1 2018-10-13 Eric Botcazou <ebotcazou@adacore.com>
3 * dwarf2cfi.c (struct dw_trace_info): Add args_size_defined_for_eh.
4 (notice_args_size): Set it in the current trace if no insn that can
5 throw internally has been seen yet.
6 (connect_traces): When connecting args_size between traces, allow the
7 incoming values not to match if there is an insn setting it before the
8 first insn that can throw internally; in that case, force the creation
9 of a CFI note on this latter insn.
11 2018-10-13 Jonathan Wakely <jwakely@redhat.com>
13 * opt-problem.h (opt_wrapper): Use template-argument-list when naming
14 the base class, because using the injected-class-name was not clearly
15 specified until DR 176.
17 2018-10-12 Paul Koning <ni1d@arrl.net>
19 * config/pdp11/pdp11.md (doloop_end): New expander.
20 (doloop_end_insn): renamed from "doloop_end".
21 (addqi3): New pattern.
22 (subqi3): New pattern.
23 * config/pdp11/predicates.md (incdec_operand): New predicate.
25 2018-10-12 Yury Gribov <tetra2005@gmail.com>
28 * real.c (format_helper::can_represent_integral_type_p): New function
29 * real.h (format_helper::can_represent_integral_type_p): Ditto.
30 * match.pd: New pattern.
32 2018-10-12 Alexandre Oliva <oliva@adacore.com>
34 * configure.ac: Introduce --enable-large-address-aware
35 to define MINGW_DEFAULT_LARGE_ADDR_AWARE.
36 * doc/install.texi: Document it.
37 * configure, config.in: Rebuilt.
38 * config/i386/mingw32.h (LINK_SPEC_LARGE_ADDR_AWARE): Define,
39 based on MINGW_DEFAULT_LARGE_ADDR_AWARE.
40 (LINK_SPEC): Insert it.
41 * config/i386/mingw-264.h: Likewise.
43 * cygming.h (HAVE_GAS_ALIGNED_COMM): Fallback-define.
45 2018-10-12 Peter Bergner <bergner@linux.ibm.com>
47 PR rtl-optimization/87600
48 * ira-lives (non_conflicting_reg_copy_p): Disable for non LRA targets.
50 2018-10-12 Paul Koning <ni1d@arrl.net>
52 * doc/md.texi (doloop_end): Document that the pattern code may
53 need to check operand mode.
55 2018-10-12 Wilco Dijkstra <wdijkstr@arm.com>
57 * config/aarch64/aarch64.md (zero_extendsidi2_aarch64): Add alternatives
58 to zero-extend between int and floating-point registers.
59 (load_pair_zero_extendsidi2_aarch64): Add alternative for zero-extended
60 ldp into floating-point registers. Add type and arch attributes.
61 (zero_extend<SHORT:mode><GPI:mode>2_aarch64): Add arch attribute.
62 Use f_loads for type attribute.
64 2018-10-11 Martin Sebor <msebor@redhat.com>
66 * doc/extend.texi (attribute packed): Correct typos.
68 2018-10-11 Martin Sebor <msebor@redhat.com>
70 * doc/extend.texi (attribute flatten): Mention interaction with
73 2018-10-11 Jan Hubicka <hubicka@ucw.cz>
76 * cgraphclones.c (cgraph_node::create_version_clone_with_body):
77 Set new_decl virtual flag to zero.
79 2018-10-11 Martin Sebor <msebor@redhat.com>
82 * doc/extend.texi (attribute format_arg): Discuss using multiple
83 attributes on a single function.
85 2018-10-11 Giuliano Belinassi <giuliano.belinassi@usp.br>
87 PR tree-optimization/86829
88 * match.pd (sin (atan (x))): New simplification rules.
89 (cos (atan (x))): Likewise.
90 * real.c (build_sinatan_real): New function.
91 * real.h (build_sinatan_real): Prototype.
93 2018-10-11 Will Schmidt <will_schmidt@vnet.ibm.com>
95 * config/rs6000/rs6000.c (map_to_integral_tree_type): New helper
97 (fold_mergeeo_helper): New helper function.
98 (rs6000_gimple_fold_builtin): Add hooks for vec_mergee and vec_mergeo
99 intrinsics. Correct some whitespace indentation issues.
101 2018-10-11 Wilco Dijkstra <wdijkstr@arm.com>
104 * config/aarch64/aarch64.c (aarch64_mask_and_shift_for_ubfiz_p):
105 Use HOST_WIDE_INT_1U for shift.
107 2018-10-11 Doug Rupp <rupp@adacore.com>
108 Olivier Hainque <hainque@adacore.com>
110 * config/rs6000/vxworks.h (VXWORKS_RELAX_LINK_SPEC): New macro.
111 Pass --relax to the linker for RTPs.
112 (LINK_SPEC): Append VXWORKS_RELAX_LINK_SPEC.
114 2018-10-11 Andrew Stubbs <ams@codesourcery.com>
115 Jan Hubicka <jh@suse.cz>
116 Martin Jambor <mjambor@suse.cz>
118 * print-rtl.c (print_rtx_operand_codes_E_and_V): Print how many times
119 the same elements are repeated rather than printing all of them.
120 * read-rtl.c (rtx_reader::read_rtx_operand): Recognize and expand
122 * read-rtl-function.c (test_loading_repeat): New function.
123 (read_rtl_function_c_tests): Call test_loading_repeat.
124 * rtl-tests.c (test_dumping_repeat): New function.
125 (rtl_tests_c_tests): Call test_dumping_repeat.
127 2018-10-11 Richard Biener <rguenther@suse.de>
129 * config/i386/x86-tune-costs.h (bdver?_memcpy, bdver?_memset,
130 bdver?_cost): Unify to ...
131 (bdver_memcpy, bdver_memset, bdver_cost): ... this.
132 * config/i386/i386.c (processor_cost_table): Adjust.
134 2018-10-10 Eric Botcazou <ebotcazou@adacore.com>
137 * cgraphunit.c (cgraph_node::expand_thunk): Force DECL_IGNORED_P on
138 the thunk when expanding to GIMPLE.
140 2018-10-10 Bernd Edlinger <bernd.edlinger@hotmail.de>
142 * varasm.c (mergeable_string_section): Don't try to move zero-length
143 strings to the merge section.
145 2018-10-10 Uros Bizjak <ubizjak@gmail.com>
148 * config/i386/mmx.md (const_vector 0 -> mem splitter): New splitter.
150 2018-10-10 Jakub Jelinek <jakub@redhat.com>
153 * config/i386/i386-builtin.def (IX86_BUILTIN_RDPMC): Move from args set
156 2018-10-10 Richard Biener <rguenther@suse.de>
158 * config/i386/sse.md (reduc_plus_scal_v8df, reduc_plus_scal_v4df,
159 reduc_plus_scal_v2df, reduc_plus_scal_v16sf, reduc_plus_scal_v8sf,
160 reduc_plus_scal_v4sf): Merge into pattern reducing to half width
161 and recursing and pattern terminating the recursion on SSE
162 vector width using ix86_expand_reduc.
163 (reduc_sminmax_scal_<mode>): Split into part reducing to half
164 width and recursing and SSE2 vector variant doing the final
165 reduction with ix86_expand_reduc.
166 (reduc_uminmax_scal_<mode>): Likewise for the AVX512 variants
167 with terminating the recursion at AVX level, splitting that
170 2018-10-09 David Malcolm <dmalcolm@redhat.com>
172 * genmatch.c (error_cb): Rename to...
173 (diagnostic_cb): ...this, converting int params to enums.
174 (fatal_at): Update for renaming.
175 (warning_at): Likewise.
177 * input.c (selftest::ebcdic_execution_charset::apply):
178 Update for renaming of...
179 (selftest::ebcdic_execution_charset::on_error): ...this, renaming
181 (selftest::ebcdic_execution_charset::on_diagnostic): ...this,
182 converting level and reason to enums.
183 (class selftest::lexer_error_sink): Rename to...
184 (class selftest::lexer_test_options): ...this, renaming field
185 "m_errors" to "m_diagnostics".
186 (selftest::lexer_test_options::apply): Update for renaming of...
187 (selftest::lexer_test_options::on_error): ...this, renaming to...
188 (selftest::lexer_test_options::on_diagnostic): ...this
189 converting level and reason to enums.
190 (selftest::test_lexer_string_locations_raw_string_unterminated):
191 Update for renamings.
192 * opth-gen.awk (struct cpp_reason_option_codes_t): Use enum for
195 2018-10-09 Paul A. Clarke <pc@us.ibm.com>
197 * config.gcc (powerpc*-*-*): Add pmmintrin.h to extra_headers.
198 * config/rs6000/pmmintrin.h: New file.
200 2018-10-09 Eric Botcazou <ebotcazou@adacore.com>
202 PR tree-optimization/86659
203 * gimple-match.h (gimple_match_op constructors): Initialize reverse.
205 2018-10-09 Richard Biener <rguenther@suse.de>
207 PR tree-optimization/63155
208 * tree-ssa-structalias.c: Include tree-ssa.h.
209 (get_constraint_for_ssa_var): For undefs return nothing_id.
210 (find_func_aliases): Cleanup PHI handling.
212 2018-10-09 Richard Biener <rguenther@suse.de>
214 * tree-vectorizer.c (dump_stmt_cost): Fix cut&paste missing
217 2018-10-09 Martin Liska <mliska@suse.cz>
219 * asan.c (asan_emit_stack_protection): If a stack variable
220 is located in a same file as current function, then emit
221 line info into variable definition string.
223 2018-10-08 Eric Botcazou <ebotcazou@adacore.com>
225 * print-rtl.c (rtx_writer::print_rtx_operand_code_i): Print column
228 2018-10-08 Eric Botcazou <ebotcazou@adacore.com>
230 * cgraphunit.c (cgraph_node::expand_thunk): Do not force DECL_IGNORED_P
233 2018-10-08 Eric Botcazou <ebotcazou@adacore.com>
235 PR tree-optimization/86659
236 * gimple-match.h (struct gimple_match_op): Add reverse field.
237 (gimple_match_op::set_op): New overloaded method.
238 * gimple-match-head.c (maybe_build_generic_op) <BIT_FIELD_REF>: Set
239 the REF_REVERSE_STORAGE_ORDER flag on the value.
240 (gimple_simplify) <GIMPLE_ASSIGN>: For BIT_FIELD_REF, propagate the
241 REF_REVERSE_STORAGE_ORDER flag and avoid simplifying if it is set.
243 2018-10-08 Richard Sandiford <richard.sandiford@arm.com>
246 * gimple-ssa-backprop.c (backprop::intersect_uses): Use
247 FOR_EACH_IMM_USE_FAST instead of FOR_EACH_IMM_USE_STMT.
249 2018-10-08 H.J. Lu <hongjiu.lu@intel.com>
252 * config/i386/avx512fintrin.h (_mm512_mask_fmaddsub_round_pd):
253 Defined with __builtin_ia32_vfmaddsubpd512_mask.
255 2018-10-08 Richard Biener <rguenther@suse.de>
257 * config/i386/x86-tune-costs.h (znver1_cost): Make AVX256 vector loads
258 cost the same as AVX128 ones.
260 2018-10-08 Paul Koning <ni1d@arrl.net>
262 * config/pdp11/pdp11-protos.h (output_block_move): Remove.
263 (expand_block_move): New function.
264 * config/pdp11/pdp11.c (output_block_move): Remove.
265 (expand_block_move): New function.
266 * config/pdp11/pdp11.h (MOVE_RATIO): New definition.
267 * config/pdp11/pdp11.md (movmemhi): Use expand_block_move.
268 (*movmemhi1): Remove.
270 2018-10-08 Robin Dapp <rdapp@linux.ibm.com>
272 * config/s390/2827.md: Increase latencies for some FP instructions.
274 2018-10-08 Richard Biener <rguenther@suse.de>
276 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost):
278 * tree-vectorizer.c (dump_stmt_cost): Add cost param and dump it.
279 * tree-vectorizer.h (dump_stmt_cost): Adjust.
280 (add_stmt_cost): Dump return value of the hook.
282 2018-10-08 Richard Biener <rguenther@suse.de>
284 PR tree-optimization/63155
285 * tree-ssa-propagate.c (add_ssa_edge): Do cheap check first.
286 (ssa_propagation_engine::ssa_propagate): Remove redundant
289 2018-10-05 Peter Bergner <bergner@linux.ibm.com>
291 PR rtl-optimization/86939
292 PR rtl-optimization/87479
293 * ira.h (non_conflicting_reg_copy_p): New prototype.
294 * ira-lives.c (ignore_reg_for_conflicts): New static variable.
295 (make_hard_regno_dead): Don't add conflicts for register
296 ignore_reg_for_conflicts.
297 (make_object_dead): Likewise.
298 (non_conflicting_reg_copy_p): New function.
299 (process_bb_node_lives): Set ignore_reg_for_conflicts for copies.
300 Remove special conflict handling of REAL_PIC_OFFSET_TABLE_REGNUM.
301 * lra-lives.c (ignore_reg_for_conflicts): New static variable.
302 (make_hard_regno_dead): Don't add conflicts for register
303 ignore_reg_for_conflicts. Remove special conflict handling of
304 REAL_PIC_OFFSET_TABLE_REGNUM. Remove now unused argument
305 check_pic_pseudo_p and update callers.
306 (mark_pseudo_dead): Don't add conflicts for register
307 ignore_reg_for_conflicts.
308 (process_bb_lives): Set ignore_reg_for_conflicts for copies.
310 2018-10-05 Andrew Waterman <andrew@sifive.com>
311 Jim Wilson <jimw@sifive.com>
313 * config/riscv/riscv.md (f<quiet_pattern>_quiet<ANYF:mode><X:mode>4):
314 Add define_expand. Add ! HONOR_SNANS check to current pattern. Add
315 new pattern using HONOR_SNANS that emits one extra instruction.
317 2018-10-05 Segher Boessenkool <segher@kernel.crashing.org>
319 * config/rs6000/rs6000.md (unnamed mfcr scc_comparison_operator
320 patterns): Merge SI and DI patterns to a GPR pattern.
321 (unnamed define_insn and define_split for record form of that): Merge
322 to a single define_insn_and_split pattern.
324 2018-10-05 David Malcolm <dmalcolm@redhat.com>
327 * input.c (expand_location_to_spelling_point): Add param "aspect"
328 and use rather than hardcoding LOCATION_ASPECT_CARET.
329 (get_substring_ranges_for_loc): Handle the case of a single token
330 within a macro expansion.
331 * input.h (expand_location_to_spelling_point): Add "aspect" param,
332 defaulting to LOCATION_ASPECT_CARET.
334 2018-10-05 Paul Koning <ni1d@arrl.net>
336 * config/pdp11/pdp11.c (TARGET_CXX_GUARD_TYPE): Define.
337 (TARGET_CXX_CLASS_DATA_ALWAYS_COMDAT): Define.
338 (TARGET_CXX_LIBRARY_RTTI_COMDAT): Define.
339 (pdp11_guard_type): New function.
341 2018-10-05 Paul Koning <ni1d@arrl.net>
343 * config/pdp11/pdp11.h (FLOAT_TYPE_SIZE): Always 32.
344 * config/pdp11/pdp11.opt (mfloat32): Remove.
346 * doc/invoke.texi (pdp11 -mfloat32): Remove:
347 (pdp11 -mfloat64): Remove.
349 2018-10-05 Uros Bizjak <ubizjak@gmail.com>
351 * config/i386/i386.md (*cmpxf_cc_i387): Remove pattern.
352 (*cmp<mode>_cc_i387): Ditto.
353 (*cmpu<mode>_cc_i387): Ditto.
354 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
355 * config/i386/i386.c (ix86_expand_fp_compare): Remove
357 <case IX86_FPCMP_SAHF>: Do not generate pattern with HImode clobber.
358 Emit x86_sahf_1 pattern.
359 (ix86_expand_compare): Update call to ix86_expand_fp_compare.
360 (ix86_expand_carry_flag_compare): Ditto.
362 2018-10-05 Uros Bizjak <ubizjak@gmail.com>
364 * config/i386/i386.md (*cmpxf_i387): Change operand 2 predicate
365 to reg_or_0_operand. Add "C" constraint.
366 (*cmpxf_cc_i387): Ditto.
367 (*cmp<mode>_i387): Change operand 2 predicate
368 to nonimm_or_0_operand. Add "C" constraint.
369 (*cmp<mode>_cc_i387): Ditto.
370 (*cmp<mode>_0_i387): Remove insn pattern.
371 (*cmp<mode>_0_cc_i387): Ditto.
373 2018-10-05 Uros Bizjak <ubizjak@gmail.com>
375 * config/i386/constraints.md ("C"): Do not depend on TARGET_SSE.
376 * config/i386/predicates.md (nonimm_or_0_operand): Rename
377 from vector_move_operand. Update all uses.
379 2018-10-05 Martin Sebor <msebor@redhat.com>
381 PR tree-optimization/87490
382 * builtins.c (expand_builtin_strnlen): Handle a null data.decl
385 2018-10-05 Richard Biener <rguenther@suse.de>
387 PR tree-optimization/63155
388 * tree-ssa-ccp.c (ccp_propagate::visit_phi): Avoid excess
389 vertical space in dumpfiles.
390 * tree-ssa-propagate.h
391 (ssa_propagation_engine::process_ssa_edge_worklist): Remove.
392 * tree-ssa-propagate.c (cfg_blocks_back): New global.
393 (ssa_edge_worklist_back): Likewise.
394 (curr_order): Likewise.
395 (cfg_blocks_get): Remove abstraction.
396 (cfg_blocks_add): Likewise.
397 (cfg_blocks_empty_p): Likewise.
398 (add_ssa_edge): Add to current or next worklist based on
400 (add_control_edge): Likewise.
401 (ssa_propagation_engine::process_ssa_edge_worklist): Fold
403 (ssa_propagation_engine::ssa_propagate): ... here. Unify
404 iteration from CFG and SSA edge worklist so we process
405 everything in RPO order, prioritizing forward progress
407 (ssa_prop_init): Allocate new worklists, do not dump
409 (ssa_prop_fini): Free new worklists.
411 2018-10-05 Richard Biener <rguenther@suse.de>
413 * tree-core.h (tree_block::abstract_flag): Remove.
414 (tree_block::block_num): Make full 32bits.
415 * tree.def (BLOCK): Remove docs about BLOCK_ABSTRACT.
416 * tree.h (BLOCK_ABSTRACT): Remove.
417 * dwarf2out.c (gen_lexical_block_die): Remove dead code
418 resulting from BLOCK_ABSTRACT being always false.
419 (gen_inlined_subroutine_die): Likewise.
420 (gen_block_die): Likewise.
421 * tree.c (block_ultimate_origin): Likewise.
422 * tree-pretty-print.c (dump_block_node): Remove code dealing
424 * tree-ssa-live.c (dump_scope_block): Likewise.
425 * tree-streamer-in.c (unpack_ts_block_value_fields): Likewise.
426 * tree-streamer-out.c (pack_ts_block_value_fields): Likewise.
428 2018-10-05 Richard Biener <rguenther@suse.de>
430 * config/i386/i386.c (ix86_add_stmt_cost): When scalar cost
431 is asked for initialize mode to the component mode of the
434 2018-10-05 H.J. Lu <hongjiu.lu@intel.com>
437 * config/i386/gnu-user.h (ASM_SPEC): Don't pass -msse2avx to
439 * config/i386/gnu-user64.h (ASM_SPEC): Likewise.
441 2018-10-05 Segher Boessenkool <segher@kernel.crashing.org>
444 * config/rs6000/rs6000-builtin.def (RS6000_BUILTIN_SET_FPSCR_DRN): Use
446 * config/rs6000/rs6000.md (rs6000_set_fpscr_rn): Require the operand
447 to be DImode. When using mffscrn, force the operand to a register.
449 2018-10-04 Uros Bizjak <ubizjak@gmail.com>
451 * config/i386/i386.md (*fop_<X87MODEF:mode>_2_i387): Macroize insn
452 from *fop_<MODEF:mode>_2_i387 and *fop_xf_2_i387 using
453 X87MODEF mode iterator.
454 (*fop_<X87MODEF:mode>_3_i387): Macroize insn from
455 *fop_<MODEF:mode>_3_i387 and *fop_xf_3_i387 using
456 X87MODEF mode iterator.
458 2018-10-04 Vinay Kumar <vinay.kumar@blackfigtech.com>
460 * doc/invoke.texi (-Wno-prio-ctor-dtor): Document new warning
463 2018-10-04 David Malcolm <dmalcolm@redhat.com>
465 * Makefile.in (OBJS): Add opt-problem.o.
466 * dump-context.h: Include "selftest.h.
467 (selftest::temp_dump_context): New forward decl.
468 (class dump_context): Make friend of class
469 selftest::temp_dump_context.
470 (dump_context::dump_loc_immediate): New decl.
471 (class dump_pretty_printer): Move here from dumpfile.c.
472 (class temp_dump_context): Move to namespace selftest.
473 (temp_dump_context::temp_dump_context): Add param
474 "forcibly_enable_dumping".
475 (selftest::verify_dumped_text):
476 (ASSERT_DUMPED_TEXT_EQ): Move here from dumpfile.c.
477 (selftest::verify_item):
478 (ASSERT_IS_TEXT): Move here from dumpfile.c.
479 (ASSERT_IS_TREE): Likewise.
480 (ASSERT_IS_GIMPLE): Likewise.
481 * dumpfile.c (dump_context::dump_loc): Move immediate dumping
483 (dump_context::dump_loc_immediate): ...this new function.
484 (class dump_pretty_printer): Move to dump-context.h.
485 (dump_switch_p_1): Don't enable MSG_PRIORITY_REEMITTED.
486 (opt_info_switch_p_1): Enable MSG_PRIORITY_REEMITTED.
487 (temp_dump_context::temp_dump_context): Move to "selftest"
488 namespace. Add param "forcibly_enable_dumping", and use it to
489 conditionalize the use of m_pp;
490 (selftest::verify_dumped_text): Make non-static.
491 (ASSERT_DUMPED_TEXT_EQ): Move to dump-context.h.
492 (selftest::verify_item): Make non-static.
493 (ASSERT_IS_TEXT): Move to dump-context.h.
494 (ASSERT_IS_TREE): Likewise.
495 (ASSERT_IS_GIMPLE): Likewise.
496 (selftest::test_capture_of_dump_calls): Pass "true" for new
497 param of temp_dump_context.
498 * dumpfile.h (enum dump_flag): Add MSG_PRIORITY_REEMITTED, adding
499 it to MSG_ALL_PRIORITIES. Update values of TDF_COMPARE_DEBUG and
501 * opt-problem.cc: New file.
502 * opt-problem.h: New file.
503 * optinfo-emit-json.cc
504 (selftest::test_building_json_from_dump_calls): Pass "true" for
505 new param of temp_dump_context.
506 * optinfo.cc (optinfo_kind_to_dump_flag): New function.
507 (optinfo::emit_for_opt_problem): New function.
508 (optinfo::emit): Clarity which emit_item is used.
509 * optinfo.h (optinfo::get_dump_location): New accessor.
510 (optinfo::emit_for_opt_problem): New decl.
511 (optinfo::emit): Make const.
512 * selftest-run-tests.c (selftest::run_tests): Call
513 selftest::opt_problem_cc_tests.
514 * selftest.h (selftest::opt_problem_cc_tests): New decl.
515 * tree-data-ref.c (dr_analyze_innermost): Convert return type from
516 bool to opt_result, converting fprintf messages to
517 opt_result::failure_at calls. Add "stmt" param for use by the
519 (create_data_ref): Pass "stmt" to the dr_analyze_innermost call.
520 (runtime_alias_check_p): Convert return type from bool to
521 opt_result, converting dump_printf calls to
522 opt_result::failure_at, using the statement DDR_A for their
524 (find_data_references_in_stmt): Convert return type from bool to
525 opt_result, converting "return false" to opt_result::failure_at
527 * tree-data-ref.h: Include "opt-problem.h".
528 (dr_analyze_innermost): Convert return type from bool to opt_result,
529 and add a const gimple * param.
530 (find_data_references_in_stmt): Convert return type from bool to
532 (runtime_alias_check_p): Likewise.
533 * tree-predcom.c (find_looparound_phi): Pass "init_stmt" to
534 dr_analyze_innermost.
535 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test):
536 Convert return type from bool to opt_result, adding a message for
537 the PARAM_VECT_MAX_VERSION_FOR_ALIAS_CHECKS zero case.
538 (vect_analyze_data_ref_dependence): Convert return type from bool
539 to opt_result. Change sense of return type from "false"
540 effectively meaning "no problems" to "false" meaning a problem,
541 so that "return false" becomes "return opt_result::success".
542 Convert "return true" calls to opt_result::failure_at, using
543 the location of statement A rather than vect_location.
544 (vect_analyze_data_ref_dependences): Convert return type from bool
546 (verify_data_ref_alignment): Likewise, converting dump_printf_loc
547 calls to opt_result::failure_at, using the stmt location rather
549 (vect_verify_datarefs_alignment): Convert return type from bool
551 (vect_enhance_data_refs_alignment): Likewise. Split local "stat"
552 into multiple more-tightly-scoped copies.
553 (vect_analyze_data_refs_alignment): Convert return type from bool
555 (vect_analyze_data_ref_accesses): Likewise, converting a
556 "return false" to a "return opt_result::failure_at", adding a
558 (vect_prune_runtime_alias_test_list): Convert return type from
559 bool to opt_result, converting dump_printf_loc to
560 opt_result::failure_at. Add a %G to show the pertinent statement,
561 and use the stmt's location rather than vect_location.
562 (vect_find_stmt_data_reference): Convert return type from
563 bool to opt_result, converting dump_printf_loc to
564 opt_result::failure_at, using stmt's location.
565 (vect_analyze_data_refs): Convert return type from bool to
566 opt_result. Convert "return false" to "return
567 opt_result::failure_at", adding messages as needed.
568 * tree-vect-loop.c (vect_determine_vf_for_stmt_1): Convert return
569 type from bool to opt_result.
570 (vect_determine_vf_for_stmt): Likewise.
571 (vect_determine_vectorization_factor): Likewise, converting
572 dump_printf_loc to opt_result::failure_at, using location of phi
573 rather than vect_location.
574 (vect_analyze_loop_form_1): Convert return type from bool to
575 opt_result, converting dump_printf_loc calls, retaining the use of
577 (vect_analyze_loop_form): Convert return type from loop_vec_info
578 to opt_loop_vec_info.
579 (vect_analyze_loop_operations): Convert return type from bool to
580 opt_result, converting dump_printf_loc calls, using the location
581 of phi/stmt rather than vect_location where available. Convert
582 various "return false" to "return opt_result::failure_at" with
583 "unsupported phi" messages.
584 (vect_get_datarefs_in_loop): Convert return type from bool to
585 opt_result. Add a message for the
586 PARAM_LOOP_MAX_DATAREFS_FOR_DATADEPS failure.
587 (vect_analyze_loop_2): Convert return type from bool to
588 opt_result. Ensure "ok" is set to a opt_result::failure_at before
589 each "goto again;", adding new messages where needed.
590 Add "unsupported grouped {store|load}" messages.
591 (vect_analyze_loop): Convert return type from loop_vec_info to
593 * tree-vect-slp.c (vect_analyze_slp): Convert return type from
595 * tree-vect-stmts.c (process_use): Likewise, converting
596 dump_printf_loc call and using stmt location, rather than
598 (vect_mark_stmts_to_be_vectorized): Likeise.
599 (vect_analyze_stmt): Likewise, adding a %G.
600 (vect_get_vector_types_for_stmt): Convert return type from bool to
601 opt_result, converting dump_printf_loc calls and using stmt
602 location, rather than vect_location.
603 (vect_get_mask_type_for_stmt): Convert return type from tree to
604 opt_tree, converting dump_printf_loc calls and using stmt location.
605 * tree-vectorizer.c: Include "opt-problem.h.
606 (try_vectorize_loop_1): Flag "Analyzing loop at" dump message as
607 MSG_PRIORITY_INTERNALS. Convert local "loop_vinfo" from
608 loop_vec_info to opt_loop_vec_info. If if fails, and dumping is
609 enabled, use it to report at the top level "couldn't vectorize
610 loop" followed by the problem.
611 * tree-vectorizer.h (opt_loop_vec_info): New typedef.
612 (vect_mark_stmts_to_be_vectorized): Convert return type from bool
614 (vect_analyze_stmt): Likewise.
615 (vect_get_vector_types_for_stmt): Likewise.
616 (tree vect_get_mask_type_for_stmt): Likewise.
617 (vect_analyze_data_ref_dependences): Likewise.
618 (vect_enhance_data_refs_alignment): Likewise.
619 (vect_analyze_data_refs_alignment): Likewise.
620 (vect_verify_datarefs_alignment): Likewise.
621 (vect_analyze_data_ref_accesses): Likewise.
622 (vect_prune_runtime_alias_test_list): Likewise.
623 (vect_find_stmt_data_reference): Likewise.
624 (vect_analyze_data_refs): Likewise.
625 (vect_analyze_loop): Convert return type from loop_vec_info to
627 (vect_analyze_loop_form): Likewise.
628 (vect_analyze_slp): Convert return type from bool to opt_result.
630 2018-10-04 David Malcolm <dmalcolm@redhat.com>
632 * doc/invoke.texi (-fopt-info): Document new "internals"
634 * dump-context.h (dump_context::apply_dump_filter_p): New decl.
635 * dumpfile.c (dump_options): Update for renaming of MSG_ALL to
637 (optinfo_verbosity_options): Add "internals".
638 (kind_as_string): Update for renaming of MSG_ALL to MSG_ALL_KINDS.
639 (dump_context::apply_dump_filter_p): New member function.
640 (dump_context::dump_loc): Use apply_dump_filter_p rather than
641 explicitly masking the dump_kind.
642 (dump_context::begin_scope): Increment the scope depth first. Use
643 apply_dump_filter_p rather than explicitly masking the dump_kind.
644 (dump_context::emit_item): Use apply_dump_filter_p rather than
645 explicitly masking the dump_kind.
646 (dump_dec): Likewise.
647 (dump_hex): Likewise.
648 (dump_switch_p_1): Default to MSG_ALL_PRIORITIES.
649 (opt_info_switch_p_1): Default to MSG_PRIORITY_USER_FACING.
650 (opt_info_switch_p): Update handling of default
651 MSG_OPTIMIZED_LOCATIONS to cope with default of
652 MSG_PRIORITY_USER_FACING.
653 (dump_basic_block): Use apply_dump_filter_p rather than explicitly
654 masking the dump_kind.
655 (selftest::test_capture_of_dump_calls): Update test_dump_context
656 instances to use MSG_ALL_KINDS | MSG_PRIORITY_USER_FACING rather
657 than MSG_ALL. Generalize scope test to be run at all four
658 combinations of with/without MSG_PRIORITY_USER_FACING and
659 MSG_PRIORITY_INTERNALS, adding examples of explicit priority
660 for each of the two values.
661 * dumpfile.h (enum dump_flag): Add comment about the MSG_* flags.
662 Rename MSG_ALL to MSG_ALL_KINDS. Add MSG_PRIORITY_USER_FACING,
663 MSG_PRIORITY_INTERNALS, and MSG_ALL_PRIORITIES, updating the
664 values for TDF_COMPARE_DEBUG and TDF_ALL_VALUES.
665 (AUTO_DUMP_SCOPE): Add a note to the comment about the interaction
667 * tree-vect-loop-manip.c (vect_loop_versioning): Mark versioning
668 dump messages as MSG_PRIORITY_USER_FACING.
669 * tree-vectorizer.h (DUMP_VECT_SCOPE): Add a note to the comment
670 about the interaction with MSG_PRIORITY_*.
672 2018-10-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
674 * varasm.c (output_constant): Add new parameter merge_strings.
675 Make strings properly zero terminated in merge string sections.
676 (mergeable_string_section): Don't fail if the last char is non-zero.
677 (assemble_variable_contents): Handle merge string sections.
678 (assemble_variable): Likewise.
679 (assemble_constant_contents): Likewise.
680 (output_constant_def_contents): Likewise.
681 (output_constructor_array_range,
682 output_constructor_regular_field): Adjust call to output_constant.
683 (output_object_block): Adjust call to assemble_constant_contents
684 and assemble_variable_contents.
686 2018-10-04 Martin Liska <mliska@suse.cz>
689 * cgraphunit.c (process_function_and_variable_attributes):
690 Warn about a function with alias attribute and a body.
692 2018-10-04 Martin Liska <mliska@suse.cz>
695 * multiple_target.c (redirect_to_specific_clone): New function.
696 (ipa_target_clone): Use it.
697 * tree-inline.c: Fix comment.
699 2018-10-04 David Malcolm <dmalcolm@redhat.com>
701 * dumpfile.c (gcc::dump_manager::dump_manager): Initialize new
703 (gcc::dump_manager::~dump_manager): Free m_optinfo_filename.
704 (gcc::dump_manager::register_pass): New member function, adapted
705 from loop body in gcc::pass_manager::register_pass, adding a
706 call to update_dfi_for_opt_info.
707 (gcc::dump_manager::opt_info_enable_passes): Store the
708 -fopt-info options into the new fields. Move the loop
710 (gcc::dump_manager::update_dfi_for_opt_info): ...this new member
712 * dumpfile.h (struct opt_pass): New forward decl.
713 (gcc::dump_manager::register_pass): New decl.
714 (gcc::dump_manager::update_dfi_for_opt_info): New decl.
715 (class gcc::dump_manager): Add fields "m_optgroup_flags",
716 "m_optinfo_flags", and "m_optinfo_filename".
717 * passes.c (gcc::pass_manager::register_pass): Move all of the
718 dump-handling code to gcc::dump_manager::register_pass.
720 2018-10-04 Peter Bergner <bergner@linux.ibm.com>
722 PR rtl-optimization/87466
723 * target.def (setjmp_preserves_nonvolatile_regs_p): New target hook.
724 * doc/tm.texi.in (TARGET_SETJMP_PRESERVES_NONVOLATILE_REGS_P): New hook.
725 * doc/tm.texi: Regenerate.
726 * ira-lives.c (process_bb_node_lives): Use the new target hook.
727 * lra-lives.c (process_bb_lives): Likewise.
728 * config/rs6000/rs6000.c (TARGET_SETJMP_PRESERVES_NONVOLATILE_REGS_P):
731 2018-10-04 Tamar Christina <tamar.christina@arm.com>
733 * params.c (add_params): Fix initialization.
735 2018-10-04 Martin Liska <mliska@suse.cz>
737 PR gcov-profile/84107
738 * tree-profile.c (init_ic_make_global_vars):
739 Remove ic_void_ptr_var and ic_gcov_type_ptr_var.
740 Come up with new ic_tuple* variables. Emit
741 __gcov_indirect_call{,_topn} variables.
742 (gimple_gen_ic_profiler): Access the variable
744 (gimple_gen_ic_func_profiler): Access
745 __gcov_indirect_call.callee field.
746 (gimple_init_gcov_profiler): Use ptr_type_node.
747 * value-prof.c (gimple_ic): Use ptr_type_node.
749 2018-10-04 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
751 PR tree-optimization/85787
752 * ipa-pure-const.c (malloc_candidate_p_1): Move most of malloc_candidate_p
753 into this function and add support for detecting multiple phis.
754 (DUMP_AND_RETURN): Move from malloc_candidate_p into top-level macro.
756 2018-10-04 Martin Liska <mliska@suse.cz>
759 * ipa-inline.c (inline_to_all_callers_1):
760 Call ultimate_alias_target for node being inlined.
762 2018-10-03 Jeff Law <law@redhat.com>
764 * gimple-ssa-sprintf.c (format_string): Do not hardcode size of
766 * tree.c (get_typenode_from_name): Moved from fortran/trans-types.c.
767 * tree.h (get_typenode_from_name): Prototype.
769 2018-10-03 Uros Bizjak <ubizjak@gmail.com>
771 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
772 Change operand 2 predicate to nonimmediate_operand.
773 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
775 2018-10-03 Martin Sebor <msebor@redhat.com>
776 Jeff Law <law@redhat.com>
778 * gimple-ssa-sprintf.c (struct fmtresult): Add new member and
780 (get_string_length): Detect unterminated arrays.
781 (format_string): Same.
782 (format_directive): Warn about unterminated arrays.
783 (handle_gimple_call): Mark statements with no_warning as needed.
785 2018-10-03 Jim Wilson <jimw@sifive.com>
787 * config/riscv/riscv-c.c (riscv_cpu_cpp_builtins): For ABI_ILP32E,
788 also define __riscv_abi_rve. Delete trailing white space.
790 2018-10-03 Paul Koning <ni1d@arrl.net>
792 Enable LRA register allocator for PDP11.
793 * config/pdp11/constraints.md (Q): Use define_memory_constraint.
796 * config/pdp11/pdp11.c (pdp11_lra_p): New function.
797 * config/pdp11/pdp11.opt (-mlra): New option.
798 * doc/invoke.texi (PDP-11 Options): Document -mlra.
800 2018-10-03 Uros Bizjak <ubizjak@gmail.com>
802 * config/i386/i386.md (*<absneg:code>extendsfdf2): Remove.
803 (*<absneg:code>extend<mode>xf2): Ditto.
805 2018-10-03 Aldy Hernandez <aldyh@redhat.com>
807 PR tree-optimization/87415
808 * tree-vrp.c (set_value_range_with_overflow): Special case one bit
811 2018-10-02 Jeff Law <law@redhat.com>
813 * gimple-fold.c (get_range_strlen): Only set *nonstr when
814 an unterminated string is discovered. Bubble up range
815 even for unterminated strings.
816 (gimple_fold_builtin_strlen): Do not fold if get_range_strlen
817 indicates the string was not terminated via NONSTR.
819 2018-10-03 Aldy Hernandez <aldyh@redhat.com>
821 * tree-vrp.c (extract_range_from_unary_expr): Special case all
823 Do not do anything special for anti-ranges.
825 2018-10-03 Jérôme Lambourg <lambourg@adacore.com>
827 * config/arm/vxworks.h (ARM_TARGET2_DWARF_FORMAT): Adjust to
828 DW_EH_PE_pcrel | DW_EH_PE_indirect for RTPs.
830 2018-10-03 Martin Liska <mliska@suse.cz>
832 PR gcov-profile/86109
833 * coverage.c (coverage_begin_function): Do not
834 mark lambdas as artificial.
835 * tree-core.h (struct GTY): Remove tm_clone_flag
836 and introduce new lambda_function.
837 * tree.h (DECL_LAMBDA_FUNCTION): New macro.
839 2018-10-02 Aaron Sawdey <acsawdey@linux.ibm.com>
842 * config/rs6000/rs6000-string.c (expand_strn_compare): Check that both
843 P8_VECTOR and VSX are enabled.
845 2018-10-02 Andreas Krebbel <krebbel@linux.ibm.com>
847 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
848 0x3907 as CPU model number.
850 2018-10-02 Andreas Krebbel <krebbel@linux.ibm.com>
852 * common/config/s390/s390-common.c: Rename PF_ARCH12 to PF_Z14.
853 * config/s390/s390.h (enum processor_flags): Rename PF_ARCH12 to
854 PF_Z14. Rename TARGET_CPU_ARCH12 to TARGET_CPU_Z14,
855 TARGET_CPU_ARCH12_P to TARGET_CPU_Z14_P, TARGET_ARCH12 to
856 TARGET_Z14, and TARGET_ARCH12_P to TARGET_Z14_P.
857 * config/s390/s390.md: Likewise. Rename also the cpu attribute
858 value from arch12 to z14.
860 2018-10-02 Uros Bizjak <ubizjak@gmail.com>
862 * config/i386/i386.md (fxam<mode>2_i387_with_temp): Remove.
864 (isinf<mode>2): Ditto.
866 2018-10-02 Uros Bizjak <ubizjak@gmail.com>
868 * config/i386/i386.c (ix86_emit_i387_round): Extend op1 to XFmode
869 before emitting fxam. Perform calculations in XFmode.
871 2018-10-02 Marc Glisse <marc.glisse@inria.fr>
873 * match.pd (((X /[ex] A) +- B) * A): New transformation.
875 2018-10-02 Marc Glisse <marc.glisse@inria.fr>
878 * fold-const.c (fold_plusminus_mult_expr): Handle complex and vectors.
879 * tree.c (signed_or_unsigned_type_for): Handle complex.
881 2018-10-02 Jeff Law <law@redhat.com>
883 * gimple-fold.c (get_range_strlen): Remove dead code.
885 2018-10-02 Martin Sebor <msebor@redhat.com>
886 Jeff Law <law@redhat.com>
888 * builtins.c (unterminated_array): Add new arguments.
889 If argument is not terminated, bubble up size and exact
891 (expand_builtin_strnlen): Detect, avoid expanding
892 and diagnose unterminated arrays.
893 (c_strlen): Fill in offset of start of unterminated strings.
894 * builtins.h (unterminated_array): Update prototype.
896 2018-10-02 Richard Biener <rguenther@suse.de>
898 * config/i386/sse.md (reduc_plus_scal_v4df): Avoid the use
899 of haddv4df, first reduce to SSE width and exploit the fact
900 that we only need element zero with the reduction result.
901 (reduc_plus_scal_v2df): Likewise.
903 2018-10-02 Eric Botcazou <ebotcazou@adacore.com>
905 * dojump.h (do_jump): Delete.
906 (do_jump_1): Likewise.
907 (split_comparison): Move around.
908 * dojump.c (do_jump): Make static.
909 (do_jump_1): Likewise.
910 (jumpifnot): Move around.
911 (jumpifnot_1): Likewise.
913 (jumpif_1): Likewise.
914 * expr.c (expand_expr_real_1): Call jumpif[not] instead of do_jump.
916 2018-10-02 Eric Botcazou <ebotcazou@adacore.com>
918 * reorg.c (make_return_insns): Use emit_copy_of_insn_after for the
919 insns in the delay slot and add_insn_after for the jump insn.
921 2018-10-02 Richard Biener <rguenther@suse.de>
923 * tree-inline.c (expand_call_inline): Use the location of
924 the callee declaration for the inline-entry marker.
925 * final.c (notice_source_line): Remove special-casing of
926 NOTE_INSN_INLINE_ENTRY.
928 2018-10-01 Carl Love <cel@us.ibm.com>
931 * config/rs6000/rs6000-builtin.def (__builtin_mffsl): New.
932 (__builtin_mtfsb0): New.
933 (__builtin_mtfsb1): New.
934 ( __builtin_set_fpscr_rn): New.
935 (__builtin_set_fpscr_drn): New.
936 * config/rs6000/rs6000.c (rs6000_expand_mtfsb_builtin): Add.
937 (rs6000_expand_set_fpscr_rn_builtin): Add.
938 (rs6000_expand_set_fpscr_drn_builtin): Add.
939 (rs6000_expand_builtin): Add case statement entries for
940 RS6000_BUILTIN_MTFSB0, RS6000_BUILTIN_MTFSB1,
941 RS6000_BUILTIN_SET_FPSCR_RN, RS6000_BUILTIN_SET_FPSCR_DRN,
942 RS6000_BUILTIN_MFFSL.
943 (rs6000_init_builtins): Add ftype initialization and def_builtin
944 calls for __builtin_mffsl, __builtin_mtfsb0, __builtin_mtfsb1,
945 __builtin_set_fpscr_rn, __builtin_set_fpscr_drn.
946 * config/rs6000.md (rs6000_mtfsb0, rs6000_mtfsb1, rs6000_mffscrn,
947 rs6000_mffscdrn): Add define_insn.
948 (rs6000_set_fpscr_rn, rs6000_set_fpscr_drn): Add define_expand.
949 * doc/extend.texi: Add documentation for the builtins.
951 2018-10-01 Richard Biener <rguenther@suse.de>
953 PR tree-optimization/87465
954 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Fix typo
955 causing branch miscounts.
957 2018-10-01 Tamar Christina <tamar.christina@arm.com>
959 * common/config/aarch64/aarch64-common.c (TARGET_OPTION_DEFAULT_PARAM,
960 aarch64_option_default_param): New.
962 (TARGET_OPTION_VALIDATE_PARAM, aarch64_option_validate_param): New.
963 * config/aarch64/aarch64.c (aarch64_override_options_internal): Simplify
964 stack-clash protection validation code.
966 2018-10-01 Tamar Christina <tamar.christina@arm.com>
968 * params.c (validate_param): New.
969 (add_params): Use it.
970 (set_param_value): Refactor param validation into validate_param.
971 (diagnostic.h): Include.
972 * diagnostic.h (diagnostic_ready_p): New.
974 2018-10-01 Tamar Christina <tamar.christina@arm.com>
976 * params.c (set_param_value):
977 Add index of parameter being validated.
978 * common/common-target.def (option_validate_param): New.
979 * common/common-targhooks.h (default_option_validate_param): New.
980 * common/common-targhooks.c (default_option_validate_param): New.
981 * doc/tm.texi.in (TARGET_OPTION_VALIDATE_PARAM): New.
982 * doc/tm.texi: Regenerate.
984 2018-10-01 Tamar Christina <tamar.christina@arm.com>
987 * config/aarch64/aarch64.c (aarch64_override_options_internal):
988 Add validation for stack-clash parameters and set defaults.
990 2018-10-01 Tamar Christina <tamar.christina@arm.com>
993 * configure.ac: Add stack-clash-protection-guard-size.
994 * doc/install.texi: Document it.
995 * config.in (DEFAULT_STK_CLASH_GUARD_SIZE): New.
996 * params.def: Update comment for guard-size.
997 (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE,
998 PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Update description.
999 * configure: Regenerate.
1001 2018-10-01 Tamar Christina <tamar.christina@arm.com>
1004 * config/aarch64/aarch64.h (STACK_CLASH_MIN_BYTES_OUTGOING_ARGS,
1005 STACK_DYNAMIC_OFFSET): New.
1006 * config/aarch64/aarch64.c (aarch64_layout_frame):
1007 Update outgoing args size.
1008 (aarch64_stack_clash_protection_alloca_probe_range,
1009 TARGET_STACK_CLASH_PROTECTION_ALLOCA_PROBE_RANGE): New.
1011 2018-10-01 Tamar Christina <tamar.christina@arm.com>
1014 * explow.c (anti_adjust_stack_and_probe_stack_clash): Support custom
1016 * target.def (stack_clash_protection_alloca_probe_range): New.
1017 (stack_clash_protection_final_dynamic_probe): Remove.
1018 * targhooks.h (default_stack_clash_protection_alloca_probe_range) New.
1019 (default_stack_clash_protection_final_dynamic_probe): Remove.
1020 * targhooks.c: Likewise.
1021 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_ALLOCA_PROBE_RANGE): New.
1022 (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE): Remove.
1023 * doc/tm.texi: Regenerate.
1025 2018-10-01 Tamar Christina <tamar.christina@arm.com>
1028 * config/aarch64/aarch64-protos.h (aarch64_output_probe_sve_stack_clash): New.
1029 * config/aarch64/aarch64.c (aarch64_output_probe_sve_stack_clash,
1030 aarch64_clamp_to_uimm12_shift): New.
1031 (aarch64_allocate_and_probe_stack_space): Add SVE specific section.
1032 * config/aarch64/aarch64.md (probe_sve_stack_clash): New.
1034 2018-10-01 Tamar Christina <tamar.christina@arm.com>
1037 * config/aarch64/aarch64.c (aarch64_layout_frame): Add assert.
1039 2018-10-01 Jeff Law <law@redhat.com>
1040 Richard Sandiford <richard.sandiford@linaro.org>
1041 Tamar Christina <tamar.christina@arm.com>
1044 * config/aarch64/aarch64.md
1045 (probe_stack_range): Add k (SP) constraint.
1046 * config/aarch64/aarch64.h (STACK_CLASH_CALLER_GUARD,
1047 STACK_CLASH_MAX_UNROLL_PAGES): New.
1048 * config/aarch64/aarch64.c (aarch64_output_probe_stack_range): Emit
1049 stack probes for stack clash.
1050 (aarch64_allocate_and_probe_stack_space): New.
1051 (aarch64_expand_prologue): Use it.
1052 (aarch64_expand_epilogue): Likewise and update IP regs re-use criteria.
1053 (aarch64_sub_sp): Add emit_move_imm optional param.
1055 2018-10-01 MCC CS <deswurstes@users.noreply.github.com>
1057 PR tree-optimization/87261
1058 * match.pd: Remove trailing whitespace.
1059 Add (x & y) | ~(x | y) -> ~(x ^ y),
1060 (~x | y) ^ (x ^ y) -> x | ~y and (x ^ y) | ~(x | y) -> ~(x & y)
1062 2018-10-01 Claudiu Zissulescu <claziss@synopsys.com>
1064 * config/arc/arc.md (*add_n): Clean up pattern, update instruction
1066 (ashlsi3_insn): Update instruction constraints.
1067 (ashrsi3_insn): Likewise.
1068 (rotrsi3): Likewise.
1069 (add_shift): Likewise.
1070 * config/arc/constraints.md (Csz): New 32 bit constraint. It
1071 avoids placing in the limm field small constants which, otherwise,
1072 could end into a small instruction.
1074 2018-10-01 Claudiu Zissulescu <claziss@synopsys.com>
1076 * config/arc/arc.md (maddsidi4_split): Don't use dmac if the
1077 destination register is not odd-even.
1078 (umaddsidi4_split): Likewise.
1080 2018-10-01 Richard Biener <rguenther@suse.de>
1082 * tree-inline.c (expand_call_inline): Store origin of fn
1083 in BLOCK_ABSTRACT_ORIGIN for the inline BLOCK.
1084 * tree.c (block_ultimate_origin): Simplify and do some
1087 2018-09-30 Uros Bizjak <ubizjak@gmail.com>
1089 * config/i386/mmx.md (EMMS): New int iterator.
1090 (emms): New int attribute.
1091 (mmx_<emms>): Macroize insn from *mmx_emms and *mmx_femms using
1092 EMMS int iterator. Explicitly declare clobbers.
1093 (mmx_emms): Remove expander.
1095 * config/i386/predicates.md (emms_operation): Remove predicate.
1096 (vzeroall_pattern): New predicate.
1097 (vzeroupper_pattern): Rename from vzeroupper_operation.
1098 * config/i386/i386.c (ix86_avx_u128_mode_after): Use
1099 vzeroupper_pattern and vzeroall_pattern predicates.
1101 2018-09-30 Peter Bergner <bergner@linux.ibm.com>
1103 PR rtl-optimization/86939
1104 * ira-lives.c (make_hard_regno_born): Rename from this...
1105 (make_hard_regno_live): ... to this. Remove update to conflict
1106 information. Update function comment.
1107 (make_hard_regno_dead): Add conflict information update. Update
1109 (make_object_born): Rename from this...
1110 (make_object_live): ... to this. Remove update to conflict information.
1111 Update function comment.
1112 (make_object_dead): Add conflict information update. Update function
1114 (mark_pseudo_regno_live): Call make_object_live.
1115 (mark_pseudo_regno_subword_live): Likewise.
1116 (mark_hard_reg_dead): Update function comment.
1117 (mark_hard_reg_live): Call make_hard_regno_live.
1118 (process_bb_node_lives): Likewise.
1119 * lra-lives.c (make_hard_regno_born): Rename from this...
1120 (make_hard_regno_live): ... to this. Remove update to conflict
1121 information. Remove now uneeded check_pic_pseudo_p argument.
1122 Update function comment.
1123 (make_hard_regno_dead): Add check_pic_pseudo_p argument and add update
1124 to conflict information. Update function comment.
1125 (mark_pseudo_live): Remove update to conflict information. Update
1127 (mark_pseudo_dead): Add conflict information update.
1128 (mark_regno_live): Call make_hard_regno_live.
1129 (mark_regno_dead): Call make_hard_regno_dead with new arguement.
1130 (process_bb_lives): Call make_hard_regno_live and make_hard_regno_dead.
1132 2018-09-29 H.J. Lu <hongjiu.lu@intel.com>
1135 * config/i386/i386.c (construct_container): Use TImode for
1136 BLKmode values in 2 integer registers.
1138 2018-09-29 Jeff Law <law@redhat.com>
1140 * builtins.c (unterminated_array): Pass in c_strlen_data * to
1141 c_strlen rather than just a tree *.
1142 (c_strlen): Change NONSTR argument to a c_strlen_data pointer.
1143 Update recursive calls appropriately. If caller did not provide a
1144 suitable data pointer, create a local one. When a non-terminated
1145 string is discovered, bubble up information about the string via the
1146 c_strlen_data object.
1147 * builtins.h (c_strlen): Update prototype.
1148 (c_strlen_data): New structure.
1149 * gimple-fold.c (get_range_strlen): Update calls to c_strlen.
1150 For a type 2 call, if c_strlen indicates a non-terminated string
1151 use the length of the non-terminated string.
1152 (gimple_fold_builtin_stpcpy): Update calls to c_strlen.
1154 2018-09-29 Jakub Jelinek <jakub@redhat.com>
1157 * config/i386/avx512fintrin.h (_mm512_abs_pd, _mm512_mask_abs_pd): Use
1158 __m512d type for __A argument rather than __m512.
1160 2018-09-28 John David Anglin <danglin@gcc.gnu.org>
1162 * match.pd (simple_comparison): Don't optimize if either operand is
1163 a function pointer when target needs function pointer canonicalization.
1165 2018-09-28 Segher Boessenkool <segher@kernel.crashing.org>
1167 * config/rs6000/driver-rs6000.c (asm_names): Adjust the entries for
1168 power5 .. power9 to remove indirection.
1169 * config/rs6000/rs6000.h (ASM_CPU_POWER5_SPEC, ASM_CPU_POWER6_SPEC,
1170 ASM_CPU_POWER7_SPEC, ASM_CPU_POWER8_SPEC, ASM_CPU_POWER9_SPEC,
1171 ASM_CPU_476_SPEC): Delete.
1172 (ASM_CPU_SPEC): Adjust.
1173 (EXTRA_SPECS): Delete asm_cpu_power5, asm_cpu_power6, asm_cpu_power7,
1174 asm_cpu_power8, asm_cpu_power9, asm_cpu_476.
1176 2018-09-28 Segher Boessenkool <segher@kernel.crashing.org>
1178 * config.in: Delete HAVE_AS_DCI.
1179 * config/powerpcspe/powerpcspe.h: Treat HAVE_AS_DCI as always true.
1180 * config/rs6000/rs6000.h: Ditto.
1181 * configure.ac: Delete HAVE_AS_DCI.
1182 * configure: Regenerate.
1184 2018-09-28 Segher Boessenkool <segher@kernel.crashing.org>
1186 * config.in (HAVE_AS_LWSYNC): Delete.
1187 * config/powerpcspe/powerpcspe.h (TARGET_LWSYNC_INSTRUCTION): Delete.
1188 * config/powerpcspe/sync.md (*lwsync): Always generate lwsync, never
1190 * config/rs6000/rs6000.h (TARGET_LWSYNC_INSTRUCTION): Delete.
1191 * config/rs6000/sync.md (*lwsync): Always generate lwsync, never do it
1193 * configure.ac: Delete HAVE_AS_LWSYNC.
1194 * configure: Regenerate.
1196 2018-09-28 Eric Botcazou <ebotcazou@adacore.com>
1197 Pierre-Marie de Rodat <derodat@adacore.com>
1199 * calls.c (expand_call): Try to do a tail call for thunks at -O0 too.
1200 * cgraph.h (struct cgraph_thunk_info): Add indirect_offset.
1201 (cgraph_node::create_thunk): Add indirect_offset parameter.
1202 (thunk_adjust): Likewise.
1203 * cgraph.c (cgraph_node::create_thunk): Add indirect_offset parameter
1204 and initialize the corresponding field with it.
1205 (cgraph_node::dump): Dump indirect_offset field.
1206 * cgraphclones.c (duplicate_thunk_for_node): Deal with indirect_offset.
1207 * cgraphunit.c (cgraph_node::analyze): Be prepared for external thunks.
1208 (thunk_adjust): Add indirect_offset parameter and deal with it.
1209 (cgraph_node::expand_thunk): Deal with the indirect_offset field and
1210 pass it to thunk_adjust. Do not call the target hook if it's non-zero
1211 or if the thunk is external or local. Fix formatting. Do not chain
1212 the RESULT_DECL to BLOCK_VARS. Pass the static chain to the target,
1213 if any, in the GIMPLE representation.
1214 * ipa-icf.c (sem_function::equals_wpa): Deal with indirect_offset.
1215 * lto-cgraph.c (lto_output_node): Write indirect_offset field.
1216 (input_node): Read indirect_offset field.
1217 * tree-inline.c (expand_call_inline): Pass indirect_offset field in the
1218 call to thunk_adjust.
1219 * tree-nested.c (struct nesting_info): Add thunk_p field.
1220 (create_nesting_tree): Set it.
1221 (convert_all_function_calls): Copy static chain from targets to thunks.
1222 (finalize_nesting_tree_1): Return early for thunks.
1223 (unnest_nesting_tree_1): Do not finalize thunks.
1224 (gimplify_all_functions): Do not gimplify thunks.
1226 2018-09-28 David Malcolm <dmalcolm@redhat.com>
1228 * opt-suggestions.c (option_proposer::build_option_suggestions):
1229 Release "option_values".
1231 2018-09-28 David Malcolm <dmalcolm@redhat.com>
1233 * coverage.c (get_coverage_counts): Convert problem-reporting dump
1234 messages from MSG_OPTIMIZED_LOCATIONS to MSG_MISSED_OPTIMIZATION.
1235 * dumpfile.c (kind_as_string): New function.
1236 (dump_loc): Rather than a hardcoded prefix of "note: ", use
1237 kind_as_string to vary the prefix based on dump_kind.
1238 (selftest::test_capture_of_dump_calls): Update for above.
1240 2018-09-28 Uros Bizjak <ubizjak@gmail.com>
1242 * config/i386/i386.h (SSE_REGNO): Fix check for FIRST_REX_SSE_REG.
1243 (GET_SSE_REGNO): Rename from SSE_REGNO. Update all uses for rename.
1245 2018-09-28 Uros Bizjak <ubizjak@gmail.com>
1247 * config/i386/i386.h (CC_REGNO): Remove FPSR_REG.
1248 * config/i386/i386.c (ix86_fixed_condition_code_regs): Use
1249 INVALID_REGNUM instead of FPSR_REG.
1250 (ix86_md_asm_adjust): Do not clobber FPSR_REG.
1251 * config/i386/i386.md: Update comment of FP compares.
1252 (fldenv): Do not clobber FPSR_REG.
1254 2018-09-28 Richard Biener <rguenther@suse.de>
1256 * tree.h (BLOCK_ORIGIN): New.
1257 * omp-expand.c (grid_expand_target_grid_body): Assign
1258 BLOCK_ORIGIN to BLOCK_ABSTRACT_ORIGIN.
1259 * tree-inline.c (remap_block): Likewise.
1260 * auto-profile.c (get_function_decl_from_block): Simplify
1261 by eliding the BLOCK_ABSTRACT_ORIGIN chasing.
1262 * langhooks.c (lhd_print_error_function): Likewise.
1263 * optinfo-emit-json.cc (optrecord_json_writer::inlining_chain_to):
1265 * tree-ssa-live.c (remove_unused_scope_block_p): Likewise.
1266 * tree.c (block_nonartificial_location): Likewise.
1267 (block_ultimate_origin): Likewise.
1268 * tree-pretty-print.c (percent_K_format): Likewise. Remove
1269 no longer needed LTO case.
1271 2018-09-28 Andrew Stubbs <ams@codesourcery.com>
1272 Jan Hubicka <jh@suse.cz>
1273 Martin Jambor <mjambor@suse.cz>
1275 * simplify-rtx.c (simplify_merge_mask): New function.
1276 (simplify_ternary_operation): Use it, also see if VEC_MERGEs with the
1277 same masks are used in op1 or op2.
1278 (test_vec_merge): New function.
1279 (test_vector_ops): Call test_vec_merge.
1281 2018-09-28 Eric Botcazou <ebotcazou@adacore.com>
1283 * config/sparc/sparc-protos.h (sparc_branch_cost): Declare.
1284 * config/sparc/sparc.h (BRANCH_COST): Call sparc_branch_cost.
1285 * config/sparc/sparc.c (struct processor_costs): Add branch_cost field.
1286 (cypress_costs): Set it.
1287 (supersparc_costs): Likewise.
1288 (hypersparc_costs): Likewise.
1289 (leon_cost): Likewise.
1290 (leon3_costs): Likewise.
1291 (sparclet_costs): Likewise.
1292 (ultrasparc_costs): Likewise.
1293 (ultrasparc_costs): Likewise.
1294 (niagara_costs): Likewise.
1295 (niagara2_costs): Likewise.
1296 (niagara3_costs): Likewise.
1297 (niagara4_costs): Likewise.
1298 (niagara7_costs): Likewise.
1299 (m8_costs): Likewise.
1300 (TARGET_CAN_FOLLOW_JUMP): Define.
1301 (pass_work_around_errata::gate): Minor tweak.
1302 (sparc_option_override): Remove MASK_FSMULD mask for V7 processors.
1303 Do not set both MASK_VIS4 and MASK_VIS4B for M8 processor.
1304 Automaitcally clear MASK_FSMULD mask for V7 processors.
1305 (sparc_can_follow_jump): New static function.
1306 (output_ubranch): Deal with CROSSING_JUMP_P.
1307 (sparc_use_sched_lookahead): Rewrite using switch statement.
1308 (sparc_issue_rate): Reorder.
1309 (sparc_branch_cost): New function.
1311 2018-09-27 Martin Sebor <msebor@redhat.com>
1313 * tree.h (tree_to_shwi): Add attribute nonnull and pure.
1314 (tree_to_poly_int64, tree_to_uhwi, tree_to_poly_uint64): Same.
1315 (int_fits_type_p): Same.
1317 2018-09-27 Uros Bizjak <ubizjak@gmail.com>
1319 * config/i386/i386.md (FPCR_REG): Remove.
1320 (UNSPEC_FLDCW): Remove.
1321 (x86_fnstcw_1): Use (const_int 0) instead of FPCR_REG.
1322 (x86_fldcw_1): Remove insn pattern.
1323 (fnstenv): Do not clobber FPCR_REG.
1325 * config/i386/i386.h (FIXED_REGISTERS) Remove fpsr register.
1326 (CALL_USED_REGISTERS): Ditto.
1327 (REG_ALLOC_ORDER): Ditto.
1328 (REG_CLASS_CONTENTS): Ditto.
1329 (HI_REGISTER_NAMES): Ditto.
1330 (ADDITIONAL_REGISTER_NAMES): Use defines instead
1331 of numerical constants.
1332 * config/i386/i386.c (regclass_map): Remove fpsr register.
1333 (dbx_register_map): Ditto.
1334 (dbx64_register_map): Ditto.
1335 (svr4_dbx_register_map): Ditto.
1336 (print_reg): Do not handle FPCR_REG.
1338 2018-09-27 Segher Boessenkool <segher@kernel.crashing.org>
1341 * config.in (HAVE_AS_CMPB, HAVE_AS_DFP, HAVE_AS_FPRND, HAVE_AS_MFPGPR,
1342 HAVE_AS_POPCNTB, HAVE_AS_POPCNTD, HAVE_AS_POWER8, HAVE_AS_POWER9):
1343 Delete, always treat as true.
1344 * config/powerpcspe/powerpcspe.c (rs6000_option_override_internal):
1345 Ditto. Simplify remaining code.
1346 * config/powerpcspe/powerpcspe.h: Ditto.
1347 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ditto.
1348 Simplify remaining code.
1349 (rs6000_expand_builtin): Ditto.
1350 * config/rs6000/rs6000.h: Ditto.
1351 * configure.ac: Ditto.
1352 * configure: Regenerate.
1354 2018-09-27 Martin Liska <mliska@suse.cz>
1356 * coverage.c (get_coverage_counts): Revert the formatting
1357 of missing profile opt info.
1359 2018-09-27 Richard Biener <rguenther@suse.de>
1363 * dwarf2out.c (set_block_origin_self): Do not mark outermost
1364 block as we do not output that.
1365 (gen_inlined_subroutine_die): Elide the originally outermost
1366 block, matching what we do for concrete instances.
1367 (decls_for_scope): Add parameter specifying whether to recurse
1370 2018-09-27 Andrew Stubbs <ams@codesourcery.com>
1371 Tom de Vries <tom@codesourcery.com>
1375 * expmed.c (emit_cstore): Fix handling of result_mode == BImode and
1376 STORE_FLAG_VALUE == 1.
1378 2018-09-27 Andreas Krebbel <krebbel@linux.ibm.com>
1380 * config/s390/s390.md (PPA_TX_ABORT, PPA_OOO_BARRIER): New
1381 constant definitions.
1382 ("tx_assist"): Replace magic number with PPA_TX_ABORT.
1383 ("*ppa"): Enable pattern also for -march=zEC12 -mno-htm.
1384 ("speculation_barrier"): New expander definition.
1386 2018-09-26 Indu Bhagat <indu.bhagat@oracle.com>
1388 PR gcov-profile/86957
1389 * common.opt: New warning option -Wmissing-profile.
1390 * coverage.c (get_coverage_counts): Add warning for missing .gcda file.
1391 * doc/invoke.texi: Document -Wmissing-profile.
1393 2018-09-26 Jim Wilson <jimw@sifive.com>
1395 * config/riscv/riscv.md (subsi3_extended2): Add J constraint.
1396 (negdi2, negsi2, negsi2_extended, negsi2_extended2): New.
1398 2018-09-26 Martin Sebor <msebor@redhat.com>
1400 * tree.c (zerop): Change return type to bool.
1401 (integer_zerop, integer_onep, integer_each_onep): Same.
1402 (integer_all_onesp, integer_minus_onep, integer_pow2p): Same.
1403 (integer_nonzerop, integer_truep, tree_ctz, real_zerop): Same.
1404 (real_onep, real_minus_onep, chain_index): Same.
1405 (print_type_hash_statistics, type_list_equal): Same.
1406 * tree.h (zerop): Same.
1407 (zerop, integer_zerop, integer_onep, integer_each_onep): Same.
1408 (integer_all_onesp, integer_minus_onep, integer_pow2p): Same.
1409 (integer_nonzerop, integer_truep, tree_ctz, real_zerop): Same.
1410 (real_onep, real_minus_onep, chain_index): Same.
1411 (print_type_hash_statistics, type_list_equal): Same.
1413 2018-09-26 Jim Wilson <jimw@sifive.com>
1415 * config/riscv/riscv.h (FUNCTION_ARG_REGNO_P): Fix comment.
1417 2018-09-26 Jakub Jelinek <jakub@redhat.com>
1420 * config/i386/i386.c: Include debug.h and dwarf2out.h.
1421 (output_indirect_thunk): Emit DW_CFA_def_cfa_offset after the
1424 2018-09-25 Andrew Stubbs <ams@codesourcery.com>
1426 * builtins.c (get_builtin_sync_mem): Force address mode conversion.
1428 2018-09-26 Uros Bizjak <ubizjak@gmail.com>
1430 * config/i386/i386.h (enum reg_class): Remove FP_TOP_SSE_REGS
1431 and FP_SECOND_SSE_REGS.
1432 (REG_CLASS_NAMES): Ditto.
1433 (REG_CLASS_CONTENTS): Ditto.
1434 * config/i386/i386.c (ix86_preferred_reload_class) Do not handle
1435 FP_TOP_SSE_REGS and FP_SECOND_SSE_REGS classes.
1436 (ix86_preferred_output_reload_class): Ditto.
1437 * config/i386/i386.md (fix_trunc<mode>_i387_fisttp): Change "=&1f"
1438 clobber constraint to "=&f".
1439 (fix_truncdi_i387): Ditto.
1440 (lrintxfdi2): Ditto.
1441 (fistdi2_<rounding>): Ditto.
1442 (fpremxf4_i387): Change "=u" constraint to "=f".
1443 (fprem1xf4_i387): Ditto.
1445 (fptanxf4_i387): Ditto.
1446 (fxtractxf3_i387): Ditto.
1447 (fscalexf4_i387): Ditto.
1448 (atan2xf3): Change "u" constraint to "f".
1449 (fyl2xxf3_i387): Ditto.
1450 (fyl2xp1xf3_i387): Ditto.
1452 2018-09-26 Uros Bizjak <ubizjak@gmail.com>
1455 * config/i386/i386.h (NUM_MODES_FOR_MODE_SWITCHING): Update
1456 for removed I387_MASK_PM entity.
1459 2018-09-26 Jeff Law <law@redhat.com>
1461 2018-09-26 Alexey Neyman <stilor@att.net>
1463 * graphite.h: Include <isl/id.h> and <isl/space.h>; these
1464 headers are no longer pulled in by <isl/val.h>.
1466 2018-09-26 Richard Biener <rguenther@suse.de>
1469 * dwarf2out.c (gen_lexical_block_die): Do not equate inline
1470 or concrete instance DIE to the tree. Create abstract origin
1471 attributes also for concrete instances.
1473 2018-09-26 Alexey Neyman <stilor@att.net>
1475 * graphite.h: Include <isl/id.h> and <isl/space.h>; these
1476 headers are no longer pulled in by <isl/val.h>.
1478 2018-09-26 Matthew Malcomson <matthew.malcomson@arm.com>
1480 * config/arm/arm.c (arm_split_compare_and_swap, arm_split_atomic_op):
1481 Use new helper functions.
1482 * config/arm/sync.md (atomic_load<mode>, atomic_store<mode>):
1483 Use new helper functions.
1484 * config/arm/aarch-common-protos.h (aarch_mm_needs_acquire,
1485 aarch_mm_needs_release): New declarations.
1486 * config/arm/aarch-common.c (aarch_mm_needs_acquire,
1487 aarch_mm_needs_release): New.
1489 2018-09-26 Eric Botcazou <ebotcazou@adacore.com>
1491 * config/arm/arm.c (arm_reorg): Skip Thumb reorg pass for thunks.
1492 (arm32_output_mi_thunk): Deal with long calls.
1494 2018-09-26 Richard Biener <rguenther@suse.de>
1498 * tree-inline.c (expand_call_inline): When the location
1499 of the call is UNKNOWN_LOCATION use DECL_SOURCE_LOCATION
1500 or BUILTINS_LOCATION for the BLOCK_SOURCE_LOCATION of
1501 the inserted BLOCK to make inlined_function_outer_scope_p
1503 * dwarf2out.c (add_call_src_coords_attributes): Do not add
1504 coords for reserved locations.
1506 2018-09-25 Segher Boessenkool <segher@kernel.crashing.org>
1508 * config/rs6000/rs6000.md (*movcc_internal1): Use set_attr_alternative.
1509 (*call_indirect_nonlocal_sysv<mode>): Ditto.
1510 (*call_value_indirect_nonlocal_sysv<mode>): Ditto.
1511 (*sibcall_nonlocal_sysv<mode>): Ditto.
1512 (*sibcall_value_nonlocal_sysv<mode>): Ditto.
1513 (<bd>_<mode>): Ditto.
1514 (<bd>tf_<mode>): Ditto.
1516 2018-09-25 Segher Boessenkool <segher@kernel.crashing.org>
1518 * config/rs6000/altivec.md (*altivec_mov<mode>): Write the output
1519 control string as a list of templates instead of as C code.
1520 (*altivec_movti): Ditto.
1521 * config/rs6000/darwin.md (movdf_low_di): Ditto.
1523 2018-09-25 Jim Wilson <jimw@sifive.com>
1525 * config/riscv/riscv.c (riscv_split_symbol): Mark auipc label as weak
1526 when target symbol is weak.
1528 2018-09-25 Bernd Edlinger <bernd.edlinger@hotmail.de>
1531 * builtins.c (unterminated_array): Simplify.
1532 * expr.c (string_constant): Handle SSA_NAME. Add more exceptions
1533 where pointer arithmetic is safe.
1535 2018-09-25 Segher Boessenkool <segher@kernel.crashing.org>
1538 * config/rs6000/altivec.md (altivec_vspltb): Use
1539 const_0_to_15_operand instead of u5bit_cint_operand.
1540 (*altivec_vspltb_internal): Ditto.
1541 (altivec_vspltb_direct): Ditto.
1542 (altivec_vsplth): Use const_0_to_7_operand instead of
1544 (*altivec_vsplth_internal): Ditto.
1545 (altivec_vsplth_direct): Ditto.
1546 (altivec_vspltw): Use const_0_to_3_operand instead of
1548 (*altivec_vspltw_internal): Ditto.
1549 (altivec_vspltw_direct): Ditto.
1550 (altivec_vspltsf): Ditto.
1551 (*altivec_vspltsf_internal): Ditto.
1552 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Handle the
1553 various splats with the proper size immediate. Reorder the various
1554 cases by ascending size of immediate, and put all such together.
1556 2018-09-25 Richard Biener <rguenther@suse.de>
1559 * dwarf2out.c (add_AT_external_die_ref): Remove now redundant
1561 (maybe_create_die_with_external_ref): Do not create
1562 DW_TAG_imported_unit here.
1563 (add_abstract_origin_attribute): Handle external BLOCK refs.
1564 (dwarf2out_abstract_function): Simplify LTO case.
1565 (dwarf2out_early_finish): Create DW_TAG_imported_unit explicitely
1566 rather than using maybe_create_die_with_external_ref.
1568 2018-09-25 Uros Bizjak <ubizjak@gmail.com>
1571 * config/i386/i386.md (frndintxf2_mask_pm): Remove.
1572 (frndintxf2_mask_pm_i387): Ditto.
1573 (nearbyintxf2): Rewrite expander pattern to match rintxf2.
1574 Enable for !flag_trapping_math.
1575 (nearbyint<mode>2): Enable x87 modes for !flag_trapping_math.
1576 Enable SSE modes for TARGET_SSE4_1 and expand them with round insn.
1577 Change operand 1 predicate to nonimmediate_operand.
1578 (attr "i387_cw"): Remove mask_pm.
1579 * config/i386/i386.h (enum ix86_stack_slot): Remove SLOT_CW_MASK_PM.
1580 (enum ix86_entity): Remove I387_MASK_PM.
1581 * config/i386/i386.c (ix86_i387_mode_needed): Do not
1582 handle I387_MASK_PM.
1583 (ix86_mode_needed): Ditto.
1584 (ix86_mode_after): Ditto.
1585 (ix86_mode_entry): Ditto.
1586 (ix86_mode_exit): Ditto.
1587 (emit_i387_cw_initialization): Do not handle I387_CW_MASK_PM.
1589 2018-09-25 Jakub Jelinek <jakub@redhat.com>
1591 * vr-values.c (vr_values::vr_values): Initialize to_remove_edges and
1592 to_update_switch_stmts to vNULL instead of calling create on them
1595 2018-09-25 Richard Biener <rguenther@suse.de>
1597 PR tree-optimization/87402
1598 * tree-ssa-sccvn.c (SSA_VISITED): Remove unused function.
1599 (visit_phi): Re-instantiate handling of supposed to be VARYING
1600 but non-VARYING backedge value.
1602 2018-09-25 Richard Biener <rguenther@suse.de>
1605 * dwarf2out.c (struct sym_off_pair): New.
1606 (external_die_map): New global.
1607 (lookup_decl_die): When in LTO create DIEs lazily from the
1609 (lookup_block_die): New function, create DIEs lazily in LTO.
1610 (equate_block_to_die): New function.
1611 (dwarf2out_die_ref_for_decl): During WPA get the association
1612 from the external DIE map.
1613 (dwarf2out_register_external_die): Record mapping into the
1615 (maybe_create_die_with_external_ref): New function split out from
1616 DIE generation part of old dwarf2out_register_external_die.
1617 (add_abstract_origin_attribute): Do not return the DIE. When
1618 in LTO reference externals directly.
1619 (dwarf2out_abstract_function): When in LTO ignore calls for
1620 decls with external DIEs (already present abstract instances).
1621 (gen_call_site_die): Adjust.
1622 (add_high_low_attributes): Likewise.
1623 (gen_lexical_block_die): Likewise.
1624 (gen_inlined_subroutine_die): Likewie.
1625 (gen_block_die): Likewise.
1626 (dwarf2out_inline_entry): Likewise.
1627 (dwarf2out_early_finish): In LTRANS phase create DW_TAG_imported_unit
1630 2018-09-25 Martin Liska <mliska@suse.cz>
1632 * ipa-fnsummary.c (estimate_node_size_and_time): Scale by two
1633 integers and not by a float value.
1635 2018-09-25 Martin Liska <mliska@suse.cz>
1638 * dbgcnt.c (dbg_cnt_process_single_pair): Return false
1640 * dumpfile.c (dump_enable_all): Remove extra parenthesis.
1641 * gcov-tool.c: Declare the function with ATTRIBUTE_NORETURN.
1642 * godump.c (go_format_type): Remove extra parenthesis.
1644 2018-09-25 Martin Liska <mliska@suse.cz>
1646 * alias.c (set_dest_equal_p): Remove unused function.
1647 * config/i386/i386.c (def_builtin_pure2): Likewise.
1648 * diagnostic-show-locus.c (class layout): Remove
1650 (layout::layout): Likewise here.
1651 * dump-context.h (class temp_dump_context): Likewise.
1652 * dwarf2out.c (add_AT_fde_ref): Remove unused function.
1653 (add_AT_loclistsptr): Likewise.
1654 (add_AT_offset): Likewise.
1655 (get_AT_hi_pc): Likewise.
1656 (is_comdat_die): Likewise.
1657 (type_is_enum): Likewise.
1658 (ceiling): Likewise.
1659 (add_AT_vms_delta): Likewise.
1660 (is_class_die): Likewise.
1661 * edit-context.c (class line_event): Remove unused field.
1662 * graphite-sese-to-poly.c (tree_int_to_gmp): Remove
1664 * ipa-cp.c (ipa_get_vr_lat): Likewise.
1665 * lra-constraints.c (ok_for_index_p_nonstrict): Likewise.
1666 (ok_for_base_p_nonstrict): Likewise.
