memrefs_conflict_p): Likewise.
10118 (init_alias_analysis): Likewise.
10119 * cfgexpand.c (expand_debug_expr): Likewise.
10120 * combine.c (combine_simplify_rtx, force_int_to_mode): Likewise.
10121 * cse.c (fold_rtx): Likewise.
10122 * explow.c (adjust_stack, anti_adjust_stack): Likewise.
10123 * expr.c (emit_block_move_hints): Likewise.
10124 (clear_storage_hints, push_block, emit_push_insn): Likewise.
10125 (store_expr_with_bounds, reduce_to_bit_field_precision): Likewise.
10126 (emit_group_load_1): Use rtx_to_poly_int64 for group offsets.
10127 (emit_group_store): Likewise.
10128 (find_args_size_adjust): Use strip_offset. Use rtx_to_poly_int64
10129 to read the PRE/POST_MODIFY increment.
10130 * calls.c (store_one_arg): Use strip_offset.
10131 * rtlanal.c (rtx_addr_can_trap_p_1): Extend CONST_INT handling to
10133 (set_noop_p): Use rtx_to_poly_int64 for the elements selected
10135 * simplify-rtx.c (avoid_constant_pool_reference): Use strip_offset.
10136 (simplify_binary_operation_1): Extend CONST_INT handling to
10138 * var-tracking.c (compute_cfa_pointer): Take a poly_int64 rather
10139 than a HOST_WIDE_INT.
10140 (hard_frame_pointer_adjustment): Change from HOST_WIDE_INT to
10142 (adjust_mems, add_stores): Update accodingly.
10143 (vt_canonicalize_addr): Track polynomial offsets.
10144 (emit_note_insn_var_location): Likewise.
10145 (vt_add_function_parameter): Likewise.
10146 (vt_initialize): Likewise.
10148 2018-06-12 Jeff Law <law@redhat.com>
10150 * config.gcc (alpha*-*-freebsd*): Remove.
10151 * config/alpha/freebsd.h: Remove.
10153 2018-06-12 David Malcolm <dmalcolm@redhat.com>
10156 * spellcheck-tree.c (levenshtein_distance): Rename to...
10157 (get_edit_distance): ...this, and update for underlying renaming.
10158 * spellcheck-tree.h (levenshtein_distance): Rename to...
10159 (get_edit_distance): ...this.
10160 * spellcheck.c (levenshtein_distance): Rename to...
10161 (get_edit_distance): ...this. Convert from Levenshtein distance
10162 to Damerau-Levenshtein distance by supporting transpositions of
10163 adjacent characters. Rename "v1" to "v_next" and "v0" to
10165 (selftest::levenshtein_distance_unit_test_oneway): Rename to...
10166 (selftest::test_edit_distance_unit_test_oneway): ...this, and
10167 update for underlying renaming.
10168 (selftest::levenshtein_distance_unit_test): Rename to...
10169 (selftest::test_get_edit_distance_unit): ...this, and update for
10170 underlying renaming.
10171 (selftest::test_find_closest_string): Add example from PR 69968
10172 where transposition helps
10173 (selftest::test_metric_conditions): Update for renaming.
10174 (selftest::test_metric_conditions): Likewise.
10175 (selftest::spellcheck_c_tests): Likewise.
10176 * spellcheck.h (levenshtein_distance): Rename both overloads to...
10177 (get_edit_distance): ...this.
10178 (best_match::consider): Update for renaming.
10180 2018-06-12 Martin Sebor <msebor@redhat.com>
10182 PR tree-optimization/85259
10183 * builtins.c (compute_objsize): Handle constant offsets.
10184 * gimple-ssa-warn-restrict.c (maybe_diag_offset_bounds): Return
10185 true iff a warning has been issued.
10186 * gimple.h (gimple_nonartificial_location): New function.
10187 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Call
10188 gimple_nonartificial_location and handle -Wno-system-headers.
10189 (handle_builtin_stxncpy): Same.
10191 2018-06-12 Martin Sebor <msebor@redhat.com>
10194 * fold-const.c (operand_equal_p): Handle SAVE_EXPR.
10196 2018-06-12 Will Schmidt <will_schmidt@vnet.ibm.com>
10198 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
10199 BUILTIN_VEC_XST entries for pointer to double and long long.
10201 2018-06-12 H.J. Lu <hongjiu.lu@intel.com>
10204 * config/i386/gnu-user.h (TARGET_THREAD_SPLIT_STACK_OFFSET):
10206 * config/i386/gnu-user64.h (TARGET_THREAD_SPLIT_STACK_OFFSET):
10209 2018-06-12 Martin Liska <mliska@suse.cz>
10211 * doc/options.texi: Document IntegerRange.
10213 2018-06-12 Martin Liska <mliska@suse.cz>
10215 * config/i386/i386.opt: Make MPX-related options as Deprecated.
10216 * opt-functions.awk: Handle Deprecated flag.
10217 * opts-common.c (decode_cmdline_option): Handle cl_deprecated
10219 (read_cmdline_option): Report warning for a deprecated option.
10220 * opts.h (struct cl_option): Add new field cl_deprecated.
10221 (CL_ERR_DEPRECATED): New.
10223 2018-06-12 Martin Liska <mliska@suse.cz>
10225 * doc/options.texi: Document Deprecated option flag.
10227 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
10229 * config/arc/arc-arch.h (arc_extras): New enum.
10230 (arc_cpu_t):Add field extra.
10231 (arc_cpu_types): Consider the extras.
10232 * config/arc/arc-cpus.def: Add extras info.
10233 * config/arc/arc-opts.h (processor_type): Consider extra field.
10234 * config/arc/arc.c (arc_override_options): Handle extra field.
10236 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
10238 * config/arc/arc-arch.h: Update ARC_OPTX macro.
10239 * config/arc/arc-options.def (ARC_OPTX): Introduce a new doc
10241 * config/arc/arc.c (arc_init): Update pic warning.
10242 (irq_range): Update irq range parsing warnings.
10243 (arc_override_options): Update various warning messages.
10244 (arc_handle_aux_attribute): Likewise.
10246 2018-06-12 Robert Suchanek <robert.suchanek@mips.com>
10248 * config/mips/i6400.md (i6400_fpu_fadd): Remove frint.
10250 2018-06-12 Jozef Lawrynowicz <jozef.l@mittosystems.com>
10252 * doc/sourcebuild.texi: Document usage of line number 0 in verify
10253 compiler messages directives.
10255 2018-06-12 Matthew Fortune <mfortune@gmail.com>
10257 * config/mips/mips-cpus.def: New MIPS_CPU for i6500.
10258 * config/mips/mips-tables.opt: Regenerate.
10259 * config/mips/mips.h (MIPS_ISA_LEVEL_SPEC): Mark i6500 as
10261 * doc/invoke.texi: Document -march=i6500.
10263 2018-06-12 Prachi Godbole <prachi.godbole@imgtec.com>
10265 * config/mips/i6400.md (i6400_gpmuldiv): Remove cpu_unit.
10266 (i6400_gpmul): Add cpu_unit.
10267 (i6400_gpdiv): Likewise.
10268 (i6400_msa_add_d): Update reservations.
10269 (i6400_msa_int_add) Likewise.
10270 (i6400_msa_short_logic3) Likewise.
10271 (i6400_msa_short_logic2) Likewise.
10272 (i6400_msa_short_logic) Likewise.
10273 (i6400_msa_move) Likewise.
10274 (i6400_msa_cmp) Likewise.
10275 (i6400_msa_short_float2) Likewise.
10276 (i6400_msa_div_d) Likewise.
10277 (i6400_msa_long_logic1) Likewise.
10278 (i6400_msa_long_logic2) Likewise.
10279 (i6400_msa_mult) Likewise.
10280 (i6400_msa_long_float2) Likewise.
10281 (i6400_msa_long_float4) Likewise.
10282 (i6400_msa_long_float5) Likewise.
10283 (i6400_msa_long_float8) Likewise.
10284 (i6400_fpu_fadd): Include frint type.
10285 (i6400_fpu_store): New define_insn_reservation.
10286 (i6400_fpu_load): Likewise.
10287 (i6400_fpu_move): Likewise.
10288 (i6400_fpu_fcmp): Likewise.
10289 (i6400_fpu_fmadd): Likewise.
10290 (i6400_int_mult): Include imul3nc type and update reservation.
10291 (i6400_int_div): Include idiv3 type and update reservation.
10292 (i6400_int_load): Update to check type not move_type.
10293 (i6400_int_store): Likewise.
10294 (i6400_int_prefetch): Set zero latency.
10296 2018-06-12 Eric Botcazou <ebotcazou@adacore.com>
10298 * gcc.c: Document new %@{...} sequence.
10299 (LINK_COMMAND_SPEC): Use it for the -L switches.
10300 (cpp_unique_options): Use it for the -I switches.
10301 (at_file_argbuf): New global variable.
10302 (in_at_file): Likewise.
10303 (alloc_args): Create at_file_argbuf.
10304 (clear_args): Truncate at_file_argbuf.
10305 (store_arg): If in_at_file, push the argument onto at_file_argbuf.
10306 (open_at_file): New function.
10307 (close_at_file): Likewise.
10308 (create_at_file): Delete.
10309 (do_spec_1) <'i'>: Use open_at_file/close_at_file.
10312 (validate_switches_from_spec): Deal with %@{...} sequence.
10313 (validate_switches): Likewise.
10314 (driver::finalize): Call clear_args.
10316 2018-06-11 Rasmus Villemoes <rasmus.villemoes@prevas.dk>
10318 * config/vx-common.h (USE_TM_CLONE_REGISTRY): #define to 0.
10320 2018-06-11 Martin Sebor <msebor@redhat.com>
10322 * doc/invoke.texi (-Wall): List -Wc++17-compat.
10323 (Wno-class-memaccess): Add @opindex.
10324 (Wno-templates, Wno-multiple-inheritance): Same.
10325 (Wno-virtual-inheritance, Wno-namespaces): Same.
10326 (Wno-pedantic, Wno-chkp, Wcoverage-mismatch): Same.
10327 (Wno-format-overflow, Wno-format-truncation): Same.
10328 (Wno-unused-local-typedefs, Walloc-size-larger-than): Same.
10329 (Wno-alloc-size-larger-than, Wframe-larger-than): Same
10330 (Wno-stack-usage, Wno-c++-compat, Wno-c++11-compat): Same.
10331 (Wno-c++14-compat, Wno-c++17-compat, Wno-openmp-simd): Same.
10332 (Wno-unsuffixed-float-constants, Wno-addr-space-convert): Same.
10333 (Wno-misspelled-isr): Same.
10335 2018-06-11 Martin Sebor <msebor@redhat.com>
10337 * PR tree-optimization/86083
10338 * tree-ssa-strlen.c (handle_char_store): Use tree_expr_nonzero_p.
10340 2018-06-11 Zhouyi Zhou <zhouzhouyi@gmail.com>
10342 * tree-eh.c (lower_eh_constructs_2): Add a comma to comment.
10344 2018-06-11 Segher Boessenkool <segher@kernel.crashing.org>
10347 * config/rs6000/rs6000.md (*movdi_internal32): Put constraint modifiers
10348 on the correct operand.
10349 (*movdi_internal64): Ditto.
10351 2018-06-11 Martin Liska <mliska@suse.cz>
10353 PR tree-optimization/86089
10354 * tree-ssa-strlen.c (get_string_length): Move back removed hunk.
10356 2018-06-11 Julia Koval <julia.koval@intel.com>
10358 * config/i386/vaesintrin.h (_mm_aesdec_epi128, _mm_aesdeclast_epi128)
10359 _mm_aesenc_epi128, _mm_aesenclast_epi128: Remove.
10360 * config/i386/vpclmulqdqintrin.h (_mm_clmulepi64_epi128): Remove.
10362 2018-06-11 Olivier Hainque <hainque@adacore.com>
10364 * dwarf2out.c (gen_compile_unit_die): Fallout to DW_LANG_Ada83
10365 for Ada with strict dwarf2.
10367 2018-06-08 Peter Bergner <bergner@vnet.ibm.com>
10370 * config/rs6000/rs6000.c (mem_operand_gpr): Enable PRE_INC and PRE_DEC
10373 2018-06-08 Jan Hubicka <hubicka@ucw.cz>
10375 * dumpfile.c (FIRST_ME_AUTO_NUMBERED_DUMP): Bump to 4.
10377 2018-06-08 David Edelsohn <dje.gcc@gmail.com>
10379 * config/rs6000/rs6000.c (rs6000_passes_ieee128): Protect with #if
10382 2018-06-08 Martin Liska <mliska@suse.cz>
10384 * tree-cfg.h (debug_function): Fix argument type to match
10387 2018-06-08 Martin Liska <mliska@suse.cz>
10389 * config/powerpcspe/powerpcspe.c (rs6000_xcoff_visibility):
10390 Remove usage of MPX-related (and removed) fields.
10391 * config/rs6000/rs6000.c (rs6000_xcoff_visibility): Likewise.
10393 2018-06-08 David Malcolm <dmalcolm@redhat.com>
10395 * cfg.c (debug): Use TDF_NONE rather than 0.
10396 * cfghooks.c (debug): Likewise.
10397 * dumpfile.c (DUMP_FILE_INFO): Likewise; also for OPTGROUP.
10398 (struct dump_option_value_info): Convert to...
10399 (struct kv_pair): ...this template type.
10400 (dump_options): Convert to kv_pair<dump_flags_t>; use TDF_NONE
10402 (optinfo_verbosity_options): Likewise.
10403 (optgroup_options): Convert to kv_pair<optgroup_flags_t>; use
10405 (gcc::dump_manager::dump_register): Use optgroup_flags_t rather
10406 than int for "optgroup_flags" param.
10407 (dump_generic_expr_loc): Use dump_flags_t rather than int for
10409 (dump_dec): Likewise.
10410 (dump_finish): Use TDF_NONE rather than 0.
10411 (gcc::dump_manager::opt_info_enable_passes): Use optgroup_flags_t
10412 rather than int for "optgroup_flags" param. Use TDF_NONE rather
10413 than 0. Update for change to option_ptr.
10414 (opt_info_switch_p_1): Convert "optgroup_flags" param from int *
10415 to optgroup_flags_t *. Use TDF_NONE and OPTGROUP_NONE rather than
10416 0. Update for changes to optinfo_verbosity_options and
10418 (opt_info_switch_p): Convert optgroup_flags from int to
10420 (dump_basic_block): Use dump_flags_t rather than int
10421 for "dump_kind" param.
10422 * dumpfile.h (TDF_ADDRESS, TDF_SLIM, TDF_RAW, TDF_DETAILS,
10423 TDF_STATS, TDF_BLOCKS, TDF_VOPS, TDF_LINENO, TDF_UID)
10424 TDF_STMTADDR, TDF_GRAPH, TDF_MEMSYMS, TDF_RHS_ONLY, TDF_ASMNAME,
10425 TDF_EH, TDF_NOUID, TDF_ALIAS, TDF_ENUMERATE_LOCALS, TDF_CSELIB,
10426 TDF_SCEV, TDF_GIMPLE, TDF_FOLDING, MSG_OPTIMIZED_LOCATIONS,
10427 MSG_MISSED_OPTIMIZATION, MSG_NOTE, MSG_ALL, TDF_COMPARE_DEBUG,
10428 TDF_NONE): Convert from macros to...
10429 (enum dump_flag): ...this new enum.
10430 (dump_flags_t): Update to use enum.
10431 (operator|, operator&, operator~, operator|=, operator&=):
10432 Implement for dump_flags_t.
10433 (OPTGROUP_NONE, OPTGROUP_IPA, OPTGROUP_LOOP, OPTGROUP_INLINE,
10434 OPTGROUP_OMP, OPTGROUP_VEC, OPTGROUP_OTHER, OPTGROUP_ALL):
10435 Convert from macros to...
10436 (enum optgroup_flag): ...this new enum.
10437 (optgroup_flags_t): New typedef.
10438 (operator|, operator|=): Implement for optgroup_flags_t.
10439 (struct dump_file_info): Convert field "alt_flags" to
10440 dump_flags_t. Convert field "optgroup_flags" to
10442 (dump_basic_block): Use dump_flags_t rather than int for param.
10443 (dump_generic_expr_loc): Likewise.
10444 (dump_dec): Likewise.
10445 (dump_register): Convert param "optgroup_flags" to
10447 (opt_info_enable_passes): Likewise.
10448 * early-remat.c (early_remat::dump_edge_list): Use TDF_NONE rather
10450 * gimple-pretty-print.c (debug): Likewise.
10451 * gimple-ssa-store-merging.c (bswap_replace): Likewise.
10452 (merged_store_group::apply_stores): Likewise.
10453 * gimple-ssa-strength-reduction.c (insert_initializers): Likewise.
10454 * gimple.c (verify_gimple_pp): Likewise.
10455 * graphite-poly.c (print_pbb_body): Likewise.
10456 * passes.c (pass_manager::register_one_dump_file): Convert
10457 local "optgroup_flags" to optgroup_flags_t.
10458 * print-tree.c (print_node): Use TDF_NONE rather than 0.
10460 (debug_body): Likewise.
10461 * tree-pass.h (struct pass_data): Convert field "optgroup_flags"
10462 to optgroup_flags_t.
10463 * tree-pretty-print.c (print_struct_decl): Use TDF_NONE rather
10465 * tree-ssa-math-opts.c (convert_mult_to_fma_1): Likewise.
10466 (convert_mult_to_fma): Likewise.
10467 * tree-ssa-reassoc.c (undistribute_ops_list): Likewise.
10468 * tree-ssa-sccvn.c (vn_eliminate): Likewise.
10469 * tree-vect-data-refs.c (dump_lower_bound): Convert param
10470 "dump_kind" to dump_flags_t.
10472 2018-06-08 Segher Boessenkool <segher@kernel.crashing.org>
10474 * config/rs6000/rs6000.c (min, max): Delete.
10476 2018-06-08 Segher Boessenkool <segher@kernel.crashing.org>
10478 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete mention of
10479 -mabi=spe and -mabi=no-spe.
10481 2018-06-08 Martin Liska <mliska@suse.cz>
10483 * ipa-pure-const.c (propagate_pure_const): Use ::get at places
10484 where we expect an existing summary.
10486 2018-06-08 Martin Liska <mliska@suse.cz>
10488 * ipa-inline-analysis.c (simple_edge_hints): Use ::get method.
10489 * ipa-inline.h (estimate_edge_growth): Likewise.
10491 2018-06-08 Martin Liska <mliska@suse.cz>
10493 * cgraph.c (function_version_hasher::hash): Use
10494 cgraph_node::get_uid ().
10495 (function_version_hasher::equal):
10496 * cgraph.h (cgraph_node::get_uid): New method.
10497 * ipa-inline.c (update_caller_keys): Use
10498 cgraph_node::get_uid ().
10499 (update_callee_keys): Likewise.
10500 * ipa-utils.c (searchc): Likewise.
10501 (ipa_reduced_postorder): Likewise.
10502 * lto-cgraph.c (input_node): Likewise.
10503 * passes.c (is_pass_explicitly_enabled_or_disabled): Likewise.
10504 * symbol-summary.h (symtab_insertion): Likewise.
10505 (symtab_removal): Likewise.
10506 (symtab_duplication): Likewise.
10507 * tree-pretty-print.c (dump_function_header): Likewise.
10508 * tree-sra.c (convert_callers_for_node): Likewise.
10510 2018-06-08 Martin Liska <mliska@suse.cz>
10512 * cgraph.c (symbol_table::create_edge): Always assign a new
10514 (symbol_table::free_edge): Do not recycle numbers.
10515 * cgraph.h (cgraph_edge::get): New method.
10516 * symbol-summary.h (symtab_removal): Use it.
10517 (symtab_duplication): Likewise.
10518 (call_summary::hashable_uid): Remove.
10520 2018-06-08 Martin Liska <mliska@suse.cz>
10522 * ipa-inline-analysis.c (inline_edge_removal_hook): Remove.
10523 (initialize_growth_caches): Remove.
10524 (free_growth_caches): Likewise.
10525 (do_estimate_edge_time): Use edge_growth_cache.
10526 (do_estimate_edge_size): Likewise.
10527 (do_estimate_edge_hints): Likewise.
10528 * ipa-inline.c (reset_edge_caches): Likewise.
10529 (recursive_inlining): Likewise.
10530 (inline_small_functions): Likewise.
10531 * ipa-inline.h (initialize_growth_caches): Remove.
10532 (estimate_edge_size): Likewise.
10533 (estimate_edge_time): Likewise.
10534 (estimate_edge_hints): Likewise.
10535 (reset_edge_growth_cache): Likewise.
10536 * symbol-summary.h (call_summary::remove): New method.
10538 2018-06-08 Martin Liska <mliska@suse.cz>
10540 * ipa-cp.c (class edge_clone_summary): New summary.
10541 (grow_edge_clone_vectors): Remove.
10542 (ipcp_edge_duplication_hook): Remove.
10543 (class edge_clone_summary_t): New call_summary class.
10544 (ipcp_edge_removal_hook): Remove.
10545 (edge_clone_summary_t::duplicate): New function.
10546 (get_next_cgraph_edge_clone): Use edge_clone_summaries.
10547 (create_specialized_node): Likewise.
10548 (ipcp_driver): Initialize edge_clone_summaries and do not
10551 2018-06-08 Martin Liska <mliska@suse.cz>
10553 * symbol-summary.h (get): New function.
10554 (call_summary::m_initialize_when_cloning): New class member.
10556 2018-06-08 Martin Liska <mliska@suse.cz>
10558 * cgraph.c (cgraph_node::remove): Do not recycle uid.
10559 * cgraph.h (symbol_table::release_symbol): Do not pass uid.
10560 (symbol_table::allocate_cgraph_symbol): Do not set uid.
10561 * passes.c (uid_hash_t): Record removed_nodes by their uids.
10562 (remove_cgraph_node_from_order): Use the removed_nodes set.
10563 (do_per_function_toporder): Likwise.
10564 * symbol-summary.h (symtab_insertion): Use cgraph_node::uid
10565 instead of summary_uid.
10566 (symtab_removal): Likewise.
10567 (symtab_duplication): Likewise.
10569 2018-06-08 Martin Liska <mliska@suse.cz>
10571 * ipa-cp.c (ipcp_store_bits_results): Use
10572 ipcp_transformation_sum.
10573 (ipcp_store_vr_results): Likewise.
10574 * ipa-prop.c (ipcp_grow_transformations_if_necessary): Renamed
10576 (ipcp_transformation_initialize): ... this.
10577 (ipa_set_node_agg_value_chain):
10578 (ipa_node_params_t::duplicate): Use ipcp_transformation_sum.
10579 (write_ipcp_transformation_info): Likewise.
10580 (read_ipcp_transformation_info): Likewise.
10581 (ipcp_update_bits): Likewise.
10582 (ipcp_update_vr): Likewise.
10583 (ipcp_transform_function): Likewise.
10584 * ipa-prop.h: Rename ipcp_transformation_summary to
10585 ipcp_transformation.
10586 (class ipcp_transformation_t): New function summary.
10587 (ipcp_get_transformation_summary): Use ipcp_transformation_sum.
10588 (ipa_get_agg_replacements_for_node): Likewise.
10590 2018-06-08 Martin Liska <mliska@suse.cz>
10592 * ipa-pure-const.c (struct funct_state_d): Do it class instead
10594 (class funct_state_summary_t): New function_summary class.
10595 (has_function_state): Remove.
10596 (get_function_state): Likewise.
10597 (set_function_state): Likewise.
10598 (add_new_function): Likewise.
10599 (funct_state_summary_t::insert): New function.
10600 (duplicate_node_data): Remove.
10601 (remove_node_data): Remove.
10602 (funct_state_summary_t::duplicate): New function.
10603 (register_hooks): Create new funct_state_summaries.
10604 (pure_const_generate_summary): Use it.
10605 (pure_const_write_summary): Likewise.
10606 (pure_const_read_summary): Likewise.
10607 (propagate_pure_const): Likewise.
10608 (propagate_nothrow): Likewise.
10609 (dump_malloc_lattice): Likewise.
10610 (propagate_malloc): Likewise.
10611 (execute): Do not register hooks, just remove summary
10613 (pass_ipa_pure_const::pass_ipa_pure_const): Simplify
10616 2018-06-08 Martin Liska <mliska@suse.cz>
10618 * ipa-reference.c (remove_node_data): Remove.
10619 (duplicate_node_data): Likewise.
10620 (class ipa_ref_var_info_summary_t): New class.
10621 (class ipa_ref_opt_summary_t): Likewise.
10622 (get_reference_vars_info): Use ipa_ref_var_info_summaries.
10623 (get_reference_optimization_summary): Use
10624 ipa_ref_opt_sum_summaries.
10625 (set_reference_vars_info): Remove.
10626 (set_reference_optimization_summary): Likewise.
10627 (ipa_init): Create summaries.
10628 (init_function_info): Use function summary.
10629 (ipa_ref_opt_summary_t::duplicate): New function.
10630 (ipa_ref_opt_summary_t::remove): New function.
10631 (get_read_write_all_from_node): Fix GNU coding style.
10632 (propagate): Use function summary.
10633 (write_node_summary_p): Fix GNU coding style.
10634 (stream_out_bitmap): Likewise.
10635 (ipa_reference_read_optimization_summary): Use function summary.
10636 (ipa_reference_c_finalize): Do not release hooks.
10638 2018-06-08 Martin Liska <mliska@suse.cz>
10640 * ipa-fnsummary.c (dump_ipa_call_summary): Use ::get method.
10641 (analyze_function_body): Extract multiple calls of get_create.
10642 * ipa-inline-analysis.c (simple_edge_hints): Likewise.
10643 * ipa-inline.c (recursive_inlining): Use ::get method.
10644 * ipa-inline.h (estimate_edge_growth): Likewise.
10646 2018-06-08 Martin Liska <mliska@suse.cz>
10648 * hsa-common.h (enum hsa_function_kind): Rename HSA_NONE to
10650 (hsa_function_summary::hsa_function_summary): Use the new enum
10652 (hsa_gpu_implementation_p): Use hsa_summaries::get.
10653 * hsa-gen.c (hsa_get_host_function): Likewise.
10654 (get_brig_function_name): Likewise.
10655 * ipa-hsa.c (process_hsa_functions): Likewise.
10656 (ipa_hsa_write_summary): Likewise.
10657 * symbol-summary.h (symtab_duplication): Use ::get function/
10658 (get): New function.
10660 2018-06-08 Martin Liska <mliska@suse.cz>
10662 * config/i386/i386.c (ix86_can_inline_p): Use get_create instead
10664 * hsa-common.c (hsa_summary_t::link_functions): Likewise.
10665 (hsa_register_kernel): Likewise.
10666 * hsa-common.h (hsa_gpu_implementation_p): Likewise.
10667 * hsa-gen.c (hsa_get_host_function): Likewise.
10668 (get_brig_function_name): Likewise.
10669 (generate_hsa): Likewise.
10670 (pass_gen_hsail::execute): Likewise.
10671 * ipa-cp.c (ipcp_cloning_candidate_p): Likewise.
10672 (devirtualization_time_bonus): Likewise.
10673 (ipcp_propagate_stage): Likewise.
10674 * ipa-fnsummary.c (redirect_to_unreachable): Likewise.
10675 (edge_set_predicate): Likewise.
10676 (evaluate_conditions_for_known_args): Likewise.
10677 (evaluate_properties_for_edge): Likewise.
10678 (ipa_fn_summary::reset): Likewise.
10679 (ipa_fn_summary_t::duplicate): Likewise.
10680 (dump_ipa_call_summary): Likewise.
10681 (ipa_dump_fn_summary): Likewise.
10682 (analyze_function_body): Likewise.
10683 (compute_fn_summary): Likewise.
10684 (estimate_edge_devirt_benefit): Likewise.
10685 (estimate_edge_size_and_time): Likewise.
10686 (estimate_calls_size_and_time): Likewise.
10687 (estimate_node_size_and_time): Likewise.
10688 (inline_update_callee_summaries): Likewise.
10689 (remap_edge_change_prob): Likewise.
10690 (remap_edge_summaries): Likewise.
10691 (ipa_merge_fn_summary_after_inlining): Likewise.
10692 (ipa_update_overall_fn_summary): Likewise.
10693 (read_ipa_call_summary): Likewise.
10694 (inline_read_section): Likewise.
10695 (write_ipa_call_summary): Likewise.
10696 (ipa_fn_summary_write): Likewise.
10697 (ipa_free_fn_summary): Likewise.
10698 * ipa-hsa.c (process_hsa_functions): Likewise.
10699 (ipa_hsa_write_summary): Likewise.
10700 (ipa_hsa_read_section): Likewise.
10701 * ipa-icf.c (sem_function::merge): Likewise.
10702 * ipa-inline-analysis.c (simple_edge_hints): Likewise.
10703 (do_estimate_edge_time): Likewise.
10704 (estimate_size_after_inlining): Likewise.
10705 (estimate_growth): Likewise.
10706 (growth_likely_positive): Likewise.
10707 * ipa-inline-transform.c (clone_inlined_nodes): Likewise.
10708 (inline_call): Likewise.
10709 * ipa-inline.c (caller_growth_limits): Likewise.
10710 (can_inline_edge_p): Likewise.
10711 (can_inline_edge_by_limits_p): Likewise.
10712 (compute_uninlined_call_time): Likewise.
10713 (compute_inlined_call_time): Likewise.
10714 (want_inline_small_function_p): Likewise.
10715 (edge_badness): Likewise.
10716 (update_caller_keys): Likewise.
10717 (update_callee_keys): Likewise.
10718 (recursive_inlining): Likewise.
10719 (inline_small_functions): Likewise.
10720 (inline_to_all_callers_1): Likewise.
10721 (dump_overall_stats): Likewise.
10722 (early_inline_small_functions): Likewise.
10723 (early_inliner): Likewise.
10724 * ipa-inline.h (estimate_edge_growth): Likewise.
10725 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
10726 * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
10727 * ipa-prop.h (IPA_NODE_REF): Likewise.
10728 (IPA_EDGE_REF): Likewise.
10729 * ipa-pure-const.c (malloc_candidate_p): Likewise.
10730 (propagate_malloc): Likewise.
10731 * ipa-split.c (execute_split_functions): Likewise.
10732 * symbol-summary.h: Rename get to get_create.
10734 (get_create): Likewise.
10735 * tree-sra.c (ipa_sra_preliminary_function_checks): Likewise.
10737 2018-06-08 Martin Liska <mliska@suse.cz>
10739 * symbol-summary.h (release): Move definition out of class
10741 (symtab_removal): Likewise.
10742 (symtab_duplication): Likewise.
10744 2018-06-08 Martin Liska <mliska@suse.cz>
10746 * symbol-summary.h (function_summary): Move constructor
10747 implementation out of class declaration.
10748 (release): Likewise.
10749 (symtab_insertion): Likewise.
10750 (symtab_removal): Likewise.
10751 (symtab_duplication): Likewise.
10754 2018-06-08 Martin Liska <mliska@suse.cz>
10756 * Makefile.in: Remove support for MPX (macros, related functions,
10757 fields in cgraph_node, ...).
10758 * builtin-types.def (BT_BND): Likewise.
10759 (BT_FN_BND_CONST_PTR): Likewise.
10760 (BT_FN_CONST_PTR_BND): Likewise.
10761 (BT_FN_VOID_PTR_BND): Likewise.
10762 (BT_FN_BND_CONST_PTR_SIZE): Likewise.
10763 (BT_FN_VOID_CONST_PTR_BND_CONST_PTR): Likewise.
10764 * builtins.c (expand_builtin_memcpy_with_bounds): Likewise.
10765 (expand_builtin_mempcpy_with_bounds): Likewise.
10766 (expand_builtin_memset_with_bounds): Likewise.
10767 (expand_builtin_memset_args): Likewise.
10768 (std_expand_builtin_va_start): Likewise.
10769 (expand_builtin): Likewise.
10770 (expand_builtin_with_bounds): Likewise.
10771 * builtins.def (DEF_BUILTIN_CHKP): Likewise.
10772 (DEF_LIB_BUILTIN_CHKP): Likewise.
10773 (DEF_EXT_LIB_BUILTIN_CHKP): Likewise.
10774 (DEF_CHKP_BUILTIN): Likewise.
10775 (BUILT_IN_MEMCPY): Likewise.
10776 (BUILT_IN_MEMMOVE): Likewise.
10777 (BUILT_IN_MEMPCPY): Likewise.
10778 (BUILT_IN_MEMSET): Likewise.
10779 (BUILT_IN_STPCPY): Likewise.
10780 (BUILT_IN_STRCAT): Likewise.
10781 (BUILT_IN_STRCHR): Likewise.
10782 (BUILT_IN_STRCPY): Likewise.
10783 (BUILT_IN_STRLEN): Likewise.
10784 (BUILT_IN_MEMCPY_CHK): Likewise.
10785 (BUILT_IN_MEMMOVE_CHK): Likewise.
10786 (BUILT_IN_MEMPCPY_CHK): Likewise.
10787 (BUILT_IN_MEMSET_CHK): Likewise.
10788 (BUILT_IN_STPCPY_CHK): Likewise.
10789 (BUILT_IN_STRCAT_CHK): Likewise.
10790 (BUILT_IN_STRCPY_CHK): Likewise.
10791 * calls.c (store_bounds): Likewise.
10792 (emit_call_1): Likewise.
10793 (special_function_p): Likewise.
10794 (maybe_warn_nonstring_arg): Likewise.
10795 (initialize_argument_information): Likewise.
10796 (finalize_must_preallocate): Likewise.
10797 (compute_argument_addresses): Likewise.
10798 (expand_call): Likewise.
10799 * cfgexpand.c (expand_call_stmt): Likewise.
10800 (expand_return): Likewise.
10801 (expand_gimple_stmt_1): Likewise.
10802 (pass_expand::execute): Likewise.
10803 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
10804 (cgraph_node::remove): Likewise.
10805 (cgraph_node::dump): Likewise.
10806 (cgraph_node::verify_node): Likewise.
10807 * cgraph.h (chkp_function_instrumented_p): Likewise.
10808 (symtab_node::get_alias_target): Likewise.
10809 (cgraph_node::can_remove_if_no_direct_calls_and_refs_p): Likewise.
10810 (cgraph_local_p): Likewise.
10811 * cgraphbuild.c (cgraph_edge::rebuild_edges): Likewise.
10812 (cgraph_edge::rebuild_references): Likewise.
10813 * cgraphunit.c (varpool_node::finalize_decl): Likewise.
10814 (walk_polymorphic_call_targets): Likewise.
10815 (cgraph_node::expand_thunk): Likewise.
10816 (symbol_table::output_weakrefs): Likewise.
10817 * common/config/i386/i386-common.c (OPTION_MASK_ISA2_GENERAL_REGS_ONLY_UNSET): Likewise.
10818 (ix86_handle_option): Likewise.
10819 * config/i386/constraints.md: Likewise.
10820 * config/i386/i386-builtin-types.def (BND): Likewise.
10824 * config/i386/i386-builtin.def (BDESC_END): Likewise.
10825 (BDESC_FIRST): Likewise.
10827 * config/i386/i386-c.c (ix86_target_macros_internal): Likewise.
10828 * config/i386/i386-protos.h (ix86_bnd_prefixed_insn_p): Likewise.
10829 * config/i386/i386.c (enum reg_class): Likewise.
10830 (ix86_target_string): Likewise.
10831 (ix86_option_override_internal): Likewise.
10832 (ix86_conditional_register_usage): Likewise.
10833 (ix86_valid_target_attribute_inner_p): Likewise.
10834 (ix86_set_indirect_branch_type): Likewise.
10835 (ix86_set_current_function): Likewise.
10836 (ix86_function_arg_regno_p): Likewise.
10837 (init_cumulative_args): Likewise.
10838 (ix86_function_arg_advance): Likewise.
10839 (ix86_function_arg): Likewise.
10840 (ix86_pass_by_reference): Likewise.
10841 (ix86_function_value_regno_p): Likewise.
10842 (ix86_function_value_1): Likewise.
10843 (ix86_function_value_bounds): Likewise.
10844 (ix86_return_in_memory): Likewise.
10845 (ix86_setup_incoming_vararg_bounds): Likewise.
10846 (ix86_va_start): Likewise.
10847 (indirect_thunk_need_prefix): Likewise.
10848 (print_reg): Likewise.
10849 (ix86_print_operand): Likewise.
10850 (ix86_expand_call): Likewise.
10851 (ix86_output_function_return): Likewise.
10852 (reg_encoded_number): Likewise.
10853 (BDESC_VERIFYS): Likewise.
10854 (ix86_init_mpx_builtins): Likewise.
10855 (ix86_init_builtins): Likewise.
10856 (ix86_emit_cmove): Likewise.
10857 (ix86_emit_move_max): Likewise.
10858 (ix86_expand_builtin): Likewise.
10859 (ix86_builtin_mpx_function): Likewise.
10860 (ix86_get_arg_address_for_bt): Likewise.
10861 (ix86_load_bounds): Likewise.
10862 (ix86_store_bounds): Likewise.
10863 (ix86_load_returned_bounds): Likewise.
10864 (ix86_store_returned_bounds): Likewise.
10865 (ix86_class_likely_spilled_p): Likewise.
10866 (ix86_hard_regno_mode_ok): Likewise.
10867 (x86_order_regs_for_local_alloc): Likewise.
10868 (ix86_mitigate_rop): Likewise.
10869 (ix86_bnd_prefixed_insn_p): Likewise.
10870 (ix86_mpx_bound_mode): Likewise.
10871 (ix86_make_bounds_constant): Likewise.
10872 (ix86_initialize_bounds): Likewise.
10873 (TARGET_LOAD_BOUNDS_FOR_ARG): Likewise.
10874 (TARGET_STORE_BOUNDS_FOR_ARG): Likewise.
10875 (TARGET_LOAD_RETURNED_BOUNDS): Likewise.
10876 (TARGET_STORE_RETURNED_BOUNDS): Likewise.
10877 (TARGET_CHKP_BOUND_MODE): Likewise.
10878 (TARGET_BUILTIN_CHKP_FUNCTION): Likewise.
10879 (TARGET_CHKP_FUNCTION_VALUE_BOUNDS): Likewise.
10880 (TARGET_CHKP_MAKE_BOUNDS_CONSTANT): Likewise.
10881 (TARGET_CHKP_INITIALIZE_BOUNDS): Likewise.
10882 * config/i386/i386.h (TARGET_MPX): Likewise.
10883 (TARGET_MPX_P): Likewise.
10884 (VALID_BND_REG_MODE): Likewise.
10885 (FIRST_BND_REG): Likewise.
10886 (LAST_BND_REG): Likewise.
10887 (enum reg_class): Likewise.
10888 (BND_REG_P): Likewise.
10889 (BND_REGNO_P): Likewise.
10890 (BNDmode): Likewise.
10891 (ADJUST_INSN_LENGTH): Likewise.
10892 * config/i386/i386.md: Likewise.
10893 * config/i386/i386.opt: Likewise.
10894 * config/i386/linux-common.h (LIBMPX_LIBS): Likewise.
10895 (defined): Likewise.
10896 (LINK_MPX): Likewise.
10897 (MPX_SPEC): Likewise.
10898 (LIBMPX_SPEC): Likewise.
10899 (LIBMPXWRAPPERS_SPEC): Likewise.
10900 (CHKP_SPEC): Likewise.
10901 * config/i386/predicates.md: Likewise.
10902 * dbxout.c (dbxout_type): Likewise.
10903 * doc/extend.texi: Likewise.
10904 * doc/invoke.texi: Likewise.
10905 * doc/md.texi: Likewise.
10906 * doc/tm.texi: Likewise.
10907 * doc/tm.texi.in: Likewise.
10908 * dwarf2out.c (is_base_type): Likewise.
10909 (gen_formal_types_die): Likewise.
10910 (gen_subprogram_die): Likewise.
10911 (gen_type_die_with_usage): Likewise.
10912 (gen_decl_die): Likewise.
10913 (dwarf2out_late_global_decl): Likewise.
10914 * expr.c (expand_assignment): Likewise.
10915 (emit_storent_insn): Likewise.
10916 (store_expr_with_bounds): Likewise.
10917 (store_expr): Likewise.
10918 (expand_expr_real_1): Likewise.
10919 * expr.h (store_expr_with_bounds): Likewise.
10920 * function.c (use_register_for_decl): Likewise.
10921 (struct bounds_parm_data): Likewise.
10922 (assign_parms_augmented_arg_list): Likewise.
10923 (assign_parm_find_entry_rtl): Likewise.
10924 (assign_parm_is_stack_parm): Likewise.
10925 (assign_parm_load_bounds): Likewise.
10926 (assign_bounds): Likewise.
10927 (assign_parms): Likewise.
10928 (expand_function_start): Likewise.
10929 * gcc.c (CHKP_SPEC): Likewise.
10930 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
10931 * gimple-ssa-warn-restrict.c (builtin_access::builtin_access): Likewise.
10932 (wrestrict_dom_walker::check_call): Likewise.
10933 * gimple.c (gimple_build_call_from_tree): Likewise.
10934 * gimple.h (enum gf_mask): Likewise.
10935 (gimple_call_with_bounds_p): Likewise.
10936 (gimple_call_set_with_bounds): Likewise.
10937 * gimplify.c (gimplify_init_constructor): Likewise.
10938 * ipa-cp.c (initialize_node_lattices): Likewise.
10939 (propagate_constants_across_call): Likewise.
10940 (find_more_scalar_values_for_callers_subset): Likewise.
10941 * ipa-hsa.c (process_hsa_functions): Likewise.
10942 * ipa-icf-gimple.c (func_checker::compare_gimple_call): Likewise.
10943 * ipa-icf.c (sem_function::merge): Likewise.
10944 * ipa-inline.c (early_inliner): Likewise.
10945 * ipa-pure-const.c (warn_function_noreturn): Likewise.
10946 (warn_function_cold): Likewise.
10947 (propagate_pure_const): Likewise.
10948 * ipa-ref.h (enum GTY): Likewise.
10949 * ipa-split.c (find_retbnd): Likewise.
10950 (consider_split): Likewise.
10951 (split_function): Likewise.
10952 * ipa-visibility.c (cgraph_externally_visible_p): Likewise.
10953 * ipa.c (walk_polymorphic_call_targets): Likewise.
10954 (symbol_table::remove_unreachable_nodes): Likewise.
10955 (process_references): Likewise.
10956 (cgraph_build_static_cdtor_1): Likewise.
10957 * lto-cgraph.c (lto_output_node): Likewise.
10958 (output_refs): Likewise.
10959 (compute_ltrans_boundary): Likewise.
10960 (input_overwrite_node): Likewise.
10961 (input_node): Likewise.
10962 (input_cgraph_1): Likewise.
10963 * params.def (PARAM_CHKP_MAX_CTOR_SIZE): Likewise.
10964 * passes.c (pass_manager::execute_early_local_passes): Likewise.
10965 (class pass_chkp_instrumentation_passes): Likewise.
10966 (make_pass_chkp_instrumentation_passes): Likewise.
10967 * passes.def: Likewise.
10968 * rtl.h (struct GTY): Likewise.
10969 (CALL_EXPR_WITH_BOUNDS_P): Likewise.
10970 * stor-layout.c (layout_type): Likewise.
10971 * symtab.c: Likewise.
10972 * target.def: Likewise.
10973 * targhooks.c (default_chkp_bound_type): Likewise.
10974 (default_chkp_bound_mode): Likewise.
10975 (default_builtin_chkp_function): Likewise.
10976 (default_chkp_function_value_bounds): Likewise.
10977 (default_chkp_make_bounds_constant): Likewise.
10978 (default_chkp_initialize_bounds): Likewise.
10979 * targhooks.h (default_chkp_bound_type): Likewise.
10980 (default_chkp_bound_mode): Likewise.
10981 (default_builtin_chkp_function): Likewise.
10982 (default_chkp_function_value_bounds): Likewise.
10983 (default_chkp_make_bounds_constant): Likewise.
10984 (default_chkp_initialize_bounds): Likewise.
10985 * toplev.c (compile_file): Likewise.
10986 (process_options): Likewise.
10987 * tree-core.h (DEF_BUILTIN): Likewise.
10988 (DEF_BUILTIN_CHKP): Likewise.
10989 * tree-inline.c (declare_return_variable): Likewise.
10990 (remap_gimple_stmt): Likewise.
10991 (copy_bb): Likewise.
10992 (initialize_inlined_parameters): Likewise.
10993 (expand_call_inline): Likewise.
10994 * tree-pass.h (make_pass_ipa_chkp_versioning): Likewise.
10995 (make_pass_ipa_chkp_early_produce_thunks): Likewise.
10996 (make_pass_ipa_chkp_produce_thunks): Likewise.
10997 (make_pass_chkp): Likewise.
10998 (make_pass_chkp_opt): Likewise.
10999 (make_pass_chkp_instrumentation_passes): Likewise.
11000 * tree-pretty-print.c (dump_generic_node): Likewise.
11001 * tree-ssa-ccp.c (insert_clobber_before_stack_restore): Likewise.
11002 * tree-ssa-dce.c (propagate_necessity): Likewise.
11003 (eliminate_unnecessary_stmts): Likewise.
11004 * tree-ssa-pre.c (create_expression_by_pieces): Likewise.
11005 * tree-ssa-sccvn.c (copy_reference_ops_from_call): Likewise.
11006 * tree-ssa-sccvn.h: Likewise.
11007 * tree-ssa-strlen.c (get_string_length): Likewise.
11008 (valid_builtin_call): Likewise.
11009 (adjust_last_stmt): Likewise.
11010 (handle_builtin_strchr): Likewise.
11011 (handle_builtin_strcpy): Likewise.
11012 (handle_builtin_stxncpy): Likewise.
11013 (handle_builtin_memcpy): Likewise.
11014 (handle_builtin_strcat): Likewise.
11015 (strlen_check_and_optimize_stmt): Likewise.
11016 * tree-stdarg.c (expand_ifn_va_arg_1): Likewise.
11017 * tree-streamer-in.c: Likewise.
11018 * tree-streamer.c (record_common_node): Likewise.
11019 * tree.c (tree_code_size): Likewise.
11020 (wide_int_to_tree_1): Likewise.
11021 (type_contains_placeholder_1): Likewise.
11022 (build_common_tree_nodes): Likewise.
11023 * tree.def (POINTER_BOUNDS_TYPE): Likewise.
11024 * tree.h (POINTER_BOUNDS_TYPE_P): Likewise.
11025 (POINTER_BOUNDS_P): Likewise.
11026 (BOUNDED_TYPE_P): Likewise.
11027 (BOUNDED_P): Likewise.
11028 (CALL_WITH_BOUNDS_P): Likewise.
11029 (pointer_bounds_type_node): Likewise.
11030 * value-prof.c (gimple_ic): Likewise.
11031 * var-tracking.c (vt_add_function_parameters): Likewise.
11032 * varasm.c (make_decl_rtl): Likewise.
11033 (assemble_start_function): Likewise.
11034 (output_constant): Likewise.
11035 (maybe_assemble_visibility): Likewise.
11036 * varpool.c (ctor_for_folding): Likewise.
11037 * chkp-builtins.def: Remove.
11038 * ipa-chkp.c: Remove.
11039 * ipa-chkp.h: Remove.
11040 * rtl-chkp.c: Remove.
11041 * rtl-chkp.h: Remove.
11042 * tree-chkp-opt.c: Remove.
11043 * tree-chkp.c: Remove.
11044 * tree-chkp.h: Remove.
11046 2018-06-07 Carl Love <cel@us.ibm.com>
11048 * config/rs6000/vsx.md (vextract_fp_from_shorth,
11049 vextract_fp_from_shortl): Add BE support.
11051 2018-06-07 Paul Koning <ni1d@arrl.net>
11053 * compare-elim.c (try_merge_compare): Don't merge compare if
11054 address contains a side effect.
11055 (try_eliminate_compare): Likewise.
11057 2018-06-07 Olga Makhotina <olga.makhotina@intel.com>
11059 * config.gcc: Support "tremont".
11060 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "tremont".
11061 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
11063 * config/i386/i386.c (m_TREMONT): Define.
11064 (processor_target_table): Add "tremont".
11065 (PTA_TREMONT): Define.
11066 (ix86_lea_outperforms): Add TARGET_TREMONT.
11067 (get_builtin_code_for_version): Handle PROCESSOR_TREMONT.
11068 (fold_builtin_cpu): Add M_INTEL_TREMONT, replace M_INTEL_GOLDMONT
11069 and M_INTEL_GOLDMONT_PLUS.
11070 (fold_builtin_cpu): Add "tremont".
11071 (ix86_add_stmt_cost): Add TARGET_TREMONT.
11072 (ix86_option_override_internal): Add "tremont".
11073 * config/i386/i386.h (processor_costs): Define TARGET_TREMONT.
11074 (processor_type): Add PROCESSOR_TREMONT.
11075 * config/i386/x86-tune.def: Add m_TREMONT.
11076 * doc/invoke.texi: Add tremont as x86 -march=/-mtune= CPU type.
11078 2018-06-07 Jozef Lawrynowicz <jozef.l@mittosystems.com>
11080 * config/msp430/msp430.c (msp430_mcu_name): Set the "i" in the
11081 symbol defined for msp430i* devices to be lower case.
11083 2018-06-07 Richard Biener <rguenther@suse.de>
11085 * graphite-sese-to-poly.c (extract_affine): Avoid unneded
11086 wrapping. Properly wrap the result of a BIT_NOT_EXPR.
11087 Properly wrap signed arithmetic if overflow wraps.
11089 2018-06-07 Jakub Jelinek <jakub@redhat.com>
11091 PR tree-optimization/69615
11092 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If rhs2 is lhs
11093 of a cast from a same precision integral SSA_NAME in a bb dominated
11094 by first_bb, retry with rhs2 set to the rhs1 of the cast. Don't emit
11095 cast to utype if rhs2 has already a compatible type.
11097 2018-06-07 Richard Biener <rguenther@suse.de>
11099 PR tree-optimization/85935
11100 * graphite-scop-detection.c (find_params_in_bb): Analyze
11101 condition operands with respect to the correct loop. Assert
11102 the analysis doesn't fail.
11104 2018-06-04 Carl Love <cel@us.ibm.com>
11106 * config/rs6000/vsx.md (first_match_index_<mode>): Calculate index
11107 using natural element order. Use gen_lshrsi3 instead of gen_ashrsi3
11108 as it is slightly cheaper.
11109 (first_match_or_eos_index_<mode>):
11110 Calculate index using natural element order.
11111 (first_match_index_<mode>):
11112 Calculate index using natural element order.
11113 (first_match_or_eos_index_<mode>):
11114 Calculate index using natural order.
11115 (define_insn vclzlsbb): Change to define_insn vclzlsbb_<mode>.
11116 for BE and LE modes.
11117 * config/rs6000/rs6000-c.c: Rename P9V_BUILTIN_VCLZLSBB,
11118 P9V_BUILTIN_VCLZLSBB_V16QI.
11119 * config/rs6000/rs6000-builtin.def: Make VCLZLSBB mode
11122 2018-06-06 Kelvin Nilsen <kelvin@gcc.gnu.org>
11124 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
11125 indentation and line wrap for many prototypes. Add missing
11126 @smallexample directives around block of prototypes for vec_xl and
11129 2018-06-05 Michael Meissner <meissner@linux.ibm.com>
11131 * config/rs6000/rs6000.c (rs6000_passes_ieee128): New boolean to
11132 track if we pass or return IEEE 128-bit floating point.
11133 (ieee128_mangling_gcc_8_1): New boolean to say whether to generate
11134 C++ mangling that is compatible with GCC 8.1.
11135 (TARGET_ASM_GLOBALIZE_DECL_NAME): Override target hook.
11136 (init_cumulative_args): Note if we pass or return IEEE 128-bit
11137 floating point types.
11138 (rs6000_function_arg_advance_1): Likewise.
11139 (rs6000_mangle_type): Optionally generate mangled names that match
11140 what GCC 8.1 generated for IEEE 128-bit floating point types.
11141 (rs6000_globalize_decl_name): If we have an external function that
11142 passes or returns IEEE 128-bit types, generate a weak reference
11143 from the mangled name used in GCC 8.1 to the current mangled
11145 (rs6000_init_builtins): Make __ibm128 use the long double type if
11146 long double is IBM extended double. Make __float128 use the long
11147 double type if long double is IEEE 128-bit.
11150 * config/rs6000/rs6000-builtin.def (BU_IBM128_2): New helper
11151 macro for __ibm128 built-in functions.
11152 (PACK_IF): Add __ibm128 pack/unpack functions.
11153 (UNPACK_IF): Likewise.
11154 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Do not
11155 enable long double built-in functions if long double is IEEE
11156 128-bit floating point.
11157 (rs6000_invalid_builtin): Update long double built-in function
11159 (rs6000_expand_builtin): For PACK_IF and UNPACK_IF built-in
11160 functions, adjust the built-in function to use the long double
11161 built-in function if __ibm128 and long double are the same type.
11162 * doc/extend.texi (PowerPC builtins): Update documention for
11163 __builtin_{,un}pack_longdouble. Add documentation for
11164 __builtin_{,un}pack_ibm128.
11166 2018-06-06 Jim Wilson <jimw@sifive.com>
11168 * config/riscv/riscv.c (enum riscv_privilege_levels): New.
11169 (struct machine_function): New field interrupt_mode.
11170 (riscv_handle_type_attribute): New function. Add forward declaration.
11171 (riscv_attribute_table) <interrupt>: Use riscv_handle_type_attribute.
11172 (riscv_expand_epilogue): Check interrupt_mode field.
11173 (riscv_set_current_function): Check interrupt attribute args and
11174 set interrupt_mode field.
11175 * config/riscv/riscv.md (UNSPECV_SRET, UNSPECV_URET): New.
11176 (riscv_sret, riscv_uret): New.
11177 * doc/extend.texi (RISC-V Function Attributes) <interrupt>: Document
11178 new arguments to interrupt attribute.
11180 2018-06-06 Peter Bergner <bergner@vnet.ibm.com>
11183 * /config/rs6000/rs6000.h (ASM_CPU_SPEC): Add support for -mpower9.
11184 Don't handle -mcpu=power8 if -mpower9-vector is also used.
11186 2018-06-06 Kelvin Nilsen <kelvin@gcc.gnu.org>
11188 * config/rs6000/rs6000-builtin.def (VSX_BUILTIN_VEC_LD,
11189 VSX_BUILTIN_VEC_ST): Add comment to explain non-traditional uses.
11190 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
11191 several redundant entries.
11193 2018-06-06 David Malcolm <dmalcolm@redhat.com>
11195 * config/i386/i386-protos.h (ix86_expand_call): Strengthen return
11196 type from "rtx" to "rtx_insn *".
11197 * config/i386/i386.c (ix86_expand_split_stack_prologue): Likewise
11198 for local "call_insn", removing cast.
11199 (ix86_expand_call): Likewise, introducing a "call_insn" local.
11201 2018-06-06 Eric Botcazou <ebotcazou@adacore.com>
11203 PR tree-optimization/86066
11204 * gimple-ssa-store-merging.c (process_store): Do not bypass BIT_NOT_EXPR
11205 for BIT_INSERT_EXPR stores.
11207 2018-06-06 Richard Biener <rguenther@suse.de>
11209 PR tree-optimization/86062
11210 * tree-ssa-sccvn.c (vn_reference_lookup_3): Handle arbitrary
11211 component refs ontop
11212 of to be offsetted base.
11214 2018-06-06 Jozef Lawrynowicz <jozef.l@mittosystems.com>
11216 * config/msp430/msp430.c (msp430_attr): Allow interrupt handlers
11217 to be static and remove check on interrupt attribute name.
11219 2018-06-05 Kelvin Nilsen <kelvin@gcc.gnu.org>
11221 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Remove
11222 volatile qualifier from vec_lvsl and vec_lvsr argument prototypes.
11224 2018-06-05 Steve Ellcey <sellcey@cavium.com>
11227 * config/aarch64/aarch64-protos.h (aarch64_err_no_fpadvsimd): Remove
11229 * config/aarch64/aarch64-protos..c (aarch64_err_no_fpadvsimd):
11230 Remove second argument, change how error is called.
11231 (aarch64_layout_arg): Remove second argument from
11232 aarch64_err_no_fpadvsimd call.
11233 (aarch64_init_cumulative_args): Ditto.
11234 (aarch64_gimplify_va_arg_expr): Ditto.
11235 * config/aarch64/aarch64.md (mov<mode>): Ditto.
11237 2018-06-05 Uros Bizjak <ubizjak@gmail.com>
11239 * config/i386/i386.md (simple_return_indirect_internal): New expander.
11240 (*simple_return_indirect_internal<mode>): Rename from
11241 simple_return_indirect_internal. Use W mode iterator.
11242 (rstorssp): New expander.
11243 (*rstorssp<mode>): Rename from rstorssp. Use P mode iterator.
11244 (clrssbsy): New expander.
11245 (*clrssbsy<mode>): Rename from clrssbsy. Use P mode iterator.
11247 2018-06-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
11249 * config/arm/arm_cmse.h (cmse_nsfptr_create): Change typeof to
11251 (cmse_check_pointed_object): Likewise.
11253 2018-06-05 Martin Liska <mliska@suse.cz>
11255 PR gcov-profile/47618
11256 * doc/invoke.texi: Document how -fprofile-dir format
11259 2018-06-05 Richard Biener <rguenther@suse.de>
11261 * tree-cfgcleanup.c (cleanup_control_flow_pre): For edge
11262 removal pretend DOM info isn't available so we do not update
11263 it and only remove edges, not dominated blocks. Actually free
11264 DOM info in case we removed something. Remove unreachable blocks.
11265 (mfb_keep_latches): Work with either DOM info or marked backedges.
11266 (cleanup_tree_cfg_noloop): Do not remove unreachable blocks
11267 first. Mark backedges if DOM info isn't available.
11268 (Re-)compute DOM info after cleanup_control_flow_pre.
11270 2018-06-05 Richard Biener <rguenther@suse.de>
11272 * tree-cfg.c (struct locus_discrim_map): Store line, not location.
11273 (locus_discrim_hasher::hash): Adjust.
11274 (locus_discrim_hasher::equal): Likewise.
11275 (next_discriminator_for_locus): Work on line directly.
11276 (same_line_p): Pass in expanded locus1 as well.
11277 (assign_discriminators): Avoid redundant location expansions.
11279 2018-06-05 Richard Biener <rguenther@suse.de>
11281 PR tree-optimization/86046
11282 * tree-ssa.c (maybe_optimize_var): Clear DECL_GIMPLE_REG_P
11283 if required after clearing TREE_ADDRESSABLE.
11285 2018-06-05 Richard Biener <rguenther@suse.de>
11287 PR tree-optimization/86047
11288 * tree-ssa-loop.c (for_each_index): Glob handling of all
11289 decls and constants and really handle all of them.
11291 2018-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11294 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
11295 qualifier_void_pointer and qualifier_const_void_pointer.
11296 (arm_ldc_qualifiers, arm_stc_qualifiers): Use the above.
11297 (arm_init_builtins): Handle the above.
11298 * config/arm/arm_acle.h (__arm_cdp, __arm_ldc, __arm_ldcl, __arm_stc,
11299 __arm_stcl, __arm_mcr, __arm_cdp2, __arm_ldc2, __arm_ldcl2, __arm_stc2,
11300 __arm_stcl2,__arm_mcr2, __arm_mcrr, __arm_mcrr2): Remove return for
11303 2018-06-05 Martin Liska <mliska@suse.cz>
11305 * auto-profile.c (read_autofdo_file): Do not use
11306 gcov_ctr_summary struct.
11307 (afdo_callsite_hot_enough_for_early_inline): Likewise.
11308 * coverage.c (struct counts_entry): Likewise.
11309 (read_counts_file): Read just single summary entry.
11310 (get_coverage_counts): Use gcov_summary struct.
11311 * coverage.h (get_coverage_counts): Likewise.
11312 * gcov-dump.c (dump_working_sets): Likewise.
11313 (tag_summary): Dump just single summary.
11314 * gcov-io.c (gcov_write_summary): Write just histogram
11316 (gcov_read_summary): Read just single summary.
11317 (compute_working_sets): Use gcov_summary struct.
11318 * gcov-io.h (GCOV_TAG_SUMMARY_LENGTH): Remove usage
11319 of GCOV_COUNTERS_SUMMABLE.
11320 (GCOV_COUNTERS_SUMMABLE): Remove.
11321 (GCOV_FIRST_VALUE_COUNTER): Replace with
11322 GCOV_COUNTER_V_INTERVAL.
11323 (struct gcov_ctr_summary): Remove.
11324 (struct gcov_summary): Directly use fields of former
11326 (compute_working_sets): Use gcov_summary struct.
11327 * gcov.c (read_count_file): Do not use ctrs fields.
11328 * lto-cgraph.c (merge_profile_summaries): Use gcov_summary
11330 * lto-streamer.h (struct GTY): Make profile_info gcov_summary
11332 * profile.c: Likewise.
11333 * profile.h: Likewise.
11335 2018-06-05 Martin Liska <mliska@suse.cz>
11337 PR gcov-profile/84846
11338 * gcov.c (output_lines): Print working directory only
11339 in intermediate format.
11341 2018-06-05 Andreas Krebbel <krebbel@linux.ibm.com>
11343 * config/s390/s390-builtin-types.def: Add void function type.
11344 * config/s390/s390-builtins.def: Use the function type for the
11347 2018-06-04 Jim Wilson <jimw@sifive.com>
11349 * config/riscv/riscv-protos.h (riscv_expand_epilogue): Change bool arg
11351 * config/riscv/riscv.c (riscv_for_each_saved_reg): New args epilogue
11352 and maybe_eh_return. Change regno to unsigned int. Use new args to
11353 handle EH_RETURN_DATA_REGNO registers properly.
11354 (riscv_expand_prologue): Pass new args to riscv_for_each_saved_reg.
11355 (riscv_expand_epilogue): Update comment. Change argument name and
11356 type. Update code to use new name and type. Pass new args to
11357 riscv_for_each_saved_reg. Only use EH_RETURN_STACKADJ_RTX when
11359 * config/riscv/riscv.md (NORMAL_RETURN): New.
11360 (SIBCALL_RETURN, EXCEPTION_RETURN): New.
11361 (epilogue, sibcall_epilogue): Update riscv_expand_epilogue arg.
11362 (eh_return): Call gen_eh_return_internal and emit barrier.
11363 (eh_return_internal): Call riscv_expand_epilogue.
11365 2018-06-04 Eric Botcazou <ebotcazou@adacore.com>
11367 * gimple-ssa-store-merging.c (struct merged_store_group): Move up
11368 bit_insertion field and declare can_be_merged_into method.
11369 (merged_store_group::can_be_merged_into): New method.
11370 (imm_store_chain_info::coalesce_immediate): Call it to decide whether
11371 consecutive non-overlapping stores can be merged. Turn MEM_REF stores
11372 into BIT_INSERT_EXPR stores if the group contains a non-MEM_REF store.
11374 2018-06-04 Richard Biener <rguenther@suse.de>
11376 PR tree-optimization/85955
11377 * builtins.c (fold_builtin_sincos): Convert pointers to
11378 destination to appropriate type before dereferencing.
11380 2018-06-04 Segher Boessenkool <segher@kernel.crashing.org>
11382 * config/rs6000/rs6000.md (abs<mode>2 for FLOAT128): Handle IFmode.
11384 2018-06-04 Richard Sandiford <richard.sandiford@linaro.org>
11386 * expr.c (expand_expr_real_1): Force the operand into memory if
11387 its TYPE_MODE is BLKmode and if there is no integer mode for
11388 the number of bits being extracted.
11390 2018-06-04 Jakub Jelinek <jakub@redhat.com>
11394 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>_1):
11395 Use vptestnm rather than vptestm in (=Yc,v,C) variant.
11397 2018-06-04 Richard Biener <rguenther@suse.de>
11399 * tree-cfgcleanup.c (cleanup_tree_cfg_1): Fold into...
11400 (cleanup_tree_cfg_noloop): ... single caller. Do
11401 start_recording_case_labels later.
11403 2018-06-04 Sebastian Peryt <sebastian.peryt@intel.com>
11405 * config/i386/cldemoteintrin.h: Change define from _X86INTRIN_H_INCLUDED
11406 to _IMMINTRIN_H_INCLUDED.
11407 * config/i386/pconfigintrin.h: Ditto.
11408 * config/i386/waitpkgintrin.h: Ditto.
11409 * config/i386/immintrin.h: Add includes for sgxintrin.h,
11410 pconfigintrin.h, waitpkgintrin.h and cldemoteintrin.h.
11411 * config/i386/x86intrin.h: Remove includes for mintrin.h, xmmintrin.h,
11412 emmintrin.h, pmmintrin.h, tmmintrin.h, smmintrin.h, wmmintrin.h,
11413 bmiintrin.h, bmi2intrin.h, lzcntintrin.h, sgxintrin.h, pconfigintrin.h,
11414 waitpkgintrin.h and cldemoteintrin.h.
11416 2018-06-04 Richard Biener <rguenther@suse.de>
11418 PR tree-optimization/86038
11419 * tracer.c (find_best_successor): Check probability for
11420 being initialized, bail out if not.
11422 2018-06-04 Richard Earnshaw <rearnsha@arm.com>
11425 * config/arm/arm-cpus.in (ALL_QUIRKS): Add xscale feature to the list
11426 of bits to ignore when comparing architectures.
11428 2018-06-04 Jakub Jelinek <jakub@redhat.com>
11430 PR tree-optimization/69615
11431 * fold-const.c (merge_ranges): If range1 is - [x, x] and x is the
11432 maximum or minimum of the type, try to merge it also as if
11433 range1 is + [-, x - 1] or + [x + 1, -].
11436 * tree.c (inchash::add_expr): Handle IDENTIFIER_NODE.
11438 2018-06-03 Eric Botcazou <ebotcazou@adacore.com>
11440 PR tree-optimization/86034
11441 * gimple-ssa-store-merging.c (output_merged_store): Convert the RHS to
11442 the unsigned bitfield type in a bit insertion sequence if it does not
11443 have a larger precision than the bitfield size.
11444 (process_store): Also bypass widening conversions for BIT_INSERT_EXPR.
11446 2018-06-03 Kito Cheng <kito.cheng@gmail.com>
11448 * config/nds32/nds32-peephole2.md: Add new patterns for code size.
11450 2018-06-03 Chung-Ju Wu <jasonwucj@gmail.com>
11452 * config/nds32/nds32-opts.h (nds32_arch_type): Add ARCH_V3J.
11453 * config/nds32/nds32.c (nds32_option_override): Consider ARCH_V3J.
11454 * config/nds32/nds32.h (TARGET_ISA_V3): Add ARCH_V3J checking.
11455 * config/nds32/nds32.opt (march): Add enum value Value(ARCH_V3J).
11457 2018-06-03 Chung-Ju Wu <jasonwucj@gmail.com>
11459 * common/config/nds32/nds32-common.c (nds32_option_optimization_table):
11460 Disable -fdelete-null-pointer-checks for ELF toolchain.
11462 2018-06-02 Chung-Ju Wu <jasonwucj@gmail.com>
11463 Kito Cheng <kito.cheng@gmail.com>
11465 * config.gcc (nds32*): Use nds32-linux.opt and nds32-elf.opt.
11466 (nds32le-*-*, nds32be-*-*): Integrate checking process.
11467 (nds32*-*-*): Add glibc and uclibc conditions.
11468 * common/config/nds32/nds32-common.c (nds32_except_unwind_info): New.
11469 (TARGET_EXCEPT_UNWIND_INFO): Define.
11470 * config/nds32/elf.h: New file.
11471 * config/nds32/linux.h: New file.
11472 * config/nds32/nds32-elf.opt: New file.
11473 * config/nds32/nds32-linux.opt: New file.
11474 * config/nds32/nds32-fp-as-gp.c
11475 (pass_nds32_fp_as_gp::gate): Consider TARGET_LINUX_ABI.
11476 * config/nds32/nds32.c (nds32_conditional_register_usage): Consider
11478 (nds32_asm_file_end): Ditto.
11479 (nds32_print_operand): Ditto.
11480 (nds32_insert_attributes): Ditto.
11481 (nds32_init_libfuncs): New function.
11482 (TARGET_HAVE_TLS): Define.
11483 (TARGET_INIT_LIBFUNCS): Define.
11484 * config/nds32/nds32.h (TARGET_DEFAULT_RELAX): Apply different relax
11486 (TARGET_ELF): Apply different mcmodel setting.
11487 (LINK_SPEC, LIB_SPEC, STARTFILE_SPEC, ENDFILE_SPEC): The content has
11488 been migrated into elf.h and linux.h files.
11489 * config/nds32/nds32.md (add_pc): Consider TARGET_LINUX_ABI.
11490 * config/nds32/nds32.opt (mvh): Consider TARGET_LINUX_ABI.
11491 (mcmodel): The content has been migrated into nds32-elf.opt and
11492 nds32-linux.opt files.
11493 * config/nds32/t-elf: New file.
11494 * config/nds32/t-linux: New file.
11496 2018-06-02 Chung-Ju Wu <jasonwucj@gmail.com>
11497 Shiva Chen <shiva0217@gmail.com>
11499 * config/nds32/constants.md (unspec_volatile_element): Add
11500 UNSPEC_VOLATILE_OMIT_FP_BEGIN and UNSPEC_VOLATILE_OMIT_FP_END.
11501 * config/nds32/nds32-fp-as-gp.c: New implementation of fp_as_gp
11503 * config/nds32/nds32-protos.h (nds32_naked_function_p): Declare.
11504 (make_pass_nds32_fp_as_gp): Declare.
11505 * config/nds32/nds32.c (nds32_register_passes): Add fp_as_gp as one
11507 (nds32_asm_function_end_prologue): Remove unused asm output.
11508 (nds32_asm_function_begin_epilogue): Remove unused asm output.
11509 (nds32_asm_file_start): Output necessary fp_as_gp information.
11510 (nds32_option_override): Adjust register usage.
11511 (nds32_expand_prologue): Consider fp_as_gp situation.
11512 (nds32_expand_prologue_v3push): Consider fp_as_gp situation.
11513 * config/nds32/nds32.md (prologue): Check fp_as_gp_p and naked_p.
11516 (simple_return): Ditto.
11517 (omit_fp_begin): Output special directive for fp_as_gp.
11518 (omit_fp_end): Output special directive for fp_as_gp.
11519 * config/nds32/nds32.opt (mfp-as-gp, mno-fp-as-gp, mforce-fp-as-gp,
11520 mforbid-fp-as-gp): New options.
11522 2018-06-01 Mark Wielaard <mark@klomp.org>
11524 * dwarf2out.c (dwarf2out_finish): Remove generation of
11525 DW_AT_loclists_base.
11527 2018-06-01 Eric Botcazou <ebotcazou@adacore.com>
11529 * gimple-ssa-store-merging.c: Include gimple-fold.h.
11530 (struct store_immediate_info): Document BIT_INSERT_EXPR stores.
11531 (struct merged_store_group): Add bit_insertion field.
11532 (dump_char_array): Use standard hexadecimal format.
11533 (merged_store_group::merged_store_group): Set bit_insertion to false.
11534 (merged_store_group::apply_stores): Use optimal buffer size. Deal
11535 with BIT_INSERT_EXPR stores. Move up code updating the mask and
11536 also print the mask in the dump file.
11537 (pass_store_merging::gate): Minor tweak.
11538 (imm_store_chain_info::coalesce_immediate): Fix wrong association
11539 of stores with groups in dump. Allow coalescing of BIT_INSERT_EXPR
11540 stores with INTEGER_CST stores.
11541 (count_multiple_uses) <BIT_INSERT_EXPR>: New case.
11542 (imm_store_chain_info::output_merged_store): Add try_bitpos variable
11543 and use it throughout. Generate bit insertion sequences if need be.
11544 (pass_store_merging::process_store): Remove redundant condition.
11545 Record stores from a SSA name to a bit-field with BIT_INSERT_EXPR.
11547 2018-06-01 Segher Boessenkool <segher@kernel.crashing.org>
11549 * config/rs6000/rs6000.c (rs6000_mangle_type): Change the mangling of
11550 the 128-bit floating point types. Fix function comment.
11552 2018-06-01 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11554 * config/aarch64/aarch64-simd.md
11555 (aarch64_simd_vec_unpack<su>_lo_<mode>): Use UXTL and SXTL assembler
11557 (aarch64_simd_vec_unpack<su>_hi_<mode>): Use UXTL2 and SXTL2 assembler
11560 2018-06-01 Richard Sandiford <richard.sandiford@linaro.org>
11562 PR tree-optimization/85989
11563 * gimple-ssa-backprop.c (backprop::m_visited_phis): New member
11565 (backprop::intersect_uses): Check it when deciding whether this
11566 is a backedge reference.
11567 (backprop::process_block): Add each phi to m_visited_phis
11568 after visiting it, then clear it at the end.
11570 2018-06-01 Richard Biener <rguenther@suse.de>
11572 * tree-vectorizer.h (vect_dr_stmt): New function.
11573 (vect_get_load_cost): Adjust.
11574 (vect_get_store_cost): Likewise.
11575 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence):
11576 Use vect_dr_stmt instead of DR_SMTT.
11577 (vect_record_base_alignments): Likewise.
11578 (vect_calculate_target_alignment): Likewise.
11579 (vect_compute_data_ref_alignment): Likewise and make static.
11580 (vect_update_misalignment_for_peel): Likewise.
11581 (vect_verify_datarefs_alignment): Likewise.
11582 (vector_alignment_reachable_p): Likewise.
11583 (vect_get_data_access_cost): Likewise. Pass down
11584 vinfo to vect_get_load_cost/vect_get_store_cost instead of DR.
11585 (vect_get_peeling_costs_all_drs): Likewise.
11586 (vect_peeling_hash_get_lowest_cost): Likewise.
11587 (vect_enhance_data_refs_alignment): Likewise.
11588 (vect_find_same_alignment_drs): Likewise.
11589 (vect_analyze_data_refs_alignment): Likewise.
11590 (vect_analyze_group_access_1): Likewise.
11591 (vect_analyze_group_access): Likewise.
11592 (vect_analyze_data_ref_access): Likewise.
11593 (vect_analyze_data_ref_accesses): Likewise.
11594 (vect_vfa_segment_size): Likewise.
11595 (vect_small_gap_p): Likewise.
11596 (vectorizable_with_step_bound_p): Likewise.
11597 (vect_prune_runtime_alias_test_list): Likewise.
11598 (vect_analyze_data_refs): Likewise.
11599 (vect_supportable_dr_alignment): Likewise.
11600 * tree-vect-loop-manip.c (get_misalign_in_elems): Likewise.
11601 (vect_gen_prolog_loop_niters): Likewise.
11602 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
11603 * tree-vect-patterns.c (vect_recog_bool_pattern): Do not
11605 (vect_recog_mask_conversion_pattern): Likewise.
11606 (vect_try_gather_scatter_pattern): Likewise.
11607 * tree-vect-stmts.c (vect_model_store_cost): Pass stmt_info
11608 to vect_get_store_cost.
11609 (vect_get_store_cost): Get stmt_info instead of DR.
11610 (vect_model_load_cost): Pass stmt_info to vect_get_load_cost.
11611 (vect_get_load_cost): Get stmt_info instead of DR.
11613 2018-06-01 Richard Biener <rguenther@suse.de>
11615 PR middle-end/86017
11616 * gimple-fold.c (var_decl_component_p): Also allow offsetted
11617 vars wrapped in MEM_REFs.
11619 2018-06-01 Richard Sandiford <richard.sandiford@linaro.org>
11621 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
11622 Fix subreg tests so that we only return a choice between
11623 GENERAL_REGS and FP_REGS if the original classes included both.
11625 2018-06-01 Richard Biener <rguenther@suse.de>
11628 * tree-ssa-structalias.c (get_function_part_constraint):
11629 Handle NULL fi->decl.
11630 (find_func_aliases_for_call): Properly handle indirect
11631 fi from direct call.
11632 (find_func_clobbers): Likewise.
11633 (ipa_pta_execute): Likewise.
11634 (create_variable_info_for): For functions that are ifunc_resolver
11635 resolve to a varinfo that contains the result of the resolver call.
11636 (associate_varinfo_to_alias): Do not treat ifunc resolvers as
11639 2018-05-31 Michael Collison <michael.collison@arm.com>
11641 * config/aarch64/aarch64.md:
11642 (*fix_to_zero_extenddfdi2): New pattern.
11643 * gcc.target/aarch64/fix_extend1.c: New testcase.
11645 2018-05-31 Qing Zhao <qing.zhao@oracle.com>
11647 PR middle-end/78809
11648 PR middle-end/83026
11649 * builtins.c (expand_builtin): Add the handling of BUILT_IN_STRCMP_EQ
11650 and BUILT_IN_STRNCMP_EQ.
11651 * builtins.def: Add new builtins BUILT_IN_STRCMP_EQ and
11652 BUILT_IN_STRNCMP_EQ.
11653 * gimple-fold.c (gimple_fold_builtin_string_compare): Add the
11654 handling of BUILTIN_IN_STRCMP_EQ and BUILT_IN_STRNCMP_EQ.
11655 (gimple_fold_builtin): Likewise.
11656 * tree-ssa-strlen.c (compute_string_length): New function.
11657 (determine_min_obsize): New function.
11658 (handle_builtin_string_cmp): New function to handle calls to
11659 string compare functions.
11660 (strlen_optimize_stmt): Add handling to builtin string compare
11662 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
11663 Add the handling of BUILT_IN_STRCMP_EQ and BUILT_IN_STRNCMP_EQ.
11664 * tree.c (build_common_builtin_nodes): Add new defines of
11665 BUILT_IN_STRNCMP_EQ and BUILT_IN_STRCMP_EQ.
11667 2018-05-31 Jakub Jelinek <jakub@redhat.com>
11670 * bb-reorder.c (pass_partition_blocks::gate): Return false for
11671 functions with naked attribute.
11673 2018-05-31 Uros Bizjak <ubizjak@gmail.com>
11675 * config/i386/sse.md (avx_vec_concat<mode>):
11676 Substitute concat_tg_mode mode attribute with xtg_mode.
11677 (<mask_codefor>avx512dq_broadcast<mode><mask_name>_1): Ditto.
11678 (concat_tg_mode): Remove mode attribute.
11680 2018-05-31 Martin Sebor <msebor@redhat.com>
11683 * calls.c (alloc_max_size): Correct a logic error/typo.
11684 Treat excessive arguments as infinite. Warn for invalid arguments.
11685 * doc/invoke.texi (-Walloc-size-larger-than): Update.
11687 2018-05-31 H.J. Lu <hongjiu.lu@intel.com>
11690 * config/i386/x86-tune.def: Re-enable partial_reg_dependency
11691 and movx for Haswell.
11693 2018-05-31 Chung-Lin Tang <cltang@codesourcery.com>
11694 Cesar Philippidis <cesar@codesourcery.com>
11696 PR middle-end/85879
11697 * gimplify.c (gimplify_adjust_omp_clauses): Add 'remove = true'
11698 when emitting error on private/firstprivate reductions.
11699 * omp-low.c (lower_omp_target): Avoid reference-type processing
11700 on pointers for firstprivate clause.
11702 2018-05-31 Sameera Deshpande <sameera.deshpande@linaro.org>
11704 * config/aarch64/aarch64-simd-builtins.def (ld1x3): New.
11707 * config/aarch64/aarch64-simd.md
11708 (aarch64_ld1x3<VALLDIF:mode>): New pattern.
11709 (aarch64_ld1_x3_<mode>): Likewise
11710 (aarch64_st1x2<VALLDIF:mode>): Likewise
11711 (aarch64_st1_x2_<mode>): Likewise
11712 (aarch64_st1x3<VALLDIF:mode>): Likewise
11713 (aarch64_st1_x3_<mode>): Likewise
11714 * config/aarch64/arm_neon.h (vld1_u8_x3): New function.
11715 (vld1_s8_x3): Likewise.
11716 (vld1_u16_x3): Likewise.
11717 (vld1_s16_x3): Likewise.
11718 (vld1_u32_x3): Likewise.
11719 (vld1_s32_x3): Likewise.
11720 (vld1_u64_x3): Likewise.
11721 (vld1_s64_x3): Likewise.
11722 (vld1_f16_x3): Likewise.
11723 (vld1_f32_x3): Likewise.
11724 (vld1_f64_x3): Likewise.
11725 (vld1_p8_x3): Likewise.
11726 (vld1_p16_x3): Likewise.
11727 (vld1_p64_x3): Likewise.
11728 (vld1q_u8_x3): Likewise.
11729 (vld1q_s8_x3): Likewise.
11730 (vld1q_u16_x3): Likewise.
11731 (vld1q_s16_x3): Likewise.
11732 (vld1q_u32_x3): Likewise.
11733 (vld1q_s32_x3): Likewise.
11734 (vld1q_u64_x3): Likewise.
11735 (vld1q_s64_x3): Likewise.
11736 (vld1q_f16_x3): Likewise.
11737 (vld1q_f32_x3): Likewise.
11738 (vld1q_f64_x3): Likewise.
11739 (vld1q_p8_x3): Likewise.
11740 (vld1q_p16_x3): Likewise.
11741 (vld1q_p64_x3): Likewise.
11742 (vst1_s64_x2): Likewise.
11743 (vst1_u64_x2): Likewise.
11744 (vst1_f64_x2): Likewise.
11745 (vst1_s8_x2): Likewise.
11746 (vst1_p8_x2): Likewise.
11747 (vst1_s16_x2): Likewise.
11748 (vst1_p16_x2): Likewise.
11749 (vst1_s32_x2): Likewise.
11750 (vst1_u8_x2): Likewise.
11751 (vst1_u16_x2): Likewise.
11752 (vst1_u32_x2): Likewise.
11753 (vst1_f16_x2): Likewise.
11754 (vst1_f32_x2): Likewise.
11755 (vst1_p64_x2): Likewise.
11756 (vst1q_s8_x2): Likewise.
11757 (vst1q_p8_x2): Likewise.
11758 (vst1q_s16_x2): Likewise.
11759 (vst1q_p16_x2): Likewise.
11760 (vst1q_s32_x2): Likewise.
11761 (vst1q_s64_x2): Likewise.
11762 (vst1q_u8_x2): Likewise.
11763 (vst1q_u16_x2): Likewise.
11764 (vst1q_u32_x2): Likewise.
11765 (vst1q_u64_x2): Likewise.
11766 (vst1q_f16_x2): Likewise.
11767 (vst1q_f32_x2): Likewise.
11768 (vst1q_f64_x2): Likewise.
11769 (vst1q_p64_x2): Likewise.
11770 (vst1_s64_x3): Likewise.
11771 (vst1_u64_x3): Likewise.
11772 (vst1_f64_x3): Likewise.
11773 (vst1_s8_x3): Likewise.
11774 (vst1_p8_x3): Likewise.
11775 (vst1_s16_x3): Likewise.
11776 (vst1_p16_x3): Likewise.
11777 (vst1_s32_x3): Likewise.
11778 (vst1_u8_x3): Likewise.
11779 (vst1_u16_x3): Likewise.
11780 (vst1_u32_x3): Likewise.
11781 (vst1_f16_x3): Likewise.
11782 (vst1_f32_x3): Likewise.
11783 (vst1_p64_x3): Likewise.
11784 (vst1q_s8_x3): Likewise.
11785 (vst1q_p8_x3): Likewise.
11786 (vst1q_s16_x3): Likewise.
11787 (vst1q_p16_x3): Likewise.
11788 (vst1q_s32_x3): Likewise.
11789 (vst1q_s64_x3): Likewise.
11790 (vst1q_u8_x3): Likewise.
11791 (vst1q_u16_x3): Likewise.
11792 (vst1q_u32_x3): Likewise.
11793 (vst1q_u64_x3): Likewise.
11794 (vst1q_f16_x3): Likewise.
11795 (vst1q_f32_x3): Likewise.
11796 (vst1q_f64_x3): Likewise.
11797 (vst1q_p64_x3): Likewise.
11799 2018-05-30 Jozef Lawrynowicz <jozef.l@mittosystems.com>
11801 * config/msp430/msp430.c (msp430_output_labelref): Prepend
11802 user_label_prefix to name.
11804 * tree-core.h: Update comment about the format of NAME string
11805 passed to handler in attribute_spec.
11807 * config/msp430/msp430.md: Remove erroneous subreg expression from
11808 zero_extendqisi2 insn pattern. Remove msp430x ISA restriction on
11809 zero_extend{q,h}isi2.
11811 2018-05-30 Borislav Petkov <bp@suse.de>
11813 * doc/extend.texi: Document some architecture specific
11814 constraints and sort entries.
11816 2018-05-30 Martin Sebor <msebor@redhat.com>
11818 PR middle-end/85369
11819 * builtins.c (expand_builtin_stpcpy_1): New function.
11820 (expand_builtin_stpcpy): Call it, and call maybe_warn_nonstring_arg
11821 only if the former succeeds.
11823 2018-05-31 Sameera Deshpande <sameera.deshpande@linaro.org>
11825 * config/aarch64/aarch64-cores.def (saphira) : Add support for ARMv8.4
11828 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
11830 * doc/invoke.texi (-flinker-output): Document
11832 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
11834 * passes.c (ipa_write_summaries): Only modify statements if body
11836 * cgraphunit.c (ipa_passes): Also produce intermeidate code when
11837 incrementally linking.
11838 (ipa_passes): Likewise.
11839 * lto-cgraph.c (lto_output_node): When incrementally linking do not
11840 pass down resolution info.
11841 * common.opt (flag_incremental_link): Update info.
11842 * gcc.c (plugin specs): Turn flinker-output=* to
11843 -plugin-opt=-linker-output-known
11844 * toplev.c (compile_file): Also cut compilation when doing incremental
11846 * flag-types. (enum lto_partition_model): Add
11847 LTO_LINKER_OUTPUT_NOLTOREL.
11848 (invoke.texi): Add -flinker-output docs.
11849 * ipa.c (symbol_table::remove_unreachable_nodes): Handle LTO incremental
11850 link same way as WPA; do not stream in dead initializers.
11852 * dwarf2out.c (dwarf2out_die_ref_for_decl,
11853 darf2out_register_external_decl): Support incremental link.
11855 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
11857 * lto-opts.c (lto_write_options): Skip OPT_dumpdir, OPT_fresolution_.
11859 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
11861 * lto-wrapper.c (debug_objcopy): Add rename parameter; pass
11862 it down to simple_object_copy_lto_debug_sections.
11863 (run_gcc): Determine incremental LTO link time and configure
11864 lto1 into non-wpa mode, disable renaming of debug sections.
11866 2018-05-30 Kelvin Nilsen <kelvin@gcc.gnu.org>
11868 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Remove
11869 descriptions of various incorrectly documented functions.
11871 2018-05-30 Andre Vieira <andre.simoesdiasvieira@arm.com>
11874 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Make
11875 address check not strict.
11877 2018-05-30 Richard Biener <rguenther@suse.de>
11879 PR tree-optimization/85964
11880 * tracer.c (better_p): Drop initialized count check, we only
11881 call the function with initialized counts now.
11882 (find_best_successor): Do find a best edge if one
11883 has uninitialized count.
11884 (find_best_predecessor): Likewise. Do BB frequency check only
11885 if count is initialized.
11887 2017-05-30 Jackson Woodruff <jackson.woodruff@arm.com>
11889 * config/aarch64/aarch64.c (aarch64_host_wide_int_compare): New.
11890 (aarch64_ldrstr_offset_compare): New.
11891 (aarch64_operands_adjust_ok_for_ldpstp): Update to consider all
11892 load/store orderings.
11893 (aarch64_gen_adjusted_ldpstp): Likewise.
11895 2018-05-30 Wilco Dijkstra <wdijkstr@arm.com>
11897 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
11898 Check for subset of GENERAL_REGS and FP_REGS.
11899 * config/aarch64/aarch64-simd.md (aarch64_get_lane): Increase cost of
11902 2018-05-30 Richard Sandiford <richard.sandiford@linaro.org>
11904 * alias.c (adjust_offset_for_component_ref): Use poly_int_tree_p
11905 and wi::to_poly_offset. Add the current offset and then check
11906 whether the sum fits, rather than using an unchecked addition of
11907 a checked term. Check for a shwi rather than a uhwi.
11908 * expr.c (get_bit_range): Use tree_to_poly_uint64.
11909 (store_constructor): Use poly_int_tree_p.
11910 (expand_expr_real_1): Likewise.
11911 * function.c (assign_temp): Likewise.
11912 * fold-const.c (const_binop): Use poly_int_tree_p and
11913 wi::to_poly_offset.
11914 (fold_indirect_ref_1): Likewise. Use multiple_p to attempt an exact
11916 * ipa-icf-gimple.c (func_checker::compare_operand): Use
11917 to_poly_offset for MEM offsets.
11918 * ipa-icf.c (sem_variable::equals): Likewise.
11919 * stor-layout.c (compute_record_mode): Use poly_int_tree_p.
11920 * tree-ssa-sccvn.c (ao_ref_init_from_vn_reference): Use
11921 wi::to_poly_offset for BIT_FIELD_REF offsets.
11922 (vn_reference_maybe_forwprop_address): Use poly_int_tree_p and
11923 wi::to_poly_offset.
11924 * var-tracking.c (emit_note_insn_var_location): Use
11925 tree_to_poly_uint64.
11927 2018-05-29 Jim Wilson <jimw@sifive.com>
11929 * config/riscv/riscv.c (riscv_interrupt_type): Fix comment typo.
11931 2018-05-29 Uros Bizjak <ubizjak@gmail.com>
11934 * config/i386/i386.md (l<rounding_insn><MODEF:mode><SWI48:mode>2):
11935 Enable for TARGET_SSE4_1 and generate rounds{s,d} and cvtts{s,d}2si{,q}
11937 (sse4_1_round<mode>2): Use nonimmediate_operand
11938 for operand 1 predicate.
11940 2018-05-29 Martin Sebor <msebor@redhat.com>
11941 Richard Biener <rguenther@suse.de>
11944 * calls.c (get_size_range): Call determine_value_range instead
11945 of get_value_range..
11946 * tree-vrp.h (determine_value_range): Declared new function.
11947 * tree-vrp.c (determine_value_range_1, determine_value_range): New.
11949 2018-05-29 Richard Biener <rguenther@suse.de>
11951 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Make
11952 sure to use non-pattern stmts for get_earlier_stmt arguments.
11953 * tree-vectorizer.h (get_earlier_stmt): Assert we do not get
11954 called on pattern stmts.
11955 (get_later_stmt): Likewise.
11957 2018-05-29 Martin Liska <mliska@suse.cz>
11959 PR gcov-profile/85759
11960 * doc/gcov.texi: Document GCOV_ERROR_FILE and GCOV_EXIT_AT_ERROR
11963 2018-05-29 Jakub Jelinek <jakub@redhat.com>
11965 * tree-cfg.c (verify_gimple_assign_unary): Add checking for
11967 (verify_gimple_assign_binary): Check TYPE_VECTOR_SUBPARTS for
11971 * tree.def (VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR,
11972 VEC_PACK_FLOAT_EXPR): New tree codes.
11973 * tree-pretty-print.c (op_code_prio): Handle
11974 VEC_UNPACK_FIX_TRUNC_HI_EXPR and VEC_UNPACK_FIX_TRUNC_LO_EXPR.
11975 (dump_generic_node): Handle VEC_UNPACK_FIX_TRUNC_HI_EXPR,
11976 VEC_UNPACK_FIX_TRUNC_LO_EXPR and VEC_PACK_FLOAT_EXPR.
11977 * tree-inline.c (estimate_operator_cost): Likewise.
11978 * gimple-pretty-print.c (dump_binary_rhs): Handle VEC_PACK_FLOAT_EXPR.
11979 * fold-const.c (const_binop): Likewise.
11980 (const_unop): Handle VEC_UNPACK_FIX_TRUNC_HI_EXPR and
11981 VEC_UNPACK_FIX_TRUNC_LO_EXPR.
11982 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
11983 (verify_gimple_assign_binary): Handle VEC_PACK_FLOAT_EXPR.
11984 * cfgexpand.c (expand_debug_expr): Handle VEC_UNPACK_FIX_TRUNC_HI_EXPR,
11985 VEC_UNPACK_FIX_TRUNC_LO_EXPR and VEC_PACK_FLOAT_EXPR.
11986 * expr.c (expand_expr_real_2): Likewise.
11987 * optabs.def (vec_packs_float_optab, vec_packu_float_optab,
11988 vec_unpack_sfix_trunc_hi_optab, vec_unpack_sfix_trunc_lo_optab,
11989 vec_unpack_ufix_trunc_hi_optab, vec_unpack_ufix_trunc_lo_optab): New
11991 * optabs.c (expand_widen_pattern_expr): For
11992 VEC_UNPACK_FIX_TRUNC_HI_EXPR and VEC_UNPACK_FIX_TRUNC_LO_EXPR use
11993 sign from result type rather than operand's type.
11994 (expand_binop_directly): For vec_packu_float_optab and
11995 vec_packs_float_optab allow result type to be different from operand's
11997 * optabs-tree.c (optab_for_tree_code): Handle
11998 VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR and
11999 VEC_PACK_FLOAT_EXPR. Formatting fixes.
12000 * tree-vect-generic.c (expand_vector_operations_1): Handle
12001 VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR and
12002 VEC_PACK_FLOAT_EXPR.
12003 * tree-vect-stmts.c (supportable_widening_operation): Handle
12005 (supportable_narrowing_operation): Handle FLOAT_EXPR.
12006 * config/i386/i386.md (fixprefix, floatprefix): New code attributes.
12007 * config/i386/sse.md (*float<floatunssuffix>v2div2sf2): Rename to ...
12008 (float<floatunssuffix>v2div2sf2): ... this. Formatting fix.
12009 (vpckfloat_concat_mode, vpckfloat_temp_mode, vpckfloat_op_mode): New
12011 (vec_pack<floatprefix>_float_<mode>): New expander.
12012 (vunpckfixt_mode, vunpckfixt_model, vunpckfixt_extract_mode): New mode
12014 (vec_unpack_<fixprefix>fix_trunc_lo_<mode>,
12015 vec_unpack_<fixprefix>fix_trunc_hi_<mode>): New expanders.
12016 * doc/md.texi (vec_packs_float_@var{m}, vec_packu_float_@var{m},
12017 vec_unpack_sfix_trunc_hi_@var{m}, vec_unpack_sfix_trunc_lo_@var{m},
12018 vec_unpack_ufix_trunc_hi_@var{m}, vec_unpack_ufix_trunc_lo_@var{m}):
12020 * doc/generic.texi (VEC_UNPACK_FLOAT_HI_EXPR,
12021 VEC_UNPACK_FLOAT_LO_EXPR): Fix pasto in description.
12022 (VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR,
12023 VEC_PACK_FLOAT_EXPR): Document.
12025 2018-05-29 Richard Biener <rguenther@suse.de>
12027 * tree-vectorizer.h (struct vec_info): Add stmt_vec_infos
12029 (stmt_vec_info_vec): Make pointer.
12030 (init_stmt_vec_info_vec): Remove.
12031 (free_stmt_vec_info_vec): Likewise.
12032 (set_stmt_vec_info_vec): New function.
12033 (free_stmt_vec_infos): Likewise.
12034 (vinfo_for_stmt): Adjust for stmt_vec_info_vec indirection.
12035 (set_vinfo_for_stmt): Likewise.
12036 (get_earlier_stmt): Likewise.
12037 (get_later_stmt): Likewise.
12038 * tree-vectorizer.c (stmt_vec_info_vec): Make pointer.
12039 (vec_info::vec_info): Allocate stmt_vec_infos and set the global.
12040 (vec_info::~vec_info): Free stmt_vec_infos.
12041 (vectorize_loops): Set the global stmt_vec_info_vec to NULL.
12042 Remove old init_stmt_vec_info_vec/free_stmt_vec_info_vec calls.
12043 (pass_slp_vectorize::execute): Likewise.
12044 * tree-vect-stmts.c (init_stmt_vec_info_vec): Remove.
12045 (free_stmt_vec_info_vec): Likewise.
12046 (set_stmt_vec_info_vec): New function.
12047 (free_stmt_vec_infos): Likewise.
12048 * tree-vect-loop.c (_loop_vec_info::~_loop_vec_info): Set
12049 the global stmt_vec_info_vec.
12050 * tree-parloops.c (gather_scalar_reductions): Use
12051 set_stmt_vec_info_vec/free_stmt_vec_infos and maintain a local
12054 2018-05-29 Richard Biener <rguenther@suse.de>
12056 * dominance.c (iterate_fix_dominators): Push/pop TV_DOMINANCE.
12058 2018-05-29 Martin Liska <mliska@suse.cz>
12059 David Malcolm <dmalcolm@redhat.com>
12061 * vec.c (test_reverse): New.
12062 (vec_c_tests): Add new test.
12063 * vec.h (vl_ptr>::reverse): New function.
12065 2018-05-29 Gerald Pfeifer <gerald@pfeifer.com>
12067 * config.gcc: Identify FreeBSD 3.x and 4.x as unsupported.
12069 * config/freebsd-spec.h (FBSD_LIB_SPEC): Only consider FreeBSD 5
12072 2018-05-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
12074 * tree-dump.c (dump_node): Use splay_tree_delete_pointers.
12076 2018-05-28 Richard Biener <rguenther@suse.de>
12078 PR tree-optimization/85933
12079 * tree-vect-data-refs.c (vect_record_base_alignments): Only
12080 look at stmts marked as vectorizable.
12082 2018-05-28 Richard Biener <rguenther@suse.de>
12084 PR tree-optimization/85934
12085 * tree-vect-generic.c (expand_vector_operations_1): Hoist
12086 vector boolean check before scalar optimization.
12088 2018-05-28 Jakub Jelinek <jakub@redhat.com>
12090 * doc/invoke.texi (ARM Options): Use @item instead of @itemx
12093 2018-05-28 Mark Wielaard <mark@klomp.org>
12095 * dwarf2asm.c (dw2_asm_output_delta_uleb128): Add brackets around lab2
12096 if it is an expression containing a minus sign.
12098 2018-05-27 John David Anglin <danglin@gcc.gnu.org>
12100 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Define to 0.
12102 2018-05-27 Paul Koning <ni1d@arrl.net>
12104 * config/pdp11/pdp11.md (truncsihi2): Remove.
12106 2018-05-27 Monk Chiang <sh.chiang04@gmail.com>
12107 Chung-Ju Wu <jasonwucj@gmail.com>
12109 * config/nds32/nds32-intrinsic.md (unaligned_storedi): Modify patterns
12111 (unaligned_store_dw): Ditto.
