1 2018-07-31 Olivier Hainque <hainque@adacore.com>
3 * gcc.c (handle_spec_function): Accept a soft_matched_part
4 argument, as do_spec_1. Pass it down to ...
5 (eval_spec_function): Accept a soft_matched_part argument,
6 and pass it down to ...
7 (do_spec_2): Accept a soft_matched_part argument, and pass
9 (do_spec_1): Pass soft_matched_part to handle_spec_function.
10 (handle_braces): Update call to handle_spec_function.
11 (driver::set_up_specs): Update calls to do_spec_2.
12 (compare_debug_dump_opt_spec_function): Likewise.
13 (compare_debug_self_opt_spec_function): Likewise.
15 2018-07-31 Olivier Hainque <hainque@adacore.com>
17 * common.opt (nolibc): New option.
18 * doc/invoke.texi (Link Options): Document it.
19 * gcc.c (LINK_GCC_C_SEQUENCE_SPEC): Honor nolibc.
20 * config/alpha/linux.h: Likewise.
21 * config/arc/elf.h: Likewise.
22 * config/arm/uclinux-elf.h: Likewise.
23 * config/arm/unknown-elf.h: Likewise.
24 * config/avr/avrlibc.h: Likewise.
25 * config/bfin/bfin.h: Likewise.
26 * config/bfin/linux.h: Likewise.
27 * config/bfin/uclinux.h: Likewise.
28 * config/darwin.h: Likewise.
29 * config/darwin10.h: Likewise.
30 * config/darwin12.h: Likewise.
31 * config/gnu-user.h: Likewise.
32 * config/lm32/uclinux-elf.h: Likewise.
33 * config/pa/pa-hpux11.h: Likewise.
34 * config/pa/pa64-hpux.h: Likewise.
35 * config/sparc/sparc.h: Likewise.
37 2018-07-31 Olivier Hainque <hainque@adacore.com>
39 * gcc.c (getenv_spec_function): Prepend '/' to value for allowed
42 2018-07-30 Segher Boessenkool <segher@kernel.crashing.org>
45 * config/arm/arm.c (arm_block_set_aligned_vect): Use gen_int_mode
48 2018-07-30 Bernd Edlinger <bernd.edlinger@hotmail.de>
50 * tree-ssa-forwprop.c (simplify_builtin_call): Don't create a not NUL
51 terminated string literal.
53 2018-07-30 Segher Boessenkool <segher@kernel.crashing.org>
55 PR rtl-optimization/85160
56 * combine.c (is_just_move): New function.
57 (try_combine): Allow combining two instructions into two if neither of
58 the original instructions was a move.
60 2018-07-30 Alexander Monakov <amonakov@ispras.ru>
63 * doc/extend.texi (Global Register Variables): Discourage use of type
65 (Local Register Variables): Likewise.
67 2018-07-30 Richard Sandiford <richard.sandiford@arm.com>
69 PR tree-optimization/86506
70 * hwint.h (ceil_log2): Resync with hwint.c implementation.
72 2018-07-30 Ilya Leoshkevich <iii@linux.ibm.com>
75 * lra-constraints.c (spill_hard_reg_in_range): When selecting the
76 hard_regno, make sure no insn between `from` and `to` clobbers it.
78 2018-07-30 Cesar Philippidis <cesar@codesourcery.com>
79 Tom de Vries <tdevries@suse.de>
81 * config/nvptx/nvptx.c (PTX_GANG_DEFAULT): Rename to ...
82 (PTX_DEFAULT_RUNTIME_DIM): ... this.
83 (nvptx_goacc_validate_dims): Set default worker and gang dims to
84 PTX_DEFAULT_RUNTIME_DIM.
85 (nvptx_dim_limit): Ignore GOMP_DIM_WORKER.
87 2018-07-29 John David Anglin <danglin@gcc.gnu.org>
89 * config/pa/pa.c (pa_output_addr_vec): Align address table.
90 * config/pa/pa.h (JUMP_TABLES_IN_TEXT_SECTION): Revise comment.
91 * config/pa/pa32-linux.h (JUMP_TABLES_IN_TEXT_SECTION): Define.
93 2018-07-27 Michael Meissner <meissner@linux.ibm.com>
95 * config/rs6000/constraints.md (wG constraint): Delete, no longer
97 * config/rs6000/predicates.md (p9_fusion_reg_operand): Rename
98 predicate to reflect toc fusion has been deleted.
99 (toc_fusion_mem_raw): Delete, no longer used.
100 (toc_fusion_mem_wrapped): Likewise.
101 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete toc
103 * config/rs6000/rs6000-protos.h (fusion_wrap_memory_address):
104 Delete, no longer used.
105 * config/rs6000/rs6000.c (struct rs6000_reg_addr): Delete fields
106 meant to be used for toc fusion.
107 (rs6000_debug_print_mode): Delete toc fusion debugging.
108 (rs6000_debug_reg_global): Likewise.
109 (rs6000_init_hard_regno_mode_ok): Delete setting up fields for toc
110 fusion and secondary reload support that were never used.
111 (rs6000_option_override_internal): Delete TOC fusion, that was only
112 partially defined, and it did not work unless you also used the
114 (rs6000_legitimate_address_p): Delete TOC fusion support.
115 (rs6000_opt_masks): Likewise.
116 (fusion_wrap_memory_address): Delete function, no longer used.
117 (fusion_split_address); Delete TOC fusion support.
118 * config/rs6000/rs6000.h (TARGET_TOC_FUSION_INT): Delete, no
119 longer used with toc fusion being deleted.
120 (TARGET_TOC_FUSION_FP): Likewise.
121 * config/rs6000/rs6000.md (UNSPEC_FUSION_ADDIS): Delete TOC fusion
123 (toc fusion spliter): Delete TOC fusion support.
124 (toc_fusionload_<mode>): Likewise.
125 (toc_fusionload_di): Likewise.
126 (fusion_gpr_load_<mode>): Delete generator function, this insn no
127 longer needs to be named. Rename predicate to delete TOC fusion.
128 (fusion_gpr_<P:mode>_<GPR_FUSION:mode>_load): Likewise.
129 (fusion_gpr_<P:mode>_<GPR_FUSION:mode>_store): Likewise.
130 (fusion_vsx_<P:mode>_<GPR_FUSION:mode>_load): Likewise.
131 (fusion_vsx_<P:mode>_<GPR_FUSION:mode>_store): Likewise.
132 (p9 fusion peephole2s): Rename predicate to delete TOC fusion.
134 2018-07-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
136 * doc/extend.texi (Basic PowerPC Built-in Functions Available on
137 ISA 2.05): Replace __uint128_t with __uint128 and __int128_t with
138 __int128 in built-in function prototypes.
139 (PowerPC AltiVec Built-in Functions on ISA 2.07): Likewise.
140 (PowerPC AltiVec Built-in Functions on ISA 3.0): Likewise.
142 2018-07-27 Martin Sebor <msebor@redhat.com>
144 PR tree-optimization/86696
145 * tree-ssa-strlen.c (get_min_string_length): Handle all integer
146 types, including enums.
147 (handle_char_store): Be prepared for the above function to fail.
149 2018-07-26 Qing Zhao <qing.zhao@oracle.com>
151 * builtins.c (inline_expand_builtin_string_cmp): Disable inlining
152 when optimization level is lower than 2 or optimize for size.
154 2018-07-26 Martin Sebor <msebor@redhat.com>
156 PR tree-optimization/86043
157 PR tree-optimization/86042
158 * tree-ssa-strlen.c (handle_builtin_memcpy): Handle strict overlaps.
159 (get_string_cst_length): Rename...
160 (get_min_string_length): ...to this. Add argument.
161 (handle_char_store): Extend to handle multi-character stores by
163 * tree.c (initializer_zerop): Use new argument. Handle MEM_REF.
164 * tree.h (initializer_zerop): Add argument.
166 2018-07-26 Jakub Jelinek <jakub@redhat.com>
169 * omp-low.c (scan_sharing_clauses): Don't ignore map clauses for
170 declare target to variables if they have always,{to,from,tofrom} map
173 2018-07-26 Martin Liska <mliska@suse.cz>
176 * lto-wrapper.c: Add linker_output as prefix
177 for ltrans_output_file.
179 2018-07-26 Segher Boessenkool <segher@kernel.crashing.org>
181 PR rtl-optimization/85805
182 * combine.c (reg_nonzero_bits_for_combine): Only use the last set
183 value for hard registers if that was written in the same mode.
185 2018-07-26 Martin Liska <mliska@suse.cz>
187 PR gcov-profile/86536
188 * gcov.c (format_gcov): Use printf format %.*f directly
189 and do not handle special values.
191 2018-07-25 Claudiu Zissulescu <claziss@synopsys.com>
193 * common/config/arc/arc-common.c (arc_option_optimization_table):
194 Update default optimizations for size.
196 2018-07-25 Claudiu Zissulescu <claziss@synopsys.com>
198 * config/arc/arc.md (movsf_insn): Add short instruction selection.
199 * config/arc/constraints.md (CfZ): New constraint.
200 * config/arc/fpu.md (addssf3_fpu): Use CfZ constraint.
201 (subsf3_fpu): Likewise.
202 (cmpsf_fpu): Likewise.
203 (cmpsf_fpu_uneq): Likewise.
205 2018-07-25 Claudiu Zissulescu <claziss@synopsys.com>
207 * config/arc/arc.c (compact_memory_operand_p): Check for uncached
209 (arc_is_uncached_mem_p): uncached applies to both the variable and
212 2018-07-25 Claudiu Zissulescu <claziss@synopsys.com>
214 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Add additional
217 2018-07-25 David Malcolm <dmalcolm@redhat.com>
219 * optinfo-emit-json.cc (class optrecord_json_writer): Convert
220 field "m_scopes" from vec to auto_vec.
222 2018-07-25 Martin Liska <mliska@suse.cz>
224 * config/powerpcspe/powerpcspe-protos.h (rs6000_loop_align): Fix
227 2018-07-25 Richard Biener <rguenther@suse.de>
230 * dwarf2out.c (dwarf2out_decl): Do not handle nested functions
231 special wrt context_die late.
232 (gen_subprogram_die): Re-use DIEs in local scope.
234 2018-07-25 Richard Sandiford <richard.sandiford@arm.com>
236 PR tree-optimization/86644
237 * hwint.c (ceil_log2): Fix comment. Return 0 for 0.
239 2018-07-25 Martin Liska <mliska@suse.cz>
242 * dumpfile.c: And excluded values with TDF_ALL_VALUES.
243 * dumpfile.h (enum dump_flag): Defince TDF_ALL_VALUES.
245 2018-07-25 Martin Liska <mliska@suse.cz>
248 * params.def: Explain ASan abbreviation and provide
249 a documentation link.
251 2018-07-24 Martin Sebor <msebor@redhat.com>
253 PR tree-optimization/86622
254 PR tree-optimization/86532
255 * builtins.h (string_length): Declare.
256 * builtins.c (c_strlen): Correct handling of non-constant offsets.
257 (check_access): Be prepared for non-constant length ranges.
258 (string_length): Make extern.
259 * expr.c (string_constant): Only handle the minor non-constant
260 array index. Use string_constant to compute the length of
261 a generic string constant.
263 2018-07-24 Richard Sandiford <richard.sandiford@arm.com>
265 PR tree-optimization/86618
266 * tree-vect-stmts.c (vectorizable_call): Don't take the address
267 of LOOP_VINFO_MASKS (loop_vinfo) when loop_vinfo is null.
269 2018-07-24 David Malcolm <dmalcolm@redhat.com>
271 PR tree-optimization/86636
272 * json.cc (json::object::set): Fix comment. Add assertions.
273 (json::array::append): Move here from json.h. Add comment and an
275 (json::string::string): Likewise.
276 * json.h (json::array::append): Move to json.cc.
277 (json::string::string): Likewise.
278 * optinfo-emit-json.cc
279 (optrecord_json_writer::impl_location_to_json): Assert that we
280 aren't attempting to write out UNKNOWN_LOCATION, or an ad-hoc
281 wrapper around it. Expand the location once, rather than three
283 (optrecord_json_writer::inlining_chain_to_json): Fix the check for
284 UNKNOWN_LOCATION, to use LOCATION_LOCUS to look through ad-hoc
286 (optrecord_json_writer::optinfo_to_json): Likewise, in four
287 places. Fix some overlong lines.
289 2018-07-24 Matthew Malcomson <matthew.malcomson@arm.com>
291 * config/aarch64/aarch64-simd.md
292 (aarch64_<ANY_EXTEND:su><ADDSUB:optab>w<mode>): Split into...
293 (aarch64_<ANY_EXTEND:su>subw<mode>): ... This...
294 (aarch64_<ANY_EXTEND:su>addw<mode>): ... And this.
295 (aarch64_<ANY_EXTEND:su><ADDSUB:optab>w<mode>_internal): Split into...
296 (aarch64_<ANY_EXTEND:su>subw<mode>_internal): ... This...
297 (aarch64_<ANY_EXTEND:su>addw<mode>_internal): ... And this.
298 (aarch64_<ANY_EXTEND:su><ADDSUB:optab>w2<mode>_internal): Split into...
299 (aarch64_<ANY_EXTEND:su>subw2<mode>_internal): ... This...
300 (aarch64_<ANY_EXTEND:su>addw2<mode>_internal): ... And this.
302 2018-07-24 Jakub Jelinek <jakub@redhat.com>
305 * expmed.c (expand_divmod): Punt if d == HOST_WIDE_INT_MIN
306 and size > HOST_BITS_PER_WIDE_INT. For size > HOST_BITS_PER_WIDE_INT
307 and abs_d == d, do the power of two handling if profitable.
309 2018-07-24 Richard Biener <rguenther@suse.de>
311 * match.pd: Add BIT_FIELD_REF canonicalizations.
313 2018-07-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
316 * genmatch.c (dt_operand::gen_match_op): Avoid folding volatile values.
318 2018-07-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
320 * gimple-fold.c (gimple_fold_builtin_printf): Don't create a not NUL
321 terminated STRING_CST object.
323 2018-07-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
325 hsa-dump.c (dump_hsa_symbol): Avoid out of scope access to buf.
327 2018-07-23 Segher Boessenkool <segher@kernel.crashing.org>
329 * config/rs6000/rs6000-p8swap.c (rtx_is_swappable_p): Adjust.
330 * config/rs6000/rs6000-protos.h (rs6000_split_v4si_init): Delete.
331 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Always force
332 the elements into a register.
333 (rs6000_split_v4si_init_di_reg): Delete.
334 (rs6000_split_v4si_init): Delete.
335 * config/rs6000/vsx.md (unspec): Delete UNSPEC_VSX_VEC_INIT.
336 (vsx_init_v4si): Rewrite as a define_expand.
338 2018-07-23 Segher Boessenkool <segher@kernel.crashing.org>
340 * config/rs6000/rs6000.md (splitters for rldimi and rlwimi with the
341 zero_extend argument from memory): New.
343 2018-07-22 Martin Sebor <msebor@redhat.com>
346 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Avoid
347 diagnosing calls with unknown arguments unless -Walloca-larger-than
348 is restricted to less than PTRDIFF_MAX bytes.
350 2018-07-22 Gerald Pfeifer <gerald@pfeifer.com>
352 * doc/gcov.texi (Invoking Gcov): Editorial changes.
354 2018-07-20 David Malcolm <dmalcolm@redhat.com>
356 * pretty-print.c (text_info::set_location): Remove redundant
357 "line_table" parameter from call to rich_location::set_range.
359 2018-07-20 Martin Sebor <msebor@redhat.com>
362 * builtins.c (expand_builtin_alloca): Adjust.
363 * calls.c (alloc_max_size): Simplify.
364 * cgraphunit.c (cgraph_node::expand): Adjust.
365 * common.opt (larger_than_size, warn_frame_larger_than): Remove
367 (frame_larger_than_size): Same.
368 (-Wframe-larger-than, -Wlarger-than, -Wstack-usage): Change options
369 to take a HOST_WIDE_INT argument and accept a byte-size suffix.
371 * doc/invoke.texi (GCC Command Options): Document option arguments.
372 Explain byte-size arguments and suffixes.
373 (-Wvla-larger-than, -Wno-alloc-size-larger-than): Update.
374 (-Wno-alloca-larger-than, -Wno-vla-larger-than): Same.
375 (-Wframe-larger-than, -Wlarger-than, -Wstack-usage): Same.
376 * doc/options.texi (UInteger): Expand.
377 (Host_Wide_Int, ByteSize): Document new properties.
378 * final.c (final_start_function_1): Include sizes in an error message.
379 * function.c (frame_offset_overflow): Same.
380 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Adjust.
381 (alloca_call_type_by_arg): Change function argument to HOST_WIDE_INT.
382 Diagnose unbounded alloca calls only for limits of less than
384 (alloca_call_type): Adjust. Diagnose possibly out-of-bounds alloca
385 calls and VLA size only for limits of less than PTRDIFF_MAX. Same
387 (pass_walloca::execute): Adjust. Diagnose alloca calls in loops
388 only for limits of less than PTRDIFF_MAX.
389 * langhooks-def.h (lhd_handle_option): Change function argument
391 * langhooks.c (lhd_handle_option): Same.
392 * langhooks.h (handle_option): Same.
393 * opt-functions.awk (switch_bit_fields): Handle Host_Wide_Int and
395 (var_type, var_type_struct): Same.
396 (var_set): Handle ByteSize flag.
397 * optc-gen.awk: Add comments to output to ease debugging. Make
398 use of HOST_WIDE_INT where appropriate.
399 * opts-gen-save.awk: Use %lx to format unsigned long.
400 * opth-gen.awk: Change function argument to HOST_WIDE_INT.
401 * opts-common.c (integral_argument): Return HOST_WIDE_INT and add
402 arguments. Parse bytes-size suffixes.
403 (enum_arg_to_value): Change function argument to HOST_WIDE_INT.
404 (enum_value_to_arg): Same.
405 (decode_cmdline_option): Handle cl_host_wide_int. Adjust.
406 (handle_option): Adjust.
407 (generate_option): Change function argument to HOST_WIDE_INT.
408 (cmdline_handle_error): Adjust.
409 (read_cmdline_option): Change function argument to HOST_WIDE_INT.
410 (set_option): Change function argument to HOST_WIDE_INT.
411 (option_enabled): Handle cl_host_wide_int.
412 (get_option_state): Handle CLVC_SIZE.
413 (control_warning_option): Same.
414 * opts.c (common_handle_option): Change function argument to
415 HOST_WIDE_INT. Remove handling of OPT_Walloca_larger_than_ and
416 OPT_Wvla_larger_than_.
417 * opts.h (enum cl_var_type): Add an enumerator.
418 * stor-layout.c (layout_decl): Print a more meaningful warning.
419 * toplev.c (output_stack_usage): Adjust.
421 2018-07-20 Qing Zhao <qing.zhao@oracle.com>
423 * builtins.c (expand_builtin_memcmp): Delete the last parameter for
424 call to inline_expand_builtin_string_cmp.
425 (expand_builtin_strcmp): Likewise.
426 (expand_builtin_strncmp): Likewise.
427 (inline_string_cmp): Delete the last parameter, change char_type_node
428 to unsigned_char_type_node for strcmp/strncmp, add conversions to the
430 (inline_expand_builtin_string_cmp): Delete the last parameter, give up
431 the inlining expansion on target where the type of the call has same or
432 narrower precision than unsigned char.
434 2018-07-20 David Malcolm <dmalcolm@redhat.com>
436 * Makefile.in (OBJS): Add json.o and optinfo-emit-json.o.
437 (CFLAGS-optinfo-emit-json.o): Define TARGET_NAME.
438 * common.opt (fsave-optimization-record): New option.
439 * coretypes.h (struct kv_pair): Move here from dumpfile.c.
440 * doc/invoke.texi (-fsave-optimization-record): New option.
441 * dumpfile.c: Include "optinfo-emit-json.h".
442 (struct kv_pair): Move to coretypes.h.
443 (optgroup_options): Make non-static.
444 (dump_context::end_scope): Call
445 optimization_records_maybe_pop_dump_scope.
446 * dumpfile.h (optgroup_options): New decl.
449 * optinfo-emit-json.cc: New file.
450 * optinfo-emit-json.h: New file.
451 * optinfo.cc: Include "optinfo-emit-json.h".
452 (optinfo::emit): Call optimization_records_maybe_record_optinfo.
453 (optinfo_enabled_p): Check optimization_records_enabled_p.
454 (optinfo_wants_inlining_info_p): Likewise.
455 * optinfo.h: Update comment.
456 * profile-count.c (profile_quality_as_string): New function.
457 * profile-count.h (profile_quality_as_string): New decl.
458 (profile_count::quality): New accessor.
459 * selftest-run-tests.c (selftest::run_tests): Call json_cc_tests
460 and optinfo_emit_json_cc_tests.
461 * selftest.h (selftest::json_cc_tests): New decl.
462 (selftest::optinfo_emit_json_cc_tests): New decl.
463 * toplev.c: Include "optinfo-emit-json.h".
464 (compile_file): Call optimization_records_finish.
465 (do_compile): Call optimization_records_start.
466 * tree-ssa-live.c: Include optinfo.h.
467 (remove_unused_scope_block_p): Retain inlining information if
468 optinfo_wants_inlining_info_p returns true.
470 2018-07-20 Richard Biener <rguenther@suse.de>
473 * dwarf2out.c (dwarf2out_die_ref_for_decl): Test in_lto_p
474 to cover -flto-partition=none.
476 2018-07-20 Martin Liska <mliska@suse.cz>
478 * tree.h (DECL_LOCATION_RANGE): Remove unused macro.
479 (get_decl_source_range): Remove unused function.
481 2018-07-20 Richard Biener <rguenther@suse.de>
483 * tree-ssa-sccvn.h (struct vn_nary_op_s): Add next member.
484 (struct vn_phi_s): Likewise.
485 (struct vn_reference_s): Likewise.
486 * tree-ssa-sccvn.c (vn_nary_op_hasher::equal): Add shortcut
487 for searching the slot of an entry known to be in the hash itself.
488 (vn_phi_hasher::equal): Likewise.
489 (vn_reference_hasher::equal): Likewise.
490 (last_inserted_ref, last_inserted_phi, last_inserted_nary): New
492 (optimistic_info, current_info): Remove, keeping only valid_info.
493 (vn_reference_lookup_1): Remove fallback lookup.
494 (vn_reference_lookup_2): Likewise.
495 (vn_nary_op_lookup_1): Likewise.
496 (vn_phi_lookup): Likewise.
497 (vn_nary_build_or_lookup_1): Make sure to not chain the built
499 (vn_reference_insert): Adjust, chain the inserted hash element
500 at last_inserted_ref.
501 (vn_reference_insert_pieces): Likewise.
502 (visit_reference_op_call): Likewise.
503 (vn_nary_op_insert_into): Chain the inserted hash element at
505 (vn_nary_op_insert_pieces): Adjust.
506 (vn_nary_op_insert): Likewise.
507 (vn_nary_op_insert_stmt): Likewise.
508 (vn_phi_insert): Adjust, chain the inserted hash element at
510 (process_scc): Remove clearing and copying the optimistic
511 table. Instead remove elements inserted during an optimistic
512 iteration from the single table we maintain.
513 (init_scc_vn): Adjust.
514 (free_scc_vn): Likewise.
515 (sccvn_dom_walker::record_cond): Likewise.
516 (sccvn_dom_walker::after_dom_children): Likewise.
518 2018-07-19 Martin Sebor <msebor@redhat.com>
520 PR tree-optimization/84047
521 PR tree-optimization/83776
522 * tree-vrp.c (vrp_prop::check_mem_ref): New function.
523 (check_array_bounds): Call it.
525 2018-07-19 Martin Sebor <msebor@redhat.com>
527 * align.h (align_flags): Use member initialization.
529 2018-07-19 David Malcolm <dmalcolm@redhat.com>
531 * Makefile.in (OBJS): Add optinfo.o.
532 * coretypes.h (class symtab_node): New forward decl.
533 (struct cgraph_node): New forward decl.
534 (class varpool_node): New forward decl.
535 * dump-context.h: New file.
536 * dumpfile.c: Include "optinfo.h", "dump-context.h", "cgraph.h",
538 (refresh_dumps_are_enabled): Use optinfo_enabled_p.
539 (set_dump_file): Call dumpfile_ensure_any_optinfo_are_flushed.
540 (set_alt_dump_file): Likewise.
541 (dump_context::~dump_context): New dtor.
542 (dump_gimple_stmt): Move implementation to...
543 (dump_context::dump_gimple_stmt): ...this new member function.
544 Add the stmt to any pending optinfo, creating one if need be.
545 (dump_gimple_stmt_loc): Move implementation to...
546 (dump_context::dump_gimple_stmt_loc): ...this new member function.
547 Start a new optinfo and add the stmt to it.
548 (dump_gimple_expr): Move implementation to...
549 (dump_context::dump_gimple_expr): ...this new member function.
550 Add the stmt to any pending optinfo, creating one if need be.
551 (dump_gimple_expr_loc): Move implementation to...
552 (dump_context::dump_gimple_expr_loc): ...this new member function.
553 Start a new optinfo and add the stmt to it.
554 (dump_generic_expr): Move implementation to...
555 (dump_context::dump_generic_expr): ...this new member function.
556 Add the tree to any pending optinfo, creating one if need be.
557 (dump_generic_expr_loc): Move implementation to...
558 (dump_context::dump_generic_expr_loc): ...this new member
559 function. Add the tree to any pending optinfo, creating one if
561 (dump_printf): Move implementation to...
562 (dump_context::dump_printf_va): ...this new member function. Add
563 the text to any pending optinfo, creating one if need be.
564 (dump_printf_loc): Move implementation to...
565 (dump_context::dump_printf_loc_va): ...this new member function.
566 Start a new optinfo and add the stmt to it.
567 (dump_dec): Move implementation to...
568 (dump_context::dump_dec): ...this new member function. Add the
569 value to any pending optinfo, creating one if need be.
570 (dump_context::dump_symtab_node): New member function.
571 (dump_context::get_scope_depth): New member function.
572 (dump_context::begin_scope): New member function.
573 (dump_context::end_scope): New member function.
574 (dump_context::ensure_pending_optinfo): New member function.
575 (dump_context::begin_next_optinfo): New member function.
576 (dump_context::end_any_optinfo): New member function.
577 (dump_context::s_current): New global.
578 (dump_context::s_default): New global.
579 (dump_scope_depth): Delete global.
580 (dumpfile_ensure_any_optinfo_are_flushed): New function.
581 (dump_symtab_node): New function.
582 (get_dump_scope_depth): Reimplement in terms of dump_context.
583 (dump_begin_scope): Likewise.
584 (dump_end_scope): Likewise.
585 (selftest::temp_dump_context::temp_dump_context): New ctor.
586 (selftest::temp_dump_context::~temp_dump_context): New dtor.
587 (selftest::verify_item): New function.
588 (ASSERT_IS_TEXT): New macro.
589 (ASSERT_IS_TREE): New macro.
590 (ASSERT_IS_GIMPLE): New macro.
591 (selftest::test_capture_of_dump_calls): New test.
592 (selftest::dumpfile_c_tests): Call it.
593 * dumpfile.h (dump_printf, dump_printf_loc, dump_basic_block)
594 (dump_generic_expr_loc, dump_generic_expr, dump_gimple_stmt_loc)
595 (dump_gimple_stmt, dump_dec): Gather these related decls and add a
597 (dump_function, print_combine_total_stats, enable_rtl_dump_file)
598 (dump_node, dump_bb): Move these unrelated decls.
599 (class dump_manager): Add leading comment.
600 * optinfo.cc: New file.
601 * optinfo.h: New file.
603 2018-07-19 Michael Collison <michael.collison@arm.com>
604 Richard Henderson <rth@redhat.com>
606 * config/aarch64/aarch64.md (subv<GPI>4, usubv<GPI>4): New patterns.
607 (subti): Handle op1 zero.
608 (subvti4, usub4ti4): New.
609 (*sub<GPI>3_compare1_imm): New.
610 (sub<GPI>3_carryinCV): New.
611 (*sub<GPI>3_carryinCV_z1_z2, *sub<GPI>3_carryinCV_z1): New.
612 (*sub<GPI>3_carryinCV_z2, *sub<GPI>3_carryinCV): New.
614 2018-07-19 Michael Collison <michael.collison@arm.com>
615 Richard Henderson <rth@redhat.com>
617 * config/aarch64/aarch64.md: (addv<GPI>4, uaddv<GPI>4): New.
618 (addti3): Create simpler code if low part is already known to be 0.
619 (addvti4, uaddvti4): New.
620 (*add<GPI>3_compareC_cconly_imm): New.
621 (*add<GPI>3_compareC_cconly): New.
622 (*add<GPI>3_compareC_imm): New.
623 (*add<GPI>3_compareC): Rename from add<GPI>3_compare1; do not
624 handle constants within this pattern..
625 (*add<GPI>3_compareV_cconly_imm): New.
626 (*add<GPI>3_compareV_cconly): New.
627 (*add<GPI>3_compareV_imm): New.
628 (add<GPI>3_compareV): New.
629 (add<GPI>3_carryinC, add<GPI>3_carryinV): New.
630 (*add<GPI>3_carryinC_zero, *add<GPI>3_carryinV_zero): New.
631 (*add<GPI>3_carryinC, *add<GPI>3_carryinV): New.
632 ((*add<GPI>3_compareC_cconly_imm): Replace 'ne' operator
633 with 'comparison' operator.
634 (*add<GPI>3_compareV_cconly_imm): Ditto.
635 (*add<GPI>3_compareV_cconly): Ditto.
636 (*add<GPI>3_compareV_imm): Ditto.
637 (add<GPI>3_compareV): Ditto.
638 (add<mode>3_carryinC): Ditto.
639 (*add<mode>3_carryinC_zero): Ditto.
640 (*add<mode>3_carryinC): Ditto.
641 (add<mode>3_carryinV): Ditto.
642 (*add<mode>3_carryinV_zero): Ditto.
643 (*add<mode>3_carryinV): Ditto.
645 2018-07-19 Michael Collison <michael.collison@arm.com>
646 Richard Henderson <rth@redhat.com>
648 * config/aarch64/aarch64-modes.def (CC_V): New.
649 * config/aarch64/aarch64-protos.h
650 (aarch64_addti_scratch_regs): Declare
651 (aarch64_subvti_scratch_regs): Declare.
652 (aarch64_expand_subvti): Declare.
653 (aarch64_gen_unlikely_cbranch): Declare
654 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Test
655 for signed overflow using CC_Vmode.
656 (aarch64_get_condition_code_1): Handle CC_Vmode.
657 (aarch64_gen_unlikely_cbranch): New function.
658 (aarch64_addti_scratch_regs): New function.
659 (aarch64_subvti_scratch_regs): New function.
660 (aarch64_expand_subvti): New function.
662 2018-07-19 Andre Vieira <andre.simoesdiasvieira@arm.com>
664 * config/aarch64/aarch64-option-extensions.def: New entry for profile
666 * config/aarch64/aarch64.h (AARCH64_FL_PROFILE): New.
667 * doc/invoke.texi (aarch64-feature-modifiers): New entry for profile
670 2018-07-19 Andre Vieira <andre.simoesdiasvieira@arm.com>
673 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Make
674 address check not strict.
676 2018-07-19 Andre Vieira <andre.simoesdiasvieira@arm.com>
678 * config/aarch64/aarch64-simd.md (aarch64_simd_mov<VQ:mode>): Replace
680 (store_pair_lanes<mode>): Likewise.
681 * config/aarch64/aarch64-protos.h (aarch64_addr_query_type): Add new
682 enum value 'ADDR_QUERY_LDP_STP_N'.
683 * config/aarch64/aarch64.c (aarch64_addr_query_type): Likewise.
684 (aarch64_print_address_internal): Add declaration.
685 (aarch64_print_ldpstp_address): Remove.
686 (aarch64_classify_address): Adapt mode for 'ADDR_QUERY_LDP_STP_N'.
687 (aarch64_print_operand): Change printing of 'y'.
688 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Use
689 new enum value 'ADDR_QUERY_LDP_STP_N', don't hardcode mode and use
690 'true' rather than '1'.
691 * config/aarch64/constraints.md (Uml): Likewise.
692 (Uml): Rename to Umn.
695 2018-07-19 Richard Biener <rguenther@suse.de>
697 * tree-ssa-sccvn.h (struct vn_phi_s): Make phiargs member
699 * tree-ssa-sccvn.c: Remove alloc-pool.h use.
700 (vn_phi_hasher): Derive from nofree_ptr_hash and remove remove method.
701 (vn_reference_hasher): Likewise.
702 (struct vn_tables_s): Remove obstack and alloc-pool members.
703 (vn_tables_obstack, vn_tables_insert_obstack): New global obstacks.
704 (vn_nary_build_or_lookup_1): Manually build in vn_tables_insert_obstack.
705 (vn_reference_insert): Allocate from obstack instead of from alloc-pool.
706 (vn_reference_insert_pieces): Likewise.
707 (alloc_vn_nary_op_noinit): Adjust.
708 (vn_nary_op_insert_stmt): Allocate phiargs in-place.
710 (shared_lookup_phiargs): Remove.
711 (vn_phi_lookup): Allocate temporary vn_phi_s on the stack.
712 (vn_phi_insert): Allocate from obstack instead of from alloc-pool.
713 (visit_reference_op_call): Likewise.
714 (copy_nary, copy_phi, copy_reference): Remove.
715 (process_scc): Rewind the obstack when iterating. Do not
716 copy the elements to valid_info but just move them from one
717 hashtable to the other.
718 (allocate_vn_table): Adjust.
719 (free_vn_table): Likewise.
720 (init_scc_vn): Likewise.
721 (free_scc_vn): Likewise.
723 2018-07-19 H.J. Lu <hongjiu.lu@intel.com>
726 * config/i386/i386.c (rest_of_insert_endbranch): Lookup
727 indirect_return as function type attribute.
728 (ix86_attribute_table): Change indirect_return to function
730 * doc/extend.texi: Update indirect_return attribute.
732 2018-07-19 Aldy Hernandez <aldyh@redhat.com>
734 * wide-int.h (widest2_int): New.
735 * gimple-fold.c (arith_overflowed_p): Use it.
736 * tree.h (widest2_int_cst): New.
737 * tree-vrp.c (wide_int_binop_overflow): Rename from
739 Rewrite to work on trees.
740 (extract_range_from_multiplicative_op_1): Abstract code to...
741 (wide_int_range_min_max): ...here.
742 (wide_int_range_cross_product): ...and here.
743 (extract_range_from_binary_expr_1): Abstract overflow code to...
744 (wide_int_range_mult_wrapping): ...here.
745 * tree-vrp.h (wide_int_range_cross_product): New.
746 (wide_int_range_mult_wrapping): New.
748 2018-07-19 Andrew Senkevich <andrew.senkevich@intel.com>
749 Julia Koval <julia.koval@intel.com>
751 * config/i386/x86-tune-costs.h (skylake_memcpy,
752 skylake_memset): Replace rep_prefix with unrolling for size 512.
754 2018-07-18 Kugan Vivekanandarajah <kuganv@linaro.org>
757 * tree-ssa-phiopt.c (cond_removal_in_popcount_pattern): Handle comparision with EQ_EXPR
760 2018-07-18 Kelvin Nilsen <kelvin@gcc.gnu.org>
762 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Rename
763 this subsection to "PowerPC AltiVec/VSX Built-in Functions".
764 (PowerPC AltiVec/VSX Built-in Functions): New name for subsection
765 previously known as "PowerPC AltiVec Built-in Functions". Move
766 some material to new subsubsections "PowerPC AltiVec Built-in
767 Functions on ISA 2.06" and "PowerPC AltiVec Built-in Functions on
769 (PowerPC Altivec Built-in Functions on ISA 2.05): New subsubsection.
770 (PowerPC Altivec Built-in Functions on ISA 2.06): Likewise.
771 (PowerPC Altivec Built-in Functions on ISA 2.07): Likewise.
772 (PowerPC Altivec Built-in Functions on ISA 3.0): Likewise.
774 2018-07-18 Richard Biener <rguenther@suse.de>
776 PR tree-optimization/86557
777 * tree-vect-patterns.c (vect_recog_divmod_pattern): Also handle
780 2018-07-18 Ilya Leoshkevich <iii@linux.ibm.com>
782 * config/s390/s390.c (s390_function_profiler): Generate CFI.
784 2018-07-17 Jeff Law <law@redhat.com>
786 * config/arm/arm.c (get_label_padding): Update for recent
787 changes to label_to_alignment.
789 PR tree-optimization/86010
790 * tree-ssa-dse.c (compute_trims): Fix typo/thinko.
792 * config/mips/mips.c (vr4130_align_insns): Update for recent
793 changes to label_to_alignment.
795 * config/frv/frv.c (frv_label_align): Update for recent changes
796 to label_to_alignment.
798 * config/nios2/nios2.c (nios2_label_align): Update for recent
799 changes which dropped ALIGN_LABELS_LOG.
801 2018-07-17 Andreas Schwab <schwab@linux-m68k.org>
803 * config/m68k/m68k.md (umulsi3_highpart+1, const_umulsi3_highpart)
804 (smulsi3_highpart+1, const_smulsi3_highpart): Add CC_STATUS_INIT.
806 2018-07-17 Claudiu Zissulescu <claziss@synopsys.com>
808 * config/arc/arc.c (arc_label_align): Use align_labels instead of
809 deprecated align_labels_log.
811 2018-07-17 Richard Biener <rguenther@suse.de>
814 * dwarf2out.c (init_sections_and_labels): Always generate
815 a debug_line_str_section for early LTO debug.
816 (dwarf2out_finish): Reset debug_line_str_hash output early.
817 Bump counter for extra dwarf5 .debug_loc labels to not conflict
819 (dwarf2out_early_finish): Output debug_line_str.
821 2018-07-17 Robin Dapp <rdapp@linux.ibm.com>
823 * config/s390/s390.c (preferred_la_operand_p): Do not use LA with
824 index register on z196 or later.
826 2018-07-17 Robin Dapp <rdapp@linux.ibm.com>
828 * config/s390/s390.c (s390_default_align): Set default function
830 (s390_override_options_after_change): Call s390_default align.
831 (s390_option_override_internal): Call s390_default_align.
832 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define.
834 2018-07-17 Jakub Jelinek <jakub@redhat.com>
837 * omp-low.c (create_task_copyfn): Copy over also fields corresponding
838 to _looptemp_ clauses, other than the first two.
840 2018-07-17 Martin Liska <mliska@suse.cz>
842 * opts.c: Do not enable OPT_falign_* for -Os.
844 2018-07-17 Martin Liska <mliska@suse.cz>
846 * align.h (MAX_CODE_ALIGN): New.
847 (MAX_CODE_ALIGN_VALUE): New.
848 * common/config/i386/i386-common.c (ix86_handle_option):
849 (MAX_CODE_ALIGN): Moved to align.h.
850 * final.c (MAX_CODE_ALIGN): Likewise.
851 * opts.c (parse_and_check_align_values):
852 (MAX_CODE_ALIGN): Likewise.
853 (MAX_CODE_ALIGN_VALUE): Likewise.
855 2018-07-17 Martin Liska <mliska@suse.cz>
857 * config/i386/att.h (ASM_OUTPUT_ALIGN): Fix spacing
858 in order to fulfil coding style.
859 * config/i386/cygming.h (ASM_OUTPUT_ALIGN): Likewise.
860 * config/i386/gas.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
861 * config/i386/x86-64.h (ASM_OUTPUT_MAX_SKIP_PAD): Likewise.
862 * config/iq2000/iq2000.h (ASM_OUTPUT_ALIGN): Likewise.
863 * config/pa/pa.h (ASM_OUTPUT_ALIGN): Likewise.
864 * config/sparc/sol2.h (ASM_OUTPUT_ALIGN_WITH_NOP): Likewise.
865 * config/sparc/sparc.h (ASM_OUTPUT_ALIGN): Likewise.
866 * config/visium/visium.h (ASM_OUTPUT_ALIGN): Likewise.
867 (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
869 2018-07-17 Martin Liska <mliska@suse.cz>
872 * config/alpha/alpha.c (alpha_align_insns_1): Use align_functions directly.
873 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Use new return type
874 align_flags of label_to_alignment.
875 * config/m32r/m32r.h (LOOP_ALIGN): Wrap returned values into align_flags
877 * config/m68k/m68k.c: Do not use removed align_labels_value and
879 * config/nds32/nds32.h (JUMP_ALIGN): Wrap result into align_flags class.
880 (LOOP_ALIGN): Likewise.
881 (LABEL_ALIGN): Likewise.
882 * config/powerpcspe/powerpcspe.c (TARGET_ASM_LOOP_ALIGN_MAX_SKIP):
883 Remove not used macro.
884 (rs6000_loop_align): Change return type to align_flags.
885 (rs6000_loop_align_max_skip): Remove.
886 * config/rs6000/rs6000-protos.h (rs6000_loop_align):
887 Change return type to align_flags.
888 * config/rs6000/rs6000.c (TARGET_ASM_LOOP_ALIGN_MAX_SKIP):
889 Remove not used macro.
890 (rs6000_loop_align): Change return type to align_flags.
891 (rs6000_loop_align_max_skip): Remove.
892 * config/rx/rx.h (JUMP_ALIGN): Wrap integer values
893 * config/rx/rx-protos.h (rx_align_for_label): Make it
895 * config/rx/rx.c (rx_align_for_label): Change return type
897 (rx_max_skip_for_label): Remove TARGET_ASM_*_ALIGN_MAX_SKIP
899 into align_flags class.
900 (LABEL_ALIGN): Likewise.
901 (LOOP_ALIGN): Likewise.
902 * config/s390/s390.c (s390_label_align): Use align_flags
904 (s390_asm_output_function_label): Likewise.
905 * config/sh/sh.c (sh_override_options_after_change):
906 Use align_flags class directly without macros.
907 (find_barrier): Likewise.
908 (barrier_align): Likewise.
909 (sh_loop_align): Likewise.
910 * config/spu/spu.c (spu_option_override):
911 Use align_flags_tuple::get_value instead of removed macros.
912 (spu_sched_init): Likewise.
913 * config/spu/spu.h (GTY): Likewise.
914 * config/visium/visium.c (visium_option_override):
915 Set "8" as default secondary alignment.
916 * config/visium/visium.h (SUBALIGN_LOG): Define to 3
917 in order to guarantee secondary alignment of 8.
918 * coretypes.h: Include align.h header file.
919 * doc/tm.texi: Remove TARGET_ASM_JUMP_ALIGN_MAX_SKIP,
920 TARGET_ASM_LOOP_ALIGN_MAX_SKIP, TARGET_ASM_LABEL_ALIGN_MAX_SKIP
921 and TARGET_ASM_LABEL_ALIGN_AFTER_BARRIER_MAX_SKIP macros.
922 * doc/tm.texi.in: Likewise.
923 * final.c (struct label_alignment): Remove not used structure.
924 (LABEL_ALIGN): Change type to align_flags.
925 (LOOP_ALIGN): Likewise.
926 (JUMP_ALIGN): Likewise.
927 (default_loop_align_max_skip): Remove.
928 (default_label_align_max_skip): Likewise.
929 (default_jump_align_max_skip): Likewise.
930 (default_label_align_after_barrier_max_skip):
931 (LABEL_TO_ALIGNMENT): Change to access label_align vector.
932 (LABEL_TO_MAX_SKIP): Remove.
933 (label_to_alignment): Return align_flags type instead of integer.
934 (label_to_max_skip): Remove.
935 (align_fuzz): Use align_flags type.
936 (compute_alignments): Use align_flags type and use align_flags::max
937 to combine multiple alignments.
938 (grow_label_align): Grow vec instead of C array.
939 (update_alignments): Assign just LABEL_TO_ALIGNMENT.
940 (shorten_branches): Use align_flags type and use align_flags::max
941 to combine multiple alignments.
942 (final_scan_insn_1): Remove usage of secondary alignment that comes
943 from label alignment, but instead use proper secondary alignment
944 which is computed in grow_label_align.
945 * flags.h (struct align_flags_tuple): Move to align.h.
946 (struct align_flags): Likewise.
947 (state_align_loops): Rename to align_loops.
948 (state_align_jumps): Rename to align_jumps.
949 (state_align_labels): Rename to align_labels.
950 (state_align_functions): Rename to align_functions.
951 (align_loops_log): Remove.
952 (align_jumps_log): Remove.
953 (align_labels_log): Remove.
954 (align_functions_log): Remove.
955 (align_loops_max_skip): Remove.
956 (align_jumps_max_skip): Remove.
957 (align_labels_max_skip): Remove.
958 (align_functions_max_skip): Remove.
959 (align_loops_value): Remove.
960 (align_jumps_value): Remove.
961 (align_labels_value): Remove.
962 (align_functions_value): Remove.
963 * output.h (label_to_alignment): Change return type to align_flags.
964 (label_to_max_skip): Remove.
965 * target.def: Remove loop_align_max_skip, label_align_max_skip,
966 jump_align_max_skip macros.
967 * targhooks.h (default_loop_align_max_skip): Remove.
968 (default_label_align_max_skip): Likewise.
969 (default_jump_align_max_skip): Likewise.
970 (default_label_align_after_barrier_max_skip): Remove.
971 * toplev.c (read_log_maxskip): Use ::normalize function.
972 (parse_N_M): Remove not used argument and also call ::normalize.
973 (parse_alignment_opts): Do not pass unused arguments.
974 * varasm.c (assemble_start_function): Use directly align_functions
975 instead of removed macros.
976 * system.h: Do not poison removed macros.
978 2018-07-17 Jakub Jelinek <jakub@redhat.com>
981 * gimplify.c (gimplify_omp_for): Ensure taskloop firstprivatized init
982 and cond temporaries don't have reference type if iterator has
983 pointer type. For init use &for_pre_body instead of pre_p if
984 for_pre_body is non-empty.
986 2018-07-16 Segher Boessenkool <segher@kernel.crashing.org>
988 * config/rs6000/rs6000.md (trunc<mode>sf2): Expand truncates of
989 double-double modes to SFmode directly directly.
990 (trunc<mode>sf2_fprs): Delete.
992 2018-07-16 Segher Boessenkool <segher@kernel.crashing.org>
994 * config/rs6000/rs6000.c (init_float128_ibm): Use the correct names
995 for conversions between IFmode and the decimal floating point modes.
996 (init_float128_ieee): Use the correct names for conversions between
997 KFmode and the decimal floating point modes.
999 2018-07-16 Segher Boessenkool <segher@kernel.crashing.org>
1001 * config/rs6000/rs6000.c (init_float128_ibm): Use more correct names
1002 for the conversions between TDmode and IFmode.
1003 (init_float128_ieee): Use more correct names for the conversions
1004 between TDmode and KFmode.
1006 2018-07-16 Jakub Jelinek <jakub@redhat.com>
1008 PR tree-optimization/86526
1009 * builtins.c (expand_builtin_memcmp): Formatting fixes.
1010 (inline_expand_builtin_string_cmp): Likewise.
1011 (inline_string_cmp): Likewise. Use c_readstr instead of
1012 builtin_memcpy_read_str. Add unit_mode temporary.
1014 2018-07-16 Bernd Edlinger <bernd.edlinger@hotmail.de>
1017 * builtins.c (check_access): Bail out if range[0] is no INTEGER_CST.
1018 * expr.c (string_constant): Fix the element size of ARRAY_TYPE.
1020 2018-07-16 Kelvin Nilsen <kelvin@gcc.gnu.org>
1022 * doc/extend.texi (PowerPC AltiVec Built-in Functions):
1023 Alphabetize prototypes of built-in functions, separating out
1024 built-in functions that are listed in this section but should be
1025 described elsewhere.
1027 2018-07-16 Uros Bizjak <ubizjak@gmail.com>
1030 * expmed.c (emit_store_flag): Do not emit setcc followed by a
1031 conditional move when trapping comparison was split to a
1032 non-trapping one (and vice versa).
1034 2018-07-16 Ilya Leoshkevich <iii@linux.ibm.com>
1036 * config/s390/s390.c (s390_function_profiler): Generate nops
1037 instead of profiler call sequences.
1038 * config/s390/s390.opt: Add the new option.
1040 2018-07-16 Ilya Leoshkevich <iii@linux.ibm.com>
1042 * config/s390/s390.c (s390_function_profiler): Generate
1043 __mcount_loc section.
1044 * config/s390/s390.opt: Add the new option.
1046 2018-07-16 Ilya Leoshkevich <iii@linux.ibm.com>
1048 * common.opt: Add the new warning.
1049 * config/s390/s390.c (s390_function_profiler): Emit "brasl
1050 %r0,__fentry__" when -mfentry is specified.
1051 (s390_option_override_internal): Disallow -mfentry for 31-bit
1053 * config/s390/s390.opt: Add the new option.
1055 2018-07-16 Richard Biener <rguenther@suse.de>
1058 * dwarf2out.c (dwarf2out_register_external_die): Assign DIE parents
1059 for function-local FUNCTION_DECL and RESULT_DECL immediately.
1061 2018-07-16 Martin Liska <mliska@suse.cz>
1064 * ipa-pure-const.c (malloc_candidate_p): Revert ::get
1067 2017-07-16 Claudiu Zissulescu <claziss@synopsys.com>
1069 * config/arc/arcHS.md: Update ARCHS scheduling rules.
1071 2017-07-16 Claudiu Zissulescu <claziss@synopsys.com>
1073 * config/arc/arc-arch.h (arc_tune_attr): Add new tune parameters
1075 * config/arc/arc-cpus.def (hs4x): New cpu.
1077 * config/arc/arc-tables.opt: Regenerate.
1078 * config/arc/arc.c (arc_sched_issue_rate): New function.
1079 (TARGET_SCHED_ISSUE_RATE): Define.
1080 (TARGET_SCHED_EXPOSED_PIPELINE): Likewise.
1081 * config/arc/arc.md (attr type): Add fpu_fuse, fpu_sdiv, fpu_ddiv,
1083 (attr tune): Add ARCHS4x tune values.
1084 (attr tune_dspmpy): Define.
1085 (*tst): Correct instruction type.
1086 * config/arc/arcHS.md: Don't use this automaton for ARCHS4x cpus.
1087 * config/arc/arcHS4x.md: New file.
1088 * config/arc/fpu.md: Update instruction type attributes.
1089 * config/arc/t-multilib: Regenerate.
1091 2018-07-16 Tom de Vries <tdevries@suse.de>
1094 * var-tracking.c (vt_initialize): Fix pre_dec handling.
1096 2018-07-16 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
1098 * config/aarch64/atomics.md (aarch64_store_execlusive<mode>): Add
1101 2018-07-16 Eric Botcazou <ebotcazou@adacore.com>
1103 PR tree-optimization/86514
1104 * tree-ssa-reassoc.c (init_range_entry) <CASE_CONVERT>: Return for a
1105 conversion to a boolean type from a type with greater precision.
1107 2018-07-16 Tom de Vries <tdevries@suse.de>
1109 * var-tracking.c (vt_initialize): Print adjusted insn slim if
1110 dump_flags request TDF_SLIM.
1112 2018-07-16 Aldy Hernandez <aldyh@redhat.com>
1114 * fold-const.c (int_const_binop_1): Abstract...
1115 (wide_int_binop): ...wide int code here.
1116 (poly_int_binop): ...poly int code here.
1117 Abstract the rest of int_const_binop_1 into int_const_binop.
1118 * fold-const.h (wide_int_binop): New.
1119 * tree-vrp.c (vrp_int_const_binop): Call wide_int_binop.
1120 Remove useless PLUS/MINUS_EXPR case.
1121 (zero_nonzero_bits_from_vr): Move wide int code...
1122 (zero_nonzero_bits_from_bounds): ...here.
1123 (extract_range_from_binary_expr_1): Move mask optimization code...
1124 (range_easy_mask_min_max): ...here.
1125 * tree-vrp.h (zero_nonzero_bits_from_bounds): New.
1126 (range_easy_mask_min_max): New.
1128 2018-07-15 Jeff Law <law@redhat.com>
1131 * config/sh/sh.c (output_mi_thunk): Remove dead conditional
1134 2018-07-14 Jim Wilson <jimw@sifive.com>
1136 * config/riscv/linux.h (TARGET_ASM_FILE_END): New.
1138 2018-07-14 Paul Koning <ni1d@arrl.net>
1140 * config/pdp11/pdp11.c (pdp11_rtx_costs): Bugfixes.
1142 2018-07-13 Jan Hubicka <hubicka@ucw.cz>
1144 * lto-streamer-out.c (copy_function_or_variable): Dump info about
1147 2018-07-13 Bill Schmidt <wschmidt@linux.ibm.com>
1148 Steve Munroe <munroesj52@gmail.com>
1150 * config/rs6000/emmintrin.h (_mm_and_si128): New function.
1151 (_mm_andnot_si128): Likewise.
1152 (_mm_or_si128): Likewise.
1153 (_mm_xor_si128): Likewise.
1155 2018-07-13 Qing Zhao <qing.zhao@oracle.com>
1158 * builtins.c (expand_builtin_memcmp): Inline the calls first
1159 when result_eq is false.
1160 (expand_builtin_strcmp): Inline the calls first.
1161 (expand_builtin_strncmp): Likewise.
1162 (inline_string_cmp): New routine. Expand a string compare
1163 call by using a sequence of char comparison.
1164 (inline_expand_builtin_string_cmp): New routine. Inline expansion
1165 a call to str(n)cmp/memcmp.
1166 * doc/invoke.texi (--param builtin-string-cmp-inline-length):
1168 * params.def (BUILTIN_STRING_CMP_INLINE_LENGTH): New.
1170 2018-07-13 Richard Earnshaw <rearnsha@arm.com>
1172 * config/arm/driver-arm.c: Include arm-native.h.
1173 (host_detect_local_cpu): Use auto-generated data tables.
1174 (vendors, arm_cpu_table): Delete. Move part information to ...
1175 * config/arm/arm-cpus.in: ... here.
1176 * config/arm/parsecpu.awk (gen_native): New function.
1177 (vendor, part): New CPU fields.
1178 (END): Add support for building the native CPU detection tables.
1179 * config/arm/t-arm (arm-native.h): Add build rule.
1180 (driver-arm.o): Add dependency on arm-native.h.
1182 2018-07-13 Richard Biener <rguenther@suse.de>
1185 * match.pd (addr1 - addr2): Allow either of the operand to
1188 2018-07-13 Tom de Vries <tdevries@suse.de>
1190 * tree-inline.c (remap_ssa_name): Save and reuse debug exprs generated
1193 2018-07-13 Jackson Woodruff <jackson.woodruff@arm.com>
1195 * config/aarch64/aarch64.c (aarch64_operands_adjust_ok_for_ldpstp): Use
1196 arrays instead of numbered variables.
1198 2018-07-13 Eric Botcazou <ebotcazou@adacore.com>
1200 * config/sparc/sparc-protos.h (sparc_compute_frame_size): Delete.
1201 * config/sparc/sparc.c (sparc_compute_frame_size): Make static.
1203 2018-07-13 Richard Biener <rguenther@suse.de>
1206 * dwarf2out.c (gen_type_die_with_usage): Use scope_die_for
1207 instead of get_context_die.
1209 2018-07-13 Kugan Vivekanandarajah <kuganv@linaro.org>
1210 Richard Biener <rguenther@suse.de>
1213 * tree-ssa-loop-niter.c (number_of_iterations_popcount): Check
1214 that the loop latch destination where phi is defined.
1216 2018-07-12 Kito Cheng <kito.cheng@gmail.com>
1218 * config/riscv/riscv.c (enum riscv_privilege_levels): Add UNKNOWN_MODE.
1219 (riscv_expand_epilogue): Add assertion to check interrupt mode.
1220 (riscv_set_current_function): Extract getting interrupt type to new
1222 (riscv_get_interrupt_type): New function.
1223 (riscv_merge_decl_attributes): New function, checking interrupt type is
1225 (TARGET_MERGE_DECL_ATTRIBUTES): Define.
1227 2018-07-12 Paul Koning <ni1d@arrl.net>
1229 * config/pdp11/pdp11.c (pdp11_output_def): Fix typo in .set
1232 2018-07-12 Paul Koning <ni1d@arrl.net>
1234 * doc/rtl.texi (REG_NONNEG): Remove decrement and branch until
1235 zero reference, add doloop_end instead.
1236 * doc/md.texi (decrement_and_branch_until_zero): Remove.
1237 (Looping patterns): Remove decrement_and_branch_until_zero. Add
1238 detail for doloop_end.
1240 2018-07-12 Martin Sebor <msebor@redhat.com>
1243 * attribs.c (decl_attributes): Reject conflicting attributes before
1244 calling attribute handlers.
1246 2018-07-12 Jan Hubicka <hubicka@ucw.cz>
1248 * dumpfile.c (gcc::dump_manager::get_dump_file_name): Add PART
1250 (gcc::dump_manager::get_dump_file_name): likewise.
1251 (dump_begin): Likewise.
1252 * dumpfile.h (dump_begin): Update prototype.
1253 (gcc::dump_manager::get_dump_file_name,
1254 gcc::dump_manager::get_dump_file_name): Update prototype.
1256 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
1258 * internal-fn.h (vectorizable_internal_fn_p): New function.
1259 * tree-vect-slp.c (compatible_calls_p): Likewise.
1260 (vect_build_slp_tree_1): Remove nops argument. Handle calls
1261 to internal functions.
1262 (vect_build_slp_tree_2): Update call to vect_build_slp_tree_1.
1264 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
1266 * fold-const.h (inverse_conditions_p): Declare.
1267 * fold-const.c (inverse_conditions_p): New function.
1268 * match.pd: Use inverse_conditions_p. Add folds of view_converts
1269 that test the inverse condition of a conditional internal function.
1270 * internal-fn.h (vectorized_internal_fn_supported_p): Declare.
1271 * internal-fn.c (internal_fn_mask_index): Handle conditional
1273 (vectorized_internal_fn_supported_p): New function.
1274 * tree-if-conv.c: Include internal-fn.h and fold-const.h.
1275 (any_pred_load_store): Replace with...
1276 (need_to_predicate): ...this new variable.
1277 (redundant_ssa_names): New variable.
1278 (ifcvt_can_use_mask_load_store): Move initial checks to...
1279 (ifcvt_can_predicate): ...this new function. Handle tree codes
1280 for which a conditional internal function exists.
1281 (if_convertible_gimple_assign_stmt_p): Use ifcvt_can_predicate
1282 instead of ifcvt_can_use_mask_load_store. Update after variable
1284 (predicate_load_or_store): New function, split out from
1285 predicate_mem_writes.
1286 (check_redundant_cond_expr): New function.
1287 (value_available_p): Likewise.
1288 (predicate_rhs_code): Likewise.
1289 (predicate_mem_writes): Rename to...
1290 (predicate_statements): ...this. Use predicate_load_or_store
1291 and predicate_rhs_code.
1292 (combine_blocks, tree_if_conversion): Update after above name changes.
1293 (ifcvt_local_dce): Handle redundant_ssa_names.
1294 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern): Handle
1295 general conditional functions.
1296 * tree-vect-stmts.c (vectorizable_call): Likewise.
1298 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
1299 Alan Hayward <alan.hayward@arm.com>
1300 David Sherwood <david.sherwood@arm.com>
1302 * internal-fn.h (can_interpret_as_conditional_op_p): Declare.
1303 * internal-fn.c (can_interpret_as_conditional_op_p): New function.
1304 * tree-ssa-math-opts.c (convert_mult_to_fma_1): Handle conditional
1305 plus and minus and convert them into IFN_COND_FMA-based sequences.
1306 (convert_mult_to_fma): Handle conditional plus and minus.
1308 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
1310 * doc/md.texi (cond_fma, cond_fms, cond_fnma, cond_fnms): Document.
1311 * optabs.def (cond_fma_optab, cond_fms_optab, cond_fnma_optab)
1312 (cond_fnms_optab): New optabs.
1313 * internal-fn.def (COND_FMA, COND_FMS, COND_FNMA, COND_FNMS): New
1315 (FMA): Use DEF_INTERNAL_FLT_FN rather than DEF_INTERNAL_FLT_FLOATN_FN.
1316 * internal-fn.h (get_conditional_internal_fn): Declare.
1317 (get_unconditional_internal_fn): Likewise.
1318 * internal-fn.c (cond_ternary_direct): New macro.
1319 (expand_cond_ternary_optab_fn): Likewise.
1320 (direct_cond_ternary_optab_supported_p): Likewise.
1321 (FOR_EACH_COND_FN_PAIR): Likewise.
1322 (get_conditional_internal_fn): New function.
1323 (get_unconditional_internal_fn): Likewise.
1324 * gimple-match.h (gimple_match_op::MAX_NUM_OPS): Bump to 5.
1325 (gimple_match_op::gimple_match_op): Add a new overload for 5
1327 (gimple_match_op::set_op): Likewise.
1328 (gimple_resimplify5): Declare.
1329 * genmatch.c (decision_tree::gen): Generate simplifications for
1331 * gimple-match-head.c (gimple_simplify): Define an overload for
1332 5 operands. Handle calls with 5 arguments in the top-level overload.
1333 (convert_conditional_op): Handle conversions from unconditional
1334 internal functions to conditional ones.
1335 (gimple_resimplify5): New function.
1336 (build_call_internal): Pass a fifth operand.
1337 (maybe_push_res_to_seq): Likewise.
1338 (try_conditional_simplification): Try converting conditional
1339 internal functions to unconditional internal functions.
1340 Handle 3-operand unconditional forms.
1341 * match.pd (UNCOND_TERNARY, COND_TERNARY): Operator lists.
1342 Define ternary equivalents of the current rules for binary conditional
1344 * config/aarch64/aarch64.c (aarch64_preferred_else_value): Handle
1346 * config/aarch64/iterators.md (UNSPEC_COND_FMLA, UNSPEC_COND_FMLS)
1347 (UNSPEC_COND_FNMLA, UNSPEC_COND_FNMLS): New unspecs.
1348 (optab): Handle them.
1349 (SVE_COND_FP_TERNARY): New int iterator.
1350 (sve_fmla_op, sve_fmad_op): New int attributes.
1351 * config/aarch64/aarch64-sve.md (cond_<optab><mode>)
1352 (*cond_<optab><mode>_2, *cond_<optab><mode_4)
1353 (*cond_<optab><mode>_any): New SVE_COND_FP_TERNARY patterns.
1355 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
1357 * target.def (preferred_else_value): New target hook.
1358 * doc/tm.texi.in (TARGET_PREFERRED_ELSE_VALUE): New hook.
1359 * doc/tm.texi: Regenerate.
1360 * targhooks.h (default_preferred_else_value): Declare.
1361 * targhooks.c (default_preferred_else_value): New function.
1362 * internal-fn.h (conditional_internal_fn_code): Declare.
1363 * internal-fn.c (FOR_EACH_CODE_MAPPING): New macro.
1364 (get_conditional_internal_fn): Use it.
1365 (conditional_internal_fn_code): New function.
1366 * gimple-match.h (gimple_match_cond): New struct.
1367 (gimple_match_op): Add a cond member function.
1368 (gimple_match_op::gimple_match_op): Update all forms to take a
1370 * genmatch.c (expr::gen_transform): Use the same condition as res_op
1371 for the suboperation, but don't specify a particular else_value.
1372 * tree-ssa-sccvn.c (vn_nary_simplify, vn_reference_lookup_3)
1373 (visit_nary_op, visit_reference_op_load): Pass
1374 gimple_match_cond::UNCOND to the gimple_match_op constructor.
1375 * gimple-match-head.c: Include tree-eh.h
1376 (convert_conditional_op): New function.
1377 (maybe_resimplify_conditional_op): Likewise.
1378 (gimple_resimplify1): Call maybe_resimplify_conditional_op.
1379 (gimple_resimplify2): Likewise.
1380 (gimple_resimplify3): Likewise.
1381 (gimple_resimplify4): Likewise.
1382 (maybe_push_res_to_seq): Return null for conditional operations.
1383 (try_conditional_simplification): New function.
1384 (gimple_simplify): Call it. Pass conditions to the gimple_match_op
1386 * match.pd: Fold VEC_COND_EXPRs of an IFN_COND_* call to a new
1388 * config/aarch64/aarch64.c (aarch64_preferred_else_value): New
1390 (TARGET_PREFERRED_ELSE_VALUE): Redefine.
1392 2018-07-12 Jan Hubicka <hubicka@ucw.cz>
1394 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not stream
1396 (hash_tree): Do not hash DECL_FCONTEXT
1397 * tree-streamer-in.c (lto_input_ts_field_decl_tree_pointers):
1398 Do not stream DECL_FCONTEXT.
1399 * tree-streamer-out.c (write_ts_field_decl_tree_pointers): Likewise.
1400 * tree.c (free_lang_data_in_decl): Free DECL_FCONTEXT.
1402 2018-07-12 Richard Biener <rguenther@suse.de>
1405 * dwarf2out.c (gen_block_die): Only output blocks when they have
1406 at least one !DECL_IGNORED_P variable.
1408 2018-07-12 Richard Biener <rguenther@suse.de>
1411 * config/gnu-user.h (GNU_USER_TARGET_NO_PTHREADS_LIB_SPEC):
1412 Remove -mieee-fp handling.
1414 2018-07-12 Richard Biener <rguenther@suse.de>
1416 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Remove bogus
1417 left-over from last patch.
1419 2018-07-12 Jakub Jelinek <jakub@redhat.com>
1421 PR tree-optimization/86492
1422 * gimple-ssa-store-merging.c
1423 (imm_store_chain_info::coalesce_immediate_stores): Call
1424 check_no_overlap even for the merge_overlapping case. Formatting fix.
1426 2018-07-12 Richard Biener <rguenther@suse.de>
1429 * fold-const.c (fold_binary_op_with_conditional_arg): Do not
1430 move possibly trapping operations into the conditional.
1432 2018-07-12 Richard Biener <rguenther@suse.de>
1434 * tree-ssa-sccvn.c (mprts_hook_cnt): Remove.
1435 (vn_lookup_simplify_result): Remove recursion limit applied
1437 (vn_nary_build_or_lookup_1): Adjust.
1438 (try_to_simplify): Likewise.
1439 * gimple-match-head.c (gimple_resimplify1): Instead apply one
1441 (gimple_resimplify2): Likewise.
1442 (gimple_resimplify3): Likewise.
1443 (gimple_resimplify4): Likewise.
1445 2018-07-11 Jakub Jelinek <jakub@redhat.com>
1447 * config/i386/avx512bitalgintrin.h (_mm512_mask_bitshuffle_epi64_mask):
1448 Use __mmask64 type instead of __mmask8 for __M argument.
1449 * config/i386/avx512fintrin.h (_mm512_mask_xor_epi64,
1450 _mm512_maskz_xor_epi64): Use __mmask8 type instead of __mmask16 for
1452 (_mm512_mask_cmpneq_epi64_mask): Use __mmask8 type instead of
1453 __mmask16 for __M argument.
1454 (_mm512_maskz_insertf32x4, _mm512_maskz_inserti32x4,
1455 _mm512_mask_insertf32x4, _mm512_mask_inserti32x4): Cast last argument
1456 to __mmask16 instead of __mmask8.
1457 * config/i386/avx512vlintrin.h (_mm_mask_add_ps, _mm_maskz_add_ps,
1458 _mm256_mask_add_ps, _mm256_maskz_add_ps, _mm_mask_sub_ps,
1459 _mm_maskz_sub_ps, _mm256_mask_sub_ps, _mm256_maskz_sub_ps,
1460 _mm256_maskz_cvtepi32_ps, _mm_maskz_cvtepi32_ps): Use __mmask8 type
1461 instead of __mmask16 for __U argument.
1462 * config/i386/avx512vlbwintrin.h (_mm_mask_cmp_epi8_mask): Use
1463 __mmask16 instead of __mmask8 for __U argument.
1464 (_mm256_mask_cmp_epi8_mask): Use __mmask32 instead of __mmask16 for
1466 (_mm256_cmp_epi8_mask): Use __mmask32 return type instead of
1468 (_mm_mask_cmp_epu8_mask): Use __mmask16 instead of __mmask8 for __U
1470 (_mm256_mask_cmp_epu8_mask): Use __mmask32 instead of __mmask16 for
1472 (_mm256_cmp_epu8_mask): Use __mmask32 return type instead of
1474 (_mm_mask_cmp_epi16_mask): Cast last argument to __mmask8 instead
1476 (_mm256_mask_cvtepi8_epi16): Use __mmask16 instead of __mmask32 for
1478 (_mm_mask_cvtepi8_epi16): Use __mmask8 instead of __mmask32 for
1480 (_mm256_mask_cvtepu8_epi16): Use __mmask16 instead of __mmask32 for
1482 (_mm_mask_cvtepu8_epi16): Use __mmask8 instead of __mmask32 for
1484 (_mm256_mask_cmpneq_epu8_mask, _mm256_mask_cmplt_epu8_mask,
1485 _mm256_mask_cmpge_epu8_mask, _mm256_mask_cmple_epu8_mask): Change
1486 return type as well as __M argument type and all casts from __mmask8
1488 (_mm256_mask_cmpneq_epu16_mask, _mm256_mask_cmplt_epu16_mask,
1489 _mm256_mask_cmpge_epu16_mask, _mm256_mask_cmple_epu16_mask): Change
1490 return type as well as __M argument type and all casts from __mmask8
1492 (_mm256_mask_cmpneq_epi8_mask, _mm256_mask_cmplt_epi8_mask,
1493 _mm256_mask_cmpge_epi8_mask, _mm256_mask_cmple_epi8_mask): Change
1494 return type as well as __M argument type and all casts from __mmask8
1496 (_mm256_mask_cmpneq_epi16_mask, _mm256_mask_cmplt_epi16_mask,
1497 _mm256_mask_cmpge_epi16_mask, _mm256_mask_cmple_epi16_mask): Change
1498 return type as well as __M argument type and all casts from __mmask8
1500 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_shrdi_epi32,
1501 _mm_mask_shldi_epi32): Cast last argument to __mmask8 instead of
1504 2018-07-11 Grazvydas Ignotas <notasas@gmail.com>
1506 * config/i386/avx512bwintrin.h: (_mm512_mask_cmp_epi8_mask,
1507 _mm512_mask_cmp_epu8_mask): Use __mmask64 type instead of __mmask32
1510 2018-07-11 Paul Koning <ni1d@arrl.net>
1512 * doc/md.texi (define_subst): Document how multiple occurrences of
1513 the same argument in the replacement pattern are handled.
1515 2018-07-11 Paul Koning <ni1d@arrl.net>
1517 * doc/extend.texi (Common Variable Attributes): Move "mode" into
1519 (Common Type Attributes): Add "mode" attribute.
1521 2018-07-11 Jan Hubicka <hubicka@ucw.cz>
1523 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not
1524 stream DECL_ORIGINAL_TYPE.
1525 (DFS::DFS_write_tree_body): Drop hack handling local external decls.
1526 (hash_tree): Do not walk DECL_ORIGINAL_TYPE.
1527 * tree-streamer-in.c (lto_input_ts_decl_non_common_tree_pointers):
1528 Do not walk original type.
1529 * tree-streamer-out.c (streamer_write_chain): Drop hack handling
1531 (write_ts_decl_non_common_tree_pointers): Do not stream
1533 * tree.c (free_lang_data_in_decl): Clear DECL_ORIGINAL_TYPE.
1534 (find_decls_types_r): Do not walk DEC_ORIGINAL_TYPE.
1536 2018-07-11 Aldy Hernandez <aldyh@redhat.com>
1538 * tree-ssa-threadupdate.c (thread_through_all_blocks): Do not jump
1539 thread twice from the same starting edge.
1541 2018-07-11 Aldy Hernandez <aldyh@redhat.com>
1543 * vr-values.c (gimple_stmt_nonzero_p): Abstract common code to...
1544 * gimple.c (gimple_call_nonnull_result_p): ...here...
1545 (gimple_call_nonnull_arg): ...and here.
1546 * gimple.h (gimple_call_nonnull_result_p): New.
1547 (gimple_call_nonnull_arg): New.
1549 2018-07-11 Richard Earnshaw <rearnsha@arm.com>
1551 * config/arm/arm-cpus.in: Move information from fpu field of each
1552 cpu definition to the isa field.
1553 * config/arm/parsecpu.awk (fpu): Delete match rule.
1554 (gen_comm_data): Don't add bits from the CPU's FPU entry.
1556 2018-07-11 Richard Biener <rguenther@suse.de>
1559 * dwarf2out.c (init_sections_and_labels): Use
1560 output_asm_line_debug_info consistently.
1561 (dwarf2out_early_finish): Likewise.
1562 (dwarf2out_finish): Remove DW_AT_stmt_list from early generated
1565 2018-07-11 Richard Biener <rguenther@suse.de>
1567 * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely_1):
1568 Rework father_bb setting in a way to avoid propagating constants
1569 multiple times on a loop body.
1571 2018-07-10 Mark Wielaard <mark@klomp.org>
1574 * dwarf2out.c (output_macinfo_op): Fix dwarf_FORM typo in gcc_assert.
1576 2018-07-10 Richard Biener <rguenther@suse.de>
1578 * hash-map.h (hash_map::iterator::operator*): Return
1579 references to key and value.
1581 2018-07-10 Jakub Jelinek <jakub@redhat.com>
1584 * gimplify.c (find_combined_omp_for): Add DATA argument, in addition
1585 to finding the inner OMP_FOR/OMP_SIMD stmt find non-trivial wrappers,
1586 BLOCKs with BLOCK_VARs, OMP_PARALLEL in between, OMP_FOR in between.
1587 (gimplify_omp_for): For composite loops, move outer
1588 OMP_{DISTRIBUTE,TASKLOOP,FOR,PARALLEL} right around innermost
1589 OMP_FOR/OMP_SIMD if there are any non-trivial wrappers. For class
1590 iterators add any needed clauses. Allow OMP_FOR_ORIG_DECLS to contain
1591 TREE_LIST for both the original class iterator and the "last" helper
1592 var. Gimplify OMP_FOR_PRE_BODY before the outermost composite
1593 loop, remember has_decl_expr from outer composite loops for the
1594 innermost OMP_SIMD in TREE_PRIVATE bit on OMP_FOR_INIT.
1596 2018-07-09 Martin Sebor <msebor@redhat.com>
1600 * builtins.c (c_strlen): Avoid out-of-bounds warnings when
1601 accessing implicitly initialized array elements.
1602 * expr.c (string_constant): Handle string initializers of
1603 character arrays within aggregates.
1604 * gimple-fold.c (fold_array_ctor_reference): Add argument.
1605 Store element offset. As a special case, handle zero size.
1606 (fold_nonarray_ctor_reference): Same.
1607 (fold_ctor_reference): Add argument. Store subobject offset.
1608 * gimple-fold.h (fold_ctor_reference): Add argument.
1610 2018-07-09 Paul Koning <ni1d@arrl.net>
1612 * config/pdp11/pdp11.c (pdp11_addr_cost): New function.
1613 (pdp11_insn_cost): New function.
1614 (pdp11_md_asm_adjust): New function.
1615 (TARGET_INVALID_WITHIN_DOLOOP): Define.
1616 (pdp11_rtx_costs): Update to match machine better.
1617 (output_addr_const_pdp11): Correct format mismatch warnings.
1618 * config/pdp11/pdp11.h (SLOW_BYTE_ACCESS): Correct definition.
1619 * config/pdp11/pdp11.md: General change to add base_cost and/or
1620 length attributes for use by new pdp11_insn_cost function.
1621 (MIN_BRANCH): Correct definition.
1623 (doloop_end): Use standard pattern name for looping pattern.
1624 (doloop_end_nocc): New.
1625 (movsf): Add another constraint alternative.
1626 (zero_extendqihi2): Add constraint alternatives for not in place
1628 (zero_extendhisi2): Remove.
1629 (shift patterns): Add CC handling variants.
1633 (define_peephole2): New peephole to recognize mov that sets CC for
1636 2018-07-09 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1638 * config/sparc/sparc.c (sparc_fold_builtin) <SPARC_BUILTIN_PDIST,
1639 SPARC_BUILTIN_PDISTN>: Adapt for signature change of wi::neg,
1642 2018-07-09 Jakub Jelinek <jakub@redhat.com>
1645 * real.c (real_nextafter): Return true if result is denormal.
1647 2018-07-09 Martin Liska <mliska@suse.cz>
1649 * common.opt: Add back wrongly removed attribute.
1651 2018-07-09 Richard Biener <rguenther@suse.de>
1654 * dwarf2out.c (gen_block_die): For an early generated DIE
1655 always output high/low PC attributes.
1657 2018-07-09 Tom de Vries <tdevries@suse.de>
1659 * cfgexpand.c (expand_debug_source_expr): Handle VAR_DECL.
1660 * tree-inline.c (remap_ssa_name): Handle default def ssa_name mapping
1661 onto VAR_DECL with abstract origin.
1663 2018-07-07 Jim Wilson <jimw@sifive.com>
1665 * config/riscv/riscv.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): New.
1667 2018-07-07 Tom de Vries <tdevries@suse.de>
1669 * tree-dfa.c (dump_enumerated_decls): Handle cfun->cfg == NULL.
1671 2018-07-07 Aldy Hernandez <aldyh@redhat.com>
1673 * tree-vrp.c (vrp_int_const_binop): Change overflow type to
1675 (combine_bound): Use wide-int overflow calculation instead of
1677 * calls.c (maybe_warn_alloc_args_overflow): Change overflow type to
1679 * fold-const.c (int_const_binop_2): Same.
1680 (extract_muldiv_1): Same.
1681 (fold_div_compare): Same.
1682 (fold_abs_const): Same.
1684 * poly-int.h (add): Same.
1688 * predict.c (predict_iv_comparison): Same.
1689 * profile-count.c (slow_safe_scale_64bit): Same.
1690 * simplify-rtx.c (simplify_const_binary_operation): Same.
1691 * tree-chrec.c (tree_fold_binomial): Same.
1692 * tree-data-ref.c (split_constant_offset_1): Same.
1693 * tree-if-conv.c (idx_within_array_bound): Same.
1694 * tree-scalar-evolution.c (iv_can_overflow_p): Same.
1695 * tree-ssa-phiopt.c (minmax_replacement): Same.
1696 * tree-vect-loop.c (is_nonwrapping_integer_induction): Same.
1697 * tree-vect-stmts.c (vect_truncate_gather_scatter_offset): Same.
1698 * vr-values.c (vr_values::adjust_range_with_scev): Same.
1699 * wide-int.cc (wi::add_large): Same.
1700 (wi::mul_internal): Same.
1701 (wi::sub_large): Same.
1702 (wi::divmod_internal): Same.
1703 * wide-int.h: Change overflow type to overflow_type for neg, add,
1704 mul, smul, umul, div_trunc, div_floor, div_ceil, div_round,
1705 mod_trunc, mod_ceil, mod_round, add_large, sub_large,
1706 mul_internal, divmod_internal.
1707 (overflow_type): New enum.
1708 (accumulate_overflow): New.
1710 2018-07-06 Kugan Vivekanandarajah <kugan.vivekanandarajah@linaro.org>
1712 * tree-ssa-phiopt.c (cond_removal_in_popcount_pattern): New.
1713 (tree_ssa_phiopt_worker): Call cond_removal_in_popcount_pattern.
1715 2018-07-06 Kugan Vivekanandarajah <kuganv@linaro.org>
1717 * tree-ssa-loop-niter.c (number_of_iterations_popcount): If popcount
1718 argument is checked for zero before entering loop, avoid checking again.
1720 2018-07-06 Kugan Vivekanandarajah <kuganv@linaro.org>
1722 * gimplify.h (generic_expr_could_trap_p): Set as global function.
1723 * gimplify.h (generic_expr_could_trap_p): Likwise.
1724 * tree-scalar-evolution.c (expression_expensive_p): Handle COND_EXPR.
1726 2018-07-06 Jakub Jelinek <jakub@redhat.com>
1728 PR tree-optimization/86401
1729 * fold-const.c (fold_binary_loc) <case BIT_AND_EXPR>: Move the
1730 ((A & N) + B) & M -> (A + B) & M etc. optimization into ...
1731 (fold_bit_and_mask): ... here. New helper function for match.pd.
1732 * fold-const.h (fold_bit_and_mask): Declare.
1733 * match.pd (((A & N) + B) & M -> (A + B) & M): New optimization.
1735 2018-07-06 Peter Bergner <bergner@linux.ibm.com>
1738 * target.def (translate_mode_attribute): New hook.
1739 * targhooks.h (default_translate_mode_attribute): Declare.
1740 * targhooks.c (default_translate_mode_attribute): New function.
1741 * doc/tm.texi.in (TARGET_TRANSLATE_MODE_ATTRIBUTE): New hook.
1742 * doc/tm.texi: Regenerate.
1743 * config/rs6000/rs6000.c (TARGET_TRANSLATE_MODE_ATTRIBUTE): Define.
1744 (rs6000_translate_mode_attribute): New function.
1746 2018-07-06 Paul Koning <ni1d@arrl.net>
1748 * doc/md.texi (define_split): Document DONE and FAIL.
1749 (define_peephole2): Ditto.
1751 2018-07-05 Jeff Law <law@redhat.com>
1753 PR tree-optimization/86010
1754 * tree-ssa-dse.c (compute_trims): More aggressively trim at
1755 both the head and tail of mem* and str* calls.
1757 2018-07-05 Jim Wilson <jimw@sifive.com>
1759 * config.gcc (riscv*-*-*): When setting xlen, handle riscv-*.
1761 2018-07-05 Indu Bhagat <indu.bhagat@oracle.com>
1763 * config/aarch64/aarch64-simd.md: correct flags text for
1764 MIN_EXPR replacement.
1766 2018-07-05 James Clarke <jrtc27@jrtc27.com>
1768 * configure: Regenerated.
1770 2018-07-05 Carl Love <cel@us.ibm.com>
1772 * config/rs6000/rs6000-c.c: Map ALTIVEC_BUILTIN_VEC_UNPACKH for
1773 float argument to VSX_BUILTIN_DOUBLEH_V4SF.
1774 Map ALTIVEC_BUILTIN_VEC_UNPACKL for float argument to
1775 VSX_BUILTIN_DOUBLEL_V4SF.
1777 2018-07-05 Martin Sebor <msebor@redhat.com>
1780 * tree-ssa-strlen.c (maybe_set_strlen_range): Use type size rather
1781 than its domain to compute its the upper bound of a char array.
1783 2018-07-05 Nathan Sidwell <nathan@acm.org>
1785 Replace NO_IMPLICIT_EXTERN_C with SYSTEM_IMPLICIT_EXTERN_C.
1786 * doc/cpp.texi: Update comment.
1787 * doc/tm.texi: Rebuilt.
1788 * doc/tm.texi.in (NO_IMPLICIT_EXTERN_C): Replace with ...
1789 (SYSTEM_IMPLICIT_EXTERN_C): ... this, opposite sense.
1790 * doc/extend.texi (Backwards Compatibility): Clarify it is system
1791 headers affected by extern "C".
1792 * system.h: Poison NO_IMPLICIT_EXTERN_C.
1793 * config/alpha/alpha.h, config/arm/uclinux-elf.h,
1794 config/bfin/elf.h, config/cris/cris.h, config/darwin.h,
1795 config/dragonfly.h, config/freebsd.h, config/gnu-user.h,
1796 config/i386/cygming.h, config/i386/djgpp.h, config/i386/nto.h,
1797 config/ia64/hpux.h, config/lm32/lm32.h, config/lm32/uclinux-elf.h,
1798 config/lynx.h, config/mips/elf.h, config/mmix/mmix.h,
1799 config/netbsd.h, config/pa/pa-hpux.h, config/powerpcspe/sysv4.h,
1800 config/riscv/elf.h, config/rs6000/sysv4.h, config/rtems.h,
1801 config/s390/tpf.h, config/sh/newlib.h, config/sol2.h,
1802 config/sparc/openbsd64.h, config/sparc/sp-elf.h,
1803 config/sparc/sp64-elf.h, config/spu/spu.h,
1804 config/stormy16/stormy16.h, config/v850/v850.h,
1805 config/visium/visium.h, config/vx-common.h, config/xtensa/elf.h: Don't
1806 define NO_IMPLICIT_EXTERN_C.
1807 * config/rs6000/aix.h: Set SYSTEM_IMPLICIT_EXTERN_C.
1809 2018-07-05 Tamar Christina <tamar.christina@arm.com>
1812 * config/arm/arm.c (arm_can_change_mode_class): Use GET_MODE_UNIT_SIZE
1813 instead of GET_MODE_SIZE when comparing Units.
1815 2018-07-05 Tamar Christina <tamar.christina@arm.com>
1818 * rtlanal.c (set_noop_p): Constrain on mode change,
1819 include hard-reg-set.h
1821 2018-07-05 Tamar Christina <tamar.christina@arm.com>
1823 * config/aarch64/aarch64.c (aarch64_expand_movmem): Fix mode size.
1825 2018-07-05 Jakub Jelinek <jakub@redhat.com>
1828 2018-07-04 Maxim Ostapenko <m.ostapenko@samsung.com>
1831 * config/gnu-user.h (LIBASAN_EARLY_SPEC): Pass -lstdc++ for static
1833 * gcc.c: Do not pass LIBUBSAN_SPEC if ASan is enabled with UBSan.
1835 2018-07-04 Maxim Ostapenko <m.ostapenko@samsung.com>
1838 * config/gnu-user.h (LIBASAN_EARLY_SPEC): Pass -lstdc++ for static
1840 * gcc.c: Do not pass LIBUBSAN_SPEC if ASan is enabled with UBSan.
1842 2018-07-04 Eric Botcazou <ebotcazou@adacore.com>
1845 * expmed.c (choose_multiplier): Fix incorrect comparison with mask.
1847 2018-07-04 Aldy Hernandez <aldyh@redhat.com>
1849 * tree-vrp.c (extract_range_from_binary_expr_1): Initialize
1850 neg_*_op* variables.
1852 2018-07-04 Martin Liska <mliska@suse.cz>
1854 * tree-switch-conversion.c: Define
1855 max_ratio_for_speed and max_ratio_for_size constants.
1857 2018-07-04 Denys Vlasenko <dvlasenk@redhat.com>
1858 Martin Liska <mliska@suse.cz>
1863 * common.opt: Rename align options with 'str_' prefix.
1864 * common/config/i386/i386-common.c (set_malign_value): New
1866 (ix86_handle_option): Use it to set -falign-* options/
1867 * config/aarch64/aarch64-protos.h (struct tune_params): Change
1868 type from int to string.
1869 * config/aarch64/aarch64.c: Update default values from int
1871 * config/alpha/alpha.c (alpha_override_options_after_change):
1873 * config/arm/arm.c (arm_override_options_after_change_1): Likewise.
1874 * config/i386/dragonfly.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
1875 max skip conditionally.
1876 * config/i386/freebsd.h (SUBALIGN_LOG): New.
1877 (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
1878 max skip conditionally.
1879 * config/i386/gas.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
1880 max skip conditionally.
1881 * config/i386/gnu-user.h (SUBALIGN_LOG): New.
1882 (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
1883 max skip conditionally.
1884 * config/i386/i386.c (struct ptt): Change type from int to
1886 (ix86_default_align): Set default values.
1887 * config/i386/i386.h (ASM_OUTPUT_MAX_SKIP_PAD): Print
1888 max skip conditionally.
1889 * config/i386/iamcu.h (SUBALIGN_LOG): New.
1890 (ASM_OUTPUT_MAX_SKIP_ALIGN):
1891 * config/i386/lynx.h (ASM_OUTPUT_MAX_SKIP_ALIGN):
1892 * config/i386/netbsd-elf.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
1893 max skip conditionally.
1894 * config/i386/openbsdelf.h (SUBALIGN_LOG): New.
1895 (ASM_OUTPUT_MAX_SKIP_ALIGN) Print max skip conditionally.:
1896 * config/i386/x86-64.h (SUBALIGN_LOG): New.
1897 (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
1898 max skip conditionally.
1899 (ASM_OUTPUT_MAX_SKIP_PAD): Likewise.
1900 * config/ia64/ia64.c (ia64_option_override): Set default values
1901 for alignment options.
1902 * config/m68k/m68k.c: Handle new str_align_* options.
1903 * config/mips/mips.c (mips_set_compression_mode): Change
1905 (mips_option_override): Set default values for options.
1906 * config/powerpcspe/powerpcspe.c (rs6000_option_override_internal):
1908 * config/rs6000/rs6000.c (rs6000_option_override_internal):
1910 * config/rx/rx.c (rx_option_override): Likewise.
1911 * config/rx/rx.h (JUMP_ALIGN): Use align_jumps_log.
1912 (LABEL_ALIGN): Use align_labels_log.
1913 (LOOP_ALIGN): Use align_loops_align.
1914 * config/s390/s390.c (s390_asm_output_function_label): Use new
1916 * config/sh/sh.c (sh_override_options_after_change):
1917 Change type of constants.
1918 * config/spu/spu.c (spu_sched_init): Likewise.
1919 * config/sparc/sparc.c (sparc_option_override): Set default
1921 * config/visium/visium.c (visium_option_override): Likewise.
1922 * config/visium/visium.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Do not
1923 emit p2align format with last argument if it's not needed.
1924 * doc/invoke.texi: Document extended format of -falign-*.
1925 * final.c: Use align_labels alignment.
1926 * flags.h (struct target_flag_state): Change type to use
1928 (struct align_flags_tuple): New.
1929 (struct align_flags): Likewise.
1930 (align_loops_log): Redefine macro to use new types.
1931 (align_loops_max_skip): Redefine macro to use new types.
1932 (align_jumps_log): Redefine macro to use new types.
1933 (align_jumps_max_skip): Redefine macro to use new types.
1934 (align_labels_log): Redefine macro to use new types.
1935 (align_labels_max_skip): Redefine macro to use new types.
1936 (align_functions_log): Redefine macro to use new types.
1937 (align_loops): Redefine macro to use new types.
1938 (align_jumps): Redefine macro to use new types.
1939 (align_labels): Redefine macro to use new types.
1940 (align_functions): Redefine macro to use new types.
1941 (align_functions_max_skip): Redefine macro to use new types.
1942 (align_loops_value): New macro.
1943 (align_jumps_value): New macro.
1944 (align_labels_value): New macro.
1945 (align_functions_value): New macro.
1946 * function.c (invoke_set_current_function_hook): Propagate
1947 alignment values from flags to global variables default in
1949 * ipa-icf.c (sem_function::equals_wpa): Use
1950 cl_optimization_option_eq instead of memcmp.
1951 * lto-streamer.h (cl_optimization_stream_out): Support streaming
1953 (cl_optimization_stream_in): Likewise.
1954 * optc-save-gen.awk: Support strings in cl_optimization.
1955 * opth-gen.awk: Likewise.
1956 * opts.c (finish_options): Remove error checking of invalid
1958 (MAX_CODE_ALIGN): Remove.
1959 (MAX_CODE_ALIGN_VALUE): Likewise.
1960 (parse_and_check_align_values): New function.
1961 (check_alignment_argument): Likewise.
1962 (common_handle_option): Use check_alignment_argument.
1963 * opts.h (parse_and_check_align_values): Declare.
1964 * toplev.c (init_alignments): Remove.
1965 (read_log_maxskip): New.
1966 (parse_N_M): Likewise.
1967 (parse_alignment_opts): Likewise.
1968 (backend_init_target): Remove usage of init_alignments.
1969 * toplev.h (parse_alignment_opts): Declare.
1970 * tree-streamer-in.c (streamer_read_tree_bitfields): Add new
1972 * tree-streamer-out.c (streamer_write_tree_bitfields): Likewise.
1973 * tree.c (cl_option_hasher::equal): New.
1974 * varasm.c: Use new global macros.
1976 2018-07-04 Denys Vlasenko <dvlasenk@redhat.com>
1978 * config/i386/dragonfly.h: (ASM_OUTPUT_MAX_SKIP_ALIGN):
1979 Use a simpler align directive also if MAXSKIP = ALIGN-1.
1980 * config/i386/gas.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
1981 * config/i386/lynx.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
1982 * config/i386/netbsd-elf.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
1983 * config/i386/i386.h (ASM_OUTPUT_MAX_SKIP_PAD): Likewise.
1984 * config/i386/freebsd.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Remove "If N
1985 is large, do at least 8 byte alignment" code. Add SUBALIGN_LOG
1986 define. Use a simpler align directive also if MAXSKIP = ALIGN-1.
1987 * config/i386/gnu-user.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
1988 * config/i386/iamcu.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
1989 * config/i386/openbsdelf.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
1990 * config/i386/x86-64.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
1992 2018-07-04 Martin Liska <mliska@suse.cz>
1993 Jonathan Wakely <jwakely@redhat.com>
1995 * coverage.c: Use correct type.
1996 * doc/invoke.texi: Language correction.
1998 2018-07-03 H.J. Lu <hongjiu.lu@intel.com>
2001 * config/i386/i386.c (rest_of_insert_endbranch): Also generate
2002 ENDBRANCH for non-tail call which may return via indirect branch.
2003 * doc/extend.texi: Document indirect_return attribute.
2005 2018-07-03 Martin Sebor <msebor@redhat.com>
2007 PR tree-optimization/86274
2008 * gimple-ssa-sprintf.c (fmtresult::type_max_digits): Verify
2010 (format_floating): Correct handling of infinities and NaNs.
2012 2018-07-03 Martin Sebor <msebor@redhat.com>
2014 * print-tree.c (print_real_cst): New function.
2015 (print_node_brief): Call it.
2016 (print_node): Ditto.
2018 2018-07-03 Jeff Law <law@redhat.com>
2020 * config/h8300/h8300.md (logical<mode>3_sn, logical<mode>3): Merge
2021 into a single pattern.
2023 * config/h8300/h8300.md (ors code_iterator): New.
2024 (bsetqi_msx, bnotqi_msx patterns and splitters): Consolidate into
2025 a single pattern and single splitter.
2026 (bsethi_msx, bnothi_msx patterns): Consolidate into a single pattern.
2027 (iorqi3_1, xorqi3_1): Likewise.
2028 (iorqi3, xorqi3 expanders): Similarly.
2030 * config/h8300/h8300.md (movmd_internal_normal): Consolidated with
2031 (movmd_internal) into a single pattern using the P mode iterator.
2032 (movmd splitters): Similarly.
2033 (stpcpy_internal_normal, stpcpy_internal): Similarly for thes patterns.
2034 (movsd splitters): Similarly.
2036 * config/h8300/h8300.c (h8300_insn_length_from_table): Consolidate
2037 ADDB, ADDW and ADDL into a single ADD attribute which selects the
2038 right table based on the size of the operand.
2039 * config/h8300/h8300.md (length_table): Corresponding changes. All
2040 references to "addb", "addw" and "addl" changed to "add".
2041 (btst patterns): Merge two variants into a single pattern.
2042 (tstqi, tsthi): Likewise.
2043 (addhi3_incdec, addsi3_incdec): Likewise.
2044 (subhi3_h8300hs, subsi3_h8300hs): Likewise.
2045 (mulhi3, mulsi3): Likewise.
2046 (udivhi3, udivsi3): Likewise.
2047 (divhi3, divsi3): Likewise.
2048 (andorqi3, andorhi3, andorsi3): Likewise.
2050 2018-07-03 Uros Bizjak <ubizjak@gmail.com>
2053 * config/i386/sse.md (uavg<mode>3_ceil): New expander.
2054 (<sse2_avx2>_uavg<mode>3<mask_name>): Simplify expander.
2056 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
2058 PR tree-optimization/85694
2059 * config/aarch64/iterators.md (HADD, RHADD): New int iterators.
2060 (u): Handle UNSPEC_SHADD, UNSPEC_UHADD, UNSPEC_SRHADD and
2062 * config/aarch64/aarch64-simd.md (<u>avg<mode>3_floor)
2063 (<u>avg<mode>3_ceil): New patterns.
2065 2018-07-03 David Malcolm <dmalcolm@redhat.com>
2067 * gcc.dg/vect/slp-perm-1.c: Remove "note: " prefix from
2068 scan-tree-dump directive.
2069 * gcc.dg/vect/slp-perm-2.c: Likewise.
2070 * gcc.dg/vect/slp-perm-3.c: Likewise.
2071 * gcc.dg/vect/slp-perm-5.c: Likewise.
2072 * gcc.dg/vect/slp-perm-6.c: Likewise.
2073 * gcc.dg/vect/slp-perm-7.c: Likewise.
2074 * gcc.dg/vect/slp-perm-8.c: Likewise.
2076 2018-07-03 Marek Polacek <polacek@redhat.com>
2079 * gimple-fold.c (size_must_be_zero_p): Check the type of the size.
2081 2018-07-03 Richard Biener <rguenther@suse.de>
2084 * tree-ssa-structalias.c (find_func_clobbers): Properly
2085 handle indirect calls.
2087 2018-07-03 Jeff Law <law@redhat.com>
2089 * config/h8300/h8300.md (HSI, QHSI, QHSIF): New mode iterators.
2090 (shifts): New code iterator.
2091 (movqi, movhi, movsi, movsf expanders): Consolidate into a single
2092 expander. Fix HImode handling on H8/SX.
2093 (addqi3, addhi3, addsi3 expanders): Consolidate into a single expander.
2094 (subqi3, subhi3, subsi3 expanders): Likewise.
2095 (andqi3, andhi3, andsi3 expanders): Likewise.
2096 (iorqi3, iorhi3, iorsi3 expanders): Likewise.
2097 (xorqi3, xorhi3, xorsi3 expanders): Likewise.
2098 (negqi2, neghi2, negsi2, negsf2 expanders): Likewise.
2099 (one_cmplqi2, one_cmplhi2, one_cmplsi2): Likewise.
2100 (zero_extendqihi2, zero_extendqisi2): Likewise.
2101 (extendqihi2, extendqisi2): Likewise.
2102 (rotlqi3, rotlhi3, rotlsi3): Likewise.
2103 (neghi2_h8300, negsi2_h8300): Likewise for these patterns.
2104 (rotlqi3_1, rotlhi3_1): Likewise.
2105 (logicalhi3_sn, logicalsi3_sn): Likewise.
2106 (logicalhi3, logicalsi3): Likewise.
2108 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
2110 * tree-vect-patterns.c (vect_recog_rotate_pattern)
2111 (vect_recog_vector_vector_shift_pattern, vect_recog_divmod_pattern)
2112 (vect_recog_mixed_size_cond_pattern, adjust_bool_pattern_cast)
2113 (adjust_bool_pattern, vect_recog_bool_pattern): Pass the vector
2114 type to append_pattern_def_seq instead of creating a stmt_vec_info
2116 (build_mask_conversion): Likewise. Remove vinfo argument.
2117 (vect_add_conversion_to_patterm): Likewise, renaming to...
2118 (vect_add_conversion_to_pattern): ...this.
2119 (vect_recog_mask_conversion_pattern): Update call to
2120 build_mask_conversion. Pass the vector type to
2121 append_pattern_def_seq here too.
2122 (vect_recog_gather_scatter_pattern): Update call to
2123 vect_add_conversion_to_pattern.
2125 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
2127 * tree-vect-patterns.c (new_pattern_def_seq): Delete.
2128 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern)
2129 (vect_recog_widen_op_pattern, vect_recog_over_widening_pattern)
2130 (vect_recog_rotate_pattern, vect_synth_mult_by_constant): Don't set
2131 STMT_VINFO_PATTERN_DEF_SEQ to null here.
2132 (vect_recog_pow_pattern, vect_recog_vector_vector_shift_pattern)
2133 (vect_recog_mixed_size_cond_pattern, vect_recog_bool_pattern): Use
2134 append_pattern_def_seq instead of new_pattern_def_seq.
2135 (vect_recog_divmod_pattern): Do both of the above.
2136 (vect_pattern_recog_1): Assert that STMT_VINO_PATTERN_DEF_SEQ
2139 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
2141 * tree-vect-patterns.c (vect_recog_dot_prod_pattern):
2142 (vect_recog_sad_pattern, vect_recog_widen_op_pattern)
2143 (vect_recog_widen_mult_pattern, vect_recog_pow_pattern):
2144 (vect_recog_widen_sum_pattern, vect_recog_over_widening_pattern)
2145 (vect_recog_average_pattern, vect_recog_cast_forwprop_pattern)
2146 (vect_recog_widen_shift_pattern, vect_recog_rotate_pattern)
2147 (vect_recog_vector_vector_shift_pattern, vect_synth_mult_by_constant)
2148 (vect_recog_mult_pattern, vect_recog_divmod_pattern)
2149 (vect_recog_mixed_size_cond_pattern, vect_recog_bool_pattern)
2150 (vect_recog_mask_conversion_pattern): Replace vec<gimple *>
2151 parameter with a single stmt_vec_info.
2152 (vect_recog_func_ptr): Likewise.
2153 (vect_recog_gather_scatter_pattern): Likewise, folding in...
2154 (vect_try_gather_scatter_pattern): ...this.
2155 (vect_pattern_recog_1): Remove stmts_to_replace and just pass
2156 the stmt_vec_info of the statement to be matched. Don't clear
2157 STMT_VINFO_RELATED_STMT.
2158 (vect_pattern_recog): Update call accordingly.
2160 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
2162 PR tree-optimization/85694
2163 * doc/md.texi (avgM3_floor, uavgM3_floor, avgM3_ceil)
2164 (uavgM3_ceil): Document new optabs.
2165 * doc/sourcebuild.texi (vect_avg_qi): Document new target selector.
2166 * internal-fn.def (IFN_AVG_FLOOR, IFN_AVG_CEIL): New internal
2168 * optabs.def (savg_floor_optab, uavg_floor_optab, savg_ceil_optab)
2169 (savg_ceil_optab): New optabs.
2170 * tree-vect-patterns.c (vect_recog_average_pattern): New function.
2171 (vect_vect_recog_func_ptrs): Add it.
2172 * tree-vect-stmts.c (vectorizable_call): Get the type of the zero
2173 constant directly from the associated lhs.
2175 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
2177 * tree-vect-patterns.c (vect_split_statement): New function.
2178 (vect_convert_input): Use it to try to split an existing cast.
2180 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
2182 * poly-int.h (print_hex): New function.
2183 * dumpfile.h (dump_dec, dump_hex): Declare.
2184 * dumpfile.c (dump_dec, dump_hex): New poly_wide_int functions.
2185 * tree-vectorizer.h (_stmt_vec_info): Add min_output_precision,
2186 min_input_precision, operation_precision and operation_sign.
2187 * tree-vect-patterns.c (vect_get_range_info): New function.
2188 (vect_same_loop_or_bb_p, vect_single_imm_use)
2189 (vect_operation_fits_smaller_type): Delete.
2190 (vect_look_through_possible_promotion): Add an optional
2191 single_use_p parameter.
2192 (vect_recog_over_widening_pattern): Rewrite to use new
2193 stmt_vec_info infomration. Handle one operation at a time.
2194 (vect_recog_cast_forwprop_pattern, vect_narrowable_type_p)
2195 (vect_truncatable_operation_p, vect_set_operation_type)
2196 (vect_set_min_input_precision): New functions.
2197 (vect_determine_min_output_precision_1): Likewise.
2198 (vect_determine_min_output_precision): Likewise.
2199 (vect_determine_precisions_from_range): Likewise.
2200 (vect_determine_precisions_from_users): Likewise.
2201 (vect_determine_stmt_precisions, vect_determine_precisions): Likewise.
2202 (vect_vect_recog_func_ptrs): Put over_widening first.
2204 (vect_pattern_recog): Call vect_determine_precisions.
2206 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
2208 * tree-vect-patterns.c (vect_mark_pattern_stmts): Remove pattern
2209 statements that have been replaced by further pattern statements.
2210 (vect_pattern_recog_1): Clear STMT_VINFO_PATTERN_DEF_SEQ on failure.
2212 2018-07-03 Richard Biener <rguenther@suse.de>
2214 * tree-vect-stmts.c (vect_is_simple_use): Consolidate dumping,
2215 always set *dt. Dump vectype in vectype overload.
2216 * dumpfile.h (dump_gimple_expr): New function.
2217 (dump_gimple_expr_loc): Likewise.
2218 * dumpfile.c (dump_gimple_expr): New function.
2219 (dump_gimple_expr_loc): Likewise.
2221 2018-07-02 Jeff Law <law@redhat.com>
2223 * config/h8300/h8300.md (movqi_h8300, movqi_h8300hs): Consolidate
2224 the H8/300, H8/300H and H8/S variants into a single pattern.
2225 (movhi_h8300, movqi_h8300hs): Similarly.
2226 (pushqi_h8300hs, pushhi_h8300hs): Consolidate into a single pattern.
2227 (QHI mode iterator): New.
2229 * config/h8300/h8300.md: Remove trailing whitespace.
2231 2018-07-02 Jim Wilson <jimw@sifive.com>
2233 * config/riscv/riscv.c (riscv_expand_epilogue): Use emit_jump_insn
2234 instead of emit_insn for interrupt returns.
2235 * config/riscv/riscv.md (riscv_met): Add (return) to rtl.
2236 (riscv_sret, riscv_uret): Likewise.
2238 2018-07-02 David Malcolm <dmalcolm@redhat.com>
2240 * pretty-print.c (selftest::test_pp_format): Move save and restore
2241 of quotes to class auto_fix_quotes, and add an instance.
2242 * selftest.c: Include "intl.h".
2243 (selftest::auto_fix_quotes::auto_fix_quotes): New ctor.
2244 (selftest::auto_fix_quotes::~auto_fix_quotes): New dtor.
2245 * selftest.h (selftest::auto_fix_quotes): New class.
2247 2018-07-02 Richard Henderson <richard.henderson@linaro.org>
2249 * config/aarch64/aarch64-protos.h, config/aarch64/aarch64.c
2250 (aarch64_sve_prepare_conditional_op): Remove.
2251 * config/aarch64/aarch64-sve.md (cond_<SVE_INT_BINARY><SVE_I>):
2252 Allow aarch64_simd_reg_or_zero as select operand; remove
2253 the aarch64_sve_prepare_conditional_op call.
2254 (cond_<SVE_INT_BINARY_SD><SVE_SDI>): Likewise.
2255 (cond_<SVE_COND_FP_BINARY><SVE_F>): Likewise.
2256 (*cond_<SVE_INT_BINARY><SVE_I>_z): New pattern.
2257 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_z): New pattern.
2258 (*cond_<SVE_COND_FP_BINARY><SVE_F>_z): New pattern.
2259 (*cond_<SVE_INT_BINARY><SVE_I>_any): New pattern.
2260 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_any): New pattern.
2261 (*cond_<SVE_COND_FP_BINARY><SVE_F>_any): New pattern
2262 and a splitters to match all of the *_any patterns.
2263 * config/aarch64/predicates.md (aarch64_sve_any_binary_operator): New.
2265 * config/aarch64/iterators.md (SVE_INT_BINARY_REV): Remove.
2266 (SVE_COND_FP_BINARY_REV): Remove.
2267 (sve_int_op_rev, sve_fp_op_rev): New.
2268 * config/aarch64/aarch64-sve.md (*cond_<SVE_INT_BINARY><SVE_I>_0): New.
2269 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_0): New.
2270 (*cond_<SVE_COND_FP_BINARY><SVE_F>_0): New.
2271 (*cond_<SVE_INT_BINARY><SVE_I>_2): Rename, add movprfx alternative.
2272 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_2): Similarly.
2273 (*cond_<SVE_COND_FP_BINARY><SVE_F>_2): Similarly.
2274 (*cond_<SVE_INT_BINARY><SVE_I>_3): Similarly; use sve_int_op_rev.
2275 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_3): Similarly.
2276 (*cond_<SVE_COND_FP_BINARY><SVE_F>_3): Similarly; use sve_fp_op_rev.
2278 * config/aarch64/aarch64-sve.md (cond_<SVE_COND_FP_BINARY><SVE_F>):
2279 Remove match_dup 1 from the inner unspec.
2280 (*cond_<SVE_COND_FP_BINARY><SVE_F>): Likewise.
2282 * config/aarch64/aarch64.md (movprfx): New attr.
2283 (length): Default movprfx to 8.
2284 * config/aarch64/aarch64-sve.md (*mul<SVE_I>3): Add movprfx alt.
2285 (*madd<SVE_I>, *msub<SVE_I): Likewise.
2286 (*<su>mul<SVE_I>3_highpart): Likewise.
2287 (*<SVE_INT_BINARY_SD><SVE_SDI>3): Likewise.
2288 (*v<ASHIFT><SVE_I>3): Likewise.
2289 (*<su><MAXMIN><SVE_I>3): Likewise.
2290 (*<su><MAXMIN><SVE_F>3): Likewise.
2291 (*fma<SVE_F>4, *fnma<SVE_F>4): Likewise.
2292 (*fms<SVE_F>4, *fnms<SVE_F>4): Likewise.
2293 (*div<SVE_F>4): Likewise.
2295 2018-07-02 Richard Sandiford <richard.sandiford@arm.com>
2297 * tree-vect-patterns.c (vect_recog_widen_shift_pattern): Fix typo
2300 2018-07-02 Richard Biener <rguenther@suse.de>
2302 PR tree-optimization/86363
2303 * tree-ssa-sccvn.c (vn_reference_lookup_3): Check the
2304 memset argument refers to a non-variable address.
2306 2018-07-02 Aldy Hernandez <aldyh@redhat.com>
2308 * tree-vrp.c (extract_range_from_binary_expr_1): Abstract a lot of the
2309 {PLUS,MINUS}_EXPR code to...
2310 (adjust_symbolic_bound): ...here,
2311 (combine_bound): ...here,
2312 (set_value_range_with_overflow): ...and here.
2314 2018-07-02 Aldy Hernandez <aldyh@redhat.com>
2316 * tree-vrp.c (extract_range_from_unary_expr): Abstract ABS_EXPR
2318 (extract_range_from_abs_expr): ...here.
2320 2018-07-02 Eric Botcazou <ebotcazou@adacore.com>
2322 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Do not overrule
2323 -fno-omit-frame-pointer when not optimizing.
2325 2018-07-02 Martin Liska <mliska@suse.cz>
2328 * ipa-pure-const.c (malloc_candidate_p): Revert usage of ::get.
2329 (propagate_nothrow): Likewise.
2331 2018-07-02 Martin Liska <mliska@suse.cz>
2334 * ipa-inline.c (early_inliner): Revert wrongly added ::get call.
2336 2018-07-02 David Malcolm <dmalcolm@redhat.com>
2338 * dumpfile.c (dump_generic_expr_loc): Undo removal of this
2339 function in r262149, changing "loc" param from source_location to
2340 const dump_location_t &.
2341 * dumpfile.h (dump_generic_expr_loc): Undo removal of this
2342 declaration, as above.
2344 2018-07-01 Paul Koning <ni1d@arrl.net>
2346 * common/config/pdp11/pdp11-common.c (pdp11_handle_option): Handle
2347 -munit-asm, -mgnu-asm, -mdec-asm.
2348 * config/pdp11/pdp11-protos.h (pdp11_gen_int_label): New.
2349 (pdp11_output_labelref): New.
2350 (pdp11_output_def): New.
2351 (pdp11_output_addr_vec_elt): New.
2352 * config/pdp11/pdp11.c: Use tab between opcode and operands. Use
2353 %# and %@ format codes.
2354 (pdp11_option_override): New.
2355 (TARGET_ASM_FILE_START_FILE_DIRECTIVE): Define.
2356 (pdp11_output_ident): New.
2357 (pdp11_asm_named_section): New.
2358 (pdp11_asm_init_sections): New.
2359 (pdp11_file_start): New.
2360 (pdp11_file_end): New.
2361 (output_ascii): Use .ascii/.asciz for -mdec-asm.
2362 (pdp11_asm_print_operand): Update %# and %$ for -mdec-asm. Add
2363 %o, like %c but octal.
2364 (pdp11_option_override): New.
2365 * config/pdp11/pdp11.h (TEXT_SECTION_ASM_OP): Update for
2367 (DATA_SECTION_ASM_OP): Ditto.
2368 (READONLY_DATA_SECTION_ASM_OP): New.
2369 (IS_ASM_LOGICAL_LINE_SEPARATOR): New.
2370 (ASM_GENERATE_INTERNAL_LABEL): Use new function.
2371 (ASM_OUTPUT_LABELREF): Ditto.
2372 (ASM_OUTPUT_DEF): Ditto.
2373 (ASM_OUTPUT_EXTERNAL): New.
2374 (ASM_OUTPUT_SOURCE_FILENAME): New.
2375 (ASM_OUTPUT_ADDR_VEC_ELT): Use new function.
2376 (ASM_OUTPUT_SKIP): Update for -mdec-asm.
2377 * config/pdp11/pdp11.md: Use tab between opcode and operands. Use
2378 %# and %@ format codes.
2379 * config/pdp11/pdp11.opt (mgnu-asm): New.
2380 (mdec-asm): Conflicts with -mgnu-asm and -munix-asm.
2381 (munix-asm): Conflicts with -mdec-asm and -mgnu-asm.
2382 * doc/invoke.txt (PDP-11 Options): Add -mgnu-asm.
2384 2018-07-01 Aldy Hernandez <aldyh@redhat.com>
2386 * tree-ssa-threadupdate.c (mark_threaded_blocks): Avoid
2387 dereferencing path[] beyond its length.
2389 (debug_all_paths): New.
2390 (rewire_first_differing_edge): New.
2391 (adjust_paths_after_duplication): New.
2392 (duplicate_thread_path): Call adjust_paths_after_duplication.
2394 (thread_through_all_blocks): Add new argument to
2395 duplicate_thread_path.
2397 2018-06-30 Jim Wilson <jimw@sifive.com>
2399 * config/riscv/predicates.md (p2m1_shift_operand): New.
2400 (high_mask_shift_operand): New.
2401 * config/riscv/riscv.md (lshrsi3_zero_extend_3+1): New combiner
2402 pattern using p2m1_shift_operand.
2403 (lshsi3_zero_extend_3+2): New combiner pattern using
2404 high_mask_shift_operand.
2406 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
2408 * tree-vect-patterns.c (vect_get_external_def_edge): New function,
2410 (vect_recog_rotate_pattern): ...here.
2411 (vect_convert_input): Try to insert casts of invariants in the
2413 * tree-vect-loop-manip.c (vect_loop_versioning): Don't require the
2414 preheader to be empty.
2416 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
2418 * tree-vect-patterns.c (append_pattern_def_seq): Take an optional
2419 vector type. If given, install it in the new statement's
2421 (vect_element_precision): New function.
2422 (vect_unpromoted_value): New struct.
2423 (vect_unpromoted_value::vect_unpromoted_value): New function.
2424 (vect_unpromoted_value::set_op): Likewise.
2425 (vect_look_through_possible_promotion): Likewise.
2426 (vect_joust_widened_integer, vect_joust_widened_type): Likewise.
2427 (vect_widened_op_tree, vect_convert_input): Likewise.
2428 (vect_convert_inputs, vect_convert_output): Likewise.
2429 (vect_recog_dot_prod_pattern): Use vect_look_through_possible_promotion
2430 to handle the optional cast of the multiplication result and
2431 vect_widened_op_tree to detect the widened multiplication itself.
2432 Do not require the input and output of promotion casts to have
2433 the same sign, but base the signedness of the operation on the
2434 input rather than the result. If the pattern includes two
2435 promotions, check that those promotions have the same sign.
2436 Do not restrict the MULT_EXPR handling to a double-width result;
2437 handle quadruple-width results and wider. Use vect_convert_inputs
2438 to convert the inputs to the common type.
2439 (vect_recog_sad_pattern): Use vect_look_through_possible_promotion
2440 to handle the optional cast of the ABS result. Also allow a sign
2441 change or a sign extension between the ABS and MINUS.
2442 Use vect_widened_op_tree to detect the widened subtraction and use
2443 vect_convert_inputs to convert the inputs to the common type.
2444 (vect_handle_widen_op_by_const): Delete.
2445 (vect_recog_widen_op_pattern): New function.
2446 (vect_recog_widen_mult_pattern): Use it.
2447 (vect_recog_widen_shift_pattern): Likewise.
2448 (vect_recog_widen_sum_pattern): Use
2449 vect_look_through_possible_promotion to handle the promoted
2452 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
2454 * gimple-iterator.c (gsi_for_stmt): Add a new overload that takes
2455 the containing gimple_seq *.
2456 * gimple-iterator.h (gsi_for_stmt): Declare it.
2457 * tree-vect-patterns.c (vect_recog_dot_prod_pattern)
2458 (vect_recog_sad_pattern, vect_recog_widen_sum_pattern)
2459 (vect_recog_widen_shift_pattern, vect_recog_rotate_pattern)
2460 (vect_recog_vector_vector_shift_pattern, vect_recog_divmod_pattern)
2461 (vect_recog_mask_conversion_pattern): Remove STMT_VINFO_IN_PATTERN_P
2463 (vect_init_pattern_stmt, vect_set_pattern_stmt): New functions,
2465 (vect_mark_pattern_stmts): ...here. Handle cases in which the
2466 statement being replaced is part of an existing pattern
2467 definition sequence, inserting the new pattern statements before
2469 (vect_pattern_recog_1): Don't return a bool. If the statement
2470 is already part of a pattern, instead apply pattern matching
2471 to the pattern definition statements. Don't clear the
2472 STMT_VINFO_RELATED_STMT if is_pattern_stmt_p.
2473 (vect_pattern_recog): Don't break after the first match;
2474 continue processing the pattern definition statements instead.
2475 Don't bail out for STMT_VINFO_IN_PATTERN_P here.
2477 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
2479 * tree-vect-patterns.c (vect_reassociating_reduction_p): New function.
2480 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern)
2481 (vect_recog_widen_sum_pattern): Use it.
2483 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
2485 * tree-vect-loop.c (vectorizable_reduction): Assert that the
2486 phi is not a pattern statement and has not been replaced by
2487 a pattern statement.
2488 * tree-vect-patterns.c (type_conversion_p): Don't check
2489 STMT_VINFO_IN_PATTERN_P.
2490 (vect_recog_vector_vector_shift_pattern): Likewise.
2491 (vect_recog_dot_prod_pattern): Expect vect_is_simple_use to return
2492 the pattern statement rather than the original statement; check
2493 directly for a WIDEN_MULT_EXPR here.
2494 * tree-vect-slp.c (vect_get_and_check_slp_defs): Expect
2495 vect_is_simple_use to return the pattern statement rather
2496 than the original statement; use is_pattern_stmt_p to check
2497 for such a pattern statement.
2498 * tree-vect-stmts.c (process_use): Expect vect_is_simple_use
2499 to return the pattern statement rather than the original statement;
2500 don't do the same transformation here.
2501 (vect_is_simple_use): If the defining statement has been replaced
2502 by a pattern statement, return the pattern statement instead.
2503 Remove the corresponding (local) transformation from the vectype
2506 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
2508 * tree-vectorizer.h (vect_is_simple_use): Move the gimple ** to the
2509 end and default to null.
2510 * tree-vect-loop.c (vect_create_epilog_for_reduction)
2511 (vectorizable_reduction): Update calls accordingly, dropping the
2512 gimple ** argument if the passed-back statement isn't needed.
2513 * tree-vect-patterns.c (vect_get_internal_def, type_conversion_p)
2514 (vect_recog_rotate_pattern): Likewise.
2515 (vect_recog_mask_conversion_pattern): Likewise.
2516 * tree-vect-slp.c (vect_get_and_check_slp_defs): Likewise.
2517 (vect_mask_constant_operand_p): Likewise.
2518 * tree-vect-stmts.c (is_simple_and_all_uses_invariant, process_use):
2519 (vect_model_simple_cost, vect_get_vec_def_for_operand): Likewise.
2520 (get_group_load_store_type, get_load_store_type): Likewise.
2521 (vect_check_load_store_mask, vect_check_store_rhs): Likewise.
2522 (vectorizable_call, vectorizable_simd_clone_call): Likewise.
2523 (vectorizable_conversion, vectorizable_assignment): Likewise.
2524 (vectorizable_shift, vectorizable_operation): Likewise.
2525 (vectorizable_store, vect_is_simple_cond): Likewise.
2526 (vectorizable_condition, vectorizable_comparison): Likewise.
2527 (get_same_sized_vectype, vect_get_mask_type_for_stmt): Likewise.
2528 (vect_is_simple_use): Rename the def_stmt argument to def_stmt_out
2529 and move it to the end. Cope with null def_stmt_outs.
2531 2018-06-30 Bernd Edlinger <bernd.edlinger@hotmail.de>
2533 * Makefile.in (FLAGS_TO_PASS): Add RANLIB_FOR_TARGET.
2535 2018-06-29 Jeff Law <law@redhat.com>
2537 * config/v850/v850.c (v850_legitimate_address_p): Handle large
2538 displacements for TARGET_V850E2V3 and newer.
2539 (TARGET_LRA_P): Remove. Defaults to LRA now.
2540 * config/v850/v850.md (sign23byte_load): Remove.
2541 (unsign23byte_load, sign23hword_load, unsign23hword_load): Likewise.
2542 (23word_load, 23byte_store, 23hword_store, 23word_store): Likewise.
2544 2018-06-29 Martin Liska <mliska@suse.cz>
2547 * coverage.c (coverage_init): Mangle full path name.
2548 * doc/invoke.texi: Document the change.
2549 * gcov-io.c (mangle_path): New.
2550 * gcov-io.h (mangle_path): Likewise.
2551 * gcov.c (mangle_name): Use mangle_path for path mangling.
2553 2018-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2555 * config/arm/arm.c (output_move_double): Don't allow STRD instructions
2556 if starting source register is not even.
2558 2018-06-29 Martin Liska <mliska@suse.cz>
2560 PR tree-optimization/86263
2561 * tree-switch-conversion.c (switch_decision_tree::try_switch_expansion):
2562 Make edge redirection.
2564 2018-06-29 David Malcolm <dmalcolm@redhat.com>
2566 * dumpfile.c (dump_loc): Add indentation based on scope depth.
2567 (dump_scope_depth): New variable.
2568 (get_dump_scope_depth): New function.
2569 (dump_begin_scope): New function.
2570 (dump_end_scope): New function.
2571 * dumpfile.h (get_dump_scope_depth): New declaration.
2572 (dump_begin_scope): New declaration.
2573 (dump_end_scope): New declaration.
2574 (class auto_dump_scope): New class.
2575 (AUTO_DUMP_SCOPE): New macro.
2576 * tree-vectorizer.h (DUMP_VECT_SCOPE): Reimplement in terms of
2579 2018-06-29 Richard Biener <rguenther@suse.de>
2581 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Assert
2582 compute_all_dependences succeeds.
2583 * tree-vect-loop.c (vect_get_datarefs_in_loop): Fail early if we
2584 exceed --param loop-max-datarefs-for-datadeps.
2586 2018-06-29 Jakub Jelinek <jakub@redhat.com>
2588 * config/rs6000/t-rs6000: Append rs6000-modes.h to TM_H.
2590 2018-06-28 Uros Bizjak <ubizjak@gmail.com>
2593 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): Use
2594 alternative 0 in preferred_for_speed attribute.
2596 2018-06-28 Paul Koning <ni1d@arrl.net>
2598 * config/pdp11/pdp11-protos.h (pdp11_shift_length): New function.
2599 * config/pdp11/pdp11.c (pdp11_shift_length): New function.
2600 * config/pdp11/pdp11.h (ADJUST_INSN_LENGTH): Remove.
2601 * config/pdp11/pdp11.md: Correct "length" attribute calculation
2602 for shift insn patterns.
2604 2018-06-28 David Malcolm <dmalcolm@redhat.com>
2606 * cgraph.c (cgraph_node::get_body): Replace assignments to
2607 "dump_file" with calls to set_dump_file.
2608 * dumpfile.c (alt_dump_file): Make static, and group with...
2609 (alt_flags): ...this definition.
2610 (dumps_are_enabled): New variable.
2611 (refresh_dumps_are_enabled): New function.
2612 (set_dump_file): New function.
2613 (set_alt_dump_file): New function.
2614 (gcc::dump_manager::dump_start): Replace assignments to
2615 "dump_file" and "alt_dump_file" with calls to set_dump_file and
2617 (gcc::dump_manager::dump_finish): Likewise.
2618 * dumpfile.h (alt_dump_file): Delete decl.
2619 (dumps_are_enabled): New variable decl.
2620 (set_dump_file): New function decl.
2621 (dump_enabled_p): Rewrite in terms of new "dumps_are_enabled"
2623 * tree-nested.c (lower_nested_functions): Replace assignments to
2624 "dump_file" with calls to set_dump_file.
2626 2018-06-28 Eric Botcazou <ebotcazou@adacore.com>
2628 * tree-cfg.c (verify_gimple_in_cfg): Call verify_location on the
2629 goto_locus of each outgoing edge of each basic block.
2631 2018-06-28 Richard Biener <rguenther@suse.de>
2633 * dwarf2out.c (decl_scope_table): Remove.
2634 (push_decl_scope): Likewise.
2635 (pop_decl_scope): Likewise.
2636 (gen_type_die_for_member): Do not call push/pop_decl_scope.
2637 (gen_struct_or_union_type_die): Likewise.
2638 (gen_tagged_type_die): Likewise.
2639 (dwarf2out_init): Do not initialize decl_scope_table.