1667 * tree-chrec.c (is_not_constant_evolution): Likewise.
1668 (chrec_fold_poly_cst): Likewise.
1669 * tree-if-conv.c (has_pred_critical_p): Likewise.
1670 * tree-ssa-coalesce.c (print_exprs): Likewise.
1671 * tree-ssa-pre.c (bitmap_set_contains_expr): Likewise.
1672 * tree-ssa-uninit.c (is_and_or_or_p): Likewise.
1673 * tree-vrp.c (value_ranges_intersect_p): Likewise.
1674 (value_range_nonnegative_p): Likewise.
1676 2018-09-25 Martin Liska <mliska@suse.cz>
1678 * config/powerpcspe/powerpcspe.c (rs6000_output_function_epilogue):
1679 Do not handle "GNU Pascal".
1680 * config/rs6000/rs6000.c (rs6000_output_function_epilogue):
1682 * config/sparc/sparc.c (sparc_pass_by_reference): Remove Pascal
1683 from documentation. Likewise.
1684 * dbxout.c (dbxout_range_type): Likewise.
1685 * doc/cpp.texi: Likewise.
1686 * doc/extend.texi: Likewise.
1687 * doc/frontends.texi: Likewise.
1688 * doc/invoke.texi: Remove Pascal entry.
1689 * tree.def (CLEANUP_POINT_EXPR): Likewise.
1690 * doc/rtl.texi (MODE_FUNCTION): Remove not used entry.
1692 2018-09-25 Martin Liska <mliska@suse.cz>
1695 * doc/invoke.texi: Document all parameters and remove default
1698 2018-09-25 Ilya Leoshkevich <iii@linux.ibm.com>
1701 * rtl.c (rtx_code_size): Take into account that EQ_ATTR_ALT
1702 contains HOST_WIDE_INTs when computing its size.
1704 2018-09-24 Jim Wilson <jimw@sifive.com>
1707 * config/riscv/riscv.h (STACK_BOUNDARY): Test riscv_abi == ABI_ILP32E
1709 (ABI_STACK_BOUNDARY, MAX_ARGS_IN_REGISTERS): Likewise.
1711 2018-09-24 Andrew Pinski <apinski@marvell.com>
1713 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Don't
1714 access prev before checking it for NULLness in the
1715 AARCH64_FUSE_CMP_BRANCH case.
1717 2018-09-24 H.J. Lu <hongjiu.lu@intel.com>
1720 * config/i386/i386.c (rest_of_insert_endbranch): Set
1721 endbr_queued_at_entrance to true and don't insert ENDBR if
1722 x86_function_profiler will be called.
1723 (x86_function_profiler): Insert ENDBR if endbr_queued_at_entrance
1725 * config/i386/i386.h (machine_function): Add
1726 endbr_queued_at_entrance.
1728 2018-09-24 Ilya Leoshkevich <iii@linux.ibm.com>
1730 * genattrtab.c (mk_attr_alt): Use alternative_mask.
1731 (attr_rtx_1): Adjust caching to match the new EQ_ATTR_ALT field
1733 (check_attr_test): Use alternative_mask.
1734 (get_attr_value): Likewise.
1735 (compute_alternative_mask): Use alternative_mask and XWINT.
1736 (make_alternative_compare): Use alternative_mask.
1737 (attr_alt_subset_p): Use XWINT.
1738 (attr_alt_subset_of_compl_p): Likewise.
1739 (attr_alt_intersection): Use alternative_mask and XWINT.
1740 (attr_alt_union): Likewise.
1741 (attr_alt_complement): Use HOST_WIDE_INT and XWINT.
1742 (mk_attr_alt): Use alternative_mask and HOST_WIDE_INT.
1743 (simplify_test_exp): Use alternative_mask and XWINT.
1744 (write_test_expr): Use alternative_mask and XWINT, adjust bit
1745 number calculation to support 64 bits. Generate code that
1746 checks 64-bit masks.
1747 (main): Use alternative_mask.
1748 * rtl.def (EQ_ATTR_ALT): Change field types from ii to ww.
1750 2018-09-24 Ilya Leoshkevich <iii@linux.ibm.com>
1753 * config/s390/s390.c (s390_emit_epilogue): Do not use PARALLEL
1754 RETURN+USE when returning via %r14.
1756 2018-09-24 Martin Liska <mliska@suse.cz>
1758 * gcov.c (output_lines): Print colorization legend
1759 for both flag_use_colors and flag_use_hotness_colors.
1762 2018-09-24 Martin Liska <mliska@suse.cz>
1764 * coverage.c (get_coverage_counts): Use warning_at
1765 with current_function_decl location. Use %qD in warning
1768 2018-09-24 Martin Liska <mliska@suse.cz>
1770 * memory-block.h (memory_block_pool::release): Annotate with
1771 valgrind that the memory is not accessible.
1773 2018-09-24 Martin Liska <mliska@suse.cz>
1776 * asan.c: Make asan_handled_variables extern.
1778 * cfgexpand.c (expand_stack_vars): Make sure
1779 a representative is unpoison if another
1780 variable in the partition is handled by
1781 use-after-scope sanitization.
1783 2018-09-24 Richard Biener <rguenther@suse.de>
1785 PR tree-optimization/63155
1786 * tree-ssa-propagate.c (add_ssa_edge): Avoid adding PHIs to
1787 the worklist when the edge of the respective argument isn't
1790 2018-09-23 Uros Bizjak <ubizjak@gmail.com>
1792 * config/i386/i386.h (enum reg_class): Rename MASK_REGS to
1793 ALL_MASK_REGS and MASK_EVEX_REGS to MASK_REGS.
1794 (MASK_CLASS_P): Update for rename.
1795 (MAYBE_MASK_CLASS_P): Ditto.
1796 (REG_CLASS_NAMES): Update.
1797 (REG_CLASS_CONTENT): Update.
1798 * config/i386/i386.c (regclass_map): Update for MASK_REG
1799 and ALL_MASK_REGS rename.
1800 * config/i386/constraints.md (Yk): Update for rename.
1803 2018-09-23 Uros Bizjak <ubizjak@gmail.com>
1805 * config/i386/i386.h (enum reg_class): Remove
1806 EVEX_SSE_REGS and MOD4_SSE_REGS.
1807 (REG_CLASS_NAMES): Update.
1808 (REG_CLASS_CONTENT): Update.
1809 * config/i386/i386.c (regclass_map): Declare AVX-512 SSE
1810 registers as ALL_SSE_REGS.
1811 (ix86_additional_allocno_class_p): Remove.
1812 (TARGET_ADDITIONAL_ALLOCNO_CLASS_P): Remove.
1813 (ix86_register_priority): Lower priority of EVEX SSE registers.
1814 Use IN_RANGE macro where appropriate.
1815 (ix86_hard_regno_mode_ok): Merge AVX-5124FMAPS and
1816 AVX-5124VNNIW checks.
1817 (ix86_modes_tieable_p): Tie 512-bit SSE modes.
1818 * config/i386/sse.md (avx5124fmaddps_4fmaddps)
1819 (avx5124fmaddps_4fmaddps_mask, avx5124fmaddps_4fmaddps_maskz)
1820 (avx5124fmaddps_4fmaddss, avx5124fmaddps_4fmaddss_mask)
1821 (avx5124fmaddps_4fmaddss_maskz, avx5124fmaddps_4fnmaddps)
1822 (avx5124fmaddps_4fnmaddps_mask, avx5124fmaddps_4fnmaddps_maskz)
1823 (avx5124fmaddps_4fnmaddss, avx5124fmaddps_4fnmaddss_mask)
1824 (avx5124fmaddps_4fnmaddss_maskz, avx5124vnniw_vp4dpwssd)
1825 (avx5124vnniw_vp4dpwssd_mask, avx5124vnniw_vp4dpwssd_maskz)
1826 (avx5124vnniw_vp4dpwssds, avx5124vnniw_vp4dpwssds_mask)
1827 (avx5124vnniw_vp4dpwssds_maskz): Use "v" instead of "Yh" constraint.
1828 * config/i386/constraints.md (Yh): Remove.
1830 2018-09-23 Uros Bizjak <ubizjak@gmail.com>
1832 * config/i386/i386.c (regclass_map): Declare integer REX registers
1835 2018-09-23 Gerald Pfeifer <gerald@pfeifer.com>
1837 * doc/service.texi (Service): Switch the fsf.org link to https.
1839 2018-09-22 Chung-Ju Wu <jasonwucj@gmail.com>
1842 * config/nds32/nds32.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
1843 Define to speculation_safe_value_not_needed.
1845 2018-09-21 Florian Weimer <fweimer@redhat.com>
1848 * doc/extend.texi (Common Function Attributes): Mention that
1849 noreturn suppresses tail call optimization.
1851 2018-09-21 Jeff Law <law@redhat.com>
1853 * gimple-ssa-evrp.c (evrp_dom_walker::cleanup): Call
1854 vr_values::cleanup_edges_and_switches.
1855 * tree-vrp.c (to_remove_edges, to_update_switch_stmts): Moved into
1857 (identify_jump_threads): Remove EDGE_IGNORE handling.
1858 (execute_vrp): Move handling of to_remove_edges and
1859 to_update_switch_stmts into vr_values class member functions.
1860 * tree-vrp.h (switch_update, to_remove_edges): Remove declarations.
1861 (to_update_switch_stmts): Likewise.
1862 * vr-values.c: Include cfghooks.h.
1863 (vr_values::vr_values): Initialize to_remove_edges and
1864 to_update_switch_stmts.
1865 (vr_values::~vr_values): Verify to_remove_edges and
1866 to_update_switch_stmts are empty.
1867 (vr_values::simplify_switch_using_ranges): Set EDGE_IGNORE as needed.
1868 (vr_values::cleanup_edges_and_switches): New member function.
1869 * vr-values.h (vr_values): Add cleanup_edges_and_switches member
1870 function. Add new data members.
1872 2018-09-21 David Malcolm <dmalcolm@redhat.com>
1874 PR tree-optimization/87309
1875 * dumpfile.c (dump_context::begin_scope): Filter the dump_loc
1876 calls with pflags and alt_flags.
1877 (selftest::test_capture_of_dump_calls): Add test of interaction of
1878 MSG_OPTIMIZED_LOCATIONS with AUTO_DUMP_SCOPE.
1880 2018-09-21 Olivier Hainque <hainque@adacore.com>
1882 * config.gcc: Factorize and comment inclusion of vxworks-dummy.h.
1884 2018-09-21 Olivier Hainque <hainque@adacore.com>
1886 * config/vxworks.h (CLEAR_INSN_CACHE): #define to 1.
1888 2018-09-21 Olivier Hainque <hainque@adacore.com>
1890 * config/vxworks.h (VXWORKS_LIBGCC_SPEC): Remove -lc_internal.
1891 Merge block comment with the one ahead of VXWORKS_LIBS_RTP. Then:
1892 (VXWORKS_LIBS_RTP): Minor reordering.
1894 2018-09-21 Olivier Hainque <hainque@adacore.com>
1896 * config/vxworks.h (STARTFILE_PREFIX_SPEC): Define.
1897 (VXWORKS_LIBS_DIR_RTP): Remove definition and use.
1899 2018-09-21 Olivier Hainque <hainque@adacore.com>
1901 * config/vxworks.h (SIZE_TYPE): Account for TARGET_VXWORKS64.
1902 (PTRDIFF_TYPE): Likewise.
1904 2018-09-21 Olivier Hainque <hainque@adacore.com>
1906 * config.gcc: Enforce def of TARGET_VXWORKS64 to 1 from
1907 triplet, similar to support for VxWorks7.
1908 * config/vxworks-dummy.h: Provide a default definition
1909 of TARGET_VXWORKS64 to 0.
1911 2018-09-21 Olivier Hainque <hainque@adacore.com>
1913 * config/vxworks.h (TARGET_VXWORKS7): Move default definition ...
1914 * config/vxworks-dummy.h: here.
1916 2018-09-21 Olivier Hainque <hainque@adacore.com>
1918 * config.gcc: Prepend vxworks-dummy.h to tm_file for powerpc*
1920 2018-09-21 Shaokun Zhang <zhangshaokun@hisilicon.com>
1921 Bo Zhou <zbo.zhou@hisilicon.com>
1923 * config/aarch64/aarch64-cores.def (tsv110): New CPU.
1924 * config/aarch64/aarch64-tune.md: Regenerated.
1925 * doc/invoke.texi (AArch64 Options/-mtune): Add "tsv110".
1926 * config/aarch64/aarch64.c (tsv110_tunings): New tuning table.
1927 * config/aarch64/aarch64-cost-tables.h: Add "tsv110" extra costs.
1929 2018-09-21 Andrew Stubbs <ams@codesourcery.com>
1930 Julian Brown <julian@codesourcery.com>
1932 * builtins.c (get_builtin_sync_mem): Handle address spaces.
1934 2018-09-21 Eric Botcazou <ebotcazou@adacore.com>
1936 * config/rs6000/rs6000.c (rs6000_function_ok_for_sibcall): Return false
1937 if the call takes a static chain.
1939 2018-09-21 Martin Liska <mliska@suse.cz>
1941 * auto-profile.c (autofdo_source_profile::read): Do not
1943 (read_profile): Do not add working sets.
1944 (read_autofdo_file): Remove sum_all.
1945 (afdo_callsite_hot_enough_for_early_inline): Remove const
1947 * coverage.c (struct counts_entry): Remove gcov_summary.
1948 (read_counts_file): Read new GCOV_TAG_OBJECT_SUMMARY,
1949 do not support GCOV_TAG_PROGRAM_SUMMARY.
1950 (get_coverage_counts): Remove summary and expected
1952 * coverage.h (get_coverage_counts): Likewise.
1953 * doc/gcov-dump.texi: Remove -w option.
1954 * gcov-dump.c (dump_working_sets): Remove.
1955 (main): Do not support '-w' option.
1956 (print_usage): Likewise.
1957 (tag_summary): Likewise.
1958 * gcov-io.c (gcov_write_summary): Do not dump
1960 (gcov_read_summary): Likewise.
1961 (gcov_histo_index): Remove.
1962 (gcov_histogram_merge): Likewise.
1963 (compute_working_sets): Likewise.
1964 * gcov-io.h (GCOV_TAG_OBJECT_SUMMARY): Mark
1966 (GCOV_TAG_PROGRAM_SUMMARY): Mark it obsolete.
1967 (GCOV_TAG_SUMMARY_LENGTH): Adjust.
1968 (GCOV_HISTOGRAM_SIZE): Remove.
1969 (GCOV_HISTOGRAM_BITVECTOR_SIZE): Likewise.
1970 (struct gcov_summary): Simplify rapidly just
1971 to runs and sum_max fields.
1972 (gcov_histo_index): Remove.
1973 (NUM_GCOV_WORKING_SETS): Likewise.
1974 (compute_working_sets): Likewise.
1975 * gcov-tool.c (print_overlap_usage_message): Remove
1976 trailing empty line.
1977 * gcov.c (read_count_file): Read GCOV_TAG_OBJECT_SUMMARY.
1978 (output_lines): Remove program related line.
1979 * ipa-profile.c (ipa_profile): Do not consider GCOV histogram.
1980 * lto-cgraph.c (output_profile_summary): Do not stream GCOV
1982 (input_profile_summary): Do not read it.
1983 (merge_profile_summaries): And do not merge it.
1984 (input_symtab): Do not call removed function.
1985 * modulo-sched.c (sms_schedule): Do not print sum_max.
1986 * params.def (HOT_BB_COUNT_FRACTION): Reincarnate param that was
1987 removed when histogram method was invented.
1988 (HOT_BB_COUNT_WS_PERMILLE): Mention that it's used only in LTO
1990 * postreload-gcse.c (eliminate_partially_redundant_load): Fix
1992 * predict.c (get_hot_bb_threshold): Use HOT_BB_COUNT_FRACTION
1993 and dump selected value.
1994 * profile.c (add_working_set): Remove.
1995 (get_working_sets): Likewise.
1996 (find_working_set): Likewise.
1997 (get_exec_counts): Do not work with working sets.
1998 (read_profile_edge_counts): Do not inform as sum_max is removed.
1999 (compute_branch_probabilities): Likewise.
2000 (compute_value_histograms): Remove argument for call of
2001 get_coverage_counts.
2002 * profile.h: Do not make gcov_summary const.
2004 2018-09-21 Monk Chiang <sh.chiang04@gmail.com>
2006 * config.gcc (nds32*-*-*): Set TARGET_DEFAULT_TLSDESC_TRAMPOLINE=0.
2008 2018-09-21 Eric Botcazou <ebotcazou@adacore.com>
2010 PR tree-optimization/86990
2011 * gimple-ssa-store-merging.c (imm_store_chain_info:coalesce_immediate):
2012 Check that the entire merged store group is made of constants only for
2015 2018-09-20 Allan Sandfeld Jensen <allan.jensen@qt.io>
2017 * gcc.c (LINK_COMMAND_SPEC): Handle -r like -nostdlib.
2018 (VTABLE_VERIFICATION_SPEC): Likewise.
2019 (SANITIZER_EARLY_SPEC): Likewise.
2020 (SANITIZER_SPEC): Likewise.
2021 * config/darwin.h (LINK_COMMAND_SPEC): Likewise.
2022 * doc/invoke.texi (Link Options): Document -r.
2024 2018-09-20 Richard Biener <rguenther@suse.de>
2027 * gimplify.c (gimplify_expr): Retain alignment of
2028 addressable lvalue in dereference.
2030 2018-09-20 Alexandre Oliva <aoliva@redhat.com>
2033 * configure.ac: Check for .loc is_stmt support.
2034 * configure, config.in: Rebuilt.
2035 * dwarf2out.c (dwarf2out_source_line): Skip is_stmt
2038 2018-09-20 Segher Boessenkool <segher@kernel.crashing.org>
2040 * config/rs6000/rs6000.opt (misel=no, misel=yes): Delete.
2041 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -misel=yes and
2044 2018-09-20 Segher Boessenkool <segher@kernel.crashing.org>
2046 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete
2048 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete
2051 2018-09-20 Marek Polacek <polacek@redhat.com>
2053 * doc/invoke.texi: Add -Wno-init-list-lifetime to C++ Language Options.
2055 2018-09-20 Richard Sandiford <richard.sandiford@arm.com>
2057 PR tree-optimization/87288
2058 * tree-vect-loop.c (vect_analyze_loop_2): Take PEELING_FOR_GAPS
2059 into account when determining PEELING_FOR_NITERS.
2061 2018-09-20 Richard Sandiford <richard.sandiford@arm.com>
2063 PR tree-optimization/86877
2064 * tree-vect-loop.c (vect_analyze_loop_2): Call
2065 vect_verify_datarefs_alignment.
2067 2018-09-19 Marek Polacek <polacek@redhat.com>
2069 * doc/invoke.texi: Document -Wclass-conversion.
2071 2018-09-19 John David Anglin <danglin@gcc.gnu.org>
2073 * config/pa/pa.c (pa_adjust_priority): Delete.
2074 (TARGET_SCHED_ADJUST_PRIORITY): Delete define.
2076 * config/pa/pa.md (atomic_storeqi): Restore deleted expander.
2077 (atomic_storehi): Likewise.
2078 (atomic_storesi): Likewise.
2079 (atomic_loaddi): Restore compare and swap exchange loop code.
2081 2018-09-19 Segher Boessenkool <segher@kernel.crashing.org>
2083 PR rtl-optimization/86902
2084 * combine.c (try_combine): When changing the CC mode used, don't change
2085 an unrelated mode in other_insn to that new CC mode.
2087 2018-09-19 David Malcolm <dmalcolm@redhat.com>
2089 * tree-data-ref.c (runtime_alias_check_p): Use formatted printing
2090 with %T in place of calls to dump_generic_expr.
2091 (prune_runtime_alias_test_list): Likewise.
2092 (create_runtime_alias_checks): Likewise.
2093 * tree-vect-data-refs.c (vect_check_nonzero_value): Likewise.
2094 (vect_analyze_data_ref_dependence): Likewise.
2095 (vect_slp_analyze_data_ref_dependence): Likewise.
2096 (vect_record_base_alignment): Likewise. Use %G in place of call
2097 to dump_gimple_stmt.
2098 (vect_compute_data_ref_alignment): Likewise.
2099 (verify_data_ref_alignment): Likewise.
2100 (vect_find_same_alignment_drs): Likewise.
2101 (vect_analyze_group_access_1): Likewise.
2102 (vect_analyze_data_ref_accesses): Likewise.
2103 (dependence_distance_ge_vf): Likewise.
2104 (dump_lower_bound): Likewise.
2105 (vect_prune_runtime_alias_test_list): Likewise.
2106 (vect_find_stmt_data_reference): Likewise.
2107 (vect_analyze_data_refs): Likewise.
2108 (vect_create_addr_base_for_vector_ref): Likewise.
2109 (vect_create_data_ref_ptr): Likewise.
2110 * tree-vect-loop-manip.c (vect_set_loop_condition): Likewise.
2111 (vect_can_advance_ivs_p): Likewise.
2112 (vect_update_ivs_after_vectorizer): Likewise.
2113 (vect_gen_prolog_loop_niters): Likewise.
2114 (vect_prepare_for_masked_peels): Likewise.
2115 * tree-vect-loop.c (vect_determine_vf_for_stmt): Likewise.
2116 (vect_determine_vectorization_factor): Likewise.
2117 (vect_is_simple_iv_evolution): Likewise.
2118 (vect_analyze_scalar_cycles_1): Likewise.
2119 (vect_analyze_loop_operations): Likewise.
2120 (report_vect_op): Likewise.
2121 (vect_is_slp_reduction): Likewise.
2122 (check_reduction_path): Likewise.
2123 (vect_is_simple_reduction): Likewise.
2124 (vect_create_epilog_for_reduction): Likewise.
2125 (vect_finalize_reduction:): Likewise.
2126 (vectorizable_induction): Likewise.
2127 (vect_transform_loop_stmt): Likewise.
2128 (vect_transform_loop): Likewise.
2129 (optimize_mask_stores): Likewise.
2130 * tree-vect-patterns.c (vect_pattern_detected): Likewise.
2131 (vect_split_statement): Likewise.
2132 (vect_recog_over_widening_pattern): Likewise.
2133 (vect_recog_average_pattern): Likewise.
2134 (vect_determine_min_output_precision_1): Likewise.
2135 (vect_determine_precisions_from_range): Likewise.
2136 (vect_determine_precisions_from_users): Likewise.
2137 (vect_mark_pattern_stmts): Likewise.
2138 (vect_pattern_recog_1): Likewise.
2139 * tree-vect-slp.c (vect_get_and_check_slp_defs): Likewise.
2140 (vect_record_max_nunits): Likewise.
2141 (vect_build_slp_tree_1): Likewise.
2142 (vect_build_slp_tree_2): Likewise.
2143 (vect_print_slp_tree): Likewise.
2144 (vect_analyze_slp_instance): Likewise.
2145 (vect_detect_hybrid_slp_stmts): Likewise.
2146 (vect_detect_hybrid_slp_1): Likewise.
2147 (vect_slp_analyze_operations): Likewise.
2148 (vect_slp_analyze_bb_1): Likewise.
2149 (vect_transform_slp_perm_load): Likewise.
2150 (vect_schedule_slp_instance): Likewise.
2151 * tree-vect-stmts.c (vect_mark_relevant): Likewise.
2152 (vect_mark_stmts_to_be_vectorized): Likewise.
2153 (vect_init_vector_1): Likewise.
2154 (vect_get_vec_def_for_operand): Likewise.
2155 (vect_finish_stmt_generation_1): Likewise.
2156 (vect_check_load_store_mask): Likewise.
2157 (vectorizable_call): Likewise.
2158 (vectorizable_conversion): Likewise.
2159 (vectorizable_operation): Likewise.
2160 (vectorizable_load): Likewise.
2161 (vect_analyze_stmt): Likewise.
2162 (vect_is_simple_use): Likewise.
2163 (vect_get_vector_types_for_stmt): Likewise.
2164 (vect_get_mask_type_for_stmt): Likewise.
2165 * tree-vectorizer.c (increase_alignment): Likewise.
2167 2018-09-19 Andrew Stubbs <ams@codesourcery.com>
2169 * doc/rtl.texi: Adjust vec_select description.
2170 * simplify-rtx.c (simplify_binary_operation_1): Allow VEC_SELECT to use
2171 non-constant selectors.
2173 2018-09-19 Matthew Malcomson <matthew.malcomson@arm.com>
2175 * config/aarch64/aarch64-protos.h
2176 (aarch64_offset_9bit_signed_unscaled_p): New declaration.
2177 * config/aarch64/aarch64.md (arches): New "rcpc8_4" attribute value.
2178 (arch_enabled): Add check for "rcpc8_4" attribute value of "arch".
2179 * config/aarch64/aarch64.h (AARCH64_FL_RCPC8_4): New bitfield.
2180 (AARCH64_FL_FOR_ARCH8_4): Include AARCH64_FL_RCPC8_4.
2181 (AARCH64_FL_PROFILE): Move index so flags are ordered.
2182 (AARCH64_ISA_RCPC8_4): New flag.
2183 * config/aarch64/aarch64.c (offset_9bit_signed_unscaled_p): Renamed
2184 to aarch64_offset_9bit_signed_unscaled_p.
2185 * config/aarch64/atomics.md (atomic_store<mode>): Allow offset
2187 * config/aarch64/constraints.md (Ust): New constraint.
2188 * config/aarch64/predicates.md.
2189 (aarch64_9bit_offset_memory_operand): New predicate.
2190 (aarch64_rcpc_memory_operand): New predicate.
2192 2018-09-19 Eric Botcazou <ebotcazou@adacore.com>
2194 PR rtl-optimization/87361
2195 * rtlanal.c (nonzero_bits1): Revert accidental change.
2197 2018-09-19 Richard Biener <rguenther@suse.de>
2199 PR tree-optimization/87349
2200 PR tree-optimization/87342
2201 * tree-ssa-sccvn.c (do_rpo_vn): Iterate max_rpo computation.
2203 2018-09-18 Marek Polacek <polacek@redhat.com>
2205 P1064R0 - Allowing Virtual Function Calls in Constant Expressions
2206 * gimple-fold.c (gimple_get_virt_method_for_vtable): Adjust assert.
2208 2018-09-18 Segher Boessenkool <segher@kernel.crashing.org>
2210 * config/rs6000/rs6000.md: Remove old "Cygnus sibcall" comment.
2212 2018-09-18 Segher Boessenkool <segher@kernel.crashing.org>
2214 PR rtl-optimization/86882
2215 * rtlanal.c (reg_overlap_mentioned_p): Handle CLOBBER.
2217 2018-09-18 Uros Bizjak <ubizjak@gmail.com>
2219 * config/i386/i386.md (*<code>extend<mode>xf2): Macroize insn from
2220 *<code>extendsfxf2 and *<code>extenddfxf2 using MODEF mode iterator.
2222 2018-09-18 Jonathan Wakely <jwakely@redhat.com>
2225 * doc/invoke.texi (Link Options): Fix formatting and grammar.
2227 2018-09-18 Richard Biener <rguenther@suse.de>
2230 * tree-ssa-coalesce.c (tree_int_map_hasher): Remove.
2231 (compute_samebase_partition_bases): Likewise.
2232 (coalesce_ssa_name): Always use compute_optimized_partition_bases.
2233 (gimple_can_coalesce_p): Simplify.
2235 2018-09-18 Hans-Peter Nilsson <hp@bitrange.com>
2237 Handle a library implementation of ffs calling __builtin_ffs.
2238 * config/mmix/mmix.c (TARGET_INIT_LIBFUNCS): Override with...
2239 (mmix_init_libfuncs): New function: make __builtin_ffs expand
2242 2018-09-17 David Malcolm <dmalcolm@redhat.com>
2244 * diagnostic-show-locus.c (class layout_range): Add field
2246 (layout_range::layout_range): Add "original_idx" param and use it
2247 to initialize new field.
2248 (make_range): Use 0 for original_idx.
2249 (layout::layout): Pass in index to calls to
2250 maybe_add_location_range.
2251 (layout::maybe_add_location_range): Add param "original_idx" and
2252 pass it on to layout_range.
2253 (layout::print_any_labels): Pass on range->m_original_idx to
2255 (gcc_rich_location::add_location_if_nearby): Use 0 for
2257 * gcc-rich-location.h (text_range_label::get_text): Update for new
2259 (range_label_for_type_mismatch::get_text): Likewise.
2261 2018-09-17 Uros Bizjak <ubizjak@gmail.com>
2263 * config/i386/i386.c (ix86_emit_i387_log1p): Emit fldln2 earlier.
2265 2018-09-17 David Malcolm <dmalcolm@redhat.com>
2267 * gimple-ssa-sprintf.c (fmtwarn): Update for introduction of
2268 format_string_diagnostic_t.
2269 (fmtwarn_n): Likewise.
2270 * substring-locations.c
2271 (format_string_diagnostic_t::format_string_diagnostic_t) New ctor.
2272 (format_warning_n_va): Convert to...
2273 (format_string_diagnostic_t::emit_warning_n_va): ...this.
2274 (format_warning_va): Convert to...
2275 (format_string_diagnostic_t::emit_warning_va): ...this.
2276 (format_warning_at_substring): Convert to...
2277 (format_string_diagnostic_t::emit_warning): ...this.
2278 (format_warning_at_substring_n): Convert to...
2279 (format_string_diagnostic_t::emit_warning_n): ...this.
2280 * substring-locations.h (class format_string_diagnostic_t): New
2282 (format_warning_va): Convert to
2283 format_string_diagnostic_t::emit_warning_va.
2284 (format_warning_n_va): Convert to
2285 format_string_diagnostic_t::emit_warning_n_va.
2286 (format_warning_at_substring): Convert to
2287 format_string_diagnostic_t::emit_warning.
2288 (format_warning_at_substring_n): Convert to
2289 format_string_diagnostic_t::emit_warning_n.
2291 2018-09-17 Cesar Philippidis <cesar@codesourcery.com>
2292 Bernd Schmidt <bernds_cb1@t-online.de>
2294 * config/nvptx/nvptx.md (atomic_fetch_<logic><mode>): Enable with
2297 2018-09-17 Uros Bizjak <ubizjak@gmail.com>
2299 * config/i386/i386.md (truncxf<mode>2_i387_noop_unspec): Change
2300 operand 0 predicate to nonimmediate operand.
2301 (rint<mode>2_frndint): Remove insn pattern.
2302 (rint<mode>2): Change operand 1 predicate to general_operand.
2303 Extend operand 1 to XFmode and generate rintxf2 insn.
2304 (frndintxf2_<rounding>): Rename from frndint<mode>2_<rounding>.
2305 Do not use X87MODEF mode macro.
2306 (frndintxf2_<rounding>_i387): Rename from
2307 frndint<mode>2_<rounding>_i387. Do not use X87MODEF mode macro.
2308 (<rounding_insn><mode>2): For non-SSE modes, extend operand 1
2309 to XFmode and generate significandxf3 insn.
2311 2018-09-17 Richard Biener <rguenther@suse.de>
2313 PR tree-optimization/87328
2314 * tree-ssa-sccvn.c (process_bb): Remove assertion about not
2315 visiting unexecutable backedges when not iterating.
2316 (do_rpo_vn): Mark all edges not executable even when not
2319 2018-09-17 Martin Jambor <mjambor@suse.cz>
2322 * doc/invoke.texi (Warning Options): Likewise.
2324 2018-09-17 Richard Biener <rguenther@suse.de>
2326 PR tree-optimization/87301
2327 * tree-ssa-sccvn.c (eliminate_dom_walker::eliminate_cleanup): Properly
2328 clean EH info from leftover copy assignments.
2330 2018-09-17 Martin Liska <mliska@suse.cz>
2332 PR gcov-profile/85871
2333 * gcov.c (output_intermediate_file): Fix out of bounds
2336 2018-09-17 Vineet Gupta <vgupta@synopsys.com>
2338 * config/arc/arc.c: Object attributes for core4 not reflected
2340 * config/arc/arc.h: Don't restrict DBNZ to core3 (core4 includes
2343 2018-09-17 Alexey Brodkin <abrodkin@synopsys.com>
2345 * config/arc/linux.h (LINK_EH_SPEC): Add missing space.
2347 2018-09-17 Martin Liska <mliska@suse.cz>
2349 * doc/gcov.texi: Document new option --use-hotness-colors.
2350 * gcov.c (struct source_info): Declare new field.
2351 (source_info::source_info): Set default for maximum_count.
2352 (print_usage): Add new -q option.
2353 (process_args): Process it.
2354 (accumulate_line_info): Save src->maximum_count.
2355 (output_line_beginning): Make color line number if
2356 flag_use_hotness_colors is set.
2357 (output_line_details): Pass default argument value.
2358 (output_lines): Pass src->maximum_count.
2360 2018-09-17 Martin Liska <mliska@suse.cz>
2362 * common/config/i386/i386-common.c (ix86_get_valid_option_values):
2363 Use processor_names table.
2364 * config/i386/i386.c (ix86_default_align): Use
2365 processor_cost_table for alignment values.
2366 (ix86_option_override_internal): Use processor_names.
2367 (ix86_function_specific_print): Likewise.
2368 * config/i386/i386.h (struct processor_costs):
2369 Add alignment values.
2370 (struct ptt): Remove and replace with const char *.
2371 * config/i386/x86-tune-costs.h (struct processor_costs):
2372 Declare default alignments for all costs.
2374 2018-09-17 Aldy Hernandez <aldyh@redhat.com>
2376 * tree-vrp.c (extract_range_from_unary_expr): Do not special case
2377 symbolics or VR_VARYING ranges for ABS_EXPR.
2378 * wide-int-range.cc (wide_int_range_abs): Return positive numbers
2379 when range will wrap.
2381 2018-09-15 Eric Botcazou <ebotcazou@adacore.com>
2384 * cfgexpand.c (expand_gimple_basic_block): Be prepared for a BARRIER
2385 before and after a JUMP_TABLE_DATA.
2387 2018-09-14 John David Anglin <danglin@gcc.gnu.org>
2390 * dojump.c (do_compare_and_jump): Canonicalize function pointers
2391 when one operand is a function pointer. Use POINTER_TYPE_P and
2392 FUNC_OR_METHOD_TYPE_P.
2393 * expr.c (do_store_flag): Use POINTER_TYPE_P and FUNC_OR_METHOD_TYPE_P.
2394 * fold-const.c (build_range_check): Likewise.
2395 * match.pd (simple_comparison): Likewise.
2397 2018-09-14 David Malcolm <dmalcolm@redhat.com>
2400 * spellcheck.c (get_edit_distance_cutoff): New function.
2401 (selftest::test_edit_distance_unit_test_oneway): Rename to...
2402 (selftest::test_get_edit_distance_one_way): ...this.
2403 (selftest::test_get_edit_distance_unit): Rename to...
2404 (selftest::test_get_edit_distance_both_ways): ...this.
2405 (selftest::test_edit_distances): Move tests to this new function,
2406 and test some more pairs of strings. Update for above renaming.
2407 (selftest::get_old_cutoff): New function.
2408 (selftest::test_get_edit_distance_cutoff): New function.
2409 (selftest::assert_suggested_for): New function.
2410 (ASSERT_SUGGESTED_FOR): New macro.
2411 (selftest::assert_not_suggested_for): New function.
2412 (ASSERT_NOT_SUGGESTED_FOR): New macro.
2413 (selftest::test_suggestions): New function.
2414 (selftest::spellcheck_c_tests): Move test_get_edit_distance_unit
2415 tests to selftest::test_edit_distances and call it. Add calls to
2416 selftest::test_get_edit_distance_cutoff and
2417 selftest::test_suggestions.
2418 * spellcheck.h (get_edit_distance_cutoff): New function declaration.
2419 (best_match::consider): Replace hard-coded cutoff calculation with
2421 (best_match::get_cutoff): New declaration.
2422 (best_match::get_best_meaningful_candidate): Likewise.
2424 2018-09-14 Bernd Edlinger <bernd.edlinger@hotmail.de>
2426 * builtins.c (fold_builtin_strlen): Remove TODO comment.
2428 2018-09-14 Bernd Edlinger <bernd.edlinger@hotmail.de>
2431 2018-07-30 Bernd Edlinger <bernd.edlinger@hotmail.de>
2433 * tree-ssa-forwprop.c (simplify_builtin_call): Don't create a not NUL
2434 terminated string literal.
2436 2018-09-14 Martin Sebor <msebor@redhat.com>
2438 * builtins.c (unterminated_array): Handle ARRAY_REF.
2439 (expand_builtin_stpcpy_1): Detect unterminated char arrays.
2440 * builtins.h (unterminated_array): Declare extern.
2441 * gimple-fold.c (gimple_fold_builtin_stpcpy): Detect unterminated
2443 (gimple_fold_builtin_sprintf): Propagate NO_WARNING to transformed
2446 2018-09-14 Martin Sebor <msebor@redhat.com>
2447 Jeff Law <law@redhat.com>
2449 * builtins.c (unterminated_array): New.
2450 (expand_builtin_strcpy): Adjust.
2451 (expand_builtin_strcpy_args): Detect unterminated arrays.
2452 * gimple-fold.c (get_maxval_strlen): Add argument. Detect
2453 unterminated arrays.
2454 * gimple-fold.h (get_maxval_strlen): Add argument.
2455 (gimple_fold_builtin_strcpy): Detec unterminated arrays.
2457 * gimple-fold.c (get_range_strlen): Add argument.
2458 (get_maxval_strlen): Adjust.
2459 * gimple-fold.h (get_range_strlen): Add argument.
2461 2018-09-14 Wei Xiao <wei3.xiao@intel.com>
2463 * config/i386/movdirintrin.h: Fix copyright year.
2465 2018-09-14 Uros Bizjak <ubizjak@gmail.com>
2467 * reg-stack.c: Include regs.h.
2468 (replace_reg): Assert that mode is MODE_FLOAT or MODE_COMPLEX_FLOAT.
2469 (emit_pop_insn): Default pop insn mode to the reg_raw_mode of
2470 FIRST_STACK_REG, not DFmode.
2471 (emit_swap_insn): Default swap insn mode to the reg_raw_mode of
2472 FIRST_STACK_REG, not XFmode. Explicitly construct swap RTX.
2473 (change stack): Default register mode to the reg_raw_mode of
2474 FIRST_STACK_REG, not DFmode.
2475 * config/i386/i386.md (*swap<mode>): Remove insn pattern.
2476 (*swapxf): Rename from swapxf.
2478 2018-09-14 Carl Love <cel@us.ibm.com>
2480 * config/rs6000/emmintrin.h: Add _MM_SHUFFLE2.
2481 * config/rs6000/xmmintrin.h: Add _MM_SHUFFLE.
2483 2018-09-14 Segher Boessenkool <segher@kernel.crashing.org>
2486 * config/rs6000/rs6000.md (*mov<mode>_hardfloat64): Add Z to the Y
2489 2018-09-14 Sam Tebbs <sam.tebbs@arm.com>
2492 * config/aarch64/aarch64.md (*aarch64_bfxilsi_uxtw): Define.
2494 2018-09-14 Jason Merrill <jason@redhat.com>
2496 Fix --enable-gather-detailed-mem-stats.
2497 * hash-table.c (hash_table_usage): Change from variable to function.
2498 * hash-table.h: Adjust.
2499 * Makefile.in: Add missing dependencies on hash-table.h.
2501 2018-09-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2503 PR tree-optimization/87259
2505 (pass_cse_reciprocals::execute): Run optimize_recip_sqrt after
2506 execute_cse_reciprocals_1 has tried transforming.
2508 2018-09-14 Aldy Hernandez <aldyh@redhat.com>
2510 * tree-vrp.c (extract_range_from_binary_expr_1): Normalize
2511 VR_VARYING for PLUS/MINUS_EXPR.
2513 2018-09-14 Ilya Leoshkevich <iii@linux.ibm.com>
2515 * config/s390/s390-passes.def (INSERT_PASS_BEFORE): Improve
2518 2018-09-14 Richard Biener <rguenther@suse.de>
2521 * tree-ssa-coalesce.c (ssa_conflicts_merge): Remove conflict
2522 bits for the merged partition.
2524 2018-09-13 Martin Sebor <msebor@redhat.com>
2525 Bernd Edlinger <bernd.edlinger@hotmail.de>
2527 * builtins.h (c_srlen): Add argument.
2528 * builtins.c (warn_string_no_nul): New function.
2529 (c_strlen): Add argument and use it. Update recursive calls.
2530 Pass DECL argument to string_constant to get info on non
2531 terminated strings. Update *NONSTR as needed.
2532 (fold_builtin_strlen): Add argument to calls to c_strlen.
2533 Warn for unterminated arrays.
2534 (warn_string_no_null): Add prototype.
2535 * expr.c (string_constant): Update arguments. Update recursive
2536 calls appropriately. Detect missing NUL terminator and outermost
2537 declaration its missing in.
2538 Improve checks for arrays with nonzero lower bound or elements
2539 that are not a single byte. Simplify offset computation.
2540 Simplify checks for non-NUL terminated strings.
2541 * gimple-fold.c (get_range_strlen): Add argument to c_strlen call.
2542 * gimple-ssa-sprintf.c (get_string_length): Remove unnecessary code.
2544 2018-09-13 Bernd Edlinger <bernd.edlinger@hotmail.de>
2546 * builtins.c (c_strlen): Handle not zero terminated STRING_CSTs
2548 * fold-const.c (c_getstr): Fix function comment. Remove unused third
2549 argument. Fix range checks.
2550 * fold-const.h (c_getstr): Adjust protoype.
2551 * gimple-fold.c (gimple_fold_builtin_memory_op): Avoid folding when
2552 string is constant but contains no NUL byte.
2554 * expr.c (string_constant): Adjust function comment.
2555 Remove bogus check for zero termination.
2557 * fold-const.c (c_getstr): Clamp STRING_LENGTH to STRING_SIZE.
2559 * varasm.c (compare_constant): Compare type size of STRING_CSTs.
2560 (get_constant_size): Don't make STRING_CSTs larger than they are.
2561 (check_string_literal): New check function for STRING_CSTs.
2562 (output_constant): Use it.
2564 2018-09-13 Eric Botcazou <ebotcazou@adacore.com>
2567 * config/visium/visium.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Define.
2569 2018-09-13 Richard Biener <rguenther@suse.de>
2571 PR tree-optimization/87263
2572 * tree-ssa-sccvn.c (visit_phi): Revert some earlier changes.
2573 (struct unwind_state): Add max_rpo field.
2574 (do_rpo_vn): Allow up-to-date loop state to be used when not iterating.
2575 Compute max_rpo, the max RPO number a block can be backwards reached
2576 from. Re-write non-iterating mode to a RPO ordered worklist approach,
2577 separating it from the iterating mode.
2579 2018-09-13 Vlad Lazar <vlad.lazar@arm.com>
2581 * haifa-sched.c (rank_for_schedule): Schedule by INSN_COST.
2582 (rfs_decision): New scheduling decision.
2584 2018-09-13 Richard Biener <rguenther@suse.de>
2587 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Fix assert.
2588 (vn_nary_op_insert_pieces_predicated): Do not write useless
2589 valid_dominated_by_p entry outside of the allocated storage.
2591 2018-09-13 Omar Sandoval <osandov@osandov.com>
2592 Tom de Vries <tdevries@suse.de>
2595 * dwarf2out.c (is_c): New function.
2596 (add_subscript_info): Add DW_AT_count of 0 for C zero-length arrays.
2598 2018-09-13 Sam Tebbs <sam.tebbs@arm.com>
2601 * config/aarch64/aarch64.md (*aarch64_bfxil):
2603 * config/aarch64/constraints.md (Ulc): Define.
2604 * config/aarch64/aarch64-protos.h (aarch64_high_bits_all_ones_p):
2606 * config/aarch64/aarch64.c (aarch64_high_bits_all_ones_p):
2609 2018-09-13 Vlad Lazar <vlad.lazar@arm.com>
2611 * config/aarch64/aarch64.h (TARGET_COMPUTE_FRAME_LAYOUT): Define.
2612 * config/aarch64/aarch64.c (aarch64_expand_prologue): Remove
2613 aarch64_layout_frame call.
2614 (aarch64_expand_epilogue): Likewise.
2615 (aarch64_initial_elimination_offset): Likewise.
2616 (aarch64_get_separate_components): Likewise.
2617 (aarch64_use_return_insn_p): Likewise.
2618 (aarch64_layout_frame): Remove unneeded check.
2620 2018-09-13 Jakub Jelinek <jakub@redhat.com>
2622 * configure.ac: Only append
2623 " : (reconfigured) $TOPLEVEL_CONFIGURE_ARGUMENTS" to
2624 gcc_config_arguments if it was never reconfigured or last reconfigure
2625 was with different arguments.
2626 * configure: Regenerated.
2628 2018-09-13 Jakub Jelinek <jakub@redhat.com>
2629 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2632 * expr.c (maybe_optimize_pow2p_mod_cmp): New function.
2633 (maybe_optimize_mod_cmp): Use it if integer_pow2p treeop1.
2635 2018-09-13 Jakub Jelinek <jakub@redhat.com>
2637 PR tree-optimization/87287
2638 * fold-const.c (fold_binary_loc) <case EQ_EXPR>: Move signed modulo
2639 X % C == 0 to X % (unsigned) C == 0 optimization to ...
2640 * match.pd (X % C == 0): ... here. New optimization.
2642 2018-09-12 Jakub Jelinek <jakub@redhat.com>
2645 * expr.h (maybe_optimize_mod_cmp): Declare.
2646 * expr.c (mod_inv): New function.
2647 (maybe_optimize_mod_cmp): New function.
2648 (do_store_flag): Use it.
2649 * cfgexpand.c (expand_gimple_cond): Likewise.
2651 2018-09-09 Cesar Philippidis <cesar@codesourcery.com>
2652 Julian Brown <julian@codesourcery.com>
2655 * gimplify.c (gimplify_scan_omp_clauses): Set
2656 target_firstprivatize_array_bases in OpenACC parallel and kernels
2657 region contexts. Remove GOMP_MAP_FIRSTPRIVATE_REFERENCE clauses from
2658 OpenACC data regions.
2660 2018-09-12 Uros Bizjak <ubizjak@gmail.com>
2662 * config/i386/i386.md (sqrt_extend<mode>xf3_i387): Remove.
2663 (sqrt<mode>2): Extend operand 1 to XFmode and generate sqrtxf3 insn.
2665 2018-09-12 Richard Biener <rguenther@suse.de>
2667 PR tree-optimization/87280
2668 * tree-ssa-sccvn.c (process_bb): Handle the case of executable
2669 edge but unreachable target.
2670 (do_rpo_vn): For conservatively handling a PHI only mark
2671 the backedge executable but not the block reachable.
2673 2018-09-12 Richard Biener <rguenther@suse.de>
2675 PR tree-optimization/87266
2676 * tree-ssa-sccvn.c (do_rpo_vn): Always iterate to not yet
2679 2018-09-12 Andreas Krebbel <krebbel@linux.ibm.com>
2681 * config/s390/s390.md (PFPO_RND_MODE_DFP, PFPO_RND_MODE_BFP): New
2683 ("trunc<BFP:mode><DFP_ALL:mode>2")
2684 ("trunc<DFP_ALL:mode><BFP:mode>2")
2685 ("extend<BFP:mode><DFP_ALL:mode>2")
2686 ("extend<DFP_ALL:mode><BFP:mode>2"): Set proper rounding mode
2687 according to the target operand type.
2689 2018-09-12 Jakub Jelinek <jakub@redhat.com>
2690 Andreas Krebbel <krebbel@linux.ibm.com>
2692 PR tree-optimization/86844
2693 * gimple-ssa-store-merging.c
2694 (imm_store_chain_info::coalesce_immediate): For overlapping stores, if
2695 there are any overlapping stores in between them, make sure they are
2696 also coalesced or we give up completely.
2698 2018-09-12 Jakub Jelinek <jakub@redhat.com>
2701 * fold-const.c (fold_ternary_loc) <case COND_EXPR>: Verify also that
2702 BIT_AND_EXPR's second operand is a power of two. Formatting fix.
2704 2018-09-12 Tom de Vries <tdevries@suse.de>
2706 * common.opt (gdescribe-dies): Add option.
2707 * dwarf2out.c (add_name_and_src_coords_attributes): Add description
2708 attribute for artifical and nameless decls.
2709 (dwarf2out_register_external_die): Add description attribute to
2710 external reference die.
2711 (add_desc_attribute): New functions.
2712 (gen_subprogram_die): Add description attribute to
2713 DW_TAG_call_site_parameter.
2714 * tree-pretty-print.c (print_generic_expr_to_str): New function.
2715 * tree-pretty-print.h (print_generic_expr_to_str): Declare.
2716 * doc/invoke.texi (@item Debugging Options): Add -gdescribe-dies and
2718 (@item -gdescribe-dies): Add.
2720 2018-09-12 Aldy Hernandez <aldyh@redhat.com>
2722 * tree-vrp.c (vrp_shift_undefined_p): Remove.
2723 (extract_range_from_binary_expr_1: Call
2724 wide_int_range_shift_undefined_p instead of vrp_shift_undefined_p.
2725 * wide-int-range.h (wide_int_range_shift_undefined_p): Do not
2728 2018-09-12 Aldy Hernandez <aldyh@redhat.com>
2730 * gimple-ssa-warn-alloca.c
2731 (alloca_type_and_limit::alloca_type_and_limit): Initialize limit
2732 field for ALLOCA_BOUND_*_LARGE.
2734 2018-09-11 Nathan Sidwell <nathan@acm.org>
2736 * gcc.c (load_specs, execute, run_attempt): Use %qs not '%s'.
2738 2018-09-11 Uros Bizjak <ubizjak@gmail.com>
2740 * reg-stack.c (subst_asm_stack_regs): Call replace_reg also
2741 for clobbers. Remove obsolete comment.
2743 2018-09-11 Uros Bizjak <ubizjak@gmail.com>
2745 * config/i386/i386.md (define_attr "type"): Remove mpxmov, mpxmk,
2746 mpxchk, mpxld and mpxst types.
2747 (define_attr length_immediate): Remove all processing of mpx types.
2748 (define_attr prefix_0f): Ditto.
2749 (define_attr memory): Ditto.
2751 2018-09-11 Uros Bizjak <ubizjak@gmail.com>
2753 * config/i386/i386.md (fyl2x_extend<mode>xf3_i387): Remove.
2754 (log<mode>2): Change operand 1 predicate to general_operand.
2755 Extend operand 1 to XFmode and generate logxf3 insn.
2756 (log10<mode>2): Change operand 1 predicate to general_operand.
2757 Extend operand 1 to XFmode and generate log10xf3 insn.
2758 (log2<mode>2): Change operand 1 predicate to general_operand.
2759 Extend operand 1 to XFmode and generate log2xf3 insn.
2760 (fyl2xp1_extend<mode>xf3_i387): Remove.
2761 (log1p<mode>2): Change operand 1 predicate to general_operand.
2762 Extend operand 1 to XFmode and generate log1pxf3 insn.
2763 (fxtract_extend<mode>xf3_i387): Remove.
2764 (logb<mode>2): Change operand 1 predicate to general_operand.
2765 Extend operand 1 to XFmode and generate logbxf3 insn.
2766 (ilogb<mode>2): Change operand 1 predicate to general_operand.
2767 Extend operand 1 to XFmode and generate fxtractxf3_i387 insn.
2768 (significand<mode>2): Change operand 1 predicate to general_operand.
2769 Extend operand 1 to XFmode and generate significandxf3 insn.
2771 2018-09-11 Nathan Sidwell <nathan@acm.org>
2773 * gcc.c (perror_with_name, pfatal_with_name): Delete.
2774 (load_specs): Use fatal_error.
2775 (DELETE_IF_ORDINARY, process_command): Use error.
2776 (execute, run_attempt): Use fatal_error.
2778 2018-09-11 Andrew Stubbs <ams@codesourcery.com>
2780 * diagnostic-core.h (sorry_at): New prototype.
2781 * diagnostic.c (sorry_at): New function.
2783 2018-09-11 Aldy Hernandez <aldyh@redhat.com>
2785 * tree-vrp.c (extract_range_from_binary_expr_1): Treat all divisions
2786 by zero as VR_UNDEFINED.
2788 2018-09-10 Uros Bizjak <ubizjak@gmail.com>
2790 * config/i386/i386.md (<sincos>xf2): Rename from *<sincos>xf2_i387.
2791 (*<sincos>_extend<mode>xf2_i387): Remove insn pattern.
2792 (<sincos>mode2): New expander.
2793 (sincos_extend<mode>xf3_i387): Remove insn pattern.
2794 (sincos -> sin, cos splitters): Remove splitter patterns.
2795 (sincos<mode>3): Change operand 2 predicate to general_operand.
2796 Extend operand 2 to XFmode and generate sincosxf3 insn.
2797 (fptanxf4_i387): Change mode of operands 0 and 3 to SFmode.
2798 Change operand 3 predicate to const1_operand.
2799 (fptan_extend<mode>xf4_i387): Remove insn pattern.
2800 (tanxf2): Update operands in the call to fptanxf4_i387.
2801 (tan<mode>2): Change operand 1 predicate to general_operand.
2802 Extend operand 1 to XFmode and generate tanxf3 insn.
2803 (atan2xf3): Rename from *fpatanxf3_i387.
2804 (fpatan_extend<mode>xf3_i387): Remove insn pattern.
2805 (atan2xf3): Remove expander.
2806 (atan2<mode<3): Change operand 1 and 2 predicates to general_operand.
2807 Extend operands 1 and 2 to XFmode and generate atan2xf3 insn.
2808 (atan<mode>2): Change operand 1 predicate to general_operand.
2809 Extend operand 1 to XFmode and generate atanxf3 insn.
2811 2018-09-10 Uros Bizjak <ubizjak@gmail.com>
2813 * config/i386/i386.md (x87/SSE constant load splitter): Use
2814 memory_operand instead of nonimmediate_operand for input operand
2817 2018-09-09 Uros Bizjak <ubizjak@gmail.com>
2819 * config/i386/i386.md (float partial SSE register stall splitter): Move
2820 splitter near its instruction pattern.
2821 (float_extend partial SSE register stall splitter): Ditto.
2822 (float_truncate partial SSE register stall splitter): Ditto.
2824 2018-09-09 Hans-Peter Nilsson <hp@bitrange.com>
2827 * config/mmix/mmix.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Redefine
2828 to speculation_safe_value_not_needed.
2831 * config/mmix/mmix.c (mmix_assemble_integer): Handle byte-size
2832 non-CONST_INT rtx:es using assemble_integer_with_op ".byte".
2833 (MMIX_CFUN_NEEDS_SAVED_EH_RETURN_ADDRESS): Don't call
2834 leaf_function_p, instead use has_hard_reg_initial_val.
2836 2018-09-09 Nathan Sidwell <nathan@acm.org>
2838 * gcc.h (pfatal_with_name): Don't declare here.
2839 * gcc.c (pfatal_with_name): Make static.
2841 2018-09-09 Xianmiao Qu <xianmiao_qu@c-sky.com>
2843 * config/csky/csky.md (*cskyv2_adddi3, *cskyv2_subdi3): Add
2846 2018-09-08 John David Anglin <danglin@gcc.gnu.org>
2848 PR rtl-optimization/85458
2849 * sel-sched.c (sel_target_adjust_priority): Allow backend adjust
2850 priority hook to reduce the priority of EXPR.
2852 2018-09-07 Uros Bizjak <ubizjak@gmail.com>
2854 * config/i386/i386.md (float<SWI48x:mode><MODEF:mode>2) Enable
2855 DImode for x87 on 32bit targets. Conditionally disable x87 modes
2856 with X87_ENABLE_FLOAT. Remove preparation code.
2857 (*float<SWI48:mode><MODEF:mode>2): Rename from
2858 *float<SWI48:mode><MODEF:mode>2_mixed. Handle x87, SSE and mixed
2859 math using "enabled" attribute.
2860 (*floatdi<MODEF:mode>2_i387): Rename from
2861 *float<SWI48x:mode><MODEF:mode>2_i387. Handle only DImode and
2862 enable for 32bit targets only.
2863 (floatdi<X87MODEF:mode>2_i387_with_xmm pre-reload splitter): New
2865 (floatdi<X87MODEF:mode>2_i387_with_xmm): Use register_operand
2866 as operand 1 predicate. Rewrite as define_insn_and_split.
2867 (floatdi<X87MODEF:mode>2_i387_with_xmm memory input splitter): Remove.
2869 2018-09-06 Uros Bizjak <ubizjak@gmail.com>
2871 * reg-stack.c (get_true_reg) <case FLOAT_TRUNCATE>: Reorder
2872 to fallthru to FLOAT case.
2874 2018-09-06 Will Schmidt <will_schmidt@vnet.ibm.com>
2877 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Update logic
2878 around folding of vec_sl to handle out of range shift values.
2880 2018-09-06 Uros Bizjak <ubizjak@gmail.com>
2882 * config/i386/i386.md (fix_trunc<mode>_fisttp_i387_1): Remove.
2883 Update callers to gen_fix_trunc<mode>_i387_fisttp
2884 (fix_trunc<mode>_i387_fisttp): Change operand 0 predicate to
2885 nonimmediate_operand.
2886 (fix_trunc<mode>_i387_fisttp_with_temp): Remove insn pattern
2887 and corresponding splitters.
2888 (*fix_trunc<mode>_i387_1): Always expand with fix_trunc<mode>_i387.
2889 (fix_truncdi_i387): Change operand 0 predicate to nonimmediate_operand.
2890 (fix_truncdi_i387_with_temp): Remove insn pattern
2891 and corresponding splitters.
2892 (fix_trunc<mode>_i387): Change operand 0 predicate to
2893 nonimmediate_operand.
2894 (fix_trunc<mode>_i387_with_temp): Remove insn pattern
2895 and corresponding splitters.
2896 (*fistdi2_1): Remove.
2898 (fistdi2_with_temp): Remove insn pattern and corresponding splitters.
2899 (lrintxfdi2): Remove expander. Reimplement as define_insn.
2900 (*fist<mode>2_1): Remove.
2901 (fist<mode>2): Ditto.
2902 (fist<mode>2_with_temp): Remove insn pattern and corresponding
2904 (lrintxf<mode>2): Remove expander. Reimplement as define_insn.
2905 (*fist<mode>2_<rounding>_1): Always expand with fist<mode>2_<rounding>.
2906 (fistdi2_<rounding>): Change operand 0 predicate to
2907 nonimmediate_operand.
2908 (fistdi2_<rounding>_with_temp): Remove insn pattern
2909 and corresponding splitters.
2910 (fist<mode>2_<rounding>): Change operand 0 predicate to
2911 nonimmediate_operand.
2912 (fist<mode>2_<rounding>_with_temp): Remove insn pattern
2913 and corresponding splitters.
2915 (*fixuns_trunc<mode>si2_avx512f_zext): Depend on TARGET_SSE_MATH.
2917 2018-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
2919 * varasm.c (output_constructor_regular_field): Check TYPE_SIZE_UNIT of
2922 2018-09-06 Will Schmidt <will_schmidt@vnet.ibm.com>
2924 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
2925 early gimple folding of vec_splat().
2926 * tree-vect-generic.c: Remove static from tree_vec_extract() definition.
2927 * gimple-fold.h: Add an extern define for tree_vec_extract().
2929 2018-09-06 Will Schmidt <will_schmidt@vnet.ibm.com>
2931 * config/rs6000/rs6000.c (fold_mergehl_helper): Add types_compatible_p
2932 wrappers around TREE_TYPE comparisons.
2934 2018-09-06 Ilya Leoshkevich <iii@linux.ibm.com>
2937 * config/s390/predicates.md: Add nonsym_memory_operand.
2938 * config/s390/s390.c (s390_legitimize_cs_operand): If operand
2939 contains a SYMBOL_REF, load it into an intermediate pseudo.
2940 (s390_emit_compare_and_swap): Legitimize operand.
2941 * config/s390/s390.md: Use the new nonsym_memory_operand
2942 with UNSPECV_CAS patterns.
2944 2018-09-06 Ilya Leoshkevich <iii@linux.ibm.com>
2947 * config/s390/s390-passes.def: New file.
2948 * config/s390/s390-protos.h (class rtl_opt_pass): Add forward
2950 (make_pass_s390_early_mach): Add declaration.
2951 * config/s390/s390.c (make_pass_s390_early_mach):
2952 (s390_option_override): Remove dynamic registration.
2953 * config/s390/t-s390: Add s390-passes.def.
2955 2018-09-06 Ilya Leoshkevich <iii@linux.ibm.com>
2957 * config/s390/s390.c (s390_decompose_constant_pool_ref):
2958 Remove UNSPEC_LTREL_BASE check.
2959 (annotate_constant_pool_refs): Likewise.
2960 (find_constant_pool_ref): Likewise.
2961 (find_ltrel_base): Removed.
2962 (replace_ltrel_base): Removed.
2963 (s390_mainpool_finish): Remove replace_ltrel_base call.
2964 (s390_chunkify_start): Remove pending LTREL_BASE logic.
2965 (s390_chunkify_finish): Remove replace_ltrel_base call.
2966 * config/s390/s390.md: Remove UNSPEC_LTREL_BASE.
2968 2018-09-06 Hans-Peter Nilsson <hp@axis.com>
2971 * config/cris/cris.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Redefine
2972 to speculation_safe_value_not_needed.
2974 2018-09-05 Cesar Philippidis <cesar@codesourcery.com>
2975 Bernd Schmidt <bernds_cb1@t-online.de>
2977 * config/nvptx/nvptx-opts.h: New file.
2978 * config/nvptx/nvptx.c (nvptx_file_start): Print the correct .target.
2979 * config/nvptx/nvptx.h: Include "nvptx-opts.h".
2981 (TARGET_SM35): New macro.
2982 * config/nvptx/nvptx.md (atomic_fetch_<logic><mode>): Enable with the
2984 * config/nvptx/nvptx.opt (ptx_isa, sm_30, sm_35): New enum and its
2986 (misa=): New option.
2987 * doc/invoke.texi (Nvidia PTX Options): Document -misa.
2989 2018-09-05 Uros Bizjak <ubizjak@gmail.com>
2991 * config/i386/i386.md (truncdfsf2): Remove expander.
2992 (truncdfsf2_with_temp): Ditto.
2993 (truncxf<mode>2): Ditto.
2994 (*truncdfsf_fast_mixed): Remove insn pattern.
2995 (*truncdfsf_fast_i387): Ditto.
2996 (*truncdfsf_mixed): Ditto.
2997 (*truncdfsf_i387): Ditto.
2998 (*truncdfsf2_i387_1): Ditto.
2999 (*truncxfsf2_mixed): Ditto.
3000 (*truncxfdf2_mixed): Ditto.
3001 (*truncxf<mode>2_i387_noop): Ditto. Update callers
3002 to call gen_truncxf<mode>2 instead.
3003 (*truncxf<mode>2_i387): Remove.
3004 (reg->reg splitters): Remove splitter pattern.
3005 (reg->mem splitters): Ditto.
3007 (truncdfsf2): New insn pattern.
3008 (truncxf<mode>2): Ditto.
3010 2018-09-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3012 * tree-ssa-math-opts.c (is_mult_by): New function.
3013 (is_square_of): Use the above.
3014 (optimize_recip_sqrt): New function.
3015 (pass_cse_reciprocals::execute): Use the above.
3017 2018-09-05 Richard Biener <rguenther@suse.de>
3020 * tree-ssa-sccvn.c (rpo_elim::eliminate_push_avail): Make sure
3021 to zero-init the emplaced vec.
3023 2018-09-05 Martin Liska <mliska@suse.cz>
3025 PR tree-optimization/87205
3026 * tree-switch-conversion.c (pass_lower_switch::execute):
3027 Group cases for switch statements.
3029 2018-09-05 Richard Biener <rguenther@suse.de>
3031 PR tree-optimization/87217
3032 * tree-ssa-sccvn.c (vuse_valueize): New.
3033 (vn_reference_lookup_pieces): Use it.
3034 (vn_reference_lookup): Likewise.
3036 2018-09-05 Nathan Sidwell <nathan@acm.org>
3039 * stor-layout.c (place_field): Scan forwards to check last
3040 bitfield when ms_bitfield_placement is in effect.
3042 2018-09-05 Richard Biener <rguenther@suse.de>
3045 * tree-vect-stmts.c (vectorizable_simd_clone_call): Fix bogus
3048 2018-09-05 Siddhesh Poyarekar <siddhesh@sourceware.org>
3049 Kugan Vivekanandarajah <kugan.vivekanandarajah@linaro.org>
3051 * config/aarch64/falkor-tag-collision-avoidance.c: New file.
3052 * config.gcc (extra_objs): Build it.
3053 * config/aarch64/t-aarch64 (falkor-tag-collision-avoidance.o):
3055 * config/aarch64/aarch64-passes.def
3056 (pass_tag_collision_avoidance): New pass.
3057 * config/aarch64/aarch64.c (qdf24xx_tunings): Add
3058 AARCH64_EXTRA_TUNE_RENAME_LOAD_REGS to tuning_flags.
3059 (aarch64_classify_address): Remove static qualifier.
3060 (aarch64_address_info, aarch64_address_type): Move to...
3061 * config/aarch64/aarch64-protos.h: ... here.
3062 (make_pass_tag_collision_avoidance): New function.
3063 * config/aarch64/aarch64-tuning-flags.def (rename_load_regs):
3066 2018-09-05 Martin Liska <mliska@suse.cz>
3068 * doc/gcov.texi: Update documentation of humar
3070 * gcov.c (format_count): Print one decimal place, it provides
3071 more fine number of situations like '1G' vs. '1.4G'.
3073 2018-09-05 Martin Liska <mliska@suse.cz>
3076 * config/rs6000/rs6000.opt: Mark the option as Deprecated.
3077 * optc-gen.awk: Allow 'Var' for Deprecated options in order
3078 to generate a MASK value.
3080 2018-09-04 H.J. Lu <hongjiu.lu@intel.com>
3083 * dwarf2out.c (based_loc_descr): Allow hard frame pointer even
3084 if frame pointer isn't used.
3085 (compute_frame_pointer_to_fb_displacement): Likewise.
3087 2018-09-04 Jakub Jelinek <jakub@redhat.com>
3090 * common/config/i386/i386-common.c (OPTION_MASK_ISA_XSAVEOPT_SET,
3091 OPTION_MASK_ISA_XSAVES_SET, OPTION_MASK_ISA_XSAVEC_SET): Use
3092 OPTION_MASK_ISA_XSAVE_SET instead of OPTION_MASK_ISA_XSAVE.
3093 (OPTION_MASK_ISA_XSAVE_UNSET): Add OPTION_MASK_ISA_XSAVES_UNSET
3094 and OPTION_MASK_ISA_XSAVEC_UNSET.
3096 2018-09-04 Max Filippov <jcmvbkbc@gmail.com>
3098 * config/xtensa/xtensa.c (xtensa_expand_atomic): Reorder AND and
3099 XOR operations in NAND case.
3101 2018-09-04 Aldy Hernandez <aldyh@redhat.com>
3103 * wide-int-range.cc (wide_int_range_convert): New.
3104 * wide-int-range.h (wide_int_range_convert): New.
3105 * tree-vrp.c (extract_range_from_unary_expr): Abstract wide int
3106 code into wide_int_range_convert.
3107 (extract_range_into_wide_ints): Do not munge anti range constants
3108 into the entire domain. Just return the range back.
3110 2018-09-04 Martin Liska <mliska@suse.cz>
3112 * genmatch.c (output_line_directive): Add new argument
3114 (dt_simplify::gen_1): Encapsulate dump within __builtin_expect.
3116 2018-09-04 Jonathan Wakely <jwakely@redhat.com>
3118 * doc/invoke.texi (Option Summary): Add whitespace.
3120 * doc/invoke.texi (Option Summary): Add -Waligned-new.
3122 2018-09-04 Richard Biener <rguenther@suse.de>
3124 PR tree-optimization/87211
3125 * tree-ssa-sccvn.c (visit_phi): When value-numbering to a
3126 backedge value we're supposed to treat as VARYING also number
3127 the PHI to VARYING in case it got a different value-number already.
3129 2018-09-04 Aldy Hernandez <aldyh@redhat.com>
3131 * tree-vrp.c (vrp_can_optimize_bit_op): Remove.
3132 (extract_range_from_binary_expr_1): Do not call
3133 vrp_can_optimize_bit_op.
3134 * wide-int-range.cc (wide_int_range_can_optimize_bit_op): Make
3136 (wide_int_range_get_mask_and_bounds): New.
3137 (wide_int_range_optimize_bit_op): New.
3138 (wide_int_range_bit_ior): Call wide_int_range_optimize_bit_op.
3139 (wide_int_range_bit_and): Same.
3140 * wide-int-range.h (wide_int_range_can_optimize_bit_op): Remove.
3141 (wide_int_range_optimize_bit_op): New.
3142 (wide_int_range_get_mask_and_bounds): New.
3144 2018-09-04 Richard Biener <rguenther@suse.de>
3146 PR tree-optimization/87176
3147 * tree-ssa-sccvn.c (visit_phi): Remove redundant allsame
3148 variable. When value-numbering a virtual PHI node make sure
3149 to not value-number to the backedge value.
3151 2018-09-04 Jonathan Wakely <jwakely@redhat.com>
3153 * doc/extend.texi (Long Long, Hex Floats): Document support for
3154 long long and hex floats in more recent versions of ISO C++.
3156 2018-09-03 Richard Biener <rguenther@suse.de>
3158 PR tree-optimization/87177
3159 * tree-ssa-sccvn.c (vuse_ssa_val): Revert previous change, keep
3162 2018-09-03 Alexander Monakov <amonakov@ispras.ru>
3164 * bb-reorder.c (edge_order): Convert to C-qsort-style
3165 tri-state comparator.
3166 (reorder_basic_blocks_simple): Change std::stable_sort to
3169 2018-09-03 Alexander Monakov <amonakov@ispras.ru>
3171 * tree-loop-distribution.c (offset_cmp): Convert to C-qsort-style
3172 tri-state comparator.
3173 (fuse_memset_builtins): Change std::stable_sort to gcc_stablesort.
3175 2018-09-03 Alexander Monakov <amonakov@ispras.ru>
3177 * sort.cc (struct sort_ctx): New field 'nlim'. Use it...
3178 (mergesort): ... here as maximum count for using netsort.
3179 (gcc_qsort): Set nlim to 3 if stable sort is requested.
3180 (gcc_stablesort): New.
3181 * system.h (gcc_stablesort): Declare.
3183 2018-09-03 Alexander Monakov <amonakov@ispras.ru>
3185 * sort.cc (gcc_qsort) [CHECKING_P]: Call qsort_chk.
3186 * system.h (qsort): Always redirect to gcc_qsort. Update comment.
3187 * vec.c (qsort_chk): Do not call gcc_qsort. Update comment.
3189 2018-09-03 Segher Boessenkool <segher@kernel.crashing.org>
3191 * config/rs6000/rs6000.md (*mov<mode>_hardfloat32): Remove %U from the
3192 lxsdx and stxsdx alternatives.
3193 (*mov<mode>_hardfloat64): Ditto.
3194 * config/rs6000/vsx.md (*vsx_extract_<mode>_store): Ditto.
3196 2018-09-03 Richard Biener <rguenther@suse.de>
3198 PR tree-optimization/87200
3199 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Valueize a
3202 2018-09-03 Martin Liska <mliska@suse.cz>
3204 PR tree-optimization/87201
3205 * tree-switch-conversion.c (switch_decision_tree::balance_case_nodes):
3206 Fix parenthesis in an expression.
3208 2018-09-03 Richard Biener <rguenther@suse.de>
3210 PR tree-optimization/87197
3211 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Mark the new def
3212 visited. CSE the VN_INFO hashtable lookup.
3214 PR tree-optimization/87169
3215 * tree-ssa-sccvn.c (do_rpo_vn): When marking loops for not
3216 iterating make sure there's no extra backedges from irreducible
3217 regions feeding the header. Mark the destination block
3220 2018-09-03 Martin Liska <mliska@suse.cz>
3223 * common/common-target.def: Add TARGET_GET_VALID_OPTION_VALUES.
3224 * common/common-targhooks.c (default_get_valid_option_values):
3226 * common/common-targhooks.h (default_get_valid_option_values):
3228 * common/config/i386/i386-common.c: Move processor_target_table
3230 (ix86_get_valid_option_values): New function.
3231 (TARGET_GET_VALID_OPTION_VALUES): New macro.
3232 * config/i386/i386.c (struct ptt): Move to i386-common.c.
3233 (PTA_*): Move all defined masks into i386-common.c.
3234 (ix86_function_specific_restore): Use new processor_cost_table.
3235 * config/i386/i386.h (struct ptt): Moved from i386.c.
3236 (struct pta): Likewise.
3237 * doc/tm.texi: Document new TARGET_GET_VALID_OPTION_VALUES.
3238 * doc/tm.texi.in: Likewise.
3239 * opt-suggestions.c (option_proposer::suggest_option):
3240 Pass prefix to build_option_suggestions.
3241 (option_proposer::get_completions): Likewise.
3242 (option_proposer::build_option_suggestions): Use the new target
3244 * opts.c (struct option_help_tuple): New struct.
3245 (print_filtered_help): Use the new target hook.
3247 2018-09-03 Martin Liska <mliska@suse.cz>
3250 * predict.c (set_even_probabilities): Add likely_edges
3251 argument and handle cases where we have precisely one
3253 (combine_predictions_for_bb): Catch also likely_edges.
3254 (tree_predict_by_opcode): Handle gswitch statements.
3255 * tree-cfg.h (find_case_label_for_value): New declaration.
3256 (find_taken_edge_switch_expr): Likewise.
3257 * tree-switch-conversion.c (switch_decision_tree::balance_case_nodes):
3258 Find pivot in decision tree based on probabily, not by number of
3261 2018-09-02 Gerald Pfeifer <gerald@pfeifer.com>
3263 * doc/standards.texi (Standards): Update Objective-C reference.
3265 2018-09-01 Gerald Pfeifer <gerald@pfeifer.com>
3267 * doc/install.texi (Prerequisites): Update link for MPC.
3269 2018-09-01 Michael Matz <matz@suse.de>
3271 PR tree-optimization/87074
3272 * gimple-loop-jam.c (unroll_jam_possible_p): Check loop exit
3273 PHIs for outer-loop uses.
3275 2018-09-01 Gerald Pfeifer <gerald@pfeifer.com>
3277 * doc/generic.texi (OpenMP): Adjust link to openmp.org.
3278 * doc/invoke.texi (C Dialect Options): Ditto.
3280 2018-09-01 Gerald Pfeifer <gerald@pfeifer.com>
3282 * doc/install.texi (Prerequisites): Adjust link mpfr.org.
3284 2018-08-31 Richard Biener <rguenther@suse.de>
3286 PR tree-optimization/87168
3287 * tree-ssa-sccvn.c (SSA_VAL): Add visited output parameter.
3288 (rpo_elim::eliminate_avail): When OP was not visited it must
3291 2018-08-31 David Malcolm <dmalcolm@redhat.com>
3293 * tree-vrp.c (copy_value_range): Convert param "from" from
3294 "value_range *" to "const value_range *".
3295 (range_is_null): Likewise for param "vr".
3296 (range_int_cst_p): Likewise.
3297 (range_int_cst_singleton_p): Likewise.
3298 (symbolic_range_p): Likewise.
3299 (value_ranges_intersect_p): Likewise for both params.
3300 (value_range_nonnegative_p): Likewise for param "vr".
3301 (value_range_constant_singleton): Likewise.
3302 (vrp_set_zero_nonzero_bits): Likewise for param "ar".
3303 (extract_range_into_wide_ints): Likewise for param "vr".
3304 (extract_range_from_multiplicative_op): Likewise for params "vr0"
3306 (vrp_can_optimize_bit_op): Likewise.
3307 (extract_range_from_binary_expr_1): Likewise for params "vr0_" and
3309 (extract_range_from_unary_expr): Likewise.
3310 (debug_value_range): Likewise for param "vr".
3311 (value_range::dump): Add "const" qualifier.
3312 (vrp_prop::check_array_ref): Convert local "vr" from
3313 "value_range *" to "const value_range *".
3314 (vrp_prop::check_mem_ref): Likewise.
3315 (vrp_prop::visit_stmt): Likewise for local "old_vr".
3316 (vrp_intersect_ranges_1): Likewise for param "vr_1".
3317 (vrp_intersect_ranges): Likewise.
3318 (simplify_stmt_for_jump_threading): Likewise for local "vr".
3319 (vrp_prop::vrp_finalize): Likewise.
3320 * tree-vrp.h (value_range::dump): Add "const" qualifier.
3321 (vrp_intersect_ranges): Add "const" qualifier to params as above.
3322 (extract_range_from_unary_expr): Likewise.
3323 (value_range_constant_singleton): Likewise.
3324 (symbolic_range_p): Likewise.
3325 (copy_value_range): Likewise.
3326 (extract_range_from_binary_expr_1): Likewise.
3327 (range_int_cst_p): Likewise.
3328 (vrp_set_zero_nonzero_bits): Likewise.
3329 (range_int_cst_singleton_p): Likewise.
3331 2018-08-31 Vlad Lazar <vlad.lazar@arm.com>
3333 * config/aarch64/arm_neon.h (vabsd_s64): New.
3334 (vnegd_s64): Likewise.
3336 2018-08-31 Martin Jambor <mjambor@suse.cz>
3338 * ipa-cp.c (estimate_local_effects): Replace wrong MAX with MIN.
3340 2018-08-31 Martin Liska <mliska@suse.cz>
3342 * ipa-icf.c (sem_item::add_type): Use
3343 sem_item::m_type_hash_cache.
3344 * ipa-icf.h: Move the cache from sem_item_optimizer
3347 2018-08-31 Nathan Sidwell <nathan@acm.org>
3349 * doc/extend.texi (Backwards Compatibility): Remove implicit
3350 extern C leeway of () being (...).