12112 * config/nds32/nds32-memory-manipulation.c
12113 (nds32_expand_movmemsi_loop_known_size): Refactoring implementation.
12114 (nds32_gen_dup_4_byte_to_word_value): Rename to ...
12115 (nds32_gen_dup_4_byte_to_word_value_aux): ... this.
12116 (emit_setmem_word_loop): Rename to ...
12117 (emit_setmem_doubleword_loop): ... this.
12118 (nds32_gen_dup_4_byte_to_word_value): New function.
12119 (nds32_gen_dup_8_byte_to_double_word_value): New function.
12120 (nds32_expand_setmem_loop): Refine implementation.
12121 (nds32_expand_setmem_loop_v3m): Ditto.
12122 * config/nds32/nds32-multiple.md (unaligned_store_update_base_dw): New
12125 2018-05-27 Chung-Ju Wu <jasonwucj@gmail.com>
12127 * config/nds32/nds32.md (bswapsi2, bswaphi2): New patterns.
12129 2018-05-27 Chung-Ju Wu <jasonwucj@gmail.com>
12131 * config/nds32/nds32.c (nds32_attribute_table): Add "no_prologue".
12132 (nds32_init_machine_status): Initialize machine->attr_naked_p and
12133 machine->attr_no_prologue_p.
12134 (nds32_compute_stack_frame): Check "naked" and "no_prologue" attributes.
12135 (nds32_naked_function_p): Handle "naked" and "no_prologue" attributes.
12136 (nds32_expand_epilogue): Consider attr_naked_p.
12137 (nds32_expand_epilogue_v3pop): Likewise.
12138 (nds32_can_use_return_insn): Likewise.
12139 * config/nds32/nds32.h (machine_function): Add attr_naked_p and
12140 attr_no_prologue_p fields.
12141 * config/nds32/nds32.opt (mret-in-naked-func): New option.
12143 2018-05-27 Jakub Jelinek <jakub@redhat.com>
12146 * config/i386/i386.md (fixunssuffix, floatunssuffix): New code
12148 * config/i386/sse.md
12149 (<floatsuffix>float<sseintvecmodelower><mode>2<mask_name><round_name>):
12151 (float<floatunssuffix><sseintvecmodelower><mode>2<mask_name><round_name>):
12153 (<floatsuffix>float<sselongvecmodelower><mode>2<mask_name><round_name>):
12155 (float<floatunssuffix><sselongvecmodelower><mode>2<mask_name><round_name>):
12157 (*<floatsuffix>floatv2div2sf2): Rename to ...
12158 (*float<floatunssuffix>v2div2sf2): ... this.
12159 (<floatsuffix>floatv2div2sf2_mask): Rename to ...
12160 (float<floatunssuffix>v2div2sf2_mask): ... this.
12161 (*<floatsuffix>floatv2div2sf2_mask_1): Rename to ...
12162 (*float<floatunssuffix>v2div2sf2_mask_1): ... this.
12163 (<fixsuffix>fix_truncv8dfv8si2<mask_name><round_saeonly_name>): Rename
12165 (fix<fixunssuffix>_truncv8dfv8si2<mask_name><round_saeonly_name>):
12167 (<fixsuffix>fix_trunc<mode><sseintvecmodelower>2<mask_name><round_saeonly_name>):
12169 (fix<fixunssuffix>_trunc<mode><sseintvecmodelower>2<mask_name><round_saeonly_name>):
12171 (<fixsuffix>fix_trunc<mode><sselongvecmodelower>2<mask_name><round_saeonly_name>):
12173 (fix<fixunssuffix>_trunc<mode><sselongvecmodelower>2<mask_name><round_saeonly_name>):
12175 (<fixsuffix>fix_truncv2sfv2di2<mask_name>): Rename to ...
12176 (fix<fixunssuffix>_truncv2sfv2di2<mask_name>): ... this.
12177 (vec_pack_ufix_trunc_<mode>): Use gen_fixuns_truncv8dfv8si2 instead of
12178 gen_ufix_truncv8dfv8si2.
12179 * config/i386/i386-builtin.def (__builtin_ia32_cvttpd2uqq256_mask,
12180 __builtin_ia32_cvttpd2uqq128_mask, __builtin_ia32_cvttps2uqq256_mask,
12181 __builtin_ia32_cvttps2uqq128_mask, __builtin_ia32_cvtuqq2ps256_mask,
12182 __builtin_ia32_cvtuqq2ps128_mask, __builtin_ia32_cvtuqq2pd256_mask,
12183 __builtin_ia32_cvtuqq2pd128_mask, __builtin_ia32_cvttpd2udq512_mask,
12184 __builtin_ia32_cvtuqq2ps512_mask, __builtin_ia32_cvtuqq2pd512_mask,
12185 __builtin_ia32_cvttps2uqq512_mask, __builtin_ia32_cvttpd2uqq512_mask):
12186 Use fixuns instead ufix or floatuns instead ufloat in CODE_FOR_ names.
12188 2018-05-24 H.J. Lu <hongjiu.lu@intel.com>
12192 * varasm.c (assemble_alias): Lookup ifunc attribute on error.
12194 2018-05-25 Jim Wilson <jimw@sifive.com>
12196 * config/riscv/riscv-protos.h (riscv_epilogue_uses): New.
12197 * config/riscv/riscv.c (struct machine_function): Add
12198 interrupt_handler_p and attribute_checked_p fields.
12199 (riscv_attribute_table): Add interrupt.
12200 (riscv_interrupt_type_p): New.
12201 (riscv_save_reg_p): Save extra regs for interrupt handler.
12202 (riscv_use_save_libcall): Return false for interrupt handler.
12203 (riscv_first_stack_step): Add forward declaration.
12204 (riscv_compute_frame_info): New local interrupt_save_t1. Set it
12205 for interrupt handler with large frame. Use it for saved reg list.
12206 (riscv_expand_prologue): Move flag_stack_usage_info support to
12207 eliminate duplication.
12208 (riscv_expand_epilogue): Generate mret for interrupt handler.
12209 (riscv_epilogue_uses): New.
12210 (riscv_can_use_return_insn): Return false for interrupt handler.
12211 (riscv_function_ok_for_sibcall): Likewise.
12212 (riscv_set_current_function): Add interrupt handler support.
12213 * config/riscv/riscv.h (EPILOGUE_USES): Call riscv_epilogue_uses.
12214 * config/riscv/riscv.md (UNSPECV_MRET): New.
12216 (riscv_frflags, riscv_fsflags): Use tab after opcode.
12218 * doc/extend.texi (RISC-V Function Attributes) <interrupt>: New.
12220 2018-05-25 Bill Schmidt <wschmidt@linux.ibm.com>
12222 PR tree-optimization/85712
12223 * gimple-ssa-strength-reduction.c (replace_one_candidate): Skip if
12224 this candidate has already been replaced in-situ by a copy.
12226 2018-05-25 Jason Merrill <jason@redhat.com>
12228 PR c++/80485 - inline function non-zero address.
12229 * symtab.c (nonzero_address): Check DECL_COMDAT.
12231 2018-05-25 Uros Bizjak <ubizjak@gmail.com>
12234 * config/alpha/alpha.md (ashlsi3): New insn pattern.
12235 (*ashlsi_se): Rename from *ashldi_se. Define as sign
12236 extension of SImode operation. Use const123_operand predicate.
12237 (*saddsi_1): Remove.
12238 (*saddl_se_1): Ditto.
12239 (*ssubsi_1): Ditto.
12240 (*ssubl_se_1): Ditto.
12241 * config/alpha/predicates.md (const123_operand): New predicate.
12242 * config/alpha/constraints.md (P): Use IN_RANGE.
12244 2018-05-25 Richard Biener <rguenther@suse.de>
12246 * tree-ssa-alias.h (refs_may_alias_p): Add tbaa_p bool parameter,
12248 (ref_maybe_used_by_stmt_p): Likewise.
12249 (stmt_may_clobber_ref_p): Likewise.
12250 (stmt_may_clobber_ref_p_1): Likewise.
12251 * tree-ssa-alias.c (refs_may_alias_p): Add tbaa_p bool parameter
12253 (ref_maybe_used_by_stmt_p): Likewise.
12254 (stmt_may_clobber_ref_p): Likewise.
12255 (stmt_may_clobber_ref_p_1): Likewise.
12256 * tree-vect-data-refs.c (vect_slp_analyze_node_dependences): Use
12257 the alias oracle to disambiguate DRs with stmts DR analysis
12259 (vect_analyze_data_refs): Do not give up on not analyzable
12260 DRs for BB vectorization. Remove code truncating the dataref
12263 2018-05-25 Jakub Jelinek <jakub@redhat.com>
12266 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>_1):
12267 Add (=Yk,v,C) variant using vptestm insn. Use TARGET_AVX512BW
12268 in test instead of TARGET_AVX512F for VI12_AVX512VL iterator.
12270 2018-05-25 Richard Biener <rguenther@suse.de>
12272 * tree-vect-data-refs.c (vect_find_stmt_data_reference): New
12273 function, combining stmt data ref gathering and fatal analysis
12275 (vect_analyze_data_refs): Remove now redudnant code and simplify.
12276 * tree-vect-loop.c (vect_get_datarefs_in_loop): Factor out from
12277 vect_analyze_loop_2 and use vect_find_stmt_data_reference.
12278 * tree-vect-slp.c (vect_slp_bb): Use vect_find_stmt_data_reference.
12279 * tree-vectorizer.h (vect_find_stmt_data_reference): Declare.
12281 2018-05-25 Bin Cheng <bin.cheng@arm.com>
12283 PR tree-optimization/85720
12284 * tree-loop-distribution.c (break_alias_scc_partitions): Don't merge
12285 SCC if all partitions are builtins.
12286 (version_loop_by_alias_check): New parameter. Generate cancelable
12287 runtime alias check if all partitions are builtins.
12288 (distribute_loop): Update call to above function.
12290 2018-05-25 Bin Cheng <bin.cheng@arm.com>
12292 * tree-outof-ssa.c (tree-ssa.h, tree-dfa.h): Include header files.
12293 (create_default_def, for_all_parms): Moved from tree-ssa-coalesce.c.
12294 (parm_default_def_partition_arg): Ditto.
12295 (set_parm_default_def_partition): Ditto.
12296 (get_parm_default_def_partitions): Ditto and make it static.
12297 (get_undefined_value_partitions): Ditto and make it static.
12298 (remove_ssa_form): Refactor call to init_var_map here.
12299 * tree-ssa-coalesce.c (build_ssa_conflict_graph): Support live range
12300 computation for loop region.
12301 (coalesce_partitions, compute_optimized_partition_bases): Ditto.
12302 (register_default_def): Delete.
12303 (for_all_parms, create_default_def): Move to tree-outof-ssa.c.
12304 (parm_default_def_partition_arg): Ditto.
12305 (set_parm_default_def_partition): Ditto.
12306 (get_parm_default_def_partitions): Ditto and make it static.
12307 (get_undefined_value_partitions): Ditto and make it static.
12308 (coalesce_with_default, coalesce_with_default): Update comment.
12309 (create_coalesce_list_for_region): New func factored out from
12310 create_outofssa_var_map.
12311 (populate_coalesce_list_for_outofssa): New func factored out from
12312 create_outofssa_var_map and coalesce_ssa_name.
12313 (create_outofssa_var_map): Delete.
12314 (coalesce_ssa_name): Refactor to support live range computation.
12315 * tree-ssa-coalesce.h (coalesce_ssa_name): Change decl.
12316 (get_parm_default_def_partitions): Delete.
12317 (get_undefined_value_partitions): Ditto.
12318 * tree-ssa-live.c (init_var_map, delete_var_map): Support live range
12319 computation for loop region.
12320 (new_tree_live_info, loe_visit_block): Ditto.
12321 (live_worklist, set_var_live_on_entry): Ditto.
12322 (calculate_live_on_exit, verify_live_on_entry): Ditto.
12323 * tree-ssa-live.h (struct _var_map): New fields.
12324 (init_var_map): Change decl.
12325 (region_contains_p): New.
12327 2018-05-25 Bin Cheng <bin.cheng@arm.com>
12329 * tree-ssa-live.h (live_merge_and_clear): Delete.
12331 2018-05-25 Richard Biener <rguenther@suse.de>
12334 * tree-dump.c (dequeue_and_dump): Remove access to removed
12335 operand 2 of a SWITCH_EXPR.
12337 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
12339 * doc/sourcebuild.texi (vect_double_cond_arith): Include
12340 multiplication and division.
12341 * doc/md.texi (cond_mul@var{m}, cond_div@var{m}, cond_mod@var{m})
12342 (cond_udiv@var{m}, cond_umod@var{m}): Document.
12343 * optabs.def (cond_smul_optab, cond_sdiv_optab, cond_smod_optab)
12344 (cond_udiv_optab, cond_umod_optab): New optabs.
12345 * internal-fn.def (IFN_COND_MUL, IFN_COND_DIV, IFN_COND_MOD)
12346 (IFN_COND_RDIV): New internal functions.
12347 * internal-fn.c (get_conditional_internal_fn): Handle TRUNC_DIV_EXPR,
12348 TRUNC_MOD_EXPR and RDIV_EXPR.
12349 * match.pd (UNCOND_BINARY, COND_BINARY): Handle them.
12350 * config/aarch64/iterators.md (UNSPEC_COND_MUL, UNSPEC_COND_DIV):
12352 (SVE_INT_BINARY): Include mult.
12353 (SVE_COND_FP_BINARY): Include UNSPEC_MUL and UNSPEC_DIV.
12354 (optab, sve_int_op): Handle mult.
12355 (optab, sve_fp_op, commutative): Handle UNSPEC_COND_MUL and
12357 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): New pattern
12358 for SVE_INT_BINARY_SD.
12360 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
12362 * config/aarch64/iterators.md (SVE_INT_BINARY_SD): New code iterator.
12363 (optab, sve_int_op): Handle div and udiv.
12364 * config/aarch64/aarch64-sve.md (<optab><mode>3): New expander
12365 for SVE_INT_BINARY_SD.
12366 (*<optab><mode>3): New insn for the same.
12368 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
12370 * tree-vect-patterns.c: Include predict.h.
12371 (vect_recog_divmod_pattern): Restrict check for division support
12372 to when optimizing for size.
12374 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
12376 * doc/sourcebuild.texi (vect_double_cond_arith: Document.
12377 * gimple-match.h (gimple_match_op::MAX_NUM_OPS): Bump to 4.
12378 (gimple_match_op::gimple_match_op): Add an overload for 4 operands.
12379 (gimple_match_op::set_op): Likewise.
12380 (gimple_resimplify4): Declare.
12381 * genmatch.c (get_operand_type): Handle CFN_COND_* functions.
12382 (expr::gen_transform): Likewise.
12383 (decision_tree::gen): Generate a simplification routine for 4 operands.
12384 * gimple-match-head.c (gimple_simplify): Add an overload for
12385 4 operands. In the top-level function, handle up to 4 call
12386 arguments and call gimple_resimplify4.
12387 (gimple_resimplify4): New function.
12388 (build_call_internal): Pass a fourth operand.
12389 (maybe_push_to_seq): Likewise.
12390 * match.pd (UNCOND_BINARY, COND_BINARY): New operator lists.
12391 Fold VEC_COND_EXPRs of an operation and a default value into
12392 an IFN_COND_* function if possible.
12393 * config/aarch64/iterators.md (UNSPEC_COND_MAX, UNSPEC_COND_MIN):
12395 (SVE_COND_FP_BINARY): Include them.
12396 (optab, sve_fp_op): Handle them.
12397 (SVE_INT_BINARY_REV): New code iterator.
12398 (SVE_COND_FP_BINARY_REV): New int iterator.
12399 (commutative): New int attribute.
12400 * config/aarch64/aarch64-protos.h (aarch64_sve_prepare_conditional_op):
12402 * config/aarch64/aarch64.c (aarch64_sve_prepare_conditional_op): New
12404 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): Use it.
12405 (*cond_<optab><mode>): New patterns for reversed operands.
12407 2018-05-25 Richard Biener <rguenther@suse.de>
12409 * tree-vectorizer.h (STMT_VINFO_GROUP_*, GROUP_*): Remove.
12410 (DR_GROUP_*): New, assert we have non-NULL ->data_ref_info.
12411 (REDUC_GROUP_*): New, assert we have NULL ->data_ref_info.
12412 (STMT_VINFO_GROUPED_ACCESS): Adjust.
12413 * tree-vect-data-refs.c (everywhere): Adjust users.
12414 * tree-vect-loop.c (everywhere): Likewise.
12415 * tree-vect-slp.c (everywhere): Likewise.
12416 * tree-vect-stmts.c (everywhere): Likewise.
12417 * tree-vect-patterns.c (vect_reassociating_reduction_p): Likewise.
12419 2018-05-25 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
12421 * configure.ac (gcc_cv_as_section_has_e): Move to common section.
12423 (gcc_cv_as_section_exclude): ... this.
12424 Try Solaris as #exclude syntax.
12425 * configure: Regenerate.
12426 * config.in: Regenerate.
12427 * config/i386/i386.c (i386_solaris_elf_named_section): Handle
12429 * config/sparc/sparc.c (sparc_solaris_elf_asm_named_section)
12430 [HAVE_GAS_SECTION_EXCLUDE]: Handle SECTION_EXCLUDE.
12432 * varasm.c (default_elf_asm_named_section): Don't check if
12433 HAVE_GAS_SECTION_EXCLUDE is defined.
12435 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
12437 * doc/md.texi: Update the documentation of the cond_* optabs
12438 to mention the new final operand. Fix GET_MODE_NUNITS call.
12439 Describe the scalar case too.
12440 * internal-fn.def (IFN_EXTRACT_LAST): Change type to fold_left.
12441 * internal-fn.c (expand_cond_unary_optab_fn): Expect 3 operands
12443 (expand_cond_binary_optab_fn): Expect 4 operands instead of 3.
12444 (get_conditional_internal_fn): Update comment.
12445 * tree-vect-loop.c (vectorizable_reduction): Pass the original
12446 accumulator value as a final argument to conditional functions.
12447 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): Turn into
12448 a define_expand and add an "else" operand. Assert for now that
12449 the else operand is equal to operand 2. Use SVE_INT_BINARY and
12450 SVE_COND_FP_BINARY instead of SVE_COND_INT_OP and SVE_COND_FP_OP.
12451 (*cond_<optab><mode>): New patterns.
12452 * config/aarch64/iterators.md (UNSPEC_COND_SMAX, UNSPEC_COND_UMAX)
12453 (UNSPEC_COND_SMIN, UNSPEC_COND_UMIN, UNSPEC_COND_AND, UNSPEC_COND_ORR)
12454 (UNSPEC_COND_EOR): Delete.
12455 (optab): Remove associated mappings.
12456 (SVE_INT_BINARY): New code iterator.
12457 (sve_int_op): Remove int attribute and add "minus" to the code
12459 (SVE_COND_INT_OP): Delete.
12460 (SVE_COND_FP_OP): Rename to...
12461 (SVE_COND_FP_BINARY): ...this.
12463 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
12465 * optabs.c (can_reuse_operands_p): New function.
12466 (maybe_legitimize_operands): Try to reuse the results for
12469 2018-05-24 Uros Bizjak <ubizjak@gmail.com>
12471 * config/i386/sse.md (cvtusi2<ssescalarmodesuffix>64<round_name>):
12472 Add {q} suffix to insn mnemonic.
12474 2018-05-23 Jozef Lawrynowicz <jozef.l@mittosystems.com>
12476 * config/msp430/msp430.c (TARGET_WARN_FUNC_RETURN): Define.
12477 (msp430_warn_func_return): New.
12479 2018-05-24 Roger Sayle <roger@nextmovesoftware.com>
12481 * fold-const.c (tree_nonzero_bits): New function.
12482 * fold-const.h (tree_nonzero_bits): Likewise.
12483 * match.pd (POPCOUNT): New patterns to fold BUILTIN_POPCOUNT and
12484 friends. POPCOUNT(x&1) => x&1, POPCOUNT(x)==0 => x==0, etc.
12486 2018-05-24 H.J. Lu <hongjiu.lu@intel.com>
12490 * varasm.c (assemble_alias): Check ifunc_resolver only on
12493 2018-05-24 Uros Bizjak <ubizjak@gmail.com>
12496 * config/i386/sse.md (movdi_to_sse): Do not generate pseudo
12497 when memory input operand is handled.
12499 2018-05-24 Luis Machado <luis.machado@linaro.org>
12501 * config/aarch64/aarch64.c (qdf24xx_addrcost_table): New static
12503 (qdf24xx_tunings) <addr_costs>: Set to qdf24xx_addrcost_table.
12505 2018-05-24 Richard Sandiford <richard.sandiford@linaro.org>
12507 * match.pd: Delay FMA folds until after vectorization.
12509 2018-05-24 Andre Vieira <andre.simoesdiasvieira@arm.com>
12512 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Make
12513 address check not strict.
12515 2018-05-24 Richard Sandiford <richard.sandiford@linaro.org>
12517 * gimple-match.h (gimple_match_op): New class.
12518 (mprts_hook): Replace parameters with a gimple_match_op *.
12519 (maybe_build_generic_op): Likewise.
12520 (gimple_simplified_result_is_gimple_val): Replace parameters with
12521 a const gimple_match_op *.
12522 (gimple_simplify): Replace code_helper * and tree * parameters with
12523 a gimple_match_op * parameter.
12524 (gimple_resimplify1): Replace code_helper *, tree and tree *
12525 parameters with a gimple_match_op * parameter.
12526 (gimple_resimplify2): Likewise.
12527 (gimple_resimplify3): Likewise.
12528 (maybe_push_res_to_seq): Replace code_helper, tree and tree *
12529 parameters with a gimple_match_op * parameter.
12530 * gimple-match-head.c (gimple_simplify): Change prototypes of
12531 auto-generated functions to take a gimple_match_op * instead of
12532 separate code_helper * and tree * parameters. Make the same
12533 change in the top-level overload and update calls to the
12534 gimple_resimplify routines. Update calls to the auto-generated
12535 functions and to maybe_push_res_to_seq in the publicly-facing
12536 operation-specific gimple_simplify overloads.
12537 (gimple_match_op::MAX_NUM_OPS): Define.
12538 (gimple_resimplify1): Replace rcode and ops with a single res_op
12539 parameter. Update call to gimple_simplify.
12540 (gimple_resimplify2): Likewise.
12541 (gimple_resimplify3): Likewise.
12542 (mprts_hook): Replace parameters with a gimple_match_op *.
12543 (maybe_build_generic_op): Likewise.
12544 (build_call_internal): Replace type, nargs and ops with
12545 a gimple_match_op *.
12546 (maybe_push_res_to_seq): Replace res_code, type and ops parameters
12547 with a single gimple_match_op *. Update calls to mprts_hook,
12548 build_call_internal and gimple_simplified_result_is_gimple_val.
12549 Factor out code that is common to the tree_code and combined_fn cases.
12550 * genmatch.c (expr::gen_transform): Replace tem_code and
12551 tem_ops with a gimple_match_op called tem_op. Update calls
12552 to the gimple_resimplify functions and maybe_push_res_to_seq.
12553 (dt_simplify::gen_1): Manipulate res_op instead of res_code and
12554 res_ops. Update call to the gimple_resimplify functions.
12555 (dt_simplify::gen): Pass res_op instead of res_code and res_ops.
12556 (decision_tree::gen): Make the functions take a gimple_match_op *
12557 called res_op instead of separate res_code and res_ops parameters.
12558 Update call accordingly.
12559 * gimple-fold.c (replace_stmt_with_simplification): Replace rcode
12560 and ops with a single res_op parameter. Update calls to
12561 maybe_build_generic_op and maybe_push_res_to_seq.
12562 (fold_stmt_1): Update calls to gimple_simplify and
12563 replace_stmt_with_simplification.
12564 (gimple_fold_stmt_to_constant_1): Update calls to gimple_simplify
12565 and gimple_simplified_result_is_gimple_val.
12566 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update call to
12568 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Replace parameters
12569 with a gimple_match_op *.
12570 (vn_nary_build_or_lookup): Likewise. Update call to
12571 vn_nary_build_or_lookup_1.
12572 (vn_nary_build_or_lookup_1): Replace rcode, type and ops with a
12573 gimple_match_op *. Update calls to the gimple_resimplify routines
12574 and to gimple_simplified_result_is_gimple_val.
12575 (vn_nary_simplify): Update call to vn_nary_build_or_lookup_1.
12576 Use gimple_match_op::MAX_NUM_OPS instead of a hard-coded 3.
12577 (vn_reference_lookup_3): Update call to vn_nary_build_or_lookup.
12578 (visit_nary_op): Likewise.
12579 (visit_reference_op_load): Likewise.
12581 2018-05-23 Luis Machado <luis.machado@linaro.org>
12583 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Use correct type
12584 modifier for printing the step amount.
12586 2018-05-23 Jozef Lawrynowicz <jozef.l@somniumtech.com>
12589 * tree.c (build_common_tree_nodes): Dont set TYPE_SIZE for __intN
12592 2018-05-23 Segher Boessenkool <segher@kernel.crashing.org>
12594 * doc/sourcebuild.texi (Endianness): New subsubsection.
12596 2018-05-23 Luis Machado <luis.machado@linaro.org>
12598 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
12599 <prefetch_dynamic_strides>: New const bool field.
12600 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
12601 prefetch_dynamic_strides.
12602 (exynosm1_prefetch_tune): Likewise.
12603 (thunderxt88_prefetch_tune): Likewise.
12604 (thunderx_prefetch_tune): Likewise.
12605 (thunderx2t99_prefetch_tune): Likewise.
12606 (qdf24xx_prefetch_tune): Likewise. Set prefetch_dynamic_strides to
12608 (aarch64_override_options_internal): Update to set
12609 PARAM_PREFETCH_DYNAMIC_STRIDES.
12610 * doc/invoke.texi (prefetch-dynamic-strides): Document new option.
12611 * params.def (PARAM_PREFETCH_DYNAMIC_STRIDES): New.
12612 * params.h (PARAM_PREFETCH_DYNAMIC_STRIDES): Define.
12613 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Account for
12614 prefetch-dynamic-strides setting.
12616 2018-05-23 Luis Machado <luis.machado@linaro.org>
12618 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
12619 <minimum_stride>: New const int field.
12620 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
12621 minimum_stride field defaulting to -1.
12622 (exynosm1_prefetch_tune): Likewise.
12623 (thunderxt88_prefetch_tune): Likewise.
12624 (thunderx_prefetch_tune): Likewise.
12625 (thunderx2t99_prefetch_tune): Likewise.
12626 (qdf24xx_prefetch_tune) <minimum_stride>: Set to 2048.
12627 <default_opt_level>: Set to 3.
12628 (aarch64_override_options_internal): Update to set
12629 PARAM_PREFETCH_MINIMUM_STRIDE.
12630 * doc/invoke.texi (prefetch-minimum-stride): Document new option.
12631 * params.def (PARAM_PREFETCH_MINIMUM_STRIDE): New.
12632 * params.h (PARAM_PREFETCH_MINIMUM_STRIDE): Define.
12633 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Return false if
12634 stride is constant and is below the minimum stride threshold.
12636 2018-05-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12638 * config/arm/arm-cpus.in (mode26): Delete.
12639 (armv4): Delete mode26 reference.
12640 * config/arm/arm.c (arm_configure_build_target): Delete use of
12643 2018-05-23 Uros Bizjak <ubizjak@gmail.com>
12645 * config/i386/i386.md (*floatuns<SWI48:mode><MODEF:mode>2_avx512):
12647 (floatunssi<mode>2): Also enable for AVX512F and TARGET_SSE_MATH.
12648 Rewrite expander pattern. Emit gen_floatunssi<mode>2_i387_with_xmm
12650 (floatunsdisf2): Rewrite expander pattern. Hanlde TARGET_AVX512F.
12651 (floatunsdidf2): Ditto.
12653 2018-05-23 Uros Bizjak <ubizjak@gmail.com>
12655 * config/i386/i386.md (fixuns_trunc<mode>di2): New insn pattern.
12656 (fixuns_trunc<mode>si2_avx512f): Ditto.
12657 (*fixuns_trunc<mode>si2_avx512f_zext): Ditto.
12658 (fixuns_trunc<mode>si2): Also enable for AVX512F and TARGET_SSE_MATH.
12659 Emit fixuns_trunc<mode>si2_avx512f for AVX512F targets.
12661 2018-05-23 Alexander Monakov <amonakov@ispras.ru>
12663 PR rtl-optimization/79985
12664 * df-scan.c (df_insn_refs_collect): Remove special case for
12665 global registers and asm statements.
12667 2018-05-23 Alexander Monakov <amonakov@ispras.ru>
12669 * extend.texi (Global Register Variables): Rewrite the bullet list.
12670 Note that the register is available for allocation. Note that access
12671 via inline asm must use constraints. Add note about async-signal
12672 handlers. Remove paragraph about automagic register selection.
12674 2018-05-23 Richard Biener <rguenther@suse.de>
12676 * tree-ssa-sccvn.c (vn_reference_lookup_3): Remove restriction
12677 of fixed offset from memset VN.
12679 2018-05-23 Bill Schmidt <wschmidt@linux.ibm.com>
12681 * gimple-ssa-strength-reduction.c (struct slsr_cand_d): Add
12682 first_interp field.
12683 (alloc_cand_and_find_basis): Initialize first_interp field.
12684 (slsr_process_mul): Modify first_interp field.
12685 (slsr_process_add): Likewise.
12686 (slsr_process_cast): Modify first_interp field for each new
12688 (slsr_process_copy): Likewise.
12689 (dump_candidate): Dump first_interp field.
12690 (replace_mult_candidate): Process all interpretations, not just
12692 (replace_rhs_if_not_dup): Likewise.
12693 (replace_one_candidate): Likewise.
12695 2018-05-23 Wilco Dijkstra <wdijkstr@arm.com>
12697 * config/aarch64/aarch64.c (aarch64_use_frame_pointer):
12699 (aarch64_needs_frame_chain): New function.
12700 (aarch64_parse_override_string): Set aarch64_use_frame_pointer.
12702 2018-05-23 Sudakshina Das <sudi.das@arm.com>
12705 * common/config/aarch64/aarch64-common.c (aarch64_handle_option):
12706 Check val before adding MASK_STRICT_ALIGN to opts->x_target_flags.
12707 * config/aarch64/aarch64.opt (mstrict-align): Remove RejectNegative.
12708 * config/aarch64/aarch64.c (aarch64_attributes): Mark allow_neg
12709 as true for strict-align.
12710 (aarch64_can_inline_p): Perform checks even when callee has no
12711 attributes to check for strict alignment.
12712 * doc/extend.texi (AArch64 Function Attributes): Document
12714 * doc/invoke.texi: (AArch64 Options): Likewise.
12716 2018-05-23 Richard Sandiford <richard.sandiford@linaro.org>
12718 PR tree-optimization/85853
12719 * tree-vect-slp.c (vect_slp_analyze_node_operations): Split out
12720 the handling of the root of the node to...
12721 (vect_slp_analyze_node_operations_1): ...this new function,
12722 and run the whole thing with the child nodes' def types
12723 set according to their SLP node's def type.
12725 2018-05-23 Richard Biener <rguenther@suse.de>
12727 PR middle-end/85874
12728 * tree-data-ref.c (create_runtime_alias_checks): Defer
12729 and ignore overflow warnings.
12731 2018-05-23 Yury Gribov <tetra2005@gmail.com>
12733 PR tree-optimization/85822
12734 * tree-vrp.c (is_masked_range_test): Fix handling of negative
12737 2018-05-23 Richard Biener <rguenther@suse.de>
12739 * tree-ssa-sccvn.c (vn_reference_lookup_3): Handle arbitrary
12740 memset constants via native_interpret_expr.
12742 2018-05-22 H.J. Lu <hongjiu.lu@intel.com>
12745 * cgraph.h (cgraph_node::create): Set ifunc_resolver for ifunc
12747 (cgraph_node::create_alias): Likewise.
12748 (cgraph_node::get_availability): Check ifunc_resolver instead
12749 of looking up ifunc attribute.
12750 * cgraphunit.c (maybe_diag_incompatible_alias): Likewise.
12751 * varasm.c (do_assemble_alias): Likewise.
12752 (assemble_alias): Likewise.
12753 (default_binds_local_p_3): Likewise.
12754 * cgraph.h (cgraph_node): Add ifunc_resolver.
12755 (cgraph_node::only_called_directly_or_aliased_p): Return false
12756 for IFUNC resolver.
12757 * lto-cgraph.c (input_node): Set ifunc_resolver for ifunc
12759 * symtab.c (symtab_node::verify_base): Verify that ifunc_resolver
12760 is equivalent to lookup_attribute ("ifunc", DECL_ATTRIBUTES (decl)).
12761 (symtab_node::binds_to_current_def_p): Check ifunc_resolver
12762 instead of looking up ifunc attribute.
12764 2018-05-22 Luis Machado <luis.machado@linaro.org>
12766 * config/aarch64/aarch64.md (*ashift<mode>_extv_bfiz): New pattern.
12768 2018-05-22 Martin Sebor <msebor@redhat.com>
12770 PR middle-end/85359
12771 * builtins.c (expand_builtin_strcpy): Call maybe_warn_nonstring_arg
12772 only when expasion succeeds.
12773 (expand_builtin_strcmp): Same.
12774 (expand_builtin_strncmp): Same.
12776 2018-05-22 Martin Sebor <msebor@redhat.com>
12778 * calls.c (maybe_warn_nonstring_arg): Fix a typo in a comment.
12780 2018-05-22 Jackson Woodruff <jackson.woodruff@arm.com>
12781 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12783 * config/aarch64/aarch64-ldpstp.md: Replace uses of
12784 aarch64_mem_pair_operand with memory_operand and delete operand swapping
12786 * config/aarch64/aarch64.c (aarch64_operands_ok_for_ldpstp):
12787 Add check for legitimate_address.
12788 (aarch64_gen_adjusted_ldpstp): Swap operands where appropriate.
12789 (aarch64_swap_ldrstr_operands): New.
12790 * config/aarch64/aarch64-protos.h (aarch64_swap_ldrstr_operands):
12793 2018-05-22 Jackson Woodruff <jackson.woodruff@arm.com>
12794 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12796 * config/aarch64/aarch64.md: New patterns to generate stp
12798 (store_pair_sw, store_pair_dw): New patterns to generate stp for
12799 single words and double words.
12800 (load_pair_sw, load_pair_dw): Likewise.
12801 (store_pair_sf, store_pair_df, store_pair_si, store_pair_di):
12803 (load_pair_sf, load_pair_df, load_pair_si, load_pair_di):
12805 * config/aarch64/aarch64-ldpstp.md: Modify peephole
12806 for different mode ldpstp and add peephole for merged zero stores.
12807 Likewise for loads.
12808 * config/aarch64/aarch64.c (aarch64_operands_ok_for_ldpstp):
12810 (aarch64_gen_store_pair): Rename calls to match new patterns.
12811 (aarch64_gen_load_pair): Rename calls to match new patterns.
12812 * config/aarch64/aarch64-simd.md (load_pair<mode>): Rename to...
12813 (load_pair<DREG:mode><DREG2:mode>): ... This.
12814 (store_pair<mode>): Rename to...
12815 (vec_store_pair<DREG:mode><DREG2:mode>): ... This.
12816 * config/aarch64/iterators.md (DREG, DREG2, DX2, SX, SX2, DSX):
12817 New mode iterators.
12818 (V_INT_EQUIV): Handle SImode.
12819 * config/aarch64/predicates.md (aarch64_reg_zero_or_fp_zero):
12822 2018-05-22 Martin Sebor <msebor@redhat.com>
12825 * calls.c (maybe_warn_nonstring_arg): Use string length to set
12826 or ajust the presumed bound on an operation to avoid unnecessary
12829 2018-05-22 Martin Sebor <msebor@redhat.com>
12831 PR tree-optimization/85826
12832 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Avoid
12833 assuming that a DECL necesarily has a constant size.
12835 2018-05-22 Richard Sandiford <richard.sandiford@linaro.org>
12837 PR middle-end/85862
12838 * internal-fn.c (expand_direct_optab_fn): Cope with a null lhs.
12840 2018-05-22 Richard Biener <rguenther@suse.de>
12842 PR tree-optimization/85834
12843 * tree-ssa-sccvn.c (vn_reference_lookup_3): Properly handle
12844 non-constant and non-zero memset arguments.
12846 2018-05-22 Martin Liska <mliska@suse.cz>
12849 * ipa-icf.c (sem_item::add_type): Do not ICE for incomplete types.
12851 2018-05-22 Richard Biener <rguenther@suse.de>
12853 PR tree-optimization/85863
12854 * tree-vect-stmts.c (vect_is_simple_cond): Only widen invariant
12855 comparisons when vectype is specified.
12856 (vectorizable_condition): Do not specify vectype for
12857 vect_is_simple_cond when SLP vectorizing.
12859 2018-05-21 Michael Meissner <meissner@linux.ibm.com>
12862 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Do not
12863 define __ibm128 as long double.
12864 * config/rs6000/rs6000.c (rs6000_init_builtins): Create __ibm128
12865 as a distinct type when IEEE 128-bit support is enabled.
12866 (init_float128_ieee): Fix up conversions between IFmode and IEEE
12867 128-bit types to use the correct functions.
12868 (rs6000_expand_float128_convert): Use explicit FLOAT_EXTEND to
12869 convert between 128-bit floating point types that have different
12870 modes but the same representation, instead of using gen_lowpart to
12872 * config/rs6000/rs6000.md (IFKF): New iterator for IFmode and
12874 (IFKF_reg): New attributes to give the register constraints for
12876 (extend<mode>tf2_internal): New insns to mark an explicit
12877 conversion between 128-bit floating point types that have a
12878 different mode but share the same representation.
12880 2018-05-21 Richard Sandiford <richard.sandiford@linaro.org>
12882 PR tree-optimization/85814
12883 * tree-ssa-strlen.c (get_stridx_plus_constant): Cope with
12884 a null return from get_strinfo when unsharing the next
12885 strinfo in the chain.
12887 2018-05-21 Vladimir Mezentsev <vladimir.mezentsev@oracle.com>
12890 * varasm.c (weak_finish): Clean up weak_decls.
12892 2018-05-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12894 * config/aarch64/aarch64.md ("unspec"): Define UNSPEC_SABAL,
12895 UNSPEC_SABDL2, UNSPEC_SADALP, UNSPEC_UABAL, UNSPEC_UABDL2,
12896 UNSPEC_UADALP values.
12897 * config/aarch64/iterators.md (ABAL): New int iterator.
12900 (sur): Add mappings for the above.
12901 * config/aarch64/aarch64-simd.md (aarch64_<sur>abdl2<mode>_3):
12903 (aarch64_<sur>abal<mode>_4): Likewise.
12904 (aarch64_<sur>adalp<mode>_3): Likewise.
12905 (<sur>sadv16qi): New define_expand.
12907 2018-05-21 Alexander Nesterovskiy <alexander.nesterovskiy@intel.com>
12909 * config/i386/i386.md (*movsf_internal): AVX falsedep fix.
12910 (*movdf_internal): Ditto.
12911 (*rcpsf2_sse): Ditto.
12912 (*rsqrtsf2_sse): Ditto.
12913 (*sqrt<mode>2_sse): Ditto.
12915 2018-05-21 Tamar Christina <tamar.christina@arm.com>
12917 * config/aarch64/aarch64-simd.md (aarch64_eor3qv8hi): Change to
12919 (aarch64_bcaxqv8hi): Change to bcaxq<mode>4.
12920 * config/aarch64/aarch64-simd-builtins.def (veor3q_u8, veor3q_u32,
12921 veor3q_u64, veor3q_s8, veor3q_s16, veor3q_s32, veor3q_s64, vbcaxq_u8,
12922 vbcaxq_u32, vbcaxq_u64, vbcaxq_s8, vbcaxq_s16, vbcaxq_s32,
12924 * config/aarch64/arm_neon.h: Likewise.
12925 * config/aarch64/iterators.md (VQ_I): New.
12927 2018-05-21 Alexey Brodkin <abrodkin@synopsys.com>
12929 * config.gcc: Add arc/t-multilib-linux to tmake_file for
12931 * config/arc/t-multilib-linux: Specify MULTILIB_OPTIONS and
12934 2018-05-20 Chung-Ju Wu <jasonwucj@gmail.com>
12936 * config/nds32/constraints.md (S): New constraint.
12937 * config/nds32/nds32.md (call_internal): Use constraint S.
12938 (call_value_internal): Likewise.
12939 (sibcall_internal): Likewise.
12940 (sibcall_value_internal): Likewise.
12942 2018-05-20 Kito Cheng <kito.cheng@gmail.com>
12943 Chung-Ju Wu <jasonwucj@gmail.com>
12945 * config/nds32/nds32.c (nds32_register_move_cost): Take garywolf cpu
12946 into consideration.
12948 2018-05-20 Kito Cheng <kito.cheng@gmail.com>
12949 Chung-Ju Wu <jasonwucj@gmail.com>
12951 * config/nds32/nds32-cost.c (rtx_cost_model_t): New structure.
12952 (insn_size_16bit, insn_size_32bit): New variables for cost evaluation.
12953 (nds32_rtx_costs_impl): Simplify.
12954 (nds32_address_cost_impl): Simplify.
12955 (nds32_init_rtx_costs): New function.
12956 (nds32_rtx_costs_speed_prefer): Likewise.
12957 (nds32_rtx_costs_size_prefer): Likewise.
12958 (nds32_address_cost_speed_prefer): Likewise.
12959 (nds32_address_cost_speed_fwprop): Likewise.
12960 (nds32_address_cost_size_prefer): Likewise.
12961 * config/nds32/nds32-protos.h (nds32_init_rtx_costs): Declare.
12962 * config/nds32/nds32.c (nds32_option_override): Use
12963 nds32_init_rtx_costs function.
12965 2018-05-20 Chung-Ju Wu <jasonwucj@gmail.com>
12967 * config/nds32/nds32.c (nds32_asm_file_start): Output pipeline model.
12968 * config/nds32/nds32.h (TARGET_PIPELINE_N7): Define.
12969 (TARGET_PIPELINE_N8): Likewise.
12970 (TARGET_PIPELINE_N10): Likewise.
12971 (TARGET_PIPELINE_N13): Likewise.
12972 (TARGET_PIPELINE_GRAYWOLF): Likewise.
12974 2018-05-19 Monk Chiang <sh.chiang04@gmail.com>
12976 * config/nds32/nds32-fpu.md: Update copyright year.
12978 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
12980 * config/nds32/nds32.h (ASM_SPEC): Adjust spec rule.
12982 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
12984 * config/nds32/nds32.c
12985 (nds32_md_asm_adjust): Consider flag_inline_asm_r15 variable.
12986 * config/nds32/nds32.opt (minline-asm-r15): New option.
12988 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
12990 * common/config/nds32/nds32-common.c (TARGET_DEFAULT_TARGET_FLAGS): Add
12992 * config/nds32/nds32.md (abssi2): New pattern.
12994 2018-05-19 Uros Bizjak <ubizjak@gmail.com>
12996 * config/i386/i386.md (rex64namesuffix): New mode attribute.
12997 * config/i386/sse.md (sse_cvtsi2ss<rex64namesuffix><round_name>):
12998 Merge insn pattern from sse_cvtsi2ss<round_name> and
12999 sse_cvtsi2ssq<round_name> using SWI48 mode iterator.
13000 (sse_cvtss2si<rex64namesuffix><round_name>): Merge insn pattern
13001 from sse_cvtss2si<round_name> and sse_cvtss2siq<round_name>
13002 using SWI48 mode iterator.
13003 (sse_cvtss2si<rex64namesuffix>_2): Merge insn pattern from
13004 sse_cvtss2si_2 and sse_cvtss2siq_2 using SWI48 mode iterator.
13005 (sse_cvttss2si<rex64namesuffix><round_saeonly_name>): Merge insn
13006 pattern from sse_cvttss2si<round_saeonly_name>
13007 and sse_cvttss2siq<round_saeonly_name> using SWI48 mode iterator.
13008 (avx512f_vcvtss2usi<rex64namesuffix><round_name>): Merge insn pattern
13009 from avx512f_vcvtss2usi<round_name> and avx512f_vcvtss2usiq<round_name>
13010 using SWI48 mode iterator.
13011 (avx512f_vcvttss2usi<rex64namesuffix><round_saeonly_name>): Merge
13012 insn pattern from avx512f_vcvttss2usi<round_saeonly_name> and
13013 avx512f_vcvttss2usiq<round_saeonly_name> using SWI48 mode iterator.
13014 (avx512f_vcvtsd2usi<rex64namesuffix><round_name>): Merge insn pattern
13015 from avx512f_vcvtsd2usi<round_name> and avx512f_vcvtsd2usiq<round_name>
13016 using SWI48 mode iterator.
13017 (avx512f_vcvttsd2usi<rex64namesuffix><round_saeonly_name>): Merge
13018 insn pattern from avx512f_vcvttsd2usi<round_saeonly_name> and
13019 avx512f_vcvttsd2usiq<round_saeonly_name> using SWI48 mode iterator.
13020 (sse2_cvtsd2si<rex64namesuffix><round_name>): Merge insn pattern from
13021 sse2_cvtsd2si<round_name> and sse2_cvtsd2siq<round_name> using
13022 SWI48 mode iterator.
13023 (sse2_cvtsd2si<rex64namesuffix>_2): Merge insn pattern from
13024 sse2_cvtsd2si_2 and sse2_cvtsd2siq_2 using SWI48 mode iterator.
13025 (sse_cvttsd2si<rex64namesuffix><round_saeonly_name>): Merge insn
13026 pattern from sse_cvttsd2si<round_saeonly_name>
13027 and sse_cvttsd2siq<round_saeonly_name> using SWI48 mode iterator.
13029 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
13031 * config/nds32/nds32-md-auxiliary.c
13032 (nds32_valid_smw_lwm_base_p): Refine.
13033 (nds32_output_smw_single_word): Refine.
13034 (nds32_output_smw_double_word): New.
13035 * config/nds32/nds32-protos.h (nds32_output_smw_double_word): New.
13037 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
13039 * config/nds32/nds32-md-auxiliary.c (nds32_output_stack_push): Refine.
13040 (nds32_output_stack_pop): Refine.
13041 (nds32_expand_unaligned_load): Refine.
13042 (nds32_expand_unaligned_store): Refine.
13044 2018-05-19 Kuan-Lin Chen <kuanlinchentw@gmail.com>
13045 Chung-Ju Wu <jasonwucj@gmail.com>
13047 * config/nds32/constants.md: Add TP_REGNUM constant.
13048 (unspec_element): Add UNSPEC_GOTINIT, UNSPEC_GOT, UNSPEC_GOTOFF,
13049 UNSPEC_PLT, UNSPEC_TLSGD, UNSPEC_TLSLD, UNSPEC_TLSIE, UNSPEC_TLSLE and
13051 * config/nds32/nds32-doubleword.md: Consider flag_pic.
13052 * config/nds32/nds32-dspext.md (mov<mode>): Expand TLS and PIC cases.
13053 * config/nds32/nds32-predicates.c (nds32_const_unspec_p): New.
13054 * config/nds32/nds32-md-auxiliary.c: Implementation that support TLS
13055 and PIC code generation.
13056 * config/nds32/nds32-protos.h: Declarations that support TLS and PIC
13058 * config/nds32/nds32-relax-opt.c: Consider TLS and PIC for relax
13060 * config/nds32/nds32.md: Support TLS and PIC.
13061 * config/nds32/nds32.c: Support TLS and PIC.
13062 * config/nds32/nds32.h (nds32_relax_insn_type): New enum type.
13063 * config/nds32/predicates.md (nds32_nonunspec_symbolic_operand): New
13066 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
13068 * config/nds32/nds32-predicates.c (const_vector_to_hwint): Use machine
13069 mode with E_ prefix.
13071 2018-05-19 Kuan-Lin Chen <kuanlinchentw@gmail.com>
13072 Chung-Ju Wu <jasonwucj@gmail.com>
13074 * config/nds32/constants.md (unspec_element): Add UNSPEC_ICT.
13075 * config/nds32/nds32-md-auxiliary.c
13076 (symbolic_reference_mentioned_p): New.
13077 (nds32_legitimize_ict_address): New.
13078 (nds32_expand_ict_move): New.
13079 (nds32_indirect_call_referenced_p): New.
13080 (nds32_symbol_binds_local_p): Delete.
13081 (nds32_long_call_p): Modify.
13082 * config/nds32/nds32-opts.h (nds32_ict_model_type): New enum type.
13083 * config/nds32/nds32-protos.h
13084 (symbolic_reference_mentioned_p): Declare.
13085 (nds32_legitimize_ict_address): Declare.
13086 (nds32_expand_ict_move): Declare.
13087 (nds32_indirect_call_referenced_p): Declare.
13088 * config/nds32/nds32-relax-opt.c (nds32_ict_const_p): New.
13089 (nds32_relax_group): Use nds32_ict_const_p as condition.
13090 * config/nds32/nds32.c (nds32_attribute_table): Add "indirect_call".
13091 (nds32_asm_file_start): Output ict_model directive in asm code.
13092 (nds32_legitimate_address_p): Consider indirect call.
13093 (nds32_print_operand): Consider indirect call.
13094 (nds32_print_operand_address): Consider indirect call.
13095 (nds32_insert_attributes): Handle "indirect_call" attribute.
13096 (TARGET_LEGITIMATE_ADDRESS_P): Define.
13097 (TARGET_LEGITIMATE_CONSTANT_P): Define.
13098 (TARGET_CANNOT_FORCE_CONST_MEM): Define.
13099 (TARGET_DELEGITIMIZE_ADDRESS): Define.
13100 (TARGET_ASM_OUTPUT_ADDR_CONST_EXTRA): Define.
13101 * config/nds32/nds32.h (SYMBOLIC_CONST_P): Define.
13102 (TARGET_ICT_MODEL_SMALL): Define.
13103 (TARGET_ICT_MODEL_LARGE): Define.
13104 * config/nds32/nds32.md (movsi): Consider ict model.
13105 (call, call_value): Consider ict model.
13106 (sibcall, sibcall_value): Consider ict model.
13107 * config/nds32/nds32.opt (mict-model): New option.
13108 * config/nds32/predicates.md (nds32_symbolic_operand): Consider ict
13111 2018-05-18 Kito Cheng <kito.cheng@gmail.com>
13112 Monk Chiang <sh.chiang04@gmail.com>
13113 Jim Wilson <jimw@sifive.com>
13115 * common/config/riscv/riscv-common.c (riscv_parse_arch_string):
13116 Add support to parse rv32e*. Clear MASK_RVE for rv32i and rv64i.
13117 * config.gcc (riscv*-*-*): Add support for rv32e* and ilp32e.
13118 * config/riscv/riscv-c.c (riscv_cpu_cpp_builtins): Define
13119 __riscv_32e when TARGET_RVE. Handle ABI_ILP32E as soft-float ABI.
13120 * config/riscv/riscv-opts.h (riscv_abi_type): Add ABI_ILP32E.
13121 * config/riscv/riscv.c (riscv_compute_frame_info): When TARGET_RVE,
13122 compute save_libcall_adjustment properly.
13123 (riscv_option_override): Call error if TARGET_RVE and not ABI_ILP32E.
13124 (riscv_conditional_register_usage): Handle TARGET_RVE and ABI_ILP32E.
13125 * config/riscv/riscv.h (UNITS_PER_FP_ARG): Handle ABI_ILP32E.
13126 (STACK_BOUNDARY, ABI_STACK_BOUNDARY): Handle TARGET_RVE.
13127 (GP_REG_LAST, MAX_ARGS_IN_REGISTERS): Likewise.
13128 (ABI_SPEC): Handle mabi=ilp32e.
13129 * config/riscv/riscv.opt (abi_type): Add ABI_ILP32E.
13130 (RVE): Add RVE mask.
13131 * doc/invoke.texi (RISC-V options) <-mabi>: Add ilp32e info.
13132 <-march>: Add rv32e as an example.
13134 2018-05-18 Marc Glisse <marc.glisse@inria.fr>
13137 * tree-ssa-structalias.c (create_variable_info_for_1): Extra argument.
13138 (intra_create_variable_infos): Handle C++ constructors.
13140 2018-05-18 Martin Liska <mliska@suse.cz>
13142 * passes.def: Remove a redundant pass.
13144 2018-05-18 Eric Botcazou <ebotcazou@adacore.com>
13147 * config/sparc/sparc.c (sparc_expand_builtin): Always initialize op[0].
13149 2018-05-18 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13151 * config/arm/arm-cpus.in (armv3m, mode32): Delete features.
13153 (ARMv2, ARMv3, ARMv3m): Delete fgroups.
13155 (armv2, armv2a, armv3, armv3m): Delete architectures.
13156 (arm2, arm250, arm3, arm6, arm60, arm600, arm610, arm620,
13157 arm7, arm7d, arm7di, arm70, arm700, arm700i, arm710, arm720,
13158 arm710c, arm7100, arm7500, arm7500fe, arm7m, arm7dm, arm7dmi):
13160 * config/arm/arm.md (maddsidi4): Remove check for arm_arch3m.
13161 (*mulsidi3adddi): Likewise.
13162 (mulsidi3): Likewise.
13163 (*mulsidi3_nov6): Likewise.
13164 (umulsidi3): Likewise.
13165 (umulsidi3_nov6): Likewise.
13166 (umaddsidi4): Likewise.
13167 (*umulsidi3adddi): Likewise.
13168 (smulsi3_highpart): Likewise.
13169 (*smulsi3_highpart_nov6): Likewise.
13170 (umulsi3_highpart): Likewise.
13171 (*umulsi3_highpart_nov6): Likewise.
13172 * config/arm/arm.h (arm_arch3m): Delete.
13173 * config/arm/arm.c (arm_arch3m): Delete.
13174 (arm_option_override_internal): Update armv3-related comment.
13175 (arm_configure_build_target): Delete use of isa_bit_mode32.
13176 (arm_option_reconfigure_globals): Delete set of arm_ach3m.
13177 (arm_rtx_costs_internal): Delete check of arm_arch3m.
13178 * config/arm/arm-fixed.md (mulsq3): Delete check for arm_arch3m.
13179 (mulsa3): Likewise.
13180 (mulusa3): Likewise.
13181 * config/arm/arm-protos.h (arm_arch3m): Delete.
13182 * config/arm/arm-tables.opt: Regenerate.
13183 * config/arm/arm-tune.md: Likewise.
13184 * config/arm/t-arm-elf (all_early_nofp): Delete mentions of
13185 deleted architectures.
13187 2018-05-18 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13189 * config/arm/arm-cpus.in (armv5, armv5e): Delete features.
13190 (armv5t, armv5te): New features.
13191 (ARMv5, ARMv5e): Delete fgroups.
13192 (ARMv5t, ARMv5te): Adjust for above changes.
13193 (ARMv6m): Likewise.
13194 (armv5, armv5e): Delete arches.
13195 * config/arm/arm.md (*call_reg_armv5): Use arm_arch5t instead of
13197 (*call_reg_arm): Likewise.
13198 (*call_value_reg_armv5): Likewise.
13199 (*call_value_reg_arm): Likewise.
13200 (*call_symbol): Likewise.
13201 (*call_value_symbol): Likewise.
13202 (*sibcall_insn): Likewise.
13203 (*sibcall_value_insn): Likewise.
13204 (clzsi2): Likewise.
13205 (prefetch): Likewise.
13206 (define_split and define_peephole2 dependent on arm_arch5):
13208 * config/arm/arm.h (TARGET_LDRD): Use arm_arch5te instead of
13210 (TARGET_ARM_QBIT): Likewise.
13211 (TARGET_DSP_MULTIPLY): Likewise.
13212 (enum base_architecture): Delete BASE_ARCH_5, BASE_ARCH_5E.
13213 (arm_arch5, arm_arch5e): Delete.
13214 (arm_arch5t, arm_arch5te): Declare.
13215 * config/arm/arm.c (arm_arch5, arm_arch5e): Delete.
13216 (arm_arch5t): Declare.
13217 (arm_option_reconfigure_globals): Update for the above.
13218 (arm_options_perform_arch_sanity_checks): Update comment, replace
13219 use of arm_arch5 with arm_arch5t.
13220 (use_return_insn): Likewise.
13221 (arm_emit_call_insn): Likewise.
13222 (output_return_instruction): Likewise.
13223 (arm_final_prescan_insn): Likewise.
13224 (arm_coproc_builtin_available): Likewise.
13225 * config/arm/arm-c.c (arm_cpu_builtins): Replace arm_arch5 and
13226 arm_arch5e with arm_arch5t and arm_arch5te.
13227 * config/arm/arm-protos.h (arm_arch5, arm_arch5e): Delete.
13228 (arm_arch5t, arm_arch5te): Declare.
13229 * config/arm/arm-tables.opt: Regenerate.
13230 * config/arm/t-arm-elf: Remove references to armv5, armv5e.
13231 * config/arm/t-multilib: Likewise.
13232 * config/arm/thumb1.md (*call_reg_thumb1_v5): Check arm_arch5t
13233 instead of arm_arch5.
13234 (*call_reg_thumb1): Likewise.
13235 (*call_value_reg_thumb1_v5): Likewise.
13236 (*call_value_reg_thumb1): Likewise.
13237 * config/arm/vxworks.h (TARGET_OS_CPP_BUILTINS): Remove now
13239 * doc/invoke.texi (ARM Options): Remove references to armv5, armv5e.
13241 2018-05-18 Martin Liska <mliska@suse.cz>
13243 PR gcov-profile/84846
13244 * doc/gcov.texi: Document -t option of gcov tool.
13246 2018-05-18 Martin Liska <mliska@suse.cz>
13248 PR gcov-profile/84846
13249 * gcov.c (print_usage): Add new -t option.
13250 (process_args): Handle the option.
13251 (generate_results): Use stdout as output when requested by
13254 2018-05-18 Martin Liska <mliska@suse.cz>
13256 PR gcov-profile/84846
13257 * coverage.c (coverage_init): Write PWD to .gcno file.
13258 * doc/gcov.texi: Document how working directory is printed.
13259 * gcov-dump.c (dump_gcov_file): Print PWD.
13260 * gcov.c (output_intermediate_file): Likewise.
13261 (read_graph_file): Read PWD string.
13262 (output_lines): Print PWD.
13264 2018-05-18 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
13266 PR middle-end/85817
13267 * ipa-pure-const.c (malloc_candidate_p): Remove the check integer_zerop
13268 for retval and return false if all args to phi are zero.
13270 2018-05-18 Richard Biener <rguenther@suse.de>
13272 * gimple-ssa-evrp.c (class evrp_folder): Add simplify_stmt_using_ranges
13274 (evrp_dom_walker::before_dom_children): Call it.
13276 2018-05-18 Richard Biener <rguenther@suse.de>
13278 * tree-dfa.c (get_ref_base_and_extent): Use range-info to refine
13279 results when processing array refs with variable index.
13281 2018-05-18 Toon Moene <toon@moene.org>
13283 * doc/invoke.texi: Move -floop-unroll-and-jam documentation
13284 directly after that of -floop-interchange. Indicate that both
13285 options are enabled by default when specifying -O3.
13287 2018-05-18 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13289 * config/aarch64/aarch64-simd.md (vec_set<mode>): Use VALL_F16 mode
13290 iterator. Delete separate integer-mode vec_set<mode> expander.
13291 (aarch64_simd_vec_setv2di): Delete.
13292 (vec_setv2di): Delete.
13293 (aarch64_simd_vec_set<mode>): Delete all other patterns with that name.
13294 Use VALL_F16 mode iterator. Add LD1 alternative and use vwcore for
13295 the "w, r" alternative.
13297 2018-05-18 Martin Liska <mliska@suse.cz>
13299 * passes.def: Add pass_lower_switch and pass_lower_switch_O0.
13300 * tree-pass.h (make_pass_lower_switch_O0): New function.
13301 * tree-switch-conversion.c (node_has_low_bound): Remove.
13302 (node_has_high_bound): Likewise.
13303 (node_is_bounded): Likewise.
13304 (class pass_lower_switch): Make it a template type and create
13306 (pass_lower_switch::execute): Add template argument.
13307 (make_pass_lower_switch): New function.
13308 (make_pass_lower_switch_O0): New function.
13309 (do_jump_if_equal): Remove.
13310 (emit_case_nodes): Simplify to just handle all 3 cases and leave
13311 all the hard work to tree optimization passes.
13313 2018-05-18 Martin Liska <mliska@suse.cz>
13315 * dbgcnt.c (limit_low): Renamed from limit.
13316 (limit_high): New variable.
13317 (dbg_cnt_is_enabled): Check for upper limit.
13318 (dbg_cnt): Adjust dumping.
13319 (dbg_cnt_set_limit_by_index): Add new argument for high
13321 (dbg_cnt_set_limit_by_name): Likewise.
13322 (dbg_cnt_process_single_pair): Parse new format.
13323 (dbg_cnt_process_opt): Use strtok.
13324 (dbg_cnt_list_all_counters): Remove 'value' and add
13326 * doc/invoke.texi: Document changes.
13328 2018-05-18 Richard Sandiford <richard.sandiford@linaro.org>
13330 * doc/sourcebuild.texi (scalar_all_fma): Document.
13331 * tree.def (FMA_EXPR): Delete.
13332 * internal-fn.def (FMA, FMS, FNMA, FNMS): New internal functions.
13333 * internal-fn.c (ternary_direct): New macro.
13334 (expand_ternary_optab_fn): Likewise.
13335 (direct_ternary_optab_supported_p): Likewise.
13336 * Makefile.in (build/genmatch.o): Depend on case-fn-macros.h.
13337 * builtins.c (fold_builtin_fma): Delete.
13338 (fold_builtin_3): Don't call it.
13339 * cfgexpand.c (expand_debug_expr): Remove FMA_EXPR handling.
13340 * expr.c (expand_expr_real_2): Likewise.
13341 * fold-const.c (operand_equal_p): Likewise.
13342 (fold_ternary_loc): Likewise.
13343 * gimple-pretty-print.c (dump_ternary_rhs): Likewise.
13344 * gimple.c (DEFTREECODE): Likewise.
13345 * gimplify.c (gimplify_expr): Likewise.
13346 * optabs-tree.c (optab_for_tree_code): Likewise.
13347 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
13348 * tree-eh.c (operation_could_trap_p): Likewise.
13349 (stmt_could_throw_1_p): Likewise.
13350 * tree-inline.c (estimate_operator_cost): Likewise.
13351 * tree-pretty-print.c (dump_generic_node): Likewise.
13352 (op_code_prio): Likewise.
13353 * tree-ssa-loop-im.c (stmt_cost): Likewise.
13354 * tree-ssa-operands.c (get_expr_operands): Likewise.
13355 * tree.c (commutative_ternary_tree_code, add_expr): Likewise.
13356 * fold-const-call.h (fold_fma): Delete.
13357 * fold-const-call.c (fold_const_call_ssss): Handle CFN_FMS,
13358 CFN_FNMA and CFN_FNMS.
13359 (fold_fma): Delete.
13360 * genmatch.c (combined_fn): New enum.
13361 (commutative_ternary_tree_code): Remove FMA_EXPR handling.
13362 (commutative_op): New function.
13363 (commutate): Use it. Handle more than 2 operands.
13364 (dt_operand::gen_gimple_expr): Use commutative_op.
13365 (parser::parse_expr): Allow :c to be used with non-binary
13366 operators if the commutative operand is known.
13367 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Handle
13368 CFN_FMS, CFN_FNMA and CFN_FNMS.
13369 (backprop::process_assign_use): Remove FMA_EXPR handling.
13370 * hsa-gen.c (gen_hsa_insns_for_operation_assignment): Likewise.
13371 (gen_hsa_fma): New function.
13372 (gen_hsa_insn_for_internal_fn_call): Use it for IFN_FMA, IFN_FMS,
13373 IFN_FNMA and IFN_FNMS.
13374 * match.pd: Add folds for IFN_FMS, IFN_FNMA and IFN_FNMS.
13375 * gimple-fold.h (follow_all_ssa_edges): Declare.
13376 * gimple-fold.c (follow_all_ssa_edges): New function.
13377 * tree-ssa-math-opts.c (convert_mult_to_fma_1): Use the
13378 gimple_build interface and use follow_all_ssa_edges to fold the result.
13379 (convert_mult_to_fma): Use direct_internal_fn_suppoerted_p
13380 instead of checking for optabs directly.
13381 * config/i386/i386.c (ix86_add_stmt_cost): Recognize FMAs as calls
13382 rather than FMA_EXPRs.
13383 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Create a
13384 call to IFN_FMA instead of an FMA_EXPR.
13386 2018-05-17 Jim Wilson <jimw@sifive.com>
13388 * expr.c (do_tablejump): When converting index to Pmode, if we have a
13389 sign extended promoted subreg, and the range does not have the sign bit
13390 set, then do a sign extend.
13392 * config/riscv/riscv.c (riscv_extend_comparands): In unsigned QImode
13393 test, check for sign extended subreg and/or constant operands, and
13394 do a sign extend in that case.
13396 2018-05-17 Steve Ellcey <sellcey@cavium.com>
13398 * config/aarch64/thunderx2t99.md (thunderx2t99_ls_both): Delete.
13399 (thunderx2t99_multiple): Delete psuedo-units from used cpus.
13401 (thunderx2t99_alu_shift): Remove alu_shift_reg, alus_shift_reg.
13402 Change logics_shift_reg to logics_shift_imm.
13403 (thunderx2t99_fp_loadpair_basic): Delete.
13404 (thunderx2t99_fp_storepair_basic): Delete.
13405 (thunderx2t99_asimd_int): Add neon_sub and neon_sub_q types.
13406 (thunderx2t99_asimd_polynomial): Delete.
13407 (thunderx2t99_asimd_fp_simple): Add neon_fp_mul_s_scalar_q
13408 and neon_fp_mul_d_scalar_q.
13409 (thunderx2t99_asimd_fp_conv): Add *int_to_fp* types.
13410 (thunderx2t99_asimd_misc): Delete neon_dup and neon_dup_q.
13411 (thunderx2t99_asimd_recip_step): Add missing *sqrt* types.
13412 (thunderx2t99_asimd_lut): Add missing tbl types.
13413 (thunderx2t99_asimd_ext): Delete.
13414 (thunderx2t99_asimd_load1_1_mult): Delete.
13415 (thunderx2t99_asimd_load1_2_mult): Delete.
13416 (thunderx2t99_asimd_load1_ldp): New.
13417 (thunderx2t99_asimd_load1): New.
13418 (thunderx2t99_asimd_load2): Add missing *load2* types.
13419 (thunderx2t99_asimd_load3): New.
13420 (thunderx2t99_asimd_load4): New.
13421 (thunderx2t99_asimd_store1_1_mult): Delete.
13422 (thunderx2t99_asimd_store1_2_mult): Delete.
13423 (thunderx2t99_asimd_store2_mult): Delete.
13424 (thunderx2t99_asimd_store2_onelane): Delete.
13425 (thunderx2t99_asimd_store_stp): New.
13426 (thunderx2t99_asimd_store1): New.
13427 (thunderx2t99_asimd_store2): New.
13428 (thunderx2t99_asimd_store3): New.
13429 (thunderx2t99_asimd_store4): New.
13431 2018-05-17 Jerome Lambourg <lambourg@adacore.com>
13433 * config/arm/arm_cmse.h (cmse_nsfptr_create, cmse_is_nsfptr): Remove
13434 #include <stdint.h>. Replace intptr_t with __INTPTR_TYPE__.
13436 2018-05-17 Pat Haugen <pthaugen@us.ibm.com>
13437 Segher Boessenkool <segher@kernel.crashing.org>
13440 * config/rs6000/rs6000.c (rs6000_output_move_128bit): Check dest
13443 2018-05-17 Richard Biener <rguenther@suse.de>
13445 * tree-ssa-dse.c (dse_classify_store): Fix iterator increment
13446 for pruning loop and prune defs feeding only already visited PHIs.
13448 2018-05-17 Richard Biener <rguenther@suse.de>
13450 * tree-ssa-sccvn.c (vn_reference_lookup_3): Improve memset handling.
13452 2018-05-17 Bin Cheng <bin.cheng@arm.com>
13453 Richard Biener <rguenther@suse.de>
13455 PR tree-optimization/85793
13456 * tree-vect-stmts.c (vectorizable_load): Handle 1 element-wise load
13457 for VMAT_ELEMENTWISE.
13459 2018-05-17 Richard Sandiford <richard.sandiford@linaro.org>
13461 * internal-fn.h (lookup_internal_fn): Declare
13462 * internal-fn.c (lookup_internal_fn): New function.
13463 * gimple.c (gimple_build_call_from_tree): Handle calls to
13464 internal functions.
13465 * gimple-pretty-print.c (dump_gimple_call): Print "." before
13466 internal function names.
13467 * tree-pretty-print.c (dump_generic_node): Likewise.
13468 * tree-ssa-scopedtables.c (expr_hash_elt::print): Likewise.
13470 2018-05-17 Richard Sandiford <richard.sandiford@linaro.org>
13472 * gimple-fold.h (gimple_build): Make the function forms take
13473 combined_fn rather than built_in_function.
13474 (gimple_simplify): Likewise.
13475 * gimple-match-head.c (gimple_simplify): Likewise.
13476 * gimple-fold.c (gimple_build): Likewise.
13477 * tree-vect-loop.c (get_initial_def_for_reduction): Use gimple_build
13478 rather than gimple_build_call_internal.
13479 (get_initial_defs_for_reduction): Likewise.
13480 (vect_create_epilog_for_reduction): Likewise.
13481 (vectorizable_live_operation): Likewise.
13483 2018-05-17 Martin Liska <mliska@suse.cz>
13485 * gimple-ssa-sprintf.c (format_directive): Do not use
13486 space in between 'G_' and '('.
13488 2018-05-17 Jakub Jelinek <jakub@redhat.com>
13491 * config/i386/i386.c (ix86_fold_builtin): Handle masked shifts
13492 even if the mask is not all ones.
13495 * config/i386/i386.c (ix86_fold_builtin): Fold shift builtins by
13497 (ix86_gimple_fold_builtin): Likewise.
13500 * config/i386/i386.c: Include tree-vector-builder.h.
13501 (ix86_vector_shift_count): New function.
13502 (ix86_fold_builtin): Fold shift builtins by scalar count.
13503 (ix86_gimple_fold_builtin): Likewise.
13505 * config/i386/avx512fintrin.h (_mm512_set_epi16, _mm512_set_epi8,
13506 _mm512_setzero): New intrinsics.
13508 2018-05-17 James Greenhalgh <james.greenhalgh@arm.com>
13509 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13511 * config/aarch64/aarch64.c (aarch64_expand_vector_init): Modify
13512 code generation for cases where splatting a value is not useful.
13513 * simplify-rtx.c (simplify_ternary_operation): Simplify
13514 vec_merge across a vec_duplicate and a paradoxical subreg forming
13515 a vector mode to a vec_concat.
13517 2018-05-17 Olga Makhotina <olga.makhotina@intel.com>
13519 * config.gcc: Support "goldmont-plus".
13520 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
13522 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
13523 PROCESSOR_GOLDMONT_PLUS.
13524 * config/i386/i386.c (m_GOLDMONT_PLUS): Define.
13525 (processor_target_table): Add "goldmont-plus".
13526 (PTA_GOLDMONT_PLUS): Define.
13527 (ix86_lea_outperforms): Add TARGET_GOLDMONT_PLUS.
13528 (get_builtin_code_for_version): Handle PROCESSOR_GOLDMONT_PLUS.
13529 (fold_builtin_cpu): Add M_INTEL_GOLDMONT_PLUS.
13530 (fold_builtin_cpu): Add "goldmont-plus".
13531 (ix86_add_stmt_cost): Add TARGET_GOLDMONT_PLUS.
13532 (ix86_option_override_internal): Add "goldmont-plus".
13533 * config/i386/i386.h (processor_costs): Define TARGET_GOLDMONT_PLUS.
13534 (processor_type): Add PROCESSOR_GOLDMONT_PLUS.
13535 * config/i386/x86-tune.def: Add m_GOLDMONT_PLUS.
13536 * doc/invoke.texi: Add goldmont-plus as x86 -march=/-mtune= CPU type.
13538 2018-05-17 Richard Biener <rguenther@suse.de>
13540 PR tree-optimization/85757
13541 * tree-ssa-dse.c (dse_classify_store): Record a PHI def and
13542 remove defs that only feed that PHI from further processing.
13544 2018-05-16 Jim Wilson <jimw@sifive.com>
13546 * config/riscv/riscv.md (<optab>si3_mask, <optab>si3_mask_1): Prepend
13548 (<optab>di3_mask, <optab>di3_mask_1): Likewise.
13550 2018-05-16 Mark Wielaard <mark@klomp.org>
13552 * dwarf2out.c (count_index_strings): New function.
13553 (output_indirect_strings): Call count_index_strings and generate
13554 header for dwarf_version >= 5.
13556 2018-05-16 Mark Wielaard <mark@klomp.org>
13558 * dwarf2out.c (dwarf_FORM): New function.
13559 (set_indirect_string): Use dwarf_FORM.
13560 (reset_indirect_string): Likewise.
13561 (size_of_die): Likewise.
13562 (value_format): Likewise.
13563 (output_die): Likewise.
13564 (add_skeleton_AT_string): Likewise.
13565 (output_macinfo_op): Likewise.
13566 (index_string): Likewise.
13567 (output_index_string_offset): Likewise.
13568 (output_index_string): Likewise.
13569 (count_index_strings): Likewise.
13571 2018-05-16 Carl Love <cel@us.ibm.com>
13573 * config/rs6000/rs6000.md (prefetch): Generate ISA 2.06 instructions
13574 dcbt and dcbtstt with TH=16 if operands[2] is 0 and Power 8 or newer.
13576 2018-05-16 Martin Jambor <mjambor@suse.cz>
13578 * ipa-prop.c (ipa_free_all_edge_args): Remove.
13579 * ipa-prop.h (ipa_free_all_edge_args): Likewise.
13581 2018-05-16 Wilco Dijkstra <wdijkstr@arm.com>
13583 * config/aarch64/aarch64.md (fma<mode>4): Change into expand pattern.
13584 (fnma<mode>4): Likewise.
13585 (fms<mode>4): Likewise.
13586 (fnms<mode>4): Likewise.
13587 (aarch64_fma<mode>4): Rename insn, reorder accumulator operand.
13588 (aarch64_fnma<mode>4): Likewise.
13589 (aarch64_fms<mode>4): Likewise.
13590 (aarch64_fnms<mode>4): Likewise.
13591 (aarch64_fnmadd<mode>4): Likewise.
13593 2018-05-16 Jason Merrill <jason@redhat.com>
13595 * tree.c (warn_deprecated_use): Return bool. Simplify logic.
13597 2018-05-16 Richard Biener <rguenther@suse.de>
13599 * tree-vectorizer.h (struct stmt_info_for_cost): Add where member.
13600 (dump_stmt_cost): Declare.
13601 (add_stmt_cost): Dump cost we add.
13602 (add_stmt_costs): New function.
13603 (vect_model_simple_cost, vect_model_store_cost, vect_model_load_cost):
13604 No longer exported.
13605 (vect_analyze_stmt): Adjust prototype.
13606 (vectorizable_condition): Likewise.
13607 (vectorizable_live_operation): Likewise.
13608 (vectorizable_reduction): Likewise.
13609 (vectorizable_induction): Likewise.
13610 * tree-vect-loop.c (vect_analyze_loop_operations): Create local
13611 cost vector to pass to vectorizable_ and record afterwards.
13612 (vect_model_reduction_cost): Take cost vector argument and adjust.
13613 (vect_model_induction_cost): Likewise.
13614 (vectorizable_reduction): Likewise.
13615 (vectorizable_induction): Likewise.
13616 (vectorizable_live_operation): Likewise.
13617 * tree-vect-slp.c (vect_create_new_slp_node): Initialize
13618 SLP_TREE_NUMBER_OF_VEC_STMTS.
13619 (vect_analyze_slp_cost_1): Remove.
13620 (vect_analyze_slp_cost): Likewise.
13621 (vect_slp_analyze_node_operations): Take visited args and
13622 a target cost vector. Avoid processing already visited stmt sets.
13623 (vect_slp_analyze_operations): Use a local cost vector to gather
13624 costs and register those of non-discarded instances.
13625 (vect_bb_vectorization_profitable_p): Use add_stmt_costs.
13626 (vect_schedule_slp_instance): Remove copying of
13627 SLP_TREE_NUMBER_OF_VEC_STMTS. Instead assert that it is not
13629 * tree-vect-stmts.c (record_stmt_cost): Remove path directly
13630 adding cost. Record cost entry location.
13631 (vect_prologue_cost_for_slp_op): Function to compute cost of
13632 a constant or invariant generated for SLP vect in the prologue,
13633 split out from vect_analyze_slp_cost_1.
13634 (vect_model_simple_cost): Make static. Adjust for SLP costing.
13635 (vect_model_promotion_demotion_cost): Likewise.
13636 (vect_model_store_cost): Likewise, make static.
13637 (vect_model_load_cost): Likewise.
13638 (vectorizable_bswap): Add cost vector arg and adjust.
13639 (vectorizable_call): Likewise.
13640 (vectorizable_simd_clone_call): Likewise.
13641 (vectorizable_conversion): Likewise.
13642 (vectorizable_assignment): Likewise.
13643 (vectorizable_shift): Likewise.
13644 (vectorizable_operation): Likewise.
13645 (vectorizable_store): Likewise.
13646 (vectorizable_load): Likewise.
13647 (vectorizable_condition): Likewise.
13648 (vectorizable_comparison): Likewise.
13649 (can_vectorize_live_stmts): Likewise.
13650 (vect_analyze_stmt): Likewise.
13651 (vect_transform_stmt): Adjust calls to vectorizable_*.
13652 * tree-vectorizer.c: Include gimple-pretty-print.h.
13653 (dump_stmt_cost): New function.
13655 2018-05-16 Richard Biener <rguenther@suse.de>
13657 * params.def (PARAM_DSE_MAX_ALIAS_QUERIES_PER_STORE): New param.
13658 * doc/invoke.texi (dse-max-alias-queries-per-store): Document.
13659 * tree-ssa-dse.c: Include tree-ssa-loop.h.
13660 (check_name): New callback.
13661 (dse_classify_store): Track cycles via a visited bitmap of PHI
13662 defs and simplify handling of in-loop and across loop dead stores
13663 and properly fail for loop-variant refs. Handle byte-tracking with
13664 multiple defs. Use PARAM_DSE_MAX_ALIAS_QUERIES_PER_STORE for
13667 2018-05-16 Richard Sandiford <richard.sandiford@linaro.org>
13669 * tree-vectorizer.h (vect_get_vector_types_for_stmt): Declare.
13670 (vect_get_mask_type_for_stmt): Likewise.
13671 * tree-vect-slp.c (vect_two_operations_perm_ok_p): New function,
13673 (vect_build_slp_tree_1): ...here. Use vect_get_vector_types_for_stmt
13674 to determine the statement's vector type and the vector type that
13675 should be used for calculating nunits. Deal with cases in which
13676 the type has to be deferred.
13677 (vect_slp_analyze_node_operations): Use vect_get_vector_types_for_stmt
13678 and vect_get_mask_type_for_stmt to calculate STMT_VINFO_VECTYPE.
13679 * tree-vect-loop.c (vect_determine_vf_for_stmt_1)
13680 (vect_determine_vf_for_stmt): New functions, split out from...
13681 (vect_determine_vectorization_factor): ...here.
13682 * tree-vect-stmts.c (vect_get_vector_types_for_stmt)
13683 (vect_get_mask_type_for_stmt): New functions, split out from
13684 vect_determine_vectorization_factor.
13686 2018-05-16 Richard Biener <rguenther@suse.de>
13688 * tree-cfg.c (verify_gimple_assign_ternary): Properly
13689 verify the [VEC_]COND_EXPR embedded comparison.
13691 2018-05-15 Martin Sebor <msebor@redhat.com>
13693 PR tree-optimization/85753
13694 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Handle
13695 RECORD_TYPE in addition to ARRAY_TYPE.
13697 2018-05-15 Martin Sebor <msebor@redhat.com>
13699 PR middle-end/85643
13700 * calls.c (get_attr_nonstring_decl): Handle MEM_REF.
13702 2018-05-15 Richard Biener <rguenther@suse.de>
13704 * tree-ssa-dse.c (dse_classify_store): Remove use_stmt parameter,
13705 add by_clobber_p one. Change algorithm to collect all defs
13706 representing uses we need to walk and try reducing them to
13707 a single one before failing.
13708 (dse_dom_walker::dse_optimize_stmt): Adjust.
13710 2018-05-13 Mark Wielaard <mark@klomp.org>
13712 * dwarf2out.c (dwarf_OP): Handle DW_OP_addrx and DW_OP_constx.
13713 (size_of_loc_descr): Likewise.
13714 (output_loc_operands): Likewise.
13715 (output_loc_operands_raw): Likewise.
13716 (dw_addr_op): Use dwarf_OP () for DW_OP_constx and DW_OP_addrx.
13717 (resolve_addr_in_expr): Handle DW_OP_addrx and DW_OP_constx.
13718 (hash_loc_operands): Likewise.
13719 (compare_loc_operands): Likewise.
13721 2018-05-14 Mark Wielaard <mark@klomp.org>
13723 * dwarf2out.c (count_index_addrs): New function.
13724 (dwarf2out_finish): Use count_index_addrs to calculate addrs_length.
13726 2018-05-15 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
13728 PR tree-optimization/83648
13729 * ipa-pure-const.c (malloc_candidate_p): Allow function with NULL
13730 return value as malloc candidate.
13732 2018-05-15 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
13735 * ipa-pure-const.c (warn_function_malloc): Pass value of known_finite
13736 param as true in call to suggest_attribute.
13738 2018-05-14 Segher Boessenkool <segher@kernel.crashing.org>
13740 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete @itemx for
13741 -mreadonly-in-sdata.
13743 2018-05-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13745 * config/aarch64/aarch64-simd.md (*aarch64_crypto_aese_fused):
13747 (aarch64_crypto_aesd_fused): Likewise.
13749 2018-05-14 Wilco Dijkstra <wdijkstr@arm.com>
13751 * config/aarch64/aarch64.md (mov<mode>): Remove '*' in alternatives.
13752 (movsi_aarch64): Likewise.
13753 (load_pairsi): Likewise.
13754 (load_pairdi): Likewise.
13755 (store_pairsi): Likewise.
13756 (store_pairdi): Likewise.
13757 (load_pairsf): Likewise.
13758 (load_pairdf): Likewise.
13759 (store_pairsf): Likewise.
13760 (store_pairdf): Likewise.
13761 (zero_extend): Likewise.
13762 (trunc): Swap alternatives.
13763 (fcvt_target): Add '?' to prefer w over r.
13765 2018-05-14 Jakub Jelinek <jakub@redhat.com>
13768 * config/i386/i386.md: Disallow non-commutative arithmetics in
13769 last twpeephole for mem {+,-,&,|,^}= x; mem != 0 after cmpelim
13770 optimization. Use COMMUTATIVE_ARITH_P test rather than != MINUS
13771 in the peephole2 before it.
13773 2018-05-14 Sebastian Peryt <sebastian.peryt@intel.com>
13775 * common/config/i386/i386-common.c (OPTION_MASK_ISA_CLDEMOTE_SET,
13776 OPTION_MASK_ISA_CLDEMOTE_UNSET): New defines.
13777 (ix86_handle_option): Handle -mcldemote.
13778 * config.gcc: New header.
13779 * config/i386/cldemoteintrin.h: New file.
13780 * config/i386/cpuid.h (bit_CLDEMOTE): New bit.
13781 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
13783 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
13784 OPTION_MASK_ISA_CLDEMOTE.
13785 * config/i386/i386.c (ix86_target_string): Add -mcldemote.
13786 (ix86_valid_target_attribute_inner_p): Ditto.
13787 (enum ix86_builtins): Add IX86_BUILTIN_CLDEMOTE.
13788 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_cldemote.
13789 (ix86_expand_builtin): Expand IX86_BUILTIN_CLDEMOTE.
13790 * config/i386/i386.h (TARGET_CLDEMOTE, TARGET_CLDEMOTE_P): New.
13791 * config/i386/i386.md (UNSPECV_CLDEMOTE): New.
13793 * config/i386/i386.opt: Add -mcldemote.
13794 * config/i386/x86intrin.h: New header.
13795 * doc/invoke.texi: Add -mcldemote.
13797 2018-05-14 Richard Biener <rguenther@suse.de>
13799 * doc/match-and-simplify.texi: Adjust :s documentation.
13801 2018-05-14 Alexander Monakov <amonakov@ispras.ru>
13803 * sort.cc (REORDER_23): Pass the type for the temporaries instead of
13804 intended memcpy size.
13805 (REORDER_45): Likewise.
13807 2018-05-13 Alexander Monakov <amonakov@ispras.ru>
13809 * sort.cc: New file.
13810 * system.h [!CHECKING_P] (qsort): Redirect to gcc_qsort.
13811 * vec.c (qsort_chk): Use gcc_qsort.
13812 * Makefile.in (OBJS-libcommon): Add sort.o.
13813 (build/sort.o): New target. Use it...
13814 (BUILD_RTL): ... here, and...
13815 (build/gencfn-macros): ... here, and...
13816 (build/genmatch): ... here.
13818 2018-05-13 Kito Cheng <kito.cheng@gmail.com>
13819 Chung-Ju Wu <jasonwucj@gmail.com>
13821 * config.gcc (nds32*-*-*): Check that n15 is valid to --with-cpu.
13822 * config/nds32/nds32-graywolf.md: New file.
13823 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_GRAYWOLF.
13824 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n15
13826 * config/nds32/nds32-protos.h: More declarations for n15 pipeline.
13827 * config/nds32/nds32-utils.c: More implementations for n15 pipeline.
13828 * config/nds32/nds32.md (pipeline_model): Add graywolf.
13829 * config/nds32/nds32.opt (mcpu): Support n15 pipeline cpus.
13830 * config/nds32/pipelines.md: Include n15 settings.
13832 2018-05-13 Kito Cheng <kito.cheng@gmail.com>
13833 Chung-Ju Wu <jasonwucj@gmail.com>
13835 * config.gcc (nds32*-*-*): Check that n12/n13 are valid to --with-cpu.
13836 * config/nds32/nds32-n13.md: New file.
13837 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N12 and CPU_N13.
13838 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n12/n13
13840 * config/nds32/nds32-protos.h: More declarations for n12/n13 pipeline.
13841 * config/nds32/nds32.md (pipeline_model): Add n13.
13842 * config/nds32/nds32.opt (mcpu): Support n12/n13 pipeline cpus.
13843 * config/nds32/pipelines.md: Include n13 settings.
13845 2018-05-13 Kito Cheng <kito.cheng@gmail.com>
13846 Chung-Ju Wu <jasonwucj@gmail.com>
13848 * config.gcc (nds32*-*-*): Check that n10/d10 are valid to --with-cpu.
13849 * config/nds32/nds32-n10.md: New file.
13850 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N10.
13851 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n10
13853 * config/nds32/nds32-protos.h: More declarations for n10 pipeline.
13854 * config/nds32/nds32-utils.c: More implementations for n10 pipeline.
13855 * config/nds32/nds32.md (pipeline_model): Add n10.
13856 * config/nds32/nds32.opt (mcpu): Support n10 pipeline cpus.
13857 * config/nds32/pipelines.md: Include n10 settings.
13859 2018-05-13 Monk Chiang <sh.chiang04@gmail.com>
13860 Kito Cheng <kito.cheng@gmail.com>
13861 Chung-Ju Wu <jasonwucj@gmail.com>
13863 * config.gcc (nds32be-*-*): Handle --with-ext-dsp.
13864 * config/nds32/constants.md (unspec_element, unspec_volatile_element):
13865 Add enum values for DSP extension instructions.
13866 * config/nds32/constraints.md (Iu06, IU06, CVp5, CVs5, CVs2, CVhi):
13868 * config/nds32/iterators.md (shifts, shiftrt, sat_plus, all_plus,
13869 sat_minus, all_minus, plus_minus, extend, sumax, sumin, sumin_max):
13870 New code iterators.
13871 (su, zs, uk, opcode, add_rsub, add_sub): New code attributes.
13872 * config/nds32/nds32-dspext.md: New file for DSP implementation.
13873 * config/nds32/nds32-intrinsic.c: Implementation of DSP extension.
13874 * config/nds32/nds32-intrinsic.md: Likewise.
13875 * config/nds32/nds32_intrinsic.h: Likewise.
13876 * config/nds32/nds32-md-auxiliary.c: Likewise.
13877 * config/nds32/nds32-memory-manipulation.c: Consider DSP extension.
13878 * config/nds32/nds32-predicates.c (const_vector_to_hwint): New.
13879 (nds32_valid_CVp5_p, nds32_valid_CVs5_p): New.
13880 (nds32_valid_CVs2_p, nds32_valid_CVhi_p): New.
13881 * config/nds32/nds32-protos.h: New declarations for DSP extension.
13882 * config/nds32/nds32-utils.c (extract_mac_non_acc_rtx): New case
13883 TYPE_DMAC in switch statement.
13884 * config/nds32/nds32.c: New checking and implementation for DSP
13885 extension instructions.
13886 * config/nds32/nds32.h: Likewise.
13887 * config/nds32/nds32.md: Likewise.
13888 * config/nds32/nds32.opt (mhw-abs, mext-dsp): New options.
13889 * config/nds32/predicates.md: Implement new predicates for DSP
13892 2018-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
13894 * config/rs6000/rs6000.md (mov<mode>_softfloat, FMOVE32):
13895 Reformat alternatives and attributes so it is easier to identify
13896 which constraints/attributes go with which instruction.
13897 (mov<mode>_hardfloat32, FMOVE64): Likewise.
13898 (mov<mode>_softfloat32, FMOVE64): Likewise.
13899 (mov<mode>_hardfloat64, FMOVE64): Likewise.
13900 (mov<mode>_softfloat64, FMOVE64): Likewise.
13902 2018-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
13904 * doc/extend.texi (PowerPC Built-in Functions): Rename this
13906 (Basic PowerPC Built-in Functions): The new name of the
13907 subsection previously known as "PowerPC Built-in Functions".
13908 (Basic PowerPC Built-in Functions Available on all Configurations):
13910 (Basic PowerPC Built-in Functions Available on ISA 2.05): Likewise.
13911 (Basic PowerPC Built-in Functions Available on ISA 2.06): Likewise.
13912 (Basic PowerPC Built-in Functions Available on ISA 2.07): Likewise.
13913 (Basic PowerPC Built-in Functions Available on ISA 3.0): Likewise.
13915 2018-05-11 Martin Jambor <mjambor@suse.cz>
13918 * ipa-cp.c (intersect_with_plats): Check that the lattice contains
13921 2018-05-11 Richard Earnshaw <rearnsha@arm.com>
13924 * config/arm/arm-cpus.in (fgroup ARMv6m): Add be8 feature.
13926 2018-05-11 Sebastian Peryt <sebastian.peryt@intel.com>
13928 * common/config/i386/i386-common.c (OPTION_MASK_ISA_WAITPKG_SET,
13929 OPTION_MASK_ISA_WAITPKG_UNSET): New defines.
13930 (ix86_handle_option): Handle -mwaitpkg.
13931 * config.gcc: New header.
13932 * config/i386/cpuid.h (bit_WAITPKG): New bit.
13933 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mwaitpkg.
13934 * config/i386/i386-builtin-types.def ((UINT8, UNSIGNED, UINT64)): New
13936 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
13937 OPTION_MASK_ISA_WAITPKG.
13938 * config/i386/i386.c (ix86_target_string): Add -mwaitpkg.
13939 (ix86_option_override_internal): Add PTA_WAITPKG.
13940 (ix86_valid_target_attribute_inner_p): Add -mwaitpkg.
13941 (enum ix86_builtins): Add IX86_BUILTIN_UMONITOR, IX86_BUILTIN_UMWAIT,
13942 IX86_BUILTIN_TPAUSE.
13943 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_umonitor,
13944 __builtin_ia32_umwait and __builtin_ia32_tpause.
13945 (ix86_expand_builtin): Expand IX86_BUILTIN_UMONITOR,
13946 IX86_BUILTIN_UMWAIT, IX86_BUILTIN_TPAUSE.
13947 * config/i386/i386.h (TARGET_WAITPKG, TARGET_WAITPKG_P): New.
13948 * config/i386/i386.md (UNSPECV_UMWAIT, UNSPECV_UMONITOR,
13949 UNSPECV_TPAUSE): New.
13950 (umwait, umwait_rex64, umonitor_<mode>, tpause, tpause_rex64): New.
13951 * config/i386/i386.opt: Add -mwaitpkg.
13952 * config/i386/waitpkgintrin.h: New file.
13953 * config/i386/x86intrin.h: New header.
13954 * doc/invoke.texi: Add -mwaitpkg.
13956 2018-05-11 Richard Earnshaw <rearnsha@arm.com>
13959 * config/arm/arm-cpus.in: Add comment that ARMv6-m and ARMv6S-m are now
13961 (cortex-m0): Use armv6s-m isa.
13962 (cortex-m0plus): Likewise.
13963 (cortex-m1): Likewise.
13964 (cortex-m0.small-multiply): Likewise.
13965 (cortex-m0plus.small-multiply): Likewise.
13966 (cortex-m1.small-multiply): Likewise.
13968 2018-05-11 Allan Sandfeld Jensen <allan.jensen@qt.io>
13969 Jakub Jelinek <jakub@redhat.com>
13971 PR tree-optimization/85692
13972 * tree-ssa-forwprop.c (simplify_vector_constructor): Try two
13973 source permute as well.
13975 2018-05-11 Martin Liska <mliska@suse.cz>
13978 * doc/extend.texi: Document LLVM style format for no_sanitize
13981 2018-05-10 Michael Meissner <meissner@linux.vnet.ibm.com>
13983 * config/rs6000/rs6000.c (mode_supports_dq_form): Rename
13984 mode_supports_vsx_dform_quad to mode_supports_dq_form.
13985 (mode_supports_vsx_dform_quad): Likewise.
13986 (mode_supports_vmx_dform): Move these functions to be next to the
13987 other mode_supports functions.
13988 (mode_supports_dq_form): Likewise.
13989 (quad_address_p): Change calls of mode_supports_vsx_dform_quad to
13990 mode_supports_dq_form.
13991 (reg_offset_addressing_ok_p): Likewise.
13992 (offsettable_ok_by_alignment): Likewise.
13993 (rs6000_legitimate_offset_address_p): Likewise.
13994 (legitimate_lo_sum_address_p): Likewise.
13995 (rs6000_legitimize_address): Likewise.
13996 (rs6000_legitimize_reload_address): Likewise.
13997 (rs6000_secondary_reload_inner): Likewise.
13998 (rs6000_preferred_reload_class): Likewise.
13999 (rs6000_output_move_128bit): Likewise.
14001 2018-05-10 Uros Bizjak <ubizjak@gmail.com>
14003 * config/i386/i386.c (ix86_expand_builtin) <case IX86_BUILTIN_RDPID>:
14004 Generate SImode target register for null target.
14005 <case IX86_BUILTIN_XGETBV>: Ditto.
14006 <case IX86_BUILTIN_XSETBV>: Optimize LSHIFTRT generation.
14007 * config/i386/xsaveintrin.h (_xgetbv): Add missing return.
14009 2018-05-10 Carl Love <cel@us.ibm.com>
14011 * config/rs6000/rs6000.md (prefetch): Generate ISA 2.06 instructions
14012 dcbtt and dcbtstt if operands[2] is 0.
14014 2018-05-10 Uros Bizjak <ubizjak@gmail.com>
14017 * config/i386/sse.md (usadv64qi): New expander.
14019 2018-05-10 Segher Boessenkool <segher@kernel.crashing.org>
14021 * config/rs6000/altivec.md (altivec_vmrghb, altivec_vmrghh,
14022 altivec_vmrghw, altivec_vmrglb, altivec_vmrglh, altivec_vmrglw): Remove
14023 -maltivec=be support.
14024 (vec_widen_umult_even_v16qi, vec_widen_smult_even_v16qi,
14025 vec_widen_umult_even_v8hi, vec_widen_smult_even_v8hi,
14026 vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
14027 vec_widen_umult_odd_v16qi, vec_widen_smult_odd_v16qi,
14028 vec_widen_umult_odd_v8hi, vec_widen_smult_odd_v8hi,
14029 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si, altivec_vpkpx,
14030 altivec_vpks<VI_char>ss, altivec_vpks<VI_char>us,
14031 altivec_vpku<VI_char>us, altivec_vpku<VI_char>um, altivec_vsum2sws,
14032 altivec_vsumsws): Adjust.
14033 (altivec_vspltb *altivec_vspltb_internal, altivec_vsplth,
14034 *altivec_vsplth_internal, altivec_vspltw, *altivec_vspltw_internal,
14035 altivec_vspltsf, *altivec_vspltsf_internal): Remove -maltivec=be
14037 (altivec_vperm_<mode>, altivec_vperm_<mode>_uns,
14038 altivec_vupkhs<VU_char>, altivec_vupkls<VU_char>, altivec_vupkhpx,
14039 altivec_vupklpx, altivec_lvsl, altivec_lvsr): Adjust.
14040 (altivec_lve<VI_char>x): Delete expand.
14041 (*altivec_lve<VI_char>x_internal): Rename to...
14042 (altivec_lve<VI_char>x): ... this.
14043 (altivec_lvxl_<mode>): Delete expand.
14044 (*altivec_lvxl_<mode>_internal): Rename to ...
14045 (altivec_lvxl_<mode>): ... this.
14046 (altivec_stvxl_<mode>): Delete expand.
14047 (*altivec_stvxl_<mode>_internal): Rename to ...
14048 (altivec_stvxl_<mode>): ... this.
14049 (altivec_stve<VI_char>x): Delete expand.
14050 (*altivec_stve<VI_char>x_internal): Rename to ...
14051 (altivec_stve<VI_char>x): ... this.
14052 (doublee<mode>2, unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2,
14053 doubleh<mode>2, unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2,
14054 reduc_plus_scal_<mode>): Adjust.
14055 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Adjust
14057 (rs6000_cpu_cpp_builtins): Adjust.
14058 (altivec_resolve_overloaded_builtin): Remove -maltivec=be support.
14059 * config/rs6000/rs6000-protos.h (altivec_expand_lvx_be,
14060 altivec_expand_stvx_be, altivec_expand_stvex_be): Delete.
14061 * config/rs6000/rs6000.c (rs6000_option_override_internal): Remove
14062 -maltivec=be support.
14063 (rs6000_split_vec_extract_var): Adjust.
14064 (rs6000_split_v4si_init): Adjust.