2640 (dwarf2out_c_finalize): Do not free it.
2642 2018-06-28 Richard Biener <rguenther@suse.de>
2644 * dwarf2out.c (gen_subprogram_die): Use is_unit_die when
2645 deciding whether to not re-use a DIE.
2647 2018-06-28 Richard Biener <rguenther@suse.de>
2649 * dwarf2out.c (gen_subprogram_die): Always re-use DIEs with an
2650 DW_AT_abstract_origin attribute.
2652 2018-06-28 Martin Liska <mliska@suse.cz>
2654 * tree-switch-conversion.c (jump_table_cluster::can_be_handled):
2655 Use newly introduced constants.
2656 * tree-switch-conversion.h (struct jump_table_cluster):
2657 Define max_ratio_for_size and max_ratio_for_speed.
2659 2018-06-28 Martin Liska <mliska@suse.cz>
2661 * tree-switch-conversion.c (jump_table_cluster::find_jump_tables):
2662 Add new checking assert to catch invalid state.
2663 (jump_table_cluster::can_be_handled): Handle single case
2665 (jump_table_cluster::is_beneficial): Bail out for such case.
2666 (bit_test_cluster::find_bit_tests):
2667 Add new checking assert to catch invalid state.
2668 (bit_test_cluster::can_be_handled): Handle single case
2670 (bit_test_cluster::is_beneficial): Bail out for such case.
2671 (switch_decision_tree::analyze_switch_statement):
2674 2018-06-28 Martin Liska <mliska@suse.cz>
2676 * common.opt: Introduce -completion option.
2677 * gcc.c (driver_handle_option): Handle it.
2678 (driver::main): Print completions if completion
2680 * opt-suggestions.c (option_proposer::get_completions):
2682 (option_proposer::suggest_completion): Likewise.
2683 (option_proposer::find_param_completions): Likewise.
2684 (verify_autocompletions): Likewise.
2685 (test_completion_valid_options): Likewise.
2686 (test_completion_valid_params): Likewise.
2687 (in_completion_p): Likewise.
2688 (empty_completion_p): Likewise.
2689 (test_completion_partial_match): Likewise.
2690 (test_completion_garbage): Likewise.
2691 (opt_proposer_c_tests): Likewise.
2692 * opt-suggestions.h: Declare new functions.
2693 * opts.c (common_handle_option): Handle OPT__completion_.
2694 * selftest-run-tests.c (selftest::run_tests): Add
2695 opt_proposer_c_tests.
2696 * selftest.c (assert_str_startswith): New.
2697 * selftest.h (assert_str_startswith): Likewise.
2698 (opt_proposer_c_tests): New.
2699 (ASSERT_STR_STARTSWITH): Likewise.
2701 2018-06-28 Martin Liska <mliska@suse.cz>
2703 * Makefile.in: Add opt-suggestions.o.
2704 * gcc-main.c: Include opt-suggestions.h.
2705 * gcc.c (driver::driver): Likewise.
2706 (driver::~driver): Remove m_option_suggestions.
2707 (driver::build_option_suggestions): Moved to option_proposer.
2708 (driver::suggest_option): Likewise.
2709 (driver::handle_unrecognized_options): Use option_proposer.
2710 * gcc.h (class driver): Add new memver m_option_proposer.
2711 * opt-suggestions.c: New file.
2712 * opt-suggestions.h: New file.
2714 2018-06-28 Martin Liska <mliska@suse.cz>
2716 * vec.h (class auto_string_vec): New (moved from auto_argvec).
2717 (auto_string_vec::~auto_string_vec): Likewise.
2719 2018-06-28 Eric Botcazou <ebotcazou@adacore.com>
2721 * tree-inline.h (struct copy_body_data): Move remapping_type_depth and
2722 prevent_decl_creation_for_types fields up and add reset_location field.
2723 * tree-inline.c (remap_gimple_stmt): Force input_location on the new
2724 statement if id->reset_location is true.
2725 (copy_edges_for_bb): Do not set goto_locus on the new edges if
2726 id->reset_location is true.
2727 (copy_phis_for_bb): Force input_location on the arguments if
2728 id->reset_location is true.
2729 (expand_call_inline): Set id->reset_location if DECL_IGNORED_P
2730 is set on the function to be inlined.
2732 2018-06-27 Stephan Bergmann <sbergman@redhat.com>
2734 * doc/invoke.texi (Debugging Options): Add -gsplit-dwarf.
2736 2018-06-27 Dimitar Dimitrov <dimitar@dinux.eu>
2738 * lra-eliminations.c (update_reg_eliminate): Mark all spanning hard
2739 registers for Pmode.
2740 * lra-lives.c (check_pseudos_live_through_calls): Mark all spanning
2741 hard registers for the clobbered pseudo.
2743 2018-06-27 Paul Koning <ni1d@arrl.net>
2745 * common/config/pdp11/pdp11-common.c (pdp11_handle_option): Handle
2746 mutually exclusive options.
2747 * config/pdp11/constraints.md (h): New constraint.
2748 (O): Update definition to match shift code generation.
2749 (D): New constraint.
2750 * config/pdp11/pdp11-modes.def (CCNZ): Define mode.
2752 * config/pdp11/pdp11-protos.h (int_no_side_effect_operand): New
2754 (output_jump): Change arguments.
2755 (pdp11_fixed_cc_regs): New function.
2756 (pdp11_cc_mode): Ditto.
2757 (pdp11_expand_shift): Ditto.
2758 (pdp11_assemble_shift): Ditto.
2759 (pdp11_small_shift): Ditto.
2760 (pdp11_branch_cost): Remove.
2761 * config/pdp11/pdp11.c (pdp11_assemble_integer): Remove comments
2763 (pdp11_register_move_cost): Update for CC registers.
2764 (pdp11_rtx_costs): Add case for LSHIFTRT.
2765 (pdp11_output_jump): Add CCNZ mode conditional branches.
2766 (notice_update_cc_on_set): Remove.
2767 (pdp11_cc_mode): New function.
2768 (simple_memory_operand): Correct pre/post decrement case.
2769 (no_side_effect_operand): New function.
2770 (pdp11_regno_reg_class): Add CC_REGS class.
2771 (pdp11_fixed_cc_regs): New function.
2772 (pdp11_small_shift): New function.
2773 (pdp11_expand_shift): New function to expand shift insns.
2774 (pdp11_assemble_shift): New function to output shifts.
2775 (pdp11_branch_cost): Remove.
2776 (pdp11_modes_tieable_p): Make QI/HI modes tieable.
2777 * config/pdp11/pdp11.h (SIZE_TYPE): Ensure 16-bit type.
2778 (WCHAR_TYPE): Ditto.
2779 (PTRDIFF_TYPE): Ditto.
2780 (ADJUST_INSN_LENGTH): New macro.
2781 (FIXED_REGISTERS): Add CC registers.
2782 (CALL_USED_REGISTERS): Ditto.
2784 (REG_CLASS_NAMES): Ditto.
2785 (REG_CLASS_CONTENTS): Ditto.
2786 (SELECT_CC_MODE): Use new function.
2787 (TARGET_FLAGS_REGNUM): New macro.
2788 (TARGET_FIXED_CONDITION_CODE_REGS): Ditto.
2789 (cc0_reg_rtx): Remove.
2790 (CC_STATUS_MDEP): Remove.
2791 (CC_STATUS_MDEFP_INIT): Remove.
2792 (CC_IN_FPU): Remove.
2793 (NOTICE_UPDATE_CC): Remove.
2794 (REGISTER_NAMES): Add CC registers.
2795 (BRANCH_COST): Change to constant 1.
2796 * config/pdp11/pdp11.md: Rewrite for CCmode condition code
2798 * config/pdp11/pdp11.opt (mbcopy): Remove.
2799 (mbcopy-builtin): Remove.
2800 (mbranch-cheap): Remove.
2801 (mbranch-expensive): Remove.
2802 * config/pdp11/predicates.md (expand_shift_operand): Update to
2803 match shift code generation.
2804 (ccnz_operator): New predicate.
2805 * doc/invoke.texi (PDP-11 Options): Remove deleted options
2806 -mbcopy, -mbcopy-builtin, -mbranch-cheap, -mbranch-expensive.
2807 Remove non-existent option -mabshi, -mno-abshi. Document mutually
2809 * doc/md.texi (PDP-11): Document new D and h constraints. Update
2810 description of O constraint.
2812 2018-06-27 Jeff Law <law@redhat.com>
2813 Austin Law <austinklaw@gmail.com>
2815 * config/v850/v850.md (addsi3_set_flags): New pattern.
2816 (subsi3_set_flags, negsi2_set_flags, andsi3_set_flags): Likewise.
2817 (iorsi3_set_flags, xorsi3_set_flags, one_cmplsi2_set_flags): Likewise.
2818 (zero_extendhisi2_v850_set_flags): Likewise.
2819 (zero_extendqisi2_v850_set_flags): Likewise.
2820 (ashlsi3_set_flags, ashlsi3_v850e2_set_flags): Likewise.
2821 (lshrsi3_set_flags, lshrsi3_v850e2_set_flags): Likewise.
2822 (ashrsi3_set_flags, ashrsi3_v850e2_set_flags): Likewise.
2824 * config/v850/v850-protos.h (notice_update_cc): Remove.
2825 * config/v850/v850.c (v850_compare_op0, v850_compare_op1): Remove.
2826 (v850_print_operand): Handle 'D' and "d".
2827 (v850_select_cc_mode): Remove ATTRIBUTE_UNUSED for last argument.
2828 Add handling of arithmetic/logical operations compared against zero.
2829 (v850_gen_float_compare): Remove ATTRIBUTE_UNUSED for last argument.
2830 Do not look at v850_compare_op, instead get mode from last argument.
2831 (v850_gen_compare): Remove
2832 (increment_stack): Use addsi3_clobber_flags to avoid splitting failure
2833 after reload for prologue insns.
2834 (expand_prologue): Account for CLOBBER of CC_REGNUM in various
2836 (construct_save_jarl): Likewise.
2837 (TARGET_FLAGS_REGNUM): Define.
2838 * config/v850/v850.h (v850_compare_op0, v850_compare_op1): Remove.
2839 (NOTICE_UPDATE_CC): Remove.
2840 * config/v850/v850.md (v850_tst1): Use (reg:CCZ CC_REGNUM) rather
2841 than cc0. Conditionalize on reload_completed.
2842 (cmpsi_insn, setfcc_insn): Likewise.
2843 (tst1 splitter): Turn into define_and_split which sets the flags
2845 (cstoresi4, cbranchsf4, cbranchdf4, cbranchsi4_insn): Likewise.
2846 (cbranchsi4, branch_normal, branch_invert): Do not expose cc0 here.
2847 (cstoresf4, cstoredf4): Clobber the flags.
2848 (cmpsi, cmpsf, cmpdf): Remove expanders.
2849 (setf_insn): Remove pattern.
2850 (addsi3): Turn into define_and_split which clobbers the flags after
2851 reload and a suitable pattern (addsi3_clobber_flags) for use after
2853 (subsi3, negsi2, andsi3, iorsi3, xorsi3, one_cmplsi2) Likewise.
2854 (ashlsi3, ashlsi3_v850e2, lshrsi3, lsh4si3_v850e2): Likewise.
2855 (ashrsi3, ashrsi3_v850e2): Likewise.
2856 (bins): Clobber the flags.
2857 (movsicc_normal_cc, movsicc_normal, movsicc_tst1): Likewise.
2858 (movsicc_tst1_revesed, sasf, swap and rotate patterns): Likewise.
2859 (fix_loop_counter, call_internal_short, call_internal_long): Likewise.
2860 (call_value_internal_short, call_value_internal_long): Likewise.
2861 (callt_save_interrupt, callt_return_interrupt): Likewise.
2862 (save_interrupt, return_interrupt): Likewise.
2863 (callt_save_all_interrupt, save_all_interrupt): Likewise.
2864 (_save_all_interrupt, callt_restore_all_interrupt): Likewise.
2865 (restore_all_interrupt, _restore_all_interrupt): Likewise.
2866 (All FP comparisons): Only allow after reload has completed.
2868 (divh, divhu): Tweak output template.
2869 (branch_z_normal, branch_z_invert): Remove
2870 (branch_nz_normal, branch_nz_invert): Likewise.
2871 (extendhisi_insn, extendqisi_insn): Do not clobber flags.
2873 * config/v850/v850-modes.def (CCZ, CCNZ): Add new modes.
2874 * config/v850/v850.c (notice_update_cc): Remove.
2875 * config/v850/v850.h (CC_OVERFLOW_UNUSABLE): Remove
2876 (CC_NO_CARRY): Likewise.
2877 (NOTICE_UPDATE_CC): Define to nothing.
2878 * config/v850/v850.md: Remove block comment on cc0 handling
2879 Remove "cc" attribute from all patterns. Remove cc_status handling
2880 from all patterns. Minor formatting fixes.
2882 2018-06-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2884 * config/aarch64/aarch64-cores.def (cortex-a76): New entry.
2885 (cortex-a76.cortex-a55): Likewise.
2886 * config/aarch64/aarch64-tune.md: Regenerate.
2887 * doc/invoke.texi (AArch64 Options): Document cortex-a76 and
2888 cortex-a76.cortex-a55.
2890 2018-06-27 Jeff Law <law@redhat.com>
2892 * config/v850/t-v850 (MULTILIB_OPTIONS): Remove 8byte-align.
2893 (MULTILIB_DIRNAMES): Similarly.
2895 2018-06-27 Eric Botcazou <ebotcazou@adacore.com>
2897 * gimple.h (gimple_return_retbnd): Delete.
2898 (gimple_return_set_retbnd): Likewise.
2899 * cgraphunit.c (cgraph_node::expand_thunk): Remove call to
2900 gimple_return_set_retbnd.
2901 * gimple-pretty-print.c (dump_gimple_return): Remove call to
2902 gimple_return_retbnd and adjust.
2903 * tree-inline.h (struct copy_body_data): Remove retbnd field.
2904 * tree-inline.c (remap_gimple_stmt): Remove handling of retbnd.
2905 Explicitly return NULL in a couple more cases. Move assertion
2906 on debug statements and remove unreachable code.
2907 (reset_debug_binding): Do not test id->retbnd.
2908 (expand_call_inline): Do not set it.
2910 2018-06-27 Rasmus Villemoes <rv@rasmusvillemoes.dk>
2912 * configure.ac: Add --disable-gcov option.
2913 * configure: Regenerate.
2914 * Makefile.in: Honour @enable_gcov@.
2915 * doc/install.texi: Document --disable-gcov.
2917 2018-06-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2919 * config/arm/arm-cpus.in (cortex-a76): New entry.
2920 (cortex-a76.cortex-a55): Likewise.
2921 * config/arm/arm-tables.opt: Regenerate.
2922 * config/arm/arm-tune.md: Likewise.
2923 * config/arm/driver-arm.c (arm_cpu_table): Add Cortex-A76 entry.
2924 * doc/invoke.texi (ARM Options): Document cortex-a76 and
2925 cortex-a76.cortex-a55.
2927 2018-06-27 Tamar Christina <tamar.christina@arm.com>
2930 * config/aarch64/aarch64.md (*movhf_aarch64): Add dup v0.4h pattern.
2932 2018-06-27 Siddhesh Poyarekar <siddhesh@sourceware.org>
2934 * config/aarch64/aarch64.h (CALL_USE_REGISTERS): Fix obsolete
2936 (EPILOGUE_USES): Likewise.
2938 2018-06-26 Eric Botcazou <ebotcazou@adacore.com>
2940 * tree-inline.c (remap_location): New function extracted from...
2941 (copy_edges_for_bb): Add ID parameter. Remap goto_locus.
2942 (copy_phis_for_bb): ...here. Call remap_location.
2943 (copy_cfg_body): Adjust call to copy_edges_for_bb.
2945 2018-06-26 Aaron Sawdey <acsawdey@linux.ibm.com>
2947 * config/rs6000/rs6000-string.c (expand_block_clear): Don't use
2948 unaligned vsx for 16B memset.
2950 2018-06-26 Segher Boessenkool <segher@kernel.crashing.org>
2953 * config/rs6000/rs6000.c (rs6000_init_builtins): Do not set
2954 ieee128_float_type_node to long_double_type_node unless
2955 TARGET_LONG_DOUBLE_128 is set.
2957 2018-06-26 David Malcolm <dmalcolm@redhat.com>
2959 * cfgloop.c (get_loop_location): Convert return type from
2960 location_t to dump_user_location_t, replacing INSN_LOCATION lookups
2961 by implicit construction from rtx_insn *, and using
2962 dump_user_location_t::from_function_decl for the fallback case.
2963 * cfgloop.h (get_loop_location): Convert return type from
2964 location_t to dump_user_location_t.
2965 * cgraphunit.c (walk_polymorphic_call_targets): Update call to
2966 dump_printf_loc to pass in a dump_location_t rather than a
2967 location_t, via the gimple stmt.
2968 * coverage.c (get_coverage_counts): Update calls to
2969 dump_printf_loc to pass in dump_location_t rather than a
2971 * doc/optinfo.texi (Dump types): Convert example of
2972 dump_printf_loc from taking "locus" to taking "insn". Update
2973 description of the "_loc" calls to cover dump_location_t.
2974 * dumpfile.c: Include "backend.h", "gimple.h", "rtl.h", and
2976 (dump_user_location_t::dump_user_location_t): New constructors,
2977 from gimple *stmt and rtx_insn *.
2978 (dump_user_location_t::from_function_decl): New function.
2979 (dump_loc): Make static.
2980 (dump_gimple_stmt_loc): Convert param "loc" from location_t to
2981 const dump_location_t &.
2982 (dump_generic_expr_loc): Delete.
2983 (dump_printf_loc): Convert param "loc" from location_t to
2984 const dump_location_t &.
2985 (selftest::test_impl_location): New function.
2986 (selftest::dumpfile_c_tests): New function.
2987 * dumpfile.h: Include "profile-count.h".
2988 (class dump_user_location_t): New class.
2989 (struct dump_impl_location_t): New struct.
2990 (class dump_location_t): New class.
2991 (dump_printf_loc): Convert 2nd param from source_location to
2992 const dump_location_t &.
2993 (dump_generic_expr_loc): Delete.
2994 (dump_gimple_stmt_loc): Convert 2nd param from source_location to
2995 const dump_location_t &.
2996 * gimple-fold.c (fold_gimple_assign): Update call to
2997 dump_printf_loc to pass in a dump_location_t rather than a
2998 location_t, via the gimple stmt.
2999 (gimple_fold_call): Likewise.
3000 * gimple-loop-interchange.cc
3001 (loop_cand::analyze_iloop_reduction_var): Update for change to
3002 check_reduction_path.
3003 (tree_loop_interchange::interchange): Update for change to
3005 * graphite-isl-ast-to-gimple.c (scop_to_isl_ast): Update for
3006 change in return-type of find_loop_location.
3007 (graphite_regenerate_ast_isl): Likewise.
3008 * graphite-optimize-isl.c (optimize_isl): Likewise.
3009 * graphite.c (graphite_transform_loops): Likewise.
3010 * ipa-devirt.c (ipa_devirt): Update call to dump_printf_loc to
3011 pass in a dump_location_t rather than a location_t, via the
3013 * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
3014 * ipa.c (walk_polymorphic_call_targets): Likewise.
3015 * loop-unroll.c (report_unroll): Convert "locus" param from
3016 location_t to dump_location_t.
3017 (decide_unrolling): Update for change to get_loop_location's
3019 * omp-grid.c (struct grid_prop): Convert field "target_loc" from
3020 location_t to dump_user_location_t.
3021 (grid_find_single_omp_among_assignments_1): Updates calls to
3022 dump_printf_loc to pass in a dump_location_t rather than a
3023 location_t, via the gimple stmt.
3024 (grid_parallel_clauses_gridifiable): Convert "tloc" from
3025 location_t to dump_location_t. Updates calls to dump_printf_loc
3026 to pass in a dump_location_t rather than a location_t, via the
3028 (grid_inner_loop_gridifiable_p): Likewise.
3029 (grid_dist_follows_simple_pattern): Likewise.
3030 (grid_gfor_follows_tiling_pattern): Likewise.
3031 (grid_target_follows_gridifiable_pattern): Likewise.
3032 (grid_attempt_target_gridification): Convert initialization
3033 of local "grid" from memset to zero-initialization; FIXME: does
3034 this require C++11? Update call to dump_printf_loc to pass in a
3035 optinfo_location rather than a location_t, via the gimple stmt.
3036 * profile.c (read_profile_edge_counts): Updates call to
3037 dump_printf_loc to pass in a dump_location_t rather than a
3039 (compute_branch_probabilities): Likewise.
3040 * selftest-run-tests.c (selftest::run_tests): Call
3042 * selftest.h (dumpfile_c_tests): New decl.
3043 * tree-loop-distribution.c (pass_loop_distribution::execute):
3044 Update for change in return type of find_loop_location.
3045 * tree-parloops.c (parallelize_loops): Likewise.
3046 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Convert
3047 "locus" from location_t to dump_user_location_t.
3048 (canonicalize_loop_induction_variables): Likewise.
3049 * tree-ssa-loop-ivopts.c (tree_ssa_iv_optimize_loop): Update
3050 for change in return type of find_loop_location.
3051 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update call
3052 to dump_printf_loc to pass in a dump_location_t rather than a
3053 location_t, via the stmt.
3054 * tree-ssa-sccvn.c (eliminate_dom_walker::before_dom_children):
3056 * tree-vect-loop-manip.c (find_loop_location): Convert return
3057 type from source_location to dump_user_location_t.
3058 (vect_do_peeling): Update for above change.
3059 (vect_loop_versioning): Update for change in type of
3061 * tree-vect-loop.c (check_reduction_path): Convert "loc" param
3062 from location_t to dump_user_location_t.
3063 (vect_estimate_min_profitable_iters): Update for change in type
3065 * tree-vect-slp.c (vect_print_slp_tree): Convert param "loc" from
3066 location_t to dump_location_t.
3067 (vect_slp_bb): Update for change in type of vect_location.
3068 * tree-vectorizer.c (vect_location): Convert from source_location
3069 to dump_user_location_t.
3070 (try_vectorize_loop_1): Update for change in vect_location's type.
3071 (vectorize_loops): Likewise.
3072 (increase_alignment): Likewise.
3073 * tree-vectorizer.h (vect_location): Convert from source_location
3074 to dump_user_location_t.
3075 (find_loop_location): Convert return type from source_location to
3076 dump_user_location_t.
3077 (check_reduction_path): Convert 1st param from location_t to
3078 dump_user_location_t.
3079 * value-prof.c (check_counter): Update call to dump_printf_loc to
3080 pass in a dump_user_location_t rather than a location_t; update
3081 call to error_at for change in type of "locus".
3082 (check_ic_target): Update call to dump_printf_loc to
3083 pass in a dump_user_location_t rather than a location_t, via the
3086 2018-06-26 Robin Dapp <rdapp@linux.vnet.ibm.com>
3088 * config/s390/s390.h (enum processor_flags): Do not use
3089 default tune parameter when -march was specified.
3091 2018-06-26 Jakub Jelinek <jakub@redhat.com>
3094 * config/i386/i386.md (setcc + movzbl to xor + setcc peephole2s):
3095 Check reg_overlap_mentioned_p in addition to reg_set_p with the same
3098 2018-06-26 Richard Biener <rguenther@suse.de>
3100 PR tree-optimization/86287
3102 * tree-vect-loop.c (vect_transform_loop_stmt): Fix read-after-free.
3103 (vect_analyze_loop): Initialize n_stmts.
3105 2018-06-26 Richard Biener <rguenther@suse.de>
3108 * fold-const.c (fold_convertible_p): Pointer extension
3111 2018-06-26 Alexandre Oliva <aoliva@redhat.com>
3114 * dwarf2out.c (loc_list_has_views): Adjust comments.
3115 (dw_loc_list): Split single cross-partition range with
3118 2018-06-25 Jeff Law <law@redhat.com>
3120 * common/config/v850/v850-common.c (TARGET_DEFAULT_TARGET_FLAGS): Turn
3121 on -mbig-switch by default.
3123 * config/v850/predicates.md (const_float_1_operand): Fix match_code
3125 (const_float_0_operand): Remove unused predicate.
3126 * config/v850/v850.md (define_constants): Remove UNSPEC_LOOP.
3127 (define_c_enum unspec): Add LOOP, RCP and RSQRT constants.
3128 (recipsf2): New expander. Original pattern now called
3130 (recipdf2, recipdf2_insn): Similarly.
3131 (rsqrtsf2, rsqrtsf2_insn): Similarly
3132 (rsqrtdf2, rsqrtdf2_insn): Similarly
3134 2018-06-26 Gerald Pfeifer <gerald@pfeifer.com>
3136 * ginclude/stddef.h: Remove an obsolete comment on FreeBSD 5.
3137 Simplify logic for FreeBSD (twice).
3139 2018-06-25 Martin Sebor <msebor@redhat.com>
3141 PR tree-optimization/86204
3142 * tree-ssa-strlen.c (handle_builtin_strlen): Avoid storing
3143 a strnlen result if it's less than the length of the string.
3145 2018-06-25 Martin Sebor <msebor@redhat.com>
3147 PR tree-optimization/85700
3148 * gimple-fold.c (gimple_fold_builtin_strncat): Adjust comment.
3149 * tree-ssa-strlen.c (is_strlen_related_p): Handle integer subtraction.
3150 (maybe_diag_stxncpy_trunc): Distinguish strncat from strncpy.
3152 2018-06-25 Martin Sebor <msebor@redhat.com>
3154 * doc/extend.texi (Zero-length arrays): Update and clarify.
3156 2018-06-25 Michael Meissner <meissner@linux.ibm.com>
3158 * config.gcc (powerpc64le*): Revert January 16th, 2018 patch that
3159 added IEEE/IBM long double multilib support on PowerPC little
3160 endian Linux systems.
3161 * config/rs6000/linux64.h (MULTILIB_DEFAULTS_IEEE): Likewise.
3162 (MULTILIB_DEFAULTS): Likewise.
3163 * config/rs6000/rs6000.c (rs6000_option_override_internal):
3165 * config/rs6000/rs6000.h (TARGET_IEEEQUAD_MULTILIB): Likewise.
3166 * config/rs6000/t-ldouble-linux64le-ibm: Delete, no longer used.
3167 * config/rs6000/t-ldouble-linux64le-ieee: Delete, no longer used.
3169 2018-06-25 Alexander Monakov <amonakov@ispras.ru>
3172 * sort.cc (REORDER_23): Avoid memcpy with same destination and source.
3173 (REORDER_45): Likewise.
3175 2018-06-25 Jeff Law <law@redhat.com>
3177 * config/v850/v850.md (divmodhi4): Make sure to sign extend the
3178 dividend to 32 bits. Adjust length.
3179 (udivmodhi4): Cleanup output template. Fix length.
3181 2018-06-25 Carl Love <cel@us.ibm.com>
3183 * config/rs6000/vsx.md: Change word selector to prefered location.
3185 2018-06-25 Richard Biener <rguenther@suse.de>
3187 PR tree-optimization/86304
3188 * tree-vectorizer.c (vectorize_loops): Walk over new possibly
3189 epilogue-if-converted loops as well.
3191 2018-06-25 Jan Hubicka <hubicka@ucw.cz>
3193 * lto-section-out.c (lto_begin_section): Do not print section
3194 name for noaddr and unnumbered dumps.
3196 2018-06-25 Richard Biener <rguenther@suse.de>
3198 * tree-vectorizer.h (struct vec_info_shared): New structure
3199 with parts split out from struct vec_info and loop_nest from
3200 struct _loop_vec_info.
3201 (struct vec_info): Adjust accordingly.
3202 (struct _loop_vec_info): Likewise.
3203 (LOOP_VINFO_LOOP_NEST): Adjust.
3204 (LOOP_VINFO_DATAREFS): Likewise.
3205 (LOOP_VINFO_DDRS): Likewise.
3206 (struct _bb_vec_info): Likewise.
3207 (BB_VINFO_DATAREFS): Likewise.
3208 (BB_VINFO_DDRS): Likewise.
3209 (struct _stmt_vec_info): Add dr_aux member.
3210 (DR_VECT_AUX): Adjust to refer to member of DR_STMTs vinfo.
3211 (DR_MISALIGNMENT_UNINITIALIZED): New.
3212 (set_dr_misalignment): Adjust.
3213 (dr_misalignment): Assert misalign isn't DR_MISALIGNMENT_UNINITIALIZED.
3214 (vect_analyze_loop): Adjust prototype.
3215 (vect_analyze_loop_form): Likewise.
3216 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences):
3217 Compute dependences lazily.
3218 (vect_record_base_alignments): Use shared datarefs/ddrs.
3219 (vect_verify_datarefs_alignment): Likewise.
3220 (vect_analyze_data_refs_alignment): Likewise.
3221 (vect_analyze_data_ref_accesses): Likewise.
3222 (vect_analyze_data_refs): Likewise.
3223 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Add
3224 constructor parameter for shared part.
3225 (vect_analyze_loop_form): Pass in shared part and adjust.
3226 (vect_analyze_loop_2): Pass in storage for the number of
3227 stmts. Move loop nest finding to the caller. Compute
3229 (vect_analyze_loop): Pass in shared part.
3230 (vect_transform_loop): Verify shared datarefs are unchanged.
3231 * tree-vect-slp.c (_bb_vec_info::_bb_vec_info): Add
3232 constructor parameter for shared part.
3233 (vect_slp_analyze_bb_1): Pass in shared part and adjust.
3234 (vect_slp_bb): Verify shared datarefs are unchanged before
3236 * tree-vect-stmts.c (ensure_base_align): Adjust for DR_AUX
3238 (new_stmt_vec_info): Initialize DR_AUX misalignment to
3239 DR_MISALIGNMENT_UNINITIALIZED.
3240 * tree-vectorizer.c (vec_info::vec_info): Add constructor
3241 parameter for shared part.
3242 (vec_info::~vec_info): Adjust.
3243 (vec_info_shared::vec_info_shared): New.
3244 (vec_info_shared::~vec_info_shared): Likewise.
3245 (vec_info_shared::save_datarefs): Likewise.
3246 (vec_info_shared::check_datarefs): Likewise.
3247 (try_vectorize_loop_1): Construct shared part live for analyses
3248 of a single loop for multiple vector sizes.
3249 * tree-parloops.c (gather_scalar_reductions): Adjust.
3251 2018-06-25 Richard Biener <rguenther@suse.de>
3253 * tree-vect-data-refs.c (vect_find_stmt_data_reference): Modify
3254 DR for SIMD lane accesses here and mark DR with (void *)-1 aux.
3255 (vect_analyze_data_refs): Remove similar code from here and
3256 simplify accordingly.
3258 2018-06-25 Richard Biener <rguenther@suse.de>
3260 * tree-vect-data-refs.c (vect_check_gather_scatter): Fail
3261 for reverse storage order accesses rather than asserting
3262 they cannot happen here.
3264 2018-06-25 Tom de Vries <tdevries@suse.de>
3267 * config/i386/i386.md (define_insn "*tls_global_dynamic_64_<mode>"):
3268 Use data16 instead of .byte for insn prefix.
3270 2018-06-25 Andreas Krebbel <krebbel@linux.ibm.com>
3273 * parser.c (make_char_string_pack): Pass this literal chars
3274 through cpp_interpret_string.
3275 (cp_parser_userdef_numeric_literal): Check the result of
3276 make_char_string_pack.
3278 2018-06-24 Maya Rashish <coypu@sdf.org>
3280 * ginclude/stddef.h: Simplify conditions around avoiding
3281 re-definition of __size_t.
3283 2018-06-22 Jan Hubicka <hubicka@ucw.cz>
3285 * lto-streamer-out.c (tree_is_indexable): Make LABEL_DECL nonindexable
3286 unless it is forced or nonlocal; assert that we stream no IMPORTED_DECL.
3288 2018-06-22 Maya Rashish <coypu@sdf.org>
3290 * doc/invoke.texi (mno-fancy-math-387): Update for changes
3291 made to OpenBSD and NetBSD through the years.
3293 2018-06-22 Kelvin Nilsen <kelvin@gcc.gnu.org>
3295 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Change
3296 behavior of vec_pack (vector double, vector double) to match
3297 behavior of vec_float2 (vector double, vector double).
3299 2018-06-22 Olivier Hainque <hainque@adacore.com>
3301 * gimplify.c (gimplify_function_tree): Prevent creation
3302 of a trampoline for the address of the current function
3303 passed to entry/exit instrumentation hooks.
3305 2018-06-22 Aaron Sawdey <acsawdey@linux.ibm.com>
3308 * config/rs6000/rs6000-string.c (expand_strn_compare): Handle -m32
3311 2018-06-22 Martin Liska <mliska@suse.cz>
3313 PR tree-optimization/86263
3314 * tree-switch-conversion.c (jump_table_cluster::find_jump_tables):
3315 Bail out if is_enabled is false.
3316 * tree-switch-conversion.h (jump_table_cluster::is_enabled):
3318 (jump_table_cluster::is_enabled): New function.
3320 2018-06-22 Jan Hubicka <hubicka@ucw.cz>
3322 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not stream
3323 BINFO_BASE_ACCESSES and BINFO_VPTR_FIELD.
3324 * tree-streamer-in.c (streamer_read_tree_bitfields): Likewise.
3325 (lto_input_ts_binfo_tree_pointers): Likewise.
3326 * tree-streamer-out.c (streamer_write_tree_bitfields,
3327 write_ts_binfo_tree_pointers): Likewise.
3328 * tree.c (free_lang_data_in_binfo): Clear BINFO_VPTR_FIELD.
3330 2018-06-22 Jan Hubicka <hubicka@ucw.cz>
3332 * tree.c (free_lang_data_in_type): Free all TYPE_VFIELDs.
3334 2018-06-22 Martin Liska <mliska@suse.cz>
3336 * symbol-summary.h (get): Make it pure and inline move
3337 functionality from ::get function.
3338 (get): Remove and inline into ::get and ::get_create.
3339 (get_create): Move code from ::get function.
3341 2018-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3344 * config/i386/sol2.h (CPP_SPEC): Don't pass -P for
3345 -x assembler-with-cpp.
3347 2018-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3349 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define
3350 _FILE_OFFSET_BITS=64 for C++.
3352 2018-06-21 Michael Meissner <meissner@linux.ibm.com>
3354 * config/rs6000/rs6000.md (extendtfif2): Add missing 128-bit
3355 conversion insn that shows up when pr85657-3.c is compiled using
3356 IEEE 128-bit long double.
3357 (neg<mode>2_internal): Use the correct mode to check whether the
3358 mode is IBM extended.
3359 * config/rs6000/rs6000.c (init_float128_ieee): Prevent complex
3360 multiply and divide external functions from being created more
3363 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
3365 * cfgrtl.c (fixup_reorder_chain): Do not emit NOPs in DECL_IGNORED_P
3367 (rtl_merge_blocks): Likewise. Do not emit a NOP if the location of
3368 the edge can be forwarded.
3369 (cfg_layout_merge_blocks): Likewise.
3371 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
3373 * except.c (finish_eh_generation): Commit edge insertions only after
3374 the EH edges have been redirected from post-landing to landing pads.
3376 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
3378 * tree-nested.c (get_frame_type): Use create_tmp_var_raw instead of
3379 create_tmp_var_for to create the FRAME decl.
3380 (finalize_nesting_tree_1): Do not unchain the FRAME decl.
3382 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
3384 * tree-inline.c (copy_edges_for_bb): Minor tweak.
3385 (maybe_move_debug_stmts_to_successors): Also reset the locus of the
3386 debug statement when resetting its value.
3387 (expand_call_inline): Copy the locus of the call onto the assignment
3388 of the return value, if any. Use local variable in more cases.
3390 2018-06-21 Martin Liska <mliska@suse.cz>
3392 * ipa-pure-const.c (propagate_nothrow): Use
3393 funct_state_summaries->get.
3394 (dump_malloc_lattice): Likewise.
3395 (propagate_malloc): Likewise.
3397 2018-06-21 Richard Biener <rguenther@suse.de>
3399 * lto-streamer-out.c (DFS::DFS_write_tree_body): Update outdated
3400 comment. Follow BLOCK_ABSTRACT_ORIGIN unconditionally.
3401 * tree-streamer-in.c (lto_input_ts_block_tree_pointers): Update
3403 * tree-streamer-out.c (write_ts_block_tree_pointers): Stream
3404 BLOCK_ABSTRACT_ORIGIN unconditionally.
3406 2018-06-21 David Malcolm <dmalcolm@redhat.com>
3408 * ipa-cp.c (ipcp_driver): Set edge_clone_summaries to NULL after
3410 * ipa-reference.c (ipa_reference_c_finalize): Delete
3411 ipa_ref_opt_sum_summaries and set it to NULL.
3413 2018-06-21 Tom de Vries <tdevries@suse.de>
3415 PR tree-optimization/85859
3416 * tree-ssa-tail-merge.c (stmt_local_def): Copy gimple_is_call
3417 test with comment from bb_no_side_effects_p.
3419 2018-06-21 Richard Biener <rguenther@suse.de>
3421 PR tree-optimization/86232
3422 * tree-ssa-loop-niter.c (number_of_iterations_popcount): Adjust
3423 max for constant niter.
3425 2018-06-21 Andre Vieira <andre.simoesdiasvieira@arm.com>
3427 * config/aarch64/aarch64-simd.md
3428 (*aarch64_crypto_aes<aes_op>v16qi_xor_combine): New.
3430 2018-06-21 Andre Vieira <andre.simoesdiasvieira@arm.com>
3432 * config/aarch64/aarch64-simd.md (aarch64_crypto_aes<aes_op>v16qi):
3433 Make opernads of the unspec commutative.
3435 2018-06-21 Richard Biener <rguenther@suse.de>
3437 * tree-data-ref.c (dr_step_indicator): Handle NULL DR_STEP.
3438 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
3439 Avoid calling vect_mark_for_runtime_alias_test with gathers or scatters.
3440 (vect_analyze_data_ref_dependence): Re-order checks to deal with
3442 (vect_record_base_alignments): Do not record base alignment
3443 for gathers or scatters.
3444 (vect_compute_data_ref_alignment): Drop return value that is always
3445 true. Bail out early for gathers or scatters.
3446 (vect_enhance_data_refs_alignment): Bail out early for gathers
3448 (vect_find_same_alignment_drs): Likewise.
3449 (vect_analyze_data_refs_alignment): Remove dead code.
3450 (vect_slp_analyze_and_verify_node_alignment): Likewise.
3451 (vect_analyze_data_refs): For possible gathers or scatters do
3452 not create an alternate DR, just check their possible validity
3453 and mark them. Adjust DECL_NONALIASED handling to not rely
3455 * tree-vect-loop-manip.c (vect_update_inits_of_drs): Do not
3456 update inits of gathers or scatters.
3457 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern):
3458 Also copy gather/scatter flag to pattern vinfo.
3460 2018-06-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
3462 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Change
3463 behavior of vec_packsu (vector unsigned long long, vector unsigned
3464 long long) to match behavior of vec_packs with same signature.
3466 2018-06-20 Chung-Lin Tang <cltang@codesourcery.com>
3467 Thomas Schwinge <thomas@codesourcery.com>
3468 Cesar Philippidis <cesar@codesourcery.com>
3470 * gimplify.c (gimplify_scan_omp_clauses): Add support for
3471 OMP_CLAUSE_{IF_PRESENT,FINALIZE}.
3472 (gimplify_adjust_omp_clauses): Likewise.
3473 (gimplify_oacc_declare_1): Add support for GOMP_MAP_RELEASE, remove
3474 support for GOMP_MAP_FORCE_{ALLOC,TO,FROM,TOFROM}.
3475 (gimplify_omp_target_update): Update handling of acc update and
3477 * omp-low.c (install_var_field): Remove unused parameter
3478 base_pointers_restrict.
3479 (scan_sharing_clauses): Remove base_pointers_restrict parameter.
3480 Update call to install_var_field. Handle OMP_CLAUSE_{IF_PRESENT,
3482 (omp_target_base_pointers_restrict_p): Delete.
3483 (scan_omp_target): Update call to scan_sharing_clauses.
3484 * tree-core.h (enum omp_clause_code): Add OMP_CLAUSE_{IF_PRESENT,
3486 * tree-nested.c (convert_nonlocal_omp_clauses): Handle
3487 OMP_CLAUSE_{IF_PRESENT,FINALIZE}.
3488 (convert_local_omp_clauses): Likewise.
3489 * tree-pretty-print.c (dump_omp_clause): Likewise.
3490 * tree.c (omp_clause_num_ops): Add entries for OMP_CLAUSE_{IF_PRESENT,
3492 (omp_clause_code_name): Likewise.
3494 2018-06-20 Jakub Jelinek <jakub@redhat.com>
3497 * var-tracking.c (use_narrower_mode_test): Check if shift amount can
3500 PR tree-optimization/86231
3501 * tree-vrp.c (union_ranges): For ( [ ) ] or ( )[ ] range and
3502 anti-range don't overwrite *vr0min before using it to compute *vr0max.
3504 2018-06-20 Tom de Vries <tdevries@suse.de>
3506 PR tree-optimization/86097
3507 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Also convert *nit to
3508 iv type if signedness of iv type is not the same as that of *nit.
3510 2018-06-20 Jakub Jelinek <jakub@redhat.com>
3512 * cfgrtl.c (rtl_verify_edges): Formatting fix. If bb->preds has any
3513 EDGE_EH edges, verify they are all EDGE_EH.
3515 2018-06-20 Maya Rashish <coypu@sdf.org>
3517 * ginclude/stddef.h: Limit #include <machine/ansi.h> to NetBSD.
3519 2018-06-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3521 * config/aarch64/aarch64-tuning-flags.def (no_ldp_stp_qregs): New.
3522 * config/aarch64/aarch64.c (xgene1_tunings): Add
3523 AARCH64_EXTRA_TUNE_NO_LDP_STP_QREGS to tune_flags.
3524 (aarch64_mode_valid_for_sched_fusion_p):
3525 Allow 16-byte modes.
3526 (aarch64_classify_address): Allow 16-byte modes for load_store_pair_p.
3527 * config/aarch64/aarch64-ldpstp.md: Add peepholes for LDP STP of
3529 * config/aarch64/aarch64-simd.md (load_pair<VQ:mode><VQ2:mode>):
3531 (vec_store_pair<VQ:mode><VQ2:mode>): Likewise.
3532 * config/aarch64/iterators.md (VQ2): New mode iterator.
3534 2018-06-20 Martin Liska <mliska@suse.cz>
3536 * tree-switch-conversion.c (jump_table_cluster::can_be_handled):
3537 Change default ratio from 10 to 8.
3539 2018-06-20 Martin Liska <mliska@suse.cz>
3541 * tree-switch-conversion.c (jump_table_cluster::find_jump_tables):
3543 (bit_test_cluster::find_bit_tests): Likewise.
3544 (switch_decision_tree::analyze_switch_statement): Find clusters.
3545 * tree-switch-conversion.h (struct jump_table_cluster): Document
3548 2018-06-20 Martin Liska <mliska@suse.cz>
3550 * tree-switch-conversion.c (switch_conversion::collect):
3551 Record m_uniq property.
3552 (switch_conversion::expand): Bail out for special conditions.
3553 (group_cluster::~group_cluster): New.
3554 (group_cluster::group_cluster): Likewise.
3555 (group_cluster::dump): Likewise.
3556 (jump_table_cluster::emit): New.
3557 (switch_decision_tree::fix_phi_operands_for_edges): New.
3558 (struct case_node): Remove struct.
3559 (jump_table_cluster::can_be_handled): New.
3560 (case_values_threshold): Moved to header.
3561 (reset_out_edges_aux): Likewise.
3562 (jump_table_cluster::is_beneficial): New.
3563 (bit_test_cluster::can_be_handled): Likewise.
3564 (add_case_node): Remove.
3565 (bit_test_cluster::is_beneficial): New.
3566 (case_bit_test::cmp): New.
3567 (bit_test_cluster::emit): New.
3568 (expand_switch_as_decision_tree_p): Remove.
3569 (bit_test_cluster::hoist_edge_and_branch_if_true): New.
3570 (fix_phi_operands_for_edge): Likewise.
3571 (switch_decision_tree::analyze_switch_statement): New.
3572 (compute_cases_per_edge): Move ...
3573 (switch_decision_tree::compute_cases_per_edge): ... here.
3574 (try_switch_expansion): Likewise.
3575 (switch_decision_tree::try_switch_expansion): Likewise.
3576 (record_phi_operand_mapping): Likewise.
3577 (switch_decision_tree::record_phi_operand_mapping): Likewise.
3578 (emit_case_decision_tree): Likewise.
3579 (switch_decision_tree::emit): Likewise.
3580 (balance_case_nodes): Likewise.
3581 (switch_decision_tree::balance_case_nodes): Likewise.
3582 (dump_case_nodes): Likewise.
3583 (switch_decision_tree::dump_case_nodes): Likewise.
3584 (emit_jump): Likewise.
3585 (switch_decision_tree::emit_jump): Likewise.
3586 (emit_cmp_and_jump_insns): Likewise.
3587 (switch_decision_tree::emit_cmp_and_jump_insns): Likewise.
3588 (emit_case_nodes): Likewise.
3589 (switch_decision_tree::emit_case_nodes): Likewise.
3590 (conditional_probability): Remove.
3591 * tree-switch-conversion.h (enum cluster_type): New.
3593 (struct cluster): Likewise.
3594 (cluster::cluster): Likewise.
3595 (struct simple_cluster): Likewise.
3596 (simple_cluster::simple_cluster): Likewise.
3597 (struct group_cluster): Likewise.
3598 (struct jump_table_cluster): Likewise.
3599 (struct bit_test_cluster): Likewise.
3600 (struct min_cluster_item): Likewise.
3601 (struct case_tree_node): Likewise.
3602 (case_tree_node::case_tree_node): Likewise.
3603 (jump_table_cluster::case_values_threshold): Likewise.
3604 (struct case_bit_test): Likewise.
3605 (struct switch_decision_tree): Likewise.
3606 (struct switch_conversion): Likewise.
3607 (switch_decision_tree::reset_out_edges_aux): Likewise.
3609 2018-06-20 Martin Liska <mliska@suse.cz>
3611 * tree-switch-conversion.c (MAX_CASE_BIT_TESTS): Remove.
3612 (hoist_edge_and_branch_if_true): Likewise.
3613 (expand_switch_using_bit_tests_p): Likewise.
3614 (struct case_bit_test): Likewise.
3615 (case_bit_test_cmp): Likewise.
3616 (emit_case_bit_tests): Likewise.
3617 (switch_conversion::switch_conversion): New class.
3618 (struct switch_conv_info): Remove old struct.
3619 (collect_switch_conv_info): More to ...
3620 (switch_conversion::collect): ... this.
3621 (check_range): Likewise.
3622 (switch_conversion::check_range): Likewise.
3623 (check_all_empty_except_final): Likewise.
3624 (switch_conversion::check_all_empty_except_final): Likewise.
3625 (check_final_bb): Likewise.
3626 (switch_conversion::check_final_bb): Likewise.
3627 (create_temp_arrays): Likewise.
3628 (switch_conversion::create_temp_arrays): Likewise.
3629 (free_temp_arrays): Likewise.
3630 (gather_default_values): Likewise.
3631 (switch_conversion::gather_default_values): Likewise.
3632 (build_constructors): Likewise.
3633 (switch_conversion::build_constructors): Likewise.
3634 (constructor_contains_same_values_p): Likewise.
3635 (switch_conversion::contains_same_values_p): Likewise.
3636 (array_value_type): Likewise.
3637 (switch_conversion::array_value_type): Likewise.
3638 (build_one_array): Likewise.
3639 (switch_conversion::build_one_array): Likewise.
3640 (build_arrays): Likewise.
3641 (switch_conversion::build_arrays): Likewise.
3642 (gen_def_assigns): Likewise.
3643 (switch_conversion::gen_def_assigns): Likewise.
3644 (prune_bbs): Likewise.
3645 (switch_conversion::prune_bbs): Likewise.
3646 (fix_phi_nodes): Likewise.
3647 (switch_conversion::fix_phi_nodes): Likewise.
3648 (gen_inbound_check): Likewise.
3649 (switch_conversion::gen_inbound_check): Likewise.
3650 (process_switch): Use the newly created class.
3651 (switch_conversion::expand): New.
3652 (switch_conversion::~switch_conversion): New.
3653 * tree-switch-conversion.h: New file.
3655 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
3657 * tree-vectorizer.h (NUM_PATTERNS, vect_recog_func_ptr): Move to
3658 tree-vect-patterns.c.
3659 * tree-vect-patterns.c (vect_supportable_direct_optab_p): New function.
3660 (vect_recog_dot_prod_pattern): Use it. Remove the type_in argument.
3661 (vect_recog_sad_pattern): Likewise.
3662 (vect_recog_widen_sum_pattern): Likewise.
3663 (vect_recog_pow_pattern): Likewise. Check for a null vectype.
3664 (vect_recog_widen_shift_pattern): Remove the type_in argument.
3665 (vect_recog_rotate_pattern): Likewise.
3666 (vect_recog_mult_pattern): Likewise.
3667 (vect_recog_vector_vector_shift_pattern): Likewise.
3668 (vect_recog_divmod_pattern): Likewise.
3669 (vect_recog_mixed_size_cond_pattern): Likewise.
3670 (vect_recog_bool_pattern): Likewise.
3671 (vect_recog_mask_conversion_pattern): Likewise.
3672 (vect_try_gather_scatter_pattern): Likewise.
3673 (vect_recog_widen_mult_pattern): Likewise. Check for a null vectype.
3674 (vect_recog_over_widening_pattern): Likewise.
3675 (vect_recog_gather_scatter_pattern): Likewise.
3676 (vect_recog_func_ptr): Move from tree-vectorizer.h
3677 (vect_vect_recog_func_ptrs): Move further down the file.
3678 (vect_recog_func): Likewise. Remove the third argument.
3679 (NUM_PATTERNS): Define based on vect_vect_recog_func_ptrs.
3680 (vect_pattern_recog_1): Expect the pattern function to do any
3681 necessary target tests. Also expect it to provide a vector type.
3682 Remove the type_in handling.
3684 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
3686 * tree-vect-patterns.c (vect_pattern_detected): New function.
3687 (vect_recog_dot_prod_patternm, vect_recog_sad_pattern)
3688 (vect_recog_widen_mult_pattern, vect_recog_widen_sum_pattern)
3689 (vect_recog_over_widening_pattern, vect_recog_widen_shift_pattern
3690 (vect_recog_rotate_pattern, vect_recog_vector_vector_shift_pattern)
3691 (vect_recog_mult_pattern, vect_recog_divmod_pattern)
3692 (vect_recog_mixed_size_cond_pattern, vect_recog_bool_pattern)
3693 (vect_recog_mask_conversion_pattern)
3694 (vect_try_gather_scatter_pattern): Likewise.
3696 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
3698 * tree-vect-patterns.c (vect_get_internal_def): New function.
3699 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern)
3700 (vect_recog_vector_vector_shift_pattern, check_bool_pattern)
3701 (search_type_for_mask_1): Use it.
3703 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
3705 * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Remove
3706 redundant WIDEN_SUM_EXPR handling.
3707 (vect_recog_sad_pattern): Likewise.
3709 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
3711 * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Remove
3712 redundant check that the types of a PLUS_EXPR or MULT_EXPR agree.
3713 (vect_recog_sad_pattern): Likewise PLUS_EXPR, ABS_EXPR and MINUS_EXPR.
3714 (vect_recog_widen_mult_pattern): Likewise MULT_EXPR.
3715 (vect_recog_widen_sum_pattern): Likewise PLUS_EXPR.
3717 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
3719 * tree-vect-stmts.c (vectorizable_call): Make sure that we
3720 use the stmt_vec_info of the original bb statement for the
3721 new zero assignment, even if the call is part of a pattern.
3723 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
3725 * tree-vectorizer.h (_stmt_vec_info): Note above pattern_def_seq
3726 that the sequence is attached to the original statement rather
3727 than the pattern statement.
3728 * tree-vect-loop.c (vect_determine_vf_for_stmt): Take the
3729 PATTERN_DEF_SEQ from the original statement rather than
3730 the main pattern statement.
3731 * tree-vect-stmts.c (free_stmt_vec_info): Likewise.
3732 * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Likewise.
3733 (vect_mark_pattern_stmts): Don't copy the PATTERN_DEF_SEQ.
3735 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
3737 * tree-vect-stmts.c (vect_analyze_stmt): Move the handling of pattern
3738 definition statements before the early exit for statements that aren't
3740 * tree-vect-loop.c (vect_transform_loop_stmt): New function,
3742 (vect_transform_loop): ...here. Process pattern definition
3743 statements without first checking whether the main pattern
3744 statement is live or relevant.
3746 2018-06-19 Eric Botcazou <ebotcazou@adacore.com>
3748 * tree-cfgcleanup.c (tree_forwarder_block_p): Do not return false at
3749 -O0 if the locus represent UNKNOWN_LOCATION but have different values.
3751 2018-06-19 Aaron Sawdey <acsawdey@linux.ibm.com>
3753 * config/rs6000/rs6000-string.c (select_block_compare_mode): Check
3754 TARGET_EFFICIENT_OVERLAPPING_UNALIGNED here instead of in caller.
3755 (do_and3, do_and3_mask, do_compb3, do_rotl3): New functions.
3756 (expand_block_compare): Change select_block_compare_mode call.
3757 (expand_strncmp_align_check): Use new functions, fix comment.
3758 (emit_final_str_compare_gpr): New function.
3759 (expand_strn_compare): Refactor and clean up code.
3760 * config/rs6000/vsx.md (vsx_mov<mode>_64bit): Remove *.
3762 2018-06-19 Tony Reix <tony.reix@atos.com>
3763 Damien Bergamini <damien.bergamini@atos.com>
3764 David Edelsohn <dje.gcc@gmail.com>
3766 * collect2.c (static_obj): New variable.
3767 (static_libs): New variable.
3768 (is_in_list): Uncomment declaration.
3769 (main): Track AIX libraries linked statically.
3770 (is_in_list): Uncomment definition.
3771 (scan_prog_file): Don't add AIX shared libraries initializer
3772 to constructor list if linking statically.
3774 2018-06-19 Max Filippov <jcmvbkbc@gmail.com>
3776 * config/xtensa/xtensa.md (UNSPEC_FRAME_BLOCKAGE): New unspec
3778 (allocate_stack, frame_blockage, *frame_blockage): New patterns.
3780 2018-06-19 Jan Hubicka <hubicka@ucw.cz>
3782 * tree.c (find_decls_types_r): Remove all non-VAR_DECLs from
3785 2018-06-19 Martin Liska <mliska@suse.cz>
3787 * config/i386/i386.c (ix86_can_inline_p): Do not use
3788 ipa_fn_summaries::get_create.
3789 * ipa-cp.c (ipcp_cloning_candidate_p): Replace get_create with
3791 (devirtualization_time_bonus): Likewise.
3792 (ipcp_propagate_stage): Likewise.
3793 * ipa-fnsummary.c (redirect_to_unreachable): Likewise.
3794 (edge_set_predicate): Likewise.
3795 (evaluate_conditions_for_known_args): Likewise.
3796 (evaluate_properties_for_edge): Likewise.
3797 (ipa_call_summary::reset): Tranform to ...
3798 (ipa_call_summary::~ipa_call_summary): ... this.
3799 (ipa_fn_summary::reset): Transform to ...
3800 (ipa_fn_summary::~ipa_fn_summary): ... this.
3801 (ipa_fn_summary_t::remove): Rename to ...
3802 (ipa_fn_summary_t::remove_callees): ... this.
3803 (ipa_fn_summary_t::duplicate): Use placement new
3804 instead of memory copy.
3805 (ipa_call_summary_t::duplicate): Likewise.
3806 (ipa_call_summary_t::remove): Remove.
3807 (dump_ipa_call_summary): Change get_create to get.
3808 (ipa_dump_fn_summary): Dump only when summary exists.
3809 (analyze_function_body): Use symbol_summary::get instead
3811 (compute_fn_summary): Likewise.
3812 (estimate_edge_devirt_benefit): Likewise.
3813 (estimate_edge_size_and_time): Likewise.
3814 (inline_update_callee_summaries): Likewise.
3815 (remap_edge_change_prob): Likewise.
3816 (remap_edge_summaries): Likewise.
3817 (ipa_merge_fn_summary_after_inlining): Likewise.
3818 (write_ipa_call_summary): Likewise.
3819 (ipa_fn_summary_write): Likewise.
3820 (ipa_free_fn_summary): Likewise.
3821 * ipa-fnsummary.h (struct GTY): Add new ctor and copy ctor.
3822 (struct ipa_call_summary): Likewise.
3823 * ipa-icf.c (sem_function::merge): Use symbol_summary::get instead
3825 * ipa-inline-analysis.c (do_estimate_edge_time): Likewise.
3826 (estimate_size_after_inlining): Likewise.
3827 (estimate_growth): Likewise.
3828 (growth_likely_positive): Likewise.
3829 * ipa-inline-transform.c (clone_inlined_nodes): Likewise.
3830 (inline_call): Likewise.
3831 * ipa-inline.c (caller_growth_limits): Likewise.
3832 (can_inline_edge_p): Likewise.
3833 (can_inline_edge_by_limits_p): Likewise.
3834 (compute_uninlined_call_time): Likewise.
3835 (compute_inlined_call_time): Likewise.
3836 (want_inline_small_function_p): Likewise.
3837 (edge_badness): Likewise.
3838 (update_caller_keys): Likewise.
3839 (update_callee_keys): Likewise.
3840 (inline_small_functions): Likewise.
3841 (inline_to_all_callers_1): Likewise.
3842 (dump_overall_stats): Likewise.
3843 (early_inline_small_functions): Likewise.
3844 (early_inliner): Likewise.
3845 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
3846 * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
3847 * ipa-pure-const.c (malloc_candidate_p): Likewise.
3848 * ipa-split.c (execute_split_functions): Likewise.
3849 * symbol-summary.h: Likewise.
3850 * tree-sra.c (ipa_sra_preliminary_function_checks): Likewise.
3852 2018-06-19 Richard Biener <rguenther@suse.de>
3854 * tree-vectorizer.c (try_vectorize_loop_1): Split out of ...
3855 (vectorize_loops): ... here. Fix dbgcnt handling.
3856 (try_vectorize_loop): Wrap try_vectorize_loop_1.
3858 2018-06-19 Segher Boessenkool <segher@kernel.crashing.org>
3861 * config/rs6000/rs6000.md (rs6000_discover_homogeneous_aggregate): An
3862 ieee128 argument takes up only one (vector) register, not two (floating
3865 2018-06-19 Eric Botcazou <ebotcazou@adacore.com>
3867 * gimplify.c (gimplify_init_constructor): Really never clear for an
3868 incomplete constructor if CONSTRUCTOR_NO_CLEARING is set.
3870 2018-06-19 Richard Biener <rguenther@suse.de>
3872 PR tree-optimization/86179
3873 * tree-vect-patterns.c (vect_pattern_recog_1): Clean up
3874 after failed recognition.
3876 2018-06-18 Martin Sebor <msebor@redhat.com>
3879 * calls.c (maybe_warn_nonstring_arg): Handle strncat.
3880 * tree-ssa-strlen.c (is_strlen_related_p): Make extern.
3881 Handle integer subtraction.
3882 (maybe_diag_stxncpy_trunc): Handle nonstring source arguments.
3883 * tree-ssa-strlen.h (is_strlen_related_p): Declare.
3885 2018-06-18 David Malcolm <dmalcolm@redhat.com>
3887 * config/frv/frv-protos.h (frv_ifcvt_modify_insn): Strengthen 3rd
3888 param from rtx to rtx_insn *.
3889 * config/frv/frv.c (frv_ifcvt_add_insn): Likewise for "insn"
3891 (frv_ifcvt_modify_insn): Likwise.
3892 (frv_ifcvt_modify_final): Likwise for local "existing_insn",
3893 adding an as_a <rtx_insn *> cast. Likewise for local "insn".
3894 * config/mips/mips.c (r10k_insert_cache_barriers): Add an
3895 as_a <rtx_insn *> cast to local "unprotected_region" once
3896 it's been established that it's not NULL or pc_rtx.
3897 * config/nds32/nds32-relax-opt.c (nds32_group_insns): Strengthen
3898 param "sethi" from rtx to rtx_insn *.
3899 (nds32_group_float_insns): Likewise for param "insn".
3900 * config/vax/vax-protos.h (vax_output_int_add): Likewise for 1st
3902 (vax_output_int_subtract): Likewise.
3903 * config/vax/vax.c (vax_output_int_add): Likewise for param
3905 (vax_output_int_subtract): Likewise.
3906 * emit-rtl.c (set_insn_deleted): Likewise, removing cast.
3907 (emit_pattern_after): Likewise for param "after".
3908 (emit_insn_after): Likewise.
3909 (emit_jump_insn_after): Likewise.
3910 (emit_call_insn_after): Likewise.
3911 (emit_debug_insn_after): Likewise.
3912 (emit_pattern_before): Likewise for param "before".
3913 (emit_insn_before): Likewise.
3914 (emit_jump_insn_before): Likewise.
3915 * final.c (get_insn_template): Likewise for param "insn", removing
3917 * output.h (get_insn_template): Likewise for 2nd param.
3918 * rtl.h (emit_insn_before): Likewise.
3919 (emit_jump_insn_before): Likewise.
3920 (emit_debug_insn_before_noloc): Likewise.
3921 (emit_insn_after): Likewise.
3922 (emit_jump_insn_after): Likewise.
3923 (emit_call_insn_after): Likewise.
3924 (emit_debug_insn_after): Likewise.
3925 (set_insn_deleted): Likewise for param.
3927 2018-06-18 Michael Meissner <meissner@linux.ibm.com>
3930 * config/rs6000/rs6000-modes.def (toplevel): Rework the 128-bit
3931 floating point modes, so that IFmode is numerically greater than
3932 TFmode, which is greater than KFmode using FRACTIONAL_FLOAT_MODE
3933 to declare the ordering. This prevents IFmode from being
3934 converted to TFmode when long double is IEEE 128-bit on an ISA 3.0
3935 machine. Include rs6000-modes.h to share the fractional values
3936 between genmodes* and the rest of the compiler.
3940 * config/rs6000/rs6000-modes.h: New file.
3941 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Change the
3942 meaning of rs6000_long_double_size so that 126..128 selects an
3943 appropriate 128-bit floating point type.
3944 (rs6000_option_override_internal): Likewise.
3945 * config/rs6000/rs6000.h (toplevel): Include rs6000-modes.h.
3946 (TARGET_LONG_DOUBLE_128): Change the meaning of
3947 rs6000_long_double_size so that 126..128 selects an appropriate
3948 128-bit floating point type.
3949 (LONG_DOUBLE_TYPE_SIZE): Update comment.
3950 * config/rs6000/rs6000.md (trunciftf2): Correct the modes of the
3951 source and destination to match the standard usage.
3952 (truncifkf2): Likewise.
3953 (copysign<mode>3, IEEE iterator): Rework copysign of float128 on
3954 ISA 2.07 to use an explicit clobber, instead of passing in a
3956 (copysign<mode>3_soft): Likewise.
3958 2018-06-18 David Malcolm <dmalcolm@redhat.com>
3960 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences):
3961 Replace dump_printf_loc call with DUMP_VECT_SCOPE.
3962 (vect_slp_analyze_instance_dependence): Likewise.
3963 (vect_enhance_data_refs_alignment): Likewise.
3964 (vect_analyze_data_refs_alignment): Likewise.
3965 (vect_slp_analyze_and_verify_instance_alignment
3966 (vect_analyze_data_ref_accesses): Likewise.
3967 (vect_prune_runtime_alias_test_list): Likewise.
3968 (vect_analyze_data_refs): Likewise.
3969 * tree-vect-loop-manip.c (vect_update_inits_of_drs): Likewise.
3970 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
3971 (vect_analyze_scalar_cycles_1): Likewise.
3972 (vect_get_loop_niters): Likewise.
3973 (vect_analyze_loop_form_1): Likewise.
3974 (vect_update_vf_for_slp): Likewise.
3975 (vect_analyze_loop_operations): Likewise.
3976 (vect_analyze_loop): Likewise.
3977 (vectorizable_induction): Likewise.
3978 (vect_transform_loop): Likewise.
3979 * tree-vect-patterns.c (vect_pattern_recog): Likewise.
3980 * tree-vect-slp.c (vect_analyze_slp): Likewise.
3981 (vect_make_slp_decision): Likewise.
3982 (vect_detect_hybrid_slp): Likewise.
3983 (vect_slp_analyze_operations): Likewise.
3984 (vect_slp_bb): Likewise.
3985 * tree-vect-stmts.c (vect_mark_stmts_to_be_vectorized): Likewise.
3986 (vectorizable_bswap): Likewise.
3987 (vectorizable_call): Likewise.
3988 (vectorizable_simd_clone_call): Likewise.
3989 (vectorizable_conversion): Likewise.
3990 (vectorizable_assignment): Likewise.
3991 (vectorizable_shift): Likewise.
3992 (vectorizable_operation): Likewise.
3993 * tree-vectorizer.h (DUMP_VECT_SCOPE): New macro.
3995 2018-06-18 Martin Sebor <msebor@redhat.com>
3997 PR tree-optimization/81384
3998 * builtin-types.def (BT_FN_SIZE_CONST_STRING_SIZE): New.
3999 * builtins.c (expand_builtin_strnlen): New function.
4000 (expand_builtin): Call it.
4001 (fold_builtin_n): Avoid setting TREE_NO_WARNING.
4002 * builtins.def (BUILT_IN_STRNLEN): New.
4003 * calls.c (maybe_warn_nonstring_arg): Handle BUILT_IN_STRNLEN.
4004 Warn for bounds in excess of maximum object size.
4005 * tree-ssa-strlen.c (maybe_set_strlen_range): Return tree representing
4006 single-value ranges. Handle strnlen.
4007 (handle_builtin_strlen): Handle strnlen.
4008 (strlen_check_and_optimize_stmt): Same.
4009 * doc/extend.texi (Other Builtins): Document strnlen.
4011 2018-06-18 Maya Rashish <coypu@sdf.org>
4013 * config/alpha/openbsd.h (TARGET_DEFAULT): Define.
4014 (LINK_SPEC, STARTFILE_SPEC, ENDFILE_SPEC): Likewise.
4015 (INTMAX_TYPE, UINTMAX_TYPE, WINT_TYPE): Likewise.
4017 * config/alpha/elf.h (STARTFILE_SPEC, ENDFILE_SPEC): Move from
4019 * config/alpha/linux.h (STARTFILE_SPEC, ENDFILE_SPEC): Here.
4021 2018-06-18 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
4023 * tree.c (escaped_string::escape): Replace cast to char * by
4024 const_cast<char *> (unescaped).
4026 2018-06-18 Nick Clifton <nickc@redhat.com>
4029 * tree.c (escaped_string): New class. Converts an unescaped
4030 string into its escaped equivalent.
4031 (warn_deprecated_use): Use the new class to convert the
4032 deprecation message, if present.
4033 (test_escaped_strings): New self test.
4034 (test_c_tests): Add test_escaped_strings.
4035 * doc/extend.texi (deprecated): Add a note that the
4036 deprecation message is affected by the -fmessage-length
4037 option, and that control characters will be escaped.
4038 (#pragma GCC error): Document this pragma.
4039 (#pragma GCC warning): Likewise.
4040 * doc/invoke.texi (-fmessage-length): Document this option's
4041 effect on the #warning and #error preprocessor directives and
4042 the deprecated attribute.
4044 2018-06-18 Eric Botcazou <ebotcazou@adacore.com>
4046 * tree.c (decl_value_expr_lookup): Revert latest change.
4047 (decl_value_expr_insert): Likewise.
4049 2018-06-17 Eric Botcazou <ebotcazou@adacore.com>
4051 * gimplify.c (nonlocal_vlas): Delete.
4052 (nonlocal_vla_vars): Likewise.
4053 (gimplify_var_or_parm_decl): Do not add debug VAR_DECLs for non-local
4055 (gimplify_body): Do not create and destroy nonlocal_vlas.
4056 * tree-nested.c: Include diagnostic.h.
4057 (use_pointer_in_frame): Tweak.
4058 (lookup_field_for_decl): Add assertion and declare the transformation.
4059 (convert_nonlocal_reference_op) <PARM_DECL>: Rework and issue an
4060 internal error when the reference is in a wrong context. Do not
4061 create a debug decl by default.
4062 (note_nonlocal_block_vlas): Delete.
4063 (convert_nonlocal_reference_stmt) <GIMPLE_BIND>: Do not call it.
4064 (convert_local_reference_op) <PARM_DECL>: Skip the frame decl. Do not
4065 create a debug decl by default.
4066 (convert_gimple_call) <GIMPLE_CALL>: Issue an internal error when the
4067 call is in a wrong context.
4068 (fixup_vla_decls): New function.
4069 (finalize_nesting_tree_1): Adjust comment. Call fixup_vla_decls if no
4070 debug variables were created.
4071 * tree.c (decl_value_expr_lookup): Add checking assertion.
4072 (decl_value_expr_insert): Likewise.
4074 2018-06-16 Kugan Vivekanandarajah <kuganv@linaro.org>
4077 * ipa-fnsummary.c (will_be_nonconstant_expr_predicate): Handle CALL_EXPR.
4078 * tree-scalar-evolution.c (interpret_expr): Likewise.
4079 (expression_expensive_p): Likewise.
4080 * tree-ssa-loop-ivopts.c (contains_abnormal_ssa_name_p): Likewise.
4081 * tree-ssa-loop-niter.c (number_of_iterations_popcount): New.
4082 (number_of_iterations_exit_assumptions): Use number_of_iterations_popcount.
4083 (ssa_defined_by_minus_one_stmt_p): New.
4085 2018-06-16 Kugan Vivekanandarajah <kuganv@linaro.org>
4088 * cfgexpand.c (expand_debug_expr): Hande ABSU_EXPR.
4089 * config/i386/i386.c (ix86_add_stmt_cost): Likewise.
4090 * dojump.c (do_jump): Likewise.
4091 * expr.c (expand_expr_real_2): Check operand type's sign.
4092 * fold-const.c (const_unop): Handle ABSU_EXPR.
4093 (fold_abs_const): Likewise.
4094 * gimple-pretty-print.c (dump_unary_rhs): Likewise.
4095 * gimple-ssa-backprop.c (backprop::process_assign_use): Likesie.
4096 (strip_sign_op_1): Likesise.
4097 * match.pd: Add new pattern to generate ABSU_EXPR.
4098 * optabs-tree.c (optab_for_tree_code): Handle ABSU_EXPR.
4099 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
4100 * tree-eh.c (operation_could_trap_helper_p): Likewise.
4101 * tree-inline.c (estimate_operator_cost): Likewise.
4102 * tree-pretty-print.c (dump_generic_node): Likewise.
4103 * tree-vect-patterns.c (vect_recog_sad_pattern): Likewise.
4104 * tree.def (ABSU_EXPR): New.
4106 2018-06-16 Jakub Jelinek <jakub@redhat.com>
4109 * common.opt (Wunsafe-loop-optimizations): Add Ignore, remove Var,
4110 documented as preserved for backward compatibility only.
4111 * doc/invoke.texi: Remove -Wunsafe-loop-optimizations documentation.
4113 PR rtl-optimization/86108
4114 * bb-reorder.c (create_forwarder_block): Renamed to ...
4115 (create_eh_forwarder_block): ... this. Split OLD_BB after labels and
4116 jump from new landing pad to the second part.
4117 (sjlj_fix_up_crossing_landing_pad, dw2_fix_up_crossing_landing_pad):
4120 2018-06-15 Jakub Jelinek <jakub@redhat.com>
4123 * expr.c (expand_assignment): Remove now redundant COMPLEX_MODE_P
4124 check from first store_expr, use to_mode instead of GET_MODE (to_rtx).
4125 Only call store_expr for halves if the mode is the same.
4128 * match.pd ((X / Y) == 0 -> X < Y): Don't transform complex divisions.
4129 Fix up comment formatting.
4131 2018-06-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
4133 * typed-splay-tree.h (typed_splay_tree::remove): New function.
4134 (typed_splay_tree::closure,
4135 typed_splay_tree::inner_foreach_fn, typed_splay_tree::m_inner): Deleted.
4136 (typed_splay_tree::typed_splay_tree,
4137 typed_splay_tree::operator =): Declared private.
4138 (typed_splay_tree::splay_tree_key, typed_splay_tree::splay_tree_value,
4139 typed_splay_tree::splay_tree_node_s, typed_splay_tree::KDEL,
4140 typed_splay_tree::VDEL, typed_splay_tree::splay_tree_delete_helper,
4141 typed_splay_tree::rotate_left, typed_splay_tree::rotate_right,
4142 typed_splay_tree::splay_tree_splay,
4143 typed_splay_tree::splay_tree_foreach_helper,
4144 typed_splay_tree::splay_tree_insert,
4145 typed_splay_tree::splay_tree_remove,
4146 typed_splay_tree::splay_tree_lookup,
4147 typed_splay_tree::splay_tree_predecessor,
4148 typed_splay_tree::splay_tree_successor,
4149 typed_splay_tree::splay_tree_min,
4150 typed_splay_tree::splay_tree_max): Took over from splay-tree.c/.h.
4151 (typed_splay_tree::root, typed_splay_tree::comp,
4152 typed_splay_tree::delete_key,
4153 typed_splay_tree::delete_value): New data members.
4154 * typed-splay-tree.c (selftest::test_str_to_int): Add a test for
4155 typed_splay_tree::remove.
4157 2018-06-15 Matthew Fortune <matthew.fortune@mips.com>
4159 * config/mips/mips.h (ASM_SPEC): Pass through -mcrc, -mno-crc,
4160 -mginv and -mno-ginv to the assembler.
4161 * config/mips/mips.opt (-mcrc): New option.
4163 * doc/invoke.text (-mcrc): Document.
4166 2018-06-15 Nick Clifton <nickc@redhat.com>
4169 * tree.c (escaped_string): New class. Converts an unescaped
4170 string into its escaped equivalent.
4171 (warn_deprecated_use): Use the new class to convert the
4172 deprecation message, if present.
4173 (test_escaped_strings): New self test.
4174 (test_c_tests): Add test_escaped_strings.
4175 * doc/extend.texi (deprecated): Add a note that the
4176 deprecation message is affected by the -fmessage-length
4177 option, and that control characters will be escaped.
4178 (#pragma GCC error): Document this pragma.
4179 (#pragma GCC warning): Likewise.
4180 * doc/invoke.texi (-fmessage-length): Document this option's
4181 effect on the #warning and #error preprocessor directives and
4182 the deprecated attribute.
4184 2018-06-15 Richard Biener <rguenther@suse.de>
4186 * tree-vect-slp.c (vect_slp_bb): Dump MSG_OPTIMIZED_LOCATIONS
4187 here, also noting vector size used.
4188 * tree-vectorizer.c (vectorize_loops): Adjust. Note vector
4189 size used in MSG_OPTIMIZED_LOCATIONS dump.
4190 (pass_slp_vectorize::execute): Adjust.
4192 2018-06-15 Claudiu Zissulescu <claziss@synopsys.com>
4195 * config/arc/arc.c (arc_return_address_register): Fix
4198 2018-06-15 Richard Biener <rguenther@suse.de>
4201 * tree-cfg.c (gimplify_build3): Do not strip sign conversions,
4202 leave useless conversion stripping to force_gimple_operand_gsi.
4203 (gimplify_build2): Likewise.
4204 (gimplify_build1): Likewise.
4206 2018-06-15 Richard Biener <rguenther@suse.de>
4209 * tree-cfg.c (move_stmt_op): unshare invariant addresses
4210 before adjusting their block.
4212 2018-06-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
4214 * config.gcc (riscv*-*-elf* | riscv*-*-rtems*): Use custom
4215 multilibs for *-*-rtems*.
4216 * config/riscv/t-rtems: New file.
4218 2018-06-14 Jakub Jelinek <jakub@redhat.com>
4221 * match.pd ((A +- CST1) +- CST2): Punt if last resort
4222 unsigned_type_for returns NULL.
4225 * lower-subreg.c (find_decomposable_subregs): Don't decompose float
4226 subregs of multi-word pseudos unless the float mode has word size.
4228 2018-06-14 Richard Biener <rguenther@suse.de>
4231 * tree-vect-generic.c (build_word_mode_vector_type): Remove
4232 duplicate and harmful type_hash_canon.
4233 * tree.c (type_hash_canon): Assert we didn't find ourselves.
4235 2018-06-14 Richard Biener <rguenther@suse.de>
4238 * tree-ssa-struct-alias.c (create_variable_info_for): Handle
4241 2018-06-14 Sebastian Huber <sebastian.huber@embedded-brains.de>
4243 * config/rtems.h (STDINT_LONG32): Define.
4245 2018-06-13 Matthew Fortune <matthew.fortune@mips.com>
4246 Prachi Godbole <prachi.godbole@imgtec.com>
4248 * config/mips/mips-cpus.def: Define P6600.
4249 * config/mips/mips-tables.opt: Regenerate.
4250 * config/mips/mips.c (mips_ucbranch_type): New enum.
4251 (mips_rtx_cost_data): Add support for P6600.
4252 (mips_issue_rate): Likewise.
4253 (mips_multipass_dfa_lookahead): Likewise.
4254 (mips_avoid_hazard): Likewise.
4255 (mips_reorg_process_insns): Likewise.
4256 (mips_classify_branch_p6600): New function.
4257 * config/mips/mips.h (TUNE_P6600): New define.
4258 (MIPS_ISA_LEVEL_SPEC): Infer mips64r6 from p6600.
4259 (ENABLE_LD_ST_PAIRS): Enable load/store bonding for p6600.
4260 * config/mips/mips.md: Include p6600.md.
4261 (processor): Add p6600.
4262 * config/mips/p6600.md: New file.
4263 * doc/invoke.texi: Add p6600 to supported architectures.
4265 2018-06-13 Martin Sebor <msebor@redhat.com>
4267 PR tree-optimization/86114
4268 * gimple-fold.c (gimple_fold_builtin_strlen): Only handle LHS
4270 * tree-ssa-strlen.c (maybe_set_strlen_range): Same.
4272 2018-06-13 Richard Biener <rguenther@suse.de>
4274 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern):
4275 Properly set vector type of the intermediate stmt.
4276 * tree-vect-stmts.c (vectorizable_operation): The destination
4277 var always has vectype_out type.
4279 2018-06-13 Jeff Law <law@redhat.com>
4281 * config/rl78/rl78.c (move_elim_pass): Use TDF_NONE rather than
4282 integer 0 for argument to print_rtl_with_bb.
4283 (rl78_reorg): Likewise.
4285 2018-06-13 David Malcolm <dmalcolm@redhat.com>
4287 * config/arc/arc.c (hwloop_optimize): Strengthen local "end_label"
4288 from rtx to rtx_insn *.
4289 * config/bfin/bfin.c (hwloop_optimize): Likewise for local
4291 (add_sched_insns_for_speculation): Likewise for local "target",
4292 converting usage of JUMP_LABEL to JUMP_LABEL_AS_INSN.
4293 * config/c6x/c6x.c (reorg_split_calls): Strengthen param "call_labels"
4294 from rtx_insn ** to rtx_code_label **.
4295 (reorg_emit_nops): Likewise.
4296 (c6x_reorg): Likewise for local "call_labels".
4297 * config/sh/sh-protos.h (get_dest_uid): Strengthen 1st param from
4299 * config/sh/sh.c (dump_table): Strengthen local "lab" from rtx to
4300 rtx_code_label *, adding safe_as_a <rtx_code_label *> casts to
4301 the loops over LABEL_REFS.
4302 (fixup_addr_diff_vecs): Add as_a <rtx_insn *> to usage of
4304 (barrier_align): Convert usage of JUMP_LABEL to JUMP_LABEL_AS_INSN.
4305 (get_dest_uid): Strengthen param "label" from rtx to rtx_insn *.
4306 (split_branches): Strengthen local "olabel" from rtx to
4307 rtx_insn *, adding a safe_as_a cast.
4308 * emit-rtl.c (next_real_insn): Strengthen param from "rtx"
4310 (add_insn_after): Likewise for first two params.
4311 (add_insn_before): Likewise.
4312 (remove_insn): Likewise for param.
4313 (emit_pattern_before_noloc): Likewise for second and third params.
4314 (emit_jump_insn_before_noloc): Convert NULL_RTX to NULL.
4315 (emit_call_insn_before_noloc): Likewise.
4316 (emit_debug_insn_before_noloc): Strengthen "before" param from "rtx"
4318 (emit_barrier_before): Likewise.
4319 (emit_label_before): Strengthen "label" param from "rtx" to
4320 "rtx_code_label *". Strengthen "before" param from "rtx" to
4322 (emit_insn_after_1): Strengthen "after" param from "rtx" to
4324 (emit_pattern_after_noloc): Likewise.
4325 (emit_insn_after_noloc): Likewise.
4326 (emit_jump_insn_after_noloc): Likewise.
4327 (emit_call_insn_after_noloc): Likewise.
4328 (emit_debug_insn_after_noloc): Likewise.
4329 (emit_barrier_after): Likewise.
4330 (emit_label_after): Likewise for both params.
4331 (emit_pattern_after_setloc): Likewise for "after" param. Convert
4332 "loc" param from "int" to "location_t".
4333 (emit_insn_after_setloc): Likewise.
4334 (emit_jump_insn_after_setloc): Likewise.
4335 (emit_call_insn_after_setloc): Likewise.
4336 (emit_debug_insn_after_setloc): Likewise.
4337 (emit_pattern_before_setloc): Likewise for "before" param. Convert
4338 "loc" param from "int" to "location_t".
4339 (emit_pattern_before): Convert NULL_RTX to NULL.
4340 (emit_insn_before_setloc): Convert "loc" param from "int" to
4342 (emit_jump_insn_before_setloc): Likewise.
4343 (emit_call_insn_before_setloc): Likewise.
4344 (emit_debug_insn_before_setloc): Strengthen "before" param from rtx to
4345 rtx_insn *. Convert "loc" param from "int" to "location_t".
4346 * rtl.h (emit_insn_before_setloc, emit_jump_insn_before_setloc,
4347 emit_call_insn_before_setloc, emit_debug_insn_before_setloc):
4348 Convert 3rd param from "int" to "location_t".
4349 (emit_barrier_before, emit_barrier_after, next_real_insn):
4350 Strengthen param from rtx to rtx_insn *.
4351 (emit_label_before): Strengthen 1st param from "rtx" to
4352 "rtx_code_label *". Strengthen 2nd param from "rtx" to
4354 (emit_insn_after_noloc, emit_jump_insn_after_noloc,
4355 emit_call_insn_after_noloc, emit_debug_insn_after_noloc):
4356 Strengthen 2nd param from "rtx" to "rtx_insn *".
4357 (emit_insn_after_setloc, emit_jump_insn_after_setloc)
4358 emit_call_insn_after_setloc, emit_debug_insn_after_setloc):
4359 Likewise. Convert 3rd param from "int" to "location_t".
4360 (emit_label_after): Strengthen 1st param from "rtx" to
4362 (next_real_insn, remove_insn): Strengthen param from "rtx" to
4364 (add_insn_before, add_insn_after): Strengthen 1st and 2nd params
4365 from "rtx" to "rtx_insn *".
4367 2018-06-13 Jan Hubicka <hubicka@gcc.gnu.org>
4369 * cgraph.c (cgraph_node::get_untransformed_body): Dump function
4370 bodies streamed in with -Q.
4371 * dumpfile.c (dump_files): Add lto-stream-out dump file.
4372 * dumpfile.h (tree_dump_index): Add lto_stream_out.
4373 * gimple-streamer-out.c: Include gimple-pretty-print.h
4374 (output_bb): Dump stmts streamed.
4375 * lto-section-out.c: Include print-tree.h
4376 (lto_begin_section): Dump sections created.
4377 (lto_output_decl_index): Dump decl encoded.
4378 * lto-streamer-out.c: Include print-tree.h
4379 (create_output_block): Dump output block created.
4380 (DFS::DFS_write_tree_body): Dump DFS SCCs streamed.
4381 (output_function): Dump function output.
4382 (output_constructor): Dump constructor streamed.
4383 (write_global_stream): Output indexes encoded.
4384 (produce_asm_for_decls): Dump streams encoded.
4385 * lto-streamer.c (streamer_dump_file): New global var.
4386 * lto-streamer.h (streamer_dump_file): Declare.
4387 * passes.c (ipa_write_summaries): Initialize streamer dump.
4388 * varpool.c (varpool_node::get_constructor): Dump constructors streamed
4391 2018-06-13 Eric Botcazou <ebotcazou@adacore.com>
4394 * config/i386/winnt.c (i386_pe_seh_cold_init): Do not emit negative
4395 offsets for register save directives. Emit a second batch of save
4396 directives, if need be, when the function accesses prior frames.
4398 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
4400 * config/arc/fpu.md (fmasf4): Force operand to register.
4401 (fnmasf4): Likewise.
4403 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
4405 * config/arc/arc-protos.h (arc_pad_return): Remove.
4406 * config/arc/arc.c (machine_function): Remove force_short_suffix
4408 (arc_print_operand): Adjust printing of '&'.
4409 (arc_verify_short): Remove conditional printing of short suffix.
4410 (arc_final_prescan_insn): Remove reference to size_reason.
4411 (pad_return): New function.
4412 (arc_reorg): Call pad_return.
4413 (arc_pad_return): Remove.
4414 (arc_init_machine_status): Remove reference to force_short_suffix.
4415 * config/arc/arc.md (vunspec): Add VUNSPEC_ARC_BLOCKAGE.
4416 (attr length): When attribute iscompact is true force to 2
4417 regardless; in the case of maybe check if we want to force the
4418 instruction to have 4 bytes length.
4419 (nopv): Change it to generate 4 byte long nop as well.
4420 (blockage): New pattern.
4421 (simple_return): Remove call to arc_pad_return.
4422 (p_return_i): Likewise.
4424 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
4426 * config/arc/elf.h (LINK_GCC_C_SEQUENCE_SPEC): Define.
4428 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
4430 * config/arc/builtins.def (SYNC): SYNC instruction is valid on all
4433 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
4435 * config/arc/arc.c (atomic_exchangesi): EX instruction is default
4436 for ARC700 and ARCv2.
4438 2018-06-13 Chenghua Xu <paul.hua.gm@gmail.com>
4441 * config/mips/loongson.md (vec_setv4hi): Gen_lowpart for
4442 operands[2] instead of operands[1].
4445 2018-06-12 Richard Sandiford <richard.sandiford@linaro.org>
4447 * lra-constraints.c (simplify_operand_subreg): In the paradoxical
4448 case, check whether the outer register overlaps an unallocatable
4449 register, not just whether it fits the required class.
4451 2018-06-12 Richard Sandiford <richard.sandiford@linaro.org>
4453 * poly-int.h (can_div_trunc_p): Add new overload in which all values
4455 * alias.c (get_addr): Extend CONST_INT handling to poly_int_rtx_p.
4456 (memrefs_conflict_p): Likewise.
4457 (init_alias_analysis): Likewise.
4458 * cfgexpand.c (expand_debug_expr): Likewise.
4459 * combine.c (combine_simplify_rtx, force_int_to_mode): Likewise.
4460 * cse.c (fold_rtx): Likewise.
4461 * explow.c (adjust_stack, anti_adjust_stack): Likewise.
4462 * expr.c (emit_block_move_hints): Likewise.
4463 (clear_storage_hints, push_block, emit_push_insn): Likewise.
4464 (store_expr_with_bounds, reduce_to_bit_field_precision): Likewise.
4465 (emit_group_load_1): Use rtx_to_poly_int64 for group offsets.
4466 (emit_group_store): Likewise.
4467 (find_args_size_adjust): Use strip_offset. Use rtx_to_poly_int64
4468 to read the PRE/POST_MODIFY increment.
4469 * calls.c (store_one_arg): Use strip_offset.
4470 * rtlanal.c (rtx_addr_can_trap_p_1): Extend CONST_INT handling to
4472 (set_noop_p): Use rtx_to_poly_int64 for the elements selected
4474 * simplify-rtx.c (avoid_constant_pool_reference): Use strip_offset.
4475 (simplify_binary_operation_1): Extend CONST_INT handling to
4477 * var-tracking.c (compute_cfa_pointer): Take a poly_int64 rather
4478 than a HOST_WIDE_INT.
4479 (hard_frame_pointer_adjustment): Change from HOST_WIDE_INT to
4481 (adjust_mems, add_stores): Update accodingly.
4482 (vt_canonicalize_addr): Track polynomial offsets.
4483 (emit_note_insn_var_location): Likewise.
4484 (vt_add_function_parameter): Likewise.
4485 (vt_initialize): Likewise.
4487 2018-06-12 Jeff Law <law@redhat.com>
4489 * config.gcc (alpha*-*-freebsd*): Remove.
4490 * config/alpha/freebsd.h: Remove.
4492 2018-06-12 David Malcolm <dmalcolm@redhat.com>
4495 * spellcheck-tree.c (levenshtein_distance): Rename to...
4496 (get_edit_distance): ...this, and update for underlying renaming.
4497 * spellcheck-tree.h (levenshtein_distance): Rename to...
4498 (get_edit_distance): ...this.
4499 * spellcheck.c (levenshtein_distance): Rename to...
4500 (get_edit_distance): ...this. Convert from Levenshtein distance
4501 to Damerau-Levenshtein distance by supporting transpositions of
4502 adjacent characters. Rename "v1" to "v_next" and "v0" to
4504 (selftest::levenshtein_distance_unit_test_oneway): Rename to...
4505 (selftest::test_edit_distance_unit_test_oneway): ...this, and
4506 update for underlying renaming.
4507 (selftest::levenshtein_distance_unit_test): Rename to...
4508 (selftest::test_get_edit_distance_unit): ...this, and update for
4509 underlying renaming.
4510 (selftest::test_find_closest_string): Add example from PR 69968
4511 where transposition helps
4512 (selftest::test_metric_conditions): Update for renaming.
4513 (selftest::test_metric_conditions): Likewise.
4514 (selftest::spellcheck_c_tests): Likewise.
4515 * spellcheck.h (levenshtein_distance): Rename both overloads to...
4516 (get_edit_distance): ...this.
4517 (best_match::consider): Update for renaming.
4519 2018-06-12 Martin Sebor <msebor@redhat.com>
4521 PR tree-optimization/85259
4522 * builtins.c (compute_objsize): Handle constant offsets.
4523 * gimple-ssa-warn-restrict.c (maybe_diag_offset_bounds): Return
4524 true iff a warning has been issued.
4525 * gimple.h (gimple_nonartificial_location): New function.
4526 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Call
4527 gimple_nonartificial_location and handle -Wno-system-headers.
4528 (handle_builtin_stxncpy): Same.
4530 2018-06-12 Martin Sebor <msebor@redhat.com>
4533 * fold-const.c (operand_equal_p): Handle SAVE_EXPR.
4535 2018-06-12 Will Schmidt <will_schmidt@vnet.ibm.com>
4537 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
4538 BUILTIN_VEC_XST entries for pointer to double and long long.
4540 2018-06-12 H.J. Lu <hongjiu.lu@intel.com>
4543 * config/i386/gnu-user.h (TARGET_THREAD_SPLIT_STACK_OFFSET):
4545 * config/i386/gnu-user64.h (TARGET_THREAD_SPLIT_STACK_OFFSET):
4548 2018-06-12 Martin Liska <mliska@suse.cz>
4550 * doc/options.texi: Document IntegerRange.
4552 2018-06-12 Martin Liska <mliska@suse.cz>
4554 * config/i386/i386.opt: Make MPX-related options as Deprecated.
4555 * opt-functions.awk: Handle Deprecated flag.
4556 * opts-common.c (decode_cmdline_option): Handle cl_deprecated
4558 (read_cmdline_option): Report warning for a deprecated option.
4559 * opts.h (struct cl_option): Add new field cl_deprecated.
4560 (CL_ERR_DEPRECATED): New.
4562 2018-06-12 Martin Liska <mliska@suse.cz>
4564 * doc/options.texi: Document Deprecated option flag.
4566 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
4568 * config/arc/arc-arch.h (arc_extras): New enum.
4569 (arc_cpu_t):Add field extra.
4570 (arc_cpu_types): Consider the extras.
4571 * config/arc/arc-cpus.def: Add extras info.
4572 * config/arc/arc-opts.h (processor_type): Consider extra field.
4573 * config/arc/arc.c (arc_override_options): Handle extra field.
4575 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
4577 * config/arc/arc-arch.h: Update ARC_OPTX macro.
4578 * config/arc/arc-options.def (ARC_OPTX): Introduce a new doc
4580 * config/arc/arc.c (arc_init): Update pic warning.
4581 (irq_range): Update irq range parsing warnings.
4582 (arc_override_options): Update various warning messages.
4583 (arc_handle_aux_attribute): Likewise.
4585 2018-06-12 Robert Suchanek <robert.suchanek@mips.com>
4587 * config/mips/i6400.md (i6400_fpu_fadd): Remove frint.
4589 2018-06-12 Jozef Lawrynowicz <jozef.l@mittosystems.com>
4591 * doc/sourcebuild.texi: Document usage of line number 0 in verify
4592 compiler messages directives.
4594 2018-06-12 Matthew Fortune <mfortune@gmail.com>
4596 * config/mips/mips-cpus.def: New MIPS_CPU for i6500.
4597 * config/mips/mips-tables.opt: Regenerate.
4598 * config/mips/mips.h (MIPS_ISA_LEVEL_SPEC): Mark i6500 as
4600 * doc/invoke.texi: Document -march=i6500.
4602 2018-06-12 Prachi Godbole <prachi.godbole@imgtec.com>
4604 * config/mips/i6400.md (i6400_gpmuldiv): Remove cpu_unit.
4605 (i6400_gpmul): Add cpu_unit.
4606 (i6400_gpdiv): Likewise.
4607 (i6400_msa_add_d): Update reservations.
4608 (i6400_msa_int_add) Likewise.
4609 (i6400_msa_short_logic3) Likewise.
4610 (i6400_msa_short_logic2) Likewise.
4611 (i6400_msa_short_logic) Likewise.
4612 (i6400_msa_move) Likewise.
4613 (i6400_msa_cmp) Likewise.
4614 (i6400_msa_short_float2) Likewise.
4615 (i6400_msa_div_d) Likewise.
4616 (i6400_msa_long_logic1) Likewise.
4617 (i6400_msa_long_logic2) Likewise.
4618 (i6400_msa_mult) Likewise.
4619 (i6400_msa_long_float2) Likewise.
4620 (i6400_msa_long_float4) Likewise.
4621 (i6400_msa_long_float5) Likewise.
4622 (i6400_msa_long_float8) Likewise.
4623 (i6400_fpu_fadd): Include frint type.
4624 (i6400_fpu_store): New define_insn_reservation.
4625 (i6400_fpu_load): Likewise.
4626 (i6400_fpu_move): Likewise.
4627 (i6400_fpu_fcmp): Likewise.
4628 (i6400_fpu_fmadd): Likewise.
4629 (i6400_int_mult): Include imul3nc type and update reservation.
4630 (i6400_int_div): Include idiv3 type and update reservation.
4631 (i6400_int_load): Update to check type not move_type.
4632 (i6400_int_store): Likewise.
4633 (i6400_int_prefetch): Set zero latency.
4635 2018-06-12 Eric Botcazou <ebotcazou@adacore.com>
4637 * gcc.c: Document new %@{...} sequence.
4638 (LINK_COMMAND_SPEC): Use it for the -L switches.
4639 (cpp_unique_options): Use it for the -I switches.
4640 (at_file_argbuf): New global variable.
4641 (in_at_file): Likewise.
4642 (alloc_args): Create at_file_argbuf.
4643 (clear_args): Truncate at_file_argbuf.
4644 (store_arg): If in_at_file, push the argument onto at_file_argbuf.
4645 (open_at_file): New function.
4646 (close_at_file): Likewise.
4647 (create_at_file): Delete.
4648 (do_spec_1) <'i'>: Use open_at_file/close_at_file.
4651 (validate_switches_from_spec): Deal with %@{...} sequence.
4652 (validate_switches): Likewise.
4653 (driver::finalize): Call clear_args.
4655 2018-06-11 Rasmus Villemoes <rasmus.villemoes@prevas.dk>
4657 * config/vx-common.h (USE_TM_CLONE_REGISTRY): #define to 0.
4659 2018-06-11 Martin Sebor <msebor@redhat.com>
4661 * doc/invoke.texi (-Wall): List -Wc++17-compat.
4662 (Wno-class-memaccess): Add @opindex.
4663 (Wno-templates, Wno-multiple-inheritance): Same.
4664 (Wno-virtual-inheritance, Wno-namespaces): Same.
4665 (Wno-pedantic, Wno-chkp, Wcoverage-mismatch): Same.
4666 (Wno-format-overflow, Wno-format-truncation): Same.
4667 (Wno-unused-local-typedefs, Walloc-size-larger-than): Same.
4668 (Wno-alloc-size-larger-than, Wframe-larger-than): Same
4669 (Wno-stack-usage, Wno-c++-compat, Wno-c++11-compat): Same.
4670 (Wno-c++14-compat, Wno-c++17-compat, Wno-openmp-simd): Same.
4671 (Wno-unsuffixed-float-constants, Wno-addr-space-convert): Same.
4672 (Wno-misspelled-isr): Same.
4674 2018-06-11 Martin Sebor <msebor@redhat.com>
4676 * PR tree-optimization/86083
4677 * tree-ssa-strlen.c (handle_char_store): Use tree_expr_nonzero_p.
4679 2018-06-11 Zhouyi Zhou <zhouzhouyi@gmail.com>
4681 * tree-eh.c (lower_eh_constructs_2): Add a comma to comment.
4683 2018-06-11 Segher Boessenkool <segher@kernel.crashing.org>
4686 * config/rs6000/rs6000.md (*movdi_internal32): Put constraint modifiers
4687 on the correct operand.
4688 (*movdi_internal64): Ditto.
4690 2018-06-11 Martin Liska <mliska@suse.cz>
4692 PR tree-optimization/86089
4693 * tree-ssa-strlen.c (get_string_length): Move back removed hunk.
4695 2018-06-11 Julia Koval <julia.koval@intel.com>
4697 * config/i386/vaesintrin.h (_mm_aesdec_epi128, _mm_aesdeclast_epi128)
4698 _mm_aesenc_epi128, _mm_aesenclast_epi128: Remove.
4699 * config/i386/vpclmulqdqintrin.h (_mm_clmulepi64_epi128): Remove.
4701 2018-06-11 Olivier Hainque <hainque@adacore.com>
4703 * dwarf2out.c (gen_compile_unit_die): Fallout to DW_LANG_Ada83
4704 for Ada with strict dwarf2.
4706 2018-06-08 Peter Bergner <bergner@vnet.ibm.com>
4709 * config/rs6000/rs6000.c (mem_operand_gpr): Enable PRE_INC and PRE_DEC
4712 2018-06-08 Jan Hubicka <hubicka@ucw.cz>
4714 * dumpfile.c (FIRST_ME_AUTO_NUMBERED_DUMP): Bump to 4.
4716 2018-06-08 David Edelsohn <dje.gcc@gmail.com>
4718 * config/rs6000/rs6000.c (rs6000_passes_ieee128): Protect with #if
4721 2018-06-08 Martin Liska <mliska@suse.cz>
4723 * tree-cfg.h (debug_function): Fix argument type to match
4726 2018-06-08 Martin Liska <mliska@suse.cz>
4728 * config/powerpcspe/powerpcspe.c (rs6000_xcoff_visibility):
4729 Remove usage of MPX-related (and removed) fields.
4730 * config/rs6000/rs6000.c (rs6000_xcoff_visibility): Likewise.
4732 2018-06-08 David Malcolm <dmalcolm@redhat.com>
4734 * cfg.c (debug): Use TDF_NONE rather than 0.
4735 * cfghooks.c (debug): Likewise.
4736 * dumpfile.c (DUMP_FILE_INFO): Likewise; also for OPTGROUP.
4737 (struct dump_option_value_info): Convert to...
4738 (struct kv_pair): ...this template type.
4739 (dump_options): Convert to kv_pair<dump_flags_t>; use TDF_NONE
4741 (optinfo_verbosity_options): Likewise.
4742 (optgroup_options): Convert to kv_pair<optgroup_flags_t>; use
4744 (gcc::dump_manager::dump_register): Use optgroup_flags_t rather
4745 than int for "optgroup_flags" param.
4746 (dump_generic_expr_loc): Use dump_flags_t rather than int for
4748 (dump_dec): Likewise.
4749 (dump_finish): Use TDF_NONE rather than 0.
4750 (gcc::dump_manager::opt_info_enable_passes): Use optgroup_flags_t
4751 rather than int for "optgroup_flags" param. Use TDF_NONE rather
4752 than 0. Update for change to option_ptr.
4753 (opt_info_switch_p_1): Convert "optgroup_flags" param from int *
4754 to optgroup_flags_t *. Use TDF_NONE and OPTGROUP_NONE rather than
4755 0. Update for changes to optinfo_verbosity_options and
4757 (opt_info_switch_p): Convert optgroup_flags from int to
4759 (dump_basic_block): Use dump_flags_t rather than int
4760 for "dump_kind" param.
4761 * dumpfile.h (TDF_ADDRESS, TDF_SLIM, TDF_RAW, TDF_DETAILS,
4762 TDF_STATS, TDF_BLOCKS, TDF_VOPS, TDF_LINENO, TDF_UID)
4763 TDF_STMTADDR, TDF_GRAPH, TDF_MEMSYMS, TDF_RHS_ONLY, TDF_ASMNAME,
4764 TDF_EH, TDF_NOUID, TDF_ALIAS, TDF_ENUMERATE_LOCALS, TDF_CSELIB,
4765 TDF_SCEV, TDF_GIMPLE, TDF_FOLDING, MSG_OPTIMIZED_LOCATIONS,
4766 MSG_MISSED_OPTIMIZATION, MSG_NOTE, MSG_ALL, TDF_COMPARE_DEBUG,
4767 TDF_NONE): Convert from macros to...
4768 (enum dump_flag): ...this new enum.
4769 (dump_flags_t): Update to use enum.
4770 (operator|, operator&, operator~, operator|=, operator&=):
4771 Implement for dump_flags_t.
4772 (OPTGROUP_NONE, OPTGROUP_IPA, OPTGROUP_LOOP, OPTGROUP_INLINE,
4773 OPTGROUP_OMP, OPTGROUP_VEC, OPTGROUP_OTHER, OPTGROUP_ALL):
4774 Convert from macros to...
4775 (enum optgroup_flag): ...this new enum.
4776 (optgroup_flags_t): New typedef.
4777 (operator|, operator|=): Implement for optgroup_flags_t.
4778 (struct dump_file_info): Convert field "alt_flags" to
4779 dump_flags_t. Convert field "optgroup_flags" to
4781 (dump_basic_block): Use dump_flags_t rather than int for param.
4782 (dump_generic_expr_loc): Likewise.
4783 (dump_dec): Likewise.
4784 (dump_register): Convert param "optgroup_flags" to
4786 (opt_info_enable_passes): Likewise.
4787 * early-remat.c (early_remat::dump_edge_list): Use TDF_NONE rather
4789 * gimple-pretty-print.c (debug): Likewise.
4790 * gimple-ssa-store-merging.c (bswap_replace): Likewise.
4791 (merged_store_group::apply_stores): Likewise.
4792 * gimple-ssa-strength-reduction.c (insert_initializers): Likewise.
4793 * gimple.c (verify_gimple_pp): Likewise.
4794 * graphite-poly.c (print_pbb_body): Likewise.
4795 * passes.c (pass_manager::register_one_dump_file): Convert
4796 local "optgroup_flags" to optgroup_flags_t.
4797 * print-tree.c (print_node): Use TDF_NONE rather than 0.
4799 (debug_body): Likewise.
4800 * tree-pass.h (struct pass_data): Convert field "optgroup_flags"
4801 to optgroup_flags_t.
4802 * tree-pretty-print.c (print_struct_decl): Use TDF_NONE rather
4804 * tree-ssa-math-opts.c (convert_mult_to_fma_1): Likewise.
4805 (convert_mult_to_fma): Likewise.
4806 * tree-ssa-reassoc.c (undistribute_ops_list): Likewise.
4807 * tree-ssa-sccvn.c (vn_eliminate): Likewise.
4808 * tree-vect-data-refs.c (dump_lower_bound): Convert param
4809 "dump_kind" to dump_flags_t.
4811 2018-06-08 Segher Boessenkool <segher@kernel.crashing.org>
4813 * config/rs6000/rs6000.c (min, max): Delete.
4815 2018-06-08 Segher Boessenkool <segher@kernel.crashing.org>
4817 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete mention of
4818 -mabi=spe and -mabi=no-spe.
4820 2018-06-08 Martin Liska <mliska@suse.cz>
4822 * ipa-pure-const.c (propagate_pure_const): Use ::get at places
4823 where we expect an existing summary.
4825 2018-06-08 Martin Liska <mliska@suse.cz>
4827 * ipa-inline-analysis.c (simple_edge_hints): Use ::get method.
4828 * ipa-inline.h (estimate_edge_growth): Likewise.
4830 2018-06-08 Martin Liska <mliska@suse.cz>
4832 * cgraph.c (function_version_hasher::hash): Use
4833 cgraph_node::get_uid ().
4834 (function_version_hasher::equal):
4835 * cgraph.h (cgraph_node::get_uid): New method.
4836 * ipa-inline.c (update_caller_keys): Use
4837 cgraph_node::get_uid ().
4838 (update_callee_keys): Likewise.
4839 * ipa-utils.c (searchc): Likewise.
4840 (ipa_reduced_postorder): Likewise.
4841 * lto-cgraph.c (input_node): Likewise.
4842 * passes.c (is_pass_explicitly_enabled_or_disabled): Likewise.
4843 * symbol-summary.h (symtab_insertion): Likewise.
4844 (symtab_removal): Likewise.
4845 (symtab_duplication): Likewise.
4846 * tree-pretty-print.c (dump_function_header): Likewise.
4847 * tree-sra.c (convert_callers_for_node): Likewise.
4849 2018-06-08 Martin Liska <mliska@suse.cz>
4851 * cgraph.c (symbol_table::create_edge): Always assign a new
4853 (symbol_table::free_edge): Do not recycle numbers.
4854 * cgraph.h (cgraph_edge::get): New method.
4855 * symbol-summary.h (symtab_removal): Use it.
4856 (symtab_duplication): Likewise.
4857 (call_summary::hashable_uid): Remove.
4859 2018-06-08 Martin Liska <mliska@suse.cz>
4861 * ipa-inline-analysis.c (inline_edge_removal_hook): Remove.
4862 (initialize_growth_caches): Remove.
4863 (free_growth_caches): Likewise.
4864 (do_estimate_edge_time): Use edge_growth_cache.
4865 (do_estimate_edge_size): Likewise.
4866 (do_estimate_edge_hints): Likewise.
4867 * ipa-inline.c (reset_edge_caches): Likewise.
4868 (recursive_inlining): Likewise.
4869 (inline_small_functions): Likewise.
4870 * ipa-inline.h (initialize_growth_caches): Remove.
4871 (estimate_edge_size): Likewise.
4872 (estimate_edge_time): Likewise.
4873 (estimate_edge_hints): Likewise.
4874 (reset_edge_growth_cache): Likewise.
4875 * symbol-summary.h (call_summary::remove): New method.
4877 2018-06-08 Martin Liska <mliska@suse.cz>
4879 * ipa-cp.c (class edge_clone_summary): New summary.
4880 (grow_edge_clone_vectors): Remove.
4881 (ipcp_edge_duplication_hook): Remove.
4882 (class edge_clone_summary_t): New call_summary class.
4883 (ipcp_edge_removal_hook): Remove.
4884 (edge_clone_summary_t::duplicate): New function.
4885 (get_next_cgraph_edge_clone): Use edge_clone_summaries.
4886 (create_specialized_node): Likewise.
4887 (ipcp_driver): Initialize edge_clone_summaries and do not
4890 2018-06-08 Martin Liska <mliska@suse.cz>
4892 * symbol-summary.h (get): New function.
4893 (call_summary::m_initialize_when_cloning): New class member.
4895 2018-06-08 Martin Liska <mliska@suse.cz>
4897 * cgraph.c (cgraph_node::remove): Do not recycle uid.
4898 * cgraph.h (symbol_table::release_symbol): Do not pass uid.
4899 (symbol_table::allocate_cgraph_symbol): Do not set uid.
4900 * passes.c (uid_hash_t): Record removed_nodes by their uids.
4901 (remove_cgraph_node_from_order): Use the removed_nodes set.
4902 (do_per_function_toporder): Likwise.
4903 * symbol-summary.h (symtab_insertion): Use cgraph_node::uid
4904 instead of summary_uid.
4905 (symtab_removal): Likewise.
4906 (symtab_duplication): Likewise.
4908 2018-06-08 Martin Liska <mliska@suse.cz>
4910 * ipa-cp.c (ipcp_store_bits_results): Use
4911 ipcp_transformation_sum.
4912 (ipcp_store_vr_results): Likewise.
4913 * ipa-prop.c (ipcp_grow_transformations_if_necessary): Renamed
4915 (ipcp_transformation_initialize): ... this.
4916 (ipa_set_node_agg_value_chain):
4917 (ipa_node_params_t::duplicate): Use ipcp_transformation_sum.
4918 (write_ipcp_transformation_info): Likewise.
4919 (read_ipcp_transformation_info): Likewise.
4920 (ipcp_update_bits): Likewise.
4921 (ipcp_update_vr): Likewise.
4922 (ipcp_transform_function): Likewise.
4923 * ipa-prop.h: Rename ipcp_transformation_summary to
4924 ipcp_transformation.
4925 (class ipcp_transformation_t): New function summary.
4926 (ipcp_get_transformation_summary): Use ipcp_transformation_sum.
4927 (ipa_get_agg_replacements_for_node): Likewise.
4929 2018-06-08 Martin Liska <mliska@suse.cz>
4931 * ipa-pure-const.c (struct funct_state_d): Do it class instead
4933 (class funct_state_summary_t): New function_summary class.
4934 (has_function_state): Remove.
4935 (get_function_state): Likewise.
4936 (set_function_state): Likewise.
4937 (add_new_function): Likewise.
4938 (funct_state_summary_t::insert): New function.
4939 (duplicate_node_data): Remove.
4940 (remove_node_data): Remove.
4941 (funct_state_summary_t::duplicate): New function.
4942 (register_hooks): Create new funct_state_summaries.
4943 (pure_const_generate_summary): Use it.
4944 (pure_const_write_summary): Likewise.
4945 (pure_const_read_summary): Likewise.
4946 (propagate_pure_const): Likewise.
4947 (propagate_nothrow): Likewise.
4948 (dump_malloc_lattice): Likewise.
4949 (propagate_malloc): Likewise.
4950 (execute): Do not register hooks, just remove summary
4952 (pass_ipa_pure_const::pass_ipa_pure_const): Simplify
4955 2018-06-08 Martin Liska <mliska@suse.cz>
4957 * ipa-reference.c (remove_node_data): Remove.
4958 (duplicate_node_data): Likewise.
4959 (class ipa_ref_var_info_summary_t): New class.
4960 (class ipa_ref_opt_summary_t): Likewise.
4961 (get_reference_vars_info): Use ipa_ref_var_info_summaries.
4962 (get_reference_optimization_summary): Use
4963 ipa_ref_opt_sum_summaries.
4964 (set_reference_vars_info): Remove.
4965 (set_reference_optimization_summary): Likewise.
4966 (ipa_init): Create summaries.
4967 (init_function_info): Use function summary.
4968 (ipa_ref_opt_summary_t::duplicate): New function.
4969 (ipa_ref_opt_summary_t::remove): New function.
4970 (get_read_write_all_from_node): Fix GNU coding style.
4971 (propagate): Use function summary.
4972 (write_node_summary_p): Fix GNU coding style.
4973 (stream_out_bitmap): Likewise.
4974 (ipa_reference_read_optimization_summary): Use function summary.
4975 (ipa_reference_c_finalize): Do not release hooks.
4977 2018-06-08 Martin Liska <mliska@suse.cz>
4979 * ipa-fnsummary.c (dump_ipa_call_summary): Use ::get method.
4980 (analyze_function_body): Extract multiple calls of get_create.
4981 * ipa-inline-analysis.c (simple_edge_hints): Likewise.
4982 * ipa-inline.c (recursive_inlining): Use ::get method.
4983 * ipa-inline.h (estimate_edge_growth): Likewise.
4985 2018-06-08 Martin Liska <mliska@suse.cz>
4987 * hsa-common.h (enum hsa_function_kind): Rename HSA_NONE to
4989 (hsa_function_summary::hsa_function_summary): Use the new enum
4991 (hsa_gpu_implementation_p): Use hsa_summaries::get.
4992 * hsa-gen.c (hsa_get_host_function): Likewise.
4993 (get_brig_function_name): Likewise.
4994 * ipa-hsa.c (process_hsa_functions): Likewise.
4995 (ipa_hsa_write_summary): Likewise.
4996 * symbol-summary.h (symtab_duplication): Use ::get function/
4997 (get): New function.
4999 2018-06-08 Martin Liska <mliska@suse.cz>
5001 * config/i386/i386.c (ix86_can_inline_p): Use get_create instead
5003 * hsa-common.c (hsa_summary_t::link_functions): Likewise.
5004 (hsa_register_kernel): Likewise.
5005 * hsa-common.h (hsa_gpu_implementation_p): Likewise.
5006 * hsa-gen.c (hsa_get_host_function): Likewise.
5007 (get_brig_function_name): Likewise.
5008 (generate_hsa): Likewise.
5009 (pass_gen_hsail::execute): Likewise.
5010 * ipa-cp.c (ipcp_cloning_candidate_p): Likewise.
5011 (devirtualization_time_bonus): Likewise.
5012 (ipcp_propagate_stage): Likewise.
5013 * ipa-fnsummary.c (redirect_to_unreachable): Likewise.
5014 (edge_set_predicate): Likewise.
5015 (evaluate_conditions_for_known_args): Likewise.
5016 (evaluate_properties_for_edge): Likewise.
5017 (ipa_fn_summary::reset): Likewise.
5018 (ipa_fn_summary_t::duplicate): Likewise.
5019 (dump_ipa_call_summary): Likewise.
5020 (ipa_dump_fn_summary): Likewise.
5021 (analyze_function_body): Likewise.
5022 (compute_fn_summary): Likewise.
5023 (estimate_edge_devirt_benefit): Likewise.
5024 (estimate_edge_size_and_time): Likewise.
5025 (estimate_calls_size_and_time): Likewise.
5026 (estimate_node_size_and_time): Likewise.
5027 (inline_update_callee_summaries): Likewise.
5028 (remap_edge_change_prob): Likewise.
5029 (remap_edge_summaries): Likewise.
5030 (ipa_merge_fn_summary_after_inlining): Likewise.
5031 (ipa_update_overall_fn_summary): Likewise.
5032 (read_ipa_call_summary): Likewise.
5033 (inline_read_section): Likewise.
5034 (write_ipa_call_summary): Likewise.
5035 (ipa_fn_summary_write): Likewise.
5036 (ipa_free_fn_summary): Likewise.
5037 * ipa-hsa.c (process_hsa_functions): Likewise.
5038 (ipa_hsa_write_summary): Likewise.
5039 (ipa_hsa_read_section): Likewise.
5040 * ipa-icf.c (sem_function::merge): Likewise.
5041 * ipa-inline-analysis.c (simple_edge_hints): Likewise.
5042 (do_estimate_edge_time): Likewise.
5043 (estimate_size_after_inlining): Likewise.
5044 (estimate_growth): Likewise.
5045 (growth_likely_positive): Likewise.
5046 * ipa-inline-transform.c (clone_inlined_nodes): Likewise.
5047 (inline_call): Likewise.
5048 * ipa-inline.c (caller_growth_limits): Likewise.
5049 (can_inline_edge_p): Likewise.
5050 (can_inline_edge_by_limits_p): Likewise.
5051 (compute_uninlined_call_time): Likewise.
5052 (compute_inlined_call_time): Likewise.
5053 (want_inline_small_function_p): Likewise.
5054 (edge_badness): Likewise.
5055 (update_caller_keys): Likewise.
5056 (update_callee_keys): Likewise.
5057 (recursive_inlining): Likewise.
5058 (inline_small_functions): Likewise.
5059 (inline_to_all_callers_1): Likewise.
5060 (dump_overall_stats): Likewise.
5061 (early_inline_small_functions): Likewise.
5062 (early_inliner): Likewise.
5063 * ipa-inline.h (estimate_edge_growth): Likewise.
5064 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
5065 * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
5066 * ipa-prop.h (IPA_NODE_REF): Likewise.
5067 (IPA_EDGE_REF): Likewise.
5068 * ipa-pure-const.c (malloc_candidate_p): Likewise.
5069 (propagate_malloc): Likewise.
5070 * ipa-split.c (execute_split_functions): Likewise.
5071 * symbol-summary.h: Rename get to get_create.
5073 (get_create): Likewise.
5074 * tree-sra.c (ipa_sra_preliminary_function_checks): Likewise.
5076 2018-06-08 Martin Liska <mliska@suse.cz>
5078 * symbol-summary.h (release): Move definition out of class
5080 (symtab_removal): Likewise.
5081 (symtab_duplication): Likewise.
5083 2018-06-08 Martin Liska <mliska@suse.cz>
5085 * symbol-summary.h (function_summary): Move constructor
5086 implementation out of class declaration.
5087 (release): Likewise.
5088 (symtab_insertion): Likewise.
5089 (symtab_removal): Likewise.
5090 (symtab_duplication): Likewise.
5093 2018-06-08 Martin Liska <mliska@suse.cz>
5095 * Makefile.in: Remove support for MPX (macros, related functions,
5096 fields in cgraph_node, ...).
5097 * builtin-types.def (BT_BND): Likewise.
5098 (BT_FN_BND_CONST_PTR): Likewise.
5099 (BT_FN_CONST_PTR_BND): Likewise.
5100 (BT_FN_VOID_PTR_BND): Likewise.
5101 (BT_FN_BND_CONST_PTR_SIZE): Likewise.
5102 (BT_FN_VOID_CONST_PTR_BND_CONST_PTR): Likewise.
5103 * builtins.c (expand_builtin_memcpy_with_bounds): Likewise.
5104 (expand_builtin_mempcpy_with_bounds): Likewise.
5105 (expand_builtin_memset_with_bounds): Likewise.
5106 (expand_builtin_memset_args): Likewise.
5107 (std_expand_builtin_va_start): Likewise.
5108 (expand_builtin): Likewise.
5109 (expand_builtin_with_bounds): Likewise.
5110 * builtins.def (DEF_BUILTIN_CHKP): Likewise.
5111 (DEF_LIB_BUILTIN_CHKP): Likewise.
5112 (DEF_EXT_LIB_BUILTIN_CHKP): Likewise.
5113 (DEF_CHKP_BUILTIN): Likewise.
5114 (BUILT_IN_MEMCPY): Likewise.
5115 (BUILT_IN_MEMMOVE): Likewise.
5116 (BUILT_IN_MEMPCPY): Likewise.
5117 (BUILT_IN_MEMSET): Likewise.
5118 (BUILT_IN_STPCPY): Likewise.
5119 (BUILT_IN_STRCAT): Likewise.
5120 (BUILT_IN_STRCHR): Likewise.
5121 (BUILT_IN_STRCPY): Likewise.
5122 (BUILT_IN_STRLEN): Likewise.
5123 (BUILT_IN_MEMCPY_CHK): Likewise.
5124 (BUILT_IN_MEMMOVE_CHK): Likewise.
5125 (BUILT_IN_MEMPCPY_CHK): Likewise.
5126 (BUILT_IN_MEMSET_CHK): Likewise.
5127 (BUILT_IN_STPCPY_CHK): Likewise.
5128 (BUILT_IN_STRCAT_CHK): Likewise.
5129 (BUILT_IN_STRCPY_CHK): Likewise.
5130 * calls.c (store_bounds): Likewise.
5131 (emit_call_1): Likewise.
5132 (special_function_p): Likewise.
5133 (maybe_warn_nonstring_arg): Likewise.
5134 (initialize_argument_information): Likewise.
5135 (finalize_must_preallocate): Likewise.
5136 (compute_argument_addresses): Likewise.
5137 (expand_call): Likewise.
5138 * cfgexpand.c (expand_call_stmt): Likewise.
5139 (expand_return): Likewise.
5140 (expand_gimple_stmt_1): Likewise.
5141 (pass_expand::execute): Likewise.
5142 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
5143 (cgraph_node::remove): Likewise.
5144 (cgraph_node::dump): Likewise.
5145 (cgraph_node::verify_node): Likewise.
5146 * cgraph.h (chkp_function_instrumented_p): Likewise.
5147 (symtab_node::get_alias_target): Likewise.
5148 (cgraph_node::can_remove_if_no_direct_calls_and_refs_p): Likewise.
5149 (cgraph_local_p): Likewise.
5150 * cgraphbuild.c (cgraph_edge::rebuild_edges): Likewise.
5151 (cgraph_edge::rebuild_references): Likewise.
5152 * cgraphunit.c (varpool_node::finalize_decl): Likewise.
5153 (walk_polymorphic_call_targets): Likewise.
5154 (cgraph_node::expand_thunk): Likewise.
5155 (symbol_table::output_weakrefs): Likewise.
5156 * common/config/i386/i386-common.c (OPTION_MASK_ISA2_GENERAL_REGS_ONLY_UNSET): Likewise.
5157 (ix86_handle_option): Likewise.
5158 * config/i386/constraints.md: Likewise.
5159 * config/i386/i386-builtin-types.def (BND): Likewise.
5163 * config/i386/i386-builtin.def (BDESC_END): Likewise.
5164 (BDESC_FIRST): Likewise.
5166 * config/i386/i386-c.c (ix86_target_macros_internal): Likewise.
5167 * config/i386/i386-protos.h (ix86_bnd_prefixed_insn_p): Likewise.
5168 * config/i386/i386.c (enum reg_class): Likewise.
5169 (ix86_target_string): Likewise.
5170 (ix86_option_override_internal): Likewise.
5171 (ix86_conditional_register_usage): Likewise.
5172 (ix86_valid_target_attribute_inner_p): Likewise.
5173 (ix86_set_indirect_branch_type): Likewise.
5174 (ix86_set_current_function): Likewise.
5175 (ix86_function_arg_regno_p): Likewise.
5176 (init_cumulative_args): Likewise.
5177 (ix86_function_arg_advance): Likewise.
5178 (ix86_function_arg): Likewise.
5179 (ix86_pass_by_reference): Likewise.
5180 (ix86_function_value_regno_p): Likewise.
5181 (ix86_function_value_1): Likewise.
5182 (ix86_function_value_bounds): Likewise.
5183 (ix86_return_in_memory): Likewise.
5184 (ix86_setup_incoming_vararg_bounds): Likewise.
5185 (ix86_va_start): Likewise.
5186 (indirect_thunk_need_prefix): Likewise.
5187 (print_reg): Likewise.
5188 (ix86_print_operand): Likewise.
5189 (ix86_expand_call): Likewise.
5190 (ix86_output_function_return): Likewise.
5191 (reg_encoded_number): Likewise.
5192 (BDESC_VERIFYS): Likewise.
5193 (ix86_init_mpx_builtins): Likewise.
5194 (ix86_init_builtins): Likewise.
5195 (ix86_emit_cmove): Likewise.
5196 (ix86_emit_move_max): Likewise.
5197 (ix86_expand_builtin): Likewise.
5198 (ix86_builtin_mpx_function): Likewise.
5199 (ix86_get_arg_address_for_bt): Likewise.
5200 (ix86_load_bounds): Likewise.
5201 (ix86_store_bounds): Likewise.
5202 (ix86_load_returned_bounds): Likewise.
5203 (ix86_store_returned_bounds): Likewise.
5204 (ix86_class_likely_spilled_p): Likewise.
5205 (ix86_hard_regno_mode_ok): Likewise.
5206 (x86_order_regs_for_local_alloc): Likewise.
5207 (ix86_mitigate_rop): Likewise.
5208 (ix86_bnd_prefixed_insn_p): Likewise.
5209 (ix86_mpx_bound_mode): Likewise.
5210 (ix86_make_bounds_constant): Likewise.
5211 (ix86_initialize_bounds): Likewise.
5212 (TARGET_LOAD_BOUNDS_FOR_ARG): Likewise.
5213 (TARGET_STORE_BOUNDS_FOR_ARG): Likewise.
5214 (TARGET_LOAD_RETURNED_BOUNDS): Likewise.
5215 (TARGET_STORE_RETURNED_BOUNDS): Likewise.
5216 (TARGET_CHKP_BOUND_MODE): Likewise.
5217 (TARGET_BUILTIN_CHKP_FUNCTION): Likewise.
5218 (TARGET_CHKP_FUNCTION_VALUE_BOUNDS): Likewise.