3352 2018-08-31 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3354 * ipa-inline.c (can_inline_edge_by_limits_p): Fix typos in comment.
3356 2018-08-31 Segher Boessenkool <segher@kernel.crashing.org>
3360 * config/rs6000/rs6000.md (lround<mode>di2): Gate on TARGET_FPRND.
3362 2018-08-31 Jakub Jelinek <jakub@redhat.com>
3365 * expmed.c (expand_mult_const): Use immed_wide_int_const instead of
3366 gen_int_mode. Formatting fixes.
3368 2018-08-30 Sandra Loosemore <sandra@codesourcery.com>
3370 * target.def (custom_function_descriptors): Improve documentation.
3371 * doc/tm.texi.in (Trampolines): Expand discussion of function
3372 descriptors and move TARGET_CUSTOM_FUNCTION_DESCRIPTORS to the
3373 beginning of the section.
3374 * doc/tm.texi: Regenerated.
3376 2018-08-30 Jose E. Marchesi <jose.marchesi@oracle.com>
3378 * cfg.h (class auto_edge_flag): Spell out the template-id of the
3379 base class in the initializer list. This is a workaround for
3380 building with older GCC.
3381 (class auto_bb_flag): Likewise.
3383 2018-08-30 Aaron Sawdey <acsawdey@linux.ibm.com>
3385 * config/rs6000/altivec.md (altivec_eq<mode>): Remove star.
3386 (altivec_vcmpequ<VI_char>_p): Remove star.
3387 * config/rs6000/rs6000-string.c (do_load_for_compare): Support
3389 (expand_strncmp_vec_sequence): New function.
3390 (emit_final_str_compare_vec): New function.
3391 (expand_strn_compare): Add support for vector strncmp.
3392 * config/rs6000/rs6000.opt (-mstring-compare-inline-limit): Change
3393 length specification to bytes.
3394 * config/rs6000/vsx.md (vsx_ld_elemrev_v16qi_internal): Remove star.
3395 (vcmpnezb_p): New pattern.
3396 * doc/invoke.texi (RS/6000 and PowerPC Options): Update documentation
3397 for option -mstring-compare-inline-limit.
3399 2018-08-30 Thiago Macieira <thiago.macieira@intel.com>
3401 * config/i386/i386.c (PTA_WESTMERE): Remove PTA_AES.
3402 (PTA_SKYLAKE): Add PTA_AES.
3403 (PTA_GOLDMONT): Likewise.
3405 2018-08-29 Jan Hubicka <jh@suse.cz>
3408 * lto-opts.c (lto_write_options): Always stream PIC/PIE mode.
3409 * lto-wrapper.c (merge_and_complain): Fix merging of PIC/PIE.
3411 2018-08-29 Jan Hubicka <jh@suse.cz>
3413 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not follow
3415 (hash_tree): Do not visit TYPE_STUB_DECL.
3416 * tree-streamer-out.c (write_ts_type_common_tree_pointers): Do not
3417 stream TYPE_STUB_DECL.
3418 * tree-streamer-in.c (lto_input_ts_type_common_tree_pointers): Likewise.
3419 * ipa-utils.h (type_with_linkage_p): Do not rely on TYPE_STUB_DECL
3420 after free_lang_data.
3421 (type_in_anonymous_namespace_p): Likewise.
3423 2018-08-29 Jan Hubicka <jh@suse.cz>
3425 * sreal.h (SREAL_PART_BITS): Change to 31; remove seemingly unnecessary
3426 comment that it has to be even number.
3427 (class sreal): Change m_sig type to int32_t.
3428 * sreal.c (sreal::dump, sreal::to_int, opreator+, operator-): Use
3429 int64_t for temporary calculations.
3430 (sreal_verify_basics): Drop one bit from minimum and maximum.
3432 2018-08-30 Richard Biener <rguenther@suse.de>
3434 PR tree-optimization/87147
3435 * tree-ssa-sccvn.c (SSA_VISITED): New function.
3436 (visit_phi): When the degenerate result is from the backedge and
3437 we didn't visit its definition yet drop to VARYING.
3438 (do_rpo_vn): Properly mark blocks with incoming backedges as executable.
3440 2018-08-29 Jan Hubicka <jh@suse.cz>
3442 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not walk
3444 (hash_tree): Likewise.
3446 2018-08-29 Jan Hubicka <jh@suse.cz>
3448 * tree.c (find_decls_types_r): Walk also TYPE_NEXT_PTR_TO
3449 and TYPE_NEXT_REF_TO.
3451 2018-08-29 Jan Hubicka <jh@suse.cz>
3453 * sreal.h (SREAL_PART_BITS): Change to 31; remove seemingly unnecessary
3454 comment that it has to be even number.
3455 (class sreal): Change m_sig type to int32_t.
3456 * sreal.c (sreal::dump, sreal::to_int, opreator+, operator-): Use
3457 int64_t for temporary calculations.
3458 (sreal_verify_basics): Drop one bit from minimum and maximum.
3460 2018-08-30 Tamar Christina <tamar.christina@arm.com>
3462 * config/aarch64/aarch64.c (aarch64_expand_movmem): Set TImode max.
3464 2018-08-30 Vlad Lazar <vlad.lazar@arm.com>
3467 * expmed.c (canonicalize_comparison): Use wi::sub instead of wi::add
3468 if to_add is negative.
3470 2018-08-29 Bernd Edlinger <bernd.edlinger@hotmail.de>
3473 * builtins.c (c_strlen): Improve range checks.
3475 2018-08-29 Martin Sebor <msebor@redhat.com>
3476 Jeff Law <law@redhat.com>
3478 PR tree-optimization/86714
3479 PR tree-optimization/86711
3480 * builtins.c (c_strlen): Add arguments to call to string_constant.
3481 * expr.c (string_constant): Add argument. Detect missing nul
3482 terminator and outermost declaration it's missing in.
3483 * expr.h (string_constant): Add argument.
3484 * fold-const.c (read_from_constant_string): Add arguments to call to
3486 (c_getstr): Likewise.
3487 * tree-ssa-forwprop.c (simplify_builtin_call): Likewise.
3489 * tree-ssa-strlen.c (get_stridx): Likewise.
3491 2018-08-29 Jan Hubicka <jh@suse.cz>
3493 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers):
3494 Do not stream DECL_VINDEX.
3495 * tree-streamer-out.c (write_ts_function_decl_tree_pointers): Likewise.
3496 * tree.c (free_lang_data_in_decl): Clear DECL_VINDEX.
3497 (decl_function_context): Use DECL_VIRTUAL_P rather than DECL_VINDEX.
3499 2018-08-29 Richard Biener <rguenther@suse.de>
3501 * tree-ssa-sccvn.c (vuse_ssa_val): Return NULL for unvisited
3502 virtual operands that are not default defs to honor region
3504 (rpo_vn_valueize): Remove ineffective code here.
3506 2018-08-29 Richard Biener <rguenther@suse.de>
3508 PR tree-optimization/87132
3509 * tree-ssa-alias.c (get_continuation_for_phi): Do not translate
3510 when skipping defs reachable over backedges.
3512 2018-08-29 Richard Biener <rguenther@suse.de>
3514 * tree-core.h: Document use of deprecated_flag in SSA_NAME.
3515 * tree.h (SSA_NAME_POINTS_TO_READONLY_MEMORY): Define.
3516 * tree-into-ssa.c (pass_build_ssa::execute): Initialize
3517 function parameters SSA_NAME_POINTS_TO_READONLY_MEMORY from fnspec.
3518 * tree-ssa-sccvn.c (const_parms, init_const_parms): Remove.
3519 (vn_reference_lookup_3): Remove use of const_parms.
3520 (free_rpo_vn): Do not free const_parms.
3521 (do_rpo_vn): Do not call init_const_parms.
3522 * tree-ssa-alias.c (refs_may_alias_p_1): Honor
3523 SSA_NAME_POINTS_TO_READONLY_MEMORY.
3524 (call_may_clobber_ref_p_1): Likewise.
3526 2018-08-29 Alexander Monakov <amonakov@ispras.ru>
3529 * invoke.texi (Optimization Options): List -ftree-scev-cprop.
3531 (-ftree-scev-cprop): Document.
3533 2018-08-29 Jan Hubicka <jh@suse.cz>
3535 * sreal.h (normalize, normalize_up, normalize_down): Add new_sig/new_exp
3537 (sreal constructor): Update.
3538 * sreal.c (sreal:operator+, sreal:operator-, sreal:operator*,
3539 sreal:operator/): Update.
3541 2018-08-29 Martin Liska <mliska@suse.cz>
3543 * tree-switch-conversion.c (switch_conversion::expand):
3544 Strenghten assumption about gswitch statements.
3546 2018-08-29 Richard Biener <rguenther@suse.de>
3548 PR tree-optimization/87117
3549 * tree-ssa-sccvn.c (eliminate_dom_walker::eliminate_stmt): Only
3550 re-value-number released SSA VDEFs.
3552 2018-08-29 Richard Biener <rguenther@suse.de>
3554 PR tree-optimization/87126
3555 * tree-ssa-sccvn.c (vn_reference_insert): Remove assert.
3557 2018-08-28 Jim Wilson <jimw@sifive.com>
3559 * config/riscv/pic.md: Rewrite.
3560 * config/riscv/riscv.c (riscv_address_insns): Return cost of 3 for
3562 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Delete.
3563 (SOFTF, default_load, softload, softstore): New.
3565 2018-08-28 Jeff Law <law@redhat.com>
3567 * fold-const.c (fold_binary_loc): Remove recently added assert.
3569 2018-08-28 Joern Rennecke <joern.rennecke@riscy-ip.com>
3571 * genpreds.c (write_predicate_subfunction): Also add ATTRIBUTE_UNUSED
3572 to OP parmeter of generated function.
3574 2018-08-28 MCC CS <deswurstes@users.noreply.github.com>
3576 PR tree-optimization/87009
3577 * match.pd: Add boolean optimizations.
3579 2018-08-28 Martin Sebor <msebor@redhat.com>
3582 * calls.c (alloc_max_size): Treat HOST_WIDE_INT special.
3583 * gimple-ssa-warn-alloca.c (adjusted_warn_limit): New function.
3584 (pass_walloca::gate): Use it.
3585 (alloca_call_type): Same.
3586 (pass_walloca::execute): Same.
3587 * stor-layout.c (layout_decl): Treat HOST_WIDE_INT special.
3589 2018-08-28 David Malcolm <dmalcolm@redhat.com>
3591 * dumpfile.h (ATTRIBUTE_GCC_DUMP_PRINTF): Change version check on
3592 GCC_VERSION for usage of "__gcc_dump_printf__" format from
3595 2018-08-28 Richard Biener <rguenther@suse.de>
3597 PR tree-optimization/87124
3598 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Guard against
3599 constants before looking up avail.
3601 2018-08-28 Jakub Jelinek <jakub@redhat.com>
3604 * calls.c (maybe_warn_nonstring_arg): Punt early if
3605 warn_stringop_overflow is zero. Don't call get_range_strlen
3606 on 3rd argument, keep iterating until lenrng[1] is INTEGER_CST.
3607 Swap comparison operands to have constants on rhs. Only use
3608 lenrng[1] if non-NULL and INTEGER_CST. Don't uselessly
3609 increment lenrng[0].
3611 2018-08-28 Richard Sandiford <richard.sandiford@arm.com>
3613 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Fix unguarded
3614 use of tree_to_shwi. Remove duplicated test for the size being
3615 a whole number of bytes.
3617 2018-08-28 Richard Biener <rguenther@suse.de>
3619 PR tree-optimization/87117
3620 * tree-ssa-sccvn.c (eliminate_dom_walker::eliminate_cleanup):
3621 Handle removed stmt without LHS (GIMPLE_NOP).
3623 2018-08-28 Richard Biener <rguenther@suse.de>
3625 PR tree-optimization/87117
3626 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Exclude
3627 void which is is_gimple_reg_type by checking for COMPLETE_TYPE_P.
3629 2018-08-28 Richard Biener <rguenther@suse.de>
3631 PR tree-optimization/87117
3632 * tree-ssa-pre.c (compute_avail): Do not make expressions
3633 with predicated values available.
3634 (get_expr_value_id): Assert we do not run into predicated value
3637 2018-08-28 Richard Biener <rguenther@suse.de>
3639 PR tree-optimization/87117
3640 * tree-ssa-operands.c (add_stmt_operand): STRING_CST may
3641 get virtual operands.
3642 (get_expr_operands): Handle STRING_CST like other decls.
3644 2018-08-28 Martin Liska <mliska@suse.cz>
3646 * tree.h: Update documentation of fndecl_built_in_p
3650 2018-08-27 Jeff Law <law@redhat.com>
3651 PR tree-optimization/87110
3652 * tree-ssa-dse.c (compute_trims): Handle non-constant
3655 2018-08-27 Martin Sebor <msebor@redhat.com>
3657 PR tree-optimization/86914
3658 * tree-ssa-strlen.c (maybe_set_strlen_range): Avoid MEM_REF.
3660 2018-08-27 Martin Sebor <msebor@redhat.com>
3662 PR tree-optimization/87112
3663 * builtins.c (expand_builtin_strnlen): Convert c_strlen result to
3664 the type of the bound argument.
3666 2018-08-27 Jeff Law <law@redhat.com>
3668 * tree-ssa-dse.c (compute_trims): Handle case where the reference's
3669 type does not have a TYPE_SIZE_UNIT.
3671 2018-08-27 Steve Ellcey <sellcey@cavium.com>
3673 * config/aarch64/aarch64-speculation.cc: Replace include of cfg.h
3674 with include of backend.h.
3676 2018-08-27 Richard Biener <rguenther@suse.de>
3678 PR tree-optimization/86927
3679 * tree-vect-loop.c (vect_create_epilog_for_reduction): Properly
3680 use const cond reduction code.
3682 2018-08-27 Alexander Monakov <amonakov@ispras.ru>
3684 PR tree-optimization/85758
3685 * match.pd ((X & Y) ^ Y): Add :s qualifier to inner expression.
3687 2018-08-27 David Malcolm <dmalcolm@redhat.com>
3690 * diagnostic-show-locus.c (class layout_range): Update for
3691 conversion of show_caret_p to a tri-state.
3692 (layout_range::layout_range): Likewise.
3693 (make_range): Likewise.
3694 (layout::maybe_add_location_range): Likewise.
3695 (layout::should_print_annotation_line_p): Don't show annotation
3696 lines for ranges that are SHOW_LINES_WITHOUT_RANGE.
3697 (layout::get_state_at_point): Update for conversion of
3698 show_caret_p to a tri-state. Bail out early for
3699 SHOW_LINES_WITHOUT_RANGE, so that such ranges don't affect
3700 underlining or source colorization.
3701 (gcc_rich_location::add_location_if_nearby): Update for conversion
3702 of show_caret_p to a tri-state.
3703 (selftest::test_one_liner_multiple_carets_and_ranges): Likewise.
3704 (selftest::test_one_liner_fixit_replace_equal_secondary_range):
3706 (selftest::test_one_liner_labels): Likewise.
3707 * gcc-rich-location.c (gcc_rich_location::add_expr): Update for
3708 conversion of show_caret_p to a tri-state.
3709 * pretty-print.c (text_info::set_location): Likewise.
3710 * pretty-print.h (text_info::set_location): Likewise.
3711 * substring-locations.c (format_warning_n_va): Likewise.
3712 * tree-diagnostic.c (default_tree_printer): Likewise.
3713 * tree-pretty-print.c (newline_and_indent): Likewise.
3715 2018-08-27 David Malcolm <dmalcolm@redhat.com>
3718 * diagnostic-show-locus.c (get_line_span_for_fixit_hint): Show the
3719 line above for line-insertion fix-it hints.
3720 (selftest::test_fixit_insert_containing_newline): Update the
3721 expected results, and add a test with line-numbering enabled.
3723 2018-08-27 Martin Liska <mliska@suse.cz>
3726 * sanopt.c (sanitize_rewrite_addressable_params): Ignore
3727 params with DECL_HAS_VALUE_EXPR_P.
3729 2018-08-27 Martin Liska <mliska@suse.cz>
3731 * config/i386/i386.c (ix86_expand_set_or_movmem): Dump
3732 selected expansion strategy.
3734 2018-08-27 Martin Liska <mliska@suse.cz>
3736 * builtins.h (is_builtin_fn): Remove and fndecl_built_in_p.
3737 * builtins.c (is_builtin_fn): Likewise.
3738 * attribs.c (diag_attr_exclusions): Use new function
3739 fndecl_built_in_p and remove check for FUNCTION_DECL if
3741 (builtin_mathfn_code): Likewise.
3742 (fold_builtin_expect): Likewise.
3743 (fold_call_expr): Likewise.
3744 (fold_builtin_call_array): Likewise.
3745 (fold_call_stmt): Likewise.
3746 (set_builtin_user_assembler_name): Likewise.
3747 (is_simple_builtin): Likewise.
3748 * calls.c (gimple_alloca_call_p): Likewise.
3749 (maybe_warn_nonstring_arg): Likewise.
3750 * cfgexpand.c (expand_call_stmt): Likewise.
3751 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Likewise.
3752 (cgraph_edge::verify_corresponds_to_fndecl): Likewise.
3753 (cgraph_node::verify_node): Likewise.
3754 * cgraphclones.c (build_function_decl_skip_args): Likewise.
3755 (cgraph_node::create_clone): Likewise.
3756 * config/arm/arm.c (arm_insert_attributes): Likewise.
3757 * config/i386/i386.c (ix86_gimple_fold_builtin): Likewise.
3758 * dse.c (scan_insn): Likewise.
3759 * expr.c (expand_expr_real_1): Likewise.
3760 * fold-const.c (operand_equal_p): Likewise.
3761 (fold_binary_loc): Likewise.
3762 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
3763 * gimple-low.c (lower_stmt): Likewise.
3764 * gimple-pretty-print.c (dump_gimple_call): Likewise.
3765 * gimple-ssa-warn-restrict.c (wrestrict_dom_walker::check_call):
3767 * gimple.c (gimple_build_call_from_tree): Likewise.
3768 (gimple_call_builtin_p): Likewise.
3769 (gimple_call_combined_fn): Likewise.
3770 * gimplify.c (gimplify_call_expr): Likewise.
3771 (gimple_boolify): Likewise.
3772 (gimplify_modify_expr): Likewise.
3773 (gimplify_addr_expr): Likewise.
3774 * hsa-gen.c (gen_hsa_insns_for_call): Likewise.
3775 * ipa-cp.c (determine_versionability): Likewise.
3776 * ipa-fnsummary.c (compute_fn_summary): Likewise.
3777 * ipa-param-manipulation.c (ipa_modify_formal_parameters): Likewise.
3778 * ipa-split.c (visit_bb): Likewise.
3779 (split_function): Likewise.
3780 * ipa-visibility.c (cgraph_externally_visible_p): Likewise.
3781 * lto-cgraph.c (input_node): Likewise.
3782 * lto-streamer-out.c (write_symbol): Likewise.
3783 * omp-low.c (setjmp_or_longjmp_p): Likewise.
3784 (lower_omp_1): Likewise.
3785 * predict.c (strip_predict_hints): Likewise.
3786 * print-tree.c (print_node): Likewise.
3787 * symtab.c (symtab_node::output_to_lto_symbol_table_p): Likewise.
3788 * trans-mem.c (is_tm_irrevocable): Likewise.
3789 (is_tm_load): Likewise.
3790 (is_tm_simple_load): Likewise.
3791 (is_tm_store): Likewise.
3792 (is_tm_simple_store): Likewise.
3793 (is_tm_abort): Likewise.
3794 (tm_region_init_1): Likewise.
3795 * tree-call-cdce.c (gen_shrink_wrap_conditions): Likewise.
3796 * tree-cfg.c (verify_gimple_call): Likewise.
3797 (move_stmt_r): Likewise.
3798 (stmt_can_terminate_bb_p): Likewise.
3799 * tree-eh.c (lower_eh_constructs_2): Likewise.
3800 * tree-if-conv.c (if_convertible_stmt_p): Likewise.
3801 * tree-inline.c (remap_gimple_stmt): Likewise.
3802 (copy_bb): Likewise.
3803 (estimate_num_insns): Likewise.
3804 (fold_marked_statements): Likewise.
3805 * tree-sra.c (scan_function): Likewise.
3806 * tree-ssa-ccp.c (surely_varying_stmt_p): Likewise.
3807 (optimize_stack_restore): Likewise.
3808 (pass_fold_builtins::execute): Likewise.
3809 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
3810 (mark_all_reaching_defs_necessary_1): Likewise.
3811 * tree-ssa-dom.c (dom_opt_dom_walker::optimize_stmt): Likewise.
3812 * tree-ssa-forwprop.c (simplify_builtin_call): Likewise.
3813 (pass_forwprop::execute): Likewise.
3814 * tree-ssa-loop-im.c (stmt_cost): Likewise.
3815 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
3816 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Likewise.
3817 * tree-ssa-strlen.c (get_string_length): Likewise.
3818 * tree-ssa-structalias.c (handle_lhs_call): Likewise.
3819 (find_func_aliases_for_call): Likewise.
3820 * tree-ssa-ter.c (find_replaceable_in_bb): Likewise.
3821 * tree-stdarg.c (optimize_va_list_gpr_fpr_size): Likewise.
3822 * tree-tailcall.c (find_tail_calls): Likewise.
3823 * tree.c (need_assembler_name_p): Likewise.
3824 (free_lang_data_in_decl): Likewise.
3825 (get_call_combined_fn): Likewise.
3826 * ubsan.c (is_ubsan_builtin_p): Likewise.
3827 * varasm.c (incorporeal_function_p): Likewise.
3828 * tree.h (DECL_BUILT_IN): Remove and replace with
3830 (DECL_BUILT_IN_P): Transfort to fndecl_built_in_p.
3831 (fndecl_built_in_p): New.
3833 2018-08-27 Martin Liska <mliska@suse.cz>
3835 PR tree-optimization/86847
3836 * tree-switch-conversion.c (switch_decision_tree::dump_case_nodes):
3837 Dump also subtree probability.
3838 (switch_decision_tree::do_jump_if_equal): New function.
3839 (switch_decision_tree::emit_case_nodes): Handle special
3840 situations in balanced tree that can be emitted much simpler.
3841 Fix calculation of probabilities that happen in tree expansion.
3842 * tree-switch-conversion.h (struct cluster): Add
3844 (struct simple_cluster): Likewise.
3845 (struct case_tree_node): Add new function has_child.
3846 (do_jump_if_equal): New.
3848 2018-08-27 Martin Liska <mliska@suse.cz>
3850 * tree-switch-conversion.c (bit_test_cluster::find_bit_tests):
3851 Add new argument to bit_test_cluster constructor.
3852 (bit_test_cluster::emit): Set bits really number of values
3854 (bit_test_cluster::hoist_edge_and_branch_if_true): Add
3855 probability argument.
3856 * tree-switch-conversion.h (struct bit_test_cluster):
3857 Add m_handles_entire_switch.
3859 2018-08-27 Martin Liska <mliska@suse.cz>
3861 PR tree-optimization/86702
3862 * tree-switch-conversion.c (jump_table_cluster::emit):
3863 Make probabilities even for values in jump table
3864 according to number of cases handled.
3865 (switch_decision_tree::compute_cases_per_edge): Pass
3866 argument to reset_out_edges_aux function.
3867 (switch_decision_tree::analyze_switch_statement): Likewise.
3868 * tree-switch-conversion.h (switch_decision_tree::reset_out_edges_aux):
3871 2018-08-27 Martin Liska <mliska@suse.cz>
3873 * cfgexpand.c (expand_asm_stmt): Use label_to_block and pass
3874 cfun argument explicitly.
3875 * gimple-pretty-print.c (dump_gimple_switch): Likewise.
3876 * hsa-gen.c (gen_hsa_insns_for_switch_stmt): Use new
3877 function gimple_switch_default_bb.
3878 (convert_switch_statements):
3880 * ipa-fnsummary.c (set_switch_stmt_execution_predicate):
3881 * stmt.c (label_to_block_fn): Use label_to_block and pass
3882 cfun argument explicitly and use gimple_switch_label_bb.
3883 (expand_case): Likewise.
3884 * tree-cfg.c (lower_phi_internal_fn): Use label_to_block and pass
3885 cfun argument explicitly. Likewise.
3886 (make_edges_bb): Likewise.
3887 (make_cond_expr_edges): Likewise.
3888 (get_cases_for_edge): Likewise.
3889 (make_gimple_switch_edges): Likewise.
3890 (label_to_block_fn): Likewise.
3891 (label_to_block): Likewise.
3892 (make_goto_expr_edges): Likewise.
3893 (make_gimple_asm_edges): Likewise.
3894 (main_block_label): Likewise.
3895 (group_case_labels_stmt): Likewise.
3896 (find_taken_edge_computed_goto): Likewise.
3897 (find_taken_edge_switch_expr): Likewise.
3898 (gimple_verify_flow_info): Likewise.
3899 (gimple_redirect_edge_and_branch): Likewise.
3900 (gimple_switch_label_bb): New function.
3901 (gimple_switch_default_bb): Likewise.
3902 (gimple_switch_edge): Likewise.
3903 (gimple_switch_default_edge): Likewise.
3904 * tree-cfg.h (label_to_block_fn): Remove and replace ...
3905 (label_to_block): ... with this.
3906 (gimple_switch_label_bb): New.
3907 (gimple_switch_default_bb): Likewise.
3908 (gimple_switch_edge): Likewise.
3909 (gimple_switch_default_edge): Likewise.
3910 * tree-cfgcleanup.c (convert_single_case_switch): Use
3911 new gimple functions and pass new argument to label_to_block.
3912 (cleanup_control_flow_bb):
3913 * tree-eh.c (make_eh_dispatch_edges): Use label_to_block and pass
3914 cfun argument explicitly.
3915 (make_eh_edges): Likewise.
3916 (redirect_eh_dispatch_edge): Likewise.
3917 (lower_resx): Likewise.
3918 (lower_eh_dispatch): Likewise.
3919 (maybe_remove_unreachable_handlers): Likewise.
3920 (unsplit_eh): Likewise.
3921 (cleanup_empty_eh): Likewise.
3922 (verify_eh_edges): Likewise.
3923 (verify_eh_dispatch_edge): Likewise.
3924 * tree-ssa-dom.c (record_edge_info): Likewise.
3925 * tree-ssa-forwprop.c (simplify_gimple_switch_label_vec): Likewise.
3926 * tree-ssa-threadedge.c (thread_around_empty_blocks): Likewise.
3927 (thread_through_normal_block): Likewise.
3928 * tree-ssa-uncprop.c (associate_equivalences_with_edges): Likewise.
3929 * tree-ssa-uninit.c (convert_control_dep_chain_into_preds):
3930 * tree-switch-conversion.c (switch_conversion::collect): Use new
3932 (switch_conversion::check_final_bb): Likewise.
3933 (switch_conversion::gather_default_values): Pass new argument
3935 (switch_conversion::build_constructors): Likewise.
3936 (switch_decision_tree::compute_cases_per_edge): Use new
3937 gimple_switch_edge function.
3938 (switch_decision_tree::analyze_switch_statement): Pass new argument
3940 (switch_decision_tree::try_switch_expansion): Use
3941 gimple_switch_default_edge.
3942 * tree-vrp.c (find_switch_asserts): Pass new argument
3944 * vr-values.c (vr_values::vrp_visit_switch_stmt): Likewise.
3945 (vr_values::simplify_switch_using_ranges): Likewise.
3947 2018-08-27 Richard Biener <rguenther@suse.de>
3949 * cfganal.h (rev_post_order_and_mark_dfs_back_seme): Declare.
3950 * cfganal.c (rev_post_order_and_mark_dfs_back_seme): New function.
3952 * tree-ssa-sccvn.h (struct vn_pval): New structure.
3953 (struct vn_nary_op_s): Add unwind_to member. Add
3954 predicated_values flag and put result into a union together
3955 with a linked list of vn_pval.
3956 (struct vn_ssa_aux): Add name member to make maintaining
3957 a map of SSA name to vn_ssa_aux possible. Remove no longer
3958 needed info, dfsnum, low, visited, on_sccstack, use_processed
3959 and range_info_anti_range_p members.
3960 (run_scc_vn, vn_eliminate, free_scc_vn, vn_valueize): Remove.
3961 (do_rpo_vn, run_rpo_vn, eliminate_with_rpo_vn, free_rpo_vn):
3963 (vn_valueize): New global.
3964 (vn_context_bb): Likewise.
3965 (VN_INFO_RANGE_INFO, VN_INFO_ANTI_RANGE_P, VN_INFO_RANGE_TYPE,
3966 VN_INFO_PTR_INFO): Remove.
3967 * tree-ssa-sccvn.c: ... (rewrite)
3968 (pass_fre::execute): For -O2+ initialize loops and run
3969 RPO VN in optimistic mode (iterating). For -O1 and -Og
3970 run RPO VN in non-optimistic mode.
3971 * params.def (PARAM_SCCVN_MAX_SCC_SIZE): Remove.
3972 (PARAM_RPO_VN_MAX_LOOP_DEPTH): Add.
3973 * doc/invoke.texi (sccvn-max-scc-size): Remove.
3974 (rpo-vn-max-loop-depth): Document.
3975 * tree-ssa-alias.c (walk_non_aliased_vuses): Stop walking
3976 when valuezing the VUSE signals we walked out of the region.
3977 * tree-ssa-pre.c (phi_translate_1): Ignore predicated values.
3978 (phi_translate): Set VN context block to use for availability
3980 (compute_avail): Likewise.
3981 (pre_valueize): New function.
3982 (pass_pre::execute): Adjust to the RPO VN API.
3984 * tree-ssa-loop-ivcanon.c: Include tree-ssa-sccvn.h.
3985 (propagate_constants_for_unrolling): Remove.
3986 (tree_unroll_loops_completely): Perform value-numbering
3987 on the unrolled bodies loop parent.
3989 2018-08-27 Richard Biener <rguenther@suse.de>
3991 * tree-ssa-pre.c (compute_antic): Re-use inverted postorder
3992 for partial antic compute.
3994 2018-08-27 Jakub Jelinek <jakub@redhat.com>
3996 PR rtl-optimization/87065
3997 * combine.c (simplify_if_then_else): Formatting fix.
3998 (if_then_else_cond): Guard MULT optimization with SCALAR_INT_MODE_P
4000 (known_cond): Don't return const_true_rtx for vector modes. Use
4001 CONST0_RTX instead of const0_rtx. Formatting fixes.
4003 2018-08-27 Martin Liska <mliska@suse.cz>
4005 PR gcov-profile/87069
4006 * gcov.c (process_file): Record files already processed
4007 and warn about a file being processed multiple times.
4009 2018-08-27 Martin Liska <mliska@suse.cz>
4012 * config/aarch64/aarch64.c (aarch64_override_options_internal):
4013 Set default values for x_aarch64_*_string strings.
4014 * config/aarch64/aarch64.opt: Remove --{march,mcpu,mtune}==
4015 prefix. For -mabi do not print '=ABI' in help and use
4016 <option_value> format for -msve-vector-bits and -moverride
4019 2018-08-26 Jeff Law <law@redhat.com>
4021 * config/mips/frame-header-opt.c: Include "backend.h" rather than
4024 2018-08-26 Marek Polacek <polacek@redhat.com>
4026 PR c++/87029, Implement -Wredundant-move.
4027 * doc/invoke.texi: Document -Wredundant-move.
4029 2018-08-25 Martin Sebor <msebor@redhat.com>
4031 PR tree-optimization/87059
4032 * builtins.c (expand_builtin_strncmp): Convert MIN_EXPR operand
4033 to the same type as the other.
4034 * fold-const.c (fold_binary_loc): Assert expectation.
4036 2018-08-25 Iain Sandoe <iain@sandoe.co.uk>
4038 * config/darwin.c (machopic_legitimize_pic_address): Clean up
4039 extraneous parentheses, dead code section and formatting.
4041 2018-08-24 David Malcolm <dmalcolm@redhat.com>
4044 * diagnostic-show-locus.c (layout::layout): Ensure the margin is
4045 wide enough for jumps in the line-numbering to be visible.
4046 (layout::print_gap_in_line_numbering): New member function.
4047 (layout::calculate_line_spans): When using line numbering, merge
4048 line spans that are only 1 line apart.
4049 (diagnostic_show_locus): When printing line numbers, show gaps in
4050 line numbering directly, rather than printing headers.
4051 (selftest::test_diagnostic_show_locus_fixit_lines): Add test of
4052 line-numbering with multiple line spans.
4053 (selftest::test_fixit_insert_containing_newline_2): Add test of
4054 line-numbering, in which the spans are close enough to be merged.
4056 2018-08-24 Aldy Hernandez <aldyh@redhat.com>
4058 * gimple-ssa-evrp-analyze.c (set_ssa_range_info): Pass value_range
4059 to range_includes_zero_p. Do not special case VR_ANTI_RANGE.
4060 * tree-vrp.c (range_is_nonnull): Remove.
4061 (range_includes_zero_p): Accept value_range instead of min/max.
4062 (extract_range_from_binary_expr_1): Do not early bail on
4064 Use range_includes_zero_p instead of range_is_nonnull.
4065 (extract_range_from_unary_expr): Use range_includes_zero_p instead
4066 of range_is_nonnull.
4067 (vrp_meet_1): Pass value_range to range_includes_zero_p. Do not
4068 special case VR_ANTI_RANGE.
4069 (vrp_finalize): Same.
4070 * tree-vrp.h (range_includes_zero_p): Pass value_range as argument
4072 (range_is_nonnull): Remove.
4073 * vr-values.c (vrp_stmt_computes_nonzero): Use
4074 range_includes_zero_p instead of range_is_nonnull.
4075 (extract_range_basic): Pass value_range to range_includes_zero_p
4076 instead of range_is_nonnull.
4078 2018-08-24 Uros Bizjak <ubizjak@gmail.com>
4080 * emit-rtl.c (init_emit_once): Do not emit MODE_POINTER_BOUNDS RTXes.
4081 * emit-rtl.h (rtl_data): Remove return_bnd.
4082 * explow.c (trunc_int_for_mode): Do not handle POINTER_BOUNDS_MODE_P.
4083 * function.c (diddle_return_value): Do not handle crtl->return_bnd.
4084 * genmodes.c (complete_mode): Do not handle MODE_POINTER_BOUNDS.
4085 (POINTER_BOUNDS_MODE): Remove definition.
4086 (make_pointer_bounds_mode): Remove.
4087 (get_mode_class): Do not handle MODE_POINTER_BOUNDS.
4088 * machmode.h (POINTER_BOUNDS_MODE_P): Remove definition.
4089 (scalare_mode::includes_p): Do not handle MODE_POINTER_BOUNDS.
4090 * mode-classes.def: Do not define MODE_POINTER_BOUNDS.
4091 * stor-layout.c (int_mode_for_mode): Do not handle MODE_POINTER_BOUNDS.
4092 * tree-core.h (enum tree_index): Remove TI_POINTER_BOUNDS_TYPE.
4093 * varasm.c (output_constant_pool_2): Do not handle MODE_POINTER_BOUNDS.
4095 * config/i386/i386-modes.def (BND32, BND64): Remove.
4096 * config/i386/i386.c (dbx_register_map): Remove bound registers.
4097 (dbx64_register_map): Ditto.
4098 (svr4_dbx_register_map): Ditto.
4099 (indirect_thunk_bnd_needed): Remove.
4100 (indirect_thunks_bnd_used): Ditto.
4101 (indirect_return_bnd_needed): Ditto.
4102 (indirect_return_via_cx_bnd): Ditto.
4103 (enum indirect_thunk_prefix): Remove indirect_thunk_prefix_bnd.
4104 (indirect_thunk_name): Remove handling of indirect_thunk_prefix_bnd.
4105 (output_indirect_thunk): Ditto. Remove need_prefix argument.
4106 (output_indirect_thunk_function): Remove handling of
4107 indirect_return_bnd_needed, indirect_return_via_cx_bnd,
4108 indirect_thunk_bnd_needed and indirect_thunks_bnd_used variables.
4109 (ix86_save_reg): Remove handling of crtl->return_bnd.
4110 (ix86_legitimate_constant_p): Remove handling of POINTER_BOUNDS_MODE_P.
4111 (ix86_print_operand_address_as): Remove handling of UNSPEC_BNDMK_ADDR
4112 and UNSPEC_BNDLX_ADDR.
4113 (ix86_output_indirect_branch_via_reg): Remove handling of
4114 indirect_thunk_prefix_bnd.
4115 (ix86_output_indirect_branch_via_push): Ditto.
4116 (ix86_output_function_return): Ditto.
4117 (ix86_output_indirect_function_return): Ditto.
4118 (avoid_func_arg_motion): Do not handle UNSPEC_BNDSTX.
4119 * config/i386/i386.h (FIXED_REGISTERS): Remove bound registers.
4120 (CALL_USED_REGISTERS): Ditto.
4121 (REG_ALLOC_ORDER): Update for removal of bound registers.
4122 (HI_REGISTER_NAMES): Ditto.
4123 * config/i386/i386.md (UNSPEC_BNDMK, UNSPEC_BNDMK_ADDR, UNSPEC_BNDSTX)
4124 (UNSPEC_BNDLDX, UNSPEC_BNDLDX_ADDR, UNSPEC_BNDCL, UNSPEC_BNDCU)
4125 (UNSPEC_BNDCN, UNSPEC_MPX_FENCE): Remove.
4126 (BND0_REG, BND1_REG, BND2_REG, BND3_REG): Remove
4127 (FIRST_PSEUDO_REG): Update.
4128 (BND): Remove mode iterator.
4129 * config/i386/predicates.md (bnd_mem_operator): Remove.
4131 2018-08-24 Richard Sandiford <richard.sandiford@arm.com>
4133 * tree-vect-stmts.c (vectorizable_bswap): Handle variable-length
4136 2018-08-24 Richard Sandiford <richard.sandiford@arm.com>
4138 * tree-vect-slp.c (vect_transform_slp_perm_load): Separate out
4139 the case in which the permute needs only a single element and
4140 repeats for every vector of the result. Extend that case to
4141 handle variable-length vectors.
4142 * tree-vect-stmts.c (vectorizable_load): Update accordingly.
4144 2018-08-24 H.J. Lu <hongjiu.lu@intel.com>
4147 * dwarf2out.c (save_macinfo_strings): Call set_indirect_string
4148 on DW_MACINFO_start_file for -gsplit-dwarf -g3.
4150 2018-08-24 Richard Biener <rguenther@suse.de>
4152 * cfg.h (struct control_flow_graph): Add edge_flags_allocated and
4153 bb_flags_allocated members.
4154 (auto_flag): New RAII class for allocating flags.
4155 (auto_edge_flag): New RAII class for allocating edge flags.
4156 (auto_bb_flag): New RAII class for allocating bb flags.
4157 * cfgloop.c (verify_loop_structure): Allocate temporary edge
4159 * cfganal.c (dfs_enumerate_from): Remove use of visited sbitmap
4160 in favor of temporarily allocated BB flag.
4161 * hsa-brig.c: Re-order includes.
4162 * hsa-dump.c: Likewise.
4163 * hsa-regalloc.c: Likewise.
4164 * print-rtl.c: Likewise.
4165 * profile-count.c: Likewise.
4167 2018-08-24 Segher Boessenkool <segher@kernel.crashing.org>
4170 * config/rs6000/rs6000.c (toc_relative_expr_p): Check that the base is
4173 2018-08-24 Aldy Hernandez <aldyh@redhat.com>
4176 * wide-int-range.cc (wide_int_range_div): Do not ignore result
4177 from wide_int_range_multiplicative_op.
4179 2018-08-23 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
4181 * tree-vect-data-refs.c (vect_grouped_store_supported): Fix typo
4184 2018-08-23 Giuliano Belinassi <giuliano.belinassi@usp.br>
4186 * genmatch.c (parser::parse_operation): Fix typo 'exapnded'
4189 2018-08-23 Alexander Monakov <amonakov@ispras.ru>
4191 * tree-scalar-evolution.c (final_value_replacement_loop): Dump
4192 full GENERIC expression used for replacement.
4194 2018-08-23 Aldy Hernandez <aldyh@redhat.com>
4196 * tree-vrp.c (abs_extent_range): Remove.
4197 (extract_range_into_wide_ints): Pass wide ints by reference.
4198 (extract_range_from_binary_expr_1): Rewrite the *DIV_EXPR code.
4199 Pass wide ints by reference in all calls to
4200 extract_range_into_wide_ints.
4201 * wide-int-range.cc (wide_int_range_div): New.
4202 * wide-int-range.h (wide_int_range_div): New.
4203 (wide_int_range_includes_zero_p): New.
4204 (wide_int_range_zero_p): New.
4206 2018-08-23 Matthew Malcomson <matthew.malcomson@arm.com>
4208 * config/aarch64/aarch64.md (arches): New enum.
4209 (arch): New enum attr.
4210 (arch_enabled): New attr.
4211 (enabled): Now uses arch_enabled only.
4212 (simd, sve, fp16): Removed attribute.
4213 (fp): Attr now defined in terms of 'arch'.
4214 (*mov<mode>_aarch64, *movsi_aarch64, *movdi_aarch64, *movti_aarch64,
4215 *movhf_aarch64, <optab><fcvt_target><GPF:mode>2,
4216 <FCVT_F2FIXED:fcvt_fixed_insn><GPF:mode>3,
4217 <FCVT_FIXED2F:fcvt_fixed_insn><GPI:mode>3): Merge 'fp' and 'simd'
4218 attributes into 'arch'.
4219 (*movsf_aarch64, *movdf_aarch64, *movtf_aarch64, *add<mode>3_aarch64,
4220 subdi3, neg<mode>2, <optab><mode>3, one_cmpl<mode>2,
4221 *<NLOGICAL:optab>_one_cmpl<mode>3, *xor_one_cmpl<mode>3,
4222 *aarch64_ashl_sisd_or_int_<mode>3, *aarch64_lshr_sisd_or_int_<mode>3,
4223 *aarch64_ashr_sisd_or_int_<mode>3, *aarch64_sisd_ushl): Convert use of
4224 'simd' attribute into 'arch'.
4225 (load_pair_sw_<SX:mode><SX2:mode>, load_pair_dw_<DX:mode><DX2:mode>,
4226 store_pair_sw_<SX:mode><SX2:mode>, store_pair_dw_<DX:mode><DX2:mode>):
4227 Convert use of 'fp' attribute to 'arch'.
4228 * config/aarch64/aarch64-simd.md (move_lo_quad_internal_<mode>,
4229 move_lo_quad_internal_<mode>): (different modes) Merge 'fp' and 'simd'
4231 (move_lo_quad_internal_be_<mode>, move_lo_quad_internal_be_<mode>):
4232 (different modes) Merge 'fp' and 'simd' into 'arch'.
4233 (*aarch64_combinez<mode>, *aarch64_combinez_be<mode>): Merge 'fp' and
4236 2018-08-23 Segher Boessenkool <segher@kernel.crashing.org>
4238 PR rtl-optimization/87026
4239 * expmed.c (canonicalize_comparison): If we can no longer create
4240 pseudoregisters, don't.
4242 2018-08-23 Richard Earnshaw <rearnsha@arm.com>
4245 * config/arm/arm-protos.h (arm_emit_speculation_barrier): New
4247 * config/arm/arm.c (speculation_barrier_libfunc): New static
4249 (arm_init_libfuncs): Initialize it.
4250 (arm_emit_speculation_barrier): New function.
4251 * config/arm/arm.md (speculation_barrier): Call
4252 arm_emit_speculation_barrier for architectures that do not have
4254 (speculation_barrier_insn): Only match on Armv7 or later.
4256 2018-08-23 Richard Biener <rguenther@suse.de>
4259 * tree-inline.c (copy_bb): Drop unused __builtin_va_arg_pack_len
4262 2018-08-23 Richard Sandiford <richard.sandiford@arm.com>
4264 * config/aarch64/aarch64.c (aarch64_evpc_sve_tbl): Fix handling
4265 of single-vector TBLs.
4266 (aarch64_vectorize_vec_perm_const): Set one_vector_p when only
4269 2018-08-23 Richard Sandiford <richard.sandiford@arm.com>
4272 * config/aarch64/aarch64.c (aarch64_expand_vec_perm_const_1): Fix
4273 aarch64_evpc_tbl guard.
4275 2018-08-22 Bernd Edlinger <bernd.edlinger@hotmail.de>
4277 * tree-ssa-dse.c (compute_trims): Avoid folding away undefined
4280 2018-08-22 Martin Sebor <msebor@redhat.com>
4283 * tree-pretty-print.c (pretty_print_string): Add argument.
4284 (dump_generic_node): Call to pretty_print_string with string size.
4286 2018-08-22 Segher Boessenkool <segher@kernel.crashing.org>
4288 PR rtl-optimization/86771
4289 * combine.c (try_combine): Do not allow splitting a resulting PARALLEL
4290 of two SETs into those two SETs, one to be placed at i2, if that SETs
4291 destination is modified between i2 and i3.
4293 2018-08-22 Richard Sandiford <richard.sandiford@arm.com>
4295 PR tree-optimization/86725
4296 * tree-vect-loop.c (vect_inner_phi_in_double_reduction_p): New
4298 (vect_analyze_scalar_cycles_1): Check it.
4300 2018-08-22 Richard Sandiford <richard.sandiford@arm.com>
4302 PR tree-optimization/86725
4303 * tree-vect-loop.c (vect_is_simple_reduction): When treating
4304 an outer loop phi as a double reduction, make sure that the
4305 single user of the phi result is an inner loop phi.
4307 2018-08-22 Richard Sandiford <richard.sandiford@arm.com>
4309 * tree-vect-data-refs.c (vect_analyze_group_access_1): Convert
4310 grouped stores with gaps to a strided group.
4312 2018-08-22 Richard Sandiford <richard.sandiford@arm.com>
4314 * tree-vect-stmts.c (get_group_load_store_type)
4315 (get_load_store_type): Only test STMT_VINFO_STRIDED_P for the
4316 first statement in a group.
4318 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
4320 * config/darwin.h (LINK_COMMAND_SPEC_A): Sync LTO options with
4321 the sequence used in gcc/gcc.c.
4323 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
4326 * gcc-ar.c (main): Don’t try to invoke the plug-in if we’re not
4329 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
4331 * config/darwin10.h (LINK_GCC_C_SEQUENCE_SPEC): Adjust to use the
4332 Darwin10-specific unwinder-shim.
4333 * config/darwin12.h (LINK_GCC_C_SEQUENCE_SPEC): Remove.
4334 * config/rs6000/darwin.h (DARWIN_CRT1_SPEC, DARWIN_DYLIB1_SPEC):
4335 New to cater for Darwin10 Rosetta.
4337 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
4339 * config/i386/i386.c (ix86_output_addr_diff_elt): Move the MACH-O
4340 specific test before the one for HAVE_AS_GOTOFF_IN_DATA.
4342 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
4347 * gcc/dwarf2out.c (FUNC_SECOND_SECT_LABEL): New.
4348 (dwarf2out_switch_text_section): Generate a local label for the
4349 second function sub-section and apply it as the second FDE start
4351 * gcc/final.c (final_scan_insn_1): Emit second FDE label after the
4352 second sub-section start.
4354 2018-08-22 Richard Biener <rguenther@suse.de>
4356 PR tree-optimization/86988
4357 * tree-vrp.c (vrp_prop::check_mem_ref): Bail out on VLAs.
4359 2018-08-22 Richard Biener <rguenther@suse.de>
4361 PR tree-optimization/86945
4362 * tree-cfg.c (generate_range_test): Use unsigned arithmetic.
4364 2018-08-22 Alexandre Oliva <oliva@adacore.com>
4366 * config/rs6000/rs6000.c (SMALL_DATA_RELOC, SMALL_DATA_REG): Add
4367 a comment about how uses of r2 for .sdata2 come about.
4369 2018-08-22 Alexandre Oliva <aoliva@redhat.com>
4371 * tree-ssa-reassoc.c (is_reassociable_op): Fix cut&pasto.
4373 2018-08-21 Marek Polacek <polacek@redhat.com>
4375 PR c++/86981, Implement -Wpessimizing-move.
4376 * doc/invoke.texi: Document -Wpessimizing-move.
4378 2018-08-21 Jan Hubicka <jh@suse.cz>
4380 * tree.c (find_decls_types_r): Do not check for redundant typedefs.
4381 * tree.h (is_redundant_typedef): Remove.
4382 * dwarf2out.c (is_redundant_typedef): Turn into static function.
4384 2018-08-21 Jan Hubicka <jh@suse.cz>
4386 * tree.c (free_lang_data_in_decl): Remove types from DECL_CONTEXT
4389 2018-08-21 Tamar Christina <tamar.christina@arm.com>
4391 * expmed.c (extract_low_bits): Reject invalid subregs early.
4393 2018-08-21 Bernd Edlinger <bernd.edlinger@hotmail.de>
4396 * tree-ssa-strlen.c (adjust_last_stmt): Avoid folding away undefined
4399 2018-08-21 Rasmus Villemoes <rv@rasmusvillemoes.dk>
4401 * config/vxworks.h: Guard vxworks_asm_out_constructor and
4402 vxworks_asm_out_destructor by !HAVE_INITFINI_ARRAY_SUPPORT
4403 * config/vxworks.c: Likewise.
4405 2018-08-21 Rasmus Villemoes <rv@rasmusvillemoes.dk>
4407 * config/vxworks.c: Set targetm.have_ctors_dtors
4408 if HAVE_INITFINI_ARRAY_SUPPORT.
4409 * config/vxworks.h: Set SUPPORTS_INIT_PRIORITY
4410 if HAVE_INITFINI_ARRAY_SUPPORT.
4412 2018-08-21 Rasmus Villemoes <rv@rasmusvillemoes.dk>
4414 * config/vxworks.h: Add $(WIND_BASE)/target/h/wrn/coreip to
4415 default search path for VxWorks < 7.
4417 2018-08-21 Bernd Edlinger <bernd.edlinger@hotmail.de>
4419 * gimple-ssa-sprintf.c (decl_constant_value): Remove.
4420 (get_format_string): Refer to c_getstr.
4422 2018-08-21 Tom de Vries <tdevries@suse.de>
4424 * cgraph.h (debuginfo_early_init, debuginfo_init, debuginfo_fini)
4425 (debuginfo_start, debuginfo_stop, debuginfo_early_start)
4426 (debuginfo_early_stop): Declare.
4427 * cgraphunit.c (debuginfo_early_init, debuginfo_init, debuginfo_fini)
4428 (debuginfo_start, debuginfo_stop, debuginfo_early_start)
4429 (debuginfo_early_stop): New function.
4430 (symbol_table::finalize_compilation_unit): Call debuginfo_early_start
4431 and debuginfo_early_stop.
4432 * dwarf2out.c (dwarf2out_finish, dwarf2out_early_finish): Dump dwarf.
4433 * toplev.c (compile_file): Call debuginfo_start and debuginfo_stop.
4434 (general_init): Call debuginfo_early_init.
4435 (finalize): Call debuginfo_fini.
4436 (do_compile): Call debuginfo_init.
4437 * doc/invoke.texi (@gccoptlist): Add -fdump-debug and
4439 (@item -fdump-debug, @item -fdump-earlydebug): Add.
4441 2018-08-21 Tom de Vries <tdevries@suse.de>
4443 * dwarf2out.c (print_dw_val, print_loc_descr, print_die): Handle
4444 flag_dump_noaddr and flag_dump_unnumbered.
4446 2018-08-21 Aldy Hernandez <aldyh@redhat.com>
4448 * wide-int-range.cc (wide_int_range_abs): New.
4449 (wide_int_range_order_set): Rename from wide_int_range_min_max.
4450 * wide-int-range.h (wide_int_range_abs): New.
4451 (wide_int_range_min_max): New.
4452 * tree-vrp.c (extract_range_from_unary_expr): Rewrite ABS_EXPR
4453 case to call wide_int_range_abs.
4454 Rewrite MIN/MAX_EXPR to call wide_int_range_min_max.
4455 (extract_range_from_abs_expr): Delete.
4457 2018-08-20 Michael Meissner <meissner@linux.ibm.com>
4460 * config/rs6000/rs6000.md (extendsi<mode>2): Change constraints
4461 from 'Y' to 'YZ' to enable the LWAX instruction to be generated
4464 2018-08-20 Nathan Sidwell <nathan@acm.org>
4465 Jeff Law <law@redhat.com>
4467 * config/s390/s390-c (s390_macro_to_expand): Use cpp_macro_p.
4468 * config/spu/spu-c.c (spu_macro_to_expand): Likewise.
4470 2018-08-20 David Malcolm <dmalcolm@redhat.com>
4473 * attribs.c (diag_attr_exclusions): Add auto_diagnostic_group instance.
4474 (decl_attributes): Likewise.
4475 * calls.c (maybe_warn_nonstring_arg): Add auto_diagnostic_group
4477 * cgraphunit.c (maybe_diag_incompatible_alias): Likewise.
4478 * diagnostic-core.h (class auto_diagnostic_group): New class.
4479 * diagnostic.c (diagnostic_initialize): Initialize the new fields.
4480 (diagnostic_report_diagnostic): Handle the first diagnostics within
4482 (emit_diagnostic): Add auto_diagnostic_group instance.
4484 (inform_n): Likewise.
4485 (warning): Likewise.
4486 (warning_at): Likewise.
4487 (warning_n): Likewise.
4488 (pedwarn): Likewise.
4489 (permerror): Likewise.
4491 (error_n): Likewise.
4492 (error_at): Likewise.
4494 (fatal_error): Likewise.
4495 (internal_error): Likewise.
4496 (internal_error_no_backtrace): Likewise.
4497 (auto_diagnostic_group::auto_diagnostic_group): New ctor.
4498 (auto_diagnostic_group::~auto_diagnostic_group): New dtor.
4499 * diagnostic.h (struct diagnostic_context): Add fields
4500 "diagnostic_group_nesting_depth",
4501 "diagnostic_group_emission_count", "begin_group_cb",
4503 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior):
4504 Add auto_diagnostic_group instance(s).
4505 (find_explicit_erroneous_behavior): Likewise.
4506 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Likewise.
4507 * gimple-ssa-warn-restrict.c (maybe_diag_offset_bounds): Likewise.
4508 * gimplify.c (warn_implicit_fallthrough_r): Likewise.
4509 (gimplify_va_arg_expr): Likewise.
4510 * hsa-gen.c (HSA_SORRY_ATV): Likewise.
4511 (HSA_SORRY_AT): Likewise.
4512 * ipa-devirt.c (compare_virtual_tables): Likewise.
4513 (warn_odr): Likewise.
4514 * multiple_target.c (expand_target_clones): Likewise.
4515 * opts-common.c (cmdline_handle_error): Likewise.
4516 * reginfo.c (globalize_reg): Likewise.
4517 * substring-locations.c (format_warning_n_va): Likewise.
4518 * tree-inline.c (expand_call_inline): Likewise.
4519 * tree-ssa-ccp.c (pass_post_ipa_warn::execute): Likewise.
4520 * tree-ssa-loop-niter.c
4521 (do_warn_aggressive_loop_optimizations): Likewise.
4522 * tree-ssa-uninit.c (warn_uninit): Likewise.
4523 * tree.c (warn_deprecated_use): Likewise.
4525 2018-08-20 H.J. Lu <hongjiu.lu@intel.com>
4528 * config/i386/i386.md (eh_return): Always update EH return
4529 address in word_mode.
4531 2018-08-20 Chung-Lin Tang <cltang@codesourcery.com>
4533 * targhooks.c (std_gimplify_va_arg_expr): Properly handle case of when
4534 TARGET_SPLIT_COMPLEX_ARG is defined.
4536 2018-08-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
4538 * expr.c (store_field): Change gcc_assert to gcc_checking_assert.
4540 2018-08-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
4543 * expr.c (expand_assignment): Assert that bitpos is positive.
4544 (store_field): Likewise
4545 (expand_expr_real_1): Make sure that bitpos is positive.
4546 * config/alpha/alpha.h (CONSTANT_ADDRESS_P): Avoid signed
4549 2018-08-20 Nathan Sidwell <nathan@acm.org>
4551 * Makefile.in (CPP_ID_DATA_H): Delete.
4552 (CPP_INTERNAL_H): Don't add it.
4553 (GTFILES): Replace CPP_ID_DATA_H with CPPLIB_H.
4554 * gengtype.c (open_base_files): Replace cpp-id-data.h with cpplib.h
4556 2018-08-20 Richard Biener <rguenther@suse.de>
4558 PR tree-optimization/78655
4559 * tree-vrp.c (extract_range_from_binary_expr_1): Make
4560 pointer + offset nonnull if either operand is nonnull work.
4562 2018-08-20 Tom de Vries <tdevries@suse.de>
4564 * dwarf2out.c (add_scalar_info): Don't add reference to existing die
4565 unless the referenced die describes the added property using
4566 DW_AT_location or DW_AT_const_value. Fall back to exprloc case.
4567 Otherwise, add a DW_AT_location to the referenced die.
4569 2018-08-19 Uros Bizjak <ubizjak@gmail.com>
4572 * config/i386/i386.c (ix86_rtx_costs) [case SET]: Check source for
4573 register_operand when calling ix86_set_reg_reg_cost.
4574 [case CONST_INT, case CONST, case LABEL_REF, case SYMBOL_REF]:
4575 Set *total to 0 for operands that satisfy x86_64_immediate_operand
4576 predicate and to 1 otherwise.
4578 2018-08-18 Iain Sandoe <iain@sandoe.co.uk>
4580 * config/darwin.c (darwin_override_options): If -gsplit-dwarf is set,
4581 emit a diagnostic that it is not supported and reset the option.
4582 * config/darwin.h (DRIVER_SELF_SPECS): Note that gsplit-dwarf is not
4583 supported and consume the option. (ASM_FINAL_SPEC): New.
4585 2018-08-17 Segher Boessenkool <segher@kernel.crashing.org>
4587 * doc/md.texi (Patterns): Use @ref instead of @xref in the middle of
4590 2018-08-17 Sandra Loosemore <sandra@codesourcery.com>
4592 C-SKY port: Documentation
4594 * doc/extend.texi (C-SKY Function Attributes): New section.
4595 * doc/invoke.texi (Option Summary): Add C-SKY options.
4596 (C-SKY Options): New section.
4597 * doc/md.texi (Machine Constraints): Document C-SKY constraints.
4599 2018-08-17 Jojo <jijie_rong@c-sky.com>
4600 Huibin Wang <huibin_wang@c-sky.com>
4601 Sandra Loosemore <sandra@codesourcery.com>
4602 Chung-Lin Tang <cltang@codesourcery.com>
4604 C-SKY port: Backend implementation
4606 * config/csky/*: New.
4607 * common/config/csky/*: New.
4609 2018-08-17 Jojo <jijie_rong@c-sky.com>
4610 Huibin Wang <huibin_wang@c-sky.com>
4611 Sandra Loosemore <sandra@codesourcery.com>
4612 Chung-Lin Tang <cltang@codesourcery.com>
4613 Andrew Jenner <andrew@codesourcery.com>
4615 C-SKY port: Configury
4617 * config.gcc (csky-*-*): New.
4618 * configure.ac: Add csky to targets for dwarf2 debug_line support.
4619 * configure: Regenerated.
4621 2018-08-17 David Malcolm <dmalcolm@redhat.com>
4623 * dump-context.h: Include "dumpfile.h".
4624 (dump_context::dump_printf_va): Convert final param from va_list
4625 to va_list *. Convert from ATTRIBUTE_PRINTF to
4626 ATTRIBUTE_GCC_DUMP_PRINTF.
4627 (dump_context::dump_printf_loc_va): Likewise.
4628 * dumpfile.c: Include "stringpool.h".
4629 (make_item_for_dump_printf_va): Delete.
4630 (make_item_for_dump_printf): Delete.
4631 (class dump_pretty_printer): New class.
4632 (dump_pretty_printer::dump_pretty_printer): New ctor.
4633 (dump_pretty_printer::emit_items): New member function.
4634 (dump_pretty_printer::emit_any_pending_textual_chunks): New member
4636 (dump_pretty_printer::emit_item): New member function.
4637 (dump_pretty_printer::stash_item): New member function.
4638 (dump_pretty_printer::format_decoder_cb): New member function.
4639 (dump_pretty_printer::decode_format): New member function.
4640 (dump_context::dump_printf_va): Reimplement in terms of
4641 dump_pretty_printer.
4642 (dump_context::dump_printf_loc_va): Convert final param from va_list
4644 (dump_context::begin_scope): Reimplement call to
4645 make_item_for_dump_printf.
4646 (dump_printf): Update for change to dump_printf_va.
4647 (dump_printf_loc): Likewise.
4648 (selftest::test_capture_of_dump_calls): Convert "stmt" from
4649 greturn * to gimple *. Add a test_decl. Add tests of dump_printf
4650 with %T, %E, and %G.
4651 * dumpfile.h (ATTRIBUTE_GCC_DUMP_PRINTF): New macro.
4652 (dump_printf): Replace ATTRIBUTE_PRINTF_2 with
4653 ATTRIBUTE_GCC_DUMP_PRINTF (2, 3).
4654 (dump_printf_loc): Replace ATTRIBUTE_PRINTF_3 with
4655 ATTRIBUTE_GCC_DUMP_PRINTF (3, 0).
4656 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Convert
4657 use of HOST_WIDE_INT_PRINT_DEC on unsigned HOST_WIDE_INT "count"
4658 within a dump_printf_loc call to "%wu".
4659 (vector_alignment_reachable_p): Merge two dump_printf[_loc] calls,
4660 converting a use of HOST_WIDE_INT_PRINT_DEC to "%wd". Add a
4661 missing space after "=".
4662 * tree-vect-loop.c (vect_analyze_loop_2) Within a dump_printf
4663 call, convert use of HOST_WIDE_INT_PRINT_DEC to "%wd".
4664 * tree-vect-slp.c (vect_slp_bb): Within a dump_printf_loc call,
4665 convert use of HOST_WIDE_INT_PRINT_UNSIGNED to "%wu".
4666 * tree-vectorizer.c (try_vectorize_loop_1): Likewise. Remove
4667 duplicate "vectorized" from message.
4669 2018-08-17 Szabolcs Nagy <szabolcs.nagy@arm.com>
4671 * config/arm/arm-builtins.c (arm_init_simd_builtin_types): Clear
4672 polyNxK_t element's TYPE_STRING_FLAG.
4674 2018-08-17 Segher Boessenkool <segher@kernel.crashing.org>
4676 * config/rs6000/rs6000.md (*cbranch, *creturn): Name these patterns
4677 (they were unnamed before). Fix comments.
4679 2018-08-17 Nathan Sidwell <nathan@acm.org>
4681 * cppbuiltin.c: Include "cpplib.h", not "cpp-id-data.h".
4683 2018-08-17 Richard Biener <rguenther@suse.de>
4685 PR tree-optimization/86841
4686 * wide-int-range.cc (wide_int_range_lshift): Use to_uhwi.
4688 2018-08-17 Martin Liska <mliska@suse.cz>
4690 * common.opt: Remove Warn, Init and Report for options with
4691 Ignore/Deprecated flag. Warning is done automatically for
4693 * config/i386/i386.opt: Likewise.
4694 * config/ia64/ia64.opt: Likewise.
4695 * config/rs6000/rs6000.opt: Likewise.
4696 * cppbuiltin.c (define_builtin_macros_for_compilation_flags):
4697 Remove usage of flag_check_pointer_bounds.
4698 * lto-wrapper.c (merge_and_complain): Do not handle
4699 OPT_fcheck_pointer_bounds.
4700 (append_compiler_options): Likewise.
4701 * opt-functions.awk: Do not handle Deprecated.
4702 * optc-gen.awk: Check that Var, Report and Init are not
4703 used for an option with Ignore/Deprecated flag.
4704 * opts-common.c (decode_cmdline_option): Do not report
4706 (read_cmdline_option): Report warning for OPT_SPECIAL_deprecated
4708 * opts.h (struct cl_option): Remove cl_deprecated flag.
4709 (CL_ERR_DEPRECATED): Remove error enum value.
4711 2018-08-17 Richard Biener <rguenther@suse.de>
4714 * tree-inline.c (copy_bb): When inlining __builtin_va_arg_pack_len ()
4715 across a va-arg-pack using call adjust its return value accordingly.
4717 2018-08-16 Martin Sebor <msebor@redhat.com>
4719 PR tree-optimization/86853
4720 * gimple-ssa-sprintf.c (struct format_result): Rename member.
4721 (struct fmtresult): Add member and initialize it in ctors.
4722 (format_character): Handle %C. Extend range to NUL. Set MAYFAIL.
4723 (format_string): Handle %S the same as %ls. Set MAYFAIL.
4724 (format_directive): Set POSUNDER4K when MAYFAIL is set.
4725 (parse_directive): Handle %C same as %c.
4726 (sprintf_dom_walker::compute_format_length): Adjust.
4727 (is_call_safe): Adjust.
4729 2018-08-16 Bernd Edlinger <bernd.edlinger@hotmail.de>
4731 * builtins.c (c_strlen): Add new parameter eltsize. Use it
4732 for determining how to count the elements.
4733 * builtins.h (c_strlen): Adjust prototype.
4734 * expr.c (string_constant): Add new parameter mem_size.
4735 Set *mem_size appropriately.
4736 * expr.h (string_constant): Adjust protoype.
4737 * gimple-fold.c (get_range_strlen): Add new parameter eltsize.
4738 * gimple-fold.h (get_range_strlen): Adjust prototype.
4739 * gimple-ssa-sprintf.c (get_string_length): Add new parameter eltsize.
4740 (format_string): Call get_string_length with eltsize.
4742 2018-08-16 David Malcolm <dmalcolm@redhat.com>
4744 * diagnostic.c (default_diagnostic_start_span_fn): Call pp_string
4745 to emit the span, rather than setting it as the prefix.
4747 2018-08-16 David Malcolm <dmalcolm@redhat.com>
4749 * diagnostic-show-locus.c (layout::start_annotation_line): Add
4750 "margin_char" parameter, defaulting to space. Use it in place
4751 of pp_space for the initial part of the margin.
4752 (layout::print_leading_fixits): Use '+' when filling the margin
4753 of line-insertion fix-it hints.
4755 2018-08-16 Segher Boessenkool <segher@kernel.crashing.org>
4757 * config/rs6000/rs6000.md (two unnamed define_insn and define_split):
4760 2018-08-16 Segher Boessenkool <segher@kernel.crashing.org>
4762 * config/rs6000/altivec.md: Don't set length attribute to the default
4764 * config/rs6000/darwin.md: Ditto.
4765 * config/rs6000/dfp.md: Ditto.
4766 * config/rs6000/htm.md: Ditto.
4767 * config/rs6000/rs6000.md: Ditto.
4768 * config/rs6000/sync.md: Ditto.
4769 * config/rs6000/vsx.md: Ditto.
4771 2018-08-16 Segher Boessenkool <segher@kernel.crashing.org>
4773 * config/rs6000/altivec.md: Don't set length attribute to the default
4774 value, for branch instructions.
4775 * config/rs6000/darwin.md: Ditto.
4776 * config/rs6000/rs6000.md: Ditto.
4778 2018-08-16 Segher Boessenkool <segher@kernel.crashing.org>
4780 * config/rs6000/rs6000.md (length): Always define as const_int 4.
4781 (unnamed conditional branch define_insn): Set length to 4 or 8
4782 depending on offset.
4783 (<bd>_<mode>): Similar, for alternative 0.
4784 (<bd>tf_<mode>): Ditto.
4786 2018-08-16 Tamar Christina <tamar.christina@arm.com>
4788 * expr.c (copy_blkmode_to_reg): Perform larger copies when safe.
4790 2018-08-16 Matthew Malcomson <matthew.malcomson@arm.com>
4792 * doc/rtl.texi: Replace old RTX class names with new names.
4795 2018-08-16 Vlad Lazar <vlad.lazar@arm.com>
4797 * expmed.h (canonicalize_comparison): New declaration.
4798 * expmed.c (canonicalize_comparison, equivalent_cmp_code): New function.
4799 * expmed.c (emit_store_flag_1): Add call to canonicalize_comparison.
4800 * optabs.c (prepare_cmp_insn): Likewise.
4801 * rtl.h (unsigned_condition_p): New function which checks if a
4802 comparison operator is unsigned.
4804 2018-08-16 Nathan Sidwell <nathan@acm.org>
4806 * config/rs6000/rs6000-c.c (rs6000_macro_to_expend): Use cpp_macro_p.
4807 * config/powerpcspc/powerpcspe-c.c (rs6000_macro_to_expend): Likewise.
4809 2018-08-16 Tamar Christina <tamar.christina@arm.com>
4812 * config/arm/arm.c (arm_can_change_mode_class): Disallow subreg.
4813 * config/arm/neon.md (movv4hf, movv8hf): Refactored to..
4814 (mov<mov>): ..this and enable unconditionally.
4816 2018-08-16 Tamar Christina <tamar.christina@arm.com>
4818 * config/arm/neon.md (*neon_mov<mode>): Remove reg-to-reg alternative.
4820 2018-08-16 Sam Tebbs <sam.tebbs@arm.com>
4822 * config/aarch64/aarch64.opt (mlow-precision-recip-sqrt)
4823 (mlow-precision-sqrt, mlow-precision-div, mverbose-cost-dump): Replace
4824 "Common" with "Target".
4826 2018-08-15 Uros Bizjak <ubizjak@gmail.com>
4828 * config/i386/i386.opt (mmitigate-rop): Mark as deprecated.
4829 * doc/invoke.texi (mmitigate-rop): Remove.
4830 * config/i386/i386.c: Do not include "regrename.h".
4831 (ix86_rop_should_change_byte_p, reg_encoded_number)
4832 (ix86_get_modrm_for_rop, set_rop_modrm_reg_bits, ix86_mitigate_rop):
4834 (ix86_reorg): Remove call to ix86_mitigate_rop.
4835 * config/i386/i386.md (attr "modrm_class"): Remove.
4836 (cmp<mode>_ccno_1, mov<mode>_xor, movstrict<mode>_xor)
4837 (x86_mov<mode>cc_0_m1. x86_mov<mode>cc_0_m1_se)
4838 (x86_mov<mode>cc_0_m1_neg): Remove modrm_class attribute override.
4840 2018-08-15 Will Schmidt <will_schmidt@vnet.ibm.com>
4842 * config/rs6000/rs600.c (rs6000_gimple_fold_builtin): Add entries to
4843 allow folding of mergeh() and mergel() for the float and double types.
4844 (fold_mergehl_helper): Rework to handle building a permute tree
4847 2018-08-15 Uros Bizjak <ubizjak@gmail.com>
4849 * config/i386/i386.c (expand_vec_perm_movs): Enable V4SFmode
4852 2018-08-15 David Malcolm <dmalcolm@redhat.com>
4854 * common.opt (fdiagnostics-show-labels): New option.
4855 * diagnostic-show-locus.c (class layout_range): Add field
4857 (class layout): Add field "m_show_labels_p".
4858 (layout_range::layout_range): Add param "label" and use it to
4860 (make_range): Pass in NULL for new "label" param of layout_range's
4862 (layout::layout): Initialize m_show_labels_p.
4863 (layout::maybe_add_location_range): Pass in loc_range->m_label
4864 when constructing layout_range instances.
4865 (struct line_label): New struct.
4866 (layout::print_any_labels): New member function.
4867 (layout::print_line): Call it if label-printing is enabled.
4868 (selftest::test_one_liner_labels): New test.
4869 (selftest::test_diagnostic_show_locus_one_liner): Call it.
4870 * diagnostic.c (diagnostic_initialize): Initialize
4871 context->show_labels_p.
4872 * diagnostic.h (struct diagnostic_context): Add field
4874 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
4875 -fno-diagnostics-show-labels.
4876 * dwarf2out.c (gen_producer_string): Add
4877 OPT_fdiagnostics_show_labels to the ignored options.
4878 * gcc-rich-location.c (gcc_rich_location::add_expr): Add "label"
4880 (gcc_rich_location::maybe_add_expr): Likewise.
4881 * gcc-rich-location.h (gcc_rich_location::gcc_rich_location): Add
4882 label" param, defaulting to NULL.
4883 (gcc_rich_location::add_expr): Add "label" param.
4884 (gcc_rich_location::maybe_add_expr): Likewise.
4885 (class text_range_label): New class.
4886 (class range_label_for_type_mismatch): New class.
4887 * gimple-ssa-sprintf.c (fmtwarn): Pass NULL for new label params
4888 of format_warning_va.
4889 (fmtwarn_n): Likewise for new params of format_warning_n_va.
4890 * lto-wrapper.c (merge_and_complain): Add
4891 OPT_fdiagnostics_show_labels to the "pick one setting" options.
4892 (append_compiler_options): Likewise to the dropped options.
4893 (append_diag_options): Likewise to the passed-on options.
4894 * opts.c (common_handle_option): Handle the new option.
4895 * selftest-diagnostic.c
4896 (test_diagnostic_context::test_diagnostic_context): Enable
4898 * substring-locations.c: Include "gcc-rich-location.h".
4899 (format_warning_n_va): Add "fmt_label" and "param_label" params
4900 and use them as appropriate.
4901 (format_warning_va): Add "fmt_label" and "param_label" params,
4902 passing them on to format_warning_n_va.
4903 (format_warning_at_substring): Likewise.
4904 (format_warning_at_substring_n): Likewise.
4905 * substring-locations.h (format_warning_va): Add "fmt_label" and
4906 "param_label" params.
4907 (format_warning_n_va): Likewise.
4908 (format_warning_at_substring): Likewise.
4909 (format_warning_at_substring_n): Likewise.
4910 * toplev.c (general_init): Initialize global_dc->show_labels_p.
4912 2018-08-15 Qing Zhao <qing.zhao@oracle.com>
4915 * builtins.c (expand_builtin_memcmp): Do not expand the call
4916 when overflow is detected.
4918 2018-08-15 Martin Sebor <msebor@redhat.com>
4920 PR tree-optimization/71625
4921 * config/aarch64/aarch64-builtins.c
4922 (aarch64_init_simd_builtin_types): Clear Poly8_t's TYPE_STRING_FLAG.
4924 2018-08-15 Ilya Leoshkevich <iii@linux.ibm.com>
4926 * config/s390/s390.c (s390_reorg): Remove loop.
4928 2018-08-15 Iain Sandoe <iain@sandoe.co.uk>
4931 (darwin_function_switched_text_sections): Delete.
4932 * gcc/config/darwin.h
4933 (TARGET_ASM_FUNCTION_SWITCHED_TEXT_SECTIONS): Likewise.
4935 2018-08-15 Iain Sandoe <iain@sandoe.co.uk>
4938 * config/darwin.h: (DEBUG_STR_OFFSETS_SECTION, DEBUG_LOCLISTS_SECTION,
4939 DEBUG_RNGLISTS_SECTION) new macros. (DEBUG_PUBNAMES_SECTION,
4940 DEBUG_PUBTYPES_SECTION) update to include GNU variant.
4942 2018-08-15 Martin Liska <mliska@suse.cz>
4944 PR tree-optimization/86925
4945 * predict.c (expr_expected_value_1): When taking
4946 later predictor, assign also probability.
4947 Use fold_build2_initializer_loc in order to fold
4948 the expression in -frounding-math.
4950 2018-08-14 Allan Sandfeld Jensen <allan.jensen@qt.io>
4952 * config/i386/i386.c (expand_vec_perm_movs): New method matching movs
4954 (expand_vec_perm_1): Try the new method.
4956 2018-08-14 Ilya Leoshkevich <iii@linux.ibm.com>
4959 * lra-lives.c (remove_some_program_points_and_update_live_ranges):
4960 Check whether lra_live_max_point is 0 before dividing.
4962 2018-08-14 Martin Sebor <msebor@redhat.com>
4964 PR tree-optimization/86650
4965 * tree-vrp.c (vrp_prop::check_array_ref): Print an inform message.
4966 (vrp_prop::check_mem_ref): Same.
4968 2018-08-13 Liu Hao <lh_mouse@126.com>
4970 * pretty-print.c (eat_esc_sequence): Swap the foreground and
4971 background colors if the COMMON_LVB_REVERSE_VIDEO flag is set,
4972 and clear it thereafter, as it only works for DBCS.
4974 2018-08-13 Liu Hao <lh_mouse@126.com>
4976 * pretty-print.c (mingw_ansi_fputs): Do not call _close() on the
4977 handle returned by _get_osf_handle().
4979 2018-08-13 Will Schmidt <will_schmidt@vnet.ibm.com>
4981 * gcc/config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support
4982 for folding vec_perm.
4984 2018-08-13 Will Schmidt <will_schmidt@vnet.ibm.com>
4986 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin):
4987 Add support for gimple-folding of vec_pack() and vec_unpack()
4990 2018-08-13 Will Schmidt <will_schmidt@vnet.ibm.com>
4992 * config/rs6000/rs6000.c (rs6000_builtin_valid_without_lhs): Add
4993 vec_xst variants to the list.
4994 (rs6000_gimple_fold_builtin): Add support for folding unaligned
4995 vector loads and stores.
4997 2018-08-13 David Edelsohn <dje.gcc@gmail.com>
4999 * config.gcc (rs6000-ibm-aix4.x): Delete.
5000 (rs6000-ibm-aix5.1): Delete.
5001 (rs6000-ibm-aix5.2): Delete.
5002 (rs6000-ibm-aix5.3): Delete.
5003 * config/rs6000/aix43.h: Delete.
5004 * config/rs6000/aix51.h: Delete.
5005 * config/rs6000/aix52.h: Delete.
5006 * config/rs6000/t-aix43: Delete.