14065 (swap_selector_for_mode): Delete.
14066 (altivec_expand_lvx_be, altivec_expand_stvx_be,
14067 altivec_expand_stvex_be): Delete.
14068 (altivec_expand_lv_builtin, altivec_expand_stv_builtin): Remove
14069 -maltivec=be support.
14070 (rs6000_gimple_fold_builtin): Ditto.
14071 (rs6000_generate_float2_double_code, rs6000_generate_float2_code):
14073 * config/rs6000/rs6000.h (VECTOR_ELT_ORDER_BIG): Delete.
14074 (TARGET_DIRECT_MOVE_64BIT): Adjust.
14075 * config/rs6000/rs6000.md (split for extendsidi2 for vectors): Adjust.
14076 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Delete.
14077 * config/rs6000/vsx.md (floate<mode>, unsfloatev2di, floato<mode>,
14078 unsfloatov2di, vsignedo_v2df, vsignede_v2df, vunsignedo_v2df,
14079 vunsignede_v2df, vsx_extract_<mode>_p9, *vsx_extract_si,
14080 *vsx_extract_<mode>_p8, *vsx_extract_si_<uns>float_df,
14081 *vsx_extract_si_<uns>float_<mode>, vsx_set_<mode>_p9, vsx_set_v4sf_p9,
14082 *vsx_insert_extract_v4sf_p9, *vsx_insert_extract_v4sf_p9_2, and an
14083 anonymous split): Adjust.
14084 (vsx_mergel_<mode>, vsx_mergeh_<mode>): Remove -maltivec=be support.
14085 (vsx_xxspltd_<mode>, extract4b, insert4b): Adjust.
14087 2018-05-10 Eric Botcazou <ebotcazou@adacore.com>
14089 * configure.ac (gcc_gxx_include_dir_add_sysroot): Set it to 1 only
14090 when --with-gxx-include-dir is also specified.
14091 * configure: Regenerate.
14093 2018-05-09 Jim Wilson <jimw@sifive.com>
14096 * config.gcc (riscv*-*-*): Handle --with-multilib-list.
14097 * config/riscv/t-withmultilib: New.
14098 * config/riscv/withmultilib.h: New.
14099 * doc/install.texi: Document RISC-V --with-multilib-list support.
14101 2018-05-09 Richard Biener <rguenther@suse.de>
14103 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Fill a cost
14105 (vect_bb_vectorization_profitable_p): Adjust. Compute
14106 actual scalar cost using the cost vector and the add_stmt_cost
14109 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
14111 PR rtl-optimization/85645
14112 * config/rs6000/rs6000.c (rs6000_emit_prologue_components): Put a SET
14113 in the REG_CFA_REGISTER note for LR, don't leave it empty.
14115 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
14117 PR rtl-optimization/85645
14118 * shrink-wrap.c (spread_components): Return a boolean saying if
14119 anything was changed.
14120 (try_shrink_wrapping_separate): Iterate spread_components until
14121 nothing changes anymore.
14123 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
14125 PR rtl-optimization/85645
14126 * regrename.c (build_def_use): Also kill the chains that include the
14127 destination of a REG_CFA_REGISTER note.
14129 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
14131 PR rtl-optimization/85645
14132 * regcprop.c (copyprop_hardreg_forward_1): Don't propagate into an
14133 insn that has a REG_CFA_REGISTER note.
14135 2018-05-09 Richard Sandiford <richard.sandiford@linaro.org>
14137 * cfgexpand.c (expand_clobber): New function.
14138 (expand_gimple_stmt_1): Use it.
14139 * tree-vect-stmts.c (vect_clobber_variable): New function,
14141 (vectorizable_simd_clone_call): ...here.
14142 (vectorizable_store): Emit a clobber either side of an
14143 IFN_STORE_LANES sequence.
14144 (vectorizable_load): Emit a clobber after an IFN_LOAD_LANES sequence.
14146 2018-05-09 Tom de Vries <tom@codesourcery.com>
14149 * config/nvptx/nvptx.md (define_insn "trap", define_insn "trap_if_true")
14150 (define_insn "trap_if_false"): Add exit after trap.
14152 2018-05-09 Eric Botcazou <ebotcazou@adacore.com>
14154 PR rtl-optimization/85638
14155 * bb-reorder.c: Include common/common-target.h.
14156 (create_forwarder_block): New function extracted from...
14157 (fix_up_crossing_landing_pad): ...here. Rename into...
14158 (dw2_fix_up_crossing_landing_pad): ...this.
14159 (sjlj_fix_up_crossing_landing_pad): New function.
14160 (find_rarely_executed_basic_blocks_and_crossing_edges): In SJLJ mode,
14161 call sjlj_fix_up_crossing_landing_pad if there are incoming EH edges
14162 from both partitions and exit the loop after one iteration.
14164 2018-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
14167 * doc/extend.texi (PowerPC Built-in Functions): Rename this
14169 (Basic PowerPC Built-in Functions): The new name of the
14170 subsection previously known as "PowerPC Built-in Functions".
14171 (Basic PowerPC Built-in Functions Available on all Configurations):
14173 (Basic PowerPC Built-in Functions Available on ISA 2.05): New
14175 (Basic PowerPC Built-in Functions Available on ISA 2.06): New
14177 (Basic PowerPC Built-in Functions Available on ISA 2.07): New
14179 (Basic PowerPC Built-in Functions Available on ISA 3.0): New
14182 2018-05-08 Jim Wilson <jimw@sifive.com>
14184 * config/riscv/linux.h (MUSL_ABI_SUFFIX): Delete unnecessary backslash.
14185 (LD_EMUL_SUFFIX): New.
14186 (LINK_SPEC): Use it.
14188 2018-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
14190 * doc/extend.texi (PowerPC Built-in Functions): Rename this
14192 (Basic PowerPC Built-in Functions): The new name of the
14193 subsection previously known as "PowerPC Built-in Functions".
14194 (Basic PowerPC Built-in Functions Available on all Configurations):
14196 (Basic PowerPC Built-in Functions Available on ISA 2.05): New
14198 (Basic PowerPC Built-in Functions Available on ISA 2.06): New
14200 (Basic PowerPC Built-in Functions Available on ISA 2.07): New
14202 (Basic PowerPC Built-in Functions Available on ISA 3.0): New
14205 2018-05-08 Jakub Jelinek <jakub@redhat.com>
14208 * config/i386/i386.md: Add peepholes for mem {+,-,&,|,^}= x; mem != 0
14209 after cmpelim optimization.
14211 2018-05-08 Olga Makhotina <olga.makhotina@intel.com>
14213 * config.gcc: Support "goldmont".
14214 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "goldmont".
14215 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
14216 PROCESSOR_GOLDMONT.
14217 * config/i386/i386.c (m_GOLDMONT): Define.
14218 (processor_target_table): Add "goldmont".
14219 (PTA_GOLDMONT): Define.
14220 (ix86_lea_outperforms): Add TARGET_GOLDMONT.
14221 (get_builtin_code_for_version): Handle PROCESSOR_GOLDMONT.
14222 (fold_builtin_cpu): Add M_INTEL_GOLDMONT.
14223 (fold_builtin_cpu): Add "goldmont".
14224 (ix86_add_stmt_cost): Add TARGET_GOLDMONT.
14225 (ix86_option_override_internal): Add "goldmont".
14226 * config/i386/i386.h (processor_costs): Define TARGET_GOLDMONT.
14227 (processor_type): Add PROCESSOR_GOLDMONT.
14228 * config/i386/i386.md: Add CPU "glm".
14229 * config/i386/glm.md: New file.
14230 * config/i386/x86-tune.def: Add m_GOLDMONT.
14231 * doc/invoke.texi: Add goldmont as x86 -march=/-mtune= CPU type.
14233 2018-05-08 Jakub Jelinek <jakub@redhat.com>
14236 * config/i386/i386.c (ix86_expand_sse2_abs): Handle E_V2DImode and
14238 * config/i386/sse.md (abs<mode>2): Use VI_AVX2 iterator instead of
14239 VI1248_AVX512VL_AVX512BW. Handle V2DImode and V4DImode if not
14240 TARGET_AVX512VL using ix86_expand_sse2_abs. Formatting fixes.
14243 * config/i386/i386.c (ix86_fold_builtin): Handle
14244 IX86_BUILTIN_{,P}MOVMSK{PS,PD,B}{,128,256}.
14247 * config/i386/sse.md (ssequaterinsnmode): New mode attribute.
14248 (*<extract_type>_vinsert<shuffletype><extract_suf>_0): New pattern.
14250 2018-05-08 Richard Earnshaw <rearnsha@arm.com>
14253 * config/arm/parsecpu.awk (check_cpu): Fix operator precedence.
14254 (check_arch): Likewise.
14255 (check_fpu): Return the result rather than printing it.
14256 (end arch): Fix operator precedence.
14257 (end cpu): Likewise.
14258 (END): Print the result from check_fpu.
14260 2018-05-08 Richard Sandiford <richard.sandiford@linaro.org>
14261 Alan Hayward <alan.hayward@arm.com>
14262 David Sherwood <david.sherwood@arm.com>
14264 * config/aarch64/aarch64-sve.md (*pred_cmp<cmp_op><mode>_combine)
14265 (*pred_cmp<cmp_op><mode>, *fcm<cmp_op><mode>_and_combine)
14266 (*fcmuo<mode>_and_combine, *fcm<cmp_op><mode>_and)
14267 (*fcmuo<mode>_and): New patterns.
14269 2018-05-08 Richard Sandiford <richard.sandiford@linaro.org>
14271 * config/aarch64/iterators.md (UNSPEC_COND_LO, UNSPEC_COND_LS)
14272 (UNSPEC_COND_HI, UNSPEC_COND_HS, UNSPEC_COND_UO): Delete.
14273 (SVE_INT_CMP, SVE_FP_CMP): New code iterators.
14274 (cmp_op, sve_imm_con): New code attributes.
14275 (SVE_COND_INT_CMP, imm_con): Delete.
14276 (cmp_op): Remove above unspecs from int attribute.
14277 * config/aarch64/aarch64-sve.md (*vec_cmp<cmp_op>_<mode>): Rename
14279 (*cmp<cmp_op><mode>): ...this. Use UNSPEC_MERGE_PTRUE instead of
14280 comparison-specific unspecs.
14281 (*vec_cmp<cmp_op>_<mode>_ptest): Rename to...
14282 (*cmp<cmp_op><mode>_ptest): ...this and adjust likewise.
14283 (*vec_cmp<cmp_op>_<mode>_cc): Rename to...
14284 (*cmp<cmp_op><mode>_cc): ...this and adjust likewise.
14285 (*vec_fcm<cmp_op><mode>): Rename to...
14286 (*fcm<cmp_op><mode>): ...this and adjust likewise.
14287 (*vec_fcmuo<mode>): Rename to...
14288 (*fcmuo<mode>): ...this and adjust likewise.
14289 (*pred_fcm<cmp_op><mode>): New pattern.
14290 * config/aarch64/aarch64.c (aarch64_emit_unop, aarch64_emit_binop)
14291 (aarch64_emit_sve_ptrue_op, aarch64_emit_sve_ptrue_op_cc): New
14293 (aarch64_unspec_cond_code): Remove handling of LTU, GTU, LEU, GEU
14295 (aarch64_gen_unspec_cond, aarch64_emit_unspec_cond): Delete.
14296 (aarch64_emit_sve_predicated_cond): New function.
14297 (aarch64_expand_sve_vec_cmp_int): Use aarch64_emit_sve_ptrue_op_cc.
14298 (aarch64_emit_unspec_cond_or): Replace with...
14299 (aarch64_emit_sve_or_conds): ...this new function. Use
14300 aarch64_emit_sve_ptrue_op for the individual comparisons and
14301 aarch64_emit_binop to OR them together.
14302 (aarch64_emit_inverted_unspec_cond): Replace with...
14303 (aarch64_emit_sve_inverted_cond): ...this new function. Use
14304 aarch64_emit_sve_ptrue_op for the comparison and
14305 aarch64_emit_unop to invert the result.
14306 (aarch64_expand_sve_vec_cmp_float): Update after the above
14307 changes. Use aarch64_emit_sve_ptrue_op for native comparisons.
14309 2018-05-07 Nathan Sidwell <nathan@acm.org>
14311 * doc/invoke.texi (C++ Dialect Options): Remove -ffor-scope.
14312 * doc/extend.texi (Deprecated Features): Remove -fno-for-scope
14313 (Backwards Compatibility): Likewise.
14315 2018-05-07 Luis Machado <luis.machado@linaro.org>
14319 2018-05-07 Luis Machado <luis.machado@linaro.org>
14321 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
14322 <prefetch_dynamic_strides>: New const bool field.
14323 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
14324 prefetch_dynamic_strides.
14325 (exynosm1_prefetch_tune): Likewise.
14326 (thunderxt88_prefetch_tune): Likewise.
14327 (thunderx_prefetch_tune): Likewise.
14328 (thunderx2t99_prefetch_tune): Likewise.
14329 (qdf24xx_prefetch_tune): Likewise. Set prefetch_dynamic_strides
14331 (aarch64_override_options_internal): Update to set
14332 PARAM_PREFETCH_DYNAMIC_STRIDES.
14333 * doc/invoke.texi (prefetch-dynamic-strides): Document new option.
14334 * params.def (PARAM_PREFETCH_DYNAMIC_STRIDES): New.
14335 * params.h (PARAM_PREFETCH_DYNAMIC_STRIDES): Define.
14336 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Account for
14337 prefetch-dynamic-strides setting.
14339 2018-05-07 Luis Machado <luis.machado@linaro.org>
14341 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
14342 <minimum_stride>: New const int field.
14343 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
14344 minimum_stride field.
14345 (exynosm1_prefetch_tune): Likewise.
14346 (thunderxt88_prefetch_tune): Likewise.
14347 (thunderx_prefetch_tune): Likewise.
14348 (thunderx2t99_prefetch_tune): Likewise.
14349 (qdf24xx_prefetch_tune): Likewise. Set minimum_stride to 2048.
14350 (aarch64_override_options_internal): Update to set
14351 PARAM_PREFETCH_MINIMUM_STRIDE.
14352 * doc/invoke.texi (prefetch-minimum-stride): Document new option.
14353 * params.def (PARAM_PREFETCH_MINIMUM_STRIDE): New.
14354 * params.h (PARAM_PREFETCH_MINIMUM_STRIDE): Define.
14355 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Return false if
14356 stride is constant and is below the minimum stride threshold.
14358 2018-05-07 Luis Machado <luis.machado@linaro.org>
14360 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune) <l2_cache_size>: Set
14363 2018-05-07 Luis Machado <luis.machado@linaro.org>
14365 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
14366 <prefetch_dynamic_strides>: New const bool field.
14367 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
14368 prefetch_dynamic_strides.
14369 (exynosm1_prefetch_tune): Likewise.
14370 (thunderxt88_prefetch_tune): Likewise.
14371 (thunderx_prefetch_tune): Likewise.
14372 (thunderx2t99_prefetch_tune): Likewise.
14373 (qdf24xx_prefetch_tune): Likewise. Set prefetch_dynamic_strides
14375 (aarch64_override_options_internal): Update to set
14376 PARAM_PREFETCH_DYNAMIC_STRIDES.
14377 * doc/invoke.texi (prefetch-dynamic-strides): Document new option.
14378 * params.def (PARAM_PREFETCH_DYNAMIC_STRIDES): New.
14379 * params.h (PARAM_PREFETCH_DYNAMIC_STRIDES): Define.
14380 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Account for
14381 prefetch-dynamic-strides setting.
14383 2018-05-07 Luis Machado <luis.machado@linaro.org>
14385 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
14386 <minimum_stride>: New const int field.
14387 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
14388 minimum_stride field.
14389 (exynosm1_prefetch_tune): Likewise.
14390 (thunderxt88_prefetch_tune): Likewise.
14391 (thunderx_prefetch_tune): Likewise.
14392 (thunderx2t99_prefetch_tune): Likewise.
14393 (qdf24xx_prefetch_tune): Likewise. Set minimum_stride to 2048.
14394 (aarch64_override_options_internal): Update to set
14395 PARAM_PREFETCH_MINIMUM_STRIDE.
14396 * doc/invoke.texi (prefetch-minimum-stride): Document new option.
14397 * params.def (PARAM_PREFETCH_MINIMUM_STRIDE): New.
14398 * params.h (PARAM_PREFETCH_MINIMUM_STRIDE): Define.
14399 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Return false if
14400 stride is constant and is below the minimum stride threshold.
14402 2018-05-06 Jakub Jelinek <jakub@redhat.com>
14405 * cfgexpand.c (expand_asm_stmt): Don't create a temporary if
14406 the type is addressable. Don't force op into register if it has
14409 2018-05-05 Roland McGrath <mcgrathr@google.com>
14412 * varasm.c (default_section_type_flags): Set SECTION_NOTYPE for
14413 any section for which we don't know a specific type it should have,
14414 regardless of name. Previously this was done only for the exact
14415 names ".init_array", ".fini_array", and ".preinit_array".
14416 (default_elf_asm_named_section): Add comment about
14417 relationship with default_section_type_flags and SECTION_NOTYPE.
14418 (get_section): Don't consider it a type conflict if one side has
14419 SECTION_NOTYPE and the other doesn't, as long as neither has the
14420 SECTION_BSS et al used in the default_section_type_flags logic.
14422 2018-05-05 Tom de Vries <tom@codesourcery.com>
14425 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG_3): Define.
14426 (workaround_barsyncs): New function.
14427 (nvptx_reorg): Use workaround_barsyncs.
14428 * config/nvptx/nvptx.md (define_c_enum "unspecv"): Add UNSPECV_MEMBAR.
14429 (define_expand "nvptx_membar_cta"): New define_expand.
14430 (define_insn "*nvptx_membar_cta"): New insn.
14432 2018-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
14434 * brig-builtins.def: Add consts to ptrs etc. in BRIG builtin defs.
14435 To improve optimization opportunities.
14436 * builtin-types.def: The new needed builtin types for the above.
14438 2018-05-04 Richard Biener <rguenther@suse.de>
14440 * bb-reorder.c (sanitize_hot_paths): Release hot_bbs_to_check.
14441 * gimple-ssa-store-merging.c
14442 (imm_store_chain_info::output_merged_store): Remove redundant create,
14443 release split_store vector contents on failure.
14444 * tree-vect-slp.c (vect_schedule_slp_instance): Avoid leaking
14445 scalar stmt vector on cache hit.
14447 2018-05-04 Segher Boessenkool <segher@kernel.crashing.org>
14449 * common/config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
14451 * config.gcc (powerpc-xilinx-eabi*): Remove.
14452 * config/rs6000/predicates.md (easy_fp_constant): Remove Xilinx FP
14454 (fusion_addis_mem_combo_load): Ditto.
14455 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove Xilinx
14457 (rs6000_cpu_cpp_builtins): Ditto.
14458 * config/rs6000/rs6000-linux.c
14459 (rs6000_linux_float_exceptions_rounding_supported_p): Ditto.
14460 * config/rs6000/rs6000-opts.h (enum fpu_type_t): Delete.
14461 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove Xilinx FP
14463 (rs6000_setup_reg_addr_masks): Ditto.
14464 (rs6000_init_hard_regno_mode_ok): Ditto.
14465 (rs6000_option_override_internal): Ditto.
14466 (legitimate_lo_sum_address_p): Ditto.
14467 (rs6000_legitimize_address): Ditto.
14468 (rs6000_legitimize_reload_address): Ditto.
14469 (rs6000_legitimate_address_p): Ditto.
14470 (abi_v4_pass_in_fpr): Ditto.
14471 (setup_incoming_varargs): Ditto.
14472 (rs6000_gimplify_va_arg): Ditto.
14473 (rs6000_split_multireg_move): Ditto.
14474 (rs6000_savres_strategy): Ditto.
14475 (rs6000_emit_prologue_components): Ditto.
14476 (rs6000_emit_epilogue_components): Ditto.
14477 (rs6000_emit_prologue): Ditto.
14478 (rs6000_emit_epilogue): Ditto.
14479 (rs6000_elf_file_end): Ditto.
14480 (rs6000_function_value): Ditto.
14481 (rs6000_libcall_value): Ditto.
14482 * config/rs6000/rs6000.h: Ditto.
14483 (TARGET_MINMAX_SF, TARGET_MINMAX_DF): Delete, merge to ...
14484 (TARGET_MINMAX): ... this. New.
14485 (TARGET_SF_FPR, TARGET_DF_FPR, TARGET_SF_INSN, TARGET_DF_INSN): Delete.
14486 * config/rs6000/rs6000.md: Remove Xilinx FP support.
14487 (*movsi_internal1_single): Delete.
14488 * config/rs6000/rs6000.opt (msingle-float, mdouble-float, msimple-fpu,
14489 mfpu=, mxilinx-fpu): Delete.
14490 * config/rs6000/singlefp.h: Delete.
14491 * config/rs6000/sysv4.h: Remove Xilinx FP support.
14492 * config/rs6000/t-rs6000: Ditto.
14493 * config/rs6000/t-xilinx: Delete.
14494 * config/rs6000/titan.md: Adjust for fp_type removal.
14495 * config/rs6000/vsx.md: Remove Xilinx FP support.
14496 (VStype_simple): Delete.
14497 (VSfptype_simple, VSfptype_mul, VSfptype_div, VSfptype_sqrt): Delete.
14498 * config/rs6000/xfpu.h: Delete.
14499 * config/rs6000/xfpu.md: Delete.
14500 * config/rs6000/xilinx.h: Delete.
14501 * config/rs6000/xilinx.opt: Delete.
14502 * doc/invoke.texi (RS/6000 and PowerPC Options): Remove
14503 -msingle-float, -mdouble-float, -msimple-fpu, -mfpu=, and -mxilinx-fpu.
14505 2018-05-04 Tom de Vries <tom@codesourcery.com>
14508 * builtins.c (expand_builtin_goacc_parlevel_id_size): Handle null target
14511 2018-05-04 Richard Biener <rguenther@suse.de>
14513 PR middle-end/85627
14514 * tree-complex.c (update_complex_assignment): We are always in SSA form.
14515 (expand_complex_div_wide): Likewise.
14516 (expand_complex_operations_1): Likewise.
14517 (expand_complex_libcall): Preserve EH info of the original stmt.
14518 (tree_lower_complex): Handle removed blocks.
14519 * tree.c (build_common_builtin_nodes): Do not set ECF_NOTRHOW
14520 on complex multiplication and division libcall builtins.
14522 2018-05-04 Richard Biener <rguenther@suse.de>
14524 PR middle-end/85574
14525 * fold-const.c (negate_expr_p): Restrict negation of operand
14526 zero of a division to when we know that can happen without
14528 (fold_negate_expr_1): Likewise.
14530 2018-05-04 Jakub Jelinek <jakub@redhat.com>
14533 * real.h (real_nextafter): Declare.
14534 * real.c (real_nextafter): New function.
14535 * fold-const-call.c (fold_const_nextafter): New function.
14536 (fold_const_call_sss): Call it for CASE_CFN_NEXTAFTER and
14537 CASE_CFN_NEXTTOWARD.
14538 (fold_const_call_1): For CASE_CFN_NEXTTOWARD call fold_const_call_sss
14539 even when arg1_mode is different from arg0_mode.
14541 2018-05-03 Nathan Sidwell <nathan@acm.org>
14543 * doc/extend.texi (Deprecated Features): Remove
14544 -ffriend-injection.
14545 (Backwards Compatibility): Likewise.
14546 * doc/invoke.texi (C++ Language Options): Likewise.
14547 (C++ Dialect Options): Likewise.
14549 2018-05-03 Jakub Jelinek <jakub@redhat.com>
14552 * config/i386/avx512fintrin.h (_mm512_mullox_epi64,
14553 _mm512_mask_mullox_epi64): New intrinsics.
14555 2018-05-03 Tom de Vries <tom@codesourcery.com>
14558 * doc/sourcebuild.texi (Commands for use in dg-final, Scan optimization
14559 dump files): Add offload-tree.
14561 2018-05-03 Richard Biener <rguenther@suse.de>
14563 PR tree-optimization/85615
14564 * tree-ssa-threadupdate.c (thread_block_1): Only allow exits
14565 to loops not nested in BBs loop father to avoid creating multi-entry
14568 2018-05-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14570 PR tree-optimization/70291
14571 * tree-complex.c (expand_complex_libcall): Add type, inplace_p
14572 arguments. Change return type to tree. Emit libcall as a new
14573 statement rather than replacing existing one when inplace_p is true.
14574 (expand_complex_multiplication_components): New function.
14575 (expand_complex_multiplication): Expand floating-point complex
14576 multiplication using the above.
14577 (expand_complex_division): Rename inner_type parameter to type.
14578 Update expand_complex_libcall call-site.
14579 (expand_complex_operations_1): Update expand_complex_multiplication
14580 and expand_complex_division call-sites.
14582 2018-05-02 Jakub Jelinek <jakub@redhat.com>
14585 * config/i386/i386.md (*ashl<dwi>3_doubleword_mask,
14586 *ashl<dwi>3_doubleword_mask_1, *<shift_insn><dwi>3_doubleword_mask,
14587 *<shift_insn><dwi>3_doubleword_mask_1): In condition require that
14588 the highest significant bit of the shift count mask is clear. In
14589 check whether and[sq]i3 is needed verify that all significant bits
14590 of the shift count other than the highest are set.
14592 2018-05-02 Tom de Vries <tom@codesourcery.com>
14595 * builtins.def (DEF_GOACC_BUILTIN_ONLY): Define.
14596 * omp-builtins.def (BUILT_IN_GOACC_PARLEVEL_ID)
14597 (BUILT_IN_GOACC_PARLEVEL_SIZE): New builtin.
14598 * builtins.c (expand_builtin_goacc_parlevel_id_size): New function.
14599 (expand_builtin): Call expand_builtin_goacc_parlevel_id_size.
14600 * doc/extend.texi (Other Builtins): Add __builtin_goacc_parlevel_id and
14601 __builtin_goacc_parlevel_size.
14603 2018-05-02 Richard Biener <rguenther@suse.de>
14605 PR tree-optimization/85597
14606 * tree-vect-stmts.c (vectorizable_operation): For ternary SLP
14607 do not use split vect_get_vec_defs call but call vect_get_slp_defs
14610 2018-05-02 Tom de Vries <tom@codesourcery.com>
14613 * doc/sourcebuild.texi (Commands for use in dg-final, Scan optimization
14614 dump files): Add ltrans-tree.
14616 2018-05-02 Tom de Vries <tom@codesourcery.com>
14619 * doc/sourcebuild.texi (Commands for use in dg-final, Scan optimization
14620 dump files): Add wpa-ipa.
14622 2018-05-02 Segher Boessenkool <segher@kernel.crashing.org>
14624 * config.gcc (powerpc*-*-*): Remove paired.h. Unsupport the
14625 powerpc*-*-linux*paired* target.
14626 * config/rs6000/750cl.h: Delete.
14627 * config/rs6000/paired.h: Delete.
14628 * config/rs6000/paired.md: Delete.
14629 * config/rs6000/predicates.md (easy_vector_constant): Remove paired
14631 * config/rs6000/rs6000-builtin.def: Remove paired float support.
14632 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
14633 comment. Remove paired float support.
14634 * config/rs6000/rs6000-modes.def: Remove V2SF and V2SI.
14635 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete
14637 * config/rs6000/rs6000-protos.h (paired_expand_vector_init,
14638 paired_emit_vector_cond_expr, paired_expand_vector_move): Delete
14640 * config/rs6000/rs6000.c: Remove paired float support.
14641 (paired_expand_vector_init, paired_expand_vector_move,
14642 paired_emit_vector_compare, paired_emit_vector_cond_expr,
14643 (paired_expand_lv_builtin, paired_expand_stv_builtin,
14644 paired_expand_builtin, paired_expand_predicate_builtin,
14645 paired_init_builtins): Delete.
14646 * config/rs6000/rs6000.h: Remove paired float support.
14647 * config/rs6000/rs6000.md: Remove paired float support.
14648 (move_from_CR_ov_bit): Delete.
14649 * config/rs6000/rs6000.opt (mpaired): Delete.
14650 * config/rs6000/t-rs6000: Remove paired.md from MD_INCLUDES.
14651 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mpaired.
14653 2018-05-02 Richard Biener <rguenther@suse.de>
14655 PR middle-end/85567
14656 * gimplify.c (gimplify_save_expr): When in SSA form allow
14657 SAVE_EXPRs to compute to SSA vars.
14659 2018-05-02 Jakub Jelinek <jakub@redhat.com>
14662 * config/i386/i386.md (*ashl<dwi>3_doubleword_mask,
14663 *ashl<dwi>3_doubleword_mask_1, *<shift_insn><dwi>3_doubleword_mask,
14664 *<shift_insn><dwi>3_doubleword_mask_1): If and[sq]i3 is needed, don't
14665 clobber operands[2], instead use a new pseudo. Formatting fixes.
14667 2018-05-02 Richard Sandiford <richard.sandiford@linaro.org>
14669 PR tree-optimization/85586
14670 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Only
14671 exit early for statements in the same group if the accesses are
14674 2018-05-02 Tom de Vries <tom@codesourcery.com>
14677 * lto-wrapper.c (compile_offload_image): Add "could not find mkoffload"
14680 2018-05-01 Marc Glisse <marc.glisse@inria.fr>
14682 PR tree-optimization/85143
14683 * match.pd (A<B&A<C): Extend to BIT_IOR_EXPR.
14685 2018-05-01 Tom de Vries <tom@codesourcery.com>
14688 * config/nvptx/mkoffload.c (main): Suggest using -B in "offload compiler
14689 not found" error message.
14691 2018-05-01 Tom de Vries <tom@codesourcery.com>
14694 * vec.h (VEC_ORDERED_REMOVE_IF, VEC_ORDERED_REMOVE_IF_FROM_TO): Define.
14695 * vec.c (test_ordered_remove_if): New function.
14696 (vec_c_tests): Call test_ordered_remove_if.
14697 * dwarf2cfi.c (connect_traces): Use VEC_ORDERED_REMOVE_IF_FROM_TO.
14698 * lto-streamer-out.c (prune_offload_funcs): Use VEC_ORDERED_REMOVE_IF.
14699 * tree-vect-patterns.c (vect_pattern_recog_1): Use
14700 VEC_ORDERED_REMOVE_IF.
14702 2018-05-01 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
14704 PR tree-optimization/82665
14705 * vr-values.c (vr_values::extract_range_from_binary_expr): Handle
14706 pointer subtraction where arguments come from a memchr call.
14708 2018-05-01 Jakub Jelinek <jakub@redhat.com>
14710 * configure.ac (LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION): Use
14711 --push-state --as-needed and --pop-state instead of --as-needed and
14712 --no-as-needed if ld supports it.
14713 * configure: Regenerated.
14716 * doc/install.texi2html: Replace _002d with - and _002a with * in
14717 generated html files using sed.
14719 2018-04-30 David Malcolm <dmalcolm@redhat.com>
14722 * gcc-rich-location.c (blank_line_before_p): New function.
14723 (use_new_line): New function.
14724 (gcc_rich_location::add_fixit_insert_formatted): New function.
14725 * gcc-rich-location.h
14726 (gcc_rich_location::add_fixit_insert_formatted): New function.
14728 2018-04-30 David Malcolm <dmalcolm@redhat.com>
14730 * selftest.c (assert_streq): Rename "expected" and "actual" to
14731 "val1" and "val2". Extend NULL-handling to cover both inputs
14732 symmetrically, while still requiring both to be non-NULL for a pass.
14733 * selftest.h (assert_streq): Rename "expected" and "actual" to
14735 (ASSERT_EQ): Likewise.
14736 (ASSERT_EQ_AT): Likewise.
14737 (ASSERT_KNOWN_EQ): Likewise.
14738 (ASSERT_KNOWN_EQ_AT): Likewise.
14739 (ASSERT_NE): Likewise.
14740 (ASSERT_MAYBE_NE): Likewise.
14741 (ASSERT_MAYBE_NE_AT): Likewise.
14742 (ASSERT_STREQ): Likewise. Clarify that both must be non-NULL for
14743 the assertion to pass.
14744 (ASSERT_STREQ_AT): Likewise.
14746 2018-04-30 Jonathan Wakely <jwakely@redhat.com>
14748 * doc/invoke.texi (-fpie, -fPIE): Fix grammar and clarify
14749 interaction with -pie.
14751 2018-04-30 David Malcolm <dmalcolm@redhat.com>
14753 * selftest.h: Fix alphabetization of per-source-file selftest
14756 2018-04-30 Jason Merrill <jason@redhat.com>
14758 PR c++/61982 - dead stores to destroyed objects.
14759 * gimplify.c (gimplify_modify_expr): Simplify complex lvalue on LHS
14762 2018-04-30 Jason Merrill <jason@redhat.com>
14764 * tree.c (build_clobber): New.
14765 * tree.h: Declare it.
14766 * gimplify.c (gimplify_bind_expr, gimplify_target_expr): Use it.
14768 2018-04-30 David Malcolm <dmalcolm@redhat.com>
14770 * diagnostic-show-locus.c (layout::layout): Update for
14771 location_get_source_line returning a char_span.
14772 (struct char_span): Move to input.h.
14773 (struct correction): Update for fields in char_span becoming
14775 (struct source_line): Update for location_get_source_line
14776 returning a char_span.
14777 (layout::print_line): Likewise.
14778 * edit-context.c (edited_file::print_content): Likewise.
14779 (edited_file::print_diff_hunk): Likewise.
14780 (edited_file::print_run_of_changed_lines): Likewise.
14781 (edited_file::get_num_lines): Likewise.
14782 (edited_line::edited_line): Likewise.
14783 * final.c (asm_show_source): Likewise.
14784 * input.c (location_get_source_line): Convert return type
14785 from const char * to char_span, losing the final "line_len"
14787 (dump_location_info): Update for the above.
14788 (get_substring_ranges_for_loc): Likewise. Use a char_span
14789 when handling the literal within the line.
14790 (test_reading_source_line): Update for location_get_source_line
14791 returning a char_span.
14792 * input.h (class char_span): Move here from
14793 diagnostic-show-locus.c, converting from a struct to a class.
14794 Make data members private.
14795 (char_span::operator bool): New.
14796 (char_span::length): New.
14797 (char_span::get_buffer): New.
14798 (char_span::operator[]): New.
14799 (char_span::subspan): Make const.
14800 (char_span::xstrdup): New.
14801 (location_get_source_line): Convert return type from const char *
14802 to char_span, losing the final "line_size" param.
14804 2018-04-30 Jan Hubicka <jh@suse.cz>
14806 * lto-wrapper.c (ltrans_priorities): New static var.
14807 (cmp_priority): New.
14808 (run_gcc): Read priorities and if doing parallel build order
14809 the Makefile by them.
14811 2018-04-30 David Malcolm <dmalcolm@redhat.com>
14813 * input.h (builtins_location_check): Convert to a STATIC_ASSERT.
14815 2018-04-30 Richard Biener <rguenther@suse.de>
14817 * tree-cfg.c (verify_address): Remove base argument, add
14818 flag whether to check TREE_ADDRESSABLE and do that.
14819 (verify_expr): Remove.
14820 (verify_types_in_gimple_reference): Add pieces from verify_expr.
14821 (verify_gimple_assign_single): Likewise.
14822 (verify_gimple_switch): Likewise.
14823 (verify_expr_location_1): Dereference tp once. Add (disabled)
14824 piece from verify_expr.
14825 (verify_gimple_in_cfg): Do not call verify_expr on all ops.
14827 2018-04-30 Claudiu Zissulescu <claziss@synopsys.com>
14829 * config/arc/linux.h (CLEAR_INSN_CACHE): Define.
14831 2018-04-30 Claudiu Zissulescu <claziss@synopsys.com>
14833 * config/arc/arc-protos.h (prepare_extend_operands): Remove.
14834 (small_data_pattern): Likewise.
14835 (arc_rewrite_small_data): Likewise.
14836 * config/arc/arc.c (LEGITIMATE_SMALL_DATA_OFFSET_P): Remove.
14837 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Likewise.
14838 (get_symbol_alignment): New function.
14839 (legitimate_small_data_address_p): Likewise.
14840 (legitimate_scaled_address): Update, call
14841 legitimate_small_data_address_p.
14842 (output_sdata): New static variable.
14843 (arc_print_operand): Update how we handle small data operands.
14844 (arc_print_operand_address): Likewise.
14845 (arc_legitimate_address_p): Update, use
14846 legitimate_small_data_address_p.
14847 (arc_rewrite_small_data_p): Remove.
14848 (arc_rewrite_small_data_1): Likewise.
14849 (arc_rewrite_small_data): Likewise.
14850 (small_data_pattern): Likewise.
14851 (compact_sda_memory_operand): Update to use
14852 legitimate_small_data_address_p and get_symbol_alignment.
14853 (prepare_move_operands): Don't rewite sdata pattern.
14854 (prepare_extend_operands): Remove.
14855 * config/arc/arc.md (zero_extendqihi2): Don't rewrite sdata
14857 (zero_extendqisi2): Likewise.
14858 (zero_extendhisi2): Likewise.
14859 (extendqihi2): Likewise.
14860 (extendqisi2): Likewise.
14861 (extendhisi2): Likewise.
14862 (addsi3): Likewise.
14863 (subsi3): Likewise.
14864 (andsi3): Likewise.
14865 * config/arc/constraints.md (Usd): Change it to memory constraint.
14867 2018-04-30 Claudiu Zissulescu <claziss@synopsys.com>
14869 * config/arc/arc.c (arc_split_move): Allow signed 6-bit constants
14870 as source of std instructions.
14871 * config/arc/arc.md (movsi_insn): Update pattern predicate to
14872 allow 6-bit constants as source for store instructions.
14873 (movdi_insn): Update instruction pattern to allow 6-bit constants
14874 as source for store instructions.
14876 2018-04-30 Jonathan Wakely <jwakely@redhat.com>
14878 * doc/invoke.texi (-fdebug-types-section): Fix grammar.
14880 2018-04-30 Nathan Sidwell <nathan@acm.org>
14881 Sandra Loosemore <sandra@codesourcery.com>
14883 * dumpfile.c (dump_open): Allow '-' for stdout.
14884 * doc/invoke.texi (Developer Options): Document dump filename
14885 determination early. Document stdin/stdout selection.
14887 2018-04-30 Andrew Sadek <andrew.sadek.se@gmail.com>
14889 Microblaze Target: PIC data text relative
14891 * config/microblaze/microblaze.opt: add new option -mpic-data-text-rel.
14892 * config/microblaze/microblaze-protos.h (microblaze_constant_address_p):
14894 * config/microblaze/microblaze.h (microblaze_constant_address_p):
14895 CONSTANT_ADDRESS_P definition to microblaze_constant_address_p.
14896 * config/microblaze/microblaze.c (TARGET_PIC_DATA_TEXT_REL):
14897 New addressing mode for data-text relative position indepenedent code.
14898 (microblaze_classify_unspec): add 'UNSPEC_TEXT' case ->
14899 'ADDRESS_SYMBOLIC_TXT_REL'.
14900 (microblaze_classify_address): Add handling for UNSPEC + CONST_INT.
14901 (microblaze_legitimate_pic_operand): Exclude function calls from
14902 pic operands in case of TARGET_PIC_DATA_TEXT_REL option.
14903 (microblaze_legitimize_address): Generate 'UNSPEC_TEXT' for all possible
14905 (microblaze_address_insns): Add 'ADDRESS_SYMBOLIC_TXT_REL' case.
14906 (print_operand): Add 'ADDRESS_SYMBOLIC_TXT_REL' case.
14907 (print_operand_address): Add 'ADDRESS_SYMBOLIC_TXT_REL' case + handling
14908 for 'address + offset'.
14909 (microblaze_expand_prologue): Add new function prologue call for
14911 (microblaze_asm_generate_pic_addr_dif_vec): Override new target hook
14912 'TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC' to disable address diff vector
14913 table in case of TARGET_PIC_DATA_TEXT_REL.
14914 (expand_pic_symbol_ref): Add handling for 'UNSPEC_TEXT'.
14915 * config/microblaze/microblaze.md (TARGET_PIC_DATA_TEXT_REL):
14916 Add new macros 'UNSPEC_TEXT',
14917 'UNSPEC_SET_TEXT' + add rule for setting r20 in function prologue
14918 + exclude function calls from 'UNSPEC_PLT' in case of data text
14920 * doc/tm.texi.in (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC): Add
14921 new target hook for generating address diff vector tables in case of
14923 * doc/tm.texi : Regenerate.
14924 * stmt.c (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC): Append new condition
14925 'targetm.asm_out.generate_pic_addr_diff_vec' to flag_pic in case
14926 of addr diff vector generation.
14927 * target.def (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC): Add
14928 target hook definition.
14929 * targhooks.h, targhooks.c (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC):
14930 Add default function for generate_pic_addr_diff_vec -> flag_pic.
14931 * doc/invoke.texi (Add new pic option): Add new microblaze pic
14932 option for data text relative.
14934 2018-04-30 Richard Biener <rguenther@suse.de>
14936 * tree-chrec.h (evolution_function_is_constant_p): Remove
14938 * tree-cfg.c (tree_node_can_be_shared): Re-order checks.
14940 2018-04-30 Richard Biener <rguenther@suse.de>
14943 * dwarf2out.c (gen_producer_string): Ignore -fchecking[=].
14945 2018-04-30 Richard Biener <rguenther@suse.de>
14947 PR tree-optimization/28364
14948 PR tree-optimization/85275
14949 * tree-ssa-loop-ch.c (ch_base::copy_headers): Stop after
14950 copying first exit test.
14952 2018-04-28 Mark Wielaard <mark@klomp.org>
14954 * dwarf2out.c (dwarf2out_finish): Add .debug_addr table header for
14955 dwarf_version >= 5.
14956 (dwarf_AT): Handle DW_AT_addr_base.
14957 (add_top_level_skeleton_die_attrs): Use dwarf_AT for DW_AT_addr_base.
14959 2018-04-28 Uros Bizjak <ubizjak@gmail.com>
14962 * config/i386/i386.md (*ashl<dwi>3_doubleword_mask): New pattern.
14963 (*ashl<dwi>3_doubleword_mask_1): Ditto.
14964 (*<shift_insn><dwi>3_doubleword_mask): Ditto.
14965 (*<shift_insn><dwi>3_doubleword_mask_1): Ditto.
14967 2018-04-28 Richard Biener <rguenther@suse.de>
14969 * tree-cfg.c (verify_gimple_phi): Take a gphi * argument.
14970 (verify_gimple_in_cfg): Rename visited_stmts to visited_throwing_stmts
14971 to reflect use. Only add interesting stmts.
14973 2018-04-27 Martin Jambor <mjambor@suse.cz>
14976 * ipa-cp.c (find_aggregate_values_for_callers_subset): Make sure
14977 the jump function allows for passing through aggregate values.
14979 2018-04-27 David Malcolm <dmalcolm@redhat.com>
14981 * input.h (in_system_header_at): Convert from macro to inline
14983 (from_macro_expansion_at): Likewise.
14984 (from_macro_definition_at): Likewise.
14986 2018-04-27 Jeff Law <law@redhat.com>
14988 * config.gcc: Mark tile* targets as deprecated/obsolete.
14990 2018-04-27 Richard Biener <rguenther@suse.de>
14992 * config/aarch64/aarch64.c: Simplify ap.__stack advance and
14995 2018-04-27 Richard Biener <rguenther@suse.de>
14997 * tree-cfg.c (verify_expr): Make dead code hit gcc_unreachable.
14999 2018-04-27 Uros Bizjak <ubizjak@gmail.com>
15001 * config/i386/i386.md (*movti_internal): Substitute Ye constraint
15002 with Yd constraint. Set "preferred_for_speed" attribute from
15003 TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC for alternatives
15004 with Yd constraint.
15005 (*movdi_internal): Ditto.
15006 (movti_interunit splitters): Remove
15007 TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC from insn condition.
15008 (movdi_interunit splitters): Ditto.
15009 * config/i386/constraints.md (Ye): Remove.
15010 (Yd): Do not depend on TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC.
15012 2018-04-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15015 * config/aarch64/constraints.md (Usg): Limit to 31.
15016 (Usj): Limit to 63.
15018 2018-04-27 Jakub Jelinek <jakub@redhat.com>
15020 PR tree-optimization/85529
15021 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Add FIRST_BB
15022 argument. Don't call get_nonzero_bits if opcode is ERROR_MARK_NODE,
15023 rhs2 def stmt's bb is dominated by first_bb and it isn't an obvious
15024 zero extension or masking of the MSB bit.
15025 (optimize_range_tests): Add FIRST_BB argument, pass it through
15026 to optimize_range_tests_var_bound.
15027 (maybe_optimize_range_tests, reassociate_bb): Adjust
15028 optimize_range_tests callers.
15030 2018-04-26 Richard Biener <rguenther@suse.de>
15031 Jakub Jelinek <jakub@redhat.com>
15033 * cgraph.h (symbol_table): Just declare debug method here.
15034 * symtab.c (symbol_table::debug): Define.
15036 2018-04-26 Eric Botcazou <ebotcazou@adacore.com>
15038 * loop-invariant.c (may_assign_reg_p): Return false for frame pointer.
15040 2018-04-26 Uros Bizjak <ubizjak@gmail.com>
15042 * config/i386/i386.md ("isa" attribute): Add x64_sse2.
15043 ("enabled" attribute): Handle x64_sse2 "isa" attribute.
15044 (*movdi_internal): Substitute Yi and Yj constraint with x
15045 and Ym and Yn constraint with y constraint. Update "isa"
15046 attribute and set "preferred_for_speed" attribute from
15047 TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC for updated alternatives.
15048 (*movsi_internal): Ditto.
15049 (*movdf_internal): Ditto.
15050 (*movsf_internal): Ditto.
15051 (*zero_extendsidi2): Ditto.
15052 * config/i386/sse.md (vec_set<mode>_0): Ditto.
15053 (sse2_loadld): Ditto.
15054 (*vec_extract<ssevecmodelower>_0): Ditto.
15055 (*vec_extractv4si_0_zext_sse4): Ditto.
15056 (vec_concatv2di): Ditto.
15057 (*vec_dup<mode>): Ditto.
15058 * config/i386/mmx.md (*mov<mode>_internal): Ditto.
15059 * config/i386/constraints.md (Yi): Remove.
15064 2018-04-26 Nathan Sidwell <nathan@acm.org>
15066 * dumpfile.c (dump_open): New.
15067 (dump_open_alternate_stream, dump_start, dump_begin): Call it.
15068 (dump_finish): Detect stdio/stderr by value not name.
15070 2018-04-26 Jonathan Wakely <jwakely@redhat.com>
15072 * doc/invoke.texi (-Wreturn-type): Document default status for C++.
15074 2018-04-26 Tom de Vries <tom@codesourcery.com>
15077 * config/nvptx/nvptx.c (verify_neutering_jumps)
15078 (verify_neutering_labels): New function
15079 (nvptx_single): Use verify_neutering_jumps and verify_neutering_labels.
15081 2018-04-26 Tom de Vries <tom@codesourcery.com>
15084 * config/nvptx/nvptx.c (needs_neutering_p): New function.
15085 (nvptx_single): Use needs_neutering_p to skip over insns that do not
15088 2018-04-26 Richard Biener <rguenther@suse.de>
15089 Tom de Vries <tom@codesourcery.com>
15092 * lto-streamer-out.c (output_function): Fixup loops if required to match
15093 discovery done in the reader.
15095 2018-04-26 Richard Biener <rguenther@suse.de>
15097 PR tree-optimization/85116
15098 * tree-ssa-loop-ch.c (do_while_loop_p): A do-while loop should
15099 have a loop exit from the single latch predecessor. Remove
15100 case of header with just condition.
15101 (ch_base::copy_headers): Exclude infinite loops from any
15103 (pass_ch::execute): Record exits.
15105 2018-04-26 Richard Biener <rguenther@suse.de>
15107 * tree-vect-data-refs.c (vect_get_data_access_cost): Get
15108 prologue cost vector and pass it to vect_get_load_cost.
15109 (vect_get_peeling_costs_all_drs): Likewise.
15110 (vect_peeling_hash_get_lowest_cost): Likewise.
15111 (vect_enhance_data_refs_alignment): Likewise.
15113 2018-04-26 Richard Biener <rguenther@suse.de>
15115 PR middle-end/85450
15116 * tree-cfg.c (verify_gimple_assign_unary): Restore proper
15117 checking of integer<->pointer conversions.
15118 * omp-expand.c (expand_omp_for_static_nochunk): Avoid
15119 sign-/zero-extending pointer types.
15120 (expand_omp_for_static_chunk): Likewise.
15122 2018-03-22 Hans-Peter Nilsson <hp@axis.com>
15123 Jean Lee <xiaoyur347@gmail.com>
15125 * config/mips/mips.c (mips_asan_shadow_offset): New function.
15126 (TARGET_ASAN_SHADOW_OFFSET): Define.
15127 * config/mips/mips.h (FRAME_GROWS_DOWNWARD): Augment to also be
15128 true for -fsanitize=address.
15130 2018-04-25 Mark Wielaard <mark@klomp.org>
15132 * dwarf2out.c (file_info_cmp): Sort longer dir prefixes before
15135 2018-04-25 Jakub Jelinek <jakub@redhat.com>
15137 * config/i386/i386.md (*x86_mov<mode>cc_0_m1): Use type "alu1" rather
15138 than "alu", remove explicit "memory" and "imm_disp" attributes.
15139 (*x86_mov<mode>cc_0_m1_se, *x86_mov<mode>cc_0_m1_neg): Likewise.
15141 PR middle-end/85414
15142 * simplify-rtx.c (simplify_unary_operation_1) <case SIGN_EXTEND,
15143 case ZERO_EXTEND>: Pass SUBREG_REG (op) rather than op to
15144 gen_lowpart_no_emit.
15146 2018-04-25 Sebastian Peryt <sebastian.peryt@intel.com>
15149 * config/i386/i386.c (ix86_expand_builtin): Change memory
15150 operand to XI, extend p0 to Pmode.
15151 * config/i386/i386.md: Change unspec volatile and operand
15152 1 mode to XI, change operand 0 mode to P.
15154 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
15156 * config/nds32/nds32-predicates.c (nds32_can_use_bclr_p): Mask with
15157 GET_MODE_MASK before any checking.
15158 (nds32_can_use_bset_p): Likewise.
15159 (nds32_can_use_btgl_p): Likewise.
15161 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
15163 * config/nds32/nds32-doubleword.md: New define_split pattern for
15164 illegal register number.
15166 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
15168 * config/nds32/nds32.c (nds32_print_operand): Set op_value ealier.
15170 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
15172 * config/nds32/nds32.h (ASM_APP_ON): Add missing newline character.
15174 2018-04-25 Richard Biener <rguenther@suse.de>
15176 * lto-streamer.h (LTO_major_version): Bump to 8.
15178 2018-04-25 Jakub Jelinek <jakub@redhat.com>
15180 * BASE-VER: Set to 9.0.0.
15182 2018-04-24 Segher Boessenkool <segher@kernel.crashing.org>
15184 * config/rs6000/rs6000.c (init_float128_ieee): Fix spelling mistakes
15185 in __abskf2 and __powikf2.
15187 2018-04-24 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15190 * config/aarch64/constraints.md (Usg, Usj): New constraints.
15191 * config/aarch64/iterators.md (cmode_simd): New mode attribute.
15192 * config/aarch64/aarch64.md (*aarch64_ashr_sisd_or_int_<mode>3):
15193 Use the above on operand 2. Reindent.
15194 (*aarch64_lshr_sisd_or_int_<mode>3): Likewise.
15196 2018-04-24 H.J. Lu <hongjiu.lu@intel.com>
15199 * common/config/i386/i386-common.c (ix86_handle_option): Don't
15201 * config/i386/i386.opt (mcet): Removed.
15202 * doc/install.texi: Remove -mcet documentation.
15203 * doc/invoke.texi: Likewise.
15205 2018-04-24 H.J. Lu <hongjiu.lu@intel.com>
15208 * doc/install.texi: Remove -mcet from bootstrap-cet.
15210 2018-04-24 Jakub Jelinek <jakub@redhat.com>
15213 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Don't define
15214 __builtin_ia32_readeflags_u32 and __builtin_ia32_writeeflags_u32
15218 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Punt if
15219 const_vector is not CONST_VECTOR or SYMBOL_REF for a constant pool
15220 containing a CONST_VECTOR.
15222 2018-04-24 Cesar Philippidis <cesar@codesourcery.com>
15224 * doc/install.texi: Update newlib dependency for nvptx.
15226 2018-04-24 Jakub Jelinek <jakub@redhat.com>
15229 * config/i386/i386.c (ix86_expand_vector_init_one_var): Use UINTVAL
15230 instead of INTVAL when shifting x left.
15232 2018-04-24 Andreas Krebbel <krebbel@linux.ibm.com>
15234 PR tree-optimization/85478
15235 * tree-vect-loop.c (vect_analyze_loop_2): Do not call
15236 vect_grouped_store_supported for single element vectors.
15238 2018-04-24 Richard Biener <rguenther@suse.de>
15241 * config/i386/i386.c (ix86_add_stmt_cost): Restrict strided
15242 load cost increase to the case of non-constant step.
15244 2018-04-24 Jakub Jelinek <jakub@redhat.com>
15247 * reg-stack.c (move_for_stack_reg): Don't fail assertion about dead
15248 destination if any_malformed_asm.
15250 2018-04-23 Eric Botcazou <ebotcazou@adacore.com>
15252 PR middle-end/85496
15253 * expr.c (store_field): In the bitfield case, if the value comes from
15254 a function call and is returned in registers by means of a PARALLEL,
15255 do not change the mode of the temporary unless BLKmode and VOIDmode.
15257 2018-04-23 Andrey Belevantsev <abel@ispras.ru>
15259 PR rtl-optimization/85423
15260 * sel-sched-ir.c (has_dependence_note_mem_dep): Only discard
15261 dependencies to debug insns when the previous insn is non-debug.
15263 2018-04-23 Claudiu Zissulescu <claziss@synopsys.com>
15265 * config/arc/arc.md ("vunspec"): Delete it, unify all the unspec
15266 enums into a single definition.
15267 (fls): Fix predicates and printing.
15270 2018-04-23 Claudiu Zissulescu <claziss@synopsys.com>
15272 * config/arc/arc-protos.h (check_if_valid_sleep_operand): Remove.
15273 * config/arc/arc.c (arc_expand_builtin): Sleep accepts registers
15274 and short u6 immediate.
15275 (check_if_valid_sleep_operand): Remove.
15276 * config/arc/arc.md (Sleep): Accepts registers and u6 immediates.
15278 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
15280 * config/nds32/nds32.c (nds32_compute_stack_frame): Consider
15281 flag_always_save_lp condition.
15282 * config/nds32/nds32.opt (malways-save-lp): New option.
15284 2018-04-22 Shiva Chen <shiva0217@gmail.com>
15286 * config/nds32/nds32-protos.h (nds32_use_load_post_increment): Declare.
15287 * config/nds32/nds32.c (nds32_use_load_post_increment): New.
15288 * config/nds32/nds32.h
15289 (USE_LOAD_POST_INCREMENT, USE_LOAD_POST_DECREMENT): Define.
15290 (USE_STORE_POST_INCREMENT, USE_STORE_POST_DECREMENT): Define.
15292 2018-04-22 Shiva Chen <shiva0217@gmail.com>
15294 * config/nds32/nds32-protos.h (nds32_ls_333_p): Remove.
15295 * config/nds32/nds32.c (nds32_ls_333_p): Remove.
15297 2018-04-22 Shiva Chen <shiva0217@gmail.com>
15298 Chung-Ju Wu <jasonwucj@gmail.com>
15300 * config/nds32/nds32-protos.h (nds32_case_vector_shorten_mode):
15302 * config/nds32/nds32.c (nds32_case_vector_shorten_mode): New function.
15303 * config/nds32/nds32.h (CASE_VECTOR_SHORTEN_MODE): Modify.
15305 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
15307 * config/nds32/nds32.c (nds32_compute_stack_frame): Fix wrong value.
15309 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
15311 * config/nds32/nds32-protos.h (nds32_data_alignment,
15312 nds32_local_alignment): Declare.
15313 * config/nds32/nds32.c (nds32_data_alignment, nds32_constant_alignment,
15314 nds32_local_alignment): New functions.
15315 (TARGET_CONSTANT_ALIGNMENT): Define.
15316 * config/nds32/nds32.h (DATA_ALIGNMENT, LOCAL_ALIGNMENT): Define.
15318 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
15320 * config/nds32/nds32.c
15321 (TARGET_HARD_REGNO_MODE_OK): Move to the bottom of file.
15322 (TARGET_MODES_TIEABLE_P): Likewise.
15324 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
15326 * config/nds32/nds32.c (nds32_asm_file_start): Display optimization
15327 level Ofast and Og.
15329 2018-04-22 Monk Chiang <sh.chiang04@gmail.com>
15330 Chung-Ju Wu <jasonwucj@gmail.com>
15332 * config/nds32/constants.md (unspec_volatile_element): Add enum values
15333 for unaligned access.
15334 * config/nds32/nds32-intrinsic.c: Implementation of expanding
15336 * config/nds32/nds32-intrinsic.md: Likewise.
15337 * config/nds32/nds32_intrinsic.h: Likewise.
15338 * config/nds32/nds32.h (nds32_builtins): Likewise.
15339 * config/nds32/nds32.opt (munaligned-access): New option.
15340 * config/nds32/nds32.c (nds32_asm_file_start): Display
15341 flag_unaligned_access status.
15343 2018-04-20 Kito Cheng <kito.cheng@gmail.com>
15345 * config/riscv/elf.h (LINK_SPEC): Pass --no-relax if
15346 -mno-relax is present.
15347 * config/riscv/linux.h (LINK_SPEC): Ditto.
15349 2018-04-20 Martin Sebor <msebor@redhat.com>
15352 * gimple-fold.c (gimple_fold_builtin_strcpy): Suppress -Wrestrict
15354 (gimple_fold_builtin_stxcpy_chk): Same.
15355 * gimple-ssa-warn-restrict.c (check_bounds_or_overlap): Same.
15357 2018-04-20 Michael Meissner <meissner@linux.ibm.com>
15360 * config/rs6000/rs6000.c (init_float128_ieee): Add support to call
15361 __powikf2 when long double is IEEE 128-bit.
15363 2018-04-20 Kito Cheng <kito.cheng@gmail.com>
15365 * config/riscv/riscv.c (riscv_first_stack_step): Round up min
15366 step to make sure stack always aligned.
15368 2018-04-20 Carl Love <cel@us.ibm.com>
15371 * config/rs6000/rs6000-c.c (rs6000_gimple_fold_builtin): Add
15372 size check for arg0.
15374 2018-04-20 Nathan Sidwell <nathan@codesourcery.com>
15375 Tom de Vries <tom@codesourcery.com>
15378 * config/nvptx/nvptx.c (nvptx_emit_forking, nvptx_emit_joining):
15379 Emit insns for calls too.
15380 (nvptx_find_par): Always look for worker-level predecessor insn.
15381 (nvptx_propagate): Add is_call parm, return bool. Copy frame for
15383 (nvptx_vpropagate, nvptx_wpropagate): Adjust.
15384 (nvptx_process_pars): Propagate frames for calls.
15386 2018-04-20 H.J. Lu <hongjiu.lu@intel.com>
15389 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET):
15391 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
15392 (ix86_handle_option): Don't handle OPT_mibt.
15393 * config/i386/cet.h: Check __CET__ instead of __IBT__ and
15395 * config/i386/driver-i386.c (host_detect_local_cpu): Remove
15397 * config/i386/i386-c.c (ix86_target_macros_internal): Don't
15398 check OPTION_MASK_ISA_IBT nor flag_cf_protection.
15399 (ix86_target_macros): Define __CET__ with flag_cf_protection
15400 for -fcf-protection.
15401 * config/i386/i386.c (isa2_opts): Remove -mibt.
15402 * config/i386/i386.h (TARGET_IBT): Removed.
15403 (TARGET_IBT_P): Likewise.
15404 (ix86_valid_target_attribute_inner_p): Don't check OPT_mibt.
15405 * config/i386/i386.md (nop_endbr): Don't check TARGET_IBT.
15406 * config/i386/i386.opt (mcet): Update help message.
15407 (mshstk): Likewise.
15409 * doc/invoke.texi: Remove -mibt. Document __CET__. Document
15410 -mcet as an alias for -mshstk.
15412 2018-04-20 Richard Biener <rguenther@suse.de>
15414 PR middle-end/85475
15415 * match.pd ((X * CST) * Y -> (X * Y) * CST): Avoid exponential
15416 complexity by forcing a single use of the multiply operand.
15418 2018-04-20 Martin Jambor <mjambor@suse.cz>
15421 * ipa-cp.c (cgraph_edge_brings_value_p): Move check for self-feeding
15422 recursion dependency to only apply to non-clones.
15424 2018-04-20 Martin Jambor <mjambor@suse.cz>
15427 * ipa-cp.c (create_specialized_node): Check that clones of
15428 self-recursive edges exist during IPA-CP.
15430 2018-04-19 Toon Moene <toon@moene.org>
15432 * doc/invoke.texi: Add -floop-unroll-and-jam to options enabled
15435 2018-04-19 Jakub Jelinek <jakub@redhat.com>
15437 PR tree-optimization/85467
15438 * fold-const.c (fold_ternary_loc) <case BIT_FIELD_REF>: Use
15439 VECTOR_TYPE_P macro. If type is vector type, VIEW_CONVERT_EXPR the
15440 VECTOR_CST element to type.
15442 2018-04-19 H.J. Lu <hongjiu.lu@intel.com>
15445 * config/i386/i386.h (STACK_SAVEAREA_MODE): New.
15446 * config/i386/i386.md (builtin_setjmp_setup): Removed.
15447 (builtin_longjmp): Likewise.
15448 (save_stack_nonlocal): New pattern.
15449 (restore_stack_nonlocal): Likewise.
15451 2018-04-19 H.J. Lu <hongjiu.lu@intel.com>
15454 * config/i386/cet.c (file_end_indicate_exec_stack_and_cet):
15455 Replace ASM_OUTPUT_LABEL with fprintf.
15457 2018-04-19 H.J. Lu <hongjiu.lu@intel.com>
15460 * config/i386/cet.c (file_end_indicate_exec_stack_and_cet):
15461 Check flag_cf_protection instead of TARGET_IBT and TARGET_SHSTK.
15462 * config/i386/i386-c.c (ix86_target_macros_internal): Also
15463 define __IBT__ and __SHSTK__ for -fcf-protection.
15464 * config/i386/i386.c (pass_insert_endbranch::gate): Don't check
15466 (ix86_trampoline_init): Likewise.
15467 (x86_output_mi_thunk): Likewise.
15468 (ix86_notrack_prefixed_insn_p): Likewise.
15469 (ix86_option_override_internal): Don't disallow -fcf-protection.
15470 * config/i386/i386.md (rdssp<mode>): Also enable for
15472 (incssp<mode>): Likewise.
15473 (nop_endbr): Likewise.
15474 * config/i386/i386.opt (mcet): Change help message to built-in
15477 (mshstk): Likewise.
15478 * doc/invoke.texi: Remove -mcet, -mibt and -mshstk condition
15479 on -fcf-protection. Change -mcet, -mibt and -mshstk to only
15480 enable CET built-in functions.
15482 2018-04-19 Sebastian Peryt <sebastian.peryt@intel.com>
15484 * common/config/i386/i386-common.c
15485 (OPTION_MASK_ISA_MOVDIRI_SET, OPTION_MASK_ISA_MOVDIR64B_SET,
15486 OPTION_MASK_ISA_MOVDIRI_UNSET,
15487 OPTION_MASK_ISA_MOVDIR64B_UNSET): New defines.
15488 (ix86_handle_option): Handle -mmovdiri and -mmovdir64b.
15489 * config.gcc (movdirintrin.h): New header.
15490 * config/i386/cpuid.h (bit_MOVDIRI,
15491 bit_MOVDIR64B): New bits.
15492 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mmovdiri
15494 * config/i386/i386-builtin-types.def ((VOID, PUNSIGNED, UNSIGNED),
15495 (VOID, PVOID, PCVOID)): New function types.
15496 * config/i386/i386-builtin.def (__builtin_ia32_directstoreu_u32,
15497 __builtin_ia32_directstoreu_u64,
15498 __builtin_ia32_movdir64b): New builtins.
15499 * config/i386/i386-c.c (__MOVDIRI__, __MOVDIR64B__): New.
15500 * config/i386/i386.c (ix86_target_string): Added -mmovdir64b
15502 (ix86_valid_target_attribute_inner_p): Ditto.
15503 (ix86_expand_special_args_builtin): Added VOID_FTYPE_PUNSIGNED_UNSIGNED
15504 and VOID_FTYPE_PUNSIGNED_UNSIGNED.
15505 (ix86_expand_builtin): Expand IX86_BUILTIN_MOVDIR64B.
15506 * config/i386/i386.h (TARGET_MOVDIRI, TARGET_MOVDIRI_P,
15507 TARGET_MOVDIR64B, TARGET_MOVDIR64B_P): New.
15508 * config/i386/i386.md (UNSPECV_MOVDIRI, UNSPECV_MOVDIR64B): New.
15509 (movdiri<mode>, movdir64b_<mode>): New.
15510 * config/i386/i386.opt: Add -mmovdiri and -mmovdir64b.
15511 * config/i386/immintrin.h: Include movdirintrin.h.
15512 * config/i386/movdirintrin.h: New file.
15513 * doc/invoke.texi: Added -mmovdiri and -mmovdir64b.
15515 2018-04-19 Richard Biener <rguenther@suse.de>
15517 PR middle-end/85455
15518 * cfg.c (clear_bb_flags): When loop state says we have
15519 marked irreducible regions also preserve BB_IRREDUCIBLE_LOOP.
15521 2018-04-19 Richard Biener <rguenther@suse.de>
15523 PR tree-optimization/84737
15524 * tree-vect-data-refs.c (vect_copy_ref_info): New function
15525 copying restrict info.
15526 (vect_setup_realignment): Use it.
15527 * tree-vectorizer.h (vect_copy_ref_info): Declare.
15528 * tree-vect-stmts.c (vectorizable_store): Copy ref info from
15529 the first DR to all generated stores.
15530 (vectorizable_load): Likewise for loads.
15532 2018-04-19 Jakub Jelinek <jakub@redhat.com>
15534 PR tree-optimization/85446
15535 * match.pd ((intptr_t) x eq/ne CST to x eq/ne (typeof x) cst): Require
15536 the integral and pointer types to have the same precision.
15538 * doc/install.texi: Document --disable-cet being the default and
15541 2018-04-18 Martin Liska <mliska@suse.cz>
15543 * ipa-devirt.c (odr_subtypes_equivalent_p): Fix GNU coding
15546 2018-04-18 Martin Liska <mliska@suse.cz>
15549 2018-03-02 Eric Botcazou <ebotcazou@adacore.com>
15552 * ipa-devirt.c (odr_subtypes_equivalent_p): Get the ODR type of both
15553 arguments if they are comparable.
15555 2018-04-18 Martin Liska <mliska@suse.cz>
15558 2018-03-13 Eric Botcazou <ebotcazou@adacore.com>
15561 * ipa-devirt.c (odr_subtypes_equivalent_p): Do not get the ODR type of
15564 2018-04-18 H.J. Lu <hongjiu.lu@intel.com>
15567 * config/i386/i386.c (ix86_expand_split_stack_prologue): Insert
15568 ENDBR after calling __morestack.
15570 2018-04-18 David Malcolm <dmalcolm@redhat.com>
15573 * configure.ac (gcc-driver-name.h): Honor --with-gcc-major-version
15574 by using gcc_base_ver to generate a gcc_driver_version, and use
15575 it when generating GCC_DRIVER_NAME.
15576 * configure: Regenerate.
15578 2018-04-18 Jakub Jelinek <jakub@redhat.com>
15581 * config.gcc: Obsolete powerpc*-*-*spe*.
15583 2018-04-17 Jakub Jelinek <jakub@redhat.com>
15586 * dbxout.c (dbxout_int): Perform negation in unsigned int type.
15587 (stabstr_D): Change type of unum from unsigned int to
15588 unsigned HOST_WIDE_INT. Perform negation in unsigned HOST_WIDE_INT
15591 2018-04-17 Jim Wilson <jimw@sifive.com>
15594 * config/riscv/riscv.c (riscv_compute_frame_info): Add calls to
15595 RISCV_STACK_ALIGN when using outgoing_args_size and pretend_args_size.
15596 Set arg_pointer_offset after using pretend_args_size.
15598 2018-04-17 Jakub Jelinek <jakub@redhat.com>
15600 PR rtl-optimization/85431
15601 * dse.c (record_store): Ignore zero width stores.
15604 * asan.c (handle_builtin_stack_restore): Adjust comment. Emit
15605 __asan_allocas_unpoison call and last_alloca_addr = new_sp before
15606 __builtin_stack_restore rather than after it.
15607 * builtins.c (expand_asan_emit_allocas_unpoison): Pass
15608 arg1 + (virtual_dynamic_stack_rtx - stack_pointer_rtx) as second
15609 argument instead of virtual_dynamic_stack_rtx.
15611 2018-04-17 Kelvin Nilsen <kelvin@gcc.gnu.org>
15613 * config/rs6000/rs6000-protos.h (rs6000_builtin_is_supported_p):
15615 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
15616 Add note to error message to explain internal mapping of overloaded
15617 built-in function name to non-overloaded built-in function name.
15618 * config/rs6000/rs6000.c (rs6000_builtin_is_supported_p): New
15621 2018-04-17 Michael Meissner <meissner@linux.vnet.ibm.com>
15624 * config/rs6000/rs6000.md (pack<mode>): Do not try handle a pack
15625 where the inputs overlap with the output.
15627 2018-04-17 Jakub Jelinek <jakub@redhat.com>
15629 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): Add
15630 (=v, v) alternative and explicit "memory" attribute.
15631 (vec_extract_lo_<mode><mask_name>): Likewise. Also add
15632 "type", "prefix", "prefix_extra", "length_immediate" and "mode"
15634 (vec_extract_lo_<mode><mask_name>): Add (=v, v) alternative and use
15635 "sselog1" type instead of "sselog".
15636 (vec_extract_hi_<mode><mask_name>): Use "sselog1" type instead of
15637 "sselog". Remove explicit "memory" attribute.
15638 (vec_extract_lo_v32hi): Add (=v, v) alternative and explicit "memory",
15639 "type", "prefix", "prefix_extra", "length_immediate" and "mode"
15641 (vec_extract_hi_v32hi): Merge all alternatives into one, use
15642 "sselog1" type instead of "sselog". Remove explicit "memory"
15644 (vec_extract_hi_v16hi): Merge each pair of alternatives into one,
15645 use "sselog1" type instead of "sselog". Remove explicit "memory"
15647 (vec_extract_lo_v64qi): Add (=v, v) alternative and explicit "memory",
15648 "type", "prefix", "prefix_extra", "length_immediate" and "mode"
15650 (vec_extract_hi_v64qi): Merge all alternatives into one, use
15651 "sselog1" type instead of "sselog". Remove explicit "memory"
15653 (vec_extract_hi_v32qi): Merge each pair of alternatives into one,
15654 use "sselog1" type instead of "sselog". Remove explicit "memory"
15658 * config/i386/i386.md (*ashlqi3_1_slp): Use alu1 type instead of alu.
15660 PR middle-end/85414
15661 * rtlhooks.c (gen_lowpart_if_possible): Don't call gen_lowpart_SUBREG
15664 2018-04-17 Martin Jambor <mjambor@suse.cz>
15667 * ipa-cp.c (create_specialized_node): Call
15668 expand_all_artificial_thunks if necessary.
15670 2018-04-17 Martin Liska <mliska@suse.cz>
15673 * ipa-devirt.c (odr_types_equivalent_p): Remove trailing
15674 in message, remote space in between '_G' and '('.
15676 2018-04-17 Jakub Jelinek <jakub@redhat.com>
15679 * config/i386/sse.md (reduces<mode><mask_scalar_name>,
15680 avx512f_vmcmp<mode>3<round_saeonly_name>,
15681 avx512f_vmcmp<mode>3_mask<round_saeonly_name>,
15682 avx512f_sgetexp<mode><mask_scalar_name><round_saeonly_scalar_name>,
15683 avx512f_rndscale<mode><round_saeonly_name>,
15684 avx512dq_ranges<mode><mask_scalar_name><round_saeonly_scalar_name>,
15685 avx512f_vgetmant<mode><mask_scalar_name><round_saeonly_scalar_name>):
15686 Use %<iptr>2 instead of %2 for -masm=intel.
15687 (avx512f_vcvtss2usi<round_name>, avx512f_vcvtss2usiq<round_name>,
15688 avx512f_vcvttss2usi<round_saeonly_name>,
15689 avx512f_vcvttss2usiq<round_saeonly_name>): Use %k1 instead of %1 for
15691 (avx512f_vcvtsd2usi<round_name>, avx512f_vcvtsd2usiq<round_name>,
15692 avx512f_vcvttsd2usi<round_saeonly_name>,
15693 avx512f_vcvttsd2usiq<round_saeonly_name>, ufloatv2siv2df2<mask_name>):
15694 Use %q1 instead of %1 for -masm=intel.
15695 (avx512f_sfixupimm<mode><sd_maskz_name><round_saeonly_name>,
15696 avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Use %<iptr>3 instead
15697 of %3 for -masm=intel.
15698 (sse2_shufpd_v2df_mask): Fix a typo, change %{6%} to %{%6%} for
15700 (*avx512vl_<code>v2div2qi2_store): Use %w0 instead of %0 for
15702 (*avx512vl_<code><mode>v4qi2_store): Use %k0 instead of %0 for
15704 (avx512vl_<code><mode>v4qi2_mask_store): Use a single pattern with
15705 %k0 and %1 for -masm=intel rather than two patterns, one with %0 and
15707 (*avx512vl_<code><mode>v8qi2_store): Use %q0 instead of %0 for
15709 (avx512vl_<code><mode>v8qi2_mask_store): Use a single pattern with
15710 %q0 and %1 for -masm=intel rather than two patterns, one with %0 and
15711 %g1 and one with %0 and %1.
15712 (avx512er_vmrcp28<mode><round_saeonly_name>,
15713 avx512er_vmrsqrt28<mode><round_saeonly_name>): Use %<iptr>1 instead of
15714 %1 for -masm=intel.
15715 (avx5124fmaddps_4fmaddps_mask, avx5124fmaddps_4fmaddss_mask,
15716 avx5124fmaddps_4fnmaddps_mask, avx5124fmaddps_4fnmaddss_mask,
15717 avx5124vnniw_vp4dpwssd_mask, avx5124vnniw_vp4dpwssds_mask): Swap order
15718 of %0 and %{%4%} for -masm=intel.
15719 (avx5124fmaddps_4fmaddps_maskz, avx5124fmaddps_4fmaddss_maskz,
15720 avx5124fmaddps_4fnmaddps_maskz, avx5124fmaddps_4fnmaddss_maskz,
15721 avx5124vnniw_vp4dpwssd_maskz, avx5124vnniw_vp4dpwssds_maskz): Swap
15722 order of %0 and %{%5%}%{z%} for -masm=intel.
15724 2018-04-17 Jan Hubicka <jh@suse.cz>
15727 * ipa-devirt.c (odr_types_equivalent_p): Handle bit fields.
15729 2018-04-17 Martin Liska <mliska@suse.cz>
15732 * multiple_target.c (create_dispatcher_calls): Set apostrophes
15733 for target_clone error message. Make default implementation
15734 clone to be a local declaration.
15735 (separate_attrs): Add new argument and check for an empty
15737 (expand_target_clones): Handle it.
15738 (ipa_target_clone): Make redirection just for target_clones
15741 2018-04-16 Cesar Philippidis <cesar@codesourcery.com>
15742 Tom de Vries <tom@codesourcery.com>
15744 PR middle-end/84955
15745 * omp-expand.c (expand_oacc_for): Add dummy false branch for
15746 tiled basic blocks without omp continue statements.
15748 2018-04-16 Aaron Sawdey <acsawdey@linux.ibm.com>
15751 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Mark
15752 vec_extract expression as having side effects to make sure it gets
15755 2018-04-16 H.J. Lu <hongjiu.lu@intel.com>
15758 * config/i386/i386.c (get_builtin_code_for_version): Check
15761 2018-04-16 Olga Makhotina <olga.makhotina@intel.com>
15764 * config.gcc: Support "skylake".
15765 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
15767 * config/i386/i386.c (m_SKYLAKE): Define.
15768 (processor_target_table): Add "skylake".
15769 (ix86_option_override_internal): Add "skylake".
15770 (get_builtin_code_for_version): Handle PROCESSOR_SKYLAKE,
15771 PROCESSOR_CANNONLAKE.
15772 (get_builtin_code_for_version): Fix priority for
15773 PROCESSOR_ICELAKE_CLIENT, PROCESSOR_ICELAKE_SERVER,
15774 PROCESSOR_SKYLAKE-AVX512.
15775 * config/i386/i386.h (processor_costs): Define TARGET_SKYLAKE.
15776 (processor_type): Add PROCESSOR_SKYLAKE.
15778 2018-04-16 Paolo Carlini <paolo.carlini@oracle.com>
15779 Jason Merrill <jason@redhat.com>
15782 * convert.c (convert_to_integer_1): Use direct recursion for
15783 enumeral types and types with a precision less than the number
15784 of bits in their mode.
15786 2018-04-16 Julia Koval <julia.koval@intel.com>
15789 * config/i386/x86-tune.def (X86_TUNE_SSE_UNALIGNED_LOAD_OPTIMAL,
15790 X86_TUNE_SSE_UNALIGNED_STORE_OPTIMAL): Add m_SKYLAKE_AVX512
15792 2018-04-14 Segher Boessenkool <segher@kernel.crashing.org>
15795 * config/rs6000/rs6000.opt (mdirect-move): Make deprecated.
15796 * doc/invoke.texi (RS/6000 and PowerPC Options): Remove -mdirect-move
15797 and -mno-direct-move.
15799 2018-04-13 Paul A. Clarke <pc@us.ibm.com>
15802 * config/rs6000/emmintrin.h (_mm_slli_epi{16,32,64}):
15803 Ensure that vec_splat_s32 is only called with 0 <= shift < 16.
15804 Ensure negative shifts result in {0}.
15806 2018-04-13 Vladimir Makarov <vmakarov@redhat.com>
15808 PR rtl-optimization/79916
15809 * config/rs6000/rs6000.c (rs6000_emit_move): Use assigned hard
15810 regs (if any) to define how to gnerate SD moves when LRA is in
15813 2018-04-13 Jakub Jelinek <jakub@redhat.com>
15815 PR rtl-optimization/85393
15816 * except.h (expand_dw2_landing_pad_for_region): Remove declaration.
15817 * except.c (expand_dw2_landing_pad_for_region): Make static.
15818 * bb-reorder.c (fix_up_crossing_landing_pad): In new_bb emit just
15819 a label and unconditional jump to old_bb, rather than
15820 expand_dw2_landing_pad_for_region insn(s) and jump to single_succ
15823 PR rtl-optimization/85376
15824 * simplify-rtx.c (simplify_const_unary_operation): For CLZ and CTZ and
15825 zero op0, if C?Z_DEFINED_VALUE_AT_ZERO is false, return NULL_RTX
15826 instead of a specific value.
15828 2018-04-13 Jan Hubicka <hubicka@ucw.cz>
15829 Bin Cheng <bin.cheng@arm.com>
15831 PR tree-optimization/82965
15832 PR tree-optimization/83991
15833 * cfgloopanal.c (expected_loop_iterations_unbounded): Add
15834 by_profile_only parameter.
15835 * cfgloopmanip.c (scale_loop_profile): Further scale loop's profile
15836 information if the loop was predicted to iterate too many times.
15837 * cfgloop.h (expected_loop_iterations_unbounded): Update prototype
15839 2018-04-13 Jan Hubicka <hubicka@ucw.cz>
15842 * config/i386/i386.c (ix86_can_inline_p): Allow safe transitions for
15845 2018-04-13 Martin Liska <mliska@suse.cz>
15846 Jakub Jelinek <jakub@redhat.com>
15848 PR middle-end/81657
15849 * expr.h (enum block_op_methods): Add BLOCK_OP_NO_LIBCALL_RET.
15850 * expr.c (emit_block_move_hints): Handle BLOCK_OP_NO_LIBCALL_RET.
15851 * builtins.c (expand_builtin_memory_copy_args): Use
15852 BLOCK_OP_NO_LIBCALL_RET method for mempcpy with non-ignored target,
15853 handle dest_addr == pc_rtx.
15855 2018-04-12 Segher Boessenkool <segher@kernel.crashing.org>
15858 * config/rs6000/rs6000.md (fix_trunc<mode>si2): Use legacy code if
15859 asked to not generate direct moves.
15860 (fix_trunc<mode>si2_stfiwx): Similar.
15861 (fix_trunc<mode>si2_internal): Similar.
15863 2018-04-12 Jakub Jelinek <jakub@redhat.com>
15866 * var-tracking.c (add_stores): Handle STRICT_LOW_PART SET_DEST.
15867 * cselib.c (cselib_record_sets): For STRICT_LOW_PART dest,
15868 lookup if dest in some wider mode is known to be const0_rtx and
15869 if so, record permanent equivalence for it to be ZERO_EXTEND of
15870 the narrower mode destination.
15872 2018-04-12 Cesar Philippidis <cesar@codesourcery.com>
15874 * lto-streamer-out.c (output_function): Revert 259346.
15875 * omp-expand.c (expand_oacc_for): Likewise.
15877 2018-04-12 Alexander Monakov <amonakov@ispras.ru>
15879 PR rtl-optimization/85354
15880 * sel-sched-ir.c (sel_init_pipelining): Move cfg_cleanup call...
15881 * sel-sched.c (sel_global_init): ... here.
15883 2018-04-12 Eric Botcazou <ebotcazou@adacore.com>
15886 * lto-wrapper.c (debug_objcopy): Open the files in binary mode.
15887 * dwarf2out.c (dwarf2out_early_finish): Do not generate assembly in LTO
15888 mode for PE-COFF targets.
15889 * config/i386/i386-protos.h (i386_pe_asm_lto_start): Declare.
15890 (i386_pe_asm_lto_end): Likewise.
15891 * config/i386/cygming.h (TARGET_ASM_LTO_START): Define.
15892 (TARGET_ASM_LTO_END): Likewise.
15893 * config/i386/winnt.c (saved_debug_info_level): New static variable.
15894 (i386_pe_asm_lto_start): New function.
15895 (i386_pe_asm_lto_end): Likewise.
15897 2018-04-12 Cesar Philippidis <cesar@codesourcery.com>
15898 Richard Biener <rguenther@suse.de>
15900 PR middle-end/84955
15901 * lto-streamer-out.c (output_function): Fix CFG loop state before
15903 * omp-expand.c (expand_oacc_for): Handle calls to internal
15904 functions like regular functions.
15906 2018-04-12 Richard Biener <rguenther@suse.de>
15909 * dwarf2out.c (init_sections_and_labels): Use debug_line_section[_label]
15910 for the early LTO debug to properly generate references to it
15911 during DIE emission. Do not re-use that for the skeleton for
15913 (dwarf2out_early_finish): Likewise.
15915 2018-04-12 Jakub Jelinek <jakub@redhat.com>
15918 * config/i386/sse.md
15919 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name> split,
15920 <mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name> split,
15921 vec_extract_lo_<mode><mask_name> split, vec_extract_lo_v32hi,
15922 vec_extract_lo_v64qi): For non-AVX512VL if input is xmm16+ reg
15923 and output is a reg, avoid creating invalid lowpart subreg, but
15924 instead split into a 512-bit move. Don't split if not AVX512VL,
15925 input is xmm16+ reg and output is a mem.
15926 (vec_extract_lo_<mode><mask_name>, vec_extract_lo_v32hi,
15927 vec_extract_lo_v64qi): Don't require split if not AVX512VL, input is
15928 xmm16+ reg and output is a mem.
15930 2018-04-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15932 * config/s390/s390.c (s390_output_indirect_thunk_function): Check
15933 also for flag_dwarf2_cfi_asm.
15935 2018-04-12 Jakub Jelinek <jakub@redhat.com>
15937 PR rtl-optimization/85342
15938 * regcprop.c (copyprop_hardreg_forward_1): Remove replaced array, use
15939 a bool scalar var inside of the loop instead. Don't try to update
15940 recog_data.operand after failed apply_change_group.
15942 2018-04-12 Tom de Vries <tom@codesourcery.com>
15945 * config/nvptx/nvptx.c (flexible_array_member_type_p): New function.
15946 (nvptx_assemble_decl_begin): Add undefined param. Declare undefined
15947 array with flexible array member as array without given dimension.
15948 (nvptx_assemble_undefined_decl): Set nvptx_assemble_decl_begin call
15949 argument for undefined param to true.
15951 2018-04-11 Aaron Sawdey <acsawdey@linux.ibm.com>
15954 * doc/invoke.texi (RS/6000 and PowerPC Options): Document options
15955 -mcall- and -mtraceback=. Remove options -mabi=spe and -mabi=no-spe
15956 from PowerPC section.
15957 * config/rs6000/sysv4.opt (mcall-): Improve help text.
15958 * config/rs6000/rs6000.opt (mblock-compare-inline-limit=): Trim
15959 help text that is too long.
15960 * config/rs6000/rs6000.opt (mblock-compare-inline-loop-limit=): Trim
15961 help text that is too long.
15962 * config/rs6000/rs6000.opt (mstring-compare-inline-limit=): Trim
15963 help text that is too long.
15965 2018-04-11 Uros Bizjak <ubizjak@gmail.com>
15967 * config/alpha/alpha.md (stack_probe_internal): Rename
15968 from "probe_stack". Update all callers.
15970 2018-04-11 Alexander Monakov <amonakov@ispras.ru>
15972 PR rtl-optimization/84566
15973 * sched-deps.c (sched_analyze_insn): Check deps->readonly when invoking
15974 sched_macro_fuse_insns.
15976 2018-04-11 Alexander Monakov <amonakov@ispras.ru>
15979 * sched-rgn.c (add_branch_dependences): Move sel_sched_p check here...
15980 (compute_block_dependences): ... from here.
15982 2018-04-11 Jakub Jelinek <jakub@redhat.com>
15984 PR tree-optimization/85331
15985 * vec-perm-indices.h (vec_perm_indices::clamp): Change input type
15986 from int to HOST_WIDE_INT.
15988 2018-04-11 Martin Jambor <mjambor@suse.cz>
15991 * ipa-cp.c (propagate_vals_across_pass_through): Expand comment.
15992 (cgraph_edge_brings_value_p): New parameter dest_val, check if it is
15993 not the same as the source val.
15994 (cgraph_edge_brings_value_p): New parameter.
15995 (gather_edges_for_value): Pass destination value to
15996 cgraph_edge_brings_value_p.
15997 (perhaps_add_new_callers): Likewise.
15998 (get_info_about_necessary_edges): Likewise and exclude values brought
15999 only by self-recursive edges.
16000 (create_specialized_node): Redirect only clones of self-calling edges.
16001 (+self_recursive_pass_through_p): New function.
16002 (find_more_scalar_values_for_callers_subset): Use it.
16003 (find_aggregate_values_for_callers_subset): Likewise.
16004 (known_aggs_to_agg_replacement_list): Removed.
16005 (decide_whether_version_node): Re-calculate known constants for all
16006 remaining context clones.
16008 2018-04-11 Richard Biener <rguenther@suse.de>
16011 * dwarf2out.c (dwarf2out_finish): Remove DW_AT_stmt_list attribute
16012 from early DWARF output.
16013 (dwarf2out_early_finish): Output line info unconditionally into
16014 early DWARF and add reference to it.
16016 2018-04-11 Jakub Jelinek <jakub@redhat.com>
16019 * config/i386/sse.md (iptr): Add V16SFmode and V8DFmode cases.
16020 (<avx512>_vec_dup<mode><mask_name>): Use a single pattern for modes
16021 other than V2DFmode using iptr mode attribute.
16022 (<avx512>_vec_dup<mode><mask_name>): Use iptr mode attribute.
16024 2018-04-11 Alexander Monakov <amonakov@ispras.ru>
16026 PR rtl-optimization/84659
16027 * sel-sched-ir.c (sel_init_pipelining): Invoke cleanup_cfg.
16029 2018-04-11 Jakub Jelinek <jakub@redhat.com>
16032 * dwarf2out.c (skip_loc_list_entry): Don't call size_of_locs if
16034 (output_loc_list): Pass address of a dummy size variable even in the
16035 locview handling loop.
16036 (index_location_lists): Add comment on why skip_loc_list_entry can't
16039 2018-04-11 Thomas Preud'homme <thomas.preudhomme@arm.com>
16042 * config/arm/arm-builtins.c (arm_expand_builtin): Force input operand
16045 2018-04-10 Aaron Sawdey <acsawdey@linux.ibm.com>
16048 * doc/invoke.texi (RS/6000 and PowerPC Options): Document options
16049 -mblock-compare-inline-limit, -mblock-compare-inline-loop-limit,
16050 and -mstring-compare-inline-limit.
16052 2018-04-10 Segher Boessenkool <segher@kernel.crashing.org>
16055 * config/rs6000/rs6000.md (allocate_stack): Put the residual size
16056 for stack clash protection in a register whenever we need it to be in
16059 2018-04-10 Segher Boessenkool <segher@kernel.crashing.org>
16061 * common/config/rs6000/rs6000-common.c (rs6000_option_init_struct):
16062 Enable -fasynchronous-unwind-tables by default if OBJECT_FORMAT_ELF.
16064 2018-04-10 Segher Boessenkool <segher@kernel.crashing.org>
16067 * config/rs6000/rs6000.opt (mtraceback=): Show the allowed values in
16069 (mlong-double-): Ditto.
16070 * config/rs6000/sysv4.opt (msdata=): Ditto.
16071 (mtls-size=): Ditto.
16073 2018-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
16075 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
16076 erroneous entries for
16077 "vector int vec_ldl (int, long int *)", and
16078 "vector unsigned int vec_ldl (int, unsigned long int *)".
16079 Add comments and entries for
16080 "vector bool char vec_ldl (int, bool char *)",
16081 "vector bool short vec_ldl (int, bool short *)",
16082 "vector bool int vec_ldl (int, bool int *)",
16083 "vector bool long long vec_ldl (int, bool long long *)",
16084 "vector pixel vec_ldl (int, pixel *)",
16085 "vector long long vec_ldl (int, long long *)",
16086 "vector unsigned long long vec_ldl (int, unsigned long long *)".
16087 * config/rs6000/rs6000.c (rs6000_init_builtins): Initialize new
16088 type tree bool_long_long_type_node and correct definition of
16089 bool_V2DI_type_node to make reference to this new type tree.
16090 (rs6000_mangle_type): Replace erroneous reference to
16091 bool_long_type_node with bool_long_long_type_node.
16092 * config/rs6000/rs6000.h (enum rs6000_builtin_type_index): Add
16093 comments to emphasize sign distinctions for char and int types and
16094 replace RS6000_BTI_bool_long constant with
16095 RS6000_BTI_bool_long_long constant. Also add comment to restrict
16096 use of RS6000_BTI_pixel.
16097 (bool_long_type_node): Remove this macro definition.
16098 (bool_long_long_type_node): New macro definition
16100 2018-04-10 Jakub Jelinek <jakub@redhat.com>
16102 PR rtl-optimization/85300
16103 * combine.c (subst): Handle subst of CONST_SCALAR_INT_P new_rtx also
16104 into FLOAT and UNSIGNED_FLOAT like ZERO_EXTEND, return a CLOBBER if
16105 simplify_unary_operation fails.
16107 2018-04-10 Martin Liska <mliska@suse.cz>
16109 * gdbhooks.py: Add pretty-printers for varpool_node, symtab_node,
16110 cgraph_edge and ipa_ref.
16112 2018-04-10 Jakub Jelinek <jakub@redhat.com>
16116 * config/i386/sse.md
16117 (<extract_type>_vinsert<shuffletype><extract_suf>_mask): Fix
16118 computation of the VEC_MERGE selector from mask.
16119 (<extract_type>_vinsert<shuffletype><extract_suf>_1<mask_name>):
16120 Fix decoding of the VEC_MERGE selector into mask.
16122 2018-04-10 Richard Sandiford <richard.sandiford@linaro.org>
16124 PR tree-optimization/85286
16125 * tree-vect-data-refs.c (vect_get_smallest_scalar_type):
16127 2018-04-10 Richard Sandiford <richard.sandiford@linaro.org>
16129 * final.c (final_1): Set insn_last_address as well as
16130 insn_current_address.
16132 2018-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16135 * explow.c (emit_stack_probe): Call validize_mem on memory location
16136 before passing it to gen_probe_stack. Create address operand and
16137 legitimize it for the probe_stack_address case.
16139 2018-04-09 Jan Hubicka <jh@suse.cz>
16142 * ipa-devirt.c (rebuild_type_inheritance-hash): New.
16143 * ipa-utils.h (rebuild_type_inheritance-hash): Declare.
16144 * tree.c (free_lang_data_in_type): Fix handling of binfos;
16146 (free_lang_data): Rebuild type inheritance graph.
16148 2018-04-09 Martin Sebor <msebor@redhat.com>
16150 * invoke.texi (-finline-small-functions): Mention other optimization
16152 (-findirect-inlining, -fpartial-inlining): Same.
16153 (-finline-functions-called-once): Same.
16154 (-freorder-blocks-and-partition): Same.
16156 2018-04-09 Jan Hubicka <jh@suse.cz>
16159 * cfgcleanup.c (try_forward_edges): Do not give up on crossing
16160 jumps; choose last target that matches the criteria (i.e.
16161 no partition changes for non-crossing jumps).
16162 * cfgrtl.c (cfg_layout_redirect_edge_and_branch): Add basic
16163 support for redirecting crossing jumps to non-crossing.
16165 2018-04-09 Alexey Brodkin <abrodkin@synopsys.com>
16167 * config/arc/arc.c (arc_expand_prologue): Set stack usage info
16168 also for naked functions.
16170 2018-04-09 Claudiu Zissulescu <claziss@synopsys.com>
16172 * config/arc/arc.md (add_shift): New pattern.
16173 (add_shift2): Likewise.
16174 (sub_shift): Likewise.
16175 (sub_shift_cmp0_noout): Likewise.
16176 (compare_si_ashiftsi): Likewise.
16177 (xbfu_cmp0_noout): New combine pattern.
16178 (xbfu_cmp0"): Likewise.
16179 (movsi_set_cc_insn): Place the predicable variant first.
16180 (commutative_binary_cmp0_noout): Remove clobber.
16181 (commutative_binary_cmp0): New pattern.
16182 (noncommutative_binary_cmp0): Likewise.
16183 (noncommutative_binary_cmp0_noout): Likewise.
16184 (noncommutative_binary_comparison_result_used): Removed.
16185 (rsub_cmp0): New pattern.
16186 (rsub_cmp0_noout): Likewise.
16187 (extzvsi): Changed, keep only meaningful variants.
16188 (SQH, SEZ): New iterators.
16189 (SQH_postfix): New mode attribute.
16190 (SEZ_prefix): New code attribute.
16191 (<SEZ_prefix>xt<SQH_postfix>_cmp0_noout): New instruction pattern.
16192 (<SEZ_prefix>xt<SQH_postfix>_cmp0): Likewise.
16193 * config/arc/predicates.md (cc_set_register): Use CC_REG instead
16194 of numerical value.
16195 (noncommutative_operator): Check the availability of barrel
16198 2018-04-09 Richard Biener <rguenther@suse.de>
16200 PR tree-optimization/85284
16201 * tree-ssa-loop-niter.c (number_of_iterations_exit_assumptions):
16202 Only use the niter constraining form of simple_iv when the exit
16203 is always executed.
16205 2018-04-09 Tom de Vries <tom@codesourcery.com>
16208 * config/nvptx/nvptx.md (define_c_enum "unspecv"): Add UNSPECV_MEMBAR.
16209 (define_expand "*memory_barrier"): New define_expand.
16210 (define_insn "memory_barrier"): New insn.
16212 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
16214 PR rtl-optimization/80463
16215 PR rtl-optimization/83972
16216 PR rtl-optimization/83480
16218 * sel-sched-ir.c (has_dependence_note_mem_dep): Take into account the
16219 correct producer for the insn.
16220 (tidy_control_flow): Fixup seqnos in case of debug insns.
16222 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
16224 PR rtl-optimization/83913
16226 * sel-sched-ir.c (merge_expr_data): Choose the middle between two
16227 different sched-times when merging exprs.
16229 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
16231 PR rtl-optimization/83962
16233 * sel-sched-ir.c (tidy_control_flow): Correct the order in which we call
16234 tidy_fallthru_edge and tidy_control_flow.
16236 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
16238 PR rtl-optimization/83530
16240 * sel-sched.c (force_next_insn): New global variable.
16241 (remove_insn_for_debug): When force_next_insn is true, also leave only
16242 next insn in the ready list.
16243 (sel_sched_region): When the region wasn't scheduled, make another pass
16244 over it with force_next_insn set to 1.
16246 2018-04-08 Monk Chiang <sh.chiang04@gmail.com>
16248 * config.gcc (nds32le-*-*, nds32be-*-*): Add nds32/nds32_intrinsic.h
16250 * config/nds32/constants.md (unspec_volatile_element): Add enum values
16251 for interrupt control.
16252 * config/nds32/nds32-intrinsic.c: Implementation of intrinsic
16253 functions for interrupt control.
16254 * config/nds32/nds32-intrinsic.md: Likewise.
16255 * config/nds32/nds32_intrinsic.h: Likewise.
16256 * config/nds32/nds32.h (nds32_builtins): Likewise.
16258 2018-04-08 Chung-Ju Wu <jasonwucj@gmail.com>
16260 * config/nds32/nds32.c (nds32_init_machine_status,
16261 nds32_legitimate_index_p, nds32_legitimate_address_p): Consider
16262 strict_aligned_p field.
16263 (nds32_expand_to_rtl_hook): New function.
16264 (TARGET_EXPAND_TO_RTL_HOOK): Define.
16265 * config/nds32/nds32.h (machine_function): Add strict_aligned_p field.
16267 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
16268 Chung-Ju Wu <jasonwucj@gmail.com>
16270 * config.gcc (nds32*-*-*): Check that n7 is valid to --with-cpu.
16271 * config/nds32/nds32-n7.md: New file.
16272 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N7.
16273 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n7
16275 * config/nds32/nds32-protos.h: More declarations for n7 pipeline.