5219 (TARGET_CHKP_MAKE_BOUNDS_CONSTANT): Likewise.
5220 (TARGET_CHKP_INITIALIZE_BOUNDS): Likewise.
5221 * config/i386/i386.h (TARGET_MPX): Likewise.
5222 (TARGET_MPX_P): Likewise.
5223 (VALID_BND_REG_MODE): Likewise.
5224 (FIRST_BND_REG): Likewise.
5225 (LAST_BND_REG): Likewise.
5226 (enum reg_class): Likewise.
5227 (BND_REG_P): Likewise.
5228 (BND_REGNO_P): Likewise.
5229 (BNDmode): Likewise.
5230 (ADJUST_INSN_LENGTH): Likewise.
5231 * config/i386/i386.md: Likewise.
5232 * config/i386/i386.opt: Likewise.
5233 * config/i386/linux-common.h (LIBMPX_LIBS): Likewise.
5234 (defined): Likewise.
5235 (LINK_MPX): Likewise.
5236 (MPX_SPEC): Likewise.
5237 (LIBMPX_SPEC): Likewise.
5238 (LIBMPXWRAPPERS_SPEC): Likewise.
5239 (CHKP_SPEC): Likewise.
5240 * config/i386/predicates.md: Likewise.
5241 * dbxout.c (dbxout_type): Likewise.
5242 * doc/extend.texi: Likewise.
5243 * doc/invoke.texi: Likewise.
5244 * doc/md.texi: Likewise.
5245 * doc/tm.texi: Likewise.
5246 * doc/tm.texi.in: Likewise.
5247 * dwarf2out.c (is_base_type): Likewise.
5248 (gen_formal_types_die): Likewise.
5249 (gen_subprogram_die): Likewise.
5250 (gen_type_die_with_usage): Likewise.
5251 (gen_decl_die): Likewise.
5252 (dwarf2out_late_global_decl): Likewise.
5253 * expr.c (expand_assignment): Likewise.
5254 (emit_storent_insn): Likewise.
5255 (store_expr_with_bounds): Likewise.
5256 (store_expr): Likewise.
5257 (expand_expr_real_1): Likewise.
5258 * expr.h (store_expr_with_bounds): Likewise.
5259 * function.c (use_register_for_decl): Likewise.
5260 (struct bounds_parm_data): Likewise.
5261 (assign_parms_augmented_arg_list): Likewise.
5262 (assign_parm_find_entry_rtl): Likewise.
5263 (assign_parm_is_stack_parm): Likewise.
5264 (assign_parm_load_bounds): Likewise.
5265 (assign_bounds): Likewise.
5266 (assign_parms): Likewise.
5267 (expand_function_start): Likewise.
5268 * gcc.c (CHKP_SPEC): Likewise.
5269 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
5270 * gimple-ssa-warn-restrict.c (builtin_access::builtin_access): Likewise.
5271 (wrestrict_dom_walker::check_call): Likewise.
5272 * gimple.c (gimple_build_call_from_tree): Likewise.
5273 * gimple.h (enum gf_mask): Likewise.
5274 (gimple_call_with_bounds_p): Likewise.
5275 (gimple_call_set_with_bounds): Likewise.
5276 * gimplify.c (gimplify_init_constructor): Likewise.
5277 * ipa-cp.c (initialize_node_lattices): Likewise.
5278 (propagate_constants_across_call): Likewise.
5279 (find_more_scalar_values_for_callers_subset): Likewise.
5280 * ipa-hsa.c (process_hsa_functions): Likewise.
5281 * ipa-icf-gimple.c (func_checker::compare_gimple_call): Likewise.
5282 * ipa-icf.c (sem_function::merge): Likewise.
5283 * ipa-inline.c (early_inliner): Likewise.
5284 * ipa-pure-const.c (warn_function_noreturn): Likewise.
5285 (warn_function_cold): Likewise.
5286 (propagate_pure_const): Likewise.
5287 * ipa-ref.h (enum GTY): Likewise.
5288 * ipa-split.c (find_retbnd): Likewise.
5289 (consider_split): Likewise.
5290 (split_function): Likewise.
5291 * ipa-visibility.c (cgraph_externally_visible_p): Likewise.
5292 * ipa.c (walk_polymorphic_call_targets): Likewise.
5293 (symbol_table::remove_unreachable_nodes): Likewise.
5294 (process_references): Likewise.
5295 (cgraph_build_static_cdtor_1): Likewise.
5296 * lto-cgraph.c (lto_output_node): Likewise.
5297 (output_refs): Likewise.
5298 (compute_ltrans_boundary): Likewise.
5299 (input_overwrite_node): Likewise.
5300 (input_node): Likewise.
5301 (input_cgraph_1): Likewise.
5302 * params.def (PARAM_CHKP_MAX_CTOR_SIZE): Likewise.
5303 * passes.c (pass_manager::execute_early_local_passes): Likewise.
5304 (class pass_chkp_instrumentation_passes): Likewise.
5305 (make_pass_chkp_instrumentation_passes): Likewise.
5306 * passes.def: Likewise.
5307 * rtl.h (struct GTY): Likewise.
5308 (CALL_EXPR_WITH_BOUNDS_P): Likewise.
5309 * stor-layout.c (layout_type): Likewise.
5310 * symtab.c: Likewise.
5311 * target.def: Likewise.
5312 * targhooks.c (default_chkp_bound_type): Likewise.
5313 (default_chkp_bound_mode): Likewise.
5314 (default_builtin_chkp_function): Likewise.
5315 (default_chkp_function_value_bounds): Likewise.
5316 (default_chkp_make_bounds_constant): Likewise.
5317 (default_chkp_initialize_bounds): Likewise.
5318 * targhooks.h (default_chkp_bound_type): Likewise.
5319 (default_chkp_bound_mode): Likewise.
5320 (default_builtin_chkp_function): Likewise.
5321 (default_chkp_function_value_bounds): Likewise.
5322 (default_chkp_make_bounds_constant): Likewise.
5323 (default_chkp_initialize_bounds): Likewise.
5324 * toplev.c (compile_file): Likewise.
5325 (process_options): Likewise.
5326 * tree-core.h (DEF_BUILTIN): Likewise.
5327 (DEF_BUILTIN_CHKP): Likewise.
5328 * tree-inline.c (declare_return_variable): Likewise.
5329 (remap_gimple_stmt): Likewise.
5330 (copy_bb): Likewise.
5331 (initialize_inlined_parameters): Likewise.
5332 (expand_call_inline): Likewise.
5333 * tree-pass.h (make_pass_ipa_chkp_versioning): Likewise.
5334 (make_pass_ipa_chkp_early_produce_thunks): Likewise.
5335 (make_pass_ipa_chkp_produce_thunks): Likewise.
5336 (make_pass_chkp): Likewise.
5337 (make_pass_chkp_opt): Likewise.
5338 (make_pass_chkp_instrumentation_passes): Likewise.
5339 * tree-pretty-print.c (dump_generic_node): Likewise.
5340 * tree-ssa-ccp.c (insert_clobber_before_stack_restore): Likewise.
5341 * tree-ssa-dce.c (propagate_necessity): Likewise.
5342 (eliminate_unnecessary_stmts): Likewise.
5343 * tree-ssa-pre.c (create_expression_by_pieces): Likewise.
5344 * tree-ssa-sccvn.c (copy_reference_ops_from_call): Likewise.
5345 * tree-ssa-sccvn.h: Likewise.
5346 * tree-ssa-strlen.c (get_string_length): Likewise.
5347 (valid_builtin_call): Likewise.
5348 (adjust_last_stmt): Likewise.
5349 (handle_builtin_strchr): Likewise.
5350 (handle_builtin_strcpy): Likewise.
5351 (handle_builtin_stxncpy): Likewise.
5352 (handle_builtin_memcpy): Likewise.
5353 (handle_builtin_strcat): Likewise.
5354 (strlen_check_and_optimize_stmt): Likewise.
5355 * tree-stdarg.c (expand_ifn_va_arg_1): Likewise.
5356 * tree-streamer-in.c: Likewise.
5357 * tree-streamer.c (record_common_node): Likewise.
5358 * tree.c (tree_code_size): Likewise.
5359 (wide_int_to_tree_1): Likewise.
5360 (type_contains_placeholder_1): Likewise.
5361 (build_common_tree_nodes): Likewise.
5362 * tree.def (POINTER_BOUNDS_TYPE): Likewise.
5363 * tree.h (POINTER_BOUNDS_TYPE_P): Likewise.
5364 (POINTER_BOUNDS_P): Likewise.
5365 (BOUNDED_TYPE_P): Likewise.
5366 (BOUNDED_P): Likewise.
5367 (CALL_WITH_BOUNDS_P): Likewise.
5368 (pointer_bounds_type_node): Likewise.
5369 * value-prof.c (gimple_ic): Likewise.
5370 * var-tracking.c (vt_add_function_parameters): Likewise.
5371 * varasm.c (make_decl_rtl): Likewise.
5372 (assemble_start_function): Likewise.
5373 (output_constant): Likewise.
5374 (maybe_assemble_visibility): Likewise.
5375 * varpool.c (ctor_for_folding): Likewise.
5376 * chkp-builtins.def: Remove.
5377 * ipa-chkp.c: Remove.
5378 * ipa-chkp.h: Remove.
5379 * rtl-chkp.c: Remove.
5380 * rtl-chkp.h: Remove.
5381 * tree-chkp-opt.c: Remove.
5382 * tree-chkp.c: Remove.
5383 * tree-chkp.h: Remove.
5385 2018-06-07 Carl Love <cel@us.ibm.com>
5387 * config/rs6000/vsx.md (vextract_fp_from_shorth,
5388 vextract_fp_from_shortl): Add BE support.
5390 2018-06-07 Paul Koning <ni1d@arrl.net>
5392 * compare-elim.c (try_merge_compare): Don't merge compare if
5393 address contains a side effect.
5394 (try_eliminate_compare): Likewise.
5396 2018-06-07 Olga Makhotina <olga.makhotina@intel.com>
5398 * config.gcc: Support "tremont".
5399 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "tremont".
5400 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
5402 * config/i386/i386.c (m_TREMONT): Define.
5403 (processor_target_table): Add "tremont".
5404 (PTA_TREMONT): Define.
5405 (ix86_lea_outperforms): Add TARGET_TREMONT.
5406 (get_builtin_code_for_version): Handle PROCESSOR_TREMONT.
5407 (fold_builtin_cpu): Add M_INTEL_TREMONT, replace M_INTEL_GOLDMONT
5408 and M_INTEL_GOLDMONT_PLUS.
5409 (fold_builtin_cpu): Add "tremont".
5410 (ix86_add_stmt_cost): Add TARGET_TREMONT.
5411 (ix86_option_override_internal): Add "tremont".
5412 * config/i386/i386.h (processor_costs): Define TARGET_TREMONT.
5413 (processor_type): Add PROCESSOR_TREMONT.
5414 * config/i386/x86-tune.def: Add m_TREMONT.
5415 * doc/invoke.texi: Add tremont as x86 -march=/-mtune= CPU type.
5417 2018-06-07 Jozef Lawrynowicz <jozef.l@mittosystems.com>
5419 * config/msp430/msp430.c (msp430_mcu_name): Set the "i" in the
5420 symbol defined for msp430i* devices to be lower case.
5422 2018-06-07 Richard Biener <rguenther@suse.de>
5424 * graphite-sese-to-poly.c (extract_affine): Avoid unneded
5425 wrapping. Properly wrap the result of a BIT_NOT_EXPR.
5426 Properly wrap signed arithmetic if overflow wraps.
5428 2018-06-07 Jakub Jelinek <jakub@redhat.com>
5430 PR tree-optimization/69615
5431 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If rhs2 is lhs
5432 of a cast from a same precision integral SSA_NAME in a bb dominated
5433 by first_bb, retry with rhs2 set to the rhs1 of the cast. Don't emit
5434 cast to utype if rhs2 has already a compatible type.
5436 2018-06-07 Richard Biener <rguenther@suse.de>
5438 PR tree-optimization/85935
5439 * graphite-scop-detection.c (find_params_in_bb): Analyze
5440 condition operands with respect to the correct loop. Assert
5441 the analysis doesn't fail.
5443 2018-06-04 Carl Love <cel@us.ibm.com>
5445 * config/rs6000/vsx.md (first_match_index_<mode>): Calculate index
5446 using natural element order. Use gen_lshrsi3 instead of gen_ashrsi3
5447 as it is slightly cheaper.
5448 (first_match_or_eos_index_<mode>):
5449 Calculate index using natural element order.
5450 (first_match_index_<mode>):
5451 Calculate index using natural element order.
5452 (first_match_or_eos_index_<mode>):
5453 Calculate index using natural order.
5454 (define_insn vclzlsbb): Change to define_insn vclzlsbb_<mode>.
5455 for BE and LE modes.
5456 * config/rs6000/rs6000-c.c: Rename P9V_BUILTIN_VCLZLSBB,
5457 P9V_BUILTIN_VCLZLSBB_V16QI.
5458 * config/rs6000/rs6000-builtin.def: Make VCLZLSBB mode
5461 2018-06-06 Kelvin Nilsen <kelvin@gcc.gnu.org>
5463 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
5464 indentation and line wrap for many prototypes. Add missing
5465 @smallexample directives around block of prototypes for vec_xl and
5468 2018-06-05 Michael Meissner <meissner@linux.ibm.com>
5470 * config/rs6000/rs6000.c (rs6000_passes_ieee128): New boolean to
5471 track if we pass or return IEEE 128-bit floating point.
5472 (ieee128_mangling_gcc_8_1): New boolean to say whether to generate
5473 C++ mangling that is compatible with GCC 8.1.
5474 (TARGET_ASM_GLOBALIZE_DECL_NAME): Override target hook.
5475 (init_cumulative_args): Note if we pass or return IEEE 128-bit
5476 floating point types.
5477 (rs6000_function_arg_advance_1): Likewise.
5478 (rs6000_mangle_type): Optionally generate mangled names that match
5479 what GCC 8.1 generated for IEEE 128-bit floating point types.
5480 (rs6000_globalize_decl_name): If we have an external function that
5481 passes or returns IEEE 128-bit types, generate a weak reference
5482 from the mangled name used in GCC 8.1 to the current mangled
5484 (rs6000_init_builtins): Make __ibm128 use the long double type if
5485 long double is IBM extended double. Make __float128 use the long
5486 double type if long double is IEEE 128-bit.
5489 * config/rs6000/rs6000-builtin.def (BU_IBM128_2): New helper
5490 macro for __ibm128 built-in functions.
5491 (PACK_IF): Add __ibm128 pack/unpack functions.
5492 (UNPACK_IF): Likewise.
5493 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Do not
5494 enable long double built-in functions if long double is IEEE
5495 128-bit floating point.
5496 (rs6000_invalid_builtin): Update long double built-in function
5498 (rs6000_expand_builtin): For PACK_IF and UNPACK_IF built-in
5499 functions, adjust the built-in function to use the long double
5500 built-in function if __ibm128 and long double are the same type.
5501 * doc/extend.texi (PowerPC builtins): Update documention for
5502 __builtin_{,un}pack_longdouble. Add documentation for
5503 __builtin_{,un}pack_ibm128.
5505 2018-06-06 Jim Wilson <jimw@sifive.com>
5507 * config/riscv/riscv.c (enum riscv_privilege_levels): New.
5508 (struct machine_function): New field interrupt_mode.
5509 (riscv_handle_type_attribute): New function. Add forward declaration.
5510 (riscv_attribute_table) <interrupt>: Use riscv_handle_type_attribute.
5511 (riscv_expand_epilogue): Check interrupt_mode field.
5512 (riscv_set_current_function): Check interrupt attribute args and
5513 set interrupt_mode field.
5514 * config/riscv/riscv.md (UNSPECV_SRET, UNSPECV_URET): New.
5515 (riscv_sret, riscv_uret): New.
5516 * doc/extend.texi (RISC-V Function Attributes) <interrupt>: Document
5517 new arguments to interrupt attribute.
5519 2018-06-06 Peter Bergner <bergner@vnet.ibm.com>
5522 * /config/rs6000/rs6000.h (ASM_CPU_SPEC): Add support for -mpower9.
5523 Don't handle -mcpu=power8 if -mpower9-vector is also used.
5525 2018-06-06 Kelvin Nilsen <kelvin@gcc.gnu.org>
5527 * config/rs6000/rs6000-builtin.def (VSX_BUILTIN_VEC_LD,
5528 VSX_BUILTIN_VEC_ST): Add comment to explain non-traditional uses.
5529 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
5530 several redundant entries.
5532 2018-06-06 David Malcolm <dmalcolm@redhat.com>
5534 * config/i386/i386-protos.h (ix86_expand_call): Strengthen return
5535 type from "rtx" to "rtx_insn *".
5536 * config/i386/i386.c (ix86_expand_split_stack_prologue): Likewise
5537 for local "call_insn", removing cast.
5538 (ix86_expand_call): Likewise, introducing a "call_insn" local.
5540 2018-06-06 Eric Botcazou <ebotcazou@adacore.com>
5542 PR tree-optimization/86066
5543 * gimple-ssa-store-merging.c (process_store): Do not bypass BIT_NOT_EXPR
5544 for BIT_INSERT_EXPR stores.
5546 2018-06-06 Richard Biener <rguenther@suse.de>
5548 PR tree-optimization/86062
5549 * tree-ssa-sccvn.c (vn_reference_lookup_3): Handle arbitrary
5550 component refs ontop
5551 of to be offsetted base.
5553 2018-06-06 Jozef Lawrynowicz <jozef.l@mittosystems.com>
5555 * config/msp430/msp430.c (msp430_attr): Allow interrupt handlers
5556 to be static and remove check on interrupt attribute name.
5558 2018-06-05 Kelvin Nilsen <kelvin@gcc.gnu.org>
5560 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Remove
5561 volatile qualifier from vec_lvsl and vec_lvsr argument prototypes.
5563 2018-06-05 Steve Ellcey <sellcey@cavium.com>
5566 * config/aarch64/aarch64-protos.h (aarch64_err_no_fpadvsimd): Remove
5568 * config/aarch64/aarch64-protos..c (aarch64_err_no_fpadvsimd):
5569 Remove second argument, change how error is called.
5570 (aarch64_layout_arg): Remove second argument from
5571 aarch64_err_no_fpadvsimd call.
5572 (aarch64_init_cumulative_args): Ditto.
5573 (aarch64_gimplify_va_arg_expr): Ditto.
5574 * config/aarch64/aarch64.md (mov<mode>): Ditto.
5576 2018-06-05 Uros Bizjak <ubizjak@gmail.com>
5578 * config/i386/i386.md (simple_return_indirect_internal): New expander.
5579 (*simple_return_indirect_internal<mode>): Rename from
5580 simple_return_indirect_internal. Use W mode iterator.
5581 (rstorssp): New expander.
5582 (*rstorssp<mode>): Rename from rstorssp. Use P mode iterator.
5583 (clrssbsy): New expander.
5584 (*clrssbsy<mode>): Rename from clrssbsy. Use P mode iterator.
5586 2018-06-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
5588 * config/arm/arm_cmse.h (cmse_nsfptr_create): Change typeof to
5590 (cmse_check_pointed_object): Likewise.
5592 2018-06-05 Martin Liska <mliska@suse.cz>
5594 PR gcov-profile/47618
5595 * doc/invoke.texi: Document how -fprofile-dir format
5598 2018-06-05 Richard Biener <rguenther@suse.de>
5600 * tree-cfgcleanup.c (cleanup_control_flow_pre): For edge
5601 removal pretend DOM info isn't available so we do not update
5602 it and only remove edges, not dominated blocks. Actually free
5603 DOM info in case we removed something. Remove unreachable blocks.
5604 (mfb_keep_latches): Work with either DOM info or marked backedges.
5605 (cleanup_tree_cfg_noloop): Do not remove unreachable blocks
5606 first. Mark backedges if DOM info isn't available.
5607 (Re-)compute DOM info after cleanup_control_flow_pre.
5609 2018-06-05 Richard Biener <rguenther@suse.de>
5611 * tree-cfg.c (struct locus_discrim_map): Store line, not location.
5612 (locus_discrim_hasher::hash): Adjust.
5613 (locus_discrim_hasher::equal): Likewise.
5614 (next_discriminator_for_locus): Work on line directly.
5615 (same_line_p): Pass in expanded locus1 as well.
5616 (assign_discriminators): Avoid redundant location expansions.
5618 2018-06-05 Richard Biener <rguenther@suse.de>
5620 PR tree-optimization/86046
5621 * tree-ssa.c (maybe_optimize_var): Clear DECL_GIMPLE_REG_P
5622 if required after clearing TREE_ADDRESSABLE.
5624 2018-06-05 Richard Biener <rguenther@suse.de>
5626 PR tree-optimization/86047
5627 * tree-ssa-loop.c (for_each_index): Glob handling of all
5628 decls and constants and really handle all of them.
5630 2018-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5633 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
5634 qualifier_void_pointer and qualifier_const_void_pointer.
5635 (arm_ldc_qualifiers, arm_stc_qualifiers): Use the above.
5636 (arm_init_builtins): Handle the above.
5637 * config/arm/arm_acle.h (__arm_cdp, __arm_ldc, __arm_ldcl, __arm_stc,
5638 __arm_stcl, __arm_mcr, __arm_cdp2, __arm_ldc2, __arm_ldcl2, __arm_stc2,
5639 __arm_stcl2,__arm_mcr2, __arm_mcrr, __arm_mcrr2): Remove return for
5642 2018-06-05 Martin Liska <mliska@suse.cz>
5644 * auto-profile.c (read_autofdo_file): Do not use
5645 gcov_ctr_summary struct.
5646 (afdo_callsite_hot_enough_for_early_inline): Likewise.
5647 * coverage.c (struct counts_entry): Likewise.
5648 (read_counts_file): Read just single summary entry.
5649 (get_coverage_counts): Use gcov_summary struct.
5650 * coverage.h (get_coverage_counts): Likewise.
5651 * gcov-dump.c (dump_working_sets): Likewise.
5652 (tag_summary): Dump just single summary.
5653 * gcov-io.c (gcov_write_summary): Write just histogram
5655 (gcov_read_summary): Read just single summary.
5656 (compute_working_sets): Use gcov_summary struct.
5657 * gcov-io.h (GCOV_TAG_SUMMARY_LENGTH): Remove usage
5658 of GCOV_COUNTERS_SUMMABLE.
5659 (GCOV_COUNTERS_SUMMABLE): Remove.
5660 (GCOV_FIRST_VALUE_COUNTER): Replace with
5661 GCOV_COUNTER_V_INTERVAL.
5662 (struct gcov_ctr_summary): Remove.
5663 (struct gcov_summary): Directly use fields of former
5665 (compute_working_sets): Use gcov_summary struct.
5666 * gcov.c (read_count_file): Do not use ctrs fields.
5667 * lto-cgraph.c (merge_profile_summaries): Use gcov_summary
5669 * lto-streamer.h (struct GTY): Make profile_info gcov_summary
5671 * profile.c: Likewise.
5672 * profile.h: Likewise.
5674 2018-06-05 Martin Liska <mliska@suse.cz>
5676 PR gcov-profile/84846
5677 * gcov.c (output_lines): Print working directory only
5678 in intermediate format.
5680 2018-06-05 Andreas Krebbel <krebbel@linux.ibm.com>
5682 * config/s390/s390-builtin-types.def: Add void function type.
5683 * config/s390/s390-builtins.def: Use the function type for the
5686 2018-06-04 Jim Wilson <jimw@sifive.com>
5688 * config/riscv/riscv-protos.h (riscv_expand_epilogue): Change bool arg
5690 * config/riscv/riscv.c (riscv_for_each_saved_reg): New args epilogue
5691 and maybe_eh_return. Change regno to unsigned int. Use new args to
5692 handle EH_RETURN_DATA_REGNO registers properly.
5693 (riscv_expand_prologue): Pass new args to riscv_for_each_saved_reg.
5694 (riscv_expand_epilogue): Update comment. Change argument name and
5695 type. Update code to use new name and type. Pass new args to
5696 riscv_for_each_saved_reg. Only use EH_RETURN_STACKADJ_RTX when
5698 * config/riscv/riscv.md (NORMAL_RETURN): New.
5699 (SIBCALL_RETURN, EXCEPTION_RETURN): New.
5700 (epilogue, sibcall_epilogue): Update riscv_expand_epilogue arg.
5701 (eh_return): Call gen_eh_return_internal and emit barrier.
5702 (eh_return_internal): Call riscv_expand_epilogue.
5704 2018-06-04 Eric Botcazou <ebotcazou@adacore.com>
5706 * gimple-ssa-store-merging.c (struct merged_store_group): Move up
5707 bit_insertion field and declare can_be_merged_into method.
5708 (merged_store_group::can_be_merged_into): New method.
5709 (imm_store_chain_info::coalesce_immediate): Call it to decide whether
5710 consecutive non-overlapping stores can be merged. Turn MEM_REF stores
5711 into BIT_INSERT_EXPR stores if the group contains a non-MEM_REF store.
5713 2018-06-04 Richard Biener <rguenther@suse.de>
5715 PR tree-optimization/85955
5716 * builtins.c (fold_builtin_sincos): Convert pointers to
5717 destination to appropriate type before dereferencing.
5719 2018-06-04 Segher Boessenkool <segher@kernel.crashing.org>
5721 * config/rs6000/rs6000.md (abs<mode>2 for FLOAT128): Handle IFmode.
5723 2018-06-04 Richard Sandiford <richard.sandiford@linaro.org>
5725 * expr.c (expand_expr_real_1): Force the operand into memory if
5726 its TYPE_MODE is BLKmode and if there is no integer mode for
5727 the number of bits being extracted.
5729 2018-06-04 Jakub Jelinek <jakub@redhat.com>
5733 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>_1):
5734 Use vptestnm rather than vptestm in (=Yc,v,C) variant.
5736 2018-06-04 Richard Biener <rguenther@suse.de>
5738 * tree-cfgcleanup.c (cleanup_tree_cfg_1): Fold into...
5739 (cleanup_tree_cfg_noloop): ... single caller. Do
5740 start_recording_case_labels later.
5742 2018-06-04 Sebastian Peryt <sebastian.peryt@intel.com>
5744 * config/i386/cldemoteintrin.h: Change define from _X86INTRIN_H_INCLUDED
5745 to _IMMINTRIN_H_INCLUDED.
5746 * config/i386/pconfigintrin.h: Ditto.
5747 * config/i386/waitpkgintrin.h: Ditto.
5748 * config/i386/immintrin.h: Add includes for sgxintrin.h,
5749 pconfigintrin.h, waitpkgintrin.h and cldemoteintrin.h.
5750 * config/i386/x86intrin.h: Remove includes for mintrin.h, xmmintrin.h,
5751 emmintrin.h, pmmintrin.h, tmmintrin.h, smmintrin.h, wmmintrin.h,
5752 bmiintrin.h, bmi2intrin.h, lzcntintrin.h, sgxintrin.h, pconfigintrin.h,
5753 waitpkgintrin.h and cldemoteintrin.h.
5755 2018-06-04 Richard Biener <rguenther@suse.de>
5757 PR tree-optimization/86038
5758 * tracer.c (find_best_successor): Check probability for
5759 being initialized, bail out if not.
5761 2018-06-04 Richard Earnshaw <rearnsha@arm.com>
5764 * config/arm/arm-cpus.in (ALL_QUIRKS): Add xscale feature to the list
5765 of bits to ignore when comparing architectures.
5767 2018-06-04 Jakub Jelinek <jakub@redhat.com>
5769 PR tree-optimization/69615
5770 * fold-const.c (merge_ranges): If range1 is - [x, x] and x is the
5771 maximum or minimum of the type, try to merge it also as if
5772 range1 is + [-, x - 1] or + [x + 1, -].
5775 * tree.c (inchash::add_expr): Handle IDENTIFIER_NODE.
5777 2018-06-03 Eric Botcazou <ebotcazou@adacore.com>
5779 PR tree-optimization/86034
5780 * gimple-ssa-store-merging.c (output_merged_store): Convert the RHS to
5781 the unsigned bitfield type in a bit insertion sequence if it does not
5782 have a larger precision than the bitfield size.
5783 (process_store): Also bypass widening conversions for BIT_INSERT_EXPR.
5785 2018-06-03 Kito Cheng <kito.cheng@gmail.com>
5787 * config/nds32/nds32-peephole2.md: Add new patterns for code size.
5789 2018-06-03 Chung-Ju Wu <jasonwucj@gmail.com>
5791 * config/nds32/nds32-opts.h (nds32_arch_type): Add ARCH_V3J.
5792 * config/nds32/nds32.c (nds32_option_override): Consider ARCH_V3J.
5793 * config/nds32/nds32.h (TARGET_ISA_V3): Add ARCH_V3J checking.
5794 * config/nds32/nds32.opt (march): Add enum value Value(ARCH_V3J).
5796 2018-06-03 Chung-Ju Wu <jasonwucj@gmail.com>
5798 * common/config/nds32/nds32-common.c (nds32_option_optimization_table):
5799 Disable -fdelete-null-pointer-checks for ELF toolchain.
5801 2018-06-02 Chung-Ju Wu <jasonwucj@gmail.com>
5802 Kito Cheng <kito.cheng@gmail.com>
5804 * config.gcc (nds32*): Use nds32-linux.opt and nds32-elf.opt.
5805 (nds32le-*-*, nds32be-*-*): Integrate checking process.
5806 (nds32*-*-*): Add glibc and uclibc conditions.
5807 * common/config/nds32/nds32-common.c (nds32_except_unwind_info): New.
5808 (TARGET_EXCEPT_UNWIND_INFO): Define.
5809 * config/nds32/elf.h: New file.
5810 * config/nds32/linux.h: New file.
5811 * config/nds32/nds32-elf.opt: New file.
5812 * config/nds32/nds32-linux.opt: New file.
5813 * config/nds32/nds32-fp-as-gp.c
5814 (pass_nds32_fp_as_gp::gate): Consider TARGET_LINUX_ABI.
5815 * config/nds32/nds32.c (nds32_conditional_register_usage): Consider
5817 (nds32_asm_file_end): Ditto.
5818 (nds32_print_operand): Ditto.
5819 (nds32_insert_attributes): Ditto.
5820 (nds32_init_libfuncs): New function.
5821 (TARGET_HAVE_TLS): Define.
5822 (TARGET_INIT_LIBFUNCS): Define.
5823 * config/nds32/nds32.h (TARGET_DEFAULT_RELAX): Apply different relax
5825 (TARGET_ELF): Apply different mcmodel setting.
5826 (LINK_SPEC, LIB_SPEC, STARTFILE_SPEC, ENDFILE_SPEC): The content has
5827 been migrated into elf.h and linux.h files.
5828 * config/nds32/nds32.md (add_pc): Consider TARGET_LINUX_ABI.
5829 * config/nds32/nds32.opt (mvh): Consider TARGET_LINUX_ABI.
5830 (mcmodel): The content has been migrated into nds32-elf.opt and
5831 nds32-linux.opt files.
5832 * config/nds32/t-elf: New file.
5833 * config/nds32/t-linux: New file.
5835 2018-06-02 Chung-Ju Wu <jasonwucj@gmail.com>
5836 Shiva Chen <shiva0217@gmail.com>
5838 * config/nds32/constants.md (unspec_volatile_element): Add
5839 UNSPEC_VOLATILE_OMIT_FP_BEGIN and UNSPEC_VOLATILE_OMIT_FP_END.
5840 * config/nds32/nds32-fp-as-gp.c: New implementation of fp_as_gp
5842 * config/nds32/nds32-protos.h (nds32_naked_function_p): Declare.
5843 (make_pass_nds32_fp_as_gp): Declare.
5844 * config/nds32/nds32.c (nds32_register_passes): Add fp_as_gp as one
5846 (nds32_asm_function_end_prologue): Remove unused asm output.
5847 (nds32_asm_function_begin_epilogue): Remove unused asm output.
5848 (nds32_asm_file_start): Output necessary fp_as_gp information.
5849 (nds32_option_override): Adjust register usage.
5850 (nds32_expand_prologue): Consider fp_as_gp situation.
5851 (nds32_expand_prologue_v3push): Consider fp_as_gp situation.
5852 * config/nds32/nds32.md (prologue): Check fp_as_gp_p and naked_p.
5855 (simple_return): Ditto.
5856 (omit_fp_begin): Output special directive for fp_as_gp.
5857 (omit_fp_end): Output special directive for fp_as_gp.
5858 * config/nds32/nds32.opt (mfp-as-gp, mno-fp-as-gp, mforce-fp-as-gp,
5859 mforbid-fp-as-gp): New options.
5861 2018-06-01 Mark Wielaard <mark@klomp.org>
5863 * dwarf2out.c (dwarf2out_finish): Remove generation of
5864 DW_AT_loclists_base.
5866 2018-06-01 Eric Botcazou <ebotcazou@adacore.com>
5868 * gimple-ssa-store-merging.c: Include gimple-fold.h.
5869 (struct store_immediate_info): Document BIT_INSERT_EXPR stores.
5870 (struct merged_store_group): Add bit_insertion field.
5871 (dump_char_array): Use standard hexadecimal format.
5872 (merged_store_group::merged_store_group): Set bit_insertion to false.
5873 (merged_store_group::apply_stores): Use optimal buffer size. Deal
5874 with BIT_INSERT_EXPR stores. Move up code updating the mask and
5875 also print the mask in the dump file.
5876 (pass_store_merging::gate): Minor tweak.
5877 (imm_store_chain_info::coalesce_immediate): Fix wrong association
5878 of stores with groups in dump. Allow coalescing of BIT_INSERT_EXPR
5879 stores with INTEGER_CST stores.
5880 (count_multiple_uses) <BIT_INSERT_EXPR>: New case.
5881 (imm_store_chain_info::output_merged_store): Add try_bitpos variable
5882 and use it throughout. Generate bit insertion sequences if need be.
5883 (pass_store_merging::process_store): Remove redundant condition.
5884 Record stores from a SSA name to a bit-field with BIT_INSERT_EXPR.
5886 2018-06-01 Segher Boessenkool <segher@kernel.crashing.org>
5888 * config/rs6000/rs6000.c (rs6000_mangle_type): Change the mangling of
5889 the 128-bit floating point types. Fix function comment.
5891 2018-06-01 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5893 * config/aarch64/aarch64-simd.md
5894 (aarch64_simd_vec_unpack<su>_lo_<mode>): Use UXTL and SXTL assembler
5896 (aarch64_simd_vec_unpack<su>_hi_<mode>): Use UXTL2 and SXTL2 assembler
5899 2018-06-01 Richard Sandiford <richard.sandiford@linaro.org>
5901 PR tree-optimization/85989
5902 * gimple-ssa-backprop.c (backprop::m_visited_phis): New member
5904 (backprop::intersect_uses): Check it when deciding whether this
5905 is a backedge reference.
5906 (backprop::process_block): Add each phi to m_visited_phis
5907 after visiting it, then clear it at the end.
5909 2018-06-01 Richard Biener <rguenther@suse.de>
5911 * tree-vectorizer.h (vect_dr_stmt): New function.
5912 (vect_get_load_cost): Adjust.
5913 (vect_get_store_cost): Likewise.
5914 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence):
5915 Use vect_dr_stmt instead of DR_SMTT.
5916 (vect_record_base_alignments): Likewise.
5917 (vect_calculate_target_alignment): Likewise.
5918 (vect_compute_data_ref_alignment): Likewise and make static.
5919 (vect_update_misalignment_for_peel): Likewise.
5920 (vect_verify_datarefs_alignment): Likewise.
5921 (vector_alignment_reachable_p): Likewise.
5922 (vect_get_data_access_cost): Likewise. Pass down
5923 vinfo to vect_get_load_cost/vect_get_store_cost instead of DR.
5924 (vect_get_peeling_costs_all_drs): Likewise.
5925 (vect_peeling_hash_get_lowest_cost): Likewise.
5926 (vect_enhance_data_refs_alignment): Likewise.
5927 (vect_find_same_alignment_drs): Likewise.
5928 (vect_analyze_data_refs_alignment): Likewise.
5929 (vect_analyze_group_access_1): Likewise.
5930 (vect_analyze_group_access): Likewise.
5931 (vect_analyze_data_ref_access): Likewise.
5932 (vect_analyze_data_ref_accesses): Likewise.
5933 (vect_vfa_segment_size): Likewise.
5934 (vect_small_gap_p): Likewise.
5935 (vectorizable_with_step_bound_p): Likewise.
5936 (vect_prune_runtime_alias_test_list): Likewise.
5937 (vect_analyze_data_refs): Likewise.
5938 (vect_supportable_dr_alignment): Likewise.
5939 * tree-vect-loop-manip.c (get_misalign_in_elems): Likewise.
5940 (vect_gen_prolog_loop_niters): Likewise.
5941 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
5942 * tree-vect-patterns.c (vect_recog_bool_pattern): Do not
5944 (vect_recog_mask_conversion_pattern): Likewise.
5945 (vect_try_gather_scatter_pattern): Likewise.
5946 * tree-vect-stmts.c (vect_model_store_cost): Pass stmt_info
5947 to vect_get_store_cost.
5948 (vect_get_store_cost): Get stmt_info instead of DR.
5949 (vect_model_load_cost): Pass stmt_info to vect_get_load_cost.
5950 (vect_get_load_cost): Get stmt_info instead of DR.
5952 2018-06-01 Richard Biener <rguenther@suse.de>
5955 * gimple-fold.c (var_decl_component_p): Also allow offsetted
5956 vars wrapped in MEM_REFs.
5958 2018-06-01 Richard Sandiford <richard.sandiford@linaro.org>
5960 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
5961 Fix subreg tests so that we only return a choice between
5962 GENERAL_REGS and FP_REGS if the original classes included both.
5964 2018-06-01 Richard Biener <rguenther@suse.de>
5967 * tree-ssa-structalias.c (get_function_part_constraint):
5968 Handle NULL fi->decl.
5969 (find_func_aliases_for_call): Properly handle indirect
5970 fi from direct call.
5971 (find_func_clobbers): Likewise.
5972 (ipa_pta_execute): Likewise.
5973 (create_variable_info_for): For functions that are ifunc_resolver
5974 resolve to a varinfo that contains the result of the resolver call.
5975 (associate_varinfo_to_alias): Do not treat ifunc resolvers as
5978 2018-05-31 Michael Collison <michael.collison@arm.com>
5980 * config/aarch64/aarch64.md:
5981 (*fix_to_zero_extenddfdi2): New pattern.
5982 * gcc.target/aarch64/fix_extend1.c: New testcase.
5984 2018-05-31 Qing Zhao <qing.zhao@oracle.com>
5988 * builtins.c (expand_builtin): Add the handling of BUILT_IN_STRCMP_EQ
5989 and BUILT_IN_STRNCMP_EQ.
5990 * builtins.def: Add new builtins BUILT_IN_STRCMP_EQ and
5991 BUILT_IN_STRNCMP_EQ.
5992 * gimple-fold.c (gimple_fold_builtin_string_compare): Add the
5993 handling of BUILTIN_IN_STRCMP_EQ and BUILT_IN_STRNCMP_EQ.
5994 (gimple_fold_builtin): Likewise.
5995 * tree-ssa-strlen.c (compute_string_length): New function.
5996 (determine_min_obsize): New function.
5997 (handle_builtin_string_cmp): New function to handle calls to
5998 string compare functions.
5999 (strlen_optimize_stmt): Add handling to builtin string compare
6001 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
6002 Add the handling of BUILT_IN_STRCMP_EQ and BUILT_IN_STRNCMP_EQ.
6003 * tree.c (build_common_builtin_nodes): Add new defines of
6004 BUILT_IN_STRNCMP_EQ and BUILT_IN_STRCMP_EQ.
6006 2018-05-31 Jakub Jelinek <jakub@redhat.com>
6009 * bb-reorder.c (pass_partition_blocks::gate): Return false for
6010 functions with naked attribute.
6012 2018-05-31 Uros Bizjak <ubizjak@gmail.com>
6014 * config/i386/sse.md (avx_vec_concat<mode>):
6015 Substitute concat_tg_mode mode attribute with xtg_mode.
6016 (<mask_codefor>avx512dq_broadcast<mode><mask_name>_1): Ditto.
6017 (concat_tg_mode): Remove mode attribute.
6019 2018-05-31 Martin Sebor <msebor@redhat.com>
6022 * calls.c (alloc_max_size): Correct a logic error/typo.
6023 Treat excessive arguments as infinite. Warn for invalid arguments.
6024 * doc/invoke.texi (-Walloc-size-larger-than): Update.
6026 2018-05-31 H.J. Lu <hongjiu.lu@intel.com>
6029 * config/i386/x86-tune.def: Re-enable partial_reg_dependency
6030 and movx for Haswell.
6032 2018-05-31 Chung-Lin Tang <cltang@codesourcery.com>
6033 Cesar Philippidis <cesar@codesourcery.com>
6036 * gimplify.c (gimplify_adjust_omp_clauses): Add 'remove = true'
6037 when emitting error on private/firstprivate reductions.
6038 * omp-low.c (lower_omp_target): Avoid reference-type processing
6039 on pointers for firstprivate clause.
6041 2018-05-31 Sameera Deshpande <sameera.deshpande@linaro.org>
6043 * config/aarch64/aarch64-simd-builtins.def (ld1x3): New.
6046 * config/aarch64/aarch64-simd.md
6047 (aarch64_ld1x3<VALLDIF:mode>): New pattern.
6048 (aarch64_ld1_x3_<mode>): Likewise
6049 (aarch64_st1x2<VALLDIF:mode>): Likewise
6050 (aarch64_st1_x2_<mode>): Likewise
6051 (aarch64_st1x3<VALLDIF:mode>): Likewise
6052 (aarch64_st1_x3_<mode>): Likewise
6053 * config/aarch64/arm_neon.h (vld1_u8_x3): New function.
6054 (vld1_s8_x3): Likewise.
6055 (vld1_u16_x3): Likewise.
6056 (vld1_s16_x3): Likewise.
6057 (vld1_u32_x3): Likewise.
6058 (vld1_s32_x3): Likewise.
6059 (vld1_u64_x3): Likewise.
6060 (vld1_s64_x3): Likewise.
6061 (vld1_f16_x3): Likewise.
6062 (vld1_f32_x3): Likewise.
6063 (vld1_f64_x3): Likewise.
6064 (vld1_p8_x3): Likewise.
6065 (vld1_p16_x3): Likewise.
6066 (vld1_p64_x3): Likewise.
6067 (vld1q_u8_x3): Likewise.
6068 (vld1q_s8_x3): Likewise.
6069 (vld1q_u16_x3): Likewise.
6070 (vld1q_s16_x3): Likewise.
6071 (vld1q_u32_x3): Likewise.
6072 (vld1q_s32_x3): Likewise.
6073 (vld1q_u64_x3): Likewise.
6074 (vld1q_s64_x3): Likewise.
6075 (vld1q_f16_x3): Likewise.
6076 (vld1q_f32_x3): Likewise.
6077 (vld1q_f64_x3): Likewise.
6078 (vld1q_p8_x3): Likewise.
6079 (vld1q_p16_x3): Likewise.
6080 (vld1q_p64_x3): Likewise.
6081 (vst1_s64_x2): Likewise.
6082 (vst1_u64_x2): Likewise.
6083 (vst1_f64_x2): Likewise.
6084 (vst1_s8_x2): Likewise.
6085 (vst1_p8_x2): Likewise.
6086 (vst1_s16_x2): Likewise.
6087 (vst1_p16_x2): Likewise.
6088 (vst1_s32_x2): Likewise.
6089 (vst1_u8_x2): Likewise.
6090 (vst1_u16_x2): Likewise.
6091 (vst1_u32_x2): Likewise.
6092 (vst1_f16_x2): Likewise.
6093 (vst1_f32_x2): Likewise.
6094 (vst1_p64_x2): Likewise.
6095 (vst1q_s8_x2): Likewise.
6096 (vst1q_p8_x2): Likewise.
6097 (vst1q_s16_x2): Likewise.
6098 (vst1q_p16_x2): Likewise.
6099 (vst1q_s32_x2): Likewise.
6100 (vst1q_s64_x2): Likewise.
6101 (vst1q_u8_x2): Likewise.
6102 (vst1q_u16_x2): Likewise.
6103 (vst1q_u32_x2): Likewise.
6104 (vst1q_u64_x2): Likewise.
6105 (vst1q_f16_x2): Likewise.
6106 (vst1q_f32_x2): Likewise.
6107 (vst1q_f64_x2): Likewise.
6108 (vst1q_p64_x2): Likewise.
6109 (vst1_s64_x3): Likewise.
6110 (vst1_u64_x3): Likewise.
6111 (vst1_f64_x3): Likewise.
6112 (vst1_s8_x3): Likewise.
6113 (vst1_p8_x3): Likewise.
6114 (vst1_s16_x3): Likewise.
6115 (vst1_p16_x3): Likewise.
6116 (vst1_s32_x3): Likewise.
6117 (vst1_u8_x3): Likewise.
6118 (vst1_u16_x3): Likewise.
6119 (vst1_u32_x3): Likewise.
6120 (vst1_f16_x3): Likewise.
6121 (vst1_f32_x3): Likewise.
6122 (vst1_p64_x3): Likewise.
6123 (vst1q_s8_x3): Likewise.
6124 (vst1q_p8_x3): Likewise.
6125 (vst1q_s16_x3): Likewise.
6126 (vst1q_p16_x3): Likewise.
6127 (vst1q_s32_x3): Likewise.
6128 (vst1q_s64_x3): Likewise.
6129 (vst1q_u8_x3): Likewise.
6130 (vst1q_u16_x3): Likewise.
6131 (vst1q_u32_x3): Likewise.
6132 (vst1q_u64_x3): Likewise.
6133 (vst1q_f16_x3): Likewise.
6134 (vst1q_f32_x3): Likewise.
6135 (vst1q_f64_x3): Likewise.
6136 (vst1q_p64_x3): Likewise.
6138 2018-05-30 Jozef Lawrynowicz <jozef.l@mittosystems.com>
6140 * config/msp430/msp430.c (msp430_output_labelref): Prepend
6141 user_label_prefix to name.
6143 * tree-core.h: Update comment about the format of NAME string
6144 passed to handler in attribute_spec.
6146 * config/msp430/msp430.md: Remove erroneous subreg expression from
6147 zero_extendqisi2 insn pattern. Remove msp430x ISA restriction on
6148 zero_extend{q,h}isi2.
6150 2018-05-30 Borislav Petkov <bp@suse.de>
6152 * doc/extend.texi: Document some architecture specific
6153 constraints and sort entries.
6155 2018-05-30 Martin Sebor <msebor@redhat.com>
6158 * builtins.c (expand_builtin_stpcpy_1): New function.
6159 (expand_builtin_stpcpy): Call it, and call maybe_warn_nonstring_arg
6160 only if the former succeeds.
6162 2018-05-31 Sameera Deshpande <sameera.deshpande@linaro.org>
6164 * config/aarch64/aarch64-cores.def (saphira) : Add support for ARMv8.4
6167 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
6169 * doc/invoke.texi (-flinker-output): Document
6171 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
6173 * passes.c (ipa_write_summaries): Only modify statements if body
6175 * cgraphunit.c (ipa_passes): Also produce intermeidate code when
6176 incrementally linking.
6177 (ipa_passes): Likewise.
6178 * lto-cgraph.c (lto_output_node): When incrementally linking do not
6179 pass down resolution info.
6180 * common.opt (flag_incremental_link): Update info.
6181 * gcc.c (plugin specs): Turn flinker-output=* to
6182 -plugin-opt=-linker-output-known
6183 * toplev.c (compile_file): Also cut compilation when doing incremental
6185 * flag-types. (enum lto_partition_model): Add
6186 LTO_LINKER_OUTPUT_NOLTOREL.
6187 (invoke.texi): Add -flinker-output docs.
6188 * ipa.c (symbol_table::remove_unreachable_nodes): Handle LTO incremental
6189 link same way as WPA; do not stream in dead initializers.
6191 * dwarf2out.c (dwarf2out_die_ref_for_decl,
6192 darf2out_register_external_decl): Support incremental link.
6194 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
6196 * lto-opts.c (lto_write_options): Skip OPT_dumpdir, OPT_fresolution_.
6198 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
6200 * lto-wrapper.c (debug_objcopy): Add rename parameter; pass
6201 it down to simple_object_copy_lto_debug_sections.
6202 (run_gcc): Determine incremental LTO link time and configure
6203 lto1 into non-wpa mode, disable renaming of debug sections.
6205 2018-05-30 Kelvin Nilsen <kelvin@gcc.gnu.org>
6207 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Remove
6208 descriptions of various incorrectly documented functions.
6210 2018-05-30 Andre Vieira <andre.simoesdiasvieira@arm.com>
6213 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Make
6214 address check not strict.
6216 2018-05-30 Richard Biener <rguenther@suse.de>
6218 PR tree-optimization/85964
6219 * tracer.c (better_p): Drop initialized count check, we only
6220 call the function with initialized counts now.
6221 (find_best_successor): Do find a best edge if one
6222 has uninitialized count.
6223 (find_best_predecessor): Likewise. Do BB frequency check only
6224 if count is initialized.
6226 2017-05-30 Jackson Woodruff <jackson.woodruff@arm.com>
6228 * config/aarch64/aarch64.c (aarch64_host_wide_int_compare): New.
6229 (aarch64_ldrstr_offset_compare): New.
6230 (aarch64_operands_adjust_ok_for_ldpstp): Update to consider all
6231 load/store orderings.
6232 (aarch64_gen_adjusted_ldpstp): Likewise.
6234 2018-05-30 Wilco Dijkstra <wdijkstr@arm.com>
6236 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
6237 Check for subset of GENERAL_REGS and FP_REGS.
6238 * config/aarch64/aarch64-simd.md (aarch64_get_lane): Increase cost of
6241 2018-05-30 Richard Sandiford <richard.sandiford@linaro.org>
6243 * alias.c (adjust_offset_for_component_ref): Use poly_int_tree_p
6244 and wi::to_poly_offset. Add the current offset and then check
6245 whether the sum fits, rather than using an unchecked addition of
6246 a checked term. Check for a shwi rather than a uhwi.
6247 * expr.c (get_bit_range): Use tree_to_poly_uint64.
6248 (store_constructor): Use poly_int_tree_p.
6249 (expand_expr_real_1): Likewise.
6250 * function.c (assign_temp): Likewise.
6251 * fold-const.c (const_binop): Use poly_int_tree_p and
6253 (fold_indirect_ref_1): Likewise. Use multiple_p to attempt an exact
6255 * ipa-icf-gimple.c (func_checker::compare_operand): Use
6256 to_poly_offset for MEM offsets.
6257 * ipa-icf.c (sem_variable::equals): Likewise.
6258 * stor-layout.c (compute_record_mode): Use poly_int_tree_p.
6259 * tree-ssa-sccvn.c (ao_ref_init_from_vn_reference): Use
6260 wi::to_poly_offset for BIT_FIELD_REF offsets.
6261 (vn_reference_maybe_forwprop_address): Use poly_int_tree_p and
6263 * var-tracking.c (emit_note_insn_var_location): Use
6264 tree_to_poly_uint64.
6266 2018-05-29 Jim Wilson <jimw@sifive.com>
6268 * config/riscv/riscv.c (riscv_interrupt_type): Fix comment typo.
6270 2018-05-29 Uros Bizjak <ubizjak@gmail.com>
6273 * config/i386/i386.md (l<rounding_insn><MODEF:mode><SWI48:mode>2):
6274 Enable for TARGET_SSE4_1 and generate rounds{s,d} and cvtts{s,d}2si{,q}
6276 (sse4_1_round<mode>2): Use nonimmediate_operand
6277 for operand 1 predicate.
6279 2018-05-29 Martin Sebor <msebor@redhat.com>
6280 Richard Biener <rguenther@suse.de>
6283 * calls.c (get_size_range): Call determine_value_range instead
6284 of get_value_range..
6285 * tree-vrp.h (determine_value_range): Declared new function.
6286 * tree-vrp.c (determine_value_range_1, determine_value_range): New.
6288 2018-05-29 Richard Biener <rguenther@suse.de>
6290 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Make
6291 sure to use non-pattern stmts for get_earlier_stmt arguments.
6292 * tree-vectorizer.h (get_earlier_stmt): Assert we do not get
6293 called on pattern stmts.
6294 (get_later_stmt): Likewise.
6296 2018-05-29 Martin Liska <mliska@suse.cz>
6298 PR gcov-profile/85759
6299 * doc/gcov.texi: Document GCOV_ERROR_FILE and GCOV_EXIT_AT_ERROR
6302 2018-05-29 Jakub Jelinek <jakub@redhat.com>
6304 * tree-cfg.c (verify_gimple_assign_unary): Add checking for
6306 (verify_gimple_assign_binary): Check TYPE_VECTOR_SUBPARTS for
6310 * tree.def (VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR,
6311 VEC_PACK_FLOAT_EXPR): New tree codes.
6312 * tree-pretty-print.c (op_code_prio): Handle
6313 VEC_UNPACK_FIX_TRUNC_HI_EXPR and VEC_UNPACK_FIX_TRUNC_LO_EXPR.
6314 (dump_generic_node): Handle VEC_UNPACK_FIX_TRUNC_HI_EXPR,
6315 VEC_UNPACK_FIX_TRUNC_LO_EXPR and VEC_PACK_FLOAT_EXPR.
6316 * tree-inline.c (estimate_operator_cost): Likewise.
6317 * gimple-pretty-print.c (dump_binary_rhs): Handle VEC_PACK_FLOAT_EXPR.
6318 * fold-const.c (const_binop): Likewise.
6319 (const_unop): Handle VEC_UNPACK_FIX_TRUNC_HI_EXPR and
6320 VEC_UNPACK_FIX_TRUNC_LO_EXPR.
6321 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
6322 (verify_gimple_assign_binary): Handle VEC_PACK_FLOAT_EXPR.
6323 * cfgexpand.c (expand_debug_expr): Handle VEC_UNPACK_FIX_TRUNC_HI_EXPR,
6324 VEC_UNPACK_FIX_TRUNC_LO_EXPR and VEC_PACK_FLOAT_EXPR.
6325 * expr.c (expand_expr_real_2): Likewise.
6326 * optabs.def (vec_packs_float_optab, vec_packu_float_optab,
6327 vec_unpack_sfix_trunc_hi_optab, vec_unpack_sfix_trunc_lo_optab,
6328 vec_unpack_ufix_trunc_hi_optab, vec_unpack_ufix_trunc_lo_optab): New
6330 * optabs.c (expand_widen_pattern_expr): For
6331 VEC_UNPACK_FIX_TRUNC_HI_EXPR and VEC_UNPACK_FIX_TRUNC_LO_EXPR use
6332 sign from result type rather than operand's type.
6333 (expand_binop_directly): For vec_packu_float_optab and
6334 vec_packs_float_optab allow result type to be different from operand's
6336 * optabs-tree.c (optab_for_tree_code): Handle
6337 VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR and
6338 VEC_PACK_FLOAT_EXPR. Formatting fixes.
6339 * tree-vect-generic.c (expand_vector_operations_1): Handle
6340 VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR and
6341 VEC_PACK_FLOAT_EXPR.
6342 * tree-vect-stmts.c (supportable_widening_operation): Handle
6344 (supportable_narrowing_operation): Handle FLOAT_EXPR.
6345 * config/i386/i386.md (fixprefix, floatprefix): New code attributes.
6346 * config/i386/sse.md (*float<floatunssuffix>v2div2sf2): Rename to ...
6347 (float<floatunssuffix>v2div2sf2): ... this. Formatting fix.
6348 (vpckfloat_concat_mode, vpckfloat_temp_mode, vpckfloat_op_mode): New
6350 (vec_pack<floatprefix>_float_<mode>): New expander.
6351 (vunpckfixt_mode, vunpckfixt_model, vunpckfixt_extract_mode): New mode
6353 (vec_unpack_<fixprefix>fix_trunc_lo_<mode>,
6354 vec_unpack_<fixprefix>fix_trunc_hi_<mode>): New expanders.
6355 * doc/md.texi (vec_packs_float_@var{m}, vec_packu_float_@var{m},
6356 vec_unpack_sfix_trunc_hi_@var{m}, vec_unpack_sfix_trunc_lo_@var{m},
6357 vec_unpack_ufix_trunc_hi_@var{m}, vec_unpack_ufix_trunc_lo_@var{m}):
6359 * doc/generic.texi (VEC_UNPACK_FLOAT_HI_EXPR,
6360 VEC_UNPACK_FLOAT_LO_EXPR): Fix pasto in description.
6361 (VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR,
6362 VEC_PACK_FLOAT_EXPR): Document.
6364 2018-05-29 Richard Biener <rguenther@suse.de>
6366 * tree-vectorizer.h (struct vec_info): Add stmt_vec_infos
6368 (stmt_vec_info_vec): Make pointer.
6369 (init_stmt_vec_info_vec): Remove.
6370 (free_stmt_vec_info_vec): Likewise.
6371 (set_stmt_vec_info_vec): New function.
6372 (free_stmt_vec_infos): Likewise.
6373 (vinfo_for_stmt): Adjust for stmt_vec_info_vec indirection.
6374 (set_vinfo_for_stmt): Likewise.
6375 (get_earlier_stmt): Likewise.
6376 (get_later_stmt): Likewise.
6377 * tree-vectorizer.c (stmt_vec_info_vec): Make pointer.
6378 (vec_info::vec_info): Allocate stmt_vec_infos and set the global.
6379 (vec_info::~vec_info): Free stmt_vec_infos.
6380 (vectorize_loops): Set the global stmt_vec_info_vec to NULL.
6381 Remove old init_stmt_vec_info_vec/free_stmt_vec_info_vec calls.
6382 (pass_slp_vectorize::execute): Likewise.
6383 * tree-vect-stmts.c (init_stmt_vec_info_vec): Remove.
6384 (free_stmt_vec_info_vec): Likewise.
6385 (set_stmt_vec_info_vec): New function.
6386 (free_stmt_vec_infos): Likewise.
6387 * tree-vect-loop.c (_loop_vec_info::~_loop_vec_info): Set
6388 the global stmt_vec_info_vec.
6389 * tree-parloops.c (gather_scalar_reductions): Use
6390 set_stmt_vec_info_vec/free_stmt_vec_infos and maintain a local
6393 2018-05-29 Richard Biener <rguenther@suse.de>
6395 * dominance.c (iterate_fix_dominators): Push/pop TV_DOMINANCE.
6397 2018-05-29 Martin Liska <mliska@suse.cz>
6398 David Malcolm <dmalcolm@redhat.com>
6400 * vec.c (test_reverse): New.
6401 (vec_c_tests): Add new test.
6402 * vec.h (vl_ptr>::reverse): New function.
6404 2018-05-29 Gerald Pfeifer <gerald@pfeifer.com>
6406 * config.gcc: Identify FreeBSD 3.x and 4.x as unsupported.
6408 * config/freebsd-spec.h (FBSD_LIB_SPEC): Only consider FreeBSD 5
6411 2018-05-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
6413 * tree-dump.c (dump_node): Use splay_tree_delete_pointers.
6415 2018-05-28 Richard Biener <rguenther@suse.de>
6417 PR tree-optimization/85933
6418 * tree-vect-data-refs.c (vect_record_base_alignments): Only
6419 look at stmts marked as vectorizable.
6421 2018-05-28 Richard Biener <rguenther@suse.de>
6423 PR tree-optimization/85934
6424 * tree-vect-generic.c (expand_vector_operations_1): Hoist
6425 vector boolean check before scalar optimization.
6427 2018-05-28 Jakub Jelinek <jakub@redhat.com>
6429 * doc/invoke.texi (ARM Options): Use @item instead of @itemx
6432 2018-05-28 Mark Wielaard <mark@klomp.org>
6434 * dwarf2asm.c (dw2_asm_output_delta_uleb128): Add brackets around lab2
6435 if it is an expression containing a minus sign.
6437 2018-05-27 John David Anglin <danglin@gcc.gnu.org>
6439 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Define to 0.
6441 2018-05-27 Paul Koning <ni1d@arrl.net>
6443 * config/pdp11/pdp11.md (truncsihi2): Remove.
6445 2018-05-27 Monk Chiang <sh.chiang04@gmail.com>
6446 Chung-Ju Wu <jasonwucj@gmail.com>
6448 * config/nds32/nds32-intrinsic.md (unaligned_storedi): Modify patterns
6450 (unaligned_store_dw): Ditto.
6451 * config/nds32/nds32-memory-manipulation.c
6452 (nds32_expand_movmemsi_loop_known_size): Refactoring implementation.
6453 (nds32_gen_dup_4_byte_to_word_value): Rename to ...
6454 (nds32_gen_dup_4_byte_to_word_value_aux): ... this.
6455 (emit_setmem_word_loop): Rename to ...
6456 (emit_setmem_doubleword_loop): ... this.
6457 (nds32_gen_dup_4_byte_to_word_value): New function.
6458 (nds32_gen_dup_8_byte_to_double_word_value): New function.
6459 (nds32_expand_setmem_loop): Refine implementation.
6460 (nds32_expand_setmem_loop_v3m): Ditto.
6461 * config/nds32/nds32-multiple.md (unaligned_store_update_base_dw): New
6464 2018-05-27 Chung-Ju Wu <jasonwucj@gmail.com>
6466 * config/nds32/nds32.md (bswapsi2, bswaphi2): New patterns.
6468 2018-05-27 Chung-Ju Wu <jasonwucj@gmail.com>
6470 * config/nds32/nds32.c (nds32_attribute_table): Add "no_prologue".
6471 (nds32_init_machine_status): Initialize machine->attr_naked_p and
6472 machine->attr_no_prologue_p.
6473 (nds32_compute_stack_frame): Check "naked" and "no_prologue" attributes.
6474 (nds32_naked_function_p): Handle "naked" and "no_prologue" attributes.
6475 (nds32_expand_epilogue): Consider attr_naked_p.
6476 (nds32_expand_epilogue_v3pop): Likewise.
6477 (nds32_can_use_return_insn): Likewise.
6478 * config/nds32/nds32.h (machine_function): Add attr_naked_p and
6479 attr_no_prologue_p fields.
6480 * config/nds32/nds32.opt (mret-in-naked-func): New option.
6482 2018-05-27 Jakub Jelinek <jakub@redhat.com>
6485 * config/i386/i386.md (fixunssuffix, floatunssuffix): New code
6487 * config/i386/sse.md
6488 (<floatsuffix>float<sseintvecmodelower><mode>2<mask_name><round_name>):
6490 (float<floatunssuffix><sseintvecmodelower><mode>2<mask_name><round_name>):
6492 (<floatsuffix>float<sselongvecmodelower><mode>2<mask_name><round_name>):
6494 (float<floatunssuffix><sselongvecmodelower><mode>2<mask_name><round_name>):
6496 (*<floatsuffix>floatv2div2sf2): Rename to ...
6497 (*float<floatunssuffix>v2div2sf2): ... this.
6498 (<floatsuffix>floatv2div2sf2_mask): Rename to ...
6499 (float<floatunssuffix>v2div2sf2_mask): ... this.
6500 (*<floatsuffix>floatv2div2sf2_mask_1): Rename to ...
6501 (*float<floatunssuffix>v2div2sf2_mask_1): ... this.
6502 (<fixsuffix>fix_truncv8dfv8si2<mask_name><round_saeonly_name>): Rename
6504 (fix<fixunssuffix>_truncv8dfv8si2<mask_name><round_saeonly_name>):
6506 (<fixsuffix>fix_trunc<mode><sseintvecmodelower>2<mask_name><round_saeonly_name>):
6508 (fix<fixunssuffix>_trunc<mode><sseintvecmodelower>2<mask_name><round_saeonly_name>):
6510 (<fixsuffix>fix_trunc<mode><sselongvecmodelower>2<mask_name><round_saeonly_name>):
6512 (fix<fixunssuffix>_trunc<mode><sselongvecmodelower>2<mask_name><round_saeonly_name>):
6514 (<fixsuffix>fix_truncv2sfv2di2<mask_name>): Rename to ...
6515 (fix<fixunssuffix>_truncv2sfv2di2<mask_name>): ... this.
6516 (vec_pack_ufix_trunc_<mode>): Use gen_fixuns_truncv8dfv8si2 instead of
6517 gen_ufix_truncv8dfv8si2.
6518 * config/i386/i386-builtin.def (__builtin_ia32_cvttpd2uqq256_mask,
6519 __builtin_ia32_cvttpd2uqq128_mask, __builtin_ia32_cvttps2uqq256_mask,
6520 __builtin_ia32_cvttps2uqq128_mask, __builtin_ia32_cvtuqq2ps256_mask,
6521 __builtin_ia32_cvtuqq2ps128_mask, __builtin_ia32_cvtuqq2pd256_mask,
6522 __builtin_ia32_cvtuqq2pd128_mask, __builtin_ia32_cvttpd2udq512_mask,
6523 __builtin_ia32_cvtuqq2ps512_mask, __builtin_ia32_cvtuqq2pd512_mask,
6524 __builtin_ia32_cvttps2uqq512_mask, __builtin_ia32_cvttpd2uqq512_mask):
6525 Use fixuns instead ufix or floatuns instead ufloat in CODE_FOR_ names.
6527 2018-05-24 H.J. Lu <hongjiu.lu@intel.com>
6531 * varasm.c (assemble_alias): Lookup ifunc attribute on error.
6533 2018-05-25 Jim Wilson <jimw@sifive.com>
6535 * config/riscv/riscv-protos.h (riscv_epilogue_uses): New.
6536 * config/riscv/riscv.c (struct machine_function): Add
6537 interrupt_handler_p and attribute_checked_p fields.
6538 (riscv_attribute_table): Add interrupt.
6539 (riscv_interrupt_type_p): New.
6540 (riscv_save_reg_p): Save extra regs for interrupt handler.
6541 (riscv_use_save_libcall): Return false for interrupt handler.
6542 (riscv_first_stack_step): Add forward declaration.
6543 (riscv_compute_frame_info): New local interrupt_save_t1. Set it
6544 for interrupt handler with large frame. Use it for saved reg list.
6545 (riscv_expand_prologue): Move flag_stack_usage_info support to
6546 eliminate duplication.
6547 (riscv_expand_epilogue): Generate mret for interrupt handler.
6548 (riscv_epilogue_uses): New.
6549 (riscv_can_use_return_insn): Return false for interrupt handler.
6550 (riscv_function_ok_for_sibcall): Likewise.
6551 (riscv_set_current_function): Add interrupt handler support.
6552 * config/riscv/riscv.h (EPILOGUE_USES): Call riscv_epilogue_uses.
6553 * config/riscv/riscv.md (UNSPECV_MRET): New.
6555 (riscv_frflags, riscv_fsflags): Use tab after opcode.
6557 * doc/extend.texi (RISC-V Function Attributes) <interrupt>: New.
6559 2018-05-25 Bill Schmidt <wschmidt@linux.ibm.com>
6561 PR tree-optimization/85712
6562 * gimple-ssa-strength-reduction.c (replace_one_candidate): Skip if
6563 this candidate has already been replaced in-situ by a copy.
6565 2018-05-25 Jason Merrill <jason@redhat.com>
6567 PR c++/80485 - inline function non-zero address.
6568 * symtab.c (nonzero_address): Check DECL_COMDAT.
6570 2018-05-25 Uros Bizjak <ubizjak@gmail.com>
6573 * config/alpha/alpha.md (ashlsi3): New insn pattern.
6574 (*ashlsi_se): Rename from *ashldi_se. Define as sign
6575 extension of SImode operation. Use const123_operand predicate.
6576 (*saddsi_1): Remove.
6577 (*saddl_se_1): Ditto.
6579 (*ssubl_se_1): Ditto.
6580 * config/alpha/predicates.md (const123_operand): New predicate.
6581 * config/alpha/constraints.md (P): Use IN_RANGE.
6583 2018-05-25 Richard Biener <rguenther@suse.de>
6585 * tree-ssa-alias.h (refs_may_alias_p): Add tbaa_p bool parameter,
6587 (ref_maybe_used_by_stmt_p): Likewise.
6588 (stmt_may_clobber_ref_p): Likewise.
6589 (stmt_may_clobber_ref_p_1): Likewise.
6590 * tree-ssa-alias.c (refs_may_alias_p): Add tbaa_p bool parameter
6592 (ref_maybe_used_by_stmt_p): Likewise.
6593 (stmt_may_clobber_ref_p): Likewise.
6594 (stmt_may_clobber_ref_p_1): Likewise.
6595 * tree-vect-data-refs.c (vect_slp_analyze_node_dependences): Use
6596 the alias oracle to disambiguate DRs with stmts DR analysis
6598 (vect_analyze_data_refs): Do not give up on not analyzable
6599 DRs for BB vectorization. Remove code truncating the dataref
6602 2018-05-25 Jakub Jelinek <jakub@redhat.com>
6605 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>_1):
6606 Add (=Yk,v,C) variant using vptestm insn. Use TARGET_AVX512BW
6607 in test instead of TARGET_AVX512F for VI12_AVX512VL iterator.
6609 2018-05-25 Richard Biener <rguenther@suse.de>
6611 * tree-vect-data-refs.c (vect_find_stmt_data_reference): New
6612 function, combining stmt data ref gathering and fatal analysis
6614 (vect_analyze_data_refs): Remove now redudnant code and simplify.
6615 * tree-vect-loop.c (vect_get_datarefs_in_loop): Factor out from
6616 vect_analyze_loop_2 and use vect_find_stmt_data_reference.
6617 * tree-vect-slp.c (vect_slp_bb): Use vect_find_stmt_data_reference.
6618 * tree-vectorizer.h (vect_find_stmt_data_reference): Declare.
6620 2018-05-25 Bin Cheng <bin.cheng@arm.com>
6622 PR tree-optimization/85720
6623 * tree-loop-distribution.c (break_alias_scc_partitions): Don't merge
6624 SCC if all partitions are builtins.
6625 (version_loop_by_alias_check): New parameter. Generate cancelable
6626 runtime alias check if all partitions are builtins.
6627 (distribute_loop): Update call to above function.
6629 2018-05-25 Bin Cheng <bin.cheng@arm.com>
6631 * tree-outof-ssa.c (tree-ssa.h, tree-dfa.h): Include header files.
6632 (create_default_def, for_all_parms): Moved from tree-ssa-coalesce.c.
6633 (parm_default_def_partition_arg): Ditto.
6634 (set_parm_default_def_partition): Ditto.
6635 (get_parm_default_def_partitions): Ditto and make it static.
6636 (get_undefined_value_partitions): Ditto and make it static.
6637 (remove_ssa_form): Refactor call to init_var_map here.
6638 * tree-ssa-coalesce.c (build_ssa_conflict_graph): Support live range
6639 computation for loop region.
6640 (coalesce_partitions, compute_optimized_partition_bases): Ditto.
6641 (register_default_def): Delete.
6642 (for_all_parms, create_default_def): Move to tree-outof-ssa.c.
6643 (parm_default_def_partition_arg): Ditto.
6644 (set_parm_default_def_partition): Ditto.
6645 (get_parm_default_def_partitions): Ditto and make it static.
6646 (get_undefined_value_partitions): Ditto and make it static.
6647 (coalesce_with_default, coalesce_with_default): Update comment.
6648 (create_coalesce_list_for_region): New func factored out from
6649 create_outofssa_var_map.
6650 (populate_coalesce_list_for_outofssa): New func factored out from
6651 create_outofssa_var_map and coalesce_ssa_name.
6652 (create_outofssa_var_map): Delete.
6653 (coalesce_ssa_name): Refactor to support live range computation.
6654 * tree-ssa-coalesce.h (coalesce_ssa_name): Change decl.
6655 (get_parm_default_def_partitions): Delete.
6656 (get_undefined_value_partitions): Ditto.
6657 * tree-ssa-live.c (init_var_map, delete_var_map): Support live range
6658 computation for loop region.
6659 (new_tree_live_info, loe_visit_block): Ditto.
6660 (live_worklist, set_var_live_on_entry): Ditto.
6661 (calculate_live_on_exit, verify_live_on_entry): Ditto.
6662 * tree-ssa-live.h (struct _var_map): New fields.
6663 (init_var_map): Change decl.
6664 (region_contains_p): New.
6666 2018-05-25 Bin Cheng <bin.cheng@arm.com>
6668 * tree-ssa-live.h (live_merge_and_clear): Delete.
6670 2018-05-25 Richard Biener <rguenther@suse.de>
6673 * tree-dump.c (dequeue_and_dump): Remove access to removed
6674 operand 2 of a SWITCH_EXPR.
6676 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
6678 * doc/sourcebuild.texi (vect_double_cond_arith): Include
6679 multiplication and division.
6680 * doc/md.texi (cond_mul@var{m}, cond_div@var{m}, cond_mod@var{m})
6681 (cond_udiv@var{m}, cond_umod@var{m}): Document.
6682 * optabs.def (cond_smul_optab, cond_sdiv_optab, cond_smod_optab)
6683 (cond_udiv_optab, cond_umod_optab): New optabs.
6684 * internal-fn.def (IFN_COND_MUL, IFN_COND_DIV, IFN_COND_MOD)
6685 (IFN_COND_RDIV): New internal functions.
6686 * internal-fn.c (get_conditional_internal_fn): Handle TRUNC_DIV_EXPR,
6687 TRUNC_MOD_EXPR and RDIV_EXPR.
6688 * match.pd (UNCOND_BINARY, COND_BINARY): Handle them.
6689 * config/aarch64/iterators.md (UNSPEC_COND_MUL, UNSPEC_COND_DIV):
6691 (SVE_INT_BINARY): Include mult.
6692 (SVE_COND_FP_BINARY): Include UNSPEC_MUL and UNSPEC_DIV.
6693 (optab, sve_int_op): Handle mult.
6694 (optab, sve_fp_op, commutative): Handle UNSPEC_COND_MUL and
6696 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): New pattern
6697 for SVE_INT_BINARY_SD.
6699 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
6701 * config/aarch64/iterators.md (SVE_INT_BINARY_SD): New code iterator.
6702 (optab, sve_int_op): Handle div and udiv.
6703 * config/aarch64/aarch64-sve.md (<optab><mode>3): New expander
6704 for SVE_INT_BINARY_SD.
6705 (*<optab><mode>3): New insn for the same.
6707 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
6709 * tree-vect-patterns.c: Include predict.h.
6710 (vect_recog_divmod_pattern): Restrict check for division support
6711 to when optimizing for size.
6713 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
6715 * doc/sourcebuild.texi (vect_double_cond_arith: Document.
6716 * gimple-match.h (gimple_match_op::MAX_NUM_OPS): Bump to 4.
6717 (gimple_match_op::gimple_match_op): Add an overload for 4 operands.
6718 (gimple_match_op::set_op): Likewise.
6719 (gimple_resimplify4): Declare.
6720 * genmatch.c (get_operand_type): Handle CFN_COND_* functions.
6721 (expr::gen_transform): Likewise.
6722 (decision_tree::gen): Generate a simplification routine for 4 operands.
6723 * gimple-match-head.c (gimple_simplify): Add an overload for
6724 4 operands. In the top-level function, handle up to 4 call
6725 arguments and call gimple_resimplify4.
6726 (gimple_resimplify4): New function.
6727 (build_call_internal): Pass a fourth operand.
6728 (maybe_push_to_seq): Likewise.
6729 * match.pd (UNCOND_BINARY, COND_BINARY): New operator lists.
6730 Fold VEC_COND_EXPRs of an operation and a default value into
6731 an IFN_COND_* function if possible.
6732 * config/aarch64/iterators.md (UNSPEC_COND_MAX, UNSPEC_COND_MIN):
6734 (SVE_COND_FP_BINARY): Include them.
6735 (optab, sve_fp_op): Handle them.
6736 (SVE_INT_BINARY_REV): New code iterator.
6737 (SVE_COND_FP_BINARY_REV): New int iterator.
6738 (commutative): New int attribute.
6739 * config/aarch64/aarch64-protos.h (aarch64_sve_prepare_conditional_op):
6741 * config/aarch64/aarch64.c (aarch64_sve_prepare_conditional_op): New
6743 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): Use it.
6744 (*cond_<optab><mode>): New patterns for reversed operands.
6746 2018-05-25 Richard Biener <rguenther@suse.de>
6748 * tree-vectorizer.h (STMT_VINFO_GROUP_*, GROUP_*): Remove.
6749 (DR_GROUP_*): New, assert we have non-NULL ->data_ref_info.
6750 (REDUC_GROUP_*): New, assert we have NULL ->data_ref_info.
6751 (STMT_VINFO_GROUPED_ACCESS): Adjust.
6752 * tree-vect-data-refs.c (everywhere): Adjust users.
6753 * tree-vect-loop.c (everywhere): Likewise.
6754 * tree-vect-slp.c (everywhere): Likewise.
6755 * tree-vect-stmts.c (everywhere): Likewise.
6756 * tree-vect-patterns.c (vect_reassociating_reduction_p): Likewise.
6758 2018-05-25 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
6760 * configure.ac (gcc_cv_as_section_has_e): Move to common section.
6762 (gcc_cv_as_section_exclude): ... this.
6763 Try Solaris as #exclude syntax.
6764 * configure: Regenerate.
6765 * config.in: Regenerate.
6766 * config/i386/i386.c (i386_solaris_elf_named_section): Handle
6768 * config/sparc/sparc.c (sparc_solaris_elf_asm_named_section)
6769 [HAVE_GAS_SECTION_EXCLUDE]: Handle SECTION_EXCLUDE.
6771 * varasm.c (default_elf_asm_named_section): Don't check if
6772 HAVE_GAS_SECTION_EXCLUDE is defined.
6774 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
6776 * doc/md.texi: Update the documentation of the cond_* optabs
6777 to mention the new final operand. Fix GET_MODE_NUNITS call.
6778 Describe the scalar case too.
6779 * internal-fn.def (IFN_EXTRACT_LAST): Change type to fold_left.
6780 * internal-fn.c (expand_cond_unary_optab_fn): Expect 3 operands
6782 (expand_cond_binary_optab_fn): Expect 4 operands instead of 3.
6783 (get_conditional_internal_fn): Update comment.
6784 * tree-vect-loop.c (vectorizable_reduction): Pass the original
6785 accumulator value as a final argument to conditional functions.
6786 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): Turn into
6787 a define_expand and add an "else" operand. Assert for now that
6788 the else operand is equal to operand 2. Use SVE_INT_BINARY and
6789 SVE_COND_FP_BINARY instead of SVE_COND_INT_OP and SVE_COND_FP_OP.
6790 (*cond_<optab><mode>): New patterns.
6791 * config/aarch64/iterators.md (UNSPEC_COND_SMAX, UNSPEC_COND_UMAX)
6792 (UNSPEC_COND_SMIN, UNSPEC_COND_UMIN, UNSPEC_COND_AND, UNSPEC_COND_ORR)
6793 (UNSPEC_COND_EOR): Delete.
6794 (optab): Remove associated mappings.
6795 (SVE_INT_BINARY): New code iterator.
6796 (sve_int_op): Remove int attribute and add "minus" to the code
6798 (SVE_COND_INT_OP): Delete.
6799 (SVE_COND_FP_OP): Rename to...
6800 (SVE_COND_FP_BINARY): ...this.
6802 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
6804 * optabs.c (can_reuse_operands_p): New function.
6805 (maybe_legitimize_operands): Try to reuse the results for
6808 2018-05-24 Uros Bizjak <ubizjak@gmail.com>
6810 * config/i386/sse.md (cvtusi2<ssescalarmodesuffix>64<round_name>):
6811 Add {q} suffix to insn mnemonic.
6813 2018-05-23 Jozef Lawrynowicz <jozef.l@mittosystems.com>
6815 * config/msp430/msp430.c (TARGET_WARN_FUNC_RETURN): Define.
6816 (msp430_warn_func_return): New.
6818 2018-05-24 Roger Sayle <roger@nextmovesoftware.com>
6820 * fold-const.c (tree_nonzero_bits): New function.
6821 * fold-const.h (tree_nonzero_bits): Likewise.
6822 * match.pd (POPCOUNT): New patterns to fold BUILTIN_POPCOUNT and
6823 friends. POPCOUNT(x&1) => x&1, POPCOUNT(x)==0 => x==0, etc.
6825 2018-05-24 H.J. Lu <hongjiu.lu@intel.com>
6829 * varasm.c (assemble_alias): Check ifunc_resolver only on
6832 2018-05-24 Uros Bizjak <ubizjak@gmail.com>
6835 * config/i386/sse.md (movdi_to_sse): Do not generate pseudo
6836 when memory input operand is handled.
6838 2018-05-24 Luis Machado <luis.machado@linaro.org>
6840 * config/aarch64/aarch64.c (qdf24xx_addrcost_table): New static
6842 (qdf24xx_tunings) <addr_costs>: Set to qdf24xx_addrcost_table.
6844 2018-05-24 Richard Sandiford <richard.sandiford@linaro.org>
6846 * match.pd: Delay FMA folds until after vectorization.
6848 2018-05-24 Andre Vieira <andre.simoesdiasvieira@arm.com>
6851 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Make
6852 address check not strict.
6854 2018-05-24 Richard Sandiford <richard.sandiford@linaro.org>
6856 * gimple-match.h (gimple_match_op): New class.
6857 (mprts_hook): Replace parameters with a gimple_match_op *.
6858 (maybe_build_generic_op): Likewise.
6859 (gimple_simplified_result_is_gimple_val): Replace parameters with
6860 a const gimple_match_op *.
6861 (gimple_simplify): Replace code_helper * and tree * parameters with
6862 a gimple_match_op * parameter.
6863 (gimple_resimplify1): Replace code_helper *, tree and tree *
6864 parameters with a gimple_match_op * parameter.
6865 (gimple_resimplify2): Likewise.
6866 (gimple_resimplify3): Likewise.
6867 (maybe_push_res_to_seq): Replace code_helper, tree and tree *
6868 parameters with a gimple_match_op * parameter.
6869 * gimple-match-head.c (gimple_simplify): Change prototypes of
6870 auto-generated functions to take a gimple_match_op * instead of
6871 separate code_helper * and tree * parameters. Make the same
6872 change in the top-level overload and update calls to the
6873 gimple_resimplify routines. Update calls to the auto-generated
6874 functions and to maybe_push_res_to_seq in the publicly-facing
6875 operation-specific gimple_simplify overloads.
6876 (gimple_match_op::MAX_NUM_OPS): Define.
6877 (gimple_resimplify1): Replace rcode and ops with a single res_op
6878 parameter. Update call to gimple_simplify.
6879 (gimple_resimplify2): Likewise.
6880 (gimple_resimplify3): Likewise.
6881 (mprts_hook): Replace parameters with a gimple_match_op *.
6882 (maybe_build_generic_op): Likewise.
6883 (build_call_internal): Replace type, nargs and ops with
6884 a gimple_match_op *.
6885 (maybe_push_res_to_seq): Replace res_code, type and ops parameters
6886 with a single gimple_match_op *. Update calls to mprts_hook,
6887 build_call_internal and gimple_simplified_result_is_gimple_val.
6888 Factor out code that is common to the tree_code and combined_fn cases.
6889 * genmatch.c (expr::gen_transform): Replace tem_code and
6890 tem_ops with a gimple_match_op called tem_op. Update calls
6891 to the gimple_resimplify functions and maybe_push_res_to_seq.
6892 (dt_simplify::gen_1): Manipulate res_op instead of res_code and
6893 res_ops. Update call to the gimple_resimplify functions.
6894 (dt_simplify::gen): Pass res_op instead of res_code and res_ops.
6895 (decision_tree::gen): Make the functions take a gimple_match_op *
6896 called res_op instead of separate res_code and res_ops parameters.
6897 Update call accordingly.
6898 * gimple-fold.c (replace_stmt_with_simplification): Replace rcode
6899 and ops with a single res_op parameter. Update calls to
6900 maybe_build_generic_op and maybe_push_res_to_seq.
6901 (fold_stmt_1): Update calls to gimple_simplify and
6902 replace_stmt_with_simplification.
6903 (gimple_fold_stmt_to_constant_1): Update calls to gimple_simplify
6904 and gimple_simplified_result_is_gimple_val.
6905 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update call to
6907 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Replace parameters
6908 with a gimple_match_op *.
6909 (vn_nary_build_or_lookup): Likewise. Update call to
6910 vn_nary_build_or_lookup_1.
6911 (vn_nary_build_or_lookup_1): Replace rcode, type and ops with a
6912 gimple_match_op *. Update calls to the gimple_resimplify routines
6913 and to gimple_simplified_result_is_gimple_val.
6914 (vn_nary_simplify): Update call to vn_nary_build_or_lookup_1.
6915 Use gimple_match_op::MAX_NUM_OPS instead of a hard-coded 3.
6916 (vn_reference_lookup_3): Update call to vn_nary_build_or_lookup.
6917 (visit_nary_op): Likewise.
6918 (visit_reference_op_load): Likewise.
6920 2018-05-23 Luis Machado <luis.machado@linaro.org>
6922 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Use correct type
6923 modifier for printing the step amount.
6925 2018-05-23 Jozef Lawrynowicz <jozef.l@somniumtech.com>
6928 * tree.c (build_common_tree_nodes): Dont set TYPE_SIZE for __intN
6931 2018-05-23 Segher Boessenkool <segher@kernel.crashing.org>
6933 * doc/sourcebuild.texi (Endianness): New subsubsection.
6935 2018-05-23 Luis Machado <luis.machado@linaro.org>
6937 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
6938 <prefetch_dynamic_strides>: New const bool field.
6939 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
6940 prefetch_dynamic_strides.
6941 (exynosm1_prefetch_tune): Likewise.
6942 (thunderxt88_prefetch_tune): Likewise.
6943 (thunderx_prefetch_tune): Likewise.
6944 (thunderx2t99_prefetch_tune): Likewise.
6945 (qdf24xx_prefetch_tune): Likewise. Set prefetch_dynamic_strides to
6947 (aarch64_override_options_internal): Update to set
6948 PARAM_PREFETCH_DYNAMIC_STRIDES.
6949 * doc/invoke.texi (prefetch-dynamic-strides): Document new option.
6950 * params.def (PARAM_PREFETCH_DYNAMIC_STRIDES): New.
6951 * params.h (PARAM_PREFETCH_DYNAMIC_STRIDES): Define.
6952 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Account for
6953 prefetch-dynamic-strides setting.
6955 2018-05-23 Luis Machado <luis.machado@linaro.org>
6957 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
6958 <minimum_stride>: New const int field.
6959 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
6960 minimum_stride field defaulting to -1.
6961 (exynosm1_prefetch_tune): Likewise.
6962 (thunderxt88_prefetch_tune): Likewise.
6963 (thunderx_prefetch_tune): Likewise.
6964 (thunderx2t99_prefetch_tune): Likewise.
6965 (qdf24xx_prefetch_tune) <minimum_stride>: Set to 2048.
6966 <default_opt_level>: Set to 3.
6967 (aarch64_override_options_internal): Update to set
6968 PARAM_PREFETCH_MINIMUM_STRIDE.
6969 * doc/invoke.texi (prefetch-minimum-stride): Document new option.
6970 * params.def (PARAM_PREFETCH_MINIMUM_STRIDE): New.
6971 * params.h (PARAM_PREFETCH_MINIMUM_STRIDE): Define.
6972 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Return false if
6973 stride is constant and is below the minimum stride threshold.
6975 2018-05-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6977 * config/arm/arm-cpus.in (mode26): Delete.
6978 (armv4): Delete mode26 reference.
6979 * config/arm/arm.c (arm_configure_build_target): Delete use of
6982 2018-05-23 Uros Bizjak <ubizjak@gmail.com>
6984 * config/i386/i386.md (*floatuns<SWI48:mode><MODEF:mode>2_avx512):
6986 (floatunssi<mode>2): Also enable for AVX512F and TARGET_SSE_MATH.
6987 Rewrite expander pattern. Emit gen_floatunssi<mode>2_i387_with_xmm
6989 (floatunsdisf2): Rewrite expander pattern. Hanlde TARGET_AVX512F.
6990 (floatunsdidf2): Ditto.
6992 2018-05-23 Uros Bizjak <ubizjak@gmail.com>
6994 * config/i386/i386.md (fixuns_trunc<mode>di2): New insn pattern.
6995 (fixuns_trunc<mode>si2_avx512f): Ditto.
6996 (*fixuns_trunc<mode>si2_avx512f_zext): Ditto.
6997 (fixuns_trunc<mode>si2): Also enable for AVX512F and TARGET_SSE_MATH.
6998 Emit fixuns_trunc<mode>si2_avx512f for AVX512F targets.
7000 2018-05-23 Alexander Monakov <amonakov@ispras.ru>
7002 PR rtl-optimization/79985
7003 * df-scan.c (df_insn_refs_collect): Remove special case for
7004 global registers and asm statements.
7006 2018-05-23 Alexander Monakov <amonakov@ispras.ru>
7008 * extend.texi (Global Register Variables): Rewrite the bullet list.
7009 Note that the register is available for allocation. Note that access
7010 via inline asm must use constraints. Add note about async-signal
7011 handlers. Remove paragraph about automagic register selection.
7013 2018-05-23 Richard Biener <rguenther@suse.de>
7015 * tree-ssa-sccvn.c (vn_reference_lookup_3): Remove restriction
7016 of fixed offset from memset VN.
7018 2018-05-23 Bill Schmidt <wschmidt@linux.ibm.com>
7020 * gimple-ssa-strength-reduction.c (struct slsr_cand_d): Add
7022 (alloc_cand_and_find_basis): Initialize first_interp field.
7023 (slsr_process_mul): Modify first_interp field.
7024 (slsr_process_add): Likewise.
7025 (slsr_process_cast): Modify first_interp field for each new
7027 (slsr_process_copy): Likewise.
7028 (dump_candidate): Dump first_interp field.
7029 (replace_mult_candidate): Process all interpretations, not just
7031 (replace_rhs_if_not_dup): Likewise.
7032 (replace_one_candidate): Likewise.
7034 2018-05-23 Wilco Dijkstra <wdijkstr@arm.com>
7036 * config/aarch64/aarch64.c (aarch64_use_frame_pointer):
7038 (aarch64_needs_frame_chain): New function.
7039 (aarch64_parse_override_string): Set aarch64_use_frame_pointer.
7041 2018-05-23 Sudakshina Das <sudi.das@arm.com>
7044 * common/config/aarch64/aarch64-common.c (aarch64_handle_option):
7045 Check val before adding MASK_STRICT_ALIGN to opts->x_target_flags.
7046 * config/aarch64/aarch64.opt (mstrict-align): Remove RejectNegative.
7047 * config/aarch64/aarch64.c (aarch64_attributes): Mark allow_neg
7048 as true for strict-align.
7049 (aarch64_can_inline_p): Perform checks even when callee has no
7050 attributes to check for strict alignment.
7051 * doc/extend.texi (AArch64 Function Attributes): Document
7053 * doc/invoke.texi: (AArch64 Options): Likewise.
7055 2018-05-23 Richard Sandiford <richard.sandiford@linaro.org>
7057 PR tree-optimization/85853
7058 * tree-vect-slp.c (vect_slp_analyze_node_operations): Split out
7059 the handling of the root of the node to...
7060 (vect_slp_analyze_node_operations_1): ...this new function,
7061 and run the whole thing with the child nodes' def types
7062 set according to their SLP node's def type.
7064 2018-05-23 Richard Biener <rguenther@suse.de>
7067 * tree-data-ref.c (create_runtime_alias_checks): Defer
7068 and ignore overflow warnings.
7070 2018-05-23 Yury Gribov <tetra2005@gmail.com>
7072 PR tree-optimization/85822
7073 * tree-vrp.c (is_masked_range_test): Fix handling of negative
7076 2018-05-23 Richard Biener <rguenther@suse.de>
7078 * tree-ssa-sccvn.c (vn_reference_lookup_3): Handle arbitrary
7079 memset constants via native_interpret_expr.
7081 2018-05-22 H.J. Lu <hongjiu.lu@intel.com>
7084 * cgraph.h (cgraph_node::create): Set ifunc_resolver for ifunc
7086 (cgraph_node::create_alias): Likewise.
7087 (cgraph_node::get_availability): Check ifunc_resolver instead
7088 of looking up ifunc attribute.
7089 * cgraphunit.c (maybe_diag_incompatible_alias): Likewise.
7090 * varasm.c (do_assemble_alias): Likewise.
7091 (assemble_alias): Likewise.
7092 (default_binds_local_p_3): Likewise.
7093 * cgraph.h (cgraph_node): Add ifunc_resolver.
7094 (cgraph_node::only_called_directly_or_aliased_p): Return false
7096 * lto-cgraph.c (input_node): Set ifunc_resolver for ifunc
7098 * symtab.c (symtab_node::verify_base): Verify that ifunc_resolver
7099 is equivalent to lookup_attribute ("ifunc", DECL_ATTRIBUTES (decl)).
7100 (symtab_node::binds_to_current_def_p): Check ifunc_resolver
7101 instead of looking up ifunc attribute.
7103 2018-05-22 Luis Machado <luis.machado@linaro.org>
7105 * config/aarch64/aarch64.md (*ashift<mode>_extv_bfiz): New pattern.
7107 2018-05-22 Martin Sebor <msebor@redhat.com>
7110 * builtins.c (expand_builtin_strcpy): Call maybe_warn_nonstring_arg
7111 only when expasion succeeds.
7112 (expand_builtin_strcmp): Same.
7113 (expand_builtin_strncmp): Same.
7115 2018-05-22 Martin Sebor <msebor@redhat.com>
7117 * calls.c (maybe_warn_nonstring_arg): Fix a typo in a comment.
7119 2018-05-22 Jackson Woodruff <jackson.woodruff@arm.com>
7120 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7122 * config/aarch64/aarch64-ldpstp.md: Replace uses of
7123 aarch64_mem_pair_operand with memory_operand and delete operand swapping
7125 * config/aarch64/aarch64.c (aarch64_operands_ok_for_ldpstp):
7126 Add check for legitimate_address.
7127 (aarch64_gen_adjusted_ldpstp): Swap operands where appropriate.
7128 (aarch64_swap_ldrstr_operands): New.
7129 * config/aarch64/aarch64-protos.h (aarch64_swap_ldrstr_operands):
7132 2018-05-22 Jackson Woodruff <jackson.woodruff@arm.com>
7133 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7135 * config/aarch64/aarch64.md: New patterns to generate stp
7137 (store_pair_sw, store_pair_dw): New patterns to generate stp for
7138 single words and double words.
7139 (load_pair_sw, load_pair_dw): Likewise.
7140 (store_pair_sf, store_pair_df, store_pair_si, store_pair_di):
7142 (load_pair_sf, load_pair_df, load_pair_si, load_pair_di):
7144 * config/aarch64/aarch64-ldpstp.md: Modify peephole
7145 for different mode ldpstp and add peephole for merged zero stores.
7147 * config/aarch64/aarch64.c (aarch64_operands_ok_for_ldpstp):
7149 (aarch64_gen_store_pair): Rename calls to match new patterns.
7150 (aarch64_gen_load_pair): Rename calls to match new patterns.
7151 * config/aarch64/aarch64-simd.md (load_pair<mode>): Rename to...
7152 (load_pair<DREG:mode><DREG2:mode>): ... This.
7153 (store_pair<mode>): Rename to...
7154 (vec_store_pair<DREG:mode><DREG2:mode>): ... This.
7155 * config/aarch64/iterators.md (DREG, DREG2, DX2, SX, SX2, DSX):
7157 (V_INT_EQUIV): Handle SImode.
7158 * config/aarch64/predicates.md (aarch64_reg_zero_or_fp_zero):
7161 2018-05-22 Martin Sebor <msebor@redhat.com>
7164 * calls.c (maybe_warn_nonstring_arg): Use string length to set
7165 or ajust the presumed bound on an operation to avoid unnecessary
7168 2018-05-22 Martin Sebor <msebor@redhat.com>
7170 PR tree-optimization/85826
7171 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Avoid
7172 assuming that a DECL necesarily has a constant size.
7174 2018-05-22 Richard Sandiford <richard.sandiford@linaro.org>
7177 * internal-fn.c (expand_direct_optab_fn): Cope with a null lhs.
7179 2018-05-22 Richard Biener <rguenther@suse.de>
7181 PR tree-optimization/85834
7182 * tree-ssa-sccvn.c (vn_reference_lookup_3): Properly handle
7183 non-constant and non-zero memset arguments.
7185 2018-05-22 Martin Liska <mliska@suse.cz>
7188 * ipa-icf.c (sem_item::add_type): Do not ICE for incomplete types.
7190 2018-05-22 Richard Biener <rguenther@suse.de>
7192 PR tree-optimization/85863
7193 * tree-vect-stmts.c (vect_is_simple_cond): Only widen invariant
7194 comparisons when vectype is specified.
7195 (vectorizable_condition): Do not specify vectype for
7196 vect_is_simple_cond when SLP vectorizing.
7198 2018-05-21 Michael Meissner <meissner@linux.ibm.com>
7201 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Do not
7202 define __ibm128 as long double.
7203 * config/rs6000/rs6000.c (rs6000_init_builtins): Create __ibm128
7204 as a distinct type when IEEE 128-bit support is enabled.
7205 (init_float128_ieee): Fix up conversions between IFmode and IEEE
7206 128-bit types to use the correct functions.
7207 (rs6000_expand_float128_convert): Use explicit FLOAT_EXTEND to
7208 convert between 128-bit floating point types that have different
7209 modes but the same representation, instead of using gen_lowpart to
7211 * config/rs6000/rs6000.md (IFKF): New iterator for IFmode and
7213 (IFKF_reg): New attributes to give the register constraints for
7215 (extend<mode>tf2_internal): New insns to mark an explicit
7216 conversion between 128-bit floating point types that have a
7217 different mode but share the same representation.
7219 2018-05-21 Richard Sandiford <richard.sandiford@linaro.org>
7221 PR tree-optimization/85814
7222 * tree-ssa-strlen.c (get_stridx_plus_constant): Cope with
7223 a null return from get_strinfo when unsharing the next
7224 strinfo in the chain.
7226 2018-05-21 Vladimir Mezentsev <vladimir.mezentsev@oracle.com>
7229 * varasm.c (weak_finish): Clean up weak_decls.
7231 2018-05-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7233 * config/aarch64/aarch64.md ("unspec"): Define UNSPEC_SABAL,
7234 UNSPEC_SABDL2, UNSPEC_SADALP, UNSPEC_UABAL, UNSPEC_UABDL2,
7235 UNSPEC_UADALP values.
7236 * config/aarch64/iterators.md (ABAL): New int iterator.
7239 (sur): Add mappings for the above.
7240 * config/aarch64/aarch64-simd.md (aarch64_<sur>abdl2<mode>_3):
7242 (aarch64_<sur>abal<mode>_4): Likewise.
7243 (aarch64_<sur>adalp<mode>_3): Likewise.
7244 (<sur>sadv16qi): New define_expand.
7246 2018-05-21 Alexander Nesterovskiy <alexander.nesterovskiy@intel.com>
7248 * config/i386/i386.md (*movsf_internal): AVX falsedep fix.
7249 (*movdf_internal): Ditto.
7250 (*rcpsf2_sse): Ditto.
7251 (*rsqrtsf2_sse): Ditto.
7252 (*sqrt<mode>2_sse): Ditto.
7254 2018-05-21 Tamar Christina <tamar.christina@arm.com>
7256 * config/aarch64/aarch64-simd.md (aarch64_eor3qv8hi): Change to
7258 (aarch64_bcaxqv8hi): Change to bcaxq<mode>4.
7259 * config/aarch64/aarch64-simd-builtins.def (veor3q_u8, veor3q_u32,
7260 veor3q_u64, veor3q_s8, veor3q_s16, veor3q_s32, veor3q_s64, vbcaxq_u8,
7261 vbcaxq_u32, vbcaxq_u64, vbcaxq_s8, vbcaxq_s16, vbcaxq_s32,
7263 * config/aarch64/arm_neon.h: Likewise.
7264 * config/aarch64/iterators.md (VQ_I): New.
7266 2018-05-21 Alexey Brodkin <abrodkin@synopsys.com>
7268 * config.gcc: Add arc/t-multilib-linux to tmake_file for
7270 * config/arc/t-multilib-linux: Specify MULTILIB_OPTIONS and
7273 2018-05-20 Chung-Ju Wu <jasonwucj@gmail.com>
7275 * config/nds32/constraints.md (S): New constraint.
7276 * config/nds32/nds32.md (call_internal): Use constraint S.
7277 (call_value_internal): Likewise.
7278 (sibcall_internal): Likewise.
7279 (sibcall_value_internal): Likewise.
7281 2018-05-20 Kito Cheng <kito.cheng@gmail.com>
7282 Chung-Ju Wu <jasonwucj@gmail.com>
7284 * config/nds32/nds32.c (nds32_register_move_cost): Take garywolf cpu
7287 2018-05-20 Kito Cheng <kito.cheng@gmail.com>
7288 Chung-Ju Wu <jasonwucj@gmail.com>
7290 * config/nds32/nds32-cost.c (rtx_cost_model_t): New structure.
7291 (insn_size_16bit, insn_size_32bit): New variables for cost evaluation.
7292 (nds32_rtx_costs_impl): Simplify.
7293 (nds32_address_cost_impl): Simplify.
7294 (nds32_init_rtx_costs): New function.
7295 (nds32_rtx_costs_speed_prefer): Likewise.
7296 (nds32_rtx_costs_size_prefer): Likewise.
7297 (nds32_address_cost_speed_prefer): Likewise.
7298 (nds32_address_cost_speed_fwprop): Likewise.
7299 (nds32_address_cost_size_prefer): Likewise.
7300 * config/nds32/nds32-protos.h (nds32_init_rtx_costs): Declare.
7301 * config/nds32/nds32.c (nds32_option_override): Use
7302 nds32_init_rtx_costs function.
7304 2018-05-20 Chung-Ju Wu <jasonwucj@gmail.com>
7306 * config/nds32/nds32.c (nds32_asm_file_start): Output pipeline model.
7307 * config/nds32/nds32.h (TARGET_PIPELINE_N7): Define.
7308 (TARGET_PIPELINE_N8): Likewise.
7309 (TARGET_PIPELINE_N10): Likewise.
7310 (TARGET_PIPELINE_N13): Likewise.
7311 (TARGET_PIPELINE_GRAYWOLF): Likewise.
7313 2018-05-19 Monk Chiang <sh.chiang04@gmail.com>
7315 * config/nds32/nds32-fpu.md: Update copyright year.
7317 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
7319 * config/nds32/nds32.h (ASM_SPEC): Adjust spec rule.
7321 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
7323 * config/nds32/nds32.c
7324 (nds32_md_asm_adjust): Consider flag_inline_asm_r15 variable.
7325 * config/nds32/nds32.opt (minline-asm-r15): New option.
7327 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
7329 * common/config/nds32/nds32-common.c (TARGET_DEFAULT_TARGET_FLAGS): Add
7331 * config/nds32/nds32.md (abssi2): New pattern.
7333 2018-05-19 Uros Bizjak <ubizjak@gmail.com>
7335 * config/i386/i386.md (rex64namesuffix): New mode attribute.
7336 * config/i386/sse.md (sse_cvtsi2ss<rex64namesuffix><round_name>):
7337 Merge insn pattern from sse_cvtsi2ss<round_name> and
7338 sse_cvtsi2ssq<round_name> using SWI48 mode iterator.
7339 (sse_cvtss2si<rex64namesuffix><round_name>): Merge insn pattern
7340 from sse_cvtss2si<round_name> and sse_cvtss2siq<round_name>
7341 using SWI48 mode iterator.
7342 (sse_cvtss2si<rex64namesuffix>_2): Merge insn pattern from
7343 sse_cvtss2si_2 and sse_cvtss2siq_2 using SWI48 mode iterator.
7344 (sse_cvttss2si<rex64namesuffix><round_saeonly_name>): Merge insn
7345 pattern from sse_cvttss2si<round_saeonly_name>
7346 and sse_cvttss2siq<round_saeonly_name> using SWI48 mode iterator.
7347 (avx512f_vcvtss2usi<rex64namesuffix><round_name>): Merge insn pattern
7348 from avx512f_vcvtss2usi<round_name> and avx512f_vcvtss2usiq<round_name>
7349 using SWI48 mode iterator.
7350 (avx512f_vcvttss2usi<rex64namesuffix><round_saeonly_name>): Merge
7351 insn pattern from avx512f_vcvttss2usi<round_saeonly_name> and
7352 avx512f_vcvttss2usiq<round_saeonly_name> using SWI48 mode iterator.
7353 (avx512f_vcvtsd2usi<rex64namesuffix><round_name>): Merge insn pattern
7354 from avx512f_vcvtsd2usi<round_name> and avx512f_vcvtsd2usiq<round_name>
7355 using SWI48 mode iterator.
7356 (avx512f_vcvttsd2usi<rex64namesuffix><round_saeonly_name>): Merge
7357 insn pattern from avx512f_vcvttsd2usi<round_saeonly_name> and
7358 avx512f_vcvttsd2usiq<round_saeonly_name> using SWI48 mode iterator.
7359 (sse2_cvtsd2si<rex64namesuffix><round_name>): Merge insn pattern from
7360 sse2_cvtsd2si<round_name> and sse2_cvtsd2siq<round_name> using
7361 SWI48 mode iterator.
7362 (sse2_cvtsd2si<rex64namesuffix>_2): Merge insn pattern from
7363 sse2_cvtsd2si_2 and sse2_cvtsd2siq_2 using SWI48 mode iterator.
7364 (sse_cvttsd2si<rex64namesuffix><round_saeonly_name>): Merge insn
7365 pattern from sse_cvttsd2si<round_saeonly_name>
7366 and sse_cvttsd2siq<round_saeonly_name> using SWI48 mode iterator.
7368 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
7370 * config/nds32/nds32-md-auxiliary.c
7371 (nds32_valid_smw_lwm_base_p): Refine.
7372 (nds32_output_smw_single_word): Refine.
7373 (nds32_output_smw_double_word): New.
7374 * config/nds32/nds32-protos.h (nds32_output_smw_double_word): New.
7376 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
7378 * config/nds32/nds32-md-auxiliary.c (nds32_output_stack_push): Refine.
7379 (nds32_output_stack_pop): Refine.
7380 (nds32_expand_unaligned_load): Refine.
7381 (nds32_expand_unaligned_store): Refine.
7383 2018-05-19 Kuan-Lin Chen <kuanlinchentw@gmail.com>
7384 Chung-Ju Wu <jasonwucj@gmail.com>
7386 * config/nds32/constants.md: Add TP_REGNUM constant.
7387 (unspec_element): Add UNSPEC_GOTINIT, UNSPEC_GOT, UNSPEC_GOTOFF,
7388 UNSPEC_PLT, UNSPEC_TLSGD, UNSPEC_TLSLD, UNSPEC_TLSIE, UNSPEC_TLSLE and
7390 * config/nds32/nds32-doubleword.md: Consider flag_pic.
7391 * config/nds32/nds32-dspext.md (mov<mode>): Expand TLS and PIC cases.
7392 * config/nds32/nds32-predicates.c (nds32_const_unspec_p): New.
7393 * config/nds32/nds32-md-auxiliary.c: Implementation that support TLS
7394 and PIC code generation.
7395 * config/nds32/nds32-protos.h: Declarations that support TLS and PIC
7397 * config/nds32/nds32-relax-opt.c: Consider TLS and PIC for relax
7399 * config/nds32/nds32.md: Support TLS and PIC.
7400 * config/nds32/nds32.c: Support TLS and PIC.
7401 * config/nds32/nds32.h (nds32_relax_insn_type): New enum type.
7402 * config/nds32/predicates.md (nds32_nonunspec_symbolic_operand): New
7405 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
7407 * config/nds32/nds32-predicates.c (const_vector_to_hwint): Use machine
7408 mode with E_ prefix.
7410 2018-05-19 Kuan-Lin Chen <kuanlinchentw@gmail.com>
7411 Chung-Ju Wu <jasonwucj@gmail.com>
7413 * config/nds32/constants.md (unspec_element): Add UNSPEC_ICT.
7414 * config/nds32/nds32-md-auxiliary.c
7415 (symbolic_reference_mentioned_p): New.
7416 (nds32_legitimize_ict_address): New.
7417 (nds32_expand_ict_move): New.
7418 (nds32_indirect_call_referenced_p): New.
7419 (nds32_symbol_binds_local_p): Delete.
7420 (nds32_long_call_p): Modify.
7421 * config/nds32/nds32-opts.h (nds32_ict_model_type): New enum type.
7422 * config/nds32/nds32-protos.h
7423 (symbolic_reference_mentioned_p): Declare.
7424 (nds32_legitimize_ict_address): Declare.
7425 (nds32_expand_ict_move): Declare.
7426 (nds32_indirect_call_referenced_p): Declare.
7427 * config/nds32/nds32-relax-opt.c (nds32_ict_const_p): New.
7428 (nds32_relax_group): Use nds32_ict_const_p as condition.
7429 * config/nds32/nds32.c (nds32_attribute_table): Add "indirect_call".
7430 (nds32_asm_file_start): Output ict_model directive in asm code.
7431 (nds32_legitimate_address_p): Consider indirect call.
7432 (nds32_print_operand): Consider indirect call.
7433 (nds32_print_operand_address): Consider indirect call.
7434 (nds32_insert_attributes): Handle "indirect_call" attribute.
7435 (TARGET_LEGITIMATE_ADDRESS_P): Define.
7436 (TARGET_LEGITIMATE_CONSTANT_P): Define.
7437 (TARGET_CANNOT_FORCE_CONST_MEM): Define.
7438 (TARGET_DELEGITIMIZE_ADDRESS): Define.
7439 (TARGET_ASM_OUTPUT_ADDR_CONST_EXTRA): Define.
7440 * config/nds32/nds32.h (SYMBOLIC_CONST_P): Define.
7441 (TARGET_ICT_MODEL_SMALL): Define.
7442 (TARGET_ICT_MODEL_LARGE): Define.
7443 * config/nds32/nds32.md (movsi): Consider ict model.
7444 (call, call_value): Consider ict model.
7445 (sibcall, sibcall_value): Consider ict model.
7446 * config/nds32/nds32.opt (mict-model): New option.
7447 * config/nds32/predicates.md (nds32_symbolic_operand): Consider ict
7450 2018-05-18 Kito Cheng <kito.cheng@gmail.com>
7451 Monk Chiang <sh.chiang04@gmail.com>
7452 Jim Wilson <jimw@sifive.com>
7454 * common/config/riscv/riscv-common.c (riscv_parse_arch_string):
7455 Add support to parse rv32e*. Clear MASK_RVE for rv32i and rv64i.
7456 * config.gcc (riscv*-*-*): Add support for rv32e* and ilp32e.
7457 * config/riscv/riscv-c.c (riscv_cpu_cpp_builtins): Define
7458 __riscv_32e when TARGET_RVE. Handle ABI_ILP32E as soft-float ABI.
7459 * config/riscv/riscv-opts.h (riscv_abi_type): Add ABI_ILP32E.
7460 * config/riscv/riscv.c (riscv_compute_frame_info): When TARGET_RVE,
7461 compute save_libcall_adjustment properly.
7462 (riscv_option_override): Call error if TARGET_RVE and not ABI_ILP32E.
7463 (riscv_conditional_register_usage): Handle TARGET_RVE and ABI_ILP32E.
7464 * config/riscv/riscv.h (UNITS_PER_FP_ARG): Handle ABI_ILP32E.
7465 (STACK_BOUNDARY, ABI_STACK_BOUNDARY): Handle TARGET_RVE.
7466 (GP_REG_LAST, MAX_ARGS_IN_REGISTERS): Likewise.
7467 (ABI_SPEC): Handle mabi=ilp32e.
7468 * config/riscv/riscv.opt (abi_type): Add ABI_ILP32E.
7469 (RVE): Add RVE mask.
7470 * doc/invoke.texi (RISC-V options) <-mabi>: Add ilp32e info.
7471 <-march>: Add rv32e as an example.
7473 2018-05-18 Marc Glisse <marc.glisse@inria.fr>
7476 * tree-ssa-structalias.c (create_variable_info_for_1): Extra argument.
7477 (intra_create_variable_infos): Handle C++ constructors.
7479 2018-05-18 Martin Liska <mliska@suse.cz>
7481 * passes.def: Remove a redundant pass.
7483 2018-05-18 Eric Botcazou <ebotcazou@adacore.com>
7486 * config/sparc/sparc.c (sparc_expand_builtin): Always initialize op[0].
7488 2018-05-18 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7490 * config/arm/arm-cpus.in (armv3m, mode32): Delete features.
7492 (ARMv2, ARMv3, ARMv3m): Delete fgroups.
7494 (armv2, armv2a, armv3, armv3m): Delete architectures.
7495 (arm2, arm250, arm3, arm6, arm60, arm600, arm610, arm620,
7496 arm7, arm7d, arm7di, arm70, arm700, arm700i, arm710, arm720,
7497 arm710c, arm7100, arm7500, arm7500fe, arm7m, arm7dm, arm7dmi):
7499 * config/arm/arm.md (maddsidi4): Remove check for arm_arch3m.
7500 (*mulsidi3adddi): Likewise.
7501 (mulsidi3): Likewise.
7502 (*mulsidi3_nov6): Likewise.
7503 (umulsidi3): Likewise.
7504 (umulsidi3_nov6): Likewise.
7505 (umaddsidi4): Likewise.
7506 (*umulsidi3adddi): Likewise.
7507 (smulsi3_highpart): Likewise.
7508 (*smulsi3_highpart_nov6): Likewise.
7509 (umulsi3_highpart): Likewise.
7510 (*umulsi3_highpart_nov6): Likewise.
7511 * config/arm/arm.h (arm_arch3m): Delete.
7512 * config/arm/arm.c (arm_arch3m): Delete.
7513 (arm_option_override_internal): Update armv3-related comment.
7514 (arm_configure_build_target): Delete use of isa_bit_mode32.
7515 (arm_option_reconfigure_globals): Delete set of arm_ach3m.
7516 (arm_rtx_costs_internal): Delete check of arm_arch3m.
7517 * config/arm/arm-fixed.md (mulsq3): Delete check for arm_arch3m.
7519 (mulusa3): Likewise.
7520 * config/arm/arm-protos.h (arm_arch3m): Delete.
7521 * config/arm/arm-tables.opt: Regenerate.
7522 * config/arm/arm-tune.md: Likewise.
7523 * config/arm/t-arm-elf (all_early_nofp): Delete mentions of
7524 deleted architectures.
7526 2018-05-18 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7528 * config/arm/arm-cpus.in (armv5, armv5e): Delete features.
7529 (armv5t, armv5te): New features.
7530 (ARMv5, ARMv5e): Delete fgroups.
7531 (ARMv5t, ARMv5te): Adjust for above changes.
7533 (armv5, armv5e): Delete arches.
7534 * config/arm/arm.md (*call_reg_armv5): Use arm_arch5t instead of
7536 (*call_reg_arm): Likewise.
7537 (*call_value_reg_armv5): Likewise.
7538 (*call_value_reg_arm): Likewise.
7539 (*call_symbol): Likewise.
7540 (*call_value_symbol): Likewise.
7541 (*sibcall_insn): Likewise.
7542 (*sibcall_value_insn): Likewise.
7544 (prefetch): Likewise.
7545 (define_split and define_peephole2 dependent on arm_arch5):
7547 * config/arm/arm.h (TARGET_LDRD): Use arm_arch5te instead of
7549 (TARGET_ARM_QBIT): Likewise.
7550 (TARGET_DSP_MULTIPLY): Likewise.
7551 (enum base_architecture): Delete BASE_ARCH_5, BASE_ARCH_5E.
7552 (arm_arch5, arm_arch5e): Delete.
7553 (arm_arch5t, arm_arch5te): Declare.
7554 * config/arm/arm.c (arm_arch5, arm_arch5e): Delete.
7555 (arm_arch5t): Declare.
7556 (arm_option_reconfigure_globals): Update for the above.
7557 (arm_options_perform_arch_sanity_checks): Update comment, replace
7558 use of arm_arch5 with arm_arch5t.
7559 (use_return_insn): Likewise.
7560 (arm_emit_call_insn): Likewise.
7561 (output_return_instruction): Likewise.
7562 (arm_final_prescan_insn): Likewise.
7563 (arm_coproc_builtin_available): Likewise.
7564 * config/arm/arm-c.c (arm_cpu_builtins): Replace arm_arch5 and
7565 arm_arch5e with arm_arch5t and arm_arch5te.
7566 * config/arm/arm-protos.h (arm_arch5, arm_arch5e): Delete.
7567 (arm_arch5t, arm_arch5te): Declare.
7568 * config/arm/arm-tables.opt: Regenerate.
7569 * config/arm/t-arm-elf: Remove references to armv5, armv5e.
7570 * config/arm/t-multilib: Likewise.
7571 * config/arm/thumb1.md (*call_reg_thumb1_v5): Check arm_arch5t
7572 instead of arm_arch5.
7573 (*call_reg_thumb1): Likewise.
7574 (*call_value_reg_thumb1_v5): Likewise.
7575 (*call_value_reg_thumb1): Likewise.
7576 * config/arm/vxworks.h (TARGET_OS_CPP_BUILTINS): Remove now
7578 * doc/invoke.texi (ARM Options): Remove references to armv5, armv5e.
7580 2018-05-18 Martin Liska <mliska@suse.cz>
7582 PR gcov-profile/84846
7583 * doc/gcov.texi: Document -t option of gcov tool.
7585 2018-05-18 Martin Liska <mliska@suse.cz>
7587 PR gcov-profile/84846
7588 * gcov.c (print_usage): Add new -t option.
7589 (process_args): Handle the option.
7590 (generate_results): Use stdout as output when requested by
7593 2018-05-18 Martin Liska <mliska@suse.cz>
7595 PR gcov-profile/84846
7596 * coverage.c (coverage_init): Write PWD to .gcno file.
7597 * doc/gcov.texi: Document how working directory is printed.
7598 * gcov-dump.c (dump_gcov_file): Print PWD.
7599 * gcov.c (output_intermediate_file): Likewise.
7600 (read_graph_file): Read PWD string.
7601 (output_lines): Print PWD.
7603 2018-05-18 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
7606 * ipa-pure-const.c (malloc_candidate_p): Remove the check integer_zerop
7607 for retval and return false if all args to phi are zero.
7609 2018-05-18 Richard Biener <rguenther@suse.de>
7611 * gimple-ssa-evrp.c (class evrp_folder): Add simplify_stmt_using_ranges
7613 (evrp_dom_walker::before_dom_children): Call it.
7615 2018-05-18 Richard Biener <rguenther@suse.de>
7617 * tree-dfa.c (get_ref_base_and_extent): Use range-info to refine
7618 results when processing array refs with variable index.
7620 2018-05-18 Toon Moene <toon@moene.org>
7622 * doc/invoke.texi: Move -floop-unroll-and-jam documentation
7623 directly after that of -floop-interchange. Indicate that both
7624 options are enabled by default when specifying -O3.
7626 2018-05-18 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7628 * config/aarch64/aarch64-simd.md (vec_set<mode>): Use VALL_F16 mode
7629 iterator. Delete separate integer-mode vec_set<mode> expander.
7630 (aarch64_simd_vec_setv2di): Delete.
7631 (vec_setv2di): Delete.
7632 (aarch64_simd_vec_set<mode>): Delete all other patterns with that name.
7633 Use VALL_F16 mode iterator. Add LD1 alternative and use vwcore for
7634 the "w, r" alternative.
7636 2018-05-18 Martin Liska <mliska@suse.cz>
7638 * passes.def: Add pass_lower_switch and pass_lower_switch_O0.
7639 * tree-pass.h (make_pass_lower_switch_O0): New function.
7640 * tree-switch-conversion.c (node_has_low_bound): Remove.
7641 (node_has_high_bound): Likewise.
7642 (node_is_bounded): Likewise.
7643 (class pass_lower_switch): Make it a template type and create
7645 (pass_lower_switch::execute): Add template argument.
7646 (make_pass_lower_switch): New function.
7647 (make_pass_lower_switch_O0): New function.
7648 (do_jump_if_equal): Remove.
7649 (emit_case_nodes): Simplify to just handle all 3 cases and leave
7650 all the hard work to tree optimization passes.
7652 2018-05-18 Martin Liska <mliska@suse.cz>
7654 * dbgcnt.c (limit_low): Renamed from limit.
7655 (limit_high): New variable.
7656 (dbg_cnt_is_enabled): Check for upper limit.
7657 (dbg_cnt): Adjust dumping.
7658 (dbg_cnt_set_limit_by_index): Add new argument for high
7660 (dbg_cnt_set_limit_by_name): Likewise.
7661 (dbg_cnt_process_single_pair): Parse new format.
7662 (dbg_cnt_process_opt): Use strtok.
7663 (dbg_cnt_list_all_counters): Remove 'value' and add
7665 * doc/invoke.texi: Document changes.
7667 2018-05-18 Richard Sandiford <richard.sandiford@linaro.org>
7669 * doc/sourcebuild.texi (scalar_all_fma): Document.
7670 * tree.def (FMA_EXPR): Delete.
7671 * internal-fn.def (FMA, FMS, FNMA, FNMS): New internal functions.
7672 * internal-fn.c (ternary_direct): New macro.
7673 (expand_ternary_optab_fn): Likewise.
7674 (direct_ternary_optab_supported_p): Likewise.
7675 * Makefile.in (build/genmatch.o): Depend on case-fn-macros.h.
7676 * builtins.c (fold_builtin_fma): Delete.
7677 (fold_builtin_3): Don't call it.
7678 * cfgexpand.c (expand_debug_expr): Remove FMA_EXPR handling.
7679 * expr.c (expand_expr_real_2): Likewise.
7680 * fold-const.c (operand_equal_p): Likewise.
7681 (fold_ternary_loc): Likewise.
7682 * gimple-pretty-print.c (dump_ternary_rhs): Likewise.
7683 * gimple.c (DEFTREECODE): Likewise.
7684 * gimplify.c (gimplify_expr): Likewise.
7685 * optabs-tree.c (optab_for_tree_code): Likewise.
7686 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
7687 * tree-eh.c (operation_could_trap_p): Likewise.
7688 (stmt_could_throw_1_p): Likewise.
7689 * tree-inline.c (estimate_operator_cost): Likewise.
7690 * tree-pretty-print.c (dump_generic_node): Likewise.
7691 (op_code_prio): Likewise.
7692 * tree-ssa-loop-im.c (stmt_cost): Likewise.
7693 * tree-ssa-operands.c (get_expr_operands): Likewise.
7694 * tree.c (commutative_ternary_tree_code, add_expr): Likewise.
7695 * fold-const-call.h (fold_fma): Delete.
7696 * fold-const-call.c (fold_const_call_ssss): Handle CFN_FMS,
7697 CFN_FNMA and CFN_FNMS.
7699 * genmatch.c (combined_fn): New enum.
7700 (commutative_ternary_tree_code): Remove FMA_EXPR handling.
7701 (commutative_op): New function.
7702 (commutate): Use it. Handle more than 2 operands.
7703 (dt_operand::gen_gimple_expr): Use commutative_op.
7704 (parser::parse_expr): Allow :c to be used with non-binary
7705 operators if the commutative operand is known.
7706 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Handle
7707 CFN_FMS, CFN_FNMA and CFN_FNMS.
7708 (backprop::process_assign_use): Remove FMA_EXPR handling.
7709 * hsa-gen.c (gen_hsa_insns_for_operation_assignment): Likewise.
7710 (gen_hsa_fma): New function.
7711 (gen_hsa_insn_for_internal_fn_call): Use it for IFN_FMA, IFN_FMS,
7712 IFN_FNMA and IFN_FNMS.
7713 * match.pd: Add folds for IFN_FMS, IFN_FNMA and IFN_FNMS.
7714 * gimple-fold.h (follow_all_ssa_edges): Declare.
7715 * gimple-fold.c (follow_all_ssa_edges): New function.
7716 * tree-ssa-math-opts.c (convert_mult_to_fma_1): Use the
7717 gimple_build interface and use follow_all_ssa_edges to fold the result.
7718 (convert_mult_to_fma): Use direct_internal_fn_suppoerted_p
7719 instead of checking for optabs directly.
7720 * config/i386/i386.c (ix86_add_stmt_cost): Recognize FMAs as calls
7721 rather than FMA_EXPRs.
7722 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Create a
7723 call to IFN_FMA instead of an FMA_EXPR.
7725 2018-05-17 Jim Wilson <jimw@sifive.com>
7727 * expr.c (do_tablejump): When converting index to Pmode, if we have a
7728 sign extended promoted subreg, and the range does not have the sign bit
7729 set, then do a sign extend.
7731 * config/riscv/riscv.c (riscv_extend_comparands): In unsigned QImode
7732 test, check for sign extended subreg and/or constant operands, and
7733 do a sign extend in that case.
7735 2018-05-17 Steve Ellcey <sellcey@cavium.com>
7737 * config/aarch64/thunderx2t99.md (thunderx2t99_ls_both): Delete.
7738 (thunderx2t99_multiple): Delete psuedo-units from used cpus.