5008 2018-08-13 Ilya Leoshkevich <iii@linux.ibm.com>
5010 * config/s390/s390.c (s390_decompose_constant_pool_ref):
5012 (s390_decompose_address): Factor out constant pool ref
5015 2018-08-12 Chung-Ju Wu <jasonwucj@gmail.com>
5017 * config/nds32/nds32-predicates.c
5018 (nds32_can_use_bclr_p): Change return type as bool.
5019 (nds32_can_use_bset_p): Ditto.
5020 (nds32_can_use_btgl_p): Ditto.
5021 (nds32_can_use_bitci_p): Ditto.
5022 * config/nds32/nds32-protos.h
5023 (nds32_can_use_bclr_p): Change declaration.
5024 (nds32_can_use_bset_p): Ditto.
5025 (nds32_can_use_btgl_p): Ditto.
5026 (nds32_can_use_bitci_p): Ditto.
5028 2018-08-12 Chung-Ju Wu <jasonwucj@gmail.com>
5030 * config/nds32/nds32.c (nds32_expand_prologue, nds32_expand_epilogue):
5031 Support -msched-prolog-epilog option.
5032 * config/nds32/nds32.opt (msched-prolog-epilog): New option.
5034 2018-08-12 Chung-Ju Wu <jasonwucj@gmail.com>
5036 * common/config/nds32/nds32-common.c
5037 (nds32_option_optimization_table): Enalbe -malways-align.
5039 2018-08-12 Chung-Ju Wu <jasonwucj@gmail.com>
5041 * config.gcc (nds32*): Add nds32_isr.h and nds32_init.inc in
5043 * common/config/nds32/nds32-common.c (nds32_handle_option): Handle
5044 OPT_misr_secure_ case.
5045 * config/nds32/nds32-isr.c: Implementation of backward compatibility.
5046 * config/nds32/nds32-protos.h (nds32_isr_function_critical_p): New.
5047 * config/nds32/nds32.c (nds32_attribute_table): Add critical and
5049 * config/nds32/nds32.h (nds32_isr_nested_type): Add NDS32_CRITICAL.
5050 (nds32_isr_info): New field security_level.
5051 (TARGET_ISR_VECTOR_SIZE_4_BYTE): New macro.
5052 * config/nds32/nds32.md (return_internal): Consider critical attribute.
5053 * config/nds32/nds32.opt (misr-secure): New option.
5054 * config/nds32/nds32_init.inc: New file.
5055 * config/nds32/nds32_isr.h: New file.
5057 2018-08-11 John David Anglin <danglin@gcc.gnu.org>
5059 * config/pa/pa.md (UNSPEC_MEMORY_BARRIER): New unspec enum.
5060 Update comment for atomic instructions.
5061 (atomic_storeqi, atomic_storehi, atomic_storesi, atomic_storesf,
5062 atomic_loaddf, atomic_loaddf_1, atomic_storedf, atomic_storedf_1):
5064 (atomic_loaddi): Revise fence expansion to only emit fence prior to
5065 load for __ATOMIC_SEQ_CST model.
5066 (atomic_loaddi_1): Remove float register target.
5067 (atomic_storedi): Handle CONST_INT values.
5068 (atomic_storedi_1): Remove float register source. Add special case
5070 (memory_barrier): New expander and insn.
5072 2018-08-11 Jakub Jelinek <jakub@redhat.com>
5074 PR tree-optimization/86835
5075 * tree-ssa-math-opts.c (insert_reciprocals): Even when inserting
5076 new_stmt after def_gsi, make sure to insert new_square_stmt after
5077 that stmt, not 2 stmts before it.
5079 2018-08-10 Alexander Monakov <amonakov@ispras.ru>
5082 * config/i386/i386.md (<s>mul<mode>3_highpart): Use DWIH mode iterator
5085 2018-08-10 Martin Liska <mliska@suse.cz>
5088 * builtin-types.def (BT_FN_LONG_LONG_LONG_DOUBLE): Add new
5090 * builtins.c (expand_builtin_expect_with_probability):
5092 (expand_builtin_expect_with_probability): New function.
5093 (build_builtin_expect_predicate): Add new argumnet probability
5094 for BUILT_IN_EXPECT_WITH_PROBABILITY.
5095 (fold_builtin_expect):
5098 * builtins.def (BUILT_IN_EXPECT_WITH_PROBABILITY):
5099 * builtins.h (fold_builtin_expect): Set new argument.
5100 * doc/extend.texi: Document __builtin_expect_with_probability.
5101 * doc/invoke.texi: Likewise.
5102 * gimple-fold.c (gimple_fold_call): Pass new argument.
5103 * ipa-fnsummary.c (find_foldable_builtin_expect): Handle
5104 also BUILT_IN_EXPECT_WITH_PROBABILITY.
5105 * predict.c (get_predictor_value): New function.
5106 (expr_expected_value): Add new argument probability. Assume
5107 that predictor and probability are always non-null.
5108 (expr_expected_value_1): Likewise. For __builtin_expect and
5109 __builtin_expect_with_probability set probability. Handle
5110 combination in binary expressions.
5111 (tree_predict_by_opcode): Simplify code by simply calling
5112 get_predictor_value.
5113 (pass_strip_predict_hints::execute): Add handling of
5114 BUILT_IN_EXPECT_WITH_PROBABILITY.
5115 * predict.def (PRED_BUILTIN_EXPECT_WITH_PROBABILITY): Add
5117 * tree.h (DECL_BUILT_IN_P): New function.
5119 2018-08-10 Martin Liska <mliska@suse.cz>
5121 PR tree-optimization/85799
5122 * passes.def: Add argument for pass_strip_predict_hints.
5123 * predict.c (class pass_strip_predict_hints): Add new argument
5125 (strip_predictor_early): New function.
5126 (pass_strip_predict_hints::execute): Call the function to
5128 (strip_predict_hints): New function.
5129 * predict.def: Fix comment.
5131 2018-08-10 Thomas Preud'homme <thomas.preudhomme@linaro.org>
5133 * Makefile.in: Clarify which tm.texi to copy over to assert the
5134 right to grant a GFDL license for all.
5136 2018-08-09 Jeff Law <law@redhat.com>
5138 * config/m68k/m68k.c (m68k_adjust_decorated_operand): Remove
5141 2018-08-09 Andreas Schwab <schwab@linux-m68k.org>
5143 * config/m68k/m68k-protos.h (m68k_final_prescan_insn): Remove
5146 2018-08-09 Richard Sandiford <richard.sandiford@arm.com>
5148 * tree-vect-loop.c (vectorizable_reduction): Allow inner-loop
5149 reductions for variable-length vectors.
5151 2018-08-09 David Malcolm <dmalcolm@redhat.com>
5154 * common.opt (fdiagnostics-show-line-numbers): New option.
5155 * diagnostic-show-locus.c (class layout): Add fields
5156 "m_show_line_numbers_p" and "m_linenum_width";
5157 (num_digits): New function.
5158 (test_num_digits): New function.
5159 (layout::layout): Initialize new fields. Update m_x_offset
5160 logic to handle any left margin.
5161 (layout::print_source_line): Print line number when requested.
5162 (layout::start_annotation_line): New member function.
5163 (layout::print_annotation_line): Call it.
5164 (layout::print_leading_fixits): Likewise.
5165 (layout::print_trailing_fixits): Likewise. Update calls to
5166 move_to_column for new parameter.
5167 (layout::get_x_bound_for_row): Add "add_left_margin" param and use
5168 it to potentially call start_annotation_line.
5169 (layout::show_ruler): Call start_annotation_line.
5170 (selftest::test_line_numbers_multiline_range): New selftest.
5171 (selftest::diagnostic_show_locus_c_tests): Call test_num_digits
5172 and selftest::test_line_numbers_multiline_range.
5173 * diagnostic.c (diagnostic_initialize): Initialize
5174 show_line_numbers_p.
5175 * diagnostic.h (struct diagnostic_context): Add field
5176 "show_line_numbers_p".
5177 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
5178 -fno-diagnostics-show-line-numbers.
5179 * dwarf2out.c (gen_producer_string): Add
5180 OPT_fdiagnostics_show_line_numbers to the ignored options.
5181 * lto-wrapper.c (merge_and_complain): Likewise to the "pick
5182 one setting" options.
5183 (append_compiler_options): Likewise to the dropped options.
5184 (append_diag_options): Likewise to the passed-on options.
5185 * opts.c (common_handle_option): Handle the new option.
5186 * toplev.c (general_init): Set up global_dc->show_line_numbers_p.
5188 2018-08-09 Kelvin Nilsen <kelvin@gcc.gnu.org>
5190 * doc/extend.texi (PowerPC AltiVec Built-in Functions Available on
5191 ISA 2.07): Correct spelling of bcdsub to be __builtin_bcdsub. Add
5192 third argument of type "const signed char" to descriptions of
5193 __builtin_bcdadd, __builtin_bcdadd_lt, __builtin_bcdadd_eq,
5194 __builtin_bcdadd_gt, __builtin_bcdadd_ov, __builtin_bcdsub,
5195 __builtin_bcdsub_lt, __builtin_bcdsub_eq, __builtin_bcdsub_gt,
5196 __builtin_bcdsub_ov functions.
5198 2018-08-09 Richard Sandiford <richard.sandiford@arm.com>
5200 PR tree-optimization/86858
5201 * tree-vect-loop.c (vect_is_simple_reduction): Restore
5202 flow_bb_inside_loop_p calls.
5204 2018-08-09 Richard Sandiford <richard.sandiford@arm.com>
5206 PR tree-optimization/86871
5207 * tree-vect-stmts.c (vect_transform_stmt): Use gimple_get_lhs
5208 instead of gimple_assign_lhs.
5210 2018-08-09 Richard Earnshaw <rearnsha@arm.com>
5213 * config/aarch64/aarch64.md (add<mode>3_carryinC_zero): Add missing
5214 register constraint to operand 0.
5215 (add<mode>3_carryinC): Likewise.
5216 (add<mode>3_carryinV_zero, add<mode>3_carryinV): Likewise.
5218 2018-08-09 Martin Liska <mliska@suse.cz>
5221 * common.opt: Remove extra line.
5223 2018-08-09 Martin Liska <mliska@suse.cz>
5225 * params.def (PARAM_ALIGN_LOOP_ITERATIONS): Remove double dots
5226 at the end of a line, make first letter capital and end up
5227 a sentence with a dot.
5228 (PARAM_LOOP_INTERCHANGE_STRIDE_RATIO): Likewise.
5229 (PARAM_LOOP_BLOCK_TILE_SIZE): Likewise.
5230 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Likewise.
5231 (PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP): Likewise.
5232 (PARAM_MAX_ISL_OPERATIONS): Likewise.
5233 (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): Likewise.
5234 (PARAM_PROFILE_FUNC_INTERNAL_ID): Likewise.
5235 (PARAM_INDIR_CALL_TOPN_PROFILE): Likewise.
5236 (PARAM_SLP_MAX_INSNS_IN_BB): Likewise.
5237 (PARAM_IPA_CP_EVAL_THRESHOLD): Likewise.
5238 (PARAM_IPA_CP_RECURSION_PENALTY): Likewise.
5239 (PARAM_IPA_CP_SINGLE_CALL_PENALTY): Likewise.
5240 (PARAM_IPA_CP_LOOP_HINT_BONUS): Likewise.
5241 (PARAM_IPA_CP_ARRAY_INDEX_HINT_BONUS): Likewise.
5242 (PARAM_TREE_REASSOC_WIDTH): Likewise.
5243 (PARAM_HSA_GEN_DEBUG_STORES): Likewise.
5244 (PARAM_MAX_SPECULATIVE_DEVIRT_MAYDEFS): Likewise.
5245 (PARAM_MAX_VRP_SWITCH_ASSERTIONS): Likewise.
5247 2018-08-09 Andreas Krebbel <krebbel@linux.ibm.com>
5250 * config/s390/s390.c (s390_option_override_internal): Reduce the
5251 stack-clash-protection-probe-interval param if it would be too big
5254 2018-08-08 Andreas Schwab <schwab@linux-m68k.org>
5257 * config/m68k/m68k.h (FINAL_PRESCAN_INSN): Don't define.
5258 * config/m68k/m68k.c (handle_move_double): Don't call
5259 m68k_final_prescan_insn.
5260 (m68k_adjust_decorated_operand): Renamed from
5261 m68k_final_prescan_insn, remove first and third operand and
5263 (print_operand): Call it.
5264 (print_operand_address): Call it.
5266 2018-08-08 Nathan Sidwell <nathan@acm.org>
5268 * diagnostic.c (diagnostic_report_current_module): Use
5269 linemap_included_from & linemap_included_from_linemap.
5271 2018-08-08 Hongbo Zhang <hongbo.zhang@linaro.org>
5273 * config/aarch64/aarch64-cores.def: Add phecda core.
5274 * config/aarch64/aarch64-tune.md: Regenerate.
5275 * doc/invoke.texi: Add phecda core.
5277 2018-08-08 Andreas Krebbel <krebbel@linux.ibm.com>
5280 * config/s390/constraints.md ("NxHD0", "NxSD0"): New constraint
5282 * config/s390/s390.md ("movti"): Add more alternatives for
5283 constant to GPR copies.
5285 2018-08-08 Andreas Krebbel <krebbel@linux.ibm.com>
5287 * config/s390/s390.c: Fix whitespace damage throughout the file.
5288 * config/s390/s390.h: Likewise.
5289 * config/s390/tpf.h: Likewise.
5291 2018-08-08 Ilya Leoshkevich <iii@linux.ibm.com>
5293 * config/s390/s390.c (s390_loadrelative_operand_p):
5294 Remove TARGET_CPU_ZARCH usages.
5295 (s390_rtx_costs): Likewise.
5296 (s390_legitimate_constant_p): Likewise.
5297 (s390_cannot_force_const_mem): Likewise.
5298 (legitimate_reload_constant_p): Likewise.
5299 (s390_preferred_reload_class): Likewise.
5300 (legitimize_pic_address): Likewise.
5301 (legitimize_tls_address): Likewise.
5302 (s390_split_branches): Removed.
5303 (s390_add_execute): Removed.
5304 (s390_dump_pool): Remove TARGET_CPU_ZARCH usages.
5305 (s390_mainpool_start): Likewise.
5306 (s390_mainpool_finish): Likewise.
5307 (s390_mainpool_cancel): Removed.
5308 (s390_chunkify_start): Remove TARGET_CPU_ZARCH usages.
5309 (s390_chunkify_cancel): Likewise.
5310 (s390_return_addr_rtx): Likewise.
5311 (s390_register_info): Remove split_branches_pending_p uages.
5312 (s390_optimize_register_info): Likewise.
5313 (s390_init_frame_layout): Remove TARGET_CPU_ZARCH and
5314 split_branches_pending_p usages.
5315 (s390_can_eliminate): Remove TARGET_CPU_ZARCH usages.
5316 (s390_load_got): Likewise.
5317 (s390_expand_split_stack_prologue): Likewise.
5318 (output_asm_nops): Likewise.
5319 (s390_function_profiler): Likewise.
5320 (s390_emit_call): Likewise.
5321 (s390_conditional_register_usage): Likewise.
5322 (s390_optimize_prologue): Likewise.
5323 (s390_reorg): Remove TARGET_CPU_ZARCH and
5324 split_branches_pending_p usages.
5325 (s390_option_override_internal): Remove TARGET_CPU_ZARCH
5327 (s390_output_indirect_thunk_function): Likewise.
5328 * config/s390/s390.h (TARGET_CPU_ZARCH): Removed.
5329 (TARGET_CPU_ZARCH_P): Removed.
5330 (struct machine_function): Remove split_branches_pending_p.
5331 * config/s390/s390.md: Remove TARGET_CPU_ZARCH usages.
5333 2018-08-08 Ilya Leoshkevich <iii@linux.ibm.com>
5335 * common/config/s390/s390-common.c (processor_flags_table):
5337 * config.gcc: Remove with_arch/with_tune support.
5338 * config/s390/2064.md: Remove cpu attribute comparisons.
5339 * config/s390/driver-native.c (s390_host_detect_local_cpu):
5341 * config/s390/linux.h (ASM_SPEC):
5342 Remove -march support.
5343 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal):
5344 Use a table to get an arch level.
5345 * config/s390/s390-opts.h (enum processor_type):
5347 * config/s390/s390.c
5348 (processor_table): Remove entries, add arch_level values.
5349 (s390_issue_rate): Remove cases.
5350 (s390_option_override): Adjust
5351 s390_option_override_internal() call.
5352 (s390_option_override_internal): Remove deprecation warning.
5353 (s390_valid_target_attribute_tree): Adjust
5354 s390_option_override_internal() call.
5355 * config/s390/s390.h (struct s390_processor):
5356 Share with s390-c.c, add arch_level field.
5357 * config/s390/s390.md:
5358 Remove occurrences in cpu attribute.
5359 * config/s390/s390.opt: Remove -march/-mtune support.
5360 * config/s390/tpf.h (ASM_SPEC): Remove -march support.
5361 * doc/invoke.texi: Remove deprecation warning.
5363 2018-08-08 Luis Machado <luis.machado@linaro.org>
5365 * config/aarch64/aarch64.c (qdf24xx_vector_cost): New static
5367 (qdf24xx_tunings): Set vector cost structure to
5368 qdf24xx_vector_cost.
5370 * config/aarch64/aarch64.c (qdf24xx_addrcost_table)
5371 <register_sextend>: Set to 3.
5373 2018-08-07 Richard Sandiford <richard.sandiford@arm.com>
5376 * config/aarch64/iterators.md (FRECP, frecp_suffix): Delete.
5377 * config/aarch64/aarch64-simd.md
5378 (aarch64_frecp<FRECP:frecp_suffix><mode>): Fold FRECPE into...
5379 (@aarch64_frecpe<mode>): ...here and the move FRECPX to...
5380 (aarch64_frecpx<mode>): ...this new pattern.
5381 * config/aarch64/aarch64-simd-builtins.def: Remove comment
5382 about aarch64_frecp<FRECP:frecp_suffix><mode>.
5384 2018-08-07 Martin Liska <mliska@suse.cz>
5387 * predict.c (expr_expected_value_1): Handle DECL_IS_MALLOC,
5388 BUILT_IN_REALLOC and DECL_IS_OPERATOR_NEW.
5389 * predict.def (PRED_MALLOC_NONNULL): New predictor.
5390 * doc/extend.texi: Document that malloc attribute adds
5393 2018-08-06 John David Anglin <danglin@gcc.gnu.org>
5396 * config/pa/pa.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
5397 Define to speculation_safe_value_not_needed.
5399 2018-08-06 Jeff Law <law@redhat.com>
5401 * tree-ssa-dom.c (dom_opt_dom_walker::optimize_stmt): Pass down
5402 the vr_values instance to cprop_into_stmt.
5403 (cprop_into_stmt): Pass vr_values instance down to cprop_operand.
5404 (cprop_operand): Also query EVRP to determine if OP is a constant.
5406 2018-08-06 Nathan Sidwell <nathan@acm.org>
5408 * diagnostic.c (diagnostic_report_current_module): Reroll
5409 included-at loop. Translate text.
5411 2018-08-06 David Malcolm <dmalcolm@redhat.com>
5413 * function-tests.c (selftest::test_expansion_to_rtl): Call
5414 free_after_compilation.
5416 2018-08-06 Alan Hayward <alan.hayward@arm.com>
5418 * config/aarch64/aarch64.md: Add clobber highs to tls_desc.
5420 2018-08-06 Andreas Krebbel <krebbel@linux.ibm.com>
5422 * config/s390/s390.c (s390_loop_unroll_adjust): Prevent small
5423 loops with memory block operations from getting unrolled.
5425 2018-08-06 Ulrich Weigand <uweigand@de.ibm.com>
5428 * config/spu/spu.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
5429 Define to speculation_safe_value_not_needed.
5431 2018-08-06 Jeff Law <law@redhat.com>
5433 * reload1.c (forget_old_reloads_1): Adjust CLOBBER_HIGH
5436 2018-08-06 Jozef Lawrynowicz <jozef.l@mittosystems.com>
5439 * gcc/tree.c (build_common_tree_nodes): Initialize integer_types array
5440 with all enabled __intN types.
5442 * gcc/testsuite/gcc.target/msp430/pr86662.c: New test.
5444 2018-08-06 Alan Hayward <alan.hayward@arm.com>
5446 * alias.c (record_set): Check for clobber high.
5447 * cfgexpand.c (expand_gimple_stmt): Likewise.
5448 * combine-stack-adj.c (single_set_for_csa): Likewise.
5449 * combine.c (find_single_use_1): Likewise.
5450 (set_nonzero_bits_and_sign_copies): Likewise.
5451 (get_combine_src_dest): Likewise.
5452 (is_parallel_of_n_reg_sets): Likewise.
5453 (try_combine): Likewise.
5454 (record_dead_and_set_regs_1): Likewise.
5455 (reg_dead_at_p_1): Likewise.
5456 (reg_dead_at_p): Likewise.
5457 * dce.c (deletable_insn_p): Likewise.
5458 (mark_nonreg_stores_1): Likewise.
5459 (mark_nonreg_stores_2): Likewise.
5460 * df-scan.c (df_find_hard_reg_defs): Likewise.
5461 (df_uses_record): Likewise.
5462 (df_get_call_refs): Likewise.
5463 * dwarf2out.c (mem_loc_descriptor): Likewise.
5464 * haifa-sched.c (haifa_classify_rtx): Likewise.
5465 * ira-build.c (create_insn_allocnos): Likewise.
5466 * ira-costs.c (scan_one_insn): Likewise.
5467 * ira.c (equiv_init_movable_p): Likewise.
5468 (rtx_moveable_p): Likewise.
5469 (interesting_dest_for_shprep): Likewise.
5470 * jump.c (mark_jump_label_1): Likewise.
5471 * postreload-gcse.c (record_opr_changes): Likewise.
5472 * postreload.c (reload_cse_simplify): Likewise.
5473 (struct reg_use): Add source expr.
5474 (reload_combine): Check for clobber high.
5475 (reload_combine_note_use): Likewise.
5476 (reload_cse_move2add): Likewise.
5477 (move2add_note_store): Likewise.
5478 * print-rtl.c (print_pattern): Likewise.
5479 * recog.c (decode_asm_operands): Likewise.
5480 (store_data_bypass_p): Likewise.
5481 (if_test_bypass_p): Likewise.
5482 * regcprop.c (kill_clobbered_value): Likewise.
5483 (kill_set_value): Likewise.
5484 * reginfo.c (reg_scan_mark_refs): Likewise.
5485 * reload1.c (maybe_fix_stack_asms): Likewise.
5486 (eliminate_regs_1): Likewise.
5487 (elimination_effects): Likewise.
5488 (mark_not_eliminable): Likewise.
5489 (scan_paradoxical_subregs): Likewise.
5490 (forget_old_reloads_1): Likewise.
5491 * reorg.c (find_end_label): Likewise.
5492 (try_merge_delay_insns): Likewise.
5493 (redundant_insn): Likewise.
5494 (own_thread_p): Likewise.
5495 (fill_simple_delay_slots): Likewise.
5496 (fill_slots_from_thread): Likewise.
5497 (dbr_schedule): Likewise.
5498 * resource.c (update_live_status): Likewise.
5499 (mark_referenced_resources): Likewise.
5500 (mark_set_resources): Likewise.
5501 * rtl.c (copy_rtx): Likewise.
5502 * rtlanal.c (reg_referenced_p): Likewise.
5503 (single_set_2): Likewise.
5504 (noop_move_p): Likewise.
5505 (note_stores): Likewise.
5506 * sched-deps.c (sched_analyze_reg): Likewise.
5507 (sched_analyze_insn): Likewise.
5509 2018-08-06 Alan Hayward <alan.hayward@arm.com>
5511 * cse.c (invalidate_reg): New function extracted from...
5512 (invalidate): ...here.
5513 (canonicalize_insn): Check for clobber high.
5514 (invalidate_from_clobbers): invalidate clobber highs.
5515 (invalidate_from_sets_and_clobbers): Likewise.
5516 (count_reg_usage): Check for clobber high.
5517 (insn_live_p): Likewise.
5518 * cselib.c (cselib_expand_value_rtx_1):Likewise.
5519 (cselib_invalidate_regno): Check for clobber in setter.
5520 (cselib_invalidate_rtx): Pass through setter.
5521 (cselib_invalidate_rtx_note_stores):
5522 (cselib_process_insn): Check for clobber high.
5523 * cselib.h (cselib_invalidate_rtx): Add operand.
5525 2018-08-06 Alan Hayward <alan.hayward@arm.com>
5527 * lra-eliminations.c (lra_eliminate_regs_1): Check for clobber high.
5528 (mark_not_eliminable): Likewise.
5529 * lra-int.h (struct lra_insn_reg): Add clobber high marker.
5530 * lra-lives.c (process_bb_lives): Check for clobber high.
5531 * lra.c (new_insn_reg): Remember clobber highs.
5532 (collect_non_operand_hard_regs): Check for clobber high.
5533 (lra_set_insn_recog_data): Likewise.
5534 (add_regs_to_insn_regno_info): Likewise.
5535 (lra_update_insn_regno_info): Likewise.
5537 2018-08-06 Alan Hayward <alan.hayward@arm.com>
5539 * rtl.h (reg_is_clobbered_by_clobber_high): Add declarations.
5540 * rtlanal.c (reg_is_clobbered_by_clobber_high): Add function.
5542 2018-08-06 Alan Hayward <alan.hayward@arm.com>
5544 * emit-rtl.c (verify_rtx_sharing): Check for CLOBBER_HIGH.
5545 (copy_insn_1): Likewise.
5546 (gen_hard_reg_clobber_high): New gen function.
5547 * genconfig.c (walk_insn_part): Check for CLOBBER_HIGH.
5548 * genemit.c (gen_exp): Likewise.
5549 (gen_emit_seq): Pass through info.
5550 (gen_insn): Check for CLOBBER_HIGH.
5551 (gen_expand): Pass through info.
5552 (gen_split): Likewise.
5553 (output_add_clobbers): Likewise.
5554 * genrecog.c (validate_pattern): Check for CLOBBER_HIGH.
5555 (remove_clobbers): Likewise.
5556 * rtl.h (gen_hard_reg_clobber_high): New declaration.
5558 2018-08-06 Alan Hayward <alan.hayward@arm.com>
5560 * doc/rtl.texi (clobber_high): Add.
5561 (parallel): Add in clobber high
5562 * rtl.c (rtl_check_failed_code3): Add function.
5563 * rtl.def (CLOBBER_HIGH): Add expression.
5564 * rtl.h (RTL_CHECKC3): Add macro.
5565 (rtl_check_failed_code3): Add declaration.
5566 (XC3EXP): Add macro.
5568 2018-08-05 H.J. Lu <hongjiu.lu@intel.com>
5571 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Set
5572 cfun->machine->max_used_stack_alignment if needed.
5574 2018-08-04 Martin Sebor <msebor@redhat.com>
5576 PR tree-optimization/86571
5577 * gimple-ssa-sprintf.c (format_floating): Extend upper bound of
5580 2018-08-03 Sandra Loosemore <sandra@codesourcery.com>
5583 * config/nios2/nios2.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
5586 2018-08-03 Jeff Law <law@redhat.com>
5589 * config/mn10300/mn10300.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
5590 Define to speculation_safe_value_not_needed.
5592 2018-08-03 David Malcolm <dmalcolm@redhat.com>
5594 * doc/gcov.texi (-x): Remove duplicate "to".
5595 * doc/invoke.texi (-Wnoexcept-type): Remove duplicate "calls".
5596 (-Wif-not-aligned): Remove duplicate "is".
5597 (-flto): Remove duplicate "the".
5598 (MicroBlaze Options): In examples of "-mcpu=cpu-type", remove
5599 duplicate "v5.00.b".
5600 (MSP430 Options): Remove duplicate "and" from the description
5601 of "-mgprel-sec=regexp".
5602 (x86 Options): Remove duplicate copies of "vmldLog102" and
5603 vmlsLog104 from description of "-mveclibabi=type".
5605 2018-08-03 Richard Sandiford <richard.sandiford@arm.com>
5607 * internal-fn.h (first_commutative_argument): Declare.
5608 * internal-fn.c (first_commutative_argument): New function.
5609 * tree-vect-slp.c (vect_get_and_check_slp_defs): Remove extra
5610 restrictions for pattern statements. Use first_commutative_argument
5611 to look for commutative operands in calls to internal functions.
5613 2018-08-03 Aldy Hernandez <aldyh@redhat.com>
5615 * Makefile.in (wide-int-range.o): New.
5616 * tree-vrp.c: Move all the wide_int_* functions to...
5617 * wide-int-range.cc: ...here.
5618 * tree-vrp.h: Move all the wide_int_* prototypes to...
5619 * wide-int-range.h: ...here.
5621 2018-08-03 Tom de Vries <tdevries@suse.de>
5623 * common/config/nvptx/nvptx-common.c (nvptx_except_unwind_info): Return
5625 * config/nvptx/nvptx.c (TARGET_ASM_BYTE_OP): Remove define.
5626 * except.c (output_function_exception_table): Do early exit if
5627 targetm_common.except_unwind_info (&global_options) == UI_NONE.
5629 2018-08-03 Martin Liska <mliska@suse.cz>
5631 * predict.c (dump_prediction): Change to 2 digits
5634 2018-08-03 Siddhesh Poyarekar <siddhesh@sourceware.org>
5636 * config/aarch64/falkor.md (falkor_am_1_vxvy_vxvy): Move
5638 (falkor_am_1_gtov_gtov): ... a new insn reservation.
5640 2018-07-19 Ilya Leoshkevich <iii@linux.ibm.com>
5642 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Replace > with >=.
5643 * df-problems.c (df_remove_dead_eq_notes): Replace > with >=.
5644 * dwarf2out.c (mem_loc_descriptor): Replace > with >=.
5645 * lra-constraints.c (spill_hard_reg_in_range): Replace <= with <.
5646 * lra-remat.c (call_used_input_regno_present_p): Replace <= with <.
5648 2018-08-02 David Malcolm <dmalcolm@redhat.com>
5650 * diagnostic-show-locus.c (diagnostic_show_locus): Use
5651 pp_take_prefix when saving the existing prefix.
5652 * diagnostic.c (diagnostic_append_note): Likewise.
5653 * langhooks.c (lhd_print_error_function): Likewise.
5654 * pretty-print.c (pp_set_prefix): Drop the "const" from "prefix"
5655 param's type. Free the existing prefix.
5656 (pp_take_prefix): New function.
5657 (pretty_printer::pretty_printer): Drop the prefix parameter.
5658 Rename the length parameter to match the comment.
5659 (pretty_printer::~pretty_printer): Free the prefix.
5660 * pretty-print.h (pretty_printer::pretty_printer): Drop the prefix
5662 (struct pretty_printer): Drop the "const" from "prefix" field's
5663 type and clarify memory management.
5664 (pp_set_prefix): Drop the "const" from the 2nd param.
5665 (pp_take_prefix): New decl.
5667 2018-08-02 Aaron Sawdey <acsawdey@linux.ibm.com>
5669 * config/rs6000/rs6000-string.c (select_block_compare_mode): Move test
5670 for word_mode_ok here instead of passing as argument.
5671 (expand_block_compare): Change select_block_compare_mode() call.
5672 (expand_strncmp_gpr_sequence): New function.
5673 (expand_strn_compare): Make use of expand_strncmp_gpr_sequence.
5675 2018-08-02 Jeff Law <law@redhat.com>
5678 * config/m68k/m68k.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
5679 Define to speculation_safe_value_not_needed.
5682 * config/h8300/h8300.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
5683 Define to speculation_safe_value_not_needed.
5685 2018-08-02 Tom de Vries <tdevries@suse.de>
5688 * common/config/nvptx/nvptx-common.c (nvptx_except_unwind_info): New
5689 function. Return UI_TARGET unconditionally.
5690 (TARGET_EXCEPT_UNWIND_INFO): Redefine to nvptx_except_unwind_info.
5691 * config/nvptx/nvptx.c (TARGET_ASM_BYTE_OP): Emit commented out '.byte'.
5693 2018-08-02 Richard Sandiford <richard.sandiford@arm.com>
5695 * genemit.c (print_overload_test): Fix typo.
5697 2018-08-02 Richard Biener <rguenther@suse.de>
5699 PR tree-optimization/86816
5700 * tree-ssa-tail-merge.c (tail_merge_valueize): New function
5701 which checks for value availability before querying it.
5702 (gvn_uses_equal): Use it.
5703 (same_succ_hash): Likewise.
5704 (gimple_equal_p): Likewise.
5706 2018-08-02 Nick Clifton <nickc@redhat.com>
5709 * config/stormy16/stormy16.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
5710 Define to speculation_safe_value_not_needed.
5713 * config/v850/v850.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
5714 Define to speculation_safe_value_not_needed.
5717 * config/v850/v850.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
5718 Define to speculation_safe_value_not_needed.
5721 * config/rx/rx.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
5722 Define to speculation_safe_value_not_needed.
5725 * config/msp430/msp430.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
5726 Define to speculation_safe_value_not_needed.
5729 * config/mcore/mcore.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
5730 Define to speculation_safe_value_not_needed.
5733 * config/m32r/m32r.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
5734 Define to speculation_safe_value_not_needed.
5737 * config/iq2000/iq2000.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
5738 Define to speculation_safe_value_not_needed.
5741 * config/frv/frv.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Define to
5742 speculation_safe_value_not_needed.
5745 * config/fr30/fr30.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Define
5746 to speculation_safe_value_not_needed.
5748 2018-08-02 Richard Sandiford <richard.sandiford@arm.com>
5750 * doc/md.texi: Expand the documentation of instruction names
5751 to mention port-local uses. Document '@' in pattern names.
5752 * read-md.h (overloaded_instance, overloaded_name): New structs.
5754 (md_reader::handle_overloaded_name): New member function.
5755 (md_reader::get_overloads): Likewise.
5756 (md_reader::m_first_overload): New member variable.
5757 (md_reader::m_next_overload_ptr): Likewise.
5758 (md_reader::m_overloads_htab): Likewise.
5759 * read-md.c (md_reader::md_reader): Initialize m_first_overload,
5760 m_next_overload_ptr and m_overloads_htab.
5761 * read-rtl.c (iterator_group): Add "type" and "get_c_token" fields.
5762 (get_mode_token, get_code_token, get_int_token): New functions.
5763 (map_attr_string): Add an optional argument that passes back
5764 the associated iterator.
5765 (overloaded_name_hash, overloaded_name_eq_p, named_rtx_p):
5766 (md_reader::handle_overloaded_name, add_overload_instance): New
5768 (apply_iterators): Handle '@' names. Report an error if '@'
5769 is used without iterators.
5770 (initialize_iterators): Initialize the new iterator_group fields.
5771 * genopinit.c (handle_overloaded_code_for)
5772 (handle_overloaded_gen): New functions.
5773 (main): Use them to print declarations of maybe_code_for_* and
5774 maybe_gen_* functions, and inline definitions of code_for_* and gen_*.
5775 * genemit.c (print_overload_arguments, print_overload_test)
5776 (handle_overloaded_code_for, handle_overloaded_gen): New functions.
5777 (main): Use it to print definitions of maybe_code_for_* and
5778 maybe_gen_* functions.
5779 * config/aarch64/aarch64.c (aarch64_split_128bit_move): Use
5780 gen_aarch64_mov{low,high}_di and gen_aarch64_movdi_{low,high}
5781 instead of explicit mode checks.
5782 (aarch64_split_simd_combine): Likewise gen_aarch64_simd_combine.
5783 (aarch64_split_simd_move): Likewise gen_aarch64_split_simd_mov.
5784 (aarch64_emit_load_exclusive): Likewise gen_aarch64_load_exclusive.
5785 (aarch64_emit_store_exclusive): Likewise gen_aarch64_store_exclusive.
5786 (aarch64_expand_compare_and_swap): Likewise
5787 gen_aarch64_compare_and_swap and gen_aarch64_compare_and_swap_lse
5788 (aarch64_gen_atomic_cas): Likewise gen_aarch64_atomic_cas.
5789 (aarch64_emit_atomic_swap): Likewise gen_aarch64_atomic_swp.
5790 (aarch64_constant_pool_reload_icode): Delete.
5791 (aarch64_secondary_reload): Use code_for_aarch64_reload_movcp
5792 instead of aarch64_constant_pool_reload_icode. Use
5793 code_for_aarch64_reload_mov instead of explicit mode checks.
5794 (rsqrte_type, get_rsqrte_type, rsqrts_type, get_rsqrts_type): Delete.
5795 (aarch64_emit_approx_sqrt): Use gen_aarch64_rsqrte instead of
5796 get_rsqrte_type and gen_aarch64_rsqrts instead of gen_rqrts_type.
5797 (recpe_type, get_recpe_type, recps_type, get_recps_type): Delete.
5798 (aarch64_emit_approx_div): Use gen_aarch64_frecpe instead of
5799 get_recpe_type and gen_aarch64_frecps instead of get_recps_type.
5800 (aarch64_atomic_load_op_code): Delete.
5801 (aarch64_emit_atomic_load_op): Likewise.
5802 (aarch64_gen_atomic_ldop): Use UNSPECV_ATOMIC_* instead of
5803 aarch64_atomic_load_op_code. Use gen_aarch64_atomic_load
5804 instead of aarch64_emit_atomic_load_op.
5805 * config/aarch64/aarch64.md (aarch64_reload_movcp<GPF_TF:mode><P:mode>)
5806 (aarch64_reload_movcp<VALL:mode><P:mode>, aarch64_reload_mov<mode>)
5807 (aarch64_movdi_<mode>low, aarch64_movdi_<mode>high)
5808 (aarch64_mov<mode>high_di, aarch64_mov<mode>low_di): Add a '@'
5809 character before the pattern name.
5810 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
5811 (aarch64_rsqrte<mode>, aarch64_rsqrts<mode>)
5812 (aarch64_simd_combine<mode>, aarch64_frecpe<mode>)
5813 (aarch64_frecps<mode>): Likewise.
5814 * config/aarch64/atomics.md (atomic_compare_and_swap<mode>)
5815 (aarch64_compare_and_swap<mode>, aarch64_compare_and_swap<mode>_lse)
5816 (aarch64_load_exclusive<mode>, aarch64_store_exclusive<mode>)
5817 (aarch64_atomic_swp<mode>, aarch64_atomic_cas<mode>)
5818 (aarch64_atomic_load<atomic_ldop><mode>): Likewise.
5820 2018-08-02 Richard Sandiford <richard.sandiford@arm.com>
5822 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
5823 Allow HFmode constants if TARGET_FP_F16INST.
5825 2018-08-02 Jackson Woodruff <jackson.woodruff@arm.com>
5828 * config/aarch64/aarch64.c (aarch64_operands_adjust_ok_for_ldpstp):
5829 No longer check last store for clobber of address register.
5831 2018-08-02 Martin Liska <mliska@suse.cz>
5833 PR gcov-profile/86817
5834 * gcov.c (process_all_functions): New function.
5836 (process_file): Move functions processing to
5837 process_all_functions.
5839 2018-08-02 David Malcolm <dmalcolm@redhat.com>
5841 * dumpfile.c (dump_user_location_t::dump_user_location_t): Add
5842 "const" to the "gimple *" and "rtx_insn *" parameters.
5843 * dumpfile.h (dump_user_location_t::dump_user_location_t):
5845 (dump_location_t::dump_location_t): Likewise.
5847 2018-08-01 Martin Sebor <msebor@redhat.com>
5849 PR tree-optimization/86650
5850 * gimple-pretty-print.c (percent_G_format): Accept a "gimple *"
5851 rather than a "gcall *". Directly pass the data of interest
5852 to percent_K_format, rather than building a temporary CALL_EXPR
5854 * gimple-fold.c (gimple_fold_builtin_strncpy): Adjust.
5855 (gimple_fold_builtin_strncat): Adjust.
5856 * gimple-ssa-warn-restrict.h (check_bounds_or_overlap): Replace
5857 gcall* argument with gimple*.
5858 * gimple-ssa-warn-restrict.c (check_call): Same.
5859 (wrestrict_dom_walker::before_dom_children): Same.
5860 (builtin_access::builtin_access): Same.
5861 (check_bounds_or_overlap): Same
5862 (maybe_diag_overlap): Same.
5863 (maybe_diag_offset_bounds): Same.
5864 * tree-diagnostic.c (default_tree_printer): Move usage of
5865 EXPR_LOCATION (t) and TREE_BLOCK (t) from within percent_K_format
5867 * tree-pretty-print.c (percent_K_format): Add argument.
5868 * tree-pretty-print.h: Add argument.
5869 * tree-ssa-ccp.c (pass_post_ipa_warn::execute): Adjust.
5870 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Adjust.
5871 (maybe_diag_stxncpy_trunc): Same.
5872 (handle_builtin_stxncpy): Same.
5873 (handle_builtin_strcat): Same.
5875 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
5877 * match.pd: Optimise pointer range checks.
5879 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
5881 PR tree-optimization/86758
5882 * tree-vect-stmts.c (vectorizable_simd_clone_call): Don't try
5883 to remove pattern statements.
5885 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
5887 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Use the
5888 result of dfs_enumerate_from when constructing stmt_vec_infos,
5889 instead of additionally calling get_loop_body.
5891 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
5893 * tree-vectorizer.h (vect_create_data_ref_ptr): Remove inv_p
5895 * tree-vect-data-refs.c (vect_create_data_ref_ptr): Likewise.
5896 When creating an iv, assert that the step is not known to be zero.
5897 (vect_setup_realignment): Update call accordingly.
5898 * tree-vect-stmts.c (vectorizable_store): Likewise.
5899 (vectorizable_load): Likewise. Handle VMAT_INVARIANT separately.
5901 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
5903 * tree-vectorizer.h (vect_stmt_to_vectorize): New function.
5904 * tree-vect-loop.c (vect_update_vf_for_slp): Use it.
5905 (vectorizable_reduction): Likewise.
5906 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
5907 (vect_detect_hybrid_slp_stmts): Likewise.
5908 * tree-vect-stmts.c (vect_is_simple_use): Likewise.
5910 2018-08-01 Aldy Hernandez <aldyh@redhat.com>
5912 * tree-vrp (zero_nonzero_bits_from_bounds): Rename to...
5913 (wide_int_set_zero_nonzero_bits): ...this.
5914 (zero_nonzero_bits_from_vr): Rename to...
5915 (vrp_set_zero_nonzero_bits): ...this.
5916 (extract_range_from_multiplicative_op_1): Abstract wide int
5918 (wide_int_range_multiplicative_op): ...here.
5919 (extract_range_from_binary_expr_1): Extract wide int binary
5920 operations into their own functions.
5921 (wide_int_range_lshift): New.
5922 (wide_int_range_can_optimize_bit_op): New.
5923 (wide_int_range_shift_undefined_p): New.
5924 (wide_int_range_bit_xor): New.
5925 (wide_int_range_bit_ior): New.
5926 (wide_int_range_bit_and): New.
5927 (wide_int_range_trunc_mod): New.
5928 (extract_range_into_wide_ints): New.
5929 (vrp_shift_undefined_p): New.
5930 (extract_range_from_multiplicative_op): New.
5931 (vrp_can_optimize_bit_op): New.
5932 * tree-vrp.h (value_range::dump): New.
5933 (wide_int_range_multiplicative_op): New.
5934 (wide_int_range_lshift):New.
5935 (wide_int_range_shift_undefined_p): New.
5936 (wide_int_range_bit_xor): New.
5937 (wide_int_range_bit_ior): New.
5938 (wide_int_range_bit_and): New.
5939 (wide_int_range_trunc_mod): New.
5940 (zero_nonzero_bits_from_bounds): Rename to...
5941 (wide_int_set_zero_nonzero_bits): ...this.
5942 (zero_nonzero_bits_from_vr): Rename to...
5943 (vrp_set_zero_nonzero_bits): ...this.
5944 (range_easy_mask_min_max): Rename to...
5945 (wide_int_range_can_optimize_bit_op): this.
5946 * vr-values.c (simplify_bit_ops_using_ranges): Rename
5947 zero_nonzero_bits_from_vr into vrp_set_zero_nonzero_bits.
5949 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
5951 * tree-vectorizer.h (vect_orig_stmt): New function.
5952 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Use it.
5953 * tree-vect-loop.c (vect_model_reduction_cost): Likewise.
5954 (vect_create_epilog_for_reduction): Likewise.
5955 (vectorizable_live_operation): Likewise.
5956 * tree-vect-slp.c (vect_find_last_scalar_stmt_in_slp): Likewise.
5957 (vect_detect_hybrid_slp_stmts, vect_schedule_slp): Likewise.
5958 * tree-vect-stmts.c (vectorizable_call): Likewise.
5959 (vectorizable_simd_clone_call, vect_remove_stores): Likewise.
5961 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
5963 * tree-vectorizer.h (vect_transform_stmt): Remove grouped_store
5965 * tree-vect-stmts.c (vect_transform_stmt): Likewise.
5966 * tree-vect-loop.c (vect_transform_loop_stmt): Update call accordingly.
5967 (vect_transform_loop): Likewise.
5968 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
5970 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
5972 * tree-vectorizer.h (vect_schedule_slp): Return void.
5973 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
5974 (vect_schedule_slp): Likewise.
5976 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
5978 * tree-vect-loop.c (vect_transform_loop_stmt): Remove slp_scheduled
5980 (vect_transform_loop): Update calls accordingly. Schedule SLP
5981 instances before the main loop, if any exist.
5983 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
5985 PR tree-optimization/86749
5986 * tree-vect-patterns.c (vect_determine_min_output_precision_1):
5987 If the lhs is used in a COND_EXPR, check that it is being used
5988 as the "then" or "else" value.
5990 2018-08-01 Tom de Vries <tdevries@suse.de>
5993 * config/nvptx/nvptx.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Define to
5994 speculation_safe_value_not_needed.
5996 2018-08-01 Richard Biener <rguenther@suse.de>
5998 * tree-ssa-sccvn.c (visit_phi): Compare invariant addresses
6001 2018-08-01 Martin Liska <mliska@suse.cz>
6003 * value-prof.c (gimple_divmod_fixed_value_transform): Unify
6004 format how successful transformation is dumped.
6005 (gimple_mod_pow2_value_transform): Likewise.
6006 (gimple_mod_subtract_transform): Likewise.
6007 (gimple_stringops_transform): Likewise.
6009 2018-08-01 Martin Liska <mliska@suse.cz>
6012 * value-prof.c (interesting_stringop_to_profile_p):
6013 Simplify the code and add BUILT_IN_MEMMOVE.
6014 (gimple_stringops_transform): Likewise.
6016 2018-08-01 Sam Tebbs <sam.tebbs@arm.com>
6018 * config/aarch64/aarch64-simd.md
6019 (*aarch64_get_lane_zero_extendsi<mode>): Rename to...
6020 (*aarch64_get_lane_zero_extend<GPI:mode><VDQQH:mode>): ... This and
6021 use GPI iterator instead of SI mode.
6023 2018-08-01 Richard Earnshaw <rearnsha@arm.com>
6025 * config/rs6000/rs6000.md (speculation_barrier): Renamed from
6026 rs6000_speculation_barrier.
6027 * config/rs6000/rs6000.c (rs6000_expand_builtin): Adjust for
6028 new barrier pattern name.
6030 2018-08-01 Richard Earnshaw <rearnsha@arm.com>
6032 * config/i386/i386.md (unspecv): Add UNSPECV_SPECULATION_BARRIER.
6033 (speculation_barrier): New insn.
6035 2018-08-01 Richard Biener <rguenther@suse.de>
6038 * graphite.h: Include isl/id.h and isl/space.h to allow build
6041 2018-08-01 Jan Willem Jagersma <jwjagersma@gmail.com>
6044 * dwarf2out.c (dwarf2out_early_finish): Do not generate assembly in LTO
6045 mode for COFF targets.
6046 * defaults.h (TARGET_COFF): Define.
6047 * config/i386/djgpp.h (TARGET_ASM_LTO_START, TARGET_ASM_LTO_END,
6048 TARGET_COFF): Define.
6049 (i386_djgpp_asm_lto_start, i386_djgpp_asm_lto_end): Declare.
6050 * config/i386/djgpp.c (saved_debug_info_level): New static variable.
6051 (i386_djgpp_asm_lto_start, i386_djgpp_asm_lto_end): New functions.
6053 2018-07-31 Alexandre Oliva <oliva@adacore.com>
6055 * gimple-streamer-in.c (input_bb): Restore BB discriminator.
6056 * gimple-streamer-out.c (output_bb): Save it.
6057 * lto-streamer-in.c (input_struct_function_base): Restore
6058 instance discriminator if available. Create map on demand.
6059 * lto-streamer-out.c (output_struct_function_base): Save it if
6061 * final.c (decl_to_instance_map): Document LTO strategy.
6063 2018-07-31 Alexandre Oliva <oliva@adacore.com>
6064 Olivier Hainque <hainque@adacore.com>
6066 * debug.h (decl_to_instance_map_t): New type.
6067 (decl_to_instance_map): Declare.
6068 (maybe_create_decl_to_instance_map): New inline function.
6069 * final.c (bb_discriminator, last_bb_discriminator): New statics,
6070 to track basic block discriminators.
6071 (final_start_function_1): Initialize them.
6072 (final_scan_insn_1): On NOTE_INSN_BASIC_BLOCK, track
6074 (decl_to_instance_map): New variable.
6075 (map_decl_to_instance, maybe_set_discriminator): New functions.
6076 (notice_source_line): Set discriminator.
6078 2018-07-31 Ian Lance Taylor <iant@golang.org>
6080 * targhooks.c (default_have_speculation_safe_value): Add
6083 2018-07-31 David Malcolm <dmalcolm@redhat.com>
6085 * dump-context.h: Include "pretty-print.h".
6086 (dump_context::refresh_dumps_are_enabled): New decl.
6087 (dump_context::emit_item): New decl.
6088 (class dump_context): Add fields "m_test_pp" and
6090 (temp_dump_context::temp_dump_context): Add param "test_pp_flags".
6091 (temp_dump_context::get_dumped_text): New decl.
6092 (class temp_dump_context): Add field "m_pp".
6093 * dumpfile.c (refresh_dumps_are_enabled): Convert to...
6094 (dump_context::refresh_dumps_are_enabled): ...and add a test for
6096 (set_dump_file): Update for above change.
6097 (set_alt_dump_file): Likewise.
6098 (dump_loc): New overload, taking a pretty_printer *.
6099 (dump_context::dump_loc): Call end_any_optinfo. Dump the location
6100 to any test pretty-printer.
6101 (make_item_for_dump_gimple_stmt): New function, adapted from
6102 optinfo::add_gimple_stmt.
6103 (dump_context::dump_gimple_stmt): Call it, and use the result,
6104 eliminating the direct usage of dump_file and alt_dump_file in
6105 favor of indirectly using them via emit_item.
6106 (make_item_for_dump_gimple_expr): New function, adapted from
6107 optinfo::add_gimple_expr.
6108 (dump_context::dump_gimple_expr): Call it, and use the result,
6109 eliminating the direct usage of dump_file and alt_dump_file in
6110 favor of indirectly using them via emit_item.
6111 (make_item_for_dump_generic_expr): New function, adapted from
6113 (dump_context::dump_generic_expr): Call it, and use the result,
6114 eliminating the direct usage of dump_file and alt_dump_file in
6115 favor of indirectly using them via emit_item.
6116 (make_item_for_dump_printf_va): New function, adapted from
6117 optinfo::add_printf_va.
6118 (make_item_for_dump_printf): New function.
6119 (dump_context::dump_printf_va): Call make_item_for_dump_printf_va,
6120 and use the result, eliminating the direct usage of dump_file and
6121 alt_dump_file in favor of indirectly using them via emit_item.
6122 (make_item_for_dump_dec): New function.
6123 (dump_context::dump_dec): Call it, and use the result,
6124 eliminating the direct usage of dump_file and alt_dump_file in
6125 favor of indirectly using them via emit_item.
6126 (make_item_for_dump_symtab_node): New function, adapted from
6127 optinfo::add_symtab_node.
6128 (dump_context::dump_symtab_node): Call it, and use the result,
6129 eliminating the direct usage of dump_file and alt_dump_file in
6130 favor of indirectly using them via emit_item.
6131 (dump_context::begin_scope): Reimplement, avoiding direct usage
6132 of dump_file and alt_dump_file in favor of indirectly using them
6134 (dump_context::emit_item): New member function.
6135 (temp_dump_context::temp_dump_context): Add param "test_pp_flags".
6136 Set up test pretty-printer on the underlying context. Call
6137 refresh_dumps_are_enabled.
6138 (temp_dump_context::~temp_dump_context): Call
6139 refresh_dumps_are_enabled.
6140 (temp_dump_context::get_dumped_text): New member function.
6141 (selftest::verify_dumped_text): New function.
6142 (ASSERT_DUMPED_TEXT_EQ): New macro.
6143 (selftest::test_capture_of_dump_calls): Run all tests twice, with
6144 and then without optinfo enabled. Add uses of
6145 ASSERT_DUMPED_TEXT_EQ to all tests. Add test of nested scopes.
6146 * dumpfile.h: Update comment for the dump_* API.
6147 * optinfo-emit-json.cc
6148 (selftest::test_building_json_from_dump_calls): Update for new
6149 param for temp_dump_context ctor.
6150 * optinfo.cc (optinfo_item::optinfo_item): Remove "owned" param
6151 and "m_owned" field.
6152 (optinfo_item::~optinfo_item): Likewise.
6153 (optinfo::add_item): New member function.
6154 (optinfo::emit): Update comment.
6155 (optinfo::add_string): Delete.
6156 (optinfo::add_printf): Delete.
6157 (optinfo::add_printf_va): Delete.
6158 (optinfo::add_gimple_stmt): Delete.
6159 (optinfo::add_gimple_expr): Delete.
6160 (optinfo::add_tree): Delete.
6161 (optinfo::add_symtab_node): Delete.
6162 (optinfo::add_dec): Delete.
6163 * optinfo.h (class dump_context): New forward decl.
6164 (optinfo::add_item): New decl.
6165 (optinfo::add_string): Delete.
6166 (optinfo::add_printf): Delete.
6167 (optinfo::add_printf_va): Delete.
6168 (optinfo::add_gimple_stmt): Delete.
6169 (optinfo::add_gimple_expr): Delete.
6170 (optinfo::add_tree): Delete.
6171 (optinfo::add_symtab_node): Delete.
6172 (optinfo::add_dec): Delete.
6173 (optinfo::add_poly_int): Delete.
6174 (optinfo_item::optinfo_item): Remove "owned" param.
6175 (class optinfo_item): Remove field "m_owned".
6177 2018-07-31 Jozef Lawrynowicz <jozef.l@mittosystems.com>
6180 * gcc/cfgexpand.c (set_parm_rtl): Use the alignment of Pmode when
6181 MAX_SUPPORTED_STACK_ALIGNMENT would otherwise be exceeded by the
6182 requested variable alignment.
6183 (expand_one_ssa_partition): Likewise.
6184 (expand_one_var): Likewise.
6186 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
6188 * config/pdp11/pdp11.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Redefine
6189 to speculation_safe_value_not_needed.
6191 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
6193 * targhooks.h (speculation_safe_value_not_needed): New prototype.
6194 * targhooks.c (speculation_safe_value_not_needed): New function.
6195 * target.def (have_speculation_safe_value): Update documentation.
6196 * doc/tm.texi: Regenerated.
6198 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
6200 * config/aarch64/iterators.md (ALLI_TI): New iterator.
6201 * config/aarch64/aarch64.md (despeculate_copy<ALLI_TI:mode>): New
6203 (despeculate_copy<ALLI:mode>_insn): New insn.
6204 (despeculate_copyti_insn): New insn.
6205 (despeculate_simple<ALLI:mode>): New insn
6206 (despeculate_simpleti): New insn.
6207 * config/aarch64/aarch64.c (aarch64_speculation_safe_value): New
6209 (TARGET_SPECULATION_SAFE_VALUE): Redefine to
6210 aarch64_speculation_safe_value.
6211 (aarch64_print_operand): Handle const0_rtx in modifier 'H'.
6213 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
6215 * config/aarch64/aarch64-speculation.cc: New file.
6216 * config/aarch64/aarch64-passes.def (pass_track_speculation): Add
6217 before pass_reorder_blocks.
6218 * config/aarch64/aarch64-protos.h (make_pass_track_speculation): Add
6220 * config/aarch64/aarch64.c (aarch64_conditional_register_usage): Fix
6221 X14 and X15 when tracking speculation.
6222 * config/aarch64/aarch64.md (register name constants): Add
6223 SPECULATION_TRACKER_REGNUM and SPECULATION_SCRATCH_REGNUM.
6224 (unspec): Add UNSPEC_SPECULATION_TRACKER.
6225 (speculation_barrier): New insn attribute.
6226 (cmp<mode>): Allow SP in comparisons.
6227 (speculation_tracker): New insn.
6228 (speculation_barrier): Add speculation_barrier attribute.
6229 * config/aarch64/t-aarch64: Add make rule for aarch64-speculation.o.
6230 * config.gcc (aarch64*-*-*): Add aarch64-speculation.o to extra_objs.
6231 * doc/invoke.texi (AArch64 Options): Document -mtrack-speculation.
6233 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
6235 * config/aarch64/aarch64.md (cb<optab><mode>1): Disable when
6236 aarch64_track_speculation is true.
6237 (tb<optab><mode>1): Likewise.
6238 * config/aarch64/aarch64.c (aarch64_split_compare_regs): Do not
6239 generate CB[N]Z when tracking speculation.
6240 (aarch64_split_compare_and_swap): Likewise.
6241 (aarch64_split_atomic_op): Likewise.
6243 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
6245 * config/aarch64/aarch64.opt (mtrack-speculation): New target option.
6247 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
6249 * config/aarch64.md (unspecv): Add UNSPECV_SPECULAION_BARRIER.
6250 (speculation_barrier): New insn.
6252 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
6254 * config/arm/unspecs.md (unspecv): Add VUNSPEC_SPECULATION_BARRIER.
6255 * config/arm/arm.md (speculation_barrier): New expand.
6256 (speculation_barrier_insn): New pattern.
6258 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
6260 * builtin-types.def (BT_FN_PTR_PTR_VAR): New function type.
6261 (BT_FN_I1_I1_VAR, BT_FN_I2_I2_VAR, BT_FN_I4_I4_VAR): Likewise.
6262 (BT_FN_I8_I8_VAR, BT_FN_I16_I16_VAR): Likewise.
6263 * builtin-attrs.def (ATTR_NOVOPS_NOTHROW_LEAF_LIST): New attribute
6265 * builtins.def (BUILT_IN_SPECULATION_SAFE_VALUE_N): New builtin.
6266 (BUILT_IN_SPECULATION_SAFE_VALUE_PTR): New internal builtin.
6267 (BUILT_IN_SPECULATION_SAFE_VALUE_1): Likewise.
6268 (BUILT_IN_SPECULATION_SAFE_VALUE_2): Likewise.
6269 (BUILT_IN_SPECULATION_SAFE_VALUE_4): Likewise.
6270 (BUILT_IN_SPECULATION_SAFE_VALUE_8): Likewise.
6271 (BUILT_IN_SPECULATION_SAFE_VALUE_16): Likewise.
6272 * builtins.c (expand_speculation_safe_value): New function.
6273 (expand_builtin): Call it.
6274 * doc/cpp.texi: Document predefine __HAVE_SPECULATION_SAFE_VALUE.
6275 * doc/extend.texi: Document __builtin_speculation_safe_value.
6276 * doc/md.texi: Document "speculation_barrier" pattern.
6277 * doc/tm.texi.in: Pull in TARGET_SPECULATION_SAFE_VALUE and
6278 TARGET_HAVE_SPECULATION_SAFE_VALUE.
6279 * doc/tm.texi: Regenerated.
6280 * target.def (have_speculation_safe_value, speculation_safe_value): New
6282 * targhooks.c (default_have_speculation_safe_value): New function.
6283 (default_speculation_safe_value): New function.
6284 * targhooks.h (default_have_speculation_safe_value): Add prototype.
6285 (default_speculation_safe_value): Add prototype.
6287 2018-07-31 David Malcolm <dmalcolm@redhat.com>
6289 * dump-context.h (dump_context::dump_loc): New decl.
6290 * dumpfile.c (dump_context::dump_loc): New member function.
6291 (dump_context::dump_gimple_stmt_loc): Reimplement using dump_loc
6292 and dump_gimple_stmt.
6293 (dump_context::dump_gimple_expr_loc): Likewise, using
6295 (dump_context::dump_generic_expr_loc): Likewise, using
6297 (dump_context::dump_printf_loc_va): Likewise, using
6299 (dump_context::begin_scope): Explicitly using the global function
6300 "dump_loc", rather than the member function.
6302 2018-07-31 Martin Sebor <msebor@redhat.com>
6304 PR tree-optimization/86741
6305 * tree-vrp.c (vrp_prop::check_mem_ref): Avoid incomplete types.
6307 2018-07-31 Andreas Krebbel <krebbel@linux.ibm.com>
6309 * config/s390/s390.c (s390_expand_setmem): Make the unrolling to
6310 depend on whether prefetch instructions will be emitted or not.
6311 Use TARGET_SETMEM_PFD for checking whether prefetch instructions
6312 will be emitted or not.
6313 * config/s390/s390.h (TARGET_SETMEM_PREFETCH_DISTANCE)
6314 (TARGET_SETMEM_PFD): New macros.
6316 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
6318 * tree-vectorizer.h (stmt_vec_info): Turn back into a typedef.
6319 (NULL_STMT_VEC_INFO): Delete.
6320 (stmt_vec_info::operator*): Likewise.
6321 (stmt_vec_info::operator gimple *): Likewise.
6322 * tree-vect-loop.c (vectorizable_reduction): Use NULL instead
6323 of NULL_STMT_VEC_INFO.
6324 * tree-vect-patterns.c (vect_init_pattern_stmt): Likewise.
6325 (vect_reassociating_reduction_p): Likewise.
6326 * tree-vect-stmts.c (vect_build_gather_load_calls): Likewise.
6327 (vectorizable_store): Likewise.
6328 * tree-vectorizer.c (vec_info::set_vinfo_for_stmt): Likewise.
6329 (vec_info::free_stmt_vec_infos): Likewise.
6331 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
6333 * tree-vectorizer.h (vect_stmt_in_region_p): Delete.
6334 * tree-vectorizer.c (vect_stmt_in_region_p): Likewise.
6336 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
6338 * tree-vectorizer.h (vec_info::new_vinfo_for_stmt)
6339 (vec_info::set_vinfo_for_stmt, vec_info::free_stmt_vec_infos)
6340 (vec_info::free_stmt_vec_info): New private member functions.
6341 (set_stmt_vec_info_vec, free_stmt_vec_infos, vinfo_for_stmt)
6342 (set_vinfo_for_stmt, new_stmt_vec_info, free_stmt_vec_info): Delete.
6343 * tree-parloops.c (gather_scalar_reductions): Remove calls to
6344 set_stmt_vec_info_vec and free_stmt_vec_infos.
6345 * tree-vect-loop.c (_loop_vec_info): Remove call to
6346 set_stmt_vec_info_vec.
6347 * tree-vect-stmts.c (new_stmt_vec_info, set_stmt_vec_info_vec)
6348 (free_stmt_vec_infos, free_stmt_vec_info): Delete in favor of...
6349 * tree-vectorizer.c (vec_info::new_stmt_vec_info)
6350 (vec_info::set_vinfo_for_stmt, vec_info::free_stmt_vec_infos)
6351 (vec_info::free_stmt_vec_info): ...these new functions. Remove
6352 assignments in {vec_info::,}new_stmt_vec_info that are redundant
6353 with the clearing in the xcalloc.
6354 (stmt_vec_info_vec): Delete.
6355 (vec_info::vec_info): Don't call set_stmt_vec_info_vec.
6356 (vectorize_loops): Likewise.
6357 (vec_info::~vec_info): Remove argument from call to
6358 free_stmt_vec_infos.
6359 (vec_info::add_stmt): Remove vinfo argument from call to
6362 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
6364 * tree-vectorizer.h (free_stmt_vec_info): Take a stmt_vec_info
6365 rather than a gimple stmt.
6366 * tree-vect-stmts.c (free_stmt_vec_info): Likewise. Don't free
6367 information for pattern statements when passed the original
6368 statement; instead wait to be passed the pattern statement itself.
6369 Don't call set_vinfo_for_stmt here.
6370 (free_stmt_vec_infos): Update call to free_stmt_vec_info.
6371 * tree-vect-loop.c (_loop_vec_info::~loop_vec_info): Don't free
6372 stmt_vec_infos here.
6373 * tree-vect-slp.c (_bb_vec_info::~bb_vec_info): Likewise.
6374 * tree-vectorizer.c (vec_info::remove_stmt): Nullify the statement's
6375 stmt_vec_infos entry.
6377 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
6379 * tree-vectorizer.h (vec_info::replace_stmt): Declare.
6380 * tree-vectorizer.c (vec_info::replace_stmt): New function.
6381 * tree-vect-slp.c (vect_remove_slp_scalar_calls): Use it.
6382 * tree-vect-stmts.c (vectorizable_call): Likewise.
6383 (vectorizable_simd_clone_call): Likewise.
6385 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
6387 * tree-vectorizer.h (vec_info::remove_stmt): Declare.
6388 * tree-vectorizer.c (vec_info::remove_stmt): New function.
6389 * tree-vect-loop-manip.c (vect_set_loop_condition): Use it.
6390 * tree-vect-loop.c (vect_transform_loop): Likewise.
6391 * tree-vect-slp.c (vect_schedule_slp): Likewise.
6392 * tree-vect-stmts.c (vect_remove_stores): Likewise.
6394 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
6396 * tree-vectorizer.h (vec_info::lookup_dr): New member function.
6397 (vect_dr_stmt): Delete.
6398 * tree-vectorizer.c (vec_info::lookup_dr): New function.
6399 * tree-vect-loop-manip.c (vect_update_inits_of_drs): Use it instead
6401 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr)
6402 (vect_analyze_data_ref_dependence, vect_record_base_alignments)
6403 (vect_verify_datarefs_alignment, vect_peeling_supportable)
6404 (vect_analyze_data_ref_accesses, vect_prune_runtime_alias_test_list)
6405 (vect_analyze_data_refs): Likewise.
6406 (vect_slp_analyze_data_ref_dependence): Likewise. Take a vec_info
6408 (vect_find_same_alignment_drs): Likewise.
6409 (vect_slp_analyze_node_dependences): Update calls accordingly.
6410 (vect_analyze_data_refs_alignment): Likewise. Use vec_info::lookup_dr
6411 instead of DR_VECT_AUX.
6412 (vect_get_peeling_costs_all_drs): Take a loop_vec_info instead
6413 of a vector data references. Use vec_info::lookup_dr instead of
6415 (vect_peeling_hash_get_lowest_cost): Update calls accordingly.
6416 (vect_enhance_data_refs_alignment): Likewise. Use vec_info::lookup_dr
6417 instead of DR_VECT_AUX.
6419 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
6421 * tree-vectorizer.h (_loop_vec_info::unaligned_dr): Change to
6423 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Update
6425 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
6426 * tree-vect-loop-manip.c (get_misalign_in_elems): Likewise.
6427 (vect_gen_prolog_loop_niters): Likewise.
6429 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
6431 * tree-vectorizer.h (set_dr_misalignment, dr_misalignment)
6432 (DR_TARGET_ALIGNMENT, aligned_access_p, known_alignment_for_access_p)
6433 (vect_known_alignment_in_bytes, vect_dr_behavior)
6434 (vect_get_scalar_dr_size): Take references as dr_vec_infos
6435 instead of data_references. Update calls to other routines for
6436 which the same change has been made.
6437 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Take
6438 dr_vec_infos instead of stmt_vec_infos.
6439 (vect_analyze_data_ref_dependence): Update call accordingly.
6440 (vect_slp_analyze_data_ref_dependence)
6441 (vect_record_base_alignments): Use DR_VECT_AUX.
6442 (vect_calculate_target_alignment, vect_compute_data_ref_alignment)
6443 (vect_update_misalignment_for_peel, verify_data_ref_alignment)
6444 (vector_alignment_reachable_p, vect_get_data_access_cost)
6445 (vect_peeling_supportable, vect_analyze_group_access_1)
6446 (vect_analyze_group_access, vect_analyze_data_ref_access)
6447 (vect_vfa_segment_size, vect_vfa_access_size, vect_vfa_align)
6448 (vect_compile_time_alias, vect_small_gap_p)
6449 (vectorizable_with_step_bound_p, vect_duplicate_ssa_name_ptr_info):
6450 (vect_supportable_dr_alignment): Take references as dr_vec_infos
6451 instead of data_references. Update calls to other routines for
6452 which the same change has been made.
6453 (vect_verify_datarefs_alignment, vect_get_peeling_costs_all_drs)
6454 (vect_find_same_alignment_drs, vect_analyze_data_refs_alignment)
6455 (vect_slp_analyze_and_verify_node_alignment)
6456 (vect_analyze_data_ref_accesses, vect_prune_runtime_alias_test_list)
6457 (vect_create_addr_base_for_vector_ref, vect_create_data_ref_ptr)
6458 (vect_setup_realignment): Use dr_vec_infos. Update calls after
6460 (_vect_peel_info::dr): Replace with...
6461 (_vect_peel_info::dr_info): ...this new field.
6462 (vect_peeling_hash_get_most_frequent)
6463 (vect_peeling_hash_choose_best_peeling): Update accordingly.
6464 (vect_peeling_hash_get_lowest_cost):
6465 (vect_enhance_data_refs_alignment): Likewise. Update calls to other
6466 routines for which the same change has been made.
6467 (vect_peeling_hash_insert): Likewise. Take a dr_vec_info instead of a
6469 * tree-vect-loop-manip.c (get_misalign_in_elems)
6470 (vect_gen_prolog_loop_niters): Use dr_vec_infos. Update calls after
6472 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
6473 * tree-vect-stmts.c (vect_get_store_cost, vect_get_load_cost)
6474 (vect_truncate_gather_scatter_offset, compare_step_with_zero)
6475 (get_group_load_store_type, get_negative_load_store_type)
6476 (vect_get_data_ptr_increment, vectorizable_store)
6477 (vectorizable_load): Likewise.
6478 (ensure_base_align): Take a dr_vec_info instead of a data_reference.
6479 Update calls to other routines for which the same change has been made.
6481 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
6483 * tree-vectorizer.h (vec_info::move_dr): New member function.
6484 (dataref_aux): Rename to...
6485 (dr_vec_info): ...this and add "dr" and "stmt" fields.
6486 (_stmt_vec_info::dr_aux): Update accordingly.
6487 (_stmt_vec_info::data_ref_info): Delete.
6488 (STMT_VINFO_GROUPED_ACCESS, DR_GROUP_FIRST_ELEMENT)
6489 (DR_GROUP_NEXT_ELEMENT, DR_GROUP_SIZE, DR_GROUP_STORE_COUNT)
6490 (DR_GROUP_GAP, DR_GROUP_SAME_DR_STMT, REDUC_GROUP_FIRST_ELEMENT):
6491 (REDUC_GROUP_NEXT_ELEMENT, REDUC_GROUP_SIZE): Use dr_aux.dr instead
6493 (STMT_VINFO_DATA_REF): Likewise. Turn into an lvalue.
6494 (STMT_VINFO_DR_INFO): New macro.
6495 (DR_VECT_AUX): Use STMT_VINFO_DR_INKFO and vect_dr_stmt.
6496 (set_dr_misalignment): Update after rename of dataref_aux.
6497 (vect_dr_stmt): Move earlier in file. Return dr_aux.stmt.
6498 * tree-vect-stmts.c (new_stmt_vec_info): Remove redundant
6499 initialization of STMT_VINFO_DATA_REF.
6500 * tree-vectorizer.c (vec_info::move_dr): New function.
6501 * tree-vect-patterns.c (vect_recog_bool_pattern)
6502 (vect_recog_mask_conversion_pattern)
6503 (vect_recog_gather_scatter_pattern): Use it.
6504 * tree-vect-data-refs.c (vect_analyze_data_refs): Initialize
6505 the "dr" and "stmt" fields of dr_vec_info instead of
6506 STMT_VINFO_DATA_REF.
6508 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
6510 * tree-vectorizer.h (_stmt_vec_info::pattern_stmt_p): New field.
6511 (is_pattern_stmt_p): Use it.
6512 * tree-vect-patterns.c (vect_init_pattern_stmt): Set pattern_stmt_p
6513 on pattern statements.
6515 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
6517 * tree-vect-patterns.c (vect_mark_pattern_stmts): Take the
6518 original stmt as a stmt_vec_info rather than a gimple stmt.
6519 (vect_pattern_recog_1): Take the statement directly as a
6520 stmt_vec_info, rather than via a gimple_stmt_iterator.
6521 Update call to vect_mark_pattern_stmts.
6522 (vect_pattern_recog): Update calls accordingly.
6524 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
6526 * tree-vectorizer.h (vect_get_vec_defs_for_stmt_copy)
6527 (vect_get_vec_def_for_stmt_copy): Take a vec_info rather than
6528 a vect_def_type for the first argument.
6529 * tree-vect-stmts.c (vect_get_vec_defs_for_stmt_copy): Likewise.
6530 (vect_get_vec_def_for_stmt_copy): Likewise. Return the original
6531 operand if it isn't defined by a vectorized statement.
6532 (vect_build_gather_load_calls): Remove the mask_dt argument and
6533 update calls to vect_get_vec_def_for_stmt_copy.
6534 (vectorizable_bswap): Likewise the dt argument.
6535 (vectorizable_call): Update calls to vectorizable_bswap and
6536 vect_get_vec_def_for_stmt_copy.
6537 (vectorizable_simd_clone_call, vectorizable_assignment)
6538 (vectorizable_shift, vectorizable_operation, vectorizable_condition)
6539 (vectorizable_comparison): Update calls to
6540 vect_get_vec_def_for_stmt_copy.
6541 (vectorizable_store): Likewise. Remove now-unnecessary calls to
6543 (vect_get_loop_based_defs): Remove dt argument and update call
6544 to vect_get_vec_def_for_stmt_copy.
6545 (vectorizable_conversion): Update calls to vect_get_loop_based_defs
6546 and vect_get_vec_def_for_stmt_copy.
6547 (vectorizable_load): Update calls to vect_build_gather_load_calls
6548 and vect_get_vec_def_for_stmt_copy.
6549 * tree-vect-loop.c (vect_create_epilog_for_reduction)
6550 (vectorizable_reduction, vectorizable_live_operation): Update calls
6551 to vect_get_vec_def_for_stmt_copy.
6553 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
6555 * tree-vect-data-refs.c (vect_record_base_alignment): Replace vec_info
6556 and gimple stmt arguments with a stmt_vec_info.
6557 (vect_record_base_alignments): Update calls accordingly.
6558 * tree-vect-slp.c (vect_record_max_nunits): Replace vec_info
6559 and gimple stmt arguments with a stmt_vec_info.
6560 (vect_build_slp_tree_1): Remove vinfo argument and update call
6561 to vect_record_max_nunits.
6562 (vect_build_slp_tree_2): Update calls to vect_build_slp_tree_1
6563 and vect_record_max_nunits.
6565 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
6567 * tree-vectorizer.h (nested_in_vect_loop_p): Move further down
6568 file and take a stmt_vec_info instead of a gimple stmt.
6569 (supportable_widening_operation, vect_finish_replace_stmt)
6570 (vect_finish_stmt_generation, vect_get_store_rhs)
6571 (vect_get_vec_def_for_operand_1, vect_get_vec_def_for_operand)
6572 (vect_get_vec_defs, vect_init_vector, vect_transform_stmt)
6573 (vect_remove_stores, vect_analyze_stmt, vectorizable_condition)
6574 (vect_get_smallest_scalar_type, vect_check_gather_scatter)
6575 (vect_create_data_ref_ptr, bump_vector_ptr)
6576 (vect_permute_store_chain, vect_setup_realignment)
6577 (vect_transform_grouped_load, vect_record_grouped_load_vectors)
6578 (vect_create_addr_base_for_vector_ref, vectorizable_live_operation)
6579 (vectorizable_reduction, vectorizable_induction)
6580 (get_initial_def_for_reduction, is_simple_and_all_uses_invariant)
6581 (vect_get_place_in_interleaving_chain): Take stmt_vec_infos rather
6582 than gimple stmts as arguments.
6583 * tree-vect-data-refs.c (vect_get_smallest_scalar_type)
6584 (vect_preserves_scalar_order_p, vect_slp_analyze_node_dependences)
6585 (can_group_stmts_p, vect_check_gather_scatter)
6586 (vect_create_addr_base_for_vector_ref, vect_create_data_ref_ptr)
6587 (bump_vector_ptr, vect_permute_store_chain, vect_setup_realignment)
6588 (vect_permute_load_chain, vect_shift_permute_load_chain)
6589 (vect_transform_grouped_load)
6590 (vect_record_grouped_load_vectors): Likewise.
6591 * tree-vect-loop.c (vect_fixup_reduc_chain)
6592 (get_initial_def_for_reduction, vect_create_epilog_for_reduction)
6593 (vectorize_fold_left_reduction, is_nonwrapping_integer_induction)
6594 (vectorizable_reduction, vectorizable_induction)
6595 (vectorizable_live_operation, vect_loop_kill_debug_uses): Likewise.
6596 * tree-vect-patterns.c (type_conversion_p, adjust_bool_stmts)
6597 (vect_get_load_store_mask): Likewise.
6598 * tree-vect-slp.c (vect_get_place_in_interleaving_chain)
6599 (vect_analyze_slp_instance, vect_mask_constant_operand_p): Likewise.