16276 * config/nds32/nds32.md (pipeline_model): Add n7.
16277 * config/nds32/nds32.opt (mcpu): Support n7 pipeline cpus.
16278 * config/nds32/pipelines.md: Include n7 settings.
16280 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
16281 Chung-Ju Wu <jasonwucj@gmail.com>
16283 * config.gcc (nds32*-*-*): Check that e8 is valid to --with-cpu.
16284 * config/nds32/nds32-e8.md: New file.
16285 * config/nds32/nds32-opts.h (nds32-cpu_type): Add CPU_E8.
16286 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for e8
16288 * config/nds32/nds32-protos.h: More declarations for e8 pipeline.
16289 * config/nds32/nds32.md (pipeline_model): Add e8.
16290 * config/nds32/nds32.opt (mcpu): Support e8 pipeline cpus.
16291 * config/nds32/pipelines.md: Include e8 settings.
16293 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
16294 Chung-Ju Wu <jasonwucj@gmail.com>
16296 * config.gcc (nds32*-*-*): Check that n6/n8/s8 are valid to --with-cpu.
16297 * config/nds32/nds32-n8.md: New file.
16298 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N6 and CPU_N8.
16299 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n8
16301 * config/nds32/nds32-protos.h: More declarations for n8 pipeline.
16302 * config/nds32/nds32-utils.c: More implementations for n8 pipeline.
16303 * config/nds32/nds32.md (pipeline_model): Add n8.
16304 * config/nds32/nds32.opt (mcpu): Support n8 pipeline cpus.
16305 * config/nds32/pipelines.md: Include n8 settings.
16307 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
16308 Chung-Ju Wu <jasonwucj@gmail.com>
16310 * config.gcc (nds32*): Add nds32-utils.o into extra_objs.
16311 * config/nds32/nds32-n9-2r1w.md: New file.
16312 * config/nds32/nds32-n9-3r2w.md: New file.
16313 * config/nds32/nds32-opts.h (nds32_cpu_type, nds32_mul_type,
16314 nds32_register_ports): New or modify for cpu n9.
16315 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n9
16317 * config/nds32/nds32-protos.h: More declarations for n9 pipeline.
16318 * config/nds32/nds32-utils.c: New file.
16319 * config/nds32/nds32.h (TARGET_PIPELINE_N9, TARGET_PIPELINE_SIMPLE,
16320 TARGET_MUL_SLOW): Define.
16321 * config/nds32/nds32.md (pipeline_model): New attribute.
16322 * config/nds32/nds32.opt (mcpu, mconfig-mul, mconfig-register-ports):
16323 New options that support cpu n9.
16324 * config/nds32/pipelines.md: Include n9 settings.
16325 * config/nds32/t-nds32 (nds32-utils.o): Add dependency.
16327 2018-04-08 Chung-Ju Wu <jasonwucj@gmail.com>
16329 * config/nds32/nds32-md-auxiliary.c (output_cond_branch): Output align
16330 information if necessary.
16331 (output_cond_branch_compare_zero): Likewise.
16332 * config/nds32/nds32.c (nds32_adjust_insn_length): Consider align case.
16333 (nds32_target_alignment): Refine for alignment.
16334 * config/nds32/nds32.h (NDS32_ALIGN_P): Define.
16335 (FUNCTION_BOUNDARY): Modify.
16336 * config/nds32/nds32.md (call_internal, call_value_internal): Consider
16338 * config/nds32/nds32.opt (malways-align, malign-functions): New.
16340 2018-04-08 Monk Chiang <sh.chiang04@gmail.com>
16342 * config/nds32/constants.md (unspec_volatile_element): Add values for
16343 TLB operation and data prefetch.
16344 * config/nds32/nds32-intrinsic.c: Implementation of intrinsic
16345 functions for TLB operation and data prefetch.
16346 * config/nds32/nds32-intrinsic.md: Likewise.
16347 * config/nds32/nds32_intrinsic.h: Likewise.
16348 * config/nds32/nds32.c (nds32_dpref_names): Likewise.
16349 (nds32_print_operand): Likewise.
16350 * config/nds32/nds32.h (nds32_builtins): Likewise.
16352 2018-04-07 Thomas Koenig <tkoenig@gcc.gnu.org>
16353 Andrew Pinski <pinsika@gcc.gnu.org>
16355 PR middle-end/82976
16356 * match.pd: Use constant_boolean_node of correct type instead of
16357 boolean_true_node or boolean_false_node for simplifying
16358 pointer comparisons to zero.
16360 2018-04-07 Jakub Jelinek <jakub@redhat.com>
16362 PR tree-optimization/80021
16363 * tree.c (verify_type_variant): Make error call in verify_variant_match
16364 translatable and remove final full stop.
16366 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
16368 * config/nds32/constants.md (unspec_volatile_element): Add
16369 UNSPEC_VOLATILE_EH_RETURN.
16370 * config/nds32/nds32-md-auxiliary.c (nds32_output_stack_push,
16371 nds32_output_stack_pop): Support dwarf exception handling process.
16372 * config/nds32/nds32-protos.h (nds32_dynamic_chain_address): Declare.
16373 * config/nds32/nds32.c (nds32_init_machine_status): Support dwarf
16374 exception handling process.
16375 (nds32_compute_stack_frame): Likewise.
16376 (nds32_return_addr_rtx): Likewise.
16377 (nds32_initial_elimination_offset): Likewise.
16378 (nds32_expand_prologue): Likewise.
16379 (nds32_expand_epilogue): Likewise.
16380 (nds32_dynamic_chain_address): New function.
16381 * config/nds32/nds32.h (machine_function): Add fields for dwarf
16382 exception handling.
16383 (DYNAMIC_CHAIN_ADDRESS): Define.
16384 (EH_RETURN_DATA_REGNO): Define.
16385 (EH_RETURN_STACKADJ_RTX): Define.
16386 * config/nds32/nds32.md (eh_return, nds32_eh_return): Implement
16387 patterns for dwarf exception handling.
16389 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
16391 * config/nds32/nds32.h: Clean up obsolete macros.
16393 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
16395 * config/nds32/constants.md (unspec_element, unspec_volatile_element):
16396 Add enum values for particular instructions.
16397 * config/nds32/nds32-intrinsic.c: Implementation of expanding
16398 particular intrinsic functions.
16399 * config/nds32/nds32-intrinsic.md: Likewise.
16400 * config/nds32/nds32_intrinsic.h: Likewise.
16401 * config/nds32/nds32.h (nds32_builtins): Likewise.
16402 * config/nds32/nds32.md (type): Add pbsad and pbsada.
16403 (btst, ave): New patterns for particular instructions.
16405 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
16407 * config/nds32/constants.md (unspec_element, unspec_volatile_element):
16408 Add enum values for atomic load/store and memory sync.
16409 * config/nds32/nds32-intrinsic.c: Implementation for atomic load/store
16411 * config/nds32/nds32-intrinsic.md: Likewise.
16412 * config/nds32/nds32_intrinsic.h: Likewise.
16413 * config/nds32/nds32.h (nds32_builtins): Likewise.
16415 2018-04-07 Jakub Jelinek <jakub@redhat.com>
16417 PR tree-optimization/85257
16418 * fold-const.c (native_encode_vector): If not all elts could fit
16419 and off is -1, return 0 rather than offset.
16420 * tree-ssa-sccvn.c (vn_reference_lookup_3): Pass
16421 (offseti - offset2) / BITS_PER_UNIT as 4th argument to
16422 native_encode_expr. Verify len * BITS_PER_UNIT >= maxsizei. Don't
16423 adjust buffer in native_interpret_expr call.
16425 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
16427 * config/nds32/constants.md (unspec_volatile_element): Add cache
16428 control enum values.
16429 * config/nds32/nds32-intrinsic.c: Add cache control expand functions.
16430 * config/nds32/nds32-intrinsic.md: Add cache control patterns.
16431 * config/nds32/nds32.c (nds32_cctl_names): New.
16432 (nds32_print_operand): Handle cache control register names.
16433 * config/nds32/nds32.h (nds32_builtins): New enum values.
16434 * config/nds32/nds32_intrinsic.h: Add cache control enum types and
16436 * config/nds32/nds32.md (type): Add mmu.
16437 * config/nds32/pipelines.md (simple_insn): Add mmu.
16439 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
16441 * config/nds32/nds32.md (type): Remove call.
16442 * config/nds32/pipelines.md (simple_insn): Likewise.
16444 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
16446 * config/nds32/constants.md (unspec_volatile_element): Add
16447 UNSPEC_VOLATILE_FMFCSR, UNSPEC_VOLATILE_FMTCSR and
16448 UNSPEC_VOLATILE_FMFCFG.
16449 * config/nds32/nds32-intrinsic.c (bdesc_noarg): New builtin
16450 description for fmfcfg and fmfcsr.
16451 (bdesc_1arg): Add fmtcsr.
16452 (bdesc_2arg): Add fcpynss, fcpyss, fcpynsd and fcpysd.
16453 (nds32_expand_builtin_impl): Deal with FPU intrinsic functions.
16454 * config/nds32/nds32-intrinsic.md (unspec_fcpynsd, unspec_fcpysd,
16455 unspec_fcpynss, unspec_fcpysd, unspec_fcpyss, unspec_fmfcsr,
16456 unspec_fmfcfg): New patterns.
16457 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_FMFCFG,
16458 NDS32_BUILTIN_FMFCSR, NDS32_BUILTIN_FMTCSR, NDS32_BUILTIN_FCPYNSS,
16459 NDS32_BUILTIN_FCPYSS,NDS32_BUILTIN_FCPYNSD and NDS32_BUILTIN_FCPYSD.
16460 * config/nds32/nds32_intrinsic.h (__nds32__fcpynsd, __nds32__fcpynss,
16461 __nds32__fcpysd, __nds32__fcpyss, __nds32__fmfcsr, __nds32__fmtcsr,
16462 __nds32__fmfcfg): Define.
16464 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
16466 * config/nds32/nds32.c (nds32_intrinsic_register_names): Add more
16467 intrinsic register names.
16468 * config/nds32/nds32_intrinsic.h (nds32_intrinsic_registers): Add more
16469 intrinsic register enum values and macros.
16471 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
16473 * config/nds32/nds32.c (nds32_legitimate_index_p): Modify condition
16474 for load/store addressing form.
16475 (nds32_print_operand_address): Likewise.
16477 2018-04-06 Eric Botcazou <ebotcazou@adacore.com>
16480 * config/sparc/sparc.c (sparc_expand_move): Deal with symbolic operands
16481 based on LABEL_REF. Remove useless assertion.
16482 (pic_address_needs_scratch): Fix formatting.
16483 (sparc_legitimize_pic_address): Minor tweaks.
16484 (sparc_delegitimize_address): Adjust assertion accordingly.
16485 * config/sparc/sparc.md (movsi_pic_label_ref): Change label_ref_operand
16486 into symbolic_operand.
16487 (movsi_high_pic_label_ref): Likewise.
16488 (movsi_lo_sum_pic_label_ref): Likewise.
16489 (movdi_pic_label_ref): Likewise.
16490 (movdi_high_pic_label_ref): Likewise.
16491 (movdi_lo_sum_pic_label_ref): Likewise.
16493 2018-04-06 Amaan Cheval <amaan.cheval@gmail.com>
16495 * config.gcc (x86_64-*-rtems*): Add rtems.h to tm_file for
16496 custom LIB_SPEC setup.
16498 2018-04-06 Ruslan Bukin <br@bsdpad.com>
16499 Kito Cheng <kito.cheng@gmail.com>
16501 * config.gcc (riscv*-*-freebsd*): Add RISC-V FreeBSD support.
16502 * config/riscv/freebsd.h: New.
16504 2018-04-06 Chung-Ju Wu <jasonwucj@gmail.com>
16506 * config/nds32/nds32.c (nds32_adjust_insn_length): Refine.
16507 * config/nds32/nds32.h (ADJUST_INSN_LENGTH): Change the location in
16510 2018-04-06 Chung-Ju Wu <jasonwucj@gmail.com>
16511 Kito Cheng <kito.cheng@gmail.com>
16513 * config/nds32/nds32-md-auxiliary.c (nds32_output_return,
16514 nds32_output_call, nds32_symbol_binds_local_p): New functions.
16515 * config/nds32/nds32-protos.h (nds32_output_call,
16516 nds32_output_return): Declare.
16517 * config/nds32/nds32.md: Refine all the call and return patterns.
16519 2018-04-06 Jakub Jelinek <jakub@redhat.com>
16522 * dwarf2out.c (rtl_for_decl_init): For STRING_CST initializer only
16523 build CONST_STRING if TYPE_MAX_VALUE is non-NULL and is INTEGER_CST.
16525 PR rtl-optimization/84872
16526 * cfgloopmanip.c (create_preheader): Use make_forwarder_block even if
16527 nentry == 1 when CP_FALLTHRU_PREHEADERS and single_entry is
16528 EDGE_CROSSING edge.
16530 2018-04-06 Tamar Christina <tamar.christina@arm.com>
16532 * expr.c (copy_blkmode_to_reg): Revert 254862.
16533 * doc/sourcebuild.texi (word_mode_no_slow_unalign): Likewise.
16535 2018-04-06 Richard Biener <rguenther@suse.de>
16537 PR middle-end/85244
16538 * tree-dfa.c (get_ref_base_and_extent): Reset seen_variable_array_ref
16539 after seeing a component reference with an adjacent field. Treat
16540 refs to arrays at struct end of external decls similar to
16541 refs to unconstrained commons.
16543 2018-04-06 Jakub Jelinek <jakub@redhat.com>
16546 * fold-const.c (twoval_comparison_p): Remove SAVE_P argument and don't
16547 look through SAVE_EXPRs with non-side-effects argument. Adjust
16549 (fold_comparison): Adjust twoval_comparison_p caller, don't handle
16552 2018-04-06 Richard Biener <rguenther@suse.de>
16554 PR middle-end/85180
16555 * alias.c (find_base_term): New wrapper around find_base_term
16556 unwinding CSELIB_VAL_PTR changes.
16557 (find_base_term): Do not restore CSELIB_VAL_PTR during the
16560 2018-04-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16562 * config/s390/s390.c (s390_z10_optimize_cmp): Expand dedicated NOP
16564 * config/s390/s390.md (UNSPECV_NOP_LR_0, UNSPECV_NOP_LR_1): New
16565 constant definitions.
16566 ("nop"): lr 0,0 -> nopr r0
16567 ("nop_lr0", "nop_lr1"): New insn definitions.
16569 2018-04-06 Chung-Ju Wu <jasonwucj@gmail.com>
16571 * config/nds32/nds32.md (*stack_push, *stack_pop): Use
16572 NDS32_V3PUSH_AVAILABLE_P macro.
16574 2018-04-06 Monk Chiang <sh.chiang04@gmail.com>
16575 Chung-Ju Wu <jasonwucj@gmail.com>
16577 * config.gcc (nds32*-*-*): Add v2j v3f v3s checking.
16578 (nds32*-*-*): Add float and fpu_config into supported_defaults.
16579 * common/config/nds32/nds32-common.c (TARGET_DEFAULT_TARGET_FLAGS):
16580 Include TARGET_DEFAULT_FPU_ISA and TARGET_DEFAULT_FPU_FMA.
16581 * config/nds32/constants.md (unspec_element): Add UNSPEC_COPYSIGN,
16582 UNSPEC_FCPYNSD, UNSPEC_FCPYNSS, UNSPEC_FCPYSD and UNSPEC_FCPYSS.
16583 * config/nds32/constraints.md: New constraints and checking for hard
16584 float configuration.
16585 * config/nds32/iterators.md: New mode iterator and attribute for hard
16586 float configuration.
16587 * config/nds32/nds32-doubleword.md: Use hard float alternatives and
16589 * config/nds32/nds32-fpu.md: New file.
16590 * config/nds32/nds32-md-auxiliary.c: New functions and checkings to
16591 deal with hard float code generation.
16592 * config/nds32/nds32-opts.h (nds32_arch_type): Add ARCH_V3F and
16594 (abi_type, float_reg_number): New enum type.
16595 * config/nds32/nds32-predicates.c: New predicates for hard float.
16596 * config/nds32/nds32-protos.h: Declare functions for hard float.
16597 * config/nds32/nds32.c: Implementation for hard float configuration.
16598 * config/nds32/nds32.h: Definitions for hard float configuration.
16599 * config/nds32/nds32.md: Include hard float machine description and
16600 modify patterns for hard float configuration.
16601 * config/nds32/nds32.opt: New options for hard float configuration.
16602 * config/nds32/predicates.md: New predicates for hard float
16605 2018-04-06 Kuan-Lin Chen <kuanlinchentw@gmail.com>
16607 * common/config/nds32/nds32-common.c
16608 (nds32_option_optimization_table): Enable -mreleax-hint by default.
16610 2018-04-05 Jakub Jelinek <jakub@redhat.com>
16612 PR middle-end/85195
16613 * match.pd (BIT_FIELD_REF CONSTRUCTOR@0 @1 @2): Use view_convert around
16614 CONSTRUCTOR_ELT (ctor, ...)->value.
16616 2018-04-05 Uros Bizjak <ubizjak@gmail.com>
16619 * config/i386/i386.md (define_attr "memory"): Handle rotate1 type.
16621 2018-04-05 Tom de Vries <tom@codesourcery.com>
16624 * config/nvptx/nvptx.c (nvptx_single): Fix neutering of bb with only
16627 2018-04-05 Shiva Chen <shiva0217@gmail.com>
16628 Kito Cheng <kito.cheng@gmail.com>
16630 * config/nds32/constraints.md (U33): Fine-tune checking condition.
16631 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Ditto.
16632 * config/nds32/nds32.h (nds32_16bit_address_type): Add
16633 ADDRESS_POST_MODIFY_LO_REG_IMM3U.
16635 2018-04-05 Shiva Chen <shiva0217@gmail.com>
16636 Kito Cheng <kito.cheng@gmail.com>
16638 * config/nds32/constraints.md (Ufe): New memory constraint.
16639 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format,
16640 nds32_output_16bit_load): Consider r8 register for lwi45.fe format.
16641 * config/nds32/nds32.c (nds32_print_operand): Output lwi45.fe
16643 * config/nds32/nds32.h (nds32_16bit_address_type): Add ADDRESS_R8_IMM7U.
16644 * config/nds32/nds32.md (*mov<mode>): Adjust pattern.
16646 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
16648 * config/nds32/nds32.md: Use optimize_size in the condition for
16649 alu-shift instructions.
16651 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
16653 * config/nds32/nds32.md (divsi4, udivsi4): New patterns.
16655 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
16657 * config/nds32/nds32.md (negsi2): Refine pattern.
16659 2018-04-05 Kito Cheng <kito.cheng@gmail.com>
16660 Chung-Ju Wu <jasonwucj@gmail.com>
16662 * config/nds32/iterators.md (shift_rotate): New code iterator.
16663 (shift): New code attribute.
16664 * config/nds32/nds32-md-auxiliary.c (nds32_expand_constant): New.
16665 * config/nds32/nds32-protos.h (nds32_expand_constant): Declare.
16666 * config/nds32/nds32.c (nds32_print_operand): Deal with more cases.
16667 * config/nds32/nds32.md (addsi3, *add_srli): Refine implementation for
16668 bit-wise operations.
16669 (andsi3, *andsi3): Ditto.
16670 (iorsi3, *iorsi3, *or_slli, *or_srli): Ditto.
16671 (xorsi3, *xorsi3, *xor_slli, *xor_srli): Ditto.
16672 (<shift>si3, *ashlsi3, *ashrsi3, *lshrsi3, *rotrsi3): Ditto.
16673 * config/nds32/predicates.md (nds32_rimm5u_operand, nds32_and_operand,
16674 nds32_ior_operand, nds32_xor_operand): New predicates.
16676 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
16678 * config/nds32/nds32.md (add<mode>3, sub<mode>3): Rename to ...
16679 (addsi3, subsi3): ... this.
16681 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
16683 * config/nds32/nds32.md (*sub_srli, *and_slli): Fine-tune predicator.
16685 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
16687 * config/nds32/nds32.md: Adjust indention.
16689 2018-04-05 Kito Cheng <kito.cheng@gmail.com>
16691 * config/nds32/nds32.md (feature): New attribute.
16693 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
16695 * config/nds32/nds32.md (subtype): New attribute.
16697 2018-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
16700 * config/arm/arm-builtins.c (arm_expand_builtin): Change
16701 expansion to perform a bitwise AND of the argument followed by a
16702 boolean negation of the result.
16704 2018-04-04 Peter Bergner <bergner@vnet.ibm.com>
16706 PR rtl-optimization/84878
16707 * ddg.c (add_cross_iteration_register_deps): Use DF_REF_BB to determine
16708 the basic block. Assert the use reference is not artificial and that
16709 it has an associated insn.
16711 2018-04-04 Michael Matz <matz@suse.de>
16713 * builtins.c (compute_objsize): Pass correct operand
16714 to array_at_struct_end_p.
16716 2018-04-04 Richard Biener <rguenther@suse.de>
16719 * dwarf2out.c (dwarf2out_register_external_die): Peel namespaces
16720 from contexts for DINFO_LEVEL_TERSE and below.
16722 2018-04-04 Kito Cheng <kito.cheng@gmail.com>
16724 * config/nds32/nds32-doubleword.md (move_<mode>): Require
16725 resiter_operand condition.
16726 * config/nds32/nds32.md (*move<mode>): Ditto.
16728 2018-04-04 Kito Cheng <kito.cheng@gmail.com>
16729 Monk Chiang <sh.chiang04@gmail.com>
16731 * config/nds32/nds32.md (movmisalign<mode>): New pattern.
16733 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
16735 * config/nds32/nds32.md (movqi, movhi): Merge into mov<mode>.
16737 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
16738 Kito Cheng <kito.cheng@gmail.com>
16740 * config/nds32/nds32-md-auxiliary.c (nds32_inverse_cond_code,
16741 nds32_cond_code_str, output_cond_branch,
16742 output_cond_branch_compare_zero, nds32_expand_cbranch,
16743 nds32_expand_cstore, nds32_expand_movcc,
16744 nds32_output_cbranchsi4_equality_zero,
16745 nds32_output_cbranchsi4_equality_reg,
16746 nds32_output_cbranchsi4_equality_reg_or_const_int,
16747 nds32_output_cbranchsi4_greater_less_zero: New functions.
16748 * config/nds32/nds32-protos.h (nds32_expand_cbranch,
16749 nds32_expand_cstore, nds32_expand_movcc,
16750 nds32_output_cbranchsi4_equality_zero,
16751 nds32_output_cbranchsi4_equality_reg,
16752 nds32_output_cbranchsi4_equality_reg_or_const_int,
16753 nds32_output_cbranchsi4_greater_less_zero): Declare.
16754 * config/nds32/predicates.md (nds32_movecc_comparison_operator,
16755 nds32_rimm11s_operand): New predicates.
16756 * config/nds32/nds32.h (nds32_expand_result_type): New enum type.
16757 * config/nds32/nds32.md: Rewrite all the branch and conditional move
16760 2018-04-04 Kito Cheng <kito.cheng@gmail.com>
16762 * config/nds32/nds32-doubleword.md: Refine all the instruction type.
16763 * config/nds32/nds32.md: Ditto.
16764 * config/nds32/pipelines.md: Ditto.
16766 2018-04-04 Richard Biener <rguenther@suse.de>
16768 PR tree-optimization/85168
16769 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address): Avoid
16770 propagating abnormals.
16772 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
16774 * config/nds32/nds32.md (enabled): Use yes/no for this attribute.
16776 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
16777 Kito Cheng <kito.cheng@gmail.com>
16779 * config/nds32/nds32-md-auxiliary.c (nds32_long_call_p): New function.
16780 * config/nds32/nds32-protos.h (nds32_long_call_p): Declare.
16781 * config/nds32/nds32.c (nds32_function_ok_for_sibcall): New function.
16782 (TARGET_FUNCTION_OK_FOR_SIBCALL): Define.
16783 * config/nds32/nds32.md (sibcall_internal): New.
16784 (sibcall_register): Remove.
16785 (sibcall_immediate): Remove.
16786 (sibcall_value_internal): New.
16787 (sibcall_value_register): Remove.
16788 (sibcall_value_immediate): Remove.
16789 * config/nds32/predicates.md (nds32_general_register_operand): New.
16790 (nds32_call_address_operand): New.
16792 2018-04-03 Jakub Jelinek <jakub@redhat.com>
16794 PR rtl-optimization/85167
16795 * shrink-wrap.c (move_insn_for_shrink_wrap): Don't set bb_uses and
16796 bb_defs if *split_p, instead preinitialize it to NULL.
16798 PR tree-optimization/85156
16799 * builtins.c (fold_builtin_expect): Use save_expr on arg1 to avoid
16800 evaluating the argument multiple times.
16802 2018-04-03 Bill Schmidt <wschmidt@linux.ibm.com>
16804 * config/rs6000/emmintrin.h (_mm_cvtpd_epi32): Use __vector rather
16806 (_mm_cvtpd_ps): Likewise.
16807 (_mm_cvttpd_epi32): Likewise.
16808 * config/rs6000/mmintrin.h (_mm_unpacklo_pi8): Likewise.
16809 * config/rs6000/xmmintrin.h: For strict-ANSI C++ or C11, undefine
16810 vector, pixel, and bool following altivec.h include.
16812 2018-04-03 Martin Sebor <msebor@redhat.com>
16814 * doc/extend.texi (Common Function Attributes): Clarify.
16815 (const attribute): Likewise.
16816 (pure attribute): Likewise.
16818 2018-04-03 Jakub Jelinek <jakub@redhat.com>
16821 * config/i386/i386.c (ix86_expand_vector_set): Use
16822 HOST_WIDE_INT_1U << elt instead of 1 << elt. Formatting fix.
16824 2018-04-03 Uros Bizjak <ubizjak@gmail.com>
16826 * config/i386/i386.c (emit_i387_cw_initialization): Always use logic
16827 instructions when changing rounding bits to preserve precision bits
16828 in the x87 control word.
16830 2018-04-03 Martin Liska <mliska@suse.cz>
16832 PR tree-optimization/82491
16833 * rtl.h (strip_offset_and_add): Replace += suboffset with
16834 poly_uint64 () + suboffset.
16836 2018-03-29 Martin Liska <mliska@suse.cz>
16837 Martin Jambor <mjambor@suse.cz>
16840 * ipa-cp.c (propagate_bits_across_jump_function): Bail out if
16841 param_type is not an integral or pointer type.
16843 2018-04-03 Richard Biener <rguenther@suse.de>
16845 * sese.h (recompute_all_dominators): Remove.
16847 2018-04-02 Martin Sebor <msebor@redhat.com>
16849 * doc/invoke.texi (-Wrestrict): Fix typos.
16851 2018-04-02 Jim Wilson <jimw@sifive.com>
16853 * config/riscv/riscv.h (SHIFT_COUNT_TRUNCATED): Set to zero.
16854 * config/riscv/riscv.md (<optab>si3): Use QImode shift count.
16855 (<optab>di3, <optab>si3_extend): Likewise.
16856 (<optab>si3_mask, <optab>si3_mask_1): New.
16857 (<optab>di3_mask, <optab>di3_mask_1): New.
16858 (<optab>si3_extend_mask, <optab>si3_extend_mask_1): New.
16859 (lshrsi3_zero_extend_1): Use VOIDmode shift count.
16860 * config/riscv/sync.md (atomic_test_and_set): Emit QImode shift count.
16862 2018-04-02 Gerald Pfeifer <gerald@pfeifer.com>
16864 * doc/cpp.texi (Variadic Macros): Fix line continuation in an
16867 2018-04-02 Chung-Ju Wu <jasonwucj@gmail.com>
16869 * config/nds32/nds32.c (TARGET_CANONICALIZE_COMPARISON): Define.
16870 (nds32_canonicalize_comparison): New function.
16872 2018-04-01 Chung-Ju Wu <jasonwucj@gmail.com>
16873 Kito Cheng <kito.cheng@gmail.com>
16874 Kuan-Lin Chen <kuanlinchentw@gmail.com>
16876 * config.gcc (nds32): Add nds32-relax-opt.o into extra_objs.
16877 * config/nds32/constants.md (unspec_volatile_element): Add
16878 UNSPEC_VOLATILE_RELAX_GROUP.
16879 * config/nds32/nds32-relax-opt.c: New file.
16880 * config/nds32/nds32-predicates.c
16881 (nds32_symbol_load_store_p): New function.
16882 * config/nds32/nds32-protos.h
16883 (nds32_symbol_load_store_p): Declare function.
16884 (make_pass_nds32_relax_opt): Declare new rtl pass function.
16885 * config/nds32/nds32.c
16886 (nds32_register_pass): New function to register pass.
16887 (nds32_register_passes): New function to register passes.
16888 * config/nds32/nds32.md (relax_group): New pattern.
16889 * config/nds32/nds32.opt (mrelax-hint): New option.
16890 * config/nds32/t-nds32 (nds32-relax-opt.o): New dependency.
16892 2018-04-01 Kito Cheng <kito.cheng@gmail.com>
16894 * config/nds32/t-nds32: Modify files dependency.
16896 2018-04-01 Chung-Ju Wu <jasonwucj@gmail.com>
16898 * config/nds32/nds32.h (FUNCTION_PROFILER): Output newline character.
16899 (PROFILE_HOOK): Define its implementation.
16901 2018-04-01 Chung-Ju Wu <jasonwucj@gmail.com>
16903 * config/nds32/nds32.h (WCHAR_TYPE, WCHAR_TYPE_SIZE): Use unsigned int
16904 type and 32-bit size.
16906 2018-04-01 Jakub Jelinek <jakub@redhat.com>
16908 PR middle-end/85090
16909 * config/i386/sse.md (V): Add V64QI and V32HI for TARGET_AVX512F.
16910 (V_128_256): New mode iterator.
16911 (*avx512dq_vextract<shuffletype>64x2_1 splitter): New define_split.
16912 (*avx512f_vextract<shuffletype>32x4_1 splitter): Likewise.
16913 (xop_pcmov_<mode><avxsizesuffix>): Use V_128_256 mode iterator instead
16915 * config/i386/i386.c (ix86_expand_vector_set): Improve V32HImode and
16916 V64QImode expansion for !TARGET_AVX512BW && TARGET_AVX512F.
16918 2018-03-31 Segher Boessenkool <segher@kernel.crashing.org>
16921 * config/rs6000/xmmintrin.h (_mm_set_ps, _mm_max_ps): Handle (quiet)
16922 NaN inputs correctly.
16924 2018-03-30 Peter Bergner <bergner@vnet.ibm.com>
16927 * config/rs6000/vsx.md (??r): New mode attribute.
16928 (*vsx_mov<mode>_64bit): Use it.
16929 (*vsx_mov<mode>_32bit): Likewise.
16931 2018-03-30 Martin Sebor <msebor@redhat.com>
16933 PR tree-optimization/84818
16934 * builtins.c (check_access): Use warning_n.
16936 2018-03-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
16939 * config/rs6000/rs6000-string.c (expand_compare_loop): Fix redundant
16941 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Fix redundant
16944 2018-03-30 Julia Koval <julia.koval@intel.com>
16947 * x86-tune.def (movx, partial_reg_dependency): Enable for
16950 2018-03-29 Vladimir Makarov <vmakarov@redhat.com>
16952 PR inline-asm/84985
16953 * lra-constraints.c (process_alt_operands): Move setting
16954 this_alternative_matches below.
16956 2018-03-29 Martin Liska <mliska@suse.cz>
16959 * doc/invoke.texi: Document how LTO works with debug info.
16960 Describe auto-load support of binutils. Mention 'x86-64'
16961 as valid option value of -march option.
16963 2018-03-29 Jakub Jelinek <jakub@redhat.com>
16965 * config/i386/sse.md (<avx512>_blendm<mode>): Use <sseintprefix>.
16968 * fold-const.c (operand_equal_p): Handle DEBUG_BEGIN_STMT.
16969 For STATEMENT_LIST, pass down OEP_LEXICOGRAPHIC and maybe
16970 OEP_NO_HASH_CHECK for recursive call, to avoid exponential
16973 2018-03-28 Peter Bergner <bergner@vnet.ibm.com>
16976 * config/rs6000/rs6000.h (RS6000_BTM_POWERPC64): New define.
16977 (RS6000_BTM_COMMON): Add RS6000_BTM_POWERPC64.
16978 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add support
16979 for RS6000_BTM_POWERPC64.
16980 (rs6000_invalid_builtin): Add handling for RS6000_BTM_POWERPC64
16981 (rs6000_builtin_mask_names): Add RS6000_BTM_POWERPC64.
16982 * config/rs6000/rs6000-builtin.def (BU_P7_POWERPC64_MISC_2): New macro
16985 (DIVDEU): Likewise.
16987 2018-03-28 Carl Love <cel@us.ibm.com>
16990 2017-09-27 Carl Love <cel@us.ibm.com>
16992 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
16993 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
16994 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
16997 2018-03-28 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16999 * config/rs6000/xmmintrin.h (_mm_max_pi16): Use __vector __bool
17000 instead of __vector bool.
17001 (_mm_max_pu8): Likewise.
17002 (_mm_min_pi16): Likewise.
17004 2018-03-28 Peter Bergner <bergner@vnet.ibm.com>
17007 * config/rs6000/rs6000-builtin.def (DIVWEO): Delete macro expansion.
17008 (DIVWEUO): Likewise.
17009 (DIVDEO): Likewise.
17010 (DIVDEUO): Likewise.
17011 * config/rs6000/rs6000.c (builtin_function_type): Remove support for
17012 DIVWEUO and DIVDEUO.
17013 * config/rs6000/rs6000.md (UNSPEC_DIVEO, UNSPEC_DIVEUO): Delete unspecs.
17014 (UNSPEC_DIV_EXTEND): Remove deleted unspecs.
17015 (div_extend): Likewise.
17016 * doc/extend.texi (__builtin_divweo): Remove documentation for deleted
17018 (__builtin_divweuo): Likewise.
17019 (__builtin_divdeo): Likewise.
17020 (__builtin_divdeuo): Likewise.
17022 2018-03-28 Jakub Jelinek <jakub@redhat.com>
17025 * config/i386/i386.md (*add<mode>3_carry_0, *addsi3_carry_zext_0,
17026 *sub<mode>3_carry_0, *subsi3_carry_zext_0): New patterns.
17028 PR tree-optimization/82004
17029 * gimple-match-head.c (optimize_pow_to_exp): New function.
17030 * match.pd (pow(C,x) -> exp(log(C)*x)): Wrap with #if GIMPLE.
17031 Don't fold to exp if optimize_pow_to_exp is false.
17033 2018-03-28 Martin Liska <mliska@suse.cz>
17036 * calls.c (initialize_argument_information): Fix trailing space.
17037 * common.opt: Fix typo and provide better explanation for
17038 -fsanitize-coverage option.
17039 * config/i386/i386.opt: Fix typo.
17041 2018-03-28 Jakub Jelinek <jakub@redhat.com>
17042 Martin Liska <mliska@suse.cz>
17045 * gimplify.c (asan_poison_variable): Don't do the check for
17046 gimplify_omp_ctxp here.
17047 (gimplify_decl_expr): Do it here.
17048 (gimplify_target_expr): Likewise.
17050 2018-03-28 Martin Liska <mliska@suse.cz>
17053 * config/i386/i386.c (ix86_function_arg_advance): Do not call
17054 chkp_type_bounds_count if MPX is not enabled.
17056 2018-03-27 Chung-Ju Wu <jasonwucj@gmail.com>
17058 * config/nds32/nds32.h (BRANCH_COST): Adjust cost.
17060 2018-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
17063 * config/rs6000/rs6000.c (create_complex_muldiv): New helper
17064 function to create the function decl for complex long double
17065 multiply and divide for -mabi=ieeelongdouble.
17066 (init_float128_ieee): Call it.
17068 2018-03-27 H.J. Lu <hongjiu.lu@intel.com>
17071 * config/i386/i386.c (ix86_trampoline_init): Insert ENDBR for
17072 -fcf-protection=branch -mibt.
17073 * config/i386/i386.h (TRAMPOLINE_SIZE): Increased by 4 bytes.
17075 2018-03-27 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
17078 * config/arm/arm.c (arm_valid_symbolic_address): Handle
17079 arm_word_relocations.
17081 2018-03-27 Cesar Philippidis <cesar@codesourcery.com>
17084 * config/nvptx/nvptx.c (nvptx_assemble_decl_begin): Add '[]' to
17085 extern array declarations.
17087 2018-03-27 Richard Biener <rguenther@suse.de>
17089 PR middle-end/84067
17090 * match.pd ((A * C) +- (B * C) -> (A+-B) * C): Guard with
17091 explicit single_use checks.
17093 2018-03-27 Richard Biener <rguenther@suse.de>
17095 PR tree-optimization/85082
17096 * tree-ssa-sccvn.c (vn_reference_lookup_or_insert_for_pieces):
17099 2018-03-27 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
17101 * config.gcc (aarch64*-*-linux*): New TARGET_DEFAULT_ASYNC_UNWIND_TABLES
17102 * common/config/aarch64/aarch64-common.c (aarch64_optimization_table[]):
17103 Turn on fasynchronous-unwind-tables and funwind-tables.
17105 2018-03-26 Uros Bizjak <ubizjak@gmail.com>
17108 * config/i386/i386.md (*bmi_blsr_<mode>_cmp): New insn pattern.
17109 (*bmi_blsr_<mode>_ccz): Ditto.
17111 2018-03-26 Tom de Vries <tom@codesourcery.com>
17113 PR tree-optimization/85063
17114 * omp-general.c (offloading_function_p): New function. Factor out
17116 * omp-offload.c (pass_omp_target_link::gate): ... here.
17117 * omp-general.h (offloading_function_p): Declare.
17118 * tree-switch-conversion.c (build_one_array): Mark CSWTCH.x variable
17119 with attribute omp declare target for offloading functions.
17121 2018-03-24 Richard Sandiford <richard.sandiford@linaro.org>
17123 PR tree-optimization/84005
17124 * tree-data-ref.h (get_base_for_alignment): Declare.
17125 * tree-data-ref.c (get_base_for_alignment_1): New function.
17126 (get_base_for_alignment): Likewise.
17127 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
17128 get_base_for_alignment to find a suitable base object, instead
17129 of always using drb->base_address.
17131 2018-03-23 Jakub Jelinek <jakub@redhat.com>
17133 PR inline-asm/85022
17134 * emit-rtl.c (init_emit_regs): Indicate that VOIDmode MEMs don't have
17135 known size by default.
17137 2018-03-23 Vladimir Makarov <vmakarov@redhat.com>
17139 PR inline-asm/85030
17140 * lra-constraints.c (process_alt_operands): Don't match BLKmode
17141 and non BLKmode operands.
17143 2018-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17146 * config/arm/arm.md (unaligned_loadhis): Remove first alternative.
17147 Clean up attributes.
17149 2018-03-23 Richard Biener <rguenther@suse.de>
17152 * dwarf2out.c (rtl_for_decl_location): Do not generate RTL early when
17153 we are going to emit early debug for LTO.
17155 2018-03-23 Jakub Jelinek <jakub@redhat.com>
17157 PR inline-asm/85034
17158 * function.c (match_asm_constraints_1): Don't optimize if input
17159 doesn't satisfy general_operand predicate for output's mode.
17161 PR inline-asm/85022
17162 * alias.c (write_dependence_p): Don't require for x_canonicalized
17163 non-VOIDmode if x has VOIDmode.
17166 * sanopt.c (maybe_optimize_ubsan_ptr_ifn): If DECL_REGISTER (base),
17167 just don't try to optimize it rather than assert it never happens.
17169 2018-03-22 Kelvin Nilsen <kelvin@gcc.gnu.org>
17171 * config/rs6000/rs6000-builtin.def: Remove various BU_ALTIVEC_X
17172 macro expansions for definition of ST_INTERNAL_<mode> and
17173 LD_INTERNAL_<mode> builtins.
17174 * config/rs6000/rs6000-protos.h (rs6000_address_for_altivec):
17176 * config/rs6000/rs6000.c (altivec_expand_ld_builtin): Delete this
17178 (altivec_expand_st_builtin): Likewise.
17179 (altivec_expand_builtin): Remove calls to deleted functions.
17180 (rs6000_address_for_altivec): Delete this function.
17181 * config/rs6000/vector.md: Remove expands for
17182 vector_altivec_load_<mode> and vector_altivec_store_<mode>.
17184 2018-03-22 Sudakshina Das <sudi.das@arm.com>
17187 * config/arm/arm.h (machine_function): Add static_chain_stack_bytes.
17188 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Avoid
17189 re-computing once computed.
17190 (arm_expand_prologue): Compute machine->static_chain_stack_bytes.
17191 (arm_init_machine_status): Initialize
17192 machine->static_chain_stack_bytes.
17194 2018-03-22 Kelvin Nilsen <kelvin@gcc.gnu.org>
17197 * doc/extend.texi: Add four new prototypes for vec_ld.
17198 * config/rs6000/rs6000-builtin.def (LVX_V1TI): Reorder symbol
17199 definitions for more logical presentation.
17200 * config/rs6000/rs6000-c.c: (altivec_overloaded_builtins): Add
17201 entries for V1TI variants of __builtin_altivec_ld builtin.
17202 * config/rs6000/rs6000.c: (altivec_expand_lv_builtin): Add test and
17203 handling of V1TI variant of LVX icode pattern.
17204 (altivec_expand_builtin): Add case for ALTIVEC_BUILTIN_LVX_V1TI.
17205 (rs6000_gimple_fold_builtin): Likewise.
17206 (altivec_init_builtins): Add code to define
17207 __builtin_altivec_lvx_v1ti function.
17209 2018-03-22 Jakub Jelinek <jakub@redhat.com>
17211 PR inline-asm/84941
17212 * function.c (match_asm_constraints_1): Don't do the optimization
17213 if input isn't a REG, SUBREG, MEM or constant.
17215 2018-03-22 Tom de Vries <tom@codesourcery.com>
17217 PR tree-optimization/84956
17218 * tree-ssa-tail-merge.c (find_clusters_1): Skip bbs with
17219 bb_has_abnormal_pred.
17221 2018-03-22 Jakub Jelinek <jakub@redhat.com>
17224 * dwarf2asm.c (dw2_output_indirect_constant_1): Set
17225 DECL_INITIAL (decl) to decl at the end.
17226 * varasm.c (use_blocks_for_decl_p): Revert the 2018-03-20 change,
17227 adjust the comment.
17229 2018-03-21 Joseph Myers <joseph@codesourcery.com>
17231 * doc/extend.texi (__builtin_tgmath): Document when complex
17232 integer types are treated as _Complex _Float64.
17234 2018-03-21 Tom de Vries <tom@codesourcery.com>
17236 * doc/extend.texi (__builtin_extend_pointer): Remove pasto.
17238 2018-03-21 Jakub Jelinek <jakub@redhat.com>
17240 PR tree-optimization/84960
17241 * tree-cfg.c (remove_bb): Don't move forced labels into bb->prev_bb
17242 if it is ENTRY block, move them into single succ of ENTRY in that case.
17244 2018-03-21 Richard Sandiford <richard.sandiford@linaro.org>
17246 PR tree-optimization/84811
17247 * poly-int.h (poly_span_traits): Remove the T3 parameter and
17248 promote HOST_WIDE_INT T2 - T1 results to unsigned HOST_WIDE_INT.
17249 (maybe_in_range_p, known_in_range_p, ranges_known_overlap_p):
17250 (known_subrange_p): Update accordingly. Cast each value involved
17251 in the size comparison, rather than casting the result of the
17254 2018-03-21 Jakub Jelinek <jakub@redhat.com>
17256 PR tree-optimization/84982
17257 * gimple-ssa-store-merging.c (invert_op): Handle boolean inversion
17258 by flipping the least significant bit rather than all bits from
17259 bitpos to bitpos + bitsize - 1.
17261 2018-03-21 Nathan Sidwell <nathan@acm.org>
17263 * doc/extend.texi (Deprecated Features): Remove mention of
17264 long-deleted deprecations.
17266 2018-03-21 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
17269 * configure.ac (gcc_cv_ld_soname) <*-*-solaris2*>: Set.
17270 * configure: Regenerate.
17272 2018-03-21 Tom de Vries <tom@codesourcery.com>
17274 PR tree-optimization/83126
17275 * tree-parloops.c (num_phis): New function.
17276 (gen_parallel_loop): Detect and handle canonicalize_loop_ivs failure.
17278 2018-03-21 Nathan Sidwell <nathan@acm.org>
17280 * doc/extend.texi (Deprecated Features): Update deprecated flags,
17281 mention anon-struct/union members and trailing attributes.
17283 2018-03-21 Bin Cheng <bin.cheng@arm.com>
17285 PR tree-optimization/84969
17286 * tree-loop-distribution.c (fuse_memset_builtins): Don't reorder
17287 builtin memset partitions if they set different rhs values.
17289 2018-03-21 Jakub Jelinek <jakub@redhat.com>
17291 PR rtl-optimization/84989
17292 * simplify-rtx.c (simplify_unary_operation_1): Don't try to simplify
17293 VEC_DUPLICATE with scalar result mode.
17295 2018-03-21 Martin Liska <mliska@suse.cz>
17298 * ipa-icf.c (sem_item_optimizer::fixup_points_to_sets): Remove
17299 not intended return statement.
17301 2018-03-21 Martin Liska <mliska@suse.cz>
17304 * tree-chkp.c (CHKP_ARRAY_MAX_CHECK_STEPS): Define a new macro.
17305 (chkp_find_bound_slots_1): Limit number of iterations.
17307 2018-03-20 David H. Gutteridge <dhgutteridge@sympatico.ca>
17310 * Minor grammar fixes for x86 options.
17312 2018-03-20 Jakub Jelinek <jakub@redhat.com>
17315 * dce.c (delete_unmarked_insns): Don't remove frame related noop moves
17316 holding REG_CFA_RESTORE notes, instead turn them into a USE.
17318 2018-03-20 Peter Bergner <bergner@vnet.ibm.com>
17321 * config/rs6000/altivec.md (altivec_lvx_<mode>_2op): Delete define_insn.
17322 (altivec_lvx_<mode>_1op): Likewise.
17323 (altivec_stvx_<mode>_2op): Likewise.
17324 (altivec_stvx_<mode>_1op): Likewise.
17325 (altivec_lvx_<VM2:mode>): New define_expand.
17326 (altivec_stvx_<VM2:mode>): Likewise.
17327 (altivec_lvx_<VM2:mode>_2op_<P:mptrsize>): New define_insn.
17328 (altivec_lvx_<VM2:mode>_1op_<P:mptrsize>): Likewise.
17329 (altivec_stvx_<VM2:mode>_2op_<P:mptrsize>): Likewise.
17330 (altivec_stvx_<VM2:mode>_1op_<P:mptrsize>): Likewise.
17331 * config/rs6000/rs6000-p8swap.c (rs6000_gen_stvx): Use new expanders.
17332 (rs6000_gen_lvx): Likewise.
17333 * config/rs6000/rs6000.c (altivec_expand_lv_builtin): Likewise.
17334 (altivec_expand_stv_builtin): Likewise.
17335 (altivec_expand_builtin): Likewise.
17336 * config/rs6000/vector.md: Likewise.
17338 2018-03-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17341 * config/arm/arm.c (arm_array_mode_supported_p): Return false for
17344 2018-03-20 Richard Biener <rguenther@suse.de>
17347 * config/i386/i386.c (ix86_add_stmt_cost): Only cost
17348 sign-conversions as zero, fall back to standard scalar_stmt
17351 2018-03-20 Martin Liska <mliska@suse.cz>
17354 * predict.c (rebuild_frequencies): Handle case when we have
17355 PROFILE_ABSENT, but flag_guess_branch_prob is false.
17357 2018-03-20 Jakub Jelinek <jakub@redhat.com>
17360 * dwarf2asm.c (dw2_output_indirect_constant_1): Temporarily turn off
17361 flag_section_anchors.
17362 * varasm.c (use_blocks_for_decl_p): Remove hack for
17363 dw2_force_const_mem.
17366 * config/aarch64/aarch64.md (*aarch64_reg_<mode>3_neg_mask2): Rename
17368 (*aarch64_<optab>_reg_<mode>3_neg_mask2): ... this. If pseudos can't
17369 be created, use lowpart_subreg of operands[0] rather than operands[0]
17371 (*aarch64_reg_<mode>3_minus_mask): Rename to ...
17372 (*aarch64_ashl_reg_<mode>3_minus_mask): ... this.
17373 (*aarch64_<optab>_reg_di3_mask2): Use const_int_operand predicate
17374 and n constraint instead of aarch64_shift_imm_di and Usd.
17375 (*aarch64_reg_<optab>_minus<mode>3): Rename to ...
17376 (*aarch64_<optab>_reg_minus<mode>3): ... this.
17378 2018-03-20 Sudakshina Das <sudi.das@arm.com>
17381 * config/arm/neon.md (ashldi3_neon): Update ?s for constraints
17382 to favor GPR over NEON registers.
17383 (<shift>di3_neon): Likewise.
17385 2018-03-20 Tom de Vries <tom@codesourcery.com>
17388 * config/nvptx/nvptx.c (nvptx_single): Don't neuter bar.sync.
17389 (nvptx_process_pars): Emit bar.sync asap and alap.
17391 2018-03-20 Tom de Vries <tom@codesourcery.com>
17394 * config/nvptx/nvptx.c (prevent_branch_around_nothing): Also update
17395 seen_label if seen_label is already set.
17397 2018-03-20 Jakub Jelinek <jakub@redhat.com>
17400 * config/i386/i386.c (fold_builtin_cpu): For features above 31
17401 use __cpu_features2 variable instead of __cpu_model.__cpu_features[0].
17402 Use 1U instead of 1. Formatting fixes.
17405 * builtins.c (fold_builtin_strpbrk): For strpbrk(x, "") use type
17406 instead of TREE_TYPE (s1) for the return value.
17408 2018-03-19 Jakub Jelinek <jakub@redhat.com>
17410 PR tree-optimization/84946
17411 * gimple-ssa-store-merging.c (mem_valid_for_store_merging): Compute
17412 bitsize + bitsize in poly_uint64 rather than poly_int64.
17415 * dwarf2asm.c: Include fold-const.c.
17416 (dw2_output_indirect_constant_1): Set DECL_INITIAL (decl) to ADDR_EXPR
17417 of decl rather than decl itself.
17419 PR rtl-optimization/84643
17420 * memmodel.h (enum memmodel): Add MEMMODEL_MAX enumerator.
17422 2018-03-19 Maxim Ostapenko <m.ostapenko@samsung.com>
17425 * dwarf2asm.c (dw2_output_indirect_constant_1): Disable ASan before
17426 calling assemble_variable.
17428 2018-03-19 Sudakshina Das <sudi.das@arm.com>
17431 * config/aarch64/aarch64-simd.md (vec_cmp<mode><v_int_equiv>): Modify
17432 instructions for UNLT, UNLE, UNGT, UNGE, UNEQ, UNORDERED and ORDERED.
17434 2018-03-19 Jim Wilson <jimw@sifive.com>
17437 * config/riscv/riscv.c (riscv_function_arg_boundary): Use
17438 PREFERRED_STACK_BOUNDARY instead of STACK_BOUNDARY.
17439 (riscv_first_stack_step): Likewise.
17440 (riscv_option_override): Use STACK_BOUNDARY instead of
17441 MIN_STACK_BOUNDARY.
17442 * config/riscv/riscv.h (STACK_BOUNDARY): Renamed from
17443 MIN_STACK_BOUNDARY.
17444 (BIGGEST_ALIGNMENT): Set to 128.
17445 (PREFERRED_STACK_BOUNDARY): Renamed from STACK_BOUNDARY.
17446 (RISCV_STACK_ALIGN): Use PREFERRED_STACK_BOUNDARY instead of
17449 2018-03-19 Richard Biener <rguenther@suse.de>
17451 PR tree-optimization/84933
17452 * tree-vrp.c (set_and_canonicalize_value_range): Treat out-of-bound
17453 values as -INF/INF when canonicalizing an ANTI_RANGE to a RANGE.
17455 2018-03-19 Richard Biener <rguenther@suse.de>
17457 PR tree-optimization/84859
17458 * tree-ssa-phiopt.c (single_trailing_store_in_bb): New function.
17459 (cond_if_else_store_replacement): Perform sinking operation on
17460 single-store BBs regardless of MAX_STORES_TO_SINK setting.
17461 Generalize what a BB with a single eligible store is.
17463 2018-03-19 Richard Biener <rguenther@suse.de>
17465 PR tree-optimization/84929
17466 * tree-data-ref.c (analyze_siv_subscript_cst_affine): Guard
17467 chrec_is_positive against non-chrec arg.
17469 2018-03-19 Tamar Christina <tamar.christina@arm.com>
17472 * config/arm/arm.c (arm_can_change_mode_class): revert r258554.
17474 2018-03-18 Martin Liska <mliska@suse.cz>
17476 PR rtl-optimization/84635
17477 * regrename.c (build_def_use): Use matches_mode only when
17480 2018-03-18 Richard Sandiford <richard.sandiford@linaro.org>
17482 PR tree-optimization/84913
17483 * tree-vect-loop.c (vectorizable_reduction): Don't try to
17484 vectorize chains of COND_EXPRs.
17486 2018-03-18 Chung-Ju Wu <jasonwucj@gmail.com>
17488 * config/nds32/nds32.h (MAX_REGS_PER_ADDRESS): Fix the value.
17490 2018-03-18 Chung-Ju Wu <jasonwucj@gmail.com>
17492 * config/nds32/nds32.h (LOGICAL_OP_NON_SHORT_CIRCUIT): Define.
17494 2018-03-18 Chung-Ju Wu <jasonwucj@gmail.com>
17496 * config/nds32/nds32.h (CLZ_DEFINED_VALUE_AT_ZERO): Define.
17498 2018-03-17 Chung-Ju Wu <jasonwucj@gmail.com>
17499 Kito Cheng <kito.cheng@gmail.com>
17501 * config/nds32/nds32-protos.h (nds32_adjust_reg_alloc_order): Declare.
17502 * config/nds32/nds32.c (nds32_reg_alloc_order_for_speed): New array.
17503 (nds32_adjust_reg_alloc_order): New function.
17504 * config/nds32/nds32.h (ADJUST_REG_ALLOC_ORDER): Define.
17506 2018-03-17 Kito Cheng <kito.cheng@gmail.com>
17508 * config/nds32/nds32.c (nds32_asm_output_mi_thunk,
17509 nds32_print_operand, nds32_print_operand_address): Use
17510 HOST_WIDE_INT_PRINT_DEC instead.
17512 2018-03-17 Chung-Ju Wu <jasonwucj@gmail.com>
17514 * config/nds32/nds32.c (nds32_register_priority): Modify cost.
17516 2018-03-17 Jakub Jelinek <jakub@redhat.com>
17519 * config/i386/i386.c (initial_ix86_tune_features,
17520 initial_ix86_arch_features): Use unsigned HOST_WIDE_INT rather than
17521 unsigned long long.
17522 (set_ix86_tune_features): Change ix86_tune_mask from unsigned int
17523 to unsigned HOST_WIDE_INT, initialize to HOST_WIDE_INT_1U << ix86_tune
17524 rather than 1u << ix86_tune. Formatting fix.
17525 (ix86_option_override_internal): Change ix86_arch_mask from
17526 unsigned int to unsigned HOST_WIDE_INT, initialize to
17527 HOST_WIDE_INT_1U << ix86_arch rather than 1u << ix86_arch.
17528 (ix86_function_specific_restore): Likewise.
17530 2018-03-16 Jakub Jelinek <jakub@redhat.com>
17533 * postreload.c (reload_combine_recognize_pattern): Perform
17534 INTVAL addition in unsigned HOST_WIDE_INT type to avoid UB and
17535 truncate_int_for_mode the result for the destination's mode.
17538 * hsa-gen.c (mem_type_for_type): Fix comment typo.
17539 * tree-vect-loop-manip.c (vect_create_cond_for_niters_checks):
17541 * gimple-ssa-warn-restrict.c (builtin_memref::set_base_and_offset):
17544 2018-03-16 Vladimir Makarov <vmakarov@redhat.com>
17547 * lra-assigns.c (lra_split_hard_reg_for): Don't use
17548 regno_allocno_class_array and sorted_pseudos.
17549 * lra-constraints.c (spill_hard_reg_in_range): Ignore hard regs in
17550 insns where regno is used.
17552 2018-03-16 Martin Liska <mliska@suse.cz>
17555 * multiple_target.c (create_dispatcher_calls): Redirect
17556 reference in the symbol table.
17558 2018-03-16 Martin Liska <mliska@suse.cz>
17561 * multiple_target.c (create_dispatcher_calls): Redirect also
17564 2018-03-16 Jakub Jelinek <jakub@redhat.com>
17568 * tree.h (TARGET_EXPR_NO_ELIDE): Define.
17569 * gimplify.c (gimplify_modify_expr_rhs): Don't elide TARGET_EXPRs with
17570 TARGET_EXPR_NO_ELIDE flag set unless *expr_p is INIT_EXPR.
17572 2018-03-16 Julia Koval <julia.koval@intel.com>
17574 * doc/invoke.texi (Skylake Server): Add CLWB.
17575 Cannonlake): Remove CLWB.
17577 2018-03-16 Jakub Jelinek <jakub@redhat.com>
17579 PR tree-optimization/84841
17580 * tree-ssa-reassoc.c (INTEGER_CONST_TYPE): Change to 1 << 4 from
17582 (FLOAT_ONE_CONST_TYPE): Define.
17583 (constant_type): Return FLOAT_ONE_CONST_TYPE for -1.0 and 1.0.
17584 (sort_by_operand_rank): Put entries with higher constant_type last
17585 rather than first to match comments.
17587 2018-03-15 Sandra Loosemore <sandra@codesourcery.com>
17589 * config/nios2/nios2.md (movsi_internal): Fix thinko in
17592 2018-03-15 Jakub Jelinek <jakub@redhat.com>
17595 * calls.c (expand_call): For TREE_ADDRESSABLE rettype ignore alignment
17596 check and use address of target always.
17598 2018-03-15 H.J. Lu <hongjiu.lu@intel.com>
17601 * config/i386/i386.c (indirect_thunk_needed): Update comments.
17602 (indirect_thunk_bnd_needed): Likewise.
17603 (indirect_thunks_used): Likewise.
17604 (indirect_thunks_bnd_used): Likewise.
17605 (indirect_return_needed): New.
17606 (indirect_return_bnd_needed): Likewise.
17607 (output_indirect_thunk_function): Add a bool argument for
17609 (output_indirect_thunk_function): Don't generate alias for
17610 function return thunk.
17611 (ix86_code_end): Call output_indirect_thunk_function to generate
17612 function return thunks.
17613 (ix86_output_function_return): Set indirect_return_bnd_needed
17614 and indirect_return_needed instead of indirect_thunk_bnd_needed
17615 and indirect_thunk_needed.
17617 2018-03-15 Olga Makhotina <olga.makhotina@intel.com>
17619 * config/i386/sgxintrin.h (_enclv_u32): New intrinsic.
17620 (__enclv_bc, __enclv_cd, __enclv_generic): New definitions.
17621 (ERDINFO, ETRACKC, ELDBC, ELDUC): New leaves.
17623 2018-03-15 David Malcolm <dmalcolm@redhat.com>
17624 Paul Hua <paul.hua.gm@gmail.com>
17627 * gcc.dg/fixits-pr84852-1.c: Fix filename in dg-regexp.
17629 2018-03-15 Segher Boessenkool <segher@kernel.crashing.org>
17631 * config/rs6000/rs6000.c (abi_v4_pass_in_fpr): Add back the
17632 TARGET_DOUBLE_FLOAT and TARGET_SINGLE_FLOAT conditions on the DFmode
17633 resp. SFmode cases.
17635 2018-03-15 Tamar Christina <tamar.christina@arm.com>
17638 * config/arm/arm.c (arm_can_change_mode_class): Use GET_MODE_UNIT_SIZE
17639 instead of GET_MODE_SIZE when comparing Units.
17641 2018-03-15 Vladimir Mezentsev <vladimir.mezentsev@oracle.com>
17644 * varasm.c (hash_section): Return an unchangeble hash value
17645 * config/aarch64/aarch64.c (aarch64_use_blocks_for_constant_p):
17646 Return !aarch64_can_use_per_function_literal_pools_p ().
17648 2018-03-15 Jakub Jelinek <jakub@redhat.com>
17651 * optabs.c (emit_conditional_move): Pass address of cmode's copy
17652 rather than address of cmode as last argument to prepare_cmp_insn.
17654 2018-03-15 Julia Koval <julia.koval@intel.com>
17656 * config/i386/i386.c (F_AVX512VBMI2, F_GFNI, F_VPCLMULQDQ,
17657 F_AVX512VNNI, F_AVX512BITALG): New.
17659 2018-03-14 John David Anglin <danglin@gcc.gnu.org>
17662 * config/pa/pa.c (pa_emit_move_sequence): Always emit secondary reload
17663 insn for floating-point loads and stores.
17665 2018-03-14 Carl Love <cel@us.ibm.com>
17667 * config/rs6000/rs6000-c.c: Add macro definitions for
17668 ALTIVEC_BUILTIN_VEC_PERMXOR.
17669 * config/rs6000/rs6000.h: Add #define for vec_permxor builtin.
17670 * config/rs6000/rs6000-builtin.def: Add macro expansions for VPERMXOR.
17671 * config/rs6000/altivec.md (altivec_vpermxor): New define expand.
17672 * config/rs6000/rs6000-p8swap.c (rtx_is_swappable_p): Add case
17674 * config/doc/extend.texi: Add prototypes for vec_permxor.
17676 2018-03-14 David Malcolm <dmalcolm@redhat.com>
17679 * diagnostic-show-locus.c (class layout_point): Convert m_line
17680 from int to linenum_type.
17681 (line_span::comparator): Use linenum "compare" function when
17682 comparing line numbers.
17683 (test_line_span): New function.
17684 (layout_range::contains_point): Convert param "row" from int to
17686 (layout_range::intersects_line_p): Likewise.
17687 (layout::will_show_line_p): Likewise.
17688 (layout::print_source_line): Likewise.
17689 (layout::should_print_annotation_line_p): Likewise.
17690 (layout::print_annotation_line): Likewise.
17691 (layout::print_leading_fixits): Likewise.
17692 (layout::annotation_line_showed_range_p): Likewise.
17693 (struct line_corrections): Likewise for field m_row.
17694 (line_corrections::line_corrections): Likewise for param "row".
17695 (layout::print_trailing_fixits): Likewise.
17696 (layout::get_state_at_point): Likewise.
17697 (layout::get_x_bound_for_row): Likewise.
17698 (layout::print_line): Likewise.
17699 (diagnostic_show_locus): Likewise for locals "last_line" and "row".
17700 (selftest::diagnostic_show_locus_c_tests): Call test_line_span.
17701 * input.c (selftest::test_linenum_comparisons): New function.
17702 (selftest::input_c_tests): Call it.
17703 * selftest.c (selftest::test_assertions): Test ASSERT_GT,
17704 ASSERT_GT_AT, ASSERT_LT, and ASSERT_LT_AT.
17705 * selftest.h (ASSERT_GT): New macro.
17706 (ASSERT_GT_AT): New macro.
17707 (ASSERT_LT): New macro.
17708 (ASSERT_LT_AT): New macro.
17710 2018-03-14 Segher Boessenkool <segher@kernel.crashing.org>
17712 PR rtl-optimization/84780
17713 * combine.c (distribute_links): Don't make a link based on pc_rtx.
17715 2018-03-14 Martin Liska <mliska@suse.cz>
17717 * tree.c (record_node_allocation_statistics): Use
17718 get_stats_node_kind.
17719 (get_stats_node_kind): New function extracted from
17720 record_node_allocation_statistics.
17721 (free_node): Use get_stats_node_kind.
17723 2018-03-14 Richard Biener <rguenther@suse.de>
17725 * tree-ssa-pre.c (compute_antic_aux): Remove code that asserts
17726 that the value-set of ANTIC_IN doesn't grow.
17729 * tree-ssa-pre.c (struct bb_bitmap_sets): Add visited_with_visited_succs
17731 (BB_VISITED_WITH_VISITED_SUCCS): New define.
17732 (compute_antic): Initialize BB_VISITED_WITH_VISITED_SUCCS.
17734 2018-03-14 Julia Koval <julia.koval@intel.com>
17736 * config.gcc (icelake-client, icelake-server): New.
17738 * config/i386/i386.c (initial_ix86_tune_features): Extend to 64 bit.
17739 (initial_ix86_arch_features): Ditto.
17740 (PTA_SKYLAKE): Add SGX.
17741 (PTA_ICELAKE): Remove.
17742 (PTA_ICELAKE_CLIENT): New.
17743 (PTA_ICELAKE_SERVER): New.
17744 (ix86_option_override_internal): Split up icelake on icelake client and
17746 (get_builtin_code_for_version): Ditto.
17747 (fold_builtin_cpu): Ditto.
17748 * config/i386/driver-i386.c (config/i386/driver-i386.c): Ditto.
17749 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto
17750 * config/i386/i386.h (processor_type): Ditto.
17751 * doc/invoke.texi: Ditto.
17753 2018-03-14 Jakub Jelinek <jakub@redhat.com>
17756 * sanopt.c (maybe_optimize_ubsan_ptr_ifn): Handle also
17757 INTEGER_CST offset, add it together with bitpos / 8 and
17758 sign extend based on POINTER_SIZE.
17762 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
17765 * config/i386/constraints.md (Yc): New register constraint.
17766 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
17767 Use Yc constraint for alternative 2 of operand 0. Remove
17768 preferred_for_speed attribute.
17770 2018-03-14 Richard Biener <rguenther@suse.de>
17772 PR tree-optimization/84830
17773 * tree-ssa-pre.c (compute_antic_aux): Intersect the new ANTIC_IN
17774 with the old one to avoid oscillations.
17776 2018-03-13 Vladimir Makarov <vmakarov@redhat.com>
17779 * lra-assigns.c (find_all_spills_for): Ignore uninteresting
17781 (assign_by_spills): Return a flag of reload assignment failure.
17782 Do not process the reload assignment failures. Do not spill other
17783 reload pseudos if they has the same reg class. Update n if
17785 (lra_assign): Add a return arg. Set up from the result of
17786 assign_by_spills call.
17787 (find_reload_regno_insns, lra_split_hard_reg_for): New functions.
17788 * lra-constraints.c (split_reg): Add a new arg. Use it instead of
17789 usage_insns if it is not NULL.
17790 (spill_hard_reg_in_range): New function.
17791 (split_if_necessary, inherit_in_ebb): Pass a new arg to split_reg.
17792 * lra-int.h (spill_hard_reg_in_range, lra_split_hard_reg_for): New
17793 function prototypes.
17794 (lra_assign): Change prototype.
17795 * lra.c (lra): Add code to deal with fails by splitting hard reg
17798 2018-03-01 Palmer Dabbelt <palmer@sifive.com>
17800 * config/riscv/riscv.opt (mrelax): New option.
17801 * config/riscv/riscv.c (riscv_file_start): Emit ".option
17802 "norelax" when riscv_mrelax is disabled.
17803 * doc/invoke.texi (RISC-V): Document "-mrelax" and "-mno-relax".
17805 2018-03-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
17808 * config/rs6000/rs6000.c (rs6000_reassociation_width): Disable parallel
17809 reassociation for int modes.
17811 2018-03-13 Richard Sandiford <richard.sandiford@linaro.org>
17813 * tree-vect-loop-manip.c (vect_maybe_permute_loop_masks):
17814 Reverse the choice between VEC_UNPACK_LO_EXPR and VEC_UNPACK_HI_EXPR
17816 * config/aarch64/iterators.md (hi_lanes_optab): New int attribute.
17817 * config/aarch64/aarch64-sve.md
17818 (*aarch64_sve_<perm_insn><perm_hilo><mode>): Rename to...
17819 (aarch64_sve_<perm_insn><perm_hilo><mode>): ...this.
17820 (*extend<mode><Vwide>2): Rename to...
17821 (aarch64_sve_extend<mode><Vwide>2): ...this.
17822 (vec_unpack<su>_<perm_hilo>_<mode>): Turn into a define_expand,
17823 renaming the old pattern to...
17824 (aarch64_sve_punpk<perm_hilo>_<mode>): ...this. Only define
17826 (vec_unpack<su>_<perm_hilo>_<SVE_BHSI:mode>): Turn into a
17827 define_expand, renaming the old pattern to...
17828 (aarch64_sve_<su>unpk<perm_hilo>_<SVE_BHSI:mode>): ...this.
17829 (*vec_unpacku_<perm_hilo>_<mode>_no_convert): Delete.
17830 (vec_unpacks_<perm_hilo>_<mode>): Take BYTES_BIG_ENDIAN into
17831 account when deciding which SVE instruction the optab should use.
17832 (vec_unpack<su_optab>_float_<perm_hilo>_vnx4si): Likewise.
17834 2018-03-13 Richard Sandiford <richard.sandiford@linaro.org>
17836 * config/aarch64/aarch64.md (V4_REGNUM, V8_REGNUM, V12_REGNUM)
17837 (V20_REGNUM, V24_REGNUM, V28_REGNUM, P1_REGNUM, P2_REGNUM, P3_REGNUM)
17838 (P4_REGNUM, P5_REGNUM, P6_REGNUM, P8_REGNUM, P9_REGNUM, P10_REGNUM)
17839 (P11_REGNUM, P12_REGNUM, P13_REGNUM, P14_REGNUM): New define_constants.
17840 (tlsdesc_small_<mode>): Turn a define_expand and use
17841 tlsdesc_small_sve_<mode> for SVE. Rename original define_insn to...
17842 (tlsdesc_small_advsimd_<mode>): ...this.
17843 (tlsdesc_small_sve_<mode>): New pattern.
17845 2018-03-13 Richard Sandiford <richard.sandiford@linaro.org>
17847 * config/aarch64/iterators.md (UNSPEC_SMUL_HIGHPART)
17848 (UNSPEC_UMUL_HIGHPART): New constants.
17849 (MUL_HIGHPART): New int iteraor.
17850 (su): Handle UNSPEC_SMUL_HIGHPART and UNSPEC_UMUL_HIGHPART.
17851 * config/aarch64/aarch64-sve.md (<su>mul<mode>3_highpart): New
17853 (*<su>mul<mode>3_highpart): New define_insn.
17855 2018-03-13 Eric Botcazou <ebotcazou@adacore.com>
17858 * ipa-devirt.c (odr_subtypes_equivalent_p): Do not get the ODR type of
17861 2018-03-13 Martin Liska <mliska@suse.cz>
17864 * (sem_item_optimizer::sem_item_optimizer): Initialize new
17866 (sem_item_optimizer::~sem_item_optimizer): Release it.
17867 (sem_item_optimizer::merge_classes): Register variable aliases.
17868 (sem_item_optimizer::fixup_pt_set): New function.
17869 (sem_item_optimizer::fixup_points_to_sets): Likewise.
17870 * ipa-icf.h: Declare new variables and functions.
17872 2018-03-13 Jakub Jelinek <jakub@redhat.com>
17874 PR middle-end/84834
17875 * match.pd ((A & C) != 0 ? D : 0): Use INTEGER_CST@2 instead of
17876 integer_pow2p@2 and test integer_pow2p in condition.
17877 (A < 0 ? C : 0): Similarly for @1.
17879 PR middle-end/84831
17880 * stmt.c (parse_output_constraint): If the CONSTRAINT_LEN (*p, p)
17881 characters starting at p contain '\0' character, don't look beyond
17885 * config/i386/i386.md (round<mode>2): For 387 fancy math, disable
17886 pattern if -ftrapping-math -fno-fp-int-builtin-inexact.
17889 * reg-stack.c (change_stack): Change update_end var from int to
17890 rtx_insn *, if non-NULL don't update just BB_END (current_block), but
17891 also call set_block_for_insn on the newly added insns and rescan.
17894 * config/i386/sse.md (sse2_loadhpd): Use Yv constraint rather than v
17895 on the last operand.
17898 * tree.c (stabilize_reference_1): Return save_expr (e) for
17899 STATEMENT_LIST even if it doesn't have side-effects.
17901 2018-03-12 Jonathan Wakely <jwakely@redhat.com>
17903 * doc/invoke.texi (-mclflushopt): Fix spelling of option.
17905 2018-03-12 Renlin Li <renlin.li@arm.com>
17907 * config/aarch64/aarch64.md (movhf_aarch64): Fix mode argument to
17908 aarch64_output_scalar_simd_mov_immediate.
17910 2018-03-12 Martin Sebor <msebor@redhat.com>
17912 PR tree-optimization/83456
17913 * gimple-fold.c (gimple_fold_builtin_memory_op): Avoid warning
17914 for perfectly overlapping calls to memcpy.
17915 (gimple_fold_builtin_memory_chk): Same.
17916 (gimple_fold_builtin_strcpy): Handle no-warning.
17917 (gimple_fold_builtin_stxcpy_chk): Same.
17918 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Handle no-warning.
17920 2018-03-12 Segher Boessenkool <segher@kernel.crashing.org>
17922 * config/rs6000/rs6000.c (abi_v4_pass_in_fpr): Add bool "named"
17923 parameter. Use it for SFmode.
17924 (rs6000_function_arg_advance_1): Adjust.
17925 (rs6000_function_arg): Adjust.
17926 (rs6000_gimplify_va_arg): Pass false for that new parameter.
17928 2018-03-12 Segher Boessenkool <segher@kernel.crashing.org>
17930 PR rtl-optimization/84169
17931 PR rtl-optimization/84780
17932 * combine.c (can_combine_p): Check for a 2-insn combination whether
17933 the destination register is used between the two insns, too.
17935 2018-03-12 Richard Biener <rguenther@suse.de>
17937 PR tree-optimization/84803
17938 * tree-if-conv.c (ifcvt_memrefs_wont_trap): Don't do anything
17939 for refs DR analysis didn't process.
17941 2018-03-12 Richard Biener <rguenther@suse.de>
17943 PR tree-optimization/84777
17944 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): For
17945 force-vectorize loops ignore whether we are optimizing for size.
17947 2018-03-12 Chung-Ju Wu <jasonwucj@gmail.com>
17949 * config/nds32/nds32.c (nds32_md_asm_adjust): New function.
17950 (TARGET_MD_ASM_ADJUST): Define.
17952 2018-03-12 Monk Chiang <sh.chiang04@gmail.com>
17953 Kito Cheng <kito.cheng@gmail.com>
17954 Chung-Ju Wu <jasonwucj@gmail.com>
17956 * config/nds32/nds32.c (nds32_compute_stack_frame,
17957 nds32_emit_stack_push_multiple, nds32_emit_stack_pop_multiple,
17958 nds32_emit_stack_v3push, nds32_emit_stack_v3pop,
17959 nds32_emit_adjust_frame, nds32_expand_prologue, nds32_expand_epilogue,
17960 nds32_expand_prologue_v3push, nds32_expand_epilogue_v3pop): Refine.
17961 * config/nds32/nds32.h (NDS32_FIRST_CALLEE_SAVE_GPR_REGNUM,
17962 NDS32_LAST_CALLEE_SAVE_GPR_REGNUM, NDS32_V3PUSH_AVAILABLE_P): New.
17963 * config/nds32/nds32.md (prologue, epilogue): Use macro
17964 NDS32_V3PUSH_AVAILABLE_P to do checking.
17966 2018-03-11 Jakub Jelinek <jakub@redhat.com>
17969 * dwarf2out.c (gen_enumeration_type_die): Don't guard adding
17970 DW_AT_declaration for ENUM_IS_OPAQUE on -gdwarf-4 or -gno-strict-dwarf,
17971 but on TYPE_SIZE. Don't do anything for ENUM_IS_OPAQUE if not creating
17972 a new die. Don't set TREE_ASM_WRITTEN if ENUM_IS_OPAQUE. Guard
17973 addition of most attributes on !orig_type_die or the attribute not
17974 being present already. Assert TYPE_VALUES is NULL for ENUM_IS_OPAQUE.
17976 2018-03-11 Kito Cheng <kito.cheng@gmail.com>
17977 Chung-Ju Wu <jasonwucj@gmail.com>
17979 * config/nds32/nds32.c (nds32_cpu_cpp_builtins): Modify to define
17980 __NDS32_VH__ macro.
17981 * config/nds32/nds32.opt (mvh): New option.
17983 2018-03-11 Kito Cheng <kito.cheng@gmail.com>
17984 Chung-Ju Wu <jasonwucj@gmail.com>
17986 * config/nds32/nds32-protos.h (nds32_cpu_cpp_builtins): Declare
17988 * config/nds32/nds32.c (nds32_cpu_cpp_builtins): New function.
17989 * config/nds32/nds32.h (TARGET_CPU_CPP_BUILTINS): Modify its
17992 2018-03-11 Kito Cheng <kito.cheng@gmail.com>
17993 Chung-Ju Wu <jasonwucj@gmail.com>
17995 * config/nds32/nds32-memory-manipulation.c (nds32_expand_strlen): New
17997 * config/nds32/nds32-multiple.md (strlensi): New pattern.
17998 * config/nds32/nds32-protos.h (nds32_expand_strlen): Declare function.
18000 2018-03-11 Monk Chiang <sh.chiang04@gmail.com>
18001 Kito Cheng <kito.cheng@gmail.com>
18002 Chung-Ju Wu <jasonwucj@gmail.com>
18004 * config/nds32/constants.md (unspec_element): Add UNSPEC_FFB,
18005 UNSPEC_FFMISM and UNSPEC_FLMISM.
18006 * config/nds32/nds32-intrinsic.c (bdesc_2arg): Add builtin description
18007 for ffb, ffmism and flmism.
18008 * config/nds32/nds32-intrinsic.md (unspec_ffb): Define new pattern.
18009 (unspec_ffmism): Ditto.
18010 (unspec_flmism): Ditto.
18011 (nds32_expand_builtin_impl): Check if string extension is available.
18012 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_FFB,
18013 NDS32_BUILTIN_FFMISM and NDS32_BUILTIN_FLMISM.
18015 2018-03-10 Vladimir Makarov <vmakarov@redhat.com>
18018 2018-03-09 Vladimir Makarov <vmakarov@redhat.com>
18021 * lra-assigns.c (assign_by_spills): Return a flag of reload
18022 assignment failure. Do not process the reload assignment
18023 failures. Do not spill other reload pseudos if they has the same
18025 (lra_assign): Add a return arg. Set up from the result of
18026 assign_by_spills call.
18027 (find_reload_regno_insns, lra_split_hard_reg_for): New functions.
18028 * lra-constraints.c (split_reg): Add a new arg. Use it instead of
18029 usage_insns if it is not NULL.
18030 (spill_hard_reg_in_range): New function.
18031 (split_if_necessary, inherit_in_ebb): Pass a new arg to split_reg.
18032 * lra-int.h (spill_hard_reg_in_range, lra_split_hard_reg_for): New
18033 function prototypes.
18034 (lra_assign): Change prototype.
18035 * lra.c (lra): Add code to deal with fails by splitting hard reg
18038 2018-03-10 H.J. Lu <hongjiu.lu@intel.com>
18041 * config/i386/i386.opt: Replace Enforcment with Enforcement.
18043 2018-03-10 Alexandre Oliva <aoliva@redhat.com>
18046 * dwarf2out.h (dw_val_class): Add dw_val_class_symview.
18047 (dw_val_node): Add val_symbolic_view.
18048 * dwarf2out.c (dw_line_info_table): Add symviews_since_reset.
18049 (symview_upper_bound): New.
18050 (new_line_info_table): Initialize symviews_since_reset.
18051 (dwarf2out_source_line): Count symviews_since_reset and set
18052 symview_upper_bound.
18053 (dw_val_equal_p): Handle symview.
18054 (add_AT_symview): New.
18055 (print_dw_val): Handle symview.
18056 (attr_checksum, attr_checksum_ordered): Likewise.
18057 (same_dw_val_p, size_of_die): Likewise.
18058 (value_format, output_die): Likewise.
18059 (add_high_low_attributes): Use add_AT_symview for entry_view.
18060 (dwarf2out_finish): Reset symview_upper_bound, clear
18063 2018-03-09 Peter Bergner <bergner@vnet.ibm.com>
18066 * config/rs6000/rs6000.c (rs6000_offsettable_memref_p): New prototype.
18067 Add strict argument and use it.
18068 (rs6000_split_multireg_move): Update for new strict argument.
18069 (mem_operand_gpr): Disallow all non-offsettable addresses.
18070 * config/rs6000/rs6000.md (*movdi_internal64): Use YZ constraint.
18072 2018-03-09 Jakub Jelinek <jakub@redhat.com>
18075 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Mark va_arg_tmp
18076 temporary TREE_ADDRESSABLE before gimplification of BUILT_IN_MEMCPY.
18077 * config/powerpcspe/powerpcspe.c (rs6000_gimplify_va_arg): Likewise.
18080 * tree-inline.c (copy_tree_body_r): For INDIRECT_REF of a remapped
18081 decl, use remap_type if we want to use the type.
18083 2018-03-09 Martin Sebor <msebor@redhat.com>
18085 PR tree-optimization/84526
18086 * gimple-ssa-warn-restrict.c (builtin_memref::set_base_and_offset):
18088 (builtin_access::generic_overlap): Be prepared to handle non-array
18091 2018-03-09 Alexandre Oliva <aoliva@redhat.com>
18093 PR rtl-optimization/84682
18094 * lra-constraints.c (process_address_1): Check is_address flag
18095 for address constraints.
18096 (process_alt_operands): Likewise.
18097 * lra.c (lra_set_insn_recog_data): Pass asm operand locs to
18098 preprocess_constraints.
18099 * recog.h (preprocess_constraints): Add oploc parameter.
18101 * recog.c (preprocess_constraints): Test address_operand for
18102 CT_ADDRESS constraints.
18104 2018-03-09 Vladimir Makarov <vmakarov@redhat.com>
18107 * lra-assigns.c (assign_by_spills): Return a flag of reload
18108 assignment failure. Do not process the reload assignment
18109 failures. Do not spill other reload pseudos if they has the same
18111 (lra_assign): Add a return arg. Set up from the result of
18112 assign_by_spills call.
18113 (find_reload_regno_insns, lra_split_hard_reg_for): New functions.
18114 * lra-constraints.c (split_reg): Add a new arg. Use it instead of
18115 usage_insns if it is not NULL.
18116 (spill_hard_reg_in_range): New function.
18117 (split_if_necessary, inherit_in_ebb): Pass a new arg to split_reg.
18118 * lra-int.h (spill_hard_reg_in_range, lra_split_hard_reg_for): New
18119 function prototypes.
18120 (lra_assign): Change prototype.
18121 * lra.c (lra): Add code to deal with fails by splitting hard reg
18124 2018-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18127 * common/config/arm/arm-common.c (arm_parse_arch_option_name):
18128 Accept complain bool parameter. Only emit errors if it is true.
18129 (arm_parse_cpu_option_name): Likewise.
18130 (arm_target_thumb_only): Adjust callers of the above.
18131 * config/arm/arm-protos.h (arm_parse_cpu_option_name): Adjust
18132 prototype to take a default true bool parameter.
18133 (arm_parse_arch_option_name): Likewise.
18135 2018-03-09 David Malcolm <dmalcolm@redhat.com>
18136 Francois-Xavier Coudert <fxcoudert@gcc.gnu.org>
18140 * Makefile.in (LD_VERSION_SCRIPT_OPTION, LD_SONAME_OPTION): New.
18141 * configure: Regenerate.
18142 * configure.ac ("linker --version-script option"): New.
18143 ("linker soname option"): New.
18145 2018-03-09 Richard Biener <rguenther@suse.de>
18147 PR tree-optimization/84775
18148 * tree-if-conv.c (add_bb_predicate_gimplified_stmts): Delink
18149 immediate uses of predicate stmts and mark them modified.
18152 PR tree-optimization/84178
18153 * tree-if-conv.c (combine_blocks): Move insert_gimplified_predicates
18155 (version_loop_for_if_conversion): Delay update_ssa call.
18156 (tree_if_conversion): Delay update_ssa until after predicate
18159 2018-03-09 Eric Botcazou <ebotcazou@adacore.com>
18162 * config/i386/winnt.c (i386_pe_seh_cold_init): Use small pre-allocation
18163 when the function accesses prior frames.
18165 2018-03-08 Jakub Jelinek <jakub@redhat.com>
18168 * dwarf2out.c (dw_loc_list): If list && loc_list->first->next, call
18169 gen_llsym, otherwise call maybe_gen_llsym.
18171 PR inline-asm/84742
18172 * recog.c (asm_operand_ok): Return 0 if multi-character constraint
18173 has ',' character inside of it.
18175 2018-03-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18178 * config/aarch64/aarch64.md (*compare_cstore<mode>_insn): Mark pattern
18179 as clobbering CC_REGNUM.
18181 2018-03-08 Richard Biener <rguenther@suse.de>
18183 PR middle-end/84552
18184 * tree-scalar-evolution.c: Include tree-into-ssa.h.
18185 (follow_copies_to_constant): Do not follow SSA names registered
18188 2018-03-08 Richard Biener <rguenther@suse.de>
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 2018-03-08 David Malcolm <dmalcolm@redhat.com>
18199 PR tree-optimization/84178
18200 * tree-if-conv.c (release_bb_predicate): Remove the
18201 the assertion that the stmts have NULL use_ops.
18202 Discard the statements, asserting that they haven't
18203 yet been added to a BB.
18205 2018-03-08 Richard Biener <rguenther@suse.de>
18207 PR tree-optimization/84746
18208 * tree-ssa-pre.c (find_leader_in_sets): Deal with SET1 being NULL.
18209 (phi_translate): Pass in destination ANTIC_OUT set.
18210 (phi_translate_1): Likewise. For a simplified result lookup
18211 a leader in ANTIC_OUT and AVAIL_OUT, not the ANTIC_IN sets.
18212 (phi_translate_set): Adjust.
18213 (do_pre_regular_insertion): Likewise.
18214 (do_pre_partial_partial_insertion): Likewise.
18216 2018-03-08 Martin Liska <mliska@suse.cz>
18218 PR gcov-profile/84735
18219 * doc/gcov.texi: Document usage of profile files.
18220 * gcov-io.h: Document changes in the format.
18222 2018-03-08 Alexandre Oliva <aoliva@redhat.com>
18226 * dwarf2out.c (struct dw_line_info_table): Update comments for
18228 (FORCE_RESET_NEXT_VIEW): New.
18229 (FORCE_RESETTING_VIEW_P): New.
18230 (RESETTING_VIEW_P): Check for -1 too.
18231 (ZERO_VIEW_P): Likewise.
18232 (new_line_info_table): Force-reset next view.
18233 (dwarf2out_begin_function): Likewise.
18234 (dwarf2out_source_line): Simplify zero_view_p initialization.
18235 Test FORCE_RESETTING_VIEW_P and RESETTING_VIEW_P instead of
18236 view directly. Omit view when omitting .loc at line 0.
18238 2018-03-08 Jakub Jelinek <jakub@redhat.com>
18240 PR tree-optimization/84740
18241 * tree-switch-conversion.c (process_switch): Call build_constructors
18242 only if info.phi_count is non-zero.
18244 PR tree-optimization/84739
18245 * tree-tailcall.c (find_tail_calls): Check call arguments against
18246 DECL_ARGUMENTS (current_function_decl) rather than
18247 DECL_ARGUMENTS (func) when checking for tail recursion.
18249 2018-03-07 Jakub Jelinek <jakub@redhat.com>
18251 * doc/contrib.texi: Add entries for Martin Liska, David Malcolm,
18252 Marek Polacek, extend Vladimir Makarov's, Jonathan Wakely's and
18253 Volker Reichelt's entry and add entries for people that perform
18254 GCC fuzzy testing and report numerous bugs.
18256 2018-03-07 Segher Boessenkool <segher@kernel.crashing.org>
18259 * config/rs6000/rs6000.c (rs6000_elf_in_small_data_p): Don't put
18260 readonly data in sdata, if that is disabled.
18261 * config/rs6000/sysv4.opt (mreadonly-in-sdata): New option.
18262 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
18263 -mreadonly-in-sdata option.
18265 2018-03-07 Martin Sebor <msebor@redhat.com>
18267 PR tree-optimization/84468
18268 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Consider successor
18269 basic block when looking for nul assignment.
18271 2018-03-07 Eric Botcazou <ebotcazou@adacore.com>
18274 * except.h (output_function_exception_table): Adjust prototype.
18275 * except.c (output_function_exception_table): Remove FNNAME parameter
18276 and add SECTION parameter. Ouput one part of the table at a time.
18277 * final.c (final_scan_insn_1) <NOTE_INSN_SWITCH_TEXT_SECTIONS>: Output
18278 the first part of the exception table and emit unwind directives.
18279 * config/i386/i386-protos.h (i386_pe_end_cold_function): Declare.
18280 (i386_pe_seh_cold_init): Likewise.
18281 * config/i386/cygming.h (ASM_DECLARE_COLD_FUNCTION_NAME): New macro.
18282 (ASM_DECLARE_COLD_FUNCTION_SIZE): Likewise.
18283 * config/i386/i386.c (x86_expand_epilogue): Fix wording in comment.
18284 (ix86_output_call_insn): Emit a nop in one more case for SEH.
18285 * config/i386/winnt.c: Include except.h.
18286 (struct seh_frame_state): Add reg_offset, after_prologue and
18287 in_cold_section fields.
18288 (i386_pe_seh_end_prologue): Set seh->after_prologue.
18289 (i386_pe_seh_cold_init): New function.
18290 (i386_pe_seh_fini): Add COLD parameter and bail out if it is not equal
18291 to seh->in_cold_section.
18292 (seh_emit_push): Record the offset of the push.
18293 (seh_emit_save): Record the offet of the save.
18294 (i386_pe_seh_unwind_emit): Deal with NOTE_INSN_SWITCH_TEXT_SECTIONS.
18295 Test seh->after_prologue to disregard the epilogue.
18296 (i386_pe_end_function): Pass FALSE to i386_pe_seh_fini.
18297 (i386_pe_end_cold_function): New function.
18299 2018-03-07 Jakub Jelinek <jakub@redhat.com>
18302 * config/aarch64/predicates.md (aarch64_simd_reg_or_zero): Use
18303 aarch64_simd_or_scalar_imm_zero rather than aarch64_simd_imm_zero.
18306 * gimple-expr.c (create_tmp_var_raw): Set DECL_NAMELESS flag
18308 * tree-pretty-print.c (dump_decl_name): For TDF_COMPARE_DEBUG,
18309 don't print names of DECL_NAMELESS DECL_IGNORED_P decls.
18311 PR middle-end/84723
18312 * multiple_target.c: Include tree-inline.h and intl.h.
18313 (expand_target_clones): Diagnose and fail if node->definition and
18314 !tree_versionable_function_p (node->decl).
18316 2018-03-06 John David Anglin <danglin@gcc.gnu.org>
18318 * config/pa/pa.h (ASM_GENERATE_INTERNAL_LABEL): Revise to use
18320 (ASM_OUTPUT_ADDR_VEC_ELT): Revise for above change.
18321 (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
18322 * config/pa/pa64-hpux.h (ASM_GENERATE_INTERNAL_LABEL): Revise as above.
18324 2018-03-06 Jakub Jelinek <jakub@redhat.com>
18327 * combine.c (try_combine): Use reg_or_subregno instead of handling
18328 just paradoxical SUBREGs and REGs.
18330 2018-03-06 Claudiu Zissulescu <claziss@synopsys.com>
18332 * config/arc/arc.c (arc_finalize_pic): Remove function.
18333 (arc_must_save_register): We use single base PIC register, remove
18334 checks to save/restore the PIC register.
18335 (arc_expand_prologue): Likewise.
18336 * config/arc/arc-protos.h (arc_set_default_type_attributes):
18338 (arc_verify_short): Likewise.
18339 (arc_attr_type): Likewise.
18340 * config/arc/arc.c (arc_set_default_type_attributes): Remove.
18341 (walk_stores): Likewise.
18342 (arc_address_cost): Make it static.
18343 (arc_verify_short): Likewise.
18344 (branch_dest): Likewise.
18345 (arc_attr_type): Likewise.
18346 * config/arc/arc.c (TARGET_ADJUST_INSN_LENGTH): Remove.
18347 (TARGET_INSN_LENGTH_PARAMETERS): Likewise.
18348 (arc_final_prescan_insn): Remove inserting the nops due to
18349 hardware hazards. It is done in reorg step.
18350 (insn_length_variant_t): Remove.
18351 (insn_length_parameters_t): Likewise.
18352 (arc_insn_length_parameters): Likewise.
18353 (arc_get_insn_variants): Likewise.
18354 * config/arc/arc.h (TARGET_UPSIZE_DBR): Remove.
18356 2018-03-06 Jakub Jelinek <jakub@redhat.com>
18358 PR inline-asm/84683
18359 * reg-stack.c (move_for_stack_reg): If any_malformed_asm, avoid
18362 PR tree-optimization/84687
18363 * omp-simd-clone.c (simd_clone_create): Clear DECL_BUILT_IN_CLASS
18365 * match.pd (pow(C,x)*expN(y) -> expN(logN(C)*x+y)): New optimization.
18367 2018-03-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18369 * config/rs6000/rs6000-builtin.def (rs6000_speculation_barrier):
18370 Rename to ppc_speculation_barrier.
18371 * config/rs6000/rs6000.c (rs6000_init_builtins): Rename builtin to
18372 __builtin_ppc_speculation_barrier.
18374 2018-03-05 Jakub Jelinek <jakub@redhat.com>
18377 * combine.c (combine_simplify_rtx): Don't try to simplify if
18378 if_then_else_cond returned non-NULL, but either true_rtx or false_rtx
18381 2018-03-05 Segher Boessenkool <segher@kernel.crashing.org>
18383 * config/rs6000/rs6000.c (rs6000_loop_align): Don't align tiny loops
18384 to 32 bytes when compiling for POWER9.
18386 2018-03-05 Jakub Jelinek <jakub@redhat.com>
18389 * config/i386/i386.c (ix86_function_ok_for_sibcall): Check for
18390 regparm >= 3 with no arg reg available also for calls with
18391 flag_force_indirect_call. Pass decl to ix86_function_regparm.
18394 * config/i386/sse.md (*<code><mode>3): Replace <mask_prefix3> with
18396 (*<plusminus_insn><mode>3): Likewise. Remove <mask_operand3> uses.
18398 2018-03-05 Peter Bergner <bergner@vnet.ibm.com>
18401 * config/rs6000/vector.md (mov<mode>): Disallow altivec memory operands.
18403 2018-03-05 Richard Biener <rguenther@suse.de>
18405 PR tree-optimization/84486
18406 * tree-ssa-pre.c (create_expression_by_pieces): Remove dead code.
18407 When inserting a __builtin_assume_aligned call set the LHS
18408 SSA name alignment info accordingly.
18410 2018-03-05 Wilco Dijkstra <wdijkstr@arm.com>
18412 PR tree-optimization/84114
18413 * config/aarch64/aarch64.c (aarch64_reassociation_width)
18414 Avoid reassociation of FLOAT_MODE addition.
18416 2018-03-05 Olga Makhotina <olga.makhotina@intel.com>
18418 * common/config/i386/i386-common.c (OPTION_MASK_ISA_PCONFIG_SET,
18419 OPTION_MASK_ISA_PCONFIG_UNSET, OPTION_MASK_ISA_WBNOINVD_SET,
18420 OPTION_MASK_ISA_WBNOINVD_UNSET): New definitions.
18421 (ix86_handle_option): Handle -mpconfig and -mwbnoinvd.
18422 * config.gcc (pconfigintrin.h, wbnoinvdintrin.h) : Add headers.
18423 * config/i386/cpuid.h (bit_PCONFIG, bit_WBNOINVD): New.
18424 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mpconfig
18426 * config/i386/i386-builtin.def (__builtin_ia32_wbnoinvd,
18427 __builtin_ia32_wbinvd): New builtins.
18428 (SPECIAL_ARGS2): New.
18429 * config/i386/i386-c.c (__WBNOINVD__, __PCONFIG__): New.
18430 (SPECIAL_ARGS2): New.
18431 * config/i386/i386.c (ix86_target_string): Add -mpconfig and -mwbnoinvd.
18432 (ix86_valid_target_attribute_inner_p): Ditto.
18433 (ix86_init_mmx_sse_builtins): Add special_args2.
18434 * config/i386/i386.h (TARGET_PCONFIG, TARGET_PCONFIG_P, TARGET_WBNOINVD,
18435 TARGET_WBNOINVD_P): New.
18436 * config/i386/i386.md (UNSPECV_WBINVD, UNSPECV_WBNOINVD): New.
18437 (define_insn "wbinvd", define_insn "wbnoinvd"): New.
18438 * config/i386/i386.opt: Add -mpconfig and -mwbnoinvd.
18439 * config/i386/immintrin.h (_wbinvd): New intrinsic.
18440 * config/i386/pconfigintrin.h: New file.
18441 * config/i386/wbnoinvdintrin.h: Ditto.
18442 * config/i386/x86intrin.h: Add headers pconfigintrin.h and
18444 * doc/invoke.texi (-mpconfig, -mwbnoinvd): New.
18446 2018-03-05 Richard Biener <rguenther@suse.de>
18448 PR tree-optimization/84670
18449 * tree-ssa-pre.c (struct bb_bitmap_sets): Add visited_with_visited_succs
18451 (BB_VISITED_WITH_VISITED_SUCCS): New define.
18452 (compute_antic): Initialize BB_VISITED_WITH_VISITED_SUCCS.
18453 (compute_antic_aux): Only assert the number of values in ANTIC_IN
18454 doesn't grow if all successors (recursively) were visited at least
18457 2018-03-05 Richard Biener <rguenther@suse.de>
18459 PR tree-optimization/84650
18460 * tree-ssa-loop-im.c (pass_lim::execute): Reset the SCEV cache
18461 if executed in the loop pipeline.
18463 2018-03-05 Sandra Loosemore <sandra@codesourcery.com>
18465 * doc/configfiles.texi (Configuration Files): Move info about
18466 conditionalizing $target-protos.h to...
18467 * doc/sourcebuild.texi (Back End): Here. Explain how $target.h
18468 differs from $target-protos.h.
18470 2018-03-05 Kito Cheng <kito.cheng@gmail.com>
18471 Chung-Ju Wu <jasonwucj@gmail.com>
18473 * config/nds32/nds32-protos.h (nds32_expand_setmem): Declare.
18474 * config/nds32/nds32-multiple.md (setmemsi): Define.
18475 * config/nds32/nds32-memory-manipulation.c
18476 (nds32_gen_dup_4_byte_to_word_value): New.
18477 (emit_setmem_word_loop): New.