7740 (thunderx2t99_alu_shift): Remove alu_shift_reg, alus_shift_reg.
7741 Change logics_shift_reg to logics_shift_imm.
7742 (thunderx2t99_fp_loadpair_basic): Delete.
7743 (thunderx2t99_fp_storepair_basic): Delete.
7744 (thunderx2t99_asimd_int): Add neon_sub and neon_sub_q types.
7745 (thunderx2t99_asimd_polynomial): Delete.
7746 (thunderx2t99_asimd_fp_simple): Add neon_fp_mul_s_scalar_q
7747 and neon_fp_mul_d_scalar_q.
7748 (thunderx2t99_asimd_fp_conv): Add *int_to_fp* types.
7749 (thunderx2t99_asimd_misc): Delete neon_dup and neon_dup_q.
7750 (thunderx2t99_asimd_recip_step): Add missing *sqrt* types.
7751 (thunderx2t99_asimd_lut): Add missing tbl types.
7752 (thunderx2t99_asimd_ext): Delete.
7753 (thunderx2t99_asimd_load1_1_mult): Delete.
7754 (thunderx2t99_asimd_load1_2_mult): Delete.
7755 (thunderx2t99_asimd_load1_ldp): New.
7756 (thunderx2t99_asimd_load1): New.
7757 (thunderx2t99_asimd_load2): Add missing *load2* types.
7758 (thunderx2t99_asimd_load3): New.
7759 (thunderx2t99_asimd_load4): New.
7760 (thunderx2t99_asimd_store1_1_mult): Delete.
7761 (thunderx2t99_asimd_store1_2_mult): Delete.
7762 (thunderx2t99_asimd_store2_mult): Delete.
7763 (thunderx2t99_asimd_store2_onelane): Delete.
7764 (thunderx2t99_asimd_store_stp): New.
7765 (thunderx2t99_asimd_store1): New.
7766 (thunderx2t99_asimd_store2): New.
7767 (thunderx2t99_asimd_store3): New.
7768 (thunderx2t99_asimd_store4): New.
7770 2018-05-17 Jerome Lambourg <lambourg@adacore.com>
7772 * config/arm/arm_cmse.h (cmse_nsfptr_create, cmse_is_nsfptr): Remove
7773 #include <stdint.h>. Replace intptr_t with __INTPTR_TYPE__.
7775 2018-05-17 Pat Haugen <pthaugen@us.ibm.com>
7776 Segher Boessenkool <segher@kernel.crashing.org>
7779 * config/rs6000/rs6000.c (rs6000_output_move_128bit): Check dest
7782 2018-05-17 Richard Biener <rguenther@suse.de>
7784 * tree-ssa-dse.c (dse_classify_store): Fix iterator increment
7785 for pruning loop and prune defs feeding only already visited PHIs.
7787 2018-05-17 Richard Biener <rguenther@suse.de>
7789 * tree-ssa-sccvn.c (vn_reference_lookup_3): Improve memset handling.
7791 2018-05-17 Bin Cheng <bin.cheng@arm.com>
7792 Richard Biener <rguenther@suse.de>
7794 PR tree-optimization/85793
7795 * tree-vect-stmts.c (vectorizable_load): Handle 1 element-wise load
7796 for VMAT_ELEMENTWISE.
7798 2018-05-17 Richard Sandiford <richard.sandiford@linaro.org>
7800 * internal-fn.h (lookup_internal_fn): Declare
7801 * internal-fn.c (lookup_internal_fn): New function.
7802 * gimple.c (gimple_build_call_from_tree): Handle calls to
7804 * gimple-pretty-print.c (dump_gimple_call): Print "." before
7805 internal function names.
7806 * tree-pretty-print.c (dump_generic_node): Likewise.
7807 * tree-ssa-scopedtables.c (expr_hash_elt::print): Likewise.
7809 2018-05-17 Richard Sandiford <richard.sandiford@linaro.org>
7811 * gimple-fold.h (gimple_build): Make the function forms take
7812 combined_fn rather than built_in_function.
7813 (gimple_simplify): Likewise.
7814 * gimple-match-head.c (gimple_simplify): Likewise.
7815 * gimple-fold.c (gimple_build): Likewise.
7816 * tree-vect-loop.c (get_initial_def_for_reduction): Use gimple_build
7817 rather than gimple_build_call_internal.
7818 (get_initial_defs_for_reduction): Likewise.
7819 (vect_create_epilog_for_reduction): Likewise.
7820 (vectorizable_live_operation): Likewise.
7822 2018-05-17 Martin Liska <mliska@suse.cz>
7824 * gimple-ssa-sprintf.c (format_directive): Do not use
7825 space in between 'G_' and '('.
7827 2018-05-17 Jakub Jelinek <jakub@redhat.com>
7830 * config/i386/i386.c (ix86_fold_builtin): Handle masked shifts
7831 even if the mask is not all ones.
7834 * config/i386/i386.c (ix86_fold_builtin): Fold shift builtins by
7836 (ix86_gimple_fold_builtin): Likewise.
7839 * config/i386/i386.c: Include tree-vector-builder.h.
7840 (ix86_vector_shift_count): New function.
7841 (ix86_fold_builtin): Fold shift builtins by scalar count.
7842 (ix86_gimple_fold_builtin): Likewise.
7844 * config/i386/avx512fintrin.h (_mm512_set_epi16, _mm512_set_epi8,
7845 _mm512_setzero): New intrinsics.
7847 2018-05-17 James Greenhalgh <james.greenhalgh@arm.com>
7848 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7850 * config/aarch64/aarch64.c (aarch64_expand_vector_init): Modify
7851 code generation for cases where splatting a value is not useful.
7852 * simplify-rtx.c (simplify_ternary_operation): Simplify
7853 vec_merge across a vec_duplicate and a paradoxical subreg forming
7854 a vector mode to a vec_concat.
7856 2018-05-17 Olga Makhotina <olga.makhotina@intel.com>
7858 * config.gcc: Support "goldmont-plus".
7859 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
7861 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
7862 PROCESSOR_GOLDMONT_PLUS.
7863 * config/i386/i386.c (m_GOLDMONT_PLUS): Define.
7864 (processor_target_table): Add "goldmont-plus".
7865 (PTA_GOLDMONT_PLUS): Define.
7866 (ix86_lea_outperforms): Add TARGET_GOLDMONT_PLUS.
7867 (get_builtin_code_for_version): Handle PROCESSOR_GOLDMONT_PLUS.
7868 (fold_builtin_cpu): Add M_INTEL_GOLDMONT_PLUS.
7869 (fold_builtin_cpu): Add "goldmont-plus".
7870 (ix86_add_stmt_cost): Add TARGET_GOLDMONT_PLUS.
7871 (ix86_option_override_internal): Add "goldmont-plus".
7872 * config/i386/i386.h (processor_costs): Define TARGET_GOLDMONT_PLUS.
7873 (processor_type): Add PROCESSOR_GOLDMONT_PLUS.
7874 * config/i386/x86-tune.def: Add m_GOLDMONT_PLUS.
7875 * doc/invoke.texi: Add goldmont-plus as x86 -march=/-mtune= CPU type.
7877 2018-05-17 Richard Biener <rguenther@suse.de>
7879 PR tree-optimization/85757
7880 * tree-ssa-dse.c (dse_classify_store): Record a PHI def and
7881 remove defs that only feed that PHI from further processing.
7883 2018-05-16 Jim Wilson <jimw@sifive.com>
7885 * config/riscv/riscv.md (<optab>si3_mask, <optab>si3_mask_1): Prepend
7887 (<optab>di3_mask, <optab>di3_mask_1): Likewise.
7889 2018-05-16 Mark Wielaard <mark@klomp.org>
7891 * dwarf2out.c (count_index_strings): New function.
7892 (output_indirect_strings): Call count_index_strings and generate
7893 header for dwarf_version >= 5.
7895 2018-05-16 Mark Wielaard <mark@klomp.org>
7897 * dwarf2out.c (dwarf_FORM): New function.
7898 (set_indirect_string): Use dwarf_FORM.
7899 (reset_indirect_string): Likewise.
7900 (size_of_die): Likewise.
7901 (value_format): Likewise.
7902 (output_die): Likewise.
7903 (add_skeleton_AT_string): Likewise.
7904 (output_macinfo_op): Likewise.
7905 (index_string): Likewise.
7906 (output_index_string_offset): Likewise.
7907 (output_index_string): Likewise.
7908 (count_index_strings): Likewise.
7910 2018-05-16 Carl Love <cel@us.ibm.com>
7912 * config/rs6000/rs6000.md (prefetch): Generate ISA 2.06 instructions
7913 dcbt and dcbtstt with TH=16 if operands[2] is 0 and Power 8 or newer.
7915 2018-05-16 Martin Jambor <mjambor@suse.cz>
7917 * ipa-prop.c (ipa_free_all_edge_args): Remove.
7918 * ipa-prop.h (ipa_free_all_edge_args): Likewise.
7920 2018-05-16 Wilco Dijkstra <wdijkstr@arm.com>
7922 * config/aarch64/aarch64.md (fma<mode>4): Change into expand pattern.
7923 (fnma<mode>4): Likewise.
7924 (fms<mode>4): Likewise.
7925 (fnms<mode>4): Likewise.
7926 (aarch64_fma<mode>4): Rename insn, reorder accumulator operand.
7927 (aarch64_fnma<mode>4): Likewise.
7928 (aarch64_fms<mode>4): Likewise.
7929 (aarch64_fnms<mode>4): Likewise.
7930 (aarch64_fnmadd<mode>4): Likewise.
7932 2018-05-16 Jason Merrill <jason@redhat.com>
7934 * tree.c (warn_deprecated_use): Return bool. Simplify logic.
7936 2018-05-16 Richard Biener <rguenther@suse.de>
7938 * tree-vectorizer.h (struct stmt_info_for_cost): Add where member.
7939 (dump_stmt_cost): Declare.
7940 (add_stmt_cost): Dump cost we add.
7941 (add_stmt_costs): New function.
7942 (vect_model_simple_cost, vect_model_store_cost, vect_model_load_cost):
7944 (vect_analyze_stmt): Adjust prototype.
7945 (vectorizable_condition): Likewise.
7946 (vectorizable_live_operation): Likewise.
7947 (vectorizable_reduction): Likewise.
7948 (vectorizable_induction): Likewise.
7949 * tree-vect-loop.c (vect_analyze_loop_operations): Create local
7950 cost vector to pass to vectorizable_ and record afterwards.
7951 (vect_model_reduction_cost): Take cost vector argument and adjust.
7952 (vect_model_induction_cost): Likewise.
7953 (vectorizable_reduction): Likewise.
7954 (vectorizable_induction): Likewise.
7955 (vectorizable_live_operation): Likewise.
7956 * tree-vect-slp.c (vect_create_new_slp_node): Initialize
7957 SLP_TREE_NUMBER_OF_VEC_STMTS.
7958 (vect_analyze_slp_cost_1): Remove.
7959 (vect_analyze_slp_cost): Likewise.
7960 (vect_slp_analyze_node_operations): Take visited args and
7961 a target cost vector. Avoid processing already visited stmt sets.
7962 (vect_slp_analyze_operations): Use a local cost vector to gather
7963 costs and register those of non-discarded instances.
7964 (vect_bb_vectorization_profitable_p): Use add_stmt_costs.
7965 (vect_schedule_slp_instance): Remove copying of
7966 SLP_TREE_NUMBER_OF_VEC_STMTS. Instead assert that it is not
7968 * tree-vect-stmts.c (record_stmt_cost): Remove path directly
7969 adding cost. Record cost entry location.
7970 (vect_prologue_cost_for_slp_op): Function to compute cost of
7971 a constant or invariant generated for SLP vect in the prologue,
7972 split out from vect_analyze_slp_cost_1.
7973 (vect_model_simple_cost): Make static. Adjust for SLP costing.
7974 (vect_model_promotion_demotion_cost): Likewise.
7975 (vect_model_store_cost): Likewise, make static.
7976 (vect_model_load_cost): Likewise.
7977 (vectorizable_bswap): Add cost vector arg and adjust.
7978 (vectorizable_call): Likewise.
7979 (vectorizable_simd_clone_call): Likewise.
7980 (vectorizable_conversion): Likewise.
7981 (vectorizable_assignment): Likewise.
7982 (vectorizable_shift): Likewise.
7983 (vectorizable_operation): Likewise.
7984 (vectorizable_store): Likewise.
7985 (vectorizable_load): Likewise.
7986 (vectorizable_condition): Likewise.
7987 (vectorizable_comparison): Likewise.
7988 (can_vectorize_live_stmts): Likewise.
7989 (vect_analyze_stmt): Likewise.
7990 (vect_transform_stmt): Adjust calls to vectorizable_*.
7991 * tree-vectorizer.c: Include gimple-pretty-print.h.
7992 (dump_stmt_cost): New function.
7994 2018-05-16 Richard Biener <rguenther@suse.de>
7996 * params.def (PARAM_DSE_MAX_ALIAS_QUERIES_PER_STORE): New param.
7997 * doc/invoke.texi (dse-max-alias-queries-per-store): Document.
7998 * tree-ssa-dse.c: Include tree-ssa-loop.h.
7999 (check_name): New callback.
8000 (dse_classify_store): Track cycles via a visited bitmap of PHI
8001 defs and simplify handling of in-loop and across loop dead stores
8002 and properly fail for loop-variant refs. Handle byte-tracking with
8003 multiple defs. Use PARAM_DSE_MAX_ALIAS_QUERIES_PER_STORE for
8006 2018-05-16 Richard Sandiford <richard.sandiford@linaro.org>
8008 * tree-vectorizer.h (vect_get_vector_types_for_stmt): Declare.
8009 (vect_get_mask_type_for_stmt): Likewise.
8010 * tree-vect-slp.c (vect_two_operations_perm_ok_p): New function,
8012 (vect_build_slp_tree_1): ...here. Use vect_get_vector_types_for_stmt
8013 to determine the statement's vector type and the vector type that
8014 should be used for calculating nunits. Deal with cases in which
8015 the type has to be deferred.
8016 (vect_slp_analyze_node_operations): Use vect_get_vector_types_for_stmt
8017 and vect_get_mask_type_for_stmt to calculate STMT_VINFO_VECTYPE.
8018 * tree-vect-loop.c (vect_determine_vf_for_stmt_1)
8019 (vect_determine_vf_for_stmt): New functions, split out from...
8020 (vect_determine_vectorization_factor): ...here.
8021 * tree-vect-stmts.c (vect_get_vector_types_for_stmt)
8022 (vect_get_mask_type_for_stmt): New functions, split out from
8023 vect_determine_vectorization_factor.
8025 2018-05-16 Richard Biener <rguenther@suse.de>
8027 * tree-cfg.c (verify_gimple_assign_ternary): Properly
8028 verify the [VEC_]COND_EXPR embedded comparison.
8030 2018-05-15 Martin Sebor <msebor@redhat.com>
8032 PR tree-optimization/85753
8033 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Handle
8034 RECORD_TYPE in addition to ARRAY_TYPE.
8036 2018-05-15 Martin Sebor <msebor@redhat.com>
8039 * calls.c (get_attr_nonstring_decl): Handle MEM_REF.
8041 2018-05-15 Richard Biener <rguenther@suse.de>
8043 * tree-ssa-dse.c (dse_classify_store): Remove use_stmt parameter,
8044 add by_clobber_p one. Change algorithm to collect all defs
8045 representing uses we need to walk and try reducing them to
8046 a single one before failing.
8047 (dse_dom_walker::dse_optimize_stmt): Adjust.
8049 2018-05-13 Mark Wielaard <mark@klomp.org>
8051 * dwarf2out.c (dwarf_OP): Handle DW_OP_addrx and DW_OP_constx.
8052 (size_of_loc_descr): Likewise.
8053 (output_loc_operands): Likewise.
8054 (output_loc_operands_raw): Likewise.
8055 (dw_addr_op): Use dwarf_OP () for DW_OP_constx and DW_OP_addrx.
8056 (resolve_addr_in_expr): Handle DW_OP_addrx and DW_OP_constx.
8057 (hash_loc_operands): Likewise.
8058 (compare_loc_operands): Likewise.
8060 2018-05-14 Mark Wielaard <mark@klomp.org>
8062 * dwarf2out.c (count_index_addrs): New function.
8063 (dwarf2out_finish): Use count_index_addrs to calculate addrs_length.
8065 2018-05-15 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
8067 PR tree-optimization/83648
8068 * ipa-pure-const.c (malloc_candidate_p): Allow function with NULL
8069 return value as malloc candidate.
8071 2018-05-15 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
8074 * ipa-pure-const.c (warn_function_malloc): Pass value of known_finite
8075 param as true in call to suggest_attribute.
8077 2018-05-14 Segher Boessenkool <segher@kernel.crashing.org>
8079 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete @itemx for
8080 -mreadonly-in-sdata.
8082 2018-05-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8084 * config/aarch64/aarch64-simd.md (*aarch64_crypto_aese_fused):
8086 (aarch64_crypto_aesd_fused): Likewise.
8088 2018-05-14 Wilco Dijkstra <wdijkstr@arm.com>
8090 * config/aarch64/aarch64.md (mov<mode>): Remove '*' in alternatives.
8091 (movsi_aarch64): Likewise.
8092 (load_pairsi): Likewise.
8093 (load_pairdi): Likewise.
8094 (store_pairsi): Likewise.
8095 (store_pairdi): Likewise.
8096 (load_pairsf): Likewise.
8097 (load_pairdf): Likewise.
8098 (store_pairsf): Likewise.
8099 (store_pairdf): Likewise.
8100 (zero_extend): Likewise.
8101 (trunc): Swap alternatives.
8102 (fcvt_target): Add '?' to prefer w over r.
8104 2018-05-14 Jakub Jelinek <jakub@redhat.com>
8107 * config/i386/i386.md: Disallow non-commutative arithmetics in
8108 last twpeephole for mem {+,-,&,|,^}= x; mem != 0 after cmpelim
8109 optimization. Use COMMUTATIVE_ARITH_P test rather than != MINUS
8110 in the peephole2 before it.
8112 2018-05-14 Sebastian Peryt <sebastian.peryt@intel.com>
8114 * common/config/i386/i386-common.c (OPTION_MASK_ISA_CLDEMOTE_SET,
8115 OPTION_MASK_ISA_CLDEMOTE_UNSET): New defines.
8116 (ix86_handle_option): Handle -mcldemote.
8117 * config.gcc: New header.
8118 * config/i386/cldemoteintrin.h: New file.
8119 * config/i386/cpuid.h (bit_CLDEMOTE): New bit.
8120 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
8122 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
8123 OPTION_MASK_ISA_CLDEMOTE.
8124 * config/i386/i386.c (ix86_target_string): Add -mcldemote.
8125 (ix86_valid_target_attribute_inner_p): Ditto.
8126 (enum ix86_builtins): Add IX86_BUILTIN_CLDEMOTE.
8127 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_cldemote.
8128 (ix86_expand_builtin): Expand IX86_BUILTIN_CLDEMOTE.
8129 * config/i386/i386.h (TARGET_CLDEMOTE, TARGET_CLDEMOTE_P): New.
8130 * config/i386/i386.md (UNSPECV_CLDEMOTE): New.
8132 * config/i386/i386.opt: Add -mcldemote.
8133 * config/i386/x86intrin.h: New header.
8134 * doc/invoke.texi: Add -mcldemote.
8136 2018-05-14 Richard Biener <rguenther@suse.de>
8138 * doc/match-and-simplify.texi: Adjust :s documentation.
8140 2018-05-14 Alexander Monakov <amonakov@ispras.ru>
8142 * sort.cc (REORDER_23): Pass the type for the temporaries instead of
8143 intended memcpy size.
8144 (REORDER_45): Likewise.
8146 2018-05-13 Alexander Monakov <amonakov@ispras.ru>
8148 * sort.cc: New file.
8149 * system.h [!CHECKING_P] (qsort): Redirect to gcc_qsort.
8150 * vec.c (qsort_chk): Use gcc_qsort.
8151 * Makefile.in (OBJS-libcommon): Add sort.o.
8152 (build/sort.o): New target. Use it...
8153 (BUILD_RTL): ... here, and...
8154 (build/gencfn-macros): ... here, and...
8155 (build/genmatch): ... here.
8157 2018-05-13 Kito Cheng <kito.cheng@gmail.com>
8158 Chung-Ju Wu <jasonwucj@gmail.com>
8160 * config.gcc (nds32*-*-*): Check that n15 is valid to --with-cpu.
8161 * config/nds32/nds32-graywolf.md: New file.
8162 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_GRAYWOLF.
8163 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n15
8165 * config/nds32/nds32-protos.h: More declarations for n15 pipeline.
8166 * config/nds32/nds32-utils.c: More implementations for n15 pipeline.
8167 * config/nds32/nds32.md (pipeline_model): Add graywolf.
8168 * config/nds32/nds32.opt (mcpu): Support n15 pipeline cpus.
8169 * config/nds32/pipelines.md: Include n15 settings.
8171 2018-05-13 Kito Cheng <kito.cheng@gmail.com>
8172 Chung-Ju Wu <jasonwucj@gmail.com>
8174 * config.gcc (nds32*-*-*): Check that n12/n13 are valid to --with-cpu.
8175 * config/nds32/nds32-n13.md: New file.
8176 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N12 and CPU_N13.
8177 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n12/n13
8179 * config/nds32/nds32-protos.h: More declarations for n12/n13 pipeline.
8180 * config/nds32/nds32.md (pipeline_model): Add n13.
8181 * config/nds32/nds32.opt (mcpu): Support n12/n13 pipeline cpus.
8182 * config/nds32/pipelines.md: Include n13 settings.
8184 2018-05-13 Kito Cheng <kito.cheng@gmail.com>
8185 Chung-Ju Wu <jasonwucj@gmail.com>
8187 * config.gcc (nds32*-*-*): Check that n10/d10 are valid to --with-cpu.
8188 * config/nds32/nds32-n10.md: New file.
8189 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N10.
8190 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n10
8192 * config/nds32/nds32-protos.h: More declarations for n10 pipeline.
8193 * config/nds32/nds32-utils.c: More implementations for n10 pipeline.
8194 * config/nds32/nds32.md (pipeline_model): Add n10.
8195 * config/nds32/nds32.opt (mcpu): Support n10 pipeline cpus.
8196 * config/nds32/pipelines.md: Include n10 settings.
8198 2018-05-13 Monk Chiang <sh.chiang04@gmail.com>
8199 Kito Cheng <kito.cheng@gmail.com>
8200 Chung-Ju Wu <jasonwucj@gmail.com>
8202 * config.gcc (nds32be-*-*): Handle --with-ext-dsp.
8203 * config/nds32/constants.md (unspec_element, unspec_volatile_element):
8204 Add enum values for DSP extension instructions.
8205 * config/nds32/constraints.md (Iu06, IU06, CVp5, CVs5, CVs2, CVhi):
8207 * config/nds32/iterators.md (shifts, shiftrt, sat_plus, all_plus,
8208 sat_minus, all_minus, plus_minus, extend, sumax, sumin, sumin_max):
8210 (su, zs, uk, opcode, add_rsub, add_sub): New code attributes.
8211 * config/nds32/nds32-dspext.md: New file for DSP implementation.
8212 * config/nds32/nds32-intrinsic.c: Implementation of DSP extension.
8213 * config/nds32/nds32-intrinsic.md: Likewise.
8214 * config/nds32/nds32_intrinsic.h: Likewise.
8215 * config/nds32/nds32-md-auxiliary.c: Likewise.
8216 * config/nds32/nds32-memory-manipulation.c: Consider DSP extension.
8217 * config/nds32/nds32-predicates.c (const_vector_to_hwint): New.
8218 (nds32_valid_CVp5_p, nds32_valid_CVs5_p): New.
8219 (nds32_valid_CVs2_p, nds32_valid_CVhi_p): New.
8220 * config/nds32/nds32-protos.h: New declarations for DSP extension.
8221 * config/nds32/nds32-utils.c (extract_mac_non_acc_rtx): New case
8222 TYPE_DMAC in switch statement.
8223 * config/nds32/nds32.c: New checking and implementation for DSP
8224 extension instructions.
8225 * config/nds32/nds32.h: Likewise.
8226 * config/nds32/nds32.md: Likewise.
8227 * config/nds32/nds32.opt (mhw-abs, mext-dsp): New options.
8228 * config/nds32/predicates.md: Implement new predicates for DSP
8231 2018-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
8233 * config/rs6000/rs6000.md (mov<mode>_softfloat, FMOVE32):
8234 Reformat alternatives and attributes so it is easier to identify
8235 which constraints/attributes go with which instruction.
8236 (mov<mode>_hardfloat32, FMOVE64): Likewise.
8237 (mov<mode>_softfloat32, FMOVE64): Likewise.
8238 (mov<mode>_hardfloat64, FMOVE64): Likewise.
8239 (mov<mode>_softfloat64, FMOVE64): Likewise.
8241 2018-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
8243 * doc/extend.texi (PowerPC Built-in Functions): Rename this
8245 (Basic PowerPC Built-in Functions): The new name of the
8246 subsection previously known as "PowerPC Built-in Functions".
8247 (Basic PowerPC Built-in Functions Available on all Configurations):
8249 (Basic PowerPC Built-in Functions Available on ISA 2.05): Likewise.
8250 (Basic PowerPC Built-in Functions Available on ISA 2.06): Likewise.
8251 (Basic PowerPC Built-in Functions Available on ISA 2.07): Likewise.
8252 (Basic PowerPC Built-in Functions Available on ISA 3.0): Likewise.
8254 2018-05-11 Martin Jambor <mjambor@suse.cz>
8257 * ipa-cp.c (intersect_with_plats): Check that the lattice contains
8260 2018-05-11 Richard Earnshaw <rearnsha@arm.com>
8263 * config/arm/arm-cpus.in (fgroup ARMv6m): Add be8 feature.
8265 2018-05-11 Sebastian Peryt <sebastian.peryt@intel.com>
8267 * common/config/i386/i386-common.c (OPTION_MASK_ISA_WAITPKG_SET,
8268 OPTION_MASK_ISA_WAITPKG_UNSET): New defines.
8269 (ix86_handle_option): Handle -mwaitpkg.
8270 * config.gcc: New header.
8271 * config/i386/cpuid.h (bit_WAITPKG): New bit.
8272 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mwaitpkg.
8273 * config/i386/i386-builtin-types.def ((UINT8, UNSIGNED, UINT64)): New
8275 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
8276 OPTION_MASK_ISA_WAITPKG.
8277 * config/i386/i386.c (ix86_target_string): Add -mwaitpkg.
8278 (ix86_option_override_internal): Add PTA_WAITPKG.
8279 (ix86_valid_target_attribute_inner_p): Add -mwaitpkg.
8280 (enum ix86_builtins): Add IX86_BUILTIN_UMONITOR, IX86_BUILTIN_UMWAIT,
8281 IX86_BUILTIN_TPAUSE.
8282 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_umonitor,
8283 __builtin_ia32_umwait and __builtin_ia32_tpause.
8284 (ix86_expand_builtin): Expand IX86_BUILTIN_UMONITOR,
8285 IX86_BUILTIN_UMWAIT, IX86_BUILTIN_TPAUSE.
8286 * config/i386/i386.h (TARGET_WAITPKG, TARGET_WAITPKG_P): New.
8287 * config/i386/i386.md (UNSPECV_UMWAIT, UNSPECV_UMONITOR,
8288 UNSPECV_TPAUSE): New.
8289 (umwait, umwait_rex64, umonitor_<mode>, tpause, tpause_rex64): New.
8290 * config/i386/i386.opt: Add -mwaitpkg.
8291 * config/i386/waitpkgintrin.h: New file.
8292 * config/i386/x86intrin.h: New header.
8293 * doc/invoke.texi: Add -mwaitpkg.
8295 2018-05-11 Richard Earnshaw <rearnsha@arm.com>
8298 * config/arm/arm-cpus.in: Add comment that ARMv6-m and ARMv6S-m are now
8300 (cortex-m0): Use armv6s-m isa.
8301 (cortex-m0plus): Likewise.
8302 (cortex-m1): Likewise.
8303 (cortex-m0.small-multiply): Likewise.
8304 (cortex-m0plus.small-multiply): Likewise.
8305 (cortex-m1.small-multiply): Likewise.
8307 2018-05-11 Allan Sandfeld Jensen <allan.jensen@qt.io>
8308 Jakub Jelinek <jakub@redhat.com>
8310 PR tree-optimization/85692
8311 * tree-ssa-forwprop.c (simplify_vector_constructor): Try two
8312 source permute as well.
8314 2018-05-11 Martin Liska <mliska@suse.cz>
8317 * doc/extend.texi: Document LLVM style format for no_sanitize
8320 2018-05-10 Michael Meissner <meissner@linux.vnet.ibm.com>
8322 * config/rs6000/rs6000.c (mode_supports_dq_form): Rename
8323 mode_supports_vsx_dform_quad to mode_supports_dq_form.
8324 (mode_supports_vsx_dform_quad): Likewise.
8325 (mode_supports_vmx_dform): Move these functions to be next to the
8326 other mode_supports functions.
8327 (mode_supports_dq_form): Likewise.
8328 (quad_address_p): Change calls of mode_supports_vsx_dform_quad to
8329 mode_supports_dq_form.
8330 (reg_offset_addressing_ok_p): Likewise.
8331 (offsettable_ok_by_alignment): Likewise.
8332 (rs6000_legitimate_offset_address_p): Likewise.
8333 (legitimate_lo_sum_address_p): Likewise.
8334 (rs6000_legitimize_address): Likewise.
8335 (rs6000_legitimize_reload_address): Likewise.
8336 (rs6000_secondary_reload_inner): Likewise.
8337 (rs6000_preferred_reload_class): Likewise.
8338 (rs6000_output_move_128bit): Likewise.
8340 2018-05-10 Uros Bizjak <ubizjak@gmail.com>
8342 * config/i386/i386.c (ix86_expand_builtin) <case IX86_BUILTIN_RDPID>:
8343 Generate SImode target register for null target.
8344 <case IX86_BUILTIN_XGETBV>: Ditto.
8345 <case IX86_BUILTIN_XSETBV>: Optimize LSHIFTRT generation.
8346 * config/i386/xsaveintrin.h (_xgetbv): Add missing return.
8348 2018-05-10 Carl Love <cel@us.ibm.com>
8350 * config/rs6000/rs6000.md (prefetch): Generate ISA 2.06 instructions
8351 dcbtt and dcbtstt if operands[2] is 0.
8353 2018-05-10 Uros Bizjak <ubizjak@gmail.com>
8356 * config/i386/sse.md (usadv64qi): New expander.
8358 2018-05-10 Segher Boessenkool <segher@kernel.crashing.org>
8360 * config/rs6000/altivec.md (altivec_vmrghb, altivec_vmrghh,
8361 altivec_vmrghw, altivec_vmrglb, altivec_vmrglh, altivec_vmrglw): Remove
8362 -maltivec=be support.
8363 (vec_widen_umult_even_v16qi, vec_widen_smult_even_v16qi,
8364 vec_widen_umult_even_v8hi, vec_widen_smult_even_v8hi,
8365 vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
8366 vec_widen_umult_odd_v16qi, vec_widen_smult_odd_v16qi,
8367 vec_widen_umult_odd_v8hi, vec_widen_smult_odd_v8hi,
8368 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si, altivec_vpkpx,
8369 altivec_vpks<VI_char>ss, altivec_vpks<VI_char>us,
8370 altivec_vpku<VI_char>us, altivec_vpku<VI_char>um, altivec_vsum2sws,
8371 altivec_vsumsws): Adjust.
8372 (altivec_vspltb *altivec_vspltb_internal, altivec_vsplth,
8373 *altivec_vsplth_internal, altivec_vspltw, *altivec_vspltw_internal,
8374 altivec_vspltsf, *altivec_vspltsf_internal): Remove -maltivec=be
8376 (altivec_vperm_<mode>, altivec_vperm_<mode>_uns,
8377 altivec_vupkhs<VU_char>, altivec_vupkls<VU_char>, altivec_vupkhpx,
8378 altivec_vupklpx, altivec_lvsl, altivec_lvsr): Adjust.
8379 (altivec_lve<VI_char>x): Delete expand.
8380 (*altivec_lve<VI_char>x_internal): Rename to...
8381 (altivec_lve<VI_char>x): ... this.
8382 (altivec_lvxl_<mode>): Delete expand.
8383 (*altivec_lvxl_<mode>_internal): Rename to ...
8384 (altivec_lvxl_<mode>): ... this.
8385 (altivec_stvxl_<mode>): Delete expand.
8386 (*altivec_stvxl_<mode>_internal): Rename to ...
8387 (altivec_stvxl_<mode>): ... this.
8388 (altivec_stve<VI_char>x): Delete expand.
8389 (*altivec_stve<VI_char>x_internal): Rename to ...
8390 (altivec_stve<VI_char>x): ... this.
8391 (doublee<mode>2, unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2,
8392 doubleh<mode>2, unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2,
8393 reduc_plus_scal_<mode>): Adjust.
8394 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Adjust
8396 (rs6000_cpu_cpp_builtins): Adjust.
8397 (altivec_resolve_overloaded_builtin): Remove -maltivec=be support.
8398 * config/rs6000/rs6000-protos.h (altivec_expand_lvx_be,
8399 altivec_expand_stvx_be, altivec_expand_stvex_be): Delete.
8400 * config/rs6000/rs6000.c (rs6000_option_override_internal): Remove
8401 -maltivec=be support.
8402 (rs6000_split_vec_extract_var): Adjust.
8403 (rs6000_split_v4si_init): Adjust.
8404 (swap_selector_for_mode): Delete.
8405 (altivec_expand_lvx_be, altivec_expand_stvx_be,
8406 altivec_expand_stvex_be): Delete.
8407 (altivec_expand_lv_builtin, altivec_expand_stv_builtin): Remove
8408 -maltivec=be support.
8409 (rs6000_gimple_fold_builtin): Ditto.
8410 (rs6000_generate_float2_double_code, rs6000_generate_float2_code):
8412 * config/rs6000/rs6000.h (VECTOR_ELT_ORDER_BIG): Delete.
8413 (TARGET_DIRECT_MOVE_64BIT): Adjust.
8414 * config/rs6000/rs6000.md (split for extendsidi2 for vectors): Adjust.
8415 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Delete.
8416 * config/rs6000/vsx.md (floate<mode>, unsfloatev2di, floato<mode>,
8417 unsfloatov2di, vsignedo_v2df, vsignede_v2df, vunsignedo_v2df,
8418 vunsignede_v2df, vsx_extract_<mode>_p9, *vsx_extract_si,
8419 *vsx_extract_<mode>_p8, *vsx_extract_si_<uns>float_df,
8420 *vsx_extract_si_<uns>float_<mode>, vsx_set_<mode>_p9, vsx_set_v4sf_p9,
8421 *vsx_insert_extract_v4sf_p9, *vsx_insert_extract_v4sf_p9_2, and an
8422 anonymous split): Adjust.
8423 (vsx_mergel_<mode>, vsx_mergeh_<mode>): Remove -maltivec=be support.
8424 (vsx_xxspltd_<mode>, extract4b, insert4b): Adjust.
8426 2018-05-10 Eric Botcazou <ebotcazou@adacore.com>
8428 * configure.ac (gcc_gxx_include_dir_add_sysroot): Set it to 1 only
8429 when --with-gxx-include-dir is also specified.
8430 * configure: Regenerate.
8432 2018-05-09 Jim Wilson <jimw@sifive.com>
8435 * config.gcc (riscv*-*-*): Handle --with-multilib-list.
8436 * config/riscv/t-withmultilib: New.
8437 * config/riscv/withmultilib.h: New.
8438 * doc/install.texi: Document RISC-V --with-multilib-list support.
8440 2018-05-09 Richard Biener <rguenther@suse.de>
8442 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Fill a cost
8444 (vect_bb_vectorization_profitable_p): Adjust. Compute
8445 actual scalar cost using the cost vector and the add_stmt_cost
8448 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
8450 PR rtl-optimization/85645
8451 * config/rs6000/rs6000.c (rs6000_emit_prologue_components): Put a SET
8452 in the REG_CFA_REGISTER note for LR, don't leave it empty.
8454 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
8456 PR rtl-optimization/85645
8457 * shrink-wrap.c (spread_components): Return a boolean saying if
8458 anything was changed.
8459 (try_shrink_wrapping_separate): Iterate spread_components until
8460 nothing changes anymore.
8462 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
8464 PR rtl-optimization/85645
8465 * regrename.c (build_def_use): Also kill the chains that include the
8466 destination of a REG_CFA_REGISTER note.
8468 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
8470 PR rtl-optimization/85645
8471 * regcprop.c (copyprop_hardreg_forward_1): Don't propagate into an
8472 insn that has a REG_CFA_REGISTER note.
8474 2018-05-09 Richard Sandiford <richard.sandiford@linaro.org>
8476 * cfgexpand.c (expand_clobber): New function.
8477 (expand_gimple_stmt_1): Use it.
8478 * tree-vect-stmts.c (vect_clobber_variable): New function,
8480 (vectorizable_simd_clone_call): ...here.
8481 (vectorizable_store): Emit a clobber either side of an
8482 IFN_STORE_LANES sequence.
8483 (vectorizable_load): Emit a clobber after an IFN_LOAD_LANES sequence.
8485 2018-05-09 Tom de Vries <tom@codesourcery.com>
8488 * config/nvptx/nvptx.md (define_insn "trap", define_insn "trap_if_true")
8489 (define_insn "trap_if_false"): Add exit after trap.
8491 2018-05-09 Eric Botcazou <ebotcazou@adacore.com>
8493 PR rtl-optimization/85638
8494 * bb-reorder.c: Include common/common-target.h.
8495 (create_forwarder_block): New function extracted from...
8496 (fix_up_crossing_landing_pad): ...here. Rename into...
8497 (dw2_fix_up_crossing_landing_pad): ...this.
8498 (sjlj_fix_up_crossing_landing_pad): New function.
8499 (find_rarely_executed_basic_blocks_and_crossing_edges): In SJLJ mode,
8500 call sjlj_fix_up_crossing_landing_pad if there are incoming EH edges
8501 from both partitions and exit the loop after one iteration.
8503 2018-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
8506 * doc/extend.texi (PowerPC Built-in Functions): Rename this
8508 (Basic PowerPC Built-in Functions): The new name of the
8509 subsection previously known as "PowerPC Built-in Functions".
8510 (Basic PowerPC Built-in Functions Available on all Configurations):
8512 (Basic PowerPC Built-in Functions Available on ISA 2.05): New
8514 (Basic PowerPC Built-in Functions Available on ISA 2.06): New
8516 (Basic PowerPC Built-in Functions Available on ISA 2.07): New
8518 (Basic PowerPC Built-in Functions Available on ISA 3.0): New
8521 2018-05-08 Jim Wilson <jimw@sifive.com>
8523 * config/riscv/linux.h (MUSL_ABI_SUFFIX): Delete unnecessary backslash.
8524 (LD_EMUL_SUFFIX): New.
8525 (LINK_SPEC): Use it.
8527 2018-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
8529 * doc/extend.texi (PowerPC Built-in Functions): Rename this
8531 (Basic PowerPC Built-in Functions): The new name of the
8532 subsection previously known as "PowerPC Built-in Functions".
8533 (Basic PowerPC Built-in Functions Available on all Configurations):
8535 (Basic PowerPC Built-in Functions Available on ISA 2.05): New
8537 (Basic PowerPC Built-in Functions Available on ISA 2.06): New
8539 (Basic PowerPC Built-in Functions Available on ISA 2.07): New
8541 (Basic PowerPC Built-in Functions Available on ISA 3.0): New
8544 2018-05-08 Jakub Jelinek <jakub@redhat.com>
8547 * config/i386/i386.md: Add peepholes for mem {+,-,&,|,^}= x; mem != 0
8548 after cmpelim optimization.
8550 2018-05-08 Olga Makhotina <olga.makhotina@intel.com>
8552 * config.gcc: Support "goldmont".
8553 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "goldmont".
8554 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
8556 * config/i386/i386.c (m_GOLDMONT): Define.
8557 (processor_target_table): Add "goldmont".
8558 (PTA_GOLDMONT): Define.
8559 (ix86_lea_outperforms): Add TARGET_GOLDMONT.
8560 (get_builtin_code_for_version): Handle PROCESSOR_GOLDMONT.
8561 (fold_builtin_cpu): Add M_INTEL_GOLDMONT.
8562 (fold_builtin_cpu): Add "goldmont".
8563 (ix86_add_stmt_cost): Add TARGET_GOLDMONT.
8564 (ix86_option_override_internal): Add "goldmont".
8565 * config/i386/i386.h (processor_costs): Define TARGET_GOLDMONT.
8566 (processor_type): Add PROCESSOR_GOLDMONT.
8567 * config/i386/i386.md: Add CPU "glm".
8568 * config/i386/glm.md: New file.
8569 * config/i386/x86-tune.def: Add m_GOLDMONT.
8570 * doc/invoke.texi: Add goldmont as x86 -march=/-mtune= CPU type.
8572 2018-05-08 Jakub Jelinek <jakub@redhat.com>
8575 * config/i386/i386.c (ix86_expand_sse2_abs): Handle E_V2DImode and
8577 * config/i386/sse.md (abs<mode>2): Use VI_AVX2 iterator instead of
8578 VI1248_AVX512VL_AVX512BW. Handle V2DImode and V4DImode if not
8579 TARGET_AVX512VL using ix86_expand_sse2_abs. Formatting fixes.
8582 * config/i386/i386.c (ix86_fold_builtin): Handle
8583 IX86_BUILTIN_{,P}MOVMSK{PS,PD,B}{,128,256}.
8586 * config/i386/sse.md (ssequaterinsnmode): New mode attribute.
8587 (*<extract_type>_vinsert<shuffletype><extract_suf>_0): New pattern.
8589 2018-05-08 Richard Earnshaw <rearnsha@arm.com>
8592 * config/arm/parsecpu.awk (check_cpu): Fix operator precedence.
8593 (check_arch): Likewise.
8594 (check_fpu): Return the result rather than printing it.
8595 (end arch): Fix operator precedence.
8596 (end cpu): Likewise.
8597 (END): Print the result from check_fpu.
8599 2018-05-08 Richard Sandiford <richard.sandiford@linaro.org>
8600 Alan Hayward <alan.hayward@arm.com>
8601 David Sherwood <david.sherwood@arm.com>
8603 * config/aarch64/aarch64-sve.md (*pred_cmp<cmp_op><mode>_combine)
8604 (*pred_cmp<cmp_op><mode>, *fcm<cmp_op><mode>_and_combine)
8605 (*fcmuo<mode>_and_combine, *fcm<cmp_op><mode>_and)
8606 (*fcmuo<mode>_and): New patterns.
8608 2018-05-08 Richard Sandiford <richard.sandiford@linaro.org>
8610 * config/aarch64/iterators.md (UNSPEC_COND_LO, UNSPEC_COND_LS)
8611 (UNSPEC_COND_HI, UNSPEC_COND_HS, UNSPEC_COND_UO): Delete.
8612 (SVE_INT_CMP, SVE_FP_CMP): New code iterators.
8613 (cmp_op, sve_imm_con): New code attributes.
8614 (SVE_COND_INT_CMP, imm_con): Delete.
8615 (cmp_op): Remove above unspecs from int attribute.
8616 * config/aarch64/aarch64-sve.md (*vec_cmp<cmp_op>_<mode>): Rename
8618 (*cmp<cmp_op><mode>): ...this. Use UNSPEC_MERGE_PTRUE instead of
8619 comparison-specific unspecs.
8620 (*vec_cmp<cmp_op>_<mode>_ptest): Rename to...
8621 (*cmp<cmp_op><mode>_ptest): ...this and adjust likewise.
8622 (*vec_cmp<cmp_op>_<mode>_cc): Rename to...
8623 (*cmp<cmp_op><mode>_cc): ...this and adjust likewise.
8624 (*vec_fcm<cmp_op><mode>): Rename to...
8625 (*fcm<cmp_op><mode>): ...this and adjust likewise.
8626 (*vec_fcmuo<mode>): Rename to...
8627 (*fcmuo<mode>): ...this and adjust likewise.
8628 (*pred_fcm<cmp_op><mode>): New pattern.
8629 * config/aarch64/aarch64.c (aarch64_emit_unop, aarch64_emit_binop)
8630 (aarch64_emit_sve_ptrue_op, aarch64_emit_sve_ptrue_op_cc): New
8632 (aarch64_unspec_cond_code): Remove handling of LTU, GTU, LEU, GEU
8634 (aarch64_gen_unspec_cond, aarch64_emit_unspec_cond): Delete.
8635 (aarch64_emit_sve_predicated_cond): New function.
8636 (aarch64_expand_sve_vec_cmp_int): Use aarch64_emit_sve_ptrue_op_cc.
8637 (aarch64_emit_unspec_cond_or): Replace with...
8638 (aarch64_emit_sve_or_conds): ...this new function. Use
8639 aarch64_emit_sve_ptrue_op for the individual comparisons and
8640 aarch64_emit_binop to OR them together.
8641 (aarch64_emit_inverted_unspec_cond): Replace with...
8642 (aarch64_emit_sve_inverted_cond): ...this new function. Use
8643 aarch64_emit_sve_ptrue_op for the comparison and
8644 aarch64_emit_unop to invert the result.
8645 (aarch64_expand_sve_vec_cmp_float): Update after the above
8646 changes. Use aarch64_emit_sve_ptrue_op for native comparisons.
8648 2018-05-07 Nathan Sidwell <nathan@acm.org>
8650 * doc/invoke.texi (C++ Dialect Options): Remove -ffor-scope.
8651 * doc/extend.texi (Deprecated Features): Remove -fno-for-scope
8652 (Backwards Compatibility): Likewise.
8654 2018-05-07 Luis Machado <luis.machado@linaro.org>
8658 2018-05-07 Luis Machado <luis.machado@linaro.org>
8660 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
8661 <prefetch_dynamic_strides>: New const bool field.
8662 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
8663 prefetch_dynamic_strides.
8664 (exynosm1_prefetch_tune): Likewise.
8665 (thunderxt88_prefetch_tune): Likewise.
8666 (thunderx_prefetch_tune): Likewise.
8667 (thunderx2t99_prefetch_tune): Likewise.
8668 (qdf24xx_prefetch_tune): Likewise. Set prefetch_dynamic_strides
8670 (aarch64_override_options_internal): Update to set
8671 PARAM_PREFETCH_DYNAMIC_STRIDES.
8672 * doc/invoke.texi (prefetch-dynamic-strides): Document new option.
8673 * params.def (PARAM_PREFETCH_DYNAMIC_STRIDES): New.
8674 * params.h (PARAM_PREFETCH_DYNAMIC_STRIDES): Define.
8675 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Account for
8676 prefetch-dynamic-strides setting.
8678 2018-05-07 Luis Machado <luis.machado@linaro.org>
8680 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
8681 <minimum_stride>: New const int field.
8682 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
8683 minimum_stride field.
8684 (exynosm1_prefetch_tune): Likewise.
8685 (thunderxt88_prefetch_tune): Likewise.
8686 (thunderx_prefetch_tune): Likewise.
8687 (thunderx2t99_prefetch_tune): Likewise.
8688 (qdf24xx_prefetch_tune): Likewise. Set minimum_stride to 2048.
8689 (aarch64_override_options_internal): Update to set
8690 PARAM_PREFETCH_MINIMUM_STRIDE.
8691 * doc/invoke.texi (prefetch-minimum-stride): Document new option.
8692 * params.def (PARAM_PREFETCH_MINIMUM_STRIDE): New.
8693 * params.h (PARAM_PREFETCH_MINIMUM_STRIDE): Define.
8694 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Return false if
8695 stride is constant and is below the minimum stride threshold.
8697 2018-05-07 Luis Machado <luis.machado@linaro.org>
8699 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune) <l2_cache_size>: Set
8702 2018-05-07 Luis Machado <luis.machado@linaro.org>
8704 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
8705 <prefetch_dynamic_strides>: New const bool field.
8706 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
8707 prefetch_dynamic_strides.
8708 (exynosm1_prefetch_tune): Likewise.
8709 (thunderxt88_prefetch_tune): Likewise.
8710 (thunderx_prefetch_tune): Likewise.
8711 (thunderx2t99_prefetch_tune): Likewise.
8712 (qdf24xx_prefetch_tune): Likewise. Set prefetch_dynamic_strides
8714 (aarch64_override_options_internal): Update to set
8715 PARAM_PREFETCH_DYNAMIC_STRIDES.
8716 * doc/invoke.texi (prefetch-dynamic-strides): Document new option.
8717 * params.def (PARAM_PREFETCH_DYNAMIC_STRIDES): New.
8718 * params.h (PARAM_PREFETCH_DYNAMIC_STRIDES): Define.
8719 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Account for
8720 prefetch-dynamic-strides setting.
8722 2018-05-07 Luis Machado <luis.machado@linaro.org>
8724 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
8725 <minimum_stride>: New const int field.
8726 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
8727 minimum_stride field.
8728 (exynosm1_prefetch_tune): Likewise.
8729 (thunderxt88_prefetch_tune): Likewise.
8730 (thunderx_prefetch_tune): Likewise.
8731 (thunderx2t99_prefetch_tune): Likewise.
8732 (qdf24xx_prefetch_tune): Likewise. Set minimum_stride to 2048.
8733 (aarch64_override_options_internal): Update to set
8734 PARAM_PREFETCH_MINIMUM_STRIDE.
8735 * doc/invoke.texi (prefetch-minimum-stride): Document new option.
8736 * params.def (PARAM_PREFETCH_MINIMUM_STRIDE): New.
8737 * params.h (PARAM_PREFETCH_MINIMUM_STRIDE): Define.
8738 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Return false if
8739 stride is constant and is below the minimum stride threshold.
8741 2018-05-06 Jakub Jelinek <jakub@redhat.com>
8744 * cfgexpand.c (expand_asm_stmt): Don't create a temporary if
8745 the type is addressable. Don't force op into register if it has
8748 2018-05-05 Roland McGrath <mcgrathr@google.com>
8751 * varasm.c (default_section_type_flags): Set SECTION_NOTYPE for
8752 any section for which we don't know a specific type it should have,
8753 regardless of name. Previously this was done only for the exact
8754 names ".init_array", ".fini_array", and ".preinit_array".
8755 (default_elf_asm_named_section): Add comment about
8756 relationship with default_section_type_flags and SECTION_NOTYPE.
8757 (get_section): Don't consider it a type conflict if one side has
8758 SECTION_NOTYPE and the other doesn't, as long as neither has the
8759 SECTION_BSS et al used in the default_section_type_flags logic.
8761 2018-05-05 Tom de Vries <tom@codesourcery.com>
8764 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG_3): Define.
8765 (workaround_barsyncs): New function.
8766 (nvptx_reorg): Use workaround_barsyncs.
8767 * config/nvptx/nvptx.md (define_c_enum "unspecv"): Add UNSPECV_MEMBAR.
8768 (define_expand "nvptx_membar_cta"): New define_expand.
8769 (define_insn "*nvptx_membar_cta"): New insn.
8771 2018-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
8773 * brig-builtins.def: Add consts to ptrs etc. in BRIG builtin defs.
8774 To improve optimization opportunities.
8775 * builtin-types.def: The new needed builtin types for the above.
8777 2018-05-04 Richard Biener <rguenther@suse.de>
8779 * bb-reorder.c (sanitize_hot_paths): Release hot_bbs_to_check.
8780 * gimple-ssa-store-merging.c
8781 (imm_store_chain_info::output_merged_store): Remove redundant create,
8782 release split_store vector contents on failure.
8783 * tree-vect-slp.c (vect_schedule_slp_instance): Avoid leaking
8784 scalar stmt vector on cache hit.
8786 2018-05-04 Segher Boessenkool <segher@kernel.crashing.org>
8788 * common/config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
8790 * config.gcc (powerpc-xilinx-eabi*): Remove.
8791 * config/rs6000/predicates.md (easy_fp_constant): Remove Xilinx FP
8793 (fusion_addis_mem_combo_load): Ditto.
8794 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove Xilinx
8796 (rs6000_cpu_cpp_builtins): Ditto.
8797 * config/rs6000/rs6000-linux.c
8798 (rs6000_linux_float_exceptions_rounding_supported_p): Ditto.
8799 * config/rs6000/rs6000-opts.h (enum fpu_type_t): Delete.
8800 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove Xilinx FP
8802 (rs6000_setup_reg_addr_masks): Ditto.
8803 (rs6000_init_hard_regno_mode_ok): Ditto.
8804 (rs6000_option_override_internal): Ditto.
8805 (legitimate_lo_sum_address_p): Ditto.
8806 (rs6000_legitimize_address): Ditto.
8807 (rs6000_legitimize_reload_address): Ditto.
8808 (rs6000_legitimate_address_p): Ditto.
8809 (abi_v4_pass_in_fpr): Ditto.
8810 (setup_incoming_varargs): Ditto.
8811 (rs6000_gimplify_va_arg): Ditto.
8812 (rs6000_split_multireg_move): Ditto.
8813 (rs6000_savres_strategy): Ditto.
8814 (rs6000_emit_prologue_components): Ditto.
8815 (rs6000_emit_epilogue_components): Ditto.
8816 (rs6000_emit_prologue): Ditto.
8817 (rs6000_emit_epilogue): Ditto.
8818 (rs6000_elf_file_end): Ditto.
8819 (rs6000_function_value): Ditto.
8820 (rs6000_libcall_value): Ditto.
8821 * config/rs6000/rs6000.h: Ditto.
8822 (TARGET_MINMAX_SF, TARGET_MINMAX_DF): Delete, merge to ...
8823 (TARGET_MINMAX): ... this. New.
8824 (TARGET_SF_FPR, TARGET_DF_FPR, TARGET_SF_INSN, TARGET_DF_INSN): Delete.
8825 * config/rs6000/rs6000.md: Remove Xilinx FP support.
8826 (*movsi_internal1_single): Delete.
8827 * config/rs6000/rs6000.opt (msingle-float, mdouble-float, msimple-fpu,
8828 mfpu=, mxilinx-fpu): Delete.
8829 * config/rs6000/singlefp.h: Delete.
8830 * config/rs6000/sysv4.h: Remove Xilinx FP support.
8831 * config/rs6000/t-rs6000: Ditto.
8832 * config/rs6000/t-xilinx: Delete.
8833 * config/rs6000/titan.md: Adjust for fp_type removal.
8834 * config/rs6000/vsx.md: Remove Xilinx FP support.
8835 (VStype_simple): Delete.
8836 (VSfptype_simple, VSfptype_mul, VSfptype_div, VSfptype_sqrt): Delete.
8837 * config/rs6000/xfpu.h: Delete.
8838 * config/rs6000/xfpu.md: Delete.
8839 * config/rs6000/xilinx.h: Delete.
8840 * config/rs6000/xilinx.opt: Delete.
8841 * doc/invoke.texi (RS/6000 and PowerPC Options): Remove
8842 -msingle-float, -mdouble-float, -msimple-fpu, -mfpu=, and -mxilinx-fpu.
8844 2018-05-04 Tom de Vries <tom@codesourcery.com>
8847 * builtins.c (expand_builtin_goacc_parlevel_id_size): Handle null target
8850 2018-05-04 Richard Biener <rguenther@suse.de>
8853 * tree-complex.c (update_complex_assignment): We are always in SSA form.
8854 (expand_complex_div_wide): Likewise.
8855 (expand_complex_operations_1): Likewise.
8856 (expand_complex_libcall): Preserve EH info of the original stmt.
8857 (tree_lower_complex): Handle removed blocks.
8858 * tree.c (build_common_builtin_nodes): Do not set ECF_NOTRHOW
8859 on complex multiplication and division libcall builtins.
8861 2018-05-04 Richard Biener <rguenther@suse.de>
8864 * fold-const.c (negate_expr_p): Restrict negation of operand
8865 zero of a division to when we know that can happen without
8867 (fold_negate_expr_1): Likewise.
8869 2018-05-04 Jakub Jelinek <jakub@redhat.com>
8872 * real.h (real_nextafter): Declare.
8873 * real.c (real_nextafter): New function.
8874 * fold-const-call.c (fold_const_nextafter): New function.
8875 (fold_const_call_sss): Call it for CASE_CFN_NEXTAFTER and
8876 CASE_CFN_NEXTTOWARD.
8877 (fold_const_call_1): For CASE_CFN_NEXTTOWARD call fold_const_call_sss
8878 even when arg1_mode is different from arg0_mode.
8880 2018-05-03 Nathan Sidwell <nathan@acm.org>
8882 * doc/extend.texi (Deprecated Features): Remove
8884 (Backwards Compatibility): Likewise.
8885 * doc/invoke.texi (C++ Language Options): Likewise.
8886 (C++ Dialect Options): Likewise.
8888 2018-05-03 Jakub Jelinek <jakub@redhat.com>
8891 * config/i386/avx512fintrin.h (_mm512_mullox_epi64,
8892 _mm512_mask_mullox_epi64): New intrinsics.
8894 2018-05-03 Tom de Vries <tom@codesourcery.com>
8897 * doc/sourcebuild.texi (Commands for use in dg-final, Scan optimization
8898 dump files): Add offload-tree.
8900 2018-05-03 Richard Biener <rguenther@suse.de>
8902 PR tree-optimization/85615
8903 * tree-ssa-threadupdate.c (thread_block_1): Only allow exits
8904 to loops not nested in BBs loop father to avoid creating multi-entry
8907 2018-05-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8909 PR tree-optimization/70291
8910 * tree-complex.c (expand_complex_libcall): Add type, inplace_p
8911 arguments. Change return type to tree. Emit libcall as a new
8912 statement rather than replacing existing one when inplace_p is true.
8913 (expand_complex_multiplication_components): New function.
8914 (expand_complex_multiplication): Expand floating-point complex
8915 multiplication using the above.
8916 (expand_complex_division): Rename inner_type parameter to type.
8917 Update expand_complex_libcall call-site.
8918 (expand_complex_operations_1): Update expand_complex_multiplication
8919 and expand_complex_division call-sites.
8921 2018-05-02 Jakub Jelinek <jakub@redhat.com>
8924 * config/i386/i386.md (*ashl<dwi>3_doubleword_mask,
8925 *ashl<dwi>3_doubleword_mask_1, *<shift_insn><dwi>3_doubleword_mask,
8926 *<shift_insn><dwi>3_doubleword_mask_1): In condition require that
8927 the highest significant bit of the shift count mask is clear. In
8928 check whether and[sq]i3 is needed verify that all significant bits
8929 of the shift count other than the highest are set.
8931 2018-05-02 Tom de Vries <tom@codesourcery.com>
8934 * builtins.def (DEF_GOACC_BUILTIN_ONLY): Define.
8935 * omp-builtins.def (BUILT_IN_GOACC_PARLEVEL_ID)
8936 (BUILT_IN_GOACC_PARLEVEL_SIZE): New builtin.
8937 * builtins.c (expand_builtin_goacc_parlevel_id_size): New function.
8938 (expand_builtin): Call expand_builtin_goacc_parlevel_id_size.
8939 * doc/extend.texi (Other Builtins): Add __builtin_goacc_parlevel_id and
8940 __builtin_goacc_parlevel_size.
8942 2018-05-02 Richard Biener <rguenther@suse.de>
8944 PR tree-optimization/85597
8945 * tree-vect-stmts.c (vectorizable_operation): For ternary SLP
8946 do not use split vect_get_vec_defs call but call vect_get_slp_defs
8949 2018-05-02 Tom de Vries <tom@codesourcery.com>
8952 * doc/sourcebuild.texi (Commands for use in dg-final, Scan optimization
8953 dump files): Add ltrans-tree.
8955 2018-05-02 Tom de Vries <tom@codesourcery.com>
8958 * doc/sourcebuild.texi (Commands for use in dg-final, Scan optimization
8959 dump files): Add wpa-ipa.
8961 2018-05-02 Segher Boessenkool <segher@kernel.crashing.org>
8963 * config.gcc (powerpc*-*-*): Remove paired.h. Unsupport the
8964 powerpc*-*-linux*paired* target.
8965 * config/rs6000/750cl.h: Delete.
8966 * config/rs6000/paired.h: Delete.
8967 * config/rs6000/paired.md: Delete.
8968 * config/rs6000/predicates.md (easy_vector_constant): Remove paired
8970 * config/rs6000/rs6000-builtin.def: Remove paired float support.
8971 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
8972 comment. Remove paired float support.
8973 * config/rs6000/rs6000-modes.def: Remove V2SF and V2SI.
8974 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete
8976 * config/rs6000/rs6000-protos.h (paired_expand_vector_init,
8977 paired_emit_vector_cond_expr, paired_expand_vector_move): Delete
8979 * config/rs6000/rs6000.c: Remove paired float support.
8980 (paired_expand_vector_init, paired_expand_vector_move,
8981 paired_emit_vector_compare, paired_emit_vector_cond_expr,
8982 (paired_expand_lv_builtin, paired_expand_stv_builtin,
8983 paired_expand_builtin, paired_expand_predicate_builtin,
8984 paired_init_builtins): Delete.
8985 * config/rs6000/rs6000.h: Remove paired float support.
8986 * config/rs6000/rs6000.md: Remove paired float support.
8987 (move_from_CR_ov_bit): Delete.
8988 * config/rs6000/rs6000.opt (mpaired): Delete.
8989 * config/rs6000/t-rs6000: Remove paired.md from MD_INCLUDES.
8990 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mpaired.
8992 2018-05-02 Richard Biener <rguenther@suse.de>
8995 * gimplify.c (gimplify_save_expr): When in SSA form allow
8996 SAVE_EXPRs to compute to SSA vars.
8998 2018-05-02 Jakub Jelinek <jakub@redhat.com>
9001 * config/i386/i386.md (*ashl<dwi>3_doubleword_mask,
9002 *ashl<dwi>3_doubleword_mask_1, *<shift_insn><dwi>3_doubleword_mask,
9003 *<shift_insn><dwi>3_doubleword_mask_1): If and[sq]i3 is needed, don't
9004 clobber operands[2], instead use a new pseudo. Formatting fixes.
9006 2018-05-02 Richard Sandiford <richard.sandiford@linaro.org>
9008 PR tree-optimization/85586
9009 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Only
9010 exit early for statements in the same group if the accesses are
9013 2018-05-02 Tom de Vries <tom@codesourcery.com>
9016 * lto-wrapper.c (compile_offload_image): Add "could not find mkoffload"
9019 2018-05-01 Marc Glisse <marc.glisse@inria.fr>
9021 PR tree-optimization/85143
9022 * match.pd (A<B&A<C): Extend to BIT_IOR_EXPR.
9024 2018-05-01 Tom de Vries <tom@codesourcery.com>
9027 * config/nvptx/mkoffload.c (main): Suggest using -B in "offload compiler
9028 not found" error message.
9030 2018-05-01 Tom de Vries <tom@codesourcery.com>
9033 * vec.h (VEC_ORDERED_REMOVE_IF, VEC_ORDERED_REMOVE_IF_FROM_TO): Define.
9034 * vec.c (test_ordered_remove_if): New function.
9035 (vec_c_tests): Call test_ordered_remove_if.
9036 * dwarf2cfi.c (connect_traces): Use VEC_ORDERED_REMOVE_IF_FROM_TO.
9037 * lto-streamer-out.c (prune_offload_funcs): Use VEC_ORDERED_REMOVE_IF.
9038 * tree-vect-patterns.c (vect_pattern_recog_1): Use
9039 VEC_ORDERED_REMOVE_IF.
9041 2018-05-01 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
9043 PR tree-optimization/82665
9044 * vr-values.c (vr_values::extract_range_from_binary_expr): Handle
9045 pointer subtraction where arguments come from a memchr call.
9047 2018-05-01 Jakub Jelinek <jakub@redhat.com>
9049 * configure.ac (LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION): Use
9050 --push-state --as-needed and --pop-state instead of --as-needed and
9051 --no-as-needed if ld supports it.
9052 * configure: Regenerated.
9055 * doc/install.texi2html: Replace _002d with - and _002a with * in
9056 generated html files using sed.
9058 2018-04-30 David Malcolm <dmalcolm@redhat.com>
9061 * gcc-rich-location.c (blank_line_before_p): New function.
9062 (use_new_line): New function.
9063 (gcc_rich_location::add_fixit_insert_formatted): New function.
9064 * gcc-rich-location.h
9065 (gcc_rich_location::add_fixit_insert_formatted): New function.
9067 2018-04-30 David Malcolm <dmalcolm@redhat.com>
9069 * selftest.c (assert_streq): Rename "expected" and "actual" to
9070 "val1" and "val2". Extend NULL-handling to cover both inputs
9071 symmetrically, while still requiring both to be non-NULL for a pass.
9072 * selftest.h (assert_streq): Rename "expected" and "actual" to
9074 (ASSERT_EQ): Likewise.
9075 (ASSERT_EQ_AT): Likewise.
9076 (ASSERT_KNOWN_EQ): Likewise.
9077 (ASSERT_KNOWN_EQ_AT): Likewise.
9078 (ASSERT_NE): Likewise.
9079 (ASSERT_MAYBE_NE): Likewise.
9080 (ASSERT_MAYBE_NE_AT): Likewise.
9081 (ASSERT_STREQ): Likewise. Clarify that both must be non-NULL for
9082 the assertion to pass.
9083 (ASSERT_STREQ_AT): Likewise.
9085 2018-04-30 Jonathan Wakely <jwakely@redhat.com>
9087 * doc/invoke.texi (-fpie, -fPIE): Fix grammar and clarify
9088 interaction with -pie.
9090 2018-04-30 David Malcolm <dmalcolm@redhat.com>
9092 * selftest.h: Fix alphabetization of per-source-file selftest
9095 2018-04-30 Jason Merrill <jason@redhat.com>
9097 PR c++/61982 - dead stores to destroyed objects.
9098 * gimplify.c (gimplify_modify_expr): Simplify complex lvalue on LHS
9101 2018-04-30 Jason Merrill <jason@redhat.com>
9103 * tree.c (build_clobber): New.
9104 * tree.h: Declare it.
9105 * gimplify.c (gimplify_bind_expr, gimplify_target_expr): Use it.
9107 2018-04-30 David Malcolm <dmalcolm@redhat.com>
9109 * diagnostic-show-locus.c (layout::layout): Update for
9110 location_get_source_line returning a char_span.
9111 (struct char_span): Move to input.h.
9112 (struct correction): Update for fields in char_span becoming
9114 (struct source_line): Update for location_get_source_line
9115 returning a char_span.
9116 (layout::print_line): Likewise.
9117 * edit-context.c (edited_file::print_content): Likewise.
9118 (edited_file::print_diff_hunk): Likewise.
9119 (edited_file::print_run_of_changed_lines): Likewise.
9120 (edited_file::get_num_lines): Likewise.
9121 (edited_line::edited_line): Likewise.
9122 * final.c (asm_show_source): Likewise.
9123 * input.c (location_get_source_line): Convert return type
9124 from const char * to char_span, losing the final "line_len"
9126 (dump_location_info): Update for the above.
9127 (get_substring_ranges_for_loc): Likewise. Use a char_span
9128 when handling the literal within the line.
9129 (test_reading_source_line): Update for location_get_source_line
9130 returning a char_span.
9131 * input.h (class char_span): Move here from
9132 diagnostic-show-locus.c, converting from a struct to a class.
9133 Make data members private.
9134 (char_span::operator bool): New.
9135 (char_span::length): New.
9136 (char_span::get_buffer): New.
9137 (char_span::operator[]): New.
9138 (char_span::subspan): Make const.
9139 (char_span::xstrdup): New.
9140 (location_get_source_line): Convert return type from const char *
9141 to char_span, losing the final "line_size" param.
9143 2018-04-30 Jan Hubicka <jh@suse.cz>
9145 * lto-wrapper.c (ltrans_priorities): New static var.
9146 (cmp_priority): New.
9147 (run_gcc): Read priorities and if doing parallel build order
9148 the Makefile by them.
9150 2018-04-30 David Malcolm <dmalcolm@redhat.com>
9152 * input.h (builtins_location_check): Convert to a STATIC_ASSERT.
9154 2018-04-30 Richard Biener <rguenther@suse.de>
9156 * tree-cfg.c (verify_address): Remove base argument, add
9157 flag whether to check TREE_ADDRESSABLE and do that.
9158 (verify_expr): Remove.
9159 (verify_types_in_gimple_reference): Add pieces from verify_expr.
9160 (verify_gimple_assign_single): Likewise.
9161 (verify_gimple_switch): Likewise.
9162 (verify_expr_location_1): Dereference tp once. Add (disabled)
9163 piece from verify_expr.
9164 (verify_gimple_in_cfg): Do not call verify_expr on all ops.
9166 2018-04-30 Claudiu Zissulescu <claziss@synopsys.com>
9168 * config/arc/linux.h (CLEAR_INSN_CACHE): Define.
9170 2018-04-30 Claudiu Zissulescu <claziss@synopsys.com>
9172 * config/arc/arc-protos.h (prepare_extend_operands): Remove.
9173 (small_data_pattern): Likewise.
9174 (arc_rewrite_small_data): Likewise.
9175 * config/arc/arc.c (LEGITIMATE_SMALL_DATA_OFFSET_P): Remove.
9176 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Likewise.
9177 (get_symbol_alignment): New function.
9178 (legitimate_small_data_address_p): Likewise.
9179 (legitimate_scaled_address): Update, call
9180 legitimate_small_data_address_p.
9181 (output_sdata): New static variable.
9182 (arc_print_operand): Update how we handle small data operands.
9183 (arc_print_operand_address): Likewise.
9184 (arc_legitimate_address_p): Update, use
9185 legitimate_small_data_address_p.
9186 (arc_rewrite_small_data_p): Remove.
9187 (arc_rewrite_small_data_1): Likewise.
9188 (arc_rewrite_small_data): Likewise.
9189 (small_data_pattern): Likewise.
9190 (compact_sda_memory_operand): Update to use
9191 legitimate_small_data_address_p and get_symbol_alignment.
9192 (prepare_move_operands): Don't rewite sdata pattern.
9193 (prepare_extend_operands): Remove.
9194 * config/arc/arc.md (zero_extendqihi2): Don't rewrite sdata
9196 (zero_extendqisi2): Likewise.
9197 (zero_extendhisi2): Likewise.
9198 (extendqihi2): Likewise.
9199 (extendqisi2): Likewise.
9200 (extendhisi2): Likewise.
9204 * config/arc/constraints.md (Usd): Change it to memory constraint.
9206 2018-04-30 Claudiu Zissulescu <claziss@synopsys.com>
9208 * config/arc/arc.c (arc_split_move): Allow signed 6-bit constants
9209 as source of std instructions.
9210 * config/arc/arc.md (movsi_insn): Update pattern predicate to
9211 allow 6-bit constants as source for store instructions.
9212 (movdi_insn): Update instruction pattern to allow 6-bit constants
9213 as source for store instructions.
9215 2018-04-30 Jonathan Wakely <jwakely@redhat.com>
9217 * doc/invoke.texi (-fdebug-types-section): Fix grammar.
9219 2018-04-30 Nathan Sidwell <nathan@acm.org>
9220 Sandra Loosemore <sandra@codesourcery.com>
9222 * dumpfile.c (dump_open): Allow '-' for stdout.
9223 * doc/invoke.texi (Developer Options): Document dump filename
9224 determination early. Document stdin/stdout selection.
9226 2018-04-30 Andrew Sadek <andrew.sadek.se@gmail.com>
9228 Microblaze Target: PIC data text relative
9230 * config/microblaze/microblaze.opt: add new option -mpic-data-text-rel.
9231 * config/microblaze/microblaze-protos.h (microblaze_constant_address_p):
9233 * config/microblaze/microblaze.h (microblaze_constant_address_p):
9234 CONSTANT_ADDRESS_P definition to microblaze_constant_address_p.
9235 * config/microblaze/microblaze.c (TARGET_PIC_DATA_TEXT_REL):
9236 New addressing mode for data-text relative position indepenedent code.
9237 (microblaze_classify_unspec): add 'UNSPEC_TEXT' case ->
9238 'ADDRESS_SYMBOLIC_TXT_REL'.
9239 (microblaze_classify_address): Add handling for UNSPEC + CONST_INT.
9240 (microblaze_legitimate_pic_operand): Exclude function calls from
9241 pic operands in case of TARGET_PIC_DATA_TEXT_REL option.
9242 (microblaze_legitimize_address): Generate 'UNSPEC_TEXT' for all possible
9244 (microblaze_address_insns): Add 'ADDRESS_SYMBOLIC_TXT_REL' case.
9245 (print_operand): Add 'ADDRESS_SYMBOLIC_TXT_REL' case.
9246 (print_operand_address): Add 'ADDRESS_SYMBOLIC_TXT_REL' case + handling
9247 for 'address + offset'.
9248 (microblaze_expand_prologue): Add new function prologue call for
9250 (microblaze_asm_generate_pic_addr_dif_vec): Override new target hook
9251 'TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC' to disable address diff vector
9252 table in case of TARGET_PIC_DATA_TEXT_REL.
9253 (expand_pic_symbol_ref): Add handling for 'UNSPEC_TEXT'.
9254 * config/microblaze/microblaze.md (TARGET_PIC_DATA_TEXT_REL):
9255 Add new macros 'UNSPEC_TEXT',
9256 'UNSPEC_SET_TEXT' + add rule for setting r20 in function prologue
9257 + exclude function calls from 'UNSPEC_PLT' in case of data text
9259 * doc/tm.texi.in (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC): Add
9260 new target hook for generating address diff vector tables in case of
9262 * doc/tm.texi : Regenerate.
9263 * stmt.c (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC): Append new condition
9264 'targetm.asm_out.generate_pic_addr_diff_vec' to flag_pic in case
9265 of addr diff vector generation.
9266 * target.def (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC): Add
9267 target hook definition.
9268 * targhooks.h, targhooks.c (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC):
9269 Add default function for generate_pic_addr_diff_vec -> flag_pic.
9270 * doc/invoke.texi (Add new pic option): Add new microblaze pic
9271 option for data text relative.
9273 2018-04-30 Richard Biener <rguenther@suse.de>
9275 * tree-chrec.h (evolution_function_is_constant_p): Remove
9277 * tree-cfg.c (tree_node_can_be_shared): Re-order checks.
9279 2018-04-30 Richard Biener <rguenther@suse.de>
9282 * dwarf2out.c (gen_producer_string): Ignore -fchecking[=].
9284 2018-04-30 Richard Biener <rguenther@suse.de>
9286 PR tree-optimization/28364
9287 PR tree-optimization/85275
9288 * tree-ssa-loop-ch.c (ch_base::copy_headers): Stop after
9289 copying first exit test.
9291 2018-04-28 Mark Wielaard <mark@klomp.org>
9293 * dwarf2out.c (dwarf2out_finish): Add .debug_addr table header for
9295 (dwarf_AT): Handle DW_AT_addr_base.
9296 (add_top_level_skeleton_die_attrs): Use dwarf_AT for DW_AT_addr_base.
9298 2018-04-28 Uros Bizjak <ubizjak@gmail.com>
9301 * config/i386/i386.md (*ashl<dwi>3_doubleword_mask): New pattern.
9302 (*ashl<dwi>3_doubleword_mask_1): Ditto.
9303 (*<shift_insn><dwi>3_doubleword_mask): Ditto.
9304 (*<shift_insn><dwi>3_doubleword_mask_1): Ditto.
9306 2018-04-28 Richard Biener <rguenther@suse.de>
9308 * tree-cfg.c (verify_gimple_phi): Take a gphi * argument.
9309 (verify_gimple_in_cfg): Rename visited_stmts to visited_throwing_stmts
9310 to reflect use. Only add interesting stmts.
9312 2018-04-27 Martin Jambor <mjambor@suse.cz>
9315 * ipa-cp.c (find_aggregate_values_for_callers_subset): Make sure
9316 the jump function allows for passing through aggregate values.
9318 2018-04-27 David Malcolm <dmalcolm@redhat.com>
9320 * input.h (in_system_header_at): Convert from macro to inline
9322 (from_macro_expansion_at): Likewise.
9323 (from_macro_definition_at): Likewise.
9325 2018-04-27 Jeff Law <law@redhat.com>
9327 * config.gcc: Mark tile* targets as deprecated/obsolete.
9329 2018-04-27 Richard Biener <rguenther@suse.de>
9331 * config/aarch64/aarch64.c: Simplify ap.__stack advance and
9334 2018-04-27 Richard Biener <rguenther@suse.de>
9336 * tree-cfg.c (verify_expr): Make dead code hit gcc_unreachable.
9338 2018-04-27 Uros Bizjak <ubizjak@gmail.com>
9340 * config/i386/i386.md (*movti_internal): Substitute Ye constraint
9341 with Yd constraint. Set "preferred_for_speed" attribute from
9342 TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC for alternatives
9344 (*movdi_internal): Ditto.
9345 (movti_interunit splitters): Remove
9346 TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC from insn condition.
9347 (movdi_interunit splitters): Ditto.
9348 * config/i386/constraints.md (Ye): Remove.
9349 (Yd): Do not depend on TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC.
9351 2018-04-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
9354 * config/aarch64/constraints.md (Usg): Limit to 31.
9357 2018-04-27 Jakub Jelinek <jakub@redhat.com>
9359 PR tree-optimization/85529
9360 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Add FIRST_BB
9361 argument. Don't call get_nonzero_bits if opcode is ERROR_MARK_NODE,
9362 rhs2 def stmt's bb is dominated by first_bb and it isn't an obvious
9363 zero extension or masking of the MSB bit.
9364 (optimize_range_tests): Add FIRST_BB argument, pass it through
9365 to optimize_range_tests_var_bound.
9366 (maybe_optimize_range_tests, reassociate_bb): Adjust
9367 optimize_range_tests callers.
9369 2018-04-26 Richard Biener <rguenther@suse.de>
9370 Jakub Jelinek <jakub@redhat.com>
9372 * cgraph.h (symbol_table): Just declare debug method here.
9373 * symtab.c (symbol_table::debug): Define.
9375 2018-04-26 Eric Botcazou <ebotcazou@adacore.com>
9377 * loop-invariant.c (may_assign_reg_p): Return false for frame pointer.
9379 2018-04-26 Uros Bizjak <ubizjak@gmail.com>
9381 * config/i386/i386.md ("isa" attribute): Add x64_sse2.
9382 ("enabled" attribute): Handle x64_sse2 "isa" attribute.
9383 (*movdi_internal): Substitute Yi and Yj constraint with x
9384 and Ym and Yn constraint with y constraint. Update "isa"
9385 attribute and set "preferred_for_speed" attribute from
9386 TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC for updated alternatives.
9387 (*movsi_internal): Ditto.
9388 (*movdf_internal): Ditto.
9389 (*movsf_internal): Ditto.
9390 (*zero_extendsidi2): Ditto.
9391 * config/i386/sse.md (vec_set<mode>_0): Ditto.
9392 (sse2_loadld): Ditto.
9393 (*vec_extract<ssevecmodelower>_0): Ditto.
9394 (*vec_extractv4si_0_zext_sse4): Ditto.
9395 (vec_concatv2di): Ditto.
9396 (*vec_dup<mode>): Ditto.
9397 * config/i386/mmx.md (*mov<mode>_internal): Ditto.
9398 * config/i386/constraints.md (Yi): Remove.
9403 2018-04-26 Nathan Sidwell <nathan@acm.org>
9405 * dumpfile.c (dump_open): New.
9406 (dump_open_alternate_stream, dump_start, dump_begin): Call it.
9407 (dump_finish): Detect stdio/stderr by value not name.
9409 2018-04-26 Jonathan Wakely <jwakely@redhat.com>
9411 * doc/invoke.texi (-Wreturn-type): Document default status for C++.
9413 2018-04-26 Tom de Vries <tom@codesourcery.com>
9416 * config/nvptx/nvptx.c (verify_neutering_jumps)
9417 (verify_neutering_labels): New function
9418 (nvptx_single): Use verify_neutering_jumps and verify_neutering_labels.
9420 2018-04-26 Tom de Vries <tom@codesourcery.com>
9423 * config/nvptx/nvptx.c (needs_neutering_p): New function.
9424 (nvptx_single): Use needs_neutering_p to skip over insns that do not
9427 2018-04-26 Richard Biener <rguenther@suse.de>
9428 Tom de Vries <tom@codesourcery.com>
9431 * lto-streamer-out.c (output_function): Fixup loops if required to match
9432 discovery done in the reader.
9434 2018-04-26 Richard Biener <rguenther@suse.de>
9436 PR tree-optimization/85116
9437 * tree-ssa-loop-ch.c (do_while_loop_p): A do-while loop should
9438 have a loop exit from the single latch predecessor. Remove
9439 case of header with just condition.
9440 (ch_base::copy_headers): Exclude infinite loops from any
9442 (pass_ch::execute): Record exits.
9444 2018-04-26 Richard Biener <rguenther@suse.de>
9446 * tree-vect-data-refs.c (vect_get_data_access_cost): Get
9447 prologue cost vector and pass it to vect_get_load_cost.
9448 (vect_get_peeling_costs_all_drs): Likewise.
9449 (vect_peeling_hash_get_lowest_cost): Likewise.
9450 (vect_enhance_data_refs_alignment): Likewise.
9452 2018-04-26 Richard Biener <rguenther@suse.de>
9455 * tree-cfg.c (verify_gimple_assign_unary): Restore proper
9456 checking of integer<->pointer conversions.
9457 * omp-expand.c (expand_omp_for_static_nochunk): Avoid
9458 sign-/zero-extending pointer types.
9459 (expand_omp_for_static_chunk): Likewise.
9461 2018-03-22 Hans-Peter Nilsson <hp@axis.com>
9462 Jean Lee <xiaoyur347@gmail.com>
9464 * config/mips/mips.c (mips_asan_shadow_offset): New function.
9465 (TARGET_ASAN_SHADOW_OFFSET): Define.
9466 * config/mips/mips.h (FRAME_GROWS_DOWNWARD): Augment to also be
9467 true for -fsanitize=address.
9469 2018-04-25 Mark Wielaard <mark@klomp.org>
9471 * dwarf2out.c (file_info_cmp): Sort longer dir prefixes before
9474 2018-04-25 Jakub Jelinek <jakub@redhat.com>
9476 * config/i386/i386.md (*x86_mov<mode>cc_0_m1): Use type "alu1" rather
9477 than "alu", remove explicit "memory" and "imm_disp" attributes.
9478 (*x86_mov<mode>cc_0_m1_se, *x86_mov<mode>cc_0_m1_neg): Likewise.
9481 * simplify-rtx.c (simplify_unary_operation_1) <case SIGN_EXTEND,
9482 case ZERO_EXTEND>: Pass SUBREG_REG (op) rather than op to
9483 gen_lowpart_no_emit.
9485 2018-04-25 Sebastian Peryt <sebastian.peryt@intel.com>
9488 * config/i386/i386.c (ix86_expand_builtin): Change memory
9489 operand to XI, extend p0 to Pmode.
9490 * config/i386/i386.md: Change unspec volatile and operand
9491 1 mode to XI, change operand 0 mode to P.
9493 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
9495 * config/nds32/nds32-predicates.c (nds32_can_use_bclr_p): Mask with
9496 GET_MODE_MASK before any checking.
9497 (nds32_can_use_bset_p): Likewise.
9498 (nds32_can_use_btgl_p): Likewise.
9500 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
9502 * config/nds32/nds32-doubleword.md: New define_split pattern for
9503 illegal register number.
9505 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
9507 * config/nds32/nds32.c (nds32_print_operand): Set op_value ealier.
9509 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
9511 * config/nds32/nds32.h (ASM_APP_ON): Add missing newline character.
9513 2018-04-25 Richard Biener <rguenther@suse.de>
9515 * lto-streamer.h (LTO_major_version): Bump to 8.
9517 2018-04-25 Jakub Jelinek <jakub@redhat.com>
9519 * BASE-VER: Set to 9.0.0.
9521 2018-04-24 Segher Boessenkool <segher@kernel.crashing.org>
9523 * config/rs6000/rs6000.c (init_float128_ieee): Fix spelling mistakes
9524 in __abskf2 and __powikf2.
9526 2018-04-24 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
9529 * config/aarch64/constraints.md (Usg, Usj): New constraints.
9530 * config/aarch64/iterators.md (cmode_simd): New mode attribute.
9531 * config/aarch64/aarch64.md (*aarch64_ashr_sisd_or_int_<mode>3):
9532 Use the above on operand 2. Reindent.
9533 (*aarch64_lshr_sisd_or_int_<mode>3): Likewise.
9535 2018-04-24 H.J. Lu <hongjiu.lu@intel.com>
9538 * common/config/i386/i386-common.c (ix86_handle_option): Don't
9540 * config/i386/i386.opt (mcet): Removed.
9541 * doc/install.texi: Remove -mcet documentation.
9542 * doc/invoke.texi: Likewise.
9544 2018-04-24 H.J. Lu <hongjiu.lu@intel.com>
9547 * doc/install.texi: Remove -mcet from bootstrap-cet.
9549 2018-04-24 Jakub Jelinek <jakub@redhat.com>
9552 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Don't define
9553 __builtin_ia32_readeflags_u32 and __builtin_ia32_writeeflags_u32
9557 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Punt if
9558 const_vector is not CONST_VECTOR or SYMBOL_REF for a constant pool
9559 containing a CONST_VECTOR.
9561 2018-04-24 Cesar Philippidis <cesar@codesourcery.com>
9563 * doc/install.texi: Update newlib dependency for nvptx.
9565 2018-04-24 Jakub Jelinek <jakub@redhat.com>
9568 * config/i386/i386.c (ix86_expand_vector_init_one_var): Use UINTVAL
9569 instead of INTVAL when shifting x left.
9571 2018-04-24 Andreas Krebbel <krebbel@linux.ibm.com>
9573 PR tree-optimization/85478
9574 * tree-vect-loop.c (vect_analyze_loop_2): Do not call
9575 vect_grouped_store_supported for single element vectors.
9577 2018-04-24 Richard Biener <rguenther@suse.de>
9580 * config/i386/i386.c (ix86_add_stmt_cost): Restrict strided
9581 load cost increase to the case of non-constant step.
9583 2018-04-24 Jakub Jelinek <jakub@redhat.com>
9586 * reg-stack.c (move_for_stack_reg): Don't fail assertion about dead
9587 destination if any_malformed_asm.
9589 2018-04-23 Eric Botcazou <ebotcazou@adacore.com>
9592 * expr.c (store_field): In the bitfield case, if the value comes from
9593 a function call and is returned in registers by means of a PARALLEL,
9594 do not change the mode of the temporary unless BLKmode and VOIDmode.
9596 2018-04-23 Andrey Belevantsev <abel@ispras.ru>
9598 PR rtl-optimization/85423
9599 * sel-sched-ir.c (has_dependence_note_mem_dep): Only discard
9600 dependencies to debug insns when the previous insn is non-debug.
9602 2018-04-23 Claudiu Zissulescu <claziss@synopsys.com>
9604 * config/arc/arc.md ("vunspec"): Delete it, unify all the unspec
9605 enums into a single definition.
9606 (fls): Fix predicates and printing.
9609 2018-04-23 Claudiu Zissulescu <claziss@synopsys.com>
9611 * config/arc/arc-protos.h (check_if_valid_sleep_operand): Remove.
9612 * config/arc/arc.c (arc_expand_builtin): Sleep accepts registers
9613 and short u6 immediate.
9614 (check_if_valid_sleep_operand): Remove.
9615 * config/arc/arc.md (Sleep): Accepts registers and u6 immediates.
9617 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
9619 * config/nds32/nds32.c (nds32_compute_stack_frame): Consider
9620 flag_always_save_lp condition.
9621 * config/nds32/nds32.opt (malways-save-lp): New option.
9623 2018-04-22 Shiva Chen <shiva0217@gmail.com>
9625 * config/nds32/nds32-protos.h (nds32_use_load_post_increment): Declare.
9626 * config/nds32/nds32.c (nds32_use_load_post_increment): New.
9627 * config/nds32/nds32.h
9628 (USE_LOAD_POST_INCREMENT, USE_LOAD_POST_DECREMENT): Define.
9629 (USE_STORE_POST_INCREMENT, USE_STORE_POST_DECREMENT): Define.
9631 2018-04-22 Shiva Chen <shiva0217@gmail.com>
9633 * config/nds32/nds32-protos.h (nds32_ls_333_p): Remove.
9634 * config/nds32/nds32.c (nds32_ls_333_p): Remove.
9636 2018-04-22 Shiva Chen <shiva0217@gmail.com>
9637 Chung-Ju Wu <jasonwucj@gmail.com>
9639 * config/nds32/nds32-protos.h (nds32_case_vector_shorten_mode):
9641 * config/nds32/nds32.c (nds32_case_vector_shorten_mode): New function.
9642 * config/nds32/nds32.h (CASE_VECTOR_SHORTEN_MODE): Modify.
9644 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
9646 * config/nds32/nds32.c (nds32_compute_stack_frame): Fix wrong value.
9648 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
9650 * config/nds32/nds32-protos.h (nds32_data_alignment,
9651 nds32_local_alignment): Declare.
9652 * config/nds32/nds32.c (nds32_data_alignment, nds32_constant_alignment,
9653 nds32_local_alignment): New functions.
9654 (TARGET_CONSTANT_ALIGNMENT): Define.
9655 * config/nds32/nds32.h (DATA_ALIGNMENT, LOCAL_ALIGNMENT): Define.
9657 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
9659 * config/nds32/nds32.c
9660 (TARGET_HARD_REGNO_MODE_OK): Move to the bottom of file.
9661 (TARGET_MODES_TIEABLE_P): Likewise.
9663 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
9665 * config/nds32/nds32.c (nds32_asm_file_start): Display optimization
9668 2018-04-22 Monk Chiang <sh.chiang04@gmail.com>
9669 Chung-Ju Wu <jasonwucj@gmail.com>
9671 * config/nds32/constants.md (unspec_volatile_element): Add enum values
9672 for unaligned access.
9673 * config/nds32/nds32-intrinsic.c: Implementation of expanding
9675 * config/nds32/nds32-intrinsic.md: Likewise.
9676 * config/nds32/nds32_intrinsic.h: Likewise.
9677 * config/nds32/nds32.h (nds32_builtins): Likewise.
9678 * config/nds32/nds32.opt (munaligned-access): New option.
9679 * config/nds32/nds32.c (nds32_asm_file_start): Display
9680 flag_unaligned_access status.
9682 2018-04-20 Kito Cheng <kito.cheng@gmail.com>
9684 * config/riscv/elf.h (LINK_SPEC): Pass --no-relax if
9685 -mno-relax is present.
9686 * config/riscv/linux.h (LINK_SPEC): Ditto.
9688 2018-04-20 Martin Sebor <msebor@redhat.com>
9691 * gimple-fold.c (gimple_fold_builtin_strcpy): Suppress -Wrestrict
9693 (gimple_fold_builtin_stxcpy_chk): Same.
9694 * gimple-ssa-warn-restrict.c (check_bounds_or_overlap): Same.
9696 2018-04-20 Michael Meissner <meissner@linux.ibm.com>
9699 * config/rs6000/rs6000.c (init_float128_ieee): Add support to call
9700 __powikf2 when long double is IEEE 128-bit.
9702 2018-04-20 Kito Cheng <kito.cheng@gmail.com>
9704 * config/riscv/riscv.c (riscv_first_stack_step): Round up min
9705 step to make sure stack always aligned.
9707 2018-04-20 Carl Love <cel@us.ibm.com>
9710 * config/rs6000/rs6000-c.c (rs6000_gimple_fold_builtin): Add
9711 size check for arg0.
9713 2018-04-20 Nathan Sidwell <nathan@codesourcery.com>
9714 Tom de Vries <tom@codesourcery.com>
9717 * config/nvptx/nvptx.c (nvptx_emit_forking, nvptx_emit_joining):
9718 Emit insns for calls too.
9719 (nvptx_find_par): Always look for worker-level predecessor insn.
9720 (nvptx_propagate): Add is_call parm, return bool. Copy frame for
9722 (nvptx_vpropagate, nvptx_wpropagate): Adjust.
9723 (nvptx_process_pars): Propagate frames for calls.
9725 2018-04-20 H.J. Lu <hongjiu.lu@intel.com>
9728 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET):
9730 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
9731 (ix86_handle_option): Don't handle OPT_mibt.
9732 * config/i386/cet.h: Check __CET__ instead of __IBT__ and
9734 * config/i386/driver-i386.c (host_detect_local_cpu): Remove
9736 * config/i386/i386-c.c (ix86_target_macros_internal): Don't
9737 check OPTION_MASK_ISA_IBT nor flag_cf_protection.
9738 (ix86_target_macros): Define __CET__ with flag_cf_protection
9739 for -fcf-protection.
9740 * config/i386/i386.c (isa2_opts): Remove -mibt.
9741 * config/i386/i386.h (TARGET_IBT): Removed.
9742 (TARGET_IBT_P): Likewise.
9743 (ix86_valid_target_attribute_inner_p): Don't check OPT_mibt.
9744 * config/i386/i386.md (nop_endbr): Don't check TARGET_IBT.
9745 * config/i386/i386.opt (mcet): Update help message.
9748 * doc/invoke.texi: Remove -mibt. Document __CET__. Document
9749 -mcet as an alias for -mshstk.
9751 2018-04-20 Richard Biener <rguenther@suse.de>
9754 * match.pd ((X * CST) * Y -> (X * Y) * CST): Avoid exponential
9755 complexity by forcing a single use of the multiply operand.
9757 2018-04-20 Martin Jambor <mjambor@suse.cz>
9760 * ipa-cp.c (cgraph_edge_brings_value_p): Move check for self-feeding
9761 recursion dependency to only apply to non-clones.
9763 2018-04-20 Martin Jambor <mjambor@suse.cz>
9766 * ipa-cp.c (create_specialized_node): Check that clones of
9767 self-recursive edges exist during IPA-CP.
9769 2018-04-19 Toon Moene <toon@moene.org>
9771 * doc/invoke.texi: Add -floop-unroll-and-jam to options enabled
9774 2018-04-19 Jakub Jelinek <jakub@redhat.com>
9776 PR tree-optimization/85467
9777 * fold-const.c (fold_ternary_loc) <case BIT_FIELD_REF>: Use
9778 VECTOR_TYPE_P macro. If type is vector type, VIEW_CONVERT_EXPR the
9779 VECTOR_CST element to type.
9781 2018-04-19 H.J. Lu <hongjiu.lu@intel.com>
9784 * config/i386/i386.h (STACK_SAVEAREA_MODE): New.
9785 * config/i386/i386.md (builtin_setjmp_setup): Removed.
9786 (builtin_longjmp): Likewise.
9787 (save_stack_nonlocal): New pattern.
9788 (restore_stack_nonlocal): Likewise.
9790 2018-04-19 H.J. Lu <hongjiu.lu@intel.com>
9793 * config/i386/cet.c (file_end_indicate_exec_stack_and_cet):
9794 Replace ASM_OUTPUT_LABEL with fprintf.
9796 2018-04-19 H.J. Lu <hongjiu.lu@intel.com>
9799 * config/i386/cet.c (file_end_indicate_exec_stack_and_cet):
9800 Check flag_cf_protection instead of TARGET_IBT and TARGET_SHSTK.
9801 * config/i386/i386-c.c (ix86_target_macros_internal): Also
9802 define __IBT__ and __SHSTK__ for -fcf-protection.
9803 * config/i386/i386.c (pass_insert_endbranch::gate): Don't check
9805 (ix86_trampoline_init): Likewise.
9806 (x86_output_mi_thunk): Likewise.
9807 (ix86_notrack_prefixed_insn_p): Likewise.
9808 (ix86_option_override_internal): Don't disallow -fcf-protection.
9809 * config/i386/i386.md (rdssp<mode>): Also enable for
9811 (incssp<mode>): Likewise.
9812 (nop_endbr): Likewise.
9813 * config/i386/i386.opt (mcet): Change help message to built-in
9817 * doc/invoke.texi: Remove -mcet, -mibt and -mshstk condition
9818 on -fcf-protection. Change -mcet, -mibt and -mshstk to only
9819 enable CET built-in functions.
9821 2018-04-19 Sebastian Peryt <sebastian.peryt@intel.com>
9823 * common/config/i386/i386-common.c
9824 (OPTION_MASK_ISA_MOVDIRI_SET, OPTION_MASK_ISA_MOVDIR64B_SET,
9825 OPTION_MASK_ISA_MOVDIRI_UNSET,
9826 OPTION_MASK_ISA_MOVDIR64B_UNSET): New defines.
9827 (ix86_handle_option): Handle -mmovdiri and -mmovdir64b.
9828 * config.gcc (movdirintrin.h): New header.
9829 * config/i386/cpuid.h (bit_MOVDIRI,
9830 bit_MOVDIR64B): New bits.
9831 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mmovdiri
9833 * config/i386/i386-builtin-types.def ((VOID, PUNSIGNED, UNSIGNED),
9834 (VOID, PVOID, PCVOID)): New function types.
9835 * config/i386/i386-builtin.def (__builtin_ia32_directstoreu_u32,
9836 __builtin_ia32_directstoreu_u64,
9837 __builtin_ia32_movdir64b): New builtins.
9838 * config/i386/i386-c.c (__MOVDIRI__, __MOVDIR64B__): New.
9839 * config/i386/i386.c (ix86_target_string): Added -mmovdir64b
9841 (ix86_valid_target_attribute_inner_p): Ditto.
9842 (ix86_expand_special_args_builtin): Added VOID_FTYPE_PUNSIGNED_UNSIGNED
9843 and VOID_FTYPE_PUNSIGNED_UNSIGNED.
9844 (ix86_expand_builtin): Expand IX86_BUILTIN_MOVDIR64B.
9845 * config/i386/i386.h (TARGET_MOVDIRI, TARGET_MOVDIRI_P,
9846 TARGET_MOVDIR64B, TARGET_MOVDIR64B_P): New.
9847 * config/i386/i386.md (UNSPECV_MOVDIRI, UNSPECV_MOVDIR64B): New.
9848 (movdiri<mode>, movdir64b_<mode>): New.
9849 * config/i386/i386.opt: Add -mmovdiri and -mmovdir64b.
9850 * config/i386/immintrin.h: Include movdirintrin.h.
9851 * config/i386/movdirintrin.h: New file.
9852 * doc/invoke.texi: Added -mmovdiri and -mmovdir64b.
9854 2018-04-19 Richard Biener <rguenther@suse.de>
9857 * cfg.c (clear_bb_flags): When loop state says we have
9858 marked irreducible regions also preserve BB_IRREDUCIBLE_LOOP.
9860 2018-04-19 Richard Biener <rguenther@suse.de>
9862 PR tree-optimization/84737
9863 * tree-vect-data-refs.c (vect_copy_ref_info): New function
9864 copying restrict info.
9865 (vect_setup_realignment): Use it.
9866 * tree-vectorizer.h (vect_copy_ref_info): Declare.
9867 * tree-vect-stmts.c (vectorizable_store): Copy ref info from
9868 the first DR to all generated stores.
9869 (vectorizable_load): Likewise for loads.
9871 2018-04-19 Jakub Jelinek <jakub@redhat.com>
9873 PR tree-optimization/85446
9874 * match.pd ((intptr_t) x eq/ne CST to x eq/ne (typeof x) cst): Require
9875 the integral and pointer types to have the same precision.
9877 * doc/install.texi: Document --disable-cet being the default and
9880 2018-04-18 Martin Liska <mliska@suse.cz>
9882 * ipa-devirt.c (odr_subtypes_equivalent_p): Fix GNU coding
9885 2018-04-18 Martin Liska <mliska@suse.cz>
9888 2018-03-02 Eric Botcazou <ebotcazou@adacore.com>
9891 * ipa-devirt.c (odr_subtypes_equivalent_p): Get the ODR type of both
9892 arguments if they are comparable.
9894 2018-04-18 Martin Liska <mliska@suse.cz>
9897 2018-03-13 Eric Botcazou <ebotcazou@adacore.com>
9900 * ipa-devirt.c (odr_subtypes_equivalent_p): Do not get the ODR type of
9903 2018-04-18 H.J. Lu <hongjiu.lu@intel.com>
9906 * config/i386/i386.c (ix86_expand_split_stack_prologue): Insert
9907 ENDBR after calling __morestack.
9909 2018-04-18 David Malcolm <dmalcolm@redhat.com>
9912 * configure.ac (gcc-driver-name.h): Honor --with-gcc-major-version
9913 by using gcc_base_ver to generate a gcc_driver_version, and use
9914 it when generating GCC_DRIVER_NAME.
9915 * configure: Regenerate.
9917 2018-04-18 Jakub Jelinek <jakub@redhat.com>
9920 * config.gcc: Obsolete powerpc*-*-*spe*.
9922 2018-04-17 Jakub Jelinek <jakub@redhat.com>
9925 * dbxout.c (dbxout_int): Perform negation in unsigned int type.
9926 (stabstr_D): Change type of unum from unsigned int to
9927 unsigned HOST_WIDE_INT. Perform negation in unsigned HOST_WIDE_INT
9930 2018-04-17 Jim Wilson <jimw@sifive.com>
9933 * config/riscv/riscv.c (riscv_compute_frame_info): Add calls to
9934 RISCV_STACK_ALIGN when using outgoing_args_size and pretend_args_size.
9935 Set arg_pointer_offset after using pretend_args_size.
9937 2018-04-17 Jakub Jelinek <jakub@redhat.com>
9939 PR rtl-optimization/85431
9940 * dse.c (record_store): Ignore zero width stores.
9943 * asan.c (handle_builtin_stack_restore): Adjust comment. Emit
9944 __asan_allocas_unpoison call and last_alloca_addr = new_sp before
9945 __builtin_stack_restore rather than after it.
9946 * builtins.c (expand_asan_emit_allocas_unpoison): Pass
9947 arg1 + (virtual_dynamic_stack_rtx - stack_pointer_rtx) as second
9948 argument instead of virtual_dynamic_stack_rtx.
9950 2018-04-17 Kelvin Nilsen <kelvin@gcc.gnu.org>
9952 * config/rs6000/rs6000-protos.h (rs6000_builtin_is_supported_p):
9954 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
9955 Add note to error message to explain internal mapping of overloaded
9956 built-in function name to non-overloaded built-in function name.
9957 * config/rs6000/rs6000.c (rs6000_builtin_is_supported_p): New
9960 2018-04-17 Michael Meissner <meissner@linux.vnet.ibm.com>
9963 * config/rs6000/rs6000.md (pack<mode>): Do not try handle a pack
9964 where the inputs overlap with the output.
9966 2018-04-17 Jakub Jelinek <jakub@redhat.com>
9968 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): Add
9969 (=v, v) alternative and explicit "memory" attribute.
9970 (vec_extract_lo_<mode><mask_name>): Likewise. Also add
9971 "type", "prefix", "prefix_extra", "length_immediate" and "mode"
9973 (vec_extract_lo_<mode><mask_name>): Add (=v, v) alternative and use
9974 "sselog1" type instead of "sselog".
9975 (vec_extract_hi_<mode><mask_name>): Use "sselog1" type instead of
9976 "sselog". Remove explicit "memory" attribute.
9977 (vec_extract_lo_v32hi): Add (=v, v) alternative and explicit "memory",
9978 "type", "prefix", "prefix_extra", "length_immediate" and "mode"
9980 (vec_extract_hi_v32hi): Merge all alternatives into one, use
9981 "sselog1" type instead of "sselog". Remove explicit "memory"
9983 (vec_extract_hi_v16hi): Merge each pair of alternatives into one,
9984 use "sselog1" type instead of "sselog". Remove explicit "memory"
9986 (vec_extract_lo_v64qi): Add (=v, v) alternative and explicit "memory",
9987 "type", "prefix", "prefix_extra", "length_immediate" and "mode"
9989 (vec_extract_hi_v64qi): Merge all alternatives into one, use
9990 "sselog1" type instead of "sselog". Remove explicit "memory"
9992 (vec_extract_hi_v32qi): Merge each pair of alternatives into one,
9993 use "sselog1" type instead of "sselog". Remove explicit "memory"
9997 * config/i386/i386.md (*ashlqi3_1_slp): Use alu1 type instead of alu.
10000 * rtlhooks.c (gen_lowpart_if_possible): Don't call gen_lowpart_SUBREG
10003 2018-04-17 Martin Jambor <mjambor@suse.cz>
10006 * ipa-cp.c (create_specialized_node): Call
10007 expand_all_artificial_thunks if necessary.
10009 2018-04-17 Martin Liska <mliska@suse.cz>
10012 * ipa-devirt.c (odr_types_equivalent_p): Remove trailing
10013 in message, remote space in between '_G' and '('.
10015 2018-04-17 Jakub Jelinek <jakub@redhat.com>
10018 * config/i386/sse.md (reduces<mode><mask_scalar_name>,
10019 avx512f_vmcmp<mode>3<round_saeonly_name>,
10020 avx512f_vmcmp<mode>3_mask<round_saeonly_name>,
10021 avx512f_sgetexp<mode><mask_scalar_name><round_saeonly_scalar_name>,
10022 avx512f_rndscale<mode><round_saeonly_name>,
10023 avx512dq_ranges<mode><mask_scalar_name><round_saeonly_scalar_name>,
10024 avx512f_vgetmant<mode><mask_scalar_name><round_saeonly_scalar_name>):
10025 Use %<iptr>2 instead of %2 for -masm=intel.
10026 (avx512f_vcvtss2usi<round_name>, avx512f_vcvtss2usiq<round_name>,
10027 avx512f_vcvttss2usi<round_saeonly_name>,
10028 avx512f_vcvttss2usiq<round_saeonly_name>): Use %k1 instead of %1 for
10030 (avx512f_vcvtsd2usi<round_name>, avx512f_vcvtsd2usiq<round_name>,
10031 avx512f_vcvttsd2usi<round_saeonly_name>,
10032 avx512f_vcvttsd2usiq<round_saeonly_name>, ufloatv2siv2df2<mask_name>):
10033 Use %q1 instead of %1 for -masm=intel.
10034 (avx512f_sfixupimm<mode><sd_maskz_name><round_saeonly_name>,
10035 avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Use %<iptr>3 instead
10036 of %3 for -masm=intel.
10037 (sse2_shufpd_v2df_mask): Fix a typo, change %{6%} to %{%6%} for
10039 (*avx512vl_<code>v2div2qi2_store): Use %w0 instead of %0 for
10041 (*avx512vl_<code><mode>v4qi2_store): Use %k0 instead of %0 for
10043 (avx512vl_<code><mode>v4qi2_mask_store): Use a single pattern with
10044 %k0 and %1 for -masm=intel rather than two patterns, one with %0 and
10046 (*avx512vl_<code><mode>v8qi2_store): Use %q0 instead of %0 for
10048 (avx512vl_<code><mode>v8qi2_mask_store): Use a single pattern with
10049 %q0 and %1 for -masm=intel rather than two patterns, one with %0 and
10050 %g1 and one with %0 and %1.
10051 (avx512er_vmrcp28<mode><round_saeonly_name>,
10052 avx512er_vmrsqrt28<mode><round_saeonly_name>): Use %<iptr>1 instead of
10053 %1 for -masm=intel.
10054 (avx5124fmaddps_4fmaddps_mask, avx5124fmaddps_4fmaddss_mask,
10055 avx5124fmaddps_4fnmaddps_mask, avx5124fmaddps_4fnmaddss_mask,
10056 avx5124vnniw_vp4dpwssd_mask, avx5124vnniw_vp4dpwssds_mask): Swap order
10057 of %0 and %{%4%} for -masm=intel.
10058 (avx5124fmaddps_4fmaddps_maskz, avx5124fmaddps_4fmaddss_maskz,
10059 avx5124fmaddps_4fnmaddps_maskz, avx5124fmaddps_4fnmaddss_maskz,
10060 avx5124vnniw_vp4dpwssd_maskz, avx5124vnniw_vp4dpwssds_maskz): Swap
10061 order of %0 and %{%5%}%{z%} for -masm=intel.
10063 2018-04-17 Jan Hubicka <jh@suse.cz>
10066 * ipa-devirt.c (odr_types_equivalent_p): Handle bit fields.
10068 2018-04-17 Martin Liska <mliska@suse.cz>
10071 * multiple_target.c (create_dispatcher_calls): Set apostrophes
10072 for target_clone error message. Make default implementation
10073 clone to be a local declaration.
10074 (separate_attrs): Add new argument and check for an empty
10076 (expand_target_clones): Handle it.
10077 (ipa_target_clone): Make redirection just for target_clones
10080 2018-04-16 Cesar Philippidis <cesar@codesourcery.com>
10081 Tom de Vries <tom@codesourcery.com>
10083 PR middle-end/84955
10084 * omp-expand.c (expand_oacc_for): Add dummy false branch for
10085 tiled basic blocks without omp continue statements.
10087 2018-04-16 Aaron Sawdey <acsawdey@linux.ibm.com>
10090 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Mark
10091 vec_extract expression as having side effects to make sure it gets
10094 2018-04-16 H.J. Lu <hongjiu.lu@intel.com>
10097 * config/i386/i386.c (get_builtin_code_for_version): Check
10100 2018-04-16 Olga Makhotina <olga.makhotina@intel.com>
10103 * config.gcc: Support "skylake".
10104 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
10106 * config/i386/i386.c (m_SKYLAKE): Define.
10107 (processor_target_table): Add "skylake".
10108 (ix86_option_override_internal): Add "skylake".
10109 (get_builtin_code_for_version): Handle PROCESSOR_SKYLAKE,
10110 PROCESSOR_CANNONLAKE.
10111 (get_builtin_code_for_version): Fix priority for
10112 PROCESSOR_ICELAKE_CLIENT, PROCESSOR_ICELAKE_SERVER,
10113 PROCESSOR_SKYLAKE-AVX512.
10114 * config/i386/i386.h (processor_costs): Define TARGET_SKYLAKE.
10115 (processor_type): Add PROCESSOR_SKYLAKE.
10117 2018-04-16 Paolo Carlini <paolo.carlini@oracle.com>
10118 Jason Merrill <jason@redhat.com>
10121 * convert.c (convert_to_integer_1): Use direct recursion for
10122 enumeral types and types with a precision less than the number
10123 of bits in their mode.
10125 2018-04-16 Julia Koval <julia.koval@intel.com>
10128 * config/i386/x86-tune.def (X86_TUNE_SSE_UNALIGNED_LOAD_OPTIMAL,
10129 X86_TUNE_SSE_UNALIGNED_STORE_OPTIMAL): Add m_SKYLAKE_AVX512
10131 2018-04-14 Segher Boessenkool <segher@kernel.crashing.org>
10134 * config/rs6000/rs6000.opt (mdirect-move): Make deprecated.
10135 * doc/invoke.texi (RS/6000 and PowerPC Options): Remove -mdirect-move
10136 and -mno-direct-move.
10138 2018-04-13 Paul A. Clarke <pc@us.ibm.com>
10141 * config/rs6000/emmintrin.h (_mm_slli_epi{16,32,64}):
10142 Ensure that vec_splat_s32 is only called with 0 <= shift < 16.
10143 Ensure negative shifts result in {0}.
10145 2018-04-13 Vladimir Makarov <vmakarov@redhat.com>
10147 PR rtl-optimization/79916
10148 * config/rs6000/rs6000.c (rs6000_emit_move): Use assigned hard
10149 regs (if any) to define how to gnerate SD moves when LRA is in
10152 2018-04-13 Jakub Jelinek <jakub@redhat.com>
10154 PR rtl-optimization/85393
10155 * except.h (expand_dw2_landing_pad_for_region): Remove declaration.
10156 * except.c (expand_dw2_landing_pad_for_region): Make static.
10157 * bb-reorder.c (fix_up_crossing_landing_pad): In new_bb emit just
10158 a label and unconditional jump to old_bb, rather than
10159 expand_dw2_landing_pad_for_region insn(s) and jump to single_succ
10162 PR rtl-optimization/85376
10163 * simplify-rtx.c (simplify_const_unary_operation): For CLZ and CTZ and
10164 zero op0, if C?Z_DEFINED_VALUE_AT_ZERO is false, return NULL_RTX
10165 instead of a specific value.
10167 2018-04-13 Jan Hubicka <hubicka@ucw.cz>
10168 Bin Cheng <bin.cheng@arm.com>
10170 PR tree-optimization/82965
10171 PR tree-optimization/83991
10172 * cfgloopanal.c (expected_loop_iterations_unbounded): Add
10173 by_profile_only parameter.
10174 * cfgloopmanip.c (scale_loop_profile): Further scale loop's profile
10175 information if the loop was predicted to iterate too many times.
10176 * cfgloop.h (expected_loop_iterations_unbounded): Update prototype
10178 2018-04-13 Jan Hubicka <hubicka@ucw.cz>
10181 * config/i386/i386.c (ix86_can_inline_p): Allow safe transitions for
10184 2018-04-13 Martin Liska <mliska@suse.cz>
10185 Jakub Jelinek <jakub@redhat.com>
10187 PR middle-end/81657
10188 * expr.h (enum block_op_methods): Add BLOCK_OP_NO_LIBCALL_RET.
10189 * expr.c (emit_block_move_hints): Handle BLOCK_OP_NO_LIBCALL_RET.
10190 * builtins.c (expand_builtin_memory_copy_args): Use
10191 BLOCK_OP_NO_LIBCALL_RET method for mempcpy with non-ignored target,
10192 handle dest_addr == pc_rtx.
10194 2018-04-12 Segher Boessenkool <segher@kernel.crashing.org>
10197 * config/rs6000/rs6000.md (fix_trunc<mode>si2): Use legacy code if
10198 asked to not generate direct moves.
10199 (fix_trunc<mode>si2_stfiwx): Similar.
10200 (fix_trunc<mode>si2_internal): Similar.
10202 2018-04-12 Jakub Jelinek <jakub@redhat.com>
10205 * var-tracking.c (add_stores): Handle STRICT_LOW_PART SET_DEST.
10206 * cselib.c (cselib_record_sets): For STRICT_LOW_PART dest,
10207 lookup if dest in some wider mode is known to be const0_rtx and
10208 if so, record permanent equivalence for it to be ZERO_EXTEND of
10209 the narrower mode destination.
10211 2018-04-12 Cesar Philippidis <cesar@codesourcery.com>
10213 * lto-streamer-out.c (output_function): Revert 259346.
10214 * omp-expand.c (expand_oacc_for): Likewise.
10216 2018-04-12 Alexander Monakov <amonakov@ispras.ru>
10218 PR rtl-optimization/85354
10219 * sel-sched-ir.c (sel_init_pipelining): Move cfg_cleanup call...
10220 * sel-sched.c (sel_global_init): ... here.
10222 2018-04-12 Eric Botcazou <ebotcazou@adacore.com>
10225 * lto-wrapper.c (debug_objcopy): Open the files in binary mode.
10226 * dwarf2out.c (dwarf2out_early_finish): Do not generate assembly in LTO
10227 mode for PE-COFF targets.
10228 * config/i386/i386-protos.h (i386_pe_asm_lto_start): Declare.
10229 (i386_pe_asm_lto_end): Likewise.
10230 * config/i386/cygming.h (TARGET_ASM_LTO_START): Define.
10231 (TARGET_ASM_LTO_END): Likewise.
10232 * config/i386/winnt.c (saved_debug_info_level): New static variable.
10233 (i386_pe_asm_lto_start): New function.
10234 (i386_pe_asm_lto_end): Likewise.
10236 2018-04-12 Cesar Philippidis <cesar@codesourcery.com>
10237 Richard Biener <rguenther@suse.de>
10239 PR middle-end/84955
10240 * lto-streamer-out.c (output_function): Fix CFG loop state before
10242 * omp-expand.c (expand_oacc_for): Handle calls to internal
10243 functions like regular functions.
10245 2018-04-12 Richard Biener <rguenther@suse.de>
10248 * dwarf2out.c (init_sections_and_labels): Use debug_line_section[_label]
10249 for the early LTO debug to properly generate references to it
10250 during DIE emission. Do not re-use that for the skeleton for
10252 (dwarf2out_early_finish): Likewise.
10254 2018-04-12 Jakub Jelinek <jakub@redhat.com>
10257 * config/i386/sse.md
10258 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name> split,
10259 <mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name> split,
10260 vec_extract_lo_<mode><mask_name> split, vec_extract_lo_v32hi,
10261 vec_extract_lo_v64qi): For non-AVX512VL if input is xmm16+ reg
10262 and output is a reg, avoid creating invalid lowpart subreg, but
10263 instead split into a 512-bit move. Don't split if not AVX512VL,
10264 input is xmm16+ reg and output is a mem.
10265 (vec_extract_lo_<mode><mask_name>, vec_extract_lo_v32hi,
10266 vec_extract_lo_v64qi): Don't require split if not AVX512VL, input is
10267 xmm16+ reg and output is a mem.
10269 2018-04-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10271 * config/s390/s390.c (s390_output_indirect_thunk_function): Check
10272 also for flag_dwarf2_cfi_asm.
10274 2018-04-12 Jakub Jelinek <jakub@redhat.com>
10276 PR rtl-optimization/85342
10277 * regcprop.c (copyprop_hardreg_forward_1): Remove replaced array, use
10278 a bool scalar var inside of the loop instead. Don't try to update
10279 recog_data.operand after failed apply_change_group.
10281 2018-04-12 Tom de Vries <tom@codesourcery.com>
10284 * config/nvptx/nvptx.c (flexible_array_member_type_p): New function.
10285 (nvptx_assemble_decl_begin): Add undefined param. Declare undefined
10286 array with flexible array member as array without given dimension.
10287 (nvptx_assemble_undefined_decl): Set nvptx_assemble_decl_begin call
10288 argument for undefined param to true.
10290 2018-04-11 Aaron Sawdey <acsawdey@linux.ibm.com>
10293 * doc/invoke.texi (RS/6000 and PowerPC Options): Document options
10294 -mcall- and -mtraceback=. Remove options -mabi=spe and -mabi=no-spe
10295 from PowerPC section.
10296 * config/rs6000/sysv4.opt (mcall-): Improve help text.
10297 * config/rs6000/rs6000.opt (mblock-compare-inline-limit=): Trim
10298 help text that is too long.
10299 * config/rs6000/rs6000.opt (mblock-compare-inline-loop-limit=): Trim
10300 help text that is too long.
10301 * config/rs6000/rs6000.opt (mstring-compare-inline-limit=): Trim
10302 help text that is too long.
10304 2018-04-11 Uros Bizjak <ubizjak@gmail.com>
10306 * config/alpha/alpha.md (stack_probe_internal): Rename
10307 from "probe_stack". Update all callers.
10309 2018-04-11 Alexander Monakov <amonakov@ispras.ru>
10311 PR rtl-optimization/84566
10312 * sched-deps.c (sched_analyze_insn): Check deps->readonly when invoking
10313 sched_macro_fuse_insns.
10315 2018-04-11 Alexander Monakov <amonakov@ispras.ru>
10318 * sched-rgn.c (add_branch_dependences): Move sel_sched_p check here...
10319 (compute_block_dependences): ... from here.
10321 2018-04-11 Jakub Jelinek <jakub@redhat.com>
10323 PR tree-optimization/85331
10324 * vec-perm-indices.h (vec_perm_indices::clamp): Change input type
10325 from int to HOST_WIDE_INT.
10327 2018-04-11 Martin Jambor <mjambor@suse.cz>
10330 * ipa-cp.c (propagate_vals_across_pass_through): Expand comment.
10331 (cgraph_edge_brings_value_p): New parameter dest_val, check if it is
10332 not the same as the source val.
10333 (cgraph_edge_brings_value_p): New parameter.
10334 (gather_edges_for_value): Pass destination value to
10335 cgraph_edge_brings_value_p.
10336 (perhaps_add_new_callers): Likewise.
10337 (get_info_about_necessary_edges): Likewise and exclude values brought
10338 only by self-recursive edges.
10339 (create_specialized_node): Redirect only clones of self-calling edges.
10340 (+self_recursive_pass_through_p): New function.
10341 (find_more_scalar_values_for_callers_subset): Use it.
10342 (find_aggregate_values_for_callers_subset): Likewise.
10343 (known_aggs_to_agg_replacement_list): Removed.
10344 (decide_whether_version_node): Re-calculate known constants for all
10345 remaining context clones.
10347 2018-04-11 Richard Biener <rguenther@suse.de>
10350 * dwarf2out.c (dwarf2out_finish): Remove DW_AT_stmt_list attribute
10351 from early DWARF output.
10352 (dwarf2out_early_finish): Output line info unconditionally into
10353 early DWARF and add reference to it.
10355 2018-04-11 Jakub Jelinek <jakub@redhat.com>
10358 * config/i386/sse.md (iptr): Add V16SFmode and V8DFmode cases.
10359 (<avx512>_vec_dup<mode><mask_name>): Use a single pattern for modes
10360 other than V2DFmode using iptr mode attribute.
10361 (<avx512>_vec_dup<mode><mask_name>): Use iptr mode attribute.
10363 2018-04-11 Alexander Monakov <amonakov@ispras.ru>
10365 PR rtl-optimization/84659
10366 * sel-sched-ir.c (sel_init_pipelining): Invoke cleanup_cfg.
10368 2018-04-11 Jakub Jelinek <jakub@redhat.com>
10371 * dwarf2out.c (skip_loc_list_entry): Don't call size_of_locs if
10373 (output_loc_list): Pass address of a dummy size variable even in the
10374 locview handling loop.
10375 (index_location_lists): Add comment on why skip_loc_list_entry can't
10378 2018-04-11 Thomas Preud'homme <thomas.preudhomme@arm.com>
10381 * config/arm/arm-builtins.c (arm_expand_builtin): Force input operand
10384 2018-04-10 Aaron Sawdey <acsawdey@linux.ibm.com>
10387 * doc/invoke.texi (RS/6000 and PowerPC Options): Document options
10388 -mblock-compare-inline-limit, -mblock-compare-inline-loop-limit,
10389 and -mstring-compare-inline-limit.
10391 2018-04-10 Segher Boessenkool <segher@kernel.crashing.org>
10394 * config/rs6000/rs6000.md (allocate_stack): Put the residual size
10395 for stack clash protection in a register whenever we need it to be in
10398 2018-04-10 Segher Boessenkool <segher@kernel.crashing.org>
10400 * common/config/rs6000/rs6000-common.c (rs6000_option_init_struct):
10401 Enable -fasynchronous-unwind-tables by default if OBJECT_FORMAT_ELF.
10403 2018-04-10 Segher Boessenkool <segher@kernel.crashing.org>
10406 * config/rs6000/rs6000.opt (mtraceback=): Show the allowed values in
10408 (mlong-double-): Ditto.
10409 * config/rs6000/sysv4.opt (msdata=): Ditto.
10410 (mtls-size=): Ditto.
10412 2018-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
10414 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
10415 erroneous entries for
10416 "vector int vec_ldl (int, long int *)", and
10417 "vector unsigned int vec_ldl (int, unsigned long int *)".
10418 Add comments and entries for
10419 "vector bool char vec_ldl (int, bool char *)",
10420 "vector bool short vec_ldl (int, bool short *)",
10421 "vector bool int vec_ldl (int, bool int *)",
10422 "vector bool long long vec_ldl (int, bool long long *)",
10423 "vector pixel vec_ldl (int, pixel *)",
10424 "vector long long vec_ldl (int, long long *)",
10425 "vector unsigned long long vec_ldl (int, unsigned long long *)".
10426 * config/rs6000/rs6000.c (rs6000_init_builtins): Initialize new
10427 type tree bool_long_long_type_node and correct definition of
10428 bool_V2DI_type_node to make reference to this new type tree.
10429 (rs6000_mangle_type): Replace erroneous reference to
10430 bool_long_type_node with bool_long_long_type_node.
10431 * config/rs6000/rs6000.h (enum rs6000_builtin_type_index): Add
10432 comments to emphasize sign distinctions for char and int types and
10433 replace RS6000_BTI_bool_long constant with
10434 RS6000_BTI_bool_long_long constant. Also add comment to restrict
10435 use of RS6000_BTI_pixel.
10436 (bool_long_type_node): Remove this macro definition.
10437 (bool_long_long_type_node): New macro definition
10439 2018-04-10 Jakub Jelinek <jakub@redhat.com>
10441 PR rtl-optimization/85300
10442 * combine.c (subst): Handle subst of CONST_SCALAR_INT_P new_rtx also
10443 into FLOAT and UNSIGNED_FLOAT like ZERO_EXTEND, return a CLOBBER if
10444 simplify_unary_operation fails.
10446 2018-04-10 Martin Liska <mliska@suse.cz>
10448 * gdbhooks.py: Add pretty-printers for varpool_node, symtab_node,
10449 cgraph_edge and ipa_ref.
10451 2018-04-10 Jakub Jelinek <jakub@redhat.com>
10455 * config/i386/sse.md
10456 (<extract_type>_vinsert<shuffletype><extract_suf>_mask): Fix
10457 computation of the VEC_MERGE selector from mask.