6600 * tree-vect-stmts.c (vect_mark_relevant)
6601 (is_simple_and_all_uses_invariant)
6602 (exist_non_indexing_operands_for_use_p, process_use)
6603 (vect_init_vector_1, vect_init_vector, vect_get_vec_def_for_operand_1)
6604 (vect_get_vec_def_for_operand, vect_get_vec_defs)
6605 (vect_finish_stmt_generation_1, vect_finish_replace_stmt)
6606 (vect_finish_stmt_generation, vect_truncate_gather_scatter_offset)
6607 (compare_step_with_zero, vect_get_store_rhs, get_group_load_store_type)
6608 (get_negative_load_store_type, get_load_store_type)
6609 (vect_check_load_store_mask, vect_check_store_rhs)
6610 (vect_build_gather_load_calls, vect_get_strided_load_store_ops)
6611 (vectorizable_bswap, vectorizable_call, vectorizable_simd_clone_call)
6612 (vect_create_vectorized_demotion_stmts, vectorizable_conversion)
6613 (vectorizable_assignment, vectorizable_shift, vectorizable_operation)
6614 (get_group_alias_ptr_type, vectorizable_store, hoist_defs_of_uses)
6615 (vectorizable_load, vectorizable_condition, vectorizable_comparison)
6616 (vect_analyze_stmt, vect_transform_stmt, vect_remove_stores)
6617 (supportable_widening_operation): Likewise.
6619 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
6621 * tree-vect-data-refs.c (vect_describe_gather_scatter_call): Take
6622 a stmt_vec_info instead of a gcall.
6623 (vect_check_gather_scatter): Update call accordingly.
6624 * tree-vect-loop-manip.c (iv_phi_p): Take a stmt_vec_info instead
6626 (vect_can_advance_ivs_p, vect_update_ivs_after_vectorizer)
6627 (slpeel_update_phi_nodes_for_loops):): Update calls accordingly.
6628 * tree-vect-loop.c (vect_transform_loop_stmt): Take a stmt_vec_info
6629 instead of a gimple stmt.
6630 (vect_transform_loop): Update calls accordingly.
6631 * tree-vect-slp.c (vect_split_slp_store_group): Take and return
6632 stmt_vec_infos instead of gimple stmts.
6633 (vect_analyze_slp_instance): Update use accordingly.
6634 * tree-vect-stmts.c (read_vector_array, write_vector_array)
6635 (vect_clobber_variable, vect_stmt_relevant_p, permute_vec_elements)
6636 (vect_use_strided_gather_scatters_p, vect_build_all_ones_mask)
6637 (vect_build_zero_merge_argument, vect_get_gather_scatter_ops)
6638 (vect_gen_widened_results_half, vect_get_loop_based_defs)
6639 (vect_create_vectorized_promotion_stmts, can_vectorize_live_stmts):
6640 Take a stmt_vec_info instead of a gimple stmt and pass stmt_vec_infos
6641 down to subroutines.
6643 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
6645 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Change the type
6646 of the worklist from a vector of gimple stmts to a vector of
6648 * tree-vect-stmts.c (vect_mark_relevant, process_use)
6649 (vect_mark_stmts_to_be_vectorized): Likewise
6651 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
6653 * tree-vect-loop.c (vect_analyze_loop_operations): Look up the
6654 statement before passing it to vect_analyze_stmt.
6655 (vect_create_epilog_for_reduction): Use a stmt_vec_info to walk
6656 the chain of phi vector definitions. Track the exit phi via its
6658 (vectorizable_reduction): Set cond_stmt_vinfo directly from the
6659 STMT_VINFO_REDUC_DEF.
6660 * tree-vect-slp.c (vect_get_place_in_interleaving_chain): Use
6661 stmt_vec_infos to handle the statement chains.
6662 (vect_get_slp_defs): Record the first statement in the node
6663 using a stmt_vec_info.
6664 * tree-vect-stmts.c (vect_mark_stmts_to_be_vectorized): Look up
6665 statements here and pass their stmt_vec_info down to subroutines.
6666 (vect_init_vector_1): Hoist call to vinfo_for_stmt and pass it
6667 down to vect_finish_stmt_generation.
6668 (vect_init_vector, vect_get_vec_defs, vect_finish_replace_stmt)
6669 (vect_finish_stmt_generation): Call vinfo_for_stmt and pass
6670 stmt_vec_infos to subroutines.
6671 (vect_remove_stores): Use stmt_vec_infos to handle the statement
6674 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
6676 * tree-vect-data-refs.c (vect_slp_analyze_node_dependences):
6677 (vect_check_gather_scatter, vect_create_data_ref_ptr, bump_vector_ptr)
6678 (vect_permute_store_chain, vect_setup_realignment)
6679 (vect_permute_load_chain, vect_shift_permute_load_chain)
6680 (vect_transform_grouped_load): Use stmt_vec_info rather than gimple
6681 stmts internally, and when passing values to other vectorizer routines.
6682 * tree-vect-loop-manip.c (vect_can_advance_ivs_p): Likewise.
6683 * tree-vect-loop.c (vect_analyze_scalar_cycles_1)
6684 (vect_analyze_loop_operations, get_initial_def_for_reduction)
6685 (vect_create_epilog_for_reduction, vectorize_fold_left_reduction)
6686 (vectorizable_reduction, vectorizable_induction)
6687 (vectorizable_live_operation, vect_transform_loop_stmt)
6688 (vect_transform_loop): Likewise.
6689 * tree-vect-patterns.c (vect_reassociating_reduction_p)
6690 (vect_recog_widen_op_pattern, vect_recog_mixed_size_cond_pattern)
6691 (vect_recog_bool_pattern, vect_recog_gather_scatter_pattern): Likewise.
6692 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
6693 (vect_slp_analyze_node_operations_1): Likewise.
6694 * tree-vect-stmts.c (vect_mark_relevant, process_use)
6695 (exist_non_indexing_operands_for_use_p, vect_init_vector_1)
6696 (vect_mark_stmts_to_be_vectorized, vect_get_vec_def_for_operand)
6697 (vect_finish_stmt_generation_1, get_group_load_store_type)
6698 (get_load_store_type, vect_build_gather_load_calls)
6699 (vectorizable_bswap, vectorizable_call, vectorizable_simd_clone_call)
6700 (vect_create_vectorized_demotion_stmts, vectorizable_conversion)
6701 (vectorizable_assignment, vectorizable_shift, vectorizable_operation)
6702 (vectorizable_store, vectorizable_load, vectorizable_condition)
6703 (vectorizable_comparison, vect_analyze_stmt, vect_transform_stmt)
6704 (supportable_widening_operation): Likewise.
6705 (vect_get_vector_types_for_stmt): Likewise.
6706 * tree-vectorizer.h (vect_dr_behavior): Likewise.
6708 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
6710 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence)
6711 (vect_slp_analyze_node_dependences, vect_analyze_data_ref_accesses)
6712 (vect_permute_store_chain, vect_permute_load_chain)
6713 (vect_shift_permute_load_chain, vect_transform_grouped_load): Avoid
6714 repeated stmt_vec_info lookups.
6715 * tree-vect-loop-manip.c (vect_can_advance_ivs_p): Likewise.
6716 (vect_update_ivs_after_vectorizer): Likewise.
6717 * tree-vect-loop.c (vect_is_simple_reduction): Likewise.
6718 (vect_create_epilog_for_reduction, vectorizable_reduction): Likewise.
6719 * tree-vect-patterns.c (adjust_bool_stmts): Likewise.
6720 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
6721 (vect_bb_slp_scalar_cost): Likewise.
6722 * tree-vect-stmts.c (get_group_alias_ptr_type): Likewise.
6724 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
6726 * tree-vect-data-refs.c (vect_check_gather_scatter): Pass the
6727 gcall rather than the generic gimple stmt to gimple_call_internal_fn.
6728 (vect_get_smallest_scalar_type, can_group_stmts_p): Use dyn_cast
6729 to get gassigns and gcalls, rather than operating on generc gimple
6731 * tree-vect-stmts.c (exist_non_indexing_operands_for_use_p)
6732 (vect_mark_stmts_to_be_vectorized, vectorizable_store)
6733 (vectorizable_load, vect_analyze_stmt): Likewise.
6734 * tree-vect-loop.c (vectorizable_reduction): Likewise gphi.
6736 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
6738 * tree-vectorizer.h (get_earlier_stmt, get_later_stmt): Take and
6739 return stmt_vec_infos rather than gimple stmts. Do not accept
6741 (vect_find_last_scalar_stmt_in_slp): Return a stmt_vec_info instead
6743 * tree-vect-slp.c (vect_find_last_scalar_stmt_in_slp): Likewise.
6744 Update use of get_later_stmt.
6745 (vect_get_constant_vectors): Update call accordingly.
6746 (vect_schedule_slp_instance): Likewise
6747 * tree-vect-data-refs.c (vect_slp_analyze_node_dependences): Likewise.
6748 (vect_slp_analyze_instance_dependence): Likewise.
6749 (vect_preserves_scalar_order_p): Update use of get_earlier_stmt.
6751 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
6753 * tree-vectorizer.h (stmt_info_for_cost::stmt): Replace with...
6754 (stmt_info_for_cost::stmt_info): ...this new field.
6755 (add_stmt_costs): Update accordingly.
6756 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost)
6757 (vect_get_known_peeling_cost): Likewise.
6758 (vect_estimate_min_profitable_iters): Likewise.
6759 * tree-vect-stmts.c (record_stmt_cost): Likewise.
6761 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
6763 * tree-vectorizer.h (_loop_vec_info::may_misalign_stmts): Change
6764 from an auto_vec<gimple *> to an auto_vec<stmt_vec_info>.
6765 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Update
6767 * tree-vect-loop-manip.c (vect_create_cond_for_align_checks): Likewise.
6769 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
6771 * tree-vectorizer.h (_stmt_vec_info::same_dr_stmt): Change from
6772 a gimple stmt to a stmt_vec_info.
6773 * tree-vect-stmts.c (vectorizable_load): Update accordingly.
6775 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
6777 * tree-vectorizer.h (vec_info::grouped_stores): Change from
6778 an auto_vec<gimple *> to an auto_vec<stmt_vec_info>.
6779 (_loop_vec_info::reduction_chains): Likewise.
6780 * tree-vect-loop.c (vect_fixup_scalar_cycles_with_patterns): Update
6782 * tree-vect-slp.c (vect_analyze_slp): Likewise.
6784 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
6786 * tree-vectorizer.h (_stmt_vec_info::first_element): Change from
6787 a gimple stmt to a stmt_vec_info.
6788 (_stmt_vec_info::next_element): Likewise.
6789 * tree-vect-data-refs.c (vect_update_misalignment_for_peel)
6790 (vect_slp_analyze_and_verify_node_alignment)
6791 (vect_analyze_group_access_1, vect_analyze_group_access)
6792 (vect_small_gap_p, vect_prune_runtime_alias_test_list)
6793 (vect_create_data_ref_ptr, vect_record_grouped_load_vectors)
6794 (vect_supportable_dr_alignment): Update accordingly.
6795 * tree-vect-loop.c (vect_fixup_reduc_chain): Likewise.
6796 (vect_fixup_scalar_cycles_with_patterns, vect_is_slp_reduction)
6797 (vect_is_simple_reduction, vectorizable_reduction): Likewise.
6798 * tree-vect-patterns.c (vect_reassociating_reduction_p): Likewise.
6799 * tree-vect-slp.c (vect_build_slp_tree_1)
6800 (vect_attempt_slp_rearrange_stmts, vect_supported_load_permutation_p)
6801 (vect_split_slp_store_group, vect_analyze_slp_instance)
6802 (vect_analyze_slp, vect_transform_slp_perm_load): Likewise.
6803 * tree-vect-stmts.c (vect_model_store_cost, vect_model_load_cost)
6804 (get_group_load_store_type, get_load_store_type)
6805 (get_group_alias_ptr_type, vectorizable_store, vectorizable_load)
6806 (vect_transform_stmt, vect_remove_stores): Likewise.
6808 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
6810 * tree-vectorizer.h (vect_dr_stmt): Return a stmt_vec_info rather
6812 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence)
6813 (vect_slp_analyze_data_ref_dependence, vect_record_base_alignments)
6814 (vect_calculate_target_alignmentm, vect_compute_data_ref_alignment)
6815 (vect_update_misalignment_for_peel, vect_verify_datarefs_alignment)
6816 (vector_alignment_reachable_p, vect_get_data_access_cost)
6817 (vect_get_peeling_costs_all_drs, vect_peeling_hash_get_lowest_cost)
6818 (vect_peeling_supportable, vect_enhance_data_refs_alignment)
6819 (vect_find_same_alignment_drs, vect_analyze_data_refs_alignment)
6820 (vect_analyze_group_access_1, vect_analyze_group_access)
6821 (vect_analyze_data_ref_access, vect_analyze_data_ref_accesses)
6822 (vect_vfa_access_size, vect_small_gap_p, vect_analyze_data_refs)
6823 (vect_supportable_dr_alignment): Remove vinfo_for_stmt from the
6824 result of vect_dr_stmt and use the stmt_vec_info instead of
6825 the associated gimple stmt.
6826 * tree-vect-loop-manip.c (get_misalign_in_elems): Likewise.
6827 (vect_gen_prolog_loop_niters): Likewise.
6828 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
6830 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
6832 * tree-vectorizer.h (_slp_tree::stmts): Change from a vec<gimple *>
6833 to a vec<stmt_vec_info>.
6834 * tree-vect-slp.c (vect_free_slp_tree): Update accordingly.
6835 (vect_create_new_slp_node): Take a vec<gimple *> instead of a
6837 (_slp_oprnd_info::def_stmts): Change from a vec<gimple *>
6838 to a vec<stmt_vec_info>.
6839 (bst_traits::value_type, bst_traits::value_type): Likewise.
6840 (bst_traits::hash): Update accordingly.
6841 (vect_get_and_check_slp_defs): Change the stmts parameter from
6842 a vec<gimple *> to a vec<stmt_vec_info>.
6843 (vect_two_operations_perm_ok_p, vect_build_slp_tree_1): Likewise.
6844 (vect_build_slp_tree): Likewise.
6845 (vect_build_slp_tree_2): Likewise. Update uses of
6846 SLP_TREE_SCALAR_STMTS.
6847 (vect_print_slp_tree): Update uses of SLP_TREE_SCALAR_STMTS.
6848 (vect_mark_slp_stmts, vect_mark_slp_stmts_relevant)
6849 (vect_slp_rearrange_stmts, vect_attempt_slp_rearrange_stmts)
6850 (vect_supported_load_permutation_p, vect_find_last_scalar_stmt_in_slp)
6851 (vect_detect_hybrid_slp_stmts, vect_slp_analyze_node_operations_1)
6852 (vect_slp_analyze_node_operations, vect_slp_analyze_operations)
6853 (vect_bb_slp_scalar_cost, vect_slp_analyze_bb_1)
6854 (vect_get_constant_vectors, vect_get_slp_defs)
6855 (vect_transform_slp_perm_load, vect_schedule_slp_instance)
6856 (vect_remove_slp_scalar_calls, vect_schedule_slp): Likewise.
6857 (vect_analyze_slp_instance): Build up a vec of stmt_vec_infos
6858 instead of gimple stmts.
6859 * tree-vect-data-refs.c (vect_slp_analyze_node_dependences): Change
6860 the stores parameter for a vec<gimple *> to a vec<stmt_vec_info>.
6861 (vect_slp_analyze_instance_dependence): Update uses of
6862 SLP_TREE_SCALAR_STMTS.
6863 (vect_slp_analyze_and_verify_node_alignment): Likewise.
6864 (vect_slp_analyze_and_verify_instance_alignment): Likewise.
6865 * tree-vect-loop.c (neutral_op_for_slp_reduction): Likewise.
6866 (get_initial_defs_for_reduction): Likewise.
6867 (vect_create_epilog_for_reduction): Likewise.
6868 (vectorize_fold_left_reduction): Likewise.
6869 * tree-vect-stmts.c (vect_prologue_cost_for_slp_op): Likewise.
6870 (vect_model_simple_cost, vectorizable_shift, vectorizable_load)
6871 (can_vectorize_live_stmts): Likewise.
6873 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
6875 * tree-vectorizer.h (_loop_vec_info::reductions): Change from an
6876 auto_vec<gimple *> to an auto_vec<stmt_vec_info>.
6877 (vect_force_simple_reduction): Take and return stmt_vec_infos rather
6879 * tree-parloops.c (valid_reduction_p): Take a stmt_vec_info instead
6881 (gather_scalar_reductions): Update after above interface changes.
6882 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Likewise.
6883 (vect_is_simple_reduction): Take and return stmt_vec_infos rather
6885 (vect_force_simple_reduction): Likewise.
6886 * tree-vect-patterns.c (vect_pattern_recog_1): Update use of
6887 LOOP_VINFO_REDUCTIONS.
6888 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
6890 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
6892 * tree-vectorizer.h (_stmt_vec_info::reduc_def): Change from
6893 a gimple stmt to a stmt_vec_info.
6894 * tree-vect-loop.c (vect_active_double_reduction_p)
6895 (vect_force_simple_reduction, vectorizable_reduction): Update
6898 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
6900 * tree-vectorizer.h (_slp_tree::vec_stmts): Change from a
6901 vec<gimple *> to a vec<stmt_vec_info>.
6902 * tree-vect-loop.c (vect_create_epilog_for_reduction): Change
6903 the reduction_phis argument from a vec<gimple *> to a
6905 (vectorizable_reduction): Likewise the phis local variable that
6906 is passed to vect_create_epilog_for_reduction. Update for new type
6907 of SLP_TREE_VEC_STMTS.
6908 (vectorizable_induction): Update for new type of SLP_TREE_VEC_STMTS.
6909 (vectorizable_live_operation): Likewise.
6910 * tree-vect-slp.c (vect_get_slp_vect_defs): Likewise.
6911 (vect_transform_slp_perm_load, vect_schedule_slp_instance): Likewise.
6913 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
6915 * tree-vectorizer.h (_stmt_vec_info::vectorized_stmt): Change from
6916 a gimple stmt to a stmt_vec_info.
6917 (vectorizable_condition, vectorizable_live_operation)
6918 (vectorizable_reduction, vectorizable_induction): Pass back the
6919 vectorized statement as a stmt_vec_info.
6920 * tree-vect-data-refs.c (vect_record_grouped_load_vectors): Update
6921 use of STMT_VINFO_VEC_STMT.
6922 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise,
6923 accumulating the inner phis that feed the STMT_VINFO_VEC_STMT
6924 as stmt_vec_infos rather than gimple stmts.
6925 (vectorize_fold_left_reduction): Change vec_stmt from a gimple stmt
6927 (vectorizable_live_operation): Likewise.
6928 (vectorizable_reduction, vectorizable_induction): Likewise,
6929 updating use of STMT_VINFO_VEC_STMT.
6930 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Update use
6931 of STMT_VINFO_VEC_STMT.
6932 (vect_build_gather_load_calls, vectorizable_bswap, vectorizable_call)
6933 (vectorizable_simd_clone_call, vectorizable_conversion)
6934 (vectorizable_assignment, vectorizable_shift, vectorizable_operation)
6935 (vectorizable_store, vectorizable_load, vectorizable_condition)
6936 (vectorizable_comparison, can_vectorize_live_stmts): Change vec_stmt
6937 from a gimple stmt to a stmt_vec_info.
6938 (vect_transform_stmt): Update use of STMT_VINFO_VEC_STMT. Pass a
6939 pointer to a stmt_vec_info to the vectorizable_* routines.
6941 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
6943 * tree-vectorizer.h (_stmt_vec_info::related_stmt): Change from
6944 a gimple stmt to a stmt_vec_info.
6945 (is_pattern_stmt_p): Update accordingly.
6946 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Likewise.
6947 (vect_record_grouped_load_vectors): Likewise.
6948 * tree-vect-loop.c (vect_determine_vf_for_stmt): Likewise.
6949 (vect_fixup_reduc_chain, vect_update_vf_for_slp): Likewise.
6950 (vect_model_reduction_cost): Likewise.
6951 (vect_create_epilog_for_reduction): Likewise.
6952 (vectorizable_reduction, vectorizable_induction): Likewise.
6953 * tree-vect-patterns.c (vect_init_pattern_stmt): Likewise.
6954 Return the stmt_vec_info for the pattern statement.
6955 (vect_set_pattern_stmt): Update use of STMT_VINFO_RELATED_STMT.
6956 (vect_split_statement, vect_mark_pattern_stmts): Likewise.
6957 * tree-vect-slp.c (vect_detect_hybrid_slp_stmts): Likewise.
6958 (vect_detect_hybrid_slp, vect_get_slp_defs): Likewise.
6959 * tree-vect-stmts.c (vect_mark_relevant): Likewise.
6960 (vect_get_vec_def_for_operand_1, vectorizable_call): Likewise.
6961 (vectorizable_simd_clone_call, vect_analyze_stmt, new_stmt_vec_info)
6962 (free_stmt_vec_info, vect_is_simple_use): Likewise.
6964 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
6966 * tree-vectorizer.h (vect_finish_replace_stmt): Return a stmt_vec_info
6967 (vect_finish_stmt_generation): Likewise.
6968 * tree-vect-stmts.c (vect_finish_stmt_generation_1): Likewise.
6969 (vect_finish_replace_stmt, vect_finish_stmt_generation): Likewise.
6970 (vect_build_gather_load_calls): Use the return value of the above
6971 functions instead of a separate call to vinfo_for_stmt. Use narrow
6972 scopes for the input gimple stmt and wider scopes for the associated
6973 stmt_vec_info. Use vec_info::lookup_def when setting these
6974 stmt_vec_infos from an SSA_NAME definition.
6975 (vectorizable_bswap, vectorizable_call, vectorizable_simd_clone_call)
6976 (vect_create_vectorized_demotion_stmts, vectorizable_conversion)
6977 (vectorizable_assignment, vectorizable_shift, vectorizable_operation)
6978 (vectorizable_store, vectorizable_load, vectorizable_condition)
6979 (vectorizable_comparison): Likewise.
6980 * tree-vect-loop.c (vectorize_fold_left_reduction): Likewise.
6981 (vectorizable_reduction): Likewise.
6983 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
6985 * tree-vectorizer.h (vect_is_simple_use): Add an optional
6986 stmt_vec_info * parameter before the optional gimple **.
6987 * tree-vect-stmts.c (vect_is_simple_use): Likewise.
6988 (process_use, vect_get_vec_def_for_operand_1): Update callers.
6989 (vect_get_vec_def_for_operand, vectorizable_shift): Likewise.
6990 * tree-vect-loop.c (vectorizable_reduction): Likewise.
6991 (vectorizable_live_operation): Likewise.
6992 * tree-vect-patterns.c (type_conversion_p): Likewise.
6993 (vect_look_through_possible_promotion): Likewise.
6994 (vect_recog_rotate_pattern): Likewise.
6995 * tree-vect-slp.c (vect_get_and_check_slp_defs): Likewise.
6997 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
6999 * tree-vectorizer.h (stmt_vec_info): Temporarily change from
7000 a typedef to a wrapper class.
7001 (NULL_STMT_VEC_INFO): New macro.
7002 (vec_info::stmt_infos): Change to vec<stmt_vec_info>.
7003 (stmt_vec_info::operator*): New function.
7004 (stmt_vec_info::operator gimple *): Likewise.
7005 (set_vinfo_for_stmt): Use NULL_STMT_VEC_INFO.
7006 (add_stmt_costs): Likewise.
7007 * tree-vect-loop-manip.c (iv_phi_p): Likewise.
7008 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost)
7009 (vect_get_known_peeling_cost): Likewise.
7010 (vect_estimate_min_profitable_iters): Likewise.
7011 * tree-vect-patterns.c (vect_init_pattern_stmt): Likewise.
7012 * tree-vect-slp.c (vect_remove_slp_scalar_calls): Likewise.
7013 * tree-vect-stmts.c (vect_build_gather_load_calls): Likewise.
7014 (vectorizable_store, free_stmt_vec_infos): Likewise.
7015 (new_stmt_vec_info): Change return type of xcalloc to
7018 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
7020 * tree-vectorizer.h (vec_info::lookup_single_use): Declare.
7021 * tree-vectorizer.c (vec_info::lookup_single_use): New function.
7022 * tree-vect-loop.c (vectorizable_reduction): Use it instead of
7023 a single_imm_use-based sequence.
7024 * tree-vect-stmts.c (supportable_widening_operation): Likewise.
7026 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
7028 * tree-vectorizer.h (vec_info::lookup_def): Declare.
7029 * tree-vectorizer.c (vec_info::lookup_def): New function.
7030 * tree-vect-patterns.c (vect_get_internal_def): Use it.
7031 (vect_widened_op_tree): Likewise.
7032 * tree-vect-stmts.c (vect_is_simple_use): Likewise.
7033 * tree-vect-loop.c (vect_analyze_loop_operations): Likewise.
7034 (vectorizable_reduction): Likewise.
7035 (vect_valid_reduction_input_p): Take a stmt_vec_info instead
7037 (vect_is_slp_reduction): Update calls accordingly. Use
7038 vec_info::lookup_def.
7039 (vect_is_simple_reduction): Likewise
7040 * tree-vect-slp.c (vect_detect_hybrid_slp_1): Use vec_info::lookup_def.
7042 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
7044 * tree-vectorizer.h (vec_info::lookup_stmt): Declare.
7045 * tree-vectorizer.c (vec_info::lookup_stmt): New function.
7046 * tree-vect-loop.c (vect_determine_vf_for_stmt): Use it instead
7048 (vect_determine_vectorization_factor, vect_analyze_scalar_cycles_1)
7049 (vect_compute_single_scalar_iteration_cost, vect_analyze_loop_form)
7050 (vect_update_vf_for_slp, vect_analyze_loop_operations)
7051 (vect_is_slp_reduction, vectorizable_induction)
7052 (vect_transform_loop_stmt, vect_transform_loop): Likewise.
7053 * tree-vect-patterns.c (vect_init_pattern_stmt):
7054 (vect_determine_min_output_precision_1, vect_determine_precisions)
7055 (vect_pattern_recog): Likewise.
7056 * tree-vect-stmts.c (vect_analyze_stmt, vect_transform_stmt): Likewise.
7057 * config/powerpcspe/powerpcspe.c (rs6000_density_test): Likewise.
7058 * config/rs6000/rs6000.c (rs6000_density_test): Likewise.
7059 * tree-vect-slp.c (vect_detect_hybrid_slp_stmts): Likewise.
7060 (vect_detect_hybrid_slp_1, vect_detect_hybrid_slp_2)
7061 (vect_detect_hybrid_slp): Likewise. Change the walk_stmt_info
7062 info field from a loop to a loop_vec_info.
7064 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
7066 * tree-vectorizer.h (stmt_vec_info): Move typedef earlier in file.
7067 (vec_info::add_stmt): Declare.
7068 * tree-vectorizer.c (vec_info::add_stmt): New function.
7069 * tree-vect-data-refs.c (vect_create_data_ref_ptr): Use it.
7070 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Likewise.
7071 (vect_create_epilog_for_reduction, vectorizable_reduction): Likewise.
7072 (vectorizable_induction): Likewise.
7073 * tree-vect-slp.c (_bb_vec_info::_bb_vec_info): Likewise.
7074 * tree-vect-stmts.c (vect_finish_stmt_generation_1): Likewise.
7075 (vectorizable_simd_clone_call, vectorizable_store): Likewise.
7076 (vectorizable_load): Likewise.
7077 * tree-vect-patterns.c (vect_init_pattern_stmt): Likewise.
7078 (vect_recog_bool_pattern, vect_recog_mask_conversion_pattern)
7079 (vect_recog_gather_scatter_pattern): Likewise.
7080 (append_pattern_def_seq): Likewise. Remove a check that is
7081 performed by add_stmt itself.
7083 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
7085 * tree-vect-loop.c (vectorizable_reduction): Fix an instance in
7086 which make_ssa_name was called with new_stmt before new_stmt
7089 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
7091 * tree-vect-loop.c (vect_valid_reduction_input_p): New function,
7093 (vect_is_slp_reduction): ...here...
7094 (vect_is_simple_reduction): ...and here. Remove repetition of tests
7095 that are already known to be false.
7097 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
7099 * tree-vectorizer.h (vect_free_slp_instance): Add a final_p parameter.
7100 * tree-vect-slp.c (vect_free_slp_tree): Likewise. Don't update
7101 STMT_VINFO_NUM_SLP_USES when it's true.
7102 (vect_free_slp_instance): Add a final_p parameter and pass it to
7104 (vect_build_slp_tree_2): Update call to vect_free_slp_instance.
7105 (vect_analyze_slp_instance): Likewise.
7106 (vect_slp_analyze_operations): Likewise.
7107 (vect_slp_analyze_bb_1): Likewise.
7108 * tree-vectorizer.c (vec_info): Likewise.
7109 * tree-vect-loop.c (vect_transform_loop): Likewise.
7111 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
7113 * tree-vect-loop.c (vectorizable_reduction): Assert that the
7114 function is not called for second and subsequent members of
7117 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
7119 * tree-vect-loop.c (get_initial_def_for_reduction): Move special
7120 cases for nested loops from here to ...
7121 (vect_create_epilog_for_reduction): ...here. Only call
7122 vect_is_simple_use for inner-loop reductions.
7124 2018-07-31 Martin Liska <mliska@suse.cz>
7126 PR gcov-profile/85338
7127 PR gcov-profile/85350
7128 PR gcov-profile/85372
7129 * profile.c (struct location_triplet): New.
7130 (struct location_triplet_hash): Likewise.
7131 (output_location): Do not output a BB that
7132 is already recorded for a line.
7133 (branch_prob): Use streamed_locations.
7135 2018-07-31 Martin Liska <mliska@suse.cz>
7137 PR gcov-profile/85370
7138 * coverage.c (coverage_begin_function): Do not mark target
7139 clones as artificial functions.
7141 2018-07-31 Martin Liska <mliska@suse.cz>
7143 PR gcov-profile/83813
7144 PR gcov-profile/84758
7145 PR gcov-profile/85217
7146 PR gcov-profile/85332
7147 * profile.c (branch_prob): Do not record GOTO expressions
7148 for GIMPLE statements which locations are already streamed.
7150 2018-07-31 Olivier Hainque <hainque@adacore.com>
7152 * gcc.c (handle_spec_function): Accept a soft_matched_part
7153 argument, as do_spec_1. Pass it down to ...
7154 (eval_spec_function): Accept a soft_matched_part argument,
7155 and pass it down to ...
7156 (do_spec_2): Accept a soft_matched_part argument, and pass
7157 it down to do_spec_1.
7158 (do_spec_1): Pass soft_matched_part to handle_spec_function.
7159 (handle_braces): Update call to handle_spec_function.
7160 (driver::set_up_specs): Update calls to do_spec_2.
7161 (compare_debug_dump_opt_spec_function): Likewise.
7162 (compare_debug_self_opt_spec_function): Likewise.
7164 2018-07-31 Olivier Hainque <hainque@adacore.com>
7166 * common.opt (nolibc): New option.
7167 * doc/invoke.texi (Link Options): Document it.
7168 * gcc.c (LINK_GCC_C_SEQUENCE_SPEC): Honor nolibc.
7169 * config/alpha/linux.h: Likewise.
7170 * config/arc/elf.h: Likewise.
7171 * config/arm/uclinux-elf.h: Likewise.
7172 * config/arm/unknown-elf.h: Likewise.
7173 * config/avr/avrlibc.h: Likewise.
7174 * config/bfin/bfin.h: Likewise.
7175 * config/bfin/linux.h: Likewise.
7176 * config/bfin/uclinux.h: Likewise.
7177 * config/darwin.h: Likewise.
7178 * config/darwin10.h: Likewise.
7179 * config/darwin12.h: Likewise.
7180 * config/gnu-user.h: Likewise.
7181 * config/lm32/uclinux-elf.h: Likewise.
7182 * config/pa/pa-hpux11.h: Likewise.
7183 * config/pa/pa64-hpux.h: Likewise.
7184 * config/sparc/sparc.h: Likewise.
7186 2018-07-31 Olivier Hainque <hainque@adacore.com>
7188 * gcc.c (getenv_spec_function): Prepend '/' to value for allowed
7189 undefined variables.
7191 2018-07-30 Segher Boessenkool <segher@kernel.crashing.org>
7194 * config/arm/arm.c (arm_block_set_aligned_vect): Use gen_int_mode
7197 2018-07-30 Bernd Edlinger <bernd.edlinger@hotmail.de>
7199 * tree-ssa-forwprop.c (simplify_builtin_call): Don't create a not NUL
7200 terminated string literal.
7202 2018-07-30 Segher Boessenkool <segher@kernel.crashing.org>
7204 PR rtl-optimization/85160
7205 * combine.c (is_just_move): New function.
7206 (try_combine): Allow combining two instructions into two if neither of
7207 the original instructions was a move.
7209 2018-07-30 Alexander Monakov <amonakov@ispras.ru>
7212 * doc/extend.texi (Global Register Variables): Discourage use of type
7214 (Local Register Variables): Likewise.
7216 2018-07-30 Richard Sandiford <richard.sandiford@arm.com>
7218 PR tree-optimization/86506
7219 * hwint.h (ceil_log2): Resync with hwint.c implementation.
7221 2018-07-30 Ilya Leoshkevich <iii@linux.ibm.com>
7224 * lra-constraints.c (spill_hard_reg_in_range): When selecting the
7225 hard_regno, make sure no insn between `from` and `to` clobbers it.
7227 2018-07-30 Cesar Philippidis <cesar@codesourcery.com>
7228 Tom de Vries <tdevries@suse.de>
7230 * config/nvptx/nvptx.c (PTX_GANG_DEFAULT): Rename to ...
7231 (PTX_DEFAULT_RUNTIME_DIM): ... this.
7232 (nvptx_goacc_validate_dims): Set default worker and gang dims to
7233 PTX_DEFAULT_RUNTIME_DIM.
7234 (nvptx_dim_limit): Ignore GOMP_DIM_WORKER.
7236 2018-07-29 John David Anglin <danglin@gcc.gnu.org>
7238 * config/pa/pa.c (pa_output_addr_vec): Align address table.
7239 * config/pa/pa.h (JUMP_TABLES_IN_TEXT_SECTION): Revise comment.
7240 * config/pa/pa32-linux.h (JUMP_TABLES_IN_TEXT_SECTION): Define.
7242 2018-07-27 Michael Meissner <meissner@linux.ibm.com>
7244 * config/rs6000/constraints.md (wG constraint): Delete, no longer
7246 * config/rs6000/predicates.md (p9_fusion_reg_operand): Rename
7247 predicate to reflect toc fusion has been deleted.
7248 (toc_fusion_mem_raw): Delete, no longer used.
7249 (toc_fusion_mem_wrapped): Likewise.
7250 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete toc
7252 * config/rs6000/rs6000-protos.h (fusion_wrap_memory_address):
7253 Delete, no longer used.
7254 * config/rs6000/rs6000.c (struct rs6000_reg_addr): Delete fields
7255 meant to be used for toc fusion.
7256 (rs6000_debug_print_mode): Delete toc fusion debugging.
7257 (rs6000_debug_reg_global): Likewise.
7258 (rs6000_init_hard_regno_mode_ok): Delete setting up fields for toc
7259 fusion and secondary reload support that were never used.
7260 (rs6000_option_override_internal): Delete TOC fusion, that was only
7261 partially defined, and it did not work unless you also used the
7263 (rs6000_legitimate_address_p): Delete TOC fusion support.
7264 (rs6000_opt_masks): Likewise.
7265 (fusion_wrap_memory_address): Delete function, no longer used.
7266 (fusion_split_address); Delete TOC fusion support.
7267 * config/rs6000/rs6000.h (TARGET_TOC_FUSION_INT): Delete, no
7268 longer used with toc fusion being deleted.
7269 (TARGET_TOC_FUSION_FP): Likewise.
7270 * config/rs6000/rs6000.md (UNSPEC_FUSION_ADDIS): Delete TOC fusion
7272 (toc fusion spliter): Delete TOC fusion support.
7273 (toc_fusionload_<mode>): Likewise.
7274 (toc_fusionload_di): Likewise.
7275 (fusion_gpr_load_<mode>): Delete generator function, this insn no
7276 longer needs to be named. Rename predicate to delete TOC fusion.
7277 (fusion_gpr_<P:mode>_<GPR_FUSION:mode>_load): Likewise.
7278 (fusion_gpr_<P:mode>_<GPR_FUSION:mode>_store): Likewise.
7279 (fusion_vsx_<P:mode>_<GPR_FUSION:mode>_load): Likewise.
7280 (fusion_vsx_<P:mode>_<GPR_FUSION:mode>_store): Likewise.
7281 (p9 fusion peephole2s): Rename predicate to delete TOC fusion.
7283 2018-07-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
7285 * doc/extend.texi (Basic PowerPC Built-in Functions Available on
7286 ISA 2.05): Replace __uint128_t with __uint128 and __int128_t with
7287 __int128 in built-in function prototypes.
7288 (PowerPC AltiVec Built-in Functions on ISA 2.07): Likewise.
7289 (PowerPC AltiVec Built-in Functions on ISA 3.0): Likewise.
7291 2018-07-27 Martin Sebor <msebor@redhat.com>
7293 PR tree-optimization/86696
7294 * tree-ssa-strlen.c (get_min_string_length): Handle all integer
7295 types, including enums.
7296 (handle_char_store): Be prepared for the above function to fail.
7298 2018-07-26 Qing Zhao <qing.zhao@oracle.com>
7300 * builtins.c (inline_expand_builtin_string_cmp): Disable inlining
7301 when optimization level is lower than 2 or optimize for size.
7303 2018-07-26 Martin Sebor <msebor@redhat.com>
7305 PR tree-optimization/86043
7306 PR tree-optimization/86042
7307 * tree-ssa-strlen.c (handle_builtin_memcpy): Handle strict overlaps.
7308 (get_string_cst_length): Rename...
7309 (get_min_string_length): ...to this. Add argument.
7310 (handle_char_store): Extend to handle multi-character stores by
7312 * tree.c (initializer_zerop): Use new argument. Handle MEM_REF.
7313 * tree.h (initializer_zerop): Add argument.
7315 2018-07-26 Jakub Jelinek <jakub@redhat.com>
7318 * omp-low.c (scan_sharing_clauses): Don't ignore map clauses for
7319 declare target to variables if they have always,{to,from,tofrom} map
7322 2018-07-26 Martin Liska <mliska@suse.cz>
7325 * lto-wrapper.c: Add linker_output as prefix
7326 for ltrans_output_file.
7328 2018-07-26 Segher Boessenkool <segher@kernel.crashing.org>
7330 PR rtl-optimization/85805
7331 * combine.c (reg_nonzero_bits_for_combine): Only use the last set
7332 value for hard registers if that was written in the same mode.
7334 2018-07-26 Martin Liska <mliska@suse.cz>
7336 PR gcov-profile/86536
7337 * gcov.c (format_gcov): Use printf format %.*f directly
7338 and do not handle special values.
7340 2018-07-25 Claudiu Zissulescu <claziss@synopsys.com>
7342 * common/config/arc/arc-common.c (arc_option_optimization_table):
7343 Update default optimizations for size.
7345 2018-07-25 Claudiu Zissulescu <claziss@synopsys.com>
7347 * config/arc/arc.md (movsf_insn): Add short instruction selection.
7348 * config/arc/constraints.md (CfZ): New constraint.
7349 * config/arc/fpu.md (addssf3_fpu): Use CfZ constraint.
7350 (subsf3_fpu): Likewise.
7351 (cmpsf_fpu): Likewise.
7352 (cmpsf_fpu_uneq): Likewise.
7354 2018-07-25 Claudiu Zissulescu <claziss@synopsys.com>
7356 * config/arc/arc.c (compact_memory_operand_p): Check for uncached
7358 (arc_is_uncached_mem_p): uncached applies to both the variable and
7361 2018-07-25 Claudiu Zissulescu <claziss@synopsys.com>
7363 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Add additional
7366 2018-07-25 David Malcolm <dmalcolm@redhat.com>
7368 * optinfo-emit-json.cc (class optrecord_json_writer): Convert
7369 field "m_scopes" from vec to auto_vec.
7371 2018-07-25 Martin Liska <mliska@suse.cz>
7373 * config/powerpcspe/powerpcspe-protos.h (rs6000_loop_align): Fix
7376 2018-07-25 Richard Biener <rguenther@suse.de>
7379 * dwarf2out.c (dwarf2out_decl): Do not handle nested functions
7380 special wrt context_die late.
7381 (gen_subprogram_die): Re-use DIEs in local scope.
7383 2018-07-25 Richard Sandiford <richard.sandiford@arm.com>
7385 PR tree-optimization/86644
7386 * hwint.c (ceil_log2): Fix comment. Return 0 for 0.
7388 2018-07-25 Martin Liska <mliska@suse.cz>
7391 * dumpfile.c: And excluded values with TDF_ALL_VALUES.
7392 * dumpfile.h (enum dump_flag): Defince TDF_ALL_VALUES.
7394 2018-07-25 Martin Liska <mliska@suse.cz>
7397 * params.def: Explain ASan abbreviation and provide
7398 a documentation link.
7400 2018-07-24 Martin Sebor <msebor@redhat.com>
7402 PR tree-optimization/86622
7403 PR tree-optimization/86532
7404 * builtins.h (string_length): Declare.
7405 * builtins.c (c_strlen): Correct handling of non-constant offsets.
7406 (check_access): Be prepared for non-constant length ranges.
7407 (string_length): Make extern.
7408 * expr.c (string_constant): Only handle the minor non-constant
7409 array index. Use string_constant to compute the length of
7410 a generic string constant.
7412 2018-07-24 Richard Sandiford <richard.sandiford@arm.com>
7414 PR tree-optimization/86618
7415 * tree-vect-stmts.c (vectorizable_call): Don't take the address
7416 of LOOP_VINFO_MASKS (loop_vinfo) when loop_vinfo is null.
7418 2018-07-24 David Malcolm <dmalcolm@redhat.com>
7420 PR tree-optimization/86636
7421 * json.cc (json::object::set): Fix comment. Add assertions.
7422 (json::array::append): Move here from json.h. Add comment and an
7424 (json::string::string): Likewise.
7425 * json.h (json::array::append): Move to json.cc.
7426 (json::string::string): Likewise.
7427 * optinfo-emit-json.cc
7428 (optrecord_json_writer::impl_location_to_json): Assert that we
7429 aren't attempting to write out UNKNOWN_LOCATION, or an ad-hoc
7430 wrapper around it. Expand the location once, rather than three
7432 (optrecord_json_writer::inlining_chain_to_json): Fix the check for
7433 UNKNOWN_LOCATION, to use LOCATION_LOCUS to look through ad-hoc
7435 (optrecord_json_writer::optinfo_to_json): Likewise, in four
7436 places. Fix some overlong lines.
7438 2018-07-24 Matthew Malcomson <matthew.malcomson@arm.com>
7440 * config/aarch64/aarch64-simd.md
7441 (aarch64_<ANY_EXTEND:su><ADDSUB:optab>w<mode>): Split into...
7442 (aarch64_<ANY_EXTEND:su>subw<mode>): ... This...
7443 (aarch64_<ANY_EXTEND:su>addw<mode>): ... And this.
7444 (aarch64_<ANY_EXTEND:su><ADDSUB:optab>w<mode>_internal): Split into...
7445 (aarch64_<ANY_EXTEND:su>subw<mode>_internal): ... This...
7446 (aarch64_<ANY_EXTEND:su>addw<mode>_internal): ... And this.
7447 (aarch64_<ANY_EXTEND:su><ADDSUB:optab>w2<mode>_internal): Split into...
7448 (aarch64_<ANY_EXTEND:su>subw2<mode>_internal): ... This...
7449 (aarch64_<ANY_EXTEND:su>addw2<mode>_internal): ... And this.
7451 2018-07-24 Jakub Jelinek <jakub@redhat.com>
7454 * expmed.c (expand_divmod): Punt if d == HOST_WIDE_INT_MIN
7455 and size > HOST_BITS_PER_WIDE_INT. For size > HOST_BITS_PER_WIDE_INT
7456 and abs_d == d, do the power of two handling if profitable.
7458 2018-07-24 Richard Biener <rguenther@suse.de>
7460 * match.pd: Add BIT_FIELD_REF canonicalizations.
7462 2018-07-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
7465 * genmatch.c (dt_operand::gen_match_op): Avoid folding volatile values.
7467 2018-07-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
7469 * gimple-fold.c (gimple_fold_builtin_printf): Don't create a not NUL
7470 terminated STRING_CST object.
7472 2018-07-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
7474 hsa-dump.c (dump_hsa_symbol): Avoid out of scope access to buf.
7476 2018-07-23 Segher Boessenkool <segher@kernel.crashing.org>
7478 * config/rs6000/rs6000-p8swap.c (rtx_is_swappable_p): Adjust.
7479 * config/rs6000/rs6000-protos.h (rs6000_split_v4si_init): Delete.
7480 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Always force
7481 the elements into a register.
7482 (rs6000_split_v4si_init_di_reg): Delete.
7483 (rs6000_split_v4si_init): Delete.
7484 * config/rs6000/vsx.md (unspec): Delete UNSPEC_VSX_VEC_INIT.
7485 (vsx_init_v4si): Rewrite as a define_expand.
7487 2018-07-23 Segher Boessenkool <segher@kernel.crashing.org>
7489 * config/rs6000/rs6000.md (splitters for rldimi and rlwimi with the
7490 zero_extend argument from memory): New.
7492 2018-07-22 Martin Sebor <msebor@redhat.com>
7495 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Avoid
7496 diagnosing calls with unknown arguments unless -Walloca-larger-than
7497 is restricted to less than PTRDIFF_MAX bytes.
7499 2018-07-22 Gerald Pfeifer <gerald@pfeifer.com>
7501 * doc/gcov.texi (Invoking Gcov): Editorial changes.
7503 2018-07-20 David Malcolm <dmalcolm@redhat.com>
7505 * pretty-print.c (text_info::set_location): Remove redundant
7506 "line_table" parameter from call to rich_location::set_range.
7508 2018-07-20 Martin Sebor <msebor@redhat.com>
7511 * builtins.c (expand_builtin_alloca): Adjust.
7512 * calls.c (alloc_max_size): Simplify.
7513 * cgraphunit.c (cgraph_node::expand): Adjust.
7514 * common.opt (larger_than_size, warn_frame_larger_than): Remove
7516 (frame_larger_than_size): Same.
7517 (-Wframe-larger-than, -Wlarger-than, -Wstack-usage): Change options
7518 to take a HOST_WIDE_INT argument and accept a byte-size suffix.
7520 * doc/invoke.texi (GCC Command Options): Document option arguments.
7521 Explain byte-size arguments and suffixes.
7522 (-Wvla-larger-than, -Wno-alloc-size-larger-than): Update.
7523 (-Wno-alloca-larger-than, -Wno-vla-larger-than): Same.
7524 (-Wframe-larger-than, -Wlarger-than, -Wstack-usage): Same.
7525 * doc/options.texi (UInteger): Expand.
7526 (Host_Wide_Int, ByteSize): Document new properties.
7527 * final.c (final_start_function_1): Include sizes in an error message.
7528 * function.c (frame_offset_overflow): Same.
7529 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Adjust.
7530 (alloca_call_type_by_arg): Change function argument to HOST_WIDE_INT.
7531 Diagnose unbounded alloca calls only for limits of less than
7533 (alloca_call_type): Adjust. Diagnose possibly out-of-bounds alloca
7534 calls and VLA size only for limits of less than PTRDIFF_MAX. Same
7536 (pass_walloca::execute): Adjust. Diagnose alloca calls in loops
7537 only for limits of less than PTRDIFF_MAX.
7538 * langhooks-def.h (lhd_handle_option): Change function argument
7540 * langhooks.c (lhd_handle_option): Same.
7541 * langhooks.h (handle_option): Same.
7542 * opt-functions.awk (switch_bit_fields): Handle Host_Wide_Int and
7544 (var_type, var_type_struct): Same.
7545 (var_set): Handle ByteSize flag.
7546 * optc-gen.awk: Add comments to output to ease debugging. Make
7547 use of HOST_WIDE_INT where appropriate.
7548 * opts-gen-save.awk: Use %lx to format unsigned long.
7549 * opth-gen.awk: Change function argument to HOST_WIDE_INT.
7550 * opts-common.c (integral_argument): Return HOST_WIDE_INT and add
7551 arguments. Parse bytes-size suffixes.
7552 (enum_arg_to_value): Change function argument to HOST_WIDE_INT.
7553 (enum_value_to_arg): Same.
7554 (decode_cmdline_option): Handle cl_host_wide_int. Adjust.
7555 (handle_option): Adjust.
7556 (generate_option): Change function argument to HOST_WIDE_INT.
7557 (cmdline_handle_error): Adjust.
7558 (read_cmdline_option): Change function argument to HOST_WIDE_INT.
7559 (set_option): Change function argument to HOST_WIDE_INT.
7560 (option_enabled): Handle cl_host_wide_int.
7561 (get_option_state): Handle CLVC_SIZE.
7562 (control_warning_option): Same.
7563 * opts.c (common_handle_option): Change function argument to
7564 HOST_WIDE_INT. Remove handling of OPT_Walloca_larger_than_ and
7565 OPT_Wvla_larger_than_.
7566 * opts.h (enum cl_var_type): Add an enumerator.
7567 * stor-layout.c (layout_decl): Print a more meaningful warning.
7568 * toplev.c (output_stack_usage): Adjust.
7570 2018-07-20 Qing Zhao <qing.zhao@oracle.com>
7572 * builtins.c (expand_builtin_memcmp): Delete the last parameter for
7573 call to inline_expand_builtin_string_cmp.
7574 (expand_builtin_strcmp): Likewise.
7575 (expand_builtin_strncmp): Likewise.
7576 (inline_string_cmp): Delete the last parameter, change char_type_node
7577 to unsigned_char_type_node for strcmp/strncmp, add conversions to the
7579 (inline_expand_builtin_string_cmp): Delete the last parameter, give up
7580 the inlining expansion on target where the type of the call has same or
7581 narrower precision than unsigned char.
7583 2018-07-20 David Malcolm <dmalcolm@redhat.com>
7585 * Makefile.in (OBJS): Add json.o and optinfo-emit-json.o.
7586 (CFLAGS-optinfo-emit-json.o): Define TARGET_NAME.
7587 * common.opt (fsave-optimization-record): New option.
7588 * coretypes.h (struct kv_pair): Move here from dumpfile.c.
7589 * doc/invoke.texi (-fsave-optimization-record): New option.
7590 * dumpfile.c: Include "optinfo-emit-json.h".
7591 (struct kv_pair): Move to coretypes.h.
7592 (optgroup_options): Make non-static.
7593 (dump_context::end_scope): Call
7594 optimization_records_maybe_pop_dump_scope.
7595 * dumpfile.h (optgroup_options): New decl.
7596 * json.cc: New file.
7598 * optinfo-emit-json.cc: New file.
7599 * optinfo-emit-json.h: New file.
7600 * optinfo.cc: Include "optinfo-emit-json.h".
7601 (optinfo::emit): Call optimization_records_maybe_record_optinfo.
7602 (optinfo_enabled_p): Check optimization_records_enabled_p.
7603 (optinfo_wants_inlining_info_p): Likewise.
7604 * optinfo.h: Update comment.
7605 * profile-count.c (profile_quality_as_string): New function.
7606 * profile-count.h (profile_quality_as_string): New decl.
7607 (profile_count::quality): New accessor.
7608 * selftest-run-tests.c (selftest::run_tests): Call json_cc_tests
7609 and optinfo_emit_json_cc_tests.
7610 * selftest.h (selftest::json_cc_tests): New decl.
7611 (selftest::optinfo_emit_json_cc_tests): New decl.
7612 * toplev.c: Include "optinfo-emit-json.h".
7613 (compile_file): Call optimization_records_finish.
7614 (do_compile): Call optimization_records_start.
7615 * tree-ssa-live.c: Include optinfo.h.
7616 (remove_unused_scope_block_p): Retain inlining information if
7617 optinfo_wants_inlining_info_p returns true.
7619 2018-07-20 Richard Biener <rguenther@suse.de>
7622 * dwarf2out.c (dwarf2out_die_ref_for_decl): Test in_lto_p
7623 to cover -flto-partition=none.
7625 2018-07-20 Martin Liska <mliska@suse.cz>
7627 * tree.h (DECL_LOCATION_RANGE): Remove unused macro.
7628 (get_decl_source_range): Remove unused function.
7630 2018-07-20 Richard Biener <rguenther@suse.de>
7632 * tree-ssa-sccvn.h (struct vn_nary_op_s): Add next member.
7633 (struct vn_phi_s): Likewise.
7634 (struct vn_reference_s): Likewise.
7635 * tree-ssa-sccvn.c (vn_nary_op_hasher::equal): Add shortcut
7636 for searching the slot of an entry known to be in the hash itself.
7637 (vn_phi_hasher::equal): Likewise.
7638 (vn_reference_hasher::equal): Likewise.
7639 (last_inserted_ref, last_inserted_phi, last_inserted_nary): New
7641 (optimistic_info, current_info): Remove, keeping only valid_info.
7642 (vn_reference_lookup_1): Remove fallback lookup.
7643 (vn_reference_lookup_2): Likewise.
7644 (vn_nary_op_lookup_1): Likewise.
7645 (vn_phi_lookup): Likewise.
7646 (vn_nary_build_or_lookup_1): Make sure to not chain the built
7648 (vn_reference_insert): Adjust, chain the inserted hash element
7649 at last_inserted_ref.
7650 (vn_reference_insert_pieces): Likewise.
7651 (visit_reference_op_call): Likewise.
7652 (vn_nary_op_insert_into): Chain the inserted hash element at
7654 (vn_nary_op_insert_pieces): Adjust.
7655 (vn_nary_op_insert): Likewise.
7656 (vn_nary_op_insert_stmt): Likewise.
7657 (vn_phi_insert): Adjust, chain the inserted hash element at
7659 (process_scc): Remove clearing and copying the optimistic
7660 table. Instead remove elements inserted during an optimistic
7661 iteration from the single table we maintain.
7662 (init_scc_vn): Adjust.
7663 (free_scc_vn): Likewise.
7664 (sccvn_dom_walker::record_cond): Likewise.
7665 (sccvn_dom_walker::after_dom_children): Likewise.
7667 2018-07-19 Martin Sebor <msebor@redhat.com>
7669 PR tree-optimization/84047
7670 PR tree-optimization/83776
7671 * tree-vrp.c (vrp_prop::check_mem_ref): New function.
7672 (check_array_bounds): Call it.
7674 2018-07-19 Martin Sebor <msebor@redhat.com>
7676 * align.h (align_flags): Use member initialization.
7678 2018-07-19 David Malcolm <dmalcolm@redhat.com>
7680 * Makefile.in (OBJS): Add optinfo.o.
7681 * coretypes.h (class symtab_node): New forward decl.
7682 (struct cgraph_node): New forward decl.
7683 (class varpool_node): New forward decl.
7684 * dump-context.h: New file.
7685 * dumpfile.c: Include "optinfo.h", "dump-context.h", "cgraph.h",
7687 (refresh_dumps_are_enabled): Use optinfo_enabled_p.
7688 (set_dump_file): Call dumpfile_ensure_any_optinfo_are_flushed.
7689 (set_alt_dump_file): Likewise.
7690 (dump_context::~dump_context): New dtor.
7691 (dump_gimple_stmt): Move implementation to...
7692 (dump_context::dump_gimple_stmt): ...this new member function.
7693 Add the stmt to any pending optinfo, creating one if need be.
7694 (dump_gimple_stmt_loc): Move implementation to...
7695 (dump_context::dump_gimple_stmt_loc): ...this new member function.
7696 Start a new optinfo and add the stmt to it.
7697 (dump_gimple_expr): Move implementation to...
7698 (dump_context::dump_gimple_expr): ...this new member function.
7699 Add the stmt to any pending optinfo, creating one if need be.
7700 (dump_gimple_expr_loc): Move implementation to...
7701 (dump_context::dump_gimple_expr_loc): ...this new member function.
7702 Start a new optinfo and add the stmt to it.
7703 (dump_generic_expr): Move implementation to...
7704 (dump_context::dump_generic_expr): ...this new member function.
7705 Add the tree to any pending optinfo, creating one if need be.
7706 (dump_generic_expr_loc): Move implementation to...
7707 (dump_context::dump_generic_expr_loc): ...this new member
7708 function. Add the tree to any pending optinfo, creating one if
7710 (dump_printf): Move implementation to...
7711 (dump_context::dump_printf_va): ...this new member function. Add
7712 the text to any pending optinfo, creating one if need be.
7713 (dump_printf_loc): Move implementation to...
7714 (dump_context::dump_printf_loc_va): ...this new member function.
7715 Start a new optinfo and add the stmt to it.
7716 (dump_dec): Move implementation to...
7717 (dump_context::dump_dec): ...this new member function. Add the
7718 value to any pending optinfo, creating one if need be.
7719 (dump_context::dump_symtab_node): New member function.
7720 (dump_context::get_scope_depth): New member function.
7721 (dump_context::begin_scope): New member function.
7722 (dump_context::end_scope): New member function.
7723 (dump_context::ensure_pending_optinfo): New member function.
7724 (dump_context::begin_next_optinfo): New member function.
7725 (dump_context::end_any_optinfo): New member function.
7726 (dump_context::s_current): New global.
7727 (dump_context::s_default): New global.
7728 (dump_scope_depth): Delete global.
7729 (dumpfile_ensure_any_optinfo_are_flushed): New function.
7730 (dump_symtab_node): New function.
7731 (get_dump_scope_depth): Reimplement in terms of dump_context.
7732 (dump_begin_scope): Likewise.
7733 (dump_end_scope): Likewise.
7734 (selftest::temp_dump_context::temp_dump_context): New ctor.
7735 (selftest::temp_dump_context::~temp_dump_context): New dtor.
7736 (selftest::verify_item): New function.
7737 (ASSERT_IS_TEXT): New macro.
7738 (ASSERT_IS_TREE): New macro.
7739 (ASSERT_IS_GIMPLE): New macro.
7740 (selftest::test_capture_of_dump_calls): New test.
7741 (selftest::dumpfile_c_tests): Call it.
7742 * dumpfile.h (dump_printf, dump_printf_loc, dump_basic_block)
7743 (dump_generic_expr_loc, dump_generic_expr, dump_gimple_stmt_loc)
7744 (dump_gimple_stmt, dump_dec): Gather these related decls and add a
7745 descriptive comment.
7746 (dump_function, print_combine_total_stats, enable_rtl_dump_file)
7747 (dump_node, dump_bb): Move these unrelated decls.
7748 (class dump_manager): Add leading comment.
7749 * optinfo.cc: New file.
7750 * optinfo.h: New file.
7752 2018-07-19 Michael Collison <michael.collison@arm.com>
7753 Richard Henderson <rth@redhat.com>
7755 * config/aarch64/aarch64.md (subv<GPI>4, usubv<GPI>4): New patterns.
7756 (subti): Handle op1 zero.
7757 (subvti4, usub4ti4): New.
7758 (*sub<GPI>3_compare1_imm): New.
7759 (sub<GPI>3_carryinCV): New.
7760 (*sub<GPI>3_carryinCV_z1_z2, *sub<GPI>3_carryinCV_z1): New.
7761 (*sub<GPI>3_carryinCV_z2, *sub<GPI>3_carryinCV): New.
7763 2018-07-19 Michael Collison <michael.collison@arm.com>
7764 Richard Henderson <rth@redhat.com>
7766 * config/aarch64/aarch64.md: (addv<GPI>4, uaddv<GPI>4): New.
7767 (addti3): Create simpler code if low part is already known to be 0.
7768 (addvti4, uaddvti4): New.
7769 (*add<GPI>3_compareC_cconly_imm): New.
7770 (*add<GPI>3_compareC_cconly): New.
7771 (*add<GPI>3_compareC_imm): New.
7772 (*add<GPI>3_compareC): Rename from add<GPI>3_compare1; do not
7773 handle constants within this pattern..
7774 (*add<GPI>3_compareV_cconly_imm): New.
7775 (*add<GPI>3_compareV_cconly): New.
7776 (*add<GPI>3_compareV_imm): New.
7777 (add<GPI>3_compareV): New.
7778 (add<GPI>3_carryinC, add<GPI>3_carryinV): New.
7779 (*add<GPI>3_carryinC_zero, *add<GPI>3_carryinV_zero): New.
7780 (*add<GPI>3_carryinC, *add<GPI>3_carryinV): New.
7781 ((*add<GPI>3_compareC_cconly_imm): Replace 'ne' operator
7782 with 'comparison' operator.
7783 (*add<GPI>3_compareV_cconly_imm): Ditto.
7784 (*add<GPI>3_compareV_cconly): Ditto.
7785 (*add<GPI>3_compareV_imm): Ditto.
7786 (add<GPI>3_compareV): Ditto.
7787 (add<mode>3_carryinC): Ditto.
7788 (*add<mode>3_carryinC_zero): Ditto.
7789 (*add<mode>3_carryinC): Ditto.
7790 (add<mode>3_carryinV): Ditto.
7791 (*add<mode>3_carryinV_zero): Ditto.
7792 (*add<mode>3_carryinV): Ditto.
7794 2018-07-19 Michael Collison <michael.collison@arm.com>
7795 Richard Henderson <rth@redhat.com>
7797 * config/aarch64/aarch64-modes.def (CC_V): New.
7798 * config/aarch64/aarch64-protos.h
7799 (aarch64_addti_scratch_regs): Declare
7800 (aarch64_subvti_scratch_regs): Declare.
7801 (aarch64_expand_subvti): Declare.
7802 (aarch64_gen_unlikely_cbranch): Declare
7803 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Test
7804 for signed overflow using CC_Vmode.
7805 (aarch64_get_condition_code_1): Handle CC_Vmode.
7806 (aarch64_gen_unlikely_cbranch): New function.
7807 (aarch64_addti_scratch_regs): New function.
7808 (aarch64_subvti_scratch_regs): New function.
7809 (aarch64_expand_subvti): New function.
7811 2018-07-19 Andre Vieira <andre.simoesdiasvieira@arm.com>
7813 * config/aarch64/aarch64-option-extensions.def: New entry for profile
7815 * config/aarch64/aarch64.h (AARCH64_FL_PROFILE): New.
7816 * doc/invoke.texi (aarch64-feature-modifiers): New entry for profile
7819 2018-07-19 Andre Vieira <andre.simoesdiasvieira@arm.com>
7822 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Make
7823 address check not strict.
7825 2018-07-19 Andre Vieira <andre.simoesdiasvieira@arm.com>
7827 * config/aarch64/aarch64-simd.md (aarch64_simd_mov<VQ:mode>): Replace
7829 (store_pair_lanes<mode>): Likewise.
7830 * config/aarch64/aarch64-protos.h (aarch64_addr_query_type): Add new
7831 enum value 'ADDR_QUERY_LDP_STP_N'.
7832 * config/aarch64/aarch64.c (aarch64_addr_query_type): Likewise.
7833 (aarch64_print_address_internal): Add declaration.
7834 (aarch64_print_ldpstp_address): Remove.
7835 (aarch64_classify_address): Adapt mode for 'ADDR_QUERY_LDP_STP_N'.
7836 (aarch64_print_operand): Change printing of 'y'.
7837 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Use
7838 new enum value 'ADDR_QUERY_LDP_STP_N', don't hardcode mode and use
7839 'true' rather than '1'.
7840 * config/aarch64/constraints.md (Uml): Likewise.
7841 (Uml): Rename to Umn.
7844 2018-07-19 Richard Biener <rguenther@suse.de>
7846 * tree-ssa-sccvn.h (struct vn_phi_s): Make phiargs member
7848 * tree-ssa-sccvn.c: Remove alloc-pool.h use.
7849 (vn_phi_hasher): Derive from nofree_ptr_hash and remove remove method.
7850 (vn_reference_hasher): Likewise.
7851 (struct vn_tables_s): Remove obstack and alloc-pool members.
7852 (vn_tables_obstack, vn_tables_insert_obstack): New global obstacks.
7853 (vn_nary_build_or_lookup_1): Manually build in vn_tables_insert_obstack.
7854 (vn_reference_insert): Allocate from obstack instead of from alloc-pool.
7855 (vn_reference_insert_pieces): Likewise.
7856 (alloc_vn_nary_op_noinit): Adjust.
7857 (vn_nary_op_insert_stmt): Allocate phiargs in-place.
7858 (vn_phi_eq): Adjust.
7859 (shared_lookup_phiargs): Remove.
7860 (vn_phi_lookup): Allocate temporary vn_phi_s on the stack.
7861 (vn_phi_insert): Allocate from obstack instead of from alloc-pool.
7862 (visit_reference_op_call): Likewise.
7863 (copy_nary, copy_phi, copy_reference): Remove.
7864 (process_scc): Rewind the obstack when iterating. Do not
7865 copy the elements to valid_info but just move them from one
7866 hashtable to the other.
7867 (allocate_vn_table): Adjust.
7868 (free_vn_table): Likewise.
7869 (init_scc_vn): Likewise.
7870 (free_scc_vn): Likewise.
7872 2018-07-19 H.J. Lu <hongjiu.lu@intel.com>
7875 * config/i386/i386.c (rest_of_insert_endbranch): Lookup
7876 indirect_return as function type attribute.
7877 (ix86_attribute_table): Change indirect_return to function
7879 * doc/extend.texi: Update indirect_return attribute.
7881 2018-07-19 Aldy Hernandez <aldyh@redhat.com>
7883 * wide-int.h (widest2_int): New.
7884 * gimple-fold.c (arith_overflowed_p): Use it.
7885 * tree.h (widest2_int_cst): New.
7886 * tree-vrp.c (wide_int_binop_overflow): Rename from
7887 vrp_int_const_binop.
7888 Rewrite to work on trees.
7889 (extract_range_from_multiplicative_op_1): Abstract code to...
7890 (wide_int_range_min_max): ...here.
7891 (wide_int_range_cross_product): ...and here.
7892 (extract_range_from_binary_expr_1): Abstract overflow code to...
7893 (wide_int_range_mult_wrapping): ...here.
7894 * tree-vrp.h (wide_int_range_cross_product): New.
7895 (wide_int_range_mult_wrapping): New.
7897 2018-07-19 Andrew Senkevich <andrew.senkevich@intel.com>
7898 Julia Koval <julia.koval@intel.com>
7900 * config/i386/x86-tune-costs.h (skylake_memcpy,
7901 skylake_memset): Replace rep_prefix with unrolling for size 512.
7903 2018-07-18 Kugan Vivekanandarajah <kuganv@linaro.org>
7906 * tree-ssa-phiopt.c (cond_removal_in_popcount_pattern): Handle
7907 comparision with EQ_EXPR in last stmt.
7909 2018-07-18 Kelvin Nilsen <kelvin@gcc.gnu.org>
7911 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Rename
7912 this subsection to "PowerPC AltiVec/VSX Built-in Functions".
7913 (PowerPC AltiVec/VSX Built-in Functions): New name for subsection
7914 previously known as "PowerPC AltiVec Built-in Functions". Move
7915 some material to new subsubsections "PowerPC AltiVec Built-in
7916 Functions on ISA 2.06" and "PowerPC AltiVec Built-in Functions on
7918 (PowerPC Altivec Built-in Functions on ISA 2.05): New subsubsection.
7919 (PowerPC Altivec Built-in Functions on ISA 2.06): Likewise.
7920 (PowerPC Altivec Built-in Functions on ISA 2.07): Likewise.
7921 (PowerPC Altivec Built-in Functions on ISA 3.0): Likewise.
7923 2018-07-18 Richard Biener <rguenther@suse.de>
7925 PR tree-optimization/86557
7926 * tree-vect-patterns.c (vect_recog_divmod_pattern): Also handle
7929 2018-07-18 Ilya Leoshkevich <iii@linux.ibm.com>
7931 * config/s390/s390.c (s390_function_profiler): Generate CFI.
7933 2018-07-17 Jeff Law <law@redhat.com>
7935 * config/arm/arm.c (get_label_padding): Update for recent
7936 changes to label_to_alignment.
7938 PR tree-optimization/86010
7939 * tree-ssa-dse.c (compute_trims): Fix typo/thinko.
7941 * config/mips/mips.c (vr4130_align_insns): Update for recent
7942 changes to label_to_alignment.
7944 * config/frv/frv.c (frv_label_align): Update for recent changes
7945 to label_to_alignment.
7947 * config/nios2/nios2.c (nios2_label_align): Update for recent
7948 changes which dropped ALIGN_LABELS_LOG.
7950 2018-07-17 Andreas Schwab <schwab@linux-m68k.org>
7952 * config/m68k/m68k.md (umulsi3_highpart+1, const_umulsi3_highpart)
7953 (smulsi3_highpart+1, const_smulsi3_highpart): Add CC_STATUS_INIT.
7955 2018-07-17 Claudiu Zissulescu <claziss@synopsys.com>
7957 * config/arc/arc.c (arc_label_align): Use align_labels instead of
7958 deprecated align_labels_log.
7960 2018-07-17 Richard Biener <rguenther@suse.de>
7963 * dwarf2out.c (init_sections_and_labels): Always generate
7964 a debug_line_str_section for early LTO debug.
7965 (dwarf2out_finish): Reset debug_line_str_hash output early.
7966 Bump counter for extra dwarf5 .debug_loc labels to not conflict
7968 (dwarf2out_early_finish): Output debug_line_str.
7970 2018-07-17 Robin Dapp <rdapp@linux.ibm.com>
7972 * config/s390/s390.c (preferred_la_operand_p): Do not use LA with
7973 index register on z196 or later.
7975 2018-07-17 Robin Dapp <rdapp@linux.ibm.com>
7977 * config/s390/s390.c (s390_default_align): Set default function
7979 (s390_override_options_after_change): Call s390_default align.
7980 (s390_option_override_internal): Call s390_default_align.
7981 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define.
7983 2018-07-17 Jakub Jelinek <jakub@redhat.com>
7986 * omp-low.c (create_task_copyfn): Copy over also fields corresponding
7987 to _looptemp_ clauses, other than the first two.
7989 2018-07-17 Martin Liska <mliska@suse.cz>
7991 * opts.c: Do not enable OPT_falign_* for -Os.
7993 2018-07-17 Martin Liska <mliska@suse.cz>
7995 * align.h (MAX_CODE_ALIGN): New.
7996 (MAX_CODE_ALIGN_VALUE): New.
7997 * common/config/i386/i386-common.c (ix86_handle_option):
7998 (MAX_CODE_ALIGN): Moved to align.h.
7999 * final.c (MAX_CODE_ALIGN): Likewise.
8000 * opts.c (parse_and_check_align_values):
8001 (MAX_CODE_ALIGN): Likewise.
8002 (MAX_CODE_ALIGN_VALUE): Likewise.
8004 2018-07-17 Martin Liska <mliska@suse.cz>
8006 * config/i386/att.h (ASM_OUTPUT_ALIGN): Fix spacing
8007 in order to fulfil coding style.
8008 * config/i386/cygming.h (ASM_OUTPUT_ALIGN): Likewise.
8009 * config/i386/gas.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
8010 * config/i386/x86-64.h (ASM_OUTPUT_MAX_SKIP_PAD): Likewise.
8011 * config/iq2000/iq2000.h (ASM_OUTPUT_ALIGN): Likewise.
8012 * config/pa/pa.h (ASM_OUTPUT_ALIGN): Likewise.
8013 * config/sparc/sol2.h (ASM_OUTPUT_ALIGN_WITH_NOP): Likewise.
8014 * config/sparc/sparc.h (ASM_OUTPUT_ALIGN): Likewise.
8015 * config/visium/visium.h (ASM_OUTPUT_ALIGN): Likewise.
8016 (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
8018 2018-07-17 Martin Liska <mliska@suse.cz>
8020 * align.h: New file.
8021 * config/alpha/alpha.c (alpha_align_insns_1): Use align_functions
8023 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Use new return type
8024 align_flags of label_to_alignment.
8025 * config/m32r/m32r.h (LOOP_ALIGN): Wrap returned values into
8027 * config/m68k/m68k.c: Do not use removed align_labels_value and
8029 * config/nds32/nds32.h (JUMP_ALIGN): Wrap result into align_flags class.
8030 (LOOP_ALIGN): Likewise.
8031 (LABEL_ALIGN): Likewise.
8032 * config/powerpcspe/powerpcspe.c (TARGET_ASM_LOOP_ALIGN_MAX_SKIP):
8033 Remove not used macro.
8034 (rs6000_loop_align): Change return type to align_flags.
8035 (rs6000_loop_align_max_skip): Remove.
8036 * config/rs6000/rs6000-protos.h (rs6000_loop_align):
8037 Change return type to align_flags.
8038 * config/rs6000/rs6000.c (TARGET_ASM_LOOP_ALIGN_MAX_SKIP):
8039 Remove not used macro.
8040 (rs6000_loop_align): Change return type to align_flags.
8041 (rs6000_loop_align_max_skip): Remove.
8042 * config/rx/rx.h (JUMP_ALIGN): Wrap integer values
8043 * config/rx/rx-protos.h (rx_align_for_label): Make it
8045 * config/rx/rx.c (rx_align_for_label): Change return type
8047 (rx_max_skip_for_label): Remove TARGET_ASM_*_ALIGN_MAX_SKIP
8049 into align_flags class.
8050 (LABEL_ALIGN): Likewise.
8051 (LOOP_ALIGN): Likewise.
8052 * config/s390/s390.c (s390_label_align): Use align_flags
8054 (s390_asm_output_function_label): Likewise.
8055 * config/sh/sh.c (sh_override_options_after_change):
8056 Use align_flags class directly without macros.
8057 (find_barrier): Likewise.
8058 (barrier_align): Likewise.
8059 (sh_loop_align): Likewise.
8060 * config/spu/spu.c (spu_option_override):
8061 Use align_flags_tuple::get_value instead of removed macros.
8062 (spu_sched_init): Likewise.
8063 * config/spu/spu.h (GTY): Likewise.
8064 * config/visium/visium.c (visium_option_override):
8065 Set "8" as default secondary alignment.
8066 * config/visium/visium.h (SUBALIGN_LOG): Define to 3
8067 in order to guarantee secondary alignment of 8.
8068 * coretypes.h: Include align.h header file.
8069 * doc/tm.texi: Remove TARGET_ASM_JUMP_ALIGN_MAX_SKIP,
8070 TARGET_ASM_LOOP_ALIGN_MAX_SKIP, TARGET_ASM_LABEL_ALIGN_MAX_SKIP
8071 and TARGET_ASM_LABEL_ALIGN_AFTER_BARRIER_MAX_SKIP macros.
8072 * doc/tm.texi.in: Likewise.
8073 * final.c (struct label_alignment): Remove not used structure.
8074 (LABEL_ALIGN): Change type to align_flags.
8075 (LOOP_ALIGN): Likewise.
8076 (JUMP_ALIGN): Likewise.
8077 (default_loop_align_max_skip): Remove.
8078 (default_label_align_max_skip): Likewise.
8079 (default_jump_align_max_skip): Likewise.
8080 (default_label_align_after_barrier_max_skip):
8081 (LABEL_TO_ALIGNMENT): Change to access label_align vector.
8082 (LABEL_TO_MAX_SKIP): Remove.
8083 (label_to_alignment): Return align_flags type instead of integer.
8084 (label_to_max_skip): Remove.
8085 (align_fuzz): Use align_flags type.
8086 (compute_alignments): Use align_flags type and use align_flags::max
8087 to combine multiple alignments.
8088 (grow_label_align): Grow vec instead of C array.
8089 (update_alignments): Assign just LABEL_TO_ALIGNMENT.
8090 (shorten_branches): Use align_flags type and use align_flags::max
8091 to combine multiple alignments.
8092 (final_scan_insn_1): Remove usage of secondary alignment that comes
8093 from label alignment, but instead use proper secondary alignment
8094 which is computed in grow_label_align.
8095 * flags.h (struct align_flags_tuple): Move to align.h.
8096 (struct align_flags): Likewise.
8097 (state_align_loops): Rename to align_loops.
8098 (state_align_jumps): Rename to align_jumps.
8099 (state_align_labels): Rename to align_labels.
8100 (state_align_functions): Rename to align_functions.
8101 (align_loops_log): Remove.
8102 (align_jumps_log): Remove.
8103 (align_labels_log): Remove.
8104 (align_functions_log): Remove.
8105 (align_loops_max_skip): Remove.
8106 (align_jumps_max_skip): Remove.
8107 (align_labels_max_skip): Remove.
8108 (align_functions_max_skip): Remove.
8109 (align_loops_value): Remove.
8110 (align_jumps_value): Remove.
8111 (align_labels_value): Remove.
8112 (align_functions_value): Remove.
8113 * output.h (label_to_alignment): Change return type to align_flags.
8114 (label_to_max_skip): Remove.
8115 * target.def: Remove loop_align_max_skip, label_align_max_skip,
8116 jump_align_max_skip macros.
8117 * targhooks.h (default_loop_align_max_skip): Remove.
8118 (default_label_align_max_skip): Likewise.
8119 (default_jump_align_max_skip): Likewise.
8120 (default_label_align_after_barrier_max_skip): Remove.
8121 * toplev.c (read_log_maxskip): Use ::normalize function.
8122 (parse_N_M): Remove not used argument and also call ::normalize.
8123 (parse_alignment_opts): Do not pass unused arguments.
8124 * varasm.c (assemble_start_function): Use directly align_functions
8125 instead of removed macros.
8126 * system.h: Do not poison removed macros.
8128 2018-07-17 Jakub Jelinek <jakub@redhat.com>
8131 * gimplify.c (gimplify_omp_for): Ensure taskloop firstprivatized init
8132 and cond temporaries don't have reference type if iterator has
8133 pointer type. For init use &for_pre_body instead of pre_p if
8134 for_pre_body is non-empty.
8136 2018-07-16 Segher Boessenkool <segher@kernel.crashing.org>
8138 * config/rs6000/rs6000.md (trunc<mode>sf2): Expand truncates of
8139 double-double modes to SFmode directly directly.
8140 (trunc<mode>sf2_fprs): Delete.