18478 (emit_setmem_byte_loop): New.
18479 (nds32_expand_setmem_loop): New.
18480 (nds32_expand_setmem_loop_v3m): New.
18481 (nds32_expand_setmem_unroll): New.
18482 (nds32_expand_setmem): New.
18484 2018-03-04 Kito Cheng <kito.cheng@gmail.com>
18485 Chung-Ju Wu <jasonwucj@gmail.com>
18487 * config/nds32/nds32-memory-manipulation.c
18488 (nds32_emit_load_store): New.
18489 (nds32_emit_post_inc_load_store): New.
18490 (nds32_emit_mem_move): New.
18491 (nds32_emit_mem_move_block): New.
18492 (nds32_expand_movmemsi_loop_unknown_size): New.
18493 (nds32_expand_movmemsi_loop_known_size): New.
18494 (nds32_expand_movmemsi_loop): New.
18495 (nds32_expand_movmemsi_unroll): New.
18496 (nds32_expand_movmemqi): Rename ...
18497 (nds32_expand_movmemsi): ... to this.
18498 * config/nds32/nds32-multiple.md (movmemqi): Rename ...
18499 (movmemsi): ... to this.
18500 * config/nds32/nds32-protos.h (nds32_expand_movmemqi): Rename ...
18501 (nds32_expand_movmemsi): ... to this.
18503 2018-03-04 Kito Cheng <kito.cheng@gmail.com>
18504 Monk Chiang <sh.chiang04@gmail.com>
18505 Chung-Ju Wu <jasonwucj@gmail.com>
18507 * config/nds32/nds32-protos.h
18508 (nds32_expand_load_multiple): New arguments.
18509 (nds32_expand_store_multiple): Ditto.
18510 (nds32_valid_multiple_load_store): Rename ...
18511 (nds32_valid_multiple_load_store_p): ... to this.
18512 * config/nds32/nds32-memory-manipulation.c
18513 (nds32_expand_load_multiple): Refine implementation.
18514 (nds32_expand_store_multiple): Ditto.
18515 * config/nds32/nds32-multiple.md
18516 (load_multiple): Update nds32_expand_load_multiple interface.
18517 (store_multiple): Update nds32_expand_store_multiple interface.
18518 * config/nds32/nds32-predicates.c
18519 (nds32_valid_multiple_load_store): Rename ...
18520 (nds32_valid_multiple_load_store_p): ... to this and refine
18522 * config/nds32/predicates.md
18523 (nds32_load_multiple_and_update_address_operation): New predicate.
18524 (nds32_store_multiple_and_update_address_operation): New predicate.
18526 2018-03-04 Kito Cheng <kito.cheng@gmail.com>
18527 Chung-Ju Wu <jasonwucj@gmail.com>
18529 * config/nds32/nds32.md (type): Add load_multiple and store_multiple.
18530 (combo): New attribute.
18531 * config/nds32/nds32-multiple.md: Refine patterns with new attributes.
18533 2018-03-03 Chung-Ju Wu <jasonwucj@gmail.com>
18535 * config/nds32/nds32.opt: Change -mcmodel= default value.
18537 2018-03-03 Kito Cheng <kito.cheng@gmail.com>
18538 Monk Chiang <sh.chiang04@gmail.com>
18539 Chung-Ju Wu <jasonwucj@gmail.com>
18541 * config/nds32/constants.md (unspec_element): New enum.
18542 * config/nds32/constraints.md (Umw): New constraint.
18543 * config/nds32/nds32-intrinsic.c: Add more builtin functions.
18544 * config/nds32/nds32-intrinsic.md: Likewise.
18545 * config/nds32/nds32-md-auxiliary.c (nds32_regno_to_enable4): New.
18546 (nds32_valid_smw_lwm_base_p): New.
18547 (nds32_output_smw_single_word): New.
18548 (nds32_output_lmw_single_word): New.
18549 (nds32_expand_unaligned_load): New.
18550 (nds32_expand_unaligned_store): New.
18551 * config/nds32/nds32-protos.h (nds32_valid_smw_lwm_base_p): Declare.
18552 (nds32_output_smw_single_word): Declare.
18553 (nds32_output_lmw_single_word): Declare.
18554 (nds32_expand_unaligned_load): Declare.
18555 (nds32_expand_unaligned_store): Declare.
18556 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_UALOAD_HW,
18557 NDS32_BUILTIN_UALOAD_W, NDS32_BUILTIN_UALOAD_DW,
18558 NDS32_BUILTIN_UASTORE_HW, NDS32_BUILTIN_UASTORE_W,
18559 NDS32_BUILTIN_UASTORE_DW.
18560 * config/nds32/predicates.md (nds32_lmw_smw_base_operand): New
18563 2018-03-03 Monk Chiang <sh.chiang04@gmail.com>
18564 Kito Cheng <kito.cheng@gmail.com>
18565 Chung-Ju Wu <jasonwucj@gmail.com>
18567 * config/nds32/nds32-intrinsic.c
18568 (nds32_expand_builtin_null_ftype_reg): Delete.
18569 (nds32_expand_builtin_reg_ftype_imm): Ditto.
18570 (nds32_expand_builtin_null_ftype_reg_imm): Ditto.
18571 (nds32_read_argument): New.
18572 (nds32_legitimize_target): Ditto.
18573 (nds32_legitimize_argument): Ditto.
18574 (nds32_check_constant_argument): Ditto.
18575 (nds32_expand_unop_builtin): Ditto.
18576 (nds32_expand_unopimm_builtin): Ditto.
18577 (nds32_expand_binop_builtin): Ditto.
18578 (nds32_builtin_decl_impl): Ditto.
18579 (builtin_description): Ditto.
18580 (nds32_expand_builtin_impl): Rewrite with new infrastructure.
18581 (nds32_init_builtins_impl): Ditto.
18582 * config/nds32/nds32.c (TARGET_BUILTIN_DECL): Define.
18583 (nds32_builtin_decl): New.
18584 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_COUNT.
18585 * config/nds32/nds32-protos.h (nds32_builtin_decl_impl): Declare.
18587 2018-03-02 Jeff Law <law@redhat.com>
18589 * reorg.c (stop_search_p): Handle DEBUG_INSN.
18590 (redundant_insn, fill_simple_delay_slots): Likewise.
18591 (fill_slots_from_thread): Likewise.
18592 * resource.c (mark_referenced_resources): Likewise.
18593 (mark_set_resources, find_dead_or_set_registers): Likewise.
18595 2018-03-02 Jakub Jelinek <jakub@redhat.com>
18597 * substring-locations.h (format_warning_va): Formatting fix for
18598 ATTRIBUTE_GCC_DIAG.
18599 (format_warning_at_substring): Fix up ATTRIBUTE_GCC_DIAG second
18601 (format_warning_n_va, format_warning_at_substring_n): New prototypes.
18602 * substring-locations.c: Include intl.h.
18603 (format_warning_va): Turned into small wrapper around
18604 format_warning_n_va, renamed to ...
18605 (format_warning_n_va): ... this, add N and PLURAL_GMSGID arguments,
18606 rename GMSGID to SINGULAR_GMSGID, if SINGULAR_GMSGID != PLURAL_GMSGID,
18608 (format_warning_at_substring_n): New function.
18609 * gimple-ssa-sprintf.c: Remove GCC diagnostic ignored pragma.
18610 (fmtwarn): Add ATTRIBUTE_GCC_DIAG. Turn into a copy of
18611 format_warning_at_substring with just a shorter name instead of
18612 const function pointer.
18613 (fmtwarn_n): New function.
18614 (maybe_warn, format_directive, parse_directive): Use fmtwarn_n where
18615 appropriate, get rid of all the fmtstr temporaries, move conditionals
18616 with G_() wrapped string literals directly into fmtwarn arguments,
18617 cast dir.len to (int), formatting fixes.
18619 2018-03-02 Thomas Schwinge <thomas@codesourcery.com>
18621 * doc/invoke.texi: Remove "Cilk Plus" references.
18623 2018-03-02 Jakub Jelinek <jakub@redhat.com>
18624 Richard Biener <rguenther@suse.de>
18627 * expr.c (expand_expr_real_1) <case CALL_EXPR>: Don't emit diagnostics
18628 for error or warning attributes if CALL_FROM_THUNK_P is set.
18631 2018-03-02 Jakub Jelinek <jakub@redhat.com>
18634 * config/pa/pa.h (TARGET_CPU_CPP_BUILTINS): Predefine
18635 __SIZEOF_128__ macro if HPUX_LONG_DOUBLE_LIBRARY.
18638 * config/ia64/ia64.h (TARGET_CPU_CPP_BUILTINS): Predefine
18639 __SIZEOF_{FPREG,FLOAT{80,128}}__ macros.
18641 * predict.c (test_prediction_value_range): Use PROB_UNINITIALIZED
18642 instead of -1U in last predictors element's probability member.
18644 2018-03-02 Eric Botcazou <ebotcazou@adacore.com>
18647 * ipa-devirt.c (odr_subtypes_equivalent_p): Get the ODR type of both
18648 arguments if they are comparable.
18650 2018-03-02 Richard Sandiford <richard.sandiford@linaro.org>
18652 PR tree-optimization/84634
18653 * tree-vect-stmts.c (vectorizable_store, vectorizable_load): Replace
18654 masks and masked_loop_p with a single loop_masks, making sure it's
18655 null for bb vectorization.
18657 2018-03-02 Richard Sandiford <richard.sandiford@linaro.org>
18659 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence)
18660 (vect_analyze_data_ref_access): Use loop->safe_len rather than
18661 loop->force_vectorize to check whether there is no alias.
18663 2018-03-02 Jakub Jelinek <jakub@redhat.com>
18666 * rtl.h (prev_real_nondebug_insn, next_real_nondebug_insn): New
18668 * emit-rtl.c (next_real_insn, prev_real_insn): Fix up function
18670 (next_real_nondebug_insn, prev_real_nondebug_insn): New functions.
18671 * cfgcleanup.c (try_head_merge_bb): Use prev_real_nondebug_insn
18672 instead of a loop around prev_real_insn.
18673 * combine.c (move_deaths): Use prev_real_nondebug_insn instead of
18676 PR inline-asm/84625
18677 * config/i386/i386.c (ix86_print_operand): Use conditional
18678 output_operand_lossage instead of gcc_assert if CONST_VECTOR is not
18681 2018-03-02 Richard Biener <rguenther@suse.de>
18683 PR tree-optimization/84427
18684 * tree-ssa-pre.c (bitmap_remove_expr_from_set): Remove.
18685 (bitmap_set_subtract_values): Rewrite to handle multiple
18688 (prune_clobbered_mems): Likewise.
18689 (phi_translate): Take edge instead of pred/phiblock.
18690 (phi_translate_1): Likewise.
18691 (phi_translate_set): Likewise. Insert all translated
18692 exprs for a value into the set, keeping possibly multiple
18693 expressions per value.
18694 (compute_antic_aux): Adjust for phi_translate changes.
18695 When intersecting union the expressions and prune those
18696 not in the final value set, keeping possibly multiple
18697 expressions per value. Do not use value-insertion
18698 for unioning ANTIC_OUT U EXP_GEN - TMP_GEN but merge
18699 all expressions. Add verification that the value-sets
18700 only shrink during iteration.
18701 (compute_partial_antic_aux): Adjust for the phi_translate changes.
18702 (do_pre_regular_insertion): Likewise.
18703 (do_pre_partial_partial_insertion): Likewise.
18705 2018-03-02 Richard Biener <rguenther@suse.de>
18708 * config/darwin.c (saved_debug_info_level): New static global.
18709 (darwin_asm_lto_start): Disable debug info generation for LTO out.
18710 (darwin_asm_lto_end): Restore debug info generation settings.
18712 2018-03-01 Martin Liska <mliska@suse.cz>
18715 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
18716 volatile arguments.
18718 2018-03-01 Richard Biener <rguenther@suse.de>
18721 * dwarf2out.c (gen_variable_die): Properly handle late VLA
18722 type annotation with LTO when debug was disabled at compile-time.
18724 2018-03-01 Matthew Fortune <mfortune@gmail.com>
18726 * config/mips/mips.c (mips_final_prescan_insn): Fix incorrect
18728 (mips_final_postscan_insn): Likewise.
18730 2018-03-01 Richard Sandiford <richard.sandiford@linaro.org>
18732 PR rtl-optimization/84528
18733 * alias.c (init_alias_target): Add commentary.
18734 (init_alias_analysis): Only give HARD_FRAME_POINTER_REGNUM
18735 a unique base value if the frame pointer is not eliminated
18736 to the stack pointer.
18738 2018-03-01 Tom de Vries <tom@codesourcery.com>
18740 PR rtl-optimization/83327
18741 * lra-int.h (hard_regs_spilled_into): Declare.
18742 * lra.c (hard_regs_spilled_into): Define.
18743 (init_reg_info): Init hard_regs_spilled_into.
18744 * lra-spills.c (assign_spill_hard_regs): Update hard_regs_spilled_into.
18745 * lra-lives.c (make_hard_regno_born, make_hard_regno_dead)
18746 (process_bb_lives): Handle hard_regs_spilled_into.
18747 (lra_create_live_ranges_1): Before doing liveness propagation, clear
18748 regs in all_hard_regs_bitmap if set in hard_regs_spilled_into.
18750 2018-02-28 David Edelsohn <dje.gcc@gmail.com>
18752 * config.gcc (powerpc-ibm-aix7.1.*): New stanza.
18753 (powerpc-ibm-aix[789]*): Default to AIX 7.2.
18754 * config/rs6000/aix71.h (TARGET_DEFAULT): Revert to Power4 ISA.
18755 * config/rs6000/aix72.h: New file.
18757 2018-02-28 Jakub Jelinek <jakub@redhat.com>
18759 * gimple-ssa-warn-restrict.c (maybe_diag_overlap): Use warning_n
18760 instead of warning_at with conditional singular and plural messages
18764 * stor-layout.c (update_alignment_for_field): For
18765 targetm.ms_bitfield_layout_p (rli->t), if !is_bitfield
18766 && !DECL_PACKED (field), do the alignment update, just use
18767 only desired_align instead of MAX (type_align, desired_align)
18769 (place_field): Don't do known_align < desired_align handling
18770 early if targetm.ms_bitfield_layout_p (rli->t) and rli->prev_field
18771 is non-NULL, instead do it after rli->prev_field handling and
18772 only if not within a bitfield word. For DECL_PACKED (field)
18773 use type_align of BITS_PER_UNIT.
18775 2018-02-28 Eric Botcazou <ebotcazou@adacore.com>
18777 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Remove
18778 superfluous parentheses and trailing spaces.
18780 2018-02-28 Richard Biener <rguenther@suse.de>
18782 PR tree-optimization/84584
18783 * graphite-scop-detection.c (scop_detection::add_scop): Discard
18784 SCoPs with fake exit edge.
18786 2018-02-28 Martin Liska <mliska@suse.cz>
18789 * timevar.c (timer::print): Fix format to properly print 100%
18792 2018-02-28 Richard Biener <rguenther@suse.de>
18794 PR middle-end/84607
18795 * genmatch.c (capture_info::walk_match): Do not mark
18796 captured expressions without operands as expr_p given
18797 they act more like predicates and should be subject to
18798 "lost tail" side-effect preserving.
18800 2018-02-28 Alexandre Oliva <aoliva@redhat.com>
18802 PR rtl-optimization/81611
18803 * auto-inc-dec.c (attempt_change): Move dead note from
18804 mem_insn if it's the next use of regno
18805 (find_address): Take address use of reg holding
18806 non-incremented value. Add parm to limit search to the named
18808 (merge_in_block): Attempt to use a mem insn that is the next
18809 use of the original regno.
18811 2018-02-27 Martin Sebor <msebor@redhat.com>
18814 * doc/invoke.texi (-Wmissing-attributes): New option.
18815 * print-tree.c (print_node): Handle DECL_UNINLINABLE.
18817 2018-02-27 Martin Sebor <msebor@redhat.com>
18819 PR translation/84207
18820 * diagnostic-core.h (warning_n, error_n, inform_n): Change
18821 n argument to unsigned HOST_WIDE_INT.
18822 * diagnostic.c (warning_n, error_n, inform_n): Ditto.
18823 (diagnostic_n_impl): Ditto. Handle arguments in excess of LONG_MAX.
18824 * gimple-ssa-sprintf.c (format_directive): Simplify inform_n call.
18825 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Use warning_n.
18827 2018-02-27 Richard Biener <rguenther@suse.de>
18829 PR tree-optimization/84512
18830 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost):
18831 Do not use the estimate returned from record_stmt_cost for
18832 the scalar iteration cost but sum properly using add_stmt_cost.
18834 2018-02-27 Richard Biener <rguenther@suse.de>
18836 PR tree-optimization/84466
18837 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
18838 Adjust last change to less strictly validate use operands.
18840 2018-02-27 Martin Liska <mliska@suse.cz>
18842 PR gcov-profile/84548
18843 * gcov.c (process_file): Allow partial overlap and consider it
18844 also as group functions.
18845 (output_lines): Properly calculate range of lines for a group.
18847 2018-02-27 Martin Liska <mliska@suse.cz>
18849 * timevar.c (timer::print_row): Remove 'usr', 'sys', 'wall' and
18850 'ggc' suffixes. Change first column width.
18851 (timer::print): Fix formatting of the column.
18853 2018-02-27 Alexandre Oliva <aoliva@redhat.com>
18855 * tree-ssa-live.c (remove_unused_scope_block_p): Do not
18856 preserve inline entry blocks for the sake of debug inline
18857 entry point markers alone.
18858 (remove_unused_locals): Suggest in comments a better place to
18859 force the preservation of inline entry blocks that are
18860 otherwise unused, but do not preserve them.
18862 2018-02-26 H.J. Lu <hongjiu.lu@intel.com>
18864 * config/i386/i386.c (ix86_output_indirect_jmp): Update comments.
18866 2018-02-26 H.J. Lu <hongjiu.lu@intel.com>
18869 * config/i386/constraints.md (Bs): Replace
18870 ix86_indirect_branch_register with
18871 TARGET_INDIRECT_BRANCH_REGISTER.
18873 * config/i386/i386.md (indirect_jump): Likewise.
18874 (tablejump): Likewise.
18875 (*sibcall_memory): Likewise.
18876 (*sibcall_value_memory): Likewise.
18877 Peepholes of indirect call and jump via memory: Likewise.
18878 (*sibcall_GOT_32): Disallowed for TARGET_INDIRECT_BRANCH_REGISTER.
18879 (*sibcall_value_GOT_32): Likewise.
18880 * config/i386/predicates.md (indirect_branch_operand): Likewise.
18881 (GOT_memory_operand): Likewise.
18882 (call_insn_operand): Likewise.
18883 (sibcall_insn_operand): Likewise.
18884 (GOT32_symbol_operand): Likewise.
18885 * config/i386/i386.h (TARGET_INDIRECT_BRANCH_REGISTER): New.
18887 2018-02-26 Eric Botcazou <ebotcazou@adacore.com>
18889 PR rtl-optimization/83496
18890 * reorg.c (steal_delay_list_from_target): Change REDUNDANT array from
18891 booleans to RTXes. Call fix_reg_dead_note on every non-null element.
18892 (steal_delay_list_from_fallthrough): Call fix_reg_dead_note on a
18893 redundant insn, if any.
18894 (relax_delay_slots): Likewise.
18895 (update_reg_unused_notes): Rename REDUNDANT_INSN to OTHER_INSN.
18897 2018-02-26 Richard Sandiford <richard.sandiford@linaro.org>
18899 PR tree-optimization/83965
18900 * tree-vect-patterns.c (vect_reassociating_reduction_p): Assume
18901 that grouped statements are part of a reduction chain. Return
18902 true if the statement is not marked as a reduction itself but
18903 is part of a group.
18904 (vect_recog_dot_prod_pattern): Don't check whether the statement
18905 is part of a group here.
18906 (vect_recog_sad_pattern): Likewise.
18907 (vect_recog_widen_sum_pattern): Likewise.
18909 2018-02-26 Eric Botcazou <ebotcazou@adacore.com>
18912 * final.c (rest_of_clean_state): Also look for calls inside sequences.
18914 2018-02-26 H.J. Lu <hongjiu.lu@intel.com>
18917 * config/i386/i386-protos.h (ix86_output_indirect_jmp): Remove
18919 (ix86_output_indirect_function_return): New prototype.
18920 (ix86_split_simple_return_pop_internal): Likewise.
18921 * config/i386/i386.c (indirect_return_via_cx): New.
18922 (indirect_return_via_cx_bnd): Likewise.
18923 (indirect_thunk_name): Handle return va CX_REG.
18924 (output_indirect_thunk_function): Create alias for
18925 __x86_return_thunk_[re]cx and __x86_return_thunk_[re]cx_bnd.
18926 (ix86_output_indirect_jmp): Remove the bool argument.
18927 (ix86_output_indirect_function_return): New function.
18928 (ix86_split_simple_return_pop_internal): Likewise.
18929 * config/i386/i386.md (*indirect_jump): Don't pass false
18930 to ix86_output_indirect_jmp.
18931 (*tablejump_1): Likewise.
18932 (simple_return_pop_internal): Change it to define_insn_and_split.
18933 Call ix86_split_simple_return_pop_internal to split it for
18934 -mfunction-return=.
18935 (simple_return_indirect_internal): Call
18936 ix86_output_indirect_function_return instead of
18937 ix86_output_indirect_jmp.
18939 2018-02-26 Jakub Jelinek <jakub@redhat.com>
18942 * vec.h (vec_default_construct): For BROKEN_VALUE_INITIALIZATION use
18943 memset and value initialization afterwards.
18945 2018-02-26 Christophe Lyon <christophe.lyon@linaro.org>
18947 * Makefile.in (lto-wrapper): Use ALL_LINKERFLAGS.
18949 2018-02-26 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
18952 * common/config/aarch64/aarch64-common.c
18953 (aarch_option_optimization_table[]): Switch
18954 off fomit-frame-pointer
18956 2018-02-26 Kito Cheng <kito.cheng@gmail.com>
18957 Chung-Ju Wu <jasonwucj@gmail.com>
18959 * config/nds32/nds32-multiple.md (load_multiple): Disallow
18961 (store_multiple): Ditto.
18963 2018-02-26 Kito Cheng <kito.cheng@gmail.com>
18965 * config.gcc: Add --with-cpu support for nds32 target.
18966 * config/nds32/nds32-opts.h (nds32_cpu_type): New.
18967 * config/nds32/nds32.opt: Add -mcpu= option.
18969 2018-02-25 Segher Boessenkool <segher@kernel.crashing.org>
18971 * config/rs6000/rs6000.opt (mvrsave=no, mvrsave=yes, isel=no,
18972 isel=yes): Warn for these deprecated options.
18974 2018-02-23 David Edelsohn <dje.gcc@gmail.com>
18976 * config/rs6000/aix71.h (TARGET_DEFAULT): Change to
18977 ISA_2_5_MASKS_EMBEDDED.
18979 2018-02-23 Jakub Jelinek <jakub@redhat.com>
18981 * ipa-prop.c (ipa_vr_ggc_hash_traits::hash): Hash p->min and
18982 p->max as pointers rather than using iterative_hash_expr.
18984 2018-02-23 Carl Love <cel@us.ibm.com>
18986 * config/rs6000/rs6000-builtin.def: Change VSIGNED2 and VUNSIGNED2
18987 macro expansions from BU_VSX_2 to BU_P8V_VSX_2 and BU_VSX_OVERLOAD_2 to
18989 * config/rs6000/rs6000-c.c: Change VSX_BUILTIN_VEC_VSIGNED2 to
18990 P8V_BUILTIN_VEC_VSIGNED2. Change VSX_BUILTIN_VEC_VUNSIGNED2 to
18991 P8V_BUILTIN_VEC_VUNSIGNED2.
18993 2018-02-22 Vladimir Makarov <vmakarov@redhat.com>
18996 * lra-int.h (LRA_UNKNOWN_ALT, LRA_NON_CLOBBERED_ALT): New macros.
18997 * lra.c (lra_set_insn_recog_data, lra_update_insn_recog_data): Use
18999 * lra-constraints.c (curr_insn_transform): Set up
19000 LRA_NON_CLOBBERED_ALT for moves processed on the fast path. Use
19002 (remove_inheritance_pseudos): Use LRA_UNKNOWN_ALT.
19003 * lra-eliminations.c (spill_pseudos): Ditto.
19004 (process_insn_for_elimination): Ditto.
19005 * lra-lives.c (reg_early_clobber_p): Use the new macros.
19006 * lra-spills.c (spill_pseudos): Use LRA_UNKNOWN_ALT and
19007 LRA_NON_CLOBBERED_ALT.
19009 2018-02-22 Martin Sebor <msebor@redhat.com>
19011 PR tree-optimization/84480
19012 * gimple-fold.c (gimple_fold_builtin_strcpy): Move warnings
19013 to maybe_diag_stxncpy_trunc. Call it.
19014 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Integrate warnings
19015 from gimple_fold_builtin_strcpy. Print inlining stack.
19016 (handle_builtin_stxncpy): Print inlining stack.
19017 * tree-ssa-strlen.h (maybe_diag_stxncpy_trunc): Declare.
19019 2018-02-22 H.J. Lu <hongjiu.lu@intel.com>
19022 * config/i386/i386.c (ix86_set_indirect_branch_type): Issue an
19023 error when -mindirect-branch=thunk-extern, -fcf-protection=branch
19024 and -fcheck-pointer-bounds are used together.
19025 (indirect_thunk_prefix): New enum.
19026 (indirect_thunk_need_prefix): New function.
19027 (indirect_thunk_name): Replace need_bnd_p with need_prefix. Use
19028 "_nt" instead of "_bnd" for NOTRACK prefix.
19029 (output_indirect_thunk): Replace need_bnd_p with need_prefix.
19030 (output_indirect_thunk_function): Likewise.
19032 (ix86_code_end): Update output_indirect_thunk_function calls.
19033 (ix86_output_indirect_branch_via_reg): Replace
19034 ix86_bnd_prefixed_insn_p with indirect_thunk_need_prefix.
19035 (ix86_output_indirect_branch_via_push): Likewise.
19036 (ix86_output_function_return): Likewise.
19037 * doc/invoke.texi: Document -mindirect-branch=thunk-extern is
19038 incompatible with -fcf-protection=branch and
19039 -fcheck-pointer-bounds.
19041 2018-02-22 Steve Ellcey <sellcey@cavium.com>
19044 * config/aarch64/aarch64.c (aarch64_print_address_internal):
19045 Change gcc_assert call to output_operand_lossage.
19047 2018-02-22 Steve Ellcey <sellcey@cavium.com>
19049 * doc/extend.texi (__builtin_extend_pointer): Document builtin.
19051 2018-02-22 DJ Delorie <dj@redhat.com>
19052 Sebastian Perta <sebastian.perta@renesas.com>
19053 Oleg Endo <olegendo@gcc.gnu.org>
19055 * config/rx/rx.c (rx_rtx_costs): New function.
19056 (TARGET_RTX_COSTS): Override to use rx_rtx_costs.
19058 2018-02-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
19060 * config/arm/t-multilib: Map Armv8-R to Armv7 multilibs.
19062 2018-02-22 Martin Liska <mliska@suse.cz>
19065 * common/config/arm/arm-common.c (arm_print_hint_for_cpu_option):
19066 Add "native" as a possible value.
19068 2018-02-22 Martin Liska <mliska@suse.cz>
19071 * config/i386/i386.c (ix86_option_override_internal):
19072 Add "native" as a possible value for -march and -mtune.
19074 2018-02-22 Jakub Jelinek <jakub@redhat.com>
19077 * stor-layout.c (finalize_type_size): Propagate TYPE_EMPTY_P flag
19078 to all type variants.
19080 PR tree-optimization/84503
19081 * gimple-ssa-store-merging.c (merged_store_group::merge_into): Compute
19082 width as info->bitpos + info->bitsize - start.
19083 (merged_store_group::merge_overlapping): Simplify width computation.
19084 (check_no_overlap): New function.
19085 (imm_store_chain_info::try_coalesce_bswap): Compute expected
19086 start + width and last_order of the group, fail if check_no_overlap
19088 (imm_store_chain_info::coalesce_immediate_stores): Don't merge info
19089 to group if check_no_overlap fails.
19091 2018-02-21 Segher Boessenkool <segher@kernel.crashing.org>
19093 * config/rs6000/altivec.md: Delete contraint arguments to
19094 define_expand, define_split, and define_peephole2, and in
19095 define_insn_and_split if always unused.
19096 * config/rs6000/darwin.md: Ditto.
19097 * config/rs6000/dfp.md: Ditto.
19098 * config/rs6000/rs6000.md: Ditto.
19099 * config/rs6000/sync.md: Ditto.
19100 * config/rs6000/vector.md: Ditto.
19101 * config/rs6000/vsx.md: Ditto.
19103 2018-02-21 Segher Boessenkool <segher@kernel.crashing.org>
19105 * config/rs6000/altivec.md: Write output control strings as braced
19106 blocks instead of double-quoted strings.
19107 * config/rs6000/darwin.md: Ditto.
19108 * config/rs6000/rs6000.md: Ditto.
19109 * config/rs6000/vector.md: Ditto.
19110 * config/rs6000/vsx.md: Ditto.
19112 2018-02-21 Jason Merrill <jason@redhat.com>
19114 PR c++/84314 - ICE with templates and fastcall attribute.
19115 * attribs.c (build_type_attribute_qual_variant): Remove assert.
19117 2018-02-21 Jan Hubicka <hubicka@ucw.cz>
19119 * ipa-cp.c (determine_versionability): Fix comment typos.
19121 2018-02-21 Jan Hubicka <hubicka@ucw.cz>
19124 * ipa-cp.c (determine_versionability): Do not version functions caling
19127 2018-02-21 Martin Liska <mliska@suse.cz>
19130 * config/aarch64/aarch64.c (aarch64_print_hint_for_core_or_arch):
19131 Add "native" as a possible value.
19132 * config/aarch64/aarch64.h (HAVE_LOCAL_CPU_DETECT): Define
19133 the macro when native cpu detection is available.
19135 2018-02-21 Martin Liska <mliska@suse.cz>
19138 * common/config/arm/arm-common.c (arm_print_hint_for_arch_option):
19139 Add "native" as a possible value.
19140 * config/arm/arm.h (HAVE_LOCAL_CPU_DETECT): Define the macro
19141 when native cpu detection is available.
19143 2018-02-21 Jakub Jelinek <jakub@redhat.com>
19144 Martin Sebor <msebor@redhat.com>
19146 PR tree-optimization/84478
19147 * gimple-fold.h (get_range_strlen): Add a bool argument defaulted to
19149 * gimple-fold.c (get_range_strlen): Make minlen const and assume it
19150 can't be NULL. Change FUZZY from bool to int, for 1 add PHI/COND_EXPR
19151 support which is conservatively correct, for 2 only stay conservative
19152 for maxlen. Formatting and comment capitalization fixes. Add STRICT
19153 argument to the 2 argument get_range_strlen, adjust 6 arg
19154 get_range_strlen caller and clear minmaxlen[0] and [1] if it returned
19156 (get_maxval_strlen): Adjust 6 arg get_range_strlen caller.
19157 (gimple_fold_builtin_strlen): Pass true as last argument to
19160 2018-02-20 Martin Sebor <msebor@redhat.com>
19162 PR middle-end/84095
19163 * gimple-ssa-warn-restrict.c (builtin_memref::extend_offset_range): New.
19164 (builtin_memref::set_base_and_offset): Same. Handle inner references.
19165 (builtin_memref::builtin_memref): Factor out parts into
19166 set_base_and_offset and call it.
19168 2018-02-20 Richard Sandiford <richard.sandiford@linaro.org>
19170 PR middle-end/84406
19171 * optabs-query.c (find_widening_optab_handler_and_mode): If from_mode
19172 is a scalar_int_mode, assert that to_mode is a scalar_int_mode with
19173 greater precision. If to_mode is a MODE_PARTIAL_INT, stop the
19174 search at the associated MODE_INT.
19176 2018-02-20 Jeff Law <law@redhat.com>
19178 PR middle-end/82123
19179 PR tree-optimization/81592
19180 PR middle-end/79257
19181 * gimple-ssa-sprintf.c (format_integer): Query EVRP range analyzer
19182 for range data rather than using global data.
19183 * gimple-ssa-sprintf.c (get_int_range): Query EVRP range analyzer for
19184 range data rather than using global data.
19185 * gimple-ssa-sprintf.c (get_int_range): Accept vr_values parameter
19186 pass it to children as needed.
19187 (struct directive::fmtresult): Similarly.
19188 (struct directive::set_width): Similarly.
19189 (struct directive::set_precision): Similarly.
19190 (format_integer, format_directive, parse_directive): Similarly.
19191 (format_none): Accept unnamed vr_values parameter.
19192 (format_percent, format_floating, format_character): Similarly.
19193 (format_string, format_plain): Similarly.
19194 * gimple-ssa-sprintf.c (sprintf_dom_walker::handle_gimple_call): Query
19195 the EVRP range analyzer for range data rather than using global data.
19196 * gimple-ssa-sprintf.c: Include alloc-pool.h, vr-values.h and
19197 gimple-ssa-evrp-analyze.h
19198 (class sprintf_dom_walker): Add after_dom_children member function.
19199 Add evrp_range_analyzer member.
19200 (sprintf_dom_walker::before_dom_children): Call into the EVRP
19201 range analyzer as needed.
19202 (sprintf_dom_walker::after_dom_children): New member function.
19203 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::enter): Do nothing
19205 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
19206 (evrp_range_analyzer::pop_to_marker): Likewise.
19208 2018-02-20 Richard Sandiford <richard.sandiford@linaro.org>
19210 PR tree-optimization/84419
19211 * internal-fn.c (expand_call_mem_ref): Create a TARGET_MEM_REF
19212 with the required type if its current type is compatible but
19215 2018-02-20 Jakub Jelinek <jakub@redhat.com>
19217 PR middle-end/82004
19218 * match.pd (pow(C,x) -> exp(log(C)*x)): Delay all folding until
19219 after vectorization.
19221 2018-02-20 Martin Liska <mliska@suse.cz>
19224 * config/aarch64/aarch64.c (aarch64_print_hint_for_core_or_arch): Print
19225 possible values if we don't have a hint.
19227 2018-02-20 Martin Liska <mliska@suse.cz>
19231 * final.c (shorten_branches): Build align_tab array with one
19233 * opts.c (finish_options): Add alignment option limit check.
19234 (MAX_CODE_ALIGN): Likewise.
19235 (MAX_CODE_ALIGN_VALUE): Likewise.
19236 * doc/invoke.texi: Document maximum allowed option value for
19237 all -falign-* options.
19239 2018-02-19 Jakub Jelinek <jakub@redhat.com>
19242 * reg-notes.def (REG_CALL_ARG_LOCATION): New reg note.
19243 * insn-notes.def (NOTE_INSN_CALL_ARG_LOCATION): Remove.
19244 * var-tracking.c (emit_note_insn_var_location): Remove all references
19245 to NOTE_INSN_CALL_ARG_LOCATION.
19246 (emit_notes_in_bb): Emit arguments as REG_CALL_ARG_LOCATION note on
19247 the CALL_INSN rather than separate NOTE_INSN_CALL_ARG_LOCATION note.
19248 Use copy_rtx_if_shared.
19249 * dwarf2out.c (gen_subprogram_die): Use XEXP with 0 instead of
19250 NOTE_VAR_LOCATION on ca_loc->call_arg_loc_note.
19251 (dwarf2out_var_location): Remove handling of
19252 NOTE_INSN_CALL_ARG_LOCATION, instead handle REG_CALL_ARG_LOCATION note
19254 * final.c (final_scan_insn): Remove all references to
19255 NOTE_INSN_CALL_ARG_LOCATION.
19256 (rest_of_clean_state): Likewise. Remove REG_CALL_ARG_LOCATION notes
19257 before dumping final insns.
19258 * except.c (emit_note_eh_region_end): Remove all references to
19259 NOTE_INSN_CALL_ARG_LOCATION.
19260 * config/alpha/alpha.c (alpha_pad_function_end): Likewise.
19261 * config/c6x/c6x.c (c6x_gen_bundles): Likewise.
19262 * config/arc/arc.c (hwloop_optimize): Likewise.
19263 * config/arm/arm.c (create_fix_barrier): Likewise.
19264 * config/s390/s390.c (s390_chunkify_start): Likewise.
19265 * config/sh/sh.c (find_barrier): Likewise.
19266 * config/i386/i386.c (rest_of_insert_endbranch,
19267 ix86_seh_fixup_eh_fallthru): Likewise.
19268 * config/xtensa/xtensa.c (hwloop_optimize): Likewise.
19269 * config/iq2000/iq2000.c (final_prescan_insn): Likewise.
19270 * config/frv/frv.c (frv_function_prologue): Likewise.
19271 * emit-rtl.c (try_split): Likewise. Copy over REG_CALL_ARG_LOCATION
19273 (note_outside_basic_block_p): Remove all references to
19274 NOTE_INSN_CALL_ARG_LOCATION.
19275 * gengtype.c (adjust_field_rtx_def): Likewise.
19276 * print-rtl.c (rtx_writer::print_rtx_operand_code_0, print_insn):
19278 * jump.c (cleanup_barriers, delete_related_insns): Likewise.
19279 * cfgrtl.c (force_nonfallthru_and_redirect): Likewise.
19282 * builtins.c (builtin_mathfn_code): Don't check if CALL_EXPR_FN (t)
19285 PR tree-optimization/84452
19286 * tree-vect-patterns.c (vect_recog_pow_pattern): Don't call
19287 expand_simd_clones if targetm.simd_clone.compute_vecsize_and_simdlen
19290 2018-02-19 Martin Liska <mliska@suse.cz>
19293 * passes.def: Put pass_sancov_O0 before pass_lower_switch with -O0.
19295 2018-02-19 Martin Liska <mliska@suse.cz>
19296 Richard Sandiford <richard.sandiford@linaro.org>
19298 PR tree-optimization/82491
19299 * gimple-fold.c (get_base_constructor): Make earlier bail out
19302 2018-02-19 Carl Love <cel@us.ibm.com>
19304 * config/rs6000/rs6000-builtin.def: Change NEG macro expansions from
19305 BU_ALTIVEC_A to BU_P8V_AV_1 and BU_ALTIVEC_OVERLOAD_1 to
19307 * config/rs6000/rs6000-c.c: Change ALTIVEC_BUILTIN_VEC_NEG to
19308 P8V_BUILTIN_VEC_NEG.
19310 2018-02-19 Sebastian Perta <sebastian.perta@renesas.com>
19312 * config/rl78/rl78.md (movdf): New define expand.
19314 2018-02-19 Martin Liska <mliska@suse.cz>
19317 * doc/invoke.texi: Fix typo.
19318 * params.def (PARAM_MAX_LOOP_HEADER_INSNS): Likewise.
19320 2018-02-18 Segher Boessenkool <segher@kernel.crashing.org>
19322 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
19323 handle rs6000_single_float and rs6000_double_float specially for
19326 2018-02-16 Jeff Law <law@redhat.com>
19328 * config/rx/rx.c (add_pop_cfi_notes): New function.;
19329 (pop_regs): Use it.
19331 2018-02-16 Jakub Jelinek <jakub@redhat.com>
19334 * ipa-inline.c (inline_small_functions): Fix a typo.
19336 2018-02-16 Nathan Sidwell <nathan@acm.org>
19338 * doc/extend.texi (Backwards Compatibility): Americanize 'behaviour'.
19340 2018-02-16 Carl Love <cel@us.ibm.com>
19342 * config/rs6000/rs6000-builtin.def: Add BU_P8V_VSX_2 macro definition.
19343 Change FLOAT2 expansions from BU_VSX_2 to BU_P8V_VSX_2 and
19344 from BU_VSX_OVERLOAD_2 to BU_P8V_OVERLOAD_2.
19345 * config/rs6000/rs6000-c.c: Changed macro VSX_BUILTIN_VEC_FLOAT2
19346 expansion to P8V_BUILTIN_VEC_FLOAT2.
19348 2018-02-16 Vladimir Makarov <vmakarov@redhat.com>
19350 PR rtl-optimization/70023
19351 * lra-constraints.c (inherit_in_ebb): Take hard reg mode of
19352 src_regno into account.
19354 2018-02-16 Carl Love <cel@us.ibm.com>
19356 * config/rs6000/altivec.h: Remove vec_vextract4b and vec_vinsert4b.
19357 * config/rs6000/rs6000-builtin.def: Remove macro expansion for
19358 VEXTRACT4B, VINSERT4B, VINSERT4B_DI and VEXTRACT4B.
19359 * config/rs6000/rs6000.c: Remove case statements for
19360 P9V_BUILTIN_VEXTRACT4B, P9V_BUILTIN_VEC_VEXTRACT4B,
19361 P9V_BUILTIN_VINSERT4B, P9V_BUILTIN_VINSERT4B_DI,
19362 and P9V_BUILTIN_VEC_VINSERT4B.
19363 * config/rs6000/rs6000-c.c (altivec_expand_builtin): Remove entries for
19364 P9V_BUILTIN_VEC_VEXTRACT4B and P9V_BUILTIN_VEC_VINSERT4B.
19365 * config/rs6000/vsx.md:
19366 * doc/extend.texi: Remove vec_vextract4b, non ABI definitions for
19369 2018-02-16 Carl Love <cel@us.ibm.com>
19371 * config/rs6000/altivec.h: Add builtin names vec_extract4b
19373 * config/rs6000/rs6000-builtin.def: Add INSERT4B and EXTRACT4B
19375 * config/rs6000/rs6000-c.c: Add the definitions for
19376 P9V_BUILTIN_VEC_EXTRACT4B and P9V_BUILTIN_VEC_INSERT4B.
19377 * config/rs6000/rs6000.c (altivec_expand_builtin): Add
19378 P9V_BUILTIN_EXTRACT4B and P9V_BUILTIN_INSERT4B case statements.
19379 * config/rs6000/vsx.md: Add define_insn extract4b. Add define_expand
19380 definition for insert4b and define insn *insert3b_internal.
19381 * doc/extend.texi: Add documentation for vec_extract4b.
19383 2018-02-16 Nathan Sidwell <nathan@acm.org>
19385 * doc/extend.texi (Backwards Compatibility): Mention friend
19386 injection. Note for-scope is deprecated.
19387 * doc/invoke.texi (-ffriend-injection): Deprecate.
19389 2018-02-16 Segher Boessenkool <segher@kernel.crashing.org>
19391 * combine.c (try_combine): When adjusting LOG_LINKS for the destination
19392 that moved to I2, also allow destinations that are a paradoxical
19393 subreg (instead of a normal reg).
19395 2018-02-16 Oleg Endo <olegendo@gcc.gnu.org>
19398 * config/rx/rx.c (rx_fuse_in_memory_bitop): Convert shift operand
19401 2018-02-16 Richard Biener <rguenther@suse.de>
19403 PR tree-optimization/84037
19404 PR tree-optimization/84016
19406 * config/i386/i386.c (ix86_builtin_vectorization_cost):
19407 Adjust vec_construct for the fact we need additional higher latency
19408 128bit inserts for AVX256 and AVX512 vector builds.
19409 (ix86_add_stmt_cost): Scale vector construction cost for
19412 2018-02-16 Richard Biener <rguenther@suse.de>
19414 PR tree-optimization/84417
19415 * tree-ssa.c (non_rewritable_mem_ref_base): Properly constrain
19416 the MEM_REF offset when conversion to BIT_FIELD_REF is desired.
19417 (non_rewritable_lvalue_p): Likewise, use poly-ints.
19419 2018-02-16 Martin Liska <mliska@suse.cz>
19422 * internal-fn.def (ASAN_CHECK): Set proper flags.
19423 (ASAN_MARK): Likewise.
19425 2018-02-16 Julia Koval <julia.koval@intel.com>
19427 * config/i386/i386.c (ix86_option_override_internal): Remove PTA_CLWB
19428 from PTA_CANNONLAKE.
19430 2018-02-16 Jakub Jelinek <jakub@redhat.com>
19433 * config/aarch64/cortex-a57-fma-steering.c (fma_forest::merge_forest):
19434 Use ++iter rather than iter++ for std::list iterators.
19435 (func_fma_steering::dfs): Likewise. Don't delete nodes right away,
19436 defer deleting them until all nodes in the forest are processed. Do
19437 free even leaf nodes. Change to_process into auto_vec.
19440 * system.h (BROKEN_VALUE_INITIALIZATION): Define for GCC < 4.3.
19441 * vec.h (vec_default_construct): Use memset instead of placement new
19442 if BROKEN_VALUE_INITIALIZATION is defined.
19443 * hash-table.h (hash_table<Descriptor, Allocator>::empty_slow): Use
19444 memset instead of value initialization if BROKEN_VALUE_INITIALIZATION
19447 PR rtl-optimization/83723
19448 * lra-int.h (lra_substitute_pseudo): Add DEBUG_P argument.
19449 * lra.c (lra_substitute_pseudo): Likewise. If true, use
19450 gen_rtx_raw_SUBREG instead of gen_rtx_SUBREG. Pass DEBUG_P to
19452 (lra_substitute_pseudo_within_insn): Adjust lra_substitute_pseudo
19454 * lra-constraints.c (inherit_reload_reg, split_reg): Likewise.
19456 2018-02-16 Eric Botcazou <ebotcazou@adacore.com>
19458 PR rtl-optimization/81443
19459 * rtlanal.c (num_sign_bit_copies1) <SUBREG>: Do not propagate results
19460 from inner REGs to paradoxical SUBREGs.
19462 2018-02-16 Richard Biener <rguenther@suse.de>
19464 PR tree-optimization/84399
19465 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
19466 For operands we can analyze at their definition make sure we can
19467 analyze them at each use as well.
19469 2018-02-16 Richard Biener <rguenther@suse.de>
19471 PR tree-optimization/84190
19472 * tree-ssa.c (non_rewritable_mem_ref_base): Do not touch
19473 volatile accesses if the decl isn't volatile.
19475 2018-02-15 Jason Merrill <jason@redhat.com>
19477 PR c++/84314 - ICE with templates and fastcall attribute.
19478 * attribs.c (build_type_attribute_qual_variant): Don't clobber
19479 TYPE_CANONICAL on an existing type.
19481 2018-02-15 Jakub Jelinek <jakub@redhat.com>
19483 PR tree-optimization/84383
19484 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Don't look at
19485 dstoff nor call operand_equal_p if dstbase is NULL.
19487 PR tree-optimization/84334
19488 * match.pd ((A +- CST1) +- CST2 -> A + CST3): If A is
19489 also a CONSTANT_CLASS_P, punt.
19491 2018-02-14 Jim Wilson <jimw@sifive.com>
19493 * config/riscv/riscv.c (riscv_first_stack_step): Move locals after
19494 first SMALL_OPERAND check. New local min_second_step. Move assert
19495 to where locals are set. Add TARGET_RVC support.
19496 * config/riscv/riscv.h (C_SxSP_BITS, SWSP_REACH, SDSP_REACH): New.
19498 2018-02-14 Indu Bhagat <indu.bhagat@oracle.com>
19500 * doc/invoke.texi: Correct -Wformat-overflow code sample.
19502 2018-02-14 Martin Sebor <msebor@redhat.com>
19504 PR tree-optimization/83698
19505 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): For
19506 arrays constrain the offset range to their bounds.
19507 (builtin_access::strcat_overlap): Adjust the bounds of overlap offset.
19508 (builtin_access::overlap): Avoid setting the size of overlap if it's
19510 (maybe_diag_overlap): Also consider arrays when deciding what values
19511 of offsets to include in diagnostics.
19513 2018-02-14 Martin Sebor <msebor@redhat.com>
19516 * attribs.c (diag_attr_exclusions): Consider the exclusion(s)
19517 that correspond to the kind of a declaration.
19519 2018-02-14 John David Anglin <danglin@gcc.gnu.org>
19522 * config/pa/pa.md: Load address of PIC label using the linkage table
19523 if the label is nonlocal.
19525 2018-02-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
19527 * config/rs6000/rs6000.c (rs6000_option_override_internal): Issue
19528 warning message if user requests -maltivec=be.
19529 * doc/invoke.texi: Document deprecation of -maltivec=be.
19531 2018-02-14 Will Schmidt <will_schmidt@vnet.ibm.com>
19534 * config/rs6000/rs6000-c.c: Update definitions for
19535 ALTIVEC_BUILTIN_VEC_SLD, ALTIVEC_BUILTIN_VEC_SLDW,
19536 VEC_XXSLDWI and ALTIVEC_BUILTIN_VEC_XXPERMDI builtins.
19538 2018-02-14 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
19541 * config/i386/cetintrin.h: Remove _rdssp[d|q] and
19542 add _get_ssp intrinsics. Remove argument from
19543 __builtin_ia32_rdssp[d|q].
19544 * config/i386/i386-builtin-types.def: Add UINT_FTYPE_VOID.
19545 * config/i386/i386-builtin.def: Remove argument from
19546 __builtin_ia32_rdssp[d|q].
19547 * config/i386/i386.c: Use UINT_FTYPE_VOID. Use
19548 ix86_expand_special_args_builtin for _rdssp[d|q].
19549 * config/i386/i386.md: Remove argument from rdssp[si|di] insn.
19550 Clear register before usage.
19551 * doc/extend.texi: Remove argument from __builtin_ia32_rdssp[d|q].
19552 Add documentation for new _get_ssp and _inc_ssp intrinsics.
19554 2018-02-14 Richard Sandiford <richard.sandiford@linaro.org>
19556 PR tree-optimization/84357
19557 * tree-data-ref.c (object_address_invariant_in_loop_p): Check
19558 operand 1 of an ARRAY_REF too.
19560 2018-02-14 Oleg Endo <olegendo@gcc.gnu.org>
19563 * config/rx/rx-protos.h (rx_reg_dead_or_unused_after_insn,
19564 rx_copy_reg_dead_or_unused_notes, rx_fuse_in_memory_bitop): New
19566 (set_of_reg): New struct.
19567 (rx_find_set_of_reg, rx_find_use_of_reg): New functions.
19568 * config/rx/rx.c (rx_reg_dead_or_unused_after_insn,
19569 rx_copy_reg_dead_or_unused_notes, rx_fuse_in_memory_bitop): New
19571 * config/rx/rx.md (andsi3, iorsi3, xorsi3): Convert to insn_and_split.
19572 Split into bitclr, bitset, bitinvert patterns if appropriate.
19573 (*bitset, *bitinvert, *bitclr): Convert to named insn_and_split and
19574 use rx_fuse_in_memory_bitop.
19575 (*bitset_in_memory, *bitinvert_in_memory, *bitclr_in_memory): Convert
19576 to named insn, correct maximum insn length.
19578 2018-02-14 Jozef Lawrynowicz <jozefl.gcc@gmail.com>
19581 * machmode.def: Define a complex mode for PARTIAL_INT.
19582 * genmodes.c (complex_class): Return MODE_COMPLEX_INT for
19584 * doc/rtl.texi: Document CSPImode.
19585 * config/msp430/msp430.c (msp430_hard_regno_nregs): Add CPSImode
19587 (msp430_hard_regno_nregs_with_padding): Likewise.
19589 2018-02-13 Peter Bergner <bergner@vnet.ibm.com>
19592 * config/rs6000/rs6000.c (mem_operand_gpr): Disallow altivec addresses.
19594 2018-02-13 Segher Boessenkool <segher@kernel.crashing.org>
19596 PR rtl-optimization/84169
19597 * combine.c (try_combine): New variable split_i2i3. Set it to true if
19598 we generated a parallel as new i3 and we split that to new i2 and i3
19599 instructions. Handle split_i2i3 similar to swap_i2i3: scan the
19600 LOG_LINKs of i3 to see which of those need to link to i2 now. Link
19601 those to i2, not i1. Partially rewrite this scan code.
19603 2018-02-13 Jakub Jelinek <jakub@redhat.com>
19606 * stor-layout.c (place_field): For variable length fields, adjust
19607 offset_align afterwards not just based on the field's alignment,
19608 but also on the size.
19610 PR middle-end/84309
19611 * match.pd (pow(C,x) -> exp(log(C)*x)): Use exp2s and log2s instead
19612 of exps and logs in the use_exp2 case.
19614 2018-02-13 Jeff Law <law@redhat.com>
19616 * config/rl/rl78.c (rl78_attribute_table): Fix terminator and
19617 entry for "vector".
19619 * config/rl78/rl78.c (rl78_handle_func_attribute): Mark
19622 2018-02-13 Alexandre Oliva <aoliva@redhat.com>
19626 * common.opt (gas-loc-support, gas-locview-support): New.
19627 (ginline-points, ginternal-reset-location-views): New.
19628 * doc/invoke.texi: Document them. Use @itemx where intended.
19629 (gvariable-location-views): Adjust.
19630 * target.def (reset_location_view): New.
19631 * doc/tm.texi.in (DWARF2_ASM_VIEW_DEBUG_INFO): New.
19632 (TARGET_RESET_LOCATION_VIEW): New.
19633 * doc/tm.texi: Rebuilt.
19634 * dwarf2out.c (dwarf2out_default_as_loc_support): New.
19635 (dwarf2out_default_as_locview_support): New.
19636 (output_asm_line_debug_info): Use option variables.
19637 (dwarf2out_maybe_output_loclist_view_pair): Likewise.
19638 (output_loc_list): Likewise.
19639 (add_high_low_attributes): Check option variables.
19640 Don't output entry view attribute in strict mode.
19641 (gen_inlined_subroutine_die): Check option variables.
19642 (dwarf2out_inline_entry): Likewise.
19643 (init_sections_and_labels): Likewise.
19644 (dwarf2out_early_finish): Likewise.
19645 (maybe_reset_location_view): New, from...
19646 (dwarf2out_var_location): ... here. Call it.
19647 * debug.h (dwarf2out_default_as_loc_support): Declare.
19648 (dwarf2out_default_as_locview_support): Declare.
19649 * hooks.c (hook_int_rtx_insn_0): New.
19650 * hooks.h (hook_int_rtx_insn_0): Declare.
19651 * toplev.c (process_options): Take -gas-loc-support and
19652 -gas-locview-support from dwarf2out. Enable
19653 -gvariable-location-views by default only with locview
19654 assembler support. Enable -ginternal-reset-location-views by
19655 default only if the target defines the corresponding hook.
19656 Enable -ginline-points by default if location views are
19657 enabled; force it disabled if statement frontiers are
19659 * tree-inline.c (expand_call_inline): Check option variables.
19660 * tree-ssa-live.c (remove_unused_scope_block_p): Likewise.
19662 2018-02-13 Richard Sandiford <richard.sandiford@linaro.org>
19664 PR tree-optimization/84321
19665 * tree-vrp.c (intersect_range_with_nonzero_bits): Fix VR_ANTI_RANGE
19666 handling. Also check whether the anti-range contains any values
19667 that satisfy the mask; switch to a VR_RANGE if not.
19669 2018-02-13 Paolo Bonzini <bonzini@gnu.org>
19672 * internal-fn.def (ASAN_CHECK, ASAN_MARK): Revert changes to fnspec.
19674 2018-02-13 Martin Jambor <mjambor@suse.cz>
19677 * ipa-param-manipulation.c (ipa_modify_call_arguments): Use location
19678 of call statements, also set location of a load to a temporary.
19680 2018-02-13 Sebastian Perta <sebastian.perta@renesas.com>
19682 * config/rl78/rl78.c (add_vector_labels): New function.
19683 * config/rl78/rl78.c (rl78_handle_vector_attribute): New function.
19684 * config/rl78/rl78.c (rl78_start_function): Call add_vector_labels.
19685 * config/rl78/rl78.c (rl78_handle_func_attribute): Removed the assert
19686 which checks that no arguments are passed.
19687 * config/rl78/rl78.c (rl78_attribute_table): Add "vector" attribute.
19688 * doc/extend.texi: Documentation for the new attribute.
19690 2018-02-13 Andreas Schwab <schwab@suse.de>
19692 * config/riscv/linux.h (CPP_SPEC): Define.
19694 2018-02-13 Jakub Jelinek <jakub@redhat.com>
19697 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Pass
19698 OPTION_MASK_ISA_AES | OPTION_MASK_ISA_SSE2 instead of
19699 OPTION_MASK_ISA_AES as first argument to def_builtin_const
19700 for AES builtins. Pass OPTION_MASK_ISA_PCLMUL | OPTION_MASK_ISA_SSE2
19701 instead of OPTION_MASK_ISA_PCLMUL as first argument to
19702 def_builtin_const for __builtin_ia32_pclmulqdq128 builtin.
19703 * config/i386/wmmintrin.h: If __SSE2__ is not defined, enable it
19704 temporarily for AES and PCLMUL builtins.
19706 PR tree-optimization/84339
19707 * gimple-fold.c (get_range_strlen): Set *FLEXP to true when handling
19708 ARRAY_REF where first operand is array_at_struct_end_p COMPONENT_REF.
19711 PR middle-end/84309
19712 * match.pd (pow(C,x) -> exp(log(C)*x)): Optimize instead into
19713 exp2(log2(C)*x) if C is a power of 2 and c99 runtime is available.
19714 * generic-match-head.c (canonicalize_math_after_vectorization_p): New
19716 * gimple-match-head.c (canonicalize_math_after_vectorization_p): New
19718 * omp-simd-clone.h: New file.
19719 * omp-simd-clone.c: Include omp-simd-clone.h.
19720 (expand_simd_clones): No longer static.
19721 * tree-vect-patterns.c: Include fold-const-call.h, attribs.h,
19722 cgraph.h and omp-simd-clone.h.
19723 (vect_recog_pow_pattern): Optimize pow(C,x) to exp(log(C)*x).
19724 (vect_recog_widen_shift_pattern): Formatting fix.
19725 (vect_pattern_recog_1): Don't check optab for calls.
19728 * config/i386/sse.md (<avx512>_vpermi2var<mode>3_mask): Force
19729 operands[2] into a REG before using gen_lowpart on it.
19731 2018-02-12 Jeff Law <law@redhat.com>
19734 * config/sh/sh.c (find_barrier): Consider a sibling call
19737 * cse.c (try_back_substitute_reg): Move any REG_ARGS_SIZE note when
19738 successfully back substituting a reg.
19740 2018-02-12 Richard Biener <rguenther@suse.de>
19742 PR tree-optimization/84037
19743 * tree-vect-slp.c (vect_analyze_slp_cost): Add visited
19744 parameter, move visited init to caller.
19745 (vect_slp_analyze_operations): Separate cost from validity
19746 check, initialize visited once for all instances.
19747 (vect_schedule_slp): Analyze map to CSE vectorized nodes once
19749 * tree-vect-stmts.c (vect_model_simple_cost): Make early
19751 (vect_model_promotion_demotion_cost): Likewise.
19752 (vectorizable_bswap): Guard cost modeling with !slp_node
19753 instead of !PURE_SLP_STMT to avoid double-counting on hybrid
19755 (vectorizable_call): Likewise.
19756 (vectorizable_conversion): Likewise.
19757 (vectorizable_assignment): Likewise.
19758 (vectorizable_shift): Likewise.
19759 (vectorizable_operation): Likewise.
19760 (vectorizable_store): Likewise.
19761 (vectorizable_load): Likewise.
19762 (vectorizable_condition): Likewise.
19763 (vectorizable_comparison): Likewise.
19765 2018-02-12 Paolo Bonzini <bonzini@gnu.org>
19768 * internal-fn.def (ASAN_CHECK): Fix fnspec to account for return value.
19769 (ASAN_MARK): Fix fnspec to account for return value, change pointer
19770 argument from 'R' to 'W' so that the pointed-to datum is clobbered.
19772 2018-02-08 Jan Hubicka <hubicka@ucw.cz>
19774 PR middle-end/83665
19775 * params.def (inline-min-speedup): Increase from 8 to 15.
19776 (max-inline-insns-auto): Decrease from 40 to 30.
19777 * ipa-split.c (consider_split): Add some buffer for function to
19778 be considered inlining candidate.
19779 * invoke.texi (max-inline-insns-auto, inline-min-speedup): UPdate
19782 2018-02-12 Richard Biener <rguenther@suse.de>
19784 PR tree-optimization/84037
19785 * tree-vect-slp.c (vect_build_slp_tree_2): Try swapping the
19786 matched stmts if we cannot swap the non-matched ones.
19788 2018-02-12 Olga Makhotina <olga.makhotina@intel.com>
19790 * config/i386/avx512fintrin.h (_mm_mask_scalef_round_sd,
19791 _mm_maskz_scalef_round_sd, _mm_mask_scalef_round_ss,
19792 _mm_maskz_scalef_round_ss): New intrinsics.
19793 (__builtin_ia32_scalefsd_round, __builtin_ia32_scalefss_round): Fix.
19794 * config/i386/i386-builtin.def (__builtin_ia32_scalefsd_round,
19795 __builtin_ia32_scalefss_round): Remove.
19796 (__builtin_ia32_scalefsd_mask_round,
19797 __builtin_ia32_scalefss_mask_round): New intrinsics.
19798 * config/i386/sse.md (vmscalef<mode><round_name>): Renamed to ...
19799 (vmscalef<mode><mask_scalar_name><round_scalar_name>): ... this.
19800 ((match_operand:VF_128 2 "<round_nimm_predicate>"
19801 "<round_constraint>")): Changed to ...
19802 ((match_operand:VF_128 2 "<round_scalar_nimm_predicate>"
19803 "<round_scalar_constraint>")): ... this.
19804 ("vscalef<ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|
19805 %0, %1, %2<round_op3>}"): Changed to ...
19806 ("vscalef<ssescalarmodesuffix>\t{<round_scalar_mask_op3>%2, %1,
19807 %0<mask_scalar_operand3>|%0<mask_scalar_operand3>, %1,
19808 %2<round_scalar_mask_op3>}"): ... this.
19809 * config/i386/subst.md (round_scalar_nimm_predicate): New.
19811 2018-02-12 Olga Makhotina <olga.makhotina@intel.com>
19813 * config/i386/avx512fintrin.h (_mm_mask_sqrt_round_sd)
19814 (_mm_maskz_sqrt_round_sd, _mm_mask_sqrt_round_ss)
19815 (_mm_maskz_sqrt_round_ss): New intrinsics.
19816 (__builtin_ia32_sqrtsd_round, __builtin_ia32_sqrtss_round): Remove.
19817 (__builtin_ia32_sqrtsd_mask_round)
19818 (__builtin_ia32_sqrtss_mask_round): New builtins.
19819 * config/i386/i386-builtin.def (__builtin_ia32_sqrtsd_round)
19820 (__builtin_ia32_sqrtss_round): Remove.
19821 (__builtin_ia32_sqrtsd_mask_round)
19822 (__builtin_ia32_sqrtss_mask_round): New builtins.
19823 * config/i386/sse.md (vmsqrt<mode>2<round_name>): Renamed to ...
19824 (vmsqrt<mode>2<mask_scalar_name><round_scalar_name>): ... this.
19825 ((match_operand:VF_128 1 "vector_operand"
19826 "xBm,<round_constraint>")): Changed to ...
19827 ((match_operand:VF_128 1 "vector_operand"
19828 "xBm,<round_scalar_constraint>")): ... this.
19829 (vsqrt<ssescalarmodesuffix>\t{<round_op3>%1, %2, %0|
19830 %0, %2, %<iptr>1<round_op3>}): Changed to ...
19831 (vsqrt<ssescalarmodesuffix>\t{<round_scalar_mask_op3>%1, %2,
19832 %0<mask_scalar_operand3>|%0<mask_scalar_operand3>, %2,
19833 %<iptr>1<round_scalar_mask_op3>}): ... this.
19834 ((set_attr "prefix" "<round_prefix>")): Changed to ...
19835 ((set_attr "prefix" "<round_scalar_prefix>")): ... this.
19837 2018-02-11 Steven Munroe <munroesj@gcc.gnu.org>
19840 * config/rs6000/mmintrin.h (_mm_cmpeq_pi32 [_ARCH_PWR9]):
19841 Cast vec_cmpeq result to correct type.
19842 * config/rs6000/mmintrin.h (_mm_cmpgt_pi32 [_ARCH_PWR9]):
19843 Cast vec_cmpgt result to correct type.
19845 2018-02-11 Alexandre Oliva <aoliva@redhat.com>
19847 * final.c (final_scan_insn_1): Renamed from...
19848 (final_scan_insn): ... this. New wrapper, to recover
19849 seen from the outermost call in recursive ones.
19850 * config/sparc/sparc.c (output_return): Drop seen from call.
19851 (output_sibcall): Likewise.
19852 * config/visium/visium.c (output_branch): Likewise.
19854 2018-02-10 John David Anglin <danglin@gcc.gnu.org>
19856 * config/pa/pa.c (hppa_profile_hook): Mark SYMBOL_REF for _mcount as
19859 2018-02-10 Alan Modra <amodra@gmail.com>
19862 * config/rs6000/rs6000.md (split_stack_return): Remove (use ..).
19863 Specify LR as an input.
19865 2018-02-10 Jakub Jelinek <jakub@redhat.com>
19868 * omp-low.c (maybe_remove_omp_member_access_dummy_vars,
19869 remove_member_access_dummy_vars): New functions.
19870 (lower_omp_for, lower_omp_taskreg, lower_omp_target,
19871 lower_omp_1, execute_lower_omp): Use them.
19873 PR rtl-optimization/84308
19874 * shrink-wrap.c (spread_components): Release todo vector.
19876 2018-02-09 Vladimir Makarov <vmakarov@redhat.com>
19878 PR rtl-optimization/57193
19879 * ira-color.c (struct allocno_color_data): Add member
19880 conflict_allocno_hard_prefs.
19881 (update_conflict_allocno_hard_prefs): New.
19882 (bucket_allocno_compare_func): Add a preference based on
19883 conflict_allocno_hard_prefs.
19884 (push_allocno_to_stack): Update conflict_allocno_hard_prefs.
19885 (color_allocnos): Remove a dead code. Initiate
19886 conflict_allocno_hard_prefs. Call update_costs_from_prefs.
19888 2018-02-09 Jakub Jelinek <jakub@redhat.com>
19891 * config/rs6000/vsx.md (p9_xxbrq_v16qi): Change input operand
19892 constraint from =wa to wa. Avoid a subreg on the output operand,
19893 instead use a pseudo and subreg it in a move.
19894 (p9_xxbrd_<mode>): Changed to ...
19895 (p9_xxbrd_v2di): ... this insn, without VSX_D iterator.
19896 (p9_xxbrd_v2df): New expander.
19897 (p9_xxbrw_<mode>): Changed to ...
19898 (p9_xxbrw_v4si): ... this insn, without VSX_W iterator.
19899 (p9_xxbrw_v4sf): New expander.
19901 2018-02-09 Sebastian Perta <sebastian.perta@renesas.com>
19903 * config/rx/rx.md (movsicc): Update expander to be matched by GCC.
19905 2018-02-09 Peter Bergner <bergner@vnet.ibm.com>
19908 * config/rs6000/vsx.md (vsx_mul_v2di): Handle generating a 64-bit
19909 multiply in 32-bit mode.
19910 (vsx_div_v2di): Handle generating a 64-bit signed divide in 32-bit mode.
19911 (vsx_udiv_v2di): Handle generating a 64-bit unsigned divide in 32-bit
19914 2018-02-09 Sebastian Perta <sebastian.perta@renesas.com>
19916 * config/rx/constraints.md (CALL_OP_SYMBOL_REF): Added new constraint
19917 to allow or block "symbol_ref" depending on the value of TARGET_JSR.
19918 * config/rx/rx.md (call_internal): Use CALL_OP_SYMBOL_REF.
19919 * config/rx/rx.md (call_value_internal): Use CALL_OP_SYMBOL_REF.
19921 2018-02-09 Pierre-Marie de Rodat <derodat@adacore.com>
19924 * dwarf2out.c (is_trivial_indirect_ref): New function.
19925 (dwarf2out_late_global_decl): Do not generate a location
19926 attribute for variables that have a non-trivial DECL_VALUE_EXPR
19927 and that are not defined in the current unit.
19929 2018-02-09 Eric Botcazou <ebotcazou@adacore.com>
19931 * optabs.c (prepare_cmp_insn): Try harder to emit a direct comparison
19932 instead of a libcall for UNORDERED.
19934 2018-02-09 Tamar Christina <tamar.christina@arm.com>
19937 * config/arm/arm-c.c (arm_cpu_builtins): Un-define __ARM_FEATURE_LDREX,
19938 __ARM_ARCH_PROFILE, __ARM_ARCH_ISA_THUMB, __ARM_FP and __ARM_NEON_FP.
19940 2018-02-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19943 * config/s390/s390.c (s390_set_current_function): Invoke
19944 s390_indirect_branch_settings also if fndecl didn't change.
19946 2018-02-09 Alexandre Oliva <aoliva@redhat.com>
19948 * config/rs6000/rs6000.md (blockage): Set length to zero.
19950 2018-02-09 Eric Botcazou <ebotcazou@adacore.com>
19952 * expr.c (optimize_bitfield_assignment_op): Remove obsolete assertion.
19954 2018-02-09 Jakub Jelinek <jakub@redhat.com>
19957 * gcc.c (STATIC_LIBASAN_LIBS, STATIC_LIBTSAN_LIBS,
19958 STATIC_LIBLSAN_LIBS, STATIC_LIBUBSAN_LIBS): Handle -static like
19962 * var-tracking.c (vt_add_function_parameter): Punt for non-onepart
19963 PARALLEL incoming that failed vt_get_decl_and_offset check.
19965 PR middle-end/84237
19966 * output.h (bss_initializer_p): Add NAMED argument, defaulted to false.
19967 * varasm.c (bss_initializer_p): Add NAMED argument, if true, ignore
19969 (get_variable_section): For decls in named .bss* sections pass true as
19970 second argument to bss_initializer_p.
19972 2018-02-09 Marek Polacek <polacek@redhat.com>
19973 Jakub Jelinek <jakub@redhat.com>
19976 * fold-const.c (fold_indirect_ref_1): Use VECTOR_TYPE_P macro.
19977 Formatting fixes. Verify first that tree_fits_poly_int64_p (op01).
19978 Sync some changes from cxx_fold_indirect_ref.
19980 2018-02-09 Alexandre Oliva <aoliva@redhat.com>
19982 * cfgexpand.c (expand_gimple_basic_block): Handle inline entry
19984 * dwarf2out.c (dwarf2_debug_hooks): Enable inline_entry hook.
19985 (BLOCK_INLINE_ENTRY_LABEL): New.
19986 (dwarf2out_var_location): Disregard inline entry markers.
19987 (inline_entry_data): New struct.
19988 (inline_entry_data_hasher): New hashtable type.
19989 (inline_entry_data_hasher::hash): New.
19990 (inline_entry_data_hasher::equal): New.
19991 (inline_entry_data_table): New variable.
19992 (add_high_low_attributes): Add DW_AT_entry_pc and
19993 DW_AT_GNU_entry_view attributes if a pending entry is found
19994 in inline_entry_data_table. Add old entry_pc attribute only
19995 if debug nonbinding markers are disabled.
19996 (gen_inlined_subroutine_die): Set BLOCK_DIE if nonbinding
19997 markers are enabled.
19998 (block_within_block_p, dwarf2out_inline_entry): New.
19999 (dwarf2out_finish): Check that no entries remained in
20000 inline_entry_data_table.
20001 * final.c (reemit_insn_block_notes): Handle inline entry notes.
20002 (final_scan_insn, notice_source_line): Likewise.
20003 (rest_of_clean_state): Skip inline entry markers.
20004 * gimple-pretty-print.c (dump_gimple_debug): Handle inline entry
20006 * gimple.c (gimple_build_debug_inline_entry): New.
20007 * gimple.h (enum gimple_debug_subcode): Add
20008 GIMPLE_DEBUG_INLINE_ENTRY.
20009 (gimple_build_debug_inline_entry): Declare.
20010 (gimple_debug_inline_entry_p): New.
20011 (gimple_debug_nonbind_marker_p): Adjust.
20012 * insn-notes.def (INLINE_ENTRY): New.
20013 * print-rtl.c (rtx_writer::print_rtx_operand_code_0): Handle
20014 inline entry marker notes.
20015 (print_insn): Likewise.
20016 * rtl.h (NOTE_MARKER_P): Add INLINE_ENTRY support.
20017 (INSN_DEBUG_MARKER_KIND): Likewise.
20018 (GEN_RTX_DEBUG_MARKER_INLINE_ENTRY_PAT): New.
20019 * tree-inline.c (expand_call_inline): Build and insert
20020 debug_inline_entry stmt.
20021 * tree-ssa-live.c (remove_unused_scope_block_p): Preserve
20022 inline entry blocks early, if nonbind markers are enabled.
20023 (dump_scope_block): Dump fragment info.
20024 * var-tracking.c (reemit_marker_as_note): Handle inline entry note.
20025 * doc/gimple.texi (gimple_debug_inline_entry_p): New.
20026 (gimple_build_debug_inline_entry): New.
20027 * doc/invoke.texi (gstatement-frontiers, gno-statement-frontiers):
20028 Enable/disable inline entry points too.
20029 * doc/rtl.texi (NOTE_INSN_INLINE_ENTRY): New.
20030 (DEBUG_INSN): Describe inline entry markers.
20032 * common.opt (gvariable-location-views): New.
20033 (gvariable-location-views=incompat5): New.
20034 * config.in: Rebuilt.
20035 * configure: Rebuilt.
20036 * configure.ac: Test assembler for view support.
20037 * dwarf2asm.c (dw2_asm_output_symname_uleb128): New.
20038 * dwarf2asm.h (dw2_asm_output_symname_uleb128): Declare.
20039 * dwarf2out.c (var_loc_view): New typedef.
20040 (struct dw_loc_list_struct): Add vl_symbol, vbegin, vend.
20041 (dwarf2out_locviews_in_attribute): New.
20042 (dwarf2out_locviews_in_loclist): New.
20043 (dw_val_equal_p): Compare val_view_list of dw_val_class_view_lists.
20044 (enum dw_line_info_opcode): Add LI_adv_address.
20045 (struct dw_line_info_table): Add view.
20046 (RESET_NEXT_VIEW, RESETTING_VIEW_P): New macros.
20047 (DWARF2_ASM_VIEW_DEBUG_INFO): Define default.
20048 (zero_view_p): New variable.
20049 (ZERO_VIEW_P): New macro.
20050 (output_asm_line_debug_info): New.
20051 (struct var_loc_node): Add view.
20052 (add_AT_view_list, AT_loc_list): New.
20053 (add_var_loc_to_decl): Add view param. Test it against last.
20054 (new_loc_list): Add view params. Record them.
20055 (AT_loc_list_ptr): Handle loc and view lists.
20056 (view_list_to_loc_list_val_node): New.
20057 (print_dw_val): Handle dw_val_class_view_list.
20058 (size_of_die): Likewise.
20059 (value_format): Likewise.
20060 (loc_list_has_views): New.
20061 (gen_llsym): Set vl_symbol too.
20062 (maybe_gen_llsym, skip_loc_list_entry): New.
20063 (dwarf2out_maybe_output_loclist_view_pair): New.
20064 (output_loc_list): Output view list or entries too.
20065 (output_view_list_offset): New.
20066 (output_die): Handle dw_val_class_view_list.
20067 (output_dwarf_version): New.
20068 (output_compilation_unit_header): Use it.
20069 (output_skeleton_debug_sections): Likewise.
20070 (output_rnglists, output_line_info): Likewise.
20071 (output_pubnames, output_aranges): Update version comments.
20072 (output_one_line_info_table): Output view numbers in asm comments.
20073 (dw_loc_list): Determine current endview, pass it to new_loc_list.
20074 Call maybe_gen_llsym.
20075 (loc_list_from_tree_1): Adjust.
20076 (add_AT_location_description): Create view list attribute if
20077 needed, check it's absent otherwise.
20078 (convert_cfa_to_fb_loc_list): Adjust.
20079 (maybe_emit_file): Call output_asm_line_debug_info for test.
20080 (dwarf2out_var_location): Reset views as needed. Precompute
20081 add_var_loc_to_decl args. Call get_attr_min_length only if we have the
20082 attribute. Set view.
20083 (new_line_info_table): Reset next view.
20084 (set_cur_line_info_table): Call output_asm_line_debug_info for test.
20085 (dwarf2out_source_line): Likewise. Output view resets and labels to
20086 the assembler, or select appropriate line info opcodes.
20087 (prune_unused_types_walk_attribs): Handle dw_val_class_view_list.
20088 (optimize_string_length): Catch it. Adjust.
20089 (resolve_addr): Copy vl_symbol along with ll_symbol. Handle
20090 dw_val_class_view_list, and remove it if no longer needed.
20091 (hash_loc_list): Hash view numbers.
20092 (loc_list_hasher::equal): Compare them.
20093 (optimize_location_lists): Check whether a view list symbol is
20094 needed, and whether the locview attribute is present, and
20095 whether they match. Remove the locview attribute if no longer
20097 (index_location_lists): Call skip_loc_list_entry for test.
20098 (dwarf2out_finish): Call output_asm_line_debug_info for test.
20099 Use output_dwarf_version.
20100 * dwarf2out.h (enum dw_val_class): Add dw_val_class_view_list.
20101 (struct dw_val_node): Add val_view_list.
20102 * final.c (SEEN_NEXT_VIEW): New.
20103 (set_next_view_needed): New.
20104 (clear_next_view_needed): New.
20105 (maybe_output_next_view): New.
20106 (final_start_function): Rename to...
20107 (final_start_function_1): ... this. Take pointer to FIRST,
20108 add SEEN parameter. Emit param bindings in the initial view.
20109 (final_start_function): Reintroduce SEEN-less interface.
20110 (final): Rename to...
20111 (final_1): ... this. Take SEEN parameter. Output final pending
20112 next view at the end.
20113 (final): Reintroduce seen-less interface.
20114 (final_scan_insn): Output pending next view before switching
20115 sections or ending a block. Mark the next view as needed when
20116 outputting variable locations. Notify debug backend of section
20117 changes, and of location view changes.
20118 (rest_of_handle_final): Adjust.
20119 * toplev.c (process_options): Autodetect value for debug variable
20120 location views option. Warn on incompat5 without -gdwarf-5.
20121 * doc/invoke.texi (gvariable-location-views): New.
20122 (gvariable-location-views=incompat5): New.
20123 (gno-variable-location-views): New.
20125 2018-02-08 David Malcolm <dmalcolm@redhat.com>
20127 PR tree-optimization/84136
20128 * tree-cfg.c (find_taken_edge_computed_goto): Remove assertion
20129 that the result of find_edge is non-NULL.
20131 2018-02-08 Sergey Shalnov <sergey.shalnov@intel.com>
20134 * config/i386/x86-tune-costs.h (skylake_cost): Fix cost of
20135 storing integer register in SImode. Fix cost of 256 and 512
20136 byte aligned SSE register store.
20138 2018-02-08 Sergey Shalnov <sergey.shalnov@intel.com>
20140 * config/i386/i386.c (ix86_multiplication_cost): Fix
20141 multiplication cost for TARGET_AVX512DQ.
20143 2018-02-08 Marek Polacek <polacek@redhat.com>
20145 PR tree-optimization/84238
20146 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Verify the result of
20149 2018-02-08 Richard Sandiford <richard.sandiford@linaro.org>
20151 PR tree-optimization/84265
20152 * tree-vect-stmts.c (vectorizable_store): Don't treat
20153 VMAT_CONTIGUOUS accesses as grouped.
20154 (vectorizable_load): Likewise.
20156 2018-02-08 Richard Sandiford <richard.sandiford@linaro.org>
20158 PR tree-optimization/81635
20159 * wide-int.h (wi::round_down_for_mask, wi::round_up_for_mask): Declare.
20160 * wide-int.cc (wi::round_down_for_mask, wi::round_up_for_mask)
20161 (test_round_for_mask): New functions.
20162 (wide_int_cc_tests): Call test_round_for_mask.
20163 * tree-vrp.h (intersect_range_with_nonzero_bits): Declare.
20164 * tree-vrp.c (intersect_range_with_nonzero_bits): New function.
20165 * tree-data-ref.c (split_constant_offset_1): Use it to refine the
20166 range returned by get_range_info.
20168 2018-02-08 Jan Hubicka <hubicka@ucw.cz>
20171 * cgraph.h (symtab_node::output_to_lto_symbol_table_p): Declare
20172 * symtab.c: Include builtins.h
20173 (symtab_node::output_to_lto_symbol_table_p): Move here
20174 from lto-streamer-out.c:output_symbol_p.
20175 * lto-streamer-out.c (write_symbol): Turn early exit to assert.
20176 (output_symbol_p): Move all logic to symtab.c
20177 (produce_symtab): Update.
20179 2018-02-08 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20181 * config/s390/s390-opts.h (enum indirect_branch): Define.
20182 * config/s390/s390-protos.h (s390_return_addr_from_memory)
20183 (s390_indirect_branch_via_thunk)
20184 (s390_indirect_branch_via_inline_thunk): Add function prototypes.
20185 (enum s390_indirect_branch_type): Define.
20186 * config/s390/s390.c (struct s390_frame_layout, struct
20187 machine_function): Remove.
20188 (indirect_branch_prez10thunk_mask, indirect_branch_z10thunk_mask)
20189 (indirect_branch_table_label_no, indirect_branch_table_name):
20191 (INDIRECT_BRANCH_NUM_OPTIONS): Define macro.
20192 (enum s390_indirect_branch_option): Define.
20193 (s390_return_addr_from_memory): New function.
20194 (s390_handle_string_attribute): New function.
20195 (s390_attribute_table): Add new attribute handler.
20196 (s390_execute_label): Handle UNSPEC_EXECUTE_JUMP patterns.
20197 (s390_indirect_branch_via_thunk): New function.
20198 (s390_indirect_branch_via_inline_thunk): New function.
20199 (s390_function_ok_for_sibcall): When jumping via thunk disallow
20200 sibling call optimization for non z10 compiles.
20201 (s390_emit_call): Force indirect branch target to be a single
20202 register. Add r1 clobber for non-z10 compiles.
20203 (s390_emit_epilogue): Emit return jump via return_use expander.
20204 (s390_reorg): Handle JUMP_INSNs as execute targets.
20205 (s390_option_override_internal): Perform validity checks for the
20206 new command line options.
20207 (s390_indirect_branch_attrvalue): New function.
20208 (s390_indirect_branch_settings): New function.
20209 (s390_set_current_function): Invoke s390_indirect_branch_settings.
20210 (s390_output_indirect_thunk_function): New function.
20211 (s390_code_end): Implement target hook.
20212 (s390_case_values_threshold): Implement target hook.
20213 (TARGET_ASM_CODE_END, TARGET_CASE_VALUES_THRESHOLD): Define target
20215 * config/s390/s390.h (struct s390_frame_layout)
20216 (struct machine_function): Move here from s390.c.
20217 (TARGET_INDIRECT_BRANCH_NOBP_RET)
20218 (TARGET_INDIRECT_BRANCH_NOBP_JUMP)
20219 (TARGET_INDIRECT_BRANCH_NOBP_JUMP_THUNK)
20220 (TARGET_INDIRECT_BRANCH_NOBP_JUMP_INLINE_THUNK)
20221 (TARGET_INDIRECT_BRANCH_NOBP_CALL)
20222 (TARGET_DEFAULT_INDIRECT_BRANCH_TABLE)
20223 (TARGET_INDIRECT_BRANCH_THUNK_NAME_EXRL)
20224 (TARGET_INDIRECT_BRANCH_THUNK_NAME_EX)
20225 (TARGET_INDIRECT_BRANCH_TABLE): Define macros.
20226 * config/s390/s390.md (UNSPEC_EXECUTE_JUMP)
20227 (INDIRECT_BRANCH_THUNK_REGNUM): Define constants.
20228 (mnemonic attribute): Add values which aren't recognized
20230 ("*cjump_long", "*icjump_long", "*basr", "*basr_r"): Disable
20231 pattern for branch conversion. Fix mnemonic attribute.
20232 ("*c<code>", "*sibcall_br", "*sibcall_value_br", "*return"): Emit
20233 indirect branch via thunk if requested.
20234 ("indirect_jump", "<code>"): Expand patterns for branch conversion.
20235 ("*indirect_jump"): Disable for branch conversion using out of
20237 ("indirect_jump_via_thunk<mode>_z10")
20238 ("indirect_jump_via_thunk<mode>")
20239 ("indirect_jump_via_inlinethunk<mode>_z10")
20240 ("indirect_jump_via_inlinethunk<mode>", "*casesi_jump")
20241 ("casesi_jump_via_thunk<mode>_z10", "casesi_jump_via_thunk<mode>")
20242 ("casesi_jump_via_inlinethunk<mode>_z10")
20243 ("casesi_jump_via_inlinethunk<mode>", "*basr_via_thunk<mode>_z10")
20244 ("*basr_via_thunk<mode>", "*basr_r_via_thunk_z10")
20245 ("*basr_r_via_thunk", "return<mode>_prez10"): New pattern.
20246 ("*indirect2_jump"): Disable for branch conversion.
20247 ("casesi_jump"): Turn into expander and expand patterns for branch
20249 ("return_use"): New expander.
20250 ("*return"): Emit return via thunk and rename it to ...
20251 ("*return<mode>"): ... this one.
20252 * config/s390/s390.opt: Add new options and and enum for the
20255 2018-02-08 Richard Sandiford <richard.sandiford@linaro.org>
20257 * lra-constraints.c (match_reload): Unconditionally use
20258 gen_lowpart_SUBREG, rather than selecting between that
20259 and equivalent gen_rtx_SUBREG code.
20261 2018-02-08 Richard Biener <rguenther@suse.de>
20263 PR tree-optimization/84233
20264 * tree-ssa-phiprop.c (propagate_with_phi): Use separate
20265 changed flag instead of boguously re-using phi_inserted.
20267 2018-02-08 Martin Jambor <mjambor@suse.cz>
20269 * hsa-gen.c (get_symbol_for_decl): Set program allocation for
20270 static local variables.
20272 2018-02-08 Richard Biener <rguenther@suse.de>
20274 PR tree-optimization/84278
20275 * tree-vect-stmts.c (vectorizable_store): When looking for
20276 smaller vector types to perform grouped strided loads/stores
20277 make sure the mode is supported by the target.
20278 (vectorizable_load): Likewise.
20280 2018-02-08 Wilco Dijkstra <wdijkstr@arm.com>
20282 * config/aarch64/aarch64.c (aarch64_components_for_bb):
20283 Increase LDP/STP opportunities by adding adjacent callee-saves.
20285 2018-02-08 Wilco Dijkstra <wdijkstr@arm.com>
20287 PR rtl-optimization/84068
20288 PR rtl-optimization/83459
20289 * haifa-sched.c (rank_for_schedule): Fix SCHED_PRESSURE_MODEL sorting.
20291 2018-02-08 Aldy Hernandez <aldyh@redhat.com>
20293 PR tree-optimization/84224
20294 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Remove assert.
20295 * calls.c (gimple_alloca_call_p): Only return TRUE when we have
20296 non-zero arguments.
20298 2018-02-07 Iain Sandoe <iain@codesourcery.com>
20301 * config/rs6000/altivec.md (*restore_world): Remove LR use.
20302 * config/rs6000/predicates.md (restore_world_operation): Adjust op
20303 count, remove one USE.
20305 2018-02-07 Michael Meissner <meissner@linux.vnet.ibm.com>
20307 * doc/install.texi (Configuration): Document the
20308 --with-long-double-format={ibm,ieee} PowerPC configuration
20312 * config/rs6000/rs6000.md (fix_trunc<SFDF:mode><QHI:mode>2):
20313 Convert from define_expand to be define_insn_and_split. Rework
20314 float/double/_Float128 conversions to QI/HI/SImode to work with
20315 both ISA 2.07 (power8) or ISA 3.0 (power9). Fix regression where
20316 conversions to QI/HImode types did a store and then a load to
20317 truncate the value. For conversions to VSX registers, don't split
20318 the insn, instead emit the code directly. Use the code iterator
20319 any_fix to combine signed and unsigned conversions.
20320 (fix<uns>_trunc<SFDF:mode>si2_p8): Likewise.
20321 (fixuns_trunc<SFDF:mode><QHI:mode>2): Likewise.
20322 (fix_trunc<IEEE128:mode><QHI:mode>2): Likewise.
20323 (fix<uns>_trunc<SFDF:mode><QHI:mode>2): Likewise.
20324 (fix_<mode>di2_hw): Likewise.
20325 (fixuns_<mode>di2_hw): Likewise.
20326 (fix_<mode>si2_hw): Likewise.
20327 (fixuns_<mode>si2_hw): Likewise.
20328 (fix<uns>_<IEEE128:mode><SDI:mode>2_hw): Likewise.
20329 (fix<uns>_trunc<IEEE128:mode><QHI:mode>2): Likewise.
20330 (fctiw<u>z_<mode>_smallint): Rename fctiw<u>z_<mode>_smallint to
20331 fix<uns>_trunc<SFDF:mode>si2_p8.
20332 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Delete, no longer used.
20333 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
20334 (fix<uns>_<mode>_mem): Likewise.
20335 (fctiw<u>z_<mode>_mem): Likewise.
20336 (fix<uns>_<mode>_mem): Likewise.
20337 (fix<uns>_trunc<SFDF:mode><QHSI:mode>2_mem): On ISA 3.0, prevent
20338 the register allocator from doing a direct move to the GPRs to do
20339 a store, and instead use the ISA 3.0 store byte/half-word from
20340 vector register instruction. For IEEE 128-bit floating point,
20341 also optimize stores of 32-bit ints.
20342 (fix<uns>_trunc<IEEE128:mode><QHSI:mode>2_mem): Likewise.
20344 2018-02-07 Alan Hayward <alan.hayward@arm.com>
20346 * genextract.c (push_pathstr_operand): New function to support
20348 (walk_rtx): Call push_pathstr_operand.
20349 (print_path): Support [a-zA-Z].
20351 2018-02-07 Richard Biener <rguenther@suse.de>
20353 PR tree-optimization/84037
20354 * tree-vectorizer.h (struct _loop_vec_info): Add ivexpr_map member.
20355 (cse_and_gimplify_to_preheader): Declare.
20356 (vect_get_place_in_interleaving_chain): Likewise.
20357 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
20359 (_loop_vec_info::~_loop_vec_info): Delete it.
20360 (cse_and_gimplify_to_preheader): New function.
20361 * tree-vect-slp.c (vect_get_place_in_interleaving_chain): Export.
20362 * tree-vect-stmts.c (vectorizable_store): CSE base and steps.
20363 (vectorizable_load): Likewise. For grouped stores always base
20364 the IV on the first element.
20365 * tree-vect-loop-manip.c (vect_loop_versioning): Unshare versioning
20366 condition before gimplifying.
20368 2018-02-07 Jakub Jelinek <jakub@redhat.com>
20370 * tree-eh.c (operation_could_trap_helper_p): Ignore honor_trapv for
20371 *DIV_EXPR and *MOD_EXPR.
20373 2018-02-07 H.J. Lu <hongjiu.lu@intel.com>
20376 * config/i386/i386.c (ix86_option_override_internal): Mask out
20377 the CF_SET bit when checking -fcf-protection.
20379 2018-02-07 Tom de Vries <tom@codesourcery.com>
20382 * omp-expand.c (expand_oacc_collapse_init): Ensure diff_type is large
20385 2018-02-07 Richard Biener <rguenther@suse.de>
20387 PR tree-optimization/84204
20388 * tree-chrec.c (chrec_fold_plus_1): Remove size limiting in
20391 PR tree-optimization/84205
20392 * graphite-isl-ast-to-gimple.c (binary_op_to_tree): Also
20393 special-case isl_ast_op_zdiv_r.
20395 PR tree-optimization/84223
20396 * graphite-scop-detection.c (gather_bbs::before_dom_children):
20397 Only add conditions from within the region.
20398 (gather_bbs::after_dom_children): Adjust.
20400 2018-02-07 Georg-Johann Lay <avr@gjlay.de>
20403 * config/avr/avr.h (GENERAL_REGNO_P, GENERAL_REG_P): New macros.
20404 * config/avr/avr.md: Only post-reload split REG-REG moves if
20405 either register is GENERAL_REG_P.
20407 2018-02-07 Jakub Jelinek <jakub@redhat.com>
20409 PR tree-optimization/84235
20410 * tree-ssa-scopedtables.c
20411 (avail_exprs_stack::simplify_binary_operation): Fir MINUS_EXPR, punt
20412 if the subtraction is performed in floating point type where NaNs are
20413 honored. For *DIV_EXPR, punt for ALL_FRACT_MODE_Ps where we can't
20414 build 1. Formatting fix.
20416 2018-02-06 Jakub Jelinek <jakub@redhat.com>
20419 * config/i386/i386.c (rest_of_insert_endbranch): Only skip
20420 NOTE_INSN_CALL_ARG_LOCATION after a call, not anything else,
20421 and skip it regardless of bb boundaries. Use CALL_P macro,
20422 don't test INSN_P (insn) together with CALL_P or JUMP_P check
20423 unnecessarily, formatting fix.
20425 2018-02-06 Michael Collison <michael.collison@arm.com>
20427 * config/arm/thumb2.md:
20428 (*thumb2_mov_negscc): Split only if TARGET_THUMB2 && !arm_restrict_it.
20429 (*thumb_mov_notscc): Ditto.
20431 2018-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
20434 * config/rs6000/rs6000.md (su code attribute): Use "u" for
20435 unsigned_fix, not "s".
20437 2018-02-06 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
20439 * configure.ac (gcc_fn_eh_frame_ro): New function.
20440 (gcc_cv_as_cfi_directive): Check both 32 and 64-bit assembler for
20441 correct .eh_frame permissions.
20442 * configure: Regenerate.
20444 2018-02-06 Andrew Jenner <andrew@codeourcery.com>
20446 * doc/invoke.texi: Add section for the PowerPC SPE backend. Remove
20447 irrelevant options.
20449 2018-02-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20451 * config/rs6000/rs6000.c (rs6000_option_override_internal):
20452 Display warning message for -mno-speculate-indirect-jumps.
20454 2018-02-06 Andrew Jenner <andrew@codesourcery.com>
20456 * config/powerpcspe/powerpcspe.opt: (msimple-fpu, mfpu) Add
20458 * config/powerpcspe/sysv4.opt (mbit-align): Likewise.
20460 2018-02-06 Aldy Hernandez <aldyh@redhat.com>
20462 PR tree-optimization/84225
20463 * tree-eh.c (find_trapping_overflow): Only call
20464 operation_no_trapping_overflow when ANY_INTEGRAL_TYPE_P.
20466 2018-02-06 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
20469 * config/i386/i386.c: Reimplement the check of possible options
20470 -mibt/-mshstk conbination. Change error messages.
20471 * doc/invoke.texi: Fix a typo: remove extra '='.
20473 2018-02-06 Marek Polacek <polacek@redhat.com>
20475 PR tree-optimization/84228
20476 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Skip debug statements.
20478 2018-02-06 Tamar Christina <tamar.christina@arm.com>
20481 * config/arm/arm.c (arm_print_asm_arch_directives): Record already
20482 emitted arch directives.
20483 * config/arm/arm-c.c (arm_cpu_builtins): Undefine __ARM_ARCH and
20484 __ARM_FEATURE_COPROC before changing architectures.
20486 2018-02-06 Richard Biener <rguenther@suse.de>
20488 * config/i386/i386.c (print_reg): Fix typo.
20489 (ix86_loop_unroll_adjust): Do not unroll beyond the original nunroll.
20491 2018-02-06 Eric Botcazou <ebotcazou@adacore.com>
20493 * configure: Regenerate.
20495 2018-02-05 Martin Sebor <msebor@redhat.com>
20497 PR tree-optimization/83369
20498 * tree-ssa-ccp.c (pass_post_ipa_warn::execute): Use %G to print
20501 2018-02-05 Martin Liska <mliska@suse.cz>
20503 * doc/invoke.texi: Cherry-pick upstream r323995.
20505 2018-02-05 Richard Sandiford <richard.sandiford@linaro.org>
20507 * ira.c (ira_init_register_move_cost): Adjust comment.
20509 2018-02-05 Martin Liska <mliska@suse.cz>
20511 PR gcov-profile/84137
20512 * doc/gcov.texi: Fix typo in documentation.
20514 2018-02-05 Martin Liska <mliska@suse.cz>
20516 PR gcov-profile/83879
20517 * doc/gcov.texi: Document necessity of --dynamic-list-data when
20518 using dlopen functionality.
20520 2018-02-05 Olga Makhotina <olga.makhotina@intel.com>
20522 * config/i386/avx512dqintrin.h (_mm_mask_range_sd, _mm_maskz_range_sd,
20523 _mm_mask_range_round_sd, _mm_maskz_range_round_sd, _mm_mask_range_ss,
20524 _mm_maskz_range_ss, _mm_mask_range_round_ss,
20525 _mm_maskz_range_round_ss): New intrinsics.
20526 (__builtin_ia32_rangesd128_round)
20527 (__builtin_ia32_rangess128_round): Remove.
20528 (__builtin_ia32_rangesd128_mask_round,
20529 __builtin_ia32_rangess128_mask_round): New builtins.
20530 * config/i386/i386-builtin.def (__builtin_ia32_rangesd128_round,
20531 __builtin_ia32_rangess128_round): Remove.
20532 (__builtin_ia32_rangesd128_mask_round,
20533 __builtin_ia32_rangess128_mask_round): New builtins.
20534 * config/i386/sse.md (ranges<mode><round_saeonly_name>): Renamed to ...
20535 (ranges<mode><mask_scalar_name><round_saeonly_scalar_name>): ... this.
20536 ((match_operand:VF_128 2 "<round_saeonly_nimm_predicate>"
20537 "<round_saeonly_constraint>")): Changed to ...
20538 ((match_operand:VF_128 2 "<round_saeonly_scalar_nimm_predicate>"
20539 "<round_saeonly_scalar_constraint>")): ... this.
20540 ("vrange<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
20541 %0, %1, %2<round_saeonly_op4>, %3}"): Changed to ...
20542 ("vrange<ssescalarmodesuffix>\t{%3, <round_saeonly_scalar_mask_op4>%2,
20543 %1, %0<mask_scalar_operand4>|%0<mask_scalar_operand4>, %1,
20544 %2<round_saeonly_scalar_mask_op4>, %3}"): ... this.
20546 2018-02-02 Andrew Jenner <andrew@codesourcery.com>
20548 * config/powerpcspe/powerpcspe.opt: Add Undocumented to irrelevant
20550 * config/powerpcspe/powerpcspe-tables.opt (rs6000_cpu_opt_value):
20551 Remove all values except native, 8540 and 8548.
20553 2018-02-02 H.J. Lu <hongjiu.lu@intel.com>
20555 * config/i386/i386.c (ix86_output_function_return): Pass
20556 INVALID_REGNUM, instead of -1, as invalid register number to
20557 indirect_thunk_name and output_indirect_thunk.