10458 (<extract_type>_vinsert<shuffletype><extract_suf>_1<mask_name>):
10459 Fix decoding of the VEC_MERGE selector into mask.
10461 2018-04-10 Richard Sandiford <richard.sandiford@linaro.org>
10463 PR tree-optimization/85286
10464 * tree-vect-data-refs.c (vect_get_smallest_scalar_type):
10466 2018-04-10 Richard Sandiford <richard.sandiford@linaro.org>
10468 * final.c (final_1): Set insn_last_address as well as
10469 insn_current_address.
10471 2018-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10474 * explow.c (emit_stack_probe): Call validize_mem on memory location
10475 before passing it to gen_probe_stack. Create address operand and
10476 legitimize it for the probe_stack_address case.
10478 2018-04-09 Jan Hubicka <jh@suse.cz>
10481 * ipa-devirt.c (rebuild_type_inheritance-hash): New.
10482 * ipa-utils.h (rebuild_type_inheritance-hash): Declare.
10483 * tree.c (free_lang_data_in_type): Fix handling of binfos;
10485 (free_lang_data): Rebuild type inheritance graph.
10487 2018-04-09 Martin Sebor <msebor@redhat.com>
10489 * invoke.texi (-finline-small-functions): Mention other optimization
10491 (-findirect-inlining, -fpartial-inlining): Same.
10492 (-finline-functions-called-once): Same.
10493 (-freorder-blocks-and-partition): Same.
10495 2018-04-09 Jan Hubicka <jh@suse.cz>
10498 * cfgcleanup.c (try_forward_edges): Do not give up on crossing
10499 jumps; choose last target that matches the criteria (i.e.
10500 no partition changes for non-crossing jumps).
10501 * cfgrtl.c (cfg_layout_redirect_edge_and_branch): Add basic
10502 support for redirecting crossing jumps to non-crossing.
10504 2018-04-09 Alexey Brodkin <abrodkin@synopsys.com>
10506 * config/arc/arc.c (arc_expand_prologue): Set stack usage info
10507 also for naked functions.
10509 2018-04-09 Claudiu Zissulescu <claziss@synopsys.com>
10511 * config/arc/arc.md (add_shift): New pattern.
10512 (add_shift2): Likewise.
10513 (sub_shift): Likewise.
10514 (sub_shift_cmp0_noout): Likewise.
10515 (compare_si_ashiftsi): Likewise.
10516 (xbfu_cmp0_noout): New combine pattern.
10517 (xbfu_cmp0"): Likewise.
10518 (movsi_set_cc_insn): Place the predicable variant first.
10519 (commutative_binary_cmp0_noout): Remove clobber.
10520 (commutative_binary_cmp0): New pattern.
10521 (noncommutative_binary_cmp0): Likewise.
10522 (noncommutative_binary_cmp0_noout): Likewise.
10523 (noncommutative_binary_comparison_result_used): Removed.
10524 (rsub_cmp0): New pattern.
10525 (rsub_cmp0_noout): Likewise.
10526 (extzvsi): Changed, keep only meaningful variants.
10527 (SQH, SEZ): New iterators.
10528 (SQH_postfix): New mode attribute.
10529 (SEZ_prefix): New code attribute.
10530 (<SEZ_prefix>xt<SQH_postfix>_cmp0_noout): New instruction pattern.
10531 (<SEZ_prefix>xt<SQH_postfix>_cmp0): Likewise.
10532 * config/arc/predicates.md (cc_set_register): Use CC_REG instead
10533 of numerical value.
10534 (noncommutative_operator): Check the availability of barrel
10537 2018-04-09 Richard Biener <rguenther@suse.de>
10539 PR tree-optimization/85284
10540 * tree-ssa-loop-niter.c (number_of_iterations_exit_assumptions):
10541 Only use the niter constraining form of simple_iv when the exit
10542 is always executed.
10544 2018-04-09 Tom de Vries <tom@codesourcery.com>
10547 * config/nvptx/nvptx.md (define_c_enum "unspecv"): Add UNSPECV_MEMBAR.
10548 (define_expand "*memory_barrier"): New define_expand.
10549 (define_insn "memory_barrier"): New insn.
10551 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
10553 PR rtl-optimization/80463
10554 PR rtl-optimization/83972
10555 PR rtl-optimization/83480
10557 * sel-sched-ir.c (has_dependence_note_mem_dep): Take into account the
10558 correct producer for the insn.
10559 (tidy_control_flow): Fixup seqnos in case of debug insns.
10561 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
10563 PR rtl-optimization/83913
10565 * sel-sched-ir.c (merge_expr_data): Choose the middle between two
10566 different sched-times when merging exprs.
10568 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
10570 PR rtl-optimization/83962
10572 * sel-sched-ir.c (tidy_control_flow): Correct the order in which we call
10573 tidy_fallthru_edge and tidy_control_flow.
10575 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
10577 PR rtl-optimization/83530
10579 * sel-sched.c (force_next_insn): New global variable.
10580 (remove_insn_for_debug): When force_next_insn is true, also leave only
10581 next insn in the ready list.
10582 (sel_sched_region): When the region wasn't scheduled, make another pass
10583 over it with force_next_insn set to 1.
10585 2018-04-08 Monk Chiang <sh.chiang04@gmail.com>
10587 * config.gcc (nds32le-*-*, nds32be-*-*): Add nds32/nds32_intrinsic.h
10589 * config/nds32/constants.md (unspec_volatile_element): Add enum values
10590 for interrupt control.
10591 * config/nds32/nds32-intrinsic.c: Implementation of intrinsic
10592 functions for interrupt control.
10593 * config/nds32/nds32-intrinsic.md: Likewise.
10594 * config/nds32/nds32_intrinsic.h: Likewise.
10595 * config/nds32/nds32.h (nds32_builtins): Likewise.
10597 2018-04-08 Chung-Ju Wu <jasonwucj@gmail.com>
10599 * config/nds32/nds32.c (nds32_init_machine_status,
10600 nds32_legitimate_index_p, nds32_legitimate_address_p): Consider
10601 strict_aligned_p field.
10602 (nds32_expand_to_rtl_hook): New function.
10603 (TARGET_EXPAND_TO_RTL_HOOK): Define.
10604 * config/nds32/nds32.h (machine_function): Add strict_aligned_p field.
10606 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
10607 Chung-Ju Wu <jasonwucj@gmail.com>
10609 * config.gcc (nds32*-*-*): Check that n7 is valid to --with-cpu.
10610 * config/nds32/nds32-n7.md: New file.
10611 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N7.
10612 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n7
10614 * config/nds32/nds32-protos.h: More declarations for n7 pipeline.
10615 * config/nds32/nds32.md (pipeline_model): Add n7.
10616 * config/nds32/nds32.opt (mcpu): Support n7 pipeline cpus.
10617 * config/nds32/pipelines.md: Include n7 settings.
10619 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
10620 Chung-Ju Wu <jasonwucj@gmail.com>
10622 * config.gcc (nds32*-*-*): Check that e8 is valid to --with-cpu.
10623 * config/nds32/nds32-e8.md: New file.
10624 * config/nds32/nds32-opts.h (nds32-cpu_type): Add CPU_E8.
10625 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for e8
10627 * config/nds32/nds32-protos.h: More declarations for e8 pipeline.
10628 * config/nds32/nds32.md (pipeline_model): Add e8.
10629 * config/nds32/nds32.opt (mcpu): Support e8 pipeline cpus.
10630 * config/nds32/pipelines.md: Include e8 settings.
10632 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
10633 Chung-Ju Wu <jasonwucj@gmail.com>
10635 * config.gcc (nds32*-*-*): Check that n6/n8/s8 are valid to --with-cpu.
10636 * config/nds32/nds32-n8.md: New file.
10637 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N6 and CPU_N8.
10638 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n8
10640 * config/nds32/nds32-protos.h: More declarations for n8 pipeline.
10641 * config/nds32/nds32-utils.c: More implementations for n8 pipeline.
10642 * config/nds32/nds32.md (pipeline_model): Add n8.
10643 * config/nds32/nds32.opt (mcpu): Support n8 pipeline cpus.
10644 * config/nds32/pipelines.md: Include n8 settings.
10646 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
10647 Chung-Ju Wu <jasonwucj@gmail.com>
10649 * config.gcc (nds32*): Add nds32-utils.o into extra_objs.
10650 * config/nds32/nds32-n9-2r1w.md: New file.
10651 * config/nds32/nds32-n9-3r2w.md: New file.
10652 * config/nds32/nds32-opts.h (nds32_cpu_type, nds32_mul_type,
10653 nds32_register_ports): New or modify for cpu n9.
10654 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n9
10656 * config/nds32/nds32-protos.h: More declarations for n9 pipeline.
10657 * config/nds32/nds32-utils.c: New file.
10658 * config/nds32/nds32.h (TARGET_PIPELINE_N9, TARGET_PIPELINE_SIMPLE,
10659 TARGET_MUL_SLOW): Define.
10660 * config/nds32/nds32.md (pipeline_model): New attribute.
10661 * config/nds32/nds32.opt (mcpu, mconfig-mul, mconfig-register-ports):
10662 New options that support cpu n9.
10663 * config/nds32/pipelines.md: Include n9 settings.
10664 * config/nds32/t-nds32 (nds32-utils.o): Add dependency.
10666 2018-04-08 Chung-Ju Wu <jasonwucj@gmail.com>
10668 * config/nds32/nds32-md-auxiliary.c (output_cond_branch): Output align
10669 information if necessary.
10670 (output_cond_branch_compare_zero): Likewise.
10671 * config/nds32/nds32.c (nds32_adjust_insn_length): Consider align case.
10672 (nds32_target_alignment): Refine for alignment.
10673 * config/nds32/nds32.h (NDS32_ALIGN_P): Define.
10674 (FUNCTION_BOUNDARY): Modify.
10675 * config/nds32/nds32.md (call_internal, call_value_internal): Consider
10677 * config/nds32/nds32.opt (malways-align, malign-functions): New.
10679 2018-04-08 Monk Chiang <sh.chiang04@gmail.com>
10681 * config/nds32/constants.md (unspec_volatile_element): Add values for
10682 TLB operation and data prefetch.
10683 * config/nds32/nds32-intrinsic.c: Implementation of intrinsic
10684 functions for TLB operation and data prefetch.
10685 * config/nds32/nds32-intrinsic.md: Likewise.
10686 * config/nds32/nds32_intrinsic.h: Likewise.
10687 * config/nds32/nds32.c (nds32_dpref_names): Likewise.
10688 (nds32_print_operand): Likewise.
10689 * config/nds32/nds32.h (nds32_builtins): Likewise.
10691 2018-04-07 Thomas Koenig <tkoenig@gcc.gnu.org>
10692 Andrew Pinski <pinsika@gcc.gnu.org>
10694 PR middle-end/82976
10695 * match.pd: Use constant_boolean_node of correct type instead of
10696 boolean_true_node or boolean_false_node for simplifying
10697 pointer comparisons to zero.
10699 2018-04-07 Jakub Jelinek <jakub@redhat.com>
10701 PR tree-optimization/80021
10702 * tree.c (verify_type_variant): Make error call in verify_variant_match
10703 translatable and remove final full stop.
10705 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
10707 * config/nds32/constants.md (unspec_volatile_element): Add
10708 UNSPEC_VOLATILE_EH_RETURN.
10709 * config/nds32/nds32-md-auxiliary.c (nds32_output_stack_push,
10710 nds32_output_stack_pop): Support dwarf exception handling process.
10711 * config/nds32/nds32-protos.h (nds32_dynamic_chain_address): Declare.
10712 * config/nds32/nds32.c (nds32_init_machine_status): Support dwarf
10713 exception handling process.
10714 (nds32_compute_stack_frame): Likewise.
10715 (nds32_return_addr_rtx): Likewise.
10716 (nds32_initial_elimination_offset): Likewise.
10717 (nds32_expand_prologue): Likewise.
10718 (nds32_expand_epilogue): Likewise.
10719 (nds32_dynamic_chain_address): New function.
10720 * config/nds32/nds32.h (machine_function): Add fields for dwarf
10721 exception handling.
10722 (DYNAMIC_CHAIN_ADDRESS): Define.
10723 (EH_RETURN_DATA_REGNO): Define.
10724 (EH_RETURN_STACKADJ_RTX): Define.
10725 * config/nds32/nds32.md (eh_return, nds32_eh_return): Implement
10726 patterns for dwarf exception handling.
10728 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
10730 * config/nds32/nds32.h: Clean up obsolete macros.
10732 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
10734 * config/nds32/constants.md (unspec_element, unspec_volatile_element):
10735 Add enum values for particular instructions.
10736 * config/nds32/nds32-intrinsic.c: Implementation of expanding
10737 particular intrinsic functions.
10738 * config/nds32/nds32-intrinsic.md: Likewise.
10739 * config/nds32/nds32_intrinsic.h: Likewise.
10740 * config/nds32/nds32.h (nds32_builtins): Likewise.
10741 * config/nds32/nds32.md (type): Add pbsad and pbsada.
10742 (btst, ave): New patterns for particular instructions.
10744 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
10746 * config/nds32/constants.md (unspec_element, unspec_volatile_element):
10747 Add enum values for atomic load/store and memory sync.
10748 * config/nds32/nds32-intrinsic.c: Implementation for atomic load/store
10750 * config/nds32/nds32-intrinsic.md: Likewise.
10751 * config/nds32/nds32_intrinsic.h: Likewise.
10752 * config/nds32/nds32.h (nds32_builtins): Likewise.
10754 2018-04-07 Jakub Jelinek <jakub@redhat.com>
10756 PR tree-optimization/85257
10757 * fold-const.c (native_encode_vector): If not all elts could fit
10758 and off is -1, return 0 rather than offset.
10759 * tree-ssa-sccvn.c (vn_reference_lookup_3): Pass
10760 (offseti - offset2) / BITS_PER_UNIT as 4th argument to
10761 native_encode_expr. Verify len * BITS_PER_UNIT >= maxsizei. Don't
10762 adjust buffer in native_interpret_expr call.
10764 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
10766 * config/nds32/constants.md (unspec_volatile_element): Add cache
10767 control enum values.
10768 * config/nds32/nds32-intrinsic.c: Add cache control expand functions.
10769 * config/nds32/nds32-intrinsic.md: Add cache control patterns.
10770 * config/nds32/nds32.c (nds32_cctl_names): New.
10771 (nds32_print_operand): Handle cache control register names.
10772 * config/nds32/nds32.h (nds32_builtins): New enum values.
10773 * config/nds32/nds32_intrinsic.h: Add cache control enum types and
10775 * config/nds32/nds32.md (type): Add mmu.
10776 * config/nds32/pipelines.md (simple_insn): Add mmu.
10778 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
10780 * config/nds32/nds32.md (type): Remove call.
10781 * config/nds32/pipelines.md (simple_insn): Likewise.
10783 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
10785 * config/nds32/constants.md (unspec_volatile_element): Add
10786 UNSPEC_VOLATILE_FMFCSR, UNSPEC_VOLATILE_FMTCSR and
10787 UNSPEC_VOLATILE_FMFCFG.
10788 * config/nds32/nds32-intrinsic.c (bdesc_noarg): New builtin
10789 description for fmfcfg and fmfcsr.
10790 (bdesc_1arg): Add fmtcsr.
10791 (bdesc_2arg): Add fcpynss, fcpyss, fcpynsd and fcpysd.
10792 (nds32_expand_builtin_impl): Deal with FPU intrinsic functions.
10793 * config/nds32/nds32-intrinsic.md (unspec_fcpynsd, unspec_fcpysd,
10794 unspec_fcpynss, unspec_fcpysd, unspec_fcpyss, unspec_fmfcsr,
10795 unspec_fmfcfg): New patterns.
10796 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_FMFCFG,
10797 NDS32_BUILTIN_FMFCSR, NDS32_BUILTIN_FMTCSR, NDS32_BUILTIN_FCPYNSS,
10798 NDS32_BUILTIN_FCPYSS,NDS32_BUILTIN_FCPYNSD and NDS32_BUILTIN_FCPYSD.
10799 * config/nds32/nds32_intrinsic.h (__nds32__fcpynsd, __nds32__fcpynss,
10800 __nds32__fcpysd, __nds32__fcpyss, __nds32__fmfcsr, __nds32__fmtcsr,
10801 __nds32__fmfcfg): Define.
10803 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
10805 * config/nds32/nds32.c (nds32_intrinsic_register_names): Add more
10806 intrinsic register names.
10807 * config/nds32/nds32_intrinsic.h (nds32_intrinsic_registers): Add more
10808 intrinsic register enum values and macros.
10810 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
10812 * config/nds32/nds32.c (nds32_legitimate_index_p): Modify condition
10813 for load/store addressing form.
10814 (nds32_print_operand_address): Likewise.
10816 2018-04-06 Eric Botcazou <ebotcazou@adacore.com>
10819 * config/sparc/sparc.c (sparc_expand_move): Deal with symbolic operands
10820 based on LABEL_REF. Remove useless assertion.
10821 (pic_address_needs_scratch): Fix formatting.
10822 (sparc_legitimize_pic_address): Minor tweaks.
10823 (sparc_delegitimize_address): Adjust assertion accordingly.
10824 * config/sparc/sparc.md (movsi_pic_label_ref): Change label_ref_operand
10825 into symbolic_operand.
10826 (movsi_high_pic_label_ref): Likewise.
10827 (movsi_lo_sum_pic_label_ref): Likewise.
10828 (movdi_pic_label_ref): Likewise.
10829 (movdi_high_pic_label_ref): Likewise.
10830 (movdi_lo_sum_pic_label_ref): Likewise.
10832 2018-04-06 Amaan Cheval <amaan.cheval@gmail.com>
10834 * config.gcc (x86_64-*-rtems*): Add rtems.h to tm_file for
10835 custom LIB_SPEC setup.
10837 2018-04-06 Ruslan Bukin <br@bsdpad.com>
10838 Kito Cheng <kito.cheng@gmail.com>
10840 * config.gcc (riscv*-*-freebsd*): Add RISC-V FreeBSD support.
10841 * config/riscv/freebsd.h: New.
10843 2018-04-06 Chung-Ju Wu <jasonwucj@gmail.com>
10845 * config/nds32/nds32.c (nds32_adjust_insn_length): Refine.
10846 * config/nds32/nds32.h (ADJUST_INSN_LENGTH): Change the location in
10849 2018-04-06 Chung-Ju Wu <jasonwucj@gmail.com>
10850 Kito Cheng <kito.cheng@gmail.com>
10852 * config/nds32/nds32-md-auxiliary.c (nds32_output_return,
10853 nds32_output_call, nds32_symbol_binds_local_p): New functions.
10854 * config/nds32/nds32-protos.h (nds32_output_call,
10855 nds32_output_return): Declare.
10856 * config/nds32/nds32.md: Refine all the call and return patterns.
10858 2018-04-06 Jakub Jelinek <jakub@redhat.com>
10861 * dwarf2out.c (rtl_for_decl_init): For STRING_CST initializer only
10862 build CONST_STRING if TYPE_MAX_VALUE is non-NULL and is INTEGER_CST.
10864 PR rtl-optimization/84872
10865 * cfgloopmanip.c (create_preheader): Use make_forwarder_block even if
10866 nentry == 1 when CP_FALLTHRU_PREHEADERS and single_entry is
10867 EDGE_CROSSING edge.
10869 2018-04-06 Tamar Christina <tamar.christina@arm.com>
10871 * expr.c (copy_blkmode_to_reg): Revert 254862.
10872 * doc/sourcebuild.texi (word_mode_no_slow_unalign): Likewise.
10874 2018-04-06 Richard Biener <rguenther@suse.de>
10876 PR middle-end/85244
10877 * tree-dfa.c (get_ref_base_and_extent): Reset seen_variable_array_ref
10878 after seeing a component reference with an adjacent field. Treat
10879 refs to arrays at struct end of external decls similar to
10880 refs to unconstrained commons.
10882 2018-04-06 Jakub Jelinek <jakub@redhat.com>
10885 * fold-const.c (twoval_comparison_p): Remove SAVE_P argument and don't
10886 look through SAVE_EXPRs with non-side-effects argument. Adjust
10888 (fold_comparison): Adjust twoval_comparison_p caller, don't handle
10891 2018-04-06 Richard Biener <rguenther@suse.de>
10893 PR middle-end/85180
10894 * alias.c (find_base_term): New wrapper around find_base_term
10895 unwinding CSELIB_VAL_PTR changes.
10896 (find_base_term): Do not restore CSELIB_VAL_PTR during the
10899 2018-04-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10901 * config/s390/s390.c (s390_z10_optimize_cmp): Expand dedicated NOP
10903 * config/s390/s390.md (UNSPECV_NOP_LR_0, UNSPECV_NOP_LR_1): New
10904 constant definitions.
10905 ("nop"): lr 0,0 -> nopr r0
10906 ("nop_lr0", "nop_lr1"): New insn definitions.
10908 2018-04-06 Chung-Ju Wu <jasonwucj@gmail.com>
10910 * config/nds32/nds32.md (*stack_push, *stack_pop): Use
10911 NDS32_V3PUSH_AVAILABLE_P macro.
10913 2018-04-06 Monk Chiang <sh.chiang04@gmail.com>
10914 Chung-Ju Wu <jasonwucj@gmail.com>
10916 * config.gcc (nds32*-*-*): Add v2j v3f v3s checking.
10917 (nds32*-*-*): Add float and fpu_config into supported_defaults.
10918 * common/config/nds32/nds32-common.c (TARGET_DEFAULT_TARGET_FLAGS):
10919 Include TARGET_DEFAULT_FPU_ISA and TARGET_DEFAULT_FPU_FMA.
10920 * config/nds32/constants.md (unspec_element): Add UNSPEC_COPYSIGN,
10921 UNSPEC_FCPYNSD, UNSPEC_FCPYNSS, UNSPEC_FCPYSD and UNSPEC_FCPYSS.
10922 * config/nds32/constraints.md: New constraints and checking for hard
10923 float configuration.
10924 * config/nds32/iterators.md: New mode iterator and attribute for hard
10925 float configuration.
10926 * config/nds32/nds32-doubleword.md: Use hard float alternatives and
10928 * config/nds32/nds32-fpu.md: New file.
10929 * config/nds32/nds32-md-auxiliary.c: New functions and checkings to
10930 deal with hard float code generation.
10931 * config/nds32/nds32-opts.h (nds32_arch_type): Add ARCH_V3F and
10933 (abi_type, float_reg_number): New enum type.
10934 * config/nds32/nds32-predicates.c: New predicates for hard float.
10935 * config/nds32/nds32-protos.h: Declare functions for hard float.
10936 * config/nds32/nds32.c: Implementation for hard float configuration.
10937 * config/nds32/nds32.h: Definitions for hard float configuration.
10938 * config/nds32/nds32.md: Include hard float machine description and
10939 modify patterns for hard float configuration.
10940 * config/nds32/nds32.opt: New options for hard float configuration.
10941 * config/nds32/predicates.md: New predicates for hard float
10944 2018-04-06 Kuan-Lin Chen <kuanlinchentw@gmail.com>
10946 * common/config/nds32/nds32-common.c
10947 (nds32_option_optimization_table): Enable -mreleax-hint by default.
10949 2018-04-05 Jakub Jelinek <jakub@redhat.com>
10951 PR middle-end/85195
10952 * match.pd (BIT_FIELD_REF CONSTRUCTOR@0 @1 @2): Use view_convert around
10953 CONSTRUCTOR_ELT (ctor, ...)->value.
10955 2018-04-05 Uros Bizjak <ubizjak@gmail.com>
10958 * config/i386/i386.md (define_attr "memory"): Handle rotate1 type.
10960 2018-04-05 Tom de Vries <tom@codesourcery.com>
10963 * config/nvptx/nvptx.c (nvptx_single): Fix neutering of bb with only
10966 2018-04-05 Shiva Chen <shiva0217@gmail.com>
10967 Kito Cheng <kito.cheng@gmail.com>
10969 * config/nds32/constraints.md (U33): Fine-tune checking condition.
10970 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Ditto.
10971 * config/nds32/nds32.h (nds32_16bit_address_type): Add
10972 ADDRESS_POST_MODIFY_LO_REG_IMM3U.
10974 2018-04-05 Shiva Chen <shiva0217@gmail.com>
10975 Kito Cheng <kito.cheng@gmail.com>
10977 * config/nds32/constraints.md (Ufe): New memory constraint.
10978 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format,
10979 nds32_output_16bit_load): Consider r8 register for lwi45.fe format.
10980 * config/nds32/nds32.c (nds32_print_operand): Output lwi45.fe
10982 * config/nds32/nds32.h (nds32_16bit_address_type): Add ADDRESS_R8_IMM7U.
10983 * config/nds32/nds32.md (*mov<mode>): Adjust pattern.
10985 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
10987 * config/nds32/nds32.md: Use optimize_size in the condition for
10988 alu-shift instructions.
10990 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
10992 * config/nds32/nds32.md (divsi4, udivsi4): New patterns.
10994 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
10996 * config/nds32/nds32.md (negsi2): Refine pattern.
10998 2018-04-05 Kito Cheng <kito.cheng@gmail.com>
10999 Chung-Ju Wu <jasonwucj@gmail.com>
11001 * config/nds32/iterators.md (shift_rotate): New code iterator.
11002 (shift): New code attribute.
11003 * config/nds32/nds32-md-auxiliary.c (nds32_expand_constant): New.
11004 * config/nds32/nds32-protos.h (nds32_expand_constant): Declare.
11005 * config/nds32/nds32.c (nds32_print_operand): Deal with more cases.
11006 * config/nds32/nds32.md (addsi3, *add_srli): Refine implementation for
11007 bit-wise operations.
11008 (andsi3, *andsi3): Ditto.
11009 (iorsi3, *iorsi3, *or_slli, *or_srli): Ditto.
11010 (xorsi3, *xorsi3, *xor_slli, *xor_srli): Ditto.
11011 (<shift>si3, *ashlsi3, *ashrsi3, *lshrsi3, *rotrsi3): Ditto.
11012 * config/nds32/predicates.md (nds32_rimm5u_operand, nds32_and_operand,
11013 nds32_ior_operand, nds32_xor_operand): New predicates.
11015 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
11017 * config/nds32/nds32.md (add<mode>3, sub<mode>3): Rename to ...
11018 (addsi3, subsi3): ... this.
11020 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
11022 * config/nds32/nds32.md (*sub_srli, *and_slli): Fine-tune predicator.
11024 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
11026 * config/nds32/nds32.md: Adjust indention.
11028 2018-04-05 Kito Cheng <kito.cheng@gmail.com>
11030 * config/nds32/nds32.md (feature): New attribute.
11032 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
11034 * config/nds32/nds32.md (subtype): New attribute.
11036 2018-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
11039 * config/arm/arm-builtins.c (arm_expand_builtin): Change
11040 expansion to perform a bitwise AND of the argument followed by a
11041 boolean negation of the result.
11043 2018-04-04 Peter Bergner <bergner@vnet.ibm.com>
11045 PR rtl-optimization/84878
11046 * ddg.c (add_cross_iteration_register_deps): Use DF_REF_BB to determine
11047 the basic block. Assert the use reference is not artificial and that
11048 it has an associated insn.
11050 2018-04-04 Michael Matz <matz@suse.de>
11052 * builtins.c (compute_objsize): Pass correct operand
11053 to array_at_struct_end_p.
11055 2018-04-04 Richard Biener <rguenther@suse.de>
11058 * dwarf2out.c (dwarf2out_register_external_die): Peel namespaces
11059 from contexts for DINFO_LEVEL_TERSE and below.
11061 2018-04-04 Kito Cheng <kito.cheng@gmail.com>
11063 * config/nds32/nds32-doubleword.md (move_<mode>): Require
11064 resiter_operand condition.
11065 * config/nds32/nds32.md (*move<mode>): Ditto.
11067 2018-04-04 Kito Cheng <kito.cheng@gmail.com>
11068 Monk Chiang <sh.chiang04@gmail.com>
11070 * config/nds32/nds32.md (movmisalign<mode>): New pattern.
11072 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
11074 * config/nds32/nds32.md (movqi, movhi): Merge into mov<mode>.
11076 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
11077 Kito Cheng <kito.cheng@gmail.com>
11079 * config/nds32/nds32-md-auxiliary.c (nds32_inverse_cond_code,
11080 nds32_cond_code_str, output_cond_branch,
11081 output_cond_branch_compare_zero, nds32_expand_cbranch,
11082 nds32_expand_cstore, nds32_expand_movcc,
11083 nds32_output_cbranchsi4_equality_zero,
11084 nds32_output_cbranchsi4_equality_reg,
11085 nds32_output_cbranchsi4_equality_reg_or_const_int,
11086 nds32_output_cbranchsi4_greater_less_zero: New functions.
11087 * config/nds32/nds32-protos.h (nds32_expand_cbranch,
11088 nds32_expand_cstore, nds32_expand_movcc,
11089 nds32_output_cbranchsi4_equality_zero,
11090 nds32_output_cbranchsi4_equality_reg,
11091 nds32_output_cbranchsi4_equality_reg_or_const_int,
11092 nds32_output_cbranchsi4_greater_less_zero): Declare.
11093 * config/nds32/predicates.md (nds32_movecc_comparison_operator,
11094 nds32_rimm11s_operand): New predicates.
11095 * config/nds32/nds32.h (nds32_expand_result_type): New enum type.
11096 * config/nds32/nds32.md: Rewrite all the branch and conditional move
11099 2018-04-04 Kito Cheng <kito.cheng@gmail.com>
11101 * config/nds32/nds32-doubleword.md: Refine all the instruction type.
11102 * config/nds32/nds32.md: Ditto.
11103 * config/nds32/pipelines.md: Ditto.
11105 2018-04-04 Richard Biener <rguenther@suse.de>
11107 PR tree-optimization/85168
11108 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address): Avoid
11109 propagating abnormals.
11111 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
11113 * config/nds32/nds32.md (enabled): Use yes/no for this attribute.
11115 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
11116 Kito Cheng <kito.cheng@gmail.com>
11118 * config/nds32/nds32-md-auxiliary.c (nds32_long_call_p): New function.
11119 * config/nds32/nds32-protos.h (nds32_long_call_p): Declare.
11120 * config/nds32/nds32.c (nds32_function_ok_for_sibcall): New function.
11121 (TARGET_FUNCTION_OK_FOR_SIBCALL): Define.
11122 * config/nds32/nds32.md (sibcall_internal): New.
11123 (sibcall_register): Remove.
11124 (sibcall_immediate): Remove.
11125 (sibcall_value_internal): New.
11126 (sibcall_value_register): Remove.
11127 (sibcall_value_immediate): Remove.
11128 * config/nds32/predicates.md (nds32_general_register_operand): New.
11129 (nds32_call_address_operand): New.
11131 2018-04-03 Jakub Jelinek <jakub@redhat.com>
11133 PR rtl-optimization/85167
11134 * shrink-wrap.c (move_insn_for_shrink_wrap): Don't set bb_uses and
11135 bb_defs if *split_p, instead preinitialize it to NULL.
11137 PR tree-optimization/85156
11138 * builtins.c (fold_builtin_expect): Use save_expr on arg1 to avoid
11139 evaluating the argument multiple times.
11141 2018-04-03 Bill Schmidt <wschmidt@linux.ibm.com>
11143 * config/rs6000/emmintrin.h (_mm_cvtpd_epi32): Use __vector rather
11145 (_mm_cvtpd_ps): Likewise.
11146 (_mm_cvttpd_epi32): Likewise.
11147 * config/rs6000/mmintrin.h (_mm_unpacklo_pi8): Likewise.
11148 * config/rs6000/xmmintrin.h: For strict-ANSI C++ or C11, undefine
11149 vector, pixel, and bool following altivec.h include.
11151 2018-04-03 Martin Sebor <msebor@redhat.com>
11153 * doc/extend.texi (Common Function Attributes): Clarify.
11154 (const attribute): Likewise.
11155 (pure attribute): Likewise.
11157 2018-04-03 Jakub Jelinek <jakub@redhat.com>
11160 * config/i386/i386.c (ix86_expand_vector_set): Use
11161 HOST_WIDE_INT_1U << elt instead of 1 << elt. Formatting fix.
11163 2018-04-03 Uros Bizjak <ubizjak@gmail.com>
11165 * config/i386/i386.c (emit_i387_cw_initialization): Always use logic
11166 instructions when changing rounding bits to preserve precision bits
11167 in the x87 control word.
11169 2018-04-03 Martin Liska <mliska@suse.cz>
11171 PR tree-optimization/82491
11172 * rtl.h (strip_offset_and_add): Replace += suboffset with
11173 poly_uint64 () + suboffset.
11175 2018-03-29 Martin Liska <mliska@suse.cz>
11176 Martin Jambor <mjambor@suse.cz>
11179 * ipa-cp.c (propagate_bits_across_jump_function): Bail out if
11180 param_type is not an integral or pointer type.
11182 2018-04-03 Richard Biener <rguenther@suse.de>
11184 * sese.h (recompute_all_dominators): Remove.
11186 2018-04-02 Martin Sebor <msebor@redhat.com>
11188 * doc/invoke.texi (-Wrestrict): Fix typos.
11190 2018-04-02 Jim Wilson <jimw@sifive.com>
11192 * config/riscv/riscv.h (SHIFT_COUNT_TRUNCATED): Set to zero.
11193 * config/riscv/riscv.md (<optab>si3): Use QImode shift count.
11194 (<optab>di3, <optab>si3_extend): Likewise.
11195 (<optab>si3_mask, <optab>si3_mask_1): New.
11196 (<optab>di3_mask, <optab>di3_mask_1): New.
11197 (<optab>si3_extend_mask, <optab>si3_extend_mask_1): New.
11198 (lshrsi3_zero_extend_1): Use VOIDmode shift count.
11199 * config/riscv/sync.md (atomic_test_and_set): Emit QImode shift count.
11201 2018-04-02 Gerald Pfeifer <gerald@pfeifer.com>
11203 * doc/cpp.texi (Variadic Macros): Fix line continuation in an
11206 2018-04-02 Chung-Ju Wu <jasonwucj@gmail.com>
11208 * config/nds32/nds32.c (TARGET_CANONICALIZE_COMPARISON): Define.
11209 (nds32_canonicalize_comparison): New function.
11211 2018-04-01 Chung-Ju Wu <jasonwucj@gmail.com>
11212 Kito Cheng <kito.cheng@gmail.com>
11213 Kuan-Lin Chen <kuanlinchentw@gmail.com>
11215 * config.gcc (nds32): Add nds32-relax-opt.o into extra_objs.
11216 * config/nds32/constants.md (unspec_volatile_element): Add
11217 UNSPEC_VOLATILE_RELAX_GROUP.
11218 * config/nds32/nds32-relax-opt.c: New file.
11219 * config/nds32/nds32-predicates.c
11220 (nds32_symbol_load_store_p): New function.
11221 * config/nds32/nds32-protos.h
11222 (nds32_symbol_load_store_p): Declare function.
11223 (make_pass_nds32_relax_opt): Declare new rtl pass function.
11224 * config/nds32/nds32.c
11225 (nds32_register_pass): New function to register pass.
11226 (nds32_register_passes): New function to register passes.
11227 * config/nds32/nds32.md (relax_group): New pattern.
11228 * config/nds32/nds32.opt (mrelax-hint): New option.
11229 * config/nds32/t-nds32 (nds32-relax-opt.o): New dependency.
11231 2018-04-01 Kito Cheng <kito.cheng@gmail.com>
11233 * config/nds32/t-nds32: Modify files dependency.
11235 2018-04-01 Chung-Ju Wu <jasonwucj@gmail.com>
11237 * config/nds32/nds32.h (FUNCTION_PROFILER): Output newline character.
11238 (PROFILE_HOOK): Define its implementation.
11240 2018-04-01 Chung-Ju Wu <jasonwucj@gmail.com>
11242 * config/nds32/nds32.h (WCHAR_TYPE, WCHAR_TYPE_SIZE): Use unsigned int
11243 type and 32-bit size.
11245 2018-04-01 Jakub Jelinek <jakub@redhat.com>
11247 PR middle-end/85090
11248 * config/i386/sse.md (V): Add V64QI and V32HI for TARGET_AVX512F.
11249 (V_128_256): New mode iterator.
11250 (*avx512dq_vextract<shuffletype>64x2_1 splitter): New define_split.
11251 (*avx512f_vextract<shuffletype>32x4_1 splitter): Likewise.
11252 (xop_pcmov_<mode><avxsizesuffix>): Use V_128_256 mode iterator instead
11254 * config/i386/i386.c (ix86_expand_vector_set): Improve V32HImode and
11255 V64QImode expansion for !TARGET_AVX512BW && TARGET_AVX512F.
11257 2018-03-31 Segher Boessenkool <segher@kernel.crashing.org>
11260 * config/rs6000/xmmintrin.h (_mm_set_ps, _mm_max_ps): Handle (quiet)
11261 NaN inputs correctly.
11263 2018-03-30 Peter Bergner <bergner@vnet.ibm.com>
11266 * config/rs6000/vsx.md (??r): New mode attribute.
11267 (*vsx_mov<mode>_64bit): Use it.
11268 (*vsx_mov<mode>_32bit): Likewise.
11270 2018-03-30 Martin Sebor <msebor@redhat.com>
11272 PR tree-optimization/84818
11273 * builtins.c (check_access): Use warning_n.
11275 2018-03-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
11278 * config/rs6000/rs6000-string.c (expand_compare_loop): Fix redundant
11280 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Fix redundant
11283 2018-03-30 Julia Koval <julia.koval@intel.com>
11286 * x86-tune.def (movx, partial_reg_dependency): Enable for
11289 2018-03-29 Vladimir Makarov <vmakarov@redhat.com>
11291 PR inline-asm/84985
11292 * lra-constraints.c (process_alt_operands): Move setting
11293 this_alternative_matches below.
11295 2018-03-29 Martin Liska <mliska@suse.cz>
11298 * doc/invoke.texi: Document how LTO works with debug info.
11299 Describe auto-load support of binutils. Mention 'x86-64'
11300 as valid option value of -march option.
11302 2018-03-29 Jakub Jelinek <jakub@redhat.com>
11304 * config/i386/sse.md (<avx512>_blendm<mode>): Use <sseintprefix>.
11307 * fold-const.c (operand_equal_p): Handle DEBUG_BEGIN_STMT.
11308 For STATEMENT_LIST, pass down OEP_LEXICOGRAPHIC and maybe
11309 OEP_NO_HASH_CHECK for recursive call, to avoid exponential
11312 2018-03-28 Peter Bergner <bergner@vnet.ibm.com>
11315 * config/rs6000/rs6000.h (RS6000_BTM_POWERPC64): New define.
11316 (RS6000_BTM_COMMON): Add RS6000_BTM_POWERPC64.
11317 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add support
11318 for RS6000_BTM_POWERPC64.
11319 (rs6000_invalid_builtin): Add handling for RS6000_BTM_POWERPC64
11320 (rs6000_builtin_mask_names): Add RS6000_BTM_POWERPC64.
11321 * config/rs6000/rs6000-builtin.def (BU_P7_POWERPC64_MISC_2): New macro
11324 (DIVDEU): Likewise.
11326 2018-03-28 Carl Love <cel@us.ibm.com>
11329 2017-09-27 Carl Love <cel@us.ibm.com>
11331 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
11332 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
11333 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
11336 2018-03-28 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11338 * config/rs6000/xmmintrin.h (_mm_max_pi16): Use __vector __bool
11339 instead of __vector bool.
11340 (_mm_max_pu8): Likewise.
11341 (_mm_min_pi16): Likewise.
11343 2018-03-28 Peter Bergner <bergner@vnet.ibm.com>
11346 * config/rs6000/rs6000-builtin.def (DIVWEO): Delete macro expansion.
11347 (DIVWEUO): Likewise.
11348 (DIVDEO): Likewise.
11349 (DIVDEUO): Likewise.
11350 * config/rs6000/rs6000.c (builtin_function_type): Remove support for
11351 DIVWEUO and DIVDEUO.
11352 * config/rs6000/rs6000.md (UNSPEC_DIVEO, UNSPEC_DIVEUO): Delete unspecs.
11353 (UNSPEC_DIV_EXTEND): Remove deleted unspecs.
11354 (div_extend): Likewise.
11355 * doc/extend.texi (__builtin_divweo): Remove documentation for deleted
11357 (__builtin_divweuo): Likewise.
11358 (__builtin_divdeo): Likewise.
11359 (__builtin_divdeuo): Likewise.
11361 2018-03-28 Jakub Jelinek <jakub@redhat.com>
11364 * config/i386/i386.md (*add<mode>3_carry_0, *addsi3_carry_zext_0,
11365 *sub<mode>3_carry_0, *subsi3_carry_zext_0): New patterns.
11367 PR tree-optimization/82004
11368 * gimple-match-head.c (optimize_pow_to_exp): New function.
11369 * match.pd (pow(C,x) -> exp(log(C)*x)): Wrap with #if GIMPLE.
11370 Don't fold to exp if optimize_pow_to_exp is false.
11372 2018-03-28 Martin Liska <mliska@suse.cz>
11375 * calls.c (initialize_argument_information): Fix trailing space.
11376 * common.opt: Fix typo and provide better explanation for
11377 -fsanitize-coverage option.
11378 * config/i386/i386.opt: Fix typo.
11380 2018-03-28 Jakub Jelinek <jakub@redhat.com>
11381 Martin Liska <mliska@suse.cz>
11384 * gimplify.c (asan_poison_variable): Don't do the check for
11385 gimplify_omp_ctxp here.
11386 (gimplify_decl_expr): Do it here.
11387 (gimplify_target_expr): Likewise.
11389 2018-03-28 Martin Liska <mliska@suse.cz>
11392 * config/i386/i386.c (ix86_function_arg_advance): Do not call
11393 chkp_type_bounds_count if MPX is not enabled.
11395 2018-03-27 Chung-Ju Wu <jasonwucj@gmail.com>
11397 * config/nds32/nds32.h (BRANCH_COST): Adjust cost.
11399 2018-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
11402 * config/rs6000/rs6000.c (create_complex_muldiv): New helper
11403 function to create the function decl for complex long double
11404 multiply and divide for -mabi=ieeelongdouble.
11405 (init_float128_ieee): Call it.
11407 2018-03-27 H.J. Lu <hongjiu.lu@intel.com>
11410 * config/i386/i386.c (ix86_trampoline_init): Insert ENDBR for
11411 -fcf-protection=branch -mibt.
11412 * config/i386/i386.h (TRAMPOLINE_SIZE): Increased by 4 bytes.
11414 2018-03-27 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
11417 * config/arm/arm.c (arm_valid_symbolic_address): Handle
11418 arm_word_relocations.
11420 2018-03-27 Cesar Philippidis <cesar@codesourcery.com>
11423 * config/nvptx/nvptx.c (nvptx_assemble_decl_begin): Add '[]' to
11424 extern array declarations.
11426 2018-03-27 Richard Biener <rguenther@suse.de>
11428 PR middle-end/84067
11429 * match.pd ((A * C) +- (B * C) -> (A+-B) * C): Guard with
11430 explicit single_use checks.
11432 2018-03-27 Richard Biener <rguenther@suse.de>
11434 PR tree-optimization/85082
11435 * tree-ssa-sccvn.c (vn_reference_lookup_or_insert_for_pieces):
11438 2018-03-27 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
11440 * config.gcc (aarch64*-*-linux*): New TARGET_DEFAULT_ASYNC_UNWIND_TABLES
11441 * common/config/aarch64/aarch64-common.c (aarch64_optimization_table[]):
11442 Turn on fasynchronous-unwind-tables and funwind-tables.
11444 2018-03-26 Uros Bizjak <ubizjak@gmail.com>
11447 * config/i386/i386.md (*bmi_blsr_<mode>_cmp): New insn pattern.
11448 (*bmi_blsr_<mode>_ccz): Ditto.
11450 2018-03-26 Tom de Vries <tom@codesourcery.com>
11452 PR tree-optimization/85063
11453 * omp-general.c (offloading_function_p): New function. Factor out
11455 * omp-offload.c (pass_omp_target_link::gate): ... here.
11456 * omp-general.h (offloading_function_p): Declare.
11457 * tree-switch-conversion.c (build_one_array): Mark CSWTCH.x variable
11458 with attribute omp declare target for offloading functions.
11460 2018-03-24 Richard Sandiford <richard.sandiford@linaro.org>
11462 PR tree-optimization/84005
11463 * tree-data-ref.h (get_base_for_alignment): Declare.
11464 * tree-data-ref.c (get_base_for_alignment_1): New function.
11465 (get_base_for_alignment): Likewise.
11466 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
11467 get_base_for_alignment to find a suitable base object, instead
11468 of always using drb->base_address.
11470 2018-03-23 Jakub Jelinek <jakub@redhat.com>
11472 PR inline-asm/85022
11473 * emit-rtl.c (init_emit_regs): Indicate that VOIDmode MEMs don't have
11474 known size by default.
11476 2018-03-23 Vladimir Makarov <vmakarov@redhat.com>
11478 PR inline-asm/85030
11479 * lra-constraints.c (process_alt_operands): Don't match BLKmode
11480 and non BLKmode operands.
11482 2018-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11485 * config/arm/arm.md (unaligned_loadhis): Remove first alternative.
11486 Clean up attributes.
11488 2018-03-23 Richard Biener <rguenther@suse.de>
11491 * dwarf2out.c (rtl_for_decl_location): Do not generate RTL early when
11492 we are going to emit early debug for LTO.
11494 2018-03-23 Jakub Jelinek <jakub@redhat.com>
11496 PR inline-asm/85034
11497 * function.c (match_asm_constraints_1): Don't optimize if input
11498 doesn't satisfy general_operand predicate for output's mode.
11500 PR inline-asm/85022
11501 * alias.c (write_dependence_p): Don't require for x_canonicalized
11502 non-VOIDmode if x has VOIDmode.
11505 * sanopt.c (maybe_optimize_ubsan_ptr_ifn): If DECL_REGISTER (base),
11506 just don't try to optimize it rather than assert it never happens.
11508 2018-03-22 Kelvin Nilsen <kelvin@gcc.gnu.org>
11510 * config/rs6000/rs6000-builtin.def: Remove various BU_ALTIVEC_X
11511 macro expansions for definition of ST_INTERNAL_<mode> and
11512 LD_INTERNAL_<mode> builtins.
11513 * config/rs6000/rs6000-protos.h (rs6000_address_for_altivec):
11515 * config/rs6000/rs6000.c (altivec_expand_ld_builtin): Delete this
11517 (altivec_expand_st_builtin): Likewise.
11518 (altivec_expand_builtin): Remove calls to deleted functions.
11519 (rs6000_address_for_altivec): Delete this function.
11520 * config/rs6000/vector.md: Remove expands for
11521 vector_altivec_load_<mode> and vector_altivec_store_<mode>.
11523 2018-03-22 Sudakshina Das <sudi.das@arm.com>
11526 * config/arm/arm.h (machine_function): Add static_chain_stack_bytes.
11527 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Avoid
11528 re-computing once computed.
11529 (arm_expand_prologue): Compute machine->static_chain_stack_bytes.
11530 (arm_init_machine_status): Initialize
11531 machine->static_chain_stack_bytes.
11533 2018-03-22 Kelvin Nilsen <kelvin@gcc.gnu.org>
11536 * doc/extend.texi: Add four new prototypes for vec_ld.
11537 * config/rs6000/rs6000-builtin.def (LVX_V1TI): Reorder symbol
11538 definitions for more logical presentation.
11539 * config/rs6000/rs6000-c.c: (altivec_overloaded_builtins): Add
11540 entries for V1TI variants of __builtin_altivec_ld builtin.
11541 * config/rs6000/rs6000.c: (altivec_expand_lv_builtin): Add test and
11542 handling of V1TI variant of LVX icode pattern.
11543 (altivec_expand_builtin): Add case for ALTIVEC_BUILTIN_LVX_V1TI.
11544 (rs6000_gimple_fold_builtin): Likewise.
11545 (altivec_init_builtins): Add code to define
11546 __builtin_altivec_lvx_v1ti function.
11548 2018-03-22 Jakub Jelinek <jakub@redhat.com>
11550 PR inline-asm/84941
11551 * function.c (match_asm_constraints_1): Don't do the optimization
11552 if input isn't a REG, SUBREG, MEM or constant.
11554 2018-03-22 Tom de Vries <tom@codesourcery.com>
11556 PR tree-optimization/84956
11557 * tree-ssa-tail-merge.c (find_clusters_1): Skip bbs with
11558 bb_has_abnormal_pred.
11560 2018-03-22 Jakub Jelinek <jakub@redhat.com>
11563 * dwarf2asm.c (dw2_output_indirect_constant_1): Set
11564 DECL_INITIAL (decl) to decl at the end.
11565 * varasm.c (use_blocks_for_decl_p): Revert the 2018-03-20 change,
11566 adjust the comment.
11568 2018-03-21 Joseph Myers <joseph@codesourcery.com>
11570 * doc/extend.texi (__builtin_tgmath): Document when complex
11571 integer types are treated as _Complex _Float64.
11573 2018-03-21 Tom de Vries <tom@codesourcery.com>
11575 * doc/extend.texi (__builtin_extend_pointer): Remove pasto.
11577 2018-03-21 Jakub Jelinek <jakub@redhat.com>
11579 PR tree-optimization/84960
11580 * tree-cfg.c (remove_bb): Don't move forced labels into bb->prev_bb
11581 if it is ENTRY block, move them into single succ of ENTRY in that case.
11583 2018-03-21 Richard Sandiford <richard.sandiford@linaro.org>
11585 PR tree-optimization/84811
11586 * poly-int.h (poly_span_traits): Remove the T3 parameter and
11587 promote HOST_WIDE_INT T2 - T1 results to unsigned HOST_WIDE_INT.
11588 (maybe_in_range_p, known_in_range_p, ranges_known_overlap_p):
11589 (known_subrange_p): Update accordingly. Cast each value involved
11590 in the size comparison, rather than casting the result of the
11593 2018-03-21 Jakub Jelinek <jakub@redhat.com>
11595 PR tree-optimization/84982
11596 * gimple-ssa-store-merging.c (invert_op): Handle boolean inversion
11597 by flipping the least significant bit rather than all bits from
11598 bitpos to bitpos + bitsize - 1.
11600 2018-03-21 Nathan Sidwell <nathan@acm.org>
11602 * doc/extend.texi (Deprecated Features): Remove mention of
11603 long-deleted deprecations.
11605 2018-03-21 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
11608 * configure.ac (gcc_cv_ld_soname) <*-*-solaris2*>: Set.
11609 * configure: Regenerate.
11611 2018-03-21 Tom de Vries <tom@codesourcery.com>
11613 PR tree-optimization/83126
11614 * tree-parloops.c (num_phis): New function.
11615 (gen_parallel_loop): Detect and handle canonicalize_loop_ivs failure.
11617 2018-03-21 Nathan Sidwell <nathan@acm.org>
11619 * doc/extend.texi (Deprecated Features): Update deprecated flags,
11620 mention anon-struct/union members and trailing attributes.
11622 2018-03-21 Bin Cheng <bin.cheng@arm.com>
11624 PR tree-optimization/84969
11625 * tree-loop-distribution.c (fuse_memset_builtins): Don't reorder
11626 builtin memset partitions if they set different rhs values.
11628 2018-03-21 Jakub Jelinek <jakub@redhat.com>
11630 PR rtl-optimization/84989
11631 * simplify-rtx.c (simplify_unary_operation_1): Don't try to simplify
11632 VEC_DUPLICATE with scalar result mode.
11634 2018-03-21 Martin Liska <mliska@suse.cz>
11637 * ipa-icf.c (sem_item_optimizer::fixup_points_to_sets): Remove
11638 not intended return statement.
11640 2018-03-21 Martin Liska <mliska@suse.cz>
11643 * tree-chkp.c (CHKP_ARRAY_MAX_CHECK_STEPS): Define a new macro.
11644 (chkp_find_bound_slots_1): Limit number of iterations.
11646 2018-03-20 David H. Gutteridge <dhgutteridge@sympatico.ca>
11649 * Minor grammar fixes for x86 options.
11651 2018-03-20 Jakub Jelinek <jakub@redhat.com>
11654 * dce.c (delete_unmarked_insns): Don't remove frame related noop moves
11655 holding REG_CFA_RESTORE notes, instead turn them into a USE.
11657 2018-03-20 Peter Bergner <bergner@vnet.ibm.com>
11660 * config/rs6000/altivec.md (altivec_lvx_<mode>_2op): Delete define_insn.
11661 (altivec_lvx_<mode>_1op): Likewise.
11662 (altivec_stvx_<mode>_2op): Likewise.
11663 (altivec_stvx_<mode>_1op): Likewise.
11664 (altivec_lvx_<VM2:mode>): New define_expand.
11665 (altivec_stvx_<VM2:mode>): Likewise.
11666 (altivec_lvx_<VM2:mode>_2op_<P:mptrsize>): New define_insn.
11667 (altivec_lvx_<VM2:mode>_1op_<P:mptrsize>): Likewise.
11668 (altivec_stvx_<VM2:mode>_2op_<P:mptrsize>): Likewise.
11669 (altivec_stvx_<VM2:mode>_1op_<P:mptrsize>): Likewise.
11670 * config/rs6000/rs6000-p8swap.c (rs6000_gen_stvx): Use new expanders.
11671 (rs6000_gen_lvx): Likewise.
11672 * config/rs6000/rs6000.c (altivec_expand_lv_builtin): Likewise.
11673 (altivec_expand_stv_builtin): Likewise.
11674 (altivec_expand_builtin): Likewise.
11675 * config/rs6000/vector.md: Likewise.
11677 2018-03-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11680 * config/arm/arm.c (arm_array_mode_supported_p): Return false for
11683 2018-03-20 Richard Biener <rguenther@suse.de>
11686 * config/i386/i386.c (ix86_add_stmt_cost): Only cost
11687 sign-conversions as zero, fall back to standard scalar_stmt
11690 2018-03-20 Martin Liska <mliska@suse.cz>
11693 * predict.c (rebuild_frequencies): Handle case when we have
11694 PROFILE_ABSENT, but flag_guess_branch_prob is false.
11696 2018-03-20 Jakub Jelinek <jakub@redhat.com>
11699 * dwarf2asm.c (dw2_output_indirect_constant_1): Temporarily turn off
11700 flag_section_anchors.
11701 * varasm.c (use_blocks_for_decl_p): Remove hack for
11702 dw2_force_const_mem.
11705 * config/aarch64/aarch64.md (*aarch64_reg_<mode>3_neg_mask2): Rename
11707 (*aarch64_<optab>_reg_<mode>3_neg_mask2): ... this. If pseudos can't
11708 be created, use lowpart_subreg of operands[0] rather than operands[0]
11710 (*aarch64_reg_<mode>3_minus_mask): Rename to ...
11711 (*aarch64_ashl_reg_<mode>3_minus_mask): ... this.
11712 (*aarch64_<optab>_reg_di3_mask2): Use const_int_operand predicate
11713 and n constraint instead of aarch64_shift_imm_di and Usd.
11714 (*aarch64_reg_<optab>_minus<mode>3): Rename to ...
11715 (*aarch64_<optab>_reg_minus<mode>3): ... this.
11717 2018-03-20 Sudakshina Das <sudi.das@arm.com>
11720 * config/arm/neon.md (ashldi3_neon): Update ?s for constraints
11721 to favor GPR over NEON registers.
11722 (<shift>di3_neon): Likewise.
11724 2018-03-20 Tom de Vries <tom@codesourcery.com>
11727 * config/nvptx/nvptx.c (nvptx_single): Don't neuter bar.sync.
11728 (nvptx_process_pars): Emit bar.sync asap and alap.
11730 2018-03-20 Tom de Vries <tom@codesourcery.com>
11733 * config/nvptx/nvptx.c (prevent_branch_around_nothing): Also update
11734 seen_label if seen_label is already set.
11736 2018-03-20 Jakub Jelinek <jakub@redhat.com>
11739 * config/i386/i386.c (fold_builtin_cpu): For features above 31
11740 use __cpu_features2 variable instead of __cpu_model.__cpu_features[0].
11741 Use 1U instead of 1. Formatting fixes.
11744 * builtins.c (fold_builtin_strpbrk): For strpbrk(x, "") use type
11745 instead of TREE_TYPE (s1) for the return value.
11747 2018-03-19 Jakub Jelinek <jakub@redhat.com>
11749 PR tree-optimization/84946
11750 * gimple-ssa-store-merging.c (mem_valid_for_store_merging): Compute
11751 bitsize + bitsize in poly_uint64 rather than poly_int64.
11754 * dwarf2asm.c: Include fold-const.c.
11755 (dw2_output_indirect_constant_1): Set DECL_INITIAL (decl) to ADDR_EXPR
11756 of decl rather than decl itself.
11758 PR rtl-optimization/84643
11759 * memmodel.h (enum memmodel): Add MEMMODEL_MAX enumerator.
11761 2018-03-19 Maxim Ostapenko <m.ostapenko@samsung.com>
11764 * dwarf2asm.c (dw2_output_indirect_constant_1): Disable ASan before
11765 calling assemble_variable.
11767 2018-03-19 Sudakshina Das <sudi.das@arm.com>
11770 * config/aarch64/aarch64-simd.md (vec_cmp<mode><v_int_equiv>): Modify
11771 instructions for UNLT, UNLE, UNGT, UNGE, UNEQ, UNORDERED and ORDERED.
11773 2018-03-19 Jim Wilson <jimw@sifive.com>
11776 * config/riscv/riscv.c (riscv_function_arg_boundary): Use
11777 PREFERRED_STACK_BOUNDARY instead of STACK_BOUNDARY.
11778 (riscv_first_stack_step): Likewise.
11779 (riscv_option_override): Use STACK_BOUNDARY instead of
11780 MIN_STACK_BOUNDARY.
11781 * config/riscv/riscv.h (STACK_BOUNDARY): Renamed from
11782 MIN_STACK_BOUNDARY.
11783 (BIGGEST_ALIGNMENT): Set to 128.
11784 (PREFERRED_STACK_BOUNDARY): Renamed from STACK_BOUNDARY.
11785 (RISCV_STACK_ALIGN): Use PREFERRED_STACK_BOUNDARY instead of
11788 2018-03-19 Richard Biener <rguenther@suse.de>
11790 PR tree-optimization/84933
11791 * tree-vrp.c (set_and_canonicalize_value_range): Treat out-of-bound
11792 values as -INF/INF when canonicalizing an ANTI_RANGE to a RANGE.
11794 2018-03-19 Richard Biener <rguenther@suse.de>
11796 PR tree-optimization/84859
11797 * tree-ssa-phiopt.c (single_trailing_store_in_bb): New function.
11798 (cond_if_else_store_replacement): Perform sinking operation on
11799 single-store BBs regardless of MAX_STORES_TO_SINK setting.
11800 Generalize what a BB with a single eligible store is.
11802 2018-03-19 Richard Biener <rguenther@suse.de>
11804 PR tree-optimization/84929
11805 * tree-data-ref.c (analyze_siv_subscript_cst_affine): Guard
11806 chrec_is_positive against non-chrec arg.
11808 2018-03-19 Tamar Christina <tamar.christina@arm.com>
11811 * config/arm/arm.c (arm_can_change_mode_class): revert r258554.
11813 2018-03-18 Martin Liska <mliska@suse.cz>
11815 PR rtl-optimization/84635
11816 * regrename.c (build_def_use): Use matches_mode only when
11819 2018-03-18 Richard Sandiford <richard.sandiford@linaro.org>
11821 PR tree-optimization/84913
11822 * tree-vect-loop.c (vectorizable_reduction): Don't try to
11823 vectorize chains of COND_EXPRs.
11825 2018-03-18 Chung-Ju Wu <jasonwucj@gmail.com>
11827 * config/nds32/nds32.h (MAX_REGS_PER_ADDRESS): Fix the value.
11829 2018-03-18 Chung-Ju Wu <jasonwucj@gmail.com>
11831 * config/nds32/nds32.h (LOGICAL_OP_NON_SHORT_CIRCUIT): Define.
11833 2018-03-18 Chung-Ju Wu <jasonwucj@gmail.com>
11835 * config/nds32/nds32.h (CLZ_DEFINED_VALUE_AT_ZERO): Define.
11837 2018-03-17 Chung-Ju Wu <jasonwucj@gmail.com>
11838 Kito Cheng <kito.cheng@gmail.com>
11840 * config/nds32/nds32-protos.h (nds32_adjust_reg_alloc_order): Declare.
11841 * config/nds32/nds32.c (nds32_reg_alloc_order_for_speed): New array.
11842 (nds32_adjust_reg_alloc_order): New function.
11843 * config/nds32/nds32.h (ADJUST_REG_ALLOC_ORDER): Define.
11845 2018-03-17 Kito Cheng <kito.cheng@gmail.com>
11847 * config/nds32/nds32.c (nds32_asm_output_mi_thunk,
11848 nds32_print_operand, nds32_print_operand_address): Use
11849 HOST_WIDE_INT_PRINT_DEC instead.
11851 2018-03-17 Chung-Ju Wu <jasonwucj@gmail.com>
11853 * config/nds32/nds32.c (nds32_register_priority): Modify cost.
11855 2018-03-17 Jakub Jelinek <jakub@redhat.com>
11858 * config/i386/i386.c (initial_ix86_tune_features,
11859 initial_ix86_arch_features): Use unsigned HOST_WIDE_INT rather than
11860 unsigned long long.
11861 (set_ix86_tune_features): Change ix86_tune_mask from unsigned int
11862 to unsigned HOST_WIDE_INT, initialize to HOST_WIDE_INT_1U << ix86_tune
11863 rather than 1u << ix86_tune. Formatting fix.
11864 (ix86_option_override_internal): Change ix86_arch_mask from
11865 unsigned int to unsigned HOST_WIDE_INT, initialize to
11866 HOST_WIDE_INT_1U << ix86_arch rather than 1u << ix86_arch.
11867 (ix86_function_specific_restore): Likewise.
11869 2018-03-16 Jakub Jelinek <jakub@redhat.com>
11872 * postreload.c (reload_combine_recognize_pattern): Perform
11873 INTVAL addition in unsigned HOST_WIDE_INT type to avoid UB and
11874 truncate_int_for_mode the result for the destination's mode.
11877 * hsa-gen.c (mem_type_for_type): Fix comment typo.
11878 * tree-vect-loop-manip.c (vect_create_cond_for_niters_checks):
11880 * gimple-ssa-warn-restrict.c (builtin_memref::set_base_and_offset):
11883 2018-03-16 Vladimir Makarov <vmakarov@redhat.com>
11886 * lra-assigns.c (lra_split_hard_reg_for): Don't use
11887 regno_allocno_class_array and sorted_pseudos.
11888 * lra-constraints.c (spill_hard_reg_in_range): Ignore hard regs in
11889 insns where regno is used.
11891 2018-03-16 Martin Liska <mliska@suse.cz>
11894 * multiple_target.c (create_dispatcher_calls): Redirect
11895 reference in the symbol table.
11897 2018-03-16 Martin Liska <mliska@suse.cz>
11900 * multiple_target.c (create_dispatcher_calls): Redirect also
11903 2018-03-16 Jakub Jelinek <jakub@redhat.com>
11907 * tree.h (TARGET_EXPR_NO_ELIDE): Define.
11908 * gimplify.c (gimplify_modify_expr_rhs): Don't elide TARGET_EXPRs with
11909 TARGET_EXPR_NO_ELIDE flag set unless *expr_p is INIT_EXPR.
11911 2018-03-16 Julia Koval <julia.koval@intel.com>
11913 * doc/invoke.texi (Skylake Server): Add CLWB.
11914 Cannonlake): Remove CLWB.
11916 2018-03-16 Jakub Jelinek <jakub@redhat.com>
11918 PR tree-optimization/84841
11919 * tree-ssa-reassoc.c (INTEGER_CONST_TYPE): Change to 1 << 4 from
11921 (FLOAT_ONE_CONST_TYPE): Define.
11922 (constant_type): Return FLOAT_ONE_CONST_TYPE for -1.0 and 1.0.
11923 (sort_by_operand_rank): Put entries with higher constant_type last
11924 rather than first to match comments.
11926 2018-03-15 Sandra Loosemore <sandra@codesourcery.com>
11928 * config/nios2/nios2.md (movsi_internal): Fix thinko in
11931 2018-03-15 Jakub Jelinek <jakub@redhat.com>
11934 * calls.c (expand_call): For TREE_ADDRESSABLE rettype ignore alignment
11935 check and use address of target always.
11937 2018-03-15 H.J. Lu <hongjiu.lu@intel.com>
11940 * config/i386/i386.c (indirect_thunk_needed): Update comments.
11941 (indirect_thunk_bnd_needed): Likewise.
11942 (indirect_thunks_used): Likewise.
11943 (indirect_thunks_bnd_used): Likewise.
11944 (indirect_return_needed): New.
11945 (indirect_return_bnd_needed): Likewise.
11946 (output_indirect_thunk_function): Add a bool argument for
11948 (output_indirect_thunk_function): Don't generate alias for
11949 function return thunk.
11950 (ix86_code_end): Call output_indirect_thunk_function to generate
11951 function return thunks.
11952 (ix86_output_function_return): Set indirect_return_bnd_needed
11953 and indirect_return_needed instead of indirect_thunk_bnd_needed
11954 and indirect_thunk_needed.
11956 2018-03-15 Olga Makhotina <olga.makhotina@intel.com>
11958 * config/i386/sgxintrin.h (_enclv_u32): New intrinsic.
11959 (__enclv_bc, __enclv_cd, __enclv_generic): New definitions.
11960 (ERDINFO, ETRACKC, ELDBC, ELDUC): New leaves.
11962 2018-03-15 David Malcolm <dmalcolm@redhat.com>
11963 Paul Hua <paul.hua.gm@gmail.com>
11966 * gcc.dg/fixits-pr84852-1.c: Fix filename in dg-regexp.
11968 2018-03-15 Segher Boessenkool <segher@kernel.crashing.org>
11970 * config/rs6000/rs6000.c (abi_v4_pass_in_fpr): Add back the
11971 TARGET_DOUBLE_FLOAT and TARGET_SINGLE_FLOAT conditions on the DFmode
11972 resp. SFmode cases.
11974 2018-03-15 Tamar Christina <tamar.christina@arm.com>
11977 * config/arm/arm.c (arm_can_change_mode_class): Use GET_MODE_UNIT_SIZE
11978 instead of GET_MODE_SIZE when comparing Units.
11980 2018-03-15 Vladimir Mezentsev <vladimir.mezentsev@oracle.com>
11983 * varasm.c (hash_section): Return an unchangeble hash value
11984 * config/aarch64/aarch64.c (aarch64_use_blocks_for_constant_p):
11985 Return !aarch64_can_use_per_function_literal_pools_p ().
11987 2018-03-15 Jakub Jelinek <jakub@redhat.com>
11990 * optabs.c (emit_conditional_move): Pass address of cmode's copy
11991 rather than address of cmode as last argument to prepare_cmp_insn.
11993 2018-03-15 Julia Koval <julia.koval@intel.com>
11995 * config/i386/i386.c (F_AVX512VBMI2, F_GFNI, F_VPCLMULQDQ,
11996 F_AVX512VNNI, F_AVX512BITALG): New.
11998 2018-03-14 John David Anglin <danglin@gcc.gnu.org>
12001 * config/pa/pa.c (pa_emit_move_sequence): Always emit secondary reload
12002 insn for floating-point loads and stores.
12004 2018-03-14 Carl Love <cel@us.ibm.com>
12006 * config/rs6000/rs6000-c.c: Add macro definitions for
12007 ALTIVEC_BUILTIN_VEC_PERMXOR.
12008 * config/rs6000/rs6000.h: Add #define for vec_permxor builtin.
12009 * config/rs6000/rs6000-builtin.def: Add macro expansions for VPERMXOR.
12010 * config/rs6000/altivec.md (altivec_vpermxor): New define expand.
12011 * config/rs6000/rs6000-p8swap.c (rtx_is_swappable_p): Add case
12013 * config/doc/extend.texi: Add prototypes for vec_permxor.
12015 2018-03-14 David Malcolm <dmalcolm@redhat.com>
12018 * diagnostic-show-locus.c (class layout_point): Convert m_line
12019 from int to linenum_type.
12020 (line_span::comparator): Use linenum "compare" function when
12021 comparing line numbers.
12022 (test_line_span): New function.
12023 (layout_range::contains_point): Convert param "row" from int to
12025 (layout_range::intersects_line_p): Likewise.
12026 (layout::will_show_line_p): Likewise.
12027 (layout::print_source_line): Likewise.
12028 (layout::should_print_annotation_line_p): Likewise.
12029 (layout::print_annotation_line): Likewise.
12030 (layout::print_leading_fixits): Likewise.
12031 (layout::annotation_line_showed_range_p): Likewise.
12032 (struct line_corrections): Likewise for field m_row.
12033 (line_corrections::line_corrections): Likewise for param "row".
12034 (layout::print_trailing_fixits): Likewise.
12035 (layout::get_state_at_point): Likewise.
12036 (layout::get_x_bound_for_row): Likewise.
12037 (layout::print_line): Likewise.
12038 (diagnostic_show_locus): Likewise for locals "last_line" and "row".
12039 (selftest::diagnostic_show_locus_c_tests): Call test_line_span.
12040 * input.c (selftest::test_linenum_comparisons): New function.
12041 (selftest::input_c_tests): Call it.
12042 * selftest.c (selftest::test_assertions): Test ASSERT_GT,
12043 ASSERT_GT_AT, ASSERT_LT, and ASSERT_LT_AT.
12044 * selftest.h (ASSERT_GT): New macro.
12045 (ASSERT_GT_AT): New macro.
12046 (ASSERT_LT): New macro.
12047 (ASSERT_LT_AT): New macro.
12049 2018-03-14 Segher Boessenkool <segher@kernel.crashing.org>
12051 PR rtl-optimization/84780
12052 * combine.c (distribute_links): Don't make a link based on pc_rtx.
12054 2018-03-14 Martin Liska <mliska@suse.cz>
12056 * tree.c (record_node_allocation_statistics): Use
12057 get_stats_node_kind.
12058 (get_stats_node_kind): New function extracted from
12059 record_node_allocation_statistics.
12060 (free_node): Use get_stats_node_kind.
12062 2018-03-14 Richard Biener <rguenther@suse.de>
12064 * tree-ssa-pre.c (compute_antic_aux): Remove code that asserts
12065 that the value-set of ANTIC_IN doesn't grow.
12068 * tree-ssa-pre.c (struct bb_bitmap_sets): Add visited_with_visited_succs
12070 (BB_VISITED_WITH_VISITED_SUCCS): New define.
12071 (compute_antic): Initialize BB_VISITED_WITH_VISITED_SUCCS.
12073 2018-03-14 Julia Koval <julia.koval@intel.com>
12075 * config.gcc (icelake-client, icelake-server): New.
12077 * config/i386/i386.c (initial_ix86_tune_features): Extend to 64 bit.
12078 (initial_ix86_arch_features): Ditto.
12079 (PTA_SKYLAKE): Add SGX.
12080 (PTA_ICELAKE): Remove.
12081 (PTA_ICELAKE_CLIENT): New.
12082 (PTA_ICELAKE_SERVER): New.
12083 (ix86_option_override_internal): Split up icelake on icelake client and
12085 (get_builtin_code_for_version): Ditto.
12086 (fold_builtin_cpu): Ditto.
12087 * config/i386/driver-i386.c (config/i386/driver-i386.c): Ditto.
12088 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto
12089 * config/i386/i386.h (processor_type): Ditto.
12090 * doc/invoke.texi: Ditto.
12092 2018-03-14 Jakub Jelinek <jakub@redhat.com>
12095 * sanopt.c (maybe_optimize_ubsan_ptr_ifn): Handle also
12096 INTEGER_CST offset, add it together with bitpos / 8 and
12097 sign extend based on POINTER_SIZE.
12101 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
12104 * config/i386/constraints.md (Yc): New register constraint.
12105 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
12106 Use Yc constraint for alternative 2 of operand 0. Remove
12107 preferred_for_speed attribute.
12109 2018-03-14 Richard Biener <rguenther@suse.de>
12111 PR tree-optimization/84830
12112 * tree-ssa-pre.c (compute_antic_aux): Intersect the new ANTIC_IN
12113 with the old one to avoid oscillations.
12115 2018-03-13 Vladimir Makarov <vmakarov@redhat.com>
12118 * lra-assigns.c (find_all_spills_for): Ignore uninteresting
12120 (assign_by_spills): Return a flag of reload assignment failure.
12121 Do not process the reload assignment failures. Do not spill other
12122 reload pseudos if they has the same reg class. Update n if
12124 (lra_assign): Add a return arg. Set up from the result of
12125 assign_by_spills call.
12126 (find_reload_regno_insns, lra_split_hard_reg_for): New functions.
12127 * lra-constraints.c (split_reg): Add a new arg. Use it instead of
12128 usage_insns if it is not NULL.
12129 (spill_hard_reg_in_range): New function.
12130 (split_if_necessary, inherit_in_ebb): Pass a new arg to split_reg.
12131 * lra-int.h (spill_hard_reg_in_range, lra_split_hard_reg_for): New
12132 function prototypes.
12133 (lra_assign): Change prototype.
12134 * lra.c (lra): Add code to deal with fails by splitting hard reg
12137 2018-03-01 Palmer Dabbelt <palmer@sifive.com>
12139 * config/riscv/riscv.opt (mrelax): New option.
12140 * config/riscv/riscv.c (riscv_file_start): Emit ".option
12141 "norelax" when riscv_mrelax is disabled.
12142 * doc/invoke.texi (RISC-V): Document "-mrelax" and "-mno-relax".
12144 2018-03-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
12147 * config/rs6000/rs6000.c (rs6000_reassociation_width): Disable parallel
12148 reassociation for int modes.
12150 2018-03-13 Richard Sandiford <richard.sandiford@linaro.org>
12152 * tree-vect-loop-manip.c (vect_maybe_permute_loop_masks):
12153 Reverse the choice between VEC_UNPACK_LO_EXPR and VEC_UNPACK_HI_EXPR
12155 * config/aarch64/iterators.md (hi_lanes_optab): New int attribute.
12156 * config/aarch64/aarch64-sve.md
12157 (*aarch64_sve_<perm_insn><perm_hilo><mode>): Rename to...
12158 (aarch64_sve_<perm_insn><perm_hilo><mode>): ...this.
12159 (*extend<mode><Vwide>2): Rename to...
12160 (aarch64_sve_extend<mode><Vwide>2): ...this.
12161 (vec_unpack<su>_<perm_hilo>_<mode>): Turn into a define_expand,
12162 renaming the old pattern to...
12163 (aarch64_sve_punpk<perm_hilo>_<mode>): ...this. Only define
12165 (vec_unpack<su>_<perm_hilo>_<SVE_BHSI:mode>): Turn into a
12166 define_expand, renaming the old pattern to...
12167 (aarch64_sve_<su>unpk<perm_hilo>_<SVE_BHSI:mode>): ...this.
12168 (*vec_unpacku_<perm_hilo>_<mode>_no_convert): Delete.
12169 (vec_unpacks_<perm_hilo>_<mode>): Take BYTES_BIG_ENDIAN into
12170 account when deciding which SVE instruction the optab should use.
12171 (vec_unpack<su_optab>_float_<perm_hilo>_vnx4si): Likewise.
12173 2018-03-13 Richard Sandiford <richard.sandiford@linaro.org>
12175 * config/aarch64/aarch64.md (V4_REGNUM, V8_REGNUM, V12_REGNUM)
12176 (V20_REGNUM, V24_REGNUM, V28_REGNUM, P1_REGNUM, P2_REGNUM, P3_REGNUM)
12177 (P4_REGNUM, P5_REGNUM, P6_REGNUM, P8_REGNUM, P9_REGNUM, P10_REGNUM)
12178 (P11_REGNUM, P12_REGNUM, P13_REGNUM, P14_REGNUM): New define_constants.
12179 (tlsdesc_small_<mode>): Turn a define_expand and use
12180 tlsdesc_small_sve_<mode> for SVE. Rename original define_insn to...
12181 (tlsdesc_small_advsimd_<mode>): ...this.
12182 (tlsdesc_small_sve_<mode>): New pattern.
12184 2018-03-13 Richard Sandiford <richard.sandiford@linaro.org>
12186 * config/aarch64/iterators.md (UNSPEC_SMUL_HIGHPART)
12187 (UNSPEC_UMUL_HIGHPART): New constants.
12188 (MUL_HIGHPART): New int iteraor.
12189 (su): Handle UNSPEC_SMUL_HIGHPART and UNSPEC_UMUL_HIGHPART.
12190 * config/aarch64/aarch64-sve.md (<su>mul<mode>3_highpart): New
12192 (*<su>mul<mode>3_highpart): New define_insn.
12194 2018-03-13 Eric Botcazou <ebotcazou@adacore.com>
12197 * ipa-devirt.c (odr_subtypes_equivalent_p): Do not get the ODR type of
12200 2018-03-13 Martin Liska <mliska@suse.cz>
12203 * (sem_item_optimizer::sem_item_optimizer): Initialize new
12205 (sem_item_optimizer::~sem_item_optimizer): Release it.
12206 (sem_item_optimizer::merge_classes): Register variable aliases.
12207 (sem_item_optimizer::fixup_pt_set): New function.
12208 (sem_item_optimizer::fixup_points_to_sets): Likewise.
12209 * ipa-icf.h: Declare new variables and functions.
12211 2018-03-13 Jakub Jelinek <jakub@redhat.com>
12213 PR middle-end/84834
12214 * match.pd ((A & C) != 0 ? D : 0): Use INTEGER_CST@2 instead of
12215 integer_pow2p@2 and test integer_pow2p in condition.
12216 (A < 0 ? C : 0): Similarly for @1.
12218 PR middle-end/84831
12219 * stmt.c (parse_output_constraint): If the CONSTRAINT_LEN (*p, p)
12220 characters starting at p contain '\0' character, don't look beyond
12224 * config/i386/i386.md (round<mode>2): For 387 fancy math, disable
12225 pattern if -ftrapping-math -fno-fp-int-builtin-inexact.
12228 * reg-stack.c (change_stack): Change update_end var from int to
12229 rtx_insn *, if non-NULL don't update just BB_END (current_block), but
12230 also call set_block_for_insn on the newly added insns and rescan.
12233 * config/i386/sse.md (sse2_loadhpd): Use Yv constraint rather than v
12234 on the last operand.
12237 * tree.c (stabilize_reference_1): Return save_expr (e) for
12238 STATEMENT_LIST even if it doesn't have side-effects.
12240 2018-03-12 Jonathan Wakely <jwakely@redhat.com>
12242 * doc/invoke.texi (-mclflushopt): Fix spelling of option.
12244 2018-03-12 Renlin Li <renlin.li@arm.com>
12246 * config/aarch64/aarch64.md (movhf_aarch64): Fix mode argument to
12247 aarch64_output_scalar_simd_mov_immediate.
12249 2018-03-12 Martin Sebor <msebor@redhat.com>
12251 PR tree-optimization/83456
12252 * gimple-fold.c (gimple_fold_builtin_memory_op): Avoid warning
12253 for perfectly overlapping calls to memcpy.
12254 (gimple_fold_builtin_memory_chk): Same.
12255 (gimple_fold_builtin_strcpy): Handle no-warning.
12256 (gimple_fold_builtin_stxcpy_chk): Same.
12257 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Handle no-warning.
12259 2018-03-12 Segher Boessenkool <segher@kernel.crashing.org>
12261 * config/rs6000/rs6000.c (abi_v4_pass_in_fpr): Add bool "named"
12262 parameter. Use it for SFmode.
12263 (rs6000_function_arg_advance_1): Adjust.
12264 (rs6000_function_arg): Adjust.
12265 (rs6000_gimplify_va_arg): Pass false for that new parameter.
12267 2018-03-12 Segher Boessenkool <segher@kernel.crashing.org>
12269 PR rtl-optimization/84169
12270 PR rtl-optimization/84780
12271 * combine.c (can_combine_p): Check for a 2-insn combination whether
12272 the destination register is used between the two insns, too.
12274 2018-03-12 Richard Biener <rguenther@suse.de>
12276 PR tree-optimization/84803
12277 * tree-if-conv.c (ifcvt_memrefs_wont_trap): Don't do anything
12278 for refs DR analysis didn't process.
12280 2018-03-12 Richard Biener <rguenther@suse.de>
12282 PR tree-optimization/84777
12283 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): For
12284 force-vectorize loops ignore whether we are optimizing for size.
12286 2018-03-12 Chung-Ju Wu <jasonwucj@gmail.com>
12288 * config/nds32/nds32.c (nds32_md_asm_adjust): New function.
12289 (TARGET_MD_ASM_ADJUST): Define.
12291 2018-03-12 Monk Chiang <sh.chiang04@gmail.com>
12292 Kito Cheng <kito.cheng@gmail.com>
12293 Chung-Ju Wu <jasonwucj@gmail.com>
12295 * config/nds32/nds32.c (nds32_compute_stack_frame,
12296 nds32_emit_stack_push_multiple, nds32_emit_stack_pop_multiple,
12297 nds32_emit_stack_v3push, nds32_emit_stack_v3pop,
12298 nds32_emit_adjust_frame, nds32_expand_prologue, nds32_expand_epilogue,
12299 nds32_expand_prologue_v3push, nds32_expand_epilogue_v3pop): Refine.
12300 * config/nds32/nds32.h (NDS32_FIRST_CALLEE_SAVE_GPR_REGNUM,
12301 NDS32_LAST_CALLEE_SAVE_GPR_REGNUM, NDS32_V3PUSH_AVAILABLE_P): New.
12302 * config/nds32/nds32.md (prologue, epilogue): Use macro
12303 NDS32_V3PUSH_AVAILABLE_P to do checking.
12305 2018-03-11 Jakub Jelinek <jakub@redhat.com>
12308 * dwarf2out.c (gen_enumeration_type_die): Don't guard adding
12309 DW_AT_declaration for ENUM_IS_OPAQUE on -gdwarf-4 or -gno-strict-dwarf,
12310 but on TYPE_SIZE. Don't do anything for ENUM_IS_OPAQUE if not creating
12311 a new die. Don't set TREE_ASM_WRITTEN if ENUM_IS_OPAQUE. Guard
12312 addition of most attributes on !orig_type_die or the attribute not
12313 being present already. Assert TYPE_VALUES is NULL for ENUM_IS_OPAQUE.
12315 2018-03-11 Kito Cheng <kito.cheng@gmail.com>
12316 Chung-Ju Wu <jasonwucj@gmail.com>
12318 * config/nds32/nds32.c (nds32_cpu_cpp_builtins): Modify to define
12319 __NDS32_VH__ macro.
12320 * config/nds32/nds32.opt (mvh): New option.
12322 2018-03-11 Kito Cheng <kito.cheng@gmail.com>
12323 Chung-Ju Wu <jasonwucj@gmail.com>
12325 * config/nds32/nds32-protos.h (nds32_cpu_cpp_builtins): Declare
12327 * config/nds32/nds32.c (nds32_cpu_cpp_builtins): New function.
12328 * config/nds32/nds32.h (TARGET_CPU_CPP_BUILTINS): Modify its
12331 2018-03-11 Kito Cheng <kito.cheng@gmail.com>
12332 Chung-Ju Wu <jasonwucj@gmail.com>
12334 * config/nds32/nds32-memory-manipulation.c (nds32_expand_strlen): New
12336 * config/nds32/nds32-multiple.md (strlensi): New pattern.
12337 * config/nds32/nds32-protos.h (nds32_expand_strlen): Declare function.
12339 2018-03-11 Monk Chiang <sh.chiang04@gmail.com>
12340 Kito Cheng <kito.cheng@gmail.com>
12341 Chung-Ju Wu <jasonwucj@gmail.com>
12343 * config/nds32/constants.md (unspec_element): Add UNSPEC_FFB,
12344 UNSPEC_FFMISM and UNSPEC_FLMISM.
12345 * config/nds32/nds32-intrinsic.c (bdesc_2arg): Add builtin description
12346 for ffb, ffmism and flmism.
12347 * config/nds32/nds32-intrinsic.md (unspec_ffb): Define new pattern.
12348 (unspec_ffmism): Ditto.
12349 (unspec_flmism): Ditto.
12350 (nds32_expand_builtin_impl): Check if string extension is available.
12351 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_FFB,
12352 NDS32_BUILTIN_FFMISM and NDS32_BUILTIN_FLMISM.
12354 2018-03-10 Vladimir Makarov <vmakarov@redhat.com>
12357 2018-03-09 Vladimir Makarov <vmakarov@redhat.com>
12360 * lra-assigns.c (assign_by_spills): Return a flag of reload
12361 assignment failure. Do not process the reload assignment
12362 failures. Do not spill other reload pseudos if they has the same
12364 (lra_assign): Add a return arg. Set up from the result of
12365 assign_by_spills call.
12366 (find_reload_regno_insns, lra_split_hard_reg_for): New functions.
12367 * lra-constraints.c (split_reg): Add a new arg. Use it instead of
12368 usage_insns if it is not NULL.
12369 (spill_hard_reg_in_range): New function.
12370 (split_if_necessary, inherit_in_ebb): Pass a new arg to split_reg.
12371 * lra-int.h (spill_hard_reg_in_range, lra_split_hard_reg_for): New
12372 function prototypes.
12373 (lra_assign): Change prototype.
12374 * lra.c (lra): Add code to deal with fails by splitting hard reg
12377 2018-03-10 H.J. Lu <hongjiu.lu@intel.com>
12380 * config/i386/i386.opt: Replace Enforcment with Enforcement.
12382 2018-03-10 Alexandre Oliva <aoliva@redhat.com>
12385 * dwarf2out.h (dw_val_class): Add dw_val_class_symview.
12386 (dw_val_node): Add val_symbolic_view.
12387 * dwarf2out.c (dw_line_info_table): Add symviews_since_reset.
12388 (symview_upper_bound): New.
12389 (new_line_info_table): Initialize symviews_since_reset.
12390 (dwarf2out_source_line): Count symviews_since_reset and set
12391 symview_upper_bound.
12392 (dw_val_equal_p): Handle symview.
12393 (add_AT_symview): New.
12394 (print_dw_val): Handle symview.
12395 (attr_checksum, attr_checksum_ordered): Likewise.
12396 (same_dw_val_p, size_of_die): Likewise.
12397 (value_format, output_die): Likewise.
12398 (add_high_low_attributes): Use add_AT_symview for entry_view.
12399 (dwarf2out_finish): Reset symview_upper_bound, clear
12402 2018-03-09 Peter Bergner <bergner@vnet.ibm.com>
12405 * config/rs6000/rs6000.c (rs6000_offsettable_memref_p): New prototype.
12406 Add strict argument and use it.
12407 (rs6000_split_multireg_move): Update for new strict argument.
12408 (mem_operand_gpr): Disallow all non-offsettable addresses.
12409 * config/rs6000/rs6000.md (*movdi_internal64): Use YZ constraint.
12411 2018-03-09 Jakub Jelinek <jakub@redhat.com>
12414 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Mark va_arg_tmp
12415 temporary TREE_ADDRESSABLE before gimplification of BUILT_IN_MEMCPY.
12416 * config/powerpcspe/powerpcspe.c (rs6000_gimplify_va_arg): Likewise.
12419 * tree-inline.c (copy_tree_body_r): For INDIRECT_REF of a remapped
12420 decl, use remap_type if we want to use the type.
12422 2018-03-09 Martin Sebor <msebor@redhat.com>
12424 PR tree-optimization/84526
12425 * gimple-ssa-warn-restrict.c (builtin_memref::set_base_and_offset):
12427 (builtin_access::generic_overlap): Be prepared to handle non-array
12430 2018-03-09 Alexandre Oliva <aoliva@redhat.com>
12432 PR rtl-optimization/84682
12433 * lra-constraints.c (process_address_1): Check is_address flag
12434 for address constraints.
12435 (process_alt_operands): Likewise.
12436 * lra.c (lra_set_insn_recog_data): Pass asm operand locs to
12437 preprocess_constraints.
12438 * recog.h (preprocess_constraints): Add oploc parameter.
12440 * recog.c (preprocess_constraints): Test address_operand for
12441 CT_ADDRESS constraints.
12443 2018-03-09 Vladimir Makarov <vmakarov@redhat.com>
12446 * lra-assigns.c (assign_by_spills): Return a flag of reload
12447 assignment failure. Do not process the reload assignment
12448 failures. Do not spill other reload pseudos if they has the same
12450 (lra_assign): Add a return arg. Set up from the result of
12451 assign_by_spills call.
12452 (find_reload_regno_insns, lra_split_hard_reg_for): New functions.
12453 * lra-constraints.c (split_reg): Add a new arg. Use it instead of
12454 usage_insns if it is not NULL.
12455 (spill_hard_reg_in_range): New function.
12456 (split_if_necessary, inherit_in_ebb): Pass a new arg to split_reg.
12457 * lra-int.h (spill_hard_reg_in_range, lra_split_hard_reg_for): New
12458 function prototypes.
12459 (lra_assign): Change prototype.
12460 * lra.c (lra): Add code to deal with fails by splitting hard reg
12463 2018-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12466 * common/config/arm/arm-common.c (arm_parse_arch_option_name):
12467 Accept complain bool parameter. Only emit errors if it is true.
12468 (arm_parse_cpu_option_name): Likewise.
12469 (arm_target_thumb_only): Adjust callers of the above.
12470 * config/arm/arm-protos.h (arm_parse_cpu_option_name): Adjust
12471 prototype to take a default true bool parameter.
12472 (arm_parse_arch_option_name): Likewise.
12474 2018-03-09 David Malcolm <dmalcolm@redhat.com>
12475 Francois-Xavier Coudert <fxcoudert@gcc.gnu.org>
12479 * Makefile.in (LD_VERSION_SCRIPT_OPTION, LD_SONAME_OPTION): New.
12480 * configure: Regenerate.
12481 * configure.ac ("linker --version-script option"): New.
12482 ("linker soname option"): New.
12484 2018-03-09 Richard Biener <rguenther@suse.de>
12486 PR tree-optimization/84775
12487 * tree-if-conv.c (add_bb_predicate_gimplified_stmts): Delink
12488 immediate uses of predicate stmts and mark them modified.
12491 PR tree-optimization/84178
12492 * tree-if-conv.c (combine_blocks): Move insert_gimplified_predicates
12494 (version_loop_for_if_conversion): Delay update_ssa call.
12495 (tree_if_conversion): Delay update_ssa until after predicate
12498 2018-03-09 Eric Botcazou <ebotcazou@adacore.com>
12501 * config/i386/winnt.c (i386_pe_seh_cold_init): Use small pre-allocation
12502 when the function accesses prior frames.
12504 2018-03-08 Jakub Jelinek <jakub@redhat.com>
12507 * dwarf2out.c (dw_loc_list): If list && loc_list->first->next, call
12508 gen_llsym, otherwise call maybe_gen_llsym.
12510 PR inline-asm/84742
12511 * recog.c (asm_operand_ok): Return 0 if multi-character constraint
12512 has ',' character inside of it.
12514 2018-03-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12517 * config/aarch64/aarch64.md (*compare_cstore<mode>_insn): Mark pattern
12518 as clobbering CC_REGNUM.
12520 2018-03-08 Richard Biener <rguenther@suse.de>
12522 PR middle-end/84552
12523 * tree-scalar-evolution.c: Include tree-into-ssa.h.
12524 (follow_copies_to_constant): Do not follow SSA names registered
12527 2018-03-08 Richard Biener <rguenther@suse.de>
12529 PR tree-optimization/84178
12530 * tree-if-conv.c (combine_blocks): Move insert_gimplified_predicates
12532 (version_loop_for_if_conversion): Delay update_ssa call.
12533 (tree_if_conversion): Delay update_ssa until after predicate
12536 2018-03-08 David Malcolm <dmalcolm@redhat.com>
12538 PR tree-optimization/84178
12539 * tree-if-conv.c (release_bb_predicate): Remove the
12540 the assertion that the stmts have NULL use_ops.
12541 Discard the statements, asserting that they haven't
12542 yet been added to a BB.
12544 2018-03-08 Richard Biener <rguenther@suse.de>
12546 PR tree-optimization/84746
12547 * tree-ssa-pre.c (find_leader_in_sets): Deal with SET1 being NULL.
12548 (phi_translate): Pass in destination ANTIC_OUT set.
12549 (phi_translate_1): Likewise. For a simplified result lookup
12550 a leader in ANTIC_OUT and AVAIL_OUT, not the ANTIC_IN sets.
12551 (phi_translate_set): Adjust.
12552 (do_pre_regular_insertion): Likewise.
12553 (do_pre_partial_partial_insertion): Likewise.
12555 2018-03-08 Martin Liska <mliska@suse.cz>
12557 PR gcov-profile/84735
12558 * doc/gcov.texi: Document usage of profile files.
12559 * gcov-io.h: Document changes in the format.
12561 2018-03-08 Alexandre Oliva <aoliva@redhat.com>
12565 * dwarf2out.c (struct dw_line_info_table): Update comments for
12567 (FORCE_RESET_NEXT_VIEW): New.
12568 (FORCE_RESETTING_VIEW_P): New.
12569 (RESETTING_VIEW_P): Check for -1 too.
12570 (ZERO_VIEW_P): Likewise.
12571 (new_line_info_table): Force-reset next view.
12572 (dwarf2out_begin_function): Likewise.
12573 (dwarf2out_source_line): Simplify zero_view_p initialization.
12574 Test FORCE_RESETTING_VIEW_P and RESETTING_VIEW_P instead of
12575 view directly. Omit view when omitting .loc at line 0.
12577 2018-03-08 Jakub Jelinek <jakub@redhat.com>
12579 PR tree-optimization/84740
12580 * tree-switch-conversion.c (process_switch): Call build_constructors
12581 only if info.phi_count is non-zero.
12583 PR tree-optimization/84739
12584 * tree-tailcall.c (find_tail_calls): Check call arguments against
12585 DECL_ARGUMENTS (current_function_decl) rather than
12586 DECL_ARGUMENTS (func) when checking for tail recursion.
12588 2018-03-07 Jakub Jelinek <jakub@redhat.com>
12590 * doc/contrib.texi: Add entries for Martin Liska, David Malcolm,
12591 Marek Polacek, extend Vladimir Makarov's, Jonathan Wakely's and
12592 Volker Reichelt's entry and add entries for people that perform
12593 GCC fuzzy testing and report numerous bugs.
12595 2018-03-07 Segher Boessenkool <segher@kernel.crashing.org>
12598 * config/rs6000/rs6000.c (rs6000_elf_in_small_data_p): Don't put
12599 readonly data in sdata, if that is disabled.
12600 * config/rs6000/sysv4.opt (mreadonly-in-sdata): New option.
12601 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
12602 -mreadonly-in-sdata option.
12604 2018-03-07 Martin Sebor <msebor@redhat.com>
12606 PR tree-optimization/84468
12607 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Consider successor
12608 basic block when looking for nul assignment.
12610 2018-03-07 Eric Botcazou <ebotcazou@adacore.com>
12613 * except.h (output_function_exception_table): Adjust prototype.
12614 * except.c (output_function_exception_table): Remove FNNAME parameter
12615 and add SECTION parameter. Ouput one part of the table at a time.
12616 * final.c (final_scan_insn_1) <NOTE_INSN_SWITCH_TEXT_SECTIONS>: Output
12617 the first part of the exception table and emit unwind directives.
12618 * config/i386/i386-protos.h (i386_pe_end_cold_function): Declare.
12619 (i386_pe_seh_cold_init): Likewise.
12620 * config/i386/cygming.h (ASM_DECLARE_COLD_FUNCTION_NAME): New macro.
12621 (ASM_DECLARE_COLD_FUNCTION_SIZE): Likewise.
12622 * config/i386/i386.c (x86_expand_epilogue): Fix wording in comment.
12623 (ix86_output_call_insn): Emit a nop in one more case for SEH.
12624 * config/i386/winnt.c: Include except.h.
12625 (struct seh_frame_state): Add reg_offset, after_prologue and
12626 in_cold_section fields.
12627 (i386_pe_seh_end_prologue): Set seh->after_prologue.
12628 (i386_pe_seh_cold_init): New function.
12629 (i386_pe_seh_fini): Add COLD parameter and bail out if it is not equal
12630 to seh->in_cold_section.
12631 (seh_emit_push): Record the offset of the push.
12632 (seh_emit_save): Record the offet of the save.
12633 (i386_pe_seh_unwind_emit): Deal with NOTE_INSN_SWITCH_TEXT_SECTIONS.
12634 Test seh->after_prologue to disregard the epilogue.
12635 (i386_pe_end_function): Pass FALSE to i386_pe_seh_fini.
12636 (i386_pe_end_cold_function): New function.
12638 2018-03-07 Jakub Jelinek <jakub@redhat.com>
12641 * config/aarch64/predicates.md (aarch64_simd_reg_or_zero): Use
12642 aarch64_simd_or_scalar_imm_zero rather than aarch64_simd_imm_zero.
12645 * gimple-expr.c (create_tmp_var_raw): Set DECL_NAMELESS flag
12647 * tree-pretty-print.c (dump_decl_name): For TDF_COMPARE_DEBUG,
12648 don't print names of DECL_NAMELESS DECL_IGNORED_P decls.
12650 PR middle-end/84723
12651 * multiple_target.c: Include tree-inline.h and intl.h.
12652 (expand_target_clones): Diagnose and fail if node->definition and
12653 !tree_versionable_function_p (node->decl).
12655 2018-03-06 John David Anglin <danglin@gcc.gnu.org>
12657 * config/pa/pa.h (ASM_GENERATE_INTERNAL_LABEL): Revise to use
12659 (ASM_OUTPUT_ADDR_VEC_ELT): Revise for above change.
12660 (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
12661 * config/pa/pa64-hpux.h (ASM_GENERATE_INTERNAL_LABEL): Revise as above.
12663 2018-03-06 Jakub Jelinek <jakub@redhat.com>
12666 * combine.c (try_combine): Use reg_or_subregno instead of handling
12667 just paradoxical SUBREGs and REGs.
12669 2018-03-06 Claudiu Zissulescu <claziss@synopsys.com>
12671 * config/arc/arc.c (arc_finalize_pic): Remove function.
12672 (arc_must_save_register): We use single base PIC register, remove
12673 checks to save/restore the PIC register.
12674 (arc_expand_prologue): Likewise.
12675 * config/arc/arc-protos.h (arc_set_default_type_attributes):
12677 (arc_verify_short): Likewise.
12678 (arc_attr_type): Likewise.
12679 * config/arc/arc.c (arc_set_default_type_attributes): Remove.
12680 (walk_stores): Likewise.
12681 (arc_address_cost): Make it static.
12682 (arc_verify_short): Likewise.
12683 (branch_dest): Likewise.
12684 (arc_attr_type): Likewise.
12685 * config/arc/arc.c (TARGET_ADJUST_INSN_LENGTH): Remove.
12686 (TARGET_INSN_LENGTH_PARAMETERS): Likewise.
12687 (arc_final_prescan_insn): Remove inserting the nops due to
12688 hardware hazards. It is done in reorg step.
12689 (insn_length_variant_t): Remove.
12690 (insn_length_parameters_t): Likewise.
12691 (arc_insn_length_parameters): Likewise.
12692 (arc_get_insn_variants): Likewise.
12693 * config/arc/arc.h (TARGET_UPSIZE_DBR): Remove.
12695 2018-03-06 Jakub Jelinek <jakub@redhat.com>
12697 PR inline-asm/84683
12698 * reg-stack.c (move_for_stack_reg): If any_malformed_asm, avoid
12701 PR tree-optimization/84687
12702 * omp-simd-clone.c (simd_clone_create): Clear DECL_BUILT_IN_CLASS
12704 * match.pd (pow(C,x)*expN(y) -> expN(logN(C)*x+y)): New optimization.
12706 2018-03-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12708 * config/rs6000/rs6000-builtin.def (rs6000_speculation_barrier):
12709 Rename to ppc_speculation_barrier.
12710 * config/rs6000/rs6000.c (rs6000_init_builtins): Rename builtin to
12711 __builtin_ppc_speculation_barrier.
12713 2018-03-05 Jakub Jelinek <jakub@redhat.com>
12716 * combine.c (combine_simplify_rtx): Don't try to simplify if
12717 if_then_else_cond returned non-NULL, but either true_rtx or false_rtx
12720 2018-03-05 Segher Boessenkool <segher@kernel.crashing.org>
12722 * config/rs6000/rs6000.c (rs6000_loop_align): Don't align tiny loops
12723 to 32 bytes when compiling for POWER9.
12725 2018-03-05 Jakub Jelinek <jakub@redhat.com>
12728 * config/i386/i386.c (ix86_function_ok_for_sibcall): Check for
12729 regparm >= 3 with no arg reg available also for calls with
12730 flag_force_indirect_call. Pass decl to ix86_function_regparm.
12733 * config/i386/sse.md (*<code><mode>3): Replace <mask_prefix3> with
12735 (*<plusminus_insn><mode>3): Likewise. Remove <mask_operand3> uses.
12737 2018-03-05 Peter Bergner <bergner@vnet.ibm.com>
12740 * config/rs6000/vector.md (mov<mode>): Disallow altivec memory operands.
12742 2018-03-05 Richard Biener <rguenther@suse.de>
12744 PR tree-optimization/84486
12745 * tree-ssa-pre.c (create_expression_by_pieces): Remove dead code.
12746 When inserting a __builtin_assume_aligned call set the LHS
12747 SSA name alignment info accordingly.
12749 2018-03-05 Wilco Dijkstra <wdijkstr@arm.com>
12751 PR tree-optimization/84114
12752 * config/aarch64/aarch64.c (aarch64_reassociation_width)
12753 Avoid reassociation of FLOAT_MODE addition.
12755 2018-03-05 Olga Makhotina <olga.makhotina@intel.com>
12757 * common/config/i386/i386-common.c (OPTION_MASK_ISA_PCONFIG_SET,
12758 OPTION_MASK_ISA_PCONFIG_UNSET, OPTION_MASK_ISA_WBNOINVD_SET,
12759 OPTION_MASK_ISA_WBNOINVD_UNSET): New definitions.
12760 (ix86_handle_option): Handle -mpconfig and -mwbnoinvd.
12761 * config.gcc (pconfigintrin.h, wbnoinvdintrin.h) : Add headers.
12762 * config/i386/cpuid.h (bit_PCONFIG, bit_WBNOINVD): New.
12763 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mpconfig
12765 * config/i386/i386-builtin.def (__builtin_ia32_wbnoinvd,
12766 __builtin_ia32_wbinvd): New builtins.
12767 (SPECIAL_ARGS2): New.
12768 * config/i386/i386-c.c (__WBNOINVD__, __PCONFIG__): New.
12769 (SPECIAL_ARGS2): New.
12770 * config/i386/i386.c (ix86_target_string): Add -mpconfig and -mwbnoinvd.
12771 (ix86_valid_target_attribute_inner_p): Ditto.
12772 (ix86_init_mmx_sse_builtins): Add special_args2.
12773 * config/i386/i386.h (TARGET_PCONFIG, TARGET_PCONFIG_P, TARGET_WBNOINVD,
12774 TARGET_WBNOINVD_P): New.
12775 * config/i386/i386.md (UNSPECV_WBINVD, UNSPECV_WBNOINVD): New.
12776 (define_insn "wbinvd", define_insn "wbnoinvd"): New.
12777 * config/i386/i386.opt: Add -mpconfig and -mwbnoinvd.
12778 * config/i386/immintrin.h (_wbinvd): New intrinsic.
12779 * config/i386/pconfigintrin.h: New file.
12780 * config/i386/wbnoinvdintrin.h: Ditto.
12781 * config/i386/x86intrin.h: Add headers pconfigintrin.h and
12783 * doc/invoke.texi (-mpconfig, -mwbnoinvd): New.
12785 2018-03-05 Richard Biener <rguenther@suse.de>
12787 PR tree-optimization/84670
12788 * tree-ssa-pre.c (struct bb_bitmap_sets): Add visited_with_visited_succs
12790 (BB_VISITED_WITH_VISITED_SUCCS): New define.
12791 (compute_antic): Initialize BB_VISITED_WITH_VISITED_SUCCS.
12792 (compute_antic_aux): Only assert the number of values in ANTIC_IN
12793 doesn't grow if all successors (recursively) were visited at least
12796 2018-03-05 Richard Biener <rguenther@suse.de>
12798 PR tree-optimization/84650
12799 * tree-ssa-loop-im.c (pass_lim::execute): Reset the SCEV cache
12800 if executed in the loop pipeline.
12802 2018-03-05 Sandra Loosemore <sandra@codesourcery.com>
12804 * doc/configfiles.texi (Configuration Files): Move info about
12805 conditionalizing $target-protos.h to...
12806 * doc/sourcebuild.texi (Back End): Here. Explain how $target.h
12807 differs from $target-protos.h.
12809 2018-03-05 Kito Cheng <kito.cheng@gmail.com>
12810 Chung-Ju Wu <jasonwucj@gmail.com>
12812 * config/nds32/nds32-protos.h (nds32_expand_setmem): Declare.
12813 * config/nds32/nds32-multiple.md (setmemsi): Define.
12814 * config/nds32/nds32-memory-manipulation.c
12815 (nds32_gen_dup_4_byte_to_word_value): New.
12816 (emit_setmem_word_loop): New.
12817 (emit_setmem_byte_loop): New.
12818 (nds32_expand_setmem_loop): New.
12819 (nds32_expand_setmem_loop_v3m): New.
12820 (nds32_expand_setmem_unroll): New.
12821 (nds32_expand_setmem): New.
12823 2018-03-04 Kito Cheng <kito.cheng@gmail.com>
12824 Chung-Ju Wu <jasonwucj@gmail.com>
12826 * config/nds32/nds32-memory-manipulation.c
12827 (nds32_emit_load_store): New.
12828 (nds32_emit_post_inc_load_store): New.
12829 (nds32_emit_mem_move): New.
12830 (nds32_emit_mem_move_block): New.
12831 (nds32_expand_movmemsi_loop_unknown_size): New.
12832 (nds32_expand_movmemsi_loop_known_size): New.
12833 (nds32_expand_movmemsi_loop): New.
12834 (nds32_expand_movmemsi_unroll): New.
12835 (nds32_expand_movmemqi): Rename ...
12836 (nds32_expand_movmemsi): ... to this.
12837 * config/nds32/nds32-multiple.md (movmemqi): Rename ...
12838 (movmemsi): ... to this.
12839 * config/nds32/nds32-protos.h (nds32_expand_movmemqi): Rename ...
12840 (nds32_expand_movmemsi): ... to this.
12842 2018-03-04 Kito Cheng <kito.cheng@gmail.com>
12843 Monk Chiang <sh.chiang04@gmail.com>
12844 Chung-Ju Wu <jasonwucj@gmail.com>
12846 * config/nds32/nds32-protos.h
12847 (nds32_expand_load_multiple): New arguments.
12848 (nds32_expand_store_multiple): Ditto.
12849 (nds32_valid_multiple_load_store): Rename ...
12850 (nds32_valid_multiple_load_store_p): ... to this.
12851 * config/nds32/nds32-memory-manipulation.c
12852 (nds32_expand_load_multiple): Refine implementation.
12853 (nds32_expand_store_multiple): Ditto.
12854 * config/nds32/nds32-multiple.md
12855 (load_multiple): Update nds32_expand_load_multiple interface.
12856 (store_multiple): Update nds32_expand_store_multiple interface.
12857 * config/nds32/nds32-predicates.c
12858 (nds32_valid_multiple_load_store): Rename ...
12859 (nds32_valid_multiple_load_store_p): ... to this and refine
12861 * config/nds32/predicates.md
12862 (nds32_load_multiple_and_update_address_operation): New predicate.
12863 (nds32_store_multiple_and_update_address_operation): New predicate.
12865 2018-03-04 Kito Cheng <kito.cheng@gmail.com>
12866 Chung-Ju Wu <jasonwucj@gmail.com>
12868 * config/nds32/nds32.md (type): Add load_multiple and store_multiple.
12869 (combo): New attribute.
12870 * config/nds32/nds32-multiple.md: Refine patterns with new attributes.
12872 2018-03-03 Chung-Ju Wu <jasonwucj@gmail.com>
12874 * config/nds32/nds32.opt: Change -mcmodel= default value.
12876 2018-03-03 Kito Cheng <kito.cheng@gmail.com>
12877 Monk Chiang <sh.chiang04@gmail.com>
12878 Chung-Ju Wu <jasonwucj@gmail.com>
12880 * config/nds32/constants.md (unspec_element): New enum.
12881 * config/nds32/constraints.md (Umw): New constraint.
12882 * config/nds32/nds32-intrinsic.c: Add more builtin functions.
12883 * config/nds32/nds32-intrinsic.md: Likewise.
12884 * config/nds32/nds32-md-auxiliary.c (nds32_regno_to_enable4): New.
12885 (nds32_valid_smw_lwm_base_p): New.
12886 (nds32_output_smw_single_word): New.
12887 (nds32_output_lmw_single_word): New.
12888 (nds32_expand_unaligned_load): New.
12889 (nds32_expand_unaligned_store): New.
12890 * config/nds32/nds32-protos.h (nds32_valid_smw_lwm_base_p): Declare.
12891 (nds32_output_smw_single_word): Declare.
12892 (nds32_output_lmw_single_word): Declare.
12893 (nds32_expand_unaligned_load): Declare.
12894 (nds32_expand_unaligned_store): Declare.
12895 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_UALOAD_HW,
12896 NDS32_BUILTIN_UALOAD_W, NDS32_BUILTIN_UALOAD_DW,
12897 NDS32_BUILTIN_UASTORE_HW, NDS32_BUILTIN_UASTORE_W,
12898 NDS32_BUILTIN_UASTORE_DW.
12899 * config/nds32/predicates.md (nds32_lmw_smw_base_operand): New
12902 2018-03-03 Monk Chiang <sh.chiang04@gmail.com>
12903 Kito Cheng <kito.cheng@gmail.com>
12904 Chung-Ju Wu <jasonwucj@gmail.com>
12906 * config/nds32/nds32-intrinsic.c
12907 (nds32_expand_builtin_null_ftype_reg): Delete.
12908 (nds32_expand_builtin_reg_ftype_imm): Ditto.
12909 (nds32_expand_builtin_null_ftype_reg_imm): Ditto.
12910 (nds32_read_argument): New.
12911 (nds32_legitimize_target): Ditto.
12912 (nds32_legitimize_argument): Ditto.
12913 (nds32_check_constant_argument): Ditto.
12914 (nds32_expand_unop_builtin): Ditto.
12915 (nds32_expand_unopimm_builtin): Ditto.
12916 (nds32_expand_binop_builtin): Ditto.
12917 (nds32_builtin_decl_impl): Ditto.
12918 (builtin_description): Ditto.
12919 (nds32_expand_builtin_impl): Rewrite with new infrastructure.
12920 (nds32_init_builtins_impl): Ditto.
12921 * config/nds32/nds32.c (TARGET_BUILTIN_DECL): Define.
12922 (nds32_builtin_decl): New.
12923 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_COUNT.
12924 * config/nds32/nds32-protos.h (nds32_builtin_decl_impl): Declare.
12926 2018-03-02 Jeff Law <law@redhat.com>
12928 * reorg.c (stop_search_p): Handle DEBUG_INSN.
12929 (redundant_insn, fill_simple_delay_slots): Likewise.
12930 (fill_slots_from_thread): Likewise.
12931 * resource.c (mark_referenced_resources): Likewise.
12932 (mark_set_resources, find_dead_or_set_registers): Likewise.
12934 2018-03-02 Jakub Jelinek <jakub@redhat.com>
12936 * substring-locations.h (format_warning_va): Formatting fix for
12937 ATTRIBUTE_GCC_DIAG.
12938 (format_warning_at_substring): Fix up ATTRIBUTE_GCC_DIAG second
12940 (format_warning_n_va, format_warning_at_substring_n): New prototypes.
12941 * substring-locations.c: Include intl.h.
12942 (format_warning_va): Turned into small wrapper around
12943 format_warning_n_va, renamed to ...
12944 (format_warning_n_va): ... this, add N and PLURAL_GMSGID arguments,
12945 rename GMSGID to SINGULAR_GMSGID, if SINGULAR_GMSGID != PLURAL_GMSGID,
12947 (format_warning_at_substring_n): New function.
12948 * gimple-ssa-sprintf.c: Remove GCC diagnostic ignored pragma.
12949 (fmtwarn): Add ATTRIBUTE_GCC_DIAG. Turn into a copy of
12950 format_warning_at_substring with just a shorter name instead of
12951 const function pointer.
12952 (fmtwarn_n): New function.
12953 (maybe_warn, format_directive, parse_directive): Use fmtwarn_n where
12954 appropriate, get rid of all the fmtstr temporaries, move conditionals
12955 with G_() wrapped string literals directly into fmtwarn arguments,
12956 cast dir.len to (int), formatting fixes.
12958 2018-03-02 Thomas Schwinge <thomas@codesourcery.com>
12960 * doc/invoke.texi: Remove "Cilk Plus" references.
12962 2018-03-02 Jakub Jelinek <jakub@redhat.com>
12963 Richard Biener <rguenther@suse.de>
12966 * expr.c (expand_expr_real_1) <case CALL_EXPR>: Don't emit diagnostics
12967 for error or warning attributes if CALL_FROM_THUNK_P is set.
12970 2018-03-02 Jakub Jelinek <jakub@redhat.com>
12973 * config/pa/pa.h (TARGET_CPU_CPP_BUILTINS): Predefine
12974 __SIZEOF_128__ macro if HPUX_LONG_DOUBLE_LIBRARY.
12977 * config/ia64/ia64.h (TARGET_CPU_CPP_BUILTINS): Predefine
12978 __SIZEOF_{FPREG,FLOAT{80,128}}__ macros.
12980 * predict.c (test_prediction_value_range): Use PROB_UNINITIALIZED
12981 instead of -1U in last predictors element's probability member.
12983 2018-03-02 Eric Botcazou <ebotcazou@adacore.com>
12986 * ipa-devirt.c (odr_subtypes_equivalent_p): Get the ODR type of both
12987 arguments if they are comparable.
12989 2018-03-02 Richard Sandiford <richard.sandiford@linaro.org>
12991 PR tree-optimization/84634
12992 * tree-vect-stmts.c (vectorizable_store, vectorizable_load): Replace
12993 masks and masked_loop_p with a single loop_masks, making sure it's
12994 null for bb vectorization.
12996 2018-03-02 Richard Sandiford <richard.sandiford@linaro.org>
12998 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence)
12999 (vect_analyze_data_ref_access): Use loop->safe_len rather than
13000 loop->force_vectorize to check whether there is no alias.
13002 2018-03-02 Jakub Jelinek <jakub@redhat.com>
13005 * rtl.h (prev_real_nondebug_insn, next_real_nondebug_insn): New
13007 * emit-rtl.c (next_real_insn, prev_real_insn): Fix up function
13009 (next_real_nondebug_insn, prev_real_nondebug_insn): New functions.
13010 * cfgcleanup.c (try_head_merge_bb): Use prev_real_nondebug_insn
13011 instead of a loop around prev_real_insn.
13012 * combine.c (move_deaths): Use prev_real_nondebug_insn instead of
13015 PR inline-asm/84625
13016 * config/i386/i386.c (ix86_print_operand): Use conditional
13017 output_operand_lossage instead of gcc_assert if CONST_VECTOR is not
13020 2018-03-02 Richard Biener <rguenther@suse.de>
13022 PR tree-optimization/84427
13023 * tree-ssa-pre.c (bitmap_remove_expr_from_set): Remove.
13024 (bitmap_set_subtract_values): Rewrite to handle multiple
13027 (prune_clobbered_mems): Likewise.
13028 (phi_translate): Take edge instead of pred/phiblock.
13029 (phi_translate_1): Likewise.
13030 (phi_translate_set): Likewise. Insert all translated
13031 exprs for a value into the set, keeping possibly multiple
13032 expressions per value.
13033 (compute_antic_aux): Adjust for phi_translate changes.
13034 When intersecting union the expressions and prune those
13035 not in the final value set, keeping possibly multiple
13036 expressions per value. Do not use value-insertion
13037 for unioning ANTIC_OUT U EXP_GEN - TMP_GEN but merge
13038 all expressions. Add verification that the value-sets
13039 only shrink during iteration.
13040 (compute_partial_antic_aux): Adjust for the phi_translate changes.
13041 (do_pre_regular_insertion): Likewise.
13042 (do_pre_partial_partial_insertion): Likewise.
13044 2018-03-02 Richard Biener <rguenther@suse.de>
13047 * config/darwin.c (saved_debug_info_level): New static global.
13048 (darwin_asm_lto_start): Disable debug info generation for LTO out.
13049 (darwin_asm_lto_end): Restore debug info generation settings.
13051 2018-03-01 Martin Liska <mliska@suse.cz>
13054 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
13055 volatile arguments.
13057 2018-03-01 Richard Biener <rguenther@suse.de>
13060 * dwarf2out.c (gen_variable_die): Properly handle late VLA
13061 type annotation with LTO when debug was disabled at compile-time.
13063 2018-03-01 Matthew Fortune <mfortune@gmail.com>
13065 * config/mips/mips.c (mips_final_prescan_insn): Fix incorrect
13067 (mips_final_postscan_insn): Likewise.
13069 2018-03-01 Richard Sandiford <richard.sandiford@linaro.org>
13071 PR rtl-optimization/84528
13072 * alias.c (init_alias_target): Add commentary.
13073 (init_alias_analysis): Only give HARD_FRAME_POINTER_REGNUM
13074 a unique base value if the frame pointer is not eliminated
13075 to the stack pointer.
13077 2018-03-01 Tom de Vries <tom@codesourcery.com>
13079 PR rtl-optimization/83327
13080 * lra-int.h (hard_regs_spilled_into): Declare.
13081 * lra.c (hard_regs_spilled_into): Define.
13082 (init_reg_info): Init hard_regs_spilled_into.
13083 * lra-spills.c (assign_spill_hard_regs): Update hard_regs_spilled_into.
13084 * lra-lives.c (make_hard_regno_born, make_hard_regno_dead)
13085 (process_bb_lives): Handle hard_regs_spilled_into.
13086 (lra_create_live_ranges_1): Before doing liveness propagation, clear
13087 regs in all_hard_regs_bitmap if set in hard_regs_spilled_into.
13089 2018-02-28 David Edelsohn <dje.gcc@gmail.com>
13091 * config.gcc (powerpc-ibm-aix7.1.*): New stanza.
13092 (powerpc-ibm-aix[789]*): Default to AIX 7.2.
13093 * config/rs6000/aix71.h (TARGET_DEFAULT): Revert to Power4 ISA.
13094 * config/rs6000/aix72.h: New file.
13096 2018-02-28 Jakub Jelinek <jakub@redhat.com>
13098 * gimple-ssa-warn-restrict.c (maybe_diag_overlap): Use warning_n
13099 instead of warning_at with conditional singular and plural messages
13103 * stor-layout.c (update_alignment_for_field): For
13104 targetm.ms_bitfield_layout_p (rli->t), if !is_bitfield
13105 && !DECL_PACKED (field), do the alignment update, just use
13106 only desired_align instead of MAX (type_align, desired_align)
13108 (place_field): Don't do known_align < desired_align handling
13109 early if targetm.ms_bitfield_layout_p (rli->t) and rli->prev_field
13110 is non-NULL, instead do it after rli->prev_field handling and
13111 only if not within a bitfield word. For DECL_PACKED (field)
13112 use type_align of BITS_PER_UNIT.
13114 2018-02-28 Eric Botcazou <ebotcazou@adacore.com>
13116 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Remove
13117 superfluous parentheses and trailing spaces.
13119 2018-02-28 Richard Biener <rguenther@suse.de>
13121 PR tree-optimization/84584
13122 * graphite-scop-detection.c (scop_detection::add_scop): Discard
13123 SCoPs with fake exit edge.
13125 2018-02-28 Martin Liska <mliska@suse.cz>
13128 * timevar.c (timer::print): Fix format to properly print 100%
13131 2018-02-28 Richard Biener <rguenther@suse.de>
13133 PR middle-end/84607
13134 * genmatch.c (capture_info::walk_match): Do not mark
13135 captured expressions without operands as expr_p given
13136 they act more like predicates and should be subject to
13137 "lost tail" side-effect preserving.
13139 2018-02-28 Alexandre Oliva <aoliva@redhat.com>
13141 PR rtl-optimization/81611
13142 * auto-inc-dec.c (attempt_change): Move dead note from
13143 mem_insn if it's the next use of regno
13144 (find_address): Take address use of reg holding
13145 non-incremented value. Add parm to limit search to the named
13147 (merge_in_block): Attempt to use a mem insn that is the next
13148 use of the original regno.
13150 2018-02-27 Martin Sebor <msebor@redhat.com>
13153 * doc/invoke.texi (-Wmissing-attributes): New option.
13154 * print-tree.c (print_node): Handle DECL_UNINLINABLE.
13156 2018-02-27 Martin Sebor <msebor@redhat.com>
13158 PR translation/84207
13159 * diagnostic-core.h (warning_n, error_n, inform_n): Change
13160 n argument to unsigned HOST_WIDE_INT.
13161 * diagnostic.c (warning_n, error_n, inform_n): Ditto.
13162 (diagnostic_n_impl): Ditto. Handle arguments in excess of LONG_MAX.
13163 * gimple-ssa-sprintf.c (format_directive): Simplify inform_n call.
13164 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Use warning_n.
13166 2018-02-27 Richard Biener <rguenther@suse.de>
13168 PR tree-optimization/84512
13169 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost):
13170 Do not use the estimate returned from record_stmt_cost for
13171 the scalar iteration cost but sum properly using add_stmt_cost.
13173 2018-02-27 Richard Biener <rguenther@suse.de>
13175 PR tree-optimization/84466
13176 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
13177 Adjust last change to less strictly validate use operands.
13179 2018-02-27 Martin Liska <mliska@suse.cz>
13181 PR gcov-profile/84548
13182 * gcov.c (process_file): Allow partial overlap and consider it
13183 also as group functions.
13184 (output_lines): Properly calculate range of lines for a group.
13186 2018-02-27 Martin Liska <mliska@suse.cz>
13188 * timevar.c (timer::print_row): Remove 'usr', 'sys', 'wall' and
13189 'ggc' suffixes. Change first column width.
13190 (timer::print): Fix formatting of the column.
13192 2018-02-27 Alexandre Oliva <aoliva@redhat.com>
13194 * tree-ssa-live.c (remove_unused_scope_block_p): Do not
13195 preserve inline entry blocks for the sake of debug inline
13196 entry point markers alone.
13197 (remove_unused_locals): Suggest in comments a better place to
13198 force the preservation of inline entry blocks that are
13199 otherwise unused, but do not preserve them.
13201 2018-02-26 H.J. Lu <hongjiu.lu@intel.com>
13203 * config/i386/i386.c (ix86_output_indirect_jmp): Update comments.
13205 2018-02-26 H.J. Lu <hongjiu.lu@intel.com>
13208 * config/i386/constraints.md (Bs): Replace
13209 ix86_indirect_branch_register with
13210 TARGET_INDIRECT_BRANCH_REGISTER.
13212 * config/i386/i386.md (indirect_jump): Likewise.
13213 (tablejump): Likewise.
13214 (*sibcall_memory): Likewise.
13215 (*sibcall_value_memory): Likewise.
13216 Peepholes of indirect call and jump via memory: Likewise.
13217 (*sibcall_GOT_32): Disallowed for TARGET_INDIRECT_BRANCH_REGISTER.
13218 (*sibcall_value_GOT_32): Likewise.
13219 * config/i386/predicates.md (indirect_branch_operand): Likewise.
13220 (GOT_memory_operand): Likewise.
13221 (call_insn_operand): Likewise.
13222 (sibcall_insn_operand): Likewise.
13223 (GOT32_symbol_operand): Likewise.
13224 * config/i386/i386.h (TARGET_INDIRECT_BRANCH_REGISTER): New.
13226 2018-02-26 Eric Botcazou <ebotcazou@adacore.com>
13228 PR rtl-optimization/83496
13229 * reorg.c (steal_delay_list_from_target): Change REDUNDANT array from
13230 booleans to RTXes. Call fix_reg_dead_note on every non-null element.
13231 (steal_delay_list_from_fallthrough): Call fix_reg_dead_note on a
13232 redundant insn, if any.
13233 (relax_delay_slots): Likewise.
13234 (update_reg_unused_notes): Rename REDUNDANT_INSN to OTHER_INSN.
13236 2018-02-26 Richard Sandiford <richard.sandiford@linaro.org>
13238 PR tree-optimization/83965
13239 * tree-vect-patterns.c (vect_reassociating_reduction_p): Assume
13240 that grouped statements are part of a reduction chain. Return
13241 true if the statement is not marked as a reduction itself but
13242 is part of a group.
13243 (vect_recog_dot_prod_pattern): Don't check whether the statement
13244 is part of a group here.
13245 (vect_recog_sad_pattern): Likewise.
13246 (vect_recog_widen_sum_pattern): Likewise.
13248 2018-02-26 Eric Botcazou <ebotcazou@adacore.com>
13251 * final.c (rest_of_clean_state): Also look for calls inside sequences.