8142 2018-07-16 Segher Boessenkool <segher@kernel.crashing.org>
8144 * config/rs6000/rs6000.c (init_float128_ibm): Use the correct names
8145 for conversions between IFmode and the decimal floating point modes.
8146 (init_float128_ieee): Use the correct names for conversions between
8147 KFmode and the decimal floating point modes.
8149 2018-07-16 Segher Boessenkool <segher@kernel.crashing.org>
8151 * config/rs6000/rs6000.c (init_float128_ibm): Use more correct names
8152 for the conversions between TDmode and IFmode.
8153 (init_float128_ieee): Use more correct names for the conversions
8154 between TDmode and KFmode.
8156 2018-07-16 Jakub Jelinek <jakub@redhat.com>
8158 PR tree-optimization/86526
8159 * builtins.c (expand_builtin_memcmp): Formatting fixes.
8160 (inline_expand_builtin_string_cmp): Likewise.
8161 (inline_string_cmp): Likewise. Use c_readstr instead of
8162 builtin_memcpy_read_str. Add unit_mode temporary.
8164 2018-07-16 Bernd Edlinger <bernd.edlinger@hotmail.de>
8167 * builtins.c (check_access): Bail out if range[0] is no INTEGER_CST.
8168 * expr.c (string_constant): Fix the element size of ARRAY_TYPE.
8170 2018-07-16 Kelvin Nilsen <kelvin@gcc.gnu.org>
8172 * doc/extend.texi (PowerPC AltiVec Built-in Functions):
8173 Alphabetize prototypes of built-in functions, separating out
8174 built-in functions that are listed in this section but should be
8175 described elsewhere.
8177 2018-07-16 Uros Bizjak <ubizjak@gmail.com>
8180 * expmed.c (emit_store_flag): Do not emit setcc followed by a
8181 conditional move when trapping comparison was split to a
8182 non-trapping one (and vice versa).
8184 2018-07-16 Ilya Leoshkevich <iii@linux.ibm.com>
8186 * config/s390/s390.c (s390_function_profiler): Generate nops
8187 instead of profiler call sequences.
8188 * config/s390/s390.opt: Add the new option.
8190 2018-07-16 Ilya Leoshkevich <iii@linux.ibm.com>
8192 * config/s390/s390.c (s390_function_profiler): Generate
8193 __mcount_loc section.
8194 * config/s390/s390.opt: Add the new option.
8196 2018-07-16 Ilya Leoshkevich <iii@linux.ibm.com>
8198 * common.opt: Add the new warning.
8199 * config/s390/s390.c (s390_function_profiler): Emit "brasl
8200 %r0,__fentry__" when -mfentry is specified.
8201 (s390_option_override_internal): Disallow -mfentry for 31-bit
8203 * config/s390/s390.opt: Add the new option.
8205 2018-07-16 Richard Biener <rguenther@suse.de>
8208 * dwarf2out.c (dwarf2out_register_external_die): Assign DIE parents
8209 for function-local FUNCTION_DECL and RESULT_DECL immediately.
8211 2018-07-16 Martin Liska <mliska@suse.cz>
8214 * ipa-pure-const.c (malloc_candidate_p): Revert ::get
8217 2017-07-16 Claudiu Zissulescu <claziss@synopsys.com>
8219 * config/arc/arcHS.md: Update ARCHS scheduling rules.
8221 2017-07-16 Claudiu Zissulescu <claziss@synopsys.com>
8223 * config/arc/arc-arch.h (arc_tune_attr): Add new tune parameters
8225 * config/arc/arc-cpus.def (hs4x): New cpu.
8227 * config/arc/arc-tables.opt: Regenerate.
8228 * config/arc/arc.c (arc_sched_issue_rate): New function.
8229 (TARGET_SCHED_ISSUE_RATE): Define.
8230 (TARGET_SCHED_EXPOSED_PIPELINE): Likewise.
8231 * config/arc/arc.md (attr type): Add fpu_fuse, fpu_sdiv, fpu_ddiv,
8233 (attr tune): Add ARCHS4x tune values.
8234 (attr tune_dspmpy): Define.
8235 (*tst): Correct instruction type.
8236 * config/arc/arcHS.md: Don't use this automaton for ARCHS4x cpus.
8237 * config/arc/arcHS4x.md: New file.
8238 * config/arc/fpu.md: Update instruction type attributes.
8239 * config/arc/t-multilib: Regenerate.
8241 2018-07-16 Tom de Vries <tdevries@suse.de>
8244 * var-tracking.c (vt_initialize): Fix pre_dec handling.
8246 2018-07-16 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
8248 * config/aarch64/atomics.md (aarch64_store_execlusive<mode>): Add
8251 2018-07-16 Eric Botcazou <ebotcazou@adacore.com>
8253 PR tree-optimization/86514
8254 * tree-ssa-reassoc.c (init_range_entry) <CASE_CONVERT>: Return for a
8255 conversion to a boolean type from a type with greater precision.
8257 2018-07-16 Tom de Vries <tdevries@suse.de>
8259 * var-tracking.c (vt_initialize): Print adjusted insn slim if
8260 dump_flags request TDF_SLIM.
8262 2018-07-16 Aldy Hernandez <aldyh@redhat.com>
8264 * fold-const.c (int_const_binop_1): Abstract...
8265 (wide_int_binop): ...wide int code here.
8266 (poly_int_binop): ...poly int code here.
8267 Abstract the rest of int_const_binop_1 into int_const_binop.
8268 * fold-const.h (wide_int_binop): New.
8269 * tree-vrp.c (vrp_int_const_binop): Call wide_int_binop.
8270 Remove useless PLUS/MINUS_EXPR case.
8271 (zero_nonzero_bits_from_vr): Move wide int code...
8272 (zero_nonzero_bits_from_bounds): ...here.
8273 (extract_range_from_binary_expr_1): Move mask optimization code...
8274 (range_easy_mask_min_max): ...here.
8275 * tree-vrp.h (zero_nonzero_bits_from_bounds): New.
8276 (range_easy_mask_min_max): New.
8278 2018-07-15 Jeff Law <law@redhat.com>
8281 * config/sh/sh.c (output_mi_thunk): Remove dead conditional
8284 2018-07-14 Jim Wilson <jimw@sifive.com>
8286 * config/riscv/linux.h (TARGET_ASM_FILE_END): New.
8288 2018-07-14 Paul Koning <ni1d@arrl.net>
8290 * config/pdp11/pdp11.c (pdp11_rtx_costs): Bugfixes.
8292 2018-07-13 Jan Hubicka <hubicka@ucw.cz>
8294 * lto-streamer-out.c (copy_function_or_variable): Dump info about
8297 2018-07-13 Bill Schmidt <wschmidt@linux.ibm.com>
8298 Steve Munroe <munroesj52@gmail.com>
8300 * config/rs6000/emmintrin.h (_mm_and_si128): New function.
8301 (_mm_andnot_si128): Likewise.
8302 (_mm_or_si128): Likewise.
8303 (_mm_xor_si128): Likewise.
8305 2018-07-13 Qing Zhao <qing.zhao@oracle.com>
8308 * builtins.c (expand_builtin_memcmp): Inline the calls first
8309 when result_eq is false.
8310 (expand_builtin_strcmp): Inline the calls first.
8311 (expand_builtin_strncmp): Likewise.
8312 (inline_string_cmp): New routine. Expand a string compare
8313 call by using a sequence of char comparison.
8314 (inline_expand_builtin_string_cmp): New routine. Inline expansion
8315 a call to str(n)cmp/memcmp.
8316 * doc/invoke.texi (--param builtin-string-cmp-inline-length):
8318 * params.def (BUILTIN_STRING_CMP_INLINE_LENGTH): New.
8320 2018-07-13 Richard Earnshaw <rearnsha@arm.com>
8322 * config/arm/driver-arm.c: Include arm-native.h.
8323 (host_detect_local_cpu): Use auto-generated data tables.
8324 (vendors, arm_cpu_table): Delete. Move part information to ...
8325 * config/arm/arm-cpus.in: ... here.
8326 * config/arm/parsecpu.awk (gen_native): New function.
8327 (vendor, part): New CPU fields.
8328 (END): Add support for building the native CPU detection tables.
8329 * config/arm/t-arm (arm-native.h): Add build rule.
8330 (driver-arm.o): Add dependency on arm-native.h.
8332 2018-07-13 Richard Biener <rguenther@suse.de>
8335 * match.pd (addr1 - addr2): Allow either of the operand to
8338 2018-07-13 Tom de Vries <tdevries@suse.de>
8340 * tree-inline.c (remap_ssa_name): Save and reuse debug exprs generated
8343 2018-07-13 Jackson Woodruff <jackson.woodruff@arm.com>
8345 * config/aarch64/aarch64.c (aarch64_operands_adjust_ok_for_ldpstp): Use
8346 arrays instead of numbered variables.
8348 2018-07-13 Eric Botcazou <ebotcazou@adacore.com>
8350 * config/sparc/sparc-protos.h (sparc_compute_frame_size): Delete.
8351 * config/sparc/sparc.c (sparc_compute_frame_size): Make static.
8353 2018-07-13 Richard Biener <rguenther@suse.de>
8356 * dwarf2out.c (gen_type_die_with_usage): Use scope_die_for
8357 instead of get_context_die.
8359 2018-07-13 Kugan Vivekanandarajah <kuganv@linaro.org>
8360 Richard Biener <rguenther@suse.de>
8363 * tree-ssa-loop-niter.c (number_of_iterations_popcount): Check
8364 that the loop latch destination where phi is defined.
8366 2018-07-12 Kito Cheng <kito.cheng@gmail.com>
8368 * config/riscv/riscv.c (enum riscv_privilege_levels): Add UNKNOWN_MODE.
8369 (riscv_expand_epilogue): Add assertion to check interrupt mode.
8370 (riscv_set_current_function): Extract getting interrupt type to new
8372 (riscv_get_interrupt_type): New function.
8373 (riscv_merge_decl_attributes): New function, checking interrupt type is
8375 (TARGET_MERGE_DECL_ATTRIBUTES): Define.
8377 2018-07-12 Paul Koning <ni1d@arrl.net>
8379 * config/pdp11/pdp11.c (pdp11_output_def): Fix typo in .set
8382 2018-07-12 Paul Koning <ni1d@arrl.net>
8384 * doc/rtl.texi (REG_NONNEG): Remove decrement and branch until
8385 zero reference, add doloop_end instead.
8386 * doc/md.texi (decrement_and_branch_until_zero): Remove.
8387 (Looping patterns): Remove decrement_and_branch_until_zero. Add
8388 detail for doloop_end.
8390 2018-07-12 Martin Sebor <msebor@redhat.com>
8393 * attribs.c (decl_attributes): Reject conflicting attributes before
8394 calling attribute handlers.
8396 2018-07-12 Jan Hubicka <hubicka@ucw.cz>
8398 * dumpfile.c (gcc::dump_manager::get_dump_file_name): Add PART
8400 (gcc::dump_manager::get_dump_file_name): likewise.
8401 (dump_begin): Likewise.
8402 * dumpfile.h (dump_begin): Update prototype.
8403 (gcc::dump_manager::get_dump_file_name,
8404 gcc::dump_manager::get_dump_file_name): Update prototype.
8406 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
8408 * internal-fn.h (vectorizable_internal_fn_p): New function.
8409 * tree-vect-slp.c (compatible_calls_p): Likewise.
8410 (vect_build_slp_tree_1): Remove nops argument. Handle calls
8411 to internal functions.
8412 (vect_build_slp_tree_2): Update call to vect_build_slp_tree_1.
8414 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
8416 * fold-const.h (inverse_conditions_p): Declare.
8417 * fold-const.c (inverse_conditions_p): New function.
8418 * match.pd: Use inverse_conditions_p. Add folds of view_converts
8419 that test the inverse condition of a conditional internal function.
8420 * internal-fn.h (vectorized_internal_fn_supported_p): Declare.
8421 * internal-fn.c (internal_fn_mask_index): Handle conditional
8423 (vectorized_internal_fn_supported_p): New function.
8424 * tree-if-conv.c: Include internal-fn.h and fold-const.h.
8425 (any_pred_load_store): Replace with...
8426 (need_to_predicate): ...this new variable.
8427 (redundant_ssa_names): New variable.
8428 (ifcvt_can_use_mask_load_store): Move initial checks to...
8429 (ifcvt_can_predicate): ...this new function. Handle tree codes
8430 for which a conditional internal function exists.
8431 (if_convertible_gimple_assign_stmt_p): Use ifcvt_can_predicate
8432 instead of ifcvt_can_use_mask_load_store. Update after variable
8434 (predicate_load_or_store): New function, split out from
8435 predicate_mem_writes.
8436 (check_redundant_cond_expr): New function.
8437 (value_available_p): Likewise.
8438 (predicate_rhs_code): Likewise.
8439 (predicate_mem_writes): Rename to...
8440 (predicate_statements): ...this. Use predicate_load_or_store
8441 and predicate_rhs_code.
8442 (combine_blocks, tree_if_conversion): Update after above name changes.
8443 (ifcvt_local_dce): Handle redundant_ssa_names.
8444 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern): Handle
8445 general conditional functions.
8446 * tree-vect-stmts.c (vectorizable_call): Likewise.
8448 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
8449 Alan Hayward <alan.hayward@arm.com>
8450 David Sherwood <david.sherwood@arm.com>
8452 * internal-fn.h (can_interpret_as_conditional_op_p): Declare.
8453 * internal-fn.c (can_interpret_as_conditional_op_p): New function.
8454 * tree-ssa-math-opts.c (convert_mult_to_fma_1): Handle conditional
8455 plus and minus and convert them into IFN_COND_FMA-based sequences.
8456 (convert_mult_to_fma): Handle conditional plus and minus.
8458 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
8460 * doc/md.texi (cond_fma, cond_fms, cond_fnma, cond_fnms): Document.
8461 * optabs.def (cond_fma_optab, cond_fms_optab, cond_fnma_optab)
8462 (cond_fnms_optab): New optabs.
8463 * internal-fn.def (COND_FMA, COND_FMS, COND_FNMA, COND_FNMS): New
8465 (FMA): Use DEF_INTERNAL_FLT_FN rather than DEF_INTERNAL_FLT_FLOATN_FN.
8466 * internal-fn.h (get_conditional_internal_fn): Declare.
8467 (get_unconditional_internal_fn): Likewise.
8468 * internal-fn.c (cond_ternary_direct): New macro.
8469 (expand_cond_ternary_optab_fn): Likewise.
8470 (direct_cond_ternary_optab_supported_p): Likewise.
8471 (FOR_EACH_COND_FN_PAIR): Likewise.
8472 (get_conditional_internal_fn): New function.
8473 (get_unconditional_internal_fn): Likewise.
8474 * gimple-match.h (gimple_match_op::MAX_NUM_OPS): Bump to 5.
8475 (gimple_match_op::gimple_match_op): Add a new overload for 5
8477 (gimple_match_op::set_op): Likewise.
8478 (gimple_resimplify5): Declare.
8479 * genmatch.c (decision_tree::gen): Generate simplifications for
8481 * gimple-match-head.c (gimple_simplify): Define an overload for
8482 5 operands. Handle calls with 5 arguments in the top-level overload.
8483 (convert_conditional_op): Handle conversions from unconditional
8484 internal functions to conditional ones.
8485 (gimple_resimplify5): New function.
8486 (build_call_internal): Pass a fifth operand.
8487 (maybe_push_res_to_seq): Likewise.
8488 (try_conditional_simplification): Try converting conditional
8489 internal functions to unconditional internal functions.
8490 Handle 3-operand unconditional forms.
8491 * match.pd (UNCOND_TERNARY, COND_TERNARY): Operator lists.
8492 Define ternary equivalents of the current rules for binary conditional
8494 * config/aarch64/aarch64.c (aarch64_preferred_else_value): Handle
8496 * config/aarch64/iterators.md (UNSPEC_COND_FMLA, UNSPEC_COND_FMLS)
8497 (UNSPEC_COND_FNMLA, UNSPEC_COND_FNMLS): New unspecs.
8498 (optab): Handle them.
8499 (SVE_COND_FP_TERNARY): New int iterator.
8500 (sve_fmla_op, sve_fmad_op): New int attributes.
8501 * config/aarch64/aarch64-sve.md (cond_<optab><mode>)
8502 (*cond_<optab><mode>_2, *cond_<optab><mode_4)
8503 (*cond_<optab><mode>_any): New SVE_COND_FP_TERNARY patterns.
8505 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
8507 * target.def (preferred_else_value): New target hook.
8508 * doc/tm.texi.in (TARGET_PREFERRED_ELSE_VALUE): New hook.
8509 * doc/tm.texi: Regenerate.
8510 * targhooks.h (default_preferred_else_value): Declare.
8511 * targhooks.c (default_preferred_else_value): New function.
8512 * internal-fn.h (conditional_internal_fn_code): Declare.
8513 * internal-fn.c (FOR_EACH_CODE_MAPPING): New macro.
8514 (get_conditional_internal_fn): Use it.
8515 (conditional_internal_fn_code): New function.
8516 * gimple-match.h (gimple_match_cond): New struct.
8517 (gimple_match_op): Add a cond member function.
8518 (gimple_match_op::gimple_match_op): Update all forms to take a
8520 * genmatch.c (expr::gen_transform): Use the same condition as res_op
8521 for the suboperation, but don't specify a particular else_value.
8522 * tree-ssa-sccvn.c (vn_nary_simplify, vn_reference_lookup_3)
8523 (visit_nary_op, visit_reference_op_load): Pass
8524 gimple_match_cond::UNCOND to the gimple_match_op constructor.
8525 * gimple-match-head.c: Include tree-eh.h
8526 (convert_conditional_op): New function.
8527 (maybe_resimplify_conditional_op): Likewise.
8528 (gimple_resimplify1): Call maybe_resimplify_conditional_op.
8529 (gimple_resimplify2): Likewise.
8530 (gimple_resimplify3): Likewise.
8531 (gimple_resimplify4): Likewise.
8532 (maybe_push_res_to_seq): Return null for conditional operations.
8533 (try_conditional_simplification): New function.
8534 (gimple_simplify): Call it. Pass conditions to the gimple_match_op
8536 * match.pd: Fold VEC_COND_EXPRs of an IFN_COND_* call to a new
8538 * config/aarch64/aarch64.c (aarch64_preferred_else_value): New
8540 (TARGET_PREFERRED_ELSE_VALUE): Redefine.
8542 2018-07-12 Jan Hubicka <hubicka@ucw.cz>
8544 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not stream
8546 (hash_tree): Do not hash DECL_FCONTEXT
8547 * tree-streamer-in.c (lto_input_ts_field_decl_tree_pointers):
8548 Do not stream DECL_FCONTEXT.
8549 * tree-streamer-out.c (write_ts_field_decl_tree_pointers): Likewise.
8550 * tree.c (free_lang_data_in_decl): Free DECL_FCONTEXT.
8552 2018-07-12 Richard Biener <rguenther@suse.de>
8555 * dwarf2out.c (gen_block_die): Only output blocks when they have
8556 at least one !DECL_IGNORED_P variable.
8558 2018-07-12 Richard Biener <rguenther@suse.de>
8561 * config/gnu-user.h (GNU_USER_TARGET_NO_PTHREADS_LIB_SPEC):
8562 Remove -mieee-fp handling.
8564 2018-07-12 Richard Biener <rguenther@suse.de>
8566 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Remove bogus
8567 left-over from last patch.
8569 2018-07-12 Jakub Jelinek <jakub@redhat.com>
8571 PR tree-optimization/86492
8572 * gimple-ssa-store-merging.c
8573 (imm_store_chain_info::coalesce_immediate_stores): Call
8574 check_no_overlap even for the merge_overlapping case. Formatting fix.
8576 2018-07-12 Richard Biener <rguenther@suse.de>
8579 * fold-const.c (fold_binary_op_with_conditional_arg): Do not
8580 move possibly trapping operations into the conditional.
8582 2018-07-12 Richard Biener <rguenther@suse.de>
8584 * tree-ssa-sccvn.c (mprts_hook_cnt): Remove.
8585 (vn_lookup_simplify_result): Remove recursion limit applied
8587 (vn_nary_build_or_lookup_1): Adjust.
8588 (try_to_simplify): Likewise.
8589 * gimple-match-head.c (gimple_resimplify1): Instead apply one
8591 (gimple_resimplify2): Likewise.
8592 (gimple_resimplify3): Likewise.
8593 (gimple_resimplify4): Likewise.
8595 2018-07-11 Jakub Jelinek <jakub@redhat.com>
8597 * config/i386/avx512bitalgintrin.h (_mm512_mask_bitshuffle_epi64_mask):
8598 Use __mmask64 type instead of __mmask8 for __M argument.
8599 * config/i386/avx512fintrin.h (_mm512_mask_xor_epi64,
8600 _mm512_maskz_xor_epi64): Use __mmask8 type instead of __mmask16 for
8602 (_mm512_mask_cmpneq_epi64_mask): Use __mmask8 type instead of
8603 __mmask16 for __M argument.
8604 (_mm512_maskz_insertf32x4, _mm512_maskz_inserti32x4,
8605 _mm512_mask_insertf32x4, _mm512_mask_inserti32x4): Cast last argument
8606 to __mmask16 instead of __mmask8.
8607 * config/i386/avx512vlintrin.h (_mm_mask_add_ps, _mm_maskz_add_ps,
8608 _mm256_mask_add_ps, _mm256_maskz_add_ps, _mm_mask_sub_ps,
8609 _mm_maskz_sub_ps, _mm256_mask_sub_ps, _mm256_maskz_sub_ps,
8610 _mm256_maskz_cvtepi32_ps, _mm_maskz_cvtepi32_ps): Use __mmask8 type
8611 instead of __mmask16 for __U argument.
8612 * config/i386/avx512vlbwintrin.h (_mm_mask_cmp_epi8_mask): Use
8613 __mmask16 instead of __mmask8 for __U argument.
8614 (_mm256_mask_cmp_epi8_mask): Use __mmask32 instead of __mmask16 for
8616 (_mm256_cmp_epi8_mask): Use __mmask32 return type instead of
8618 (_mm_mask_cmp_epu8_mask): Use __mmask16 instead of __mmask8 for __U
8620 (_mm256_mask_cmp_epu8_mask): Use __mmask32 instead of __mmask16 for
8622 (_mm256_cmp_epu8_mask): Use __mmask32 return type instead of
8624 (_mm_mask_cmp_epi16_mask): Cast last argument to __mmask8 instead
8626 (_mm256_mask_cvtepi8_epi16): Use __mmask16 instead of __mmask32 for
8628 (_mm_mask_cvtepi8_epi16): Use __mmask8 instead of __mmask32 for
8630 (_mm256_mask_cvtepu8_epi16): Use __mmask16 instead of __mmask32 for
8632 (_mm_mask_cvtepu8_epi16): Use __mmask8 instead of __mmask32 for
8634 (_mm256_mask_cmpneq_epu8_mask, _mm256_mask_cmplt_epu8_mask,
8635 _mm256_mask_cmpge_epu8_mask, _mm256_mask_cmple_epu8_mask): Change
8636 return type as well as __M argument type and all casts from __mmask8
8638 (_mm256_mask_cmpneq_epu16_mask, _mm256_mask_cmplt_epu16_mask,
8639 _mm256_mask_cmpge_epu16_mask, _mm256_mask_cmple_epu16_mask): Change
8640 return type as well as __M argument type and all casts from __mmask8
8642 (_mm256_mask_cmpneq_epi8_mask, _mm256_mask_cmplt_epi8_mask,
8643 _mm256_mask_cmpge_epi8_mask, _mm256_mask_cmple_epi8_mask): Change
8644 return type as well as __M argument type and all casts from __mmask8
8646 (_mm256_mask_cmpneq_epi16_mask, _mm256_mask_cmplt_epi16_mask,
8647 _mm256_mask_cmpge_epi16_mask, _mm256_mask_cmple_epi16_mask): Change
8648 return type as well as __M argument type and all casts from __mmask8
8650 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_shrdi_epi32,
8651 _mm_mask_shldi_epi32): Cast last argument to __mmask8 instead of
8654 2018-07-11 Grazvydas Ignotas <notasas@gmail.com>
8656 * config/i386/avx512bwintrin.h: (_mm512_mask_cmp_epi8_mask,
8657 _mm512_mask_cmp_epu8_mask): Use __mmask64 type instead of __mmask32
8660 2018-07-11 Paul Koning <ni1d@arrl.net>
8662 * doc/md.texi (define_subst): Document how multiple occurrences of
8663 the same argument in the replacement pattern are handled.
8665 2018-07-11 Paul Koning <ni1d@arrl.net>
8667 * doc/extend.texi (Common Variable Attributes): Move "mode" into
8669 (Common Type Attributes): Add "mode" attribute.
8671 2018-07-11 Jan Hubicka <hubicka@ucw.cz>
8673 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not
8674 stream DECL_ORIGINAL_TYPE.
8675 (DFS::DFS_write_tree_body): Drop hack handling local external decls.
8676 (hash_tree): Do not walk DECL_ORIGINAL_TYPE.
8677 * tree-streamer-in.c (lto_input_ts_decl_non_common_tree_pointers):
8678 Do not walk original type.
8679 * tree-streamer-out.c (streamer_write_chain): Drop hack handling
8681 (write_ts_decl_non_common_tree_pointers): Do not stream
8683 * tree.c (free_lang_data_in_decl): Clear DECL_ORIGINAL_TYPE.
8684 (find_decls_types_r): Do not walk DEC_ORIGINAL_TYPE.
8686 2018-07-11 Aldy Hernandez <aldyh@redhat.com>
8688 * tree-ssa-threadupdate.c (thread_through_all_blocks): Do not jump
8689 thread twice from the same starting edge.
8691 2018-07-11 Aldy Hernandez <aldyh@redhat.com>
8693 * vr-values.c (gimple_stmt_nonzero_p): Abstract common code to...
8694 * gimple.c (gimple_call_nonnull_result_p): ...here...
8695 (gimple_call_nonnull_arg): ...and here.
8696 * gimple.h (gimple_call_nonnull_result_p): New.
8697 (gimple_call_nonnull_arg): New.
8699 2018-07-11 Richard Earnshaw <rearnsha@arm.com>
8701 * config/arm/arm-cpus.in: Move information from fpu field of each
8702 cpu definition to the isa field.
8703 * config/arm/parsecpu.awk (fpu): Delete match rule.
8704 (gen_comm_data): Don't add bits from the CPU's FPU entry.
8706 2018-07-11 Richard Biener <rguenther@suse.de>
8709 * dwarf2out.c (init_sections_and_labels): Use
8710 output_asm_line_debug_info consistently.
8711 (dwarf2out_early_finish): Likewise.
8712 (dwarf2out_finish): Remove DW_AT_stmt_list from early generated
8715 2018-07-11 Richard Biener <rguenther@suse.de>
8717 * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely_1):
8718 Rework father_bb setting in a way to avoid propagating constants
8719 multiple times on a loop body.
8721 2018-07-10 Mark Wielaard <mark@klomp.org>
8724 * dwarf2out.c (output_macinfo_op): Fix dwarf_FORM typo in gcc_assert.
8726 2018-07-10 Richard Biener <rguenther@suse.de>
8728 * hash-map.h (hash_map::iterator::operator*): Return
8729 references to key and value.
8731 2018-07-10 Jakub Jelinek <jakub@redhat.com>
8734 * gimplify.c (find_combined_omp_for): Add DATA argument, in addition
8735 to finding the inner OMP_FOR/OMP_SIMD stmt find non-trivial wrappers,
8736 BLOCKs with BLOCK_VARs, OMP_PARALLEL in between, OMP_FOR in between.
8737 (gimplify_omp_for): For composite loops, move outer
8738 OMP_{DISTRIBUTE,TASKLOOP,FOR,PARALLEL} right around innermost
8739 OMP_FOR/OMP_SIMD if there are any non-trivial wrappers. For class
8740 iterators add any needed clauses. Allow OMP_FOR_ORIG_DECLS to contain
8741 TREE_LIST for both the original class iterator and the "last" helper
8742 var. Gimplify OMP_FOR_PRE_BODY before the outermost composite
8743 loop, remember has_decl_expr from outer composite loops for the
8744 innermost OMP_SIMD in TREE_PRIVATE bit on OMP_FOR_INIT.
8746 2018-07-09 Martin Sebor <msebor@redhat.com>
8750 * builtins.c (c_strlen): Avoid out-of-bounds warnings when
8751 accessing implicitly initialized array elements.
8752 * expr.c (string_constant): Handle string initializers of
8753 character arrays within aggregates.
8754 * gimple-fold.c (fold_array_ctor_reference): Add argument.
8755 Store element offset. As a special case, handle zero size.
8756 (fold_nonarray_ctor_reference): Same.
8757 (fold_ctor_reference): Add argument. Store subobject offset.
8758 * gimple-fold.h (fold_ctor_reference): Add argument.
8760 2018-07-09 Paul Koning <ni1d@arrl.net>
8762 * config/pdp11/pdp11.c (pdp11_addr_cost): New function.
8763 (pdp11_insn_cost): New function.
8764 (pdp11_md_asm_adjust): New function.
8765 (TARGET_INVALID_WITHIN_DOLOOP): Define.
8766 (pdp11_rtx_costs): Update to match machine better.
8767 (output_addr_const_pdp11): Correct format mismatch warnings.
8768 * config/pdp11/pdp11.h (SLOW_BYTE_ACCESS): Correct definition.
8769 * config/pdp11/pdp11.md: General change to add base_cost and/or
8770 length attributes for use by new pdp11_insn_cost function.
8771 (MIN_BRANCH): Correct definition.
8773 (doloop_end): Use standard pattern name for looping pattern.
8774 (doloop_end_nocc): New.
8775 (movsf): Add another constraint alternative.
8776 (zero_extendqihi2): Add constraint alternatives for not in place
8778 (zero_extendhisi2): Remove.
8779 (shift patterns): Add CC handling variants.
8783 (define_peephole2): New peephole to recognize mov that sets CC for
8786 2018-07-09 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
8788 * config/sparc/sparc.c (sparc_fold_builtin) <SPARC_BUILTIN_PDIST,
8789 SPARC_BUILTIN_PDISTN>: Adapt for signature change of wi::neg,
8792 2018-07-09 Jakub Jelinek <jakub@redhat.com>
8795 * real.c (real_nextafter): Return true if result is denormal.
8797 2018-07-09 Martin Liska <mliska@suse.cz>
8799 * common.opt: Add back wrongly removed attribute.
8801 2018-07-09 Richard Biener <rguenther@suse.de>
8804 * dwarf2out.c (gen_block_die): For an early generated DIE
8805 always output high/low PC attributes.
8807 2018-07-09 Tom de Vries <tdevries@suse.de>
8809 * cfgexpand.c (expand_debug_source_expr): Handle VAR_DECL.
8810 * tree-inline.c (remap_ssa_name): Handle default def ssa_name mapping
8811 onto VAR_DECL with abstract origin.
8813 2018-07-07 Jim Wilson <jimw@sifive.com>
8815 * config/riscv/riscv.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): New.
8817 2018-07-07 Tom de Vries <tdevries@suse.de>
8819 * tree-dfa.c (dump_enumerated_decls): Handle cfun->cfg == NULL.
8821 2018-07-07 Aldy Hernandez <aldyh@redhat.com>
8823 * tree-vrp.c (vrp_int_const_binop): Change overflow type to
8825 (combine_bound): Use wide-int overflow calculation instead of
8827 * calls.c (maybe_warn_alloc_args_overflow): Change overflow type to
8829 * fold-const.c (int_const_binop_2): Same.
8830 (extract_muldiv_1): Same.
8831 (fold_div_compare): Same.
8832 (fold_abs_const): Same.
8834 * poly-int.h (add): Same.
8838 * predict.c (predict_iv_comparison): Same.
8839 * profile-count.c (slow_safe_scale_64bit): Same.
8840 * simplify-rtx.c (simplify_const_binary_operation): Same.
8841 * tree-chrec.c (tree_fold_binomial): Same.
8842 * tree-data-ref.c (split_constant_offset_1): Same.
8843 * tree-if-conv.c (idx_within_array_bound): Same.
8844 * tree-scalar-evolution.c (iv_can_overflow_p): Same.
8845 * tree-ssa-phiopt.c (minmax_replacement): Same.
8846 * tree-vect-loop.c (is_nonwrapping_integer_induction): Same.
8847 * tree-vect-stmts.c (vect_truncate_gather_scatter_offset): Same.
8848 * vr-values.c (vr_values::adjust_range_with_scev): Same.
8849 * wide-int.cc (wi::add_large): Same.
8850 (wi::mul_internal): Same.
8851 (wi::sub_large): Same.
8852 (wi::divmod_internal): Same.
8853 * wide-int.h: Change overflow type to overflow_type for neg, add,
8854 mul, smul, umul, div_trunc, div_floor, div_ceil, div_round,
8855 mod_trunc, mod_ceil, mod_round, add_large, sub_large,
8856 mul_internal, divmod_internal.
8857 (overflow_type): New enum.
8858 (accumulate_overflow): New.
8860 2018-07-06 Kugan Vivekanandarajah <kugan.vivekanandarajah@linaro.org>
8862 * tree-ssa-phiopt.c (cond_removal_in_popcount_pattern): New.
8863 (tree_ssa_phiopt_worker): Call cond_removal_in_popcount_pattern.
8865 2018-07-06 Kugan Vivekanandarajah <kuganv@linaro.org>
8867 * tree-ssa-loop-niter.c (number_of_iterations_popcount): If popcount
8868 argument is checked for zero before entering loop, avoid checking again.
8870 2018-07-06 Kugan Vivekanandarajah <kuganv@linaro.org>
8872 * gimplify.h (generic_expr_could_trap_p): Set as global function.
8873 * gimplify.h (generic_expr_could_trap_p): Likwise.
8874 * tree-scalar-evolution.c (expression_expensive_p): Handle COND_EXPR.
8876 2018-07-06 Jakub Jelinek <jakub@redhat.com>
8878 PR tree-optimization/86401
8879 * fold-const.c (fold_binary_loc) <case BIT_AND_EXPR>: Move the
8880 ((A & N) + B) & M -> (A + B) & M etc. optimization into ...
8881 (fold_bit_and_mask): ... here. New helper function for match.pd.
8882 * fold-const.h (fold_bit_and_mask): Declare.
8883 * match.pd (((A & N) + B) & M -> (A + B) & M): New optimization.
8885 2018-07-06 Peter Bergner <bergner@linux.ibm.com>
8888 * target.def (translate_mode_attribute): New hook.
8889 * targhooks.h (default_translate_mode_attribute): Declare.
8890 * targhooks.c (default_translate_mode_attribute): New function.
8891 * doc/tm.texi.in (TARGET_TRANSLATE_MODE_ATTRIBUTE): New hook.
8892 * doc/tm.texi: Regenerate.
8893 * config/rs6000/rs6000.c (TARGET_TRANSLATE_MODE_ATTRIBUTE): Define.
8894 (rs6000_translate_mode_attribute): New function.
8896 2018-07-06 Paul Koning <ni1d@arrl.net>
8898 * doc/md.texi (define_split): Document DONE and FAIL.
8899 (define_peephole2): Ditto.
8901 2018-07-05 Jeff Law <law@redhat.com>
8903 PR tree-optimization/86010
8904 * tree-ssa-dse.c (compute_trims): More aggressively trim at
8905 both the head and tail of mem* and str* calls.
8907 2018-07-05 Jim Wilson <jimw@sifive.com>
8909 * config.gcc (riscv*-*-*): When setting xlen, handle riscv-*.
8911 2018-07-05 Indu Bhagat <indu.bhagat@oracle.com>
8913 * config/aarch64/aarch64-simd.md: correct flags text for
8914 MIN_EXPR replacement.
8916 2018-07-05 James Clarke <jrtc27@jrtc27.com>
8918 * configure: Regenerated.
8920 2018-07-05 Carl Love <cel@us.ibm.com>
8922 * config/rs6000/rs6000-c.c: Map ALTIVEC_BUILTIN_VEC_UNPACKH for
8923 float argument to VSX_BUILTIN_DOUBLEH_V4SF.
8924 Map ALTIVEC_BUILTIN_VEC_UNPACKL for float argument to
8925 VSX_BUILTIN_DOUBLEL_V4SF.
8927 2018-07-05 Martin Sebor <msebor@redhat.com>
8930 * tree-ssa-strlen.c (maybe_set_strlen_range): Use type size rather
8931 than its domain to compute its the upper bound of a char array.
8933 2018-07-05 Nathan Sidwell <nathan@acm.org>
8935 Replace NO_IMPLICIT_EXTERN_C with SYSTEM_IMPLICIT_EXTERN_C.
8936 * doc/cpp.texi: Update comment.
8937 * doc/tm.texi: Rebuilt.
8938 * doc/tm.texi.in (NO_IMPLICIT_EXTERN_C): Replace with ...
8939 (SYSTEM_IMPLICIT_EXTERN_C): ... this, opposite sense.
8940 * doc/extend.texi (Backwards Compatibility): Clarify it is system
8941 headers affected by extern "C".
8942 * system.h: Poison NO_IMPLICIT_EXTERN_C.
8943 * config/alpha/alpha.h, config/arm/uclinux-elf.h,
8944 config/bfin/elf.h, config/cris/cris.h, config/darwin.h,
8945 config/dragonfly.h, config/freebsd.h, config/gnu-user.h,
8946 config/i386/cygming.h, config/i386/djgpp.h, config/i386/nto.h,
8947 config/ia64/hpux.h, config/lm32/lm32.h, config/lm32/uclinux-elf.h,
8948 config/lynx.h, config/mips/elf.h, config/mmix/mmix.h,
8949 config/netbsd.h, config/pa/pa-hpux.h, config/powerpcspe/sysv4.h,
8950 config/riscv/elf.h, config/rs6000/sysv4.h, config/rtems.h,
8951 config/s390/tpf.h, config/sh/newlib.h, config/sol2.h,
8952 config/sparc/openbsd64.h, config/sparc/sp-elf.h,
8953 config/sparc/sp64-elf.h, config/spu/spu.h,
8954 config/stormy16/stormy16.h, config/v850/v850.h,
8955 config/visium/visium.h, config/vx-common.h, config/xtensa/elf.h: Don't
8956 define NO_IMPLICIT_EXTERN_C.
8957 * config/rs6000/aix.h: Set SYSTEM_IMPLICIT_EXTERN_C.
8959 2018-07-05 Tamar Christina <tamar.christina@arm.com>
8962 * config/arm/arm.c (arm_can_change_mode_class): Use GET_MODE_UNIT_SIZE
8963 instead of GET_MODE_SIZE when comparing Units.
8965 2018-07-05 Tamar Christina <tamar.christina@arm.com>
8968 * rtlanal.c (set_noop_p): Constrain on mode change,
8969 include hard-reg-set.h
8971 2018-07-05 Tamar Christina <tamar.christina@arm.com>
8973 * config/aarch64/aarch64.c (aarch64_expand_movmem): Fix mode size.
8975 2018-07-05 Jakub Jelinek <jakub@redhat.com>
8978 2018-07-04 Maxim Ostapenko <m.ostapenko@samsung.com>
8981 * config/gnu-user.h (LIBASAN_EARLY_SPEC): Pass -lstdc++ for static
8983 * gcc.c: Do not pass LIBUBSAN_SPEC if ASan is enabled with UBSan.
8985 2018-07-04 Maxim Ostapenko <m.ostapenko@samsung.com>
8988 * config/gnu-user.h (LIBASAN_EARLY_SPEC): Pass -lstdc++ for static
8990 * gcc.c: Do not pass LIBUBSAN_SPEC if ASan is enabled with UBSan.
8992 2018-07-04 Eric Botcazou <ebotcazou@adacore.com>
8995 * expmed.c (choose_multiplier): Fix incorrect comparison with mask.
8997 2018-07-04 Aldy Hernandez <aldyh@redhat.com>
8999 * tree-vrp.c (extract_range_from_binary_expr_1): Initialize
9000 neg_*_op* variables.
9002 2018-07-04 Martin Liska <mliska@suse.cz>
9004 * tree-switch-conversion.c: Define
9005 max_ratio_for_speed and max_ratio_for_size constants.
9007 2018-07-04 Denys Vlasenko <dvlasenk@redhat.com>
9008 Martin Liska <mliska@suse.cz>
9013 * common.opt: Rename align options with 'str_' prefix.
9014 * common/config/i386/i386-common.c (set_malign_value): New
9016 (ix86_handle_option): Use it to set -falign-* options/
9017 * config/aarch64/aarch64-protos.h (struct tune_params): Change
9018 type from int to string.
9019 * config/aarch64/aarch64.c: Update default values from int
9021 * config/alpha/alpha.c (alpha_override_options_after_change):
9023 * config/arm/arm.c (arm_override_options_after_change_1): Likewise.
9024 * config/i386/dragonfly.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
9025 max skip conditionally.
9026 * config/i386/freebsd.h (SUBALIGN_LOG): New.
9027 (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
9028 max skip conditionally.
9029 * config/i386/gas.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
9030 max skip conditionally.
9031 * config/i386/gnu-user.h (SUBALIGN_LOG): New.
9032 (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
9033 max skip conditionally.
9034 * config/i386/i386.c (struct ptt): Change type from int to
9036 (ix86_default_align): Set default values.
9037 * config/i386/i386.h (ASM_OUTPUT_MAX_SKIP_PAD): Print
9038 max skip conditionally.
9039 * config/i386/iamcu.h (SUBALIGN_LOG): New.
9040 (ASM_OUTPUT_MAX_SKIP_ALIGN):
9041 * config/i386/lynx.h (ASM_OUTPUT_MAX_SKIP_ALIGN):
9042 * config/i386/netbsd-elf.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
9043 max skip conditionally.
9044 * config/i386/openbsdelf.h (SUBALIGN_LOG): New.
9045 (ASM_OUTPUT_MAX_SKIP_ALIGN) Print max skip conditionally.:
9046 * config/i386/x86-64.h (SUBALIGN_LOG): New.
9047 (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
9048 max skip conditionally.
9049 (ASM_OUTPUT_MAX_SKIP_PAD): Likewise.
9050 * config/ia64/ia64.c (ia64_option_override): Set default values
9051 for alignment options.
9052 * config/m68k/m68k.c: Handle new str_align_* options.
9053 * config/mips/mips.c (mips_set_compression_mode): Change
9055 (mips_option_override): Set default values for options.
9056 * config/powerpcspe/powerpcspe.c (rs6000_option_override_internal):
9058 * config/rs6000/rs6000.c (rs6000_option_override_internal):
9060 * config/rx/rx.c (rx_option_override): Likewise.
9061 * config/rx/rx.h (JUMP_ALIGN): Use align_jumps_log.
9062 (LABEL_ALIGN): Use align_labels_log.
9063 (LOOP_ALIGN): Use align_loops_align.
9064 * config/s390/s390.c (s390_asm_output_function_label): Use new
9066 * config/sh/sh.c (sh_override_options_after_change):
9067 Change type of constants.
9068 * config/spu/spu.c (spu_sched_init): Likewise.
9069 * config/sparc/sparc.c (sparc_option_override): Set default
9071 * config/visium/visium.c (visium_option_override): Likewise.
9072 * config/visium/visium.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Do not
9073 emit p2align format with last argument if it's not needed.
9074 * doc/invoke.texi: Document extended format of -falign-*.
9075 * final.c: Use align_labels alignment.
9076 * flags.h (struct target_flag_state): Change type to use
9078 (struct align_flags_tuple): New.
9079 (struct align_flags): Likewise.
9080 (align_loops_log): Redefine macro to use new types.
9081 (align_loops_max_skip): Redefine macro to use new types.
9082 (align_jumps_log): Redefine macro to use new types.
9083 (align_jumps_max_skip): Redefine macro to use new types.
9084 (align_labels_log): Redefine macro to use new types.
9085 (align_labels_max_skip): Redefine macro to use new types.
9086 (align_functions_log): Redefine macro to use new types.
9087 (align_loops): Redefine macro to use new types.
9088 (align_jumps): Redefine macro to use new types.
9089 (align_labels): Redefine macro to use new types.
9090 (align_functions): Redefine macro to use new types.
9091 (align_functions_max_skip): Redefine macro to use new types.
9092 (align_loops_value): New macro.
9093 (align_jumps_value): New macro.
9094 (align_labels_value): New macro.
9095 (align_functions_value): New macro.
9096 * function.c (invoke_set_current_function_hook): Propagate
9097 alignment values from flags to global variables default in
9099 * ipa-icf.c (sem_function::equals_wpa): Use
9100 cl_optimization_option_eq instead of memcmp.
9101 * lto-streamer.h (cl_optimization_stream_out): Support streaming
9103 (cl_optimization_stream_in): Likewise.
9104 * optc-save-gen.awk: Support strings in cl_optimization.
9105 * opth-gen.awk: Likewise.
9106 * opts.c (finish_options): Remove error checking of invalid
9108 (MAX_CODE_ALIGN): Remove.
9109 (MAX_CODE_ALIGN_VALUE): Likewise.
9110 (parse_and_check_align_values): New function.
9111 (check_alignment_argument): Likewise.
9112 (common_handle_option): Use check_alignment_argument.
9113 * opts.h (parse_and_check_align_values): Declare.
9114 * toplev.c (init_alignments): Remove.
9115 (read_log_maxskip): New.
9116 (parse_N_M): Likewise.
9117 (parse_alignment_opts): Likewise.
9118 (backend_init_target): Remove usage of init_alignments.
9119 * toplev.h (parse_alignment_opts): Declare.
9120 * tree-streamer-in.c (streamer_read_tree_bitfields): Add new
9122 * tree-streamer-out.c (streamer_write_tree_bitfields): Likewise.
9123 * tree.c (cl_option_hasher::equal): New.
9124 * varasm.c: Use new global macros.
9126 2018-07-04 Denys Vlasenko <dvlasenk@redhat.com>
9128 * config/i386/dragonfly.h: (ASM_OUTPUT_MAX_SKIP_ALIGN):
9129 Use a simpler align directive also if MAXSKIP = ALIGN-1.
9130 * config/i386/gas.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
9131 * config/i386/lynx.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
9132 * config/i386/netbsd-elf.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
9133 * config/i386/i386.h (ASM_OUTPUT_MAX_SKIP_PAD): Likewise.
9134 * config/i386/freebsd.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Remove "If N
9135 is large, do at least 8 byte alignment" code. Add SUBALIGN_LOG
9136 define. Use a simpler align directive also if MAXSKIP = ALIGN-1.
9137 * config/i386/gnu-user.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
9138 * config/i386/iamcu.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
9139 * config/i386/openbsdelf.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
9140 * config/i386/x86-64.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
9142 2018-07-04 Martin Liska <mliska@suse.cz>
9143 Jonathan Wakely <jwakely@redhat.com>
9145 * coverage.c: Use correct type.
9146 * doc/invoke.texi: Language correction.
9148 2018-07-03 H.J. Lu <hongjiu.lu@intel.com>
9151 * config/i386/i386.c (rest_of_insert_endbranch): Also generate
9152 ENDBRANCH for non-tail call which may return via indirect branch.
9153 * doc/extend.texi: Document indirect_return attribute.
9155 2018-07-03 Martin Sebor <msebor@redhat.com>
9157 PR tree-optimization/86274
9158 * gimple-ssa-sprintf.c (fmtresult::type_max_digits): Verify
9160 (format_floating): Correct handling of infinities and NaNs.
9162 2018-07-03 Martin Sebor <msebor@redhat.com>
9164 * print-tree.c (print_real_cst): New function.
9165 (print_node_brief): Call it.
9166 (print_node): Ditto.
9168 2018-07-03 Jeff Law <law@redhat.com>
9170 * config/h8300/h8300.md (logical<mode>3_sn, logical<mode>3): Merge
9171 into a single pattern.
9173 * config/h8300/h8300.md (ors code_iterator): New.
9174 (bsetqi_msx, bnotqi_msx patterns and splitters): Consolidate into
9175 a single pattern and single splitter.
9176 (bsethi_msx, bnothi_msx patterns): Consolidate into a single pattern.
9177 (iorqi3_1, xorqi3_1): Likewise.
9178 (iorqi3, xorqi3 expanders): Similarly.
9180 * config/h8300/h8300.md (movmd_internal_normal): Consolidated with
9181 (movmd_internal) into a single pattern using the P mode iterator.
9182 (movmd splitters): Similarly.
9183 (stpcpy_internal_normal, stpcpy_internal): Similarly for thes patterns.
9184 (movsd splitters): Similarly.
9186 * config/h8300/h8300.c (h8300_insn_length_from_table): Consolidate
9187 ADDB, ADDW and ADDL into a single ADD attribute which selects the
9188 right table based on the size of the operand.
9189 * config/h8300/h8300.md (length_table): Corresponding changes. All
9190 references to "addb", "addw" and "addl" changed to "add".
9191 (btst patterns): Merge two variants into a single pattern.
9192 (tstqi, tsthi): Likewise.
9193 (addhi3_incdec, addsi3_incdec): Likewise.
9194 (subhi3_h8300hs, subsi3_h8300hs): Likewise.
9195 (mulhi3, mulsi3): Likewise.
9196 (udivhi3, udivsi3): Likewise.
9197 (divhi3, divsi3): Likewise.
9198 (andorqi3, andorhi3, andorsi3): Likewise.
9200 2018-07-03 Uros Bizjak <ubizjak@gmail.com>
9203 * config/i386/sse.md (uavg<mode>3_ceil): New expander.
9204 (<sse2_avx2>_uavg<mode>3<mask_name>): Simplify expander.
9206 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
9208 PR tree-optimization/85694
9209 * config/aarch64/iterators.md (HADD, RHADD): New int iterators.
9210 (u): Handle UNSPEC_SHADD, UNSPEC_UHADD, UNSPEC_SRHADD and
9212 * config/aarch64/aarch64-simd.md (<u>avg<mode>3_floor)
9213 (<u>avg<mode>3_ceil): New patterns.
9215 2018-07-03 David Malcolm <dmalcolm@redhat.com>
9217 * gcc.dg/vect/slp-perm-1.c: Remove "note: " prefix from
9218 scan-tree-dump directive.
9219 * gcc.dg/vect/slp-perm-2.c: Likewise.
9220 * gcc.dg/vect/slp-perm-3.c: Likewise.
9221 * gcc.dg/vect/slp-perm-5.c: Likewise.
9222 * gcc.dg/vect/slp-perm-6.c: Likewise.
9223 * gcc.dg/vect/slp-perm-7.c: Likewise.
9224 * gcc.dg/vect/slp-perm-8.c: Likewise.
9226 2018-07-03 Marek Polacek <polacek@redhat.com>
9229 * gimple-fold.c (size_must_be_zero_p): Check the type of the size.
9231 2018-07-03 Richard Biener <rguenther@suse.de>
9234 * tree-ssa-structalias.c (find_func_clobbers): Properly
9235 handle indirect calls.
9237 2018-07-03 Jeff Law <law@redhat.com>
9239 * config/h8300/h8300.md (HSI, QHSI, QHSIF): New mode iterators.
9240 (shifts): New code iterator.
9241 (movqi, movhi, movsi, movsf expanders): Consolidate into a single
9242 expander. Fix HImode handling on H8/SX.
9243 (addqi3, addhi3, addsi3 expanders): Consolidate into a single expander.
9244 (subqi3, subhi3, subsi3 expanders): Likewise.
9245 (andqi3, andhi3, andsi3 expanders): Likewise.
9246 (iorqi3, iorhi3, iorsi3 expanders): Likewise.
9247 (xorqi3, xorhi3, xorsi3 expanders): Likewise.
9248 (negqi2, neghi2, negsi2, negsf2 expanders): Likewise.
9249 (one_cmplqi2, one_cmplhi2, one_cmplsi2): Likewise.
9250 (zero_extendqihi2, zero_extendqisi2): Likewise.
9251 (extendqihi2, extendqisi2): Likewise.
9252 (rotlqi3, rotlhi3, rotlsi3): Likewise.
9253 (neghi2_h8300, negsi2_h8300): Likewise for these patterns.
9254 (rotlqi3_1, rotlhi3_1): Likewise.
9255 (logicalhi3_sn, logicalsi3_sn): Likewise.
9256 (logicalhi3, logicalsi3): Likewise.
9258 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
9260 * tree-vect-patterns.c (vect_recog_rotate_pattern)
9261 (vect_recog_vector_vector_shift_pattern, vect_recog_divmod_pattern)
9262 (vect_recog_mixed_size_cond_pattern, adjust_bool_pattern_cast)
9263 (adjust_bool_pattern, vect_recog_bool_pattern): Pass the vector
9264 type to append_pattern_def_seq instead of creating a stmt_vec_info
9266 (build_mask_conversion): Likewise. Remove vinfo argument.
9267 (vect_add_conversion_to_patterm): Likewise, renaming to...
9268 (vect_add_conversion_to_pattern): ...this.
9269 (vect_recog_mask_conversion_pattern): Update call to
9270 build_mask_conversion. Pass the vector type to
9271 append_pattern_def_seq here too.
9272 (vect_recog_gather_scatter_pattern): Update call to
9273 vect_add_conversion_to_pattern.
9275 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
9277 * tree-vect-patterns.c (new_pattern_def_seq): Delete.
9278 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern)
9279 (vect_recog_widen_op_pattern, vect_recog_over_widening_pattern)
9280 (vect_recog_rotate_pattern, vect_synth_mult_by_constant): Don't set
9281 STMT_VINFO_PATTERN_DEF_SEQ to null here.
9282 (vect_recog_pow_pattern, vect_recog_vector_vector_shift_pattern)
9283 (vect_recog_mixed_size_cond_pattern, vect_recog_bool_pattern): Use
9284 append_pattern_def_seq instead of new_pattern_def_seq.
9285 (vect_recog_divmod_pattern): Do both of the above.
9286 (vect_pattern_recog_1): Assert that STMT_VINO_PATTERN_DEF_SEQ
9289 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
9291 * tree-vect-patterns.c (vect_recog_dot_prod_pattern):
9292 (vect_recog_sad_pattern, vect_recog_widen_op_pattern)
9293 (vect_recog_widen_mult_pattern, vect_recog_pow_pattern):
9294 (vect_recog_widen_sum_pattern, vect_recog_over_widening_pattern)
9295 (vect_recog_average_pattern, vect_recog_cast_forwprop_pattern)
9296 (vect_recog_widen_shift_pattern, vect_recog_rotate_pattern)
9297 (vect_recog_vector_vector_shift_pattern, vect_synth_mult_by_constant)
9298 (vect_recog_mult_pattern, vect_recog_divmod_pattern)
9299 (vect_recog_mixed_size_cond_pattern, vect_recog_bool_pattern)
9300 (vect_recog_mask_conversion_pattern): Replace vec<gimple *>
9301 parameter with a single stmt_vec_info.
9302 (vect_recog_func_ptr): Likewise.
9303 (vect_recog_gather_scatter_pattern): Likewise, folding in...
9304 (vect_try_gather_scatter_pattern): ...this.
9305 (vect_pattern_recog_1): Remove stmts_to_replace and just pass
9306 the stmt_vec_info of the statement to be matched. Don't clear
9307 STMT_VINFO_RELATED_STMT.
9308 (vect_pattern_recog): Update call accordingly.
9310 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
9312 PR tree-optimization/85694
9313 * doc/md.texi (avgM3_floor, uavgM3_floor, avgM3_ceil)
9314 (uavgM3_ceil): Document new optabs.
9315 * doc/sourcebuild.texi (vect_avg_qi): Document new target selector.
9316 * internal-fn.def (IFN_AVG_FLOOR, IFN_AVG_CEIL): New internal
9318 * optabs.def (savg_floor_optab, uavg_floor_optab, savg_ceil_optab)
9319 (savg_ceil_optab): New optabs.
9320 * tree-vect-patterns.c (vect_recog_average_pattern): New function.
9321 (vect_vect_recog_func_ptrs): Add it.
9322 * tree-vect-stmts.c (vectorizable_call): Get the type of the zero
9323 constant directly from the associated lhs.
9325 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
9327 * tree-vect-patterns.c (vect_split_statement): New function.
9328 (vect_convert_input): Use it to try to split an existing cast.
9330 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
9332 * poly-int.h (print_hex): New function.
9333 * dumpfile.h (dump_dec, dump_hex): Declare.
9334 * dumpfile.c (dump_dec, dump_hex): New poly_wide_int functions.
9335 * tree-vectorizer.h (_stmt_vec_info): Add min_output_precision,
9336 min_input_precision, operation_precision and operation_sign.
9337 * tree-vect-patterns.c (vect_get_range_info): New function.
9338 (vect_same_loop_or_bb_p, vect_single_imm_use)
9339 (vect_operation_fits_smaller_type): Delete.
9340 (vect_look_through_possible_promotion): Add an optional
9341 single_use_p parameter.
9342 (vect_recog_over_widening_pattern): Rewrite to use new
9343 stmt_vec_info infomration. Handle one operation at a time.
9344 (vect_recog_cast_forwprop_pattern, vect_narrowable_type_p)
9345 (vect_truncatable_operation_p, vect_set_operation_type)
9346 (vect_set_min_input_precision): New functions.
9347 (vect_determine_min_output_precision_1): Likewise.
9348 (vect_determine_min_output_precision): Likewise.
9349 (vect_determine_precisions_from_range): Likewise.
9350 (vect_determine_precisions_from_users): Likewise.
9351 (vect_determine_stmt_precisions, vect_determine_precisions): Likewise.
9352 (vect_vect_recog_func_ptrs): Put over_widening first.
9354 (vect_pattern_recog): Call vect_determine_precisions.
9356 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
9358 * tree-vect-patterns.c (vect_mark_pattern_stmts): Remove pattern
9359 statements that have been replaced by further pattern statements.
9360 (vect_pattern_recog_1): Clear STMT_VINFO_PATTERN_DEF_SEQ on failure.
9362 2018-07-03 Richard Biener <rguenther@suse.de>
9364 * tree-vect-stmts.c (vect_is_simple_use): Consolidate dumping,
9365 always set *dt. Dump vectype in vectype overload.
9366 * dumpfile.h (dump_gimple_expr): New function.
9367 (dump_gimple_expr_loc): Likewise.
9368 * dumpfile.c (dump_gimple_expr): New function.
9369 (dump_gimple_expr_loc): Likewise.
9371 2018-07-02 Jeff Law <law@redhat.com>
9373 * config/h8300/h8300.md (movqi_h8300, movqi_h8300hs): Consolidate
9374 the H8/300, H8/300H and H8/S variants into a single pattern.
9375 (movhi_h8300, movqi_h8300hs): Similarly.
9376 (pushqi_h8300hs, pushhi_h8300hs): Consolidate into a single pattern.
9377 (QHI mode iterator): New.
9379 * config/h8300/h8300.md: Remove trailing whitespace.
9381 2018-07-02 Jim Wilson <jimw@sifive.com>
9383 * config/riscv/riscv.c (riscv_expand_epilogue): Use emit_jump_insn
9384 instead of emit_insn for interrupt returns.
9385 * config/riscv/riscv.md (riscv_met): Add (return) to rtl.
9386 (riscv_sret, riscv_uret): Likewise.
9388 2018-07-02 David Malcolm <dmalcolm@redhat.com>
9390 * pretty-print.c (selftest::test_pp_format): Move save and restore
9391 of quotes to class auto_fix_quotes, and add an instance.
9392 * selftest.c: Include "intl.h".
9393 (selftest::auto_fix_quotes::auto_fix_quotes): New ctor.
9394 (selftest::auto_fix_quotes::~auto_fix_quotes): New dtor.
9395 * selftest.h (selftest::auto_fix_quotes): New class.
9397 2018-07-02 Richard Henderson <richard.henderson@linaro.org>
9399 * config/aarch64/aarch64-protos.h, config/aarch64/aarch64.c
9400 (aarch64_sve_prepare_conditional_op): Remove.
9401 * config/aarch64/aarch64-sve.md (cond_<SVE_INT_BINARY><SVE_I>):
9402 Allow aarch64_simd_reg_or_zero as select operand; remove
9403 the aarch64_sve_prepare_conditional_op call.
9404 (cond_<SVE_INT_BINARY_SD><SVE_SDI>): Likewise.
9405 (cond_<SVE_COND_FP_BINARY><SVE_F>): Likewise.
9406 (*cond_<SVE_INT_BINARY><SVE_I>_z): New pattern.
9407 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_z): New pattern.
9408 (*cond_<SVE_COND_FP_BINARY><SVE_F>_z): New pattern.
9409 (*cond_<SVE_INT_BINARY><SVE_I>_any): New pattern.
9410 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_any): New pattern.
9411 (*cond_<SVE_COND_FP_BINARY><SVE_F>_any): New pattern
9412 and a splitters to match all of the *_any patterns.
9413 * config/aarch64/predicates.md (aarch64_sve_any_binary_operator): New.
9415 * config/aarch64/iterators.md (SVE_INT_BINARY_REV): Remove.
9416 (SVE_COND_FP_BINARY_REV): Remove.
9417 (sve_int_op_rev, sve_fp_op_rev): New.
9418 * config/aarch64/aarch64-sve.md (*cond_<SVE_INT_BINARY><SVE_I>_0): New.
9419 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_0): New.
9420 (*cond_<SVE_COND_FP_BINARY><SVE_F>_0): New.
9421 (*cond_<SVE_INT_BINARY><SVE_I>_2): Rename, add movprfx alternative.
9422 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_2): Similarly.
9423 (*cond_<SVE_COND_FP_BINARY><SVE_F>_2): Similarly.
9424 (*cond_<SVE_INT_BINARY><SVE_I>_3): Similarly; use sve_int_op_rev.
9425 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_3): Similarly.
9426 (*cond_<SVE_COND_FP_BINARY><SVE_F>_3): Similarly; use sve_fp_op_rev.
9428 * config/aarch64/aarch64-sve.md (cond_<SVE_COND_FP_BINARY><SVE_F>):
9429 Remove match_dup 1 from the inner unspec.
9430 (*cond_<SVE_COND_FP_BINARY><SVE_F>): Likewise.
9432 * config/aarch64/aarch64.md (movprfx): New attr.
9433 (length): Default movprfx to 8.
9434 * config/aarch64/aarch64-sve.md (*mul<SVE_I>3): Add movprfx alt.
9435 (*madd<SVE_I>, *msub<SVE_I): Likewise.
9436 (*<su>mul<SVE_I>3_highpart): Likewise.
9437 (*<SVE_INT_BINARY_SD><SVE_SDI>3): Likewise.
9438 (*v<ASHIFT><SVE_I>3): Likewise.
9439 (*<su><MAXMIN><SVE_I>3): Likewise.
9440 (*<su><MAXMIN><SVE_F>3): Likewise.
9441 (*fma<SVE_F>4, *fnma<SVE_F>4): Likewise.
9442 (*fms<SVE_F>4, *fnms<SVE_F>4): Likewise.
9443 (*div<SVE_F>4): Likewise.
9445 2018-07-02 Richard Sandiford <richard.sandiford@arm.com>
9447 * tree-vect-patterns.c (vect_recog_widen_shift_pattern): Fix typo
9450 2018-07-02 Richard Biener <rguenther@suse.de>
9452 PR tree-optimization/86363
9453 * tree-ssa-sccvn.c (vn_reference_lookup_3): Check the
9454 memset argument refers to a non-variable address.
9456 2018-07-02 Aldy Hernandez <aldyh@redhat.com>
9458 * tree-vrp.c (extract_range_from_binary_expr_1): Abstract a lot of the
9459 {PLUS,MINUS}_EXPR code to...
9460 (adjust_symbolic_bound): ...here,
9461 (combine_bound): ...here,
9462 (set_value_range_with_overflow): ...and here.
9464 2018-07-02 Aldy Hernandez <aldyh@redhat.com>
9466 * tree-vrp.c (extract_range_from_unary_expr): Abstract ABS_EXPR
9468 (extract_range_from_abs_expr): ...here.
9470 2018-07-02 Eric Botcazou <ebotcazou@adacore.com>
9472 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Do not overrule
9473 -fno-omit-frame-pointer when not optimizing.
9475 2018-07-02 Martin Liska <mliska@suse.cz>
9478 * ipa-pure-const.c (malloc_candidate_p): Revert usage of ::get.
9479 (propagate_nothrow): Likewise.
9481 2018-07-02 Martin Liska <mliska@suse.cz>
9484 * ipa-inline.c (early_inliner): Revert wrongly added ::get call.
9486 2018-07-02 David Malcolm <dmalcolm@redhat.com>
9488 * dumpfile.c (dump_generic_expr_loc): Undo removal of this
9489 function in r262149, changing "loc" param from source_location to
9490 const dump_location_t &.
9491 * dumpfile.h (dump_generic_expr_loc): Undo removal of this
9492 declaration, as above.
9494 2018-07-01 Paul Koning <ni1d@arrl.net>
9496 * common/config/pdp11/pdp11-common.c (pdp11_handle_option): Handle
9497 -munit-asm, -mgnu-asm, -mdec-asm.
9498 * config/pdp11/pdp11-protos.h (pdp11_gen_int_label): New.
9499 (pdp11_output_labelref): New.
9500 (pdp11_output_def): New.
9501 (pdp11_output_addr_vec_elt): New.
9502 * config/pdp11/pdp11.c: Use tab between opcode and operands. Use
9503 %# and %@ format codes.
9504 (pdp11_option_override): New.
9505 (TARGET_ASM_FILE_START_FILE_DIRECTIVE): Define.
9506 (pdp11_output_ident): New.
9507 (pdp11_asm_named_section): New.
9508 (pdp11_asm_init_sections): New.
9509 (pdp11_file_start): New.
9510 (pdp11_file_end): New.
9511 (output_ascii): Use .ascii/.asciz for -mdec-asm.
9512 (pdp11_asm_print_operand): Update %# and %$ for -mdec-asm. Add
9513 %o, like %c but octal.
9514 (pdp11_option_override): New.
9515 * config/pdp11/pdp11.h (TEXT_SECTION_ASM_OP): Update for
9517 (DATA_SECTION_ASM_OP): Ditto.
9518 (READONLY_DATA_SECTION_ASM_OP): New.
9519 (IS_ASM_LOGICAL_LINE_SEPARATOR): New.
9520 (ASM_GENERATE_INTERNAL_LABEL): Use new function.
9521 (ASM_OUTPUT_LABELREF): Ditto.
9522 (ASM_OUTPUT_DEF): Ditto.
9523 (ASM_OUTPUT_EXTERNAL): New.
9524 (ASM_OUTPUT_SOURCE_FILENAME): New.
9525 (ASM_OUTPUT_ADDR_VEC_ELT): Use new function.
9526 (ASM_OUTPUT_SKIP): Update for -mdec-asm.
9527 * config/pdp11/pdp11.md: Use tab between opcode and operands. Use
9528 %# and %@ format codes.
9529 * config/pdp11/pdp11.opt (mgnu-asm): New.
9530 (mdec-asm): Conflicts with -mgnu-asm and -munix-asm.
9531 (munix-asm): Conflicts with -mdec-asm and -mgnu-asm.
9532 * doc/invoke.txt (PDP-11 Options): Add -mgnu-asm.
9534 2018-07-01 Aldy Hernandez <aldyh@redhat.com>
9536 * tree-ssa-threadupdate.c (mark_threaded_blocks): Avoid
9537 dereferencing path[] beyond its length.
9539 (debug_all_paths): New.
9540 (rewire_first_differing_edge): New.
9541 (adjust_paths_after_duplication): New.
9542 (duplicate_thread_path): Call adjust_paths_after_duplication.
9544 (thread_through_all_blocks): Add new argument to
9545 duplicate_thread_path.
9547 2018-06-30 Jim Wilson <jimw@sifive.com>
9549 * config/riscv/predicates.md (p2m1_shift_operand): New.
9550 (high_mask_shift_operand): New.
9551 * config/riscv/riscv.md (lshrsi3_zero_extend_3+1): New combiner
9552 pattern using p2m1_shift_operand.
9553 (lshsi3_zero_extend_3+2): New combiner pattern using
9554 high_mask_shift_operand.
9556 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
9558 * tree-vect-patterns.c (vect_get_external_def_edge): New function,
9560 (vect_recog_rotate_pattern): ...here.
9561 (vect_convert_input): Try to insert casts of invariants in the
9563 * tree-vect-loop-manip.c (vect_loop_versioning): Don't require the
9564 preheader to be empty.
9566 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
9568 * tree-vect-patterns.c (append_pattern_def_seq): Take an optional
9569 vector type. If given, install it in the new statement's
9571 (vect_element_precision): New function.
9572 (vect_unpromoted_value): New struct.
9573 (vect_unpromoted_value::vect_unpromoted_value): New function.
9574 (vect_unpromoted_value::set_op): Likewise.
9575 (vect_look_through_possible_promotion): Likewise.
9576 (vect_joust_widened_integer, vect_joust_widened_type): Likewise.
9577 (vect_widened_op_tree, vect_convert_input): Likewise.
9578 (vect_convert_inputs, vect_convert_output): Likewise.
9579 (vect_recog_dot_prod_pattern): Use vect_look_through_possible_promotion
9580 to handle the optional cast of the multiplication result and
9581 vect_widened_op_tree to detect the widened multiplication itself.
9582 Do not require the input and output of promotion casts to have
9583 the same sign, but base the signedness of the operation on the
9584 input rather than the result. If the pattern includes two
9585 promotions, check that those promotions have the same sign.
9586 Do not restrict the MULT_EXPR handling to a double-width result;
9587 handle quadruple-width results and wider. Use vect_convert_inputs
9588 to convert the inputs to the common type.
9589 (vect_recog_sad_pattern): Use vect_look_through_possible_promotion
9590 to handle the optional cast of the ABS result. Also allow a sign
9591 change or a sign extension between the ABS and MINUS.
9592 Use vect_widened_op_tree to detect the widened subtraction and use
9593 vect_convert_inputs to convert the inputs to the common type.
9594 (vect_handle_widen_op_by_const): Delete.
9595 (vect_recog_widen_op_pattern): New function.
9596 (vect_recog_widen_mult_pattern): Use it.
9597 (vect_recog_widen_shift_pattern): Likewise.
9598 (vect_recog_widen_sum_pattern): Use
9599 vect_look_through_possible_promotion to handle the promoted
9602 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
9604 * gimple-iterator.c (gsi_for_stmt): Add a new overload that takes
9605 the containing gimple_seq *.
9606 * gimple-iterator.h (gsi_for_stmt): Declare it.
9607 * tree-vect-patterns.c (vect_recog_dot_prod_pattern)
9608 (vect_recog_sad_pattern, vect_recog_widen_sum_pattern)
9609 (vect_recog_widen_shift_pattern, vect_recog_rotate_pattern)
9610 (vect_recog_vector_vector_shift_pattern, vect_recog_divmod_pattern)
9611 (vect_recog_mask_conversion_pattern): Remove STMT_VINFO_IN_PATTERN_P
9613 (vect_init_pattern_stmt, vect_set_pattern_stmt): New functions,
9615 (vect_mark_pattern_stmts): ...here. Handle cases in which the
9616 statement being replaced is part of an existing pattern
9617 definition sequence, inserting the new pattern statements before
9619 (vect_pattern_recog_1): Don't return a bool. If the statement
9620 is already part of a pattern, instead apply pattern matching
9621 to the pattern definition statements. Don't clear the
9622 STMT_VINFO_RELATED_STMT if is_pattern_stmt_p.
9623 (vect_pattern_recog): Don't break after the first match;
9624 continue processing the pattern definition statements instead.
9625 Don't bail out for STMT_VINFO_IN_PATTERN_P here.
9627 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
9629 * tree-vect-patterns.c (vect_reassociating_reduction_p): New function.
9630 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern)
9631 (vect_recog_widen_sum_pattern): Use it.
9633 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
9635 * tree-vect-loop.c (vectorizable_reduction): Assert that the
9636 phi is not a pattern statement and has not been replaced by
9637 a pattern statement.
9638 * tree-vect-patterns.c (type_conversion_p): Don't check
9639 STMT_VINFO_IN_PATTERN_P.
9640 (vect_recog_vector_vector_shift_pattern): Likewise.
9641 (vect_recog_dot_prod_pattern): Expect vect_is_simple_use to return
9642 the pattern statement rather than the original statement; check
9643 directly for a WIDEN_MULT_EXPR here.
9644 * tree-vect-slp.c (vect_get_and_check_slp_defs): Expect
9645 vect_is_simple_use to return the pattern statement rather
9646 than the original statement; use is_pattern_stmt_p to check
9647 for such a pattern statement.
9648 * tree-vect-stmts.c (process_use): Expect vect_is_simple_use
9649 to return the pattern statement rather than the original statement;
9650 don't do the same transformation here.
9651 (vect_is_simple_use): If the defining statement has been replaced
9652 by a pattern statement, return the pattern statement instead.
9653 Remove the corresponding (local) transformation from the vectype
9656 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
9658 * tree-vectorizer.h (vect_is_simple_use): Move the gimple ** to the
9659 end and default to null.
9660 * tree-vect-loop.c (vect_create_epilog_for_reduction)
9661 (vectorizable_reduction): Update calls accordingly, dropping the
9662 gimple ** argument if the passed-back statement isn't needed.
9663 * tree-vect-patterns.c (vect_get_internal_def, type_conversion_p)
9664 (vect_recog_rotate_pattern): Likewise.
9665 (vect_recog_mask_conversion_pattern): Likewise.
9666 * tree-vect-slp.c (vect_get_and_check_slp_defs): Likewise.
9667 (vect_mask_constant_operand_p): Likewise.
9668 * tree-vect-stmts.c (is_simple_and_all_uses_invariant, process_use):
9669 (vect_model_simple_cost, vect_get_vec_def_for_operand): Likewise.
9670 (get_group_load_store_type, get_load_store_type): Likewise.
9671 (vect_check_load_store_mask, vect_check_store_rhs): Likewise.
9672 (vectorizable_call, vectorizable_simd_clone_call): Likewise.
9673 (vectorizable_conversion, vectorizable_assignment): Likewise.
9674 (vectorizable_shift, vectorizable_operation): Likewise.
9675 (vectorizable_store, vect_is_simple_cond): Likewise.
9676 (vectorizable_condition, vectorizable_comparison): Likewise.
9677 (get_same_sized_vectype, vect_get_mask_type_for_stmt): Likewise.
9678 (vect_is_simple_use): Rename the def_stmt argument to def_stmt_out
9679 and move it to the end. Cope with null def_stmt_outs.
9681 2018-06-30 Bernd Edlinger <bernd.edlinger@hotmail.de>
9683 * Makefile.in (FLAGS_TO_PASS): Add RANLIB_FOR_TARGET.
9685 2018-06-29 Jeff Law <law@redhat.com>
9687 * config/v850/v850.c (v850_legitimate_address_p): Handle large
9688 displacements for TARGET_V850E2V3 and newer.
9689 (TARGET_LRA_P): Remove. Defaults to LRA now.
9690 * config/v850/v850.md (sign23byte_load): Remove.
9691 (unsign23byte_load, sign23hword_load, unsign23hword_load): Likewise.
9692 (23word_load, 23byte_store, 23hword_store, 23word_store): Likewise.
9694 2018-06-29 Martin Liska <mliska@suse.cz>
9697 * coverage.c (coverage_init): Mangle full path name.
9698 * doc/invoke.texi: Document the change.
9699 * gcov-io.c (mangle_path): New.
9700 * gcov-io.h (mangle_path): Likewise.
9701 * gcov.c (mangle_name): Use mangle_path for path mangling.
9703 2018-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
9705 * config/arm/arm.c (output_move_double): Don't allow STRD instructions
9706 if starting source register is not even.
9708 2018-06-29 Martin Liska <mliska@suse.cz>
9710 PR tree-optimization/86263
9711 * tree-switch-conversion.c (switch_decision_tree::try_switch_expansion):
9712 Make edge redirection.
9714 2018-06-29 David Malcolm <dmalcolm@redhat.com>
9716 * dumpfile.c (dump_loc): Add indentation based on scope depth.
9717 (dump_scope_depth): New variable.
9718 (get_dump_scope_depth): New function.
9719 (dump_begin_scope): New function.
9720 (dump_end_scope): New function.
9721 * dumpfile.h (get_dump_scope_depth): New declaration.
9722 (dump_begin_scope): New declaration.
9723 (dump_end_scope): New declaration.
9724 (class auto_dump_scope): New class.
9725 (AUTO_DUMP_SCOPE): New macro.
9726 * tree-vectorizer.h (DUMP_VECT_SCOPE): Reimplement in terms of
9729 2018-06-29 Richard Biener <rguenther@suse.de>
9731 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Assert
9732 compute_all_dependences succeeds.
9733 * tree-vect-loop.c (vect_get_datarefs_in_loop): Fail early if we
9734 exceed --param loop-max-datarefs-for-datadeps.
9736 2018-06-29 Jakub Jelinek <jakub@redhat.com>
9738 * config/rs6000/t-rs6000: Append rs6000-modes.h to TM_H.
9740 2018-06-28 Uros Bizjak <ubizjak@gmail.com>
9743 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): Use
9744 alternative 0 in preferred_for_speed attribute.
9746 2018-06-28 Paul Koning <ni1d@arrl.net>
9748 * config/pdp11/pdp11-protos.h (pdp11_shift_length): New function.
9749 * config/pdp11/pdp11.c (pdp11_shift_length): New function.
9750 * config/pdp11/pdp11.h (ADJUST_INSN_LENGTH): Remove.
9751 * config/pdp11/pdp11.md: Correct "length" attribute calculation
9752 for shift insn patterns.
9754 2018-06-28 David Malcolm <dmalcolm@redhat.com>
9756 * cgraph.c (cgraph_node::get_body): Replace assignments to
9757 "dump_file" with calls to set_dump_file.
9758 * dumpfile.c (alt_dump_file): Make static, and group with...
9759 (alt_flags): ...this definition.
9760 (dumps_are_enabled): New variable.