20559 2018-02-02 Julia Koval <julia.koval@intel.com>
20561 * config.gcc: Add -march=icelake.
20562 * config/i386/driver-i386.c (host_detect_local_cpu): Detect icelake.
20563 * config/i386/i386-c.c (ix86_target_macros_internal): Handle icelake.
20564 * config/i386/i386.c (processor_costs): Add m_ICELAKE.
20565 (PTA_ICELAKE, PTA_AVX512VNNI, PTA_GFNI, PTA_VAES, PTA_AVX512VBMI2,
20566 PTA_VPCLMULQDQ, PTA_RDPID, PTA_AVX512BITALG): New.
20567 (processor_target_table): Add icelake.
20568 (ix86_option_override_internal): Handle new PTAs.
20569 (get_builtin_code_for_version): Handle icelake.
20570 (M_INTEL_COREI7_ICELAKE): New.
20571 (fold_builtin_cpu): Handle icelake.
20572 * config/i386/i386.h (TARGET_ICELAKE, PROCESSOR_ICELAKE): New.
20573 * doc/invoke.texi: Add -march=icelake.
20575 2018-02-02 Julia Koval <julia.koval@intel.com>
20577 * config/i386/i386.c (ix86_option_override_internal): Change flags type
20578 to wide_int_bitmask.
20579 * wide-int-bitmask.h: New.
20581 2018-02-02 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
20584 * config/i386/i386.md: Replace Pmode with word_mode in
20585 builtin_setjmp_setup and builtin_longjmp to support x32.
20587 2018-02-01 Peter Bergner <bergner@vnet.ibm.com>
20591 * config/rs6000/driver-rs6000.c: #include "diagnostic.h".
20593 (rs6000_supported_cpu_names): New static variable.
20594 (linux_cpu_translation_table): Likewise.
20595 (elf_platform) <cpu>: Define new static variable and use it.
20596 Translate kernel AT_PLATFORM name to canonical name if needed.
20597 Error if platform name is unknown.
20599 2018-02-01 Aldy Hernandez <aldyh@redhat.com>
20602 * config/pa/predicates.md (base14_operand): Handle E_VOIDmode.
20604 2018-02-01 Jeff Law <law@redhat.com>
20607 * config/i386/i386.c (release_scratch_register_on_entry): Add new
20608 OFFSET and RELEASE_VIA_POP arguments. Use SP+OFFSET to restore
20609 the scratch if RELEASE_VIA_POP is false.
20610 (ix86_adjust_stack_and_probe_stack_clash): Un-constify SIZE.
20611 If we have to save a temporary register, decrement SIZE appropriately.
20612 Pass new arguments to release_scratch_register_on_entry.
20613 (ix86_adjust_stack_and_probe): Likewise.
20614 (ix86_emit_probe_stack_range): Pass new arguments to
20615 release_scratch_register_on_entry.
20617 2018-02-01 Uros Bizjak <ubizjak@gmail.com>
20619 PR rtl-optimization/84157
20620 * combine.c (change_zero_ext): Use REG_P predicate in
20621 front of HARD_REGISTER_P predicate.
20623 2018-02-01 Georg-Johann Lay <avr@gjlay.de>
20625 * config/avr/avr.c (avr_option_override): Move disabling of
20626 -fdelete-null-pointer-checks to...
20627 * common/config/avr/avr-common.c (avr_option_optimization_table):
20630 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
20632 PR tree-optimization/81635
20633 * tree-data-ref.c (split_constant_offset_1): For types that
20634 wrap on overflow, try to use range info to prove that wrapping
20637 2018-02-01 Renlin Li <renlin.li@arm.com>
20640 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
20641 TAILCALL_ADDR_REGS.
20642 (aarch64_register_move_cost): Likewise.
20643 * config/aarch64/aarch64.h (reg_class): Rename CALLER_SAVE_REGS to
20644 TAILCALL_ADDR_REGS.
20645 (REG_CLASS_NAMES): Likewise.
20646 (REG_CLASS_CONTENTS): Rename CALLER_SAVE_REGS to
20647 TAILCALL_ADDR_REGS. Remove IP registers.
20648 * config/aarch64/aarch64.md (Ucs): Update register constraint.
20650 2018-02-01 Richard Biener <rguenther@suse.de>
20652 * domwalk.h (dom_walker::dom_walker): Add additional constructor
20653 for specifying RPO order and allow NULL for that.
20654 * domwalk.c (dom_walker::dom_walker): Likewise.
20655 (dom_walker::walk): Handle NULL RPO order.
20656 * tree-into-ssa.c (rewrite_dom_walker): Do not walk dom children
20658 (rewrite_update_dom_walker): Likewise.
20659 (mark_def_dom_walker): Likewise.
20661 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
20663 * config/aarch64/aarch64-protos.h (aarch64_split_sve_subreg_move)
20664 (aarch64_maybe_expand_sve_subreg_move): Declare.
20665 * config/aarch64/aarch64.md (UNSPEC_REV_SUBREG): New unspec.
20666 * config/aarch64/predicates.md (aarch64_any_register_operand): New
20668 * config/aarch64/aarch64-sve.md (mov<mode>): Optimize subreg moves
20669 that are semantically a reverse operation.
20670 (*aarch64_sve_mov<mode>_subreg_be): New pattern.
20671 * config/aarch64/aarch64.c (aarch64_maybe_expand_sve_subreg_move):
20672 (aarch64_replace_reg_mode, aarch64_split_sve_subreg_move): New
20674 (aarch64_can_change_mode_class): For big-endian, forbid changes
20675 between two SVE modes if they have different element sizes.
20677 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
20679 * config/aarch64/aarch64.c (aarch64_expand_sve_const_vector): Prefer
20680 the TImode handling for big-endian targets.
20682 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
20684 * config/aarch64/aarch64-sve.md (sve_ld1rq): Replace with...
20685 (*sve_ld1rq<Vesize>): ... this new pattern. Handle all element sizes,
20687 * config/aarch64/aarch64.c (aarch64_expand_sve_widened_duplicate):
20688 Remove BSWAP handing for big-endian targets and use the form of
20689 LD1RQ appropariate for the mode.
20691 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
20693 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Handle
20694 all CONST_VECTOR_DUPLICATE_P vectors, not just those with a single
20695 duplicated element.
20697 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
20700 * config/aarch64/aarch64.c (aarch64_secondary_reload): Tighten
20701 check for operands that need to go through aarch64_sve_reload_be.
20703 2018-02-01 Jakub Jelinek <jakub@redhat.com>
20705 PR tree-optimization/81661
20706 PR tree-optimization/84117
20707 * tree-eh.h (rewrite_to_non_trapping_overflow): Declare.
20708 * tree-eh.c: Include gimplify.h.
20709 (find_trapping_overflow, replace_trapping_overflow,
20710 rewrite_to_non_trapping_overflow): New functions.
20711 * tree-vect-loop.c: Include tree-eh.h.
20712 (vect_get_loop_niters): Use rewrite_to_non_trapping_overflow.
20713 * tree-data-ref.c: Include tree-eh.h.
20714 (get_segment_min_max): Use rewrite_to_non_trapping_overflow.
20716 2018-01-31 Uros Bizjak <ubizjak@gmail.com>
20718 PR rtl-optimization/84123
20719 * combine.c (change_zero_ext): Check if hard register satisfies
20720 can_change_dest_mode before calling gen_lowpart_SUBREG.
20722 2018-01-31 Vladimir Makarov <vmakarov@redhat.com>
20725 * ira.c (ira_init_register_move_cost): Remove assert.
20727 2018-01-31 Eric Botcazou <ebotcazou@adacore.com>
20729 PR rtl-optimization/84071
20730 * doc/tm.texi.in (WORD_REGISTER_OPERATIONS): Add explicit case.
20731 * doc/tm.texi: Regenerate.
20733 2018-01-31 Richard Biener <rguenther@suse.de>
20735 PR tree-optimization/84132
20736 * tree-data-ref.c (analyze_miv_subscript): Properly
20737 check whether evolution_function_is_affine_multivariate_p
20738 before calling gcd_of_steps_may_divide_p.
20740 2018-01-31 Julia Koval <julia.koval@intel.com>
20743 * config/i386/i386.c (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
20744 * config/i386/i386.md (rdpid_rex64) New.
20745 (rdpid): Make 32bit only.
20747 2018-01-29 Aldy Hernandez <aldyh@redhat.com>
20750 * tree-pretty-print.c (dump_generic_node): Handle a TYPE_NAME with
20751 an IDENTIFIER_NODE for FUNCTION_TYPE's.
20753 2018-01-31 Eric Botcazou <ebotcazou@adacore.com>
20756 2018-01-12 Eric Botcazou <ebotcazou@adacore.com>
20758 * config/sparc/sparc.md (vxworks_load_got): Set the GOT register.
20760 2018-01-31 Eric Botcazou <ebotcazou@adacore.com>
20762 PR rtl-optimization/84071
20763 * combine.c (record_dead_and_set_regs_1): Record the source unmodified
20764 for a paradoxical SUBREG on a WORD_REGISTER_OPERATIONS target.
20766 2018-01-31 Claudiu Zissulescu <claziss@synopsys.com>
20768 * config/arc/arc.c (arc_handle_aux_attribute): New function.
20769 (arc_attribute_table): Add 'aux' attribute.
20770 (arc_in_small_data_p): Consider aux like variables.
20771 (arc_is_aux_reg_p): New function.
20772 (arc_asm_output_aligned_decl_local): Ignore 'aux' like variables.
20773 (arc_get_aux_arg): New function.
20774 (prepare_move_operands): Handle aux-register access.
20775 (arc_handle_aux_attribute): New function.
20776 * doc/extend.texi (ARC Variable attributes): Add subsection.
20778 2018-01-31 Claudiu Zissulescu <claziss@synopsys.com>
20780 * config/arc/arc-protos.h (arc_is_uncached_mem_p): Function proto.
20781 * config/arc/arc.c (arc_handle_uncached_attribute): New function.
20782 (arc_attribute_table): Add 'uncached' attribute.
20783 (arc_print_operand): Print '.di' flag for uncached memory
20785 (arc_in_small_data_p): Do not consider for small data the uncached
20787 (arc_is_uncached_mem_p): New function.
20788 * config/arc/predicates.md (compact_store_memory_operand): Check
20789 for uncached memory accesses.
20790 (nonvol_nonimm_operand): Likewise.
20791 * doc/extend.texi (ARC Type Attribute): New subsection.
20793 2018-01-31 Jakub Jelinek <jakub@redhat.com>
20796 * common.opt (falign-functions=, falign-jumps=, falign-labels=,
20797 falign-loops=): Add Optimization flag.
20799 2018-01-30 Jeff Law <law@redhat.com>
20802 * i386.c (ix86_adjust_stack_and_probe_stack_clash): New argument
20803 INT_REGISTERS_SAVED. Check it prior to calling
20804 get_scratch_register_on_entry.
20805 (ix86_adjust_stack_and_probe): Similarly.
20806 (ix86_emit_probe_stack_range): Similarly.
20807 (ix86_expand_prologue): Corresponding changes.
20809 2018-01-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
20812 * config/sol2.h (STARTFILE_ARCH_SPEC): Use -std=c*,
20813 -std=iso9899:199409 instead of -pedantic to select values-Xc.o.
20815 2018-01-30 Vladimir Makarov <vmakarov@redhat.com>
20818 * lra-constraints.c (curr_insn_transform): Process AND in the
20821 2018-01-30 Jakub Jelinek <jakub@redhat.com>
20823 PR rtl-optimization/83986
20824 * sched-deps.c (sched_analyze_insn): For frame related insns, add anti
20825 dependence against last_pending_memory_flush in addition to
20826 pending_jump_insns.
20828 2018-01-30 Alexandre Oliva <aoliva@redhat.com>
20830 PR tree-optimization/81611
20831 * tree-ssa-dom.c (simple_iv_increment_p): Skip intervening
20834 2018-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
20837 * config/rs6000/rs6000.c (rs6000_internal_arg_pointer): Only return
20840 2018-01-30 Richard Biener <rguenther@suse.de>
20841 Jakub Jelinek <jakub@redhat.com>
20843 PR tree-optimization/84111
20844 * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely_1): Skip
20845 inner loops added during recursion, as they don't have up-to-date
20848 2018-01-30 Jan Hubicka <hubicka@ucw.cz>
20851 * ipa-inline.c (can_inline_edge_p): Break out late tests to...
20852 (can_inline_edge_by_limits_p): ... here.
20853 (can_early_inline_edge_p, check_callers,
20854 update_caller_keys, update_callee_keys, recursive_inlining,
20855 add_new_edges_to_heap, speculation_useful_p,
20856 inline_small_functions,
20857 inline_small_functions, flatten_function,
20858 inline_to_all_callers_1): Update.
20860 2018-01-30 Jan Hubicka <hubicka@ucw.cz>
20862 * profile-count.c (profile_count::combine_with_ipa_count): Handle
20865 2018-01-30 Richard Biener <rguenther@suse.de>
20867 PR tree-optimization/83008
20868 * tree-vect-slp.c (vect_analyze_slp_cost_1): Properly cost
20869 invariant and constant vector uses in stmts when they need
20870 more than one stmt.
20872 2018-01-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
20875 * configure.ac (gcc_cv_as_shf_merge): Disable on Solaris 10/x86.
20876 * configure: Regenerate.
20878 2018-01-30 Richard Sandiford <richard.sandiford@linaro.org>
20880 * config/aarch64/aarch64-sve.md (*vec_extract<mode><Vel>_0): New
20882 (*vec_extract<mode><Vel>_v128): Require a nonzero lane number.
20883 Use gen_rtx_REG rather than gen_lowpart.
20885 2018-01-30 Richard Sandiford <richard.sandiford@linaro.org>
20887 * lra-constraints.c (match_reload): Use subreg_lowpart_offset
20888 rather than 0 when creating partial subregs.
20890 2018-01-30 Richard Sandiford <richard.sandiford@linaro.org>
20892 * vec-perm-indices.c (vec_perm_indices::series_p): Give examples
20895 2018-01-29 Michael Meissner <meissner@linux.vnet.ibm.com>
20898 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): If DFmode
20899 and SFmode can go in Altivec registers (-mcpu=power7 for DFmode,
20900 -mcpu=power8 for SFmode) don't set the PRE_INCDEC or PRE_MODIFY
20901 flags. This restores the settings used before the 2017-07-24.
20902 Turning off pre increment/decrement/modify allows IVOPTS to
20903 optimize DF/SF loops where the index is an int.
20905 2018-01-29 Richard Biener <rguenther@suse.de>
20906 Kelvin Nilsen <kelvin@gcc.gnu.org>
20909 * tree-vect-stmts.c (vectorizable_call): Don't call
20910 targetm.vectorize_builtin_md_vectorized_function if callee is
20913 2018-01-22 Carl Love <cel@us.ibm.com>
20915 * doc/extend.tex: Fix typo in second arg in
20916 __builtin_bcdadd_{lt|eq|gt|ov} and __builtin_bcdsub_{lt|eq|gt|ov}.
20918 2018-01-29 Richard Biener <rguenther@suse.de>
20920 PR tree-optimization/84086
20921 * tree-ssanames.c: Include cfgloop.h and tree-scalar-evolution.h.
20922 (flush_ssaname_freelist): When SSA names were released reset
20923 the SCEV hash table.
20925 2018-01-29 Richard Biener <rguenther@suse.de>
20927 PR tree-optimization/84057
20928 * tree-ssa-loop-ivcanon.c (unloop_loops): Deal with already
20929 removed paths when removing edges.
20931 2018-01-27 H.J. Lu <hongjiu.lu@intel.com>
20933 * doc/invoke.texi: Replace -mfunction-return==@var{choice} with
20934 -mfunction-return=@var{choice}.
20936 2018-01-27 Bernd Edlinger <bernd.edlinger@hotmail.de>
20938 PR diagnostic/84034
20939 * diagnostic-show-locus.c (get_line_width_without_trailing_whitespace):
20940 Handle CR like TAB.
20941 (layout::print_source_line): Likewise.
20942 (test_get_line_width_without_trailing_whitespace): Add test cases.
20944 2018-01-27 Jakub Jelinek <jakub@redhat.com>
20946 PR middle-end/84040
20947 * sched-deps.c (sched_macro_fuse_insns): Return immediately for
20950 2018-01-26 Jim Wilson <jimw@sifive.com>
20952 * config/riscv/riscv.h (MAX_FIXED_MODE_SIZE): New.
20954 * config/riscv/elf.h (LIB_SPEC): Don't include -lgloss when nosys.specs
20957 2018-01-26 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20959 * config/aarch64/aarch64.md: Add peepholes for CMP + SUB -> SUBS
20960 and CMP + SUB-immediate -> SUBS.
20962 2018-01-26 Martin Sebor <msebor@redhat.com>
20964 PR tree-optimization/83896
20965 * tree-ssa-strlen.c (get_string_len): Rename...
20966 (get_string_cst_length): ...to this. Return HOST_WIDE_INT.
20967 Avoid assuming length is constant.
20968 (handle_char_store): Use HOST_WIDE_INT for string length.
20970 2018-01-26 Uros Bizjak <ubizjak@gmail.com>
20973 * config/i386/i386.md (*andndi3_doubleword): Add earlyclobber
20974 to (=&r,r,rm) alternative. Add (=r,0,rm) and (=r,r,0) alternatives.
20976 2018-01-26 Richard Biener <rguenther@suse.de>
20978 PR rtl-optimization/84003
20979 * dse.c (record_store): Only record redundant stores when
20980 the earlier store aliases at least all accesses the later one does.
20982 2018-01-26 Jakub Jelinek <jakub@redhat.com>
20984 PR rtl-optimization/83985
20985 * dce.c (deletable_insn_p): Return false for separate shrink wrapping
20986 REG_CFA_RESTORE insns.
20987 (delete_unmarked_insns): Don't ignore separate shrink wrapping
20988 REG_CFA_RESTORE insns here.
20991 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Don't
20992 use SSA_NAME_VAR as base for SSA_NAMEs with non-NULL SSA_NAME_VAR.
20994 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
20996 * config/arc/arc-arch.h (arc_tune_attr): Add ARC_TUNE_CORE_3.
20997 * config/arc/arc.c (arc_sched_issue_rate): Use ARC_TUNE_... .
20998 (arc_init): Likewise.
20999 (arc_override_options): Likewise.
21000 (arc_file_start): Choose Tag_ARC_CPU_variation based on arc_tune
21002 (hwloop_fail): Use TARGET_DBNZ when we want to check for dbnz insn
21004 * config/arc/arc.h (TARGET_DBNZ): Define.
21005 * config/arc/arc.md (attr tune): Add core_3, use ARC_TUNE_... to
21006 properly set the tune attribute.
21007 (dbnz): Use TARGET_DBNZ guard.
21008 * config/arc/arc.opt (mtune): Add core3 option.
21010 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
21012 * config/arc/arc.c (arc_delegitimize_address_0): Refactored to
21013 recognize new pic like addresses.
21014 (arc_delegitimize_address): Clean up.
21016 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
21018 * config/arc/arc-arches.def: Option mrf16 valid for all
21020 * config/arc/arc-c.def (__ARC_RF16__): New predefined macro.
21021 * config/arc/arc-cpus.def (em_mini): New cpu with rf16 on.
21022 * config/arc/arc-options.def (FL_RF16): Add mrf16 option.
21023 * config/arc/arc-tables.opt: Regenerate.
21024 * config/arc/arc.c (arc_conditional_register_usage): Handle
21025 reduced register file case.
21026 (arc_file_start): Set must have build attributes.
21027 * config/arc/arc.h (MAX_ARC_PARM_REGS): Conditional define using
21028 mrf16 option value.
21029 * config/arc/arc.opt (mrf16): Add new option.
21030 * config/arc/elf.h (ATTRIBUTE_PCS): Define.
21031 * config/arc/genmultilib.awk: Handle new mrf16 option.
21032 * config/arc/linux.h (ATTRIBUTE_PCS): Define.
21033 * config/arc/t-multilib: Regenerate.
21034 * doc/invoke.texi (ARC Options): Document mrf16 option.
21036 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
21038 * config/arc/arc-protos.h: Add arc_is_secure_call_p proto.
21039 * config/arc/arc.c (arc_handle_secure_attribute): New function.
21040 (arc_attribute_table): Add 'secure_call' attribute.
21041 (arc_print_operand): Print secure call operand.
21042 (arc_function_ok_for_sibcall): Don't optimize tail calls when
21044 (arc_is_secure_call_p): New function. * config/arc/arc.md
21045 (call_i): Add support for sjli instruction.
21046 (call_value_i): Likewise.
21047 * config/arc/constraints.md (Csc): New constraint.
21049 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
21050 John Eric Martin <John.Martin@emmicro-us.com>
21052 * config/arc/arc-protos.h: Add arc_is_jli_call_p proto.
21053 * config/arc/arc.c (_arc_jli_section): New struct.
21054 (arc_jli_section): New type.
21055 (rc_jli_sections): New static variable.
21056 (arc_handle_jli_attribute): New function.
21057 (arc_attribute_table): Add jli_always and jli_fixed attribute.
21058 (arc_file_end): New function.
21059 (TARGET_ASM_FILE_END): Define.
21060 (arc_print_operand): Reuse 'S' letter for JLI output instruction.
21061 (arc_add_jli_section): New function.
21062 (jli_call_scan): Likewise.
21063 (arc_reorg): Call jli_call_scan.
21064 (arc_output_addsi): Remove 'S' from printing asm operand.
21065 (arc_is_jli_call_p): New function.
21066 * config/arc/arc.md (movqi_insn): Remove 'S' from printing asm
21068 (movhi_insn): Likewise.
21069 (movsi_insn): Likewise.
21070 (movsi_set_cc_insn): Likewise.
21071 (loadqi_update): Likewise.
21072 (load_zeroextendqisi_update): Likewise.
21073 (load_signextendqisi_update): Likewise.
21074 (loadhi_update): Likewise.
21075 (load_zeroextendhisi_update): Likewise.
21076 (load_signextendhisi_update): Likewise.
21077 (loadsi_update): Likewise.
21078 (loadsf_update): Likewise.
21079 (movsicc_insn): Likewise.
21080 (bset_insn): Likewise.
21081 (bxor_insn): Likewise.
21082 (bclr_insn): Likewise.
21083 (bmsk_insn): Likewise.
21084 (bicsi3_insn): Likewise.
21085 (cmpsi_cc_c_insn): Likewise.
21086 (movsi_ne): Likewise.
21087 (movsi_cond_exec): Likewise.
21088 (clrsbsi2): Likewise.
21089 (norm_f): Likewise.
21099 (call_i): Remove 'S' asm letter, add jli instruction.
21100 (call_value_i): Likewise.
21101 * config/arc/arc.op (mjli-always): New option.
21102 * config/arc/constraints.md (Cji): New constraint.
21103 * config/arc/fpx.md (addsf3_fpx): Remove 'S' from printing asm
21105 (subsf3_fpx): Likewise.
21106 (mulsf3_fpx): Likewise.
21107 * config/arc/simdext.md (vendrec_insn): Remove 'S' from printing
21109 * doc/extend.texi (ARC): Document 'jli-always' and 'jli-fixed'
21110 function attrbutes.
21111 * doc/invoke.texi (ARC): Document mjli-always option.
21113 2018-01-26 Sebastian Perta <sebastian.perta@renesas.com>
21115 * config/rl78/rl78.c (rl78_addsi3_internal): If operand 2 is const
21116 avoid addition with 0 and use incw and decw where possible.
21118 2018-01-26 Richard Biener <rguenther@suse.de>
21120 PR tree-optimization/81082
21121 * fold-const.c (fold_plusminus_mult_expr): Do not perform the
21122 association if it requires casting to unsigned.
21123 * match.pd ((A * C) +- (B * C) -> (A+-B)): New patterns derived
21124 from fold_plusminus_mult_expr to catch important cases late when
21125 range info is available.
21127 2018-01-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
21129 * config/i386/sol2.h (USE_HIDDEN_LINKONCE): Remove.
21130 * configure.ac (hidden_linkonce): New test.
21131 * configure: Regenerate.
21132 * config.in: Regenerate.
21134 2018-01-26 Julia Koval <julia.koval@intel.com>
21136 * config/i386/avx512bitalgintrin.h (_mm512_bitshuffle_epi64_mask,
21137 _mm512_mask_bitshuffle_epi64_mask, _mm256_bitshuffle_epi64_mask,
21138 _mm256_mask_bitshuffle_epi64_mask, _mm_bitshuffle_epi64_mask,
21139 _mm_mask_bitshuffle_epi64_mask): Fix type.
21140 * config/i386/i386-builtin-types.def (UHI_FTYPE_V2DI_V2DI_UHI,
21141 USI_FTYPE_V4DI_V4DI_USI): Remove.
21142 * config/i386/i386-builtin.def (__builtin_ia32_vpshufbitqmb512_mask,
21143 __builtin_ia32_vpshufbitqmb256_mask,
21144 __builtin_ia32_vpshufbitqmb128_mask): Fix types.
21145 * config/i386/i386.c (ix86_expand_args_builtin): Remove old types.
21146 * config/i386/sse.md (VI1_AVX512VLBW): Change types.
21148 2018-01-26 Alan Modra <amodra@gmail.com>
21151 * config/rs6000/rs6000-p8swap.c (rtx_is_swappable_p): Exclude
21152 UNSPEC_VBPERMQ. Sort other unspecs.
21154 2018-01-25 David Edelsohn <dje.gcc@gmail.com>
21156 * doc/invoke.texi (PowerPC Options): Document 'native' cpu type.
21158 2018-01-25 Jan Hubicka <hubicka@ucw.cz>
21160 PR middle-end/83055
21161 * predict.c (drop_profile): Do not push/pop cfun; update also
21163 (handle_missing_profiles): Fix logic looking for zero profiles.
21165 2018-01-25 Jakub Jelinek <jakub@redhat.com>
21167 PR middle-end/83977
21168 * ipa-fnsummary.c (compute_fn_summary): Clear can_change_signature
21169 on functions with #pragma omp declare simd or functions with simd
21171 * omp-simd-clone.c (expand_simd_clones): Revert 2018-01-24 change.
21172 * config/i386/i386.c (ix86_simd_clone_compute_vecsize_and_simdlen):
21173 Remove trailing \n from warning_at calls.
21175 2018-01-25 Tom de Vries <tom@codesourcery.com>
21178 * config/nvptx/nvptx.c (nvptx_single): Add exit insn after noreturn call
21179 for neutered workers.
21181 2018-01-24 Joseph Myers <joseph@codesourcery.com>
21184 * config/m68k/m68k.c (m68k_promote_function_mode): New function.
21185 (TARGET_PROMOTE_FUNCTION_MODE): New macro.
21187 2018-01-24 Jeff Law <law@redhat.com>
21190 * i386.c (get_probe_interval): Move to earlier point.
21191 (ix86_compute_frame_layout): If -fstack-clash-protection and
21192 the frame is larger than the probe interval, then use pushes
21193 to save registers rather than reg->mem moves.
21194 (ix86_expand_prologue): Remove conditional for int_registers_saved
21197 2018-01-24 Vladimir Makarov <vmakarov@redhat.com>
21200 * ira-build.c (setup_min_max_allocno_live_range_point): Set up
21201 min/max for never referenced object.
21203 2018-01-24 Jakub Jelinek <jakub@redhat.com>
21205 PR middle-end/83977
21206 * tree.c (free_lang_data_in_decl): Don't clear DECL_ABSTRACT_ORIGIN
21208 * omp-low.c (create_omp_child_function): Remove "omp declare simd"
21209 attributes from DECL_ATTRIBUTES (decl) without affecting
21210 DECL_ATTRIBUTES (current_function_decl).
21211 * omp-simd-clone.c (expand_simd_clones): Ignore DECL_ARTIFICIAL
21212 functions with non-NULL DECL_ABSTRACT_ORIGIN.
21214 2018-01-24 Richard Sandiford <richard.sandiford@linaro.org>
21216 PR tree-optimization/83979
21217 * fold-const.c (fold_comparison): Use constant_boolean_node
21218 instead of boolean_{true,false}_node.
21220 2018-01-24 Jan Hubicka <hubicka@ucw.cz>
21222 * ipa-profile.c (ipa_propagate_frequency_1): Fix logic skipping calls
21225 2018-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21227 * config/rs6000/rs6000.md (*call_indirect_nonlocal_sysv<mode>):
21228 Simplify the clause that sets the length attribute.
21229 (*call_value_indirect_nonlocal_sysv<mode>): Likewise.
21230 (*sibcall_nonlocal_sysv<mode>): Clean up code block; simplify the
21231 clause that sets the length attribute.
21232 (*sibcall_value_nonlocal_sysv<mode>): Likewise.
21234 2018-01-24 Tom de Vries <tom@codesourcery.com>
21237 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG_2): Define to 1.
21238 (nvptx_pc_set, nvptx_condjump_label): New function. Copy from jump.c.
21239 Add strict parameter.
21240 (prevent_branch_around_nothing): Insert dummy insn between branch to
21241 label and label with no ptx insn inbetween.
21242 * config/nvptx/nvptx.md (define_insn "fake_nop"): New insn.
21244 2018-01-24 Tom de Vries <tom@codesourcery.com>
21247 * config/nvptx/nvptx.c (nvptx_single): Add exit insn after noreturn call
21248 for neutered threads in warp.
21249 * config/nvptx/nvptx.md (define_insn "exit"): New insn.
21251 2018-01-24 Richard Biener <rguenther@suse.de>
21253 PR tree-optimization/83176
21254 * tree-chrec.c (chrec_fold_plus_1): Handle (signed T){(T) .. }
21257 2018-01-24 Richard Biener <rguenther@suse.de>
21259 PR tree-optimization/82819
21260 * graphite-isl-ast-to-gimple.c (binary_op_to_tree): Avoid
21261 code generating pluses that are no-ops in the target precision.
21263 2018-01-24 Richard Biener <rguenther@suse.de>
21265 PR middle-end/84000
21266 * tree-cfg.c (replace_loop_annotate): Handle annot_expr_parallel_kind.
21268 2018-01-23 Jan Hubicka <hubicka@ucw.cz>
21270 * cfgcleanup.c (try_crossjump_to_edge): Use combine_with_count
21271 to merge probabilities.
21272 * predict.c (probably_never_executed): Also mark as cold functions
21273 with global 0 profile and guessed local profile.
21274 * profile-count.c (profile_probability::combine_with_count): New
21276 * profile-count.h (profile_probability::operator*,
21277 profile_probability::operator*=, profile_probability::operator/,
21278 profile_probability::operator/=): Reduce precision to adjusted
21279 and set value to guessed on contradictory divisions.
21280 (profile_probability::combine_with_freq): Remove.
21281 (profile_probability::combine_wiht_count): Declare.
21282 (profile_count::force_nonzero):: Set to adjusted.
21283 (profile_count::probability_in):: Set quality to adjusted.
21284 * tree-ssa-tail-merge.c (replace_block_by): Use
21285 combine_with_count.
21287 2018-01-23 Andrew Waterman <andrew@sifive.com>
21288 Jim Wilson <jimw@sifive.com>
21290 * config/riscv/riscv.c (riscv_stack_boundary): New.
21291 (riscv_option_override): Set riscv_stack_boundary. Handle
21292 riscv_preferred_stack_boundary_arg.
21293 * config/riscv/riscv.h (MIN_STACK_BOUNDARY, ABI_STACK_BOUNDARY): New.
21294 (BIGGEST_ALIGNMENT): Set to STACK_BOUNDARY.
21295 (STACK_BOUNDARY): Set to riscv_stack_boundary.
21296 (RISCV_STACK_ALIGN): Use STACK_BOUNDARY.
21297 * config/riscv/riscv.opt (mpreferred-stack-boundary): New.
21298 * doc/invoke.tex (RISC-V Options): Add -mpreferred-stack-boundary.
21300 2018-01-23 H.J. Lu <hongjiu.lu@intel.com>
21303 * config/i386/i386.c (ix86_expand_prologue): Use cost reference
21304 of struct ix86_frame.
21305 (ix86_expand_epilogue): Likewise. Add a local variable for
21306 the reg_save_offset field in struct ix86_frame.
21308 2018-01-23 Bin Cheng <bin.cheng@arm.com>
21310 PR tree-optimization/82604
21311 * tree-loop-distribution.c (enum partition_kind): New enum item
21312 PKIND_PARTIAL_MEMSET.
21313 (partition_builtin_p): Support above new enum item.
21314 (generate_code_for_partition): Ditto.
21315 (compute_access_range): Differentiate cases that equality can be
21316 proven at all loops, the innermost loops or no loops.
21317 (classify_builtin_st, classify_builtin_ldst): Adjust call to above
21318 function. Set PKIND_PARTIAL_MEMSET for partition appropriately.
21319 (finalize_partitions, distribute_loop): Don't fuse partition of
21320 PKIND_PARTIAL_MEMSET kind when distributing 3-level loop nest.
21321 (prepare_perfect_loop_nest): Distribute 3-level loop nest only if
21322 parloop is enabled.
21324 2018-01-23 Martin Liska <mliska@suse.cz>
21326 * predict.def (PRED_INDIR_CALL): Set probability to PROB_EVEN in
21327 order to ignore the predictor.
21328 (PRED_POLYMORPHIC_CALL): Likewise.
21329 (PRED_RECURSIVE_CALL): Likewise.
21331 2018-01-23 Martin Liska <mliska@suse.cz>
21333 * tree-profile.c (tree_profiling): Print function header to
21334 aware reader which function we are working on.
21335 * value-prof.c (gimple_find_values_to_profile): Do not print
21336 not interesting value histograms.
21338 2018-01-23 Martin Liska <mliska@suse.cz>
21340 * profile-count.h (enum profile_quality): Add
21341 profile_uninitialized as the first value. Do not number values
21342 as they are zero based.
21343 (profile_count::verify): Update sanity check.
21344 (profile_probability::verify): Likewise.
21346 2018-01-23 Nathan Sidwell <nathan@acm.org>
21348 * doc/invoke.texi (ffor-scope): Deprecate.
21350 2018-01-23 David Malcolm <dmalcolm@redhat.com>
21352 PR tree-optimization/83510
21353 * domwalk.c (set_all_edges_as_executable): New function.
21354 (dom_walker::dom_walker): Convert bool param
21355 "skip_unreachable_blocks" to enum reachability. Move setup of
21356 edge flags to set_all_edges_as_executable and only do it when
21357 reachability is REACHABLE_BLOCKS.
21358 * domwalk.h (enum dom_walker::reachability): New enum.
21359 (dom_walker::dom_walker): Convert bool param
21360 "skip_unreachable_blocks" to enum reachability.
21361 (set_all_edges_as_executable): New decl.
21362 * graphite-scop-detection.c (gather_bbs::gather_bbs): Convert
21363 from false for "skip_unreachable_blocks" to ALL_BLOCKS for
21365 * tree-ssa-dom.c (dom_opt_dom_walker::dom_opt_dom_walker): Likewise,
21366 but converting true to REACHABLE_BLOCKS.
21367 * tree-ssa-sccvn.c (sccvn_dom_walker::sccvn_dom_walker): Likewise.
21369 (check_array_bounds_dom_walker::check_array_bounds_dom_walker):
21370 Likewise, but converting it to REACHABLE_BLOCKS_PRESERVING_FLAGS.
21371 (vrp_dom_walker::vrp_dom_walker): Likewise, but converting it to
21373 (vrp_prop::vrp_finalize): Call set_all_edges_as_executable
21374 if check_all_array_refs will be called.
21376 2018-01-23 David Malcolm <dmalcolm@redhat.com>
21378 * tree.c (selftest::test_location_wrappers): Add more test
21381 2018-01-23 David Malcolm <dmalcolm@redhat.com>
21383 * sbitmap.c (selftest::test_set_range): Fix memory leaks.
21384 (selftest::test_bit_in_range): Likewise.
21386 2018-01-23 Richard Sandiford <richard.sandiford@linaro.org>
21389 * doc/sourcebuild.texi (vect_float): Say that the selector
21390 only describes the situation when -funsafe-math-optimizations is on.
21391 (vect_float_strict): Document.
21393 2018-01-23 Richard Sandiford <richard.sandiford@linaro.org>
21395 PR tree-optimization/83965
21396 * tree-vect-patterns.c (vect_reassociating_reduction_p): New function.
21397 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern): Use it
21398 instead of checking only for a reduction.
21399 (vect_recog_widen_sum_pattern): Likewise.
21401 2018-01-23 Jan Hubicka <hubicka@ucw.cz>
21403 * predict.c (probably_never_executed): Only use precise profile info.
21404 (compute_function_frequency): Skip after inlining hack since we now
21405 have quality checking.
21407 2018-01-23 Jan Hubicka <hubicka@ucw.cz>
21409 * profile-count.h (profile_probability::very_unlikely,
21410 profile_probability::unlikely, profile_probability::even): Set
21411 precision to guessed.
21413 2018-01-23 Richard Biener <rguenther@suse.de>
21415 PR tree-optimization/83963
21416 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
21417 Properly terminate dominator walk when crossing the exit edge not
21418 when visiting its source block.
21420 2018-01-23 Jakub Jelinek <jakub@redhat.com>
21423 * tree.c (maybe_wrap_with_location): Use NON_LVALUE_EXPR rather than
21424 VIEW_CONVERT_EXPR to wrap CONST_DECLs.
21426 2018-01-22 Jakub Jelinek <jakub@redhat.com>
21428 PR tree-optimization/83957
21429 * omp-expand.c (expand_omp_for_generic): Ignore virtual PHIs. Remove
21430 semicolon after for body surrounded by braces.
21432 PR tree-optimization/83081
21433 * profile-count.h (profile_probability::split): New method.
21434 * dojump.c (do_jump_1) <case TRUTH_ANDIF_EXPR, case TRUTH_ORIF_EXPR>:
21435 Use profile_probability::split.
21436 (do_compare_rtx_and_jump): Fix adjustment of probabilities
21437 when splitting a single conditional jump into 2.
21439 2018-01-22 David Malcolm <dmalcolm@redhat.com>
21441 PR tree-optimization/69452
21442 * tree-ssa-loop-im.c (class move_computations_dom_walker): Remove
21445 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
21447 * config/rl78/rl78-expand.md (bswaphi2): New define_expand.
21448 * config/rl78/rl78-virt.md (*bswaphi2_virt): New define_insn.
21449 * config/rl78/rl78-real.md (*bswaphi2_real): New define_insn.
21451 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
21453 * config/rl78/rl78-protos.h (rl78_split_movdi): New function
21455 * config/rl78/rl78.md (movdi): New define_expand.
21456 * config/rl78/rl78.c (rl78_split_movdi): New function.
21458 2018-01-22 Michael Meissner <meissner@linux.vnet.ibm.com>
21461 * config/rs6000/rs6000-protos.h (rs6000_split_signbit): Delete,
21463 * config/rs6000/rs6000.c (rs6000_split_signbit): Likewise.
21464 * config/rs6000/rs6000.md (signbit<mode>2): Change code for IEEE
21465 128-bit to produce an UNSPEC move to get the double word with the
21466 signbit and then a shift directly to do signbit.
21467 (signbit<mode>2_dm): Replace old IEEE 128-bit signbit
21468 implementation with a new version that just does either a direct
21469 move or a regular move. Move memory interface to separate insns.
21470 Move insns so they are next to the expander.
21471 (signbit<mode>2_dm_mem_be): New combiner insns to combine load
21472 with signbit move. Split big and little endian case.
21473 (signbit<mode>2_dm_mem_le): Likewise.
21474 (signbit<mode>2_dm_<su>ext): Delete, no longer used.
21475 (signbit<mode>2_dm2): Likewise.
21477 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
21479 * config/rl78/rl78.md (anddi3): New define_expand.
21481 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
21483 * config/rl78/rl78.md (umindi3): New define_expand.
21485 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
21487 * config/rl78/rl78.md (smindi3): New define_expand.
21489 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
21491 * config/rl78/rl78.md (smaxdi3): New define_expand.
21493 2018-01-22 Carl Love <cel@us.ibm.com>
21495 * config/rs6000/rs6000-builtin.def (ST_ELEMREV_V1TI, LD_ELEMREV_V1TI,
21496 LVX_V1TI): Add macro expansion.
21497 * config/rs6000/rs6000-c.c (altivec_builtin_types): Add argument
21498 definitions for VSX_BUILTIN_VEC_XST_BE, VSX_BUILTIN_VEC_ST,
21499 VSX_BUILTIN_VEC_XL, LD_ELEMREV_V1TI builtins.
21500 * config/rs6000/rs6000-p8swap.c (insn_is_swappable_p);
21501 Change check to determine if the instruction is a byte reversing
21502 entry. Fix typo in comment.
21503 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case entry
21504 for VSX_BUILTIN_ST_ELEMREV_V1TI and VSX_BUILTIN_LD_ELEMREV_V1TI.
21505 Add def_builtin calls for new builtins.
21506 * config/rs6000/vsx.md (vsx_st_elemrev_v1ti, vsx_ld_elemrev_v1ti):
21507 Add define_insn expansion.
21509 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
21511 * config/rl78/rl78.md (umaxdi3): New define_expand.
21513 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
21515 * config/rl78/rl78.c (rl78_note_reg_set): Fixed dead reg check
21516 for non-QImode registers.
21518 2018-01-22 Richard Biener <rguenther@suse.de>
21520 PR tree-optimization/83963
21521 * graphite-scop-detection.c (scop_detection::get_sese): Delay
21522 including the loop exit block.
21523 (scop_detection::merge_sese): Likewise.
21524 (scop_detection::add_scop): Do it here instead.
21526 2018-01-22 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21528 * doc/sourcebuild.texi (arm_softfloat): Document.
21530 2018-01-21 John David Anglin <danglin@gcc.gnu.org>
21533 * config/pa/pa.c (pa_function_ok_for_sibcall): Use
21534 targetm.binds_local_p instead of TREE_PUBLIC to check local binding.
21535 Move TARGET_PORTABLE_RUNTIME check after TARGET_64BIT check.
21537 2018-01-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21538 David Edelsohn <dje.gcc@gmail.com>
21541 * config/rs6000/rs6000.md (*call_indirect_nonlocal_sysv<mode>):
21542 Change "crset eq" to "crset 2".
21543 (*call_value_indirect_nonlocal_sysv<mode>): Likewise.
21544 (*call_indirect_aix<mode>_nospec): Likewise.
21545 (*call_value_indirect_aix<mode>_nospec): Likewise.
21546 (*call_indirect_elfv2<mode>_nospec): Likewise.
21547 (*call_value_indirect_elfv2<mode>_nospec): Likewise.
21548 (*sibcall_nonlocal_sysv<mode>): Change "crset eq" to "crset 2";
21549 change assembly output from . to $.
21550 (*sibcall_value_nonlocal_sysv<mode>): Likewise.
21551 (indirect_jump<mode>_nospec): Change assembly output from . to $.
21552 (*tablejump<mode>_internal1_nospec): Likewise.
21554 2018-01-21 Oleg Endo <olegendo@gcc.gnu.org>
21557 * config/sh/sh_optimize_sett_clrt.cc:
21558 Use INCLUDE_ALGORITHM and INCLUDE_VECTOR instead of direct includes.
21560 2018-01-20 Richard Sandiford <richard.sandiford@linaro.org>
21562 PR tree-optimization/83940
21563 * tree-vect-stmts.c (vect_truncate_gather_scatter_offset): Set
21564 offset_dt to vect_constant_def rather than vect_unknown_def_type.
21565 (vect_check_load_store_mask): Add a mask_dt_out parameter and
21566 use it to pass back the definition type.
21567 (vect_check_store_rhs): Likewise rhs_dt_out.
21568 (vect_build_gather_load_calls): Add a mask_dt argument and use
21569 it instead of a call to vect_is_simple_use.
21570 (vectorizable_store): Update calls to vect_check_load_store_mask
21571 and vect_check_store_rhs. Use the dt returned by the latter instead
21572 of scatter_src_dt. Use the cached mask_dt and gs_info.offset_dt
21573 instead of calls to vect_is_simple_use. Pass the scalar rather
21574 than the vector operand to vect_is_simple_use when handling
21575 second and subsequent copies of an rhs value.
21576 (vectorizable_load): Update calls to vect_check_load_store_mask
21577 and vect_build_gather_load_calls. Use the cached mask_dt and
21578 gs_info.offset_dt instead of calls to vect_is_simple_use.
21580 2018-01-20 Jakub Jelinek <jakub@redhat.com>
21582 PR middle-end/83945
21583 * tree-emutls.c: Include gimplify.h.
21584 (lower_emutls_2): New function.
21585 (lower_emutls_1): If ADDR_EXPR is a gimple invariant and walk_tree
21586 with lower_emutls_2 callback finds some TLS decl in it, unshare_expr
21587 it before further processing.
21590 * simplify-rtx.c (simplify_binary_operation_1) <case UMOD>: Use
21591 UINTVAL (trueop1) instead of INTVAL (op1).
21593 2018-01-19 Jakub Jelinek <jakub@redhat.com>
21597 * dwarf2cfi.c (DEFAULT_INCOMING_FRAME_SP_OFFSET): Define to
21598 INCOMING_FRAME_SP_OFFSET if not defined.
21599 (scan_trace): Add ENTRY argument. If true and
21600 DEFAULT_INCOMING_FRAME_SP_OFFSET != INCOMING_FRAME_SP_OFFSET,
21601 emit a note to adjust the CFA offset.
21602 (create_cfi_notes): Adjust scan_trace callers.
21603 (create_cie_data): Use DEFAULT_INCOMING_FRAME_SP_OFFSET rather than
21604 INCOMING_FRAME_SP_OFFSET in the CIE.
21605 * config/i386/i386.h (DEFAULT_INCOMING_FRAME_SP_OFFSET): Define.
21606 * config/stormy16/stormy16.h (DEFAULT_INCOMING_FRAME_SP_OFFSET):
21608 * doc/tm.texi.in (DEFAULT_INCOMING_FRAME_SP_OFFSET): Document.
21609 * doc/tm.texi: Regenerated.
21611 2018-01-19 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21613 PR rtl-optimization/83147
21614 * lra-constraints.c (remove_inheritance_pseudos): Use
21615 lra_substitute_pseudo_within_insn.
21617 2018-01-19 Tom de Vries <tom@codesourcery.com>
21618 Cesar Philippidis <cesar@codesourcery.com>
21621 * config/nvptx/nvptx.c (nvptx_single): Fix jit workaround.
21623 2018-01-19 Cesar Philippidis <cesar@codesourcery.com>
21626 * config/nvptx/nvptx.c (output_init_frag): Don't use generic address
21627 spaces for function labels.
21629 2018-01-19 Martin Liska <mliska@suse.cz>
21631 * predict.def (PRED_LOOP_EXIT): Change from 85 to 89.
21632 (PRED_LOOP_EXIT_WITH_RECURSION): Change from 72 to 78.
21633 (PRED_LOOP_EXTRA_EXIT): Change from 83 to 67.
21634 (PRED_OPCODE_POSITIVE): Change from 64 to 59.
21635 (PRED_TREE_OPCODE_POSITIVE): Change from 64 to 59.
21636 (PRED_CONST_RETURN): Change from 69 to 65.
21637 (PRED_NULL_RETURN): Change from 91 to 71.
21638 (PRED_LOOP_IV_COMPARE_GUESS): Change from 98 to 64.
21639 (PRED_LOOP_GUARD): Change from 66 to 73.
21641 2018-01-19 Martin Liska <mliska@suse.cz>
21643 * predict.c (predict_insn_def): Add new assert.
21644 (struct branch_predictor): Change type to signed integer.
21645 (test_prediction_value_range): Amend test to cover
21646 PROB_UNINITIALIZED.
21647 * predict.def (PRED_LOOP_ITERATIONS): Use the new constant.
21648 (PRED_LOOP_ITERATIONS_GUESSED): Likewise.
21649 (PRED_LOOP_ITERATIONS_MAX): Likewise.
21650 (PRED_LOOP_IV_COMPARE): Likewise.
21651 * predict.h (PROB_UNINITIALIZED): Define new constant.
21653 2018-01-19 Martin Liska <mliska@suse.cz>
21655 * predict.c (dump_prediction): Add new format for
21656 analyze_brprob.py script which is enabled with -details
21658 * profile-count.h (precise_p): New function.
21660 2018-01-19 Richard Sandiford <richard.sandiford@linaro.org>
21662 PR tree-optimization/83922
21663 * tree-vect-loop.c (vect_verify_full_masking): Return false if
21664 there are no statements that need masking.
21665 (vect_active_double_reduction_p): New function.
21666 (vect_analyze_loop_operations): Use it when handling phis that
21667 are not in the loop header.
21669 2018-01-19 Richard Sandiford <richard.sandiford@linaro.org>
21671 PR tree-optimization/83914
21672 * tree-vect-loop.c (vectorizable_induction): Don't convert
21673 init_expr or apply the peeling adjustment for inductions
21674 that are nested within the vectorized loop.
21676 2018-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21678 * config/arm/thumb2.md (*thumb2_negsi2_short): Use RSB mnemonic
21681 2018-01-18 Jakub Jelinek <jakub@redhat.com>
21685 * function.h (gimplify_parameters): Add gimple_seq * argument.
21686 * function.c: Include gimple.h and options.h.
21687 (gimplify_parameters): Add cleanup argument, add CLOBBER stmts
21688 for the added local temporaries if needed.
21689 * gimplify.c (gimplify_body): Adjust gimplify_parameters caller,
21690 if there are any parameter cleanups, wrap whole body into a
21691 try/finally with the cleanups.
21693 2018-01-18 Wilco Dijkstra <wdijkstr@arm.com>
21696 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
21697 Use GET_MODE_CLASS for scalar floating point.
21699 2018-01-18 Jan Hubicka <hubicka@ucw.cz>
21703 * cgraphclones.c (cgraph_node::create_version_clone_with_body):
21704 Fix call of call_cgraph_insertion_hooks.
21706 2018-01-18 Martin Sebor <msebor@redhat.com>
21708 * doc/invoke.texi (-Wclass-memaccess): Tweak text.
21710 2018-01-18 Jan Hubicka <hubicka@ucw.cz>
21713 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Update edge
21716 2018-01-18 Boris Kolpackov <boris@codesynthesis.com>
21719 * common.opt: (-ffile-prefix-map): New option.
21720 * opts.c (common_handle_option): Defer it.
21721 * opts-global.c (handle_common_deferred_options): Handle it.
21722 * debug.h (remap_debug_filename, add_debug_prefix_map): Move to...
21723 * file-prefix-map.h: New file.
21724 (remap_debug_filename, add_debug_prefix_map): ...here.
21725 (add_macro_prefix_map, add_file_prefix_map, remap_macro_filename): New.
21726 * final.c (debug_prefix_map, add_debug_prefix_map
21727 remap_debug_filename): Move to...
21728 * file-prefix-map.c: New file.
21729 (file_prefix_map, add_prefix_map, remap_filename) ...here and rename,
21730 generalize, get rid of alloca(), use strrchr() instead of strchr().
21731 (add_macro_prefix_map, add_debug_prefix_map, add_file_prefix_map):
21732 Implement in terms of add_prefix_map().
21733 (remap_macro_filename, remap_debug_filename): Implement in term of
21735 * Makefile.in (OBJS, PLUGIN_HEADERS): Add new files.
21736 * builtins.c (fold_builtin_FILE): Call remap_macro_filename().
21737 * dbxout.c: Include file-prefix-map.h.
21738 * varasm.c: Likewise.
21739 * vmsdbgout.c: Likewise.
21740 * xcoffout.c: Likewise.
21741 * dwarf2out.c: Likewise plus omit new options from DW_AT_producer.
21742 * doc/cppopts.texi (-fmacro-prefix-map): Document.
21743 * doc/invoke.texi (-ffile-prefix-map): Document.
21744 (-fdebug-prefix-map): Update description.
21746 2018-01-18 Martin Liska <mliska@suse.cz>
21748 * config/i386/i386.c (indirect_thunk_name): Document that also
21750 (output_indirect_thunk): Document why both instructions
21751 (pause and lfence) are generated.
21753 2018-01-18 Richard Biener <rguenther@suse.de>
21755 PR tree-optimization/83887
21756 * graphite-scop-detection.c
21757 (scop_detection::get_nearest_dom_with_single_entry): Remove.
21758 (scop_detection::get_nearest_pdom_with_single_exit): Likewise.
21759 (scop_detection::merge_sese): Re-implement with a flood-fill
21760 algorithm that properly finds a SESE region if it exists.
21762 2018-01-18 Jakub Jelinek <jakub@redhat.com>
21765 * match.pd ((P + A) - P, P - (P + A), (P + A) - (P + B)): For
21766 pointer_diff optimizations use view_convert instead of convert.
21768 2018-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21770 * config/rs6000/rs6000.md (*call_indirect_nonlocal_sysv<mode>):
21771 Generate different code for -mno-speculate-indirect-jumps.
21772 (*call_value_indirect_nonlocal_sysv<mode>): Likewise.
21773 (*call_indirect_aix<mode>): Disable for
21774 -mno-speculate-indirect-jumps.
21775 (*call_indirect_aix<mode>_nospec): New define_insn.
21776 (*call_value_indirect_aix<mode>): Disable for
21777 -mno-speculate-indirect-jumps.
21778 (*call_value_indirect_aix<mode>_nospec): New define_insn.
21779 (*sibcall_nonlocal_sysv<mode>): Generate different code for
21780 -mno-speculate-indirect-jumps.
21781 (*sibcall_value_nonlocal_sysv<mode>): Likewise.
21783 2018-01-17 Michael Meissner <meissner@linux.vnet.ibm.com>
21785 * config/rs6000/rs6000.c (rs6000_emit_move): If we load or store a
21786 long double type, set the flags for noting the default long double
21787 type, even if we don't pass or return a long double type.
21789 2018-01-17 Jan Hubicka <hubicka@ucw.cz>
21792 * ipa-inline.c (flatten_function): Do not overwrite final inlining
21795 2018-01-17 Will Schmidt <will_schmidt@vnet.ibm.com>
21797 * config/rs6000/rs6000.c (rs6000_gimple_builtin): Add gimple folding
21798 support for merge[hl].
21799 (fold_mergehl_helper): New helper function.
21800 (tree-vector-builder.h): New #include for tree_vector_builder usage.
21801 * config/rs6000/altivec.md (altivec_vmrghw_direct): Add xxmrghw insn.
21802 (altivec_vmrglw_direct): Add xxmrglw insn.
21804 2018-01-17 Andrew Waterman <andrew@sifive.com>
21806 * config/riscv/riscv.c (riscv_conditional_register_usage): If
21807 UNITS_PER_FP_ARG is 0, set call_used_regs to 1 for all FP regs.
21809 2018-01-17 David Malcolm <dmalcolm@redhat.com>
21812 * ipa-devirt.c (add_type_duplicate): When comparing memory layout,
21813 call the lto_location_cache before reading the
21814 DECL_SOURCE_LOCATION of the types.
21816 2018-01-17 Wilco Dijkstra <wdijkstr@arm.com>
21817 Richard Sandiford <richard.sandiford@linaro.org>
21819 * config/aarch64/aarch64.md (movti_aarch64): Use Uti constraint.
21820 * config/aarch64/aarch64.c (aarch64_mov128_immediate): New function.
21821 (aarch64_legitimate_constant_p): Just support CONST_DOUBLE
21822 SF/DF/TF mode to avoid creating illegal CONST_WIDE_INT immediates.
21823 * config/aarch64/aarch64-protos.h (aarch64_mov128_immediate):
21825 * config/aarch64/constraints.md (aarch64_movti_operand):
21827 * config/aarch64/predicates.md (Uti): Add new constraint.
21829 2018-01-17 Carl Love <cel@us.ibm.com>
21831 * config/rs6000/vsx.md (define_expand xl_len_r,
21832 define_expand stxvl, define_expand *stxvl): Add match_dup argument.
21833 (define_insn): Add, match_dup 1 argument to define_insn stxvll and
21835 (define_expand, define_insn): Move the shift left from the
21836 define_insn to the define_expand for lxvl and stxvl instructions.
21837 * config/rs6000/rs6000-builtin.def (BU_P9V_64BIT_VSX_2): Change LXVL
21838 and XL_LEN_R definitions to PURE.
21840 2018-01-17 Uros Bizjak <ubizjak@gmail.com>
21842 * config/i386/i386.c (indirect_thunk_name): Declare regno
21843 as unsigned int. Compare regno with INVALID_REGNUM.
21844 (output_indirect_thunk): Ditto.
21845 (output_indirect_thunk_function): Ditto.
21846 (ix86_code_end): Declare regno as unsigned int. Use INVALID_REGNUM
21847 in the call to output_indirect_thunk_function.
21849 2018-01-17 Richard Sandiford <richard.sandiford@linaro.org>
21851 PR middle-end/83884
21852 * expr.c (expand_expr_real_1): Use the size of GET_MODE (op0)
21853 rather than the size of inner_type to determine the stack slot size
21854 when handling VIEW_CONVERT_EXPRs on strict-alignment targets.
21856 2018-01-16 Sebastian Peryt <sebastian.peryt@intel.com>
21859 * config/i386/i386.c (ix86_option_override_internal): Add PTA_RDRND
21862 2018-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
21864 * config.gcc (powerpc*-linux*-*): Add support for 64-bit little
21865 endian Linux systems to optionally enable multilibs for selecting
21866 the long double type if the user configured an explicit type.
21867 * config/rs6000/rs6000.h (TARGET_IEEEQUAD_MULTILIB): Indicate we
21868 have no long double multilibs if not defined.
21869 * config/rs6000/rs6000.c (rs6000_option_override_internal): Do not
21870 warn if the user used -mabi={ieee,ibm}longdouble and we built
21871 multilibs for long double.
21872 * config/rs6000/linux64.h (MULTILIB_DEFAULTS_IEEE): Define as the
21873 appropriate multilib option.
21874 (MULTILIB_DEFAULTS): Add MULTILIB_DEFAULTS_IEEE to the default
21876 * config/rs6000/t-ldouble-linux64le-ibm: New configuration files
21877 for building long double multilibs.
21878 * config/rs6000/t-ldouble-linux64le-ieee: Likewise.
21880 2018-01-16 John David Anglin <danglin@gcc.gnu.org>
21882 * config.gcc (hppa*-*-linux*): Change callee copies ABI to caller
21885 * config/pa.h (MALLOC_ABI_ALIGNMENT): Set 32-bit alignment default to
21887 * config/pa/pa32-linux.h (MALLOC_ABI_ALIGNMENT): Set alignment to
21890 * config/pa/som.h (ASM_DECLARE_FUNCTION_NAME): Cleanup type and mode
21893 * config/pa/pa.c (pa_function_arg_size): Apply CEIL to GET_MODE_SIZE
21896 2018-01-16 Eric Botcazou <ebotcazou@adacore.com>
21898 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): For an
21899 ADDR_EXPR, do not count the offset of a COMPONENT_REF twice.
21901 2018-01-16 Kelvin Nilsen <kelvin@gcc.gnu.org>
21903 * config/rs6000/rs6000-p8swap.c (rs6000_gen_stvx): Generate
21904 different rtl trees depending on TARGET_64BIT.
21905 (rs6000_gen_lvx): Likewise.
21907 2018-01-16 Eric Botcazou <ebotcazou@adacore.com>
21909 * config/visium/visium.md (nop): Tweak comment.
21910 (hazard_nop): Likewise.
21912 2018-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21914 * config/rs6000/rs6000.c (rs6000_opt_vars): Add entry for
21915 -mspeculate-indirect-jumps.
21916 * config/rs6000/rs6000.md (*call_indirect_elfv2<mode>): Disable
21917 for -mno-speculate-indirect-jumps.
21918 (*call_indirect_elfv2<mode>_nospec): New define_insn.
21919 (*call_value_indirect_elfv2<mode>): Disable for
21920 -mno-speculate-indirect-jumps.
21921 (*call_value_indirect_elfv2<mode>_nospec): New define_insn.
21922 (indirect_jump): Emit different RTL for
21923 -mno-speculate-indirect-jumps.
21924 (*indirect_jump<mode>): Disable for
21925 -mno-speculate-indirect-jumps.
21926 (*indirect_jump<mode>_nospec): New define_insn.
21927 (tablejump): Emit different RTL for
21928 -mno-speculate-indirect-jumps.
21929 (tablejumpsi): Disable for -mno-speculate-indirect-jumps.
21930 (tablejumpsi_nospec): New define_expand.
21931 (tablejumpdi): Disable for -mno-speculate-indirect-jumps.
21932 (tablejumpdi_nospec): New define_expand.
21933 (*tablejump<mode>_internal1): Disable for
21934 -mno-speculate-indirect-jumps.
21935 (*tablejump<mode>_internal1_nospec): New define_insn.
21936 * config/rs6000/rs6000.opt (mspeculate-indirect-jumps): New
21939 2018-01-16 Artyom Skrobov tyomitch@gmail.com
21941 * caller-save.c (insert_save): Drop unnecessary parameter. All
21944 2018-01-16 Jakub Jelinek <jakub@redhat.com>
21945 Richard Biener <rguenth@suse.de>
21948 * gimplify.c (gimplify_one_sizepos): For is_gimple_constant (expr)
21949 return early, inline manually is_gimple_sizepos. Make sure if we
21950 call gimplify_expr we don't end up with a gimple constant.
21951 * tree.c (variably_modified_type_p): Don't return true for
21952 is_gimple_constant (_t). Inline manually is_gimple_sizepos.
21953 * gimplify.h (is_gimple_sizepos): Remove.
21955 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
21957 PR tree-optimization/83857
21958 * tree-vect-loop.c (vect_analyze_loop_operations): Don't call
21959 vectorizable_live_operation for pure SLP statements.
21960 (vectorizable_live_operation): Handle PHIs.
21962 2018-01-16 Richard Biener <rguenther@suse.de>
21964 PR tree-optimization/83867
21965 * tree-vect-stmts.c (vect_transform_stmt): Precompute
21966 nested_in_vect_loop_p since the scalar stmt may get invalidated.
21968 2018-01-16 Jakub Jelinek <jakub@redhat.com>
21971 * stor-layout.c (handle_warn_if_not_align): Use byte_position and
21972 multiple_of_p instead of unchecked tree_to_uhwi and UHWI check.
21973 If off is not INTEGER_CST, issue a may not be aligned warning
21974 rather than isn't aligned. Use isn%'t rather than isn't.
21975 * fold-const.c (multiple_of_p) <case BIT_AND_EXPR>: Don't fall through
21977 <case MULT_EXPR>: Improve the case when bottom and one of the
21978 MULT_EXPR operands are INTEGER_CSTs and bottom is multiple of that
21979 operand, in that case check if the other operand is multiple of
21980 bottom divided by the INTEGER_CST operand.
21982 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
21985 * config/pa/pa.h (FUNCTION_ARG_SIZE): Delete.
21986 * config/pa/pa-protos.h (pa_function_arg_size): Declare.
21987 * config/pa/som.h (ASM_DECLARE_FUNCTION_NAME): Use
21988 pa_function_arg_size instead of FUNCTION_ARG_SIZE.
21989 * config/pa/pa.c (pa_function_arg_advance): Likewise.
21990 (pa_function_arg, pa_arg_partial_bytes): Likewise.
21991 (pa_function_arg_size): New function.
21993 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
21995 * fold-const.c (fold_ternary_loc): Construct the vec_perm_indices
21996 in a separate statement.
21998 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
22000 PR tree-optimization/83847
22001 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Don't
22002 group gathers and scatters.
22004 2018-01-16 Jakub Jelinek <jakub@redhat.com>
22006 PR rtl-optimization/86620
22007 * params.def (max-sched-ready-insns): Bump minimum value to 1.
22009 PR rtl-optimization/83213
22010 * recog.c (peep2_attempt): Copy over CROSSING_JUMP_P from peepinsn
22011 to last if both are JUMP_INSNs.
22013 PR tree-optimization/83843
22014 * gimple-ssa-store-merging.c
22015 (imm_store_chain_info::output_merged_store): Handle bit_not_p on
22016 store_immediate_info for bswap/nop orig_stores.
22018 2018-01-15 Andrew Waterman <andrew@sifive.com>
22020 * config/riscv/riscv.c (riscv_rtx_costs) <MULT>: Increase cost if
22022 <UDIV>: Increase cost if !TARGET_DIV.
22024 2018-01-15 Segher Boessenkool <segher@kernel.crashing.org>
22026 * config/rs6000/rs6000.md (define_attr "type"): Remove delayed_cr.
22027 (define_attr "cr_logical_3op"): New.
22028 (cceq_ior_compare): Adjust.
22029 (cceq_ior_compare_complement): Adjust.
22030 (*cceq_rev_compare): Adjust.
22031 * config/rs6000/rs6000.c (rs6000_adjust_cost): Adjust.
22032 (is_cracked_insn): Adjust.
22033 (insn_must_be_first_in_group): Adjust.
22034 * config/rs6000/40x.md: Adjust.
22035 * config/rs6000/440.md: Adjust.
22036 * config/rs6000/476.md: Adjust.
22037 * config/rs6000/601.md: Adjust.
22038 * config/rs6000/603.md: Adjust.
22039 * config/rs6000/6xx.md: Adjust.
22040 * config/rs6000/7450.md: Adjust.
22041 * config/rs6000/7xx.md: Adjust.
22042 * config/rs6000/8540.md: Adjust.
22043 * config/rs6000/cell.md: Adjust.
22044 * config/rs6000/e300c2c3.md: Adjust.
22045 * config/rs6000/e500mc.md: Adjust.
22046 * config/rs6000/e500mc64.md: Adjust.
22047 * config/rs6000/e5500.md: Adjust.
22048 * config/rs6000/e6500.md: Adjust.
22049 * config/rs6000/mpc.md: Adjust.
22050 * config/rs6000/power4.md: Adjust.
22051 * config/rs6000/power5.md: Adjust.
22052 * config/rs6000/power6.md: Adjust.
22053 * config/rs6000/power7.md: Adjust.
22054 * config/rs6000/power8.md: Adjust.
22055 * config/rs6000/power9.md: Adjust.
22056 * config/rs6000/rs64.md: Adjust.
22057 * config/rs6000/titan.md: Adjust.
22059 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
22061 * config/i386/predicates.md (indirect_branch_operand): Rewrite
22062 ix86_indirect_branch_register logic.
22064 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
22066 * config/i386/constraints.md (Bs): Update
22067 ix86_indirect_branch_register check. Don't check
22068 ix86_indirect_branch_register with GOT_memory_operand.
22070 * config/i386/predicates.md (GOT_memory_operand): Don't check
22071 ix86_indirect_branch_register here.
22072 (GOT32_symbol_operand): Likewise.
22074 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
22076 * config/i386/predicates.md (constant_call_address_operand):
22077 Rewrite ix86_indirect_branch_register logic.
22078 (sibcall_insn_operand): Likewise.
22080 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
22082 * config/i386/constraints.md (Bs): Replace
22083 ix86_indirect_branch_thunk_register with
22084 ix86_indirect_branch_register.
22086 * config/i386/i386.md (indirect_jump): Likewise.
22087 (tablejump): Likewise.
22088 (*sibcall_memory): Likewise.
22089 (*sibcall_value_memory): Likewise.
22090 Peepholes of indirect call and jump via memory: Likewise.
22091 * config/i386/i386.opt: Likewise.
22092 * config/i386/predicates.md (indirect_branch_operand): Likewise.
22093 (GOT_memory_operand): Likewise.
22094 (call_insn_operand): Likewise.
22095 (sibcall_insn_operand): Likewise.
22096 (GOT32_symbol_operand): Likewise.
22098 2018-01-15 Jakub Jelinek <jakub@redhat.com>
22100 PR middle-end/83837
22101 * omp-expand.c (expand_omp_atomic_pipeline): Use loaded_val
22102 type rather than type addr's type points to.
22103 (expand_omp_atomic_mutex): Likewise.
22104 (expand_omp_atomic): Likewise.
22106 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
22109 * config/i386/i386.c (output_indirect_thunk_function): Use
22110 ASM_OUTPUT_LABEL, instead of ASM_OUTPUT_DEF, for TARGET_MACHO
22111 for __x86_return_thunk.
22113 2018-01-15 Richard Biener <rguenther@suse.de>
22115 PR middle-end/83850
22116 * expmed.c (extract_bit_field_1): Fix typo.
22118 2018-01-15 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22121 * config/arm/iterators.md (VF): New mode iterator.
22122 * config/arm/neon.md (neon_vabd<mode>_2): Use the above.
22123 Remove integer-related logic from pattern.
22124 (neon_vabd<mode>_3): Likewise.
22126 2018-01-15 Jakub Jelinek <jakub@redhat.com>
22128 PR middle-end/82694
22129 * common.opt (fstrict-overflow): No longer an alias.
22130 (fwrapv-pointer): New option.
22131 * tree.h (TYPE_OVERFLOW_WRAPS, TYPE_OVERFLOW_UNDEFINED): Define
22132 also for pointer types based on flag_wrapv_pointer.
22133 * opts.c (common_handle_option) <case OPT_fstrict_overflow>: Set
22134 opts->x_flag_wrap[pv] to !value, clear opts->x_flag_trapv if
22135 opts->x_flag_wrapv got set.
22136 * fold-const.c (fold_comparison, fold_binary_loc): Revert 2017-08-01
22137 changes, just use TYPE_OVERFLOW_UNDEFINED on pointer type instead of
22138 POINTER_TYPE_OVERFLOW_UNDEFINED.
22139 * match.pd: Likewise in address comparison pattern.
22140 * doc/invoke.texi: Document -fwrapv and -fstrict-overflow.
22142 2018-01-15 Richard Biener <rguenther@suse.de>
22145 * tree.c (free_lang_data_in_type): Always unlink TYPE_DECLs
22146 from TYPE_FIELDS. Free TYPE_BINFO if not used by devirtualization.
22147 Reset type names to their identifier if their TYPE_DECL doesn't
22148 have linkage (and thus is used for ODR and devirt).
22149 (save_debug_info_for_decl): Remove.
22150 (save_debug_info_for_type): Likewise.
22151 (add_tree_to_fld_list): Adjust.
22152 * tree-pretty-print.c (dump_generic_node): Make dumping of
22153 type names more robust.
22155 2018-01-15 Richard Biener <rguenther@suse.de>
22157 * BASE-VER: Bump to 8.0.1.
22159 2018-01-14 Martin Sebor <msebor@redhat.com>
22162 * builtins.c (check_access): Avoid warning when the no-warning bit
22165 2018-01-14 Cory Fields <cory-nospam-@coryfields.com>
22167 * tree-ssa-loop-im.c (sort_bbs_in_loop_postorder_cmp): Stabilize sort.
22168 * ira-color (allocno_hard_regs_compare): Likewise.
22170 2018-01-14 Nathan Rossi <nathan@nathanrossi.com>
22173 * config/microblaze/microblaze.c (microblaze_asm_output_ident):
22174 Use .pushsection/.popsection.
22176 2018-01-14 Martin Sebor <msebor@redhat.com>
22179 * doc/invoke.texi (-Wlass-memaccess): Document suppression by casting.
22181 2018-01-14 Jakub Jelinek <jakub@redhat.com>
22183 * config.gcc (i[34567]86-*-*): Remove one duplicate gfniintrin.h
22184 entry from extra_headers.
22185 (x86_64-*-*): Remove two duplicate gfniintrin.h entries from
22186 extra_headers, make the list bitwise identical to the i?86-*-* one.
22188 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
22190 * config/i386/i386.c (ix86_set_indirect_branch_type): Disallow
22191 -mcmodel=large with -mindirect-branch=thunk,
22192 -mindirect-branch=thunk-extern, -mfunction-return=thunk and
22193 -mfunction-return=thunk-extern.
22194 * doc/invoke.texi: Document -mcmodel=large is incompatible with
22195 -mindirect-branch=thunk, -mindirect-branch=thunk-extern,
22196 -mfunction-return=thunk and -mfunction-return=thunk-extern.
22198 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
22200 * config/i386/i386.c (print_reg): Print the name of the full
22201 integer register without '%'.
22202 (ix86_print_operand): Handle 'V'.
22203 * doc/extend.texi: Document 'V' modifier.
22205 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
22207 * config/i386/constraints.md (Bs): Disallow memory operand for
22208 -mindirect-branch-register.
22210 * config/i386/predicates.md (indirect_branch_operand): Likewise.
22211 (GOT_memory_operand): Likewise.
22212 (call_insn_operand): Likewise.
22213 (sibcall_insn_operand): Likewise.
22214 (GOT32_symbol_operand): Likewise.
22215 * config/i386/i386.md (indirect_jump): Call convert_memory_address
22216 for -mindirect-branch-register.
22217 (tablejump): Likewise.
22218 (*sibcall_memory): Likewise.
22219 (*sibcall_value_memory): Likewise.
22220 Disallow peepholes of indirect call and jump via memory for
22221 -mindirect-branch-register.
22222 (*call_pop): Replace m with Bw.
22223 (*call_value_pop): Likewise.
22224 (*sibcall_pop_memory): Replace m with Bs.
22225 * config/i386/i386.opt (mindirect-branch-register): New option.
22226 * doc/invoke.texi: Document -mindirect-branch-register option.
22228 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
22230 * config/i386/i386-protos.h (ix86_output_function_return): New.
22231 * config/i386/i386.c (ix86_set_indirect_branch_type): Also
22232 set function_return_type.
22233 (indirect_thunk_name): Add ret_p to indicate thunk for function
22235 (output_indirect_thunk_function): Pass false to
22236 indirect_thunk_name.
22237 (ix86_output_indirect_branch_via_reg): Likewise.
22238 (ix86_output_indirect_branch_via_push): Likewise.
22239 (output_indirect_thunk_function): Create alias for function
22240 return thunk if regno < 0.
22241 (ix86_output_function_return): New function.
22242 (ix86_handle_fndecl_attribute): Handle function_return.
22243 (ix86_attribute_table): Add function_return.
22244 * config/i386/i386.h (machine_function): Add
22245 function_return_type.
22246 * config/i386/i386.md (simple_return_internal): Use
22247 ix86_output_function_return.
22248 (simple_return_internal_long): Likewise.
22249 * config/i386/i386.opt (mfunction-return=): New option.
22250 (indirect_branch): Mention -mfunction-return=.
22251 * doc/extend.texi: Document function_return function attribute.
22252 * doc/invoke.texi: Document -mfunction-return= option.
22254 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
22256 * config/i386/i386-opts.h (indirect_branch): New.
22257 * config/i386/i386-protos.h (ix86_output_indirect_jmp): Likewise.
22258 * config/i386/i386.c (ix86_using_red_zone): Disallow red-zone
22259 with local indirect jump when converting indirect call and jump.
22260 (ix86_set_indirect_branch_type): New.
22261 (ix86_set_current_function): Call ix86_set_indirect_branch_type.
22262 (indirectlabelno): New.
22263 (indirect_thunk_needed): Likewise.
22264 (indirect_thunk_bnd_needed): Likewise.
22265 (indirect_thunks_used): Likewise.
22266 (indirect_thunks_bnd_used): Likewise.
22267 (INDIRECT_LABEL): Likewise.
22268 (indirect_thunk_name): Likewise.
22269 (output_indirect_thunk): Likewise.
22270 (output_indirect_thunk_function): Likewise.
22271 (ix86_output_indirect_branch_via_reg): Likewise.
22272 (ix86_output_indirect_branch_via_push): Likewise.
22273 (ix86_output_indirect_branch): Likewise.
22274 (ix86_output_indirect_jmp): Likewise.
22275 (ix86_code_end): Call output_indirect_thunk_function if needed.
22276 (ix86_output_call_insn): Call ix86_output_indirect_branch if
22278 (ix86_handle_fndecl_attribute): Handle indirect_branch.
22279 (ix86_attribute_table): Add indirect_branch.
22280 * config/i386/i386.h (machine_function): Add indirect_branch_type
22281 and has_local_indirect_jump.
22282 * config/i386/i386.md (indirect_jump): Set has_local_indirect_jump
22284 (tablejump): Likewise.
22285 (*indirect_jump): Use ix86_output_indirect_jmp.
22286 (*tablejump_1): Likewise.
22287 (simple_return_indirect_internal): Likewise.
22288 * config/i386/i386.opt (mindirect-branch=): New option.
22289 (indirect_branch): New.
22292 (thunk-inline): Likewise.
22293 (thunk-extern): Likewise.
22294 * doc/extend.texi: Document indirect_branch function attribute.
22295 * doc/invoke.texi: Document -mindirect-branch= option.
22297 2018-01-14 Jan Hubicka <hubicka@ucw.cz>
22300 * ipa-inline.c (edge_badness): Tolerate roundoff errors.
22302 2018-01-14 Richard Sandiford <richard.sandiford@linaro.org>
22304 * ipa-inline.c (want_inline_small_function_p): Return false if
22305 inlining has already failed with CIF_FINAL_ERROR.
22306 (update_caller_keys): Call want_inline_small_function_p before
22308 (update_callee_keys): Likewise.
22310 2018-01-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
22312 * config/rs6000/rs6000-p8swap.c (rs6000_sum_of_two_registers_p):
22314 (rs6000_quadword_masked_address_p): Likewise.
22315 (quad_aligned_load_p): Likewise.
22316 (quad_aligned_store_p): Likewise.
22317 (const_load_sequence_p): Add comment to describe the outer-most loop.
22318 (mimic_memory_attributes_and_flags): New function.
22319 (rs6000_gen_stvx): Likewise.
22320 (replace_swapped_aligned_store): Likewise.
22321 (rs6000_gen_lvx): Likewise.
22322 (replace_swapped_aligned_load): Likewise.
22323 (replace_swapped_load_constant): Capitalize argument name in
22324 comment describing this function.
22325 (rs6000_analyze_swaps): Add a third pass to search for vector loads
22326 and stores that access quad-word aligned addresses and replace
22327 with stvx or lvx instructions when appropriate.
22328 * config/rs6000/rs6000-protos.h (rs6000_sum_of_two_registers_p):
22329 New function prototype.
22330 (rs6000_quadword_masked_address_p): Likewise.
22331 (rs6000_gen_lvx): Likewise.
22332 (rs6000_gen_stvx): Likewise.
22333 * config/rs6000/vsx.md (*vsx_le_perm_load_<mode>): For modes
22334 VSX_D (V2DF, V2DI), modify this split to select lvx instruction
22335 when memory address is aligned.
22336 (*vsx_le_perm_load_<mode>): For modes VSX_W (V4SF, V4SI), modify
22337 this split to select lvx instruction when memory address is aligned.
22338 (*vsx_le_perm_load_v8hi): Modify this split to select lvx
22339 instruction when memory address is aligned.
22340 (*vsx_le_perm_load_v16qi): Likewise.
22341 (four unnamed splitters): Modify to select the stvx instruction
22342 when memory is aligned.
22344 2018-01-13 Jan Hubicka <hubicka@ucw.cz>
22346 * predict.c (determine_unlikely_bbs): Handle correctly BBs
22347 which appears in the queue multiple times.
22349 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
22350 Alan Hayward <alan.hayward@arm.com>
22351 David Sherwood <david.sherwood@arm.com>
22353 * tree-vectorizer.h (vec_lower_bound): New structure.
22354 (_loop_vec_info): Add check_nonzero and lower_bounds.
22355 (LOOP_VINFO_CHECK_NONZERO): New macro.
22356 (LOOP_VINFO_LOWER_BOUNDS): Likewise.
22357 (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Check lower_bounds too.
22358 * tree-data-ref.h (dr_with_seg_len): Add access_size and align
22359 fields. Make seg_len the distance travelled, not including the
22361 (dr_direction_indicator): Declare.
22362 (dr_zero_step_indicator): Likewise.
22363 (dr_known_forward_stride_p): Likewise.
22364 * tree-data-ref.c: Include stringpool.h, tree-vrp.h and
22366 (runtime_alias_check_p): Allow runtime alias checks with
22368 (operator ==): Compare access_size and align.
22369 (prune_runtime_alias_test_list): Rework for new distinction between
22370 the access_size and seg_len.
22371 (create_intersect_range_checks_index): Likewise. Cope with polynomial
22373 (get_segment_min_max): New function.
22374 (create_intersect_range_checks): Use it.
22375 (dr_step_indicator): New function.
22376 (dr_direction_indicator): Likewise.
22377 (dr_zero_step_indicator): Likewise.
22378 (dr_known_forward_stride_p): Likewise.
22379 * tree-loop-distribution.c (data_ref_segment_size): Return
22380 DR_STEP * (niters - 1).
22381 (compute_alias_check_pairs): Update call to the dr_with_seg_len
22383 * tree-vect-data-refs.c (vect_check_nonzero_value): New function.
22384 (vect_preserves_scalar_order_p): New function, split out from...
22385 (vect_analyze_data_ref_dependence): ...here. Check for zero steps.
22386 (vect_vfa_segment_size): Return DR_STEP * (length_factor - 1).
22387 (vect_vfa_access_size): New function.
22388 (vect_vfa_align): Likewise.
22389 (vect_compile_time_alias): Take access_size_a and access_b arguments.
22390 (dump_lower_bound): New function.
22391 (vect_check_lower_bound): Likewise.
22392 (vect_small_gap_p): Likewise.
22393 (vectorizable_with_step_bound_p): Likewise.
22394 (vect_prune_runtime_alias_test_list): Ignore cross-iteration
22395 depencies if the vectorization factor is 1. Convert the checks
22396 for nonzero steps into checks on the bounds of DR_STEP. Try using
22397 a bunds check for variable steps if the minimum required step is
22398 relatively small. Update calls to the dr_with_seg_len
22399 constructor and to vect_compile_time_alias.
22400 * tree-vect-loop-manip.c (vect_create_cond_for_lower_bounds): New
22402 (vect_loop_versioning): Call it.
22403 * tree-vect-loop.c (vect_analyze_loop_2): Clear LOOP_VINFO_LOWER_BOUNDS
22405 (vect_estimate_min_profitable_iters): Account for any bounds checks.
22407 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
22408 Alan Hayward <alan.hayward@arm.com>
22409 David Sherwood <david.sherwood@arm.com>
22411 * doc/sourcebuild.texi (vect_scatter_store): Document.
22412 * optabs.def (scatter_store_optab, mask_scatter_store_optab): New
22414 * doc/md.texi (scatter_store@var{m}, mask_scatter_store@var{m}):
22416 * genopinit.c (main): Add supports_vec_scatter_store and
22417 supports_vec_scatter_store_cached to target_optabs.
22418 * gimple.h (gimple_expr_type): Handle IFN_SCATTER_STORE and
22419 IFN_MASK_SCATTER_STORE.
22420 * internal-fn.def (SCATTER_STORE, MASK_SCATTER_STORE): New internal
22422 * internal-fn.h (internal_store_fn_p): Declare.
22423 (internal_fn_stored_value_index): Likewise.
22424 * internal-fn.c (scatter_store_direct): New macro.
22425 (expand_scatter_store_optab_fn): New function.
22426 (direct_scatter_store_optab_supported_p): New macro.
22427 (internal_store_fn_p): New function.
22428 (internal_gather_scatter_fn_p): Handle IFN_SCATTER_STORE and
22429 IFN_MASK_SCATTER_STORE.
22430 (internal_fn_mask_index): Likewise.
22431 (internal_fn_stored_value_index): New function.
22432 (internal_gather_scatter_fn_supported_p): Adjust operand numbers
22433 for scatter stores.
22434 * optabs-query.h (supports_vec_scatter_store_p): Declare.
22435 * optabs-query.c (supports_vec_scatter_store_p): New function.
22436 * tree-vectorizer.h (vect_get_store_rhs): Declare.
22437 * tree-vect-data-refs.c (vect_analyze_data_ref_access): Return
22438 true for scatter stores.
22439 (vect_gather_scatter_fn_p): Handle scatter stores too.
22440 (vect_check_gather_scatter): Consider using scatter stores if
22441 supports_vec_scatter_store_p.
22442 * tree-vect-patterns.c (vect_try_gather_scatter_pattern): Handle
22443 scatter stores too.
22444 * tree-vect-stmts.c (exist_non_indexing_operands_for_use_p): Use
22445 internal_fn_stored_value_index.
22446 (check_load_store_masking): Handle scatter stores too.
22447 (vect_get_store_rhs): Make public.
22448 (vectorizable_call): Use internal_store_fn_p.
22449 (vectorizable_store): Handle scatter store internal functions.
22450 (vect_transform_stmt): Compare GROUP_STORE_COUNT with GROUP_SIZE
22451 when deciding whether the end of the group has been reached.
22452 * config/aarch64/aarch64.md (UNSPEC_ST1_SCATTER): New unspec.
22453 * config/aarch64/aarch64-sve.md (scatter_store<mode>): New expander.
22454 (mask_scatter_store<mode>): New insns.
22456 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
22457 Alan Hayward <alan.hayward@arm.com>
22458 David Sherwood <david.sherwood@arm.com>
22460 * tree-vectorizer.h (vect_gather_scatter_fn_p): Declare.
22461 * tree-vect-data-refs.c (vect_gather_scatter_fn_p): Make public.
22462 * tree-vect-stmts.c (vect_truncate_gather_scatter_offset): New
22464 (vect_use_strided_gather_scatters_p): Take a masked_p argument.
22465 Use vect_truncate_gather_scatter_offset if we can't treat the
22466 operation as a normal gather load or scatter store.
22467 (get_group_load_store_type): Take the gather_scatter_info
22468 as argument. Try using a gather load or scatter store for
22469 single-element groups.
22470 (get_load_store_type): Update calls to get_group_load_store_type
22471 and vect_use_strided_gather_scatters_p.
22473 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
22474 Alan Hayward <alan.hayward@arm.com>
22475 David Sherwood <david.sherwood@arm.com>
22477 * tree-vectorizer.h (vect_create_data_ref_ptr): Take an extra
22478 optional tree argument.
22479 * tree-vect-data-refs.c (vect_check_gather_scatter): Check for
22481 (vect_create_data_ref_ptr): Take the iv_step as an optional argument,
22482 but continue to use the current value as a fallback.
22483 (bump_vector_ptr): Use operand_equal_p rather than tree_int_cst_compare
22484 to compare the updates.
22485 * tree-vect-stmts.c (vect_use_strided_gather_scatters_p): New function.
22486 (get_load_store_type): Use it when handling a strided access.
22487 (vect_get_strided_load_store_ops): New function.
22488 (vect_get_data_ptr_increment): Likewise.
22489 (vectorizable_load): Handle strided gather loads. Always pass
22490 a step to vect_create_data_ref_ptr and bump_vector_ptr.
22492 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
22493 Alan Hayward <alan.hayward@arm.com>
22494 David Sherwood <david.sherwood@arm.com>
22496 * doc/md.texi (gather_load@var{m}): Document.
22497 (mask_gather_load@var{m}): Likewise.
22498 * genopinit.c (main): Add supports_vec_gather_load and
22499 supports_vec_gather_load_cached to target_optabs.
22500 * optabs-tree.c (init_tree_optimization_optabs): Use
22501 ggc_cleared_alloc to allocate target_optabs.
22502 * optabs.def (gather_load_optab, mask_gather_laod_optab): New optabs.
22503 * internal-fn.def (GATHER_LOAD, MASK_GATHER_LOAD): New internal
22505 * internal-fn.h (internal_load_fn_p): Declare.
22506 (internal_gather_scatter_fn_p): Likewise.
22507 (internal_fn_mask_index): Likewise.
22508 (internal_gather_scatter_fn_supported_p): Likewise.
22509 * internal-fn.c (gather_load_direct): New macro.
22510 (expand_gather_load_optab_fn): New function.
22511 (direct_gather_load_optab_supported_p): New macro.
22512 (direct_internal_fn_optab): New function.
22513 (internal_load_fn_p): Likewise.
22514 (internal_gather_scatter_fn_p): Likewise.
22515 (internal_fn_mask_index): Likewise.
22516 (internal_gather_scatter_fn_supported_p): Likewise.
22517 * optabs-query.c (supports_at_least_one_mode_p): New function.
22518 (supports_vec_gather_load_p): Likewise.
22519 * optabs-query.h (supports_vec_gather_load_p): Declare.
22520 * tree-vectorizer.h (gather_scatter_info): Add ifn, element_type
22521 and memory_type field.
22522 (NUM_PATTERNS): Bump to 15.
22523 * tree-vect-data-refs.c: Include internal-fn.h.
22524 (vect_gather_scatter_fn_p): New function.
22525 (vect_describe_gather_scatter_call): Likewise.
22526 (vect_check_gather_scatter): Try using internal functions for
22527 gather loads. Recognize existing calls to a gather load function.
22528 (vect_analyze_data_refs): Consider using gather loads if
22529 supports_vec_gather_load_p.
22530 * tree-vect-patterns.c (vect_get_load_store_mask): New function.
22531 (vect_get_gather_scatter_offset_type): Likewise.
22532 (vect_convert_mask_for_vectype): Likewise.
22533 (vect_add_conversion_to_patterm): Likewise.
22534 (vect_try_gather_scatter_pattern): Likewise.
22535 (vect_recog_gather_scatter_pattern): New pattern recognizer.
22536 (vect_vect_recog_func_ptrs): Add it.
22537 * tree-vect-stmts.c (exist_non_indexing_operands_for_use_p): Use
22538 internal_fn_mask_index and internal_gather_scatter_fn_p.
22539 (check_load_store_masking): Take the gather_scatter_info as an
22540 argument and handle gather loads.
22541 (vect_get_gather_scatter_ops): New function.
22542 (vectorizable_call): Check internal_load_fn_p.
22543 (vectorizable_load): Likewise. Handle gather load internal
22545 (vectorizable_store): Update call to check_load_store_masking.
22546 * config/aarch64/aarch64.md (UNSPEC_LD1_GATHER): New unspec.
22547 * config/aarch64/iterators.md (SVE_S, SVE_D): New mode iterators.
22548 * config/aarch64/predicates.md (aarch64_gather_scale_operand_w)
22549 (aarch64_gather_scale_operand_d): New predicates.
22550 * config/aarch64/aarch64-sve.md (gather_load<mode>): New expander.
22551 (mask_gather_load<mode>): New insns.
22553 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
22554 Alan Hayward <alan.hayward@arm.com>
22555 David Sherwood <david.sherwood@arm.com>
22557 * optabs.def (fold_left_plus_optab): New optab.
22558 * doc/md.texi (fold_left_plus_@var{m}): Document.
22559 * internal-fn.def (IFN_FOLD_LEFT_PLUS): New internal function.
22560 * internal-fn.c (fold_left_direct): Define.
22561 (expand_fold_left_optab_fn): Likewise.
22562 (direct_fold_left_optab_supported_p): Likewise.
22563 * fold-const-call.c (fold_const_fold_left): New function.
22564 (fold_const_call): Use it to fold CFN_FOLD_LEFT_PLUS.
22565 * tree-parloops.c (valid_reduction_p): New function.
22566 (gather_scalar_reductions): Use it.
22567 * tree-vectorizer.h (FOLD_LEFT_REDUCTION): New vect_reduction_type.
22568 (vect_finish_replace_stmt): Declare.
22569 * tree-vect-loop.c (fold_left_reduction_fn): New function.
22570 (needs_fold_left_reduction_p): New function, split out from...
22571 (vect_is_simple_reduction): ...here. Accept reductions that
22572 forbid reassociation, but give them type FOLD_LEFT_REDUCTION.
22573 (vect_force_simple_reduction): Also store the reduction type in
22574 the assignment's STMT_VINFO_REDUC_TYPE.
22575 (vect_model_reduction_cost): Handle FOLD_LEFT_REDUCTION.
22576 (merge_with_identity): New function.
22577 (vect_expand_fold_left): Likewise.
22578 (vectorize_fold_left_reduction): Likewise.
22579 (vectorizable_reduction): Handle FOLD_LEFT_REDUCTION. Leave the
22580 scalar phi in place for it. Check for target support and reject
22581 cases that would reassociate the operation. Defer the transform
22582 phase to vectorize_fold_left_reduction.
22583 * config/aarch64/aarch64.md (UNSPEC_FADDA): New unspec.
22584 * config/aarch64/aarch64-sve.md (fold_left_plus_<mode>): New expander.
22585 (*fold_left_plus_<mode>, *pred_fold_left_plus_<mode>): New insns.
22587 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
22589 * tree-if-conv.c (predicate_mem_writes): Remove redundant
22590 call to ifc_temp_var.
22592 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
22593 Alan Hayward <alan.hayward@arm.com>
22594 David Sherwood <david.sherwood@arm.com>
22596 * target.def (legitimize_address_displacement): Take the original
22597 offset as a poly_int.
22598 * targhooks.h (default_legitimize_address_displacement): Update
22600 * targhooks.c (default_legitimize_address_displacement): Likewise.
22601 * doc/tm.texi: Regenerate.
22602 * lra-constraints.c (base_plus_disp_to_reg): Take the displacement
22603 as an argument, moving assert of ad->disp == ad->disp_term to...
22604 (process_address_1): ...here. Update calls to base_plus_disp_to_reg.
22605 Try calling targetm.legitimize_address_displacement before expanding
22606 the address rather than afterwards, and adjust for the new interface.
22607 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
22608 Match the new hook interface. Handle SVE addresses.
22609 * config/sh/sh.c (sh_legitimize_address_displacement): Make the
22610 new hook interface.
22612 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
22614 * Makefile.in (OBJS): Add early-remat.o.
22615 * target.def (select_early_remat_modes): New hook.
22616 * doc/tm.texi.in (TARGET_SELECT_EARLY_REMAT_MODES): New hook.
22617 * doc/tm.texi: Regenerate.
22618 * targhooks.h (default_select_early_remat_modes): Declare.
22619 * targhooks.c (default_select_early_remat_modes): New function.
22620 * timevar.def (TV_EARLY_REMAT): New timevar.
22621 * passes.def (pass_early_remat): New pass.
22622 * tree-pass.h (make_pass_early_remat): Declare.
22623 * early-remat.c: New file.
22624 * config/aarch64/aarch64.c (aarch64_select_early_remat_modes): New
22626 (TARGET_SELECT_EARLY_REMAT_MODES): Define.
22628 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
22629 Alan Hayward <alan.hayward@arm.com>
22630 David Sherwood <david.sherwood@arm.com>
22632 * tree-vect-loop-manip.c (vect_gen_scalar_loop_niters): Replace
22633 vfm1 with a bound_epilog parameter.
22634 (vect_do_peeling): Update calls accordingly, and move the prologue
22635 call earlier in the function. Treat the base bound_epilog as 0 for
22636 fully-masked loops and retain vf - 1 for other loops. Add 1 to
22637 this base when peeling for gaps.
22638 * tree-vect-loop.c (vect_analyze_loop_2): Allow peeling for gaps
22639 with fully-masked loops.
22640 (vect_estimate_min_profitable_iters): Handle the single peeled
22641 iteration in that case.
22643 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
22644 Alan Hayward <alan.hayward@arm.com>
22645 David Sherwood <david.sherwood@arm.com>
22647 * tree-vect-data-refs.c (vect_analyze_group_access_1): Allow
22648 single-element interleaving even if the size is not a power of 2.
22649 * tree-vect-stmts.c (get_load_store_type): Disallow elementwise
22650 accesses for single-element interleaving if the group size is
22653 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
22654 Alan Hayward <alan.hayward@arm.com>
22655 David Sherwood <david.sherwood@arm.com>
22657 * doc/md.texi (fold_extract_last_@var{m}): Document.
22658 * doc/sourcebuild.texi (vect_fold_extract_last): Likewise.
22659 * optabs.def (fold_extract_last_optab): New optab.
22660 * internal-fn.def (FOLD_EXTRACT_LAST): New internal function.
22661 * internal-fn.c (fold_extract_direct): New macro.
22662 (expand_fold_extract_optab_fn): Likewise.
22663 (direct_fold_extract_optab_supported_p): Likewise.
22664 * tree-vectorizer.h (EXTRACT_LAST_REDUCTION): New vect_reduction_type.
22665 * tree-vect-loop.c (vect_model_reduction_cost): Handle
22666 EXTRACT_LAST_REDUCTION.
22667 (get_initial_def_for_reduction): Do not create an initial vector
22668 for EXTRACT_LAST_REDUCTION reductions.
22669 (vectorizable_reduction): Leave the scalar phi in place for
22670 EXTRACT_LAST_REDUCTIONs. Try using EXTRACT_LAST_REDUCTION
22671 ahead of INTEGER_INDUC_COND_REDUCTION. Do not check for an
22672 epilogue code for EXTRACT_LAST_REDUCTION and defer the
22673 transform phase to vectorizable_condition.
22674 * tree-vect-stmts.c (vect_finish_stmt_generation_1): New function,
22676 (vect_finish_stmt_generation): ...here.
22677 (vect_finish_replace_stmt): New function.
22678 (vectorizable_condition): Handle EXTRACT_LAST_REDUCTION.
22679 * config/aarch64/aarch64-sve.md (fold_extract_last_<mode>): New
22681 * config/aarch64/aarch64.md (UNSPEC_CLASTB): New unspec.
22683 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
22684 Alan Hayward <alan.hayward@arm.com>
22685 David Sherwood <david.sherwood@arm.com>
22687 * doc/md.texi (extract_last_@var{m}): Document.
22688 * optabs.def (extract_last_optab): New optab.
22689 * internal-fn.def (EXTRACT_LAST): New internal function.
22690 * internal-fn.c (cond_unary_direct): New macro.
22691 (expand_cond_unary_optab_fn): Likewise.
22692 (direct_cond_unary_optab_supported_p): Likewise.
22693 * tree-vect-loop.c (vectorizable_live_operation): Allow fully-masked
22694 loops using EXTRACT_LAST.
22695 * config/aarch64/aarch64-sve.md (aarch64_sve_lastb<mode>): Rename to...
22696 (extract_last_<mode>): ...this optab.
22697 (vec_extract<mode><Vel>): Update accordingly.
22699 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
22700 Alan Hayward <alan.hayward@arm.com>
22701 David Sherwood <david.sherwood@arm.com>
22703 * target.def (empty_mask_is_expensive): New hook.
22704 * doc/tm.texi.in (TARGET_VECTORIZE_EMPTY_MASK_IS_EXPENSIVE): New hook.
22705 * doc/tm.texi: Regenerate.
22706 * targhooks.h (default_empty_mask_is_expensive): Declare.
22707 * targhooks.c (default_empty_mask_is_expensive): New function.
22708 * tree-vectorizer.c (vectorize_loops): Only call optimize_mask_stores
22709 if the target says that empty masks are expensive.