13253 2018-02-26 H.J. Lu <hongjiu.lu@intel.com>
13256 * config/i386/i386-protos.h (ix86_output_indirect_jmp): Remove
13258 (ix86_output_indirect_function_return): New prototype.
13259 (ix86_split_simple_return_pop_internal): Likewise.
13260 * config/i386/i386.c (indirect_return_via_cx): New.
13261 (indirect_return_via_cx_bnd): Likewise.
13262 (indirect_thunk_name): Handle return va CX_REG.
13263 (output_indirect_thunk_function): Create alias for
13264 __x86_return_thunk_[re]cx and __x86_return_thunk_[re]cx_bnd.
13265 (ix86_output_indirect_jmp): Remove the bool argument.
13266 (ix86_output_indirect_function_return): New function.
13267 (ix86_split_simple_return_pop_internal): Likewise.
13268 * config/i386/i386.md (*indirect_jump): Don't pass false
13269 to ix86_output_indirect_jmp.
13270 (*tablejump_1): Likewise.
13271 (simple_return_pop_internal): Change it to define_insn_and_split.
13272 Call ix86_split_simple_return_pop_internal to split it for
13273 -mfunction-return=.
13274 (simple_return_indirect_internal): Call
13275 ix86_output_indirect_function_return instead of
13276 ix86_output_indirect_jmp.
13278 2018-02-26 Jakub Jelinek <jakub@redhat.com>
13281 * vec.h (vec_default_construct): For BROKEN_VALUE_INITIALIZATION use
13282 memset and value initialization afterwards.
13284 2018-02-26 Christophe Lyon <christophe.lyon@linaro.org>
13286 * Makefile.in (lto-wrapper): Use ALL_LINKERFLAGS.
13288 2018-02-26 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
13291 * common/config/aarch64/aarch64-common.c
13292 (aarch_option_optimization_table[]): Switch
13293 off fomit-frame-pointer
13295 2018-02-26 Kito Cheng <kito.cheng@gmail.com>
13296 Chung-Ju Wu <jasonwucj@gmail.com>
13298 * config/nds32/nds32-multiple.md (load_multiple): Disallow
13300 (store_multiple): Ditto.
13302 2018-02-26 Kito Cheng <kito.cheng@gmail.com>
13304 * config.gcc: Add --with-cpu support for nds32 target.
13305 * config/nds32/nds32-opts.h (nds32_cpu_type): New.
13306 * config/nds32/nds32.opt: Add -mcpu= option.
13308 2018-02-25 Segher Boessenkool <segher@kernel.crashing.org>
13310 * config/rs6000/rs6000.opt (mvrsave=no, mvrsave=yes, isel=no,
13311 isel=yes): Warn for these deprecated options.
13313 2018-02-23 David Edelsohn <dje.gcc@gmail.com>
13315 * config/rs6000/aix71.h (TARGET_DEFAULT): Change to
13316 ISA_2_5_MASKS_EMBEDDED.
13318 2018-02-23 Jakub Jelinek <jakub@redhat.com>
13320 * ipa-prop.c (ipa_vr_ggc_hash_traits::hash): Hash p->min and
13321 p->max as pointers rather than using iterative_hash_expr.
13323 2018-02-23 Carl Love <cel@us.ibm.com>
13325 * config/rs6000/rs6000-builtin.def: Change VSIGNED2 and VUNSIGNED2
13326 macro expansions from BU_VSX_2 to BU_P8V_VSX_2 and BU_VSX_OVERLOAD_2 to
13328 * config/rs6000/rs6000-c.c: Change VSX_BUILTIN_VEC_VSIGNED2 to
13329 P8V_BUILTIN_VEC_VSIGNED2. Change VSX_BUILTIN_VEC_VUNSIGNED2 to
13330 P8V_BUILTIN_VEC_VUNSIGNED2.
13332 2018-02-22 Vladimir Makarov <vmakarov@redhat.com>
13335 * lra-int.h (LRA_UNKNOWN_ALT, LRA_NON_CLOBBERED_ALT): New macros.
13336 * lra.c (lra_set_insn_recog_data, lra_update_insn_recog_data): Use
13338 * lra-constraints.c (curr_insn_transform): Set up
13339 LRA_NON_CLOBBERED_ALT for moves processed on the fast path. Use
13341 (remove_inheritance_pseudos): Use LRA_UNKNOWN_ALT.
13342 * lra-eliminations.c (spill_pseudos): Ditto.
13343 (process_insn_for_elimination): Ditto.
13344 * lra-lives.c (reg_early_clobber_p): Use the new macros.
13345 * lra-spills.c (spill_pseudos): Use LRA_UNKNOWN_ALT and
13346 LRA_NON_CLOBBERED_ALT.
13348 2018-02-22 Martin Sebor <msebor@redhat.com>
13350 PR tree-optimization/84480
13351 * gimple-fold.c (gimple_fold_builtin_strcpy): Move warnings
13352 to maybe_diag_stxncpy_trunc. Call it.
13353 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Integrate warnings
13354 from gimple_fold_builtin_strcpy. Print inlining stack.
13355 (handle_builtin_stxncpy): Print inlining stack.
13356 * tree-ssa-strlen.h (maybe_diag_stxncpy_trunc): Declare.
13358 2018-02-22 H.J. Lu <hongjiu.lu@intel.com>
13361 * config/i386/i386.c (ix86_set_indirect_branch_type): Issue an
13362 error when -mindirect-branch=thunk-extern, -fcf-protection=branch
13363 and -fcheck-pointer-bounds are used together.
13364 (indirect_thunk_prefix): New enum.
13365 (indirect_thunk_need_prefix): New function.
13366 (indirect_thunk_name): Replace need_bnd_p with need_prefix. Use
13367 "_nt" instead of "_bnd" for NOTRACK prefix.
13368 (output_indirect_thunk): Replace need_bnd_p with need_prefix.
13369 (output_indirect_thunk_function): Likewise.
13371 (ix86_code_end): Update output_indirect_thunk_function calls.
13372 (ix86_output_indirect_branch_via_reg): Replace
13373 ix86_bnd_prefixed_insn_p with indirect_thunk_need_prefix.
13374 (ix86_output_indirect_branch_via_push): Likewise.
13375 (ix86_output_function_return): Likewise.
13376 * doc/invoke.texi: Document -mindirect-branch=thunk-extern is
13377 incompatible with -fcf-protection=branch and
13378 -fcheck-pointer-bounds.
13380 2018-02-22 Steve Ellcey <sellcey@cavium.com>
13383 * config/aarch64/aarch64.c (aarch64_print_address_internal):
13384 Change gcc_assert call to output_operand_lossage.
13386 2018-02-22 Steve Ellcey <sellcey@cavium.com>
13388 * doc/extend.texi (__builtin_extend_pointer): Document builtin.
13390 2018-02-22 DJ Delorie <dj@redhat.com>
13391 Sebastian Perta <sebastian.perta@renesas.com>
13392 Oleg Endo <olegendo@gcc.gnu.org>
13394 * config/rx/rx.c (rx_rtx_costs): New function.
13395 (TARGET_RTX_COSTS): Override to use rx_rtx_costs.
13397 2018-02-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
13399 * config/arm/t-multilib: Map Armv8-R to Armv7 multilibs.
13401 2018-02-22 Martin Liska <mliska@suse.cz>
13404 * common/config/arm/arm-common.c (arm_print_hint_for_cpu_option):
13405 Add "native" as a possible value.
13407 2018-02-22 Martin Liska <mliska@suse.cz>
13410 * config/i386/i386.c (ix86_option_override_internal):
13411 Add "native" as a possible value for -march and -mtune.
13413 2018-02-22 Jakub Jelinek <jakub@redhat.com>
13416 * stor-layout.c (finalize_type_size): Propagate TYPE_EMPTY_P flag
13417 to all type variants.
13419 PR tree-optimization/84503
13420 * gimple-ssa-store-merging.c (merged_store_group::merge_into): Compute
13421 width as info->bitpos + info->bitsize - start.
13422 (merged_store_group::merge_overlapping): Simplify width computation.
13423 (check_no_overlap): New function.
13424 (imm_store_chain_info::try_coalesce_bswap): Compute expected
13425 start + width and last_order of the group, fail if check_no_overlap
13427 (imm_store_chain_info::coalesce_immediate_stores): Don't merge info
13428 to group if check_no_overlap fails.
13430 2018-02-21 Segher Boessenkool <segher@kernel.crashing.org>
13432 * config/rs6000/altivec.md: Delete contraint arguments to
13433 define_expand, define_split, and define_peephole2, and in
13434 define_insn_and_split if always unused.
13435 * config/rs6000/darwin.md: Ditto.
13436 * config/rs6000/dfp.md: Ditto.
13437 * config/rs6000/rs6000.md: Ditto.
13438 * config/rs6000/sync.md: Ditto.
13439 * config/rs6000/vector.md: Ditto.
13440 * config/rs6000/vsx.md: Ditto.
13442 2018-02-21 Segher Boessenkool <segher@kernel.crashing.org>
13444 * config/rs6000/altivec.md: Write output control strings as braced
13445 blocks instead of double-quoted strings.
13446 * config/rs6000/darwin.md: Ditto.
13447 * config/rs6000/rs6000.md: Ditto.
13448 * config/rs6000/vector.md: Ditto.
13449 * config/rs6000/vsx.md: Ditto.
13451 2018-02-21 Jason Merrill <jason@redhat.com>
13453 PR c++/84314 - ICE with templates and fastcall attribute.
13454 * attribs.c (build_type_attribute_qual_variant): Remove assert.
13456 2018-02-21 Jan Hubicka <hubicka@ucw.cz>
13458 * ipa-cp.c (determine_versionability): Fix comment typos.
13460 2018-02-21 Jan Hubicka <hubicka@ucw.cz>
13463 * ipa-cp.c (determine_versionability): Do not version functions caling
13466 2018-02-21 Martin Liska <mliska@suse.cz>
13469 * config/aarch64/aarch64.c (aarch64_print_hint_for_core_or_arch):
13470 Add "native" as a possible value.
13471 * config/aarch64/aarch64.h (HAVE_LOCAL_CPU_DETECT): Define
13472 the macro when native cpu detection is available.
13474 2018-02-21 Martin Liska <mliska@suse.cz>
13477 * common/config/arm/arm-common.c (arm_print_hint_for_arch_option):
13478 Add "native" as a possible value.
13479 * config/arm/arm.h (HAVE_LOCAL_CPU_DETECT): Define the macro
13480 when native cpu detection is available.
13482 2018-02-21 Jakub Jelinek <jakub@redhat.com>
13483 Martin Sebor <msebor@redhat.com>
13485 PR tree-optimization/84478
13486 * gimple-fold.h (get_range_strlen): Add a bool argument defaulted to
13488 * gimple-fold.c (get_range_strlen): Make minlen const and assume it
13489 can't be NULL. Change FUZZY from bool to int, for 1 add PHI/COND_EXPR
13490 support which is conservatively correct, for 2 only stay conservative
13491 for maxlen. Formatting and comment capitalization fixes. Add STRICT
13492 argument to the 2 argument get_range_strlen, adjust 6 arg
13493 get_range_strlen caller and clear minmaxlen[0] and [1] if it returned
13495 (get_maxval_strlen): Adjust 6 arg get_range_strlen caller.
13496 (gimple_fold_builtin_strlen): Pass true as last argument to
13499 2018-02-20 Martin Sebor <msebor@redhat.com>
13501 PR middle-end/84095
13502 * gimple-ssa-warn-restrict.c (builtin_memref::extend_offset_range): New.
13503 (builtin_memref::set_base_and_offset): Same. Handle inner references.
13504 (builtin_memref::builtin_memref): Factor out parts into
13505 set_base_and_offset and call it.
13507 2018-02-20 Richard Sandiford <richard.sandiford@linaro.org>
13509 PR middle-end/84406
13510 * optabs-query.c (find_widening_optab_handler_and_mode): If from_mode
13511 is a scalar_int_mode, assert that to_mode is a scalar_int_mode with
13512 greater precision. If to_mode is a MODE_PARTIAL_INT, stop the
13513 search at the associated MODE_INT.
13515 2018-02-20 Jeff Law <law@redhat.com>
13517 PR middle-end/82123
13518 PR tree-optimization/81592
13519 PR middle-end/79257
13520 * gimple-ssa-sprintf.c (format_integer): Query EVRP range analyzer
13521 for range data rather than using global data.
13522 * gimple-ssa-sprintf.c (get_int_range): Query EVRP range analyzer for
13523 range data rather than using global data.
13524 * gimple-ssa-sprintf.c (get_int_range): Accept vr_values parameter
13525 pass it to children as needed.
13526 (struct directive::fmtresult): Similarly.
13527 (struct directive::set_width): Similarly.
13528 (struct directive::set_precision): Similarly.
13529 (format_integer, format_directive, parse_directive): Similarly.
13530 (format_none): Accept unnamed vr_values parameter.
13531 (format_percent, format_floating, format_character): Similarly.
13532 (format_string, format_plain): Similarly.
13533 * gimple-ssa-sprintf.c (sprintf_dom_walker::handle_gimple_call): Query
13534 the EVRP range analyzer for range data rather than using global data.
13535 * gimple-ssa-sprintf.c: Include alloc-pool.h, vr-values.h and
13536 gimple-ssa-evrp-analyze.h
13537 (class sprintf_dom_walker): Add after_dom_children member function.
13538 Add evrp_range_analyzer member.
13539 (sprintf_dom_walker::before_dom_children): Call into the EVRP
13540 range analyzer as needed.
13541 (sprintf_dom_walker::after_dom_children): New member function.
13542 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::enter): Do nothing
13544 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
13545 (evrp_range_analyzer::pop_to_marker): Likewise.
13547 2018-02-20 Richard Sandiford <richard.sandiford@linaro.org>
13549 PR tree-optimization/84419
13550 * internal-fn.c (expand_call_mem_ref): Create a TARGET_MEM_REF
13551 with the required type if its current type is compatible but
13554 2018-02-20 Jakub Jelinek <jakub@redhat.com>
13556 PR middle-end/82004
13557 * match.pd (pow(C,x) -> exp(log(C)*x)): Delay all folding until
13558 after vectorization.
13560 2018-02-20 Martin Liska <mliska@suse.cz>
13563 * config/aarch64/aarch64.c (aarch64_print_hint_for_core_or_arch): Print
13564 possible values if we don't have a hint.
13566 2018-02-20 Martin Liska <mliska@suse.cz>
13570 * final.c (shorten_branches): Build align_tab array with one
13572 * opts.c (finish_options): Add alignment option limit check.
13573 (MAX_CODE_ALIGN): Likewise.
13574 (MAX_CODE_ALIGN_VALUE): Likewise.
13575 * doc/invoke.texi: Document maximum allowed option value for
13576 all -falign-* options.
13578 2018-02-19 Jakub Jelinek <jakub@redhat.com>
13581 * reg-notes.def (REG_CALL_ARG_LOCATION): New reg note.
13582 * insn-notes.def (NOTE_INSN_CALL_ARG_LOCATION): Remove.
13583 * var-tracking.c (emit_note_insn_var_location): Remove all references
13584 to NOTE_INSN_CALL_ARG_LOCATION.
13585 (emit_notes_in_bb): Emit arguments as REG_CALL_ARG_LOCATION note on
13586 the CALL_INSN rather than separate NOTE_INSN_CALL_ARG_LOCATION note.
13587 Use copy_rtx_if_shared.
13588 * dwarf2out.c (gen_subprogram_die): Use XEXP with 0 instead of
13589 NOTE_VAR_LOCATION on ca_loc->call_arg_loc_note.
13590 (dwarf2out_var_location): Remove handling of
13591 NOTE_INSN_CALL_ARG_LOCATION, instead handle REG_CALL_ARG_LOCATION note
13593 * final.c (final_scan_insn): Remove all references to
13594 NOTE_INSN_CALL_ARG_LOCATION.
13595 (rest_of_clean_state): Likewise. Remove REG_CALL_ARG_LOCATION notes
13596 before dumping final insns.
13597 * except.c (emit_note_eh_region_end): Remove all references to
13598 NOTE_INSN_CALL_ARG_LOCATION.
13599 * config/alpha/alpha.c (alpha_pad_function_end): Likewise.
13600 * config/c6x/c6x.c (c6x_gen_bundles): Likewise.
13601 * config/arc/arc.c (hwloop_optimize): Likewise.
13602 * config/arm/arm.c (create_fix_barrier): Likewise.
13603 * config/s390/s390.c (s390_chunkify_start): Likewise.
13604 * config/sh/sh.c (find_barrier): Likewise.
13605 * config/i386/i386.c (rest_of_insert_endbranch,
13606 ix86_seh_fixup_eh_fallthru): Likewise.
13607 * config/xtensa/xtensa.c (hwloop_optimize): Likewise.
13608 * config/iq2000/iq2000.c (final_prescan_insn): Likewise.
13609 * config/frv/frv.c (frv_function_prologue): Likewise.
13610 * emit-rtl.c (try_split): Likewise. Copy over REG_CALL_ARG_LOCATION
13612 (note_outside_basic_block_p): Remove all references to
13613 NOTE_INSN_CALL_ARG_LOCATION.
13614 * gengtype.c (adjust_field_rtx_def): Likewise.
13615 * print-rtl.c (rtx_writer::print_rtx_operand_code_0, print_insn):
13617 * jump.c (cleanup_barriers, delete_related_insns): Likewise.
13618 * cfgrtl.c (force_nonfallthru_and_redirect): Likewise.
13621 * builtins.c (builtin_mathfn_code): Don't check if CALL_EXPR_FN (t)
13624 PR tree-optimization/84452
13625 * tree-vect-patterns.c (vect_recog_pow_pattern): Don't call
13626 expand_simd_clones if targetm.simd_clone.compute_vecsize_and_simdlen
13629 2018-02-19 Martin Liska <mliska@suse.cz>
13632 * passes.def: Put pass_sancov_O0 before pass_lower_switch with -O0.
13634 2018-02-19 Martin Liska <mliska@suse.cz>
13635 Richard Sandiford <richard.sandiford@linaro.org>
13637 PR tree-optimization/82491
13638 * gimple-fold.c (get_base_constructor): Make earlier bail out
13641 2018-02-19 Carl Love <cel@us.ibm.com>
13643 * config/rs6000/rs6000-builtin.def: Change NEG macro expansions from
13644 BU_ALTIVEC_A to BU_P8V_AV_1 and BU_ALTIVEC_OVERLOAD_1 to
13646 * config/rs6000/rs6000-c.c: Change ALTIVEC_BUILTIN_VEC_NEG to
13647 P8V_BUILTIN_VEC_NEG.
13649 2018-02-19 Sebastian Perta <sebastian.perta@renesas.com>
13651 * config/rl78/rl78.md (movdf): New define expand.
13653 2018-02-19 Martin Liska <mliska@suse.cz>
13656 * doc/invoke.texi: Fix typo.
13657 * params.def (PARAM_MAX_LOOP_HEADER_INSNS): Likewise.
13659 2018-02-18 Segher Boessenkool <segher@kernel.crashing.org>
13661 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
13662 handle rs6000_single_float and rs6000_double_float specially for
13665 2018-02-16 Jeff Law <law@redhat.com>
13667 * config/rx/rx.c (add_pop_cfi_notes): New function.;
13668 (pop_regs): Use it.
13670 2018-02-16 Jakub Jelinek <jakub@redhat.com>
13673 * ipa-inline.c (inline_small_functions): Fix a typo.
13675 2018-02-16 Nathan Sidwell <nathan@acm.org>
13677 * doc/extend.texi (Backwards Compatibility): Americanize 'behaviour'.
13679 2018-02-16 Carl Love <cel@us.ibm.com>
13681 * config/rs6000/rs6000-builtin.def: Add BU_P8V_VSX_2 macro definition.
13682 Change FLOAT2 expansions from BU_VSX_2 to BU_P8V_VSX_2 and
13683 from BU_VSX_OVERLOAD_2 to BU_P8V_OVERLOAD_2.
13684 * config/rs6000/rs6000-c.c: Changed macro VSX_BUILTIN_VEC_FLOAT2
13685 expansion to P8V_BUILTIN_VEC_FLOAT2.
13687 2018-02-16 Vladimir Makarov <vmakarov@redhat.com>
13689 PR rtl-optimization/70023
13690 * lra-constraints.c (inherit_in_ebb): Take hard reg mode of
13691 src_regno into account.
13693 2018-02-16 Carl Love <cel@us.ibm.com>
13695 * config/rs6000/altivec.h: Remove vec_vextract4b and vec_vinsert4b.
13696 * config/rs6000/rs6000-builtin.def: Remove macro expansion for
13697 VEXTRACT4B, VINSERT4B, VINSERT4B_DI and VEXTRACT4B.
13698 * config/rs6000/rs6000.c: Remove case statements for
13699 P9V_BUILTIN_VEXTRACT4B, P9V_BUILTIN_VEC_VEXTRACT4B,
13700 P9V_BUILTIN_VINSERT4B, P9V_BUILTIN_VINSERT4B_DI,
13701 and P9V_BUILTIN_VEC_VINSERT4B.
13702 * config/rs6000/rs6000-c.c (altivec_expand_builtin): Remove entries for
13703 P9V_BUILTIN_VEC_VEXTRACT4B and P9V_BUILTIN_VEC_VINSERT4B.
13704 * config/rs6000/vsx.md:
13705 * doc/extend.texi: Remove vec_vextract4b, non ABI definitions for
13708 2018-02-16 Carl Love <cel@us.ibm.com>
13710 * config/rs6000/altivec.h: Add builtin names vec_extract4b
13712 * config/rs6000/rs6000-builtin.def: Add INSERT4B and EXTRACT4B
13714 * config/rs6000/rs6000-c.c: Add the definitions for
13715 P9V_BUILTIN_VEC_EXTRACT4B and P9V_BUILTIN_VEC_INSERT4B.
13716 * config/rs6000/rs6000.c (altivec_expand_builtin): Add
13717 P9V_BUILTIN_EXTRACT4B and P9V_BUILTIN_INSERT4B case statements.
13718 * config/rs6000/vsx.md: Add define_insn extract4b. Add define_expand
13719 definition for insert4b and define insn *insert3b_internal.
13720 * doc/extend.texi: Add documentation for vec_extract4b.
13722 2018-02-16 Nathan Sidwell <nathan@acm.org>
13724 * doc/extend.texi (Backwards Compatibility): Mention friend
13725 injection. Note for-scope is deprecated.
13726 * doc/invoke.texi (-ffriend-injection): Deprecate.
13728 2018-02-16 Segher Boessenkool <segher@kernel.crashing.org>
13730 * combine.c (try_combine): When adjusting LOG_LINKS for the destination
13731 that moved to I2, also allow destinations that are a paradoxical
13732 subreg (instead of a normal reg).
13734 2018-02-16 Oleg Endo <olegendo@gcc.gnu.org>
13737 * config/rx/rx.c (rx_fuse_in_memory_bitop): Convert shift operand
13740 2018-02-16 Richard Biener <rguenther@suse.de>
13742 PR tree-optimization/84037
13743 PR tree-optimization/84016
13745 * config/i386/i386.c (ix86_builtin_vectorization_cost):
13746 Adjust vec_construct for the fact we need additional higher latency
13747 128bit inserts for AVX256 and AVX512 vector builds.
13748 (ix86_add_stmt_cost): Scale vector construction cost for
13751 2018-02-16 Richard Biener <rguenther@suse.de>
13753 PR tree-optimization/84417
13754 * tree-ssa.c (non_rewritable_mem_ref_base): Properly constrain
13755 the MEM_REF offset when conversion to BIT_FIELD_REF is desired.
13756 (non_rewritable_lvalue_p): Likewise, use poly-ints.
13758 2018-02-16 Martin Liska <mliska@suse.cz>
13761 * internal-fn.def (ASAN_CHECK): Set proper flags.
13762 (ASAN_MARK): Likewise.
13764 2018-02-16 Julia Koval <julia.koval@intel.com>
13766 * config/i386/i386.c (ix86_option_override_internal): Remove PTA_CLWB
13767 from PTA_CANNONLAKE.
13769 2018-02-16 Jakub Jelinek <jakub@redhat.com>
13772 * config/aarch64/cortex-a57-fma-steering.c (fma_forest::merge_forest):
13773 Use ++iter rather than iter++ for std::list iterators.
13774 (func_fma_steering::dfs): Likewise. Don't delete nodes right away,
13775 defer deleting them until all nodes in the forest are processed. Do
13776 free even leaf nodes. Change to_process into auto_vec.
13779 * system.h (BROKEN_VALUE_INITIALIZATION): Define for GCC < 4.3.
13780 * vec.h (vec_default_construct): Use memset instead of placement new
13781 if BROKEN_VALUE_INITIALIZATION is defined.
13782 * hash-table.h (hash_table<Descriptor, Allocator>::empty_slow): Use
13783 memset instead of value initialization if BROKEN_VALUE_INITIALIZATION
13786 PR rtl-optimization/83723
13787 * lra-int.h (lra_substitute_pseudo): Add DEBUG_P argument.
13788 * lra.c (lra_substitute_pseudo): Likewise. If true, use
13789 gen_rtx_raw_SUBREG instead of gen_rtx_SUBREG. Pass DEBUG_P to
13791 (lra_substitute_pseudo_within_insn): Adjust lra_substitute_pseudo
13793 * lra-constraints.c (inherit_reload_reg, split_reg): Likewise.
13795 2018-02-16 Eric Botcazou <ebotcazou@adacore.com>
13797 PR rtl-optimization/81443
13798 * rtlanal.c (num_sign_bit_copies1) <SUBREG>: Do not propagate results
13799 from inner REGs to paradoxical SUBREGs.
13801 2018-02-16 Richard Biener <rguenther@suse.de>
13803 PR tree-optimization/84399
13804 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
13805 For operands we can analyze at their definition make sure we can
13806 analyze them at each use as well.
13808 2018-02-16 Richard Biener <rguenther@suse.de>
13810 PR tree-optimization/84190
13811 * tree-ssa.c (non_rewritable_mem_ref_base): Do not touch
13812 volatile accesses if the decl isn't volatile.
13814 2018-02-15 Jason Merrill <jason@redhat.com>
13816 PR c++/84314 - ICE with templates and fastcall attribute.
13817 * attribs.c (build_type_attribute_qual_variant): Don't clobber
13818 TYPE_CANONICAL on an existing type.
13820 2018-02-15 Jakub Jelinek <jakub@redhat.com>
13822 PR tree-optimization/84383
13823 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Don't look at
13824 dstoff nor call operand_equal_p if dstbase is NULL.
13826 PR tree-optimization/84334
13827 * match.pd ((A +- CST1) +- CST2 -> A + CST3): If A is
13828 also a CONSTANT_CLASS_P, punt.
13830 2018-02-14 Jim Wilson <jimw@sifive.com>
13832 * config/riscv/riscv.c (riscv_first_stack_step): Move locals after
13833 first SMALL_OPERAND check. New local min_second_step. Move assert
13834 to where locals are set. Add TARGET_RVC support.
13835 * config/riscv/riscv.h (C_SxSP_BITS, SWSP_REACH, SDSP_REACH): New.
13837 2018-02-14 Indu Bhagat <indu.bhagat@oracle.com>
13839 * doc/invoke.texi: Correct -Wformat-overflow code sample.
13841 2018-02-14 Martin Sebor <msebor@redhat.com>
13843 PR tree-optimization/83698
13844 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): For
13845 arrays constrain the offset range to their bounds.
13846 (builtin_access::strcat_overlap): Adjust the bounds of overlap offset.
13847 (builtin_access::overlap): Avoid setting the size of overlap if it's
13849 (maybe_diag_overlap): Also consider arrays when deciding what values
13850 of offsets to include in diagnostics.
13852 2018-02-14 Martin Sebor <msebor@redhat.com>
13855 * attribs.c (diag_attr_exclusions): Consider the exclusion(s)
13856 that correspond to the kind of a declaration.
13858 2018-02-14 John David Anglin <danglin@gcc.gnu.org>
13861 * config/pa/pa.md: Load address of PIC label using the linkage table
13862 if the label is nonlocal.
13864 2018-02-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
13866 * config/rs6000/rs6000.c (rs6000_option_override_internal): Issue
13867 warning message if user requests -maltivec=be.
13868 * doc/invoke.texi: Document deprecation of -maltivec=be.
13870 2018-02-14 Will Schmidt <will_schmidt@vnet.ibm.com>
13873 * config/rs6000/rs6000-c.c: Update definitions for
13874 ALTIVEC_BUILTIN_VEC_SLD, ALTIVEC_BUILTIN_VEC_SLDW,
13875 VEC_XXSLDWI and ALTIVEC_BUILTIN_VEC_XXPERMDI builtins.
13877 2018-02-14 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
13880 * config/i386/cetintrin.h: Remove _rdssp[d|q] and
13881 add _get_ssp intrinsics. Remove argument from
13882 __builtin_ia32_rdssp[d|q].
13883 * config/i386/i386-builtin-types.def: Add UINT_FTYPE_VOID.
13884 * config/i386/i386-builtin.def: Remove argument from
13885 __builtin_ia32_rdssp[d|q].
13886 * config/i386/i386.c: Use UINT_FTYPE_VOID. Use
13887 ix86_expand_special_args_builtin for _rdssp[d|q].
13888 * config/i386/i386.md: Remove argument from rdssp[si|di] insn.
13889 Clear register before usage.
13890 * doc/extend.texi: Remove argument from __builtin_ia32_rdssp[d|q].
13891 Add documentation for new _get_ssp and _inc_ssp intrinsics.
13893 2018-02-14 Richard Sandiford <richard.sandiford@linaro.org>
13895 PR tree-optimization/84357
13896 * tree-data-ref.c (object_address_invariant_in_loop_p): Check
13897 operand 1 of an ARRAY_REF too.
13899 2018-02-14 Oleg Endo <olegendo@gcc.gnu.org>
13902 * config/rx/rx-protos.h (rx_reg_dead_or_unused_after_insn,
13903 rx_copy_reg_dead_or_unused_notes, rx_fuse_in_memory_bitop): New
13905 (set_of_reg): New struct.
13906 (rx_find_set_of_reg, rx_find_use_of_reg): New functions.
13907 * config/rx/rx.c (rx_reg_dead_or_unused_after_insn,
13908 rx_copy_reg_dead_or_unused_notes, rx_fuse_in_memory_bitop): New
13910 * config/rx/rx.md (andsi3, iorsi3, xorsi3): Convert to insn_and_split.
13911 Split into bitclr, bitset, bitinvert patterns if appropriate.
13912 (*bitset, *bitinvert, *bitclr): Convert to named insn_and_split and
13913 use rx_fuse_in_memory_bitop.
13914 (*bitset_in_memory, *bitinvert_in_memory, *bitclr_in_memory): Convert
13915 to named insn, correct maximum insn length.
13917 2018-02-14 Jozef Lawrynowicz <jozefl.gcc@gmail.com>
13920 * machmode.def: Define a complex mode for PARTIAL_INT.
13921 * genmodes.c (complex_class): Return MODE_COMPLEX_INT for
13923 * doc/rtl.texi: Document CSPImode.
13924 * config/msp430/msp430.c (msp430_hard_regno_nregs): Add CPSImode
13926 (msp430_hard_regno_nregs_with_padding): Likewise.
13928 2018-02-13 Peter Bergner <bergner@vnet.ibm.com>
13931 * config/rs6000/rs6000.c (mem_operand_gpr): Disallow altivec addresses.
13933 2018-02-13 Segher Boessenkool <segher@kernel.crashing.org>
13935 PR rtl-optimization/84169
13936 * combine.c (try_combine): New variable split_i2i3. Set it to true if
13937 we generated a parallel as new i3 and we split that to new i2 and i3
13938 instructions. Handle split_i2i3 similar to swap_i2i3: scan the
13939 LOG_LINKs of i3 to see which of those need to link to i2 now. Link
13940 those to i2, not i1. Partially rewrite this scan code.
13942 2018-02-13 Jakub Jelinek <jakub@redhat.com>
13945 * stor-layout.c (place_field): For variable length fields, adjust
13946 offset_align afterwards not just based on the field's alignment,
13947 but also on the size.
13949 PR middle-end/84309
13950 * match.pd (pow(C,x) -> exp(log(C)*x)): Use exp2s and log2s instead
13951 of exps and logs in the use_exp2 case.
13953 2018-02-13 Jeff Law <law@redhat.com>
13955 * config/rl/rl78.c (rl78_attribute_table): Fix terminator and
13956 entry for "vector".
13958 * config/rl78/rl78.c (rl78_handle_func_attribute): Mark
13961 2018-02-13 Alexandre Oliva <aoliva@redhat.com>
13965 * common.opt (gas-loc-support, gas-locview-support): New.
13966 (ginline-points, ginternal-reset-location-views): New.
13967 * doc/invoke.texi: Document them. Use @itemx where intended.
13968 (gvariable-location-views): Adjust.
13969 * target.def (reset_location_view): New.
13970 * doc/tm.texi.in (DWARF2_ASM_VIEW_DEBUG_INFO): New.
13971 (TARGET_RESET_LOCATION_VIEW): New.
13972 * doc/tm.texi: Rebuilt.
13973 * dwarf2out.c (dwarf2out_default_as_loc_support): New.
13974 (dwarf2out_default_as_locview_support): New.
13975 (output_asm_line_debug_info): Use option variables.
13976 (dwarf2out_maybe_output_loclist_view_pair): Likewise.
13977 (output_loc_list): Likewise.
13978 (add_high_low_attributes): Check option variables.
13979 Don't output entry view attribute in strict mode.
13980 (gen_inlined_subroutine_die): Check option variables.
13981 (dwarf2out_inline_entry): Likewise.
13982 (init_sections_and_labels): Likewise.
13983 (dwarf2out_early_finish): Likewise.
13984 (maybe_reset_location_view): New, from...
13985 (dwarf2out_var_location): ... here. Call it.
13986 * debug.h (dwarf2out_default_as_loc_support): Declare.
13987 (dwarf2out_default_as_locview_support): Declare.
13988 * hooks.c (hook_int_rtx_insn_0): New.
13989 * hooks.h (hook_int_rtx_insn_0): Declare.
13990 * toplev.c (process_options): Take -gas-loc-support and
13991 -gas-locview-support from dwarf2out. Enable
13992 -gvariable-location-views by default only with locview
13993 assembler support. Enable -ginternal-reset-location-views by
13994 default only if the target defines the corresponding hook.
13995 Enable -ginline-points by default if location views are
13996 enabled; force it disabled if statement frontiers are
13998 * tree-inline.c (expand_call_inline): Check option variables.
13999 * tree-ssa-live.c (remove_unused_scope_block_p): Likewise.
14001 2018-02-13 Richard Sandiford <richard.sandiford@linaro.org>
14003 PR tree-optimization/84321
14004 * tree-vrp.c (intersect_range_with_nonzero_bits): Fix VR_ANTI_RANGE
14005 handling. Also check whether the anti-range contains any values
14006 that satisfy the mask; switch to a VR_RANGE if not.
14008 2018-02-13 Paolo Bonzini <bonzini@gnu.org>
14011 * internal-fn.def (ASAN_CHECK, ASAN_MARK): Revert changes to fnspec.
14013 2018-02-13 Martin Jambor <mjambor@suse.cz>
14016 * ipa-param-manipulation.c (ipa_modify_call_arguments): Use location
14017 of call statements, also set location of a load to a temporary.
14019 2018-02-13 Sebastian Perta <sebastian.perta@renesas.com>
14021 * config/rl78/rl78.c (add_vector_labels): New function.
14022 * config/rl78/rl78.c (rl78_handle_vector_attribute): New function.
14023 * config/rl78/rl78.c (rl78_start_function): Call add_vector_labels.
14024 * config/rl78/rl78.c (rl78_handle_func_attribute): Removed the assert
14025 which checks that no arguments are passed.
14026 * config/rl78/rl78.c (rl78_attribute_table): Add "vector" attribute.
14027 * doc/extend.texi: Documentation for the new attribute.
14029 2018-02-13 Andreas Schwab <schwab@suse.de>
14031 * config/riscv/linux.h (CPP_SPEC): Define.
14033 2018-02-13 Jakub Jelinek <jakub@redhat.com>
14036 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Pass
14037 OPTION_MASK_ISA_AES | OPTION_MASK_ISA_SSE2 instead of
14038 OPTION_MASK_ISA_AES as first argument to def_builtin_const
14039 for AES builtins. Pass OPTION_MASK_ISA_PCLMUL | OPTION_MASK_ISA_SSE2
14040 instead of OPTION_MASK_ISA_PCLMUL as first argument to
14041 def_builtin_const for __builtin_ia32_pclmulqdq128 builtin.
14042 * config/i386/wmmintrin.h: If __SSE2__ is not defined, enable it
14043 temporarily for AES and PCLMUL builtins.
14045 PR tree-optimization/84339
14046 * gimple-fold.c (get_range_strlen): Set *FLEXP to true when handling
14047 ARRAY_REF where first operand is array_at_struct_end_p COMPONENT_REF.
14050 PR middle-end/84309
14051 * match.pd (pow(C,x) -> exp(log(C)*x)): Optimize instead into
14052 exp2(log2(C)*x) if C is a power of 2 and c99 runtime is available.
14053 * generic-match-head.c (canonicalize_math_after_vectorization_p): New
14055 * gimple-match-head.c (canonicalize_math_after_vectorization_p): New
14057 * omp-simd-clone.h: New file.
14058 * omp-simd-clone.c: Include omp-simd-clone.h.
14059 (expand_simd_clones): No longer static.
14060 * tree-vect-patterns.c: Include fold-const-call.h, attribs.h,
14061 cgraph.h and omp-simd-clone.h.
14062 (vect_recog_pow_pattern): Optimize pow(C,x) to exp(log(C)*x).
14063 (vect_recog_widen_shift_pattern): Formatting fix.
14064 (vect_pattern_recog_1): Don't check optab for calls.
14067 * config/i386/sse.md (<avx512>_vpermi2var<mode>3_mask): Force
14068 operands[2] into a REG before using gen_lowpart on it.
14070 2018-02-12 Jeff Law <law@redhat.com>
14073 * config/sh/sh.c (find_barrier): Consider a sibling call
14076 * cse.c (try_back_substitute_reg): Move any REG_ARGS_SIZE note when
14077 successfully back substituting a reg.
14079 2018-02-12 Richard Biener <rguenther@suse.de>
14081 PR tree-optimization/84037
14082 * tree-vect-slp.c (vect_analyze_slp_cost): Add visited
14083 parameter, move visited init to caller.
14084 (vect_slp_analyze_operations): Separate cost from validity
14085 check, initialize visited once for all instances.
14086 (vect_schedule_slp): Analyze map to CSE vectorized nodes once
14088 * tree-vect-stmts.c (vect_model_simple_cost): Make early
14090 (vect_model_promotion_demotion_cost): Likewise.
14091 (vectorizable_bswap): Guard cost modeling with !slp_node
14092 instead of !PURE_SLP_STMT to avoid double-counting on hybrid
14094 (vectorizable_call): Likewise.
14095 (vectorizable_conversion): Likewise.
14096 (vectorizable_assignment): Likewise.
14097 (vectorizable_shift): Likewise.
14098 (vectorizable_operation): Likewise.
14099 (vectorizable_store): Likewise.
14100 (vectorizable_load): Likewise.
14101 (vectorizable_condition): Likewise.
14102 (vectorizable_comparison): Likewise.
14104 2018-02-12 Paolo Bonzini <bonzini@gnu.org>
14107 * internal-fn.def (ASAN_CHECK): Fix fnspec to account for return value.
14108 (ASAN_MARK): Fix fnspec to account for return value, change pointer
14109 argument from 'R' to 'W' so that the pointed-to datum is clobbered.
14111 2018-02-08 Jan Hubicka <hubicka@ucw.cz>
14113 PR middle-end/83665
14114 * params.def (inline-min-speedup): Increase from 8 to 15.
14115 (max-inline-insns-auto): Decrease from 40 to 30.
14116 * ipa-split.c (consider_split): Add some buffer for function to
14117 be considered inlining candidate.
14118 * invoke.texi (max-inline-insns-auto, inline-min-speedup): UPdate
14121 2018-02-12 Richard Biener <rguenther@suse.de>
14123 PR tree-optimization/84037
14124 * tree-vect-slp.c (vect_build_slp_tree_2): Try swapping the
14125 matched stmts if we cannot swap the non-matched ones.
14127 2018-02-12 Olga Makhotina <olga.makhotina@intel.com>
14129 * config/i386/avx512fintrin.h (_mm_mask_scalef_round_sd,
14130 _mm_maskz_scalef_round_sd, _mm_mask_scalef_round_ss,
14131 _mm_maskz_scalef_round_ss): New intrinsics.
14132 (__builtin_ia32_scalefsd_round, __builtin_ia32_scalefss_round): Fix.
14133 * config/i386/i386-builtin.def (__builtin_ia32_scalefsd_round,
14134 __builtin_ia32_scalefss_round): Remove.
14135 (__builtin_ia32_scalefsd_mask_round,
14136 __builtin_ia32_scalefss_mask_round): New intrinsics.
14137 * config/i386/sse.md (vmscalef<mode><round_name>): Renamed to ...
14138 (vmscalef<mode><mask_scalar_name><round_scalar_name>): ... this.
14139 ((match_operand:VF_128 2 "<round_nimm_predicate>"
14140 "<round_constraint>")): Changed to ...
14141 ((match_operand:VF_128 2 "<round_scalar_nimm_predicate>"
14142 "<round_scalar_constraint>")): ... this.
14143 ("vscalef<ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|
14144 %0, %1, %2<round_op3>}"): Changed to ...
14145 ("vscalef<ssescalarmodesuffix>\t{<round_scalar_mask_op3>%2, %1,
14146 %0<mask_scalar_operand3>|%0<mask_scalar_operand3>, %1,
14147 %2<round_scalar_mask_op3>}"): ... this.
14148 * config/i386/subst.md (round_scalar_nimm_predicate): New.
14150 2018-02-12 Olga Makhotina <olga.makhotina@intel.com>
14152 * config/i386/avx512fintrin.h (_mm_mask_sqrt_round_sd)
14153 (_mm_maskz_sqrt_round_sd, _mm_mask_sqrt_round_ss)
14154 (_mm_maskz_sqrt_round_ss): New intrinsics.
14155 (__builtin_ia32_sqrtsd_round, __builtin_ia32_sqrtss_round): Remove.
14156 (__builtin_ia32_sqrtsd_mask_round)
14157 (__builtin_ia32_sqrtss_mask_round): New builtins.
14158 * config/i386/i386-builtin.def (__builtin_ia32_sqrtsd_round)
14159 (__builtin_ia32_sqrtss_round): Remove.
14160 (__builtin_ia32_sqrtsd_mask_round)
14161 (__builtin_ia32_sqrtss_mask_round): New builtins.
14162 * config/i386/sse.md (vmsqrt<mode>2<round_name>): Renamed to ...
14163 (vmsqrt<mode>2<mask_scalar_name><round_scalar_name>): ... this.
14164 ((match_operand:VF_128 1 "vector_operand"
14165 "xBm,<round_constraint>")): Changed to ...
14166 ((match_operand:VF_128 1 "vector_operand"
14167 "xBm,<round_scalar_constraint>")): ... this.
14168 (vsqrt<ssescalarmodesuffix>\t{<round_op3>%1, %2, %0|
14169 %0, %2, %<iptr>1<round_op3>}): Changed to ...
14170 (vsqrt<ssescalarmodesuffix>\t{<round_scalar_mask_op3>%1, %2,
14171 %0<mask_scalar_operand3>|%0<mask_scalar_operand3>, %2,
14172 %<iptr>1<round_scalar_mask_op3>}): ... this.
14173 ((set_attr "prefix" "<round_prefix>")): Changed to ...
14174 ((set_attr "prefix" "<round_scalar_prefix>")): ... this.
14176 2018-02-11 Steven Munroe <munroesj@gcc.gnu.org>
14179 * config/rs6000/mmintrin.h (_mm_cmpeq_pi32 [_ARCH_PWR9]):
14180 Cast vec_cmpeq result to correct type.
14181 * config/rs6000/mmintrin.h (_mm_cmpgt_pi32 [_ARCH_PWR9]):
14182 Cast vec_cmpgt result to correct type.
14184 2018-02-11 Alexandre Oliva <aoliva@redhat.com>
14186 * final.c (final_scan_insn_1): Renamed from...
14187 (final_scan_insn): ... this. New wrapper, to recover
14188 seen from the outermost call in recursive ones.
14189 * config/sparc/sparc.c (output_return): Drop seen from call.
14190 (output_sibcall): Likewise.
14191 * config/visium/visium.c (output_branch): Likewise.
14193 2018-02-10 John David Anglin <danglin@gcc.gnu.org>
14195 * config/pa/pa.c (hppa_profile_hook): Mark SYMBOL_REF for _mcount as
14198 2018-02-10 Alan Modra <amodra@gmail.com>
14201 * config/rs6000/rs6000.md (split_stack_return): Remove (use ..).
14202 Specify LR as an input.
14204 2018-02-10 Jakub Jelinek <jakub@redhat.com>
14207 * omp-low.c (maybe_remove_omp_member_access_dummy_vars,
14208 remove_member_access_dummy_vars): New functions.
14209 (lower_omp_for, lower_omp_taskreg, lower_omp_target,
14210 lower_omp_1, execute_lower_omp): Use them.
14212 PR rtl-optimization/84308
14213 * shrink-wrap.c (spread_components): Release todo vector.
14215 2018-02-09 Vladimir Makarov <vmakarov@redhat.com>
14217 PR rtl-optimization/57193
14218 * ira-color.c (struct allocno_color_data): Add member
14219 conflict_allocno_hard_prefs.
14220 (update_conflict_allocno_hard_prefs): New.
14221 (bucket_allocno_compare_func): Add a preference based on
14222 conflict_allocno_hard_prefs.
14223 (push_allocno_to_stack): Update conflict_allocno_hard_prefs.
14224 (color_allocnos): Remove a dead code. Initiate
14225 conflict_allocno_hard_prefs. Call update_costs_from_prefs.
14227 2018-02-09 Jakub Jelinek <jakub@redhat.com>
14230 * config/rs6000/vsx.md (p9_xxbrq_v16qi): Change input operand
14231 constraint from =wa to wa. Avoid a subreg on the output operand,
14232 instead use a pseudo and subreg it in a move.
14233 (p9_xxbrd_<mode>): Changed to ...
14234 (p9_xxbrd_v2di): ... this insn, without VSX_D iterator.
14235 (p9_xxbrd_v2df): New expander.
14236 (p9_xxbrw_<mode>): Changed to ...
14237 (p9_xxbrw_v4si): ... this insn, without VSX_W iterator.
14238 (p9_xxbrw_v4sf): New expander.
14240 2018-02-09 Sebastian Perta <sebastian.perta@renesas.com>
14242 * config/rx/rx.md (movsicc): Update expander to be matched by GCC.
14244 2018-02-09 Peter Bergner <bergner@vnet.ibm.com>
14247 * config/rs6000/vsx.md (vsx_mul_v2di): Handle generating a 64-bit
14248 multiply in 32-bit mode.
14249 (vsx_div_v2di): Handle generating a 64-bit signed divide in 32-bit mode.
14250 (vsx_udiv_v2di): Handle generating a 64-bit unsigned divide in 32-bit
14253 2018-02-09 Sebastian Perta <sebastian.perta@renesas.com>
14255 * config/rx/constraints.md (CALL_OP_SYMBOL_REF): Added new constraint
14256 to allow or block "symbol_ref" depending on the value of TARGET_JSR.
14257 * config/rx/rx.md (call_internal): Use CALL_OP_SYMBOL_REF.
14258 * config/rx/rx.md (call_value_internal): Use CALL_OP_SYMBOL_REF.
14260 2018-02-09 Pierre-Marie de Rodat <derodat@adacore.com>
14263 * dwarf2out.c (is_trivial_indirect_ref): New function.
14264 (dwarf2out_late_global_decl): Do not generate a location
14265 attribute for variables that have a non-trivial DECL_VALUE_EXPR
14266 and that are not defined in the current unit.
14268 2018-02-09 Eric Botcazou <ebotcazou@adacore.com>
14270 * optabs.c (prepare_cmp_insn): Try harder to emit a direct comparison
14271 instead of a libcall for UNORDERED.
14273 2018-02-09 Tamar Christina <tamar.christina@arm.com>
14276 * config/arm/arm-c.c (arm_cpu_builtins): Un-define __ARM_FEATURE_LDREX,
14277 __ARM_ARCH_PROFILE, __ARM_ARCH_ISA_THUMB, __ARM_FP and __ARM_NEON_FP.
14279 2018-02-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14282 * config/s390/s390.c (s390_set_current_function): Invoke
14283 s390_indirect_branch_settings also if fndecl didn't change.
14285 2018-02-09 Alexandre Oliva <aoliva@redhat.com>
14287 * config/rs6000/rs6000.md (blockage): Set length to zero.
14289 2018-02-09 Eric Botcazou <ebotcazou@adacore.com>
14291 * expr.c (optimize_bitfield_assignment_op): Remove obsolete assertion.
14293 2018-02-09 Jakub Jelinek <jakub@redhat.com>
14296 * gcc.c (STATIC_LIBASAN_LIBS, STATIC_LIBTSAN_LIBS,
14297 STATIC_LIBLSAN_LIBS, STATIC_LIBUBSAN_LIBS): Handle -static like
14301 * var-tracking.c (vt_add_function_parameter): Punt for non-onepart
14302 PARALLEL incoming that failed vt_get_decl_and_offset check.
14304 PR middle-end/84237
14305 * output.h (bss_initializer_p): Add NAMED argument, defaulted to false.
14306 * varasm.c (bss_initializer_p): Add NAMED argument, if true, ignore
14308 (get_variable_section): For decls in named .bss* sections pass true as
14309 second argument to bss_initializer_p.
14311 2018-02-09 Marek Polacek <polacek@redhat.com>
14312 Jakub Jelinek <jakub@redhat.com>
14315 * fold-const.c (fold_indirect_ref_1): Use VECTOR_TYPE_P macro.
14316 Formatting fixes. Verify first that tree_fits_poly_int64_p (op01).
14317 Sync some changes from cxx_fold_indirect_ref.
14319 2018-02-09 Alexandre Oliva <aoliva@redhat.com>
14321 * cfgexpand.c (expand_gimple_basic_block): Handle inline entry
14323 * dwarf2out.c (dwarf2_debug_hooks): Enable inline_entry hook.
14324 (BLOCK_INLINE_ENTRY_LABEL): New.
14325 (dwarf2out_var_location): Disregard inline entry markers.
14326 (inline_entry_data): New struct.
14327 (inline_entry_data_hasher): New hashtable type.
14328 (inline_entry_data_hasher::hash): New.
14329 (inline_entry_data_hasher::equal): New.
14330 (inline_entry_data_table): New variable.
14331 (add_high_low_attributes): Add DW_AT_entry_pc and
14332 DW_AT_GNU_entry_view attributes if a pending entry is found
14333 in inline_entry_data_table. Add old entry_pc attribute only
14334 if debug nonbinding markers are disabled.
14335 (gen_inlined_subroutine_die): Set BLOCK_DIE if nonbinding
14336 markers are enabled.
14337 (block_within_block_p, dwarf2out_inline_entry): New.
14338 (dwarf2out_finish): Check that no entries remained in
14339 inline_entry_data_table.
14340 * final.c (reemit_insn_block_notes): Handle inline entry notes.
14341 (final_scan_insn, notice_source_line): Likewise.
14342 (rest_of_clean_state): Skip inline entry markers.
14343 * gimple-pretty-print.c (dump_gimple_debug): Handle inline entry
14345 * gimple.c (gimple_build_debug_inline_entry): New.
14346 * gimple.h (enum gimple_debug_subcode): Add
14347 GIMPLE_DEBUG_INLINE_ENTRY.
14348 (gimple_build_debug_inline_entry): Declare.
14349 (gimple_debug_inline_entry_p): New.
14350 (gimple_debug_nonbind_marker_p): Adjust.
14351 * insn-notes.def (INLINE_ENTRY): New.
14352 * print-rtl.c (rtx_writer::print_rtx_operand_code_0): Handle
14353 inline entry marker notes.
14354 (print_insn): Likewise.
14355 * rtl.h (NOTE_MARKER_P): Add INLINE_ENTRY support.
14356 (INSN_DEBUG_MARKER_KIND): Likewise.
14357 (GEN_RTX_DEBUG_MARKER_INLINE_ENTRY_PAT): New.
14358 * tree-inline.c (expand_call_inline): Build and insert
14359 debug_inline_entry stmt.
14360 * tree-ssa-live.c (remove_unused_scope_block_p): Preserve
14361 inline entry blocks early, if nonbind markers are enabled.
14362 (dump_scope_block): Dump fragment info.
14363 * var-tracking.c (reemit_marker_as_note): Handle inline entry note.
14364 * doc/gimple.texi (gimple_debug_inline_entry_p): New.
14365 (gimple_build_debug_inline_entry): New.
14366 * doc/invoke.texi (gstatement-frontiers, gno-statement-frontiers):
14367 Enable/disable inline entry points too.
14368 * doc/rtl.texi (NOTE_INSN_INLINE_ENTRY): New.
14369 (DEBUG_INSN): Describe inline entry markers.
14371 * common.opt (gvariable-location-views): New.
14372 (gvariable-location-views=incompat5): New.
14373 * config.in: Rebuilt.
14374 * configure: Rebuilt.
14375 * configure.ac: Test assembler for view support.
14376 * dwarf2asm.c (dw2_asm_output_symname_uleb128): New.
14377 * dwarf2asm.h (dw2_asm_output_symname_uleb128): Declare.
14378 * dwarf2out.c (var_loc_view): New typedef.
14379 (struct dw_loc_list_struct): Add vl_symbol, vbegin, vend.
14380 (dwarf2out_locviews_in_attribute): New.
14381 (dwarf2out_locviews_in_loclist): New.
14382 (dw_val_equal_p): Compare val_view_list of dw_val_class_view_lists.
14383 (enum dw_line_info_opcode): Add LI_adv_address.
14384 (struct dw_line_info_table): Add view.
14385 (RESET_NEXT_VIEW, RESETTING_VIEW_P): New macros.
14386 (DWARF2_ASM_VIEW_DEBUG_INFO): Define default.
14387 (zero_view_p): New variable.
14388 (ZERO_VIEW_P): New macro.
14389 (output_asm_line_debug_info): New.
14390 (struct var_loc_node): Add view.
14391 (add_AT_view_list, AT_loc_list): New.
14392 (add_var_loc_to_decl): Add view param. Test it against last.
14393 (new_loc_list): Add view params. Record them.
14394 (AT_loc_list_ptr): Handle loc and view lists.
14395 (view_list_to_loc_list_val_node): New.
14396 (print_dw_val): Handle dw_val_class_view_list.
14397 (size_of_die): Likewise.
14398 (value_format): Likewise.
14399 (loc_list_has_views): New.
14400 (gen_llsym): Set vl_symbol too.
14401 (maybe_gen_llsym, skip_loc_list_entry): New.
14402 (dwarf2out_maybe_output_loclist_view_pair): New.
14403 (output_loc_list): Output view list or entries too.
14404 (output_view_list_offset): New.
14405 (output_die): Handle dw_val_class_view_list.
14406 (output_dwarf_version): New.
14407 (output_compilation_unit_header): Use it.
14408 (output_skeleton_debug_sections): Likewise.
14409 (output_rnglists, output_line_info): Likewise.
14410 (output_pubnames, output_aranges): Update version comments.
14411 (output_one_line_info_table): Output view numbers in asm comments.
14412 (dw_loc_list): Determine current endview, pass it to new_loc_list.
14413 Call maybe_gen_llsym.
14414 (loc_list_from_tree_1): Adjust.
14415 (add_AT_location_description): Create view list attribute if
14416 needed, check it's absent otherwise.
14417 (convert_cfa_to_fb_loc_list): Adjust.
14418 (maybe_emit_file): Call output_asm_line_debug_info for test.
14419 (dwarf2out_var_location): Reset views as needed. Precompute
14420 add_var_loc_to_decl args. Call get_attr_min_length only if we have the
14421 attribute. Set view.
14422 (new_line_info_table): Reset next view.
14423 (set_cur_line_info_table): Call output_asm_line_debug_info for test.
14424 (dwarf2out_source_line): Likewise. Output view resets and labels to
14425 the assembler, or select appropriate line info opcodes.
14426 (prune_unused_types_walk_attribs): Handle dw_val_class_view_list.
14427 (optimize_string_length): Catch it. Adjust.
14428 (resolve_addr): Copy vl_symbol along with ll_symbol. Handle
14429 dw_val_class_view_list, and remove it if no longer needed.
14430 (hash_loc_list): Hash view numbers.
14431 (loc_list_hasher::equal): Compare them.
14432 (optimize_location_lists): Check whether a view list symbol is
14433 needed, and whether the locview attribute is present, and
14434 whether they match. Remove the locview attribute if no longer
14436 (index_location_lists): Call skip_loc_list_entry for test.
14437 (dwarf2out_finish): Call output_asm_line_debug_info for test.
14438 Use output_dwarf_version.
14439 * dwarf2out.h (enum dw_val_class): Add dw_val_class_view_list.
14440 (struct dw_val_node): Add val_view_list.
14441 * final.c (SEEN_NEXT_VIEW): New.
14442 (set_next_view_needed): New.
14443 (clear_next_view_needed): New.
14444 (maybe_output_next_view): New.
14445 (final_start_function): Rename to...
14446 (final_start_function_1): ... this. Take pointer to FIRST,
14447 add SEEN parameter. Emit param bindings in the initial view.
14448 (final_start_function): Reintroduce SEEN-less interface.
14449 (final): Rename to...
14450 (final_1): ... this. Take SEEN parameter. Output final pending
14451 next view at the end.
14452 (final): Reintroduce seen-less interface.
14453 (final_scan_insn): Output pending next view before switching
14454 sections or ending a block. Mark the next view as needed when
14455 outputting variable locations. Notify debug backend of section
14456 changes, and of location view changes.
14457 (rest_of_handle_final): Adjust.
14458 * toplev.c (process_options): Autodetect value for debug variable
14459 location views option. Warn on incompat5 without -gdwarf-5.
14460 * doc/invoke.texi (gvariable-location-views): New.
14461 (gvariable-location-views=incompat5): New.
14462 (gno-variable-location-views): New.
14464 2018-02-08 David Malcolm <dmalcolm@redhat.com>
14466 PR tree-optimization/84136
14467 * tree-cfg.c (find_taken_edge_computed_goto): Remove assertion
14468 that the result of find_edge is non-NULL.
14470 2018-02-08 Sergey Shalnov <sergey.shalnov@intel.com>
14473 * config/i386/x86-tune-costs.h (skylake_cost): Fix cost of
14474 storing integer register in SImode. Fix cost of 256 and 512
14475 byte aligned SSE register store.
14477 2018-02-08 Sergey Shalnov <sergey.shalnov@intel.com>
14479 * config/i386/i386.c (ix86_multiplication_cost): Fix
14480 multiplication cost for TARGET_AVX512DQ.
14482 2018-02-08 Marek Polacek <polacek@redhat.com>
14484 PR tree-optimization/84238
14485 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Verify the result of
14488 2018-02-08 Richard Sandiford <richard.sandiford@linaro.org>
14490 PR tree-optimization/84265
14491 * tree-vect-stmts.c (vectorizable_store): Don't treat
14492 VMAT_CONTIGUOUS accesses as grouped.
14493 (vectorizable_load): Likewise.
14495 2018-02-08 Richard Sandiford <richard.sandiford@linaro.org>
14497 PR tree-optimization/81635
14498 * wide-int.h (wi::round_down_for_mask, wi::round_up_for_mask): Declare.
14499 * wide-int.cc (wi::round_down_for_mask, wi::round_up_for_mask)
14500 (test_round_for_mask): New functions.
14501 (wide_int_cc_tests): Call test_round_for_mask.
14502 * tree-vrp.h (intersect_range_with_nonzero_bits): Declare.
14503 * tree-vrp.c (intersect_range_with_nonzero_bits): New function.
14504 * tree-data-ref.c (split_constant_offset_1): Use it to refine the
14505 range returned by get_range_info.
14507 2018-02-08 Jan Hubicka <hubicka@ucw.cz>
14510 * cgraph.h (symtab_node::output_to_lto_symbol_table_p): Declare
14511 * symtab.c: Include builtins.h
14512 (symtab_node::output_to_lto_symbol_table_p): Move here
14513 from lto-streamer-out.c:output_symbol_p.
14514 * lto-streamer-out.c (write_symbol): Turn early exit to assert.
14515 (output_symbol_p): Move all logic to symtab.c
14516 (produce_symtab): Update.
14518 2018-02-08 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14520 * config/s390/s390-opts.h (enum indirect_branch): Define.
14521 * config/s390/s390-protos.h (s390_return_addr_from_memory)
14522 (s390_indirect_branch_via_thunk)
14523 (s390_indirect_branch_via_inline_thunk): Add function prototypes.
14524 (enum s390_indirect_branch_type): Define.
14525 * config/s390/s390.c (struct s390_frame_layout, struct
14526 machine_function): Remove.
14527 (indirect_branch_prez10thunk_mask, indirect_branch_z10thunk_mask)
14528 (indirect_branch_table_label_no, indirect_branch_table_name):
14530 (INDIRECT_BRANCH_NUM_OPTIONS): Define macro.
14531 (enum s390_indirect_branch_option): Define.
14532 (s390_return_addr_from_memory): New function.
14533 (s390_handle_string_attribute): New function.
14534 (s390_attribute_table): Add new attribute handler.
14535 (s390_execute_label): Handle UNSPEC_EXECUTE_JUMP patterns.
14536 (s390_indirect_branch_via_thunk): New function.
14537 (s390_indirect_branch_via_inline_thunk): New function.
14538 (s390_function_ok_for_sibcall): When jumping via thunk disallow
14539 sibling call optimization for non z10 compiles.
14540 (s390_emit_call): Force indirect branch target to be a single
14541 register. Add r1 clobber for non-z10 compiles.
14542 (s390_emit_epilogue): Emit return jump via return_use expander.
14543 (s390_reorg): Handle JUMP_INSNs as execute targets.
14544 (s390_option_override_internal): Perform validity checks for the
14545 new command line options.
14546 (s390_indirect_branch_attrvalue): New function.
14547 (s390_indirect_branch_settings): New function.
14548 (s390_set_current_function): Invoke s390_indirect_branch_settings.
14549 (s390_output_indirect_thunk_function): New function.
14550 (s390_code_end): Implement target hook.
14551 (s390_case_values_threshold): Implement target hook.
14552 (TARGET_ASM_CODE_END, TARGET_CASE_VALUES_THRESHOLD): Define target
14554 * config/s390/s390.h (struct s390_frame_layout)
14555 (struct machine_function): Move here from s390.c.
14556 (TARGET_INDIRECT_BRANCH_NOBP_RET)
14557 (TARGET_INDIRECT_BRANCH_NOBP_JUMP)
14558 (TARGET_INDIRECT_BRANCH_NOBP_JUMP_THUNK)
14559 (TARGET_INDIRECT_BRANCH_NOBP_JUMP_INLINE_THUNK)
14560 (TARGET_INDIRECT_BRANCH_NOBP_CALL)
14561 (TARGET_DEFAULT_INDIRECT_BRANCH_TABLE)
14562 (TARGET_INDIRECT_BRANCH_THUNK_NAME_EXRL)
14563 (TARGET_INDIRECT_BRANCH_THUNK_NAME_EX)
14564 (TARGET_INDIRECT_BRANCH_TABLE): Define macros.
14565 * config/s390/s390.md (UNSPEC_EXECUTE_JUMP)
14566 (INDIRECT_BRANCH_THUNK_REGNUM): Define constants.
14567 (mnemonic attribute): Add values which aren't recognized
14569 ("*cjump_long", "*icjump_long", "*basr", "*basr_r"): Disable
14570 pattern for branch conversion. Fix mnemonic attribute.
14571 ("*c<code>", "*sibcall_br", "*sibcall_value_br", "*return"): Emit
14572 indirect branch via thunk if requested.
14573 ("indirect_jump", "<code>"): Expand patterns for branch conversion.
14574 ("*indirect_jump"): Disable for branch conversion using out of
14576 ("indirect_jump_via_thunk<mode>_z10")
14577 ("indirect_jump_via_thunk<mode>")
14578 ("indirect_jump_via_inlinethunk<mode>_z10")
14579 ("indirect_jump_via_inlinethunk<mode>", "*casesi_jump")
14580 ("casesi_jump_via_thunk<mode>_z10", "casesi_jump_via_thunk<mode>")
14581 ("casesi_jump_via_inlinethunk<mode>_z10")
14582 ("casesi_jump_via_inlinethunk<mode>", "*basr_via_thunk<mode>_z10")
14583 ("*basr_via_thunk<mode>", "*basr_r_via_thunk_z10")
14584 ("*basr_r_via_thunk", "return<mode>_prez10"): New pattern.
14585 ("*indirect2_jump"): Disable for branch conversion.
14586 ("casesi_jump"): Turn into expander and expand patterns for branch
14588 ("return_use"): New expander.
14589 ("*return"): Emit return via thunk and rename it to ...
14590 ("*return<mode>"): ... this one.
14591 * config/s390/s390.opt: Add new options and and enum for the
14594 2018-02-08 Richard Sandiford <richard.sandiford@linaro.org>
14596 * lra-constraints.c (match_reload): Unconditionally use
14597 gen_lowpart_SUBREG, rather than selecting between that
14598 and equivalent gen_rtx_SUBREG code.
14600 2018-02-08 Richard Biener <rguenther@suse.de>
14602 PR tree-optimization/84233
14603 * tree-ssa-phiprop.c (propagate_with_phi): Use separate
14604 changed flag instead of boguously re-using phi_inserted.
14606 2018-02-08 Martin Jambor <mjambor@suse.cz>
14608 * hsa-gen.c (get_symbol_for_decl): Set program allocation for
14609 static local variables.
14611 2018-02-08 Richard Biener <rguenther@suse.de>
14613 PR tree-optimization/84278
14614 * tree-vect-stmts.c (vectorizable_store): When looking for
14615 smaller vector types to perform grouped strided loads/stores
14616 make sure the mode is supported by the target.
14617 (vectorizable_load): Likewise.
14619 2018-02-08 Wilco Dijkstra <wdijkstr@arm.com>
14621 * config/aarch64/aarch64.c (aarch64_components_for_bb):
14622 Increase LDP/STP opportunities by adding adjacent callee-saves.
14624 2018-02-08 Wilco Dijkstra <wdijkstr@arm.com>
14626 PR rtl-optimization/84068
14627 PR rtl-optimization/83459
14628 * haifa-sched.c (rank_for_schedule): Fix SCHED_PRESSURE_MODEL sorting.
14630 2018-02-08 Aldy Hernandez <aldyh@redhat.com>
14632 PR tree-optimization/84224
14633 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Remove assert.
14634 * calls.c (gimple_alloca_call_p): Only return TRUE when we have
14635 non-zero arguments.
14637 2018-02-07 Iain Sandoe <iain@codesourcery.com>
14640 * config/rs6000/altivec.md (*restore_world): Remove LR use.
14641 * config/rs6000/predicates.md (restore_world_operation): Adjust op
14642 count, remove one USE.
14644 2018-02-07 Michael Meissner <meissner@linux.vnet.ibm.com>
14646 * doc/install.texi (Configuration): Document the
14647 --with-long-double-format={ibm,ieee} PowerPC configuration
14651 * config/rs6000/rs6000.md (fix_trunc<SFDF:mode><QHI:mode>2):
14652 Convert from define_expand to be define_insn_and_split. Rework
14653 float/double/_Float128 conversions to QI/HI/SImode to work with
14654 both ISA 2.07 (power8) or ISA 3.0 (power9). Fix regression where
14655 conversions to QI/HImode types did a store and then a load to
14656 truncate the value. For conversions to VSX registers, don't split
14657 the insn, instead emit the code directly. Use the code iterator
14658 any_fix to combine signed and unsigned conversions.
14659 (fix<uns>_trunc<SFDF:mode>si2_p8): Likewise.
14660 (fixuns_trunc<SFDF:mode><QHI:mode>2): Likewise.
14661 (fix_trunc<IEEE128:mode><QHI:mode>2): Likewise.
14662 (fix<uns>_trunc<SFDF:mode><QHI:mode>2): Likewise.
14663 (fix_<mode>di2_hw): Likewise.
14664 (fixuns_<mode>di2_hw): Likewise.
14665 (fix_<mode>si2_hw): Likewise.
14666 (fixuns_<mode>si2_hw): Likewise.
14667 (fix<uns>_<IEEE128:mode><SDI:mode>2_hw): Likewise.
14668 (fix<uns>_trunc<IEEE128:mode><QHI:mode>2): Likewise.
14669 (fctiw<u>z_<mode>_smallint): Rename fctiw<u>z_<mode>_smallint to
14670 fix<uns>_trunc<SFDF:mode>si2_p8.
14671 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Delete, no longer used.
14672 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
14673 (fix<uns>_<mode>_mem): Likewise.
14674 (fctiw<u>z_<mode>_mem): Likewise.
14675 (fix<uns>_<mode>_mem): Likewise.
14676 (fix<uns>_trunc<SFDF:mode><QHSI:mode>2_mem): On ISA 3.0, prevent
14677 the register allocator from doing a direct move to the GPRs to do
14678 a store, and instead use the ISA 3.0 store byte/half-word from
14679 vector register instruction. For IEEE 128-bit floating point,
14680 also optimize stores of 32-bit ints.
14681 (fix<uns>_trunc<IEEE128:mode><QHSI:mode>2_mem): Likewise.
14683 2018-02-07 Alan Hayward <alan.hayward@arm.com>
14685 * genextract.c (push_pathstr_operand): New function to support
14687 (walk_rtx): Call push_pathstr_operand.
14688 (print_path): Support [a-zA-Z].
14690 2018-02-07 Richard Biener <rguenther@suse.de>
14692 PR tree-optimization/84037
14693 * tree-vectorizer.h (struct _loop_vec_info): Add ivexpr_map member.
14694 (cse_and_gimplify_to_preheader): Declare.
14695 (vect_get_place_in_interleaving_chain): Likewise.
14696 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
14698 (_loop_vec_info::~_loop_vec_info): Delete it.
14699 (cse_and_gimplify_to_preheader): New function.
14700 * tree-vect-slp.c (vect_get_place_in_interleaving_chain): Export.
14701 * tree-vect-stmts.c (vectorizable_store): CSE base and steps.
14702 (vectorizable_load): Likewise. For grouped stores always base
14703 the IV on the first element.
14704 * tree-vect-loop-manip.c (vect_loop_versioning): Unshare versioning
14705 condition before gimplifying.
14707 2018-02-07 Jakub Jelinek <jakub@redhat.com>
14709 * tree-eh.c (operation_could_trap_helper_p): Ignore honor_trapv for
14710 *DIV_EXPR and *MOD_EXPR.
14712 2018-02-07 H.J. Lu <hongjiu.lu@intel.com>
14715 * config/i386/i386.c (ix86_option_override_internal): Mask out
14716 the CF_SET bit when checking -fcf-protection.
14718 2018-02-07 Tom de Vries <tom@codesourcery.com>
14721 * omp-expand.c (expand_oacc_collapse_init): Ensure diff_type is large
14724 2018-02-07 Richard Biener <rguenther@suse.de>
14726 PR tree-optimization/84204
14727 * tree-chrec.c (chrec_fold_plus_1): Remove size limiting in
14730 PR tree-optimization/84205
14731 * graphite-isl-ast-to-gimple.c (binary_op_to_tree): Also
14732 special-case isl_ast_op_zdiv_r.
14734 PR tree-optimization/84223
14735 * graphite-scop-detection.c (gather_bbs::before_dom_children):
14736 Only add conditions from within the region.
14737 (gather_bbs::after_dom_children): Adjust.
14739 2018-02-07 Georg-Johann Lay <avr@gjlay.de>
14742 * config/avr/avr.h (GENERAL_REGNO_P, GENERAL_REG_P): New macros.
14743 * config/avr/avr.md: Only post-reload split REG-REG moves if
14744 either register is GENERAL_REG_P.
14746 2018-02-07 Jakub Jelinek <jakub@redhat.com>
14748 PR tree-optimization/84235
14749 * tree-ssa-scopedtables.c
14750 (avail_exprs_stack::simplify_binary_operation): Fir MINUS_EXPR, punt
14751 if the subtraction is performed in floating point type where NaNs are
14752 honored. For *DIV_EXPR, punt for ALL_FRACT_MODE_Ps where we can't
14753 build 1. Formatting fix.
14755 2018-02-06 Jakub Jelinek <jakub@redhat.com>
14758 * config/i386/i386.c (rest_of_insert_endbranch): Only skip
14759 NOTE_INSN_CALL_ARG_LOCATION after a call, not anything else,
14760 and skip it regardless of bb boundaries. Use CALL_P macro,
14761 don't test INSN_P (insn) together with CALL_P or JUMP_P check
14762 unnecessarily, formatting fix.
14764 2018-02-06 Michael Collison <michael.collison@arm.com>
14766 * config/arm/thumb2.md:
14767 (*thumb2_mov_negscc): Split only if TARGET_THUMB2 && !arm_restrict_it.
14768 (*thumb_mov_notscc): Ditto.
14770 2018-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
14773 * config/rs6000/rs6000.md (su code attribute): Use "u" for
14774 unsigned_fix, not "s".
14776 2018-02-06 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
14778 * configure.ac (gcc_fn_eh_frame_ro): New function.
14779 (gcc_cv_as_cfi_directive): Check both 32 and 64-bit assembler for
14780 correct .eh_frame permissions.
14781 * configure: Regenerate.
14783 2018-02-06 Andrew Jenner <andrew@codeourcery.com>
14785 * doc/invoke.texi: Add section for the PowerPC SPE backend. Remove
14786 irrelevant options.
14788 2018-02-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14790 * config/rs6000/rs6000.c (rs6000_option_override_internal):
14791 Display warning message for -mno-speculate-indirect-jumps.
14793 2018-02-06 Andrew Jenner <andrew@codesourcery.com>
14795 * config/powerpcspe/powerpcspe.opt: (msimple-fpu, mfpu) Add
14797 * config/powerpcspe/sysv4.opt (mbit-align): Likewise.
14799 2018-02-06 Aldy Hernandez <aldyh@redhat.com>
14801 PR tree-optimization/84225
14802 * tree-eh.c (find_trapping_overflow): Only call
14803 operation_no_trapping_overflow when ANY_INTEGRAL_TYPE_P.
14805 2018-02-06 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
14808 * config/i386/i386.c: Reimplement the check of possible options
14809 -mibt/-mshstk conbination. Change error messages.
14810 * doc/invoke.texi: Fix a typo: remove extra '='.
14812 2018-02-06 Marek Polacek <polacek@redhat.com>
14814 PR tree-optimization/84228
14815 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Skip debug statements.
14817 2018-02-06 Tamar Christina <tamar.christina@arm.com>
14820 * config/arm/arm.c (arm_print_asm_arch_directives): Record already
14821 emitted arch directives.
14822 * config/arm/arm-c.c (arm_cpu_builtins): Undefine __ARM_ARCH and
14823 __ARM_FEATURE_COPROC before changing architectures.
14825 2018-02-06 Richard Biener <rguenther@suse.de>
14827 * config/i386/i386.c (print_reg): Fix typo.
14828 (ix86_loop_unroll_adjust): Do not unroll beyond the original nunroll.
14830 2018-02-06 Eric Botcazou <ebotcazou@adacore.com>
14832 * configure: Regenerate.
14834 2018-02-05 Martin Sebor <msebor@redhat.com>
14836 PR tree-optimization/83369
14837 * tree-ssa-ccp.c (pass_post_ipa_warn::execute): Use %G to print
14840 2018-02-05 Martin Liska <mliska@suse.cz>
14842 * doc/invoke.texi: Cherry-pick upstream r323995.
14844 2018-02-05 Richard Sandiford <richard.sandiford@linaro.org>
14846 * ira.c (ira_init_register_move_cost): Adjust comment.
14848 2018-02-05 Martin Liska <mliska@suse.cz>
14850 PR gcov-profile/84137
14851 * doc/gcov.texi: Fix typo in documentation.
14853 2018-02-05 Martin Liska <mliska@suse.cz>
14855 PR gcov-profile/83879
14856 * doc/gcov.texi: Document necessity of --dynamic-list-data when
14857 using dlopen functionality.
14859 2018-02-05 Olga Makhotina <olga.makhotina@intel.com>
14861 * config/i386/avx512dqintrin.h (_mm_mask_range_sd, _mm_maskz_range_sd,
14862 _mm_mask_range_round_sd, _mm_maskz_range_round_sd, _mm_mask_range_ss,
14863 _mm_maskz_range_ss, _mm_mask_range_round_ss,
14864 _mm_maskz_range_round_ss): New intrinsics.
14865 (__builtin_ia32_rangesd128_round)
14866 (__builtin_ia32_rangess128_round): Remove.
14867 (__builtin_ia32_rangesd128_mask_round,
14868 __builtin_ia32_rangess128_mask_round): New builtins.
14869 * config/i386/i386-builtin.def (__builtin_ia32_rangesd128_round,
14870 __builtin_ia32_rangess128_round): Remove.
14871 (__builtin_ia32_rangesd128_mask_round,
14872 __builtin_ia32_rangess128_mask_round): New builtins.
14873 * config/i386/sse.md (ranges<mode><round_saeonly_name>): Renamed to ...
14874 (ranges<mode><mask_scalar_name><round_saeonly_scalar_name>): ... this.
14875 ((match_operand:VF_128 2 "<round_saeonly_nimm_predicate>"
14876 "<round_saeonly_constraint>")): Changed to ...
14877 ((match_operand:VF_128 2 "<round_saeonly_scalar_nimm_predicate>"
14878 "<round_saeonly_scalar_constraint>")): ... this.
14879 ("vrange<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
14880 %0, %1, %2<round_saeonly_op4>, %3}"): Changed to ...
14881 ("vrange<ssescalarmodesuffix>\t{%3, <round_saeonly_scalar_mask_op4>%2,
14882 %1, %0<mask_scalar_operand4>|%0<mask_scalar_operand4>, %1,
14883 %2<round_saeonly_scalar_mask_op4>, %3}"): ... this.
14885 2018-02-02 Andrew Jenner <andrew@codesourcery.com>
14887 * config/powerpcspe/powerpcspe.opt: Add Undocumented to irrelevant
14889 * config/powerpcspe/powerpcspe-tables.opt (rs6000_cpu_opt_value):
14890 Remove all values except native, 8540 and 8548.
14892 2018-02-02 H.J. Lu <hongjiu.lu@intel.com>
14894 * config/i386/i386.c (ix86_output_function_return): Pass
14895 INVALID_REGNUM, instead of -1, as invalid register number to
14896 indirect_thunk_name and output_indirect_thunk.
14898 2018-02-02 Julia Koval <julia.koval@intel.com>
14900 * config.gcc: Add -march=icelake.
14901 * config/i386/driver-i386.c (host_detect_local_cpu): Detect icelake.
14902 * config/i386/i386-c.c (ix86_target_macros_internal): Handle icelake.
14903 * config/i386/i386.c (processor_costs): Add m_ICELAKE.
14904 (PTA_ICELAKE, PTA_AVX512VNNI, PTA_GFNI, PTA_VAES, PTA_AVX512VBMI2,
14905 PTA_VPCLMULQDQ, PTA_RDPID, PTA_AVX512BITALG): New.
14906 (processor_target_table): Add icelake.
14907 (ix86_option_override_internal): Handle new PTAs.
14908 (get_builtin_code_for_version): Handle icelake.
14909 (M_INTEL_COREI7_ICELAKE): New.
14910 (fold_builtin_cpu): Handle icelake.
14911 * config/i386/i386.h (TARGET_ICELAKE, PROCESSOR_ICELAKE): New.
14912 * doc/invoke.texi: Add -march=icelake.
14914 2018-02-02 Julia Koval <julia.koval@intel.com>
14916 * config/i386/i386.c (ix86_option_override_internal): Change flags type
14917 to wide_int_bitmask.
14918 * wide-int-bitmask.h: New.
14920 2018-02-02 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
14923 * config/i386/i386.md: Replace Pmode with word_mode in
14924 builtin_setjmp_setup and builtin_longjmp to support x32.
14926 2018-02-01 Peter Bergner <bergner@vnet.ibm.com>
14930 * config/rs6000/driver-rs6000.c: #include "diagnostic.h".
14932 (rs6000_supported_cpu_names): New static variable.
14933 (linux_cpu_translation_table): Likewise.
14934 (elf_platform) <cpu>: Define new static variable and use it.
14935 Translate kernel AT_PLATFORM name to canonical name if needed.
14936 Error if platform name is unknown.
14938 2018-02-01 Aldy Hernandez <aldyh@redhat.com>
14941 * config/pa/predicates.md (base14_operand): Handle E_VOIDmode.
14943 2018-02-01 Jeff Law <law@redhat.com>
14946 * config/i386/i386.c (release_scratch_register_on_entry): Add new
14947 OFFSET and RELEASE_VIA_POP arguments. Use SP+OFFSET to restore
14948 the scratch if RELEASE_VIA_POP is false.
14949 (ix86_adjust_stack_and_probe_stack_clash): Un-constify SIZE.
14950 If we have to save a temporary register, decrement SIZE appropriately.
14951 Pass new arguments to release_scratch_register_on_entry.
14952 (ix86_adjust_stack_and_probe): Likewise.
14953 (ix86_emit_probe_stack_range): Pass new arguments to
14954 release_scratch_register_on_entry.
14956 2018-02-01 Uros Bizjak <ubizjak@gmail.com>
14958 PR rtl-optimization/84157
14959 * combine.c (change_zero_ext): Use REG_P predicate in
14960 front of HARD_REGISTER_P predicate.
14962 2018-02-01 Georg-Johann Lay <avr@gjlay.de>
14964 * config/avr/avr.c (avr_option_override): Move disabling of
14965 -fdelete-null-pointer-checks to...
14966 * common/config/avr/avr-common.c (avr_option_optimization_table):
14969 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
14971 PR tree-optimization/81635
14972 * tree-data-ref.c (split_constant_offset_1): For types that
14973 wrap on overflow, try to use range info to prove that wrapping
14976 2018-02-01 Renlin Li <renlin.li@arm.com>
14979 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
14980 TAILCALL_ADDR_REGS.
14981 (aarch64_register_move_cost): Likewise.
14982 * config/aarch64/aarch64.h (reg_class): Rename CALLER_SAVE_REGS to
14983 TAILCALL_ADDR_REGS.
14984 (REG_CLASS_NAMES): Likewise.
14985 (REG_CLASS_CONTENTS): Rename CALLER_SAVE_REGS to
14986 TAILCALL_ADDR_REGS. Remove IP registers.
14987 * config/aarch64/aarch64.md (Ucs): Update register constraint.
14989 2018-02-01 Richard Biener <rguenther@suse.de>
14991 * domwalk.h (dom_walker::dom_walker): Add additional constructor
14992 for specifying RPO order and allow NULL for that.
14993 * domwalk.c (dom_walker::dom_walker): Likewise.
14994 (dom_walker::walk): Handle NULL RPO order.
14995 * tree-into-ssa.c (rewrite_dom_walker): Do not walk dom children
14997 (rewrite_update_dom_walker): Likewise.
14998 (mark_def_dom_walker): Likewise.
15000 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
15002 * config/aarch64/aarch64-protos.h (aarch64_split_sve_subreg_move)
15003 (aarch64_maybe_expand_sve_subreg_move): Declare.
15004 * config/aarch64/aarch64.md (UNSPEC_REV_SUBREG): New unspec.
15005 * config/aarch64/predicates.md (aarch64_any_register_operand): New
15007 * config/aarch64/aarch64-sve.md (mov<mode>): Optimize subreg moves
15008 that are semantically a reverse operation.
15009 (*aarch64_sve_mov<mode>_subreg_be): New pattern.
15010 * config/aarch64/aarch64.c (aarch64_maybe_expand_sve_subreg_move):
15011 (aarch64_replace_reg_mode, aarch64_split_sve_subreg_move): New
15013 (aarch64_can_change_mode_class): For big-endian, forbid changes
15014 between two SVE modes if they have different element sizes.
15016 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
15018 * config/aarch64/aarch64.c (aarch64_expand_sve_const_vector): Prefer
15019 the TImode handling for big-endian targets.
15021 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
15023 * config/aarch64/aarch64-sve.md (sve_ld1rq): Replace with...
15024 (*sve_ld1rq<Vesize>): ... this new pattern. Handle all element sizes,
15026 * config/aarch64/aarch64.c (aarch64_expand_sve_widened_duplicate):
15027 Remove BSWAP handing for big-endian targets and use the form of
15028 LD1RQ appropariate for the mode.
15030 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
15032 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Handle
15033 all CONST_VECTOR_DUPLICATE_P vectors, not just those with a single
15034 duplicated element.
15036 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
15039 * config/aarch64/aarch64.c (aarch64_secondary_reload): Tighten
15040 check for operands that need to go through aarch64_sve_reload_be.
15042 2018-02-01 Jakub Jelinek <jakub@redhat.com>
15044 PR tree-optimization/81661
15045 PR tree-optimization/84117
15046 * tree-eh.h (rewrite_to_non_trapping_overflow): Declare.
15047 * tree-eh.c: Include gimplify.h.
15048 (find_trapping_overflow, replace_trapping_overflow,
15049 rewrite_to_non_trapping_overflow): New functions.
15050 * tree-vect-loop.c: Include tree-eh.h.
15051 (vect_get_loop_niters): Use rewrite_to_non_trapping_overflow.
15052 * tree-data-ref.c: Include tree-eh.h.
15053 (get_segment_min_max): Use rewrite_to_non_trapping_overflow.
15055 2018-01-31 Uros Bizjak <ubizjak@gmail.com>
15057 PR rtl-optimization/84123
15058 * combine.c (change_zero_ext): Check if hard register satisfies
15059 can_change_dest_mode before calling gen_lowpart_SUBREG.
15061 2018-01-31 Vladimir Makarov <vmakarov@redhat.com>
15064 * ira.c (ira_init_register_move_cost): Remove assert.
15066 2018-01-31 Eric Botcazou <ebotcazou@adacore.com>
15068 PR rtl-optimization/84071
15069 * doc/tm.texi.in (WORD_REGISTER_OPERATIONS): Add explicit case.
15070 * doc/tm.texi: Regenerate.
15072 2018-01-31 Richard Biener <rguenther@suse.de>
15074 PR tree-optimization/84132
15075 * tree-data-ref.c (analyze_miv_subscript): Properly
15076 check whether evolution_function_is_affine_multivariate_p
15077 before calling gcd_of_steps_may_divide_p.
15079 2018-01-31 Julia Koval <julia.koval@intel.com>
15082 * config/i386/i386.c (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
15083 * config/i386/i386.md (rdpid_rex64) New.
15084 (rdpid): Make 32bit only.
15086 2018-01-29 Aldy Hernandez <aldyh@redhat.com>
15089 * tree-pretty-print.c (dump_generic_node): Handle a TYPE_NAME with
15090 an IDENTIFIER_NODE for FUNCTION_TYPE's.
15092 2018-01-31 Eric Botcazou <ebotcazou@adacore.com>
15095 2018-01-12 Eric Botcazou <ebotcazou@adacore.com>
15097 * config/sparc/sparc.md (vxworks_load_got): Set the GOT register.
15099 2018-01-31 Eric Botcazou <ebotcazou@adacore.com>
15101 PR rtl-optimization/84071
15102 * combine.c (record_dead_and_set_regs_1): Record the source unmodified
15103 for a paradoxical SUBREG on a WORD_REGISTER_OPERATIONS target.
15105 2018-01-31 Claudiu Zissulescu <claziss@synopsys.com>
15107 * config/arc/arc.c (arc_handle_aux_attribute): New function.
15108 (arc_attribute_table): Add 'aux' attribute.
15109 (arc_in_small_data_p): Consider aux like variables.
15110 (arc_is_aux_reg_p): New function.
15111 (arc_asm_output_aligned_decl_local): Ignore 'aux' like variables.
15112 (arc_get_aux_arg): New function.
15113 (prepare_move_operands): Handle aux-register access.
15114 (arc_handle_aux_attribute): New function.
15115 * doc/extend.texi (ARC Variable attributes): Add subsection.
15117 2018-01-31 Claudiu Zissulescu <claziss@synopsys.com>
15119 * config/arc/arc-protos.h (arc_is_uncached_mem_p): Function proto.
15120 * config/arc/arc.c (arc_handle_uncached_attribute): New function.
15121 (arc_attribute_table): Add 'uncached' attribute.
15122 (arc_print_operand): Print '.di' flag for uncached memory
15124 (arc_in_small_data_p): Do not consider for small data the uncached
15126 (arc_is_uncached_mem_p): New function.
15127 * config/arc/predicates.md (compact_store_memory_operand): Check
15128 for uncached memory accesses.
15129 (nonvol_nonimm_operand): Likewise.
15130 * doc/extend.texi (ARC Type Attribute): New subsection.
15132 2018-01-31 Jakub Jelinek <jakub@redhat.com>
15135 * common.opt (falign-functions=, falign-jumps=, falign-labels=,
15136 falign-loops=): Add Optimization flag.
15138 2018-01-30 Jeff Law <law@redhat.com>
15141 * i386.c (ix86_adjust_stack_and_probe_stack_clash): New argument
15142 INT_REGISTERS_SAVED. Check it prior to calling
15143 get_scratch_register_on_entry.
15144 (ix86_adjust_stack_and_probe): Similarly.
15145 (ix86_emit_probe_stack_range): Similarly.
15146 (ix86_expand_prologue): Corresponding changes.
15148 2018-01-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
15151 * config/sol2.h (STARTFILE_ARCH_SPEC): Use -std=c*,
15152 -std=iso9899:199409 instead of -pedantic to select values-Xc.o.
15154 2018-01-30 Vladimir Makarov <vmakarov@redhat.com>
15157 * lra-constraints.c (curr_insn_transform): Process AND in the
15160 2018-01-30 Jakub Jelinek <jakub@redhat.com>
15162 PR rtl-optimization/83986
15163 * sched-deps.c (sched_analyze_insn): For frame related insns, add anti
15164 dependence against last_pending_memory_flush in addition to
15165 pending_jump_insns.
15167 2018-01-30 Alexandre Oliva <aoliva@redhat.com>
15169 PR tree-optimization/81611
15170 * tree-ssa-dom.c (simple_iv_increment_p): Skip intervening
15173 2018-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
15176 * config/rs6000/rs6000.c (rs6000_internal_arg_pointer): Only return
15179 2018-01-30 Richard Biener <rguenther@suse.de>
15180 Jakub Jelinek <jakub@redhat.com>
15182 PR tree-optimization/84111
15183 * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely_1): Skip
15184 inner loops added during recursion, as they don't have up-to-date
15187 2018-01-30 Jan Hubicka <hubicka@ucw.cz>
15190 * ipa-inline.c (can_inline_edge_p): Break out late tests to...
15191 (can_inline_edge_by_limits_p): ... here.
15192 (can_early_inline_edge_p, check_callers,
15193 update_caller_keys, update_callee_keys, recursive_inlining,
15194 add_new_edges_to_heap, speculation_useful_p,
15195 inline_small_functions,
15196 inline_small_functions, flatten_function,
15197 inline_to_all_callers_1): Update.
15199 2018-01-30 Jan Hubicka <hubicka@ucw.cz>
15201 * profile-count.c (profile_count::combine_with_ipa_count): Handle
15204 2018-01-30 Richard Biener <rguenther@suse.de>
15206 PR tree-optimization/83008
15207 * tree-vect-slp.c (vect_analyze_slp_cost_1): Properly cost
15208 invariant and constant vector uses in stmts when they need
15209 more than one stmt.
15211 2018-01-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
15214 * configure.ac (gcc_cv_as_shf_merge): Disable on Solaris 10/x86.
15215 * configure: Regenerate.
15217 2018-01-30 Richard Sandiford <richard.sandiford@linaro.org>
15219 * config/aarch64/aarch64-sve.md (*vec_extract<mode><Vel>_0): New
15221 (*vec_extract<mode><Vel>_v128): Require a nonzero lane number.
15222 Use gen_rtx_REG rather than gen_lowpart.
15224 2018-01-30 Richard Sandiford <richard.sandiford@linaro.org>
15226 * lra-constraints.c (match_reload): Use subreg_lowpart_offset
15227 rather than 0 when creating partial subregs.
15229 2018-01-30 Richard Sandiford <richard.sandiford@linaro.org>
15231 * vec-perm-indices.c (vec_perm_indices::series_p): Give examples
15234 2018-01-29 Michael Meissner <meissner@linux.vnet.ibm.com>
15237 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): If DFmode
15238 and SFmode can go in Altivec registers (-mcpu=power7 for DFmode,
15239 -mcpu=power8 for SFmode) don't set the PRE_INCDEC or PRE_MODIFY
15240 flags. This restores the settings used before the 2017-07-24.
15241 Turning off pre increment/decrement/modify allows IVOPTS to
15242 optimize DF/SF loops where the index is an int.
15244 2018-01-29 Richard Biener <rguenther@suse.de>
15245 Kelvin Nilsen <kelvin@gcc.gnu.org>
15248 * tree-vect-stmts.c (vectorizable_call): Don't call
15249 targetm.vectorize_builtin_md_vectorized_function if callee is
15252 2018-01-22 Carl Love <cel@us.ibm.com>
15254 * doc/extend.tex: Fix typo in second arg in
15255 __builtin_bcdadd_{lt|eq|gt|ov} and __builtin_bcdsub_{lt|eq|gt|ov}.
15257 2018-01-29 Richard Biener <rguenther@suse.de>
15259 PR tree-optimization/84086
15260 * tree-ssanames.c: Include cfgloop.h and tree-scalar-evolution.h.
15261 (flush_ssaname_freelist): When SSA names were released reset
15262 the SCEV hash table.
15264 2018-01-29 Richard Biener <rguenther@suse.de>
15266 PR tree-optimization/84057
15267 * tree-ssa-loop-ivcanon.c (unloop_loops): Deal with already
15268 removed paths when removing edges.
15270 2018-01-27 H.J. Lu <hongjiu.lu@intel.com>
15272 * doc/invoke.texi: Replace -mfunction-return==@var{choice} with
15273 -mfunction-return=@var{choice}.
15275 2018-01-27 Bernd Edlinger <bernd.edlinger@hotmail.de>
15277 PR diagnostic/84034
15278 * diagnostic-show-locus.c (get_line_width_without_trailing_whitespace):
15279 Handle CR like TAB.
15280 (layout::print_source_line): Likewise.
15281 (test_get_line_width_without_trailing_whitespace): Add test cases.
15283 2018-01-27 Jakub Jelinek <jakub@redhat.com>
15285 PR middle-end/84040
15286 * sched-deps.c (sched_macro_fuse_insns): Return immediately for
15289 2018-01-26 Jim Wilson <jimw@sifive.com>
15291 * config/riscv/riscv.h (MAX_FIXED_MODE_SIZE): New.
15293 * config/riscv/elf.h (LIB_SPEC): Don't include -lgloss when nosys.specs
15296 2018-01-26 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15298 * config/aarch64/aarch64.md: Add peepholes for CMP + SUB -> SUBS
15299 and CMP + SUB-immediate -> SUBS.
15301 2018-01-26 Martin Sebor <msebor@redhat.com>
15303 PR tree-optimization/83896
15304 * tree-ssa-strlen.c (get_string_len): Rename...
15305 (get_string_cst_length): ...to this. Return HOST_WIDE_INT.
15306 Avoid assuming length is constant.
15307 (handle_char_store): Use HOST_WIDE_INT for string length.
15309 2018-01-26 Uros Bizjak <ubizjak@gmail.com>
15312 * config/i386/i386.md (*andndi3_doubleword): Add earlyclobber
15313 to (=&r,r,rm) alternative. Add (=r,0,rm) and (=r,r,0) alternatives.
15315 2018-01-26 Richard Biener <rguenther@suse.de>
15317 PR rtl-optimization/84003
15318 * dse.c (record_store): Only record redundant stores when
15319 the earlier store aliases at least all accesses the later one does.
15321 2018-01-26 Jakub Jelinek <jakub@redhat.com>
15323 PR rtl-optimization/83985
15324 * dce.c (deletable_insn_p): Return false for separate shrink wrapping
15325 REG_CFA_RESTORE insns.
15326 (delete_unmarked_insns): Don't ignore separate shrink wrapping
15327 REG_CFA_RESTORE insns here.
15330 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Don't
15331 use SSA_NAME_VAR as base for SSA_NAMEs with non-NULL SSA_NAME_VAR.
15333 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
15335 * config/arc/arc-arch.h (arc_tune_attr): Add ARC_TUNE_CORE_3.
15336 * config/arc/arc.c (arc_sched_issue_rate): Use ARC_TUNE_... .
15337 (arc_init): Likewise.
15338 (arc_override_options): Likewise.
15339 (arc_file_start): Choose Tag_ARC_CPU_variation based on arc_tune
15341 (hwloop_fail): Use TARGET_DBNZ when we want to check for dbnz insn
15343 * config/arc/arc.h (TARGET_DBNZ): Define.
15344 * config/arc/arc.md (attr tune): Add core_3, use ARC_TUNE_... to
15345 properly set the tune attribute.
15346 (dbnz): Use TARGET_DBNZ guard.
15347 * config/arc/arc.opt (mtune): Add core3 option.
15349 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
15351 * config/arc/arc.c (arc_delegitimize_address_0): Refactored to
15352 recognize new pic like addresses.
15353 (arc_delegitimize_address): Clean up.
15355 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
15357 * config/arc/arc-arches.def: Option mrf16 valid for all
15359 * config/arc/arc-c.def (__ARC_RF16__): New predefined macro.
15360 * config/arc/arc-cpus.def (em_mini): New cpu with rf16 on.
15361 * config/arc/arc-options.def (FL_RF16): Add mrf16 option.
15362 * config/arc/arc-tables.opt: Regenerate.
15363 * config/arc/arc.c (arc_conditional_register_usage): Handle
15364 reduced register file case.
15365 (arc_file_start): Set must have build attributes.
15366 * config/arc/arc.h (MAX_ARC_PARM_REGS): Conditional define using
15367 mrf16 option value.
15368 * config/arc/arc.opt (mrf16): Add new option.
15369 * config/arc/elf.h (ATTRIBUTE_PCS): Define.
15370 * config/arc/genmultilib.awk: Handle new mrf16 option.
15371 * config/arc/linux.h (ATTRIBUTE_PCS): Define.
15372 * config/arc/t-multilib: Regenerate.
15373 * doc/invoke.texi (ARC Options): Document mrf16 option.
15375 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
15377 * config/arc/arc-protos.h: Add arc_is_secure_call_p proto.
15378 * config/arc/arc.c (arc_handle_secure_attribute): New function.
15379 (arc_attribute_table): Add 'secure_call' attribute.
15380 (arc_print_operand): Print secure call operand.
15381 (arc_function_ok_for_sibcall): Don't optimize tail calls when
15383 (arc_is_secure_call_p): New function. * config/arc/arc.md
15384 (call_i): Add support for sjli instruction.
15385 (call_value_i): Likewise.
15386 * config/arc/constraints.md (Csc): New constraint.
15388 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
15389 John Eric Martin <John.Martin@emmicro-us.com>
15391 * config/arc/arc-protos.h: Add arc_is_jli_call_p proto.
15392 * config/arc/arc.c (_arc_jli_section): New struct.
15393 (arc_jli_section): New type.
15394 (rc_jli_sections): New static variable.
15395 (arc_handle_jli_attribute): New function.
15396 (arc_attribute_table): Add jli_always and jli_fixed attribute.
15397 (arc_file_end): New function.
15398 (TARGET_ASM_FILE_END): Define.
15399 (arc_print_operand): Reuse 'S' letter for JLI output instruction.
15400 (arc_add_jli_section): New function.
15401 (jli_call_scan): Likewise.
15402 (arc_reorg): Call jli_call_scan.
15403 (arc_output_addsi): Remove 'S' from printing asm operand.
15404 (arc_is_jli_call_p): New function.
15405 * config/arc/arc.md (movqi_insn): Remove 'S' from printing asm
15407 (movhi_insn): Likewise.
15408 (movsi_insn): Likewise.
15409 (movsi_set_cc_insn): Likewise.
15410 (loadqi_update): Likewise.
15411 (load_zeroextendqisi_update): Likewise.
15412 (load_signextendqisi_update): Likewise.
15413 (loadhi_update): Likewise.
15414 (load_zeroextendhisi_update): Likewise.
15415 (load_signextendhisi_update): Likewise.
15416 (loadsi_update): Likewise.
15417 (loadsf_update): Likewise.
15418 (movsicc_insn): Likewise.
15419 (bset_insn): Likewise.
15420 (bxor_insn): Likewise.
15421 (bclr_insn): Likewise.
15422 (bmsk_insn): Likewise.
15423 (bicsi3_insn): Likewise.
15424 (cmpsi_cc_c_insn): Likewise.
15425 (movsi_ne): Likewise.
15426 (movsi_cond_exec): Likewise.
15427 (clrsbsi2): Likewise.
15428 (norm_f): Likewise.
15438 (call_i): Remove 'S' asm letter, add jli instruction.
15439 (call_value_i): Likewise.
15440 * config/arc/arc.op (mjli-always): New option.
15441 * config/arc/constraints.md (Cji): New constraint.
15442 * config/arc/fpx.md (addsf3_fpx): Remove 'S' from printing asm
15444 (subsf3_fpx): Likewise.
15445 (mulsf3_fpx): Likewise.
15446 * config/arc/simdext.md (vendrec_insn): Remove 'S' from printing
15448 * doc/extend.texi (ARC): Document 'jli-always' and 'jli-fixed'
15449 function attrbutes.
15450 * doc/invoke.texi (ARC): Document mjli-always option.
15452 2018-01-26 Sebastian Perta <sebastian.perta@renesas.com>
15454 * config/rl78/rl78.c (rl78_addsi3_internal): If operand 2 is const
15455 avoid addition with 0 and use incw and decw where possible.
15457 2018-01-26 Richard Biener <rguenther@suse.de>
15459 PR tree-optimization/81082
15460 * fold-const.c (fold_plusminus_mult_expr): Do not perform the
15461 association if it requires casting to unsigned.
15462 * match.pd ((A * C) +- (B * C) -> (A+-B)): New patterns derived
15463 from fold_plusminus_mult_expr to catch important cases late when
15464 range info is available.
15466 2018-01-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
15468 * config/i386/sol2.h (USE_HIDDEN_LINKONCE): Remove.
15469 * configure.ac (hidden_linkonce): New test.
15470 * configure: Regenerate.
15471 * config.in: Regenerate.
15473 2018-01-26 Julia Koval <julia.koval@intel.com>
15475 * config/i386/avx512bitalgintrin.h (_mm512_bitshuffle_epi64_mask,
15476 _mm512_mask_bitshuffle_epi64_mask, _mm256_bitshuffle_epi64_mask,
15477 _mm256_mask_bitshuffle_epi64_mask, _mm_bitshuffle_epi64_mask,
15478 _mm_mask_bitshuffle_epi64_mask): Fix type.
15479 * config/i386/i386-builtin-types.def (UHI_FTYPE_V2DI_V2DI_UHI,
15480 USI_FTYPE_V4DI_V4DI_USI): Remove.
15481 * config/i386/i386-builtin.def (__builtin_ia32_vpshufbitqmb512_mask,
15482 __builtin_ia32_vpshufbitqmb256_mask,
15483 __builtin_ia32_vpshufbitqmb128_mask): Fix types.
15484 * config/i386/i386.c (ix86_expand_args_builtin): Remove old types.
15485 * config/i386/sse.md (VI1_AVX512VLBW): Change types.
15487 2018-01-26 Alan Modra <amodra@gmail.com>
15490 * config/rs6000/rs6000-p8swap.c (rtx_is_swappable_p): Exclude
15491 UNSPEC_VBPERMQ. Sort other unspecs.
15493 2018-01-25 David Edelsohn <dje.gcc@gmail.com>
15495 * doc/invoke.texi (PowerPC Options): Document 'native' cpu type.
15497 2018-01-25 Jan Hubicka <hubicka@ucw.cz>
15499 PR middle-end/83055
15500 * predict.c (drop_profile): Do not push/pop cfun; update also
15502 (handle_missing_profiles): Fix logic looking for zero profiles.
15504 2018-01-25 Jakub Jelinek <jakub@redhat.com>
15506 PR middle-end/83977
15507 * ipa-fnsummary.c (compute_fn_summary): Clear can_change_signature
15508 on functions with #pragma omp declare simd or functions with simd
15510 * omp-simd-clone.c (expand_simd_clones): Revert 2018-01-24 change.
15511 * config/i386/i386.c (ix86_simd_clone_compute_vecsize_and_simdlen):
15512 Remove trailing \n from warning_at calls.
15514 2018-01-25 Tom de Vries <tom@codesourcery.com>
15517 * config/nvptx/nvptx.c (nvptx_single): Add exit insn after noreturn call
15518 for neutered workers.
15520 2018-01-24 Joseph Myers <joseph@codesourcery.com>
15523 * config/m68k/m68k.c (m68k_promote_function_mode): New function.
15524 (TARGET_PROMOTE_FUNCTION_MODE): New macro.
15526 2018-01-24 Jeff Law <law@redhat.com>
15529 * i386.c (get_probe_interval): Move to earlier point.
15530 (ix86_compute_frame_layout): If -fstack-clash-protection and
15531 the frame is larger than the probe interval, then use pushes
15532 to save registers rather than reg->mem moves.
15533 (ix86_expand_prologue): Remove conditional for int_registers_saved
15536 2018-01-24 Vladimir Makarov <vmakarov@redhat.com>
15539 * ira-build.c (setup_min_max_allocno_live_range_point): Set up
15540 min/max for never referenced object.
15542 2018-01-24 Jakub Jelinek <jakub@redhat.com>
15544 PR middle-end/83977
15545 * tree.c (free_lang_data_in_decl): Don't clear DECL_ABSTRACT_ORIGIN
15547 * omp-low.c (create_omp_child_function): Remove "omp declare simd"
15548 attributes from DECL_ATTRIBUTES (decl) without affecting
15549 DECL_ATTRIBUTES (current_function_decl).
15550 * omp-simd-clone.c (expand_simd_clones): Ignore DECL_ARTIFICIAL
15551 functions with non-NULL DECL_ABSTRACT_ORIGIN.
15553 2018-01-24 Richard Sandiford <richard.sandiford@linaro.org>
15555 PR tree-optimization/83979
15556 * fold-const.c (fold_comparison): Use constant_boolean_node
15557 instead of boolean_{true,false}_node.
15559 2018-01-24 Jan Hubicka <hubicka@ucw.cz>
15561 * ipa-profile.c (ipa_propagate_frequency_1): Fix logic skipping calls
15564 2018-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15566 * config/rs6000/rs6000.md (*call_indirect_nonlocal_sysv<mode>):
15567 Simplify the clause that sets the length attribute.
15568 (*call_value_indirect_nonlocal_sysv<mode>): Likewise.
15569 (*sibcall_nonlocal_sysv<mode>): Clean up code block; simplify the
15570 clause that sets the length attribute.
15571 (*sibcall_value_nonlocal_sysv<mode>): Likewise.
15573 2018-01-24 Tom de Vries <tom@codesourcery.com>
15576 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG_2): Define to 1.
15577 (nvptx_pc_set, nvptx_condjump_label): New function. Copy from jump.c.
15578 Add strict parameter.
15579 (prevent_branch_around_nothing): Insert dummy insn between branch to
15580 label and label with no ptx insn inbetween.
15581 * config/nvptx/nvptx.md (define_insn "fake_nop"): New insn.
15583 2018-01-24 Tom de Vries <tom@codesourcery.com>
15586 * config/nvptx/nvptx.c (nvptx_single): Add exit insn after noreturn call
15587 for neutered threads in warp.
15588 * config/nvptx/nvptx.md (define_insn "exit"): New insn.
15590 2018-01-24 Richard Biener <rguenther@suse.de>
15592 PR tree-optimization/83176
15593 * tree-chrec.c (chrec_fold_plus_1): Handle (signed T){(T) .. }
15596 2018-01-24 Richard Biener <rguenther@suse.de>
15598 PR tree-optimization/82819
15599 * graphite-isl-ast-to-gimple.c (binary_op_to_tree): Avoid
15600 code generating pluses that are no-ops in the target precision.
15602 2018-01-24 Richard Biener <rguenther@suse.de>
15604 PR middle-end/84000
15605 * tree-cfg.c (replace_loop_annotate): Handle annot_expr_parallel_kind.
15607 2018-01-23 Jan Hubicka <hubicka@ucw.cz>
15609 * cfgcleanup.c (try_crossjump_to_edge): Use combine_with_count
15610 to merge probabilities.
15611 * predict.c (probably_never_executed): Also mark as cold functions
15612 with global 0 profile and guessed local profile.
15613 * profile-count.c (profile_probability::combine_with_count): New
15615 * profile-count.h (profile_probability::operator*,
15616 profile_probability::operator*=, profile_probability::operator/,
15617 profile_probability::operator/=): Reduce precision to adjusted
15618 and set value to guessed on contradictory divisions.
15619 (profile_probability::combine_with_freq): Remove.
15620 (profile_probability::combine_wiht_count): Declare.
15621 (profile_count::force_nonzero):: Set to adjusted.
15622 (profile_count::probability_in):: Set quality to adjusted.
15623 * tree-ssa-tail-merge.c (replace_block_by): Use
15624 combine_with_count.
15626 2018-01-23 Andrew Waterman <andrew@sifive.com>
15627 Jim Wilson <jimw@sifive.com>
15629 * config/riscv/riscv.c (riscv_stack_boundary): New.
15630 (riscv_option_override): Set riscv_stack_boundary. Handle
15631 riscv_preferred_stack_boundary_arg.
15632 * config/riscv/riscv.h (MIN_STACK_BOUNDARY, ABI_STACK_BOUNDARY): New.
15633 (BIGGEST_ALIGNMENT): Set to STACK_BOUNDARY.
15634 (STACK_BOUNDARY): Set to riscv_stack_boundary.
15635 (RISCV_STACK_ALIGN): Use STACK_BOUNDARY.
15636 * config/riscv/riscv.opt (mpreferred-stack-boundary): New.
15637 * doc/invoke.tex (RISC-V Options): Add -mpreferred-stack-boundary.
15639 2018-01-23 H.J. Lu <hongjiu.lu@intel.com>
15642 * config/i386/i386.c (ix86_expand_prologue): Use cost reference
15643 of struct ix86_frame.
15644 (ix86_expand_epilogue): Likewise. Add a local variable for
15645 the reg_save_offset field in struct ix86_frame.
15647 2018-01-23 Bin Cheng <bin.cheng@arm.com>
15649 PR tree-optimization/82604
15650 * tree-loop-distribution.c (enum partition_kind): New enum item
15651 PKIND_PARTIAL_MEMSET.
15652 (partition_builtin_p): Support above new enum item.
15653 (generate_code_for_partition): Ditto.
15654 (compute_access_range): Differentiate cases that equality can be
15655 proven at all loops, the innermost loops or no loops.
15656 (classify_builtin_st, classify_builtin_ldst): Adjust call to above
15657 function. Set PKIND_PARTIAL_MEMSET for partition appropriately.
15658 (finalize_partitions, distribute_loop): Don't fuse partition of
15659 PKIND_PARTIAL_MEMSET kind when distributing 3-level loop nest.
15660 (prepare_perfect_loop_nest): Distribute 3-level loop nest only if
15661 parloop is enabled.
15663 2018-01-23 Martin Liska <mliska@suse.cz>
15665 * predict.def (PRED_INDIR_CALL): Set probability to PROB_EVEN in
15666 order to ignore the predictor.
15667 (PRED_POLYMORPHIC_CALL): Likewise.
15668 (PRED_RECURSIVE_CALL): Likewise.
15670 2018-01-23 Martin Liska <mliska@suse.cz>
15672 * tree-profile.c (tree_profiling): Print function header to
15673 aware reader which function we are working on.
15674 * value-prof.c (gimple_find_values_to_profile): Do not print
15675 not interesting value histograms.
15677 2018-01-23 Martin Liska <mliska@suse.cz>
15679 * profile-count.h (enum profile_quality): Add
15680 profile_uninitialized as the first value. Do not number values
15681 as they are zero based.
15682 (profile_count::verify): Update sanity check.
15683 (profile_probability::verify): Likewise.
15685 2018-01-23 Nathan Sidwell <nathan@acm.org>
15687 * doc/invoke.texi (ffor-scope): Deprecate.
15689 2018-01-23 David Malcolm <dmalcolm@redhat.com>
15691 PR tree-optimization/83510
15692 * domwalk.c (set_all_edges_as_executable): New function.
15693 (dom_walker::dom_walker): Convert bool param
15694 "skip_unreachable_blocks" to enum reachability. Move setup of
15695 edge flags to set_all_edges_as_executable and only do it when
15696 reachability is REACHABLE_BLOCKS.
15697 * domwalk.h (enum dom_walker::reachability): New enum.
15698 (dom_walker::dom_walker): Convert bool param
15699 "skip_unreachable_blocks" to enum reachability.
15700 (set_all_edges_as_executable): New decl.
15701 * graphite-scop-detection.c (gather_bbs::gather_bbs): Convert
15702 from false for "skip_unreachable_blocks" to ALL_BLOCKS for
15704 * tree-ssa-dom.c (dom_opt_dom_walker::dom_opt_dom_walker): Likewise,
15705 but converting true to REACHABLE_BLOCKS.
15706 * tree-ssa-sccvn.c (sccvn_dom_walker::sccvn_dom_walker): Likewise.
15708 (check_array_bounds_dom_walker::check_array_bounds_dom_walker):
15709 Likewise, but converting it to REACHABLE_BLOCKS_PRESERVING_FLAGS.
15710 (vrp_dom_walker::vrp_dom_walker): Likewise, but converting it to
15712 (vrp_prop::vrp_finalize): Call set_all_edges_as_executable
15713 if check_all_array_refs will be called.
15715 2018-01-23 David Malcolm <dmalcolm@redhat.com>
15717 * tree.c (selftest::test_location_wrappers): Add more test
15720 2018-01-23 David Malcolm <dmalcolm@redhat.com>
15722 * sbitmap.c (selftest::test_set_range): Fix memory leaks.
15723 (selftest::test_bit_in_range): Likewise.
15725 2018-01-23 Richard Sandiford <richard.sandiford@linaro.org>
15728 * doc/sourcebuild.texi (vect_float): Say that the selector
15729 only describes the situation when -funsafe-math-optimizations is on.
15730 (vect_float_strict): Document.
15732 2018-01-23 Richard Sandiford <richard.sandiford@linaro.org>
15734 PR tree-optimization/83965
15735 * tree-vect-patterns.c (vect_reassociating_reduction_p): New function.
15736 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern): Use it
15737 instead of checking only for a reduction.
15738 (vect_recog_widen_sum_pattern): Likewise.
15740 2018-01-23 Jan Hubicka <hubicka@ucw.cz>
15742 * predict.c (probably_never_executed): Only use precise profile info.
15743 (compute_function_frequency): Skip after inlining hack since we now
15744 have quality checking.
15746 2018-01-23 Jan Hubicka <hubicka@ucw.cz>
15748 * profile-count.h (profile_probability::very_unlikely,
15749 profile_probability::unlikely, profile_probability::even): Set
15750 precision to guessed.
15752 2018-01-23 Richard Biener <rguenther@suse.de>
15754 PR tree-optimization/83963
15755 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
15756 Properly terminate dominator walk when crossing the exit edge not
15757 when visiting its source block.
15759 2018-01-23 Jakub Jelinek <jakub@redhat.com>
15762 * tree.c (maybe_wrap_with_location): Use NON_LVALUE_EXPR rather than
15763 VIEW_CONVERT_EXPR to wrap CONST_DECLs.
15765 2018-01-22 Jakub Jelinek <jakub@redhat.com>
15767 PR tree-optimization/83957
15768 * omp-expand.c (expand_omp_for_generic): Ignore virtual PHIs. Remove
15769 semicolon after for body surrounded by braces.
15771 PR tree-optimization/83081
15772 * profile-count.h (profile_probability::split): New method.
15773 * dojump.c (do_jump_1) <case TRUTH_ANDIF_EXPR, case TRUTH_ORIF_EXPR>:
15774 Use profile_probability::split.
15775 (do_compare_rtx_and_jump): Fix adjustment of probabilities
15776 when splitting a single conditional jump into 2.
15778 2018-01-22 David Malcolm <dmalcolm@redhat.com>
15780 PR tree-optimization/69452
15781 * tree-ssa-loop-im.c (class move_computations_dom_walker): Remove
15784 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
15786 * config/rl78/rl78-expand.md (bswaphi2): New define_expand.
15787 * config/rl78/rl78-virt.md (*bswaphi2_virt): New define_insn.
15788 * config/rl78/rl78-real.md (*bswaphi2_real): New define_insn.
15790 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
15792 * config/rl78/rl78-protos.h (rl78_split_movdi): New function
15794 * config/rl78/rl78.md (movdi): New define_expand.
15795 * config/rl78/rl78.c (rl78_split_movdi): New function.
15797 2018-01-22 Michael Meissner <meissner@linux.vnet.ibm.com>
15800 * config/rs6000/rs6000-protos.h (rs6000_split_signbit): Delete,
15802 * config/rs6000/rs6000.c (rs6000_split_signbit): Likewise.
15803 * config/rs6000/rs6000.md (signbit<mode>2): Change code for IEEE
15804 128-bit to produce an UNSPEC move to get the double word with the
15805 signbit and then a shift directly to do signbit.
15806 (signbit<mode>2_dm): Replace old IEEE 128-bit signbit
15807 implementation with a new version that just does either a direct
15808 move or a regular move. Move memory interface to separate insns.
15809 Move insns so they are next to the expander.
15810 (signbit<mode>2_dm_mem_be): New combiner insns to combine load
15811 with signbit move. Split big and little endian case.
15812 (signbit<mode>2_dm_mem_le): Likewise.
15813 (signbit<mode>2_dm_<su>ext): Delete, no longer used.
15814 (signbit<mode>2_dm2): Likewise.
15816 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
15818 * config/rl78/rl78.md (anddi3): New define_expand.
15820 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
15822 * config/rl78/rl78.md (umindi3): New define_expand.
15824 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
15826 * config/rl78/rl78.md (smindi3): New define_expand.
15828 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
15830 * config/rl78/rl78.md (smaxdi3): New define_expand.
15832 2018-01-22 Carl Love <cel@us.ibm.com>
15834 * config/rs6000/rs6000-builtin.def (ST_ELEMREV_V1TI, LD_ELEMREV_V1TI,
15835 LVX_V1TI): Add macro expansion.
15836 * config/rs6000/rs6000-c.c (altivec_builtin_types): Add argument
15837 definitions for VSX_BUILTIN_VEC_XST_BE, VSX_BUILTIN_VEC_ST,
15838 VSX_BUILTIN_VEC_XL, LD_ELEMREV_V1TI builtins.
15839 * config/rs6000/rs6000-p8swap.c (insn_is_swappable_p);
15840 Change check to determine if the instruction is a byte reversing
15841 entry. Fix typo in comment.
15842 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case entry
15843 for VSX_BUILTIN_ST_ELEMREV_V1TI and VSX_BUILTIN_LD_ELEMREV_V1TI.
15844 Add def_builtin calls for new builtins.
15845 * config/rs6000/vsx.md (vsx_st_elemrev_v1ti, vsx_ld_elemrev_v1ti):
15846 Add define_insn expansion.
15848 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
15850 * config/rl78/rl78.md (umaxdi3): New define_expand.
15852 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
15854 * config/rl78/rl78.c (rl78_note_reg_set): Fixed dead reg check
15855 for non-QImode registers.
15857 2018-01-22 Richard Biener <rguenther@suse.de>
15859 PR tree-optimization/83963
15860 * graphite-scop-detection.c (scop_detection::get_sese): Delay
15861 including the loop exit block.
15862 (scop_detection::merge_sese): Likewise.
15863 (scop_detection::add_scop): Do it here instead.
15865 2018-01-22 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15867 * doc/sourcebuild.texi (arm_softfloat): Document.
15869 2018-01-21 John David Anglin <danglin@gcc.gnu.org>
15872 * config/pa/pa.c (pa_function_ok_for_sibcall): Use
15873 targetm.binds_local_p instead of TREE_PUBLIC to check local binding.
15874 Move TARGET_PORTABLE_RUNTIME check after TARGET_64BIT check.
15876 2018-01-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15877 David Edelsohn <dje.gcc@gmail.com>
15880 * config/rs6000/rs6000.md (*call_indirect_nonlocal_sysv<mode>):
15881 Change "crset eq" to "crset 2".
15882 (*call_value_indirect_nonlocal_sysv<mode>): Likewise.
15883 (*call_indirect_aix<mode>_nospec): Likewise.
15884 (*call_value_indirect_aix<mode>_nospec): Likewise.
15885 (*call_indirect_elfv2<mode>_nospec): Likewise.
15886 (*call_value_indirect_elfv2<mode>_nospec): Likewise.
15887 (*sibcall_nonlocal_sysv<mode>): Change "crset eq" to "crset 2";
15888 change assembly output from . to $.
15889 (*sibcall_value_nonlocal_sysv<mode>): Likewise.
15890 (indirect_jump<mode>_nospec): Change assembly output from . to $.
15891 (*tablejump<mode>_internal1_nospec): Likewise.
15893 2018-01-21 Oleg Endo <olegendo@gcc.gnu.org>
15896 * config/sh/sh_optimize_sett_clrt.cc:
15897 Use INCLUDE_ALGORITHM and INCLUDE_VECTOR instead of direct includes.
15899 2018-01-20 Richard Sandiford <richard.sandiford@linaro.org>
15901 PR tree-optimization/83940
15902 * tree-vect-stmts.c (vect_truncate_gather_scatter_offset): Set
15903 offset_dt to vect_constant_def rather than vect_unknown_def_type.
15904 (vect_check_load_store_mask): Add a mask_dt_out parameter and
15905 use it to pass back the definition type.
15906 (vect_check_store_rhs): Likewise rhs_dt_out.
15907 (vect_build_gather_load_calls): Add a mask_dt argument and use
15908 it instead of a call to vect_is_simple_use.
15909 (vectorizable_store): Update calls to vect_check_load_store_mask
15910 and vect_check_store_rhs. Use the dt returned by the latter instead
15911 of scatter_src_dt. Use the cached mask_dt and gs_info.offset_dt
15912 instead of calls to vect_is_simple_use. Pass the scalar rather
15913 than the vector operand to vect_is_simple_use when handling
15914 second and subsequent copies of an rhs value.
15915 (vectorizable_load): Update calls to vect_check_load_store_mask
15916 and vect_build_gather_load_calls. Use the cached mask_dt and
15917 gs_info.offset_dt instead of calls to vect_is_simple_use.
15919 2018-01-20 Jakub Jelinek <jakub@redhat.com>
15921 PR middle-end/83945
15922 * tree-emutls.c: Include gimplify.h.
15923 (lower_emutls_2): New function.
15924 (lower_emutls_1): If ADDR_EXPR is a gimple invariant and walk_tree
15925 with lower_emutls_2 callback finds some TLS decl in it, unshare_expr
15926 it before further processing.
15929 * simplify-rtx.c (simplify_binary_operation_1) <case UMOD>: Use
15930 UINTVAL (trueop1) instead of INTVAL (op1).
15932 2018-01-19 Jakub Jelinek <jakub@redhat.com>
15936 * dwarf2cfi.c (DEFAULT_INCOMING_FRAME_SP_OFFSET): Define to
15937 INCOMING_FRAME_SP_OFFSET if not defined.
15938 (scan_trace): Add ENTRY argument. If true and
15939 DEFAULT_INCOMING_FRAME_SP_OFFSET != INCOMING_FRAME_SP_OFFSET,
15940 emit a note to adjust the CFA offset.
15941 (create_cfi_notes): Adjust scan_trace callers.
15942 (create_cie_data): Use DEFAULT_INCOMING_FRAME_SP_OFFSET rather than
15943 INCOMING_FRAME_SP_OFFSET in the CIE.
15944 * config/i386/i386.h (DEFAULT_INCOMING_FRAME_SP_OFFSET): Define.
15945 * config/stormy16/stormy16.h (DEFAULT_INCOMING_FRAME_SP_OFFSET):
15947 * doc/tm.texi.in (DEFAULT_INCOMING_FRAME_SP_OFFSET): Document.
15948 * doc/tm.texi: Regenerated.
15950 2018-01-19 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15952 PR rtl-optimization/83147
15953 * lra-constraints.c (remove_inheritance_pseudos): Use
15954 lra_substitute_pseudo_within_insn.
15956 2018-01-19 Tom de Vries <tom@codesourcery.com>
15957 Cesar Philippidis <cesar@codesourcery.com>
15960 * config/nvptx/nvptx.c (nvptx_single): Fix jit workaround.