9761 (refresh_dumps_are_enabled): New function.
9762 (set_dump_file): New function.
9763 (set_alt_dump_file): New function.
9764 (gcc::dump_manager::dump_start): Replace assignments to
9765 "dump_file" and "alt_dump_file" with calls to set_dump_file and
9767 (gcc::dump_manager::dump_finish): Likewise.
9768 * dumpfile.h (alt_dump_file): Delete decl.
9769 (dumps_are_enabled): New variable decl.
9770 (set_dump_file): New function decl.
9771 (dump_enabled_p): Rewrite in terms of new "dumps_are_enabled"
9773 * tree-nested.c (lower_nested_functions): Replace assignments to
9774 "dump_file" with calls to set_dump_file.
9776 2018-06-28 Eric Botcazou <ebotcazou@adacore.com>
9778 * tree-cfg.c (verify_gimple_in_cfg): Call verify_location on the
9779 goto_locus of each outgoing edge of each basic block.
9781 2018-06-28 Richard Biener <rguenther@suse.de>
9783 * dwarf2out.c (decl_scope_table): Remove.
9784 (push_decl_scope): Likewise.
9785 (pop_decl_scope): Likewise.
9786 (gen_type_die_for_member): Do not call push/pop_decl_scope.
9787 (gen_struct_or_union_type_die): Likewise.
9788 (gen_tagged_type_die): Likewise.
9789 (dwarf2out_init): Do not initialize decl_scope_table.
9790 (dwarf2out_c_finalize): Do not free it.
9792 2018-06-28 Richard Biener <rguenther@suse.de>
9794 * dwarf2out.c (gen_subprogram_die): Use is_unit_die when
9795 deciding whether to not re-use a DIE.
9797 2018-06-28 Richard Biener <rguenther@suse.de>
9799 * dwarf2out.c (gen_subprogram_die): Always re-use DIEs with an
9800 DW_AT_abstract_origin attribute.
9802 2018-06-28 Martin Liska <mliska@suse.cz>
9804 * tree-switch-conversion.c (jump_table_cluster::can_be_handled):
9805 Use newly introduced constants.
9806 * tree-switch-conversion.h (struct jump_table_cluster):
9807 Define max_ratio_for_size and max_ratio_for_speed.
9809 2018-06-28 Martin Liska <mliska@suse.cz>
9811 * tree-switch-conversion.c (jump_table_cluster::find_jump_tables):
9812 Add new checking assert to catch invalid state.
9813 (jump_table_cluster::can_be_handled): Handle single case
9815 (jump_table_cluster::is_beneficial): Bail out for such case.
9816 (bit_test_cluster::find_bit_tests):
9817 Add new checking assert to catch invalid state.
9818 (bit_test_cluster::can_be_handled): Handle single case
9820 (bit_test_cluster::is_beneficial): Bail out for such case.
9821 (switch_decision_tree::analyze_switch_statement):
9824 2018-06-28 Martin Liska <mliska@suse.cz>
9826 * common.opt: Introduce -completion option.
9827 * gcc.c (driver_handle_option): Handle it.
9828 (driver::main): Print completions if completion
9830 * opt-suggestions.c (option_proposer::get_completions):
9832 (option_proposer::suggest_completion): Likewise.
9833 (option_proposer::find_param_completions): Likewise.
9834 (verify_autocompletions): Likewise.
9835 (test_completion_valid_options): Likewise.
9836 (test_completion_valid_params): Likewise.
9837 (in_completion_p): Likewise.
9838 (empty_completion_p): Likewise.
9839 (test_completion_partial_match): Likewise.
9840 (test_completion_garbage): Likewise.
9841 (opt_proposer_c_tests): Likewise.
9842 * opt-suggestions.h: Declare new functions.
9843 * opts.c (common_handle_option): Handle OPT__completion_.
9844 * selftest-run-tests.c (selftest::run_tests): Add
9845 opt_proposer_c_tests.
9846 * selftest.c (assert_str_startswith): New.
9847 * selftest.h (assert_str_startswith): Likewise.
9848 (opt_proposer_c_tests): New.
9849 (ASSERT_STR_STARTSWITH): Likewise.
9851 2018-06-28 Martin Liska <mliska@suse.cz>
9853 * Makefile.in: Add opt-suggestions.o.
9854 * gcc-main.c: Include opt-suggestions.h.
9855 * gcc.c (driver::driver): Likewise.
9856 (driver::~driver): Remove m_option_suggestions.
9857 (driver::build_option_suggestions): Moved to option_proposer.
9858 (driver::suggest_option): Likewise.
9859 (driver::handle_unrecognized_options): Use option_proposer.
9860 * gcc.h (class driver): Add new memver m_option_proposer.
9861 * opt-suggestions.c: New file.
9862 * opt-suggestions.h: New file.
9864 2018-06-28 Martin Liska <mliska@suse.cz>
9866 * vec.h (class auto_string_vec): New (moved from auto_argvec).
9867 (auto_string_vec::~auto_string_vec): Likewise.
9869 2018-06-28 Eric Botcazou <ebotcazou@adacore.com>
9871 * tree-inline.h (struct copy_body_data): Move remapping_type_depth and
9872 prevent_decl_creation_for_types fields up and add reset_location field.
9873 * tree-inline.c (remap_gimple_stmt): Force input_location on the new
9874 statement if id->reset_location is true.
9875 (copy_edges_for_bb): Do not set goto_locus on the new edges if
9876 id->reset_location is true.
9877 (copy_phis_for_bb): Force input_location on the arguments if
9878 id->reset_location is true.
9879 (expand_call_inline): Set id->reset_location if DECL_IGNORED_P
9880 is set on the function to be inlined.
9882 2018-06-27 Stephan Bergmann <sbergman@redhat.com>
9884 * doc/invoke.texi (Debugging Options): Add -gsplit-dwarf.
9886 2018-06-27 Dimitar Dimitrov <dimitar@dinux.eu>
9888 * lra-eliminations.c (update_reg_eliminate): Mark all spanning hard
9889 registers for Pmode.
9890 * lra-lives.c (check_pseudos_live_through_calls): Mark all spanning
9891 hard registers for the clobbered pseudo.
9893 2018-06-27 Paul Koning <ni1d@arrl.net>
9895 * common/config/pdp11/pdp11-common.c (pdp11_handle_option): Handle
9896 mutually exclusive options.
9897 * config/pdp11/constraints.md (h): New constraint.
9898 (O): Update definition to match shift code generation.
9899 (D): New constraint.
9900 * config/pdp11/pdp11-modes.def (CCNZ): Define mode.
9902 * config/pdp11/pdp11-protos.h (int_no_side_effect_operand): New
9904 (output_jump): Change arguments.
9905 (pdp11_fixed_cc_regs): New function.
9906 (pdp11_cc_mode): Ditto.
9907 (pdp11_expand_shift): Ditto.
9908 (pdp11_assemble_shift): Ditto.
9909 (pdp11_small_shift): Ditto.
9910 (pdp11_branch_cost): Remove.
9911 * config/pdp11/pdp11.c (pdp11_assemble_integer): Remove comments
9913 (pdp11_register_move_cost): Update for CC registers.
9914 (pdp11_rtx_costs): Add case for LSHIFTRT.
9915 (pdp11_output_jump): Add CCNZ mode conditional branches.
9916 (notice_update_cc_on_set): Remove.
9917 (pdp11_cc_mode): New function.
9918 (simple_memory_operand): Correct pre/post decrement case.
9919 (no_side_effect_operand): New function.
9920 (pdp11_regno_reg_class): Add CC_REGS class.
9921 (pdp11_fixed_cc_regs): New function.
9922 (pdp11_small_shift): New function.
9923 (pdp11_expand_shift): New function to expand shift insns.
9924 (pdp11_assemble_shift): New function to output shifts.
9925 (pdp11_branch_cost): Remove.
9926 (pdp11_modes_tieable_p): Make QI/HI modes tieable.
9927 * config/pdp11/pdp11.h (SIZE_TYPE): Ensure 16-bit type.
9928 (WCHAR_TYPE): Ditto.
9929 (PTRDIFF_TYPE): Ditto.
9930 (ADJUST_INSN_LENGTH): New macro.
9931 (FIXED_REGISTERS): Add CC registers.
9932 (CALL_USED_REGISTERS): Ditto.
9934 (REG_CLASS_NAMES): Ditto.
9935 (REG_CLASS_CONTENTS): Ditto.
9936 (SELECT_CC_MODE): Use new function.
9937 (TARGET_FLAGS_REGNUM): New macro.
9938 (TARGET_FIXED_CONDITION_CODE_REGS): Ditto.
9939 (cc0_reg_rtx): Remove.
9940 (CC_STATUS_MDEP): Remove.
9941 (CC_STATUS_MDEFP_INIT): Remove.
9942 (CC_IN_FPU): Remove.
9943 (NOTICE_UPDATE_CC): Remove.
9944 (REGISTER_NAMES): Add CC registers.
9945 (BRANCH_COST): Change to constant 1.
9946 * config/pdp11/pdp11.md: Rewrite for CCmode condition code
9948 * config/pdp11/pdp11.opt (mbcopy): Remove.
9949 (mbcopy-builtin): Remove.
9950 (mbranch-cheap): Remove.
9951 (mbranch-expensive): Remove.
9952 * config/pdp11/predicates.md (expand_shift_operand): Update to
9953 match shift code generation.
9954 (ccnz_operator): New predicate.
9955 * doc/invoke.texi (PDP-11 Options): Remove deleted options
9956 -mbcopy, -mbcopy-builtin, -mbranch-cheap, -mbranch-expensive.
9957 Remove non-existent option -mabshi, -mno-abshi. Document mutually
9959 * doc/md.texi (PDP-11): Document new D and h constraints. Update
9960 description of O constraint.
9962 2018-06-27 Jeff Law <law@redhat.com>
9963 Austin Law <austinklaw@gmail.com>
9965 * config/v850/v850.md (addsi3_set_flags): New pattern.
9966 (subsi3_set_flags, negsi2_set_flags, andsi3_set_flags): Likewise.
9967 (iorsi3_set_flags, xorsi3_set_flags, one_cmplsi2_set_flags): Likewise.
9968 (zero_extendhisi2_v850_set_flags): Likewise.
9969 (zero_extendqisi2_v850_set_flags): Likewise.
9970 (ashlsi3_set_flags, ashlsi3_v850e2_set_flags): Likewise.
9971 (lshrsi3_set_flags, lshrsi3_v850e2_set_flags): Likewise.
9972 (ashrsi3_set_flags, ashrsi3_v850e2_set_flags): Likewise.
9974 * config/v850/v850-protos.h (notice_update_cc): Remove.
9975 * config/v850/v850.c (v850_compare_op0, v850_compare_op1): Remove.
9976 (v850_print_operand): Handle 'D' and "d".
9977 (v850_select_cc_mode): Remove ATTRIBUTE_UNUSED for last argument.
9978 Add handling of arithmetic/logical operations compared against zero.
9979 (v850_gen_float_compare): Remove ATTRIBUTE_UNUSED for last argument.
9980 Do not look at v850_compare_op, instead get mode from last argument.
9981 (v850_gen_compare): Remove
9982 (increment_stack): Use addsi3_clobber_flags to avoid splitting failure
9983 after reload for prologue insns.
9984 (expand_prologue): Account for CLOBBER of CC_REGNUM in various
9986 (construct_save_jarl): Likewise.
9987 (TARGET_FLAGS_REGNUM): Define.
9988 * config/v850/v850.h (v850_compare_op0, v850_compare_op1): Remove.
9989 (NOTICE_UPDATE_CC): Remove.
9990 * config/v850/v850.md (v850_tst1): Use (reg:CCZ CC_REGNUM) rather
9991 than cc0. Conditionalize on reload_completed.
9992 (cmpsi_insn, setfcc_insn): Likewise.
9993 (tst1 splitter): Turn into define_and_split which sets the flags
9995 (cstoresi4, cbranchsf4, cbranchdf4, cbranchsi4_insn): Likewise.
9996 (cbranchsi4, branch_normal, branch_invert): Do not expose cc0 here.
9997 (cstoresf4, cstoredf4): Clobber the flags.
9998 (cmpsi, cmpsf, cmpdf): Remove expanders.
9999 (setf_insn): Remove pattern.
10000 (addsi3): Turn into define_and_split which clobbers the flags after
10001 reload and a suitable pattern (addsi3_clobber_flags) for use after
10003 (subsi3, negsi2, andsi3, iorsi3, xorsi3, one_cmplsi2) Likewise.
10004 (ashlsi3, ashlsi3_v850e2, lshrsi3, lsh4si3_v850e2): Likewise.
10005 (ashrsi3, ashrsi3_v850e2): Likewise.
10006 (bins): Clobber the flags.
10007 (movsicc_normal_cc, movsicc_normal, movsicc_tst1): Likewise.
10008 (movsicc_tst1_revesed, sasf, swap and rotate patterns): Likewise.
10009 (fix_loop_counter, call_internal_short, call_internal_long): Likewise.
10010 (call_value_internal_short, call_value_internal_long): Likewise.
10011 (callt_save_interrupt, callt_return_interrupt): Likewise.
10012 (save_interrupt, return_interrupt): Likewise.
10013 (callt_save_all_interrupt, save_all_interrupt): Likewise.
10014 (_save_all_interrupt, callt_restore_all_interrupt): Likewise.
10015 (restore_all_interrupt, _restore_all_interrupt): Likewise.
10016 (All FP comparisons): Only allow after reload has completed.
10018 (divh, divhu): Tweak output template.
10019 (branch_z_normal, branch_z_invert): Remove
10020 (branch_nz_normal, branch_nz_invert): Likewise.
10021 (extendhisi_insn, extendqisi_insn): Do not clobber flags.
10023 * config/v850/v850-modes.def (CCZ, CCNZ): Add new modes.
10024 * config/v850/v850.c (notice_update_cc): Remove.
10025 * config/v850/v850.h (CC_OVERFLOW_UNUSABLE): Remove
10026 (CC_NO_CARRY): Likewise.
10027 (NOTICE_UPDATE_CC): Define to nothing.
10028 * config/v850/v850.md: Remove block comment on cc0 handling
10029 Remove "cc" attribute from all patterns. Remove cc_status handling
10030 from all patterns. Minor formatting fixes.
10032 2018-06-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10034 * config/aarch64/aarch64-cores.def (cortex-a76): New entry.
10035 (cortex-a76.cortex-a55): Likewise.
10036 * config/aarch64/aarch64-tune.md: Regenerate.
10037 * doc/invoke.texi (AArch64 Options): Document cortex-a76 and
10038 cortex-a76.cortex-a55.
10040 2018-06-27 Jeff Law <law@redhat.com>
10042 * config/v850/t-v850 (MULTILIB_OPTIONS): Remove 8byte-align.
10043 (MULTILIB_DIRNAMES): Similarly.
10045 2018-06-27 Eric Botcazou <ebotcazou@adacore.com>
10047 * gimple.h (gimple_return_retbnd): Delete.
10048 (gimple_return_set_retbnd): Likewise.
10049 * cgraphunit.c (cgraph_node::expand_thunk): Remove call to
10050 gimple_return_set_retbnd.
10051 * gimple-pretty-print.c (dump_gimple_return): Remove call to
10052 gimple_return_retbnd and adjust.
10053 * tree-inline.h (struct copy_body_data): Remove retbnd field.
10054 * tree-inline.c (remap_gimple_stmt): Remove handling of retbnd.
10055 Explicitly return NULL in a couple more cases. Move assertion
10056 on debug statements and remove unreachable code.
10057 (reset_debug_binding): Do not test id->retbnd.
10058 (expand_call_inline): Do not set it.
10060 2018-06-27 Rasmus Villemoes <rv@rasmusvillemoes.dk>
10062 * configure.ac: Add --disable-gcov option.
10063 * configure: Regenerate.
10064 * Makefile.in: Honour @enable_gcov@.
10065 * doc/install.texi: Document --disable-gcov.
10067 2018-06-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10069 * config/arm/arm-cpus.in (cortex-a76): New entry.
10070 (cortex-a76.cortex-a55): Likewise.
10071 * config/arm/arm-tables.opt: Regenerate.
10072 * config/arm/arm-tune.md: Likewise.
10073 * config/arm/driver-arm.c (arm_cpu_table): Add Cortex-A76 entry.
10074 * doc/invoke.texi (ARM Options): Document cortex-a76 and
10075 cortex-a76.cortex-a55.
10077 2018-06-27 Tamar Christina <tamar.christina@arm.com>
10080 * config/aarch64/aarch64.md (*movhf_aarch64): Add dup v0.4h pattern.
10082 2018-06-27 Siddhesh Poyarekar <siddhesh@sourceware.org>
10084 * config/aarch64/aarch64.h (CALL_USE_REGISTERS): Fix obsolete
10086 (EPILOGUE_USES): Likewise.
10088 2018-06-26 Eric Botcazou <ebotcazou@adacore.com>
10090 * tree-inline.c (remap_location): New function extracted from...
10091 (copy_edges_for_bb): Add ID parameter. Remap goto_locus.
10092 (copy_phis_for_bb): ...here. Call remap_location.
10093 (copy_cfg_body): Adjust call to copy_edges_for_bb.
10095 2018-06-26 Aaron Sawdey <acsawdey@linux.ibm.com>
10097 * config/rs6000/rs6000-string.c (expand_block_clear): Don't use
10098 unaligned vsx for 16B memset.
10100 2018-06-26 Segher Boessenkool <segher@kernel.crashing.org>
10103 * config/rs6000/rs6000.c (rs6000_init_builtins): Do not set
10104 ieee128_float_type_node to long_double_type_node unless
10105 TARGET_LONG_DOUBLE_128 is set.
10107 2018-06-26 David Malcolm <dmalcolm@redhat.com>
10109 * cfgloop.c (get_loop_location): Convert return type from
10110 location_t to dump_user_location_t, replacing INSN_LOCATION lookups
10111 by implicit construction from rtx_insn *, and using
10112 dump_user_location_t::from_function_decl for the fallback case.
10113 * cfgloop.h (get_loop_location): Convert return type from
10114 location_t to dump_user_location_t.
10115 * cgraphunit.c (walk_polymorphic_call_targets): Update call to
10116 dump_printf_loc to pass in a dump_location_t rather than a
10117 location_t, via the gimple stmt.
10118 * coverage.c (get_coverage_counts): Update calls to
10119 dump_printf_loc to pass in dump_location_t rather than a
10121 * doc/optinfo.texi (Dump types): Convert example of
10122 dump_printf_loc from taking "locus" to taking "insn". Update
10123 description of the "_loc" calls to cover dump_location_t.
10124 * dumpfile.c: Include "backend.h", "gimple.h", "rtl.h", and
10126 (dump_user_location_t::dump_user_location_t): New constructors,
10127 from gimple *stmt and rtx_insn *.
10128 (dump_user_location_t::from_function_decl): New function.
10129 (dump_loc): Make static.
10130 (dump_gimple_stmt_loc): Convert param "loc" from location_t to
10131 const dump_location_t &.
10132 (dump_generic_expr_loc): Delete.
10133 (dump_printf_loc): Convert param "loc" from location_t to
10134 const dump_location_t &.
10135 (selftest::test_impl_location): New function.
10136 (selftest::dumpfile_c_tests): New function.
10137 * dumpfile.h: Include "profile-count.h".
10138 (class dump_user_location_t): New class.
10139 (struct dump_impl_location_t): New struct.
10140 (class dump_location_t): New class.
10141 (dump_printf_loc): Convert 2nd param from source_location to
10142 const dump_location_t &.
10143 (dump_generic_expr_loc): Delete.
10144 (dump_gimple_stmt_loc): Convert 2nd param from source_location to
10145 const dump_location_t &.
10146 * gimple-fold.c (fold_gimple_assign): Update call to
10147 dump_printf_loc to pass in a dump_location_t rather than a
10148 location_t, via the gimple stmt.
10149 (gimple_fold_call): Likewise.
10150 * gimple-loop-interchange.cc
10151 (loop_cand::analyze_iloop_reduction_var): Update for change to
10152 check_reduction_path.
10153 (tree_loop_interchange::interchange): Update for change to
10154 find_loop_location.
10155 * graphite-isl-ast-to-gimple.c (scop_to_isl_ast): Update for
10156 change in return-type of find_loop_location.
10157 (graphite_regenerate_ast_isl): Likewise.
10158 * graphite-optimize-isl.c (optimize_isl): Likewise.
10159 * graphite.c (graphite_transform_loops): Likewise.
10160 * ipa-devirt.c (ipa_devirt): Update call to dump_printf_loc to
10161 pass in a dump_location_t rather than a location_t, via the
10163 * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
10164 * ipa.c (walk_polymorphic_call_targets): Likewise.
10165 * loop-unroll.c (report_unroll): Convert "locus" param from
10166 location_t to dump_location_t.
10167 (decide_unrolling): Update for change to get_loop_location's
10169 * omp-grid.c (struct grid_prop): Convert field "target_loc" from
10170 location_t to dump_user_location_t.
10171 (grid_find_single_omp_among_assignments_1): Updates calls to
10172 dump_printf_loc to pass in a dump_location_t rather than a
10173 location_t, via the gimple stmt.
10174 (grid_parallel_clauses_gridifiable): Convert "tloc" from
10175 location_t to dump_location_t. Updates calls to dump_printf_loc
10176 to pass in a dump_location_t rather than a location_t, via the
10178 (grid_inner_loop_gridifiable_p): Likewise.
10179 (grid_dist_follows_simple_pattern): Likewise.
10180 (grid_gfor_follows_tiling_pattern): Likewise.
10181 (grid_target_follows_gridifiable_pattern): Likewise.
10182 (grid_attempt_target_gridification): Convert initialization
10183 of local "grid" from memset to zero-initialization; FIXME: does
10184 this require C++11? Update call to dump_printf_loc to pass in a
10185 optinfo_location rather than a location_t, via the gimple stmt.
10186 * profile.c (read_profile_edge_counts): Updates call to
10187 dump_printf_loc to pass in a dump_location_t rather than a
10189 (compute_branch_probabilities): Likewise.
10190 * selftest-run-tests.c (selftest::run_tests): Call
10192 * selftest.h (dumpfile_c_tests): New decl.
10193 * tree-loop-distribution.c (pass_loop_distribution::execute):
10194 Update for change in return type of find_loop_location.
10195 * tree-parloops.c (parallelize_loops): Likewise.
10196 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Convert
10197 "locus" from location_t to dump_user_location_t.
10198 (canonicalize_loop_induction_variables): Likewise.
10199 * tree-ssa-loop-ivopts.c (tree_ssa_iv_optimize_loop): Update
10200 for change in return type of find_loop_location.
10201 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update call
10202 to dump_printf_loc to pass in a dump_location_t rather than a
10203 location_t, via the stmt.
10204 * tree-ssa-sccvn.c (eliminate_dom_walker::before_dom_children):
10206 * tree-vect-loop-manip.c (find_loop_location): Convert return
10207 type from source_location to dump_user_location_t.
10208 (vect_do_peeling): Update for above change.
10209 (vect_loop_versioning): Update for change in type of
10211 * tree-vect-loop.c (check_reduction_path): Convert "loc" param
10212 from location_t to dump_user_location_t.
10213 (vect_estimate_min_profitable_iters): Update for change in type
10215 * tree-vect-slp.c (vect_print_slp_tree): Convert param "loc" from
10216 location_t to dump_location_t.
10217 (vect_slp_bb): Update for change in type of vect_location.
10218 * tree-vectorizer.c (vect_location): Convert from source_location
10219 to dump_user_location_t.
10220 (try_vectorize_loop_1): Update for change in vect_location's type.
10221 (vectorize_loops): Likewise.
10222 (increase_alignment): Likewise.
10223 * tree-vectorizer.h (vect_location): Convert from source_location
10224 to dump_user_location_t.
10225 (find_loop_location): Convert return type from source_location to
10226 dump_user_location_t.
10227 (check_reduction_path): Convert 1st param from location_t to
10228 dump_user_location_t.
10229 * value-prof.c (check_counter): Update call to dump_printf_loc to
10230 pass in a dump_user_location_t rather than a location_t; update
10231 call to error_at for change in type of "locus".
10232 (check_ic_target): Update call to dump_printf_loc to
10233 pass in a dump_user_location_t rather than a location_t, via the
10236 2018-06-26 Robin Dapp <rdapp@linux.vnet.ibm.com>
10238 * config/s390/s390.h (enum processor_flags): Do not use
10239 default tune parameter when -march was specified.
10241 2018-06-26 Jakub Jelinek <jakub@redhat.com>
10244 * config/i386/i386.md (setcc + movzbl to xor + setcc peephole2s):
10245 Check reg_overlap_mentioned_p in addition to reg_set_p with the same
10248 2018-06-26 Richard Biener <rguenther@suse.de>
10250 PR tree-optimization/86287
10252 * tree-vect-loop.c (vect_transform_loop_stmt): Fix read-after-free.
10253 (vect_analyze_loop): Initialize n_stmts.
10255 2018-06-26 Richard Biener <rguenther@suse.de>
10257 PR middle-end/86271
10258 * fold-const.c (fold_convertible_p): Pointer extension
10261 2018-06-26 Alexandre Oliva <aoliva@redhat.com>
10264 * dwarf2out.c (loc_list_has_views): Adjust comments.
10265 (dw_loc_list): Split single cross-partition range with
10268 2018-06-25 Jeff Law <law@redhat.com>
10270 * common/config/v850/v850-common.c (TARGET_DEFAULT_TARGET_FLAGS): Turn
10271 on -mbig-switch by default.
10273 * config/v850/predicates.md (const_float_1_operand): Fix match_code
10275 (const_float_0_operand): Remove unused predicate.
10276 * config/v850/v850.md (define_constants): Remove UNSPEC_LOOP.
10277 (define_c_enum unspec): Add LOOP, RCP and RSQRT constants.
10278 (recipsf2): New expander. Original pattern now called
10280 (recipdf2, recipdf2_insn): Similarly.
10281 (rsqrtsf2, rsqrtsf2_insn): Similarly
10282 (rsqrtdf2, rsqrtdf2_insn): Similarly
10284 2018-06-26 Gerald Pfeifer <gerald@pfeifer.com>
10286 * ginclude/stddef.h: Remove an obsolete comment on FreeBSD 5.
10287 Simplify logic for FreeBSD (twice).
10289 2018-06-25 Martin Sebor <msebor@redhat.com>
10291 PR tree-optimization/86204
10292 * tree-ssa-strlen.c (handle_builtin_strlen): Avoid storing
10293 a strnlen result if it's less than the length of the string.
10295 2018-06-25 Martin Sebor <msebor@redhat.com>
10297 PR tree-optimization/85700
10298 * gimple-fold.c (gimple_fold_builtin_strncat): Adjust comment.
10299 * tree-ssa-strlen.c (is_strlen_related_p): Handle integer subtraction.
10300 (maybe_diag_stxncpy_trunc): Distinguish strncat from strncpy.
10302 2018-06-25 Martin Sebor <msebor@redhat.com>
10304 * doc/extend.texi (Zero-length arrays): Update and clarify.
10306 2018-06-25 Michael Meissner <meissner@linux.ibm.com>
10308 * config.gcc (powerpc64le*): Revert January 16th, 2018 patch that
10309 added IEEE/IBM long double multilib support on PowerPC little
10310 endian Linux systems.
10311 * config/rs6000/linux64.h (MULTILIB_DEFAULTS_IEEE): Likewise.
10312 (MULTILIB_DEFAULTS): Likewise.
10313 * config/rs6000/rs6000.c (rs6000_option_override_internal):
10315 * config/rs6000/rs6000.h (TARGET_IEEEQUAD_MULTILIB): Likewise.
10316 * config/rs6000/t-ldouble-linux64le-ibm: Delete, no longer used.
10317 * config/rs6000/t-ldouble-linux64le-ieee: Delete, no longer used.
10319 2018-06-25 Alexander Monakov <amonakov@ispras.ru>
10321 PR middle-end/86311
10322 * sort.cc (REORDER_23): Avoid memcpy with same destination and source.
10323 (REORDER_45): Likewise.
10325 2018-06-25 Jeff Law <law@redhat.com>
10327 * config/v850/v850.md (divmodhi4): Make sure to sign extend the
10328 dividend to 32 bits. Adjust length.
10329 (udivmodhi4): Cleanup output template. Fix length.
10331 2018-06-25 Carl Love <cel@us.ibm.com>
10333 * config/rs6000/vsx.md: Change word selector to prefered location.
10335 2018-06-25 Richard Biener <rguenther@suse.de>
10337 PR tree-optimization/86304
10338 * tree-vectorizer.c (vectorize_loops): Walk over new possibly
10339 epilogue-if-converted loops as well.
10341 2018-06-25 Jan Hubicka <hubicka@ucw.cz>
10343 * lto-section-out.c (lto_begin_section): Do not print section
10344 name for noaddr and unnumbered dumps.
10346 2018-06-25 Richard Biener <rguenther@suse.de>
10348 * tree-vectorizer.h (struct vec_info_shared): New structure
10349 with parts split out from struct vec_info and loop_nest from
10350 struct _loop_vec_info.
10351 (struct vec_info): Adjust accordingly.
10352 (struct _loop_vec_info): Likewise.
10353 (LOOP_VINFO_LOOP_NEST): Adjust.
10354 (LOOP_VINFO_DATAREFS): Likewise.
10355 (LOOP_VINFO_DDRS): Likewise.
10356 (struct _bb_vec_info): Likewise.
10357 (BB_VINFO_DATAREFS): Likewise.
10358 (BB_VINFO_DDRS): Likewise.
10359 (struct _stmt_vec_info): Add dr_aux member.
10360 (DR_VECT_AUX): Adjust to refer to member of DR_STMTs vinfo.
10361 (DR_MISALIGNMENT_UNINITIALIZED): New.
10362 (set_dr_misalignment): Adjust.
10363 (dr_misalignment): Assert misalign isn't DR_MISALIGNMENT_UNINITIALIZED.
10364 (vect_analyze_loop): Adjust prototype.
10365 (vect_analyze_loop_form): Likewise.
10366 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences):
10367 Compute dependences lazily.
10368 (vect_record_base_alignments): Use shared datarefs/ddrs.
10369 (vect_verify_datarefs_alignment): Likewise.
10370 (vect_analyze_data_refs_alignment): Likewise.
10371 (vect_analyze_data_ref_accesses): Likewise.
10372 (vect_analyze_data_refs): Likewise.
10373 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Add
10374 constructor parameter for shared part.
10375 (vect_analyze_loop_form): Pass in shared part and adjust.
10376 (vect_analyze_loop_2): Pass in storage for the number of
10377 stmts. Move loop nest finding to the caller. Compute
10379 (vect_analyze_loop): Pass in shared part.
10380 (vect_transform_loop): Verify shared datarefs are unchanged.
10381 * tree-vect-slp.c (_bb_vec_info::_bb_vec_info): Add
10382 constructor parameter for shared part.
10383 (vect_slp_analyze_bb_1): Pass in shared part and adjust.
10384 (vect_slp_bb): Verify shared datarefs are unchanged before
10386 * tree-vect-stmts.c (ensure_base_align): Adjust for DR_AUX
10388 (new_stmt_vec_info): Initialize DR_AUX misalignment to
10389 DR_MISALIGNMENT_UNINITIALIZED.
10390 * tree-vectorizer.c (vec_info::vec_info): Add constructor
10391 parameter for shared part.
10392 (vec_info::~vec_info): Adjust.
10393 (vec_info_shared::vec_info_shared): New.
10394 (vec_info_shared::~vec_info_shared): Likewise.
10395 (vec_info_shared::save_datarefs): Likewise.
10396 (vec_info_shared::check_datarefs): Likewise.
10397 (try_vectorize_loop_1): Construct shared part live for analyses
10398 of a single loop for multiple vector sizes.
10399 * tree-parloops.c (gather_scalar_reductions): Adjust.
10401 2018-06-25 Richard Biener <rguenther@suse.de>
10403 * tree-vect-data-refs.c (vect_find_stmt_data_reference): Modify
10404 DR for SIMD lane accesses here and mark DR with (void *)-1 aux.
10405 (vect_analyze_data_refs): Remove similar code from here and
10406 simplify accordingly.
10408 2018-06-25 Richard Biener <rguenther@suse.de>
10410 * tree-vect-data-refs.c (vect_check_gather_scatter): Fail
10411 for reverse storage order accesses rather than asserting
10412 they cannot happen here.
10414 2018-06-25 Tom de Vries <tdevries@suse.de>
10417 * config/i386/i386.md (define_insn "*tls_global_dynamic_64_<mode>"):
10418 Use data16 instead of .byte for insn prefix.
10420 2018-06-25 Andreas Krebbel <krebbel@linux.ibm.com>
10423 * parser.c (make_char_string_pack): Pass this literal chars
10424 through cpp_interpret_string.
10425 (cp_parser_userdef_numeric_literal): Check the result of
10426 make_char_string_pack.
10428 2018-06-24 Maya Rashish <coypu@sdf.org>
10430 * ginclude/stddef.h: Simplify conditions around avoiding
10431 re-definition of __size_t.
10433 2018-06-22 Jan Hubicka <hubicka@ucw.cz>
10435 * lto-streamer-out.c (tree_is_indexable): Make LABEL_DECL nonindexable
10436 unless it is forced or nonlocal; assert that we stream no IMPORTED_DECL.
10438 2018-06-22 Maya Rashish <coypu@sdf.org>
10440 * doc/invoke.texi (mno-fancy-math-387): Update for changes
10441 made to OpenBSD and NetBSD through the years.
10443 2018-06-22 Kelvin Nilsen <kelvin@gcc.gnu.org>
10445 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Change
10446 behavior of vec_pack (vector double, vector double) to match
10447 behavior of vec_float2 (vector double, vector double).
10449 2018-06-22 Olivier Hainque <hainque@adacore.com>
10451 * gimplify.c (gimplify_function_tree): Prevent creation
10452 of a trampoline for the address of the current function
10453 passed to entry/exit instrumentation hooks.
10455 2018-06-22 Aaron Sawdey <acsawdey@linux.ibm.com>
10458 * config/rs6000/rs6000-string.c (expand_strn_compare): Handle -m32
10461 2018-06-22 Martin Liska <mliska@suse.cz>
10463 PR tree-optimization/86263
10464 * tree-switch-conversion.c (jump_table_cluster::find_jump_tables):
10465 Bail out if is_enabled is false.
10466 * tree-switch-conversion.h (jump_table_cluster::is_enabled):
10468 (jump_table_cluster::is_enabled): New function.
10470 2018-06-22 Jan Hubicka <hubicka@ucw.cz>
10472 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not stream
10473 BINFO_BASE_ACCESSES and BINFO_VPTR_FIELD.
10474 * tree-streamer-in.c (streamer_read_tree_bitfields): Likewise.
10475 (lto_input_ts_binfo_tree_pointers): Likewise.
10476 * tree-streamer-out.c (streamer_write_tree_bitfields,
10477 write_ts_binfo_tree_pointers): Likewise.
10478 * tree.c (free_lang_data_in_binfo): Clear BINFO_VPTR_FIELD.
10480 2018-06-22 Jan Hubicka <hubicka@ucw.cz>
10482 * tree.c (free_lang_data_in_type): Free all TYPE_VFIELDs.
10484 2018-06-22 Martin Liska <mliska@suse.cz>
10486 * symbol-summary.h (get): Make it pure and inline move
10487 functionality from ::get function.
10488 (get): Remove and inline into ::get and ::get_create.
10489 (get_create): Move code from ::get function.
10491 2018-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
10494 * config/i386/sol2.h (CPP_SPEC): Don't pass -P for
10495 -x assembler-with-cpp.
10497 2018-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
10499 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define
10500 _FILE_OFFSET_BITS=64 for C++.
10502 2018-06-21 Michael Meissner <meissner@linux.ibm.com>
10504 * config/rs6000/rs6000.md (extendtfif2): Add missing 128-bit
10505 conversion insn that shows up when pr85657-3.c is compiled using
10506 IEEE 128-bit long double.
10507 (neg<mode>2_internal): Use the correct mode to check whether the
10508 mode is IBM extended.
10509 * config/rs6000/rs6000.c (init_float128_ieee): Prevent complex
10510 multiply and divide external functions from being created more
10513 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
10515 * cfgrtl.c (fixup_reorder_chain): Do not emit NOPs in DECL_IGNORED_P
10517 (rtl_merge_blocks): Likewise. Do not emit a NOP if the location of
10518 the edge can be forwarded.
10519 (cfg_layout_merge_blocks): Likewise.
10521 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
10523 * except.c (finish_eh_generation): Commit edge insertions only after
10524 the EH edges have been redirected from post-landing to landing pads.
10526 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
10528 * tree-nested.c (get_frame_type): Use create_tmp_var_raw instead of
10529 create_tmp_var_for to create the FRAME decl.
10530 (finalize_nesting_tree_1): Do not unchain the FRAME decl.
10532 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
10534 * tree-inline.c (copy_edges_for_bb): Minor tweak.
10535 (maybe_move_debug_stmts_to_successors): Also reset the locus of the
10536 debug statement when resetting its value.
10537 (expand_call_inline): Copy the locus of the call onto the assignment
10538 of the return value, if any. Use local variable in more cases.
10540 2018-06-21 Martin Liska <mliska@suse.cz>
10542 * ipa-pure-const.c (propagate_nothrow): Use
10543 funct_state_summaries->get.
10544 (dump_malloc_lattice): Likewise.
10545 (propagate_malloc): Likewise.
10547 2018-06-21 Richard Biener <rguenther@suse.de>
10549 * lto-streamer-out.c (DFS::DFS_write_tree_body): Update outdated
10550 comment. Follow BLOCK_ABSTRACT_ORIGIN unconditionally.
10551 * tree-streamer-in.c (lto_input_ts_block_tree_pointers): Update
10553 * tree-streamer-out.c (write_ts_block_tree_pointers): Stream
10554 BLOCK_ABSTRACT_ORIGIN unconditionally.
10556 2018-06-21 David Malcolm <dmalcolm@redhat.com>
10558 * ipa-cp.c (ipcp_driver): Set edge_clone_summaries to NULL after
10560 * ipa-reference.c (ipa_reference_c_finalize): Delete
10561 ipa_ref_opt_sum_summaries and set it to NULL.
10563 2018-06-21 Tom de Vries <tdevries@suse.de>
10565 PR tree-optimization/85859
10566 * tree-ssa-tail-merge.c (stmt_local_def): Copy gimple_is_call
10567 test with comment from bb_no_side_effects_p.
10569 2018-06-21 Richard Biener <rguenther@suse.de>
10571 PR tree-optimization/86232
10572 * tree-ssa-loop-niter.c (number_of_iterations_popcount): Adjust
10573 max for constant niter.
10575 2018-06-21 Andre Vieira <andre.simoesdiasvieira@arm.com>
10577 * config/aarch64/aarch64-simd.md
10578 (*aarch64_crypto_aes<aes_op>v16qi_xor_combine): New.
10580 2018-06-21 Andre Vieira <andre.simoesdiasvieira@arm.com>
10582 * config/aarch64/aarch64-simd.md (aarch64_crypto_aes<aes_op>v16qi):
10583 Make opernads of the unspec commutative.
10585 2018-06-21 Richard Biener <rguenther@suse.de>
10587 * tree-data-ref.c (dr_step_indicator): Handle NULL DR_STEP.
10588 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
10589 Avoid calling vect_mark_for_runtime_alias_test with gathers or scatters.
10590 (vect_analyze_data_ref_dependence): Re-order checks to deal with
10592 (vect_record_base_alignments): Do not record base alignment
10593 for gathers or scatters.
10594 (vect_compute_data_ref_alignment): Drop return value that is always
10595 true. Bail out early for gathers or scatters.
10596 (vect_enhance_data_refs_alignment): Bail out early for gathers
10598 (vect_find_same_alignment_drs): Likewise.
10599 (vect_analyze_data_refs_alignment): Remove dead code.
10600 (vect_slp_analyze_and_verify_node_alignment): Likewise.
10601 (vect_analyze_data_refs): For possible gathers or scatters do
10602 not create an alternate DR, just check their possible validity
10603 and mark them. Adjust DECL_NONALIASED handling to not rely
10604 on DR_BASE_ADDRESS.
10605 * tree-vect-loop-manip.c (vect_update_inits_of_drs): Do not
10606 update inits of gathers or scatters.
10607 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern):
10608 Also copy gather/scatter flag to pattern vinfo.
10610 2018-06-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
10612 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Change
10613 behavior of vec_packsu (vector unsigned long long, vector unsigned
10614 long long) to match behavior of vec_packs with same signature.
10616 2018-06-20 Chung-Lin Tang <cltang@codesourcery.com>
10617 Thomas Schwinge <thomas@codesourcery.com>
10618 Cesar Philippidis <cesar@codesourcery.com>
10620 * gimplify.c (gimplify_scan_omp_clauses): Add support for
10621 OMP_CLAUSE_{IF_PRESENT,FINALIZE}.
10622 (gimplify_adjust_omp_clauses): Likewise.
10623 (gimplify_oacc_declare_1): Add support for GOMP_MAP_RELEASE, remove
10624 support for GOMP_MAP_FORCE_{ALLOC,TO,FROM,TOFROM}.
10625 (gimplify_omp_target_update): Update handling of acc update and
10627 * omp-low.c (install_var_field): Remove unused parameter
10628 base_pointers_restrict.
10629 (scan_sharing_clauses): Remove base_pointers_restrict parameter.
10630 Update call to install_var_field. Handle OMP_CLAUSE_{IF_PRESENT,
10632 (omp_target_base_pointers_restrict_p): Delete.
10633 (scan_omp_target): Update call to scan_sharing_clauses.
10634 * tree-core.h (enum omp_clause_code): Add OMP_CLAUSE_{IF_PRESENT,
10636 * tree-nested.c (convert_nonlocal_omp_clauses): Handle
10637 OMP_CLAUSE_{IF_PRESENT,FINALIZE}.
10638 (convert_local_omp_clauses): Likewise.
10639 * tree-pretty-print.c (dump_omp_clause): Likewise.
10640 * tree.c (omp_clause_num_ops): Add entries for OMP_CLAUSE_{IF_PRESENT,
10642 (omp_clause_code_name): Likewise.
10644 2018-06-20 Jakub Jelinek <jakub@redhat.com>
10647 * var-tracking.c (use_narrower_mode_test): Check if shift amount can
10650 PR tree-optimization/86231
10651 * tree-vrp.c (union_ranges): For ( [ ) ] or ( )[ ] range and
10652 anti-range don't overwrite *vr0min before using it to compute *vr0max.
10654 2018-06-20 Tom de Vries <tdevries@suse.de>
10656 PR tree-optimization/86097
10657 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Also convert *nit to
10658 iv type if signedness of iv type is not the same as that of *nit.
10660 2018-06-20 Jakub Jelinek <jakub@redhat.com>
10662 * cfgrtl.c (rtl_verify_edges): Formatting fix. If bb->preds has any
10663 EDGE_EH edges, verify they are all EDGE_EH.
10665 2018-06-20 Maya Rashish <coypu@sdf.org>
10667 * ginclude/stddef.h: Limit #include <machine/ansi.h> to NetBSD.
10669 2018-06-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10671 * config/aarch64/aarch64-tuning-flags.def (no_ldp_stp_qregs): New.
10672 * config/aarch64/aarch64.c (xgene1_tunings): Add
10673 AARCH64_EXTRA_TUNE_NO_LDP_STP_QREGS to tune_flags.
10674 (aarch64_mode_valid_for_sched_fusion_p):
10675 Allow 16-byte modes.
10676 (aarch64_classify_address): Allow 16-byte modes for load_store_pair_p.
10677 * config/aarch64/aarch64-ldpstp.md: Add peepholes for LDP STP of
10679 * config/aarch64/aarch64-simd.md (load_pair<VQ:mode><VQ2:mode>):
10681 (vec_store_pair<VQ:mode><VQ2:mode>): Likewise.
10682 * config/aarch64/iterators.md (VQ2): New mode iterator.
10684 2018-06-20 Martin Liska <mliska@suse.cz>
10686 * tree-switch-conversion.c (jump_table_cluster::can_be_handled):
10687 Change default ratio from 10 to 8.
10689 2018-06-20 Martin Liska <mliska@suse.cz>
10691 * tree-switch-conversion.c (jump_table_cluster::find_jump_tables):
10693 (bit_test_cluster::find_bit_tests): Likewise.
10694 (switch_decision_tree::analyze_switch_statement): Find clusters.
10695 * tree-switch-conversion.h (struct jump_table_cluster): Document
10698 2018-06-20 Martin Liska <mliska@suse.cz>
10700 * tree-switch-conversion.c (switch_conversion::collect):
10701 Record m_uniq property.
10702 (switch_conversion::expand): Bail out for special conditions.
10703 (group_cluster::~group_cluster): New.
10704 (group_cluster::group_cluster): Likewise.
10705 (group_cluster::dump): Likewise.
10706 (jump_table_cluster::emit): New.
10707 (switch_decision_tree::fix_phi_operands_for_edges): New.
10708 (struct case_node): Remove struct.
10709 (jump_table_cluster::can_be_handled): New.
10710 (case_values_threshold): Moved to header.
10711 (reset_out_edges_aux): Likewise.
10712 (jump_table_cluster::is_beneficial): New.
10713 (bit_test_cluster::can_be_handled): Likewise.
10714 (add_case_node): Remove.
10715 (bit_test_cluster::is_beneficial): New.
10716 (case_bit_test::cmp): New.
10717 (bit_test_cluster::emit): New.
10718 (expand_switch_as_decision_tree_p): Remove.
10719 (bit_test_cluster::hoist_edge_and_branch_if_true): New.
10720 (fix_phi_operands_for_edge): Likewise.
10721 (switch_decision_tree::analyze_switch_statement): New.
10722 (compute_cases_per_edge): Move ...
10723 (switch_decision_tree::compute_cases_per_edge): ... here.
10724 (try_switch_expansion): Likewise.
10725 (switch_decision_tree::try_switch_expansion): Likewise.
10726 (record_phi_operand_mapping): Likewise.
10727 (switch_decision_tree::record_phi_operand_mapping): Likewise.
10728 (emit_case_decision_tree): Likewise.
10729 (switch_decision_tree::emit): Likewise.
10730 (balance_case_nodes): Likewise.
10731 (switch_decision_tree::balance_case_nodes): Likewise.
10732 (dump_case_nodes): Likewise.
10733 (switch_decision_tree::dump_case_nodes): Likewise.
10734 (emit_jump): Likewise.
10735 (switch_decision_tree::emit_jump): Likewise.
10736 (emit_cmp_and_jump_insns): Likewise.
10737 (switch_decision_tree::emit_cmp_and_jump_insns): Likewise.
10738 (emit_case_nodes): Likewise.
10739 (switch_decision_tree::emit_case_nodes): Likewise.
10740 (conditional_probability): Remove.
10741 * tree-switch-conversion.h (enum cluster_type): New.
10743 (struct cluster): Likewise.
10744 (cluster::cluster): Likewise.
10745 (struct simple_cluster): Likewise.
10746 (simple_cluster::simple_cluster): Likewise.
10747 (struct group_cluster): Likewise.
10748 (struct jump_table_cluster): Likewise.
10749 (struct bit_test_cluster): Likewise.
10750 (struct min_cluster_item): Likewise.
10751 (struct case_tree_node): Likewise.
10752 (case_tree_node::case_tree_node): Likewise.
10753 (jump_table_cluster::case_values_threshold): Likewise.
10754 (struct case_bit_test): Likewise.
10755 (struct switch_decision_tree): Likewise.
10756 (struct switch_conversion): Likewise.
10757 (switch_decision_tree::reset_out_edges_aux): Likewise.
10759 2018-06-20 Martin Liska <mliska@suse.cz>
10761 * tree-switch-conversion.c (MAX_CASE_BIT_TESTS): Remove.
10762 (hoist_edge_and_branch_if_true): Likewise.
10763 (expand_switch_using_bit_tests_p): Likewise.
10764 (struct case_bit_test): Likewise.
10765 (case_bit_test_cmp): Likewise.
10766 (emit_case_bit_tests): Likewise.
10767 (switch_conversion::switch_conversion): New class.
10768 (struct switch_conv_info): Remove old struct.
10769 (collect_switch_conv_info): More to ...
10770 (switch_conversion::collect): ... this.
10771 (check_range): Likewise.
10772 (switch_conversion::check_range): Likewise.
10773 (check_all_empty_except_final): Likewise.
10774 (switch_conversion::check_all_empty_except_final): Likewise.
10775 (check_final_bb): Likewise.
10776 (switch_conversion::check_final_bb): Likewise.
10777 (create_temp_arrays): Likewise.
10778 (switch_conversion::create_temp_arrays): Likewise.
10779 (free_temp_arrays): Likewise.
10780 (gather_default_values): Likewise.
10781 (switch_conversion::gather_default_values): Likewise.
10782 (build_constructors): Likewise.
10783 (switch_conversion::build_constructors): Likewise.
10784 (constructor_contains_same_values_p): Likewise.
10785 (switch_conversion::contains_same_values_p): Likewise.
10786 (array_value_type): Likewise.
10787 (switch_conversion::array_value_type): Likewise.
10788 (build_one_array): Likewise.
10789 (switch_conversion::build_one_array): Likewise.
10790 (build_arrays): Likewise.
10791 (switch_conversion::build_arrays): Likewise.
10792 (gen_def_assigns): Likewise.
10793 (switch_conversion::gen_def_assigns): Likewise.
10794 (prune_bbs): Likewise.
10795 (switch_conversion::prune_bbs): Likewise.
10796 (fix_phi_nodes): Likewise.
10797 (switch_conversion::fix_phi_nodes): Likewise.
10798 (gen_inbound_check): Likewise.
10799 (switch_conversion::gen_inbound_check): Likewise.
10800 (process_switch): Use the newly created class.
10801 (switch_conversion::expand): New.
10802 (switch_conversion::~switch_conversion): New.
10803 * tree-switch-conversion.h: New file.
10805 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
10807 * tree-vectorizer.h (NUM_PATTERNS, vect_recog_func_ptr): Move to
10808 tree-vect-patterns.c.
10809 * tree-vect-patterns.c (vect_supportable_direct_optab_p): New function.
10810 (vect_recog_dot_prod_pattern): Use it. Remove the type_in argument.
10811 (vect_recog_sad_pattern): Likewise.
10812 (vect_recog_widen_sum_pattern): Likewise.
10813 (vect_recog_pow_pattern): Likewise. Check for a null vectype.
10814 (vect_recog_widen_shift_pattern): Remove the type_in argument.
10815 (vect_recog_rotate_pattern): Likewise.
10816 (vect_recog_mult_pattern): Likewise.
10817 (vect_recog_vector_vector_shift_pattern): Likewise.
10818 (vect_recog_divmod_pattern): Likewise.
10819 (vect_recog_mixed_size_cond_pattern): Likewise.
10820 (vect_recog_bool_pattern): Likewise.
10821 (vect_recog_mask_conversion_pattern): Likewise.
10822 (vect_try_gather_scatter_pattern): Likewise.
10823 (vect_recog_widen_mult_pattern): Likewise. Check for a null vectype.
10824 (vect_recog_over_widening_pattern): Likewise.
10825 (vect_recog_gather_scatter_pattern): Likewise.
10826 (vect_recog_func_ptr): Move from tree-vectorizer.h
10827 (vect_vect_recog_func_ptrs): Move further down the file.
10828 (vect_recog_func): Likewise. Remove the third argument.
10829 (NUM_PATTERNS): Define based on vect_vect_recog_func_ptrs.
10830 (vect_pattern_recog_1): Expect the pattern function to do any
10831 necessary target tests. Also expect it to provide a vector type.
10832 Remove the type_in handling.
10834 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
10836 * tree-vect-patterns.c (vect_pattern_detected): New function.
10837 (vect_recog_dot_prod_patternm, vect_recog_sad_pattern)
10838 (vect_recog_widen_mult_pattern, vect_recog_widen_sum_pattern)
10839 (vect_recog_over_widening_pattern, vect_recog_widen_shift_pattern
10840 (vect_recog_rotate_pattern, vect_recog_vector_vector_shift_pattern)
10841 (vect_recog_mult_pattern, vect_recog_divmod_pattern)
10842 (vect_recog_mixed_size_cond_pattern, vect_recog_bool_pattern)
10843 (vect_recog_mask_conversion_pattern)
10844 (vect_try_gather_scatter_pattern): Likewise.
10846 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
10848 * tree-vect-patterns.c (vect_get_internal_def): New function.
10849 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern)
10850 (vect_recog_vector_vector_shift_pattern, check_bool_pattern)
10851 (search_type_for_mask_1): Use it.
10853 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
10855 * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Remove
10856 redundant WIDEN_SUM_EXPR handling.
10857 (vect_recog_sad_pattern): Likewise.
10859 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
10861 * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Remove
10862 redundant check that the types of a PLUS_EXPR or MULT_EXPR agree.
10863 (vect_recog_sad_pattern): Likewise PLUS_EXPR, ABS_EXPR and MINUS_EXPR.
10864 (vect_recog_widen_mult_pattern): Likewise MULT_EXPR.
10865 (vect_recog_widen_sum_pattern): Likewise PLUS_EXPR.
10867 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
10869 * tree-vect-stmts.c (vectorizable_call): Make sure that we
10870 use the stmt_vec_info of the original bb statement for the
10871 new zero assignment, even if the call is part of a pattern.
10873 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
10875 * tree-vectorizer.h (_stmt_vec_info): Note above pattern_def_seq
10876 that the sequence is attached to the original statement rather
10877 than the pattern statement.
10878 * tree-vect-loop.c (vect_determine_vf_for_stmt): Take the
10879 PATTERN_DEF_SEQ from the original statement rather than
10880 the main pattern statement.
10881 * tree-vect-stmts.c (free_stmt_vec_info): Likewise.
10882 * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Likewise.
10883 (vect_mark_pattern_stmts): Don't copy the PATTERN_DEF_SEQ.
10885 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
10887 * tree-vect-stmts.c (vect_analyze_stmt): Move the handling of pattern
10888 definition statements before the early exit for statements that aren't
10890 * tree-vect-loop.c (vect_transform_loop_stmt): New function,
10892 (vect_transform_loop): ...here. Process pattern definition
10893 statements without first checking whether the main pattern
10894 statement is live or relevant.
10896 2018-06-19 Eric Botcazou <ebotcazou@adacore.com>
10898 * tree-cfgcleanup.c (tree_forwarder_block_p): Do not return false at
10899 -O0 if the locus represent UNKNOWN_LOCATION but have different values.
10901 2018-06-19 Aaron Sawdey <acsawdey@linux.ibm.com>
10903 * config/rs6000/rs6000-string.c (select_block_compare_mode): Check
10904 TARGET_EFFICIENT_OVERLAPPING_UNALIGNED here instead of in caller.
10905 (do_and3, do_and3_mask, do_compb3, do_rotl3): New functions.
10906 (expand_block_compare): Change select_block_compare_mode call.
10907 (expand_strncmp_align_check): Use new functions, fix comment.
10908 (emit_final_str_compare_gpr): New function.
10909 (expand_strn_compare): Refactor and clean up code.
10910 * config/rs6000/vsx.md (vsx_mov<mode>_64bit): Remove *.
10912 2018-06-19 Tony Reix <tony.reix@atos.com>
10913 Damien Bergamini <damien.bergamini@atos.com>
10914 David Edelsohn <dje.gcc@gmail.com>
10916 * collect2.c (static_obj): New variable.
10917 (static_libs): New variable.
10918 (is_in_list): Uncomment declaration.
10919 (main): Track AIX libraries linked statically.
10920 (is_in_list): Uncomment definition.
10921 (scan_prog_file): Don't add AIX shared libraries initializer
10922 to constructor list if linking statically.
10924 2018-06-19 Max Filippov <jcmvbkbc@gmail.com>
10926 * config/xtensa/xtensa.md (UNSPEC_FRAME_BLOCKAGE): New unspec
10928 (allocate_stack, frame_blockage, *frame_blockage): New patterns.
10930 2018-06-19 Jan Hubicka <hubicka@ucw.cz>
10932 * tree.c (find_decls_types_r): Remove all non-VAR_DECLs from
10935 2018-06-19 Martin Liska <mliska@suse.cz>
10937 * config/i386/i386.c (ix86_can_inline_p): Do not use
10938 ipa_fn_summaries::get_create.
10939 * ipa-cp.c (ipcp_cloning_candidate_p): Replace get_create with
10941 (devirtualization_time_bonus): Likewise.
10942 (ipcp_propagate_stage): Likewise.
10943 * ipa-fnsummary.c (redirect_to_unreachable): Likewise.
10944 (edge_set_predicate): Likewise.
10945 (evaluate_conditions_for_known_args): Likewise.
10946 (evaluate_properties_for_edge): Likewise.
10947 (ipa_call_summary::reset): Tranform to ...
10948 (ipa_call_summary::~ipa_call_summary): ... this.
10949 (ipa_fn_summary::reset): Transform to ...
10950 (ipa_fn_summary::~ipa_fn_summary): ... this.
10951 (ipa_fn_summary_t::remove): Rename to ...
10952 (ipa_fn_summary_t::remove_callees): ... this.
10953 (ipa_fn_summary_t::duplicate): Use placement new
10954 instead of memory copy.
10955 (ipa_call_summary_t::duplicate): Likewise.
10956 (ipa_call_summary_t::remove): Remove.
10957 (dump_ipa_call_summary): Change get_create to get.
10958 (ipa_dump_fn_summary): Dump only when summary exists.
10959 (analyze_function_body): Use symbol_summary::get instead
10961 (compute_fn_summary): Likewise.
10962 (estimate_edge_devirt_benefit): Likewise.
10963 (estimate_edge_size_and_time): Likewise.
10964 (inline_update_callee_summaries): Likewise.
10965 (remap_edge_change_prob): Likewise.
10966 (remap_edge_summaries): Likewise.
10967 (ipa_merge_fn_summary_after_inlining): Likewise.
10968 (write_ipa_call_summary): Likewise.
10969 (ipa_fn_summary_write): Likewise.
10970 (ipa_free_fn_summary): Likewise.
10971 * ipa-fnsummary.h (struct GTY): Add new ctor and copy ctor.
10972 (struct ipa_call_summary): Likewise.
10973 * ipa-icf.c (sem_function::merge): Use symbol_summary::get instead
10975 * ipa-inline-analysis.c (do_estimate_edge_time): Likewise.
10976 (estimate_size_after_inlining): Likewise.
10977 (estimate_growth): Likewise.
10978 (growth_likely_positive): Likewise.
10979 * ipa-inline-transform.c (clone_inlined_nodes): Likewise.
10980 (inline_call): Likewise.
10981 * ipa-inline.c (caller_growth_limits): Likewise.
10982 (can_inline_edge_p): Likewise.
10983 (can_inline_edge_by_limits_p): Likewise.
10984 (compute_uninlined_call_time): Likewise.
10985 (compute_inlined_call_time): Likewise.
10986 (want_inline_small_function_p): Likewise.
10987 (edge_badness): Likewise.
10988 (update_caller_keys): Likewise.
10989 (update_callee_keys): Likewise.
10990 (inline_small_functions): Likewise.
10991 (inline_to_all_callers_1): Likewise.
10992 (dump_overall_stats): Likewise.
10993 (early_inline_small_functions): Likewise.
10994 (early_inliner): Likewise.
10995 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
10996 * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
10997 * ipa-pure-const.c (malloc_candidate_p): Likewise.
10998 * ipa-split.c (execute_split_functions): Likewise.
10999 * symbol-summary.h: Likewise.
11000 * tree-sra.c (ipa_sra_preliminary_function_checks): Likewise.
11002 2018-06-19 Richard Biener <rguenther@suse.de>
11004 * tree-vectorizer.c (try_vectorize_loop_1): Split out of ...
11005 (vectorize_loops): ... here. Fix dbgcnt handling.
11006 (try_vectorize_loop): Wrap try_vectorize_loop_1.
11008 2018-06-19 Segher Boessenkool <segher@kernel.crashing.org>
11011 * config/rs6000/rs6000.md (rs6000_discover_homogeneous_aggregate): An
11012 ieee128 argument takes up only one (vector) register, not two (floating
11015 2018-06-19 Eric Botcazou <ebotcazou@adacore.com>
11017 * gimplify.c (gimplify_init_constructor): Really never clear for an
11018 incomplete constructor if CONSTRUCTOR_NO_CLEARING is set.
11020 2018-06-19 Richard Biener <rguenther@suse.de>
11022 PR tree-optimization/86179
11023 * tree-vect-patterns.c (vect_pattern_recog_1): Clean up
11024 after failed recognition.
11026 2018-06-18 Martin Sebor <msebor@redhat.com>
11028 PR middle-end/85602
11029 * calls.c (maybe_warn_nonstring_arg): Handle strncat.
11030 * tree-ssa-strlen.c (is_strlen_related_p): Make extern.
11031 Handle integer subtraction.
11032 (maybe_diag_stxncpy_trunc): Handle nonstring source arguments.
11033 * tree-ssa-strlen.h (is_strlen_related_p): Declare.
11035 2018-06-18 David Malcolm <dmalcolm@redhat.com>
11037 * config/frv/frv-protos.h (frv_ifcvt_modify_insn): Strengthen 3rd
11038 param from rtx to rtx_insn *.
11039 * config/frv/frv.c (frv_ifcvt_add_insn): Likewise for "insn"
11041 (frv_ifcvt_modify_insn): Likwise.
11042 (frv_ifcvt_modify_final): Likwise for local "existing_insn",
11043 adding an as_a <rtx_insn *> cast. Likewise for local "insn".
11044 * config/mips/mips.c (r10k_insert_cache_barriers): Add an
11045 as_a <rtx_insn *> cast to local "unprotected_region" once
11046 it's been established that it's not NULL or pc_rtx.
11047 * config/nds32/nds32-relax-opt.c (nds32_group_insns): Strengthen
11048 param "sethi" from rtx to rtx_insn *.
11049 (nds32_group_float_insns): Likewise for param "insn".
11050 * config/vax/vax-protos.h (vax_output_int_add): Likewise for 1st
11052 (vax_output_int_subtract): Likewise.
11053 * config/vax/vax.c (vax_output_int_add): Likewise for param
11055 (vax_output_int_subtract): Likewise.
11056 * emit-rtl.c (set_insn_deleted): Likewise, removing cast.
11057 (emit_pattern_after): Likewise for param "after".
11058 (emit_insn_after): Likewise.
11059 (emit_jump_insn_after): Likewise.
11060 (emit_call_insn_after): Likewise.
11061 (emit_debug_insn_after): Likewise.
11062 (emit_pattern_before): Likewise for param "before".
11063 (emit_insn_before): Likewise.
11064 (emit_jump_insn_before): Likewise.
11065 * final.c (get_insn_template): Likewise for param "insn", removing
11067 * output.h (get_insn_template): Likewise for 2nd param.
11068 * rtl.h (emit_insn_before): Likewise.
11069 (emit_jump_insn_before): Likewise.
11070 (emit_debug_insn_before_noloc): Likewise.
11071 (emit_insn_after): Likewise.
11072 (emit_jump_insn_after): Likewise.
11073 (emit_call_insn_after): Likewise.
11074 (emit_debug_insn_after): Likewise.
11075 (set_insn_deleted): Likewise for param.
11077 2018-06-18 Michael Meissner <meissner@linux.ibm.com>
11080 * config/rs6000/rs6000-modes.def (toplevel): Rework the 128-bit
11081 floating point modes, so that IFmode is numerically greater than
11082 TFmode, which is greater than KFmode using FRACTIONAL_FLOAT_MODE
11083 to declare the ordering. This prevents IFmode from being
11084 converted to TFmode when long double is IEEE 128-bit on an ISA 3.0
11085 machine. Include rs6000-modes.h to share the fractional values
11086 between genmodes* and the rest of the compiler.
11087 (IFmode): Likewise.
11088 (KFmode): Likewise.
11089 (TFmode): Likewise.
11090 * config/rs6000/rs6000-modes.h: New file.
11091 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Change the
11092 meaning of rs6000_long_double_size so that 126..128 selects an
11093 appropriate 128-bit floating point type.
11094 (rs6000_option_override_internal): Likewise.
11095 * config/rs6000/rs6000.h (toplevel): Include rs6000-modes.h.
11096 (TARGET_LONG_DOUBLE_128): Change the meaning of
11097 rs6000_long_double_size so that 126..128 selects an appropriate
11098 128-bit floating point type.
11099 (LONG_DOUBLE_TYPE_SIZE): Update comment.
11100 * config/rs6000/rs6000.md (trunciftf2): Correct the modes of the
11101 source and destination to match the standard usage.
11102 (truncifkf2): Likewise.
11103 (copysign<mode>3, IEEE iterator): Rework copysign of float128 on
11104 ISA 2.07 to use an explicit clobber, instead of passing in a
11106 (copysign<mode>3_soft): Likewise.
11108 2018-06-18 David Malcolm <dmalcolm@redhat.com>
11110 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences):
11111 Replace dump_printf_loc call with DUMP_VECT_SCOPE.
11112 (vect_slp_analyze_instance_dependence): Likewise.
11113 (vect_enhance_data_refs_alignment): Likewise.
11114 (vect_analyze_data_refs_alignment): Likewise.
11115 (vect_slp_analyze_and_verify_instance_alignment
11116 (vect_analyze_data_ref_accesses): Likewise.
11117 (vect_prune_runtime_alias_test_list): Likewise.
11118 (vect_analyze_data_refs): Likewise.
11119 * tree-vect-loop-manip.c (vect_update_inits_of_drs): Likewise.
11120 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
11121 (vect_analyze_scalar_cycles_1): Likewise.
11122 (vect_get_loop_niters): Likewise.
11123 (vect_analyze_loop_form_1): Likewise.
11124 (vect_update_vf_for_slp): Likewise.
11125 (vect_analyze_loop_operations): Likewise.
11126 (vect_analyze_loop): Likewise.
11127 (vectorizable_induction): Likewise.
11128 (vect_transform_loop): Likewise.
11129 * tree-vect-patterns.c (vect_pattern_recog): Likewise.
11130 * tree-vect-slp.c (vect_analyze_slp): Likewise.
11131 (vect_make_slp_decision): Likewise.
11132 (vect_detect_hybrid_slp): Likewise.
11133 (vect_slp_analyze_operations): Likewise.
11134 (vect_slp_bb): Likewise.
11135 * tree-vect-stmts.c (vect_mark_stmts_to_be_vectorized): Likewise.
11136 (vectorizable_bswap): Likewise.
11137 (vectorizable_call): Likewise.
11138 (vectorizable_simd_clone_call): Likewise.
11139 (vectorizable_conversion): Likewise.
11140 (vectorizable_assignment): Likewise.
11141 (vectorizable_shift): Likewise.
11142 (vectorizable_operation): Likewise.
11143 * tree-vectorizer.h (DUMP_VECT_SCOPE): New macro.
11145 2018-06-18 Martin Sebor <msebor@redhat.com>
11147 PR tree-optimization/81384
11148 * builtin-types.def (BT_FN_SIZE_CONST_STRING_SIZE): New.
11149 * builtins.c (expand_builtin_strnlen): New function.
11150 (expand_builtin): Call it.
11151 (fold_builtin_n): Avoid setting TREE_NO_WARNING.
11152 * builtins.def (BUILT_IN_STRNLEN): New.
11153 * calls.c (maybe_warn_nonstring_arg): Handle BUILT_IN_STRNLEN.
11154 Warn for bounds in excess of maximum object size.
11155 * tree-ssa-strlen.c (maybe_set_strlen_range): Return tree representing
11156 single-value ranges. Handle strnlen.
11157 (handle_builtin_strlen): Handle strnlen.
11158 (strlen_check_and_optimize_stmt): Same.
11159 * doc/extend.texi (Other Builtins): Document strnlen.
11161 2018-06-18 Maya Rashish <coypu@sdf.org>
11163 * config/alpha/openbsd.h (TARGET_DEFAULT): Define.
11164 (LINK_SPEC, STARTFILE_SPEC, ENDFILE_SPEC): Likewise.
11165 (INTMAX_TYPE, UINTMAX_TYPE, WINT_TYPE): Likewise.
11167 * config/alpha/elf.h (STARTFILE_SPEC, ENDFILE_SPEC): Move from
11169 * config/alpha/linux.h (STARTFILE_SPEC, ENDFILE_SPEC): Here.
11171 2018-06-18 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
11173 * tree.c (escaped_string::escape): Replace cast to char * by
11174 const_cast<char *> (unescaped).
11176 2018-06-18 Nick Clifton <nickc@redhat.com>
11179 * tree.c (escaped_string): New class. Converts an unescaped
11180 string into its escaped equivalent.
11181 (warn_deprecated_use): Use the new class to convert the
11182 deprecation message, if present.
11183 (test_escaped_strings): New self test.
11184 (test_c_tests): Add test_escaped_strings.
11185 * doc/extend.texi (deprecated): Add a note that the
11186 deprecation message is affected by the -fmessage-length
11187 option, and that control characters will be escaped.
11188 (#pragma GCC error): Document this pragma.
11189 (#pragma GCC warning): Likewise.
11190 * doc/invoke.texi (-fmessage-length): Document this option's
11191 effect on the #warning and #error preprocessor directives and
11192 the deprecated attribute.
11194 2018-06-18 Eric Botcazou <ebotcazou@adacore.com>
11196 * tree.c (decl_value_expr_lookup): Revert latest change.
11197 (decl_value_expr_insert): Likewise.
11199 2018-06-17 Eric Botcazou <ebotcazou@adacore.com>
11201 * gimplify.c (nonlocal_vlas): Delete.
11202 (nonlocal_vla_vars): Likewise.
11203 (gimplify_var_or_parm_decl): Do not add debug VAR_DECLs for non-local
11205 (gimplify_body): Do not create and destroy nonlocal_vlas.
11206 * tree-nested.c: Include diagnostic.h.
11207 (use_pointer_in_frame): Tweak.
11208 (lookup_field_for_decl): Add assertion and declare the transformation.
11209 (convert_nonlocal_reference_op) <PARM_DECL>: Rework and issue an
11210 internal error when the reference is in a wrong context. Do not
11211 create a debug decl by default.
11212 (note_nonlocal_block_vlas): Delete.
11213 (convert_nonlocal_reference_stmt) <GIMPLE_BIND>: Do not call it.
11214 (convert_local_reference_op) <PARM_DECL>: Skip the frame decl. Do not
11215 create a debug decl by default.
11216 (convert_gimple_call) <GIMPLE_CALL>: Issue an internal error when the
11217 call is in a wrong context.
11218 (fixup_vla_decls): New function.
11219 (finalize_nesting_tree_1): Adjust comment. Call fixup_vla_decls if no
11220 debug variables were created.
11221 * tree.c (decl_value_expr_lookup): Add checking assertion.
11222 (decl_value_expr_insert): Likewise.
11224 2018-06-16 Kugan Vivekanandarajah <kuganv@linaro.org>
11226 PR middle-end/82479
11227 * ipa-fnsummary.c (will_be_nonconstant_expr_predicate): Handle CALL_EXPR.
11228 * tree-scalar-evolution.c (interpret_expr): Likewise.
11229 (expression_expensive_p): Likewise.
11230 * tree-ssa-loop-ivopts.c (contains_abnormal_ssa_name_p): Likewise.
11231 * tree-ssa-loop-niter.c (number_of_iterations_popcount): New.
11232 (number_of_iterations_exit_assumptions): Use number_of_iterations_popcount.
11233 (ssa_defined_by_minus_one_stmt_p): New.
11235 2018-06-16 Kugan Vivekanandarajah <kuganv@linaro.org>
11237 PR middle-end/64946
11238 * cfgexpand.c (expand_debug_expr): Hande ABSU_EXPR.
11239 * config/i386/i386.c (ix86_add_stmt_cost): Likewise.
11240 * dojump.c (do_jump): Likewise.
11241 * expr.c (expand_expr_real_2): Check operand type's sign.
11242 * fold-const.c (const_unop): Handle ABSU_EXPR.
11243 (fold_abs_const): Likewise.
11244 * gimple-pretty-print.c (dump_unary_rhs): Likewise.
11245 * gimple-ssa-backprop.c (backprop::process_assign_use): Likesie.
11246 (strip_sign_op_1): Likesise.
11247 * match.pd: Add new pattern to generate ABSU_EXPR.
11248 * optabs-tree.c (optab_for_tree_code): Handle ABSU_EXPR.
11249 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
11250 * tree-eh.c (operation_could_trap_helper_p): Likewise.
11251 * tree-inline.c (estimate_operator_cost): Likewise.
11252 * tree-pretty-print.c (dump_generic_node): Likewise.
11253 * tree-vect-patterns.c (vect_recog_sad_pattern): Likewise.
11254 * tree.def (ABSU_EXPR): New.
11256 2018-06-16 Jakub Jelinek <jakub@redhat.com>
11258 PR middle-end/86095
11259 * common.opt (Wunsafe-loop-optimizations): Add Ignore, remove Var,
11260 documented as preserved for backward compatibility only.
11261 * doc/invoke.texi: Remove -Wunsafe-loop-optimizations documentation.
11263 PR rtl-optimization/86108
11264 * bb-reorder.c (create_forwarder_block): Renamed to ...
11265 (create_eh_forwarder_block): ... this. Split OLD_BB after labels and
11266 jump from new landing pad to the second part.
11267 (sjlj_fix_up_crossing_landing_pad, dw2_fix_up_crossing_landing_pad):
11270 2018-06-15 Jakub Jelinek <jakub@redhat.com>
11272 PR middle-end/85878
11273 * expr.c (expand_assignment): Remove now redundant COMPLEX_MODE_P
11274 check from first store_expr, use to_mode instead of GET_MODE (to_rtx).
11275 Only call store_expr for halves if the mode is the same.
11277 PR middle-end/86123
11278 * match.pd ((X / Y) == 0 -> X < Y): Don't transform complex divisions.
11279 Fix up comment formatting.
11281 2018-06-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
11283 * typed-splay-tree.h (typed_splay_tree::remove): New function.
11284 (typed_splay_tree::closure,
11285 typed_splay_tree::inner_foreach_fn, typed_splay_tree::m_inner): Deleted.
11286 (typed_splay_tree::typed_splay_tree,
11287 typed_splay_tree::operator =): Declared private.
11288 (typed_splay_tree::splay_tree_key, typed_splay_tree::splay_tree_value,
11289 typed_splay_tree::splay_tree_node_s, typed_splay_tree::KDEL,
11290 typed_splay_tree::VDEL, typed_splay_tree::splay_tree_delete_helper,
11291 typed_splay_tree::rotate_left, typed_splay_tree::rotate_right,
11292 typed_splay_tree::splay_tree_splay,
11293 typed_splay_tree::splay_tree_foreach_helper,
11294 typed_splay_tree::splay_tree_insert,
11295 typed_splay_tree::splay_tree_remove,
11296 typed_splay_tree::splay_tree_lookup,
11297 typed_splay_tree::splay_tree_predecessor,
11298 typed_splay_tree::splay_tree_successor,
11299 typed_splay_tree::splay_tree_min,
11300 typed_splay_tree::splay_tree_max): Took over from splay-tree.c/.h.
11301 (typed_splay_tree::root, typed_splay_tree::comp,
11302 typed_splay_tree::delete_key,
11303 typed_splay_tree::delete_value): New data members.
11304 * typed-splay-tree.c (selftest::test_str_to_int): Add a test for
11305 typed_splay_tree::remove.
11307 2018-06-15 Matthew Fortune <matthew.fortune@mips.com>
11309 * config/mips/mips.h (ASM_SPEC): Pass through -mcrc, -mno-crc,
11310 -mginv and -mno-ginv to the assembler.
11311 * config/mips/mips.opt (-mcrc): New option.
11312 (-mginv): Likewise.
11313 * doc/invoke.text (-mcrc): Document.
11314 (-mginv): Likewise.
11316 2018-06-15 Nick Clifton <nickc@redhat.com>
11319 * tree.c (escaped_string): New class. Converts an unescaped
11320 string into its escaped equivalent.
11321 (warn_deprecated_use): Use the new class to convert the
11322 deprecation message, if present.
11323 (test_escaped_strings): New self test.
11324 (test_c_tests): Add test_escaped_strings.
11325 * doc/extend.texi (deprecated): Add a note that the
11326 deprecation message is affected by the -fmessage-length
11327 option, and that control characters will be escaped.
11328 (#pragma GCC error): Document this pragma.
11329 (#pragma GCC warning): Likewise.
11330 * doc/invoke.texi (-fmessage-length): Document this option's
11331 effect on the #warning and #error preprocessor directives and
11332 the deprecated attribute.
11334 2018-06-15 Richard Biener <rguenther@suse.de>
11336 * tree-vect-slp.c (vect_slp_bb): Dump MSG_OPTIMIZED_LOCATIONS
11337 here, also noting vector size used.
11338 * tree-vectorizer.c (vectorize_loops): Adjust. Note vector
11339 size used in MSG_OPTIMIZED_LOCATIONS dump.
11340 (pass_slp_vectorize::execute): Adjust.
11342 2018-06-15 Claudiu Zissulescu <claziss@synopsys.com>
11345 * config/arc/arc.c (arc_return_address_register): Fix
11348 2018-06-15 Richard Biener <rguenther@suse.de>
11350 PR middle-end/86159
11351 * tree-cfg.c (gimplify_build3): Do not strip sign conversions,
11352 leave useless conversion stripping to force_gimple_operand_gsi.
11353 (gimplify_build2): Likewise.
11354 (gimplify_build1): Likewise.
11356 2018-06-15 Richard Biener <rguenther@suse.de>
11358 PR middle-end/86076
11359 * tree-cfg.c (move_stmt_op): unshare invariant addresses
11360 before adjusting their block.
11362 2018-06-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
11364 * config.gcc (riscv*-*-elf* | riscv*-*-rtems*): Use custom
11365 multilibs for *-*-rtems*.