22710 * config/aarch64/aarch64.c (aarch64_empty_mask_is_expensive):
22712 (TARGET_VECTORIZE_EMPTY_MASK_IS_EXPENSIVE): Redefine.
22714 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
22715 Alan Hayward <alan.hayward@arm.com>
22716 David Sherwood <david.sherwood@arm.com>
22718 * tree-vectorizer.h (_loop_vec_info::mask_skip_niters): New field.
22719 (LOOP_VINFO_MASK_SKIP_NITERS): New macro.
22720 (vect_use_loop_mask_for_alignment_p): New function.
22721 (vect_prepare_for_masked_peels, vect_gen_while_not): Declare.
22722 * tree-vect-loop-manip.c (vect_set_loop_masks_directly): Add an
22723 niters_skip argument. Make sure that the first niters_skip elements
22724 of the first iteration are inactive.
22725 (vect_set_loop_condition_masked): Handle LOOP_VINFO_MASK_SKIP_NITERS.
22726 Update call to vect_set_loop_masks_directly.
22727 (get_misalign_in_elems): New function, split out from...
22728 (vect_gen_prolog_loop_niters): ...here.
22729 (vect_update_init_of_dr): Take a code argument that specifies whether
22730 the adjustment should be added or subtracted.
22731 (vect_update_init_of_drs): Likewise.
22732 (vect_prepare_for_masked_peels): New function.
22733 (vect_do_peeling): Skip prologue peeling if we're using a mask
22734 instead. Update call to vect_update_inits_of_drs.
22735 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
22737 (vect_analyze_loop_2): Allow fully-masked loops with peeling for
22738 alignment. Do not include the number of peeled iterations in
22739 the minimum threshold in that case.
22740 (vectorizable_induction): Adjust the start value down by
22741 LOOP_VINFO_MASK_SKIP_NITERS iterations.
22742 (vect_transform_loop): Call vect_prepare_for_masked_peels.
22743 Take the number of skipped iterations into account when calculating
22745 * tree-vect-stmts.c (vect_gen_while_not): New function.
22747 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
22748 Alan Hayward <alan.hayward@arm.com>
22749 David Sherwood <david.sherwood@arm.com>
22751 * doc/sourcebuild.texi (vect_fully_masked): Document.
22752 * params.def (PARAM_MIN_VECT_LOOP_BOUND): Change minimum and
22753 default value to 0.
22754 * tree-vect-loop.c (vect_analyze_loop_costing): New function,
22756 (vect_analyze_loop_2): ...here. Don't check the vectorization
22757 factor against the number of loop iterations if the loop is
22760 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
22761 Alan Hayward <alan.hayward@arm.com>
22762 David Sherwood <david.sherwood@arm.com>
22764 * tree-ssa-loop-ivopts.c (USE_ADDRESS): Split into...
22765 (USE_REF_ADDRESS, USE_PTR_ADDRESS): ...these new use types.
22766 (dump_groups): Update accordingly.
22767 (iv_use::mem_type): New member variable.
22768 (address_p): New function.
22769 (record_use): Add a mem_type argument and initialize the new
22771 (record_group_use): Add a mem_type argument. Use address_p.
22772 Remove obsolete null checks of base_object. Update call to record_use.
22773 (find_interesting_uses_op): Update call to record_group_use.
22774 (find_interesting_uses_cond): Likewise.
22775 (find_interesting_uses_address): Likewise.
22776 (get_mem_type_for_internal_fn): New function.
22777 (find_address_like_use): Likewise.
22778 (find_interesting_uses_stmt): Try find_address_like_use before
22779 calling find_interesting_uses_op.
22780 (addr_offset_valid_p): Use the iv mem_type field as the type
22781 of the addressed memory.
22782 (add_autoinc_candidates): Likewise.
22783 (get_address_cost): Likewise.
22784 (split_small_address_groups_p): Use address_p.
22785 (split_address_groups): Likewise.
22786 (add_iv_candidate_for_use): Likewise.
22787 (autoinc_possible_for_pair): Likewise.
22788 (rewrite_groups): Likewise.
22789 (get_use_type): Check for USE_REF_ADDRESS instead of USE_ADDRESS.
22790 (determine_group_iv_cost): Update after split of USE_ADDRESS.
22791 (get_alias_ptr_type_for_ptr_address): New function.
22792 (rewrite_use_address): Rewrite address uses in calls that were
22793 identified by find_address_like_use.
22795 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
22796 Alan Hayward <alan.hayward@arm.com>
22797 David Sherwood <david.sherwood@arm.com>
22799 * expr.c (expand_expr_addr_expr_1): Handle ADDR_EXPRs of
22801 * gimple-expr.h (is_gimple_addressable: Likewise.
22802 * gimple-expr.c (is_gimple_address): Likewise.
22803 * internal-fn.c (expand_call_mem_ref): New function.
22804 (expand_mask_load_optab_fn): Use it.
22805 (expand_mask_store_optab_fn): Likewise.
22807 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
22808 Alan Hayward <alan.hayward@arm.com>
22809 David Sherwood <david.sherwood@arm.com>
22811 * doc/md.texi (cond_add@var{mode}, cond_sub@var{mode})
22812 (cond_and@var{mode}, cond_ior@var{mode}, cond_xor@var{mode})
22813 (cond_smin@var{mode}, cond_smax@var{mode}, cond_umin@var{mode})
22814 (cond_umax@var{mode}): Document.
22815 * optabs.def (cond_add_optab, cond_sub_optab, cond_and_optab)
22816 (cond_ior_optab, cond_xor_optab, cond_smin_optab, cond_smax_optab)
22817 (cond_umin_optab, cond_umax_optab): New optabs.
22818 * internal-fn.def (COND_ADD, COND_SUB, COND_MIN, COND_MAX, COND_AND)
22819 (COND_IOR, COND_XOR): New internal functions.
22820 * internal-fn.h (get_conditional_internal_fn): Declare.
22821 * internal-fn.c (cond_binary_direct): New macro.
22822 (expand_cond_binary_optab_fn): Likewise.
22823 (direct_cond_binary_optab_supported_p): Likewise.
22824 (get_conditional_internal_fn): New function.
22825 * tree-vect-loop.c (vectorizable_reduction): Handle fully-masked loops.
22826 Cope with reduction statements that are vectorized as calls rather
22828 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): New insns.
22829 * config/aarch64/iterators.md (UNSPEC_COND_ADD, UNSPEC_COND_SUB)
22830 (UNSPEC_COND_SMAX, UNSPEC_COND_UMAX, UNSPEC_COND_SMIN)
22831 (UNSPEC_COND_UMIN, UNSPEC_COND_AND, UNSPEC_COND_ORR)
22832 (UNSPEC_COND_EOR): New unspecs.
22833 (optab): Add mappings for them.
22834 (SVE_COND_INT_OP, SVE_COND_FP_OP): New int iterators.
22835 (sve_int_op, sve_fp_op): New int attributes.
22837 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
22838 Alan Hayward <alan.hayward@arm.com>
22839 David Sherwood <david.sherwood@arm.com>
22841 * optabs.def (while_ult_optab): New optab.
22842 * doc/md.texi (while_ult@var{m}@var{n}): Document.
22843 * internal-fn.def (WHILE_ULT): New internal function.
22844 * internal-fn.h (direct_internal_fn_supported_p): New override
22845 that takes two types as argument.
22846 * internal-fn.c (while_direct): New macro.
22847 (expand_while_optab_fn): New function.
22848 (convert_optab_supported_p): Likewise.
22849 (direct_while_optab_supported_p): New macro.
22850 * wide-int.h (wi::udiv_ceil): New function.
22851 * tree-vectorizer.h (rgroup_masks): New structure.
22852 (vec_loop_masks): New typedef.
22853 (_loop_vec_info): Add masks, mask_compare_type, can_fully_mask_p
22854 and fully_masked_p.
22855 (LOOP_VINFO_CAN_FULLY_MASK_P, LOOP_VINFO_FULLY_MASKED_P)
22856 (LOOP_VINFO_MASKS, LOOP_VINFO_MASK_COMPARE_TYPE): New macros.
22857 (vect_max_vf): New function.
22858 (slpeel_make_loop_iterate_ntimes): Delete.
22859 (vect_set_loop_condition, vect_get_loop_mask_type, vect_gen_while)
22860 (vect_halve_mask_nunits, vect_double_mask_nunits): Declare.
22861 (vect_record_loop_mask, vect_get_loop_mask): Likewise.
22862 * tree-vect-loop-manip.c: Include tree-ssa-loop-niter.h,
22863 internal-fn.h, stor-layout.h and optabs-query.h.
22864 (vect_set_loop_mask): New function.
22865 (add_preheader_seq): Likewise.
22866 (add_header_seq): Likewise.
22867 (interleave_supported_p): Likewise.
22868 (vect_maybe_permute_loop_masks): Likewise.
22869 (vect_set_loop_masks_directly): Likewise.
22870 (vect_set_loop_condition_masked): Likewise.
22871 (vect_set_loop_condition_unmasked): New function, split out from
22872 slpeel_make_loop_iterate_ntimes.
22873 (slpeel_make_loop_iterate_ntimes): Rename to..
22874 (vect_set_loop_condition): ...this. Use vect_set_loop_condition_masked
22875 for fully-masked loops and vect_set_loop_condition_unmasked otherwise.
22876 (vect_do_peeling): Update call accordingly.
22877 (vect_gen_vector_loop_niters): Use VF as the step for fully-masked
22879 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
22880 mask_compare_type, can_fully_mask_p and fully_masked_p.
22881 (release_vec_loop_masks): New function.
22882 (_loop_vec_info): Use it to free the loop masks.
22883 (can_produce_all_loop_masks_p): New function.
22884 (vect_get_max_nscalars_per_iter): Likewise.
22885 (vect_verify_full_masking): Likewise.
22886 (vect_analyze_loop_2): Save LOOP_VINFO_CAN_FULLY_MASK_P around
22887 retries, and free the mask rgroups before retrying. Check loop-wide
22888 reasons for disallowing fully-masked loops. Make the final decision
22889 about whether use a fully-masked loop or not.
22890 (vect_estimate_min_profitable_iters): Do not assume that peeling
22891 for the number of iterations will be needed for fully-masked loops.
22892 (vectorizable_reduction): Disable fully-masked loops.
22893 (vectorizable_live_operation): Likewise.
22894 (vect_halve_mask_nunits): New function.
22895 (vect_double_mask_nunits): Likewise.
22896 (vect_record_loop_mask): Likewise.
22897 (vect_get_loop_mask): Likewise.
22898 (vect_transform_loop): Handle the case in which the final loop
22899 iteration might handle a partial vector. Call vect_set_loop_condition
22900 instead of slpeel_make_loop_iterate_ntimes.
22901 * tree-vect-stmts.c: Include tree-ssa-loop-niter.h and gimple-fold.h.
22902 (check_load_store_masking): New function.
22903 (prepare_load_store_mask): Likewise.
22904 (vectorizable_store): Handle fully-masked loops.
22905 (vectorizable_load): Likewise.
22906 (supportable_widening_operation): Use vect_halve_mask_nunits for
22908 (supportable_narrowing_operation): Likewise vect_double_mask_nunits.
22909 (vect_gen_while): New function.
22910 * config/aarch64/aarch64.md (umax<mode>3): New expander.
22911 (aarch64_uqdec<mode>): New insn.
22913 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
22914 Alan Hayward <alan.hayward@arm.com>
22915 David Sherwood <david.sherwood@arm.com>
22917 * optabs.def (reduc_and_scal_optab, reduc_ior_scal_optab)
22918 (reduc_xor_scal_optab): New optabs.
22919 * doc/md.texi (reduc_and_scal_@var{m}, reduc_ior_scal_@var{m})
22920 (reduc_xor_scal_@var{m}): Document.
22921 * doc/sourcebuild.texi (vect_logical_reduc): Likewise.
22922 * internal-fn.def (IFN_REDUC_AND, IFN_REDUC_IOR, IFN_REDUC_XOR): New
22923 internal functions.
22924 * fold-const-call.c (fold_const_call): Handle them.
22925 * tree-vect-loop.c (reduction_fn_for_scalar_code): Return the new
22926 internal functions for BIT_AND_EXPR, BIT_IOR_EXPR and BIT_XOR_EXPR.
22927 * config/aarch64/aarch64-sve.md (reduc_<bit_reduc>_scal_<mode>):
22928 (*reduc_<bit_reduc>_scal_<mode>): New patterns.
22929 * config/aarch64/iterators.md (UNSPEC_ANDV, UNSPEC_ORV)
22930 (UNSPEC_XORV): New unspecs.
22931 (optab): Add entries for them.
22932 (BITWISEV): New int iterator.
22933 (bit_reduc_op): New int attributes.
22935 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
22936 Alan Hayward <alan.hayward@arm.com>
22937 David Sherwood <david.sherwood@arm.com>
22939 * doc/md.texi (vec_shl_insert_@var{m}): New optab.
22940 * internal-fn.def (VEC_SHL_INSERT): New internal function.
22941 * optabs.def (vec_shl_insert_optab): New optab.
22942 * tree-vectorizer.h (can_duplicate_and_interleave_p): Declare.
22943 (duplicate_and_interleave): Likewise.
22944 * tree-vect-loop.c: Include internal-fn.h.
22945 (neutral_op_for_slp_reduction): New function, split out from
22946 get_initial_defs_for_reduction.
22947 (get_initial_def_for_reduction): Handle option 2 for variable-length
22948 vectors by loading the neutral value into a vector and then shifting
22949 the initial value into element 0.
22950 (get_initial_defs_for_reduction): Replace the code argument with
22951 the neutral value calculated by neutral_op_for_slp_reduction.
22952 Use gimple_build_vector for constant-length vectors.
22953 Use IFN_VEC_SHL_INSERT for variable-length vectors if all
22954 but the first group_size elements have a neutral value.
22955 Use duplicate_and_interleave otherwise.
22956 (vect_create_epilog_for_reduction): Take a neutral_op parameter.
22957 Update call to get_initial_defs_for_reduction. Handle SLP
22958 reductions for variable-length vectors by creating one vector
22959 result for each scalar result, with the elements associated
22960 with other scalar results stubbed out with the neutral value.
22961 (vectorizable_reduction): Call neutral_op_for_slp_reduction.
22962 Require IFN_VEC_SHL_INSERT for double reductions on
22963 variable-length vectors, or SLP reductions that have
22964 a neutral value. Require can_duplicate_and_interleave_p
22965 support for variable-length unchained SLP reductions if there
22966 is no neutral value, such as for MIN/MAX reductions. Also require
22967 the number of vector elements to be a multiple of the number of
22968 SLP statements when doing variable-length unchained SLP reductions.
22969 Update call to vect_create_epilog_for_reduction.
22970 * tree-vect-slp.c (can_duplicate_and_interleave_p): Make public
22971 and remove initial values.
22972 (duplicate_and_interleave): Make public.
22973 * config/aarch64/aarch64.md (UNSPEC_INSR): New unspec.
22974 * config/aarch64/aarch64-sve.md (vec_shl_insert_<mode>): New insn.
22976 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
22977 Alan Hayward <alan.hayward@arm.com>
22978 David Sherwood <david.sherwood@arm.com>
22980 * tree-vect-slp.c: Include gimple-fold.h and internal-fn.h
22981 (can_duplicate_and_interleave_p): New function.
22982 (vect_get_and_check_slp_defs): Take the vector of statements
22983 rather than just the current one. Remove excess parentheses.
22984 Restriction rejectinon of vect_constant_def and vect_external_def
22985 for variable-length vectors to boolean types, or types for which
22986 can_duplicate_and_interleave_p is false.
22987 (vect_build_slp_tree_2): Update call to vect_get_and_check_slp_defs.
22988 (duplicate_and_interleave): New function.
22989 (vect_get_constant_vectors): Use gimple_build_vector for
22990 constant-length vectors and suitable variable-length constant
22991 vectors. Use duplicate_and_interleave for other variable-length
22992 vectors. Don't defer the update when inserting new statements.
22994 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
22995 Alan Hayward <alan.hayward@arm.com>
22996 David Sherwood <david.sherwood@arm.com>
22998 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Make sure
22999 min_profitable_iters doesn't go negative.
23001 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
23002 Alan Hayward <alan.hayward@arm.com>
23003 David Sherwood <david.sherwood@arm.com>
23005 * doc/md.texi (vec_mask_load_lanes@var{m}@var{n}): Document.
23006 (vec_mask_store_lanes@var{m}@var{n}): Likewise.
23007 * optabs.def (vec_mask_load_lanes_optab): New optab.
23008 (vec_mask_store_lanes_optab): Likewise.
23009 * internal-fn.def (MASK_LOAD_LANES): New internal function.
23010 (MASK_STORE_LANES): Likewise.
23011 * internal-fn.c (mask_load_lanes_direct): New macro.
23012 (mask_store_lanes_direct): Likewise.
23013 (expand_mask_load_optab_fn): Handle masked operations.
23014 (expand_mask_load_lanes_optab_fn): New macro.
23015 (expand_mask_store_optab_fn): Handle masked operations.
23016 (expand_mask_store_lanes_optab_fn): New macro.
23017 (direct_mask_load_lanes_optab_supported_p): Likewise.
23018 (direct_mask_store_lanes_optab_supported_p): Likewise.
23019 * tree-vectorizer.h (vect_store_lanes_supported): Take a masked_p
23021 (vect_load_lanes_supported): Likewise.
23022 * tree-vect-data-refs.c (strip_conversion): New function.
23023 (can_group_stmts_p): Likewise.
23024 (vect_analyze_data_ref_accesses): Use it instead of checking
23025 for a pair of assignments.
23026 (vect_store_lanes_supported): Take a masked_p parameter.
23027 (vect_load_lanes_supported): Likewise.
23028 * tree-vect-loop.c (vect_analyze_loop_2): Update calls to
23029 vect_store_lanes_supported and vect_load_lanes_supported.
23030 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
23031 * tree-vect-stmts.c (get_group_load_store_type): Take a masked_p
23032 parameter. Don't allow gaps for masked accesses.
23033 Use vect_get_store_rhs. Update calls to vect_store_lanes_supported
23034 and vect_load_lanes_supported.
23035 (get_load_store_type): Take a masked_p parameter and update
23036 call to get_group_load_store_type.
23037 (vectorizable_store): Update call to get_load_store_type.
23038 Handle IFN_MASK_STORE_LANES.
23039 (vectorizable_load): Update call to get_load_store_type.
23040 Handle IFN_MASK_LOAD_LANES.
23042 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
23043 Alan Hayward <alan.hayward@arm.com>
23044 David Sherwood <david.sherwood@arm.com>
23046 * config/aarch64/aarch64-modes.def: Define x2, x3 and x4 vector
23048 * config/aarch64/aarch64-protos.h
23049 (aarch64_sve_struct_memory_operand_p): Declare.
23050 * config/aarch64/iterators.md (SVE_STRUCT): New mode iterator.
23051 (vector_count, insn_length, VSINGLE, vsingle): New mode attributes.
23052 (VPRED, vpred): Handle SVE structure modes.
23053 * config/aarch64/constraints.md (Utx): New constraint.
23054 * config/aarch64/predicates.md (aarch64_sve_struct_memory_operand)
23055 (aarch64_sve_struct_nonimmediate_operand): New predicates.
23056 * config/aarch64/aarch64.md (UNSPEC_LDN, UNSPEC_STN): New unspecs.
23057 * config/aarch64/aarch64-sve.md (mov<mode>, *aarch64_sve_mov<mode>_le)
23058 (*aarch64_sve_mov<mode>_be, pred_mov<mode>): New patterns for
23059 structure modes. Split into pieces after RA.
23060 (vec_load_lanes<mode><vsingle>, vec_mask_load_lanes<mode><vsingle>)
23061 (vec_store_lanes<mode><vsingle>, vec_mask_store_lanes<mode><vsingle>):
23063 * config/aarch64/aarch64.c (aarch64_classify_vector_mode): Handle
23064 SVE structure modes.
23065 (aarch64_classify_address): Likewise.
23066 (sizetochar): Move earlier in file.
23067 (aarch64_print_operand): Handle SVE register lists.
23068 (aarch64_array_mode): New function.
23069 (aarch64_sve_struct_memory_operand_p): Likewise.
23070 (TARGET_ARRAY_MODE): Redefine.
23072 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
23073 Alan Hayward <alan.hayward@arm.com>
23074 David Sherwood <david.sherwood@arm.com>
23076 * target.def (array_mode): New target hook.
23077 * doc/tm.texi.in (TARGET_ARRAY_MODE): New hook.
23078 * doc/tm.texi: Regenerate.
23079 * hooks.h (hook_optmode_mode_uhwi_none): Declare.
23080 * hooks.c (hook_optmode_mode_uhwi_none): New function.
23081 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Use
23082 targetm.array_mode.
23083 * stor-layout.c (mode_for_array): Likewise. Support polynomial
23086 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
23087 Alan Hayward <alan.hayward@arm.com>
23088 David Sherwood <david.sherwood@arm.com>
23090 * fold-const.c (fold_binary_loc): Check the argument types
23091 rather than the result type when testing for a vector operation.
23093 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
23095 * doc/tm.texi.in (DWARF_LAZY_REGISTER_VALUE): Document.
23096 * doc/tm.texi: Regenerate.
23098 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
23099 Alan Hayward <alan.hayward@arm.com>
23100 David Sherwood <david.sherwood@arm.com>
23102 * doc/invoke.texi (-msve-vector-bits=): Document new option.
23103 (sve): Document new AArch64 extension.
23104 * doc/md.texi (w): Extend the description of the AArch64
23105 constraint to include SVE vectors.
23106 (Upl, Upa): Document new AArch64 predicate constraints.
23107 * config/aarch64/aarch64-opts.h (aarch64_sve_vector_bits_enum): New
23109 * config/aarch64/aarch64.opt (sve_vector_bits): New enum.
23110 (msve-vector-bits=): New option.
23111 * config/aarch64/aarch64-option-extensions.def (fp, simd): Disable
23112 SVE when these are disabled.
23113 (sve): New extension.
23114 * config/aarch64/aarch64-modes.def: Define SVE vector and predicate
23115 modes. Adjust their number of units based on aarch64_sve_vg.
23116 (MAX_BITSIZE_MODE_ANY_MODE): Define.
23117 * config/aarch64/aarch64-protos.h (ADDR_QUERY_ANY): New
23118 aarch64_addr_query_type.
23119 (aarch64_const_vec_all_same_in_range_p, aarch64_sve_pred_mode)
23120 (aarch64_sve_cnt_immediate_p, aarch64_sve_addvl_addpl_immediate_p)
23121 (aarch64_sve_inc_dec_immediate_p, aarch64_add_offset_temporaries)
23122 (aarch64_split_add_offset, aarch64_output_sve_cnt_immediate)
23123 (aarch64_output_sve_addvl_addpl, aarch64_output_sve_inc_dec_immediate)
23124 (aarch64_output_sve_mov_immediate, aarch64_output_ptrue): Declare.
23125 (aarch64_simd_imm_zero_p): Delete.
23126 (aarch64_check_zero_based_sve_index_immediate): Declare.
23127 (aarch64_sve_index_immediate_p, aarch64_sve_arith_immediate_p)
23128 (aarch64_sve_bitmask_immediate_p, aarch64_sve_dup_immediate_p)
23129 (aarch64_sve_cmp_immediate_p, aarch64_sve_float_arith_immediate_p)
23130 (aarch64_sve_float_mul_immediate_p): Likewise.
23131 (aarch64_classify_symbol): Take the offset as a HOST_WIDE_INT
23132 rather than an rtx.
23133 (aarch64_sve_ld1r_operand_p, aarch64_sve_ldr_operand_p): Declare.
23134 (aarch64_expand_mov_immediate): Take a gen_vec_duplicate callback.
23135 (aarch64_emit_sve_pred_move, aarch64_expand_sve_mem_move): Declare.
23136 (aarch64_expand_sve_vec_cmp_int, aarch64_expand_sve_vec_cmp_float)
23137 (aarch64_expand_sve_vcond, aarch64_expand_sve_vec_perm): Declare.
23138 (aarch64_regmode_natural_size): Likewise.
23139 * config/aarch64/aarch64.h (AARCH64_FL_SVE): New macro.
23140 (AARCH64_FL_V8_3, AARCH64_FL_RCPC, AARCH64_FL_DOTPROD): Shift
23142 (AARCH64_ISA_SVE, TARGET_SVE): New macros.
23143 (FIXED_REGISTERS, CALL_USED_REGISTERS, REGISTER_NAMES): Add entries
23144 for VG and the SVE predicate registers.
23145 (V_ALIASES): Add a "z"-prefixed alias.
23146 (FIRST_PSEUDO_REGISTER): Change to P15_REGNUM + 1.
23147 (AARCH64_DWARF_VG, AARCH64_DWARF_P0): New macros.
23148 (PR_REGNUM_P, PR_LO_REGNUM_P): Likewise.
23149 (PR_LO_REGS, PR_HI_REGS, PR_REGS): New reg_classes.
23150 (REG_CLASS_NAMES): Add entries for them.
23151 (REG_CLASS_CONTENTS): Likewise. Update ALL_REGS to include VG
23152 and the predicate registers.
23153 (aarch64_sve_vg): Declare.
23154 (BITS_PER_SVE_VECTOR, BYTES_PER_SVE_VECTOR, BYTES_PER_SVE_PRED)
23155 (SVE_BYTE_MODE, MAX_COMPILE_TIME_VEC_BYTES): New macros.
23156 (REGMODE_NATURAL_SIZE): Define.
23157 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins): Handle
23159 * config/aarch64/aarch64.c: Include cfgrtl.h.
23160 (simd_immediate_info): Add a constructor for series vectors,
23161 and an associated step field.
23162 (aarch64_sve_vg): New variable.
23163 (aarch64_dbx_register_number): Handle VG and the predicate registers.
23164 (aarch64_vect_struct_mode_p, aarch64_vector_mode_p): Delete.
23165 (VEC_ADVSIMD, VEC_SVE_DATA, VEC_SVE_PRED, VEC_STRUCT, VEC_ANY_SVE)
23166 (VEC_ANY_DATA, VEC_STRUCT): New constants.
23167 (aarch64_advsimd_struct_mode_p, aarch64_sve_pred_mode_p)
23168 (aarch64_classify_vector_mode, aarch64_vector_data_mode_p)
23169 (aarch64_sve_data_mode_p, aarch64_sve_pred_mode)
23170 (aarch64_get_mask_mode): New functions.
23171 (aarch64_hard_regno_nregs): Handle SVE data modes for FP_REGS
23172 and FP_LO_REGS. Handle PR_REGS, PR_LO_REGS and PR_HI_REGS.
23173 (aarch64_hard_regno_mode_ok): Handle VG. Also handle the SVE
23174 predicate modes and predicate registers. Explicitly restrict
23175 GPRs to modes of 16 bytes or smaller. Only allow FP registers
23176 to store a vector mode if it is recognized by
23177 aarch64_classify_vector_mode.
23178 (aarch64_regmode_natural_size): New function.
23179 (aarch64_hard_regno_caller_save_mode): Return the original mode
23181 (aarch64_sve_cnt_immediate_p, aarch64_output_sve_cnt_immediate)
23182 (aarch64_sve_addvl_addpl_immediate_p, aarch64_output_sve_addvl_addpl)
23183 (aarch64_sve_inc_dec_immediate_p, aarch64_output_sve_inc_dec_immediate)
23184 (aarch64_add_offset_1_temporaries, aarch64_offset_temporaries): New
23186 (aarch64_add_offset): Add a temp2 parameter. Assert that temp1
23187 does not overlap dest if the function is frame-related. Handle
23189 (aarch64_split_add_offset): New function.
23190 (aarch64_add_sp, aarch64_sub_sp): Add temp2 parameters and pass
23191 them aarch64_add_offset.
23192 (aarch64_allocate_and_probe_stack_space): Add a temp2 parameter
23193 and update call to aarch64_sub_sp.
23194 (aarch64_add_cfa_expression): New function.
23195 (aarch64_expand_prologue): Pass extra temporary registers to the
23196 functions above. Handle the case in which we need to emit new
23197 DW_CFA_expressions for registers that were originally saved
23198 relative to the stack pointer, but now have to be expressed
23199 relative to the frame pointer.
23200 (aarch64_output_mi_thunk): Pass extra temporary registers to the
23202 (aarch64_expand_epilogue): Likewise. Prevent inheritance of
23203 IP0 and IP1 values for SVE frames.
23204 (aarch64_expand_vec_series): New function.
23205 (aarch64_expand_sve_widened_duplicate): Likewise.
23206 (aarch64_expand_sve_const_vector): Likewise.
23207 (aarch64_expand_mov_immediate): Add a gen_vec_duplicate parameter.
23208 Handle SVE constants. Use emit_move_insn to move a force_const_mem
23209 into the register, rather than emitting a SET directly.
23210 (aarch64_emit_sve_pred_move, aarch64_expand_sve_mem_move)
23211 (aarch64_get_reg_raw_mode, offset_4bit_signed_scaled_p)
23212 (offset_6bit_unsigned_scaled_p, aarch64_offset_7bit_signed_scaled_p)
23213 (offset_9bit_signed_scaled_p): New functions.
23214 (aarch64_replicate_bitmask_imm): New function.
23215 (aarch64_bitmask_imm): Use it.
23216 (aarch64_cannot_force_const_mem): Reject expressions involving
23217 a CONST_POLY_INT. Update call to aarch64_classify_symbol.
23218 (aarch64_classify_index): Handle SVE indices, by requiring
23219 a plain register index with a scale that matches the element size.
23220 (aarch64_classify_address): Handle SVE addresses. Assert that
23221 the mode of the address is VOIDmode or an integer mode.
23222 Update call to aarch64_classify_symbol.
23223 (aarch64_classify_symbolic_expression): Update call to
23224 aarch64_classify_symbol.
23225 (aarch64_const_vec_all_in_range_p): New function.
23226 (aarch64_print_vector_float_operand): Likewise.
23227 (aarch64_print_operand): Handle 'N' and 'C'. Use "zN" rather than
23228 "vN" for FP registers with SVE modes. Handle (const ...) vectors
23229 and the FP immediates 1.0 and 0.5.
23230 (aarch64_print_address_internal): Handle SVE addresses.
23231 (aarch64_print_operand_address): Use ADDR_QUERY_ANY.
23232 (aarch64_regno_regclass): Handle predicate registers.
23233 (aarch64_secondary_reload): Handle big-endian reloads of SVE
23235 (aarch64_class_max_nregs): Handle SVE modes and predicate registers.
23236 (aarch64_rtx_costs): Check for ADDVL and ADDPL instructions.
23237 (aarch64_convert_sve_vector_bits): New function.
23238 (aarch64_override_options): Use it to handle -msve-vector-bits=.
23239 (aarch64_classify_symbol): Take the offset as a HOST_WIDE_INT
23240 rather than an rtx.
23241 (aarch64_legitimate_constant_p): Use aarch64_classify_vector_mode.
23242 Handle SVE vector and predicate modes. Accept VL-based constants
23243 that need only one temporary register, and VL offsets that require
23244 no temporary registers.
23245 (aarch64_conditional_register_usage): Mark the predicate registers
23246 as fixed if SVE isn't available.
23247 (aarch64_vector_mode_supported_p): Use aarch64_classify_vector_mode.
23248 Return true for SVE vector and predicate modes.
23249 (aarch64_simd_container_mode): Take the number of bits as a poly_int64
23250 rather than an unsigned int. Handle SVE modes.
23251 (aarch64_preferred_simd_mode): Update call accordingly. Handle
23253 (aarch64_autovectorize_vector_sizes): Add BYTES_PER_SVE_VECTOR
23255 (aarch64_sve_index_immediate_p, aarch64_sve_arith_immediate_p)
23256 (aarch64_sve_bitmask_immediate_p, aarch64_sve_dup_immediate_p)
23257 (aarch64_sve_cmp_immediate_p, aarch64_sve_float_arith_immediate_p)
23258 (aarch64_sve_float_mul_immediate_p): New functions.
23259 (aarch64_sve_valid_immediate): New function.
23260 (aarch64_simd_valid_immediate): Use it as the fallback for SVE vectors.
23261 Explicitly reject structure modes. Check for INDEX constants.
23262 Handle PTRUE and PFALSE constants.
23263 (aarch64_check_zero_based_sve_index_immediate): New function.
23264 (aarch64_simd_imm_zero_p): Delete.
23265 (aarch64_mov_operand_p): Use aarch64_simd_valid_immediate for
23266 vector modes. Accept constants in the range of CNT[BHWD].
23267 (aarch64_simd_scalar_immediate_valid_for_move): Explicitly
23268 ask for an Advanced SIMD mode.
23269 (aarch64_sve_ld1r_operand_p, aarch64_sve_ldr_operand_p): New functions.
23270 (aarch64_simd_vector_alignment): Handle SVE predicates.
23271 (aarch64_vectorize_preferred_vector_alignment): New function.
23272 (aarch64_simd_vector_alignment_reachable): Use it instead of
23274 (aarch64_shift_truncation_mask): Use aarch64_vector_data_mode_p.
23275 (aarch64_output_sve_mov_immediate, aarch64_output_ptrue): New
23277 (MAX_VECT_LEN): Delete.
23278 (expand_vec_perm_d): Add a vec_flags field.
23279 (emit_unspec2, aarch64_expand_sve_vec_perm): New functions.
23280 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
23281 (aarch64_evpc_ext): Don't apply a big-endian lane correction
23283 (aarch64_evpc_rev): Rename to...
23284 (aarch64_evpc_rev_local): ...this. Use a predicated operation for SVE.
23285 (aarch64_evpc_rev_global): New function.
23286 (aarch64_evpc_dup): Enforce a 64-byte range for SVE DUP.
23287 (aarch64_evpc_tbl): Use MAX_COMPILE_TIME_VEC_BYTES instead of
23289 (aarch64_evpc_sve_tbl): New function.
23290 (aarch64_expand_vec_perm_const_1): Update after rename of
23291 aarch64_evpc_rev. Handle SVE permutes too, trying
23292 aarch64_evpc_rev_global and using aarch64_evpc_sve_tbl rather
23293 than aarch64_evpc_tbl.
23294 (aarch64_vectorize_vec_perm_const): Initialize vec_flags.
23295 (aarch64_sve_cmp_operand_p, aarch64_unspec_cond_code)
23296 (aarch64_gen_unspec_cond, aarch64_expand_sve_vec_cmp_int)
23297 (aarch64_emit_unspec_cond, aarch64_emit_unspec_cond_or)
23298 (aarch64_emit_inverted_unspec_cond, aarch64_expand_sve_vec_cmp_float)
23299 (aarch64_expand_sve_vcond): New functions.
23300 (aarch64_modes_tieable_p): Use aarch64_vector_data_mode_p instead
23301 of aarch64_vector_mode_p.
23302 (aarch64_dwarf_poly_indeterminate_value): New function.
23303 (aarch64_compute_pressure_classes): Likewise.
23304 (aarch64_can_change_mode_class): Likewise.
23305 (TARGET_GET_RAW_RESULT_MODE, TARGET_GET_RAW_ARG_MODE): Redefine.
23306 (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): Likewise.
23307 (TARGET_VECTORIZE_GET_MASK_MODE): Likewise.
23308 (TARGET_DWARF_POLY_INDETERMINATE_VALUE): Likewise.
23309 (TARGET_COMPUTE_PRESSURE_CLASSES): Likewise.
23310 (TARGET_CAN_CHANGE_MODE_CLASS): Likewise.
23311 * config/aarch64/constraints.md (Upa, Upl, Uav, Uat, Usv, Usi, Utr)
23312 (Uty, Dm, vsa, vsc, vsd, vsi, vsn, vsl, vsm, vsA, vsM, vsN): New
23314 (Dn, Dl, Dr): Accept const as well as const_vector.
23315 (Dz): Likewise. Compare against CONST0_RTX.
23316 * config/aarch64/iterators.md: Refer to "Advanced SIMD" instead
23317 of "vector" where appropriate.
23318 (SVE_ALL, SVE_BH, SVE_BHS, SVE_BHSI, SVE_HSDI, SVE_HSF, SVE_SD)
23319 (SVE_SDI, SVE_I, SVE_F, PRED_ALL, PRED_BHS): New mode iterators.
23320 (UNSPEC_SEL, UNSPEC_ANDF, UNSPEC_IORF, UNSPEC_XORF, UNSPEC_COND_LT)
23321 (UNSPEC_COND_LE, UNSPEC_COND_EQ, UNSPEC_COND_NE, UNSPEC_COND_GE)
23322 (UNSPEC_COND_GT, UNSPEC_COND_LO, UNSPEC_COND_LS, UNSPEC_COND_HS)
23323 (UNSPEC_COND_HI, UNSPEC_COND_UO): New unspecs.
23324 (Vetype, VEL, Vel, VWIDE, Vwide, vw, vwcore, V_INT_EQUIV)
23325 (v_int_equiv): Extend to SVE modes.
23326 (Vesize, V128, v128, Vewtype, V_FP_EQUIV, v_fp_equiv, VPRED): New
23328 (LOGICAL_OR, SVE_INT_UNARY, SVE_FP_UNARY): New code iterators.
23329 (optab): Handle popcount, smin, smax, umin, umax, abs and sqrt.
23330 (logical_nn, lr, sve_int_op, sve_fp_op): New code attributs.
23331 (LOGICALF, OPTAB_PERMUTE, UNPACK, UNPACK_UNSIGNED, SVE_COND_INT_CMP)
23332 (SVE_COND_FP_CMP): New int iterators.
23333 (perm_hilo): Handle the new unpack unspecs.
23334 (optab, logicalf_op, su, perm_optab, cmp_op, imm_con): New int
23336 * config/aarch64/predicates.md (aarch64_sve_cnt_immediate)
23337 (aarch64_sve_addvl_addpl_immediate, aarch64_split_add_offset_immediate)
23338 (aarch64_pluslong_or_poly_operand, aarch64_nonmemory_operand)
23339 (aarch64_equality_operator, aarch64_constant_vector_operand)
23340 (aarch64_sve_ld1r_operand, aarch64_sve_ldr_operand): New predicates.
23341 (aarch64_sve_nonimmediate_operand): Likewise.
23342 (aarch64_sve_general_operand): Likewise.
23343 (aarch64_sve_dup_operand, aarch64_sve_arith_immediate): Likewise.
23344 (aarch64_sve_sub_arith_immediate, aarch64_sve_inc_dec_immediate)
23345 (aarch64_sve_logical_immediate, aarch64_sve_mul_immediate): Likewise.
23346 (aarch64_sve_dup_immediate, aarch64_sve_cmp_vsc_immediate): Likewise.
23347 (aarch64_sve_cmp_vsd_immediate, aarch64_sve_index_immediate): Likewise.
23348 (aarch64_sve_float_arith_immediate): Likewise.
23349 (aarch64_sve_float_arith_with_sub_immediate): Likewise.
23350 (aarch64_sve_float_mul_immediate, aarch64_sve_arith_operand): Likewise.
23351 (aarch64_sve_add_operand, aarch64_sve_logical_operand): Likewise.
23352 (aarch64_sve_lshift_operand, aarch64_sve_rshift_operand): Likewise.
23353 (aarch64_sve_mul_operand, aarch64_sve_cmp_vsc_operand): Likewise.
23354 (aarch64_sve_cmp_vsd_operand, aarch64_sve_index_operand): Likewise.
23355 (aarch64_sve_float_arith_operand): Likewise.
23356 (aarch64_sve_float_arith_with_sub_operand): Likewise.
23357 (aarch64_sve_float_mul_operand): Likewise.
23358 (aarch64_sve_vec_perm_operand): Likewise.
23359 (aarch64_pluslong_operand): Include aarch64_sve_addvl_addpl_immediate.
23360 (aarch64_mov_operand): Accept const_poly_int and const_vector.
23361 (aarch64_simd_lshift_imm, aarch64_simd_rshift_imm): Accept const
23362 as well as const_vector.
23363 (aarch64_simd_imm_zero, aarch64_simd_imm_minus_one): Move earlier
23364 in file. Use CONST0_RTX and CONSTM1_RTX.
23365 (aarch64_simd_or_scalar_imm_zero): Likewise. Add match_codes.
23366 (aarch64_simd_reg_or_zero): Accept const as well as const_vector.
23367 Use aarch64_simd_imm_zero.
23368 * config/aarch64/aarch64-sve.md: New file.
23369 * config/aarch64/aarch64.md: Include it.
23370 (VG_REGNUM, P0_REGNUM, P7_REGNUM, P15_REGNUM): New register numbers.
23371 (UNSPEC_REV, UNSPEC_LD1_SVE, UNSPEC_ST1_SVE, UNSPEC_MERGE_PTRUE)
23372 (UNSPEC_PTEST_PTRUE, UNSPEC_UNPACKSHI, UNSPEC_UNPACKUHI)
23373 (UNSPEC_UNPACKSLO, UNSPEC_UNPACKULO, UNSPEC_PACK)
23374 (UNSPEC_FLOAT_CONVERT, UNSPEC_WHILE_LO): New unspec constants.
23375 (sve): New attribute.
23376 (enabled): Disable instructions with the sve attribute unless
23378 (movqi, movhi): Pass CONST_POLY_INT operaneds through
23379 aarch64_expand_mov_immediate.
23380 (*mov<mode>_aarch64, *movsi_aarch64, *movdi_aarch64): Handle
23381 CNT[BHSD] immediates.
23382 (movti): Split CONST_POLY_INT moves into two halves.
23383 (add<mode>3): Accept aarch64_pluslong_or_poly_operand.
23384 Split additions that need a temporary here if the destination
23385 is the stack pointer.
23386 (*add<mode>3_aarch64): Handle ADDVL and ADDPL immediates.
23387 (*add<mode>3_poly_1): New instruction.
23388 (set_clobber_cc): New expander.
23390 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
23392 * simplify-rtx.c (simplify_immed_subreg): Add an inner_bytes
23393 parameter and use it instead of GET_MODE_SIZE (innermode). Use
23394 inner_bytes * BITS_PER_UNIT instead of GET_MODE_BITSIZE (innermode).
23395 Use CEIL (inner_bytes, GET_MODE_UNIT_SIZE (innermode)) instead of
23396 GET_MODE_NUNITS (innermode). Also add a first_elem parameter.
23397 Change innermode from fixed_mode_size to machine_mode.
23398 (simplify_subreg): Update call accordingly. Handle a constant-sized
23399 subreg of a variable-length CONST_VECTOR.
23401 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
23402 Alan Hayward <alan.hayward@arm.com>
23403 David Sherwood <david.sherwood@arm.com>
23405 * tree-ssa-address.c (mem_ref_valid_without_offset_p): New function.
23406 (add_offset_to_base): New function, split out from...
23407 (create_mem_ref): ...here. When handling a scale other than 1,
23408 check first whether the address is valid without the offset.
23409 Add it into the base if so, leaving the index and scale as-is.
23411 2018-01-12 Jakub Jelinek <jakub@redhat.com>
23414 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Call
23415 fold_for_warn before checking if arg2 is INTEGER_CST.
23417 2018-01-12 Segher Boessenkool <segher@kernel.crashing.org>
23419 * config/rs6000/predicates.md (load_multiple_operation): Delete.
23420 (store_multiple_operation): Delete.
23421 * config/rs6000/rs6000-cpus.def (601): Remove MASK_STRING.
23422 * config/rs6000/rs6000-protos.h (rs6000_output_load_multiple): Delete.
23423 * config/rs6000/rs6000-string.c (expand_block_move): Delete everything
23424 guarded by TARGET_STRING.
23425 (rs6000_output_load_multiple): Delete.
23426 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
23427 OPTION_MASK_STRING / TARGET_STRING handling.
23428 (print_operand) <'N', 'O'>: Add comment that these are unused now.
23429 (const rs6000_opt_masks) <"string">: Change mask to 0.
23430 * config/rs6000/rs6000.h (TARGET_DEFAULT): Remove MASK_STRING.
23431 (MASK_STRING): Delete.
23432 * config/rs6000/rs6000.md (*mov<mode>_string): Delete TARGET_STRING
23434 (load_multiple): Delete.
23441 (store_multiple): Delete.
23448 (movmemsi_8reg): Delete.
23449 (corresponding unnamed define_insn): Delete.
23450 (movmemsi_6reg): Delete.
23451 (corresponding unnamed define_insn): Delete.
23452 (movmemsi_4reg): Delete.
23453 (corresponding unnamed define_insn): Delete.
23454 (movmemsi_2reg): Delete.
23455 (corresponding unnamed define_insn): Delete.
23456 (movmemsi_1reg): Delete.
23457 (corresponding unnamed define_insn): Delete.
23458 * config/rs6000/rs6000.opt (mno-string): New.
23459 (mstring): Replace by deprecation warning stub.
23460 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mstring.
23462 2018-01-12 Jakub Jelinek <jakub@redhat.com>
23464 * regrename.c (regrename_do_replace): If replacing the same
23465 reg multiple times, try to reuse last created gen_raw_REG.
23468 * bb-reorder.c (pass_partition_blocks::gate): In lto don't partition
23469 main to workaround a bug in GDB.
23471 2018-01-12 Tom de Vries <tom@codesourcery.com>
23474 * config.gcc (nvptx*-*-*): Set use_gcc_stdint=wrap.
23476 2018-01-12 Vladimir Makarov <vmakarov@redhat.com>
23478 PR rtl-optimization/80481
23479 * ira-color.c (get_cap_member): New function.
23480 (allocnos_conflict_by_live_ranges_p): Use it.
23481 (slot_coalesced_allocno_live_ranges_intersect_p): Add assert.
23482 (setup_slot_coalesced_allocno_live_ranges): Ditto.
23484 2018-01-12 Uros Bizjak <ubizjak@gmail.com>
23487 * config/alpha/alpha.md (*saddsi_1): New insn_ans_split pattern.
23488 (*saddl_se_1): Ditto.
23489 (*ssubsi_1): Ditto.
23490 (*ssubl_se_1): Ditto.
23492 2018-01-12 Richard Sandiford <richard.sandiford@linaro.org>
23494 * tree-predcom.c (aff_combination_dr_offset): Use wi::to_poly_widest
23495 rather than wi::to_widest for DR_INITs.
23496 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Use
23497 wi::to_poly_offset rather than wi::to_offset for DR_INIT.
23498 (vect_analyze_data_ref_accesses): Require both DR_INITs to be
23500 (vect_analyze_group_access_1): Note that here.
23502 2018-01-12 Richard Sandiford <richard.sandiford@linaro.org>
23504 * tree-vectorizer.c (get_vec_alignment_for_array_type): Handle
23505 polynomial type sizes.
23507 2018-01-12 Richard Sandiford <richard.sandiford@linaro.org>
23509 * gimplify.c (gimple_add_tmp_var_fn): Allow variables to have a
23510 poly_uint64 size, rather than requiring an unsigned HOST_WIDE_INT size.
23511 (gimple_add_tmp_var): Likewise.
23513 2018-01-12 Martin Liska <mliska@suse.cz>
23515 * gimple.c (gimple_alloc_counts): Use uint64_t instead of int.
23516 (gimple_alloc_sizes): Likewise.
23517 (dump_gimple_statistics): Use PRIu64 in printf format.
23518 * gimple.h: Change uint64_t to int.
23520 2018-01-12 Martin Liska <mliska@suse.cz>
23522 * tree-core.h: Use uint64_t instead of int.
23523 * tree.c (tree_node_counts): Likewise.
23524 (tree_node_sizes): Likewise.
23525 (dump_tree_statistics): Use PRIu64 in printf format.
23527 2018-01-12 Martin Liska <mliska@suse.cz>
23529 * Makefile.in: As qsort_chk is implemented in vec.c, add
23530 vec.o to linkage of gencfn-macros.
23531 * tree.c (build_new_poly_int_cst): Add CXX_MEM_STAT_INFO as it's
23532 passing the info to record_node_allocation_statistics.
23533 (test_vector_cst_patterns): Add CXX_MEM_STAT_INFO to declaration
23535 * ggc-common.c (struct ggc_usage): Add operator== and use
23536 it in operator< and compare function.
23537 * mem-stats.h (struct mem_usage): Likewise.
23538 * vec.c (struct vec_usage): Remove operator< and compare
23539 function. Can be simply inherited.
23541 2018-01-12 Martin Jambor <mjambor@suse.cz>
23544 * params.def: New parameter PARAM_AVOID_FMA_MAX_BITS.
23545 * tree-ssa-math-opts.c: Include domwalk.h.
23546 (convert_mult_to_fma_1): New function.
23547 (fma_transformation_info): New type.
23548 (fma_deferring_state): Likewise.
23549 (cancel_fma_deferring): New function.
23550 (result_of_phi): Likewise.
23551 (last_fma_candidate_feeds_initial_phi): Likewise.
23552 (convert_mult_to_fma): Added deferring logic, split actual
23553 transformation to convert_mult_to_fma_1.
23554 (math_opts_dom_walker): New type.
23555 (math_opts_dom_walker::after_dom_children): New method, body moved
23556 here from pass_optimize_widening_mul::execute, added deferring logic
23558 (pass_optimize_widening_mul::execute): Moved most of code to
23559 math_opts_dom_walker::after_dom_children.
23560 * config/i386/x86-tune.def (X86_TUNE_AVOID_128FMA_CHAINS): New.
23561 * config/i386/i386.c (ix86_option_override_internal): Added
23562 maybe_setting of PARAM_AVOID_FMA_MAX_BITS.
23564 2018-01-12 Richard Biener <rguenther@suse.de>
23567 * dwarf2out.c (gen_variable_die): Do not reset old_die for
23568 inline instance vars.
23570 2018-01-12 Oleg Endo <olegendo@gcc.gnu.org>
23573 * config/rx/rx.c (rx_is_restricted_memory_address):
23574 Handle SUBREG case.
23576 2018-01-12 Richard Biener <rguenther@suse.de>
23578 PR tree-optimization/80846
23579 * target.def (split_reduction): New target hook.
23580 * targhooks.c (default_split_reduction): New function.
23581 * targhooks.h (default_split_reduction): Declare.
23582 * tree-vect-loop.c (vect_create_epilog_for_reduction): If the
23583 target requests first reduce vectors by combining low and high
23585 * tree-vect-stmts.c (vect_gen_perm_mask_any): Adjust.
23586 (get_vectype_for_scalar_type_and_size): Export.
23587 * tree-vectorizer.h (get_vectype_for_scalar_type_and_size): Declare.
23588 * doc/tm.texi.in (TARGET_VECTORIZE_SPLIT_REDUCTION): Document.
23589 * doc/tm.texi: Regenerate.
23590 * config/i386/i386.c (ix86_split_reduction): Implement
23591 TARGET_VECTORIZE_SPLIT_REDUCTION.
23593 2018-01-12 Eric Botcazou <ebotcazou@adacore.com>
23596 * config/sparc/sparc.h (PIC_OFFSET_TABLE_REGNUM): Set to INVALID_REGNUM
23597 in PIC mode except for TARGET_VXWORKS_RTP.
23598 * config/sparc/sparc.c: Include cfgrtl.h.
23599 (TARGET_INIT_PIC_REG): Define.
23600 (TARGET_USE_PSEUDO_PIC_REG): Likewise.
23601 (sparc_pic_register_p): New predicate.
23602 (sparc_legitimate_address_p): Use it.
23603 (sparc_legitimize_pic_address): Likewise.
23604 (sparc_delegitimize_address): Likewise.
23605 (sparc_mode_dependent_address_p): Likewise.
23606 (gen_load_pcrel_sym): Remove 4th parameter.
23607 (load_got_register): Adjust call to above. Remove obsolete stuff.
23608 (sparc_expand_prologue): Do not call load_got_register here.
23609 (sparc_flat_expand_prologue): Likewise.
23610 (sparc_output_mi_thunk): Set the pic_offset_table_rtx object.
23611 (sparc_use_pseudo_pic_reg): New function.
23612 (sparc_init_pic_reg): Likewise.
23613 * config/sparc/sparc.md (vxworks_load_got): Set the GOT register.
23614 (builtin_setjmp_receiver): Enable only for TARGET_VXWORKS_RTP.
23616 2018-01-12 Christophe Lyon <christophe.lyon@linaro.org>
23618 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
23619 Add item for branch_cost.
23621 2018-01-12 Eric Botcazou <ebotcazou@adacore.com>
23623 PR rtl-optimization/83565
23624 * rtlanal.c (nonzero_bits1): On WORD_REGISTER_OPERATIONS machines, do
23625 not extend the result to a larger mode for rotate operations.
23626 (num_sign_bit_copies1): Likewise.
23628 2018-01-12 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
23631 * config/sol2.h (STARTFILE_ARCH_SPEC): Don't use with -shared or
23633 Use values-Xc.o for -pedantic.
23634 Link with values-xpg4.o for C90, values-xpg6.o otherwise.
23636 2018-01-12 Martin Liska <mliska@suse.cz>
23639 * ipa-devirt.c (final_warning_record::grow_type_warnings):
23641 (possible_polymorphic_call_targets): Use it.
23642 (ipa_devirt): Likewise.
23644 2018-01-12 Martin Liska <mliska@suse.cz>
23646 * profile-count.h (enum profile_quality): Use 0 as invalid
23647 enum value of profile_quality.
23649 2018-01-12 Chung-Ju Wu <jasonwucj@gmail.com>
23651 * doc/invoke.texi (NDS32 Options): Add -mext-perf, -mext-perf2 and
23652 -mext-string options.
23654 2018-01-12 Richard Biener <rguenther@suse.de>
23656 * lto-streamer-out.c (DFS::DFS_write_tree_body): Process
23657 DECL_DEBUG_EXPR conditional on DECL_HAS_DEBUG_EXPR_P.
23658 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
23660 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
23662 2018-01-11 Michael Meissner <meissner@linux.vnet.ibm.com>
23664 * configure.ac (--with-long-double-format): Add support for the
23665 configuration option to change the default long double format on
23667 * config.gcc (powerpc*-linux*-*): Likewise.
23668 * configure: Regenerate.
23669 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If long
23670 double is IEEE, define __KC__ and __KF__ to allow floatn.h to be
23671 used without modification.
23673 2018-01-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23675 * config/rs6000/rs6000-builtin.def (BU_P7_MISC_X): New #define.
23676 (SPEC_BARRIER): New instantiation of BU_P7_MISC_X.
23677 * config/rs6000/rs6000.c (rs6000_expand_builtin): Handle
23678 MISC_BUILTIN_SPEC_BARRIER.
23679 (rs6000_init_builtins): Likewise.
23680 * config/rs6000/rs6000.md (UNSPECV_SPEC_BARRIER): New UNSPECV
23682 (speculation_barrier): New define_insn.
23683 * doc/extend.texi: Document __builtin_speculation_barrier.
23685 2018-01-11 Jakub Jelinek <jakub@redhat.com>
23688 * config/i386/i386.c (ix86_expand_vector_init_one_nonzero): If one_var
23689 is 0, for V{8,16}S[IF] and V[48]D[IF]mode use gen_vec_set<mode>_0.
23690 * config/i386/sse.md (VI8_AVX_AVX512F, VI4F_256_512): New mode
23692 (ssescalarmodesuffix): Add 512-bit vectors. Use "d" or "q" for
23693 integral modes instead of "ss" and "sd".
23694 (vec_set<mode>_0): New define_insns for 256-bit and 512-bit
23695 vectors with 32-bit and 64-bit elements.
23696 (vecdupssescalarmodesuffix): New mode attribute.
23697 (vec_dup<mode>): Use it.
23699 2018-01-11 H.J. Lu <hongjiu.lu@intel.com>
23702 * config/i386/i386.c (ix86_compute_frame_layout): Align stack
23703 frame if argument is passed on stack.
23705 2018-01-11 Jakub Jelinek <jakub@redhat.com>
23708 * ree.c (combine_reaching_defs): Optimize also
23709 reg2=exp; reg1=reg2; reg2=any_extend(reg1); into
23710 reg2=any_extend(exp); reg1=reg2;, formatting fix.
23712 2018-01-11 Jan Hubicka <hubicka@ucw.cz>
23714 PR middle-end/83189
23715 * gimple-ssa-isolate-paths.c (isolate_path): Fix profile update.
23717 2018-01-11 Jan Hubicka <hubicka@ucw.cz>
23719 PR middle-end/83718
23720 * tree-inline.c (copy_cfg_body): Adjust num&den for scaling
23721 after they are computed.
23723 2018-01-11 Bin Cheng <bin.cheng@arm.com>
23725 PR tree-optimization/83695
23726 * gimple-loop-linterchange.cc
23727 (tree_loop_interchange::interchange_loops): Call scev_reset_htab to
23728 reset cached scev information after interchange.
23729 (pass_linterchange::execute): Remove call to scev_reset_htab.
23731 2018-01-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23733 * config/arm/arm_neon.h (vfmlal_lane_low_u32, vfmlal_lane_high_u32,
23734 vfmlalq_laneq_low_u32, vfmlalq_lane_low_u32, vfmlal_laneq_low_u32,
23735 vfmlalq_laneq_high_u32, vfmlalq_lane_high_u32, vfmlal_laneq_high_u32,
23736 vfmlsl_lane_low_u32, vfmlsl_lane_high_u32, vfmlslq_laneq_low_u32,
23737 vfmlslq_lane_low_u32, vfmlsl_laneq_low_u32, vfmlslq_laneq_high_u32,
23738 vfmlslq_lane_high_u32, vfmlsl_laneq_high_u32): Define.
23739 * config/arm/arm_neon_builtins.def (vfmal_lane_low,
23740 vfmal_lane_lowv4hf, vfmal_lane_lowv8hf, vfmal_lane_high,
23741 vfmal_lane_highv4hf, vfmal_lane_highv8hf, vfmsl_lane_low,
23742 vfmsl_lane_lowv4hf, vfmsl_lane_lowv8hf, vfmsl_lane_high,
23743 vfmsl_lane_highv4hf, vfmsl_lane_highv8hf): New sets of builtins.
23744 * config/arm/iterators.md (VFMLSEL2, vfmlsel2): New mode attributes.
23745 (V_lane_reg): Likewise.
23746 * config/arm/neon.md (neon_vfm<vfml_op>l_lane_<vfml_half><VCVTF:mode>):
23748 (neon_vfm<vfml_op>l_lane_<vfml_half><vfmlsel2><mode>): Likewise.
23749 (vfmal_lane_low<mode>_intrinsic,
23750 vfmal_lane_low<vfmlsel2><mode>_intrinsic,
23751 vfmal_lane_high<vfmlsel2><mode>_intrinsic,
23752 vfmal_lane_high<mode>_intrinsic, vfmsl_lane_low<mode>_intrinsic,
23753 vfmsl_lane_low<vfmlsel2><mode>_intrinsic,
23754 vfmsl_lane_high<vfmlsel2><mode>_intrinsic,
23755 vfmsl_lane_high<mode>_intrinsic): New define_insns.
23757 2018-01-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23759 * config/arm/arm-cpus.in (fp16fml): New feature.
23760 (ALL_SIMD): Add fp16fml.
23761 (armv8.2-a): Add fp16fml as an option.
23762 (armv8.3-a): Likewise.
23763 (armv8.4-a): Add fp16fml as part of fp16.
23764 * config/arm/arm.h (TARGET_FP16FML): Define.
23765 * config/arm/arm-c.c (arm_cpu_builtins): Define __ARM_FEATURE_FP16_FML
23767 * config/arm/arm-modes.def (V2HF): Define.
23768 * config/arm/arm_neon.h (vfmlal_low_u32, vfmlsl_low_u32,
23769 vfmlal_high_u32, vfmlsl_high_u32, vfmlalq_low_u32,
23770 vfmlslq_low_u32, vfmlalq_high_u32, vfmlslq_high_u32): Define.
23771 * config/arm/arm_neon_builtins.def (vfmal_low, vfmal_high,
23772 vfmsl_low, vfmsl_high): New set of builtins.
23773 * config/arm/iterators.md (PLUSMINUS): New code iterator.
23774 (vfml_op): New code attribute.
23775 (VFMLHALVES): New int iterator.
23776 (VFML, VFMLSEL): New mode attributes.
23777 (V_reg): Define mapping for V2HF.
23778 (V_hi, V_lo): New mode attributes.
23779 (VF_constraint): Likewise.
23780 (vfml_half, vfml_half_selector): New int attributes.
23781 * config/arm/neon.md (neon_vfm<vfml_op>l_<vfml_half><mode>): New
23783 (vfmal_low<mode>_intrinsic, vfmsl_high<mode>_intrinsic,
23784 vfmal_high<mode>_intrinsic, vfmsl_low<mode>_intrinsic):
23786 * config/arm/t-arm-elf (v8_fps): Add fp16fml.
23787 * config/arm/t-multilib (v8_2_a_simd_variants): Add fp16fml.
23788 * config/arm/unspecs.md (UNSPEC_VFML_LO, UNSPEC_VFML_HI): New unspecs.
23789 * doc/invoke.texi (ARM Options): Document fp16fml. Update armv8.4-a
23791 * doc/sourcebuild.texi (arm_fp16fml_neon_ok, arm_fp16fml_neon):
23792 Document new effective target and option set.
23794 2018-01-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23796 * config/arm/arm-cpus.in (armv8_4): New feature.
23797 (ARMv8_4a): New fgroup.
23798 (armv8.4-a): New arch.
23799 * config/arm/arm-tables.opt: Regenerate.
23800 * config/arm/t-aprofile: Add matching rules for -march=armv8.4-a.
23801 * config/arm/t-arm-elf (all_v8_archs): Add armv8.4-a.
23802 * config/arm/t-multilib (v8_4_a_simd_variants): New variable.
23803 Add matching rules for -march=armv8.4-a and extensions.
23804 * doc/invoke.texi (ARM Options): Document -march=armv8.4-a.
23806 2018-01-11 Oleg Endo <olegendo@gcc.gnu.org>
23809 * config/rx/rx.md (BW): New mode attribute.
23810 (sync_lock_test_and_setsi): Add mode suffix to insn output.
23812 2018-01-11 Richard Biener <rguenther@suse.de>
23814 PR tree-optimization/83435
23815 * graphite.c (canonicalize_loop_form): Ignore fake loop exit edges.
23816 * graphite-scop-detection.c (scop_detection::get_sese): Likewise.
23817 * tree-vrp.c (add_assert_info): Drop TREE_OVERFLOW if they appear.
23819 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
23820 Alan Hayward <alan.hayward@arm.com>
23821 David Sherwood <david.sherwood@arm.com>
23823 * config/aarch64/aarch64.c (aarch64_address_info): Add a const_offset
23825 (aarch64_classify_address): Initialize it. Track polynomial offsets.
23826 (aarch64_print_address_internal): Use it to check for a zero offset.
23828 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
23829 Alan Hayward <alan.hayward@arm.com>
23830 David Sherwood <david.sherwood@arm.com>
23832 * config/aarch64/aarch64-modes.def (NUM_POLY_INT_COEFFS): Set to 2.
23833 * config/aarch64/aarch64-protos.h (aarch64_initial_elimination_offset):
23834 Return a poly_int64 rather than a HOST_WIDE_INT.
23835 (aarch64_offset_7bit_signed_scaled_p): Take the offset as a poly_int64
23836 rather than a HOST_WIDE_INT.
23837 * config/aarch64/aarch64.h (aarch64_frame): Protect with
23838 HAVE_POLY_INT_H rather than HOST_WIDE_INT. Change locals_offset,
23839 hard_fp_offset, frame_size, initial_adjust, callee_offset and
23840 final_offset from HOST_WIDE_INT to poly_int64.
23841 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args): Use
23842 to_constant when getting the number of units in an Advanced SIMD
23844 (aarch64_builtin_vectorized_function): Check for a constant number
23846 * config/aarch64/aarch64-simd.md (mov<mode>): Handle polynomial
23848 (aarch64_ld<VSTRUCT:nregs>_lane<VALLDIF:mode>): Use the nunits
23849 attribute instead of GET_MODE_NUNITS.
23850 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs)
23851 (aarch64_class_max_nregs): Use the constant_lowest_bound of the
23852 GET_MODE_SIZE for fixed-size registers.
23853 (aarch64_const_vec_all_same_in_range_p): Use const_vec_duplicate_p.
23854 (aarch64_hard_regno_call_part_clobbered, aarch64_classify_index)
23855 (aarch64_mode_valid_for_sched_fusion_p, aarch64_classify_address)
23856 (aarch64_legitimize_address_displacement, aarch64_secondary_reload)
23857 (aarch64_print_operand, aarch64_print_address_internal)
23858 (aarch64_address_cost, aarch64_rtx_costs, aarch64_register_move_cost)
23859 (aarch64_short_vector_p, aapcs_vfp_sub_candidate)
23860 (aarch64_simd_attr_length_rglist, aarch64_operands_ok_for_ldpstp):
23861 Handle polynomial GET_MODE_SIZE.
23862 (aarch64_hard_regno_caller_save_mode): Likewise. Return modes
23863 wider than SImode without modification.
23864 (tls_symbolic_operand_type): Use strip_offset instead of split_const.
23865 (aarch64_pass_by_reference, aarch64_layout_arg, aarch64_pad_reg_upward)
23866 (aarch64_gimplify_va_arg_expr): Assert that we don't yet handle
23867 passing and returning SVE modes.
23868 (aarch64_function_value, aarch64_layout_arg): Use gen_int_mode
23869 rather than GEN_INT.
23870 (aarch64_emit_probe_stack_range): Take the size as a poly_int64
23871 rather than a HOST_WIDE_INT, but call sorry if it isn't constant.
23872 (aarch64_allocate_and_probe_stack_space): Likewise.
23873 (aarch64_layout_frame): Cope with polynomial offsets.
23874 (aarch64_save_callee_saves, aarch64_restore_callee_saves): Take the
23875 start_offset as a poly_int64 rather than a HOST_WIDE_INT. Track
23876 polynomial offsets.
23877 (offset_9bit_signed_unscaled_p, offset_12bit_unsigned_scaled_p)
23878 (aarch64_offset_7bit_signed_scaled_p): Take the offset as a
23879 poly_int64 rather than a HOST_WIDE_INT.
23880 (aarch64_get_separate_components, aarch64_process_components)
23881 (aarch64_expand_prologue, aarch64_expand_epilogue)
23882 (aarch64_use_return_insn_p): Handle polynomial frame offsets.
23883 (aarch64_anchor_offset): New function, split out from...
23884 (aarch64_legitimize_address): ...here.
23885 (aarch64_builtin_vectorization_cost): Handle polynomial
23886 TYPE_VECTOR_SUBPARTS.
23887 (aarch64_simd_check_vect_par_cnst_half): Handle polynomial
23889 (aarch64_simd_make_constant, aarch64_expand_vector_init): Get the
23890 number of elements from the PARALLEL rather than the mode.
23891 (aarch64_shift_truncation_mask): Use GET_MODE_UNIT_BITSIZE
23892 rather than GET_MODE_BITSIZE.
23893 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_ext)
23894 (aarch64_evpc_rev, aarch64_evpc_dup, aarch64_evpc_zip)
23895 (aarch64_expand_vec_perm_const_1): Handle polynomial
23896 d->perm.length () and d->perm elements.
23897 (aarch64_evpc_tbl): Likewise. Use nelt rather than GET_MODE_NUNITS.
23898 Apply to_constant to d->perm elements.
23899 (aarch64_simd_valid_immediate, aarch64_vec_fpconst_pow_of_2): Handle
23900 polynomial CONST_VECTOR_NUNITS.
23901 (aarch64_move_pointer): Take amount as a poly_int64 rather
23903 (aarch64_progress_pointer): Avoid temporary variable.
23904 * config/aarch64/aarch64.md (aarch64_<crc_variant>): Use
23905 the mode attribute instead of GET_MODE.
23907 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
23908 Alan Hayward <alan.hayward@arm.com>
23909 David Sherwood <david.sherwood@arm.com>
23911 * config/aarch64/aarch64.c (aarch64_force_temporary): Assert that
23912 x exists before using it.
23913 (aarch64_add_constant_internal): Rename to...
23914 (aarch64_add_offset_1): ...this. Replace regnum with separate
23915 src and dest rtxes. Handle the case in which they're different,
23916 including when the offset is zero. Replace scratchreg with an rtx.
23917 Use 2 additions if there is no spare register into which we can
23918 move a 16-bit constant.
23919 (aarch64_add_constant): Delete.
23920 (aarch64_add_offset): Replace reg with separate src and dest
23921 rtxes. Take a poly_int64 offset instead of a HOST_WIDE_INT.
23922 Use aarch64_add_offset_1.
23923 (aarch64_add_sp, aarch64_sub_sp): Take the scratch register as
23924 an rtx rather than an int. Take the delta as a poly_int64
23925 rather than a HOST_WIDE_INT. Use aarch64_add_offset.
23926 (aarch64_expand_mov_immediate): Update uses of aarch64_add_offset.
23927 (aarch64_expand_prologue): Update calls to aarch64_sub_sp,
23928 aarch64_allocate_and_probe_stack_space and aarch64_add_offset.
23929 (aarch64_expand_epilogue): Update calls to aarch64_add_offset
23930 and aarch64_add_sp.
23931 (aarch64_output_mi_thunk): Use aarch64_add_offset rather than
23932 aarch64_add_constant.
23934 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
23936 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
23937 Use scalar_float_mode.
23939 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
23941 * config/aarch64/aarch64-simd.md
23942 (aarch64_fml<f16mac1>l<f16quad>_low<mode>): Avoid GET_MODE_NUNITS.
23943 (aarch64_fml<f16mac1>l<f16quad>_high<mode>): Likewise.
23944 (aarch64_fml<f16mac1>l_lane_lowv2sf): Likewise.
23945 (aarch64_fml<f16mac1>l_lane_highv2sf): Likewise.
23946 (aarch64_fml<f16mac1>lq_laneq_lowv4sf): Likewise.
23947 (aarch64_fml<f16mac1>lq_laneq_highv4sf): Likewise.
23948 (aarch64_fml<f16mac1>l_laneq_lowv2sf): Likewise.
23949 (aarch64_fml<f16mac1>l_laneq_highv2sf): Likewise.
23950 (aarch64_fml<f16mac1>lq_lane_lowv4sf): Likewise.
23951 (aarch64_fml<f16mac1>lq_lane_highv4sf): Likewise.
23953 2018-01-11 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
23956 * config/arm/arm.c (arm_declare_function_name): Set arch_to_print if
23957 targ_options->x_arm_arch_string is non NULL.
23959 2018-01-11 Tamar Christina <tamar.christina@arm.com>
23961 * config/aarch64/aarch64.h
23962 (AARCH64_FL_FOR_ARCH8_4): Add AARCH64_FL_DOTPROD.
23964 2018-01-11 Sudakshina Das <sudi.das@arm.com>
23967 * expmed.c (emit_store_flag_force): Swap if const op0
23968 and change VOIDmode to mode of op0.
23970 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
23972 PR rtl-optimization/83761
23973 * caller-save.c (replace_reg_with_saved_mem): Pass bits rather
23974 than bytes to mode_for_size.
23976 2018-01-10 Jan Hubicka <hubicka@ucw.cz>
23978 PR middle-end/83189
23979 * gfortran.fortran-torture/compile/pr83189.f90: New testcase.
23980 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Handle zero
23983 2018-01-10 Jan Hubicka <hubicka@ucw.cz>
23985 PR middle-end/83575
23986 * cfgrtl.c (rtl_verify_edges): Only verify fixability of partition
23987 when in layout mode.
23988 (cfg_layout_finalize): Do not verify cfg before we are out of layout.
23989 * cfgcleanup.c (try_optimize_cfg): Only verify flow info when doing
23992 2018-01-10 Michael Collison <michael.collison@arm.com>
23994 * config/aarch64/aarch64-modes.def (V2HF): New VECTOR_MODE.
23995 * config/aarch64/aarch64-option-extension.def: Add
23996 AARCH64_OPT_EXTENSION of 'fp16fml'.
23997 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
23998 (__ARM_FEATURE_FP16_FML): Define if TARGET_F16FML is true.
23999 * config/aarch64/predicates.md (aarch64_lane_imm3): New predicate.
24000 * config/aarch64/constraints.md (Ui7): New constraint.
24001 * config/aarch64/iterators.md (VFMLA_W): New mode iterator.
24002 (VFMLA_SEL_W): Ditto.
24005 (VFMLA16_LOW): New int iterator.
24006 (VFMLA16_HIGH): Ditto.
24007 (UNSPEC_FMLAL): New unspec.
24008 (UNSPEC_FMLSL): Ditto.
24009 (UNSPEC_FMLAL2): Ditto.
24010 (UNSPEC_FMLSL2): Ditto.
24011 (f16mac): New code attribute.
24012 * config/aarch64/aarch64-simd-builtins.def
24013 (aarch64_fmlal_lowv2sf): Ditto.
24014 (aarch64_fmlsl_lowv2sf): Ditto.
24015 (aarch64_fmlalq_lowv4sf): Ditto.
24016 (aarch64_fmlslq_lowv4sf): Ditto.
24017 (aarch64_fmlal_highv2sf): Ditto.
24018 (aarch64_fmlsl_highv2sf): Ditto.
24019 (aarch64_fmlalq_highv4sf): Ditto.
24020 (aarch64_fmlslq_highv4sf): Ditto.
24021 (aarch64_fmlal_lane_lowv2sf): Ditto.
24022 (aarch64_fmlsl_lane_lowv2sf): Ditto.
24023 (aarch64_fmlal_laneq_lowv2sf): Ditto.
24024 (aarch64_fmlsl_laneq_lowv2sf): Ditto.
24025 (aarch64_fmlalq_lane_lowv4sf): Ditto.
24026 (aarch64_fmlsl_lane_lowv4sf): Ditto.
24027 (aarch64_fmlalq_laneq_lowv4sf): Ditto.
24028 (aarch64_fmlsl_laneq_lowv4sf): Ditto.
24029 (aarch64_fmlal_lane_highv2sf): Ditto.
24030 (aarch64_fmlsl_lane_highv2sf): Ditto.
24031 (aarch64_fmlal_laneq_highv2sf): Ditto.
24032 (aarch64_fmlsl_laneq_highv2sf): Ditto.
24033 (aarch64_fmlalq_lane_highv4sf): Ditto.
24034 (aarch64_fmlsl_lane_highv4sf): Ditto.
24035 (aarch64_fmlalq_laneq_highv4sf): Ditto.
24036 (aarch64_fmlsl_laneq_highv4sf): Ditto.
24037 * config/aarch64/aarch64-simd.md:
24038 (aarch64_fml<f16mac1>l<f16quad>_low<mode>): New pattern.
24039 (aarch64_fml<f16mac1>l<f16quad>_high<mode>): Ditto.
24040 (aarch64_simd_fml<f16mac1>l<f16quad>_low<mode>): Ditto.
24041 (aarch64_simd_fml<f16mac1>l<f16quad>_high<mode>): Ditto.
24042 (aarch64_fml<f16mac1>l_lane_lowv2sf): Ditto.
24043 (aarch64_fml<f16mac1>l_lane_highv2sf): Ditto.
24044 (aarch64_simd_fml<f16mac>l_lane_lowv2sf): Ditto.
24045 (aarch64_simd_fml<f16mac>l_lane_highv2sf): Ditto.
24046 (aarch64_fml<f16mac1>lq_laneq_lowv4sf): Ditto.
24047 (aarch64_fml<f16mac1>lq_laneq_highv4sf): Ditto.
24048 (aarch64_simd_fml<f16mac>lq_laneq_lowv4sf): Ditto.
24049 (aarch64_simd_fml<f16mac>lq_laneq_highv4sf): Ditto.
24050 (aarch64_fml<f16mac1>l_laneq_lowv2sf): Ditto.
24051 (aarch64_fml<f16mac1>l_laneq_highv2sf): Ditto.
24052 (aarch64_simd_fml<f16mac>l_laneq_lowv2sf): Ditto.
24053 (aarch64_simd_fml<f16mac>l_laneq_highv2sf): Ditto.
24054 (aarch64_fml<f16mac1>lq_lane_lowv4sf): Ditto.
24055 (aarch64_fml<f16mac1>lq_lane_highv4sf): Ditto.
24056 (aarch64_simd_fml<f16mac>lq_lane_lowv4sf): Ditto.
24057 (aarch64_simd_fml<f16mac>lq_lane_highv4sf): Ditto.
24058 * config/aarch64/arm_neon.h (vfmlal_low_u32): New intrinsic.
24059 (vfmlsl_low_u32): Ditto.
24060 (vfmlalq_low_u32): Ditto.
24061 (vfmlslq_low_u32): Ditto.
24062 (vfmlal_high_u32): Ditto.
24063 (vfmlsl_high_u32): Ditto.
24064 (vfmlalq_high_u32): Ditto.
24065 (vfmlslq_high_u32): Ditto.
24066 (vfmlal_lane_low_u32): Ditto.
24067 (vfmlsl_lane_low_u32): Ditto.
24068 (vfmlal_laneq_low_u32): Ditto.
24069 (vfmlsl_laneq_low_u32): Ditto.
24070 (vfmlalq_lane_low_u32): Ditto.
24071 (vfmlslq_lane_low_u32): Ditto.
24072 (vfmlalq_laneq_low_u32): Ditto.
24073 (vfmlslq_laneq_low_u32): Ditto.
24074 (vfmlal_lane_high_u32): Ditto.
24075 (vfmlsl_lane_high_u32): Ditto.
24076 (vfmlal_laneq_high_u32): Ditto.
24077 (vfmlsl_laneq_high_u32): Ditto.
24078 (vfmlalq_lane_high_u32): Ditto.
24079 (vfmlslq_lane_high_u32): Ditto.
24080 (vfmlalq_laneq_high_u32): Ditto.
24081 (vfmlslq_laneq_high_u32): Ditto.
24082 * config/aarch64/aarch64.h (AARCH64_FL_F16SML): New flag.
24083 (AARCH64_FL_FOR_ARCH8_4): New.
24084 (AARCH64_ISA_F16FML): New ISA flag.
24085 (TARGET_F16FML): New feature flag for fp16fml.
24086 (doc/invoke.texi): Document new fp16fml option.
24088 2018-01-10 Michael Collison <michael.collison@arm.com>
24090 * config/aarch64/aarch64-builtins.c:
24091 (aarch64_types_ternopu_imm_qualifiers, TYPES_TERNOPUI): New.
24092 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
24093 (__ARM_FEATURE_SHA3): Define if TARGET_SHA3 is true.
24094 * config/aarch64/aarch64.h (AARCH64_FL_SHA3): New flags.
24095 (AARCH64_ISA_SHA3): New ISA flag.
24096 (TARGET_SHA3): New feature flag for sha3.
24097 * config/aarch64/iterators.md (sha512_op): New int attribute.
24098 (CRYPTO_SHA512): New int iterator.
24099 (UNSPEC_SHA512H): New unspec.
24100 (UNSPEC_SHA512H2): Ditto.
24101 (UNSPEC_SHA512SU0): Ditto.
24102 (UNSPEC_SHA512SU1): Ditto.
24103 * config/aarch64/aarch64-simd-builtins.def
24104 (aarch64_crypto_sha512hqv2di): New builtin.
24105 (aarch64_crypto_sha512h2qv2di): Ditto.
24106 (aarch64_crypto_sha512su0qv2di): Ditto.
24107 (aarch64_crypto_sha512su1qv2di): Ditto.
24108 (aarch64_eor3qv8hi): Ditto.
24109 (aarch64_rax1qv2di): Ditto.
24110 (aarch64_xarqv2di): Ditto.
24111 (aarch64_bcaxqv8hi): Ditto.
24112 * config/aarch64/aarch64-simd.md:
24113 (aarch64_crypto_sha512h<sha512_op>qv2di): New pattern.
24114 (aarch64_crypto_sha512su0qv2di): Ditto.
24115 (aarch64_crypto_sha512su1qv2di): Ditto.
24116 (aarch64_eor3qv8hi): Ditto.
24117 (aarch64_rax1qv2di): Ditto.
24118 (aarch64_xarqv2di): Ditto.
24119 (aarch64_bcaxqv8hi): Ditto.
24120 * config/aarch64/arm_neon.h (vsha512hq_u64): New intrinsic.
24121 (vsha512h2q_u64): Ditto.
24122 (vsha512su0q_u64): Ditto.
24123 (vsha512su1q_u64): Ditto.
24124 (veor3q_u16): Ditto.
24125 (vrax1q_u64): Ditto.
24126 (vxarq_u64): Ditto.
24127 (vbcaxq_u16): Ditto.
24128 * config/arm/types.md (crypto_sha512): New type attribute.
24129 (crypto_sha3): Ditto.
24130 (doc/invoke.texi): Document new sha3 option.
24132 2018-01-10 Michael Collison <michael.collison@arm.com>
24134 * config/aarch64/aarch64-builtins.c:
24135 (aarch64_types_quadopu_imm_qualifiers, TYPES_QUADOPUI): New.
24136 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
24137 (__ARM_FEATURE_SM3): Define if TARGET_SM4 is true.
24138 (__ARM_FEATURE_SM4): Define if TARGET_SM4 is true.
24139 * config/aarch64/aarch64.h (AARCH64_FL_SM4): New flags.
24140 (AARCH64_ISA_SM4): New ISA flag.
24141 (TARGET_SM4): New feature flag for sm4.
24142 * config/aarch64/aarch64-simd-builtins.def
24143 (aarch64_sm3ss1qv4si): Ditto.
24144 (aarch64_sm3tt1aq4si): Ditto.
24145 (aarch64_sm3tt1bq4si): Ditto.
24146 (aarch64_sm3tt2aq4si): Ditto.
24147 (aarch64_sm3tt2bq4si): Ditto.
24148 (aarch64_sm3partw1qv4si): Ditto.
24149 (aarch64_sm3partw2qv4si): Ditto.
24150 (aarch64_sm4eqv4si): Ditto.
24151 (aarch64_sm4ekeyqv4si): Ditto.
24152 * config/aarch64/aarch64-simd.md:
24153 (aarch64_sm3ss1qv4si): Ditto.
24154 (aarch64_sm3tt<sm3tt_op>qv4si): Ditto.
24155 (aarch64_sm3partw<sm3part_op>qv4si): Ditto.
24156 (aarch64_sm4eqv4si): Ditto.
24157 (aarch64_sm4ekeyqv4si): Ditto.
24158 * config/aarch64/iterators.md (sm3tt_op): New int iterator.
24159 (sm3part_op): Ditto.
24160 (CRYPTO_SM3TT): Ditto.
24161 (CRYPTO_SM3PART): Ditto.
24162 (UNSPEC_SM3SS1): New unspec.
24163 (UNSPEC_SM3TT1A): Ditto.
24164 (UNSPEC_SM3TT1B): Ditto.
24165 (UNSPEC_SM3TT2A): Ditto.
24166 (UNSPEC_SM3TT2B): Ditto.
24167 (UNSPEC_SM3PARTW1): Ditto.
24168 (UNSPEC_SM3PARTW2): Ditto.
24169 (UNSPEC_SM4E): Ditto.
24170 (UNSPEC_SM4EKEY): Ditto.
24171 * config/aarch64/constraints.md (Ui2): New constraint.
24172 * config/aarch64/predicates.md (aarch64_imm2): New predicate.
24173 * config/arm/types.md (crypto_sm3): New type attribute.
24174 (crypto_sm4): Ditto.
24175 * config/aarch64/arm_neon.h (vsm3ss1q_u32): New intrinsic.
24176 (vsm3tt1aq_u32): Ditto.
24177 (vsm3tt1bq_u32): Ditto.
24178 (vsm3tt2aq_u32): Ditto.
24179 (vsm3tt2bq_u32): Ditto.
24180 (vsm3partw1q_u32): Ditto.
24181 (vsm3partw2q_u32): Ditto.
24182 (vsm4eq_u32): Ditto.
24183 (vsm4ekeyq_u32): Ditto.
24184 (doc/invoke.texi): Document new sm4 option.
24186 2018-01-10 Michael Collison <michael.collison@arm.com>
24188 * config/aarch64/aarch64-arches.def (armv8.4-a): New architecture.
24189 * config/aarch64/aarch64.h (AARCH64_ISA_V8_4): New ISA flag.
24190 (AARCH64_FL_FOR_ARCH8_4): New.
24191 (AARCH64_FL_V8_4): New flag.
24192 (doc/invoke.texi): Document new armv8.4-a option.
24194 2018-01-10 Michael Collison <michael.collison@arm.com>
24196 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
24197 (__ARM_FEATURE_AES): Define if TARGET_AES is true.
24198 (__ARM_FEATURE_SHA2): Define if TARGET_SHA2 is true.
24199 * config/aarch64/aarch64-option-extension.def: Add
24200 AARCH64_OPT_EXTENSION of 'sha2'.
24201 (aes): Add AARCH64_OPT_EXTENSION of 'aes'.
24202 (crypto): Disable sha2 and aes if crypto disabled.
24203 (crypto): Enable aes and sha2 if enabled.
24204 (simd): Disable sha2 and aes if simd disabled.
24205 * config/aarch64/aarch64.h (AARCH64_FL_AES, AARCH64_FL_SHA2):
24207 (AARCH64_ISA_AES, AARCH64_ISA_SHA2): New ISA flags.
24208 (TARGET_SHA2): New feature flag for sha2.
24209 (TARGET_AES): New feature flag for aes.
24210 * config/aarch64/aarch64-simd.md:
24211 (aarch64_crypto_aes<aes_op>v16qi): Make pattern
24212 conditional on TARGET_AES.
24213 (aarch64_crypto_aes<aesmc_op>v16qi): Ditto.
24214 (aarch64_crypto_sha1hsi): Make pattern conditional
24216 (aarch64_crypto_sha1hv4si): Ditto.
24217 (aarch64_be_crypto_sha1hv4si): Ditto.
24218 (aarch64_crypto_sha1su1v4si): Ditto.
24219 (aarch64_crypto_sha1<sha1_op>v4si): Ditto.
24220 (aarch64_crypto_sha1su0v4si): Ditto.
24221 (aarch64_crypto_sha256h<sha256_op>v4si): Ditto.
24222 (aarch64_crypto_sha256su0v4si): Ditto.
24223 (aarch64_crypto_sha256su1v4si): Ditto.
24224 (doc/invoke.texi): Document new aes and sha2 options.
24226 2018-01-10 Martin Sebor <msebor@redhat.com>
24228 PR tree-optimization/83781
24229 * gimple-fold.c (get_range_strlen): Avoid treating arrays of pointers
24232 2018-01-11 Martin Sebor <msebor@gmail.com>
24233 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
24235 PR tree-optimization/83501
24236 PR tree-optimization/81703
24238 * tree-ssa-strlen.c (get_string_cst): Rename...
24239 (get_string_len): ...to this. Handle global constants.
24240 (handle_char_store): Adjust.
24242 2018-01-10 Kito Cheng <kito.cheng@gmail.com>
24243 Jim Wilson <jimw@sifive.com>
24245 * config/riscv/riscv-protos.h (riscv_output_return): New.
24246 * config/riscv/riscv.c (struct machine_function): New naked_p field.
24247 (riscv_attribute_table, riscv_output_return),
24248 (riscv_handle_fndecl_attribute, riscv_naked_function_p),
24249 (riscv_allocate_stack_slots_for_args, riscv_warn_func_return): New.
24250 (riscv_compute_frame_info): Only compute frame->mask if not a naked
24252 (riscv_expand_prologue): Add early return for naked function.
24253 (riscv_expand_epilogue): Likewise.
24254 (riscv_function_ok_for_sibcall): Return false for naked function.
24255 (riscv_set_current_function): New.
24256 (TARGET_SET_CURRENT_FUNCTION, TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS),
24257 (TARGET_ATTRIBUTE_TABLE, TARGET_WARN_FUNC_RETURN): New.
24258 * config/riscv/riscv.md (simple_return): Call riscv_output_return.
24259 * doc/extend.texi (RISC-V Function Attributes): New.
24261 2018-01-10 Michael Meissner <meissner@linux.vnet.ibm.com>
24263 * config/rs6000/rs6000.c (is_complex_IBM_long_double): Explicitly
24264 check for 128-bit long double before checking TCmode.
24265 * config/rs6000/rs6000.h (FLOAT128_IEEE_P): Explicitly check for
24266 128-bit long doubles before checking TFmode or TCmode.
24267 (FLOAT128_IBM_P): Likewise.
24269 2018-01-10 Martin Sebor <msebor@redhat.com>
24271 PR tree-optimization/83671
24272 * builtins.c (c_strlen): Unconditionally return zero for the empty
24274 Use -Warray-bounds for warnings.
24275 * gimple-fold.c (get_range_strlen): Handle non-constant lengths
24276 for non-constant array indices with COMPONENT_REF, arrays of
24277 arrays, and pointers to arrays.
24278 (gimple_fold_builtin_strlen): Determine and set length range for
24279 non-constant character arrays.
24281 2018-01-10 Aldy Hernandez <aldyh@redhat.com>
24283 PR middle-end/81897
24284 * tree-ssa-uninit.c (convert_control_dep_chain_into_preds): Skip
24287 2018-01-10 Eric Botcazou <ebotcazou@adacore.com>
24289 * dwarf2out.c (dwarf2out_var_location): Do not pass NULL to fprintf.
24291 2018-01-10 Peter Bergner <bergner@vnet.ibm.com>
24294 * config/rs6000/rs6000.c (print_operand) <'y'>: Use
24295 VECTOR_MEM_ALTIVEC_OR_VSX_P.
24296 * config/rs6000/vsx.md (*vsx_le_perm_load_<mode> for VSX_D): Use
24297 indexed_or_indirect_operand predicate.
24298 (*vsx_le_perm_load_<mode> for VSX_W): Likewise.
24299 (*vsx_le_perm_load_v8hi): Likewise.
24300 (*vsx_le_perm_load_v16qi): Likewise.
24301 (*vsx_le_perm_store_<mode> for VSX_D): Likewise.
24302 (*vsx_le_perm_store_<mode> for VSX_W): Likewise.
24303 (*vsx_le_perm_store_v8hi): Likewise.
24304 (*vsx_le_perm_store_v16qi): Likewise.
24305 (eight unnamed splitters): Likewise.
24307 2018-01-10 Peter Bergner <bergner@vnet.ibm.com>
24309 * config/rs6000/x86intrin.h: Change #warning to #error. Update message.
24310 * config/rs6000/emmintrin.h: Likewise.
24311 * config/rs6000/mmintrin.h: Likewise.
24312 * config/rs6000/xmmintrin.h: Likewise.
24314 2018-01-10 David Malcolm <dmalcolm@redhat.com>
24317 * tree-core.h: Document EXPR_LOCATION_WRAPPER_P's usage of
24319 * tree.c (tree_nop_conversion): Return true for location wrapper
24321 (maybe_wrap_with_location): New function.
24322 (selftest::check_strip_nops): New function.
24323 (selftest::test_location_wrappers): New function.
24324 (selftest::tree_c_tests): Call it.
24325 * tree.h (STRIP_ANY_LOCATION_WRAPPER): New macro.
24326 (maybe_wrap_with_location): New decl.
24327 (EXPR_LOCATION_WRAPPER_P): New macro.
24328 (location_wrapper_p): New inline function.
24329 (tree_strip_any_location_wrapper): New inline function.
24331 2018-01-10 H.J. Lu <hongjiu.lu@intel.com>
24334 * config/i386/i386.c (ix86_compute_frame_layout): Always adjust
24335 stack_realign_offset for the largest alignment of stack slot
24337 (ix86_find_max_used_stack_alignment): New function.
24338 (ix86_finalize_stack_frame_flags): Use it. Set
24339 max_used_stack_alignment if we don't realign stack.
24340 * config/i386/i386.h (machine_function): Add
24341 max_used_stack_alignment.
24343 2018-01-10 Christophe Lyon <christophe.lyon@linaro.org>
24345 * config/arm/arm.opt (-mbranch-cost): New option.
24346 * config/arm/arm.h (BRANCH_COST): Take arm_branch_cost into
24349 2018-01-10 Segher Boessenkool <segher@kernel.crashing.org>
24352 * config/rs6000/rs6000.md (load_toc_v4_PIC_2, load_toc_v4_PIC_3b,
24353 load_toc_v4_PIC_3c): Wrap const term in CONST RTL.
24355 2018-01-10 Richard Biener <rguenther@suse.de>
24358 * dwarf2out.c (gen_subprogram_die): Hoist old_die && declaration
24359 early out so it also covers the case where we have a non-NULL
24362 2018-01-10 Richard Sandiford <richard.sandiford@linaro.org>
24364 PR tree-optimization/83753
24365 * tree-vect-stmts.c (get_group_load_store_type): Use VMAT_CONTIGUOUS
24366 for non-strided grouped accesses if the number of elements is 1.
24368 2018-01-10 Jan Hubicka <hubicka@ucw.cz>
24371 * i386.c (ix86_vectorize_builtin_gather): Check TARGET_USE_GATHER.
24372 * i386.h (TARGET_USE_GATHER): Define.
24373 * x86-tune.def (X86_TUNE_USE_GATHER): New.
24375 2018-01-10 Martin Liska <mliska@suse.cz>
24378 * basic-block.h (CLEANUP_NO_PARTITIONING): New define.
24379 * bb-reorder.c (pass_reorder_blocks::execute): Do not clean up
24381 * cfgcleanup.c (try_optimize_cfg): Fix up partitioning if
24382 CLEANUP_NO_PARTITIONING is not set.
24384 2018-01-10 Richard Sandiford <richard.sandiford@linaro.org>
24386 * doc/rtl.texi: Remove documentation of (const ...) wrappers
24387 for vectors, as a partial revert of r254296.
24388 * rtl.h (const_vec_p): Delete.
24389 (const_vec_duplicate_p): Don't test for vector CONSTs.
24390 (unwrap_const_vec_duplicate, const_vec_series_p): Likewise.
24391 * expmed.c (make_tree): Likewise.
24394 * common.md (E, F): Use CONSTANT_P instead of checking for
24396 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
24397 checking for CONST_VECTOR.
24399 2018-01-09 Jan Hubicka <hubicka@ucw.cz>
24401 PR middle-end/83575
24402 * predict.c (force_edge_cold): Handle in more sane way edges
24403 with no prediction.
24405 2018-01-09 Carl Love <cel@us.ibm.com>
24407 * config/rs6002/altivec.md (p8_vmrgow): Add support for V2DI, V2DF,
24409 (p8_vmrgew): Add support for V2DI, V2DF, V4SF types.
24410 * config/rs6000/rs6000-builtin.def: Add definitions for FLOAT2_V2DF,
24411 VMRGEW_V2DI, VMRGEW_V2DF, VMRGEW_V4SF, VMRGOW_V4SI, VMRGOW_V4SF,
24412 VMRGOW_V2DI, VMRGOW_V2DF. Remove definition for VMRGOW.
24413 * config/rs6000/rs6000-c.c (VSX_BUILTIN_VEC_FLOAT2,
24414 P8V_BUILTIN_VEC_VMRGEW, P8V_BUILTIN_VEC_VMRGOW): Add definitions.
24415 * config/rs6000/rs6000-protos.h: Add extern defition for
24416 rs6000_generate_float2_double_code.
24417 * config/rs6000/rs6000.c (rs6000_generate_float2_double_code): Add
24419 * config/rs6000/vsx.md (vsx_xvcdpsp): Add define_insn.
24420 (float2_v2df): Add define_expand.
24422 2018-01-09 Uros Bizjak <ubizjak@gmail.com>
24425 * combine.c (force_int_to_mode) <case ASHIFT>: Use mode instead of
24426 op_mode in the force_to_mode call.
24428 2018-01-09 Richard Sandiford <richard.sandiford@linaro.org>
24430 * config/aarch64/aarch64.c (aarch64_evpc_trn): Use d.perm.series_p
24431 instead of checking each element individually.
24432 (aarch64_evpc_uzp): Likewise.
24433 (aarch64_evpc_zip): Likewise.
24434 (aarch64_evpc_ext): Likewise.
24435 (aarch64_evpc_rev): Likewise.
24436 (aarch64_evpc_dup): Test the encoding for a single duplicated element,
24437 instead of checking each element individually. Return true without
24439 (aarch64_vectorize_vec_perm_const): Use all_from_input_p to test
24440 whether all selected elements come from the same input, instead of
24441 checking each element individually. Remove calls to gen_rtx_REG,
24442 start_sequence and end_sequence and instead assert that no rtl is
24445 2018-01-09 Richard Sandiford <richard.sandiford@linaro.org>
24447 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p): Fix
24448 order of HIGH and CONST checks.
24450 2018-01-09 Richard Sandiford <richard.sandiford@linaro.org>
24452 * tree-vect-stmts.c (permute_vec_elements): Create a fresh variable
24453 if the destination isn't an SSA_NAME.
24455 2018-01-09 Richard Biener <rguenther@suse.de>
24457 PR tree-optimization/83668
24458 * graphite.c (canonicalize_loop_closed_ssa): Add edge argument,
24460 (canonicalize_loop_form): ... here, renamed from ...
24461 (canonicalize_loop_closed_ssa_form): ... this and amended to
24462 swap successor edges for loop exit blocks to make us use
24463 the RPO order we need for initial schedule generation.
24465 2018-01-09 Joseph Myers <joseph@codesourcery.com>
24467 PR tree-optimization/64811
24468 * match.pd: When optimizing comparisons with Inf, avoid
24469 introducing or losing exceptions from comparisons with NaN.
24471 2018-01-09 Martin Liska <mliska@suse.cz>
24474 * asan.c (shadow_mem_size): Add gcc_assert.
24476 2018-01-09 Georg-Johann Lay <avr@gjlay.de>
24478 Don't save registers in main().
24481 * doc/invoke.texi (AVR Options) [-mmain-is-OS_task]: Document it.
24482 * config/avr/avr.opt (-mmain-is-OS_task): New target option.
24483 * config/avr/avr.c (avr_set_current_function): Don't error if
24484 naked, OS_task or OS_main are specified at the same time.
24485 (avr_function_ok_for_sibcall): Don't disable sibcalls for OS_task,
24487 (avr_insert_attributes) [-mmain-is-OS_task] <main>: Add OS_task
24489 * common/config/avr/avr-common.c (avr_option_optimization_table):
24490 Switch on -mmain-is-OS_task for optimizing compilations.
24492 2018-01-09 Richard Biener <rguenther@suse.de>
24494 PR tree-optimization/83572
24495 * graphite.c: Include cfganal.h.
24496 (graphite_transform_loops): Connect infinite loops to exit
24497 and remove fake edges at the end.
24499 2018-01-09 Jan Hubicka <hubicka@ucw.cz>
24501 * ipa-inline.c (edge_badness): Revert accidental checkin.
24503 2018-01-09 Jan Hubicka <hubicka@ucw.cz>
24506 * ipa-comdats.c (set_comdat_group): Only set comdat group of real
24507 symbols; not inline clones.