15962 2018-01-19 Cesar Philippidis <cesar@codesourcery.com>
15965 * config/nvptx/nvptx.c (output_init_frag): Don't use generic address
15966 spaces for function labels.
15968 2018-01-19 Martin Liska <mliska@suse.cz>
15970 * predict.def (PRED_LOOP_EXIT): Change from 85 to 89.
15971 (PRED_LOOP_EXIT_WITH_RECURSION): Change from 72 to 78.
15972 (PRED_LOOP_EXTRA_EXIT): Change from 83 to 67.
15973 (PRED_OPCODE_POSITIVE): Change from 64 to 59.
15974 (PRED_TREE_OPCODE_POSITIVE): Change from 64 to 59.
15975 (PRED_CONST_RETURN): Change from 69 to 65.
15976 (PRED_NULL_RETURN): Change from 91 to 71.
15977 (PRED_LOOP_IV_COMPARE_GUESS): Change from 98 to 64.
15978 (PRED_LOOP_GUARD): Change from 66 to 73.
15980 2018-01-19 Martin Liska <mliska@suse.cz>
15982 * predict.c (predict_insn_def): Add new assert.
15983 (struct branch_predictor): Change type to signed integer.
15984 (test_prediction_value_range): Amend test to cover
15985 PROB_UNINITIALIZED.
15986 * predict.def (PRED_LOOP_ITERATIONS): Use the new constant.
15987 (PRED_LOOP_ITERATIONS_GUESSED): Likewise.
15988 (PRED_LOOP_ITERATIONS_MAX): Likewise.
15989 (PRED_LOOP_IV_COMPARE): Likewise.
15990 * predict.h (PROB_UNINITIALIZED): Define new constant.
15992 2018-01-19 Martin Liska <mliska@suse.cz>
15994 * predict.c (dump_prediction): Add new format for
15995 analyze_brprob.py script which is enabled with -details
15997 * profile-count.h (precise_p): New function.
15999 2018-01-19 Richard Sandiford <richard.sandiford@linaro.org>
16001 PR tree-optimization/83922
16002 * tree-vect-loop.c (vect_verify_full_masking): Return false if
16003 there are no statements that need masking.
16004 (vect_active_double_reduction_p): New function.
16005 (vect_analyze_loop_operations): Use it when handling phis that
16006 are not in the loop header.
16008 2018-01-19 Richard Sandiford <richard.sandiford@linaro.org>
16010 PR tree-optimization/83914
16011 * tree-vect-loop.c (vectorizable_induction): Don't convert
16012 init_expr or apply the peeling adjustment for inductions
16013 that are nested within the vectorized loop.
16015 2018-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16017 * config/arm/thumb2.md (*thumb2_negsi2_short): Use RSB mnemonic
16020 2018-01-18 Jakub Jelinek <jakub@redhat.com>
16024 * function.h (gimplify_parameters): Add gimple_seq * argument.
16025 * function.c: Include gimple.h and options.h.
16026 (gimplify_parameters): Add cleanup argument, add CLOBBER stmts
16027 for the added local temporaries if needed.
16028 * gimplify.c (gimplify_body): Adjust gimplify_parameters caller,
16029 if there are any parameter cleanups, wrap whole body into a
16030 try/finally with the cleanups.
16032 2018-01-18 Wilco Dijkstra <wdijkstr@arm.com>
16035 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
16036 Use GET_MODE_CLASS for scalar floating point.
16038 2018-01-18 Jan Hubicka <hubicka@ucw.cz>
16042 * cgraphclones.c (cgraph_node::create_version_clone_with_body):
16043 Fix call of call_cgraph_insertion_hooks.
16045 2018-01-18 Martin Sebor <msebor@redhat.com>
16047 * doc/invoke.texi (-Wclass-memaccess): Tweak text.
16049 2018-01-18 Jan Hubicka <hubicka@ucw.cz>
16052 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Update edge
16055 2018-01-18 Boris Kolpackov <boris@codesynthesis.com>
16058 * common.opt: (-ffile-prefix-map): New option.
16059 * opts.c (common_handle_option): Defer it.
16060 * opts-global.c (handle_common_deferred_options): Handle it.
16061 * debug.h (remap_debug_filename, add_debug_prefix_map): Move to...
16062 * file-prefix-map.h: New file.
16063 (remap_debug_filename, add_debug_prefix_map): ...here.
16064 (add_macro_prefix_map, add_file_prefix_map, remap_macro_filename): New.
16065 * final.c (debug_prefix_map, add_debug_prefix_map
16066 remap_debug_filename): Move to...
16067 * file-prefix-map.c: New file.
16068 (file_prefix_map, add_prefix_map, remap_filename) ...here and rename,
16069 generalize, get rid of alloca(), use strrchr() instead of strchr().
16070 (add_macro_prefix_map, add_debug_prefix_map, add_file_prefix_map):
16071 Implement in terms of add_prefix_map().
16072 (remap_macro_filename, remap_debug_filename): Implement in term of
16074 * Makefile.in (OBJS, PLUGIN_HEADERS): Add new files.
16075 * builtins.c (fold_builtin_FILE): Call remap_macro_filename().
16076 * dbxout.c: Include file-prefix-map.h.
16077 * varasm.c: Likewise.
16078 * vmsdbgout.c: Likewise.
16079 * xcoffout.c: Likewise.
16080 * dwarf2out.c: Likewise plus omit new options from DW_AT_producer.
16081 * doc/cppopts.texi (-fmacro-prefix-map): Document.
16082 * doc/invoke.texi (-ffile-prefix-map): Document.
16083 (-fdebug-prefix-map): Update description.
16085 2018-01-18 Martin Liska <mliska@suse.cz>
16087 * config/i386/i386.c (indirect_thunk_name): Document that also
16089 (output_indirect_thunk): Document why both instructions
16090 (pause and lfence) are generated.
16092 2018-01-18 Richard Biener <rguenther@suse.de>
16094 PR tree-optimization/83887
16095 * graphite-scop-detection.c
16096 (scop_detection::get_nearest_dom_with_single_entry): Remove.
16097 (scop_detection::get_nearest_pdom_with_single_exit): Likewise.
16098 (scop_detection::merge_sese): Re-implement with a flood-fill
16099 algorithm that properly finds a SESE region if it exists.
16101 2018-01-18 Jakub Jelinek <jakub@redhat.com>
16104 * match.pd ((P + A) - P, P - (P + A), (P + A) - (P + B)): For
16105 pointer_diff optimizations use view_convert instead of convert.
16107 2018-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16109 * config/rs6000/rs6000.md (*call_indirect_nonlocal_sysv<mode>):
16110 Generate different code for -mno-speculate-indirect-jumps.
16111 (*call_value_indirect_nonlocal_sysv<mode>): Likewise.
16112 (*call_indirect_aix<mode>): Disable for
16113 -mno-speculate-indirect-jumps.
16114 (*call_indirect_aix<mode>_nospec): New define_insn.
16115 (*call_value_indirect_aix<mode>): Disable for
16116 -mno-speculate-indirect-jumps.
16117 (*call_value_indirect_aix<mode>_nospec): New define_insn.
16118 (*sibcall_nonlocal_sysv<mode>): Generate different code for
16119 -mno-speculate-indirect-jumps.
16120 (*sibcall_value_nonlocal_sysv<mode>): Likewise.
16122 2018-01-17 Michael Meissner <meissner@linux.vnet.ibm.com>
16124 * config/rs6000/rs6000.c (rs6000_emit_move): If we load or store a
16125 long double type, set the flags for noting the default long double
16126 type, even if we don't pass or return a long double type.
16128 2018-01-17 Jan Hubicka <hubicka@ucw.cz>
16131 * ipa-inline.c (flatten_function): Do not overwrite final inlining
16134 2018-01-17 Will Schmidt <will_schmidt@vnet.ibm.com>
16136 * config/rs6000/rs6000.c (rs6000_gimple_builtin): Add gimple folding
16137 support for merge[hl].
16138 (fold_mergehl_helper): New helper function.
16139 (tree-vector-builder.h): New #include for tree_vector_builder usage.
16140 * config/rs6000/altivec.md (altivec_vmrghw_direct): Add xxmrghw insn.
16141 (altivec_vmrglw_direct): Add xxmrglw insn.
16143 2018-01-17 Andrew Waterman <andrew@sifive.com>
16145 * config/riscv/riscv.c (riscv_conditional_register_usage): If
16146 UNITS_PER_FP_ARG is 0, set call_used_regs to 1 for all FP regs.
16148 2018-01-17 David Malcolm <dmalcolm@redhat.com>
16151 * ipa-devirt.c (add_type_duplicate): When comparing memory layout,
16152 call the lto_location_cache before reading the
16153 DECL_SOURCE_LOCATION of the types.
16155 2018-01-17 Wilco Dijkstra <wdijkstr@arm.com>
16156 Richard Sandiford <richard.sandiford@linaro.org>
16158 * config/aarch64/aarch64.md (movti_aarch64): Use Uti constraint.
16159 * config/aarch64/aarch64.c (aarch64_mov128_immediate): New function.
16160 (aarch64_legitimate_constant_p): Just support CONST_DOUBLE
16161 SF/DF/TF mode to avoid creating illegal CONST_WIDE_INT immediates.
16162 * config/aarch64/aarch64-protos.h (aarch64_mov128_immediate):
16164 * config/aarch64/constraints.md (aarch64_movti_operand):
16166 * config/aarch64/predicates.md (Uti): Add new constraint.
16168 2018-01-17 Carl Love <cel@us.ibm.com>
16170 * config/rs6000/vsx.md (define_expand xl_len_r,
16171 define_expand stxvl, define_expand *stxvl): Add match_dup argument.
16172 (define_insn): Add, match_dup 1 argument to define_insn stxvll and
16174 (define_expand, define_insn): Move the shift left from the
16175 define_insn to the define_expand for lxvl and stxvl instructions.
16176 * config/rs6000/rs6000-builtin.def (BU_P9V_64BIT_VSX_2): Change LXVL
16177 and XL_LEN_R definitions to PURE.
16179 2018-01-17 Uros Bizjak <ubizjak@gmail.com>
16181 * config/i386/i386.c (indirect_thunk_name): Declare regno
16182 as unsigned int. Compare regno with INVALID_REGNUM.
16183 (output_indirect_thunk): Ditto.
16184 (output_indirect_thunk_function): Ditto.
16185 (ix86_code_end): Declare regno as unsigned int. Use INVALID_REGNUM
16186 in the call to output_indirect_thunk_function.
16188 2018-01-17 Richard Sandiford <richard.sandiford@linaro.org>
16190 PR middle-end/83884
16191 * expr.c (expand_expr_real_1): Use the size of GET_MODE (op0)
16192 rather than the size of inner_type to determine the stack slot size
16193 when handling VIEW_CONVERT_EXPRs on strict-alignment targets.
16195 2018-01-16 Sebastian Peryt <sebastian.peryt@intel.com>
16198 * config/i386/i386.c (ix86_option_override_internal): Add PTA_RDRND
16201 2018-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
16203 * config.gcc (powerpc*-linux*-*): Add support for 64-bit little
16204 endian Linux systems to optionally enable multilibs for selecting
16205 the long double type if the user configured an explicit type.
16206 * config/rs6000/rs6000.h (TARGET_IEEEQUAD_MULTILIB): Indicate we
16207 have no long double multilibs if not defined.
16208 * config/rs6000/rs6000.c (rs6000_option_override_internal): Do not
16209 warn if the user used -mabi={ieee,ibm}longdouble and we built
16210 multilibs for long double.
16211 * config/rs6000/linux64.h (MULTILIB_DEFAULTS_IEEE): Define as the
16212 appropriate multilib option.
16213 (MULTILIB_DEFAULTS): Add MULTILIB_DEFAULTS_IEEE to the default
16215 * config/rs6000/t-ldouble-linux64le-ibm: New configuration files
16216 for building long double multilibs.
16217 * config/rs6000/t-ldouble-linux64le-ieee: Likewise.
16219 2018-01-16 John David Anglin <danglin@gcc.gnu.org>
16221 * config.gcc (hppa*-*-linux*): Change callee copies ABI to caller
16224 * config/pa.h (MALLOC_ABI_ALIGNMENT): Set 32-bit alignment default to
16226 * config/pa/pa32-linux.h (MALLOC_ABI_ALIGNMENT): Set alignment to
16229 * config/pa/som.h (ASM_DECLARE_FUNCTION_NAME): Cleanup type and mode
16232 * config/pa/pa.c (pa_function_arg_size): Apply CEIL to GET_MODE_SIZE
16235 2018-01-16 Eric Botcazou <ebotcazou@adacore.com>
16237 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): For an
16238 ADDR_EXPR, do not count the offset of a COMPONENT_REF twice.
16240 2018-01-16 Kelvin Nilsen <kelvin@gcc.gnu.org>
16242 * config/rs6000/rs6000-p8swap.c (rs6000_gen_stvx): Generate
16243 different rtl trees depending on TARGET_64BIT.
16244 (rs6000_gen_lvx): Likewise.
16246 2018-01-16 Eric Botcazou <ebotcazou@adacore.com>
16248 * config/visium/visium.md (nop): Tweak comment.
16249 (hazard_nop): Likewise.
16251 2018-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16253 * config/rs6000/rs6000.c (rs6000_opt_vars): Add entry for
16254 -mspeculate-indirect-jumps.
16255 * config/rs6000/rs6000.md (*call_indirect_elfv2<mode>): Disable
16256 for -mno-speculate-indirect-jumps.
16257 (*call_indirect_elfv2<mode>_nospec): New define_insn.
16258 (*call_value_indirect_elfv2<mode>): Disable for
16259 -mno-speculate-indirect-jumps.
16260 (*call_value_indirect_elfv2<mode>_nospec): New define_insn.
16261 (indirect_jump): Emit different RTL for
16262 -mno-speculate-indirect-jumps.
16263 (*indirect_jump<mode>): Disable for
16264 -mno-speculate-indirect-jumps.
16265 (*indirect_jump<mode>_nospec): New define_insn.
16266 (tablejump): Emit different RTL for
16267 -mno-speculate-indirect-jumps.
16268 (tablejumpsi): Disable for -mno-speculate-indirect-jumps.
16269 (tablejumpsi_nospec): New define_expand.
16270 (tablejumpdi): Disable for -mno-speculate-indirect-jumps.
16271 (tablejumpdi_nospec): New define_expand.
16272 (*tablejump<mode>_internal1): Disable for
16273 -mno-speculate-indirect-jumps.
16274 (*tablejump<mode>_internal1_nospec): New define_insn.
16275 * config/rs6000/rs6000.opt (mspeculate-indirect-jumps): New
16278 2018-01-16 Artyom Skrobov tyomitch@gmail.com
16280 * caller-save.c (insert_save): Drop unnecessary parameter. All
16283 2018-01-16 Jakub Jelinek <jakub@redhat.com>
16284 Richard Biener <rguenth@suse.de>
16287 * gimplify.c (gimplify_one_sizepos): For is_gimple_constant (expr)
16288 return early, inline manually is_gimple_sizepos. Make sure if we
16289 call gimplify_expr we don't end up with a gimple constant.
16290 * tree.c (variably_modified_type_p): Don't return true for
16291 is_gimple_constant (_t). Inline manually is_gimple_sizepos.
16292 * gimplify.h (is_gimple_sizepos): Remove.
16294 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
16296 PR tree-optimization/83857
16297 * tree-vect-loop.c (vect_analyze_loop_operations): Don't call
16298 vectorizable_live_operation for pure SLP statements.
16299 (vectorizable_live_operation): Handle PHIs.
16301 2018-01-16 Richard Biener <rguenther@suse.de>
16303 PR tree-optimization/83867
16304 * tree-vect-stmts.c (vect_transform_stmt): Precompute
16305 nested_in_vect_loop_p since the scalar stmt may get invalidated.
16307 2018-01-16 Jakub Jelinek <jakub@redhat.com>
16310 * stor-layout.c (handle_warn_if_not_align): Use byte_position and
16311 multiple_of_p instead of unchecked tree_to_uhwi and UHWI check.
16312 If off is not INTEGER_CST, issue a may not be aligned warning
16313 rather than isn't aligned. Use isn%'t rather than isn't.
16314 * fold-const.c (multiple_of_p) <case BIT_AND_EXPR>: Don't fall through
16316 <case MULT_EXPR>: Improve the case when bottom and one of the
16317 MULT_EXPR operands are INTEGER_CSTs and bottom is multiple of that
16318 operand, in that case check if the other operand is multiple of
16319 bottom divided by the INTEGER_CST operand.
16321 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
16324 * config/pa/pa.h (FUNCTION_ARG_SIZE): Delete.
16325 * config/pa/pa-protos.h (pa_function_arg_size): Declare.
16326 * config/pa/som.h (ASM_DECLARE_FUNCTION_NAME): Use
16327 pa_function_arg_size instead of FUNCTION_ARG_SIZE.
16328 * config/pa/pa.c (pa_function_arg_advance): Likewise.
16329 (pa_function_arg, pa_arg_partial_bytes): Likewise.
16330 (pa_function_arg_size): New function.
16332 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
16334 * fold-const.c (fold_ternary_loc): Construct the vec_perm_indices
16335 in a separate statement.
16337 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
16339 PR tree-optimization/83847
16340 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Don't
16341 group gathers and scatters.
16343 2018-01-16 Jakub Jelinek <jakub@redhat.com>
16345 PR rtl-optimization/86620
16346 * params.def (max-sched-ready-insns): Bump minimum value to 1.
16348 PR rtl-optimization/83213
16349 * recog.c (peep2_attempt): Copy over CROSSING_JUMP_P from peepinsn
16350 to last if both are JUMP_INSNs.
16352 PR tree-optimization/83843
16353 * gimple-ssa-store-merging.c
16354 (imm_store_chain_info::output_merged_store): Handle bit_not_p on
16355 store_immediate_info for bswap/nop orig_stores.
16357 2018-01-15 Andrew Waterman <andrew@sifive.com>
16359 * config/riscv/riscv.c (riscv_rtx_costs) <MULT>: Increase cost if
16361 <UDIV>: Increase cost if !TARGET_DIV.
16363 2018-01-15 Segher Boessenkool <segher@kernel.crashing.org>
16365 * config/rs6000/rs6000.md (define_attr "type"): Remove delayed_cr.
16366 (define_attr "cr_logical_3op"): New.
16367 (cceq_ior_compare): Adjust.
16368 (cceq_ior_compare_complement): Adjust.
16369 (*cceq_rev_compare): Adjust.
16370 * config/rs6000/rs6000.c (rs6000_adjust_cost): Adjust.
16371 (is_cracked_insn): Adjust.
16372 (insn_must_be_first_in_group): Adjust.
16373 * config/rs6000/40x.md: Adjust.
16374 * config/rs6000/440.md: Adjust.
16375 * config/rs6000/476.md: Adjust.
16376 * config/rs6000/601.md: Adjust.
16377 * config/rs6000/603.md: Adjust.
16378 * config/rs6000/6xx.md: Adjust.
16379 * config/rs6000/7450.md: Adjust.
16380 * config/rs6000/7xx.md: Adjust.
16381 * config/rs6000/8540.md: Adjust.
16382 * config/rs6000/cell.md: Adjust.
16383 * config/rs6000/e300c2c3.md: Adjust.
16384 * config/rs6000/e500mc.md: Adjust.
16385 * config/rs6000/e500mc64.md: Adjust.
16386 * config/rs6000/e5500.md: Adjust.
16387 * config/rs6000/e6500.md: Adjust.
16388 * config/rs6000/mpc.md: Adjust.
16389 * config/rs6000/power4.md: Adjust.
16390 * config/rs6000/power5.md: Adjust.
16391 * config/rs6000/power6.md: Adjust.
16392 * config/rs6000/power7.md: Adjust.
16393 * config/rs6000/power8.md: Adjust.
16394 * config/rs6000/power9.md: Adjust.
16395 * config/rs6000/rs64.md: Adjust.
16396 * config/rs6000/titan.md: Adjust.
16398 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
16400 * config/i386/predicates.md (indirect_branch_operand): Rewrite
16401 ix86_indirect_branch_register logic.
16403 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
16405 * config/i386/constraints.md (Bs): Update
16406 ix86_indirect_branch_register check. Don't check
16407 ix86_indirect_branch_register with GOT_memory_operand.
16409 * config/i386/predicates.md (GOT_memory_operand): Don't check
16410 ix86_indirect_branch_register here.
16411 (GOT32_symbol_operand): Likewise.
16413 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
16415 * config/i386/predicates.md (constant_call_address_operand):
16416 Rewrite ix86_indirect_branch_register logic.
16417 (sibcall_insn_operand): Likewise.
16419 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
16421 * config/i386/constraints.md (Bs): Replace
16422 ix86_indirect_branch_thunk_register with
16423 ix86_indirect_branch_register.
16425 * config/i386/i386.md (indirect_jump): Likewise.
16426 (tablejump): Likewise.
16427 (*sibcall_memory): Likewise.
16428 (*sibcall_value_memory): Likewise.
16429 Peepholes of indirect call and jump via memory: Likewise.
16430 * config/i386/i386.opt: Likewise.
16431 * config/i386/predicates.md (indirect_branch_operand): Likewise.
16432 (GOT_memory_operand): Likewise.
16433 (call_insn_operand): Likewise.
16434 (sibcall_insn_operand): Likewise.
16435 (GOT32_symbol_operand): Likewise.
16437 2018-01-15 Jakub Jelinek <jakub@redhat.com>
16439 PR middle-end/83837
16440 * omp-expand.c (expand_omp_atomic_pipeline): Use loaded_val
16441 type rather than type addr's type points to.
16442 (expand_omp_atomic_mutex): Likewise.
16443 (expand_omp_atomic): Likewise.
16445 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
16448 * config/i386/i386.c (output_indirect_thunk_function): Use
16449 ASM_OUTPUT_LABEL, instead of ASM_OUTPUT_DEF, for TARGET_MACHO
16450 for __x86_return_thunk.
16452 2018-01-15 Richard Biener <rguenther@suse.de>
16454 PR middle-end/83850
16455 * expmed.c (extract_bit_field_1): Fix typo.
16457 2018-01-15 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16460 * config/arm/iterators.md (VF): New mode iterator.
16461 * config/arm/neon.md (neon_vabd<mode>_2): Use the above.
16462 Remove integer-related logic from pattern.
16463 (neon_vabd<mode>_3): Likewise.
16465 2018-01-15 Jakub Jelinek <jakub@redhat.com>
16467 PR middle-end/82694
16468 * common.opt (fstrict-overflow): No longer an alias.
16469 (fwrapv-pointer): New option.
16470 * tree.h (TYPE_OVERFLOW_WRAPS, TYPE_OVERFLOW_UNDEFINED): Define
16471 also for pointer types based on flag_wrapv_pointer.
16472 * opts.c (common_handle_option) <case OPT_fstrict_overflow>: Set
16473 opts->x_flag_wrap[pv] to !value, clear opts->x_flag_trapv if
16474 opts->x_flag_wrapv got set.
16475 * fold-const.c (fold_comparison, fold_binary_loc): Revert 2017-08-01
16476 changes, just use TYPE_OVERFLOW_UNDEFINED on pointer type instead of
16477 POINTER_TYPE_OVERFLOW_UNDEFINED.
16478 * match.pd: Likewise in address comparison pattern.
16479 * doc/invoke.texi: Document -fwrapv and -fstrict-overflow.
16481 2018-01-15 Richard Biener <rguenther@suse.de>
16484 * tree.c (free_lang_data_in_type): Always unlink TYPE_DECLs
16485 from TYPE_FIELDS. Free TYPE_BINFO if not used by devirtualization.
16486 Reset type names to their identifier if their TYPE_DECL doesn't
16487 have linkage (and thus is used for ODR and devirt).
16488 (save_debug_info_for_decl): Remove.
16489 (save_debug_info_for_type): Likewise.
16490 (add_tree_to_fld_list): Adjust.
16491 * tree-pretty-print.c (dump_generic_node): Make dumping of
16492 type names more robust.
16494 2018-01-15 Richard Biener <rguenther@suse.de>
16496 * BASE-VER: Bump to 8.0.1.
16498 2018-01-14 Martin Sebor <msebor@redhat.com>
16501 * builtins.c (check_access): Avoid warning when the no-warning bit
16504 2018-01-14 Cory Fields <cory-nospam-@coryfields.com>
16506 * tree-ssa-loop-im.c (sort_bbs_in_loop_postorder_cmp): Stabilize sort.
16507 * ira-color (allocno_hard_regs_compare): Likewise.
16509 2018-01-14 Nathan Rossi <nathan@nathanrossi.com>
16512 * config/microblaze/microblaze.c (microblaze_asm_output_ident):
16513 Use .pushsection/.popsection.
16515 2018-01-14 Martin Sebor <msebor@redhat.com>
16518 * doc/invoke.texi (-Wlass-memaccess): Document suppression by casting.
16520 2018-01-14 Jakub Jelinek <jakub@redhat.com>
16522 * config.gcc (i[34567]86-*-*): Remove one duplicate gfniintrin.h
16523 entry from extra_headers.
16524 (x86_64-*-*): Remove two duplicate gfniintrin.h entries from
16525 extra_headers, make the list bitwise identical to the i?86-*-* one.
16527 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
16529 * config/i386/i386.c (ix86_set_indirect_branch_type): Disallow
16530 -mcmodel=large with -mindirect-branch=thunk,
16531 -mindirect-branch=thunk-extern, -mfunction-return=thunk and
16532 -mfunction-return=thunk-extern.
16533 * doc/invoke.texi: Document -mcmodel=large is incompatible with
16534 -mindirect-branch=thunk, -mindirect-branch=thunk-extern,
16535 -mfunction-return=thunk and -mfunction-return=thunk-extern.
16537 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
16539 * config/i386/i386.c (print_reg): Print the name of the full
16540 integer register without '%'.
16541 (ix86_print_operand): Handle 'V'.
16542 * doc/extend.texi: Document 'V' modifier.
16544 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
16546 * config/i386/constraints.md (Bs): Disallow memory operand for
16547 -mindirect-branch-register.
16549 * config/i386/predicates.md (indirect_branch_operand): Likewise.
16550 (GOT_memory_operand): Likewise.
16551 (call_insn_operand): Likewise.
16552 (sibcall_insn_operand): Likewise.
16553 (GOT32_symbol_operand): Likewise.
16554 * config/i386/i386.md (indirect_jump): Call convert_memory_address
16555 for -mindirect-branch-register.
16556 (tablejump): Likewise.
16557 (*sibcall_memory): Likewise.
16558 (*sibcall_value_memory): Likewise.
16559 Disallow peepholes of indirect call and jump via memory for
16560 -mindirect-branch-register.
16561 (*call_pop): Replace m with Bw.
16562 (*call_value_pop): Likewise.
16563 (*sibcall_pop_memory): Replace m with Bs.
16564 * config/i386/i386.opt (mindirect-branch-register): New option.
16565 * doc/invoke.texi: Document -mindirect-branch-register option.
16567 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
16569 * config/i386/i386-protos.h (ix86_output_function_return): New.
16570 * config/i386/i386.c (ix86_set_indirect_branch_type): Also
16571 set function_return_type.
16572 (indirect_thunk_name): Add ret_p to indicate thunk for function
16574 (output_indirect_thunk_function): Pass false to
16575 indirect_thunk_name.
16576 (ix86_output_indirect_branch_via_reg): Likewise.
16577 (ix86_output_indirect_branch_via_push): Likewise.
16578 (output_indirect_thunk_function): Create alias for function
16579 return thunk if regno < 0.
16580 (ix86_output_function_return): New function.
16581 (ix86_handle_fndecl_attribute): Handle function_return.
16582 (ix86_attribute_table): Add function_return.
16583 * config/i386/i386.h (machine_function): Add
16584 function_return_type.
16585 * config/i386/i386.md (simple_return_internal): Use
16586 ix86_output_function_return.
16587 (simple_return_internal_long): Likewise.
16588 * config/i386/i386.opt (mfunction-return=): New option.
16589 (indirect_branch): Mention -mfunction-return=.
16590 * doc/extend.texi: Document function_return function attribute.
16591 * doc/invoke.texi: Document -mfunction-return= option.
16593 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
16595 * config/i386/i386-opts.h (indirect_branch): New.
16596 * config/i386/i386-protos.h (ix86_output_indirect_jmp): Likewise.
16597 * config/i386/i386.c (ix86_using_red_zone): Disallow red-zone
16598 with local indirect jump when converting indirect call and jump.
16599 (ix86_set_indirect_branch_type): New.
16600 (ix86_set_current_function): Call ix86_set_indirect_branch_type.
16601 (indirectlabelno): New.
16602 (indirect_thunk_needed): Likewise.
16603 (indirect_thunk_bnd_needed): Likewise.
16604 (indirect_thunks_used): Likewise.
16605 (indirect_thunks_bnd_used): Likewise.
16606 (INDIRECT_LABEL): Likewise.
16607 (indirect_thunk_name): Likewise.
16608 (output_indirect_thunk): Likewise.
16609 (output_indirect_thunk_function): Likewise.
16610 (ix86_output_indirect_branch_via_reg): Likewise.
16611 (ix86_output_indirect_branch_via_push): Likewise.
16612 (ix86_output_indirect_branch): Likewise.
16613 (ix86_output_indirect_jmp): Likewise.
16614 (ix86_code_end): Call output_indirect_thunk_function if needed.
16615 (ix86_output_call_insn): Call ix86_output_indirect_branch if
16617 (ix86_handle_fndecl_attribute): Handle indirect_branch.
16618 (ix86_attribute_table): Add indirect_branch.
16619 * config/i386/i386.h (machine_function): Add indirect_branch_type
16620 and has_local_indirect_jump.
16621 * config/i386/i386.md (indirect_jump): Set has_local_indirect_jump
16623 (tablejump): Likewise.
16624 (*indirect_jump): Use ix86_output_indirect_jmp.
16625 (*tablejump_1): Likewise.
16626 (simple_return_indirect_internal): Likewise.
16627 * config/i386/i386.opt (mindirect-branch=): New option.
16628 (indirect_branch): New.
16631 (thunk-inline): Likewise.
16632 (thunk-extern): Likewise.
16633 * doc/extend.texi: Document indirect_branch function attribute.
16634 * doc/invoke.texi: Document -mindirect-branch= option.
16636 2018-01-14 Jan Hubicka <hubicka@ucw.cz>
16639 * ipa-inline.c (edge_badness): Tolerate roundoff errors.
16641 2018-01-14 Richard Sandiford <richard.sandiford@linaro.org>
16643 * ipa-inline.c (want_inline_small_function_p): Return false if
16644 inlining has already failed with CIF_FINAL_ERROR.
16645 (update_caller_keys): Call want_inline_small_function_p before
16647 (update_callee_keys): Likewise.
16649 2018-01-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
16651 * config/rs6000/rs6000-p8swap.c (rs6000_sum_of_two_registers_p):
16653 (rs6000_quadword_masked_address_p): Likewise.
16654 (quad_aligned_load_p): Likewise.
16655 (quad_aligned_store_p): Likewise.
16656 (const_load_sequence_p): Add comment to describe the outer-most loop.
16657 (mimic_memory_attributes_and_flags): New function.
16658 (rs6000_gen_stvx): Likewise.
16659 (replace_swapped_aligned_store): Likewise.
16660 (rs6000_gen_lvx): Likewise.
16661 (replace_swapped_aligned_load): Likewise.
16662 (replace_swapped_load_constant): Capitalize argument name in
16663 comment describing this function.
16664 (rs6000_analyze_swaps): Add a third pass to search for vector loads
16665 and stores that access quad-word aligned addresses and replace
16666 with stvx or lvx instructions when appropriate.
16667 * config/rs6000/rs6000-protos.h (rs6000_sum_of_two_registers_p):
16668 New function prototype.
16669 (rs6000_quadword_masked_address_p): Likewise.
16670 (rs6000_gen_lvx): Likewise.
16671 (rs6000_gen_stvx): Likewise.
16672 * config/rs6000/vsx.md (*vsx_le_perm_load_<mode>): For modes
16673 VSX_D (V2DF, V2DI), modify this split to select lvx instruction
16674 when memory address is aligned.
16675 (*vsx_le_perm_load_<mode>): For modes VSX_W (V4SF, V4SI), modify
16676 this split to select lvx instruction when memory address is aligned.
16677 (*vsx_le_perm_load_v8hi): Modify this split to select lvx
16678 instruction when memory address is aligned.
16679 (*vsx_le_perm_load_v16qi): Likewise.
16680 (four unnamed splitters): Modify to select the stvx instruction
16681 when memory is aligned.
16683 2018-01-13 Jan Hubicka <hubicka@ucw.cz>
16685 * predict.c (determine_unlikely_bbs): Handle correctly BBs
16686 which appears in the queue multiple times.
16688 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
16689 Alan Hayward <alan.hayward@arm.com>
16690 David Sherwood <david.sherwood@arm.com>
16692 * tree-vectorizer.h (vec_lower_bound): New structure.
16693 (_loop_vec_info): Add check_nonzero and lower_bounds.
16694 (LOOP_VINFO_CHECK_NONZERO): New macro.
16695 (LOOP_VINFO_LOWER_BOUNDS): Likewise.
16696 (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Check lower_bounds too.
16697 * tree-data-ref.h (dr_with_seg_len): Add access_size and align
16698 fields. Make seg_len the distance travelled, not including the
16700 (dr_direction_indicator): Declare.
16701 (dr_zero_step_indicator): Likewise.
16702 (dr_known_forward_stride_p): Likewise.
16703 * tree-data-ref.c: Include stringpool.h, tree-vrp.h and
16705 (runtime_alias_check_p): Allow runtime alias checks with
16707 (operator ==): Compare access_size and align.
16708 (prune_runtime_alias_test_list): Rework for new distinction between
16709 the access_size and seg_len.
16710 (create_intersect_range_checks_index): Likewise. Cope with polynomial
16712 (get_segment_min_max): New function.
16713 (create_intersect_range_checks): Use it.
16714 (dr_step_indicator): New function.
16715 (dr_direction_indicator): Likewise.
16716 (dr_zero_step_indicator): Likewise.
16717 (dr_known_forward_stride_p): Likewise.
16718 * tree-loop-distribution.c (data_ref_segment_size): Return
16719 DR_STEP * (niters - 1).
16720 (compute_alias_check_pairs): Update call to the dr_with_seg_len
16722 * tree-vect-data-refs.c (vect_check_nonzero_value): New function.
16723 (vect_preserves_scalar_order_p): New function, split out from...
16724 (vect_analyze_data_ref_dependence): ...here. Check for zero steps.
16725 (vect_vfa_segment_size): Return DR_STEP * (length_factor - 1).
16726 (vect_vfa_access_size): New function.
16727 (vect_vfa_align): Likewise.
16728 (vect_compile_time_alias): Take access_size_a and access_b arguments.
16729 (dump_lower_bound): New function.
16730 (vect_check_lower_bound): Likewise.
16731 (vect_small_gap_p): Likewise.
16732 (vectorizable_with_step_bound_p): Likewise.
16733 (vect_prune_runtime_alias_test_list): Ignore cross-iteration
16734 depencies if the vectorization factor is 1. Convert the checks
16735 for nonzero steps into checks on the bounds of DR_STEP. Try using
16736 a bunds check for variable steps if the minimum required step is
16737 relatively small. Update calls to the dr_with_seg_len
16738 constructor and to vect_compile_time_alias.
16739 * tree-vect-loop-manip.c (vect_create_cond_for_lower_bounds): New
16741 (vect_loop_versioning): Call it.
16742 * tree-vect-loop.c (vect_analyze_loop_2): Clear LOOP_VINFO_LOWER_BOUNDS
16744 (vect_estimate_min_profitable_iters): Account for any bounds checks.
16746 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
16747 Alan Hayward <alan.hayward@arm.com>
16748 David Sherwood <david.sherwood@arm.com>
16750 * doc/sourcebuild.texi (vect_scatter_store): Document.
16751 * optabs.def (scatter_store_optab, mask_scatter_store_optab): New
16753 * doc/md.texi (scatter_store@var{m}, mask_scatter_store@var{m}):
16755 * genopinit.c (main): Add supports_vec_scatter_store and
16756 supports_vec_scatter_store_cached to target_optabs.
16757 * gimple.h (gimple_expr_type): Handle IFN_SCATTER_STORE and
16758 IFN_MASK_SCATTER_STORE.
16759 * internal-fn.def (SCATTER_STORE, MASK_SCATTER_STORE): New internal
16761 * internal-fn.h (internal_store_fn_p): Declare.
16762 (internal_fn_stored_value_index): Likewise.
16763 * internal-fn.c (scatter_store_direct): New macro.
16764 (expand_scatter_store_optab_fn): New function.
16765 (direct_scatter_store_optab_supported_p): New macro.
16766 (internal_store_fn_p): New function.
16767 (internal_gather_scatter_fn_p): Handle IFN_SCATTER_STORE and
16768 IFN_MASK_SCATTER_STORE.
16769 (internal_fn_mask_index): Likewise.
16770 (internal_fn_stored_value_index): New function.
16771 (internal_gather_scatter_fn_supported_p): Adjust operand numbers
16772 for scatter stores.
16773 * optabs-query.h (supports_vec_scatter_store_p): Declare.
16774 * optabs-query.c (supports_vec_scatter_store_p): New function.
16775 * tree-vectorizer.h (vect_get_store_rhs): Declare.
16776 * tree-vect-data-refs.c (vect_analyze_data_ref_access): Return
16777 true for scatter stores.
16778 (vect_gather_scatter_fn_p): Handle scatter stores too.
16779 (vect_check_gather_scatter): Consider using scatter stores if
16780 supports_vec_scatter_store_p.
16781 * tree-vect-patterns.c (vect_try_gather_scatter_pattern): Handle
16782 scatter stores too.
16783 * tree-vect-stmts.c (exist_non_indexing_operands_for_use_p): Use
16784 internal_fn_stored_value_index.
16785 (check_load_store_masking): Handle scatter stores too.
16786 (vect_get_store_rhs): Make public.
16787 (vectorizable_call): Use internal_store_fn_p.
16788 (vectorizable_store): Handle scatter store internal functions.
16789 (vect_transform_stmt): Compare GROUP_STORE_COUNT with GROUP_SIZE
16790 when deciding whether the end of the group has been reached.
16791 * config/aarch64/aarch64.md (UNSPEC_ST1_SCATTER): New unspec.
16792 * config/aarch64/aarch64-sve.md (scatter_store<mode>): New expander.
16793 (mask_scatter_store<mode>): New insns.
16795 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
16796 Alan Hayward <alan.hayward@arm.com>
16797 David Sherwood <david.sherwood@arm.com>
16799 * tree-vectorizer.h (vect_gather_scatter_fn_p): Declare.
16800 * tree-vect-data-refs.c (vect_gather_scatter_fn_p): Make public.
16801 * tree-vect-stmts.c (vect_truncate_gather_scatter_offset): New
16803 (vect_use_strided_gather_scatters_p): Take a masked_p argument.
16804 Use vect_truncate_gather_scatter_offset if we can't treat the
16805 operation as a normal gather load or scatter store.
16806 (get_group_load_store_type): Take the gather_scatter_info
16807 as argument. Try using a gather load or scatter store for
16808 single-element groups.
16809 (get_load_store_type): Update calls to get_group_load_store_type
16810 and vect_use_strided_gather_scatters_p.
16812 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
16813 Alan Hayward <alan.hayward@arm.com>
16814 David Sherwood <david.sherwood@arm.com>
16816 * tree-vectorizer.h (vect_create_data_ref_ptr): Take an extra
16817 optional tree argument.
16818 * tree-vect-data-refs.c (vect_check_gather_scatter): Check for
16820 (vect_create_data_ref_ptr): Take the iv_step as an optional argument,
16821 but continue to use the current value as a fallback.
16822 (bump_vector_ptr): Use operand_equal_p rather than tree_int_cst_compare
16823 to compare the updates.
16824 * tree-vect-stmts.c (vect_use_strided_gather_scatters_p): New function.
16825 (get_load_store_type): Use it when handling a strided access.
16826 (vect_get_strided_load_store_ops): New function.
16827 (vect_get_data_ptr_increment): Likewise.
16828 (vectorizable_load): Handle strided gather loads. Always pass
16829 a step to vect_create_data_ref_ptr and bump_vector_ptr.
16831 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
16832 Alan Hayward <alan.hayward@arm.com>
16833 David Sherwood <david.sherwood@arm.com>
16835 * doc/md.texi (gather_load@var{m}): Document.
16836 (mask_gather_load@var{m}): Likewise.
16837 * genopinit.c (main): Add supports_vec_gather_load and
16838 supports_vec_gather_load_cached to target_optabs.
16839 * optabs-tree.c (init_tree_optimization_optabs): Use
16840 ggc_cleared_alloc to allocate target_optabs.
16841 * optabs.def (gather_load_optab, mask_gather_laod_optab): New optabs.
16842 * internal-fn.def (GATHER_LOAD, MASK_GATHER_LOAD): New internal
16844 * internal-fn.h (internal_load_fn_p): Declare.
16845 (internal_gather_scatter_fn_p): Likewise.
16846 (internal_fn_mask_index): Likewise.
16847 (internal_gather_scatter_fn_supported_p): Likewise.
16848 * internal-fn.c (gather_load_direct): New macro.
16849 (expand_gather_load_optab_fn): New function.
16850 (direct_gather_load_optab_supported_p): New macro.
16851 (direct_internal_fn_optab): New function.
16852 (internal_load_fn_p): Likewise.
16853 (internal_gather_scatter_fn_p): Likewise.
16854 (internal_fn_mask_index): Likewise.
16855 (internal_gather_scatter_fn_supported_p): Likewise.
16856 * optabs-query.c (supports_at_least_one_mode_p): New function.
16857 (supports_vec_gather_load_p): Likewise.
16858 * optabs-query.h (supports_vec_gather_load_p): Declare.
16859 * tree-vectorizer.h (gather_scatter_info): Add ifn, element_type
16860 and memory_type field.
16861 (NUM_PATTERNS): Bump to 15.
16862 * tree-vect-data-refs.c: Include internal-fn.h.
16863 (vect_gather_scatter_fn_p): New function.
16864 (vect_describe_gather_scatter_call): Likewise.
16865 (vect_check_gather_scatter): Try using internal functions for
16866 gather loads. Recognize existing calls to a gather load function.
16867 (vect_analyze_data_refs): Consider using gather loads if
16868 supports_vec_gather_load_p.
16869 * tree-vect-patterns.c (vect_get_load_store_mask): New function.
16870 (vect_get_gather_scatter_offset_type): Likewise.
16871 (vect_convert_mask_for_vectype): Likewise.
16872 (vect_add_conversion_to_patterm): Likewise.
16873 (vect_try_gather_scatter_pattern): Likewise.
16874 (vect_recog_gather_scatter_pattern): New pattern recognizer.
16875 (vect_vect_recog_func_ptrs): Add it.
16876 * tree-vect-stmts.c (exist_non_indexing_operands_for_use_p): Use
16877 internal_fn_mask_index and internal_gather_scatter_fn_p.
16878 (check_load_store_masking): Take the gather_scatter_info as an
16879 argument and handle gather loads.
16880 (vect_get_gather_scatter_ops): New function.
16881 (vectorizable_call): Check internal_load_fn_p.
16882 (vectorizable_load): Likewise. Handle gather load internal
16884 (vectorizable_store): Update call to check_load_store_masking.
16885 * config/aarch64/aarch64.md (UNSPEC_LD1_GATHER): New unspec.
16886 * config/aarch64/iterators.md (SVE_S, SVE_D): New mode iterators.
16887 * config/aarch64/predicates.md (aarch64_gather_scale_operand_w)
16888 (aarch64_gather_scale_operand_d): New predicates.
16889 * config/aarch64/aarch64-sve.md (gather_load<mode>): New expander.
16890 (mask_gather_load<mode>): New insns.
16892 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
16893 Alan Hayward <alan.hayward@arm.com>
16894 David Sherwood <david.sherwood@arm.com>
16896 * optabs.def (fold_left_plus_optab): New optab.
16897 * doc/md.texi (fold_left_plus_@var{m}): Document.
16898 * internal-fn.def (IFN_FOLD_LEFT_PLUS): New internal function.
16899 * internal-fn.c (fold_left_direct): Define.
16900 (expand_fold_left_optab_fn): Likewise.
16901 (direct_fold_left_optab_supported_p): Likewise.
16902 * fold-const-call.c (fold_const_fold_left): New function.
16903 (fold_const_call): Use it to fold CFN_FOLD_LEFT_PLUS.
16904 * tree-parloops.c (valid_reduction_p): New function.
16905 (gather_scalar_reductions): Use it.
16906 * tree-vectorizer.h (FOLD_LEFT_REDUCTION): New vect_reduction_type.
16907 (vect_finish_replace_stmt): Declare.
16908 * tree-vect-loop.c (fold_left_reduction_fn): New function.
16909 (needs_fold_left_reduction_p): New function, split out from...
16910 (vect_is_simple_reduction): ...here. Accept reductions that
16911 forbid reassociation, but give them type FOLD_LEFT_REDUCTION.
16912 (vect_force_simple_reduction): Also store the reduction type in
16913 the assignment's STMT_VINFO_REDUC_TYPE.
16914 (vect_model_reduction_cost): Handle FOLD_LEFT_REDUCTION.
16915 (merge_with_identity): New function.
16916 (vect_expand_fold_left): Likewise.
16917 (vectorize_fold_left_reduction): Likewise.
16918 (vectorizable_reduction): Handle FOLD_LEFT_REDUCTION. Leave the
16919 scalar phi in place for it. Check for target support and reject
16920 cases that would reassociate the operation. Defer the transform
16921 phase to vectorize_fold_left_reduction.
16922 * config/aarch64/aarch64.md (UNSPEC_FADDA): New unspec.
16923 * config/aarch64/aarch64-sve.md (fold_left_plus_<mode>): New expander.
16924 (*fold_left_plus_<mode>, *pred_fold_left_plus_<mode>): New insns.
16926 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
16928 * tree-if-conv.c (predicate_mem_writes): Remove redundant
16929 call to ifc_temp_var.
16931 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
16932 Alan Hayward <alan.hayward@arm.com>
16933 David Sherwood <david.sherwood@arm.com>
16935 * target.def (legitimize_address_displacement): Take the original
16936 offset as a poly_int.
16937 * targhooks.h (default_legitimize_address_displacement): Update
16939 * targhooks.c (default_legitimize_address_displacement): Likewise.
16940 * doc/tm.texi: Regenerate.
16941 * lra-constraints.c (base_plus_disp_to_reg): Take the displacement
16942 as an argument, moving assert of ad->disp == ad->disp_term to...
16943 (process_address_1): ...here. Update calls to base_plus_disp_to_reg.
16944 Try calling targetm.legitimize_address_displacement before expanding
16945 the address rather than afterwards, and adjust for the new interface.
16946 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
16947 Match the new hook interface. Handle SVE addresses.
16948 * config/sh/sh.c (sh_legitimize_address_displacement): Make the
16949 new hook interface.
16951 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
16953 * Makefile.in (OBJS): Add early-remat.o.
16954 * target.def (select_early_remat_modes): New hook.
16955 * doc/tm.texi.in (TARGET_SELECT_EARLY_REMAT_MODES): New hook.
16956 * doc/tm.texi: Regenerate.
16957 * targhooks.h (default_select_early_remat_modes): Declare.
16958 * targhooks.c (default_select_early_remat_modes): New function.
16959 * timevar.def (TV_EARLY_REMAT): New timevar.
16960 * passes.def (pass_early_remat): New pass.
16961 * tree-pass.h (make_pass_early_remat): Declare.
16962 * early-remat.c: New file.
16963 * config/aarch64/aarch64.c (aarch64_select_early_remat_modes): New
16965 (TARGET_SELECT_EARLY_REMAT_MODES): Define.
16967 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
16968 Alan Hayward <alan.hayward@arm.com>
16969 David Sherwood <david.sherwood@arm.com>
16971 * tree-vect-loop-manip.c (vect_gen_scalar_loop_niters): Replace
16972 vfm1 with a bound_epilog parameter.
16973 (vect_do_peeling): Update calls accordingly, and move the prologue
16974 call earlier in the function. Treat the base bound_epilog as 0 for
16975 fully-masked loops and retain vf - 1 for other loops. Add 1 to
16976 this base when peeling for gaps.
16977 * tree-vect-loop.c (vect_analyze_loop_2): Allow peeling for gaps
16978 with fully-masked loops.
16979 (vect_estimate_min_profitable_iters): Handle the single peeled
16980 iteration in that case.
16982 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
16983 Alan Hayward <alan.hayward@arm.com>
16984 David Sherwood <david.sherwood@arm.com>
16986 * tree-vect-data-refs.c (vect_analyze_group_access_1): Allow
16987 single-element interleaving even if the size is not a power of 2.
16988 * tree-vect-stmts.c (get_load_store_type): Disallow elementwise
16989 accesses for single-element interleaving if the group size is
16992 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
16993 Alan Hayward <alan.hayward@arm.com>
16994 David Sherwood <david.sherwood@arm.com>
16996 * doc/md.texi (fold_extract_last_@var{m}): Document.
16997 * doc/sourcebuild.texi (vect_fold_extract_last): Likewise.
16998 * optabs.def (fold_extract_last_optab): New optab.
16999 * internal-fn.def (FOLD_EXTRACT_LAST): New internal function.
17000 * internal-fn.c (fold_extract_direct): New macro.
17001 (expand_fold_extract_optab_fn): Likewise.
17002 (direct_fold_extract_optab_supported_p): Likewise.
17003 * tree-vectorizer.h (EXTRACT_LAST_REDUCTION): New vect_reduction_type.
17004 * tree-vect-loop.c (vect_model_reduction_cost): Handle
17005 EXTRACT_LAST_REDUCTION.
17006 (get_initial_def_for_reduction): Do not create an initial vector
17007 for EXTRACT_LAST_REDUCTION reductions.
17008 (vectorizable_reduction): Leave the scalar phi in place for
17009 EXTRACT_LAST_REDUCTIONs. Try using EXTRACT_LAST_REDUCTION
17010 ahead of INTEGER_INDUC_COND_REDUCTION. Do not check for an
17011 epilogue code for EXTRACT_LAST_REDUCTION and defer the
17012 transform phase to vectorizable_condition.
17013 * tree-vect-stmts.c (vect_finish_stmt_generation_1): New function,
17015 (vect_finish_stmt_generation): ...here.
17016 (vect_finish_replace_stmt): New function.
17017 (vectorizable_condition): Handle EXTRACT_LAST_REDUCTION.
17018 * config/aarch64/aarch64-sve.md (fold_extract_last_<mode>): New
17020 * config/aarch64/aarch64.md (UNSPEC_CLASTB): New unspec.
17022 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
17023 Alan Hayward <alan.hayward@arm.com>
17024 David Sherwood <david.sherwood@arm.com>
17026 * doc/md.texi (extract_last_@var{m}): Document.
17027 * optabs.def (extract_last_optab): New optab.
17028 * internal-fn.def (EXTRACT_LAST): New internal function.
17029 * internal-fn.c (cond_unary_direct): New macro.
17030 (expand_cond_unary_optab_fn): Likewise.
17031 (direct_cond_unary_optab_supported_p): Likewise.
17032 * tree-vect-loop.c (vectorizable_live_operation): Allow fully-masked
17033 loops using EXTRACT_LAST.
17034 * config/aarch64/aarch64-sve.md (aarch64_sve_lastb<mode>): Rename to...
17035 (extract_last_<mode>): ...this optab.
17036 (vec_extract<mode><Vel>): Update accordingly.
17038 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
17039 Alan Hayward <alan.hayward@arm.com>
17040 David Sherwood <david.sherwood@arm.com>
17042 * target.def (empty_mask_is_expensive): New hook.
17043 * doc/tm.texi.in (TARGET_VECTORIZE_EMPTY_MASK_IS_EXPENSIVE): New hook.
17044 * doc/tm.texi: Regenerate.
17045 * targhooks.h (default_empty_mask_is_expensive): Declare.
17046 * targhooks.c (default_empty_mask_is_expensive): New function.
17047 * tree-vectorizer.c (vectorize_loops): Only call optimize_mask_stores
17048 if the target says that empty masks are expensive.
17049 * config/aarch64/aarch64.c (aarch64_empty_mask_is_expensive):
17051 (TARGET_VECTORIZE_EMPTY_MASK_IS_EXPENSIVE): Redefine.
17053 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
17054 Alan Hayward <alan.hayward@arm.com>
17055 David Sherwood <david.sherwood@arm.com>
17057 * tree-vectorizer.h (_loop_vec_info::mask_skip_niters): New field.
17058 (LOOP_VINFO_MASK_SKIP_NITERS): New macro.
17059 (vect_use_loop_mask_for_alignment_p): New function.
17060 (vect_prepare_for_masked_peels, vect_gen_while_not): Declare.
17061 * tree-vect-loop-manip.c (vect_set_loop_masks_directly): Add an
17062 niters_skip argument. Make sure that the first niters_skip elements
17063 of the first iteration are inactive.
17064 (vect_set_loop_condition_masked): Handle LOOP_VINFO_MASK_SKIP_NITERS.
17065 Update call to vect_set_loop_masks_directly.
17066 (get_misalign_in_elems): New function, split out from...
17067 (vect_gen_prolog_loop_niters): ...here.
17068 (vect_update_init_of_dr): Take a code argument that specifies whether
17069 the adjustment should be added or subtracted.
17070 (vect_update_init_of_drs): Likewise.
17071 (vect_prepare_for_masked_peels): New function.
17072 (vect_do_peeling): Skip prologue peeling if we're using a mask
17073 instead. Update call to vect_update_inits_of_drs.
17074 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
17076 (vect_analyze_loop_2): Allow fully-masked loops with peeling for
17077 alignment. Do not include the number of peeled iterations in
17078 the minimum threshold in that case.
17079 (vectorizable_induction): Adjust the start value down by
17080 LOOP_VINFO_MASK_SKIP_NITERS iterations.
17081 (vect_transform_loop): Call vect_prepare_for_masked_peels.
17082 Take the number of skipped iterations into account when calculating
17084 * tree-vect-stmts.c (vect_gen_while_not): New function.
17086 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
17087 Alan Hayward <alan.hayward@arm.com>
17088 David Sherwood <david.sherwood@arm.com>
17090 * doc/sourcebuild.texi (vect_fully_masked): Document.
17091 * params.def (PARAM_MIN_VECT_LOOP_BOUND): Change minimum and
17092 default value to 0.
17093 * tree-vect-loop.c (vect_analyze_loop_costing): New function,
17095 (vect_analyze_loop_2): ...here. Don't check the vectorization
17096 factor against the number of loop iterations if the loop is
17099 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
17100 Alan Hayward <alan.hayward@arm.com>
17101 David Sherwood <david.sherwood@arm.com>
17103 * tree-ssa-loop-ivopts.c (USE_ADDRESS): Split into...
17104 (USE_REF_ADDRESS, USE_PTR_ADDRESS): ...these new use types.
17105 (dump_groups): Update accordingly.
17106 (iv_use::mem_type): New member variable.
17107 (address_p): New function.
17108 (record_use): Add a mem_type argument and initialize the new
17110 (record_group_use): Add a mem_type argument. Use address_p.
17111 Remove obsolete null checks of base_object. Update call to record_use.
17112 (find_interesting_uses_op): Update call to record_group_use.
17113 (find_interesting_uses_cond): Likewise.
17114 (find_interesting_uses_address): Likewise.
17115 (get_mem_type_for_internal_fn): New function.
17116 (find_address_like_use): Likewise.
17117 (find_interesting_uses_stmt): Try find_address_like_use before
17118 calling find_interesting_uses_op.
17119 (addr_offset_valid_p): Use the iv mem_type field as the type
17120 of the addressed memory.
17121 (add_autoinc_candidates): Likewise.
17122 (get_address_cost): Likewise.
17123 (split_small_address_groups_p): Use address_p.
17124 (split_address_groups): Likewise.
17125 (add_iv_candidate_for_use): Likewise.
17126 (autoinc_possible_for_pair): Likewise.
17127 (rewrite_groups): Likewise.
17128 (get_use_type): Check for USE_REF_ADDRESS instead of USE_ADDRESS.
17129 (determine_group_iv_cost): Update after split of USE_ADDRESS.
17130 (get_alias_ptr_type_for_ptr_address): New function.
17131 (rewrite_use_address): Rewrite address uses in calls that were
17132 identified by find_address_like_use.
17134 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
17135 Alan Hayward <alan.hayward@arm.com>
17136 David Sherwood <david.sherwood@arm.com>
17138 * expr.c (expand_expr_addr_expr_1): Handle ADDR_EXPRs of
17140 * gimple-expr.h (is_gimple_addressable: Likewise.
17141 * gimple-expr.c (is_gimple_address): Likewise.
17142 * internal-fn.c (expand_call_mem_ref): New function.
17143 (expand_mask_load_optab_fn): Use it.
17144 (expand_mask_store_optab_fn): Likewise.
17146 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
17147 Alan Hayward <alan.hayward@arm.com>
17148 David Sherwood <david.sherwood@arm.com>
17150 * doc/md.texi (cond_add@var{mode}, cond_sub@var{mode})
17151 (cond_and@var{mode}, cond_ior@var{mode}, cond_xor@var{mode})
17152 (cond_smin@var{mode}, cond_smax@var{mode}, cond_umin@var{mode})
17153 (cond_umax@var{mode}): Document.
17154 * optabs.def (cond_add_optab, cond_sub_optab, cond_and_optab)
17155 (cond_ior_optab, cond_xor_optab, cond_smin_optab, cond_smax_optab)
17156 (cond_umin_optab, cond_umax_optab): New optabs.
17157 * internal-fn.def (COND_ADD, COND_SUB, COND_MIN, COND_MAX, COND_AND)
17158 (COND_IOR, COND_XOR): New internal functions.
17159 * internal-fn.h (get_conditional_internal_fn): Declare.
17160 * internal-fn.c (cond_binary_direct): New macro.
17161 (expand_cond_binary_optab_fn): Likewise.
17162 (direct_cond_binary_optab_supported_p): Likewise.
17163 (get_conditional_internal_fn): New function.
17164 * tree-vect-loop.c (vectorizable_reduction): Handle fully-masked loops.
17165 Cope with reduction statements that are vectorized as calls rather
17167 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): New insns.
17168 * config/aarch64/iterators.md (UNSPEC_COND_ADD, UNSPEC_COND_SUB)
17169 (UNSPEC_COND_SMAX, UNSPEC_COND_UMAX, UNSPEC_COND_SMIN)
17170 (UNSPEC_COND_UMIN, UNSPEC_COND_AND, UNSPEC_COND_ORR)
17171 (UNSPEC_COND_EOR): New unspecs.
17172 (optab): Add mappings for them.
17173 (SVE_COND_INT_OP, SVE_COND_FP_OP): New int iterators.
17174 (sve_int_op, sve_fp_op): New int attributes.
17176 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
17177 Alan Hayward <alan.hayward@arm.com>
17178 David Sherwood <david.sherwood@arm.com>
17180 * optabs.def (while_ult_optab): New optab.
17181 * doc/md.texi (while_ult@var{m}@var{n}): Document.
17182 * internal-fn.def (WHILE_ULT): New internal function.
17183 * internal-fn.h (direct_internal_fn_supported_p): New override
17184 that takes two types as argument.
17185 * internal-fn.c (while_direct): New macro.
17186 (expand_while_optab_fn): New function.
17187 (convert_optab_supported_p): Likewise.
17188 (direct_while_optab_supported_p): New macro.
17189 * wide-int.h (wi::udiv_ceil): New function.
17190 * tree-vectorizer.h (rgroup_masks): New structure.
17191 (vec_loop_masks): New typedef.
17192 (_loop_vec_info): Add masks, mask_compare_type, can_fully_mask_p
17193 and fully_masked_p.
17194 (LOOP_VINFO_CAN_FULLY_MASK_P, LOOP_VINFO_FULLY_MASKED_P)
17195 (LOOP_VINFO_MASKS, LOOP_VINFO_MASK_COMPARE_TYPE): New macros.
17196 (vect_max_vf): New function.
17197 (slpeel_make_loop_iterate_ntimes): Delete.
17198 (vect_set_loop_condition, vect_get_loop_mask_type, vect_gen_while)
17199 (vect_halve_mask_nunits, vect_double_mask_nunits): Declare.
17200 (vect_record_loop_mask, vect_get_loop_mask): Likewise.
17201 * tree-vect-loop-manip.c: Include tree-ssa-loop-niter.h,
17202 internal-fn.h, stor-layout.h and optabs-query.h.
17203 (vect_set_loop_mask): New function.
17204 (add_preheader_seq): Likewise.
17205 (add_header_seq): Likewise.
17206 (interleave_supported_p): Likewise.
17207 (vect_maybe_permute_loop_masks): Likewise.
17208 (vect_set_loop_masks_directly): Likewise.
17209 (vect_set_loop_condition_masked): Likewise.
17210 (vect_set_loop_condition_unmasked): New function, split out from
17211 slpeel_make_loop_iterate_ntimes.
17212 (slpeel_make_loop_iterate_ntimes): Rename to..
17213 (vect_set_loop_condition): ...this. Use vect_set_loop_condition_masked
17214 for fully-masked loops and vect_set_loop_condition_unmasked otherwise.
17215 (vect_do_peeling): Update call accordingly.
17216 (vect_gen_vector_loop_niters): Use VF as the step for fully-masked
17218 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
17219 mask_compare_type, can_fully_mask_p and fully_masked_p.
17220 (release_vec_loop_masks): New function.
17221 (_loop_vec_info): Use it to free the loop masks.
17222 (can_produce_all_loop_masks_p): New function.
17223 (vect_get_max_nscalars_per_iter): Likewise.
17224 (vect_verify_full_masking): Likewise.
17225 (vect_analyze_loop_2): Save LOOP_VINFO_CAN_FULLY_MASK_P around
17226 retries, and free the mask rgroups before retrying. Check loop-wide
17227 reasons for disallowing fully-masked loops. Make the final decision
17228 about whether use a fully-masked loop or not.
17229 (vect_estimate_min_profitable_iters): Do not assume that peeling
17230 for the number of iterations will be needed for fully-masked loops.
17231 (vectorizable_reduction): Disable fully-masked loops.
17232 (vectorizable_live_operation): Likewise.
17233 (vect_halve_mask_nunits): New function.
17234 (vect_double_mask_nunits): Likewise.
17235 (vect_record_loop_mask): Likewise.
17236 (vect_get_loop_mask): Likewise.
17237 (vect_transform_loop): Handle the case in which the final loop
17238 iteration might handle a partial vector. Call vect_set_loop_condition
17239 instead of slpeel_make_loop_iterate_ntimes.
17240 * tree-vect-stmts.c: Include tree-ssa-loop-niter.h and gimple-fold.h.
17241 (check_load_store_masking): New function.
17242 (prepare_load_store_mask): Likewise.
17243 (vectorizable_store): Handle fully-masked loops.
17244 (vectorizable_load): Likewise.
17245 (supportable_widening_operation): Use vect_halve_mask_nunits for
17247 (supportable_narrowing_operation): Likewise vect_double_mask_nunits.
17248 (vect_gen_while): New function.
17249 * config/aarch64/aarch64.md (umax<mode>3): New expander.
17250 (aarch64_uqdec<mode>): New insn.
17252 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
17253 Alan Hayward <alan.hayward@arm.com>
17254 David Sherwood <david.sherwood@arm.com>
17256 * optabs.def (reduc_and_scal_optab, reduc_ior_scal_optab)
17257 (reduc_xor_scal_optab): New optabs.
17258 * doc/md.texi (reduc_and_scal_@var{m}, reduc_ior_scal_@var{m})
17259 (reduc_xor_scal_@var{m}): Document.
17260 * doc/sourcebuild.texi (vect_logical_reduc): Likewise.
17261 * internal-fn.def (IFN_REDUC_AND, IFN_REDUC_IOR, IFN_REDUC_XOR): New
17262 internal functions.
17263 * fold-const-call.c (fold_const_call): Handle them.
17264 * tree-vect-loop.c (reduction_fn_for_scalar_code): Return the new
17265 internal functions for BIT_AND_EXPR, BIT_IOR_EXPR and BIT_XOR_EXPR.
17266 * config/aarch64/aarch64-sve.md (reduc_<bit_reduc>_scal_<mode>):
17267 (*reduc_<bit_reduc>_scal_<mode>): New patterns.
17268 * config/aarch64/iterators.md (UNSPEC_ANDV, UNSPEC_ORV)
17269 (UNSPEC_XORV): New unspecs.
17270 (optab): Add entries for them.
17271 (BITWISEV): New int iterator.
17272 (bit_reduc_op): New int attributes.
17274 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
17275 Alan Hayward <alan.hayward@arm.com>
17276 David Sherwood <david.sherwood@arm.com>
17278 * doc/md.texi (vec_shl_insert_@var{m}): New optab.
17279 * internal-fn.def (VEC_SHL_INSERT): New internal function.
17280 * optabs.def (vec_shl_insert_optab): New optab.
17281 * tree-vectorizer.h (can_duplicate_and_interleave_p): Declare.
17282 (duplicate_and_interleave): Likewise.
17283 * tree-vect-loop.c: Include internal-fn.h.
17284 (neutral_op_for_slp_reduction): New function, split out from
17285 get_initial_defs_for_reduction.
17286 (get_initial_def_for_reduction): Handle option 2 for variable-length
17287 vectors by loading the neutral value into a vector and then shifting
17288 the initial value into element 0.
17289 (get_initial_defs_for_reduction): Replace the code argument with
17290 the neutral value calculated by neutral_op_for_slp_reduction.
17291 Use gimple_build_vector for constant-length vectors.
17292 Use IFN_VEC_SHL_INSERT for variable-length vectors if all
17293 but the first group_size elements have a neutral value.
17294 Use duplicate_and_interleave otherwise.
17295 (vect_create_epilog_for_reduction): Take a neutral_op parameter.
17296 Update call to get_initial_defs_for_reduction. Handle SLP
17297 reductions for variable-length vectors by creating one vector
17298 result for each scalar result, with the elements associated
17299 with other scalar results stubbed out with the neutral value.
17300 (vectorizable_reduction): Call neutral_op_for_slp_reduction.
17301 Require IFN_VEC_SHL_INSERT for double reductions on
17302 variable-length vectors, or SLP reductions that have
17303 a neutral value. Require can_duplicate_and_interleave_p
17304 support for variable-length unchained SLP reductions if there
17305 is no neutral value, such as for MIN/MAX reductions. Also require
17306 the number of vector elements to be a multiple of the number of
17307 SLP statements when doing variable-length unchained SLP reductions.
17308 Update call to vect_create_epilog_for_reduction.
17309 * tree-vect-slp.c (can_duplicate_and_interleave_p): Make public
17310 and remove initial values.
17311 (duplicate_and_interleave): Make public.
17312 * config/aarch64/aarch64.md (UNSPEC_INSR): New unspec.
17313 * config/aarch64/aarch64-sve.md (vec_shl_insert_<mode>): New insn.
17315 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
17316 Alan Hayward <alan.hayward@arm.com>
17317 David Sherwood <david.sherwood@arm.com>
17319 * tree-vect-slp.c: Include gimple-fold.h and internal-fn.h
17320 (can_duplicate_and_interleave_p): New function.
17321 (vect_get_and_check_slp_defs): Take the vector of statements
17322 rather than just the current one. Remove excess parentheses.
17323 Restriction rejectinon of vect_constant_def and vect_external_def
17324 for variable-length vectors to boolean types, or types for which
17325 can_duplicate_and_interleave_p is false.
17326 (vect_build_slp_tree_2): Update call to vect_get_and_check_slp_defs.
17327 (duplicate_and_interleave): New function.
17328 (vect_get_constant_vectors): Use gimple_build_vector for
17329 constant-length vectors and suitable variable-length constant
17330 vectors. Use duplicate_and_interleave for other variable-length
17331 vectors. Don't defer the update when inserting new statements.
17333 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
17334 Alan Hayward <alan.hayward@arm.com>
17335 David Sherwood <david.sherwood@arm.com>
17337 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Make sure
17338 min_profitable_iters doesn't go negative.
17340 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
17341 Alan Hayward <alan.hayward@arm.com>
17342 David Sherwood <david.sherwood@arm.com>
17344 * doc/md.texi (vec_mask_load_lanes@var{m}@var{n}): Document.
17345 (vec_mask_store_lanes@var{m}@var{n}): Likewise.
17346 * optabs.def (vec_mask_load_lanes_optab): New optab.
17347 (vec_mask_store_lanes_optab): Likewise.
17348 * internal-fn.def (MASK_LOAD_LANES): New internal function.
17349 (MASK_STORE_LANES): Likewise.
17350 * internal-fn.c (mask_load_lanes_direct): New macro.
17351 (mask_store_lanes_direct): Likewise.
17352 (expand_mask_load_optab_fn): Handle masked operations.
17353 (expand_mask_load_lanes_optab_fn): New macro.
17354 (expand_mask_store_optab_fn): Handle masked operations.
17355 (expand_mask_store_lanes_optab_fn): New macro.
17356 (direct_mask_load_lanes_optab_supported_p): Likewise.
17357 (direct_mask_store_lanes_optab_supported_p): Likewise.
17358 * tree-vectorizer.h (vect_store_lanes_supported): Take a masked_p
17360 (vect_load_lanes_supported): Likewise.
17361 * tree-vect-data-refs.c (strip_conversion): New function.
17362 (can_group_stmts_p): Likewise.
17363 (vect_analyze_data_ref_accesses): Use it instead of checking
17364 for a pair of assignments.
17365 (vect_store_lanes_supported): Take a masked_p parameter.
17366 (vect_load_lanes_supported): Likewise.
17367 * tree-vect-loop.c (vect_analyze_loop_2): Update calls to
17368 vect_store_lanes_supported and vect_load_lanes_supported.
17369 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
17370 * tree-vect-stmts.c (get_group_load_store_type): Take a masked_p
17371 parameter. Don't allow gaps for masked accesses.
17372 Use vect_get_store_rhs. Update calls to vect_store_lanes_supported
17373 and vect_load_lanes_supported.
17374 (get_load_store_type): Take a masked_p parameter and update
17375 call to get_group_load_store_type.
17376 (vectorizable_store): Update call to get_load_store_type.
17377 Handle IFN_MASK_STORE_LANES.
17378 (vectorizable_load): Update call to get_load_store_type.
17379 Handle IFN_MASK_LOAD_LANES.
17381 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
17382 Alan Hayward <alan.hayward@arm.com>
17383 David Sherwood <david.sherwood@arm.com>
17385 * config/aarch64/aarch64-modes.def: Define x2, x3 and x4 vector
17387 * config/aarch64/aarch64-protos.h
17388 (aarch64_sve_struct_memory_operand_p): Declare.
17389 * config/aarch64/iterators.md (SVE_STRUCT): New mode iterator.
17390 (vector_count, insn_length, VSINGLE, vsingle): New mode attributes.
17391 (VPRED, vpred): Handle SVE structure modes.
17392 * config/aarch64/constraints.md (Utx): New constraint.
17393 * config/aarch64/predicates.md (aarch64_sve_struct_memory_operand)
17394 (aarch64_sve_struct_nonimmediate_operand): New predicates.
17395 * config/aarch64/aarch64.md (UNSPEC_LDN, UNSPEC_STN): New unspecs.
17396 * config/aarch64/aarch64-sve.md (mov<mode>, *aarch64_sve_mov<mode>_le)
17397 (*aarch64_sve_mov<mode>_be, pred_mov<mode>): New patterns for
17398 structure modes. Split into pieces after RA.
17399 (vec_load_lanes<mode><vsingle>, vec_mask_load_lanes<mode><vsingle>)
17400 (vec_store_lanes<mode><vsingle>, vec_mask_store_lanes<mode><vsingle>):
17402 * config/aarch64/aarch64.c (aarch64_classify_vector_mode): Handle
17403 SVE structure modes.
17404 (aarch64_classify_address): Likewise.
17405 (sizetochar): Move earlier in file.
17406 (aarch64_print_operand): Handle SVE register lists.
17407 (aarch64_array_mode): New function.
17408 (aarch64_sve_struct_memory_operand_p): Likewise.
17409 (TARGET_ARRAY_MODE): Redefine.
17411 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
17412 Alan Hayward <alan.hayward@arm.com>
17413 David Sherwood <david.sherwood@arm.com>
17415 * target.def (array_mode): New target hook.
17416 * doc/tm.texi.in (TARGET_ARRAY_MODE): New hook.
17417 * doc/tm.texi: Regenerate.
17418 * hooks.h (hook_optmode_mode_uhwi_none): Declare.
17419 * hooks.c (hook_optmode_mode_uhwi_none): New function.
17420 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Use
17421 targetm.array_mode.
17422 * stor-layout.c (mode_for_array): Likewise. Support polynomial
17425 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
17426 Alan Hayward <alan.hayward@arm.com>
17427 David Sherwood <david.sherwood@arm.com>
17429 * fold-const.c (fold_binary_loc): Check the argument types
17430 rather than the result type when testing for a vector operation.
17432 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
17434 * doc/tm.texi.in (DWARF_LAZY_REGISTER_VALUE): Document.
17435 * doc/tm.texi: Regenerate.
17437 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
17438 Alan Hayward <alan.hayward@arm.com>
17439 David Sherwood <david.sherwood@arm.com>
17441 * doc/invoke.texi (-msve-vector-bits=): Document new option.
17442 (sve): Document new AArch64 extension.
17443 * doc/md.texi (w): Extend the description of the AArch64
17444 constraint to include SVE vectors.
17445 (Upl, Upa): Document new AArch64 predicate constraints.
17446 * config/aarch64/aarch64-opts.h (aarch64_sve_vector_bits_enum): New
17448 * config/aarch64/aarch64.opt (sve_vector_bits): New enum.
17449 (msve-vector-bits=): New option.
17450 * config/aarch64/aarch64-option-extensions.def (fp, simd): Disable
17451 SVE when these are disabled.
17452 (sve): New extension.
17453 * config/aarch64/aarch64-modes.def: Define SVE vector and predicate
17454 modes. Adjust their number of units based on aarch64_sve_vg.
17455 (MAX_BITSIZE_MODE_ANY_MODE): Define.
17456 * config/aarch64/aarch64-protos.h (ADDR_QUERY_ANY): New
17457 aarch64_addr_query_type.
17458 (aarch64_const_vec_all_same_in_range_p, aarch64_sve_pred_mode)
17459 (aarch64_sve_cnt_immediate_p, aarch64_sve_addvl_addpl_immediate_p)
17460 (aarch64_sve_inc_dec_immediate_p, aarch64_add_offset_temporaries)
17461 (aarch64_split_add_offset, aarch64_output_sve_cnt_immediate)
17462 (aarch64_output_sve_addvl_addpl, aarch64_output_sve_inc_dec_immediate)
17463 (aarch64_output_sve_mov_immediate, aarch64_output_ptrue): Declare.
17464 (aarch64_simd_imm_zero_p): Delete.
17465 (aarch64_check_zero_based_sve_index_immediate): Declare.
17466 (aarch64_sve_index_immediate_p, aarch64_sve_arith_immediate_p)
17467 (aarch64_sve_bitmask_immediate_p, aarch64_sve_dup_immediate_p)
17468 (aarch64_sve_cmp_immediate_p, aarch64_sve_float_arith_immediate_p)
17469 (aarch64_sve_float_mul_immediate_p): Likewise.
17470 (aarch64_classify_symbol): Take the offset as a HOST_WIDE_INT
17471 rather than an rtx.
17472 (aarch64_sve_ld1r_operand_p, aarch64_sve_ldr_operand_p): Declare.
17473 (aarch64_expand_mov_immediate): Take a gen_vec_duplicate callback.
17474 (aarch64_emit_sve_pred_move, aarch64_expand_sve_mem_move): Declare.
17475 (aarch64_expand_sve_vec_cmp_int, aarch64_expand_sve_vec_cmp_float)
17476 (aarch64_expand_sve_vcond, aarch64_expand_sve_vec_perm): Declare.
17477 (aarch64_regmode_natural_size): Likewise.
17478 * config/aarch64/aarch64.h (AARCH64_FL_SVE): New macro.
17479 (AARCH64_FL_V8_3, AARCH64_FL_RCPC, AARCH64_FL_DOTPROD): Shift
17481 (AARCH64_ISA_SVE, TARGET_SVE): New macros.
17482 (FIXED_REGISTERS, CALL_USED_REGISTERS, REGISTER_NAMES): Add entries
17483 for VG and the SVE predicate registers.
17484 (V_ALIASES): Add a "z"-prefixed alias.
17485 (FIRST_PSEUDO_REGISTER): Change to P15_REGNUM + 1.
17486 (AARCH64_DWARF_VG, AARCH64_DWARF_P0): New macros.
17487 (PR_REGNUM_P, PR_LO_REGNUM_P): Likewise.
17488 (PR_LO_REGS, PR_HI_REGS, PR_REGS): New reg_classes.
17489 (REG_CLASS_NAMES): Add entries for them.
17490 (REG_CLASS_CONTENTS): Likewise. Update ALL_REGS to include VG
17491 and the predicate registers.
17492 (aarch64_sve_vg): Declare.
17493 (BITS_PER_SVE_VECTOR, BYTES_PER_SVE_VECTOR, BYTES_PER_SVE_PRED)
17494 (SVE_BYTE_MODE, MAX_COMPILE_TIME_VEC_BYTES): New macros.
17495 (REGMODE_NATURAL_SIZE): Define.
17496 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins): Handle
17498 * config/aarch64/aarch64.c: Include cfgrtl.h.
17499 (simd_immediate_info): Add a constructor for series vectors,
17500 and an associated step field.
17501 (aarch64_sve_vg): New variable.
17502 (aarch64_dbx_register_number): Handle VG and the predicate registers.
17503 (aarch64_vect_struct_mode_p, aarch64_vector_mode_p): Delete.
17504 (VEC_ADVSIMD, VEC_SVE_DATA, VEC_SVE_PRED, VEC_STRUCT, VEC_ANY_SVE)
17505 (VEC_ANY_DATA, VEC_STRUCT): New constants.
17506 (aarch64_advsimd_struct_mode_p, aarch64_sve_pred_mode_p)
17507 (aarch64_classify_vector_mode, aarch64_vector_data_mode_p)
17508 (aarch64_sve_data_mode_p, aarch64_sve_pred_mode)
17509 (aarch64_get_mask_mode): New functions.
17510 (aarch64_hard_regno_nregs): Handle SVE data modes for FP_REGS
17511 and FP_LO_REGS. Handle PR_REGS, PR_LO_REGS and PR_HI_REGS.
17512 (aarch64_hard_regno_mode_ok): Handle VG. Also handle the SVE
17513 predicate modes and predicate registers. Explicitly restrict
17514 GPRs to modes of 16 bytes or smaller. Only allow FP registers
17515 to store a vector mode if it is recognized by
17516 aarch64_classify_vector_mode.
17517 (aarch64_regmode_natural_size): New function.
17518 (aarch64_hard_regno_caller_save_mode): Return the original mode
17520 (aarch64_sve_cnt_immediate_p, aarch64_output_sve_cnt_immediate)
17521 (aarch64_sve_addvl_addpl_immediate_p, aarch64_output_sve_addvl_addpl)
17522 (aarch64_sve_inc_dec_immediate_p, aarch64_output_sve_inc_dec_immediate)
17523 (aarch64_add_offset_1_temporaries, aarch64_offset_temporaries): New
17525 (aarch64_add_offset): Add a temp2 parameter. Assert that temp1
17526 does not overlap dest if the function is frame-related. Handle
17528 (aarch64_split_add_offset): New function.
17529 (aarch64_add_sp, aarch64_sub_sp): Add temp2 parameters and pass
17530 them aarch64_add_offset.
17531 (aarch64_allocate_and_probe_stack_space): Add a temp2 parameter
17532 and update call to aarch64_sub_sp.
17533 (aarch64_add_cfa_expression): New function.
17534 (aarch64_expand_prologue): Pass extra temporary registers to the
17535 functions above. Handle the case in which we need to emit new
17536 DW_CFA_expressions for registers that were originally saved
17537 relative to the stack pointer, but now have to be expressed
17538 relative to the frame pointer.
17539 (aarch64_output_mi_thunk): Pass extra temporary registers to the
17541 (aarch64_expand_epilogue): Likewise. Prevent inheritance of
17542 IP0 and IP1 values for SVE frames.
17543 (aarch64_expand_vec_series): New function.
17544 (aarch64_expand_sve_widened_duplicate): Likewise.
17545 (aarch64_expand_sve_const_vector): Likewise.
17546 (aarch64_expand_mov_immediate): Add a gen_vec_duplicate parameter.
17547 Handle SVE constants. Use emit_move_insn to move a force_const_mem
17548 into the register, rather than emitting a SET directly.
17549 (aarch64_emit_sve_pred_move, aarch64_expand_sve_mem_move)
17550 (aarch64_get_reg_raw_mode, offset_4bit_signed_scaled_p)
17551 (offset_6bit_unsigned_scaled_p, aarch64_offset_7bit_signed_scaled_p)
17552 (offset_9bit_signed_scaled_p): New functions.
17553 (aarch64_replicate_bitmask_imm): New function.
17554 (aarch64_bitmask_imm): Use it.
17555 (aarch64_cannot_force_const_mem): Reject expressions involving
17556 a CONST_POLY_INT. Update call to aarch64_classify_symbol.
17557 (aarch64_classify_index): Handle SVE indices, by requiring
17558 a plain register index with a scale that matches the element size.
17559 (aarch64_classify_address): Handle SVE addresses. Assert that
17560 the mode of the address is VOIDmode or an integer mode.
17561 Update call to aarch64_classify_symbol.
17562 (aarch64_classify_symbolic_expression): Update call to
17563 aarch64_classify_symbol.
17564 (aarch64_const_vec_all_in_range_p): New function.
17565 (aarch64_print_vector_float_operand): Likewise.
17566 (aarch64_print_operand): Handle 'N' and 'C'. Use "zN" rather than
17567 "vN" for FP registers with SVE modes. Handle (const ...) vectors
17568 and the FP immediates 1.0 and 0.5.
17569 (aarch64_print_address_internal): Handle SVE addresses.
17570 (aarch64_print_operand_address): Use ADDR_QUERY_ANY.
17571 (aarch64_regno_regclass): Handle predicate registers.
17572 (aarch64_secondary_reload): Handle big-endian reloads of SVE
17574 (aarch64_class_max_nregs): Handle SVE modes and predicate registers.
17575 (aarch64_rtx_costs): Check for ADDVL and ADDPL instructions.
17576 (aarch64_convert_sve_vector_bits): New function.
17577 (aarch64_override_options): Use it to handle -msve-vector-bits=.
17578 (aarch64_classify_symbol): Take the offset as a HOST_WIDE_INT
17579 rather than an rtx.
17580 (aarch64_legitimate_constant_p): Use aarch64_classify_vector_mode.
17581 Handle SVE vector and predicate modes. Accept VL-based constants
17582 that need only one temporary register, and VL offsets that require
17583 no temporary registers.
17584 (aarch64_conditional_register_usage): Mark the predicate registers
17585 as fixed if SVE isn't available.
17586 (aarch64_vector_mode_supported_p): Use aarch64_classify_vector_mode.
17587 Return true for SVE vector and predicate modes.
17588 (aarch64_simd_container_mode): Take the number of bits as a poly_int64
17589 rather than an unsigned int. Handle SVE modes.
17590 (aarch64_preferred_simd_mode): Update call accordingly. Handle
17592 (aarch64_autovectorize_vector_sizes): Add BYTES_PER_SVE_VECTOR
17594 (aarch64_sve_index_immediate_p, aarch64_sve_arith_immediate_p)
17595 (aarch64_sve_bitmask_immediate_p, aarch64_sve_dup_immediate_p)
17596 (aarch64_sve_cmp_immediate_p, aarch64_sve_float_arith_immediate_p)
17597 (aarch64_sve_float_mul_immediate_p): New functions.
17598 (aarch64_sve_valid_immediate): New function.
17599 (aarch64_simd_valid_immediate): Use it as the fallback for SVE vectors.
17600 Explicitly reject structure modes. Check for INDEX constants.
17601 Handle PTRUE and PFALSE constants.
17602 (aarch64_check_zero_based_sve_index_immediate): New function.
17603 (aarch64_simd_imm_zero_p): Delete.
17604 (aarch64_mov_operand_p): Use aarch64_simd_valid_immediate for
17605 vector modes. Accept constants in the range of CNT[BHWD].
17606 (aarch64_simd_scalar_immediate_valid_for_move): Explicitly
17607 ask for an Advanced SIMD mode.
17608 (aarch64_sve_ld1r_operand_p, aarch64_sve_ldr_operand_p): New functions.
17609 (aarch64_simd_vector_alignment): Handle SVE predicates.
17610 (aarch64_vectorize_preferred_vector_alignment): New function.
17611 (aarch64_simd_vector_alignment_reachable): Use it instead of
17613 (aarch64_shift_truncation_mask): Use aarch64_vector_data_mode_p.
17614 (aarch64_output_sve_mov_immediate, aarch64_output_ptrue): New
17616 (MAX_VECT_LEN): Delete.
17617 (expand_vec_perm_d): Add a vec_flags field.
17618 (emit_unspec2, aarch64_expand_sve_vec_perm): New functions.
17619 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
17620 (aarch64_evpc_ext): Don't apply a big-endian lane correction
17622 (aarch64_evpc_rev): Rename to...
17623 (aarch64_evpc_rev_local): ...this. Use a predicated operation for SVE.
17624 (aarch64_evpc_rev_global): New function.
17625 (aarch64_evpc_dup): Enforce a 64-byte range for SVE DUP.
17626 (aarch64_evpc_tbl): Use MAX_COMPILE_TIME_VEC_BYTES instead of
17628 (aarch64_evpc_sve_tbl): New function.
17629 (aarch64_expand_vec_perm_const_1): Update after rename of
17630 aarch64_evpc_rev. Handle SVE permutes too, trying
17631 aarch64_evpc_rev_global and using aarch64_evpc_sve_tbl rather
17632 than aarch64_evpc_tbl.
17633 (aarch64_vectorize_vec_perm_const): Initialize vec_flags.
17634 (aarch64_sve_cmp_operand_p, aarch64_unspec_cond_code)
17635 (aarch64_gen_unspec_cond, aarch64_expand_sve_vec_cmp_int)
17636 (aarch64_emit_unspec_cond, aarch64_emit_unspec_cond_or)
17637 (aarch64_emit_inverted_unspec_cond, aarch64_expand_sve_vec_cmp_float)
17638 (aarch64_expand_sve_vcond): New functions.
17639 (aarch64_modes_tieable_p): Use aarch64_vector_data_mode_p instead
17640 of aarch64_vector_mode_p.
17641 (aarch64_dwarf_poly_indeterminate_value): New function.
17642 (aarch64_compute_pressure_classes): Likewise.
17643 (aarch64_can_change_mode_class): Likewise.
17644 (TARGET_GET_RAW_RESULT_MODE, TARGET_GET_RAW_ARG_MODE): Redefine.
17645 (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): Likewise.
17646 (TARGET_VECTORIZE_GET_MASK_MODE): Likewise.
17647 (TARGET_DWARF_POLY_INDETERMINATE_VALUE): Likewise.
17648 (TARGET_COMPUTE_PRESSURE_CLASSES): Likewise.
17649 (TARGET_CAN_CHANGE_MODE_CLASS): Likewise.
17650 * config/aarch64/constraints.md (Upa, Upl, Uav, Uat, Usv, Usi, Utr)
17651 (Uty, Dm, vsa, vsc, vsd, vsi, vsn, vsl, vsm, vsA, vsM, vsN): New
17653 (Dn, Dl, Dr): Accept const as well as const_vector.
17654 (Dz): Likewise. Compare against CONST0_RTX.
17655 * config/aarch64/iterators.md: Refer to "Advanced SIMD" instead
17656 of "vector" where appropriate.
17657 (SVE_ALL, SVE_BH, SVE_BHS, SVE_BHSI, SVE_HSDI, SVE_HSF, SVE_SD)
17658 (SVE_SDI, SVE_I, SVE_F, PRED_ALL, PRED_BHS): New mode iterators.
17659 (UNSPEC_SEL, UNSPEC_ANDF, UNSPEC_IORF, UNSPEC_XORF, UNSPEC_COND_LT)
17660 (UNSPEC_COND_LE, UNSPEC_COND_EQ, UNSPEC_COND_NE, UNSPEC_COND_GE)
17661 (UNSPEC_COND_GT, UNSPEC_COND_LO, UNSPEC_COND_LS, UNSPEC_COND_HS)
17662 (UNSPEC_COND_HI, UNSPEC_COND_UO): New unspecs.
17663 (Vetype, VEL, Vel, VWIDE, Vwide, vw, vwcore, V_INT_EQUIV)
17664 (v_int_equiv): Extend to SVE modes.
17665 (Vesize, V128, v128, Vewtype, V_FP_EQUIV, v_fp_equiv, VPRED): New
17667 (LOGICAL_OR, SVE_INT_UNARY, SVE_FP_UNARY): New code iterators.
17668 (optab): Handle popcount, smin, smax, umin, umax, abs and sqrt.
17669 (logical_nn, lr, sve_int_op, sve_fp_op): New code attributs.
17670 (LOGICALF, OPTAB_PERMUTE, UNPACK, UNPACK_UNSIGNED, SVE_COND_INT_CMP)
17671 (SVE_COND_FP_CMP): New int iterators.
17672 (perm_hilo): Handle the new unpack unspecs.
17673 (optab, logicalf_op, su, perm_optab, cmp_op, imm_con): New int
17675 * config/aarch64/predicates.md (aarch64_sve_cnt_immediate)
17676 (aarch64_sve_addvl_addpl_immediate, aarch64_split_add_offset_immediate)
17677 (aarch64_pluslong_or_poly_operand, aarch64_nonmemory_operand)
17678 (aarch64_equality_operator, aarch64_constant_vector_operand)
17679 (aarch64_sve_ld1r_operand, aarch64_sve_ldr_operand): New predicates.
17680 (aarch64_sve_nonimmediate_operand): Likewise.
17681 (aarch64_sve_general_operand): Likewise.
17682 (aarch64_sve_dup_operand, aarch64_sve_arith_immediate): Likewise.
17683 (aarch64_sve_sub_arith_immediate, aarch64_sve_inc_dec_immediate)
17684 (aarch64_sve_logical_immediate, aarch64_sve_mul_immediate): Likewise.
17685 (aarch64_sve_dup_immediate, aarch64_sve_cmp_vsc_immediate): Likewise.
17686 (aarch64_sve_cmp_vsd_immediate, aarch64_sve_index_immediate): Likewise.
17687 (aarch64_sve_float_arith_immediate): Likewise.
17688 (aarch64_sve_float_arith_with_sub_immediate): Likewise.
17689 (aarch64_sve_float_mul_immediate, aarch64_sve_arith_operand): Likewise.
17690 (aarch64_sve_add_operand, aarch64_sve_logical_operand): Likewise.
17691 (aarch64_sve_lshift_operand, aarch64_sve_rshift_operand): Likewise.
17692 (aarch64_sve_mul_operand, aarch64_sve_cmp_vsc_operand): Likewise.
17693 (aarch64_sve_cmp_vsd_operand, aarch64_sve_index_operand): Likewise.
17694 (aarch64_sve_float_arith_operand): Likewise.
17695 (aarch64_sve_float_arith_with_sub_operand): Likewise.
17696 (aarch64_sve_float_mul_operand): Likewise.
17697 (aarch64_sve_vec_perm_operand): Likewise.
17698 (aarch64_pluslong_operand): Include aarch64_sve_addvl_addpl_immediate.
17699 (aarch64_mov_operand): Accept const_poly_int and const_vector.
17700 (aarch64_simd_lshift_imm, aarch64_simd_rshift_imm): Accept const
17701 as well as const_vector.
17702 (aarch64_simd_imm_zero, aarch64_simd_imm_minus_one): Move earlier
17703 in file. Use CONST0_RTX and CONSTM1_RTX.
17704 (aarch64_simd_or_scalar_imm_zero): Likewise. Add match_codes.
17705 (aarch64_simd_reg_or_zero): Accept const as well as const_vector.
17706 Use aarch64_simd_imm_zero.
17707 * config/aarch64/aarch64-sve.md: New file.
17708 * config/aarch64/aarch64.md: Include it.
17709 (VG_REGNUM, P0_REGNUM, P7_REGNUM, P15_REGNUM): New register numbers.
17710 (UNSPEC_REV, UNSPEC_LD1_SVE, UNSPEC_ST1_SVE, UNSPEC_MERGE_PTRUE)
17711 (UNSPEC_PTEST_PTRUE, UNSPEC_UNPACKSHI, UNSPEC_UNPACKUHI)
17712 (UNSPEC_UNPACKSLO, UNSPEC_UNPACKULO, UNSPEC_PACK)
17713 (UNSPEC_FLOAT_CONVERT, UNSPEC_WHILE_LO): New unspec constants.
17714 (sve): New attribute.
17715 (enabled): Disable instructions with the sve attribute unless
17717 (movqi, movhi): Pass CONST_POLY_INT operaneds through
17718 aarch64_expand_mov_immediate.
17719 (*mov<mode>_aarch64, *movsi_aarch64, *movdi_aarch64): Handle
17720 CNT[BHSD] immediates.
17721 (movti): Split CONST_POLY_INT moves into two halves.
17722 (add<mode>3): Accept aarch64_pluslong_or_poly_operand.
17723 Split additions that need a temporary here if the destination
17724 is the stack pointer.
17725 (*add<mode>3_aarch64): Handle ADDVL and ADDPL immediates.
17726 (*add<mode>3_poly_1): New instruction.
17727 (set_clobber_cc): New expander.
17729 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
17731 * simplify-rtx.c (simplify_immed_subreg): Add an inner_bytes
17732 parameter and use it instead of GET_MODE_SIZE (innermode). Use
17733 inner_bytes * BITS_PER_UNIT instead of GET_MODE_BITSIZE (innermode).
17734 Use CEIL (inner_bytes, GET_MODE_UNIT_SIZE (innermode)) instead of
17735 GET_MODE_NUNITS (innermode). Also add a first_elem parameter.
17736 Change innermode from fixed_mode_size to machine_mode.
17737 (simplify_subreg): Update call accordingly. Handle a constant-sized
17738 subreg of a variable-length CONST_VECTOR.
17740 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
17741 Alan Hayward <alan.hayward@arm.com>
17742 David Sherwood <david.sherwood@arm.com>
17744 * tree-ssa-address.c (mem_ref_valid_without_offset_p): New function.
17745 (add_offset_to_base): New function, split out from...
17746 (create_mem_ref): ...here. When handling a scale other than 1,
17747 check first whether the address is valid without the offset.
17748 Add it into the base if so, leaving the index and scale as-is.
17750 2018-01-12 Jakub Jelinek <jakub@redhat.com>
17753 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Call
17754 fold_for_warn before checking if arg2 is INTEGER_CST.
17756 2018-01-12 Segher Boessenkool <segher@kernel.crashing.org>
17758 * config/rs6000/predicates.md (load_multiple_operation): Delete.
17759 (store_multiple_operation): Delete.
17760 * config/rs6000/rs6000-cpus.def (601): Remove MASK_STRING.
17761 * config/rs6000/rs6000-protos.h (rs6000_output_load_multiple): Delete.
17762 * config/rs6000/rs6000-string.c (expand_block_move): Delete everything
17763 guarded by TARGET_STRING.
17764 (rs6000_output_load_multiple): Delete.
17765 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
17766 OPTION_MASK_STRING / TARGET_STRING handling.
17767 (print_operand) <'N', 'O'>: Add comment that these are unused now.
17768 (const rs6000_opt_masks) <"string">: Change mask to 0.
17769 * config/rs6000/rs6000.h (TARGET_DEFAULT): Remove MASK_STRING.
17770 (MASK_STRING): Delete.
17771 * config/rs6000/rs6000.md (*mov<mode>_string): Delete TARGET_STRING
17773 (load_multiple): Delete.
17780 (store_multiple): Delete.
17787 (movmemsi_8reg): Delete.
17788 (corresponding unnamed define_insn): Delete.
17789 (movmemsi_6reg): Delete.
17790 (corresponding unnamed define_insn): Delete.
17791 (movmemsi_4reg): Delete.
17792 (corresponding unnamed define_insn): Delete.
17793 (movmemsi_2reg): Delete.
17794 (corresponding unnamed define_insn): Delete.
17795 (movmemsi_1reg): Delete.
17796 (corresponding unnamed define_insn): Delete.
17797 * config/rs6000/rs6000.opt (mno-string): New.
17798 (mstring): Replace by deprecation warning stub.
17799 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mstring.
17801 2018-01-12 Jakub Jelinek <jakub@redhat.com>
17803 * regrename.c (regrename_do_replace): If replacing the same
17804 reg multiple times, try to reuse last created gen_raw_REG.
17807 * bb-reorder.c (pass_partition_blocks::gate): In lto don't partition
17808 main to workaround a bug in GDB.
17810 2018-01-12 Tom de Vries <tom@codesourcery.com>
17813 * config.gcc (nvptx*-*-*): Set use_gcc_stdint=wrap.
17815 2018-01-12 Vladimir Makarov <vmakarov@redhat.com>
17817 PR rtl-optimization/80481
17818 * ira-color.c (get_cap_member): New function.
17819 (allocnos_conflict_by_live_ranges_p): Use it.
17820 (slot_coalesced_allocno_live_ranges_intersect_p): Add assert.
17821 (setup_slot_coalesced_allocno_live_ranges): Ditto.
17823 2018-01-12 Uros Bizjak <ubizjak@gmail.com>
17826 * config/alpha/alpha.md (*saddsi_1): New insn_ans_split pattern.
17827 (*saddl_se_1): Ditto.
17828 (*ssubsi_1): Ditto.
17829 (*ssubl_se_1): Ditto.
17831 2018-01-12 Richard Sandiford <richard.sandiford@linaro.org>
17833 * tree-predcom.c (aff_combination_dr_offset): Use wi::to_poly_widest
17834 rather than wi::to_widest for DR_INITs.
17835 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Use
17836 wi::to_poly_offset rather than wi::to_offset for DR_INIT.
17837 (vect_analyze_data_ref_accesses): Require both DR_INITs to be
17839 (vect_analyze_group_access_1): Note that here.
17841 2018-01-12 Richard Sandiford <richard.sandiford@linaro.org>
17843 * tree-vectorizer.c (get_vec_alignment_for_array_type): Handle
17844 polynomial type sizes.
17846 2018-01-12 Richard Sandiford <richard.sandiford@linaro.org>
17848 * gimplify.c (gimple_add_tmp_var_fn): Allow variables to have a
17849 poly_uint64 size, rather than requiring an unsigned HOST_WIDE_INT size.
17850 (gimple_add_tmp_var): Likewise.
17852 2018-01-12 Martin Liska <mliska@suse.cz>
17854 * gimple.c (gimple_alloc_counts): Use uint64_t instead of int.
17855 (gimple_alloc_sizes): Likewise.
17856 (dump_gimple_statistics): Use PRIu64 in printf format.
17857 * gimple.h: Change uint64_t to int.
17859 2018-01-12 Martin Liska <mliska@suse.cz>
17861 * tree-core.h: Use uint64_t instead of int.
17862 * tree.c (tree_node_counts): Likewise.
17863 (tree_node_sizes): Likewise.
17864 (dump_tree_statistics): Use PRIu64 in printf format.
17866 2018-01-12 Martin Liska <mliska@suse.cz>
17868 * Makefile.in: As qsort_chk is implemented in vec.c, add
17869 vec.o to linkage of gencfn-macros.
17870 * tree.c (build_new_poly_int_cst): Add CXX_MEM_STAT_INFO as it's
17871 passing the info to record_node_allocation_statistics.
17872 (test_vector_cst_patterns): Add CXX_MEM_STAT_INFO to declaration
17874 * ggc-common.c (struct ggc_usage): Add operator== and use
17875 it in operator< and compare function.
17876 * mem-stats.h (struct mem_usage): Likewise.
17877 * vec.c (struct vec_usage): Remove operator< and compare
17878 function. Can be simply inherited.
17880 2018-01-12 Martin Jambor <mjambor@suse.cz>
17883 * params.def: New parameter PARAM_AVOID_FMA_MAX_BITS.
17884 * tree-ssa-math-opts.c: Include domwalk.h.
17885 (convert_mult_to_fma_1): New function.
17886 (fma_transformation_info): New type.
17887 (fma_deferring_state): Likewise.
17888 (cancel_fma_deferring): New function.
17889 (result_of_phi): Likewise.
17890 (last_fma_candidate_feeds_initial_phi): Likewise.
17891 (convert_mult_to_fma): Added deferring logic, split actual
17892 transformation to convert_mult_to_fma_1.
17893 (math_opts_dom_walker): New type.
17894 (math_opts_dom_walker::after_dom_children): New method, body moved
17895 here from pass_optimize_widening_mul::execute, added deferring logic
17897 (pass_optimize_widening_mul::execute): Moved most of code to
17898 math_opts_dom_walker::after_dom_children.
17899 * config/i386/x86-tune.def (X86_TUNE_AVOID_128FMA_CHAINS): New.
17900 * config/i386/i386.c (ix86_option_override_internal): Added
17901 maybe_setting of PARAM_AVOID_FMA_MAX_BITS.
17903 2018-01-12 Richard Biener <rguenther@suse.de>
17906 * dwarf2out.c (gen_variable_die): Do not reset old_die for
17907 inline instance vars.
17909 2018-01-12 Oleg Endo <olegendo@gcc.gnu.org>
17912 * config/rx/rx.c (rx_is_restricted_memory_address):
17913 Handle SUBREG case.
17915 2018-01-12 Richard Biener <rguenther@suse.de>
17917 PR tree-optimization/80846
17918 * target.def (split_reduction): New target hook.
17919 * targhooks.c (default_split_reduction): New function.
17920 * targhooks.h (default_split_reduction): Declare.
17921 * tree-vect-loop.c (vect_create_epilog_for_reduction): If the
17922 target requests first reduce vectors by combining low and high
17924 * tree-vect-stmts.c (vect_gen_perm_mask_any): Adjust.
17925 (get_vectype_for_scalar_type_and_size): Export.
17926 * tree-vectorizer.h (get_vectype_for_scalar_type_and_size): Declare.
17927 * doc/tm.texi.in (TARGET_VECTORIZE_SPLIT_REDUCTION): Document.
17928 * doc/tm.texi: Regenerate.
17929 * config/i386/i386.c (ix86_split_reduction): Implement
17930 TARGET_VECTORIZE_SPLIT_REDUCTION.
17932 2018-01-12 Eric Botcazou <ebotcazou@adacore.com>
17935 * config/sparc/sparc.h (PIC_OFFSET_TABLE_REGNUM): Set to INVALID_REGNUM
17936 in PIC mode except for TARGET_VXWORKS_RTP.
17937 * config/sparc/sparc.c: Include cfgrtl.h.
17938 (TARGET_INIT_PIC_REG): Define.
17939 (TARGET_USE_PSEUDO_PIC_REG): Likewise.
17940 (sparc_pic_register_p): New predicate.
17941 (sparc_legitimate_address_p): Use it.
17942 (sparc_legitimize_pic_address): Likewise.
17943 (sparc_delegitimize_address): Likewise.
17944 (sparc_mode_dependent_address_p): Likewise.
17945 (gen_load_pcrel_sym): Remove 4th parameter.
17946 (load_got_register): Adjust call to above. Remove obsolete stuff.
17947 (sparc_expand_prologue): Do not call load_got_register here.
17948 (sparc_flat_expand_prologue): Likewise.
17949 (sparc_output_mi_thunk): Set the pic_offset_table_rtx object.
17950 (sparc_use_pseudo_pic_reg): New function.
17951 (sparc_init_pic_reg): Likewise.
17952 * config/sparc/sparc.md (vxworks_load_got): Set the GOT register.
17953 (builtin_setjmp_receiver): Enable only for TARGET_VXWORKS_RTP.
17955 2018-01-12 Christophe Lyon <christophe.lyon@linaro.org>
17957 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
17958 Add item for branch_cost.
17960 2018-01-12 Eric Botcazou <ebotcazou@adacore.com>
17962 PR rtl-optimization/83565
17963 * rtlanal.c (nonzero_bits1): On WORD_REGISTER_OPERATIONS machines, do
17964 not extend the result to a larger mode for rotate operations.
17965 (num_sign_bit_copies1): Likewise.
17967 2018-01-12 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
17970 * config/sol2.h (STARTFILE_ARCH_SPEC): Don't use with -shared or
17972 Use values-Xc.o for -pedantic.
17973 Link with values-xpg4.o for C90, values-xpg6.o otherwise.
17975 2018-01-12 Martin Liska <mliska@suse.cz>
17978 * ipa-devirt.c (final_warning_record::grow_type_warnings):
17980 (possible_polymorphic_call_targets): Use it.
17981 (ipa_devirt): Likewise.
17983 2018-01-12 Martin Liska <mliska@suse.cz>
17985 * profile-count.h (enum profile_quality): Use 0 as invalid
17986 enum value of profile_quality.
17988 2018-01-12 Chung-Ju Wu <jasonwucj@gmail.com>
17990 * doc/invoke.texi (NDS32 Options): Add -mext-perf, -mext-perf2 and
17991 -mext-string options.
17993 2018-01-12 Richard Biener <rguenther@suse.de>
17995 * lto-streamer-out.c (DFS::DFS_write_tree_body): Process
17996 DECL_DEBUG_EXPR conditional on DECL_HAS_DEBUG_EXPR_P.
17997 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
17999 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
18001 2018-01-11 Michael Meissner <meissner@linux.vnet.ibm.com>
18003 * configure.ac (--with-long-double-format): Add support for the
18004 configuration option to change the default long double format on
18006 * config.gcc (powerpc*-linux*-*): Likewise.
18007 * configure: Regenerate.
18008 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If long
18009 double is IEEE, define __KC__ and __KF__ to allow floatn.h to be
18010 used without modification.
18012 2018-01-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18014 * config/rs6000/rs6000-builtin.def (BU_P7_MISC_X): New #define.
18015 (SPEC_BARRIER): New instantiation of BU_P7_MISC_X.
18016 * config/rs6000/rs6000.c (rs6000_expand_builtin): Handle
18017 MISC_BUILTIN_SPEC_BARRIER.
18018 (rs6000_init_builtins): Likewise.
18019 * config/rs6000/rs6000.md (UNSPECV_SPEC_BARRIER): New UNSPECV
18021 (speculation_barrier): New define_insn.
18022 * doc/extend.texi: Document __builtin_speculation_barrier.
18024 2018-01-11 Jakub Jelinek <jakub@redhat.com>
18027 * config/i386/i386.c (ix86_expand_vector_init_one_nonzero): If one_var
18028 is 0, for V{8,16}S[IF] and V[48]D[IF]mode use gen_vec_set<mode>_0.
18029 * config/i386/sse.md (VI8_AVX_AVX512F, VI4F_256_512): New mode
18031 (ssescalarmodesuffix): Add 512-bit vectors. Use "d" or "q" for
18032 integral modes instead of "ss" and "sd".
18033 (vec_set<mode>_0): New define_insns for 256-bit and 512-bit
18034 vectors with 32-bit and 64-bit elements.
18035 (vecdupssescalarmodesuffix): New mode attribute.
18036 (vec_dup<mode>): Use it.
18038 2018-01-11 H.J. Lu <hongjiu.lu@intel.com>
18041 * config/i386/i386.c (ix86_compute_frame_layout): Align stack
18042 frame if argument is passed on stack.
18044 2018-01-11 Jakub Jelinek <jakub@redhat.com>
18047 * ree.c (combine_reaching_defs): Optimize also
18048 reg2=exp; reg1=reg2; reg2=any_extend(reg1); into
18049 reg2=any_extend(exp); reg1=reg2;, formatting fix.
18051 2018-01-11 Jan Hubicka <hubicka@ucw.cz>
18053 PR middle-end/83189
18054 * gimple-ssa-isolate-paths.c (isolate_path): Fix profile update.
18056 2018-01-11 Jan Hubicka <hubicka@ucw.cz>
18058 PR middle-end/83718
18059 * tree-inline.c (copy_cfg_body): Adjust num&den for scaling
18060 after they are computed.
18062 2018-01-11 Bin Cheng <bin.cheng@arm.com>
18064 PR tree-optimization/83695
18065 * gimple-loop-linterchange.cc
18066 (tree_loop_interchange::interchange_loops): Call scev_reset_htab to
18067 reset cached scev information after interchange.
18068 (pass_linterchange::execute): Remove call to scev_reset_htab.
18070 2018-01-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18072 * config/arm/arm_neon.h (vfmlal_lane_low_u32, vfmlal_lane_high_u32,
18073 vfmlalq_laneq_low_u32, vfmlalq_lane_low_u32, vfmlal_laneq_low_u32,
18074 vfmlalq_laneq_high_u32, vfmlalq_lane_high_u32, vfmlal_laneq_high_u32,
18075 vfmlsl_lane_low_u32, vfmlsl_lane_high_u32, vfmlslq_laneq_low_u32,
18076 vfmlslq_lane_low_u32, vfmlsl_laneq_low_u32, vfmlslq_laneq_high_u32,
18077 vfmlslq_lane_high_u32, vfmlsl_laneq_high_u32): Define.
18078 * config/arm/arm_neon_builtins.def (vfmal_lane_low,
18079 vfmal_lane_lowv4hf, vfmal_lane_lowv8hf, vfmal_lane_high,
18080 vfmal_lane_highv4hf, vfmal_lane_highv8hf, vfmsl_lane_low,
18081 vfmsl_lane_lowv4hf, vfmsl_lane_lowv8hf, vfmsl_lane_high,
18082 vfmsl_lane_highv4hf, vfmsl_lane_highv8hf): New sets of builtins.
18083 * config/arm/iterators.md (VFMLSEL2, vfmlsel2): New mode attributes.
18084 (V_lane_reg): Likewise.
18085 * config/arm/neon.md (neon_vfm<vfml_op>l_lane_<vfml_half><VCVTF:mode>):
18087 (neon_vfm<vfml_op>l_lane_<vfml_half><vfmlsel2><mode>): Likewise.
18088 (vfmal_lane_low<mode>_intrinsic,
18089 vfmal_lane_low<vfmlsel2><mode>_intrinsic,
18090 vfmal_lane_high<vfmlsel2><mode>_intrinsic,
18091 vfmal_lane_high<mode>_intrinsic, vfmsl_lane_low<mode>_intrinsic,
18092 vfmsl_lane_low<vfmlsel2><mode>_intrinsic,
18093 vfmsl_lane_high<vfmlsel2><mode>_intrinsic,
18094 vfmsl_lane_high<mode>_intrinsic): New define_insns.
18096 2018-01-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18098 * config/arm/arm-cpus.in (fp16fml): New feature.
18099 (ALL_SIMD): Add fp16fml.
18100 (armv8.2-a): Add fp16fml as an option.
18101 (armv8.3-a): Likewise.
18102 (armv8.4-a): Add fp16fml as part of fp16.
18103 * config/arm/arm.h (TARGET_FP16FML): Define.
18104 * config/arm/arm-c.c (arm_cpu_builtins): Define __ARM_FEATURE_FP16_FML
18106 * config/arm/arm-modes.def (V2HF): Define.
18107 * config/arm/arm_neon.h (vfmlal_low_u32, vfmlsl_low_u32,
18108 vfmlal_high_u32, vfmlsl_high_u32, vfmlalq_low_u32,
18109 vfmlslq_low_u32, vfmlalq_high_u32, vfmlslq_high_u32): Define.
18110 * config/arm/arm_neon_builtins.def (vfmal_low, vfmal_high,
18111 vfmsl_low, vfmsl_high): New set of builtins.
18112 * config/arm/iterators.md (PLUSMINUS): New code iterator.
18113 (vfml_op): New code attribute.
18114 (VFMLHALVES): New int iterator.
18115 (VFML, VFMLSEL): New mode attributes.
18116 (V_reg): Define mapping for V2HF.
18117 (V_hi, V_lo): New mode attributes.
18118 (VF_constraint): Likewise.
18119 (vfml_half, vfml_half_selector): New int attributes.
18120 * config/arm/neon.md (neon_vfm<vfml_op>l_<vfml_half><mode>): New
18122 (vfmal_low<mode>_intrinsic, vfmsl_high<mode>_intrinsic,
18123 vfmal_high<mode>_intrinsic, vfmsl_low<mode>_intrinsic):
18125 * config/arm/t-arm-elf (v8_fps): Add fp16fml.
18126 * config/arm/t-multilib (v8_2_a_simd_variants): Add fp16fml.
18127 * config/arm/unspecs.md (UNSPEC_VFML_LO, UNSPEC_VFML_HI): New unspecs.
18128 * doc/invoke.texi (ARM Options): Document fp16fml. Update armv8.4-a
18130 * doc/sourcebuild.texi (arm_fp16fml_neon_ok, arm_fp16fml_neon):
18131 Document new effective target and option set.
18133 2018-01-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18135 * config/arm/arm-cpus.in (armv8_4): New feature.
18136 (ARMv8_4a): New fgroup.
18137 (armv8.4-a): New arch.
18138 * config/arm/arm-tables.opt: Regenerate.
18139 * config/arm/t-aprofile: Add matching rules for -march=armv8.4-a.
18140 * config/arm/t-arm-elf (all_v8_archs): Add armv8.4-a.
18141 * config/arm/t-multilib (v8_4_a_simd_variants): New variable.
18142 Add matching rules for -march=armv8.4-a and extensions.
18143 * doc/invoke.texi (ARM Options): Document -march=armv8.4-a.
18145 2018-01-11 Oleg Endo <olegendo@gcc.gnu.org>
18148 * config/rx/rx.md (BW): New mode attribute.
18149 (sync_lock_test_and_setsi): Add mode suffix to insn output.
18151 2018-01-11 Richard Biener <rguenther@suse.de>
18153 PR tree-optimization/83435
18154 * graphite.c (canonicalize_loop_form): Ignore fake loop exit edges.
18155 * graphite-scop-detection.c (scop_detection::get_sese): Likewise.
18156 * tree-vrp.c (add_assert_info): Drop TREE_OVERFLOW if they appear.
18158 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
18159 Alan Hayward <alan.hayward@arm.com>
18160 David Sherwood <david.sherwood@arm.com>
18162 * config/aarch64/aarch64.c (aarch64_address_info): Add a const_offset
18164 (aarch64_classify_address): Initialize it. Track polynomial offsets.
18165 (aarch64_print_address_internal): Use it to check for a zero offset.
18167 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
18168 Alan Hayward <alan.hayward@arm.com>
18169 David Sherwood <david.sherwood@arm.com>
18171 * config/aarch64/aarch64-modes.def (NUM_POLY_INT_COEFFS): Set to 2.
18172 * config/aarch64/aarch64-protos.h (aarch64_initial_elimination_offset):
18173 Return a poly_int64 rather than a HOST_WIDE_INT.
18174 (aarch64_offset_7bit_signed_scaled_p): Take the offset as a poly_int64
18175 rather than a HOST_WIDE_INT.
18176 * config/aarch64/aarch64.h (aarch64_frame): Protect with
18177 HAVE_POLY_INT_H rather than HOST_WIDE_INT. Change locals_offset,
18178 hard_fp_offset, frame_size, initial_adjust, callee_offset and
18179 final_offset from HOST_WIDE_INT to poly_int64.
18180 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args): Use
18181 to_constant when getting the number of units in an Advanced SIMD
18183 (aarch64_builtin_vectorized_function): Check for a constant number
18185 * config/aarch64/aarch64-simd.md (mov<mode>): Handle polynomial
18187 (aarch64_ld<VSTRUCT:nregs>_lane<VALLDIF:mode>): Use the nunits
18188 attribute instead of GET_MODE_NUNITS.
18189 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs)
18190 (aarch64_class_max_nregs): Use the constant_lowest_bound of the
18191 GET_MODE_SIZE for fixed-size registers.
18192 (aarch64_const_vec_all_same_in_range_p): Use const_vec_duplicate_p.
18193 (aarch64_hard_regno_call_part_clobbered, aarch64_classify_index)
18194 (aarch64_mode_valid_for_sched_fusion_p, aarch64_classify_address)
18195 (aarch64_legitimize_address_displacement, aarch64_secondary_reload)
18196 (aarch64_print_operand, aarch64_print_address_internal)
18197 (aarch64_address_cost, aarch64_rtx_costs, aarch64_register_move_cost)
18198 (aarch64_short_vector_p, aapcs_vfp_sub_candidate)
18199 (aarch64_simd_attr_length_rglist, aarch64_operands_ok_for_ldpstp):
18200 Handle polynomial GET_MODE_SIZE.
18201 (aarch64_hard_regno_caller_save_mode): Likewise. Return modes
18202 wider than SImode without modification.
18203 (tls_symbolic_operand_type): Use strip_offset instead of split_const.
18204 (aarch64_pass_by_reference, aarch64_layout_arg, aarch64_pad_reg_upward)
18205 (aarch64_gimplify_va_arg_expr): Assert that we don't yet handle
18206 passing and returning SVE modes.
18207 (aarch64_function_value, aarch64_layout_arg): Use gen_int_mode
18208 rather than GEN_INT.
18209 (aarch64_emit_probe_stack_range): Take the size as a poly_int64
18210 rather than a HOST_WIDE_INT, but call sorry if it isn't constant.
18211 (aarch64_allocate_and_probe_stack_space): Likewise.
18212 (aarch64_layout_frame): Cope with polynomial offsets.
18213 (aarch64_save_callee_saves, aarch64_restore_callee_saves): Take the
18214 start_offset as a poly_int64 rather than a HOST_WIDE_INT. Track
18215 polynomial offsets.
18216 (offset_9bit_signed_unscaled_p, offset_12bit_unsigned_scaled_p)
18217 (aarch64_offset_7bit_signed_scaled_p): Take the offset as a
18218 poly_int64 rather than a HOST_WIDE_INT.
18219 (aarch64_get_separate_components, aarch64_process_components)
18220 (aarch64_expand_prologue, aarch64_expand_epilogue)
18221 (aarch64_use_return_insn_p): Handle polynomial frame offsets.
18222 (aarch64_anchor_offset): New function, split out from...
18223 (aarch64_legitimize_address): ...here.
18224 (aarch64_builtin_vectorization_cost): Handle polynomial
18225 TYPE_VECTOR_SUBPARTS.
18226 (aarch64_simd_check_vect_par_cnst_half): Handle polynomial
18228 (aarch64_simd_make_constant, aarch64_expand_vector_init): Get the
18229 number of elements from the PARALLEL rather than the mode.
18230 (aarch64_shift_truncation_mask): Use GET_MODE_UNIT_BITSIZE
18231 rather than GET_MODE_BITSIZE.
18232 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_ext)
18233 (aarch64_evpc_rev, aarch64_evpc_dup, aarch64_evpc_zip)
18234 (aarch64_expand_vec_perm_const_1): Handle polynomial
18235 d->perm.length () and d->perm elements.
18236 (aarch64_evpc_tbl): Likewise. Use nelt rather than GET_MODE_NUNITS.
18237 Apply to_constant to d->perm elements.
18238 (aarch64_simd_valid_immediate, aarch64_vec_fpconst_pow_of_2): Handle
18239 polynomial CONST_VECTOR_NUNITS.
18240 (aarch64_move_pointer): Take amount as a poly_int64 rather
18242 (aarch64_progress_pointer): Avoid temporary variable.
18243 * config/aarch64/aarch64.md (aarch64_<crc_variant>): Use
18244 the mode attribute instead of GET_MODE.
18246 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
18247 Alan Hayward <alan.hayward@arm.com>
18248 David Sherwood <david.sherwood@arm.com>
18250 * config/aarch64/aarch64.c (aarch64_force_temporary): Assert that
18251 x exists before using it.
18252 (aarch64_add_constant_internal): Rename to...
18253 (aarch64_add_offset_1): ...this. Replace regnum with separate
18254 src and dest rtxes. Handle the case in which they're different,
18255 including when the offset is zero. Replace scratchreg with an rtx.
18256 Use 2 additions if there is no spare register into which we can
18257 move a 16-bit constant.
18258 (aarch64_add_constant): Delete.
18259 (aarch64_add_offset): Replace reg with separate src and dest
18260 rtxes. Take a poly_int64 offset instead of a HOST_WIDE_INT.
18261 Use aarch64_add_offset_1.
18262 (aarch64_add_sp, aarch64_sub_sp): Take the scratch register as
18263 an rtx rather than an int. Take the delta as a poly_int64
18264 rather than a HOST_WIDE_INT. Use aarch64_add_offset.
18265 (aarch64_expand_mov_immediate): Update uses of aarch64_add_offset.
18266 (aarch64_expand_prologue): Update calls to aarch64_sub_sp,
18267 aarch64_allocate_and_probe_stack_space and aarch64_add_offset.
18268 (aarch64_expand_epilogue): Update calls to aarch64_add_offset
18269 and aarch64_add_sp.
18270 (aarch64_output_mi_thunk): Use aarch64_add_offset rather than
18271 aarch64_add_constant.
18273 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
18275 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
18276 Use scalar_float_mode.
18278 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
18280 * config/aarch64/aarch64-simd.md
18281 (aarch64_fml<f16mac1>l<f16quad>_low<mode>): Avoid GET_MODE_NUNITS.
18282 (aarch64_fml<f16mac1>l<f16quad>_high<mode>): Likewise.
18283 (aarch64_fml<f16mac1>l_lane_lowv2sf): Likewise.
18284 (aarch64_fml<f16mac1>l_lane_highv2sf): Likewise.
18285 (aarch64_fml<f16mac1>lq_laneq_lowv4sf): Likewise.
18286 (aarch64_fml<f16mac1>lq_laneq_highv4sf): Likewise.
18287 (aarch64_fml<f16mac1>l_laneq_lowv2sf): Likewise.
18288 (aarch64_fml<f16mac1>l_laneq_highv2sf): Likewise.
18289 (aarch64_fml<f16mac1>lq_lane_lowv4sf): Likewise.
18290 (aarch64_fml<f16mac1>lq_lane_highv4sf): Likewise.
18292 2018-01-11 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
18295 * config/arm/arm.c (arm_declare_function_name): Set arch_to_print if
18296 targ_options->x_arm_arch_string is non NULL.
18298 2018-01-11 Tamar Christina <tamar.christina@arm.com>
18300 * config/aarch64/aarch64.h
18301 (AARCH64_FL_FOR_ARCH8_4): Add AARCH64_FL_DOTPROD.
18303 2018-01-11 Sudakshina Das <sudi.das@arm.com>
18306 * expmed.c (emit_store_flag_force): Swap if const op0
18307 and change VOIDmode to mode of op0.
18309 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
18311 PR rtl-optimization/83761
18312 * caller-save.c (replace_reg_with_saved_mem): Pass bits rather
18313 than bytes to mode_for_size.
18315 2018-01-10 Jan Hubicka <hubicka@ucw.cz>
18317 PR middle-end/83189
18318 * gfortran.fortran-torture/compile/pr83189.f90: New testcase.
18319 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Handle zero
18322 2018-01-10 Jan Hubicka <hubicka@ucw.cz>
18324 PR middle-end/83575
18325 * cfgrtl.c (rtl_verify_edges): Only verify fixability of partition
18326 when in layout mode.
18327 (cfg_layout_finalize): Do not verify cfg before we are out of layout.
18328 * cfgcleanup.c (try_optimize_cfg): Only verify flow info when doing
18331 2018-01-10 Michael Collison <michael.collison@arm.com>
18333 * config/aarch64/aarch64-modes.def (V2HF): New VECTOR_MODE.
18334 * config/aarch64/aarch64-option-extension.def: Add
18335 AARCH64_OPT_EXTENSION of 'fp16fml'.
18336 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
18337 (__ARM_FEATURE_FP16_FML): Define if TARGET_F16FML is true.
18338 * config/aarch64/predicates.md (aarch64_lane_imm3): New predicate.
18339 * config/aarch64/constraints.md (Ui7): New constraint.
18340 * config/aarch64/iterators.md (VFMLA_W): New mode iterator.
18341 (VFMLA_SEL_W): Ditto.
18344 (VFMLA16_LOW): New int iterator.
18345 (VFMLA16_HIGH): Ditto.
18346 (UNSPEC_FMLAL): New unspec.
18347 (UNSPEC_FMLSL): Ditto.
18348 (UNSPEC_FMLAL2): Ditto.
18349 (UNSPEC_FMLSL2): Ditto.
18350 (f16mac): New code attribute.
18351 * config/aarch64/aarch64-simd-builtins.def
18352 (aarch64_fmlal_lowv2sf): Ditto.
18353 (aarch64_fmlsl_lowv2sf): Ditto.
18354 (aarch64_fmlalq_lowv4sf): Ditto.
18355 (aarch64_fmlslq_lowv4sf): Ditto.
18356 (aarch64_fmlal_highv2sf): Ditto.
18357 (aarch64_fmlsl_highv2sf): Ditto.