24509 2018-01-09 Jakub Jelinek <jakub@redhat.com>
24512 * modulo-sched.c (schedule_reg_moves): Punt if we'd need to move
24513 hard registers. Formatting fixes.
24515 PR preprocessor/83722
24516 * gcc.c (try_generate_repro): Pass
24517 &temp_stderr_files[RETRY_ICE_ATTEMPTS - 1] rather than
24518 &temp_stdout_files[RETRY_ICE_ATTEMPTS - 1] as last argument to
24521 2018-01-08 Monk Chiang <sh.chiang04@gmail.com>
24522 Kito Cheng <kito.cheng@gmail.com>
24524 * config/riscv/riscv.c (machine_function::is_leaf): Remove field.
24525 (riscv_leaf_function_p): Delete.
24526 (riscv_function_ok_for_sibcall): Return false when TARGET_SAVE_RESTORE.
24528 2018-01-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
24530 * config/rs6000/rs6000-string.c (do_load_for_compare_from_addr): New
24532 (do_ifelse): New function.
24533 (do_isel): New function.
24534 (do_sub3): New function.
24535 (do_add3): New function.
24536 (do_load_mask_compare): New function.
24537 (do_overlap_load_compare): New function.
24538 (expand_compare_loop): New function.
24539 (expand_block_compare): Call expand_compare_loop() when appropriate.
24540 * config/rs6000/rs6000.opt (-mblock-compare-inline-limit): Change
24541 option description.
24542 (-mblock-compare-inline-loop-limit): New option.
24544 2018-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24547 * config/rs6000/altivec.md (*altivec_vpermr_<mode>_internal):
24548 Reverse order of second and third operands in first alternative.
24549 * config/rs6000/rs6000.c (rs6000_expand_vector_set): Reverse order
24550 of first and second elements in UNSPEC_VPERMR vector.
24551 (altivec_expand_vec_perm_le): Likewise.
24553 2018-01-08 Jeff Law <law@redhat.com>
24555 PR rtl-optimizatin/81308
24556 * tree-switch-conversion.c (cfg_altered): New file scoped static.
24557 (process_switch): If group_case_labels makes a change, then set
24559 (pass_convert_switch::execute): If a switch is converted, then
24560 set cfg_altered. Return TODO_cfg_cleanup if cfg_altered is true.
24562 PR rtl-optimization/81308
24563 * recog.c (split_all_insns): Conditionally cleanup the CFG after
24566 2018-01-08 Vidya Praveen <vidyapraveen@arm.com>
24568 PR target/83663 - Revert r255946
24569 * config/aarch64/aarch64.c (aarch64_expand_vector_init): Modify code
24570 generation for cases where splatting a value is not useful.
24571 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
24572 across a vec_duplicate and a paradoxical subreg forming a vector
24573 mode to a vec_concat.
24575 2018-01-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24577 * config/arm/t-aprofile (MULTILIB_MATCHES): Add mapping rules for
24578 -march=armv8.3-a variants.
24579 * config/arm/t-multilib: Likewise.
24580 * config/arm/t-arm-elf: Likewise. Handle dotprod extension.
24582 2018-01-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
24584 * config/rs6000/rs6000.md (cceq_ior_compare): Remove * so I can use it
24586 (cceq_ior_compare_complement): Give it a name so I can use it, and
24587 change boolean_or_operator predicate to boolean_operator so it can
24588 be used to generate a crand.
24589 (eqne): New code iterator.
24590 (bd/bd_neg): New code_attrs.
24591 (<bd>_<mode>): New name for ctr<mode>_internal[12] now combined into
24592 a single define_insn.
24593 (<bd>tf_<mode>): A new insn pattern for the conditional form branch
24594 decrement (bdnzt/bdnzf/bdzt/bdzf).
24595 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Updated
24596 with the new names of the branch decrement patterns, and added the
24597 names of the branch decrement conditional patterns.
24599 2018-01-08 Richard Biener <rguenther@suse.de>
24601 PR tree-optimization/83563
24602 * graphite.c (canonicalize_loop_closed_ssa_form): Reset the SCEV
24605 2018-01-08 Richard Biener <rguenther@suse.de>
24607 PR middle-end/83713
24608 * convert.c (do_narrow): Properly guard TYPE_OVERFLOW_WRAPS checks.
24610 2018-01-08 Richard Biener <rguenther@suse.de>
24612 PR tree-optimization/83685
24613 * tree-ssa-pre.c (create_expression_by_pieces): Do not insert
24614 references to abnormals.
24616 2018-01-08 Richard Biener <rguenther@suse.de>
24619 * dwarf2out.c (output_indirect_strings): Handle empty
24620 skeleton_debug_str_hash.
24621 (dwarf2out_early_finish): Index strings for -gsplit-dwarf.
24623 2018-01-08 Claudiu Zissulescu <claziss@synopsys.com>
24625 * config/arc/arc.c (TARGET_TRAMPOLINE_ADJUST_ADDRESS): Delete.
24626 (emit_store_direct): Likewise.
24627 (arc_trampoline_adjust_address): Likewise.
24628 (arc_asm_trampoline_template): New function.
24629 (arc_initialize_trampoline): Use asm_trampoline_template.
24630 (TARGET_ASM_TRAMPOLINE_TEMPLATE): Define.
24631 * config/arc/arc.h (TRAMPOLINE_SIZE): Adjust to 16.
24632 * config/arc/arc.md (flush_icache): Delete pattern.
24634 2018-01-08 Claudiu Zissulescu <claziss@synopsys.com>
24636 * config/arc/arc-c.def (__ARC_UNALIGNED__): New define.
24637 * config/arc/arc.h (STRICT_ALIGNMENT): Control this macro using
24640 2018-01-08 Sebastian Huber <sebastian.huber@embedded-brains.de>
24643 * config/epiphany/epiphany.h (make_pass_mode_switch_use): Guard
24644 by not USED_FOR_TARGET.
24645 (make_pass_resolve_sw_modes): Likewise.
24647 2018-01-08 Sebastian Huber <sebastian.huber@embedded-brains.de>
24649 * config/nios2/nios2.h (nios2_section_threshold): Guard by not
24652 2018-01-08 Richard Biener <rguenther@suse.de>
24654 PR middle-end/83580
24655 * tree-data-ref.c (split_constant_offset): Remove STRIP_NOPS.
24657 2018-01-08 Richard Biener <rguenther@suse.de>
24659 PR middle-end/83517
24660 * match.pd ((t * 2) / 2) -> t): Add missing :c.
24662 2018-01-06 Aldy Hernandez <aldyh@redhat.com>
24664 PR middle-end/81897
24665 * tree-ssa-uninit.c (compute_control_dep_chain): Do not bail on
24666 basic blocks with a small number of successors.
24667 (convert_control_dep_chain_into_preds): Improve handling of
24669 (dump_predicates): Split apart into...
24670 (dump_pred_chain): ...here...
24671 (dump_pred_info): ...and here.
24672 (can_one_predicate_be_invalidated_p): Add debugging printfs.
24673 (can_chain_union_be_invalidated_p): Improve check for invalidation
24675 (uninit_uses_cannot_happen): Avoid unnecessary if
24676 convert_control_dep_chain_into_preds yielded nothing.
24678 2018-01-06 Martin Sebor <msebor@redhat.com>
24680 PR tree-optimization/83640
24681 * gimple-ssa-warn-restrict.c (builtin_access::builtin_access): Avoid
24682 subtracting negative offset from size.
24683 (builtin_access::overlap): Adjust offset bounds of the access to fall
24684 within the size of the object if possible.
24686 2018-01-06 Richard Sandiford <richard.sandiford@linaro.org>
24688 PR rtl-optimization/83699
24689 * expmed.c (extract_bit_field_1): Restrict the vector usage of
24690 extract_bit_field_as_subreg to cases in which the extracted
24691 value is also a vector.
24693 * lra-constraints.c (process_alt_operands): Test for the equivalence
24694 substitutions when detecting a possible reload cycle.
24696 2018-01-06 Jakub Jelinek <jakub@redhat.com>
24699 * toplev.c (process_options): Don't enable debug_nonbind_markers_p
24700 by default if flag_selective_schedling{,2}. Formatting fixes.
24702 PR rtl-optimization/83682
24703 * rtl.h (const_vec_duplicate_p): Only return true for VEC_DUPLICATE
24704 if it has non-VECTOR_MODE element mode.
24705 (vec_duplicate_p): Likewise.
24707 PR middle-end/83694
24708 * cfgexpand.c (expand_debug_expr): Punt if mode1 is VOIDmode
24709 and bitsize might be greater than MAX_BITSIZE_MODE_ANY_INT.
24711 2018-01-05 Jakub Jelinek <jakub@redhat.com>
24714 * config/i386/i386-builtin.def
24715 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
24716 __builtin_ia32_vgf2p8affineqb_v64qi, __builtin_ia32_vgf2p8mulb_v64qi):
24717 Require also OPTION_MASK_ISA_AVX512F in addition to
24718 OPTION_MASK_ISA_GFNI.
24719 (__builtin_ia32_vgf2p8affineinvqb_v16qi_mask,
24720 __builtin_ia32_vgf2p8affineqb_v16qi_mask): Require
24721 OPTION_MASK_ISA_AVX512VL instead of OPTION_MASK_ISA_SSE in addition
24722 to OPTION_MASK_ISA_GFNI.
24723 (__builtin_ia32_vgf2p8mulb_v32qi_mask): Require
24724 OPTION_MASK_ISA_AVX512VL in addition to OPTION_MASK_ISA_GFNI and
24725 OPTION_MASK_ISA_AVX512BW.
24726 (__builtin_ia32_vgf2p8mulb_v16qi_mask): Require
24727 OPTION_MASK_ISA_AVX512VL instead of OPTION_MASK_ISA_AVX512BW in
24728 addition to OPTION_MASK_ISA_GFNI.
24729 (__builtin_ia32_vgf2p8affineinvqb_v16qi,
24730 __builtin_ia32_vgf2p8affineqb_v16qi, __builtin_ia32_vgf2p8mulb_v16qi):
24731 Require OPTION_MASK_ISA_SSE2 instead of OPTION_MASK_ISA_SSE in addition
24732 to OPTION_MASK_ISA_GFNI.
24733 * config/i386/i386.c (def_builtin): Change to builtin isa/isa2 being
24734 a requirement for all ISAs rather than any of them with a few
24736 (ix86_add_new_builtins): Clear OPTION_MASK_ISA_64BIT from isa before
24738 (ix86_expand_builtin): Require all ISAs from builtin's isa and isa2
24739 bitmasks to be enabled with 3 exceptions, instead of requiring any
24740 enabled ISA with lots of exceptions.
24741 * config/i386/sse.md (vgf2p8affineinvqb_<mode><mask_name>,
24742 vgf2p8affineqb_<mode><mask_name>, vgf2p8mulb_<mode><mask_name>):
24743 Change avx512bw in isa attribute to avx512f.
24744 * config/i386/sgxintrin.h: Add license boilerplate.
24745 * config/i386/vaesintrin.h: Likewise. Fix macro spelling __AVX512F
24746 to __AVX512F__ and __AVX512VL to __AVX512VL__.
24747 (_mm256_aesdec_epi128, _mm256_aesdeclast_epi128, _mm256_aesenc_epi128,
24748 _mm256_aesenclast_epi128): Enable temporarily avx if __AVX__ is not
24750 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8,
24751 _mm_gf2p8affineinv_epi64_epi8, _mm_gf2p8affine_epi64_epi8): Enable
24752 temporarily sse2 rather than sse if not enabled already.
24755 * config/i386/sse.md (VI248_VLBW): Rename to ...
24756 (VI248_AVX512VL): ... this. Don't guard V32HI with TARGET_AVX512BW.
24757 (vpshrd_<mode><mask_name>, vpshld_<mode><mask_name>,
24758 vpshrdv_<mode>, vpshrdv_<mode>_mask, vpshrdv_<mode>_maskz,
24759 vpshrdv_<mode>_maskz_1, vpshldv_<mode>, vpshldv_<mode>_mask,
24760 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1): Use VI248_AVX512VL
24761 mode iterator instead of VI248_VLBW.
24763 2018-01-05 Jan Hubicka <hubicka@ucw.cz>
24765 * ipa-fnsummary.c (record_modified_bb_info): Add OP.
24766 (record_modified): Skip clobbers; add debug output.
24767 (param_change_prob): Use sreal frequencies.
24769 2018-01-05 Richard Sandiford <richard.sandiford@linaro.org>
24771 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
24772 punt for user-aligned variables.
24774 2018-01-05 Richard Sandiford <richard.sandiford@linaro.org>
24776 * tree-chrec.c (chrec_contains_symbols): Return true for
24779 2018-01-05 Sudakshina Das <sudi.das@arm.com>
24782 * simplify-rtx.c (simplify_relational_operation_1): Add simplifications
24783 of (x|y) == x for BICS pattern.
24785 2018-01-05 Jakub Jelinek <jakub@redhat.com>
24787 PR tree-optimization/83605
24788 * gimple-ssa-strength-reduction.c: Include tree-eh.h.
24789 (find_candidates_dom_walker::before_dom_children): Ignore stmts that
24792 2018-01-05 Sebastian Huber <sebastian.huber@embedded-brains.de>
24794 * config.gcc (epiphany-*-elf*): Add (epiphany-*-rtems*) configuration.
24795 * config/epiphany/rtems.h: New file.
24797 2018-01-04 Jakub Jelinek <jakub@redhat.com>
24798 Uros Bizjak <ubizjak@gmail.com>
24801 * config/i386/i386.md (*<rotate_insn>hi3_1 splitter): Use
24802 QIreg_operand instead of register_operand predicate.
24803 * config/i386/i386.c (ix86_rop_should_change_byte_p,
24804 set_rop_modrm_reg_bits, ix86_mitigate_rop): Use -mmitigate-rop in
24805 comments instead of -fmitigate[-_]rop.
24807 2018-01-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
24810 * cgraphunit.c (symbol_table::compile): Switch to text_section
24811 before calling assembly_start debug hook.
24812 * run-rtl-passes.c (run_rtl_passes): Likewise.
24815 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
24817 * tree-vrp.c (extract_range_from_binary_expr_1): Check
24818 range_int_cst_p rather than !symbolic_range_p before calling
24819 extract_range_from_multiplicative_op_1.
24821 2018-01-04 Jeff Law <law@redhat.com>
24823 * tree-ssa-math-opts.c (execute_cse_reciprocals_1): Remove
24824 redundant test in assertion.
24826 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
24828 * doc/rtl.texi: Document machine_mode wrapper classes.
24830 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
24832 * fold-const.c (fold_ternary_loc): Check tree_fits_uhwi_p before
24833 using tree_to_uhwi.
24835 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
24837 * tree-ssa-forwprop.c (is_combined_permutation_identity): Allow
24838 the VEC_PERM_EXPR fold to fail.
24840 2018-01-04 Jakub Jelinek <jakub@redhat.com>
24843 * bb-reorder.c (insert_section_boundary_note): Set has_bb_partition
24844 to switched_sections.
24846 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
24849 * config/arm/arm.c (arm_vectorize_vec_perm_const): Fix inverted
24850 test for d.testing.
24852 2018-01-04 Peter Bergner <bergner@vnet.ibm.com>
24855 * config/rs6000/rs6000.c (rs6000_discover_homogeneous_aggregate): Do not
24856 allow arguments in FP registers if TARGET_HARD_FLOAT is false.
24858 2018-01-04 Jakub Jelinek <jakub@redhat.com>
24861 * cfgexpand.c (expand_debug_expr) <case BIT_FIELD_REF>: Punt if mode
24862 is BLKmode and bitpos not zero or mode change is needed.
24864 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
24867 * config/sparc/sparc.c (sparc_vectorize_vec_perm_const): Require
24870 2018-01-04 Uros Bizjak <ubizjak@gmail.com>
24873 * config/alpha/alpha.md (*sadd<modesuffix>): Use ASHIFT
24874 instead of MULT rtx. Update all corresponding splitters.
24875 (*saddl_se): Ditto.
24876 (*ssub<modesuffix>): Ditto.
24877 (*ssubl_se): Ditto.
24878 (*cmp_sadd_di): Update split patterns.
24879 (*cmp_sadd_si): Ditto.
24880 (*cmp_sadd_sidi): Ditto.
24881 (*cmp_ssub_di): Ditto.
24882 (*cmp_ssub_si): Ditto.
24883 (*cmp_ssub_sidi): Ditto.
24884 * config/alpha/predicates.md (const23_operand): New predicate.
24885 * config/alpha/alpha.c (alpha_rtx_costs) [PLUS, MINUS]:
24886 Look for ASHIFT, not MULT inner operand.
24887 (alpha_split_conditional_move): Update for *sadd<modesuffix> change.
24889 2018-01-04 Martin Liska <mliska@suse.cz>
24891 PR gcov-profile/83669
24892 * gcov.c (output_intermediate_file): Add version to intermediate
24894 * doc/gcov.texi: Document new field 'version' in intermediate
24895 file format. Fix location of '-k' option of gcov command.
24897 2018-01-04 Martin Liska <mliska@suse.cz>
24900 * ipa-icf.c (sem_function::merge): Do not cross comdat boundary.
24902 2018-01-04 Jakub Jelinek <jakub@redhat.com>
24904 * gimple-ssa-sprintf.c (parse_directive): Cast second dir.len to uhwi.
24906 2018-01-03 Martin Sebor <msebor@redhat.com>
24908 PR tree-optimization/83655
24909 * gimple-ssa-warn-restrict.c (wrestrict_dom_walker::check_call): Avoid
24910 checking calls with invalid arguments.
24912 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24914 * tree-vect-stmts.c (vect_get_store_rhs): New function.
24915 (vectorizable_mask_load_store): Delete.
24916 (vectorizable_call): Return false for masked loads and stores.
24917 (vectorizable_store): Handle IFN_MASK_STORE. Use vect_get_store_rhs
24918 instead of gimple_assign_rhs1.
24919 (vectorizable_load): Handle IFN_MASK_LOAD.
24920 (vect_transform_stmt): Don't set is_store for call_vec_info_type.
24922 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24924 * tree-vect-stmts.c (vect_build_gather_load_calls): New function,
24926 (vectorizable_mask_load_store): ...here.
24927 (vectorizable_load): ...and here.
24929 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24931 * tree-vect-stmts.c (vect_build_all_ones_mask)
24932 (vect_build_zero_merge_argument): New functions, split out from...
24933 (vectorizable_load): ...here.
24935 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24937 * tree-vect-stmts.c (vect_check_store_rhs): New function,
24939 (vectorizable_mask_load_store): ...here.
24940 (vectorizable_store): ...and here.
24942 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24944 * tree-vect-stmts.c (vect_check_load_store_mask): New function,
24946 (vectorizable_mask_load_store): ...here.
24948 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24950 * tree-vectorizer.h (vec_load_store_type): Moved from tree-vec-stmts.c
24951 (vect_model_store_cost): Take a vec_load_store_type instead of a
24953 * tree-vect-stmts.c (vec_load_store_type): Move to tree-vectorizer.h.
24954 (vect_model_store_cost): Take a vec_load_store_type instead of a
24956 (vectorizable_mask_load_store): Update accordingly.
24957 (vectorizable_store): Likewise.
24958 * tree-vect-slp.c (vect_analyze_slp_cost_1): Update accordingly.
24960 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24962 * tree-vect-loop.c (vect_transform_loop): Stub out scalar
24963 IFN_MASK_LOAD calls here rather than...
24964 * tree-vect-stmts.c (vectorizable_mask_load_store): ...here.
24966 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24967 Alan Hayward <alan.hayward@arm.com>
24968 David Sherwood <david.sherwood@arm.com>
24970 * expmed.c (extract_bit_field_1): For vector extracts,
24971 fall back to extract_bit_field_as_subreg if vec_extract
24974 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24975 Alan Hayward <alan.hayward@arm.com>
24976 David Sherwood <david.sherwood@arm.com>
24978 * lra-spills.c (pseudo_reg_slot_compare): Sort slots by whether
24979 they are variable or constant sized.
24980 (assign_stack_slot_num_and_sort_pseudos): Don't reuse variable-sized
24981 slots for constant-sized data.
24983 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24984 Alan Hayward <alan.hayward@arm.com>
24985 David Sherwood <david.sherwood@arm.com>
24987 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern): When
24988 handling COND_EXPRs with boolean comparisons, try to find a better
24989 basis for the mask type than the boolean itself.
24991 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24993 * doc/rtl.texi (MAX_BITSIZE_MODE_ANY_MODE): Describe how the default
24994 is calculated and how it can be overridden.
24995 * genmodes.c (max_bitsize_mode_any_mode): New variable.
24996 (create_modes): Initialize it from MAX_BITSIZE_MODE_ANY_MODE,
24998 (emit_max_int): Use it to set the output MAX_BITSIZE_MODE_ANY_MODE,
25001 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25002 Alan Hayward <alan.hayward@arm.com>
25003 David Sherwood <david.sherwood@arm.com>
25005 * config/aarch64/aarch64-protos.h (aarch64_output_simd_mov_immediate):
25006 Remove the mode argument.
25007 (aarch64_simd_valid_immediate): Remove the mode and inverse
25009 * config/aarch64/iterators.md (bitsize): New iterator.
25010 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>, and<mode>3)
25011 (ior<mode>3): Update calls to aarch64_output_simd_mov_immediate.
25012 * config/aarch64/constraints.md (Do, Db, Dn): Update calls to
25013 aarch64_simd_valid_immediate.
25014 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Likewise.
25015 (aarch64_reg_or_bic_imm): Likewise.
25016 * config/aarch64/aarch64.c (simd_immediate_info): Replace mvn
25017 with an insn_type enum and msl with a modifier_type enum.
25018 Replace element_width with a scalar_mode. Change the shift
25019 to unsigned int. Add constructors for scalar_float_mode and
25020 scalar_int_mode elements.
25021 (aarch64_vect_float_const_representable_p): Delete.
25022 (aarch64_can_const_movi_rtx_p)
25023 (aarch64_simd_scalar_immediate_valid_for_move)
25024 (aarch64_simd_make_constant): Update call to
25025 aarch64_simd_valid_immediate.
25026 (aarch64_advsimd_valid_immediate_hs): New function.
25027 (aarch64_advsimd_valid_immediate): Likewise.
25028 (aarch64_simd_valid_immediate): Remove mode and inverse
25029 arguments. Rewrite to use the above. Use const_vec_duplicate_p
25030 to detect duplicated constants and use aarch64_float_const_zero_rtx_p
25031 and aarch64_float_const_representable_p on the result.
25032 (aarch64_output_simd_mov_immediate): Remove mode argument.
25033 Update call to aarch64_simd_valid_immediate and use of
25034 simd_immediate_info.
25035 (aarch64_output_scalar_simd_mov_immediate): Update call
25038 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25039 Alan Hayward <alan.hayward@arm.com>
25040 David Sherwood <david.sherwood@arm.com>
25042 * machmode.h (mode_precision): Prefix with CONST_MODE_PRECISION.
25043 (mode_nunits): Likewise CONST_MODE_NUNITS.
25044 * machmode.def (ADJUST_NUNITS): Document.
25045 * genmodes.c (mode_data::need_nunits_adj): New field.
25046 (blank_mode): Update accordingly.
25047 (adj_nunits): New variable.
25048 (print_maybe_const_decl): Replace CATEGORY with a NEEDS_ADJ
25050 (emit_mode_size_inline): Set need_bytesize_adj for all modes
25051 listed in adj_nunits.
25052 (emit_mode_nunits_inline): Set need_nunits_adj for all modes
25053 listed in adj_nunits. Don't emit case statements for such modes.
25054 (emit_insn_modes_h): Emit definitions of CONST_MODE_NUNITS
25055 and CONST_MODE_PRECISION. Make CONST_MODE_SIZE expand to
25056 nothing if adj_nunits is nonnull.
25057 (emit_mode_precision, emit_mode_nunits): Use print_maybe_const_decl.
25058 (emit_mode_unit_size, emit_mode_base_align, emit_mode_ibit)
25059 (emit_mode_fbit): Update use of print_maybe_const_decl.
25060 (emit_move_size): Likewise. Treat the array as non-const
25062 (emit_mode_adjustments): Handle adj_nunits.
25064 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25066 * machmode.def (VECTOR_MODES_WITH_PREFIX): Document.
25067 * genmodes.c (VECTOR_MODES_WITH_PREFIX): New macro.
25068 (VECTOR_MODES): Use it.
25069 (make_vector_modes): Take the prefix as an argument.
25071 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25072 Alan Hayward <alan.hayward@arm.com>
25073 David Sherwood <david.sherwood@arm.com>
25075 * mode-classes.def (MODE_VECTOR_BOOL): New mode class.
25076 * machmode.h (INTEGRAL_MODE_P, VECTOR_MODE_P): Return true
25077 for MODE_VECTOR_BOOL.
25078 * machmode.def (VECTOR_BOOL_MODE): Document.
25079 * genmodes.c (VECTOR_BOOL_MODE): New macro.
25080 (make_vector_bool_mode): New function.
25081 (complete_mode, emit_mode_wider, emit_mode_adjustments): Handle
25083 * lto-streamer-in.c (lto_input_mode_table): Likewise.
25084 * rtx-vector-builder.c (rtx_vector_builder::find_cached_value):
25086 * stor-layout.c (int_mode_for_mode): Likewise.
25087 * tree.c (build_vector_type_for_mode): Likewise.
25088 * varasm.c (output_constant_pool_2): Likewise.
25089 * emit-rtl.c (init_emit_once): Make sure that CONST1_RTX (BImode) and
25090 CONSTM1_RTX (BImode) are the same thing. Initialize const_tiny_rtx
25091 for MODE_VECTOR_BOOL.
25092 * expr.c (expand_expr_real_1): Use VECTOR_MODE_P instead of a list
25093 of mode class checks.
25094 * tree-vect-generic.c (expand_vector_operation): Use VECTOR_MODE_P
25095 instead of a list of mode class checks.
25096 (expand_vector_scalar_condition): Likewise.
25097 (type_for_widest_vector_mode): Handle BImode as an inner mode.
25099 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25100 Alan Hayward <alan.hayward@arm.com>
25101 David Sherwood <david.sherwood@arm.com>
25103 * machmode.h (mode_size): Change from unsigned short to
25105 (mode_to_bytes): Return a poly_uint16 rather than an unsigned short.
25106 (GET_MODE_SIZE): Return a constant if ONLY_FIXED_SIZE_MODES,
25107 or if measurement_type is not polynomial.
25108 (fixed_size_mode::includes_p): Check for constant-sized modes.
25109 * genmodes.c (emit_mode_size_inline): Make mode_size_inline
25110 return a poly_uint16 rather than an unsigned short.
25111 (emit_mode_size): Change the type of mode_size from unsigned short
25112 to poly_uint16_pod. Use ZERO_COEFFS for the initializer.
25113 (emit_mode_adjustments): Cope with polynomial vector sizes.
25114 * lto-streamer-in.c (lto_input_mode_table): Use bp_unpack_poly_value
25116 * lto-streamer-out.c (lto_write_mode_table): Use bp_pack_poly_value
25118 * auto-inc-dec.c (try_merge): Treat GET_MODE_SIZE as polynomial.
25119 * builtins.c (expand_ifn_atomic_compare_exchange_into_call): Likewise.
25120 * caller-save.c (setup_save_areas): Likewise.
25121 (replace_reg_with_saved_mem): Likewise.
25122 * calls.c (emit_library_call_value_1): Likewise.
25123 * combine-stack-adj.c (combine_stack_adjustments_for_block): Likewise.
25124 * combine.c (simplify_set, make_extraction, simplify_shift_const_1)
25125 (gen_lowpart_for_combine): Likewise.
25126 * convert.c (convert_to_integer_1): Likewise.
25127 * cse.c (equiv_constant, cse_insn): Likewise.
25128 * cselib.c (autoinc_split, cselib_hash_rtx): Likewise.
25129 (cselib_subst_to_values): Likewise.
25130 * dce.c (word_dce_process_block): Likewise.
25131 * df-problems.c (df_word_lr_mark_ref): Likewise.
25132 * dwarf2cfi.c (init_one_dwarf_reg_size): Likewise.
25133 * dwarf2out.c (multiple_reg_loc_descriptor, mem_loc_descriptor)
25134 (concat_loc_descriptor, concatn_loc_descriptor, loc_descriptor)
25135 (rtl_for_decl_location): Likewise.
25136 * emit-rtl.c (gen_highpart, widen_memory_access): Likewise.
25137 * expmed.c (extract_bit_field_1, extract_integral_bit_field): Likewise.
25138 * expr.c (emit_group_load_1, clear_storage_hints): Likewise.
25139 (emit_move_complex, emit_move_multi_word, emit_push_insn): Likewise.
25140 (expand_expr_real_1): Likewise.
25141 * function.c (assign_parm_setup_block_p, assign_parm_setup_block)
25142 (pad_below): Likewise.
25143 * gimple-fold.c (optimize_atomic_compare_exchange_p): Likewise.
25144 * gimple-ssa-store-merging.c (rhs_valid_for_store_merging_p): Likewise.
25145 * ira.c (get_subreg_tracking_sizes): Likewise.
25146 * ira-build.c (ira_create_allocno_objects): Likewise.
25147 * ira-color.c (coalesced_pseudo_reg_slot_compare): Likewise.
25148 (ira_sort_regnos_for_alter_reg): Likewise.
25149 * ira-costs.c (record_operand_costs): Likewise.
25150 * lower-subreg.c (interesting_mode_p, simplify_gen_subreg_concatn)
25151 (resolve_simple_move): Likewise.
25152 * lra-constraints.c (get_reload_reg, operands_match_p): Likewise.
25153 (process_addr_reg, simplify_operand_subreg, curr_insn_transform)
25154 (lra_constraints): Likewise.
25155 (CONST_POOL_OK_P): Reject variable-sized modes.
25156 * lra-spills.c (slot, assign_mem_slot, pseudo_reg_slot_compare)
25157 (add_pseudo_to_slot, lra_spill): Likewise.
25158 * omp-low.c (omp_clause_aligned_alignment): Likewise.
25159 * optabs-query.c (get_best_extraction_insn): Likewise.
25160 * optabs-tree.c (expand_vec_cond_expr_p): Likewise.
25161 * optabs.c (expand_vec_perm_var, expand_vec_cond_expr): Likewise.
25162 (expand_mult_highpart, valid_multiword_target_p): Likewise.
25163 * recog.c (offsettable_address_addr_space_p): Likewise.
25164 * regcprop.c (maybe_mode_change): Likewise.
25165 * reginfo.c (choose_hard_reg_mode, record_subregs_of_mode): Likewise.
25166 * regrename.c (build_def_use): Likewise.
25167 * regstat.c (dump_reg_info): Likewise.
25168 * reload.c (complex_word_subreg_p, push_reload, find_dummy_reload)
25169 (find_reloads, find_reloads_subreg_address): Likewise.
25170 * reload1.c (eliminate_regs_1): Likewise.
25171 * rtlanal.c (for_each_inc_dec_find_inc_dec, rtx_cost): Likewise.
25172 * simplify-rtx.c (avoid_constant_pool_reference): Likewise.
25173 (simplify_binary_operation_1, simplify_subreg): Likewise.
25174 * targhooks.c (default_function_arg_padding): Likewise.
25175 (default_hard_regno_nregs, default_class_max_nregs): Likewise.
25176 * tree-cfg.c (verify_gimple_assign_binary): Likewise.
25177 (verify_gimple_assign_ternary): Likewise.
25178 * tree-inline.c (estimate_move_cost): Likewise.
25179 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
25180 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Likewise.
25181 (get_address_cost_ainc): Likewise.
25182 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
25183 (vect_supportable_dr_alignment): Likewise.
25184 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
25185 (vectorizable_reduction): Likewise.
25186 * tree-vect-stmts.c (vectorizable_assignment, vectorizable_shift)
25187 (vectorizable_operation, vectorizable_load): Likewise.
25188 * tree.c (build_same_sized_truth_vector_type): Likewise.
25189 * valtrack.c (cleanup_auto_inc_dec): Likewise.
25190 * var-tracking.c (emit_note_insn_var_location): Likewise.
25191 * config/arc/arc.h (ASM_OUTPUT_CASE_END): Use as_a <scalar_int_mode>.
25192 (ADDR_VEC_ALIGN): Likewise.
25194 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25195 Alan Hayward <alan.hayward@arm.com>
25196 David Sherwood <david.sherwood@arm.com>
25198 * machmode.h (mode_to_bits): Return a poly_uint16 rather than an
25200 (GET_MODE_BITSIZE): Return a constant if ONLY_FIXED_SIZE_MODES,
25201 or if measurement_type is polynomial.
25202 * calls.c (shift_return_value): Treat GET_MODE_BITSIZE as polynomial.
25203 * combine.c (make_extraction): Likewise.
25204 * dse.c (find_shift_sequence): Likewise.
25205 * dwarf2out.c (mem_loc_descriptor): Likewise.
25206 * expmed.c (store_integral_bit_field, extract_bit_field_1): Likewise.
25207 (extract_bit_field, extract_low_bits): Likewise.
25208 * expr.c (convert_move, convert_modes, emit_move_insn_1): Likewise.
25209 (optimize_bitfield_assignment_op, expand_assignment): Likewise.
25210 (store_expr_with_bounds, store_field, expand_expr_real_1): Likewise.
25211 * fold-const.c (optimize_bit_field_compare, merge_ranges): Likewise.
25212 * gimple-fold.c (optimize_atomic_compare_exchange_p): Likewise.
25213 * reload.c (find_reloads): Likewise.
25214 * reload1.c (alter_reg): Likewise.
25215 * stor-layout.c (bitwise_mode_for_mode, compute_record_mode): Likewise.
25216 * targhooks.c (default_secondary_memory_needed_mode): Likewise.
25217 * tree-if-conv.c (predicate_mem_writes): Likewise.
25218 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
25219 * tree-vect-patterns.c (adjust_bool_pattern): Likewise.
25220 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
25221 * valtrack.c (dead_debug_insert_temp): Likewise.
25222 * varasm.c (mergeable_constant_section): Likewise.
25223 * config/sh/sh.h (LOCAL_ALIGNMENT): Use as_a <fixed_size_mode>.
25225 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25226 Alan Hayward <alan.hayward@arm.com>
25227 David Sherwood <david.sherwood@arm.com>
25229 * expr.c (expand_assignment): Cope with polynomial mode sizes
25230 when assigning to a CONCAT.
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_precision): Change from unsigned short to
25238 (mode_to_precision): Return a poly_uint16 rather than an unsigned
25240 (GET_MODE_PRECISION): Return a constant if ONLY_FIXED_SIZE_MODES,
25241 or if measurement_type is not polynomial.
25242 (HWI_COMPUTABLE_MODE_P): Turn into a function. Optimize the case
25243 in which the mode is already known to be a scalar_int_mode.
25244 * genmodes.c (emit_mode_precision): Change the type of mode_precision
25245 from unsigned short to poly_uint16_pod. Use ZERO_COEFFS for the
25247 * lto-streamer-in.c (lto_input_mode_table): Use bp_unpack_poly_value
25248 for GET_MODE_PRECISION.
25249 * lto-streamer-out.c (lto_write_mode_table): Use bp_pack_poly_value
25250 for GET_MODE_PRECISION.
25251 * combine.c (update_rsp_from_reg_equal): Treat GET_MODE_PRECISION
25253 (try_combine, find_split_point, combine_simplify_rtx): Likewise.
25254 (expand_field_assignment, make_extraction): Likewise.
25255 (make_compound_operation_int, record_dead_and_set_regs_1): Likewise.
25256 (get_last_value): Likewise.
25257 * convert.c (convert_to_integer_1): Likewise.
25258 * cse.c (cse_insn): Likewise.
25259 * expr.c (expand_expr_real_1): Likewise.
25260 * lra-constraints.c (simplify_operand_subreg): Likewise.
25261 * optabs-query.c (can_atomic_load_p): Likewise.
25262 * optabs.c (expand_atomic_load): Likewise.
25263 (expand_atomic_store): Likewise.
25264 * ree.c (combine_reaching_defs): Likewise.
25265 * rtl.h (partial_subreg_p, paradoxical_subreg_p): Likewise.
25266 * rtlanal.c (nonzero_bits1, lsb_bitfield_op_p): Likewise.
25267 * tree.h (type_has_mode_precision_p): Likewise.
25268 * ubsan.c (instrument_si_overflow): Likewise.
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 * tree.h (TYPE_VECTOR_SUBPARTS): Turn into a function and handle
25275 polynomial numbers of units.
25276 (SET_TYPE_VECTOR_SUBPARTS): Likewise.
25277 (valid_vector_subparts_p): New function.
25278 (build_vector_type): Remove temporary shim and take the number
25279 of units as a poly_uint64 rather than an int.
25280 (build_opaque_vector_type): Take the number of units as a
25281 poly_uint64 rather than an int.
25282 * tree.c (build_vector_from_ctor): Handle polynomial
25283 TYPE_VECTOR_SUBPARTS.
25284 (type_hash_canon_hash, type_cache_hasher::equal): Likewise.
25285 (uniform_vector_p, vector_type_mode, build_vector): Likewise.
25286 (build_vector_from_val): If the number of units is variable,
25287 use build_vec_duplicate_cst for constant operands and
25288 VEC_DUPLICATE_EXPR otherwise.
25289 (make_vector_type): Remove temporary is_constant ().
25290 (build_vector_type, build_opaque_vector_type): Take the number of
25291 units as a poly_uint64 rather than an int.
25292 (check_vector_cst): Handle polynomial TYPE_VECTOR_SUBPARTS and
25294 * cfgexpand.c (expand_debug_expr): Likewise.
25295 * expr.c (count_type_elements, categorize_ctor_elements_1): Likewise.
25296 (store_constructor, expand_expr_real_1): Likewise.
25297 (const_scalar_mask_from_tree): Likewise.
25298 * fold-const-call.c (fold_const_reduction): Likewise.
25299 * fold-const.c (const_binop, const_unop, fold_convert_const): Likewise.
25300 (operand_equal_p, fold_vec_perm, fold_ternary_loc): Likewise.
25301 (native_encode_vector, vec_cst_ctor_to_array): Likewise.
25302 (fold_relational_const): Likewise.
25303 (native_interpret_vector): Likewise. Change the size from an
25304 int to an unsigned int.
25305 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Handle polynomial
25306 TYPE_VECTOR_SUBPARTS.
25307 (gimple_fold_indirect_ref, gimple_build_vector): Likewise.
25308 (gimple_build_vector_from_val): Use VEC_DUPLICATE_EXPR when
25309 duplicating a non-constant operand into a variable-length vector.
25310 * hsa-brig.c (hsa_op_immed::emit_to_buffer): Handle polynomial
25311 TYPE_VECTOR_SUBPARTS and VECTOR_CST_NELTS.
25312 * ipa-icf.c (sem_variable::equals): Likewise.
25313 * match.pd: Likewise.
25314 * omp-simd-clone.c (simd_clone_subparts): Likewise.
25315 * print-tree.c (print_node): Likewise.
25316 * stor-layout.c (layout_type): Likewise.
25317 * targhooks.c (default_builtin_vectorization_cost): Likewise.
25318 * tree-cfg.c (verify_gimple_comparison): Likewise.
25319 (verify_gimple_assign_binary): Likewise.
25320 (verify_gimple_assign_ternary): Likewise.
25321 (verify_gimple_assign_single): Likewise.
25322 * tree-pretty-print.c (dump_generic_node): Likewise.
25323 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
25324 (simplify_bitfield_ref, is_combined_permutation_identity): Likewise.
25325 * tree-vect-data-refs.c (vect_permute_store_chain): Likewise.
25326 (vect_grouped_load_supported, vect_permute_load_chain): Likewise.
25327 (vect_shift_permute_load_chain): Likewise.
25328 * tree-vect-generic.c (nunits_for_known_piecewise_op): Likewise.
25329 (expand_vector_condition, optimize_vector_constructor): Likewise.
25330 (lower_vec_perm, get_compute_type): Likewise.
25331 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
25332 (get_initial_defs_for_reduction, vect_transform_loop): Likewise.
25333 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
25334 (vect_recog_mask_conversion_pattern): Likewise.
25335 * tree-vect-slp.c (vect_supported_load_permutation_p): Likewise.
25336 (vect_get_constant_vectors, vect_transform_slp_perm_load): Likewise.
25337 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
25338 (get_group_load_store_type, vectorizable_mask_load_store): Likewise.
25339 (vectorizable_bswap, simd_clone_subparts, vectorizable_assignment)
25340 (vectorizable_shift, vectorizable_operation, vectorizable_store)
25341 (vectorizable_load, vect_is_simple_cond, vectorizable_comparison)
25342 (supportable_widening_operation): Likewise.
25343 (supportable_narrowing_operation): Likewise.
25344 * tree-vector-builder.c (tree_vector_builder::binary_encoded_nelts):
25346 * varasm.c (output_constant): Likewise.
25348 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25349 Alan Hayward <alan.hayward@arm.com>
25350 David Sherwood <david.sherwood@arm.com>
25352 * tree-vect-data-refs.c (vect_permute_store_chain): Reorganize
25353 so that both the length == 3 and length != 3 cases set up their
25354 own permute vectors. Add comments explaining why we know the
25355 number of elements is constant.
25356 (vect_permute_load_chain): Likewise.
25358 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25359 Alan Hayward <alan.hayward@arm.com>
25360 David Sherwood <david.sherwood@arm.com>
25362 * machmode.h (mode_nunits): Change from unsigned char to
25364 (ONLY_FIXED_SIZE_MODES): New macro.
25365 (pod_mode::measurement_type, scalar_int_mode::measurement_type)
25366 (scalar_float_mode::measurement_type, scalar_mode::measurement_type)
25367 (complex_mode::measurement_type, fixed_size_mode::measurement_type):
25369 (mode_to_nunits): Return a poly_uint16 rather than an unsigned short.
25370 (GET_MODE_NUNITS): Return a constant if ONLY_FIXED_SIZE_MODES,
25371 or if measurement_type is not polynomial.
25372 * genmodes.c (ZERO_COEFFS): New macro.
25373 (emit_mode_nunits_inline): Make mode_nunits_inline return a
25375 (emit_mode_nunits): Change the type of mode_nunits to poly_uint16_pod.
25376 Use ZERO_COEFFS when emitting initializers.
25377 * data-streamer.h (bp_pack_poly_value): New function.
25378 (bp_unpack_poly_value): Likewise.
25379 * lto-streamer-in.c (lto_input_mode_table): Use bp_unpack_poly_value
25380 for GET_MODE_NUNITS.
25381 * lto-streamer-out.c (lto_write_mode_table): Use bp_pack_poly_value
25382 for GET_MODE_NUNITS.
25383 * tree.c (make_vector_type): Remove temporary shim and make
25384 the real function take the number of units as a poly_uint64
25385 rather than an int.
25386 (build_vector_type_for_mode): Handle polynomial nunits.
25387 * dwarf2out.c (loc_descriptor, add_const_value_attribute): Likewise.
25388 * emit-rtl.c (const_vec_series_p_1): Likewise.
25389 (gen_rtx_CONST_VECTOR): Likewise.
25390 * fold-const.c (test_vec_duplicate_folding): Likewise.
25391 * genrecog.c (validate_pattern): Likewise.
25392 * optabs-query.c (can_vec_perm_var_p, can_mult_highpart_p): Likewise.
25393 * optabs-tree.c (expand_vec_cond_expr_p): Likewise.
25394 * optabs.c (expand_vector_broadcast, expand_binop_directly): Likewise.
25395 (shift_amt_for_vec_perm_mask, expand_vec_perm_var): Likewise.
25396 (expand_vec_cond_expr, expand_mult_highpart): Likewise.
25397 * rtlanal.c (subreg_get_info): Likewise.
25398 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
25399 (vect_grouped_load_supported): Likewise.
25400 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
25401 * tree-vect-loop.c (have_whole_vector_shift): Likewise.
25402 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
25403 (simplify_const_unary_operation, simplify_binary_operation_1)
25404 (simplify_const_binary_operation, simplify_ternary_operation)
25405 (test_vector_ops_duplicate, test_vector_ops): Likewise.
25406 (simplify_immed_subreg): Use GET_MODE_NUNITS on a fixed_size_mode
25407 instead of CONST_VECTOR_NUNITS.
25408 * varasm.c (output_constant_pool_2): Likewise.
25409 * rtx-vector-builder.c (rtx_vector_builder::build): Only include the
25410 explicit-encoded elements in the XVEC for variable-length vectors.
25412 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25414 * lra-constraints.c (curr_insn_transform): Use partial_subreg_p.
25416 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25417 Alan Hayward <alan.hayward@arm.com>
25418 David Sherwood <david.sherwood@arm.com>
25420 * coretypes.h (fixed_size_mode): Declare.
25421 (fixed_size_mode_pod): New typedef.
25422 * builtins.h (target_builtins::x_apply_args_mode)
25423 (target_builtins::x_apply_result_mode): Change type to
25424 fixed_size_mode_pod.
25425 * builtins.c (apply_args_size, apply_result_size, result_vector)
25426 (expand_builtin_apply_args_1, expand_builtin_apply)
25427 (expand_builtin_return): Update accordingly.
25429 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25431 * cse.c (hash_rtx_cb): Hash only the encoded elements.
25432 * cselib.c (cselib_hash_rtx): Likewise.
25433 * expmed.c (make_tree): Build VECTOR_CSTs directly from the
25434 CONST_VECTOR encoding.
25436 2018-01-03 Jakub Jelinek <jakub@redhat.com>
25437 Jeff Law <law@redhat.com>
25440 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): For
25441 noreturn probe, use gen_pop instead of ix86_emit_restore_reg_using_pop,
25442 only set RTX_FRAME_RELATED_P on both the push and pop if cfa_reg is sp
25443 and add REG_CFA_ADJUST_CFA notes in that case to both insns.
25446 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Do not
25447 explicitly probe *sp in a noreturn function if there were any callee
25448 register saves or frame pointer is needed.
25450 2018-01-03 Jakub Jelinek <jakub@redhat.com>
25453 * cfgexpand.c (expand_debug_expr): Return NULL if mode is
25454 BLKmode for ternary, binary or unary expressions.
25457 * var-tracking.c (delete_vta_debug_insn): New inline function.
25458 (delete_vta_debug_insns): Add USE_CFG argument, if true, walk just
25459 insns from get_insns () to NULL instead of each bb separately.
25460 Use delete_vta_debug_insn. No longer static.
25461 (vt_debug_insns_local, variable_tracking_main_1): Adjust
25462 delete_vta_debug_insns callers.
25463 * rtl.h (delete_vta_debug_insns): Declare.
25464 * final.c (rest_of_handle_final): Call delete_vta_debug_insns
25465 instead of variable_tracking_main.
25467 2018-01-03 Martin Sebor <msebor@redhat.com>
25469 PR tree-optimization/83603
25470 * calls.c (maybe_warn_nonstring_arg): Avoid accessing function
25471 arguments past the endof the argument list in functions declared
25472 without a prototype.
25473 * gimple-ssa-warn-restrict.c (wrestrict_dom_walker::check_call):
25474 Avoid checking when arguments are null.
25476 2018-01-03 Martin Sebor <msebor@redhat.com>
25479 * doc/extend.texi (attribute const): Fix a typo.
25480 * ipa-pure-const.c ((warn_function_const, warn_function_pure): Avoid
25481 issuing -Wsuggest-attribute for void functions.
25483 2018-01-03 Martin Sebor <msebor@redhat.com>
25485 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Use
25486 offset_int::from instead of wide_int::to_shwi.
25487 (maybe_diag_overlap): Remove assertion.
25488 Use HOST_WIDE_INT_PRINT_DEC instead of %lli.
25489 * gimple-ssa-sprintf.c (format_directive): Same.
25490 (parse_directive): Same.
25491 (sprintf_dom_walker::compute_format_length): Same.
25492 (try_substitute_return_value): Same.
25494 2018-01-03 Jeff Law <law@redhat.com>
25496 PR middle-end/83654
25497 * explow.c (anti_adjust_stack_and_probe_stack_clash): Test a
25498 non-constant residual for zero at runtime and avoid probing in
25499 that case. Reorganize code for trailing problem to mirror handling
25502 2018-01-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
25504 PR tree-optimization/83501
25505 * tree-ssa-strlen.c (get_string_cst): New.
25506 (handle_char_store): Call get_string_cst.
25508 2018-01-03 Martin Liska <mliska@suse.cz>
25510 PR tree-optimization/83593
25511 * tree-ssa-strlen.c: Include tree-cfg.h.
25512 (strlen_check_and_optimize_stmt): Add new argument cleanup_eh.
25513 (strlen_dom_walker): Add new member variable m_cleanup_cfg.
25514 (strlen_dom_walker::strlen_dom_walker): Initialize m_cleanup_cfg
25516 (strlen_dom_walker::before_dom_children): Call
25517 gimple_purge_dead_eh_edges. Dump tranformation with details
25519 (strlen_dom_walker::before_dom_children): Update call by adding
25520 new argument cleanup_eh.
25521 (pass_strlen::execute): Return TODO_cleanup_cfg if needed.
25523 2018-01-03 Martin Liska <mliska@suse.cz>
25526 * cif-code.def (VARIADIC_THUNK): New enum value.
25527 * ipa-fnsummary.c (compute_fn_summary): Do not inline variadic
25530 2018-01-03 Jan Beulich <jbeulich@suse.com>
25532 * sse.md (mov<mode>_internal): Tighten condition for when to use
25533 vmovdqu<ssescalarsize> for TI and OI modes.
25535 2018-01-03 Jakub Jelinek <jakub@redhat.com>
25537 Update copyright years.
25539 2018-01-03 Martin Liska <mliska@suse.cz>
25542 * ipa-visibility.c (function_and_variable_visibility): Skip
25543 functions with noipa attribure.
25545 2018-01-03 Jakub Jelinek <jakub@redhat.com>
25547 * gcc.c (process_command): Update copyright notice dates.
25548 * gcov-dump.c (print_version): Ditto.
25549 * gcov.c (print_version): Ditto.
25550 * gcov-tool.c (print_version): Ditto.
25551 * gengtype.c (create_file): Ditto.
25552 * doc/cpp.texi: Bump @copying's copyright year.
25553 * doc/cppinternals.texi: Ditto.
25554 * doc/gcc.texi: Ditto.
25555 * doc/gccint.texi: Ditto.
25556 * doc/gcov.texi: Ditto.
25557 * doc/install.texi: Ditto.
25558 * doc/invoke.texi: Ditto.
25560 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25562 * vector-builder.h (vector_builder::m_full_nelts): Change from
25563 unsigned int to poly_uint64.
25564 (vector_builder::full_nelts): Update prototype accordingly.
25565 (vector_builder::new_vector): Likewise.
25566 (vector_builder::encoded_full_vector_p): Handle polynomial full_nelts.
25567 (vector_builder::operator ==): Likewise.
25568 (vector_builder::finalize): Likewise.
25569 * int-vector-builder.h (int_vector_builder::int_vector_builder):
25570 Take the number of elements as a poly_uint64 rather than an
25572 * vec-perm-indices.h (vec_perm_indices::m_nelts_per_input): Change
25573 from unsigned int to poly_uint64.
25574 (vec_perm_indices::vec_perm_indices): Update prototype accordingly.
25575 (vec_perm_indices::new_vector): Likewise.
25576 (vec_perm_indices::length): Likewise.
25577 (vec_perm_indices::nelts_per_input): Likewise.
25578 (vec_perm_indices::input_nelts): Likewise.
25579 * vec-perm-indices.c (vec_perm_indices::new_vector): Take the
25580 number of elements per input as a poly_uint64 rather than an
25581 unsigned int. Use the original encoding for variable-length
25582 vectors, rather than clamping each individual element.
25583 For the second and subsequent elements in each pattern,
25584 clamp the step and base before clamping their sum.
25585 (vec_perm_indices::series_p): Handle polynomial element counts.
25586 (vec_perm_indices::all_in_range_p): Likewise.
25587 (vec_perm_indices_to_tree): Likewise.
25588 (vec_perm_indices_to_rtx): Likewise.
25589 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
25590 * tree-vector-builder.c (tree_vector_builder::new_unary_operation)
25591 (tree_vector_builder::new_binary_operation): Handle polynomial
25592 element counts. Return false if we need to know the number
25593 of elements at compile time.
25594 * fold-const.c (fold_vec_perm): Punt if the number of elements
25595 isn't known at compile time.
25597 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25599 * vec-perm-indices.h (vec_perm_builder): Change element type
25600 from HOST_WIDE_INT to poly_int64.
25601 (vec_perm_indices::element_type): Update accordingly.
25602 (vec_perm_indices::clamp): Handle polynomial element_types.
25603 * vec-perm-indices.c (vec_perm_indices::series_p): Likewise.
25604 (vec_perm_indices::all_in_range_p): Likewise.
25605 (tree_to_vec_perm_builder): Check for poly_int64 trees rather
25607 * vector-builder.h (vector_builder::stepped_sequence_p): Handle
25608 polynomial vec_perm_indices element types.
25609 * int-vector-builder.h (int_vector_builder::equal_p): Likewise.
25610 * fold-const.c (fold_vec_perm): Likewise.
25611 * optabs.c (shift_amt_for_vec_perm_mask): Likewise.
25612 * tree-vect-generic.c (lower_vec_perm): Likewise.
25613 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
25614 * config/aarch64/aarch64.c (aarch64_evpc_tbl): Cast d->perm
25615 element type to HOST_WIDE_INT.
25617 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25618 Alan Hayward <alan.hayward@arm.com>
25619 David Sherwood <david.sherwood@arm.com>
25621 * alias.c (addr_side_effect_eval): Take the size as a poly_int64
25622 rather than an int. Use plus_constant.
25623 (memrefs_conflict_p): Take the sizes as poly_int64s rather than ints.
25624 Take the offset "c" as a poly_int64 rather than a HOST_WIDE_INT.
25626 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25627 Alan Hayward <alan.hayward@arm.com>
25628 David Sherwood <david.sherwood@arm.com>
25630 * calls.c (emit_call_1, expand_call): Change struct_value_size from
25631 a HOST_WIDE_INT to a poly_int64.
25633 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25634 Alan Hayward <alan.hayward@arm.com>
25635 David Sherwood <david.sherwood@arm.com>
25637 * calls.c (load_register_parameters): Cope with polynomial
25638 mode sizes. Require a constant size for BLKmode parameters
25639 that aren't described by a PARALLEL. If BLOCK_REG_PADDING
25640 forces a parameter to be padded at the lsb end in order to
25641 fill a complete number of words, require the parameter size
25642 to be ordered wrt UNITS_PER_WORD.
25644 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25645 Alan Hayward <alan.hayward@arm.com>
25646 David Sherwood <david.sherwood@arm.com>
25648 * reload1.c (spill_stack_slot_width): Change element type
25649 from unsigned int to poly_uint64_pod.
25650 (alter_reg): Treat mode sizes as polynomial.
25652 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25653 Alan Hayward <alan.hayward@arm.com>
25654 David Sherwood <david.sherwood@arm.com>
25656 * reload.c (complex_word_subreg_p): New function.
25657 (reload_inner_reg_of_subreg, push_reload): Use it.
25659 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25660 Alan Hayward <alan.hayward@arm.com>
25661 David Sherwood <david.sherwood@arm.com>
25663 * lra-constraints.c (process_alt_operands): Reject matched
25664 operands whose sizes aren't ordered.
25665 (match_reload): Refer to this check here.
25667 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25668 Alan Hayward <alan.hayward@arm.com>
25669 David Sherwood <david.sherwood@arm.com>
25671 * builtins.c (expand_ifn_atomic_compare_exchange_into_call): Assert
25672 that the mode size is in the set {1, 2, 4, 8, 16}.
25674 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25675 Alan Hayward <alan.hayward@arm.com>
25676 David Sherwood <david.sherwood@arm.com>
25678 * var-tracking.c (adjust_mems): Treat mode sizes as polynomial.
25679 Use plus_constant instead of gen_rtx_PLUS.
25681 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25682 Alan Hayward <alan.hayward@arm.com>
25683 David Sherwood <david.sherwood@arm.com>
25685 * config/cr16/cr16-protos.h (cr16_push_rounding): Declare.
25686 * config/cr16/cr16.h (PUSH_ROUNDING): Move implementation to...
25687 * config/cr16/cr16.c (cr16_push_rounding): ...this new function.
25688 * config/h8300/h8300-protos.h (h8300_push_rounding): Declare.
25689 * config/h8300/h8300.h (PUSH_ROUNDING): Move implementation to...
25690 * config/h8300/h8300.c (h8300_push_rounding): ...this new function.
25691 * config/i386/i386-protos.h (ix86_push_rounding): Declare.
25692 * config/i386/i386.h (PUSH_ROUNDING): Move implementation to...
25693 * config/i386/i386.c (ix86_push_rounding): ...this new function.
25694 * config/m32c/m32c-protos.h (m32c_push_rounding): Take and return
25696 * config/m32c/m32c.c (m32c_push_rounding): Likewise.
25697 * config/m68k/m68k-protos.h (m68k_push_rounding): Declare.
25698 * config/m68k/m68k.h (PUSH_ROUNDING): Move implementation to...
25699 * config/m68k/m68k.c (m68k_push_rounding): ...this new function.
25700 * config/pdp11/pdp11-protos.h (pdp11_push_rounding): Declare.
25701 * config/pdp11/pdp11.h (PUSH_ROUNDING): Move implementation to...
25702 * config/pdp11/pdp11.c (pdp11_push_rounding): ...this new function.
25703 * config/stormy16/stormy16-protos.h (xstormy16_push_rounding): Declare.
25704 * config/stormy16/stormy16.h (PUSH_ROUNDING): Move implementation to...
25705 * config/stormy16/stormy16.c (xstormy16_push_rounding): ...this new
25707 * expr.c (emit_move_resolve_push): Treat the input and result
25708 of PUSH_ROUNDING as a poly_int64.
25709 (emit_move_complex_push, emit_single_push_insn_1): Likewise.
25710 (emit_push_insn): Likewise.
25711 * lra-eliminations.c (mark_not_eliminable): Likewise.
25712 * recog.c (push_operand): Likewise.
25713 * reload1.c (elimination_effects): Likewise.
25714 * rtlanal.c (nonzero_bits1): Likewise.
25715 * calls.c (store_one_arg): Likewise. Require the padding to be
25716 known at compile time.
25718 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25719 Alan Hayward <alan.hayward@arm.com>
25720 David Sherwood <david.sherwood@arm.com>
25722 * expr.c (emit_single_push_insn_1): Treat mode sizes as polynomial.
25723 Use plus_constant instead of gen_rtx_PLUS.
25725 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25726 Alan Hayward <alan.hayward@arm.com>
25727 David Sherwood <david.sherwood@arm.com>
25729 * auto-inc-dec.c (set_inc_state): Take the mode size as a poly_int64
25730 rather than an int.
25732 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25733 Alan Hayward <alan.hayward@arm.com>
25734 David Sherwood <david.sherwood@arm.com>
25736 * expr.c (expand_expr_real_1): Use tree_to_poly_uint64
25737 instead of int_size_in_bytes when handling VIEW_CONVERT_EXPRs
25738 via stack temporaries. Treat the mode size as polynomial too.
25740 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25741 Alan Hayward <alan.hayward@arm.com>
25742 David Sherwood <david.sherwood@arm.com>
25744 * expr.c (expand_expr_real_2): When handling conversions involving
25745 unions, apply tree_to_poly_uint64 to the TYPE_SIZE rather than
25746 multiplying int_size_in_bytes by BITS_PER_UNIT. Treat GET_MODE_BISIZE
25747 as a poly_uint64 too.
25749 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25750 Alan Hayward <alan.hayward@arm.com>
25751 David Sherwood <david.sherwood@arm.com>
25753 * rtlanal.c (subreg_get_info): Handle polynomial mode sizes.
25755 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25756 Alan Hayward <alan.hayward@arm.com>
25757 David Sherwood <david.sherwood@arm.com>
25759 * combine.c (can_change_dest_mode): Handle polynomial
25760 REGMODE_NATURAL_SIZE.
25761 * expmed.c (store_bit_field_1): Likewise.
25762 * expr.c (store_constructor): Likewise.
25763 * emit-rtl.c (validate_subreg): Operate on polynomial mode sizes
25764 and polynomial REGMODE_NATURAL_SIZE.
25765 (gen_lowpart_common): Likewise.
25766 * reginfo.c (record_subregs_of_mode): Likewise.
25767 * rtlanal.c (read_modify_subreg_p): Likewise.
25769 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25770 Alan Hayward <alan.hayward@arm.com>
25771 David Sherwood <david.sherwood@arm.com>
25773 * internal-fn.c (expand_vector_ubsan_overflow): Handle polynomial
25774 numbers of elements.
25776 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25777 Alan Hayward <alan.hayward@arm.com>
25778 David Sherwood <david.sherwood@arm.com>
25780 * match.pd: Cope with polynomial numbers of vector elements.
25782 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25783 Alan Hayward <alan.hayward@arm.com>
25784 David Sherwood <david.sherwood@arm.com>
25786 * fold-const.c (fold_indirect_ref_1): Handle polynomial offsets
25787 in a POINTER_PLUS_EXPR.
25789 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25790 Alan Hayward <alan.hayward@arm.com>
25791 David Sherwood <david.sherwood@arm.com>
25793 * omp-simd-clone.c (simd_clone_subparts): New function.
25794 (simd_clone_init_simd_arrays): Use it instead of TYPE_VECTOR_SUBPARTS.
25795 (ipa_simd_modify_function_body): Likewise.
25797 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25798 Alan Hayward <alan.hayward@arm.com>
25799 David Sherwood <david.sherwood@arm.com>
25801 * tree-vect-generic.c (nunits_for_known_piecewise_op): New function.
25802 (expand_vector_piecewise): Use it instead of TYPE_VECTOR_SUBPARTS.
25803 (expand_vector_addition, add_rshift, expand_vector_divmod): Likewise.
25804 (expand_vector_condition, vector_element): Likewise.
25805 (subparts_gt): New function.
25806 (get_compute_type): Use subparts_gt.
25807 (count_type_subparts): Delete.
25808 (expand_vector_operations_1): Use subparts_gt instead of
25809 count_type_subparts.
25811 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25812 Alan Hayward <alan.hayward@arm.com>
25813 David Sherwood <david.sherwood@arm.com>
25815 * tree-vect-data-refs.c (vect_no_alias_p): Replace with...
25816 (vect_compile_time_alias): ...this new function. Do the calculation
25817 on poly_ints rather than trees.
25818 (vect_prune_runtime_alias_test_list): Update call accordingly.
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 * tree-vect-slp.c (vect_build_slp_tree_1): Handle polynomial
25826 (vect_schedule_slp_instance): Likewise.
25828 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25829 Alan Hayward <alan.hayward@arm.com>
25830 David Sherwood <david.sherwood@arm.com>
25832 * tree-vect-slp.c (vect_get_and_check_slp_defs): Reject
25833 constant and extern definitions for variable-length vectors.
25834 (vect_get_constant_vectors): Note that the number of units
25835 is known to be constant.
25837 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25838 Alan Hayward <alan.hayward@arm.com>
25839 David Sherwood <david.sherwood@arm.com>
25841 * tree-vect-stmts.c (vectorizable_conversion): Treat the number
25842 of units as polynomial. Choose between WIDE and NARROW based
25845 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25846 Alan Hayward <alan.hayward@arm.com>
25847 David Sherwood <david.sherwood@arm.com>
25849 * tree-vect-stmts.c (simd_clone_subparts): New function.
25850 (vectorizable_simd_clone_call): Use it instead of TYPE_VECTOR_SUBPARTS.
25852 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25853 Alan Hayward <alan.hayward@arm.com>
25854 David Sherwood <david.sherwood@arm.com>
25856 * tree-vect-stmts.c (vectorizable_call): Treat the number of
25857 vectors as polynomial. Use build_index_vector for
25858 IFN_GOMP_SIMD_LANE.
25860 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25861 Alan Hayward <alan.hayward@arm.com>
25862 David Sherwood <david.sherwood@arm.com>
25864 * tree-vect-stmts.c (get_load_store_type): Treat the number of
25865 units as polynomial. Reject VMAT_ELEMENTWISE and VMAT_STRIDED_SLP
25866 for variable-length vectors.
25867 (vectorizable_mask_load_store): Treat the number of units as
25868 polynomial, asserting that it is constant if the condition has
25869 already been enforced.
25870 (vectorizable_store, vectorizable_load): Likewise.
25872 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25873 Alan Hayward <alan.hayward@arm.com>
25874 David Sherwood <david.sherwood@arm.com>
25876 * tree-vect-loop.c (vectorizable_live_operation): Treat the number
25877 of units as polynomial. Punt if we can't tell at compile time
25878 which vector contains the final result.
25880 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25881 Alan Hayward <alan.hayward@arm.com>
25882 David Sherwood <david.sherwood@arm.com>
25884 * tree-vect-loop.c (vectorizable_induction): Treat the number
25885 of units as polynomial. Punt on SLP inductions. Use an integer
25886 VEC_SERIES_EXPR for variable-length integer reductions. Use a
25887 cast of such a series for variable-length floating-point
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.h (build_index_vector): Declare.
25895 * tree.c (build_index_vector): New function.
25896 * tree-vect-loop.c (get_initial_defs_for_reduction): Treat the number
25897 of units as polynomial, forcibly converting it to a constant if
25898 vectorizable_reduction has already enforced the condition.
25899 (vect_create_epilog_for_reduction): Likewise. Use build_index_vector
25900 to create a {1,2,3,...} vector.
25901 (vectorizable_reduction): Treat the number of units as polynomial.
25902 Choose vectype_in based on the largest scalar element size rather
25903 than the smallest number of units. Enforce the restrictions
25906 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25907 Alan Hayward <alan.hayward@arm.com>
25908 David Sherwood <david.sherwood@arm.com>
25910 * tree-vect-data-refs.c (vector_alignment_reachable_p): Treat the
25911 number of units as polynomial.
25913 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25914 Alan Hayward <alan.hayward@arm.com>
25915 David Sherwood <david.sherwood@arm.com>
25917 * target.h (vector_sizes, auto_vector_sizes): New typedefs.
25918 * target.def (autovectorize_vector_sizes): Return the vector sizes
25919 by pointer, using vector_sizes rather than a bitmask.
25920 * targhooks.h (default_autovectorize_vector_sizes): Update accordingly.
25921 * targhooks.c (default_autovectorize_vector_sizes): Likewise.
25922 * config/aarch64/aarch64.c (aarch64_autovectorize_vector_sizes):
25924 * config/arc/arc.c (arc_autovectorize_vector_sizes): Likewise.
25925 * config/arm/arm.c (arm_autovectorize_vector_sizes): Likewise.
25926 * config/i386/i386.c (ix86_autovectorize_vector_sizes): Likewise.
25927 * config/mips/mips.c (mips_autovectorize_vector_sizes): Likewise.
25928 * omp-general.c (omp_max_vf): Likewise.
25929 * omp-low.c (omp_clause_aligned_alignment): Likewise.
25930 * optabs-query.c (can_vec_mask_load_store_p): Likewise.
25931 * tree-vect-loop.c (vect_analyze_loop): Likewise.
25932 * tree-vect-slp.c (vect_slp_bb): Likewise.
25933 * doc/tm.texi: Regenerate.
25934 * tree-vectorizer.h (current_vector_size): Change from an unsigned int
25936 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Take
25937 the vector size as a poly_uint64 rather than an unsigned int.
25938 (current_vector_size): Change from an unsigned int to a poly_uint64.
25939 (get_vectype_for_scalar_type): Update accordingly.
25940 * tree.h (build_truth_vector_type): Take the size and number of
25941 units as a poly_uint64 rather than an unsigned int.
25942 (build_vector_type): Add a temporary overload that takes
25943 the number of units as a poly_uint64 rather than an unsigned int.
25944 * tree.c (make_vector_type): Likewise.
25945 (build_truth_vector_type): Take the number of units as a poly_uint64
25946 rather than an unsigned int.
25948 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25949 Alan Hayward <alan.hayward@arm.com>
25950 David Sherwood <david.sherwood@arm.com>
25952 * target.def (get_mask_mode): Take the number of units and length
25953 as poly_uint64s rather than unsigned ints.
25954 * targhooks.h (default_get_mask_mode): Update accordingly.
25955 * targhooks.c (default_get_mask_mode): Likewise.
25956 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
25957 * doc/tm.texi: Regenerate.
25959 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25960 Alan Hayward <alan.hayward@arm.com>
25961 David Sherwood <david.sherwood@arm.com>
25963 * omp-general.h (omp_max_vf): Return a poly_uint64 instead of an int.
25964 * omp-general.c (omp_max_vf): Likewise.
25965 * omp-expand.c (omp_adjust_chunk_size): Update call to omp_max_vf.
25966 (expand_omp_simd): Handle polynomial safelen.
25967 * omp-low.c (omplow_simd_context): Add a default constructor.
25968 (omplow_simd_context::max_vf): Change from int to poly_uint64.
25969 (lower_rec_simd_input_clauses): Update accordingly.
25970 (lower_rec_input_clauses): Likewise.
25972 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25973 Alan Hayward <alan.hayward@arm.com>
25974 David Sherwood <david.sherwood@arm.com>
25976 * tree-vectorizer.h (vect_nunits_for_cost): New function.
25977 * tree-vect-loop.c (vect_model_reduction_cost): Use it.
25978 * tree-vect-slp.c (vect_analyze_slp_cost_1): Likewise.
25979 (vect_analyze_slp_cost): Likewise.
25980 * tree-vect-stmts.c (vect_model_store_cost): Likewise.
25981 (vect_model_load_cost): Likewise.
25983 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25984 Alan Hayward <alan.hayward@arm.com>
25985 David Sherwood <david.sherwood@arm.com>
25987 * tree-vect-slp.c (vect_record_max_nunits, vect_build_slp_tree_1)
25988 (vect_build_slp_tree_2, vect_build_slp_tree): Change max_nunits
25989 from an unsigned int * to a poly_uint64_pod *.
25990 (calculate_unrolling_factor): New function.
25991 (vect_analyze_slp_instance): Use it. Track polynomial max_nunits.
25993 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25994 Alan Hayward <alan.hayward@arm.com>
25995 David Sherwood <david.sherwood@arm.com>
25997 * tree-vectorizer.h (_slp_instance::unrolling_factor): Change
25998 from an unsigned int to a poly_uint64.
25999 (_loop_vec_info::slp_unrolling_factor): Likewise.
26000 (_loop_vec_info::vectorization_factor): Change from an int
26002 (MAX_VECTORIZATION_FACTOR): Bump from 64 to INT_MAX.
26003 (vect_get_num_vectors): New function.
26004 (vect_update_max_nunits, vect_vf_for_cost): Likewise.
26005 (vect_get_num_copies): Use vect_get_num_vectors.
26006 (vect_analyze_data_ref_dependences): Change max_vf from an int *
26007 to an unsigned int *.
26008 (vect_analyze_data_refs): Change min_vf from an int * to a
26010 (vect_transform_slp_perm_load): Take the vf as a poly_uint64 rather
26011 than an unsigned HOST_WIDE_INT.
26012 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr)
26013 (vect_analyze_data_ref_dependence): Change max_vf from an int *
26014 to an unsigned int *.
26015 (vect_analyze_data_ref_dependences): Likewise.
26016 (vect_compute_data_ref_alignment): Handle polynomial vf.
26017 (vect_enhance_data_refs_alignment): Likewise.
26018 (vect_prune_runtime_alias_test_list): Likewise.
26019 (vect_shift_permute_load_chain): Likewise.
26020 (vect_supportable_dr_alignment): Likewise.
26021 (dependence_distance_ge_vf): Take the vectorization factor as a
26022 poly_uint64 rather than an unsigned HOST_WIDE_INT.
26023 (vect_analyze_data_refs): Change min_vf from an int * to a
26025 * tree-vect-loop-manip.c (vect_gen_scalar_loop_niters): Take
26026 vfm1 as a poly_uint64 rather than an int. Make the same change
26027 for the returned bound_scalar.
26028 (vect_gen_vector_loop_niters): Handle polynomial vf.
26029 (vect_do_peeling): Likewise. Update call to
26030 vect_gen_scalar_loop_niters and handle polynomial bound_scalars.
26031 (vect_gen_vector_loop_niters_mult_vf): Assert that the vf must
26033 * tree-vect-loop.c (vect_determine_vectorization_factor)
26034 (vect_update_vf_for_slp, vect_analyze_loop_2): Handle polynomial vf.
26035 (vect_get_known_peeling_cost): Likewise.
26036 (vect_estimate_min_profitable_iters, vectorizable_reduction): Likewise.
26037 (vect_worthwhile_without_simd_p, vectorizable_induction): Likewise.
26038 (vect_transform_loop): Likewise. Use the lowest possible VF when
26039 updating the upper bounds of the loop.
26040 (vect_min_worthwhile_factor): Make static. Return an unsigned int
26041 rather than an int.
26042 * tree-vect-slp.c (vect_attempt_slp_rearrange_stmts): Cope with
26043 polynomial unroll factors.
26044 (vect_analyze_slp_cost_1, vect_analyze_slp_instance): Likewise.
26045 (vect_make_slp_decision): Likewise.
26046 (vect_supported_load_permutation_p): Likewise, and polynomial
26048 (vect_analyze_slp_cost): Handle polynomial vf.
26049 (vect_slp_analyze_node_operations): Likewise.
26050 (vect_slp_analyze_bb_1): Likewise.
26051 (vect_transform_slp_perm_load): Take the vf as a poly_uint64 rather
26052 than an unsigned HOST_WIDE_INT.
26053 * tree-vect-stmts.c (vectorizable_simd_clone_call, vectorizable_store)
26054 (vectorizable_load): Handle polynomial vf.
26055 * tree-vectorizer.c (simduid_to_vf::vf): Change from an int to
26057 (adjust_simduid_builtins, shrink_simd_arrays): Update accordingly.
26059 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
26060 Alan Hayward <alan.hayward@arm.com>
26061 David Sherwood <david.sherwood@arm.com>
26063 * match.pd: Handle bit operations involving three constants
26064 and try to fold one pair.
26066 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
26068 * tree-vect-loop-manip.c: Include gimple-fold.h.
26069 (slpeel_make_loop_iterate_ntimes): Add step, final_iv and
26070 niters_maybe_zero parameters. Handle other cases besides a step of 1.
26071 (vect_gen_vector_loop_niters): Add a step_vector_ptr parameter.
26072 Add a path that uses a step of VF instead of 1, but disable it
26074 (vect_do_peeling): Add step_vector, niters_vector_mult_vf_var
26075 and niters_no_overflow parameters. Update calls to
26076 slpeel_make_loop_iterate_ntimes and vect_gen_vector_loop_niters.
26077 Create a new SSA name if the latter choses to use a ste other
26078 than zero, and return it via niters_vector_mult_vf_var.
26079 * tree-vect-loop.c (vect_transform_loop): Update calls to
26080 vect_do_peeling, vect_gen_vector_loop_niters and
26081 slpeel_make_loop_iterate_ntimes.
26082 * tree-vectorizer.h (slpeel_make_loop_iterate_ntimes, vect_do_peeling)
26083 (vect_gen_vector_loop_niters): Update declarations after above changes.
26085 2018-01-02 Michael Meissner <meissner@linux.vnet.ibm.com>
26087 * config/rs6000/rs6000.md (floor<mode>2): Add support for IEEE
26088 128-bit round to integer instructions.
26089 (ceil<mode>2): Likewise.
26090 (btrunc<mode>2): Likewise.
26091 (round<mode>2): Likewise.
26093 2018-01-02 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
26095 * config/rs6000/rs6000-string.c (expand_block_move): Allow the use of
26096 unaligned VSX load/store on P8/P9.
26097 (expand_block_clear): Allow the use of unaligned VSX
26098 load/store on P8/P9.
26100 2018-01-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26102 * config/rs6000/rs6000-p8swap.c (swap_feeds_both_load_and_store):
26104 (rs6000_analyze_swaps): Mark a web unoptimizable if it contains a
26105 swap associated with both a load and a store.
26107 2018-01-02 Andrew Waterman <andrew@sifive.com>
26109 * config/riscv/linux.h (ICACHE_FLUSH_FUNC): New.
26110 * config/riscv/riscv.md (clear_cache): Use it.
26112 2018-01-02 Artyom Skrobov <tyomitch@gmail.com>
26114 * web.c: Remove out-of-date comment.
26116 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
26118 * expr.c (fixup_args_size_notes): Check that any existing
26119 REG_ARGS_SIZE notes are correct, and don't try to re-add them.
26120 (emit_single_push_insn_1): Move stack_pointer_delta adjustment to...
26121 (emit_single_push_insn): ...here.
26123 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
26125 * rtl.h (CONST_VECTOR_ELT): Redefine to const_vector_elt.
26126 (const_vector_encoded_nelts): New function.
26127 (CONST_VECTOR_NUNITS): Redefine to use GET_MODE_NUNITS.
26128 (const_vector_int_elt, const_vector_elt): Declare.
26129 * emit-rtl.c (const_vector_int_elt_1): New function.
26130 (const_vector_elt): Likewise.
26131 * simplify-rtx.c (simplify_immed_subreg): Avoid taking the address
26132 of CONST_VECTOR_ELT.
26134 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
26136 * expr.c: Include rtx-vector-builder.h.
26137 (const_vector_mask_from_tree): Use rtx_vector_builder and operate
26138 directly on the tree encoding.
26139 (const_vector_from_tree): Likewise.
26140 * optabs.c: Include rtx-vector-builder.h.
26141 (expand_vec_perm_var): Use rtx_vector_builder and create a repeating
26142 sequence of "u" values.
26143 * vec-perm-indices.c: Include rtx-vector-builder.h.
26144 (vec_perm_indices_to_rtx): Use rtx_vector_builder and operate
26145 directly on the vec_perm_indices encoding.
26147 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
26149 * doc/rtl.texi (const_vector): Describe new encoding scheme.
26150 * Makefile.in (OBJS): Add rtx-vector-builder.o.
26151 * rtx-vector-builder.h: New file.
26152 * rtx-vector-builder.c: Likewise.
26153 * rtl.h (rtx_def::u2): Add a const_vector field.
26154 (CONST_VECTOR_NPATTERNS): New macro.
26155 (CONST_VECTOR_NELTS_PER_PATTERN): Likewise.
26156 (CONST_VECTOR_DUPLICATE_P): Likewise.
26157 (CONST_VECTOR_STEPPED_P): Likewise.
26158 (CONST_VECTOR_ENCODED_ELT): Likewise.
26159 (const_vec_duplicate_p): Check for a duplicated vector encoding.
26160 (unwrap_const_vec_duplicate): Likewise.
26161 (const_vec_series_p): Check for a non-duplicated vector encoding.
26162 Say that the function only returns true for integer vectors.
26163 * emit-rtl.c: Include rtx-vector-builder.h.
26164 (gen_const_vec_duplicate_1): Delete.
26165 (gen_const_vector): Call gen_const_vec_duplicate instead of
26166 gen_const_vec_duplicate_1.
26167 (const_vec_series_p_1): Operate directly on the CONST_VECTOR encoding.
26168 (gen_const_vec_duplicate): Use rtx_vector_builder.
26169 (gen_const_vec_series): Likewise.
26170 (gen_rtx_CONST_VECTOR): Likewise.
26171 * config/powerpcspe/powerpcspe.c: Include rtx-vector-builder.h.
26172 (swap_const_vector_halves): Take an rtx pointer rather than rtx.
26173 Build a new vector rather than modifying a CONST_VECTOR in-place.
26174 (handle_special_swappables): Update call accordingly.
26175 * config/rs6000/rs6000-p8swap.c: Include rtx-vector-builder.h.
26176 (swap_const_vector_halves): Take an rtx pointer rather than rtx.
26177 Build a new vector rather than modifying a CONST_VECTOR in-place.
26178 (handle_special_swappables): Update call accordingly.
26180 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
26182 * simplify-rtx.c (simplify_const_binary_operation): Use
26183 CONST_VECTOR_ELT instead of XVECEXP.
26185 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
26187 * tree-cfg.c (verify_gimple_assign_ternary): Allow the size of
26188 the selector elements to be different from the data elements
26189 if the selector is a VECTOR_CST.
26190 * tree-vect-stmts.c (vect_gen_perm_mask_any): Use a vector of
26191 ssizetype for the selector.
26193 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
26195 * optabs.c (shift_amt_for_vec_perm_mask): Try using series_p
26196 before testing each element individually.
26197 * tree-vect-generic.c (lower_vec_perm): Likewise.
26199 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
26201 * selftest.h (selftest::vec_perm_indices_c_tests): Declare.
26202 * selftest-run-tests.c (selftest::run_tests): Call it.
26203 * vector-builder.h (vector_builder::operator ==): New function.
26204 (vector_builder::operator !=): Likewise.
26205 * vec-perm-indices.h (vec_perm_indices::series_p): Declare.
26206 (vec_perm_indices::all_from_input_p): New function.
26207 * vec-perm-indices.c (vec_perm_indices::series_p): Likewise.
26208 (test_vec_perm_12, selftest::vec_perm_indices_c_tests): Likewise.
26209 * fold-const.c (fold_ternary_loc): Use tree_to_vec_perm_builder
26210 instead of reading the VECTOR_CST directly. Detect whether both
26211 vector inputs are the same before constructing the vec_perm_indices,
26212 and update the number of inputs argument accordingly. Use the
26213 utility functions added above. Only construct sel2 if we need to.
26215 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
26217 * optabs.c (expand_vec_perm_var): Use an explicit encoding for
26218 the broadcast of the low byte.
26219 (expand_mult_highpart): Use an explicit encoding for the permutes.
26220 * optabs-query.c (can_mult_highpart_p): Likewise.
26221 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
26222 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
26223 (vectorizable_bswap): Likewise.
26224 * tree-vect-data-refs.c (vect_grouped_store_supported): Use an
26225 explicit encoding for the power-of-2 permutes.
26226 (vect_permute_store_chain): Likewise.
26227 (vect_grouped_load_supported): Likewise.
26228 (vect_permute_load_chain): Likewise.
26230 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
26232 * vec-perm-indices.h (vec_perm_indices_to_tree): Declare.
26233 * vec-perm-indices.c (vec_perm_indices_to_tree): New function.
26234 * tree-ssa-forwprop.c (simplify_vector_constructor): Use it.
26235 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
26236 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
26237 (vect_gen_perm_mask_any): Likewise.
26239 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
26241 * int-vector-builder.h: New file.
26242 * vec-perm-indices.h: Include int-vector-builder.h.
26243 (vec_perm_indices): Redefine as an int_vector_builder.
26244 (auto_vec_perm_indices): Delete.
26245 (vec_perm_builder): Redefine as a stand-alone class.
26246 (vec_perm_indices::vec_perm_indices): New function.
26247 (vec_perm_indices::clamp): Likewise.
26248 * vec-perm-indices.c: Include fold-const.h and tree-vector-builder.h.
26249 (vec_perm_indices::new_vector): New function.
26250 (vec_perm_indices::new_expanded_vector): Update for new
26251 vec_perm_indices class.
26252 (vec_perm_indices::rotate_inputs): New function.
26253 (vec_perm_indices::all_in_range_p): Operate directly on the
26254 encoded form, without computing elided elements.
26255 (tree_to_vec_perm_builder): Operate directly on the VECTOR_CST
26256 encoding. Update for new vec_perm_indices class.
26257 * optabs.c (expand_vec_perm_const): Create a vec_perm_indices for
26258 the given vec_perm_builder.
26259 (expand_vec_perm_var): Update vec_perm_builder constructor.
26260 (expand_mult_highpart): Use vec_perm_builder instead of
26261 auto_vec_perm_indices.
26262 * optabs-query.c (can_mult_highpart_p): Use vec_perm_builder and
26263 vec_perm_indices instead of auto_vec_perm_indices. Use a single
26264 or double series encoding as appropriate.
26265 * fold-const.c (fold_ternary_loc): Use vec_perm_builder and
26266 vec_perm_indices instead of auto_vec_perm_indices.
26267 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
26268 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
26269 (vect_permute_store_chain): Likewise.
26270 (vect_grouped_load_supported): Likewise.
26271 (vect_permute_load_chain): Likewise.
26272 (vect_shift_permute_load_chain): Likewise.
26273 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
26274 (vect_transform_slp_perm_load): Likewise.
26275 (vect_schedule_slp_instance): Likewise.
26276 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
26277 (vectorizable_mask_load_store): Likewise.
26278 (vectorizable_bswap): Likewise.
26279 (vectorizable_store): Likewise.
26280 (vectorizable_load): Likewise.
26281 * tree-vect-generic.c (lower_vec_perm): Use vec_perm_builder and
26282 vec_perm_indices instead of auto_vec_perm_indices. Use
26283 tree_to_vec_perm_builder to read the vector from a tree.
26284 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Take a
26285 vec_perm_builder instead of a vec_perm_indices.
26286 (have_whole_vector_shift): Use vec_perm_builder and
26287 vec_perm_indices instead of auto_vec_perm_indices. Leave the
26288 truncation to calc_vec_perm_mask_for_shift.
26289 (vect_create_epilog_for_reduction): Likewise.
26290 * config/aarch64/aarch64.c (expand_vec_perm_d::perm): Change
26291 from auto_vec_perm_indices to vec_perm_indices.
26292 (aarch64_expand_vec_perm_const_1): Use rotate_inputs on d.perm
26293 instead of changing individual elements.
26294 (aarch64_vectorize_vec_perm_const): Use new_vector to install
26295 the vector in d.perm.
26296 * config/arm/arm.c (expand_vec_perm_d::perm): Change
26297 from auto_vec_perm_indices to vec_perm_indices.
26298 (arm_expand_vec_perm_const_1): Use rotate_inputs on d.perm
26299 instead of changing individual elements.
26300 (arm_vectorize_vec_perm_const): Use new_vector to install
26301 the vector in d.perm.
26302 * config/powerpcspe/powerpcspe.c (rs6000_expand_extract_even):
26303 Update vec_perm_builder constructor.
26304 (rs6000_expand_interleave): Likewise.
26305 * config/rs6000/rs6000.c (rs6000_expand_extract_even): Likewise.
26306 (rs6000_expand_interleave): Likewise.
26308 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
26310 * optabs-query.c (can_vec_perm_var_p): Check whether lowering
26311 to qimode could truncate the indices.
26312 * optabs.c (expand_vec_perm_var): Likewise.
26314 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
26316 * Makefile.in (OBJS): Add vec-perm-indices.o.
26317 * vec-perm-indices.h: New file.
26318 * vec-perm-indices.c: Likewise.
26319 * target.h (vec_perm_indices): Replace with a forward class
26321 (auto_vec_perm_indices): Move to vec-perm-indices.h.
26322 * optabs.h: Include vec-perm-indices.h.
26323 (expand_vec_perm): Delete.
26324 (selector_fits_mode_p, expand_vec_perm_var): Declare.
26325 (expand_vec_perm_const): Declare.
26326 * target.def (vec_perm_const_ok): Replace with...
26327 (vec_perm_const): ...this new hook.
26328 * doc/tm.texi.in (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Replace with...
26329 (TARGET_VECTORIZE_VEC_PERM_CONST): ...this new hook.
26330 * doc/tm.texi: Regenerate.
26331 * optabs.def (vec_perm_const): Delete.
26332 * doc/md.texi (vec_perm_const): Likewise.
26333 (vec_perm): Refer to TARGET_VECTORIZE_VEC_PERM_CONST.
26334 * expr.c (expand_expr_real_2): Use expand_vec_perm_const rather than
26335 expand_vec_perm for constant permutation vectors. Assert that
26336 the mode of variable permutation vectors is the integer equivalent
26337 of the mode that is being permuted.
26338 * optabs-query.h (selector_fits_mode_p): Declare.
26339 * optabs-query.c: Include vec-perm-indices.h.
26340 (selector_fits_mode_p): New function.
26341 (can_vec_perm_const_p): Check whether targetm.vectorize.vec_perm_const
26342 is defined, instead of checking whether the vec_perm_const_optab
26343 exists. Use targetm.vectorize.vec_perm_const instead of
26344 targetm.vectorize.vec_perm_const_ok. Check whether the indices
26345 fit in the vector mode before using a variable permute.
26346 * optabs.c (shift_amt_for_vec_perm_mask): Take a mode and a
26347 vec_perm_indices instead of an rtx.
26348 (expand_vec_perm): Replace with...
26349 (expand_vec_perm_const): ...this new function. Take the selector
26350 as a vec_perm_indices rather than an rtx. Also take the mode of
26351 the selector. Update call to shift_amt_for_vec_perm_mask.
26352 Use targetm.vectorize.vec_perm_const instead of vec_perm_const_optab.
26353 Use vec_perm_indices::new_expanded_vector to expand the original
26354 selector into bytes. Check whether the indices fit in the vector
26355 mode before using a variable permute.
26356 (expand_vec_perm_var): Make global.
26357 (expand_mult_highpart): Use expand_vec_perm_const.
26358 * fold-const.c: Includes vec-perm-indices.h.
26359 * tree-ssa-forwprop.c: Likewise.
26360 * tree-vect-data-refs.c: Likewise.
26361 * tree-vect-generic.c: Likewise.
26362 * tree-vect-loop.c: Likewise.
26363 * tree-vect-slp.c: Likewise.
26364 * tree-vect-stmts.c: Likewise.
26365 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm_const):
26367 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>): Delete.
26368 * config/aarch64/aarch64.c (aarch64_expand_vec_perm_const)
26369 (aarch64_vectorize_vec_perm_const_ok): Fuse into...
26370 (aarch64_vectorize_vec_perm_const): ...this new function.
26371 (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
26372 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
26373 * config/arm/arm-protos.h (arm_expand_vec_perm_const): Delete.
26374 * config/arm/vec-common.md (vec_perm_const<mode>): Delete.
26375 * config/arm/arm.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
26376 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
26377 (arm_expand_vec_perm_const, arm_vectorize_vec_perm_const_ok): Merge
26379 (arm_vectorize_vec_perm_const): ...this new function. Explicitly
26380 check for NEON modes.
26381 * config/i386/i386-protos.h (ix86_expand_vec_perm_const): Delete.
26382 * config/i386/sse.md (VEC_PERM_CONST, vec_perm_const<mode>): Delete.
26383 * config/i386/i386.c (ix86_expand_vec_perm_const_1): Update comment.
26384 (ix86_expand_vec_perm_const, ix86_vectorize_vec_perm_const_ok): Merge
26386 (ix86_vectorize_vec_perm_const): ...this new function. Incorporate
26387 the old VEC_PERM_CONST conditions.
26388 * config/ia64/ia64-protos.h (ia64_expand_vec_perm_const): Delete.
26389 * config/ia64/vect.md (vec_perm_const<mode>): Delete.
26390 * config/ia64/ia64.c (ia64_expand_vec_perm_const)
26391 (ia64_vectorize_vec_perm_const_ok): Merge into...
26392 (ia64_vectorize_vec_perm_const): ...this new function.
26393 * config/mips/loongson.md (vec_perm_const<mode>): Delete.
26394 * config/mips/mips-msa.md (vec_perm_const<mode>): Delete.
26395 * config/mips/mips-ps-3d.md (vec_perm_constv2sf): Delete.
26396 * config/mips/mips-protos.h (mips_expand_vec_perm_const): Delete.
26397 * config/mips/mips.c (mips_expand_vec_perm_const)
26398 (mips_vectorize_vec_perm_const_ok): Merge into...
26399 (mips_vectorize_vec_perm_const): ...this new function.
26400 * config/powerpcspe/altivec.md (vec_perm_constv16qi): Delete.
26401 * config/powerpcspe/paired.md (vec_perm_constv2sf): Delete.
26402 * config/powerpcspe/spe.md (vec_perm_constv2si): Delete.
26403 * config/powerpcspe/vsx.md (vec_perm_const<mode>): Delete.
26404 * config/powerpcspe/powerpcspe-protos.h (altivec_expand_vec_perm_const)
26405 (rs6000_expand_vec_perm_const): Delete.
26406 * config/powerpcspe/powerpcspe.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK):
26408 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
26409 (altivec_expand_vec_perm_const_le): Take each operand individually.
26410 Operate on constant selectors rather than rtxes.
26411 (altivec_expand_vec_perm_const): Likewise. Update call to
26412 altivec_expand_vec_perm_const_le.
26413 (rs6000_expand_vec_perm_const): Delete.
26414 (rs6000_vectorize_vec_perm_const_ok): Delete.
26415 (rs6000_vectorize_vec_perm_const): New function.
26416 (rs6000_do_expand_vec_perm): Take a vec_perm_builder instead of
26417 an element count and rtx array.
26418 (rs6000_expand_extract_even): Update call accordingly.
26419 (rs6000_expand_interleave): Likewise.
26420 * config/rs6000/altivec.md (vec_perm_constv16qi): Delete.
26421 * config/rs6000/paired.md (vec_perm_constv2sf): Delete.
26422 * config/rs6000/vsx.md (vec_perm_const<mode>): Delete.
26423 * config/rs6000/rs6000-protos.h (altivec_expand_vec_perm_const)
26424 (rs6000_expand_vec_perm_const): Delete.
26425 * config/rs6000/rs6000.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
26426 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
26427 (altivec_expand_vec_perm_const_le): Take each operand individually.
26428 Operate on constant selectors rather than rtxes.
26429 (altivec_expand_vec_perm_const): Likewise. Update call to
26430 altivec_expand_vec_perm_const_le.
26431 (rs6000_expand_vec_perm_const): Delete.
26432 (rs6000_vectorize_vec_perm_const_ok): Delete.
26433 (rs6000_vectorize_vec_perm_const): New function. Remove stray
26434 reference to the SPE evmerge intructions.
26435 (rs6000_do_expand_vec_perm): Take a vec_perm_builder instead of
26436 an element count and rtx array.
26437 (rs6000_expand_extract_even): Update call accordingly.
26438 (rs6000_expand_interleave): Likewise.
26439 * config/sparc/sparc.md (vec_perm_constv8qi): Delete in favor of...
26440 * config/sparc/sparc.c (sparc_vectorize_vec_perm_const): ...this
26442 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
26444 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
26446 * optabs.c (expand_vec_perm_1): Assert that SEL has an integer
26447 vector mode and that that mode matches the mode of the data
26449 (expand_vec_perm): Split handling of non-CONST_VECTOR selectors
26450 out into expand_vec_perm_var. Do all CONST_VECTOR handling here,
26451 directly using expand_vec_perm_1 when forcing selectors into
26453 (expand_vec_perm_var): New function, split out from expand_vec_perm.
26455 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
26457 * optabs-query.h (can_vec_perm_p): Delete.
26458 (can_vec_perm_var_p, can_vec_perm_const_p): Declare.
26459 * optabs-query.c (can_vec_perm_p): Split into...
26460 (can_vec_perm_var_p, can_vec_perm_const_p): ...these two functions.
26461 (can_mult_highpart_p): Use can_vec_perm_const_p to test whether a
26462 particular selector is valid.
26463 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
26464 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
26465 (vect_grouped_load_supported): Likewise.
26466 (vect_shift_permute_load_chain): Likewise.
26467 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
26468 (vect_transform_slp_perm_load): Likewise.
26469 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
26470 (vectorizable_bswap): Likewise.
26471 (vect_gen_perm_mask_checked): Likewise.
26472 * fold-const.c (fold_ternary_loc): Likewise. Don't take
26473 implementations of variable permutation vectors into account
26474 when deciding which selector to use.
26475 * tree-vect-loop.c (have_whole_vector_shift): Don't check whether
26476 vec_perm_const_optab is supported; instead use can_vec_perm_const_p
26477 with a false third argument.
26478 * tree-vect-generic.c (lower_vec_perm): Use can_vec_perm_const_p
26479 to test whether the constant selector is valid and can_vec_perm_var_p
26480 to test whether a variable selector is valid.
26482 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
26484 * optabs-query.h (can_vec_perm_p): Take a const vec_perm_indices *.
26485 * optabs-query.c (can_vec_perm_p): Likewise.
26486 * fold-const.c (fold_vec_perm): Take a const vec_perm_indices &
26487 instead of vec_perm_indices.
26488 * tree-vectorizer.h (vect_gen_perm_mask_any): Likewise,
26489 (vect_gen_perm_mask_checked): Likewise,
26490 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise,
26491 (vect_gen_perm_mask_checked): Likewise,
26493 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
26495 * optabs-query.h (qimode_for_vec_perm): Declare.
26496 * optabs-query.c (can_vec_perm_p): Split out qimode search to...
26497 (qimode_for_vec_perm): ...this new function.
26498 * optabs.c (expand_vec_perm): Use qimode_for_vec_perm.
26500 2018-01-02 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
26502 * rtlanal.c (canonicalize_condition): Return 0 if final rtx
26503 does not have a conditional at the top.
26505 2018-01-02 Richard Biener <rguenther@suse.de>
26507 * ipa-inline.c (big_speedup_p): Fix expression.
26509 2018-01-02 Jan Hubicka <hubicka@ucw.cz>
26512 * config/i386/x86-tune-costs.h: Increase cost of integer load costs
26515 2018-01-02 Jan Hubicka <hubicka@ucw.cz>
26519 * x86-tune-costs.h (generic_cost): Reduce cost of FDIV 20->17,
26520 cost of sqrt 20->14, DIVSS 18->13, DIVSD 32->17, SQRtSS 30->14
26521 and SQRTsD 58->18, cond_not_taken_branch_cost. 2->1. Increase
26522 cond_taken_branch_cost 3->4.
26524 2018-01-01 Jakub Jelinek <jakub@redhat.com>
26526 PR tree-optimization/83581
26527 * tree-loop-distribution.c (pass_loop_distribution::execute): Return
26528 TODO_cleanup_cfg if any changes have been made.
26530 PR middle-end/83608
26531 * expr.c (store_expr_with_bounds): Use simplify_gen_subreg instead of
26532 convert_modes if target mode has the right side, but different mode
26535 PR middle-end/83609
26536 * expr.c (expand_assignment): Fix up a typo in simplify_gen_subreg
26537 last argument when extracting from CONCAT. If either from_real or
26538 from_imag is NULL, use expansion through memory. If result is not
26539 a CONCAT and simplify_gen_subreg fails, try to simplify_gen_subreg
26540 the parts directly to inner mode, if even that fails, use expansion
26543 PR middle-end/83623
26544 * expmed.c (expand_shift_1): For 2-byte rotates by BITS_PER_UNIT,
26545 check for bswap in mode rather than HImode and use that in expand_unop
26548 Copyright (C) 2018 Free Software Foundation, Inc.
26550 Copying and distribution of this file, with or without modification,
26551 are permitted in any medium without royalty provided the copyright
26552 notice and this notice are preserved.