18358 (aarch64_fmlalq_highv4sf): Ditto.
18359 (aarch64_fmlslq_highv4sf): Ditto.
18360 (aarch64_fmlal_lane_lowv2sf): Ditto.
18361 (aarch64_fmlsl_lane_lowv2sf): Ditto.
18362 (aarch64_fmlal_laneq_lowv2sf): Ditto.
18363 (aarch64_fmlsl_laneq_lowv2sf): Ditto.
18364 (aarch64_fmlalq_lane_lowv4sf): Ditto.
18365 (aarch64_fmlsl_lane_lowv4sf): Ditto.
18366 (aarch64_fmlalq_laneq_lowv4sf): Ditto.
18367 (aarch64_fmlsl_laneq_lowv4sf): Ditto.
18368 (aarch64_fmlal_lane_highv2sf): Ditto.
18369 (aarch64_fmlsl_lane_highv2sf): Ditto.
18370 (aarch64_fmlal_laneq_highv2sf): Ditto.
18371 (aarch64_fmlsl_laneq_highv2sf): Ditto.
18372 (aarch64_fmlalq_lane_highv4sf): Ditto.
18373 (aarch64_fmlsl_lane_highv4sf): Ditto.
18374 (aarch64_fmlalq_laneq_highv4sf): Ditto.
18375 (aarch64_fmlsl_laneq_highv4sf): Ditto.
18376 * config/aarch64/aarch64-simd.md:
18377 (aarch64_fml<f16mac1>l<f16quad>_low<mode>): New pattern.
18378 (aarch64_fml<f16mac1>l<f16quad>_high<mode>): Ditto.
18379 (aarch64_simd_fml<f16mac1>l<f16quad>_low<mode>): Ditto.
18380 (aarch64_simd_fml<f16mac1>l<f16quad>_high<mode>): Ditto.
18381 (aarch64_fml<f16mac1>l_lane_lowv2sf): Ditto.
18382 (aarch64_fml<f16mac1>l_lane_highv2sf): Ditto.
18383 (aarch64_simd_fml<f16mac>l_lane_lowv2sf): Ditto.
18384 (aarch64_simd_fml<f16mac>l_lane_highv2sf): Ditto.
18385 (aarch64_fml<f16mac1>lq_laneq_lowv4sf): Ditto.
18386 (aarch64_fml<f16mac1>lq_laneq_highv4sf): Ditto.
18387 (aarch64_simd_fml<f16mac>lq_laneq_lowv4sf): Ditto.
18388 (aarch64_simd_fml<f16mac>lq_laneq_highv4sf): Ditto.
18389 (aarch64_fml<f16mac1>l_laneq_lowv2sf): Ditto.
18390 (aarch64_fml<f16mac1>l_laneq_highv2sf): Ditto.
18391 (aarch64_simd_fml<f16mac>l_laneq_lowv2sf): Ditto.
18392 (aarch64_simd_fml<f16mac>l_laneq_highv2sf): Ditto.
18393 (aarch64_fml<f16mac1>lq_lane_lowv4sf): Ditto.
18394 (aarch64_fml<f16mac1>lq_lane_highv4sf): Ditto.
18395 (aarch64_simd_fml<f16mac>lq_lane_lowv4sf): Ditto.
18396 (aarch64_simd_fml<f16mac>lq_lane_highv4sf): Ditto.
18397 * config/aarch64/arm_neon.h (vfmlal_low_u32): New intrinsic.
18398 (vfmlsl_low_u32): Ditto.
18399 (vfmlalq_low_u32): Ditto.
18400 (vfmlslq_low_u32): Ditto.
18401 (vfmlal_high_u32): Ditto.
18402 (vfmlsl_high_u32): Ditto.
18403 (vfmlalq_high_u32): Ditto.
18404 (vfmlslq_high_u32): Ditto.
18405 (vfmlal_lane_low_u32): Ditto.
18406 (vfmlsl_lane_low_u32): Ditto.
18407 (vfmlal_laneq_low_u32): Ditto.
18408 (vfmlsl_laneq_low_u32): Ditto.
18409 (vfmlalq_lane_low_u32): Ditto.
18410 (vfmlslq_lane_low_u32): Ditto.
18411 (vfmlalq_laneq_low_u32): Ditto.
18412 (vfmlslq_laneq_low_u32): Ditto.
18413 (vfmlal_lane_high_u32): Ditto.
18414 (vfmlsl_lane_high_u32): Ditto.
18415 (vfmlal_laneq_high_u32): Ditto.
18416 (vfmlsl_laneq_high_u32): Ditto.
18417 (vfmlalq_lane_high_u32): Ditto.
18418 (vfmlslq_lane_high_u32): Ditto.
18419 (vfmlalq_laneq_high_u32): Ditto.
18420 (vfmlslq_laneq_high_u32): Ditto.
18421 * config/aarch64/aarch64.h (AARCH64_FL_F16SML): New flag.
18422 (AARCH64_FL_FOR_ARCH8_4): New.
18423 (AARCH64_ISA_F16FML): New ISA flag.
18424 (TARGET_F16FML): New feature flag for fp16fml.
18425 (doc/invoke.texi): Document new fp16fml option.
18427 2018-01-10 Michael Collison <michael.collison@arm.com>
18429 * config/aarch64/aarch64-builtins.c:
18430 (aarch64_types_ternopu_imm_qualifiers, TYPES_TERNOPUI): New.
18431 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
18432 (__ARM_FEATURE_SHA3): Define if TARGET_SHA3 is true.
18433 * config/aarch64/aarch64.h (AARCH64_FL_SHA3): New flags.
18434 (AARCH64_ISA_SHA3): New ISA flag.
18435 (TARGET_SHA3): New feature flag for sha3.
18436 * config/aarch64/iterators.md (sha512_op): New int attribute.
18437 (CRYPTO_SHA512): New int iterator.
18438 (UNSPEC_SHA512H): New unspec.
18439 (UNSPEC_SHA512H2): Ditto.
18440 (UNSPEC_SHA512SU0): Ditto.
18441 (UNSPEC_SHA512SU1): Ditto.
18442 * config/aarch64/aarch64-simd-builtins.def
18443 (aarch64_crypto_sha512hqv2di): New builtin.
18444 (aarch64_crypto_sha512h2qv2di): Ditto.
18445 (aarch64_crypto_sha512su0qv2di): Ditto.
18446 (aarch64_crypto_sha512su1qv2di): Ditto.
18447 (aarch64_eor3qv8hi): Ditto.
18448 (aarch64_rax1qv2di): Ditto.
18449 (aarch64_xarqv2di): Ditto.
18450 (aarch64_bcaxqv8hi): Ditto.
18451 * config/aarch64/aarch64-simd.md:
18452 (aarch64_crypto_sha512h<sha512_op>qv2di): New pattern.
18453 (aarch64_crypto_sha512su0qv2di): Ditto.
18454 (aarch64_crypto_sha512su1qv2di): Ditto.
18455 (aarch64_eor3qv8hi): Ditto.
18456 (aarch64_rax1qv2di): Ditto.
18457 (aarch64_xarqv2di): Ditto.
18458 (aarch64_bcaxqv8hi): Ditto.
18459 * config/aarch64/arm_neon.h (vsha512hq_u64): New intrinsic.
18460 (vsha512h2q_u64): Ditto.
18461 (vsha512su0q_u64): Ditto.
18462 (vsha512su1q_u64): Ditto.
18463 (veor3q_u16): Ditto.
18464 (vrax1q_u64): Ditto.
18465 (vxarq_u64): Ditto.
18466 (vbcaxq_u16): Ditto.
18467 * config/arm/types.md (crypto_sha512): New type attribute.
18468 (crypto_sha3): Ditto.
18469 (doc/invoke.texi): Document new sha3 option.
18471 2018-01-10 Michael Collison <michael.collison@arm.com>
18473 * config/aarch64/aarch64-builtins.c:
18474 (aarch64_types_quadopu_imm_qualifiers, TYPES_QUADOPUI): New.
18475 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
18476 (__ARM_FEATURE_SM3): Define if TARGET_SM4 is true.
18477 (__ARM_FEATURE_SM4): Define if TARGET_SM4 is true.
18478 * config/aarch64/aarch64.h (AARCH64_FL_SM4): New flags.
18479 (AARCH64_ISA_SM4): New ISA flag.
18480 (TARGET_SM4): New feature flag for sm4.
18481 * config/aarch64/aarch64-simd-builtins.def
18482 (aarch64_sm3ss1qv4si): Ditto.
18483 (aarch64_sm3tt1aq4si): Ditto.
18484 (aarch64_sm3tt1bq4si): Ditto.
18485 (aarch64_sm3tt2aq4si): Ditto.
18486 (aarch64_sm3tt2bq4si): Ditto.
18487 (aarch64_sm3partw1qv4si): Ditto.
18488 (aarch64_sm3partw2qv4si): Ditto.
18489 (aarch64_sm4eqv4si): Ditto.
18490 (aarch64_sm4ekeyqv4si): Ditto.
18491 * config/aarch64/aarch64-simd.md:
18492 (aarch64_sm3ss1qv4si): Ditto.
18493 (aarch64_sm3tt<sm3tt_op>qv4si): Ditto.
18494 (aarch64_sm3partw<sm3part_op>qv4si): Ditto.
18495 (aarch64_sm4eqv4si): Ditto.
18496 (aarch64_sm4ekeyqv4si): Ditto.
18497 * config/aarch64/iterators.md (sm3tt_op): New int iterator.
18498 (sm3part_op): Ditto.
18499 (CRYPTO_SM3TT): Ditto.
18500 (CRYPTO_SM3PART): Ditto.
18501 (UNSPEC_SM3SS1): New unspec.
18502 (UNSPEC_SM3TT1A): Ditto.
18503 (UNSPEC_SM3TT1B): Ditto.
18504 (UNSPEC_SM3TT2A): Ditto.
18505 (UNSPEC_SM3TT2B): Ditto.
18506 (UNSPEC_SM3PARTW1): Ditto.
18507 (UNSPEC_SM3PARTW2): Ditto.
18508 (UNSPEC_SM4E): Ditto.
18509 (UNSPEC_SM4EKEY): Ditto.
18510 * config/aarch64/constraints.md (Ui2): New constraint.
18511 * config/aarch64/predicates.md (aarch64_imm2): New predicate.
18512 * config/arm/types.md (crypto_sm3): New type attribute.
18513 (crypto_sm4): Ditto.
18514 * config/aarch64/arm_neon.h (vsm3ss1q_u32): New intrinsic.
18515 (vsm3tt1aq_u32): Ditto.
18516 (vsm3tt1bq_u32): Ditto.
18517 (vsm3tt2aq_u32): Ditto.
18518 (vsm3tt2bq_u32): Ditto.
18519 (vsm3partw1q_u32): Ditto.
18520 (vsm3partw2q_u32): Ditto.
18521 (vsm4eq_u32): Ditto.
18522 (vsm4ekeyq_u32): Ditto.
18523 (doc/invoke.texi): Document new sm4 option.
18525 2018-01-10 Michael Collison <michael.collison@arm.com>
18527 * config/aarch64/aarch64-arches.def (armv8.4-a): New architecture.
18528 * config/aarch64/aarch64.h (AARCH64_ISA_V8_4): New ISA flag.
18529 (AARCH64_FL_FOR_ARCH8_4): New.
18530 (AARCH64_FL_V8_4): New flag.
18531 (doc/invoke.texi): Document new armv8.4-a option.
18533 2018-01-10 Michael Collison <michael.collison@arm.com>
18535 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
18536 (__ARM_FEATURE_AES): Define if TARGET_AES is true.
18537 (__ARM_FEATURE_SHA2): Define if TARGET_SHA2 is true.
18538 * config/aarch64/aarch64-option-extension.def: Add
18539 AARCH64_OPT_EXTENSION of 'sha2'.
18540 (aes): Add AARCH64_OPT_EXTENSION of 'aes'.
18541 (crypto): Disable sha2 and aes if crypto disabled.
18542 (crypto): Enable aes and sha2 if enabled.
18543 (simd): Disable sha2 and aes if simd disabled.
18544 * config/aarch64/aarch64.h (AARCH64_FL_AES, AARCH64_FL_SHA2):
18546 (AARCH64_ISA_AES, AARCH64_ISA_SHA2): New ISA flags.
18547 (TARGET_SHA2): New feature flag for sha2.
18548 (TARGET_AES): New feature flag for aes.
18549 * config/aarch64/aarch64-simd.md:
18550 (aarch64_crypto_aes<aes_op>v16qi): Make pattern
18551 conditional on TARGET_AES.
18552 (aarch64_crypto_aes<aesmc_op>v16qi): Ditto.
18553 (aarch64_crypto_sha1hsi): Make pattern conditional
18555 (aarch64_crypto_sha1hv4si): Ditto.
18556 (aarch64_be_crypto_sha1hv4si): Ditto.
18557 (aarch64_crypto_sha1su1v4si): Ditto.
18558 (aarch64_crypto_sha1<sha1_op>v4si): Ditto.
18559 (aarch64_crypto_sha1su0v4si): Ditto.
18560 (aarch64_crypto_sha256h<sha256_op>v4si): Ditto.
18561 (aarch64_crypto_sha256su0v4si): Ditto.
18562 (aarch64_crypto_sha256su1v4si): Ditto.
18563 (doc/invoke.texi): Document new aes and sha2 options.
18565 2018-01-10 Martin Sebor <msebor@redhat.com>
18567 PR tree-optimization/83781
18568 * gimple-fold.c (get_range_strlen): Avoid treating arrays of pointers
18571 2018-01-11 Martin Sebor <msebor@gmail.com>
18572 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
18574 PR tree-optimization/83501
18575 PR tree-optimization/81703
18577 * tree-ssa-strlen.c (get_string_cst): Rename...
18578 (get_string_len): ...to this. Handle global constants.
18579 (handle_char_store): Adjust.
18581 2018-01-10 Kito Cheng <kito.cheng@gmail.com>
18582 Jim Wilson <jimw@sifive.com>
18584 * config/riscv/riscv-protos.h (riscv_output_return): New.
18585 * config/riscv/riscv.c (struct machine_function): New naked_p field.
18586 (riscv_attribute_table, riscv_output_return),
18587 (riscv_handle_fndecl_attribute, riscv_naked_function_p),
18588 (riscv_allocate_stack_slots_for_args, riscv_warn_func_return): New.
18589 (riscv_compute_frame_info): Only compute frame->mask if not a naked
18591 (riscv_expand_prologue): Add early return for naked function.
18592 (riscv_expand_epilogue): Likewise.
18593 (riscv_function_ok_for_sibcall): Return false for naked function.
18594 (riscv_set_current_function): New.
18595 (TARGET_SET_CURRENT_FUNCTION, TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS),
18596 (TARGET_ATTRIBUTE_TABLE, TARGET_WARN_FUNC_RETURN): New.
18597 * config/riscv/riscv.md (simple_return): Call riscv_output_return.
18598 * doc/extend.texi (RISC-V Function Attributes): New.
18600 2018-01-10 Michael Meissner <meissner@linux.vnet.ibm.com>
18602 * config/rs6000/rs6000.c (is_complex_IBM_long_double): Explicitly
18603 check for 128-bit long double before checking TCmode.
18604 * config/rs6000/rs6000.h (FLOAT128_IEEE_P): Explicitly check for
18605 128-bit long doubles before checking TFmode or TCmode.
18606 (FLOAT128_IBM_P): Likewise.
18608 2018-01-10 Martin Sebor <msebor@redhat.com>
18610 PR tree-optimization/83671
18611 * builtins.c (c_strlen): Unconditionally return zero for the empty
18613 Use -Warray-bounds for warnings.
18614 * gimple-fold.c (get_range_strlen): Handle non-constant lengths
18615 for non-constant array indices with COMPONENT_REF, arrays of
18616 arrays, and pointers to arrays.
18617 (gimple_fold_builtin_strlen): Determine and set length range for
18618 non-constant character arrays.
18620 2018-01-10 Aldy Hernandez <aldyh@redhat.com>
18622 PR middle-end/81897
18623 * tree-ssa-uninit.c (convert_control_dep_chain_into_preds): Skip
18626 2018-01-10 Eric Botcazou <ebotcazou@adacore.com>
18628 * dwarf2out.c (dwarf2out_var_location): Do not pass NULL to fprintf.
18630 2018-01-10 Peter Bergner <bergner@vnet.ibm.com>
18633 * config/rs6000/rs6000.c (print_operand) <'y'>: Use
18634 VECTOR_MEM_ALTIVEC_OR_VSX_P.
18635 * config/rs6000/vsx.md (*vsx_le_perm_load_<mode> for VSX_D): Use
18636 indexed_or_indirect_operand predicate.
18637 (*vsx_le_perm_load_<mode> for VSX_W): Likewise.
18638 (*vsx_le_perm_load_v8hi): Likewise.
18639 (*vsx_le_perm_load_v16qi): Likewise.
18640 (*vsx_le_perm_store_<mode> for VSX_D): Likewise.
18641 (*vsx_le_perm_store_<mode> for VSX_W): Likewise.
18642 (*vsx_le_perm_store_v8hi): Likewise.
18643 (*vsx_le_perm_store_v16qi): Likewise.
18644 (eight unnamed splitters): Likewise.
18646 2018-01-10 Peter Bergner <bergner@vnet.ibm.com>
18648 * config/rs6000/x86intrin.h: Change #warning to #error. Update message.
18649 * config/rs6000/emmintrin.h: Likewise.
18650 * config/rs6000/mmintrin.h: Likewise.
18651 * config/rs6000/xmmintrin.h: Likewise.
18653 2018-01-10 David Malcolm <dmalcolm@redhat.com>
18656 * tree-core.h: Document EXPR_LOCATION_WRAPPER_P's usage of
18658 * tree.c (tree_nop_conversion): Return true for location wrapper
18660 (maybe_wrap_with_location): New function.
18661 (selftest::check_strip_nops): New function.
18662 (selftest::test_location_wrappers): New function.
18663 (selftest::tree_c_tests): Call it.
18664 * tree.h (STRIP_ANY_LOCATION_WRAPPER): New macro.
18665 (maybe_wrap_with_location): New decl.
18666 (EXPR_LOCATION_WRAPPER_P): New macro.
18667 (location_wrapper_p): New inline function.
18668 (tree_strip_any_location_wrapper): New inline function.
18670 2018-01-10 H.J. Lu <hongjiu.lu@intel.com>
18673 * config/i386/i386.c (ix86_compute_frame_layout): Always adjust
18674 stack_realign_offset for the largest alignment of stack slot
18676 (ix86_find_max_used_stack_alignment): New function.
18677 (ix86_finalize_stack_frame_flags): Use it. Set
18678 max_used_stack_alignment if we don't realign stack.
18679 * config/i386/i386.h (machine_function): Add
18680 max_used_stack_alignment.
18682 2018-01-10 Christophe Lyon <christophe.lyon@linaro.org>
18684 * config/arm/arm.opt (-mbranch-cost): New option.
18685 * config/arm/arm.h (BRANCH_COST): Take arm_branch_cost into
18688 2018-01-10 Segher Boessenkool <segher@kernel.crashing.org>
18691 * config/rs6000/rs6000.md (load_toc_v4_PIC_2, load_toc_v4_PIC_3b,
18692 load_toc_v4_PIC_3c): Wrap const term in CONST RTL.
18694 2018-01-10 Richard Biener <rguenther@suse.de>
18697 * dwarf2out.c (gen_subprogram_die): Hoist old_die && declaration
18698 early out so it also covers the case where we have a non-NULL
18701 2018-01-10 Richard Sandiford <richard.sandiford@linaro.org>
18703 PR tree-optimization/83753
18704 * tree-vect-stmts.c (get_group_load_store_type): Use VMAT_CONTIGUOUS
18705 for non-strided grouped accesses if the number of elements is 1.
18707 2018-01-10 Jan Hubicka <hubicka@ucw.cz>
18710 * i386.c (ix86_vectorize_builtin_gather): Check TARGET_USE_GATHER.
18711 * i386.h (TARGET_USE_GATHER): Define.
18712 * x86-tune.def (X86_TUNE_USE_GATHER): New.
18714 2018-01-10 Martin Liska <mliska@suse.cz>
18717 * basic-block.h (CLEANUP_NO_PARTITIONING): New define.
18718 * bb-reorder.c (pass_reorder_blocks::execute): Do not clean up
18720 * cfgcleanup.c (try_optimize_cfg): Fix up partitioning if
18721 CLEANUP_NO_PARTITIONING is not set.
18723 2018-01-10 Richard Sandiford <richard.sandiford@linaro.org>
18725 * doc/rtl.texi: Remove documentation of (const ...) wrappers
18726 for vectors, as a partial revert of r254296.
18727 * rtl.h (const_vec_p): Delete.
18728 (const_vec_duplicate_p): Don't test for vector CONSTs.
18729 (unwrap_const_vec_duplicate, const_vec_series_p): Likewise.
18730 * expmed.c (make_tree): Likewise.
18733 * common.md (E, F): Use CONSTANT_P instead of checking for
18735 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
18736 checking for CONST_VECTOR.
18738 2018-01-09 Jan Hubicka <hubicka@ucw.cz>
18740 PR middle-end/83575
18741 * predict.c (force_edge_cold): Handle in more sane way edges
18742 with no prediction.
18744 2018-01-09 Carl Love <cel@us.ibm.com>
18746 * config/rs6002/altivec.md (p8_vmrgow): Add support for V2DI, V2DF,
18748 (p8_vmrgew): Add support for V2DI, V2DF, V4SF types.
18749 * config/rs6000/rs6000-builtin.def: Add definitions for FLOAT2_V2DF,
18750 VMRGEW_V2DI, VMRGEW_V2DF, VMRGEW_V4SF, VMRGOW_V4SI, VMRGOW_V4SF,
18751 VMRGOW_V2DI, VMRGOW_V2DF. Remove definition for VMRGOW.
18752 * config/rs6000/rs6000-c.c (VSX_BUILTIN_VEC_FLOAT2,
18753 P8V_BUILTIN_VEC_VMRGEW, P8V_BUILTIN_VEC_VMRGOW): Add definitions.
18754 * config/rs6000/rs6000-protos.h: Add extern defition for
18755 rs6000_generate_float2_double_code.
18756 * config/rs6000/rs6000.c (rs6000_generate_float2_double_code): Add
18758 * config/rs6000/vsx.md (vsx_xvcdpsp): Add define_insn.
18759 (float2_v2df): Add define_expand.
18761 2018-01-09 Uros Bizjak <ubizjak@gmail.com>
18764 * combine.c (force_int_to_mode) <case ASHIFT>: Use mode instead of
18765 op_mode in the force_to_mode call.
18767 2018-01-09 Richard Sandiford <richard.sandiford@linaro.org>
18769 * config/aarch64/aarch64.c (aarch64_evpc_trn): Use d.perm.series_p
18770 instead of checking each element individually.
18771 (aarch64_evpc_uzp): Likewise.
18772 (aarch64_evpc_zip): Likewise.
18773 (aarch64_evpc_ext): Likewise.
18774 (aarch64_evpc_rev): Likewise.
18775 (aarch64_evpc_dup): Test the encoding for a single duplicated element,
18776 instead of checking each element individually. Return true without
18778 (aarch64_vectorize_vec_perm_const): Use all_from_input_p to test
18779 whether all selected elements come from the same input, instead of
18780 checking each element individually. Remove calls to gen_rtx_REG,
18781 start_sequence and end_sequence and instead assert that no rtl is
18784 2018-01-09 Richard Sandiford <richard.sandiford@linaro.org>
18786 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p): Fix
18787 order of HIGH and CONST checks.
18789 2018-01-09 Richard Sandiford <richard.sandiford@linaro.org>
18791 * tree-vect-stmts.c (permute_vec_elements): Create a fresh variable
18792 if the destination isn't an SSA_NAME.
18794 2018-01-09 Richard Biener <rguenther@suse.de>
18796 PR tree-optimization/83668
18797 * graphite.c (canonicalize_loop_closed_ssa): Add edge argument,
18799 (canonicalize_loop_form): ... here, renamed from ...
18800 (canonicalize_loop_closed_ssa_form): ... this and amended to
18801 swap successor edges for loop exit blocks to make us use
18802 the RPO order we need for initial schedule generation.
18804 2018-01-09 Joseph Myers <joseph@codesourcery.com>
18806 PR tree-optimization/64811
18807 * match.pd: When optimizing comparisons with Inf, avoid
18808 introducing or losing exceptions from comparisons with NaN.
18810 2018-01-09 Martin Liska <mliska@suse.cz>
18813 * asan.c (shadow_mem_size): Add gcc_assert.
18815 2018-01-09 Georg-Johann Lay <avr@gjlay.de>
18817 Don't save registers in main().
18820 * doc/invoke.texi (AVR Options) [-mmain-is-OS_task]: Document it.
18821 * config/avr/avr.opt (-mmain-is-OS_task): New target option.
18822 * config/avr/avr.c (avr_set_current_function): Don't error if
18823 naked, OS_task or OS_main are specified at the same time.
18824 (avr_function_ok_for_sibcall): Don't disable sibcalls for OS_task,
18826 (avr_insert_attributes) [-mmain-is-OS_task] <main>: Add OS_task
18828 * common/config/avr/avr-common.c (avr_option_optimization_table):
18829 Switch on -mmain-is-OS_task for optimizing compilations.
18831 2018-01-09 Richard Biener <rguenther@suse.de>
18833 PR tree-optimization/83572
18834 * graphite.c: Include cfganal.h.
18835 (graphite_transform_loops): Connect infinite loops to exit
18836 and remove fake edges at the end.
18838 2018-01-09 Jan Hubicka <hubicka@ucw.cz>
18840 * ipa-inline.c (edge_badness): Revert accidental checkin.
18842 2018-01-09 Jan Hubicka <hubicka@ucw.cz>
18845 * ipa-comdats.c (set_comdat_group): Only set comdat group of real
18846 symbols; not inline clones.
18848 2018-01-09 Jakub Jelinek <jakub@redhat.com>
18851 * modulo-sched.c (schedule_reg_moves): Punt if we'd need to move
18852 hard registers. Formatting fixes.
18854 PR preprocessor/83722
18855 * gcc.c (try_generate_repro): Pass
18856 &temp_stderr_files[RETRY_ICE_ATTEMPTS - 1] rather than
18857 &temp_stdout_files[RETRY_ICE_ATTEMPTS - 1] as last argument to
18860 2018-01-08 Monk Chiang <sh.chiang04@gmail.com>
18861 Kito Cheng <kito.cheng@gmail.com>
18863 * config/riscv/riscv.c (machine_function::is_leaf): Remove field.
18864 (riscv_leaf_function_p): Delete.
18865 (riscv_function_ok_for_sibcall): Return false when TARGET_SAVE_RESTORE.
18867 2018-01-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
18869 * config/rs6000/rs6000-string.c (do_load_for_compare_from_addr): New
18871 (do_ifelse): New function.
18872 (do_isel): New function.
18873 (do_sub3): New function.
18874 (do_add3): New function.
18875 (do_load_mask_compare): New function.
18876 (do_overlap_load_compare): New function.
18877 (expand_compare_loop): New function.
18878 (expand_block_compare): Call expand_compare_loop() when appropriate.
18879 * config/rs6000/rs6000.opt (-mblock-compare-inline-limit): Change
18880 option description.
18881 (-mblock-compare-inline-loop-limit): New option.
18883 2018-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18886 * config/rs6000/altivec.md (*altivec_vpermr_<mode>_internal):
18887 Reverse order of second and third operands in first alternative.
18888 * config/rs6000/rs6000.c (rs6000_expand_vector_set): Reverse order
18889 of first and second elements in UNSPEC_VPERMR vector.
18890 (altivec_expand_vec_perm_le): Likewise.
18892 2018-01-08 Jeff Law <law@redhat.com>
18894 PR rtl-optimizatin/81308
18895 * tree-switch-conversion.c (cfg_altered): New file scoped static.
18896 (process_switch): If group_case_labels makes a change, then set
18898 (pass_convert_switch::execute): If a switch is converted, then
18899 set cfg_altered. Return TODO_cfg_cleanup if cfg_altered is true.
18901 PR rtl-optimization/81308
18902 * recog.c (split_all_insns): Conditionally cleanup the CFG after
18905 2018-01-08 Vidya Praveen <vidyapraveen@arm.com>
18907 PR target/83663 - Revert r255946
18908 * config/aarch64/aarch64.c (aarch64_expand_vector_init): Modify code
18909 generation for cases where splatting a value is not useful.
18910 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
18911 across a vec_duplicate and a paradoxical subreg forming a vector
18912 mode to a vec_concat.
18914 2018-01-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18916 * config/arm/t-aprofile (MULTILIB_MATCHES): Add mapping rules for
18917 -march=armv8.3-a variants.
18918 * config/arm/t-multilib: Likewise.
18919 * config/arm/t-arm-elf: Likewise. Handle dotprod extension.
18921 2018-01-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
18923 * config/rs6000/rs6000.md (cceq_ior_compare): Remove * so I can use it
18925 (cceq_ior_compare_complement): Give it a name so I can use it, and
18926 change boolean_or_operator predicate to boolean_operator so it can
18927 be used to generate a crand.
18928 (eqne): New code iterator.
18929 (bd/bd_neg): New code_attrs.
18930 (<bd>_<mode>): New name for ctr<mode>_internal[12] now combined into
18931 a single define_insn.
18932 (<bd>tf_<mode>): A new insn pattern for the conditional form branch
18933 decrement (bdnzt/bdnzf/bdzt/bdzf).
18934 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Updated
18935 with the new names of the branch decrement patterns, and added the
18936 names of the branch decrement conditional patterns.
18938 2018-01-08 Richard Biener <rguenther@suse.de>
18940 PR tree-optimization/83563
18941 * graphite.c (canonicalize_loop_closed_ssa_form): Reset the SCEV
18944 2018-01-08 Richard Biener <rguenther@suse.de>
18946 PR middle-end/83713
18947 * convert.c (do_narrow): Properly guard TYPE_OVERFLOW_WRAPS checks.
18949 2018-01-08 Richard Biener <rguenther@suse.de>
18951 PR tree-optimization/83685
18952 * tree-ssa-pre.c (create_expression_by_pieces): Do not insert
18953 references to abnormals.
18955 2018-01-08 Richard Biener <rguenther@suse.de>
18958 * dwarf2out.c (output_indirect_strings): Handle empty
18959 skeleton_debug_str_hash.
18960 (dwarf2out_early_finish): Index strings for -gsplit-dwarf.
18962 2018-01-08 Claudiu Zissulescu <claziss@synopsys.com>
18964 * config/arc/arc.c (TARGET_TRAMPOLINE_ADJUST_ADDRESS): Delete.
18965 (emit_store_direct): Likewise.
18966 (arc_trampoline_adjust_address): Likewise.
18967 (arc_asm_trampoline_template): New function.
18968 (arc_initialize_trampoline): Use asm_trampoline_template.
18969 (TARGET_ASM_TRAMPOLINE_TEMPLATE): Define.
18970 * config/arc/arc.h (TRAMPOLINE_SIZE): Adjust to 16.
18971 * config/arc/arc.md (flush_icache): Delete pattern.
18973 2018-01-08 Claudiu Zissulescu <claziss@synopsys.com>
18975 * config/arc/arc-c.def (__ARC_UNALIGNED__): New define.
18976 * config/arc/arc.h (STRICT_ALIGNMENT): Control this macro using
18979 2018-01-08 Sebastian Huber <sebastian.huber@embedded-brains.de>
18982 * config/epiphany/epiphany.h (make_pass_mode_switch_use): Guard
18983 by not USED_FOR_TARGET.
18984 (make_pass_resolve_sw_modes): Likewise.
18986 2018-01-08 Sebastian Huber <sebastian.huber@embedded-brains.de>
18988 * config/nios2/nios2.h (nios2_section_threshold): Guard by not
18991 2018-01-08 Richard Biener <rguenther@suse.de>
18993 PR middle-end/83580
18994 * tree-data-ref.c (split_constant_offset): Remove STRIP_NOPS.
18996 2018-01-08 Richard Biener <rguenther@suse.de>
18998 PR middle-end/83517
18999 * match.pd ((t * 2) / 2) -> t): Add missing :c.
19001 2018-01-06 Aldy Hernandez <aldyh@redhat.com>
19003 PR middle-end/81897
19004 * tree-ssa-uninit.c (compute_control_dep_chain): Do not bail on
19005 basic blocks with a small number of successors.
19006 (convert_control_dep_chain_into_preds): Improve handling of
19008 (dump_predicates): Split apart into...
19009 (dump_pred_chain): ...here...
19010 (dump_pred_info): ...and here.
19011 (can_one_predicate_be_invalidated_p): Add debugging printfs.
19012 (can_chain_union_be_invalidated_p): Improve check for invalidation
19014 (uninit_uses_cannot_happen): Avoid unnecessary if
19015 convert_control_dep_chain_into_preds yielded nothing.
19017 2018-01-06 Martin Sebor <msebor@redhat.com>
19019 PR tree-optimization/83640
19020 * gimple-ssa-warn-restrict.c (builtin_access::builtin_access): Avoid
19021 subtracting negative offset from size.
19022 (builtin_access::overlap): Adjust offset bounds of the access to fall
19023 within the size of the object if possible.
19025 2018-01-06 Richard Sandiford <richard.sandiford@linaro.org>
19027 PR rtl-optimization/83699
19028 * expmed.c (extract_bit_field_1): Restrict the vector usage of
19029 extract_bit_field_as_subreg to cases in which the extracted
19030 value is also a vector.
19032 * lra-constraints.c (process_alt_operands): Test for the equivalence
19033 substitutions when detecting a possible reload cycle.
19035 2018-01-06 Jakub Jelinek <jakub@redhat.com>
19038 * toplev.c (process_options): Don't enable debug_nonbind_markers_p
19039 by default if flag_selective_schedling{,2}. Formatting fixes.
19041 PR rtl-optimization/83682
19042 * rtl.h (const_vec_duplicate_p): Only return true for VEC_DUPLICATE
19043 if it has non-VECTOR_MODE element mode.
19044 (vec_duplicate_p): Likewise.
19046 PR middle-end/83694
19047 * cfgexpand.c (expand_debug_expr): Punt if mode1 is VOIDmode
19048 and bitsize might be greater than MAX_BITSIZE_MODE_ANY_INT.
19050 2018-01-05 Jakub Jelinek <jakub@redhat.com>
19053 * config/i386/i386-builtin.def
19054 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
19055 __builtin_ia32_vgf2p8affineqb_v64qi, __builtin_ia32_vgf2p8mulb_v64qi):
19056 Require also OPTION_MASK_ISA_AVX512F in addition to
19057 OPTION_MASK_ISA_GFNI.
19058 (__builtin_ia32_vgf2p8affineinvqb_v16qi_mask,
19059 __builtin_ia32_vgf2p8affineqb_v16qi_mask): Require
19060 OPTION_MASK_ISA_AVX512VL instead of OPTION_MASK_ISA_SSE in addition
19061 to OPTION_MASK_ISA_GFNI.
19062 (__builtin_ia32_vgf2p8mulb_v32qi_mask): Require
19063 OPTION_MASK_ISA_AVX512VL in addition to OPTION_MASK_ISA_GFNI and
19064 OPTION_MASK_ISA_AVX512BW.
19065 (__builtin_ia32_vgf2p8mulb_v16qi_mask): Require
19066 OPTION_MASK_ISA_AVX512VL instead of OPTION_MASK_ISA_AVX512BW in
19067 addition to OPTION_MASK_ISA_GFNI.
19068 (__builtin_ia32_vgf2p8affineinvqb_v16qi,
19069 __builtin_ia32_vgf2p8affineqb_v16qi, __builtin_ia32_vgf2p8mulb_v16qi):
19070 Require OPTION_MASK_ISA_SSE2 instead of OPTION_MASK_ISA_SSE in addition
19071 to OPTION_MASK_ISA_GFNI.
19072 * config/i386/i386.c (def_builtin): Change to builtin isa/isa2 being
19073 a requirement for all ISAs rather than any of them with a few
19075 (ix86_add_new_builtins): Clear OPTION_MASK_ISA_64BIT from isa before
19077 (ix86_expand_builtin): Require all ISAs from builtin's isa and isa2
19078 bitmasks to be enabled with 3 exceptions, instead of requiring any
19079 enabled ISA with lots of exceptions.
19080 * config/i386/sse.md (vgf2p8affineinvqb_<mode><mask_name>,
19081 vgf2p8affineqb_<mode><mask_name>, vgf2p8mulb_<mode><mask_name>):
19082 Change avx512bw in isa attribute to avx512f.
19083 * config/i386/sgxintrin.h: Add license boilerplate.
19084 * config/i386/vaesintrin.h: Likewise. Fix macro spelling __AVX512F
19085 to __AVX512F__ and __AVX512VL to __AVX512VL__.
19086 (_mm256_aesdec_epi128, _mm256_aesdeclast_epi128, _mm256_aesenc_epi128,
19087 _mm256_aesenclast_epi128): Enable temporarily avx if __AVX__ is not
19089 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8,
19090 _mm_gf2p8affineinv_epi64_epi8, _mm_gf2p8affine_epi64_epi8): Enable
19091 temporarily sse2 rather than sse if not enabled already.
19094 * config/i386/sse.md (VI248_VLBW): Rename to ...
19095 (VI248_AVX512VL): ... this. Don't guard V32HI with TARGET_AVX512BW.
19096 (vpshrd_<mode><mask_name>, vpshld_<mode><mask_name>,
19097 vpshrdv_<mode>, vpshrdv_<mode>_mask, vpshrdv_<mode>_maskz,
19098 vpshrdv_<mode>_maskz_1, vpshldv_<mode>, vpshldv_<mode>_mask,
19099 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1): Use VI248_AVX512VL
19100 mode iterator instead of VI248_VLBW.
19102 2018-01-05 Jan Hubicka <hubicka@ucw.cz>
19104 * ipa-fnsummary.c (record_modified_bb_info): Add OP.
19105 (record_modified): Skip clobbers; add debug output.
19106 (param_change_prob): Use sreal frequencies.
19108 2018-01-05 Richard Sandiford <richard.sandiford@linaro.org>
19110 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
19111 punt for user-aligned variables.
19113 2018-01-05 Richard Sandiford <richard.sandiford@linaro.org>
19115 * tree-chrec.c (chrec_contains_symbols): Return true for
19118 2018-01-05 Sudakshina Das <sudi.das@arm.com>
19121 * simplify-rtx.c (simplify_relational_operation_1): Add simplifications
19122 of (x|y) == x for BICS pattern.
19124 2018-01-05 Jakub Jelinek <jakub@redhat.com>
19126 PR tree-optimization/83605
19127 * gimple-ssa-strength-reduction.c: Include tree-eh.h.
19128 (find_candidates_dom_walker::before_dom_children): Ignore stmts that
19131 2018-01-05 Sebastian Huber <sebastian.huber@embedded-brains.de>
19133 * config.gcc (epiphany-*-elf*): Add (epiphany-*-rtems*) configuration.
19134 * config/epiphany/rtems.h: New file.
19136 2018-01-04 Jakub Jelinek <jakub@redhat.com>
19137 Uros Bizjak <ubizjak@gmail.com>
19140 * config/i386/i386.md (*<rotate_insn>hi3_1 splitter): Use
19141 QIreg_operand instead of register_operand predicate.
19142 * config/i386/i386.c (ix86_rop_should_change_byte_p,
19143 set_rop_modrm_reg_bits, ix86_mitigate_rop): Use -mmitigate-rop in
19144 comments instead of -fmitigate[-_]rop.
19146 2018-01-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
19149 * cgraphunit.c (symbol_table::compile): Switch to text_section
19150 before calling assembly_start debug hook.
19151 * run-rtl-passes.c (run_rtl_passes): Likewise.
19154 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
19156 * tree-vrp.c (extract_range_from_binary_expr_1): Check
19157 range_int_cst_p rather than !symbolic_range_p before calling
19158 extract_range_from_multiplicative_op_1.
19160 2018-01-04 Jeff Law <law@redhat.com>
19162 * tree-ssa-math-opts.c (execute_cse_reciprocals_1): Remove
19163 redundant test in assertion.
19165 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
19167 * doc/rtl.texi: Document machine_mode wrapper classes.
19169 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
19171 * fold-const.c (fold_ternary_loc): Check tree_fits_uhwi_p before
19172 using tree_to_uhwi.
19174 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
19176 * tree-ssa-forwprop.c (is_combined_permutation_identity): Allow
19177 the VEC_PERM_EXPR fold to fail.
19179 2018-01-04 Jakub Jelinek <jakub@redhat.com>
19182 * bb-reorder.c (insert_section_boundary_note): Set has_bb_partition
19183 to switched_sections.
19185 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
19188 * config/arm/arm.c (arm_vectorize_vec_perm_const): Fix inverted
19189 test for d.testing.
19191 2018-01-04 Peter Bergner <bergner@vnet.ibm.com>
19194 * config/rs6000/rs6000.c (rs6000_discover_homogeneous_aggregate): Do not
19195 allow arguments in FP registers if TARGET_HARD_FLOAT is false.
19197 2018-01-04 Jakub Jelinek <jakub@redhat.com>
19200 * cfgexpand.c (expand_debug_expr) <case BIT_FIELD_REF>: Punt if mode
19201 is BLKmode and bitpos not zero or mode change is needed.
19203 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
19206 * config/sparc/sparc.c (sparc_vectorize_vec_perm_const): Require
19209 2018-01-04 Uros Bizjak <ubizjak@gmail.com>
19212 * config/alpha/alpha.md (*sadd<modesuffix>): Use ASHIFT
19213 instead of MULT rtx. Update all corresponding splitters.
19214 (*saddl_se): Ditto.
19215 (*ssub<modesuffix>): Ditto.
19216 (*ssubl_se): Ditto.
19217 (*cmp_sadd_di): Update split patterns.
19218 (*cmp_sadd_si): Ditto.
19219 (*cmp_sadd_sidi): Ditto.
19220 (*cmp_ssub_di): Ditto.
19221 (*cmp_ssub_si): Ditto.
19222 (*cmp_ssub_sidi): Ditto.
19223 * config/alpha/predicates.md (const23_operand): New predicate.
19224 * config/alpha/alpha.c (alpha_rtx_costs) [PLUS, MINUS]:
19225 Look for ASHIFT, not MULT inner operand.
19226 (alpha_split_conditional_move): Update for *sadd<modesuffix> change.
19228 2018-01-04 Martin Liska <mliska@suse.cz>
19230 PR gcov-profile/83669
19231 * gcov.c (output_intermediate_file): Add version to intermediate
19233 * doc/gcov.texi: Document new field 'version' in intermediate
19234 file format. Fix location of '-k' option of gcov command.
19236 2018-01-04 Martin Liska <mliska@suse.cz>
19239 * ipa-icf.c (sem_function::merge): Do not cross comdat boundary.
19241 2018-01-04 Jakub Jelinek <jakub@redhat.com>
19243 * gimple-ssa-sprintf.c (parse_directive): Cast second dir.len to uhwi.
19245 2018-01-03 Martin Sebor <msebor@redhat.com>
19247 PR tree-optimization/83655
19248 * gimple-ssa-warn-restrict.c (wrestrict_dom_walker::check_call): Avoid
19249 checking calls with invalid arguments.
19251 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
19253 * tree-vect-stmts.c (vect_get_store_rhs): New function.
19254 (vectorizable_mask_load_store): Delete.
19255 (vectorizable_call): Return false for masked loads and stores.
19256 (vectorizable_store): Handle IFN_MASK_STORE. Use vect_get_store_rhs
19257 instead of gimple_assign_rhs1.
19258 (vectorizable_load): Handle IFN_MASK_LOAD.
19259 (vect_transform_stmt): Don't set is_store for call_vec_info_type.
19261 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
19263 * tree-vect-stmts.c (vect_build_gather_load_calls): New function,
19265 (vectorizable_mask_load_store): ...here.
19266 (vectorizable_load): ...and here.
19268 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
19270 * tree-vect-stmts.c (vect_build_all_ones_mask)
19271 (vect_build_zero_merge_argument): New functions, split out from...
19272 (vectorizable_load): ...here.
19274 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
19276 * tree-vect-stmts.c (vect_check_store_rhs): New function,
19278 (vectorizable_mask_load_store): ...here.
19279 (vectorizable_store): ...and here.
19281 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
19283 * tree-vect-stmts.c (vect_check_load_store_mask): New function,
19285 (vectorizable_mask_load_store): ...here.
19287 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
19289 * tree-vectorizer.h (vec_load_store_type): Moved from tree-vec-stmts.c
19290 (vect_model_store_cost): Take a vec_load_store_type instead of a
19292 * tree-vect-stmts.c (vec_load_store_type): Move to tree-vectorizer.h.
19293 (vect_model_store_cost): Take a vec_load_store_type instead of a
19295 (vectorizable_mask_load_store): Update accordingly.
19296 (vectorizable_store): Likewise.
19297 * tree-vect-slp.c (vect_analyze_slp_cost_1): Update accordingly.
19299 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
19301 * tree-vect-loop.c (vect_transform_loop): Stub out scalar
19302 IFN_MASK_LOAD calls here rather than...
19303 * tree-vect-stmts.c (vectorizable_mask_load_store): ...here.
19305 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
19306 Alan Hayward <alan.hayward@arm.com>
19307 David Sherwood <david.sherwood@arm.com>
19309 * expmed.c (extract_bit_field_1): For vector extracts,
19310 fall back to extract_bit_field_as_subreg if vec_extract
19313 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
19314 Alan Hayward <alan.hayward@arm.com>
19315 David Sherwood <david.sherwood@arm.com>
19317 * lra-spills.c (pseudo_reg_slot_compare): Sort slots by whether
19318 they are variable or constant sized.
19319 (assign_stack_slot_num_and_sort_pseudos): Don't reuse variable-sized
19320 slots for constant-sized data.
19322 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
19323 Alan Hayward <alan.hayward@arm.com>
19324 David Sherwood <david.sherwood@arm.com>
19326 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern): When
19327 handling COND_EXPRs with boolean comparisons, try to find a better
19328 basis for the mask type than the boolean itself.
19330 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
19332 * doc/rtl.texi (MAX_BITSIZE_MODE_ANY_MODE): Describe how the default
19333 is calculated and how it can be overridden.
19334 * genmodes.c (max_bitsize_mode_any_mode): New variable.
19335 (create_modes): Initialize it from MAX_BITSIZE_MODE_ANY_MODE,
19337 (emit_max_int): Use it to set the output MAX_BITSIZE_MODE_ANY_MODE,
19340 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
19341 Alan Hayward <alan.hayward@arm.com>
19342 David Sherwood <david.sherwood@arm.com>
19344 * config/aarch64/aarch64-protos.h (aarch64_output_simd_mov_immediate):
19345 Remove the mode argument.
19346 (aarch64_simd_valid_immediate): Remove the mode and inverse
19348 * config/aarch64/iterators.md (bitsize): New iterator.
19349 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>, and<mode>3)
19350 (ior<mode>3): Update calls to aarch64_output_simd_mov_immediate.
19351 * config/aarch64/constraints.md (Do, Db, Dn): Update calls to
19352 aarch64_simd_valid_immediate.
19353 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Likewise.
19354 (aarch64_reg_or_bic_imm): Likewise.
19355 * config/aarch64/aarch64.c (simd_immediate_info): Replace mvn
19356 with an insn_type enum and msl with a modifier_type enum.
19357 Replace element_width with a scalar_mode. Change the shift
19358 to unsigned int. Add constructors for scalar_float_mode and
19359 scalar_int_mode elements.
19360 (aarch64_vect_float_const_representable_p): Delete.
19361 (aarch64_can_const_movi_rtx_p)
19362 (aarch64_simd_scalar_immediate_valid_for_move)
19363 (aarch64_simd_make_constant): Update call to
19364 aarch64_simd_valid_immediate.
19365 (aarch64_advsimd_valid_immediate_hs): New function.
19366 (aarch64_advsimd_valid_immediate): Likewise.
19367 (aarch64_simd_valid_immediate): Remove mode and inverse
19368 arguments. Rewrite to use the above. Use const_vec_duplicate_p
19369 to detect duplicated constants and use aarch64_float_const_zero_rtx_p
19370 and aarch64_float_const_representable_p on the result.
19371 (aarch64_output_simd_mov_immediate): Remove mode argument.
19372 Update call to aarch64_simd_valid_immediate and use of
19373 simd_immediate_info.
19374 (aarch64_output_scalar_simd_mov_immediate): Update call
19377 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
19378 Alan Hayward <alan.hayward@arm.com>
19379 David Sherwood <david.sherwood@arm.com>
19381 * machmode.h (mode_precision): Prefix with CONST_MODE_PRECISION.
19382 (mode_nunits): Likewise CONST_MODE_NUNITS.
19383 * machmode.def (ADJUST_NUNITS): Document.
19384 * genmodes.c (mode_data::need_nunits_adj): New field.
19385 (blank_mode): Update accordingly.
19386 (adj_nunits): New variable.
19387 (print_maybe_const_decl): Replace CATEGORY with a NEEDS_ADJ
19389 (emit_mode_size_inline): Set need_bytesize_adj for all modes
19390 listed in adj_nunits.
19391 (emit_mode_nunits_inline): Set need_nunits_adj for all modes
19392 listed in adj_nunits. Don't emit case statements for such modes.
19393 (emit_insn_modes_h): Emit definitions of CONST_MODE_NUNITS
19394 and CONST_MODE_PRECISION. Make CONST_MODE_SIZE expand to
19395 nothing if adj_nunits is nonnull.
19396 (emit_mode_precision, emit_mode_nunits): Use print_maybe_const_decl.
19397 (emit_mode_unit_size, emit_mode_base_align, emit_mode_ibit)
19398 (emit_mode_fbit): Update use of print_maybe_const_decl.
19399 (emit_move_size): Likewise. Treat the array as non-const
19401 (emit_mode_adjustments): Handle adj_nunits.
19403 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
19405 * machmode.def (VECTOR_MODES_WITH_PREFIX): Document.
19406 * genmodes.c (VECTOR_MODES_WITH_PREFIX): New macro.
19407 (VECTOR_MODES): Use it.
19408 (make_vector_modes): Take the prefix as an argument.
19410 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
19411 Alan Hayward <alan.hayward@arm.com>
19412 David Sherwood <david.sherwood@arm.com>
19414 * mode-classes.def (MODE_VECTOR_BOOL): New mode class.
19415 * machmode.h (INTEGRAL_MODE_P, VECTOR_MODE_P): Return true
19416 for MODE_VECTOR_BOOL.
19417 * machmode.def (VECTOR_BOOL_MODE): Document.
19418 * genmodes.c (VECTOR_BOOL_MODE): New macro.
19419 (make_vector_bool_mode): New function.
19420 (complete_mode, emit_mode_wider, emit_mode_adjustments): Handle
19422 * lto-streamer-in.c (lto_input_mode_table): Likewise.
19423 * rtx-vector-builder.c (rtx_vector_builder::find_cached_value):
19425 * stor-layout.c (int_mode_for_mode): Likewise.
19426 * tree.c (build_vector_type_for_mode): Likewise.
19427 * varasm.c (output_constant_pool_2): Likewise.
19428 * emit-rtl.c (init_emit_once): Make sure that CONST1_RTX (BImode) and
19429 CONSTM1_RTX (BImode) are the same thing. Initialize const_tiny_rtx
19430 for MODE_VECTOR_BOOL.
19431 * expr.c (expand_expr_real_1): Use VECTOR_MODE_P instead of a list
19432 of mode class checks.
19433 * tree-vect-generic.c (expand_vector_operation): Use VECTOR_MODE_P
19434 instead of a list of mode class checks.
19435 (expand_vector_scalar_condition): Likewise.
19436 (type_for_widest_vector_mode): Handle BImode as an inner mode.
19438 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
19439 Alan Hayward <alan.hayward@arm.com>
19440 David Sherwood <david.sherwood@arm.com>
19442 * machmode.h (mode_size): Change from unsigned short to
19444 (mode_to_bytes): Return a poly_uint16 rather than an unsigned short.
19445 (GET_MODE_SIZE): Return a constant if ONLY_FIXED_SIZE_MODES,
19446 or if measurement_type is not polynomial.
19447 (fixed_size_mode::includes_p): Check for constant-sized modes.
19448 * genmodes.c (emit_mode_size_inline): Make mode_size_inline
19449 return a poly_uint16 rather than an unsigned short.
19450 (emit_mode_size): Change the type of mode_size from unsigned short
19451 to poly_uint16_pod. Use ZERO_COEFFS for the initializer.
19452 (emit_mode_adjustments): Cope with polynomial vector sizes.
19453 * lto-streamer-in.c (lto_input_mode_table): Use bp_unpack_poly_value
19455 * lto-streamer-out.c (lto_write_mode_table): Use bp_pack_poly_value
19457 * auto-inc-dec.c (try_merge): Treat GET_MODE_SIZE as polynomial.
19458 * builtins.c (expand_ifn_atomic_compare_exchange_into_call): Likewise.
19459 * caller-save.c (setup_save_areas): Likewise.
19460 (replace_reg_with_saved_mem): Likewise.
19461 * calls.c (emit_library_call_value_1): Likewise.
19462 * combine-stack-adj.c (combine_stack_adjustments_for_block): Likewise.
19463 * combine.c (simplify_set, make_extraction, simplify_shift_const_1)
19464 (gen_lowpart_for_combine): Likewise.
19465 * convert.c (convert_to_integer_1): Likewise.
19466 * cse.c (equiv_constant, cse_insn): Likewise.
19467 * cselib.c (autoinc_split, cselib_hash_rtx): Likewise.
19468 (cselib_subst_to_values): Likewise.
19469 * dce.c (word_dce_process_block): Likewise.
19470 * df-problems.c (df_word_lr_mark_ref): Likewise.
19471 * dwarf2cfi.c (init_one_dwarf_reg_size): Likewise.
19472 * dwarf2out.c (multiple_reg_loc_descriptor, mem_loc_descriptor)
19473 (concat_loc_descriptor, concatn_loc_descriptor, loc_descriptor)
19474 (rtl_for_decl_location): Likewise.
19475 * emit-rtl.c (gen_highpart, widen_memory_access): Likewise.
19476 * expmed.c (extract_bit_field_1, extract_integral_bit_field): Likewise.
19477 * expr.c (emit_group_load_1, clear_storage_hints): Likewise.
19478 (emit_move_complex, emit_move_multi_word, emit_push_insn): Likewise.
19479 (expand_expr_real_1): Likewise.
19480 * function.c (assign_parm_setup_block_p, assign_parm_setup_block)
19481 (pad_below): Likewise.
19482 * gimple-fold.c (optimize_atomic_compare_exchange_p): Likewise.
19483 * gimple-ssa-store-merging.c (rhs_valid_for_store_merging_p): Likewise.
19484 * ira.c (get_subreg_tracking_sizes): Likewise.
19485 * ira-build.c (ira_create_allocno_objects): Likewise.
19486 * ira-color.c (coalesced_pseudo_reg_slot_compare): Likewise.
19487 (ira_sort_regnos_for_alter_reg): Likewise.
19488 * ira-costs.c (record_operand_costs): Likewise.
19489 * lower-subreg.c (interesting_mode_p, simplify_gen_subreg_concatn)
19490 (resolve_simple_move): Likewise.
19491 * lra-constraints.c (get_reload_reg, operands_match_p): Likewise.
19492 (process_addr_reg, simplify_operand_subreg, curr_insn_transform)
19493 (lra_constraints): Likewise.
19494 (CONST_POOL_OK_P): Reject variable-sized modes.
19495 * lra-spills.c (slot, assign_mem_slot, pseudo_reg_slot_compare)
19496 (add_pseudo_to_slot, lra_spill): Likewise.
19497 * omp-low.c (omp_clause_aligned_alignment): Likewise.
19498 * optabs-query.c (get_best_extraction_insn): Likewise.
19499 * optabs-tree.c (expand_vec_cond_expr_p): Likewise.
19500 * optabs.c (expand_vec_perm_var, expand_vec_cond_expr): Likewise.
19501 (expand_mult_highpart, valid_multiword_target_p): Likewise.
19502 * recog.c (offsettable_address_addr_space_p): Likewise.
19503 * regcprop.c (maybe_mode_change): Likewise.
19504 * reginfo.c (choose_hard_reg_mode, record_subregs_of_mode): Likewise.
19505 * regrename.c (build_def_use): Likewise.
19506 * regstat.c (dump_reg_info): Likewise.
19507 * reload.c (complex_word_subreg_p, push_reload, find_dummy_reload)
19508 (find_reloads, find_reloads_subreg_address): Likewise.
19509 * reload1.c (eliminate_regs_1): Likewise.
19510 * rtlanal.c (for_each_inc_dec_find_inc_dec, rtx_cost): Likewise.
19511 * simplify-rtx.c (avoid_constant_pool_reference): Likewise.
19512 (simplify_binary_operation_1, simplify_subreg): Likewise.
19513 * targhooks.c (default_function_arg_padding): Likewise.
19514 (default_hard_regno_nregs, default_class_max_nregs): Likewise.
19515 * tree-cfg.c (verify_gimple_assign_binary): Likewise.
19516 (verify_gimple_assign_ternary): Likewise.
19517 * tree-inline.c (estimate_move_cost): Likewise.
19518 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
19519 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Likewise.
19520 (get_address_cost_ainc): Likewise.
19521 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
19522 (vect_supportable_dr_alignment): Likewise.
19523 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
19524 (vectorizable_reduction): Likewise.
19525 * tree-vect-stmts.c (vectorizable_assignment, vectorizable_shift)
19526 (vectorizable_operation, vectorizable_load): Likewise.
19527 * tree.c (build_same_sized_truth_vector_type): Likewise.
19528 * valtrack.c (cleanup_auto_inc_dec): Likewise.
19529 * var-tracking.c (emit_note_insn_var_location): Likewise.
19530 * config/arc/arc.h (ASM_OUTPUT_CASE_END): Use as_a <scalar_int_mode>.
19531 (ADDR_VEC_ALIGN): Likewise.
19533 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
19534 Alan Hayward <alan.hayward@arm.com>
19535 David Sherwood <david.sherwood@arm.com>
19537 * machmode.h (mode_to_bits): Return a poly_uint16 rather than an
19539 (GET_MODE_BITSIZE): Return a constant if ONLY_FIXED_SIZE_MODES,
19540 or if measurement_type is polynomial.
19541 * calls.c (shift_return_value): Treat GET_MODE_BITSIZE as polynomial.
19542 * combine.c (make_extraction): Likewise.
19543 * dse.c (find_shift_sequence): Likewise.
19544 * dwarf2out.c (mem_loc_descriptor): Likewise.
19545 * expmed.c (store_integral_bit_field, extract_bit_field_1): Likewise.
19546 (extract_bit_field, extract_low_bits): Likewise.
19547 * expr.c (convert_move, convert_modes, emit_move_insn_1): Likewise.
19548 (optimize_bitfield_assignment_op, expand_assignment): Likewise.
19549 (store_expr_with_bounds, store_field, expand_expr_real_1): Likewise.
19550 * fold-const.c (optimize_bit_field_compare, merge_ranges): Likewise.
19551 * gimple-fold.c (optimize_atomic_compare_exchange_p): Likewise.
19552 * reload.c (find_reloads): Likewise.
19553 * reload1.c (alter_reg): Likewise.
19554 * stor-layout.c (bitwise_mode_for_mode, compute_record_mode): Likewise.
19555 * targhooks.c (default_secondary_memory_needed_mode): Likewise.
19556 * tree-if-conv.c (predicate_mem_writes): Likewise.
19557 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
19558 * tree-vect-patterns.c (adjust_bool_pattern): Likewise.
19559 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
19560 * valtrack.c (dead_debug_insert_temp): Likewise.
19561 * varasm.c (mergeable_constant_section): Likewise.
19562 * config/sh/sh.h (LOCAL_ALIGNMENT): Use as_a <fixed_size_mode>.
19564 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
19565 Alan Hayward <alan.hayward@arm.com>
19566 David Sherwood <david.sherwood@arm.com>
19568 * expr.c (expand_assignment): Cope with polynomial mode sizes
19569 when assigning to a CONCAT.
19571 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
19572 Alan Hayward <alan.hayward@arm.com>
19573 David Sherwood <david.sherwood@arm.com>
19575 * machmode.h (mode_precision): Change from unsigned short to
19577 (mode_to_precision): Return a poly_uint16 rather than an unsigned
19579 (GET_MODE_PRECISION): Return a constant if ONLY_FIXED_SIZE_MODES,
19580 or if measurement_type is not polynomial.
19581 (HWI_COMPUTABLE_MODE_P): Turn into a function. Optimize the case
19582 in which the mode is already known to be a scalar_int_mode.
19583 * genmodes.c (emit_mode_precision): Change the type of mode_precision
19584 from unsigned short to poly_uint16_pod. Use ZERO_COEFFS for the
19586 * lto-streamer-in.c (lto_input_mode_table): Use bp_unpack_poly_value
19587 for GET_MODE_PRECISION.
19588 * lto-streamer-out.c (lto_write_mode_table): Use bp_pack_poly_value
19589 for GET_MODE_PRECISION.
19590 * combine.c (update_rsp_from_reg_equal): Treat GET_MODE_PRECISION
19592 (try_combine, find_split_point, combine_simplify_rtx): Likewise.
19593 (expand_field_assignment, make_extraction): Likewise.
19594 (make_compound_operation_int, record_dead_and_set_regs_1): Likewise.
19595 (get_last_value): Likewise.
19596 * convert.c (convert_to_integer_1): Likewise.
19597 * cse.c (cse_insn): Likewise.
19598 * expr.c (expand_expr_real_1): Likewise.
19599 * lra-constraints.c (simplify_operand_subreg): Likewise.
19600 * optabs-query.c (can_atomic_load_p): Likewise.
19601 * optabs.c (expand_atomic_load): Likewise.
19602 (expand_atomic_store): Likewise.
19603 * ree.c (combine_reaching_defs): Likewise.
19604 * rtl.h (partial_subreg_p, paradoxical_subreg_p): Likewise.
19605 * rtlanal.c (nonzero_bits1, lsb_bitfield_op_p): Likewise.
19606 * tree.h (type_has_mode_precision_p): Likewise.
19607 * ubsan.c (instrument_si_overflow): Likewise.
19609 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
19610 Alan Hayward <alan.hayward@arm.com>
19611 David Sherwood <david.sherwood@arm.com>
19613 * tree.h (TYPE_VECTOR_SUBPARTS): Turn into a function and handle
19614 polynomial numbers of units.
19615 (SET_TYPE_VECTOR_SUBPARTS): Likewise.
19616 (valid_vector_subparts_p): New function.
19617 (build_vector_type): Remove temporary shim and take the number
19618 of units as a poly_uint64 rather than an int.
19619 (build_opaque_vector_type): Take the number of units as a
19620 poly_uint64 rather than an int.
19621 * tree.c (build_vector_from_ctor): Handle polynomial
19622 TYPE_VECTOR_SUBPARTS.
19623 (type_hash_canon_hash, type_cache_hasher::equal): Likewise.
19624 (uniform_vector_p, vector_type_mode, build_vector): Likewise.
19625 (build_vector_from_val): If the number of units is variable,
19626 use build_vec_duplicate_cst for constant operands and
19627 VEC_DUPLICATE_EXPR otherwise.
19628 (make_vector_type): Remove temporary is_constant ().
19629 (build_vector_type, build_opaque_vector_type): Take the number of
19630 units as a poly_uint64 rather than an int.
19631 (check_vector_cst): Handle polynomial TYPE_VECTOR_SUBPARTS and
19633 * cfgexpand.c (expand_debug_expr): Likewise.
19634 * expr.c (count_type_elements, categorize_ctor_elements_1): Likewise.
19635 (store_constructor, expand_expr_real_1): Likewise.
19636 (const_scalar_mask_from_tree): Likewise.
19637 * fold-const-call.c (fold_const_reduction): Likewise.
19638 * fold-const.c (const_binop, const_unop, fold_convert_const): Likewise.
19639 (operand_equal_p, fold_vec_perm, fold_ternary_loc): Likewise.
19640 (native_encode_vector, vec_cst_ctor_to_array): Likewise.
19641 (fold_relational_const): Likewise.
19642 (native_interpret_vector): Likewise. Change the size from an
19643 int to an unsigned int.
19644 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Handle polynomial
19645 TYPE_VECTOR_SUBPARTS.
19646 (gimple_fold_indirect_ref, gimple_build_vector): Likewise.
19647 (gimple_build_vector_from_val): Use VEC_DUPLICATE_EXPR when
19648 duplicating a non-constant operand into a variable-length vector.
19649 * hsa-brig.c (hsa_op_immed::emit_to_buffer): Handle polynomial
19650 TYPE_VECTOR_SUBPARTS and VECTOR_CST_NELTS.
19651 * ipa-icf.c (sem_variable::equals): Likewise.
19652 * match.pd: Likewise.
19653 * omp-simd-clone.c (simd_clone_subparts): Likewise.
19654 * print-tree.c (print_node): Likewise.
19655 * stor-layout.c (layout_type): Likewise.
19656 * targhooks.c (default_builtin_vectorization_cost): Likewise.
19657 * tree-cfg.c (verify_gimple_comparison): Likewise.
19658 (verify_gimple_assign_binary): Likewise.
19659 (verify_gimple_assign_ternary): Likewise.
19660 (verify_gimple_assign_single): Likewise.
19661 * tree-pretty-print.c (dump_generic_node): Likewise.
19662 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
19663 (simplify_bitfield_ref, is_combined_permutation_identity): Likewise.
19664 * tree-vect-data-refs.c (vect_permute_store_chain): Likewise.
19665 (vect_grouped_load_supported, vect_permute_load_chain): Likewise.
19666 (vect_shift_permute_load_chain): Likewise.
19667 * tree-vect-generic.c (nunits_for_known_piecewise_op): Likewise.
19668 (expand_vector_condition, optimize_vector_constructor): Likewise.
19669 (lower_vec_perm, get_compute_type): Likewise.
19670 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
19671 (get_initial_defs_for_reduction, vect_transform_loop): Likewise.
19672 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
19673 (vect_recog_mask_conversion_pattern): Likewise.
19674 * tree-vect-slp.c (vect_supported_load_permutation_p): Likewise.
19675 (vect_get_constant_vectors, vect_transform_slp_perm_load): Likewise.
19676 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
19677 (get_group_load_store_type, vectorizable_mask_load_store): Likewise.
19678 (vectorizable_bswap, simd_clone_subparts, vectorizable_assignment)
19679 (vectorizable_shift, vectorizable_operation, vectorizable_store)
19680 (vectorizable_load, vect_is_simple_cond, vectorizable_comparison)
19681 (supportable_widening_operation): Likewise.
19682 (supportable_narrowing_operation): Likewise.
19683 * tree-vector-builder.c (tree_vector_builder::binary_encoded_nelts):
19685 * varasm.c (output_constant): Likewise.
19687 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
19688 Alan Hayward <alan.hayward@arm.com>
19689 David Sherwood <david.sherwood@arm.com>
19691 * tree-vect-data-refs.c (vect_permute_store_chain): Reorganize
19692 so that both the length == 3 and length != 3 cases set up their
19693 own permute vectors. Add comments explaining why we know the
19694 number of elements is constant.
19695 (vect_permute_load_chain): Likewise.
19697 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
19698 Alan Hayward <alan.hayward@arm.com>
19699 David Sherwood <david.sherwood@arm.com>
19701 * machmode.h (mode_nunits): Change from unsigned char to
19703 (ONLY_FIXED_SIZE_MODES): New macro.
19704 (pod_mode::measurement_type, scalar_int_mode::measurement_type)
19705 (scalar_float_mode::measurement_type, scalar_mode::measurement_type)
19706 (complex_mode::measurement_type, fixed_size_mode::measurement_type):
19708 (mode_to_nunits): Return a poly_uint16 rather than an unsigned short.
19709 (GET_MODE_NUNITS): Return a constant if ONLY_FIXED_SIZE_MODES,
19710 or if measurement_type is not polynomial.
19711 * genmodes.c (ZERO_COEFFS): New macro.
19712 (emit_mode_nunits_inline): Make mode_nunits_inline return a
19714 (emit_mode_nunits): Change the type of mode_nunits to poly_uint16_pod.
19715 Use ZERO_COEFFS when emitting initializers.
19716 * data-streamer.h (bp_pack_poly_value): New function.
19717 (bp_unpack_poly_value): Likewise.
19718 * lto-streamer-in.c (lto_input_mode_table): Use bp_unpack_poly_value
19719 for GET_MODE_NUNITS.
19720 * lto-streamer-out.c (lto_write_mode_table): Use bp_pack_poly_value
19721 for GET_MODE_NUNITS.
19722 * tree.c (make_vector_type): Remove temporary shim and make
19723 the real function take the number of units as a poly_uint64
19724 rather than an int.
19725 (build_vector_type_for_mode): Handle polynomial nunits.
19726 * dwarf2out.c (loc_descriptor, add_const_value_attribute): Likewise.
19727 * emit-rtl.c (const_vec_series_p_1): Likewise.
19728 (gen_rtx_CONST_VECTOR): Likewise.
19729 * fold-const.c (test_vec_duplicate_folding): Likewise.
19730 * genrecog.c (validate_pattern): Likewise.
19731 * optabs-query.c (can_vec_perm_var_p, can_mult_highpart_p): Likewise.
19732 * optabs-tree.c (expand_vec_cond_expr_p): Likewise.
19733 * optabs.c (expand_vector_broadcast, expand_binop_directly): Likewise.
19734 (shift_amt_for_vec_perm_mask, expand_vec_perm_var): Likewise.
19735 (expand_vec_cond_expr, expand_mult_highpart): Likewise.
19736 * rtlanal.c (subreg_get_info): Likewise.
19737 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
19738 (vect_grouped_load_supported): Likewise.
19739 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
19740 * tree-vect-loop.c (have_whole_vector_shift): Likewise.
19741 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
19742 (simplify_const_unary_operation, simplify_binary_operation_1)
19743 (simplify_const_binary_operation, simplify_ternary_operation)
19744 (test_vector_ops_duplicate, test_vector_ops): Likewise.
19745 (simplify_immed_subreg): Use GET_MODE_NUNITS on a fixed_size_mode
19746 instead of CONST_VECTOR_NUNITS.
19747 * varasm.c (output_constant_pool_2): Likewise.
19748 * rtx-vector-builder.c (rtx_vector_builder::build): Only include the
19749 explicit-encoded elements in the XVEC for variable-length vectors.
19751 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
19753 * lra-constraints.c (curr_insn_transform): Use partial_subreg_p.
19755 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
19756 Alan Hayward <alan.hayward@arm.com>
19757 David Sherwood <david.sherwood@arm.com>
19759 * coretypes.h (fixed_size_mode): Declare.
19760 (fixed_size_mode_pod): New typedef.
19761 * builtins.h (target_builtins::x_apply_args_mode)
19762 (target_builtins::x_apply_result_mode): Change type to
19763 fixed_size_mode_pod.
19764 * builtins.c (apply_args_size, apply_result_size, result_vector)
19765 (expand_builtin_apply_args_1, expand_builtin_apply)
19766 (expand_builtin_return): Update accordingly.
19768 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
19770 * cse.c (hash_rtx_cb): Hash only the encoded elements.
19771 * cselib.c (cselib_hash_rtx): Likewise.
19772 * expmed.c (make_tree): Build VECTOR_CSTs directly from the
19773 CONST_VECTOR encoding.
19775 2018-01-03 Jakub Jelinek <jakub@redhat.com>
19776 Jeff Law <law@redhat.com>
19779 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): For
19780 noreturn probe, use gen_pop instead of ix86_emit_restore_reg_using_pop,
19781 only set RTX_FRAME_RELATED_P on both the push and pop if cfa_reg is sp
19782 and add REG_CFA_ADJUST_CFA notes in that case to both insns.
19785 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Do not
19786 explicitly probe *sp in a noreturn function if there were any callee
19787 register saves or frame pointer is needed.
19789 2018-01-03 Jakub Jelinek <jakub@redhat.com>
19792 * cfgexpand.c (expand_debug_expr): Return NULL if mode is
19793 BLKmode for ternary, binary or unary expressions.
19796 * var-tracking.c (delete_vta_debug_insn): New inline function.
19797 (delete_vta_debug_insns): Add USE_CFG argument, if true, walk just
19798 insns from get_insns () to NULL instead of each bb separately.
19799 Use delete_vta_debug_insn. No longer static.
19800 (vt_debug_insns_local, variable_tracking_main_1): Adjust
19801 delete_vta_debug_insns callers.
19802 * rtl.h (delete_vta_debug_insns): Declare.
19803 * final.c (rest_of_handle_final): Call delete_vta_debug_insns
19804 instead of variable_tracking_main.
19806 2018-01-03 Martin Sebor <msebor@redhat.com>
19808 PR tree-optimization/83603
19809 * calls.c (maybe_warn_nonstring_arg): Avoid accessing function
19810 arguments past the endof the argument list in functions declared
19811 without a prototype.
19812 * gimple-ssa-warn-restrict.c (wrestrict_dom_walker::check_call):
19813 Avoid checking when arguments are null.
19815 2018-01-03 Martin Sebor <msebor@redhat.com>
19818 * doc/extend.texi (attribute const): Fix a typo.
19819 * ipa-pure-const.c ((warn_function_const, warn_function_pure): Avoid
19820 issuing -Wsuggest-attribute for void functions.
19822 2018-01-03 Martin Sebor <msebor@redhat.com>
19824 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Use
19825 offset_int::from instead of wide_int::to_shwi.
19826 (maybe_diag_overlap): Remove assertion.
19827 Use HOST_WIDE_INT_PRINT_DEC instead of %lli.
19828 * gimple-ssa-sprintf.c (format_directive): Same.
19829 (parse_directive): Same.
19830 (sprintf_dom_walker::compute_format_length): Same.
19831 (try_substitute_return_value): Same.
19833 2018-01-03 Jeff Law <law@redhat.com>
19835 PR middle-end/83654
19836 * explow.c (anti_adjust_stack_and_probe_stack_clash): Test a
19837 non-constant residual for zero at runtime and avoid probing in
19838 that case. Reorganize code for trailing problem to mirror handling
19841 2018-01-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
19843 PR tree-optimization/83501
19844 * tree-ssa-strlen.c (get_string_cst): New.
19845 (handle_char_store): Call get_string_cst.
19847 2018-01-03 Martin Liska <mliska@suse.cz>
19849 PR tree-optimization/83593
19850 * tree-ssa-strlen.c: Include tree-cfg.h.
19851 (strlen_check_and_optimize_stmt): Add new argument cleanup_eh.
19852 (strlen_dom_walker): Add new member variable m_cleanup_cfg.
19853 (strlen_dom_walker::strlen_dom_walker): Initialize m_cleanup_cfg
19855 (strlen_dom_walker::before_dom_children): Call
19856 gimple_purge_dead_eh_edges. Dump tranformation with details
19858 (strlen_dom_walker::before_dom_children): Update call by adding
19859 new argument cleanup_eh.
19860 (pass_strlen::execute): Return TODO_cleanup_cfg if needed.
19862 2018-01-03 Martin Liska <mliska@suse.cz>
19865 * cif-code.def (VARIADIC_THUNK): New enum value.
19866 * ipa-fnsummary.c (compute_fn_summary): Do not inline variadic
19869 2018-01-03 Jan Beulich <jbeulich@suse.com>
19871 * sse.md (mov<mode>_internal): Tighten condition for when to use
19872 vmovdqu<ssescalarsize> for TI and OI modes.
19874 2018-01-03 Jakub Jelinek <jakub@redhat.com>
19876 Update copyright years.
19878 2018-01-03 Martin Liska <mliska@suse.cz>
19881 * ipa-visibility.c (function_and_variable_visibility): Skip
19882 functions with noipa attribure.
19884 2018-01-03 Jakub Jelinek <jakub@redhat.com>
19886 * gcc.c (process_command): Update copyright notice dates.
19887 * gcov-dump.c (print_version): Ditto.
19888 * gcov.c (print_version): Ditto.
19889 * gcov-tool.c (print_version): Ditto.
19890 * gengtype.c (create_file): Ditto.
19891 * doc/cpp.texi: Bump @copying's copyright year.
19892 * doc/cppinternals.texi: Ditto.
19893 * doc/gcc.texi: Ditto.
19894 * doc/gccint.texi: Ditto.
19895 * doc/gcov.texi: Ditto.
19896 * doc/install.texi: Ditto.
19897 * doc/invoke.texi: Ditto.
19899 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
19901 * vector-builder.h (vector_builder::m_full_nelts): Change from
19902 unsigned int to poly_uint64.
19903 (vector_builder::full_nelts): Update prototype accordingly.
19904 (vector_builder::new_vector): Likewise.
19905 (vector_builder::encoded_full_vector_p): Handle polynomial full_nelts.
19906 (vector_builder::operator ==): Likewise.
19907 (vector_builder::finalize): Likewise.
19908 * int-vector-builder.h (int_vector_builder::int_vector_builder):
19909 Take the number of elements as a poly_uint64 rather than an
19911 * vec-perm-indices.h (vec_perm_indices::m_nelts_per_input): Change
19912 from unsigned int to poly_uint64.
19913 (vec_perm_indices::vec_perm_indices): Update prototype accordingly.
19914 (vec_perm_indices::new_vector): Likewise.
19915 (vec_perm_indices::length): Likewise.
19916 (vec_perm_indices::nelts_per_input): Likewise.
19917 (vec_perm_indices::input_nelts): Likewise.
19918 * vec-perm-indices.c (vec_perm_indices::new_vector): Take the
19919 number of elements per input as a poly_uint64 rather than an
19920 unsigned int. Use the original encoding for variable-length
19921 vectors, rather than clamping each individual element.
19922 For the second and subsequent elements in each pattern,
19923 clamp the step and base before clamping their sum.
19924 (vec_perm_indices::series_p): Handle polynomial element counts.
19925 (vec_perm_indices::all_in_range_p): Likewise.
19926 (vec_perm_indices_to_tree): Likewise.
19927 (vec_perm_indices_to_rtx): Likewise.
19928 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
19929 * tree-vector-builder.c (tree_vector_builder::new_unary_operation)
19930 (tree_vector_builder::new_binary_operation): Handle polynomial
19931 element counts. Return false if we need to know the number
19932 of elements at compile time.
19933 * fold-const.c (fold_vec_perm): Punt if the number of elements
19934 isn't known at compile time.
19936 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
19938 * vec-perm-indices.h (vec_perm_builder): Change element type
19939 from HOST_WIDE_INT to poly_int64.
19940 (vec_perm_indices::element_type): Update accordingly.
19941 (vec_perm_indices::clamp): Handle polynomial element_types.
19942 * vec-perm-indices.c (vec_perm_indices::series_p): Likewise.
19943 (vec_perm_indices::all_in_range_p): Likewise.
19944 (tree_to_vec_perm_builder): Check for poly_int64 trees rather
19946 * vector-builder.h (vector_builder::stepped_sequence_p): Handle
19947 polynomial vec_perm_indices element types.
19948 * int-vector-builder.h (int_vector_builder::equal_p): Likewise.
19949 * fold-const.c (fold_vec_perm): Likewise.
19950 * optabs.c (shift_amt_for_vec_perm_mask): Likewise.
19951 * tree-vect-generic.c (lower_vec_perm): Likewise.
19952 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
19953 * config/aarch64/aarch64.c (aarch64_evpc_tbl): Cast d->perm
19954 element type to HOST_WIDE_INT.
19956 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
19957 Alan Hayward <alan.hayward@arm.com>
19958 David Sherwood <david.sherwood@arm.com>
19960 * alias.c (addr_side_effect_eval): Take the size as a poly_int64
19961 rather than an int. Use plus_constant.
19962 (memrefs_conflict_p): Take the sizes as poly_int64s rather than ints.
19963 Take the offset "c" as a poly_int64 rather than a HOST_WIDE_INT.
19965 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
19966 Alan Hayward <alan.hayward@arm.com>
19967 David Sherwood <david.sherwood@arm.com>
19969 * calls.c (emit_call_1, expand_call): Change struct_value_size from
19970 a HOST_WIDE_INT to a poly_int64.
19972 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
19973 Alan Hayward <alan.hayward@arm.com>
19974 David Sherwood <david.sherwood@arm.com>
19976 * calls.c (load_register_parameters): Cope with polynomial
19977 mode sizes. Require a constant size for BLKmode parameters
19978 that aren't described by a PARALLEL. If BLOCK_REG_PADDING
19979 forces a parameter to be padded at the lsb end in order to
19980 fill a complete number of words, require the parameter size
19981 to be ordered wrt UNITS_PER_WORD.
19983 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
19984 Alan Hayward <alan.hayward@arm.com>
19985 David Sherwood <david.sherwood@arm.com>
19987 * reload1.c (spill_stack_slot_width): Change element type
19988 from unsigned int to poly_uint64_pod.
19989 (alter_reg): Treat mode sizes as polynomial.
19991 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
19992 Alan Hayward <alan.hayward@arm.com>
19993 David Sherwood <david.sherwood@arm.com>
19995 * reload.c (complex_word_subreg_p): New function.
19996 (reload_inner_reg_of_subreg, push_reload): Use it.
19998 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
19999 Alan Hayward <alan.hayward@arm.com>
20000 David Sherwood <david.sherwood@arm.com>
20002 * lra-constraints.c (process_alt_operands): Reject matched
20003 operands whose sizes aren't ordered.
20004 (match_reload): Refer to this check here.
20006 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
20007 Alan Hayward <alan.hayward@arm.com>
20008 David Sherwood <david.sherwood@arm.com>
20010 * builtins.c (expand_ifn_atomic_compare_exchange_into_call): Assert
20011 that the mode size is in the set {1, 2, 4, 8, 16}.
20013 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
20014 Alan Hayward <alan.hayward@arm.com>
20015 David Sherwood <david.sherwood@arm.com>
20017 * var-tracking.c (adjust_mems): Treat mode sizes as polynomial.
20018 Use plus_constant instead of gen_rtx_PLUS.
20020 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
20021 Alan Hayward <alan.hayward@arm.com>
20022 David Sherwood <david.sherwood@arm.com>
20024 * config/cr16/cr16-protos.h (cr16_push_rounding): Declare.
20025 * config/cr16/cr16.h (PUSH_ROUNDING): Move implementation to...
20026 * config/cr16/cr16.c (cr16_push_rounding): ...this new function.
20027 * config/h8300/h8300-protos.h (h8300_push_rounding): Declare.
20028 * config/h8300/h8300.h (PUSH_ROUNDING): Move implementation to...
20029 * config/h8300/h8300.c (h8300_push_rounding): ...this new function.
20030 * config/i386/i386-protos.h (ix86_push_rounding): Declare.
20031 * config/i386/i386.h (PUSH_ROUNDING): Move implementation to...
20032 * config/i386/i386.c (ix86_push_rounding): ...this new function.
20033 * config/m32c/m32c-protos.h (m32c_push_rounding): Take and return
20035 * config/m32c/m32c.c (m32c_push_rounding): Likewise.
20036 * config/m68k/m68k-protos.h (m68k_push_rounding): Declare.
20037 * config/m68k/m68k.h (PUSH_ROUNDING): Move implementation to...
20038 * config/m68k/m68k.c (m68k_push_rounding): ...this new function.
20039 * config/pdp11/pdp11-protos.h (pdp11_push_rounding): Declare.
20040 * config/pdp11/pdp11.h (PUSH_ROUNDING): Move implementation to...
20041 * config/pdp11/pdp11.c (pdp11_push_rounding): ...this new function.
20042 * config/stormy16/stormy16-protos.h (xstormy16_push_rounding): Declare.
20043 * config/stormy16/stormy16.h (PUSH_ROUNDING): Move implementation to...
20044 * config/stormy16/stormy16.c (xstormy16_push_rounding): ...this new
20046 * expr.c (emit_move_resolve_push): Treat the input and result
20047 of PUSH_ROUNDING as a poly_int64.
20048 (emit_move_complex_push, emit_single_push_insn_1): Likewise.
20049 (emit_push_insn): Likewise.
20050 * lra-eliminations.c (mark_not_eliminable): Likewise.
20051 * recog.c (push_operand): Likewise.
20052 * reload1.c (elimination_effects): Likewise.
20053 * rtlanal.c (nonzero_bits1): Likewise.
20054 * calls.c (store_one_arg): Likewise. Require the padding to be
20055 known at compile time.
20057 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
20058 Alan Hayward <alan.hayward@arm.com>
20059 David Sherwood <david.sherwood@arm.com>
20061 * expr.c (emit_single_push_insn_1): Treat mode sizes as polynomial.
20062 Use plus_constant instead of gen_rtx_PLUS.
20064 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
20065 Alan Hayward <alan.hayward@arm.com>
20066 David Sherwood <david.sherwood@arm.com>
20068 * auto-inc-dec.c (set_inc_state): Take the mode size as a poly_int64
20069 rather than an int.
20071 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
20072 Alan Hayward <alan.hayward@arm.com>
20073 David Sherwood <david.sherwood@arm.com>
20075 * expr.c (expand_expr_real_1): Use tree_to_poly_uint64
20076 instead of int_size_in_bytes when handling VIEW_CONVERT_EXPRs
20077 via stack temporaries. Treat the mode size as polynomial too.
20079 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
20080 Alan Hayward <alan.hayward@arm.com>
20081 David Sherwood <david.sherwood@arm.com>
20083 * expr.c (expand_expr_real_2): When handling conversions involving
20084 unions, apply tree_to_poly_uint64 to the TYPE_SIZE rather than
20085 multiplying int_size_in_bytes by BITS_PER_UNIT. Treat GET_MODE_BISIZE
20086 as a poly_uint64 too.
20088 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
20089 Alan Hayward <alan.hayward@arm.com>
20090 David Sherwood <david.sherwood@arm.com>
20092 * rtlanal.c (subreg_get_info): Handle polynomial mode sizes.
20094 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
20095 Alan Hayward <alan.hayward@arm.com>
20096 David Sherwood <david.sherwood@arm.com>
20098 * combine.c (can_change_dest_mode): Handle polynomial
20099 REGMODE_NATURAL_SIZE.
20100 * expmed.c (store_bit_field_1): Likewise.
20101 * expr.c (store_constructor): Likewise.
20102 * emit-rtl.c (validate_subreg): Operate on polynomial mode sizes
20103 and polynomial REGMODE_NATURAL_SIZE.
20104 (gen_lowpart_common): Likewise.
20105 * reginfo.c (record_subregs_of_mode): Likewise.
20106 * rtlanal.c (read_modify_subreg_p): Likewise.
20108 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
20109 Alan Hayward <alan.hayward@arm.com>
20110 David Sherwood <david.sherwood@arm.com>
20112 * internal-fn.c (expand_vector_ubsan_overflow): Handle polynomial
20113 numbers of elements.
20115 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
20116 Alan Hayward <alan.hayward@arm.com>
20117 David Sherwood <david.sherwood@arm.com>
20119 * match.pd: Cope with polynomial numbers of vector elements.
20121 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
20122 Alan Hayward <alan.hayward@arm.com>
20123 David Sherwood <david.sherwood@arm.com>
20125 * fold-const.c (fold_indirect_ref_1): Handle polynomial offsets
20126 in a POINTER_PLUS_EXPR.
20128 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
20129 Alan Hayward <alan.hayward@arm.com>
20130 David Sherwood <david.sherwood@arm.com>
20132 * omp-simd-clone.c (simd_clone_subparts): New function.
20133 (simd_clone_init_simd_arrays): Use it instead of TYPE_VECTOR_SUBPARTS.
20134 (ipa_simd_modify_function_body): Likewise.
20136 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
20137 Alan Hayward <alan.hayward@arm.com>
20138 David Sherwood <david.sherwood@arm.com>
20140 * tree-vect-generic.c (nunits_for_known_piecewise_op): New function.
20141 (expand_vector_piecewise): Use it instead of TYPE_VECTOR_SUBPARTS.
20142 (expand_vector_addition, add_rshift, expand_vector_divmod): Likewise.
20143 (expand_vector_condition, vector_element): Likewise.
20144 (subparts_gt): New function.
20145 (get_compute_type): Use subparts_gt.
20146 (count_type_subparts): Delete.
20147 (expand_vector_operations_1): Use subparts_gt instead of
20148 count_type_subparts.
20150 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
20151 Alan Hayward <alan.hayward@arm.com>
20152 David Sherwood <david.sherwood@arm.com>
20154 * tree-vect-data-refs.c (vect_no_alias_p): Replace with...
20155 (vect_compile_time_alias): ...this new function. Do the calculation
20156 on poly_ints rather than trees.
20157 (vect_prune_runtime_alias_test_list): Update call accordingly.
20159 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
20160 Alan Hayward <alan.hayward@arm.com>
20161 David Sherwood <david.sherwood@arm.com>
20163 * tree-vect-slp.c (vect_build_slp_tree_1): Handle polynomial
20165 (vect_schedule_slp_instance): Likewise.
20167 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
20168 Alan Hayward <alan.hayward@arm.com>
20169 David Sherwood <david.sherwood@arm.com>
20171 * tree-vect-slp.c (vect_get_and_check_slp_defs): Reject
20172 constant and extern definitions for variable-length vectors.
20173 (vect_get_constant_vectors): Note that the number of units
20174 is known to be constant.
20176 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
20177 Alan Hayward <alan.hayward@arm.com>
20178 David Sherwood <david.sherwood@arm.com>
20180 * tree-vect-stmts.c (vectorizable_conversion): Treat the number
20181 of units as polynomial. Choose between WIDE and NARROW based
20184 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
20185 Alan Hayward <alan.hayward@arm.com>
20186 David Sherwood <david.sherwood@arm.com>
20188 * tree-vect-stmts.c (simd_clone_subparts): New function.
20189 (vectorizable_simd_clone_call): Use it instead of TYPE_VECTOR_SUBPARTS.
20191 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
20192 Alan Hayward <alan.hayward@arm.com>
20193 David Sherwood <david.sherwood@arm.com>
20195 * tree-vect-stmts.c (vectorizable_call): Treat the number of
20196 vectors as polynomial. Use build_index_vector for
20197 IFN_GOMP_SIMD_LANE.
20199 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
20200 Alan Hayward <alan.hayward@arm.com>
20201 David Sherwood <david.sherwood@arm.com>
20203 * tree-vect-stmts.c (get_load_store_type): Treat the number of
20204 units as polynomial. Reject VMAT_ELEMENTWISE and VMAT_STRIDED_SLP
20205 for variable-length vectors.
20206 (vectorizable_mask_load_store): Treat the number of units as
20207 polynomial, asserting that it is constant if the condition has
20208 already been enforced.
20209 (vectorizable_store, vectorizable_load): Likewise.
20211 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
20212 Alan Hayward <alan.hayward@arm.com>
20213 David Sherwood <david.sherwood@arm.com>
20215 * tree-vect-loop.c (vectorizable_live_operation): Treat the number
20216 of units as polynomial. Punt if we can't tell at compile time
20217 which vector contains the final result.
20219 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
20220 Alan Hayward <alan.hayward@arm.com>
20221 David Sherwood <david.sherwood@arm.com>
20223 * tree-vect-loop.c (vectorizable_induction): Treat the number
20224 of units as polynomial. Punt on SLP inductions. Use an integer
20225 VEC_SERIES_EXPR for variable-length integer reductions. Use a
20226 cast of such a series for variable-length floating-point
20229 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
20230 Alan Hayward <alan.hayward@arm.com>
20231 David Sherwood <david.sherwood@arm.com>
20233 * tree.h (build_index_vector): Declare.
20234 * tree.c (build_index_vector): New function.
20235 * tree-vect-loop.c (get_initial_defs_for_reduction): Treat the number
20236 of units as polynomial, forcibly converting it to a constant if
20237 vectorizable_reduction has already enforced the condition.
20238 (vect_create_epilog_for_reduction): Likewise. Use build_index_vector
20239 to create a {1,2,3,...} vector.
20240 (vectorizable_reduction): Treat the number of units as polynomial.
20241 Choose vectype_in based on the largest scalar element size rather
20242 than the smallest number of units. Enforce the restrictions
20245 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
20246 Alan Hayward <alan.hayward@arm.com>
20247 David Sherwood <david.sherwood@arm.com>
20249 * tree-vect-data-refs.c (vector_alignment_reachable_p): Treat the
20250 number of units as polynomial.
20252 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
20253 Alan Hayward <alan.hayward@arm.com>
20254 David Sherwood <david.sherwood@arm.com>
20256 * target.h (vector_sizes, auto_vector_sizes): New typedefs.
20257 * target.def (autovectorize_vector_sizes): Return the vector sizes
20258 by pointer, using vector_sizes rather than a bitmask.
20259 * targhooks.h (default_autovectorize_vector_sizes): Update accordingly.
20260 * targhooks.c (default_autovectorize_vector_sizes): Likewise.
20261 * config/aarch64/aarch64.c (aarch64_autovectorize_vector_sizes):
20263 * config/arc/arc.c (arc_autovectorize_vector_sizes): Likewise.
20264 * config/arm/arm.c (arm_autovectorize_vector_sizes): Likewise.
20265 * config/i386/i386.c (ix86_autovectorize_vector_sizes): Likewise.
20266 * config/mips/mips.c (mips_autovectorize_vector_sizes): Likewise.
20267 * omp-general.c (omp_max_vf): Likewise.
20268 * omp-low.c (omp_clause_aligned_alignment): Likewise.
20269 * optabs-query.c (can_vec_mask_load_store_p): Likewise.
20270 * tree-vect-loop.c (vect_analyze_loop): Likewise.
20271 * tree-vect-slp.c (vect_slp_bb): Likewise.
20272 * doc/tm.texi: Regenerate.
20273 * tree-vectorizer.h (current_vector_size): Change from an unsigned int
20275 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Take
20276 the vector size as a poly_uint64 rather than an unsigned int.
20277 (current_vector_size): Change from an unsigned int to a poly_uint64.
20278 (get_vectype_for_scalar_type): Update accordingly.
20279 * tree.h (build_truth_vector_type): Take the size and number of
20280 units as a poly_uint64 rather than an unsigned int.
20281 (build_vector_type): Add a temporary overload that takes
20282 the number of units as a poly_uint64 rather than an unsigned int.
20283 * tree.c (make_vector_type): Likewise.
20284 (build_truth_vector_type): Take the number of units as a poly_uint64
20285 rather than an unsigned int.
20287 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
20288 Alan Hayward <alan.hayward@arm.com>
20289 David Sherwood <david.sherwood@arm.com>
20291 * target.def (get_mask_mode): Take the number of units and length
20292 as poly_uint64s rather than unsigned ints.
20293 * targhooks.h (default_get_mask_mode): Update accordingly.
20294 * targhooks.c (default_get_mask_mode): Likewise.
20295 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
20296 * doc/tm.texi: Regenerate.
20298 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
20299 Alan Hayward <alan.hayward@arm.com>
20300 David Sherwood <david.sherwood@arm.com>
20302 * omp-general.h (omp_max_vf): Return a poly_uint64 instead of an int.
20303 * omp-general.c (omp_max_vf): Likewise.
20304 * omp-expand.c (omp_adjust_chunk_size): Update call to omp_max_vf.
20305 (expand_omp_simd): Handle polynomial safelen.
20306 * omp-low.c (omplow_simd_context): Add a default constructor.
20307 (omplow_simd_context::max_vf): Change from int to poly_uint64.
20308 (lower_rec_simd_input_clauses): Update accordingly.
20309 (lower_rec_input_clauses): Likewise.
20311 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
20312 Alan Hayward <alan.hayward@arm.com>
20313 David Sherwood <david.sherwood@arm.com>
20315 * tree-vectorizer.h (vect_nunits_for_cost): New function.
20316 * tree-vect-loop.c (vect_model_reduction_cost): Use it.
20317 * tree-vect-slp.c (vect_analyze_slp_cost_1): Likewise.
20318 (vect_analyze_slp_cost): Likewise.
20319 * tree-vect-stmts.c (vect_model_store_cost): Likewise.
20320 (vect_model_load_cost): Likewise.
20322 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
20323 Alan Hayward <alan.hayward@arm.com>
20324 David Sherwood <david.sherwood@arm.com>
20326 * tree-vect-slp.c (vect_record_max_nunits, vect_build_slp_tree_1)
20327 (vect_build_slp_tree_2, vect_build_slp_tree): Change max_nunits
20328 from an unsigned int * to a poly_uint64_pod *.
20329 (calculate_unrolling_factor): New function.
20330 (vect_analyze_slp_instance): Use it. Track polynomial max_nunits.
20332 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
20333 Alan Hayward <alan.hayward@arm.com>
20334 David Sherwood <david.sherwood@arm.com>
20336 * tree-vectorizer.h (_slp_instance::unrolling_factor): Change
20337 from an unsigned int to a poly_uint64.
20338 (_loop_vec_info::slp_unrolling_factor): Likewise.
20339 (_loop_vec_info::vectorization_factor): Change from an int
20341 (MAX_VECTORIZATION_FACTOR): Bump from 64 to INT_MAX.
20342 (vect_get_num_vectors): New function.
20343 (vect_update_max_nunits, vect_vf_for_cost): Likewise.
20344 (vect_get_num_copies): Use vect_get_num_vectors.
20345 (vect_analyze_data_ref_dependences): Change max_vf from an int *
20346 to an unsigned int *.
20347 (vect_analyze_data_refs): Change min_vf from an int * to a
20349 (vect_transform_slp_perm_load): Take the vf as a poly_uint64 rather
20350 than an unsigned HOST_WIDE_INT.
20351 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr)
20352 (vect_analyze_data_ref_dependence): Change max_vf from an int *
20353 to an unsigned int *.
20354 (vect_analyze_data_ref_dependences): Likewise.
20355 (vect_compute_data_ref_alignment): Handle polynomial vf.
20356 (vect_enhance_data_refs_alignment): Likewise.
20357 (vect_prune_runtime_alias_test_list): Likewise.
20358 (vect_shift_permute_load_chain): Likewise.
20359 (vect_supportable_dr_alignment): Likewise.
20360 (dependence_distance_ge_vf): Take the vectorization factor as a
20361 poly_uint64 rather than an unsigned HOST_WIDE_INT.
20362 (vect_analyze_data_refs): Change min_vf from an int * to a
20364 * tree-vect-loop-manip.c (vect_gen_scalar_loop_niters): Take
20365 vfm1 as a poly_uint64 rather than an int. Make the same change
20366 for the returned bound_scalar.
20367 (vect_gen_vector_loop_niters): Handle polynomial vf.
20368 (vect_do_peeling): Likewise. Update call to
20369 vect_gen_scalar_loop_niters and handle polynomial bound_scalars.
20370 (vect_gen_vector_loop_niters_mult_vf): Assert that the vf must
20372 * tree-vect-loop.c (vect_determine_vectorization_factor)
20373 (vect_update_vf_for_slp, vect_analyze_loop_2): Handle polynomial vf.
20374 (vect_get_known_peeling_cost): Likewise.
20375 (vect_estimate_min_profitable_iters, vectorizable_reduction): Likewise.
20376 (vect_worthwhile_without_simd_p, vectorizable_induction): Likewise.
20377 (vect_transform_loop): Likewise. Use the lowest possible VF when
20378 updating the upper bounds of the loop.
20379 (vect_min_worthwhile_factor): Make static. Return an unsigned int
20380 rather than an int.
20381 * tree-vect-slp.c (vect_attempt_slp_rearrange_stmts): Cope with
20382 polynomial unroll factors.
20383 (vect_analyze_slp_cost_1, vect_analyze_slp_instance): Likewise.
20384 (vect_make_slp_decision): Likewise.
20385 (vect_supported_load_permutation_p): Likewise, and polynomial
20387 (vect_analyze_slp_cost): Handle polynomial vf.
20388 (vect_slp_analyze_node_operations): Likewise.
20389 (vect_slp_analyze_bb_1): Likewise.
20390 (vect_transform_slp_perm_load): Take the vf as a poly_uint64 rather
20391 than an unsigned HOST_WIDE_INT.
20392 * tree-vect-stmts.c (vectorizable_simd_clone_call, vectorizable_store)
20393 (vectorizable_load): Handle polynomial vf.
20394 * tree-vectorizer.c (simduid_to_vf::vf): Change from an int to
20396 (adjust_simduid_builtins, shrink_simd_arrays): Update accordingly.
20398 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
20399 Alan Hayward <alan.hayward@arm.com>
20400 David Sherwood <david.sherwood@arm.com>
20402 * match.pd: Handle bit operations involving three constants
20403 and try to fold one pair.
20405 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
20407 * tree-vect-loop-manip.c: Include gimple-fold.h.
20408 (slpeel_make_loop_iterate_ntimes): Add step, final_iv and
20409 niters_maybe_zero parameters. Handle other cases besides a step of 1.
20410 (vect_gen_vector_loop_niters): Add a step_vector_ptr parameter.
20411 Add a path that uses a step of VF instead of 1, but disable it
20413 (vect_do_peeling): Add step_vector, niters_vector_mult_vf_var
20414 and niters_no_overflow parameters. Update calls to
20415 slpeel_make_loop_iterate_ntimes and vect_gen_vector_loop_niters.
20416 Create a new SSA name if the latter choses to use a ste other
20417 than zero, and return it via niters_vector_mult_vf_var.
20418 * tree-vect-loop.c (vect_transform_loop): Update calls to
20419 vect_do_peeling, vect_gen_vector_loop_niters and
20420 slpeel_make_loop_iterate_ntimes.
20421 * tree-vectorizer.h (slpeel_make_loop_iterate_ntimes, vect_do_peeling)
20422 (vect_gen_vector_loop_niters): Update declarations after above changes.
20424 2018-01-02 Michael Meissner <meissner@linux.vnet.ibm.com>
20426 * config/rs6000/rs6000.md (floor<mode>2): Add support for IEEE
20427 128-bit round to integer instructions.
20428 (ceil<mode>2): Likewise.
20429 (btrunc<mode>2): Likewise.
20430 (round<mode>2): Likewise.
20432 2018-01-02 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
20434 * config/rs6000/rs6000-string.c (expand_block_move): Allow the use of
20435 unaligned VSX load/store on P8/P9.
20436 (expand_block_clear): Allow the use of unaligned VSX
20437 load/store on P8/P9.
20439 2018-01-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20441 * config/rs6000/rs6000-p8swap.c (swap_feeds_both_load_and_store):
20443 (rs6000_analyze_swaps): Mark a web unoptimizable if it contains a
20444 swap associated with both a load and a store.
20446 2018-01-02 Andrew Waterman <andrew@sifive.com>
20448 * config/riscv/linux.h (ICACHE_FLUSH_FUNC): New.
20449 * config/riscv/riscv.md (clear_cache): Use it.
20451 2018-01-02 Artyom Skrobov <tyomitch@gmail.com>
20453 * web.c: Remove out-of-date comment.
20455 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
20457 * expr.c (fixup_args_size_notes): Check that any existing
20458 REG_ARGS_SIZE notes are correct, and don't try to re-add them.
20459 (emit_single_push_insn_1): Move stack_pointer_delta adjustment to...
20460 (emit_single_push_insn): ...here.
20462 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
20464 * rtl.h (CONST_VECTOR_ELT): Redefine to const_vector_elt.
20465 (const_vector_encoded_nelts): New function.
20466 (CONST_VECTOR_NUNITS): Redefine to use GET_MODE_NUNITS.
20467 (const_vector_int_elt, const_vector_elt): Declare.
20468 * emit-rtl.c (const_vector_int_elt_1): New function.
20469 (const_vector_elt): Likewise.
20470 * simplify-rtx.c (simplify_immed_subreg): Avoid taking the address
20471 of CONST_VECTOR_ELT.
20473 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
20475 * expr.c: Include rtx-vector-builder.h.
20476 (const_vector_mask_from_tree): Use rtx_vector_builder and operate
20477 directly on the tree encoding.
20478 (const_vector_from_tree): Likewise.
20479 * optabs.c: Include rtx-vector-builder.h.
20480 (expand_vec_perm_var): Use rtx_vector_builder and create a repeating
20481 sequence of "u" values.
20482 * vec-perm-indices.c: Include rtx-vector-builder.h.
20483 (vec_perm_indices_to_rtx): Use rtx_vector_builder and operate
20484 directly on the vec_perm_indices encoding.
20486 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
20488 * doc/rtl.texi (const_vector): Describe new encoding scheme.
20489 * Makefile.in (OBJS): Add rtx-vector-builder.o.
20490 * rtx-vector-builder.h: New file.
20491 * rtx-vector-builder.c: Likewise.
20492 * rtl.h (rtx_def::u2): Add a const_vector field.
20493 (CONST_VECTOR_NPATTERNS): New macro.
20494 (CONST_VECTOR_NELTS_PER_PATTERN): Likewise.
20495 (CONST_VECTOR_DUPLICATE_P): Likewise.
20496 (CONST_VECTOR_STEPPED_P): Likewise.
20497 (CONST_VECTOR_ENCODED_ELT): Likewise.
20498 (const_vec_duplicate_p): Check for a duplicated vector encoding.
20499 (unwrap_const_vec_duplicate): Likewise.
20500 (const_vec_series_p): Check for a non-duplicated vector encoding.
20501 Say that the function only returns true for integer vectors.
20502 * emit-rtl.c: Include rtx-vector-builder.h.
20503 (gen_const_vec_duplicate_1): Delete.
20504 (gen_const_vector): Call gen_const_vec_duplicate instead of
20505 gen_const_vec_duplicate_1.
20506 (const_vec_series_p_1): Operate directly on the CONST_VECTOR encoding.
20507 (gen_const_vec_duplicate): Use rtx_vector_builder.
20508 (gen_const_vec_series): Likewise.
20509 (gen_rtx_CONST_VECTOR): Likewise.
20510 * config/powerpcspe/powerpcspe.c: Include rtx-vector-builder.h.
20511 (swap_const_vector_halves): Take an rtx pointer rather than rtx.
20512 Build a new vector rather than modifying a CONST_VECTOR in-place.
20513 (handle_special_swappables): Update call accordingly.
20514 * config/rs6000/rs6000-p8swap.c: Include rtx-vector-builder.h.
20515 (swap_const_vector_halves): Take an rtx pointer rather than rtx.
20516 Build a new vector rather than modifying a CONST_VECTOR in-place.
20517 (handle_special_swappables): Update call accordingly.
20519 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
20521 * simplify-rtx.c (simplify_const_binary_operation): Use
20522 CONST_VECTOR_ELT instead of XVECEXP.
20524 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
20526 * tree-cfg.c (verify_gimple_assign_ternary): Allow the size of
20527 the selector elements to be different from the data elements
20528 if the selector is a VECTOR_CST.
20529 * tree-vect-stmts.c (vect_gen_perm_mask_any): Use a vector of
20530 ssizetype for the selector.
20532 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
20534 * optabs.c (shift_amt_for_vec_perm_mask): Try using series_p
20535 before testing each element individually.
20536 * tree-vect-generic.c (lower_vec_perm): Likewise.
20538 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
20540 * selftest.h (selftest::vec_perm_indices_c_tests): Declare.
20541 * selftest-run-tests.c (selftest::run_tests): Call it.
20542 * vector-builder.h (vector_builder::operator ==): New function.
20543 (vector_builder::operator !=): Likewise.
20544 * vec-perm-indices.h (vec_perm_indices::series_p): Declare.
20545 (vec_perm_indices::all_from_input_p): New function.
20546 * vec-perm-indices.c (vec_perm_indices::series_p): Likewise.
20547 (test_vec_perm_12, selftest::vec_perm_indices_c_tests): Likewise.
20548 * fold-const.c (fold_ternary_loc): Use tree_to_vec_perm_builder
20549 instead of reading the VECTOR_CST directly. Detect whether both
20550 vector inputs are the same before constructing the vec_perm_indices,
20551 and update the number of inputs argument accordingly. Use the
20552 utility functions added above. Only construct sel2 if we need to.
20554 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
20556 * optabs.c (expand_vec_perm_var): Use an explicit encoding for
20557 the broadcast of the low byte.
20558 (expand_mult_highpart): Use an explicit encoding for the permutes.
20559 * optabs-query.c (can_mult_highpart_p): Likewise.
20560 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
20561 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
20562 (vectorizable_bswap): Likewise.
20563 * tree-vect-data-refs.c (vect_grouped_store_supported): Use an
20564 explicit encoding for the power-of-2 permutes.
20565 (vect_permute_store_chain): Likewise.
20566 (vect_grouped_load_supported): Likewise.
20567 (vect_permute_load_chain): Likewise.
20569 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
20571 * vec-perm-indices.h (vec_perm_indices_to_tree): Declare.
20572 * vec-perm-indices.c (vec_perm_indices_to_tree): New function.
20573 * tree-ssa-forwprop.c (simplify_vector_constructor): Use it.
20574 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
20575 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
20576 (vect_gen_perm_mask_any): Likewise.
20578 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
20580 * int-vector-builder.h: New file.
20581 * vec-perm-indices.h: Include int-vector-builder.h.
20582 (vec_perm_indices): Redefine as an int_vector_builder.
20583 (auto_vec_perm_indices): Delete.
20584 (vec_perm_builder): Redefine as a stand-alone class.
20585 (vec_perm_indices::vec_perm_indices): New function.
20586 (vec_perm_indices::clamp): Likewise.
20587 * vec-perm-indices.c: Include fold-const.h and tree-vector-builder.h.
20588 (vec_perm_indices::new_vector): New function.
20589 (vec_perm_indices::new_expanded_vector): Update for new
20590 vec_perm_indices class.
20591 (vec_perm_indices::rotate_inputs): New function.
20592 (vec_perm_indices::all_in_range_p): Operate directly on the
20593 encoded form, without computing elided elements.
20594 (tree_to_vec_perm_builder): Operate directly on the VECTOR_CST
20595 encoding. Update for new vec_perm_indices class.
20596 * optabs.c (expand_vec_perm_const): Create a vec_perm_indices for
20597 the given vec_perm_builder.
20598 (expand_vec_perm_var): Update vec_perm_builder constructor.
20599 (expand_mult_highpart): Use vec_perm_builder instead of
20600 auto_vec_perm_indices.
20601 * optabs-query.c (can_mult_highpart_p): Use vec_perm_builder and
20602 vec_perm_indices instead of auto_vec_perm_indices. Use a single
20603 or double series encoding as appropriate.
20604 * fold-const.c (fold_ternary_loc): Use vec_perm_builder and
20605 vec_perm_indices instead of auto_vec_perm_indices.
20606 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
20607 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
20608 (vect_permute_store_chain): Likewise.
20609 (vect_grouped_load_supported): Likewise.
20610 (vect_permute_load_chain): Likewise.
20611 (vect_shift_permute_load_chain): Likewise.
20612 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
20613 (vect_transform_slp_perm_load): Likewise.
20614 (vect_schedule_slp_instance): Likewise.
20615 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
20616 (vectorizable_mask_load_store): Likewise.
20617 (vectorizable_bswap): Likewise.
20618 (vectorizable_store): Likewise.
20619 (vectorizable_load): Likewise.
20620 * tree-vect-generic.c (lower_vec_perm): Use vec_perm_builder and
20621 vec_perm_indices instead of auto_vec_perm_indices. Use
20622 tree_to_vec_perm_builder to read the vector from a tree.
20623 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Take a
20624 vec_perm_builder instead of a vec_perm_indices.
20625 (have_whole_vector_shift): Use vec_perm_builder and
20626 vec_perm_indices instead of auto_vec_perm_indices. Leave the
20627 truncation to calc_vec_perm_mask_for_shift.
20628 (vect_create_epilog_for_reduction): Likewise.
20629 * config/aarch64/aarch64.c (expand_vec_perm_d::perm): Change
20630 from auto_vec_perm_indices to vec_perm_indices.
20631 (aarch64_expand_vec_perm_const_1): Use rotate_inputs on d.perm
20632 instead of changing individual elements.
20633 (aarch64_vectorize_vec_perm_const): Use new_vector to install
20634 the vector in d.perm.
20635 * config/arm/arm.c (expand_vec_perm_d::perm): Change
20636 from auto_vec_perm_indices to vec_perm_indices.
20637 (arm_expand_vec_perm_const_1): Use rotate_inputs on d.perm
20638 instead of changing individual elements.
20639 (arm_vectorize_vec_perm_const): Use new_vector to install
20640 the vector in d.perm.
20641 * config/powerpcspe/powerpcspe.c (rs6000_expand_extract_even):
20642 Update vec_perm_builder constructor.
20643 (rs6000_expand_interleave): Likewise.
20644 * config/rs6000/rs6000.c (rs6000_expand_extract_even): Likewise.
20645 (rs6000_expand_interleave): Likewise.
20647 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
20649 * optabs-query.c (can_vec_perm_var_p): Check whether lowering
20650 to qimode could truncate the indices.
20651 * optabs.c (expand_vec_perm_var): Likewise.
20653 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
20655 * Makefile.in (OBJS): Add vec-perm-indices.o.
20656 * vec-perm-indices.h: New file.
20657 * vec-perm-indices.c: Likewise.
20658 * target.h (vec_perm_indices): Replace with a forward class
20660 (auto_vec_perm_indices): Move to vec-perm-indices.h.
20661 * optabs.h: Include vec-perm-indices.h.
20662 (expand_vec_perm): Delete.
20663 (selector_fits_mode_p, expand_vec_perm_var): Declare.
20664 (expand_vec_perm_const): Declare.
20665 * target.def (vec_perm_const_ok): Replace with...
20666 (vec_perm_const): ...this new hook.
20667 * doc/tm.texi.in (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Replace with...
20668 (TARGET_VECTORIZE_VEC_PERM_CONST): ...this new hook.
20669 * doc/tm.texi: Regenerate.
20670 * optabs.def (vec_perm_const): Delete.
20671 * doc/md.texi (vec_perm_const): Likewise.
20672 (vec_perm): Refer to TARGET_VECTORIZE_VEC_PERM_CONST.
20673 * expr.c (expand_expr_real_2): Use expand_vec_perm_const rather than
20674 expand_vec_perm for constant permutation vectors. Assert that
20675 the mode of variable permutation vectors is the integer equivalent
20676 of the mode that is being permuted.
20677 * optabs-query.h (selector_fits_mode_p): Declare.
20678 * optabs-query.c: Include vec-perm-indices.h.
20679 (selector_fits_mode_p): New function.
20680 (can_vec_perm_const_p): Check whether targetm.vectorize.vec_perm_const
20681 is defined, instead of checking whether the vec_perm_const_optab
20682 exists. Use targetm.vectorize.vec_perm_const instead of
20683 targetm.vectorize.vec_perm_const_ok. Check whether the indices
20684 fit in the vector mode before using a variable permute.
20685 * optabs.c (shift_amt_for_vec_perm_mask): Take a mode and a
20686 vec_perm_indices instead of an rtx.
20687 (expand_vec_perm): Replace with...
20688 (expand_vec_perm_const): ...this new function. Take the selector
20689 as a vec_perm_indices rather than an rtx. Also take the mode of
20690 the selector. Update call to shift_amt_for_vec_perm_mask.
20691 Use targetm.vectorize.vec_perm_const instead of vec_perm_const_optab.
20692 Use vec_perm_indices::new_expanded_vector to expand the original
20693 selector into bytes. Check whether the indices fit in the vector
20694 mode before using a variable permute.
20695 (expand_vec_perm_var): Make global.
20696 (expand_mult_highpart): Use expand_vec_perm_const.
20697 * fold-const.c: Includes vec-perm-indices.h.
20698 * tree-ssa-forwprop.c: Likewise.
20699 * tree-vect-data-refs.c: Likewise.
20700 * tree-vect-generic.c: Likewise.
20701 * tree-vect-loop.c: Likewise.
20702 * tree-vect-slp.c: Likewise.
20703 * tree-vect-stmts.c: Likewise.
20704 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm_const):
20706 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>): Delete.
20707 * config/aarch64/aarch64.c (aarch64_expand_vec_perm_const)
20708 (aarch64_vectorize_vec_perm_const_ok): Fuse into...
20709 (aarch64_vectorize_vec_perm_const): ...this new function.
20710 (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
20711 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
20712 * config/arm/arm-protos.h (arm_expand_vec_perm_const): Delete.
20713 * config/arm/vec-common.md (vec_perm_const<mode>): Delete.
20714 * config/arm/arm.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
20715 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
20716 (arm_expand_vec_perm_const, arm_vectorize_vec_perm_const_ok): Merge
20718 (arm_vectorize_vec_perm_const): ...this new function. Explicitly
20719 check for NEON modes.
20720 * config/i386/i386-protos.h (ix86_expand_vec_perm_const): Delete.
20721 * config/i386/sse.md (VEC_PERM_CONST, vec_perm_const<mode>): Delete.
20722 * config/i386/i386.c (ix86_expand_vec_perm_const_1): Update comment.
20723 (ix86_expand_vec_perm_const, ix86_vectorize_vec_perm_const_ok): Merge
20725 (ix86_vectorize_vec_perm_const): ...this new function. Incorporate
20726 the old VEC_PERM_CONST conditions.
20727 * config/ia64/ia64-protos.h (ia64_expand_vec_perm_const): Delete.
20728 * config/ia64/vect.md (vec_perm_const<mode>): Delete.
20729 * config/ia64/ia64.c (ia64_expand_vec_perm_const)
20730 (ia64_vectorize_vec_perm_const_ok): Merge into...
20731 (ia64_vectorize_vec_perm_const): ...this new function.
20732 * config/mips/loongson.md (vec_perm_const<mode>): Delete.
20733 * config/mips/mips-msa.md (vec_perm_const<mode>): Delete.
20734 * config/mips/mips-ps-3d.md (vec_perm_constv2sf): Delete.
20735 * config/mips/mips-protos.h (mips_expand_vec_perm_const): Delete.
20736 * config/mips/mips.c (mips_expand_vec_perm_const)
20737 (mips_vectorize_vec_perm_const_ok): Merge into...
20738 (mips_vectorize_vec_perm_const): ...this new function.
20739 * config/powerpcspe/altivec.md (vec_perm_constv16qi): Delete.
20740 * config/powerpcspe/paired.md (vec_perm_constv2sf): Delete.
20741 * config/powerpcspe/spe.md (vec_perm_constv2si): Delete.
20742 * config/powerpcspe/vsx.md (vec_perm_const<mode>): Delete.
20743 * config/powerpcspe/powerpcspe-protos.h (altivec_expand_vec_perm_const)
20744 (rs6000_expand_vec_perm_const): Delete.
20745 * config/powerpcspe/powerpcspe.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK):
20747 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
20748 (altivec_expand_vec_perm_const_le): Take each operand individually.
20749 Operate on constant selectors rather than rtxes.
20750 (altivec_expand_vec_perm_const): Likewise. Update call to
20751 altivec_expand_vec_perm_const_le.
20752 (rs6000_expand_vec_perm_const): Delete.
20753 (rs6000_vectorize_vec_perm_const_ok): Delete.
20754 (rs6000_vectorize_vec_perm_const): New function.
20755 (rs6000_do_expand_vec_perm): Take a vec_perm_builder instead of
20756 an element count and rtx array.
20757 (rs6000_expand_extract_even): Update call accordingly.
20758 (rs6000_expand_interleave): Likewise.
20759 * config/rs6000/altivec.md (vec_perm_constv16qi): Delete.
20760 * config/rs6000/paired.md (vec_perm_constv2sf): Delete.
20761 * config/rs6000/vsx.md (vec_perm_const<mode>): Delete.
20762 * config/rs6000/rs6000-protos.h (altivec_expand_vec_perm_const)
20763 (rs6000_expand_vec_perm_const): Delete.
20764 * config/rs6000/rs6000.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
20765 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
20766 (altivec_expand_vec_perm_const_le): Take each operand individually.
20767 Operate on constant selectors rather than rtxes.
20768 (altivec_expand_vec_perm_const): Likewise. Update call to
20769 altivec_expand_vec_perm_const_le.
20770 (rs6000_expand_vec_perm_const): Delete.
20771 (rs6000_vectorize_vec_perm_const_ok): Delete.
20772 (rs6000_vectorize_vec_perm_const): New function. Remove stray
20773 reference to the SPE evmerge intructions.
20774 (rs6000_do_expand_vec_perm): Take a vec_perm_builder instead of
20775 an element count and rtx array.
20776 (rs6000_expand_extract_even): Update call accordingly.
20777 (rs6000_expand_interleave): Likewise.
20778 * config/sparc/sparc.md (vec_perm_constv8qi): Delete in favor of...
20779 * config/sparc/sparc.c (sparc_vectorize_vec_perm_const): ...this
20781 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
20783 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
20785 * optabs.c (expand_vec_perm_1): Assert that SEL has an integer
20786 vector mode and that that mode matches the mode of the data
20788 (expand_vec_perm): Split handling of non-CONST_VECTOR selectors
20789 out into expand_vec_perm_var. Do all CONST_VECTOR handling here,
20790 directly using expand_vec_perm_1 when forcing selectors into
20792 (expand_vec_perm_var): New function, split out from expand_vec_perm.
20794 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
20796 * optabs-query.h (can_vec_perm_p): Delete.
20797 (can_vec_perm_var_p, can_vec_perm_const_p): Declare.
20798 * optabs-query.c (can_vec_perm_p): Split into...
20799 (can_vec_perm_var_p, can_vec_perm_const_p): ...these two functions.
20800 (can_mult_highpart_p): Use can_vec_perm_const_p to test whether a
20801 particular selector is valid.
20802 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
20803 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
20804 (vect_grouped_load_supported): Likewise.
20805 (vect_shift_permute_load_chain): Likewise.
20806 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
20807 (vect_transform_slp_perm_load): Likewise.
20808 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
20809 (vectorizable_bswap): Likewise.
20810 (vect_gen_perm_mask_checked): Likewise.
20811 * fold-const.c (fold_ternary_loc): Likewise. Don't take
20812 implementations of variable permutation vectors into account
20813 when deciding which selector to use.
20814 * tree-vect-loop.c (have_whole_vector_shift): Don't check whether
20815 vec_perm_const_optab is supported; instead use can_vec_perm_const_p
20816 with a false third argument.
20817 * tree-vect-generic.c (lower_vec_perm): Use can_vec_perm_const_p
20818 to test whether the constant selector is valid and can_vec_perm_var_p
20819 to test whether a variable selector is valid.
20821 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
20823 * optabs-query.h (can_vec_perm_p): Take a const vec_perm_indices *.
20824 * optabs-query.c (can_vec_perm_p): Likewise.
20825 * fold-const.c (fold_vec_perm): Take a const vec_perm_indices &
20826 instead of vec_perm_indices.
20827 * tree-vectorizer.h (vect_gen_perm_mask_any): Likewise,
20828 (vect_gen_perm_mask_checked): Likewise,
20829 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise,
20830 (vect_gen_perm_mask_checked): Likewise,
20832 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
20834 * optabs-query.h (qimode_for_vec_perm): Declare.
20835 * optabs-query.c (can_vec_perm_p): Split out qimode search to...
20836 (qimode_for_vec_perm): ...this new function.
20837 * optabs.c (expand_vec_perm): Use qimode_for_vec_perm.
20839 2018-01-02 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
20841 * rtlanal.c (canonicalize_condition): Return 0 if final rtx
20842 does not have a conditional at the top.
20844 2018-01-02 Richard Biener <rguenther@suse.de>
20846 * ipa-inline.c (big_speedup_p): Fix expression.
20848 2018-01-02 Jan Hubicka <hubicka@ucw.cz>
20851 * config/i386/x86-tune-costs.h: Increase cost of integer load costs
20854 2018-01-02 Jan Hubicka <hubicka@ucw.cz>
20858 * x86-tune-costs.h (generic_cost): Reduce cost of FDIV 20->17,
20859 cost of sqrt 20->14, DIVSS 18->13, DIVSD 32->17, SQRtSS 30->14
20860 and SQRTsD 58->18, cond_not_taken_branch_cost. 2->1. Increase
20861 cond_taken_branch_cost 3->4.
20863 2018-01-01 Jakub Jelinek <jakub@redhat.com>
20865 PR tree-optimization/83581
20866 * tree-loop-distribution.c (pass_loop_distribution::execute): Return
20867 TODO_cleanup_cfg if any changes have been made.
20869 PR middle-end/83608
20870 * expr.c (store_expr_with_bounds): Use simplify_gen_subreg instead of
20871 convert_modes if target mode has the right side, but different mode
20874 PR middle-end/83609
20875 * expr.c (expand_assignment): Fix up a typo in simplify_gen_subreg
20876 last argument when extracting from CONCAT. If either from_real or
20877 from_imag is NULL, use expansion through memory. If result is not
20878 a CONCAT and simplify_gen_subreg fails, try to simplify_gen_subreg
20879 the parts directly to inner mode, if even that fails, use expansion
20882 PR middle-end/83623
20883 * expmed.c (expand_shift_1): For 2-byte rotates by BITS_PER_UNIT,
20884 check for bswap in mode rather than HImode and use that in expand_unop
20887 Copyright (C) 2018 Free Software Foundation, Inc.
20889 Copying and distribution of this file, with or without modification,
20890 are permitted in any medium without royalty provided the copyright
20891 notice and this notice are preserved.