11366 * config/riscv/t-rtems: New file.
11368 2018-06-14 Jakub Jelinek <jakub@redhat.com>
11370 PR middle-end/86122
11371 * match.pd ((A +- CST1) +- CST2): Punt if last resort
11372 unsigned_type_for returns NULL.
11375 * lower-subreg.c (find_decomposable_subregs): Don't decompose float
11376 subregs of multi-word pseudos unless the float mode has word size.
11378 2018-06-14 Richard Biener <rguenther@suse.de>
11380 PR middle-end/86139
11381 * tree-vect-generic.c (build_word_mode_vector_type): Remove
11382 duplicate and harmful type_hash_canon.
11383 * tree.c (type_hash_canon): Assert we didn't find ourselves.
11385 2018-06-14 Richard Biener <rguenther@suse.de>
11388 * tree-ssa-struct-alias.c (create_variable_info_for): Handle
11391 2018-06-14 Sebastian Huber <sebastian.huber@embedded-brains.de>
11393 * config/rtems.h (STDINT_LONG32): Define.
11395 2018-06-13 Matthew Fortune <matthew.fortune@mips.com>
11396 Prachi Godbole <prachi.godbole@imgtec.com>
11398 * config/mips/mips-cpus.def: Define P6600.
11399 * config/mips/mips-tables.opt: Regenerate.
11400 * config/mips/mips.c (mips_ucbranch_type): New enum.
11401 (mips_rtx_cost_data): Add support for P6600.
11402 (mips_issue_rate): Likewise.
11403 (mips_multipass_dfa_lookahead): Likewise.
11404 (mips_avoid_hazard): Likewise.
11405 (mips_reorg_process_insns): Likewise.
11406 (mips_classify_branch_p6600): New function.
11407 * config/mips/mips.h (TUNE_P6600): New define.
11408 (MIPS_ISA_LEVEL_SPEC): Infer mips64r6 from p6600.
11409 (ENABLE_LD_ST_PAIRS): Enable load/store bonding for p6600.
11410 * config/mips/mips.md: Include p6600.md.
11411 (processor): Add p6600.
11412 * config/mips/p6600.md: New file.
11413 * doc/invoke.texi: Add p6600 to supported architectures.
11415 2018-06-13 Martin Sebor <msebor@redhat.com>
11417 PR tree-optimization/86114
11418 * gimple-fold.c (gimple_fold_builtin_strlen): Only handle LHS
11420 * tree-ssa-strlen.c (maybe_set_strlen_range): Same.
11422 2018-06-13 Richard Biener <rguenther@suse.de>
11424 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern):
11425 Properly set vector type of the intermediate stmt.
11426 * tree-vect-stmts.c (vectorizable_operation): The destination
11427 var always has vectype_out type.
11429 2018-06-13 Jeff Law <law@redhat.com>
11431 * config/rl78/rl78.c (move_elim_pass): Use TDF_NONE rather than
11432 integer 0 for argument to print_rtl_with_bb.
11433 (rl78_reorg): Likewise.
11435 2018-06-13 David Malcolm <dmalcolm@redhat.com>
11437 * config/arc/arc.c (hwloop_optimize): Strengthen local "end_label"
11438 from rtx to rtx_insn *.
11439 * config/bfin/bfin.c (hwloop_optimize): Likewise for local
11441 (add_sched_insns_for_speculation): Likewise for local "target",
11442 converting usage of JUMP_LABEL to JUMP_LABEL_AS_INSN.
11443 * config/c6x/c6x.c (reorg_split_calls): Strengthen param "call_labels"
11444 from rtx_insn ** to rtx_code_label **.
11445 (reorg_emit_nops): Likewise.
11446 (c6x_reorg): Likewise for local "call_labels".
11447 * config/sh/sh-protos.h (get_dest_uid): Strengthen 1st param from
11449 * config/sh/sh.c (dump_table): Strengthen local "lab" from rtx to
11450 rtx_code_label *, adding safe_as_a <rtx_code_label *> casts to
11451 the loops over LABEL_REFS.
11452 (fixup_addr_diff_vecs): Add as_a <rtx_insn *> to usage of
11454 (barrier_align): Convert usage of JUMP_LABEL to JUMP_LABEL_AS_INSN.
11455 (get_dest_uid): Strengthen param "label" from rtx to rtx_insn *.
11456 (split_branches): Strengthen local "olabel" from rtx to
11457 rtx_insn *, adding a safe_as_a cast.
11458 * emit-rtl.c (next_real_insn): Strengthen param from "rtx"
11460 (add_insn_after): Likewise for first two params.
11461 (add_insn_before): Likewise.
11462 (remove_insn): Likewise for param.
11463 (emit_pattern_before_noloc): Likewise for second and third params.
11464 (emit_jump_insn_before_noloc): Convert NULL_RTX to NULL.
11465 (emit_call_insn_before_noloc): Likewise.
11466 (emit_debug_insn_before_noloc): Strengthen "before" param from "rtx"
11468 (emit_barrier_before): Likewise.
11469 (emit_label_before): Strengthen "label" param from "rtx" to
11470 "rtx_code_label *". Strengthen "before" param from "rtx" to
11472 (emit_insn_after_1): Strengthen "after" param from "rtx" to
11474 (emit_pattern_after_noloc): Likewise.
11475 (emit_insn_after_noloc): Likewise.
11476 (emit_jump_insn_after_noloc): Likewise.
11477 (emit_call_insn_after_noloc): Likewise.
11478 (emit_debug_insn_after_noloc): Likewise.
11479 (emit_barrier_after): Likewise.
11480 (emit_label_after): Likewise for both params.
11481 (emit_pattern_after_setloc): Likewise for "after" param. Convert
11482 "loc" param from "int" to "location_t".
11483 (emit_insn_after_setloc): Likewise.
11484 (emit_jump_insn_after_setloc): Likewise.
11485 (emit_call_insn_after_setloc): Likewise.
11486 (emit_debug_insn_after_setloc): Likewise.
11487 (emit_pattern_before_setloc): Likewise for "before" param. Convert
11488 "loc" param from "int" to "location_t".
11489 (emit_pattern_before): Convert NULL_RTX to NULL.
11490 (emit_insn_before_setloc): Convert "loc" param from "int" to
11492 (emit_jump_insn_before_setloc): Likewise.
11493 (emit_call_insn_before_setloc): Likewise.
11494 (emit_debug_insn_before_setloc): Strengthen "before" param from rtx to
11495 rtx_insn *. Convert "loc" param from "int" to "location_t".
11496 * rtl.h (emit_insn_before_setloc, emit_jump_insn_before_setloc,
11497 emit_call_insn_before_setloc, emit_debug_insn_before_setloc):
11498 Convert 3rd param from "int" to "location_t".
11499 (emit_barrier_before, emit_barrier_after, next_real_insn):
11500 Strengthen param from rtx to rtx_insn *.
11501 (emit_label_before): Strengthen 1st param from "rtx" to
11502 "rtx_code_label *". Strengthen 2nd param from "rtx" to
11504 (emit_insn_after_noloc, emit_jump_insn_after_noloc,
11505 emit_call_insn_after_noloc, emit_debug_insn_after_noloc):
11506 Strengthen 2nd param from "rtx" to "rtx_insn *".
11507 (emit_insn_after_setloc, emit_jump_insn_after_setloc)
11508 emit_call_insn_after_setloc, emit_debug_insn_after_setloc):
11509 Likewise. Convert 3rd param from "int" to "location_t".
11510 (emit_label_after): Strengthen 1st param from "rtx" to
11511 "rtx_code_label *".
11512 (next_real_insn, remove_insn): Strengthen param from "rtx" to
11514 (add_insn_before, add_insn_after): Strengthen 1st and 2nd params
11515 from "rtx" to "rtx_insn *".
11517 2018-06-13 Jan Hubicka <hubicka@gcc.gnu.org>
11519 * cgraph.c (cgraph_node::get_untransformed_body): Dump function
11520 bodies streamed in with -Q.
11521 * dumpfile.c (dump_files): Add lto-stream-out dump file.
11522 * dumpfile.h (tree_dump_index): Add lto_stream_out.
11523 * gimple-streamer-out.c: Include gimple-pretty-print.h
11524 (output_bb): Dump stmts streamed.
11525 * lto-section-out.c: Include print-tree.h
11526 (lto_begin_section): Dump sections created.
11527 (lto_output_decl_index): Dump decl encoded.
11528 * lto-streamer-out.c: Include print-tree.h
11529 (create_output_block): Dump output block created.
11530 (DFS::DFS_write_tree_body): Dump DFS SCCs streamed.
11531 (output_function): Dump function output.
11532 (output_constructor): Dump constructor streamed.
11533 (write_global_stream): Output indexes encoded.
11534 (produce_asm_for_decls): Dump streams encoded.
11535 * lto-streamer.c (streamer_dump_file): New global var.
11536 * lto-streamer.h (streamer_dump_file): Declare.
11537 * passes.c (ipa_write_summaries): Initialize streamer dump.
11538 * varpool.c (varpool_node::get_constructor): Dump constructors streamed
11541 2018-06-13 Eric Botcazou <ebotcazou@adacore.com>
11544 * config/i386/winnt.c (i386_pe_seh_cold_init): Do not emit negative
11545 offsets for register save directives. Emit a second batch of save
11546 directives, if need be, when the function accesses prior frames.
11548 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
11550 * config/arc/fpu.md (fmasf4): Force operand to register.
11551 (fnmasf4): Likewise.
11553 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
11555 * config/arc/arc-protos.h (arc_pad_return): Remove.
11556 * config/arc/arc.c (machine_function): Remove force_short_suffix
11558 (arc_print_operand): Adjust printing of '&'.
11559 (arc_verify_short): Remove conditional printing of short suffix.
11560 (arc_final_prescan_insn): Remove reference to size_reason.
11561 (pad_return): New function.
11562 (arc_reorg): Call pad_return.
11563 (arc_pad_return): Remove.
11564 (arc_init_machine_status): Remove reference to force_short_suffix.
11565 * config/arc/arc.md (vunspec): Add VUNSPEC_ARC_BLOCKAGE.
11566 (attr length): When attribute iscompact is true force to 2
11567 regardless; in the case of maybe check if we want to force the
11568 instruction to have 4 bytes length.
11569 (nopv): Change it to generate 4 byte long nop as well.
11570 (blockage): New pattern.
11571 (simple_return): Remove call to arc_pad_return.
11572 (p_return_i): Likewise.
11574 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
11576 * config/arc/elf.h (LINK_GCC_C_SEQUENCE_SPEC): Define.
11578 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
11580 * config/arc/builtins.def (SYNC): SYNC instruction is valid on all
11583 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
11585 * config/arc/arc.c (atomic_exchangesi): EX instruction is default
11586 for ARC700 and ARCv2.
11588 2018-06-13 Chenghua Xu <paul.hua.gm@gmail.com>
11591 * config/mips/loongson.md (vec_setv4hi): Gen_lowpart for
11592 operands[2] instead of operands[1].
11595 2018-06-12 Richard Sandiford <richard.sandiford@linaro.org>
11597 * lra-constraints.c (simplify_operand_subreg): In the paradoxical
11598 case, check whether the outer register overlaps an unallocatable
11599 register, not just whether it fits the required class.
11601 2018-06-12 Richard Sandiford <richard.sandiford@linaro.org>
11603 * poly-int.h (can_div_trunc_p): Add new overload in which all values
11605 * alias.c (get_addr): Extend CONST_INT handling to poly_int_rtx_p.
11606 (memrefs_conflict_p): Likewise.
11607 (init_alias_analysis): Likewise.
11608 * cfgexpand.c (expand_debug_expr): Likewise.
11609 * combine.c (combine_simplify_rtx, force_int_to_mode): Likewise.
11610 * cse.c (fold_rtx): Likewise.
11611 * explow.c (adjust_stack, anti_adjust_stack): Likewise.
11612 * expr.c (emit_block_move_hints): Likewise.
11613 (clear_storage_hints, push_block, emit_push_insn): Likewise.
11614 (store_expr_with_bounds, reduce_to_bit_field_precision): Likewise.
11615 (emit_group_load_1): Use rtx_to_poly_int64 for group offsets.
11616 (emit_group_store): Likewise.
11617 (find_args_size_adjust): Use strip_offset. Use rtx_to_poly_int64
11618 to read the PRE/POST_MODIFY increment.
11619 * calls.c (store_one_arg): Use strip_offset.
11620 * rtlanal.c (rtx_addr_can_trap_p_1): Extend CONST_INT handling to
11622 (set_noop_p): Use rtx_to_poly_int64 for the elements selected
11624 * simplify-rtx.c (avoid_constant_pool_reference): Use strip_offset.
11625 (simplify_binary_operation_1): Extend CONST_INT handling to
11627 * var-tracking.c (compute_cfa_pointer): Take a poly_int64 rather
11628 than a HOST_WIDE_INT.
11629 (hard_frame_pointer_adjustment): Change from HOST_WIDE_INT to
11631 (adjust_mems, add_stores): Update accodingly.
11632 (vt_canonicalize_addr): Track polynomial offsets.
11633 (emit_note_insn_var_location): Likewise.
11634 (vt_add_function_parameter): Likewise.
11635 (vt_initialize): Likewise.
11637 2018-06-12 Jeff Law <law@redhat.com>
11639 * config.gcc (alpha*-*-freebsd*): Remove.
11640 * config/alpha/freebsd.h: Remove.
11642 2018-06-12 David Malcolm <dmalcolm@redhat.com>
11645 * spellcheck-tree.c (levenshtein_distance): Rename to...
11646 (get_edit_distance): ...this, and update for underlying renaming.
11647 * spellcheck-tree.h (levenshtein_distance): Rename to...
11648 (get_edit_distance): ...this.
11649 * spellcheck.c (levenshtein_distance): Rename to...
11650 (get_edit_distance): ...this. Convert from Levenshtein distance
11651 to Damerau-Levenshtein distance by supporting transpositions of
11652 adjacent characters. Rename "v1" to "v_next" and "v0" to
11654 (selftest::levenshtein_distance_unit_test_oneway): Rename to...
11655 (selftest::test_edit_distance_unit_test_oneway): ...this, and
11656 update for underlying renaming.
11657 (selftest::levenshtein_distance_unit_test): Rename to...
11658 (selftest::test_get_edit_distance_unit): ...this, and update for
11659 underlying renaming.
11660 (selftest::test_find_closest_string): Add example from PR 69968
11661 where transposition helps
11662 (selftest::test_metric_conditions): Update for renaming.
11663 (selftest::test_metric_conditions): Likewise.
11664 (selftest::spellcheck_c_tests): Likewise.
11665 * spellcheck.h (levenshtein_distance): Rename both overloads to...
11666 (get_edit_distance): ...this.
11667 (best_match::consider): Update for renaming.
11669 2018-06-12 Martin Sebor <msebor@redhat.com>
11671 PR tree-optimization/85259
11672 * builtins.c (compute_objsize): Handle constant offsets.
11673 * gimple-ssa-warn-restrict.c (maybe_diag_offset_bounds): Return
11674 true iff a warning has been issued.
11675 * gimple.h (gimple_nonartificial_location): New function.
11676 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Call
11677 gimple_nonartificial_location and handle -Wno-system-headers.
11678 (handle_builtin_stxncpy): Same.
11680 2018-06-12 Martin Sebor <msebor@redhat.com>
11683 * fold-const.c (operand_equal_p): Handle SAVE_EXPR.
11685 2018-06-12 Will Schmidt <will_schmidt@vnet.ibm.com>
11687 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
11688 BUILTIN_VEC_XST entries for pointer to double and long long.
11690 2018-06-12 H.J. Lu <hongjiu.lu@intel.com>
11693 * config/i386/gnu-user.h (TARGET_THREAD_SPLIT_STACK_OFFSET):
11695 * config/i386/gnu-user64.h (TARGET_THREAD_SPLIT_STACK_OFFSET):
11698 2018-06-12 Martin Liska <mliska@suse.cz>
11700 * doc/options.texi: Document IntegerRange.
11702 2018-06-12 Martin Liska <mliska@suse.cz>
11704 * config/i386/i386.opt: Make MPX-related options as Deprecated.
11705 * opt-functions.awk: Handle Deprecated flag.
11706 * opts-common.c (decode_cmdline_option): Handle cl_deprecated
11708 (read_cmdline_option): Report warning for a deprecated option.
11709 * opts.h (struct cl_option): Add new field cl_deprecated.
11710 (CL_ERR_DEPRECATED): New.
11712 2018-06-12 Martin Liska <mliska@suse.cz>
11714 * doc/options.texi: Document Deprecated option flag.
11716 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
11718 * config/arc/arc-arch.h (arc_extras): New enum.
11719 (arc_cpu_t):Add field extra.
11720 (arc_cpu_types): Consider the extras.
11721 * config/arc/arc-cpus.def: Add extras info.
11722 * config/arc/arc-opts.h (processor_type): Consider extra field.
11723 * config/arc/arc.c (arc_override_options): Handle extra field.
11725 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
11727 * config/arc/arc-arch.h: Update ARC_OPTX macro.
11728 * config/arc/arc-options.def (ARC_OPTX): Introduce a new doc
11730 * config/arc/arc.c (arc_init): Update pic warning.
11731 (irq_range): Update irq range parsing warnings.
11732 (arc_override_options): Update various warning messages.
11733 (arc_handle_aux_attribute): Likewise.
11735 2018-06-12 Robert Suchanek <robert.suchanek@mips.com>
11737 * config/mips/i6400.md (i6400_fpu_fadd): Remove frint.
11739 2018-06-12 Jozef Lawrynowicz <jozef.l@mittosystems.com>
11741 * doc/sourcebuild.texi: Document usage of line number 0 in verify
11742 compiler messages directives.
11744 2018-06-12 Matthew Fortune <mfortune@gmail.com>
11746 * config/mips/mips-cpus.def: New MIPS_CPU for i6500.
11747 * config/mips/mips-tables.opt: Regenerate.
11748 * config/mips/mips.h (MIPS_ISA_LEVEL_SPEC): Mark i6500 as
11750 * doc/invoke.texi: Document -march=i6500.
11752 2018-06-12 Prachi Godbole <prachi.godbole@imgtec.com>
11754 * config/mips/i6400.md (i6400_gpmuldiv): Remove cpu_unit.
11755 (i6400_gpmul): Add cpu_unit.
11756 (i6400_gpdiv): Likewise.
11757 (i6400_msa_add_d): Update reservations.
11758 (i6400_msa_int_add) Likewise.
11759 (i6400_msa_short_logic3) Likewise.
11760 (i6400_msa_short_logic2) Likewise.
11761 (i6400_msa_short_logic) Likewise.
11762 (i6400_msa_move) Likewise.
11763 (i6400_msa_cmp) Likewise.
11764 (i6400_msa_short_float2) Likewise.
11765 (i6400_msa_div_d) Likewise.
11766 (i6400_msa_long_logic1) Likewise.
11767 (i6400_msa_long_logic2) Likewise.
11768 (i6400_msa_mult) Likewise.
11769 (i6400_msa_long_float2) Likewise.
11770 (i6400_msa_long_float4) Likewise.
11771 (i6400_msa_long_float5) Likewise.
11772 (i6400_msa_long_float8) Likewise.
11773 (i6400_fpu_fadd): Include frint type.
11774 (i6400_fpu_store): New define_insn_reservation.
11775 (i6400_fpu_load): Likewise.
11776 (i6400_fpu_move): Likewise.
11777 (i6400_fpu_fcmp): Likewise.
11778 (i6400_fpu_fmadd): Likewise.
11779 (i6400_int_mult): Include imul3nc type and update reservation.
11780 (i6400_int_div): Include idiv3 type and update reservation.
11781 (i6400_int_load): Update to check type not move_type.
11782 (i6400_int_store): Likewise.
11783 (i6400_int_prefetch): Set zero latency.
11785 2018-06-12 Eric Botcazou <ebotcazou@adacore.com>
11787 * gcc.c: Document new %@{...} sequence.
11788 (LINK_COMMAND_SPEC): Use it for the -L switches.
11789 (cpp_unique_options): Use it for the -I switches.
11790 (at_file_argbuf): New global variable.
11791 (in_at_file): Likewise.
11792 (alloc_args): Create at_file_argbuf.
11793 (clear_args): Truncate at_file_argbuf.
11794 (store_arg): If in_at_file, push the argument onto at_file_argbuf.
11795 (open_at_file): New function.
11796 (close_at_file): Likewise.
11797 (create_at_file): Delete.
11798 (do_spec_1) <'i'>: Use open_at_file/close_at_file.
11801 (validate_switches_from_spec): Deal with %@{...} sequence.
11802 (validate_switches): Likewise.
11803 (driver::finalize): Call clear_args.
11805 2018-06-11 Rasmus Villemoes <rasmus.villemoes@prevas.dk>
11807 * config/vx-common.h (USE_TM_CLONE_REGISTRY): #define to 0.
11809 2018-06-11 Martin Sebor <msebor@redhat.com>
11811 * doc/invoke.texi (-Wall): List -Wc++17-compat.
11812 (Wno-class-memaccess): Add @opindex.
11813 (Wno-templates, Wno-multiple-inheritance): Same.
11814 (Wno-virtual-inheritance, Wno-namespaces): Same.
11815 (Wno-pedantic, Wno-chkp, Wcoverage-mismatch): Same.
11816 (Wno-format-overflow, Wno-format-truncation): Same.
11817 (Wno-unused-local-typedefs, Walloc-size-larger-than): Same.
11818 (Wno-alloc-size-larger-than, Wframe-larger-than): Same
11819 (Wno-stack-usage, Wno-c++-compat, Wno-c++11-compat): Same.
11820 (Wno-c++14-compat, Wno-c++17-compat, Wno-openmp-simd): Same.
11821 (Wno-unsuffixed-float-constants, Wno-addr-space-convert): Same.
11822 (Wno-misspelled-isr): Same.
11824 2018-06-11 Martin Sebor <msebor@redhat.com>
11826 * PR tree-optimization/86083
11827 * tree-ssa-strlen.c (handle_char_store): Use tree_expr_nonzero_p.
11829 2018-06-11 Zhouyi Zhou <zhouzhouyi@gmail.com>
11831 * tree-eh.c (lower_eh_constructs_2): Add a comma to comment.
11833 2018-06-11 Segher Boessenkool <segher@kernel.crashing.org>
11836 * config/rs6000/rs6000.md (*movdi_internal32): Put constraint modifiers
11837 on the correct operand.
11838 (*movdi_internal64): Ditto.
11840 2018-06-11 Martin Liska <mliska@suse.cz>
11842 PR tree-optimization/86089
11843 * tree-ssa-strlen.c (get_string_length): Move back removed hunk.
11845 2018-06-11 Julia Koval <julia.koval@intel.com>
11847 * config/i386/vaesintrin.h (_mm_aesdec_epi128, _mm_aesdeclast_epi128)
11848 _mm_aesenc_epi128, _mm_aesenclast_epi128: Remove.
11849 * config/i386/vpclmulqdqintrin.h (_mm_clmulepi64_epi128): Remove.
11851 2018-06-11 Olivier Hainque <hainque@adacore.com>
11853 * dwarf2out.c (gen_compile_unit_die): Fallout to DW_LANG_Ada83
11854 for Ada with strict dwarf2.
11856 2018-06-08 Peter Bergner <bergner@vnet.ibm.com>
11859 * config/rs6000/rs6000.c (mem_operand_gpr): Enable PRE_INC and PRE_DEC
11862 2018-06-08 Jan Hubicka <hubicka@ucw.cz>
11864 * dumpfile.c (FIRST_ME_AUTO_NUMBERED_DUMP): Bump to 4.
11866 2018-06-08 David Edelsohn <dje.gcc@gmail.com>
11868 * config/rs6000/rs6000.c (rs6000_passes_ieee128): Protect with #if
11871 2018-06-08 Martin Liska <mliska@suse.cz>
11873 * tree-cfg.h (debug_function): Fix argument type to match
11876 2018-06-08 Martin Liska <mliska@suse.cz>
11878 * config/powerpcspe/powerpcspe.c (rs6000_xcoff_visibility):
11879 Remove usage of MPX-related (and removed) fields.
11880 * config/rs6000/rs6000.c (rs6000_xcoff_visibility): Likewise.
11882 2018-06-08 David Malcolm <dmalcolm@redhat.com>
11884 * cfg.c (debug): Use TDF_NONE rather than 0.
11885 * cfghooks.c (debug): Likewise.
11886 * dumpfile.c (DUMP_FILE_INFO): Likewise; also for OPTGROUP.
11887 (struct dump_option_value_info): Convert to...
11888 (struct kv_pair): ...this template type.
11889 (dump_options): Convert to kv_pair<dump_flags_t>; use TDF_NONE
11891 (optinfo_verbosity_options): Likewise.
11892 (optgroup_options): Convert to kv_pair<optgroup_flags_t>; use
11894 (gcc::dump_manager::dump_register): Use optgroup_flags_t rather
11895 than int for "optgroup_flags" param.
11896 (dump_generic_expr_loc): Use dump_flags_t rather than int for
11898 (dump_dec): Likewise.
11899 (dump_finish): Use TDF_NONE rather than 0.
11900 (gcc::dump_manager::opt_info_enable_passes): Use optgroup_flags_t
11901 rather than int for "optgroup_flags" param. Use TDF_NONE rather
11902 than 0. Update for change to option_ptr.
11903 (opt_info_switch_p_1): Convert "optgroup_flags" param from int *
11904 to optgroup_flags_t *. Use TDF_NONE and OPTGROUP_NONE rather than
11905 0. Update for changes to optinfo_verbosity_options and
11907 (opt_info_switch_p): Convert optgroup_flags from int to
11909 (dump_basic_block): Use dump_flags_t rather than int
11910 for "dump_kind" param.
11911 * dumpfile.h (TDF_ADDRESS, TDF_SLIM, TDF_RAW, TDF_DETAILS,
11912 TDF_STATS, TDF_BLOCKS, TDF_VOPS, TDF_LINENO, TDF_UID)
11913 TDF_STMTADDR, TDF_GRAPH, TDF_MEMSYMS, TDF_RHS_ONLY, TDF_ASMNAME,
11914 TDF_EH, TDF_NOUID, TDF_ALIAS, TDF_ENUMERATE_LOCALS, TDF_CSELIB,
11915 TDF_SCEV, TDF_GIMPLE, TDF_FOLDING, MSG_OPTIMIZED_LOCATIONS,
11916 MSG_MISSED_OPTIMIZATION, MSG_NOTE, MSG_ALL, TDF_COMPARE_DEBUG,
11917 TDF_NONE): Convert from macros to...
11918 (enum dump_flag): ...this new enum.
11919 (dump_flags_t): Update to use enum.
11920 (operator|, operator&, operator~, operator|=, operator&=):
11921 Implement for dump_flags_t.
11922 (OPTGROUP_NONE, OPTGROUP_IPA, OPTGROUP_LOOP, OPTGROUP_INLINE,
11923 OPTGROUP_OMP, OPTGROUP_VEC, OPTGROUP_OTHER, OPTGROUP_ALL):
11924 Convert from macros to...
11925 (enum optgroup_flag): ...this new enum.
11926 (optgroup_flags_t): New typedef.
11927 (operator|, operator|=): Implement for optgroup_flags_t.
11928 (struct dump_file_info): Convert field "alt_flags" to
11929 dump_flags_t. Convert field "optgroup_flags" to
11931 (dump_basic_block): Use dump_flags_t rather than int for param.
11932 (dump_generic_expr_loc): Likewise.
11933 (dump_dec): Likewise.
11934 (dump_register): Convert param "optgroup_flags" to
11936 (opt_info_enable_passes): Likewise.
11937 * early-remat.c (early_remat::dump_edge_list): Use TDF_NONE rather
11939 * gimple-pretty-print.c (debug): Likewise.
11940 * gimple-ssa-store-merging.c (bswap_replace): Likewise.
11941 (merged_store_group::apply_stores): Likewise.
11942 * gimple-ssa-strength-reduction.c (insert_initializers): Likewise.
11943 * gimple.c (verify_gimple_pp): Likewise.
11944 * graphite-poly.c (print_pbb_body): Likewise.
11945 * passes.c (pass_manager::register_one_dump_file): Convert
11946 local "optgroup_flags" to optgroup_flags_t.
11947 * print-tree.c (print_node): Use TDF_NONE rather than 0.
11949 (debug_body): Likewise.
11950 * tree-pass.h (struct pass_data): Convert field "optgroup_flags"
11951 to optgroup_flags_t.
11952 * tree-pretty-print.c (print_struct_decl): Use TDF_NONE rather
11954 * tree-ssa-math-opts.c (convert_mult_to_fma_1): Likewise.
11955 (convert_mult_to_fma): Likewise.
11956 * tree-ssa-reassoc.c (undistribute_ops_list): Likewise.
11957 * tree-ssa-sccvn.c (vn_eliminate): Likewise.
11958 * tree-vect-data-refs.c (dump_lower_bound): Convert param
11959 "dump_kind" to dump_flags_t.
11961 2018-06-08 Segher Boessenkool <segher@kernel.crashing.org>
11963 * config/rs6000/rs6000.c (min, max): Delete.
11965 2018-06-08 Segher Boessenkool <segher@kernel.crashing.org>
11967 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete mention of
11968 -mabi=spe and -mabi=no-spe.
11970 2018-06-08 Martin Liska <mliska@suse.cz>
11972 * ipa-pure-const.c (propagate_pure_const): Use ::get at places
11973 where we expect an existing summary.
11975 2018-06-08 Martin Liska <mliska@suse.cz>
11977 * ipa-inline-analysis.c (simple_edge_hints): Use ::get method.
11978 * ipa-inline.h (estimate_edge_growth): Likewise.
11980 2018-06-08 Martin Liska <mliska@suse.cz>
11982 * cgraph.c (function_version_hasher::hash): Use
11983 cgraph_node::get_uid ().
11984 (function_version_hasher::equal):
11985 * cgraph.h (cgraph_node::get_uid): New method.
11986 * ipa-inline.c (update_caller_keys): Use
11987 cgraph_node::get_uid ().
11988 (update_callee_keys): Likewise.
11989 * ipa-utils.c (searchc): Likewise.
11990 (ipa_reduced_postorder): Likewise.
11991 * lto-cgraph.c (input_node): Likewise.
11992 * passes.c (is_pass_explicitly_enabled_or_disabled): Likewise.
11993 * symbol-summary.h (symtab_insertion): Likewise.
11994 (symtab_removal): Likewise.
11995 (symtab_duplication): Likewise.
11996 * tree-pretty-print.c (dump_function_header): Likewise.
11997 * tree-sra.c (convert_callers_for_node): Likewise.
11999 2018-06-08 Martin Liska <mliska@suse.cz>
12001 * cgraph.c (symbol_table::create_edge): Always assign a new
12003 (symbol_table::free_edge): Do not recycle numbers.
12004 * cgraph.h (cgraph_edge::get): New method.
12005 * symbol-summary.h (symtab_removal): Use it.
12006 (symtab_duplication): Likewise.
12007 (call_summary::hashable_uid): Remove.
12009 2018-06-08 Martin Liska <mliska@suse.cz>
12011 * ipa-inline-analysis.c (inline_edge_removal_hook): Remove.
12012 (initialize_growth_caches): Remove.
12013 (free_growth_caches): Likewise.
12014 (do_estimate_edge_time): Use edge_growth_cache.
12015 (do_estimate_edge_size): Likewise.
12016 (do_estimate_edge_hints): Likewise.
12017 * ipa-inline.c (reset_edge_caches): Likewise.
12018 (recursive_inlining): Likewise.
12019 (inline_small_functions): Likewise.
12020 * ipa-inline.h (initialize_growth_caches): Remove.
12021 (estimate_edge_size): Likewise.
12022 (estimate_edge_time): Likewise.
12023 (estimate_edge_hints): Likewise.
12024 (reset_edge_growth_cache): Likewise.
12025 * symbol-summary.h (call_summary::remove): New method.
12027 2018-06-08 Martin Liska <mliska@suse.cz>
12029 * ipa-cp.c (class edge_clone_summary): New summary.
12030 (grow_edge_clone_vectors): Remove.
12031 (ipcp_edge_duplication_hook): Remove.
12032 (class edge_clone_summary_t): New call_summary class.
12033 (ipcp_edge_removal_hook): Remove.
12034 (edge_clone_summary_t::duplicate): New function.
12035 (get_next_cgraph_edge_clone): Use edge_clone_summaries.
12036 (create_specialized_node): Likewise.
12037 (ipcp_driver): Initialize edge_clone_summaries and do not
12040 2018-06-08 Martin Liska <mliska@suse.cz>
12042 * symbol-summary.h (get): New function.
12043 (call_summary::m_initialize_when_cloning): New class member.
12045 2018-06-08 Martin Liska <mliska@suse.cz>
12047 * cgraph.c (cgraph_node::remove): Do not recycle uid.
12048 * cgraph.h (symbol_table::release_symbol): Do not pass uid.
12049 (symbol_table::allocate_cgraph_symbol): Do not set uid.
12050 * passes.c (uid_hash_t): Record removed_nodes by their uids.
12051 (remove_cgraph_node_from_order): Use the removed_nodes set.
12052 (do_per_function_toporder): Likwise.
12053 * symbol-summary.h (symtab_insertion): Use cgraph_node::uid
12054 instead of summary_uid.
12055 (symtab_removal): Likewise.
12056 (symtab_duplication): Likewise.
12058 2018-06-08 Martin Liska <mliska@suse.cz>
12060 * ipa-cp.c (ipcp_store_bits_results): Use
12061 ipcp_transformation_sum.
12062 (ipcp_store_vr_results): Likewise.
12063 * ipa-prop.c (ipcp_grow_transformations_if_necessary): Renamed
12065 (ipcp_transformation_initialize): ... this.
12066 (ipa_set_node_agg_value_chain):
12067 (ipa_node_params_t::duplicate): Use ipcp_transformation_sum.
12068 (write_ipcp_transformation_info): Likewise.
12069 (read_ipcp_transformation_info): Likewise.
12070 (ipcp_update_bits): Likewise.
12071 (ipcp_update_vr): Likewise.
12072 (ipcp_transform_function): Likewise.
12073 * ipa-prop.h: Rename ipcp_transformation_summary to
12074 ipcp_transformation.
12075 (class ipcp_transformation_t): New function summary.
12076 (ipcp_get_transformation_summary): Use ipcp_transformation_sum.
12077 (ipa_get_agg_replacements_for_node): Likewise.
12079 2018-06-08 Martin Liska <mliska@suse.cz>
12081 * ipa-pure-const.c (struct funct_state_d): Do it class instead
12083 (class funct_state_summary_t): New function_summary class.
12084 (has_function_state): Remove.
12085 (get_function_state): Likewise.
12086 (set_function_state): Likewise.
12087 (add_new_function): Likewise.
12088 (funct_state_summary_t::insert): New function.
12089 (duplicate_node_data): Remove.
12090 (remove_node_data): Remove.
12091 (funct_state_summary_t::duplicate): New function.
12092 (register_hooks): Create new funct_state_summaries.
12093 (pure_const_generate_summary): Use it.
12094 (pure_const_write_summary): Likewise.
12095 (pure_const_read_summary): Likewise.
12096 (propagate_pure_const): Likewise.
12097 (propagate_nothrow): Likewise.
12098 (dump_malloc_lattice): Likewise.
12099 (propagate_malloc): Likewise.
12100 (execute): Do not register hooks, just remove summary
12102 (pass_ipa_pure_const::pass_ipa_pure_const): Simplify
12105 2018-06-08 Martin Liska <mliska@suse.cz>
12107 * ipa-reference.c (remove_node_data): Remove.
12108 (duplicate_node_data): Likewise.
12109 (class ipa_ref_var_info_summary_t): New class.
12110 (class ipa_ref_opt_summary_t): Likewise.
12111 (get_reference_vars_info): Use ipa_ref_var_info_summaries.
12112 (get_reference_optimization_summary): Use
12113 ipa_ref_opt_sum_summaries.
12114 (set_reference_vars_info): Remove.
12115 (set_reference_optimization_summary): Likewise.
12116 (ipa_init): Create summaries.
12117 (init_function_info): Use function summary.
12118 (ipa_ref_opt_summary_t::duplicate): New function.
12119 (ipa_ref_opt_summary_t::remove): New function.
12120 (get_read_write_all_from_node): Fix GNU coding style.
12121 (propagate): Use function summary.
12122 (write_node_summary_p): Fix GNU coding style.
12123 (stream_out_bitmap): Likewise.
12124 (ipa_reference_read_optimization_summary): Use function summary.
12125 (ipa_reference_c_finalize): Do not release hooks.
12127 2018-06-08 Martin Liska <mliska@suse.cz>
12129 * ipa-fnsummary.c (dump_ipa_call_summary): Use ::get method.
12130 (analyze_function_body): Extract multiple calls of get_create.
12131 * ipa-inline-analysis.c (simple_edge_hints): Likewise.
12132 * ipa-inline.c (recursive_inlining): Use ::get method.
12133 * ipa-inline.h (estimate_edge_growth): Likewise.
12135 2018-06-08 Martin Liska <mliska@suse.cz>
12137 * hsa-common.h (enum hsa_function_kind): Rename HSA_NONE to
12139 (hsa_function_summary::hsa_function_summary): Use the new enum
12141 (hsa_gpu_implementation_p): Use hsa_summaries::get.
12142 * hsa-gen.c (hsa_get_host_function): Likewise.
12143 (get_brig_function_name): Likewise.
12144 * ipa-hsa.c (process_hsa_functions): Likewise.
12145 (ipa_hsa_write_summary): Likewise.
12146 * symbol-summary.h (symtab_duplication): Use ::get function/
12147 (get): New function.
12149 2018-06-08 Martin Liska <mliska@suse.cz>
12151 * config/i386/i386.c (ix86_can_inline_p): Use get_create instead
12153 * hsa-common.c (hsa_summary_t::link_functions): Likewise.
12154 (hsa_register_kernel): Likewise.
12155 * hsa-common.h (hsa_gpu_implementation_p): Likewise.
12156 * hsa-gen.c (hsa_get_host_function): Likewise.
12157 (get_brig_function_name): Likewise.
12158 (generate_hsa): Likewise.
12159 (pass_gen_hsail::execute): Likewise.
12160 * ipa-cp.c (ipcp_cloning_candidate_p): Likewise.
12161 (devirtualization_time_bonus): Likewise.
12162 (ipcp_propagate_stage): Likewise.
12163 * ipa-fnsummary.c (redirect_to_unreachable): Likewise.
12164 (edge_set_predicate): Likewise.
12165 (evaluate_conditions_for_known_args): Likewise.
12166 (evaluate_properties_for_edge): Likewise.
12167 (ipa_fn_summary::reset): Likewise.
12168 (ipa_fn_summary_t::duplicate): Likewise.
12169 (dump_ipa_call_summary): Likewise.
12170 (ipa_dump_fn_summary): Likewise.
12171 (analyze_function_body): Likewise.
12172 (compute_fn_summary): Likewise.
12173 (estimate_edge_devirt_benefit): Likewise.
12174 (estimate_edge_size_and_time): Likewise.
12175 (estimate_calls_size_and_time): Likewise.
12176 (estimate_node_size_and_time): Likewise.
12177 (inline_update_callee_summaries): Likewise.
12178 (remap_edge_change_prob): Likewise.
12179 (remap_edge_summaries): Likewise.
12180 (ipa_merge_fn_summary_after_inlining): Likewise.
12181 (ipa_update_overall_fn_summary): Likewise.
12182 (read_ipa_call_summary): Likewise.
12183 (inline_read_section): Likewise.
12184 (write_ipa_call_summary): Likewise.
12185 (ipa_fn_summary_write): Likewise.
12186 (ipa_free_fn_summary): Likewise.
12187 * ipa-hsa.c (process_hsa_functions): Likewise.
12188 (ipa_hsa_write_summary): Likewise.
12189 (ipa_hsa_read_section): Likewise.
12190 * ipa-icf.c (sem_function::merge): Likewise.
12191 * ipa-inline-analysis.c (simple_edge_hints): Likewise.
12192 (do_estimate_edge_time): Likewise.
12193 (estimate_size_after_inlining): Likewise.
12194 (estimate_growth): Likewise.
12195 (growth_likely_positive): Likewise.
12196 * ipa-inline-transform.c (clone_inlined_nodes): Likewise.
12197 (inline_call): Likewise.
12198 * ipa-inline.c (caller_growth_limits): Likewise.
12199 (can_inline_edge_p): Likewise.
12200 (can_inline_edge_by_limits_p): Likewise.
12201 (compute_uninlined_call_time): Likewise.
12202 (compute_inlined_call_time): Likewise.
12203 (want_inline_small_function_p): Likewise.
12204 (edge_badness): Likewise.
12205 (update_caller_keys): Likewise.
12206 (update_callee_keys): Likewise.
12207 (recursive_inlining): Likewise.
12208 (inline_small_functions): Likewise.
12209 (inline_to_all_callers_1): Likewise.
12210 (dump_overall_stats): Likewise.
12211 (early_inline_small_functions): Likewise.
12212 (early_inliner): Likewise.
12213 * ipa-inline.h (estimate_edge_growth): Likewise.
12214 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
12215 * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
12216 * ipa-prop.h (IPA_NODE_REF): Likewise.
12217 (IPA_EDGE_REF): Likewise.
12218 * ipa-pure-const.c (malloc_candidate_p): Likewise.
12219 (propagate_malloc): Likewise.
12220 * ipa-split.c (execute_split_functions): Likewise.
12221 * symbol-summary.h: Rename get to get_create.
12223 (get_create): Likewise.
12224 * tree-sra.c (ipa_sra_preliminary_function_checks): Likewise.
12226 2018-06-08 Martin Liska <mliska@suse.cz>
12228 * symbol-summary.h (release): Move definition out of class
12230 (symtab_removal): Likewise.
12231 (symtab_duplication): Likewise.
12233 2018-06-08 Martin Liska <mliska@suse.cz>
12235 * symbol-summary.h (function_summary): Move constructor
12236 implementation out of class declaration.
12237 (release): Likewise.
12238 (symtab_insertion): Likewise.
12239 (symtab_removal): Likewise.
12240 (symtab_duplication): Likewise.
12243 2018-06-08 Martin Liska <mliska@suse.cz>
12245 * Makefile.in: Remove support for MPX (macros, related functions,
12246 fields in cgraph_node, ...).
12247 * builtin-types.def (BT_BND): Likewise.
12248 (BT_FN_BND_CONST_PTR): Likewise.
12249 (BT_FN_CONST_PTR_BND): Likewise.
12250 (BT_FN_VOID_PTR_BND): Likewise.
12251 (BT_FN_BND_CONST_PTR_SIZE): Likewise.
12252 (BT_FN_VOID_CONST_PTR_BND_CONST_PTR): Likewise.
12253 * builtins.c (expand_builtin_memcpy_with_bounds): Likewise.
12254 (expand_builtin_mempcpy_with_bounds): Likewise.
12255 (expand_builtin_memset_with_bounds): Likewise.
12256 (expand_builtin_memset_args): Likewise.
12257 (std_expand_builtin_va_start): Likewise.
12258 (expand_builtin): Likewise.
12259 (expand_builtin_with_bounds): Likewise.
12260 * builtins.def (DEF_BUILTIN_CHKP): Likewise.
12261 (DEF_LIB_BUILTIN_CHKP): Likewise.
12262 (DEF_EXT_LIB_BUILTIN_CHKP): Likewise.
12263 (DEF_CHKP_BUILTIN): Likewise.
12264 (BUILT_IN_MEMCPY): Likewise.
12265 (BUILT_IN_MEMMOVE): Likewise.
12266 (BUILT_IN_MEMPCPY): Likewise.
12267 (BUILT_IN_MEMSET): Likewise.
12268 (BUILT_IN_STPCPY): Likewise.
12269 (BUILT_IN_STRCAT): Likewise.
12270 (BUILT_IN_STRCHR): Likewise.
12271 (BUILT_IN_STRCPY): Likewise.
12272 (BUILT_IN_STRLEN): Likewise.
12273 (BUILT_IN_MEMCPY_CHK): Likewise.
12274 (BUILT_IN_MEMMOVE_CHK): Likewise.
12275 (BUILT_IN_MEMPCPY_CHK): Likewise.
12276 (BUILT_IN_MEMSET_CHK): Likewise.
12277 (BUILT_IN_STPCPY_CHK): Likewise.
12278 (BUILT_IN_STRCAT_CHK): Likewise.
12279 (BUILT_IN_STRCPY_CHK): Likewise.
12280 * calls.c (store_bounds): Likewise.
12281 (emit_call_1): Likewise.
12282 (special_function_p): Likewise.
12283 (maybe_warn_nonstring_arg): Likewise.
12284 (initialize_argument_information): Likewise.
12285 (finalize_must_preallocate): Likewise.
12286 (compute_argument_addresses): Likewise.
12287 (expand_call): Likewise.
12288 * cfgexpand.c (expand_call_stmt): Likewise.
12289 (expand_return): Likewise.
12290 (expand_gimple_stmt_1): Likewise.
12291 (pass_expand::execute): Likewise.
12292 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
12293 (cgraph_node::remove): Likewise.
12294 (cgraph_node::dump): Likewise.
12295 (cgraph_node::verify_node): Likewise.
12296 * cgraph.h (chkp_function_instrumented_p): Likewise.
12297 (symtab_node::get_alias_target): Likewise.
12298 (cgraph_node::can_remove_if_no_direct_calls_and_refs_p): Likewise.
12299 (cgraph_local_p): Likewise.
12300 * cgraphbuild.c (cgraph_edge::rebuild_edges): Likewise.
12301 (cgraph_edge::rebuild_references): Likewise.
12302 * cgraphunit.c (varpool_node::finalize_decl): Likewise.
12303 (walk_polymorphic_call_targets): Likewise.
12304 (cgraph_node::expand_thunk): Likewise.
12305 (symbol_table::output_weakrefs): Likewise.
12306 * common/config/i386/i386-common.c (OPTION_MASK_ISA2_GENERAL_REGS_ONLY_UNSET): Likewise.
12307 (ix86_handle_option): Likewise.
12308 * config/i386/constraints.md: Likewise.
12309 * config/i386/i386-builtin-types.def (BND): Likewise.
12313 * config/i386/i386-builtin.def (BDESC_END): Likewise.
12314 (BDESC_FIRST): Likewise.
12316 * config/i386/i386-c.c (ix86_target_macros_internal): Likewise.
12317 * config/i386/i386-protos.h (ix86_bnd_prefixed_insn_p): Likewise.
12318 * config/i386/i386.c (enum reg_class): Likewise.
12319 (ix86_target_string): Likewise.
12320 (ix86_option_override_internal): Likewise.
12321 (ix86_conditional_register_usage): Likewise.
12322 (ix86_valid_target_attribute_inner_p): Likewise.
12323 (ix86_set_indirect_branch_type): Likewise.
12324 (ix86_set_current_function): Likewise.
12325 (ix86_function_arg_regno_p): Likewise.
12326 (init_cumulative_args): Likewise.
12327 (ix86_function_arg_advance): Likewise.
12328 (ix86_function_arg): Likewise.
12329 (ix86_pass_by_reference): Likewise.
12330 (ix86_function_value_regno_p): Likewise.
12331 (ix86_function_value_1): Likewise.
12332 (ix86_function_value_bounds): Likewise.
12333 (ix86_return_in_memory): Likewise.
12334 (ix86_setup_incoming_vararg_bounds): Likewise.
12335 (ix86_va_start): Likewise.
12336 (indirect_thunk_need_prefix): Likewise.
12337 (print_reg): Likewise.
12338 (ix86_print_operand): Likewise.
12339 (ix86_expand_call): Likewise.
12340 (ix86_output_function_return): Likewise.
12341 (reg_encoded_number): Likewise.
12342 (BDESC_VERIFYS): Likewise.
12343 (ix86_init_mpx_builtins): Likewise.
12344 (ix86_init_builtins): Likewise.
12345 (ix86_emit_cmove): Likewise.
12346 (ix86_emit_move_max): Likewise.
12347 (ix86_expand_builtin): Likewise.
12348 (ix86_builtin_mpx_function): Likewise.
12349 (ix86_get_arg_address_for_bt): Likewise.
12350 (ix86_load_bounds): Likewise.
12351 (ix86_store_bounds): Likewise.
12352 (ix86_load_returned_bounds): Likewise.
12353 (ix86_store_returned_bounds): Likewise.
12354 (ix86_class_likely_spilled_p): Likewise.
12355 (ix86_hard_regno_mode_ok): Likewise.
12356 (x86_order_regs_for_local_alloc): Likewise.
12357 (ix86_mitigate_rop): Likewise.
12358 (ix86_bnd_prefixed_insn_p): Likewise.
12359 (ix86_mpx_bound_mode): Likewise.
12360 (ix86_make_bounds_constant): Likewise.
12361 (ix86_initialize_bounds): Likewise.
12362 (TARGET_LOAD_BOUNDS_FOR_ARG): Likewise.
12363 (TARGET_STORE_BOUNDS_FOR_ARG): Likewise.
12364 (TARGET_LOAD_RETURNED_BOUNDS): Likewise.
12365 (TARGET_STORE_RETURNED_BOUNDS): Likewise.
12366 (TARGET_CHKP_BOUND_MODE): Likewise.
12367 (TARGET_BUILTIN_CHKP_FUNCTION): Likewise.
12368 (TARGET_CHKP_FUNCTION_VALUE_BOUNDS): Likewise.
12369 (TARGET_CHKP_MAKE_BOUNDS_CONSTANT): Likewise.
12370 (TARGET_CHKP_INITIALIZE_BOUNDS): Likewise.
12371 * config/i386/i386.h (TARGET_MPX): Likewise.
12372 (TARGET_MPX_P): Likewise.
12373 (VALID_BND_REG_MODE): Likewise.
12374 (FIRST_BND_REG): Likewise.
12375 (LAST_BND_REG): Likewise.
12376 (enum reg_class): Likewise.
12377 (BND_REG_P): Likewise.
12378 (BND_REGNO_P): Likewise.
12379 (BNDmode): Likewise.
12380 (ADJUST_INSN_LENGTH): Likewise.
12381 * config/i386/i386.md: Likewise.
12382 * config/i386/i386.opt: Likewise.
12383 * config/i386/linux-common.h (LIBMPX_LIBS): Likewise.
12384 (defined): Likewise.
12385 (LINK_MPX): Likewise.
12386 (MPX_SPEC): Likewise.
12387 (LIBMPX_SPEC): Likewise.
12388 (LIBMPXWRAPPERS_SPEC): Likewise.
12389 (CHKP_SPEC): Likewise.
12390 * config/i386/predicates.md: Likewise.
12391 * dbxout.c (dbxout_type): Likewise.
12392 * doc/extend.texi: Likewise.
12393 * doc/invoke.texi: Likewise.
12394 * doc/md.texi: Likewise.
12395 * doc/tm.texi: Likewise.
12396 * doc/tm.texi.in: Likewise.
12397 * dwarf2out.c (is_base_type): Likewise.
12398 (gen_formal_types_die): Likewise.
12399 (gen_subprogram_die): Likewise.
12400 (gen_type_die_with_usage): Likewise.
12401 (gen_decl_die): Likewise.
12402 (dwarf2out_late_global_decl): Likewise.
12403 * expr.c (expand_assignment): Likewise.
12404 (emit_storent_insn): Likewise.
12405 (store_expr_with_bounds): Likewise.
12406 (store_expr): Likewise.
12407 (expand_expr_real_1): Likewise.
12408 * expr.h (store_expr_with_bounds): Likewise.
12409 * function.c (use_register_for_decl): Likewise.
12410 (struct bounds_parm_data): Likewise.
12411 (assign_parms_augmented_arg_list): Likewise.
12412 (assign_parm_find_entry_rtl): Likewise.
12413 (assign_parm_is_stack_parm): Likewise.
12414 (assign_parm_load_bounds): Likewise.
12415 (assign_bounds): Likewise.
12416 (assign_parms): Likewise.
12417 (expand_function_start): Likewise.
12418 * gcc.c (CHKP_SPEC): Likewise.
12419 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
12420 * gimple-ssa-warn-restrict.c (builtin_access::builtin_access): Likewise.
12421 (wrestrict_dom_walker::check_call): Likewise.
12422 * gimple.c (gimple_build_call_from_tree): Likewise.
12423 * gimple.h (enum gf_mask): Likewise.
12424 (gimple_call_with_bounds_p): Likewise.
12425 (gimple_call_set_with_bounds): Likewise.
12426 * gimplify.c (gimplify_init_constructor): Likewise.
12427 * ipa-cp.c (initialize_node_lattices): Likewise.
12428 (propagate_constants_across_call): Likewise.
12429 (find_more_scalar_values_for_callers_subset): Likewise.
12430 * ipa-hsa.c (process_hsa_functions): Likewise.
12431 * ipa-icf-gimple.c (func_checker::compare_gimple_call): Likewise.
12432 * ipa-icf.c (sem_function::merge): Likewise.
12433 * ipa-inline.c (early_inliner): Likewise.
12434 * ipa-pure-const.c (warn_function_noreturn): Likewise.
12435 (warn_function_cold): Likewise.
12436 (propagate_pure_const): Likewise.
12437 * ipa-ref.h (enum GTY): Likewise.
12438 * ipa-split.c (find_retbnd): Likewise.
12439 (consider_split): Likewise.
12440 (split_function): Likewise.
12441 * ipa-visibility.c (cgraph_externally_visible_p): Likewise.
12442 * ipa.c (walk_polymorphic_call_targets): Likewise.
12443 (symbol_table::remove_unreachable_nodes): Likewise.
12444 (process_references): Likewise.
12445 (cgraph_build_static_cdtor_1): Likewise.
12446 * lto-cgraph.c (lto_output_node): Likewise.
12447 (output_refs): Likewise.
12448 (compute_ltrans_boundary): Likewise.
12449 (input_overwrite_node): Likewise.
12450 (input_node): Likewise.
12451 (input_cgraph_1): Likewise.
12452 * params.def (PARAM_CHKP_MAX_CTOR_SIZE): Likewise.
12453 * passes.c (pass_manager::execute_early_local_passes): Likewise.
12454 (class pass_chkp_instrumentation_passes): Likewise.
12455 (make_pass_chkp_instrumentation_passes): Likewise.
12456 * passes.def: Likewise.
12457 * rtl.h (struct GTY): Likewise.
12458 (CALL_EXPR_WITH_BOUNDS_P): Likewise.
12459 * stor-layout.c (layout_type): Likewise.
12460 * symtab.c: Likewise.
12461 * target.def: Likewise.
12462 * targhooks.c (default_chkp_bound_type): Likewise.
12463 (default_chkp_bound_mode): Likewise.
12464 (default_builtin_chkp_function): Likewise.
12465 (default_chkp_function_value_bounds): Likewise.
12466 (default_chkp_make_bounds_constant): Likewise.
12467 (default_chkp_initialize_bounds): Likewise.
12468 * targhooks.h (default_chkp_bound_type): Likewise.
12469 (default_chkp_bound_mode): Likewise.
12470 (default_builtin_chkp_function): Likewise.
12471 (default_chkp_function_value_bounds): Likewise.
12472 (default_chkp_make_bounds_constant): Likewise.
12473 (default_chkp_initialize_bounds): Likewise.
12474 * toplev.c (compile_file): Likewise.
12475 (process_options): Likewise.
12476 * tree-core.h (DEF_BUILTIN): Likewise.
12477 (DEF_BUILTIN_CHKP): Likewise.
12478 * tree-inline.c (declare_return_variable): Likewise.
12479 (remap_gimple_stmt): Likewise.
12480 (copy_bb): Likewise.
12481 (initialize_inlined_parameters): Likewise.
12482 (expand_call_inline): Likewise.
12483 * tree-pass.h (make_pass_ipa_chkp_versioning): Likewise.
12484 (make_pass_ipa_chkp_early_produce_thunks): Likewise.
12485 (make_pass_ipa_chkp_produce_thunks): Likewise.
12486 (make_pass_chkp): Likewise.
12487 (make_pass_chkp_opt): Likewise.
12488 (make_pass_chkp_instrumentation_passes): Likewise.
12489 * tree-pretty-print.c (dump_generic_node): Likewise.
12490 * tree-ssa-ccp.c (insert_clobber_before_stack_restore): Likewise.
12491 * tree-ssa-dce.c (propagate_necessity): Likewise.
12492 (eliminate_unnecessary_stmts): Likewise.
12493 * tree-ssa-pre.c (create_expression_by_pieces): Likewise.
12494 * tree-ssa-sccvn.c (copy_reference_ops_from_call): Likewise.
12495 * tree-ssa-sccvn.h: Likewise.
12496 * tree-ssa-strlen.c (get_string_length): Likewise.
12497 (valid_builtin_call): Likewise.
12498 (adjust_last_stmt): Likewise.
12499 (handle_builtin_strchr): Likewise.
12500 (handle_builtin_strcpy): Likewise.
12501 (handle_builtin_stxncpy): Likewise.
12502 (handle_builtin_memcpy): Likewise.
12503 (handle_builtin_strcat): Likewise.
12504 (strlen_check_and_optimize_stmt): Likewise.
12505 * tree-stdarg.c (expand_ifn_va_arg_1): Likewise.
12506 * tree-streamer-in.c: Likewise.
12507 * tree-streamer.c (record_common_node): Likewise.
12508 * tree.c (tree_code_size): Likewise.
12509 (wide_int_to_tree_1): Likewise.
12510 (type_contains_placeholder_1): Likewise.
12511 (build_common_tree_nodes): Likewise.
12512 * tree.def (POINTER_BOUNDS_TYPE): Likewise.
12513 * tree.h (POINTER_BOUNDS_TYPE_P): Likewise.
12514 (POINTER_BOUNDS_P): Likewise.
12515 (BOUNDED_TYPE_P): Likewise.
12516 (BOUNDED_P): Likewise.
12517 (CALL_WITH_BOUNDS_P): Likewise.
12518 (pointer_bounds_type_node): Likewise.
12519 * value-prof.c (gimple_ic): Likewise.
12520 * var-tracking.c (vt_add_function_parameters): Likewise.
12521 * varasm.c (make_decl_rtl): Likewise.
12522 (assemble_start_function): Likewise.
12523 (output_constant): Likewise.
12524 (maybe_assemble_visibility): Likewise.
12525 * varpool.c (ctor_for_folding): Likewise.
12526 * chkp-builtins.def: Remove.
12527 * ipa-chkp.c: Remove.
12528 * ipa-chkp.h: Remove.
12529 * rtl-chkp.c: Remove.
12530 * rtl-chkp.h: Remove.
12531 * tree-chkp-opt.c: Remove.
12532 * tree-chkp.c: Remove.
12533 * tree-chkp.h: Remove.
12535 2018-06-07 Carl Love <cel@us.ibm.com>
12537 * config/rs6000/vsx.md (vextract_fp_from_shorth,
12538 vextract_fp_from_shortl): Add BE support.
12540 2018-06-07 Paul Koning <ni1d@arrl.net>
12542 * compare-elim.c (try_merge_compare): Don't merge compare if
12543 address contains a side effect.
12544 (try_eliminate_compare): Likewise.
12546 2018-06-07 Olga Makhotina <olga.makhotina@intel.com>
12548 * config.gcc: Support "tremont".
12549 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "tremont".
12550 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
12552 * config/i386/i386.c (m_TREMONT): Define.
12553 (processor_target_table): Add "tremont".
12554 (PTA_TREMONT): Define.
12555 (ix86_lea_outperforms): Add TARGET_TREMONT.
12556 (get_builtin_code_for_version): Handle PROCESSOR_TREMONT.
12557 (fold_builtin_cpu): Add M_INTEL_TREMONT, replace M_INTEL_GOLDMONT
12558 and M_INTEL_GOLDMONT_PLUS.
12559 (fold_builtin_cpu): Add "tremont".
12560 (ix86_add_stmt_cost): Add TARGET_TREMONT.
12561 (ix86_option_override_internal): Add "tremont".
12562 * config/i386/i386.h (processor_costs): Define TARGET_TREMONT.
12563 (processor_type): Add PROCESSOR_TREMONT.
12564 * config/i386/x86-tune.def: Add m_TREMONT.
12565 * doc/invoke.texi: Add tremont as x86 -march=/-mtune= CPU type.
12567 2018-06-07 Jozef Lawrynowicz <jozef.l@mittosystems.com>
12569 * config/msp430/msp430.c (msp430_mcu_name): Set the "i" in the
12570 symbol defined for msp430i* devices to be lower case.
12572 2018-06-07 Richard Biener <rguenther@suse.de>
12574 * graphite-sese-to-poly.c (extract_affine): Avoid unneded
12575 wrapping. Properly wrap the result of a BIT_NOT_EXPR.
12576 Properly wrap signed arithmetic if overflow wraps.
12578 2018-06-07 Jakub Jelinek <jakub@redhat.com>
12580 PR tree-optimization/69615
12581 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If rhs2 is lhs
12582 of a cast from a same precision integral SSA_NAME in a bb dominated
12583 by first_bb, retry with rhs2 set to the rhs1 of the cast. Don't emit
12584 cast to utype if rhs2 has already a compatible type.
12586 2018-06-07 Richard Biener <rguenther@suse.de>
12588 PR tree-optimization/85935
12589 * graphite-scop-detection.c (find_params_in_bb): Analyze
12590 condition operands with respect to the correct loop. Assert
12591 the analysis doesn't fail.
12593 2018-06-04 Carl Love <cel@us.ibm.com>
12595 * config/rs6000/vsx.md (first_match_index_<mode>): Calculate index
12596 using natural element order. Use gen_lshrsi3 instead of gen_ashrsi3
12597 as it is slightly cheaper.
12598 (first_match_or_eos_index_<mode>):
12599 Calculate index using natural element order.
12600 (first_match_index_<mode>):
12601 Calculate index using natural element order.
12602 (first_match_or_eos_index_<mode>):
12603 Calculate index using natural order.
12604 (define_insn vclzlsbb): Change to define_insn vclzlsbb_<mode>.
12605 for BE and LE modes.
12606 * config/rs6000/rs6000-c.c: Rename P9V_BUILTIN_VCLZLSBB,
12607 P9V_BUILTIN_VCLZLSBB_V16QI.
12608 * config/rs6000/rs6000-builtin.def: Make VCLZLSBB mode
12611 2018-06-06 Kelvin Nilsen <kelvin@gcc.gnu.org>
12613 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
12614 indentation and line wrap for many prototypes. Add missing
12615 @smallexample directives around block of prototypes for vec_xl and
12618 2018-06-05 Michael Meissner <meissner@linux.ibm.com>
12620 * config/rs6000/rs6000.c (rs6000_passes_ieee128): New boolean to
12621 track if we pass or return IEEE 128-bit floating point.
12622 (ieee128_mangling_gcc_8_1): New boolean to say whether to generate
12623 C++ mangling that is compatible with GCC 8.1.
12624 (TARGET_ASM_GLOBALIZE_DECL_NAME): Override target hook.
12625 (init_cumulative_args): Note if we pass or return IEEE 128-bit
12626 floating point types.
12627 (rs6000_function_arg_advance_1): Likewise.
12628 (rs6000_mangle_type): Optionally generate mangled names that match
12629 what GCC 8.1 generated for IEEE 128-bit floating point types.
12630 (rs6000_globalize_decl_name): If we have an external function that
12631 passes or returns IEEE 128-bit types, generate a weak reference
12632 from the mangled name used in GCC 8.1 to the current mangled
12634 (rs6000_init_builtins): Make __ibm128 use the long double type if
12635 long double is IBM extended double. Make __float128 use the long
12636 double type if long double is IEEE 128-bit.
12639 * config/rs6000/rs6000-builtin.def (BU_IBM128_2): New helper
12640 macro for __ibm128 built-in functions.
12641 (PACK_IF): Add __ibm128 pack/unpack functions.
12642 (UNPACK_IF): Likewise.
12643 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Do not
12644 enable long double built-in functions if long double is IEEE
12645 128-bit floating point.
12646 (rs6000_invalid_builtin): Update long double built-in function
12648 (rs6000_expand_builtin): For PACK_IF and UNPACK_IF built-in
12649 functions, adjust the built-in function to use the long double
12650 built-in function if __ibm128 and long double are the same type.
12651 * doc/extend.texi (PowerPC builtins): Update documention for
12652 __builtin_{,un}pack_longdouble. Add documentation for
12653 __builtin_{,un}pack_ibm128.
12655 2018-06-06 Jim Wilson <jimw@sifive.com>
12657 * config/riscv/riscv.c (enum riscv_privilege_levels): New.
12658 (struct machine_function): New field interrupt_mode.
12659 (riscv_handle_type_attribute): New function. Add forward declaration.
12660 (riscv_attribute_table) <interrupt>: Use riscv_handle_type_attribute.
12661 (riscv_expand_epilogue): Check interrupt_mode field.
12662 (riscv_set_current_function): Check interrupt attribute args and
12663 set interrupt_mode field.
12664 * config/riscv/riscv.md (UNSPECV_SRET, UNSPECV_URET): New.
12665 (riscv_sret, riscv_uret): New.
12666 * doc/extend.texi (RISC-V Function Attributes) <interrupt>: Document
12667 new arguments to interrupt attribute.
12669 2018-06-06 Peter Bergner <bergner@vnet.ibm.com>
12672 * /config/rs6000/rs6000.h (ASM_CPU_SPEC): Add support for -mpower9.
12673 Don't handle -mcpu=power8 if -mpower9-vector is also used.
12675 2018-06-06 Kelvin Nilsen <kelvin@gcc.gnu.org>
12677 * config/rs6000/rs6000-builtin.def (VSX_BUILTIN_VEC_LD,
12678 VSX_BUILTIN_VEC_ST): Add comment to explain non-traditional uses.
12679 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
12680 several redundant entries.
12682 2018-06-06 David Malcolm <dmalcolm@redhat.com>
12684 * config/i386/i386-protos.h (ix86_expand_call): Strengthen return
12685 type from "rtx" to "rtx_insn *".
12686 * config/i386/i386.c (ix86_expand_split_stack_prologue): Likewise
12687 for local "call_insn", removing cast.
12688 (ix86_expand_call): Likewise, introducing a "call_insn" local.
12690 2018-06-06 Eric Botcazou <ebotcazou@adacore.com>
12692 PR tree-optimization/86066
12693 * gimple-ssa-store-merging.c (process_store): Do not bypass BIT_NOT_EXPR
12694 for BIT_INSERT_EXPR stores.
12696 2018-06-06 Richard Biener <rguenther@suse.de>
12698 PR tree-optimization/86062
12699 * tree-ssa-sccvn.c (vn_reference_lookup_3): Handle arbitrary
12700 component refs ontop
12701 of to be offsetted base.
12703 2018-06-06 Jozef Lawrynowicz <jozef.l@mittosystems.com>
12705 * config/msp430/msp430.c (msp430_attr): Allow interrupt handlers
12706 to be static and remove check on interrupt attribute name.
12708 2018-06-05 Kelvin Nilsen <kelvin@gcc.gnu.org>
12710 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Remove
12711 volatile qualifier from vec_lvsl and vec_lvsr argument prototypes.
12713 2018-06-05 Steve Ellcey <sellcey@cavium.com>
12716 * config/aarch64/aarch64-protos.h (aarch64_err_no_fpadvsimd): Remove
12718 * config/aarch64/aarch64-protos..c (aarch64_err_no_fpadvsimd):
12719 Remove second argument, change how error is called.
12720 (aarch64_layout_arg): Remove second argument from
12721 aarch64_err_no_fpadvsimd call.
12722 (aarch64_init_cumulative_args): Ditto.
12723 (aarch64_gimplify_va_arg_expr): Ditto.
12724 * config/aarch64/aarch64.md (mov<mode>): Ditto.
12726 2018-06-05 Uros Bizjak <ubizjak@gmail.com>
12728 * config/i386/i386.md (simple_return_indirect_internal): New expander.
12729 (*simple_return_indirect_internal<mode>): Rename from
12730 simple_return_indirect_internal. Use W mode iterator.
12731 (rstorssp): New expander.
12732 (*rstorssp<mode>): Rename from rstorssp. Use P mode iterator.
12733 (clrssbsy): New expander.
12734 (*clrssbsy<mode>): Rename from clrssbsy. Use P mode iterator.
12736 2018-06-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
12738 * config/arm/arm_cmse.h (cmse_nsfptr_create): Change typeof to
12740 (cmse_check_pointed_object): Likewise.
12742 2018-06-05 Martin Liska <mliska@suse.cz>
12744 PR gcov-profile/47618
12745 * doc/invoke.texi: Document how -fprofile-dir format
12748 2018-06-05 Richard Biener <rguenther@suse.de>
12750 * tree-cfgcleanup.c (cleanup_control_flow_pre): For edge
12751 removal pretend DOM info isn't available so we do not update
12752 it and only remove edges, not dominated blocks. Actually free
12753 DOM info in case we removed something. Remove unreachable blocks.
12754 (mfb_keep_latches): Work with either DOM info or marked backedges.
12755 (cleanup_tree_cfg_noloop): Do not remove unreachable blocks
12756 first. Mark backedges if DOM info isn't available.
12757 (Re-)compute DOM info after cleanup_control_flow_pre.
12759 2018-06-05 Richard Biener <rguenther@suse.de>
12761 * tree-cfg.c (struct locus_discrim_map): Store line, not location.
12762 (locus_discrim_hasher::hash): Adjust.
12763 (locus_discrim_hasher::equal): Likewise.
12764 (next_discriminator_for_locus): Work on line directly.
12765 (same_line_p): Pass in expanded locus1 as well.
12766 (assign_discriminators): Avoid redundant location expansions.
12768 2018-06-05 Richard Biener <rguenther@suse.de>
12770 PR tree-optimization/86046
12771 * tree-ssa.c (maybe_optimize_var): Clear DECL_GIMPLE_REG_P
12772 if required after clearing TREE_ADDRESSABLE.
12774 2018-06-05 Richard Biener <rguenther@suse.de>
12776 PR tree-optimization/86047
12777 * tree-ssa-loop.c (for_each_index): Glob handling of all
12778 decls and constants and really handle all of them.
12780 2018-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12783 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
12784 qualifier_void_pointer and qualifier_const_void_pointer.
12785 (arm_ldc_qualifiers, arm_stc_qualifiers): Use the above.
12786 (arm_init_builtins): Handle the above.
12787 * config/arm/arm_acle.h (__arm_cdp, __arm_ldc, __arm_ldcl, __arm_stc,
12788 __arm_stcl, __arm_mcr, __arm_cdp2, __arm_ldc2, __arm_ldcl2, __arm_stc2,
12789 __arm_stcl2,__arm_mcr2, __arm_mcrr, __arm_mcrr2): Remove return for
12792 2018-06-05 Martin Liska <mliska@suse.cz>
12794 * auto-profile.c (read_autofdo_file): Do not use
12795 gcov_ctr_summary struct.
12796 (afdo_callsite_hot_enough_for_early_inline): Likewise.
12797 * coverage.c (struct counts_entry): Likewise.
12798 (read_counts_file): Read just single summary entry.
12799 (get_coverage_counts): Use gcov_summary struct.
12800 * coverage.h (get_coverage_counts): Likewise.
12801 * gcov-dump.c (dump_working_sets): Likewise.
12802 (tag_summary): Dump just single summary.
12803 * gcov-io.c (gcov_write_summary): Write just histogram
12805 (gcov_read_summary): Read just single summary.
12806 (compute_working_sets): Use gcov_summary struct.
12807 * gcov-io.h (GCOV_TAG_SUMMARY_LENGTH): Remove usage
12808 of GCOV_COUNTERS_SUMMABLE.
12809 (GCOV_COUNTERS_SUMMABLE): Remove.
12810 (GCOV_FIRST_VALUE_COUNTER): Replace with
12811 GCOV_COUNTER_V_INTERVAL.
12812 (struct gcov_ctr_summary): Remove.
12813 (struct gcov_summary): Directly use fields of former
12815 (compute_working_sets): Use gcov_summary struct.
12816 * gcov.c (read_count_file): Do not use ctrs fields.
12817 * lto-cgraph.c (merge_profile_summaries): Use gcov_summary
12819 * lto-streamer.h (struct GTY): Make profile_info gcov_summary
12821 * profile.c: Likewise.
12822 * profile.h: Likewise.
12824 2018-06-05 Martin Liska <mliska@suse.cz>
12826 PR gcov-profile/84846
12827 * gcov.c (output_lines): Print working directory only
12828 in intermediate format.
12830 2018-06-05 Andreas Krebbel <krebbel@linux.ibm.com>
12832 * config/s390/s390-builtin-types.def: Add void function type.
12833 * config/s390/s390-builtins.def: Use the function type for the
12836 2018-06-04 Jim Wilson <jimw@sifive.com>
12838 * config/riscv/riscv-protos.h (riscv_expand_epilogue): Change bool arg
12840 * config/riscv/riscv.c (riscv_for_each_saved_reg): New args epilogue
12841 and maybe_eh_return. Change regno to unsigned int. Use new args to
12842 handle EH_RETURN_DATA_REGNO registers properly.
12843 (riscv_expand_prologue): Pass new args to riscv_for_each_saved_reg.
12844 (riscv_expand_epilogue): Update comment. Change argument name and
12845 type. Update code to use new name and type. Pass new args to
12846 riscv_for_each_saved_reg. Only use EH_RETURN_STACKADJ_RTX when
12848 * config/riscv/riscv.md (NORMAL_RETURN): New.
12849 (SIBCALL_RETURN, EXCEPTION_RETURN): New.
12850 (epilogue, sibcall_epilogue): Update riscv_expand_epilogue arg.
12851 (eh_return): Call gen_eh_return_internal and emit barrier.
12852 (eh_return_internal): Call riscv_expand_epilogue.
12854 2018-06-04 Eric Botcazou <ebotcazou@adacore.com>
12856 * gimple-ssa-store-merging.c (struct merged_store_group): Move up
12857 bit_insertion field and declare can_be_merged_into method.
12858 (merged_store_group::can_be_merged_into): New method.
12859 (imm_store_chain_info::coalesce_immediate): Call it to decide whether
12860 consecutive non-overlapping stores can be merged. Turn MEM_REF stores
12861 into BIT_INSERT_EXPR stores if the group contains a non-MEM_REF store.
12863 2018-06-04 Richard Biener <rguenther@suse.de>
12865 PR tree-optimization/85955
12866 * builtins.c (fold_builtin_sincos): Convert pointers to
12867 destination to appropriate type before dereferencing.
12869 2018-06-04 Segher Boessenkool <segher@kernel.crashing.org>
12871 * config/rs6000/rs6000.md (abs<mode>2 for FLOAT128): Handle IFmode.
12873 2018-06-04 Richard Sandiford <richard.sandiford@linaro.org>
12875 * expr.c (expand_expr_real_1): Force the operand into memory if
12876 its TYPE_MODE is BLKmode and if there is no integer mode for
12877 the number of bits being extracted.
12879 2018-06-04 Jakub Jelinek <jakub@redhat.com>
12883 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>_1):
12884 Use vptestnm rather than vptestm in (=Yc,v,C) variant.
12886 2018-06-04 Richard Biener <rguenther@suse.de>
12888 * tree-cfgcleanup.c (cleanup_tree_cfg_1): Fold into...
12889 (cleanup_tree_cfg_noloop): ... single caller. Do
12890 start_recording_case_labels later.
12892 2018-06-04 Sebastian Peryt <sebastian.peryt@intel.com>
12894 * config/i386/cldemoteintrin.h: Change define from _X86INTRIN_H_INCLUDED
12895 to _IMMINTRIN_H_INCLUDED.
12896 * config/i386/pconfigintrin.h: Ditto.
12897 * config/i386/waitpkgintrin.h: Ditto.
12898 * config/i386/immintrin.h: Add includes for sgxintrin.h,
12899 pconfigintrin.h, waitpkgintrin.h and cldemoteintrin.h.
12900 * config/i386/x86intrin.h: Remove includes for mintrin.h, xmmintrin.h,
12901 emmintrin.h, pmmintrin.h, tmmintrin.h, smmintrin.h, wmmintrin.h,
12902 bmiintrin.h, bmi2intrin.h, lzcntintrin.h, sgxintrin.h, pconfigintrin.h,
12903 waitpkgintrin.h and cldemoteintrin.h.
12905 2018-06-04 Richard Biener <rguenther@suse.de>
12907 PR tree-optimization/86038
12908 * tracer.c (find_best_successor): Check probability for
12909 being initialized, bail out if not.
12911 2018-06-04 Richard Earnshaw <rearnsha@arm.com>
12914 * config/arm/arm-cpus.in (ALL_QUIRKS): Add xscale feature to the list
12915 of bits to ignore when comparing architectures.
12917 2018-06-04 Jakub Jelinek <jakub@redhat.com>
12919 PR tree-optimization/69615
12920 * fold-const.c (merge_ranges): If range1 is - [x, x] and x is the
12921 maximum or minimum of the type, try to merge it also as if
12922 range1 is + [-, x - 1] or + [x + 1, -].
12925 * tree.c (inchash::add_expr): Handle IDENTIFIER_NODE.
12927 2018-06-03 Eric Botcazou <ebotcazou@adacore.com>
12929 PR tree-optimization/86034
12930 * gimple-ssa-store-merging.c (output_merged_store): Convert the RHS to
12931 the unsigned bitfield type in a bit insertion sequence if it does not
12932 have a larger precision than the bitfield size.
12933 (process_store): Also bypass widening conversions for BIT_INSERT_EXPR.
12935 2018-06-03 Kito Cheng <kito.cheng@gmail.com>
12937 * config/nds32/nds32-peephole2.md: Add new patterns for code size.
12939 2018-06-03 Chung-Ju Wu <jasonwucj@gmail.com>
12941 * config/nds32/nds32-opts.h (nds32_arch_type): Add ARCH_V3J.
12942 * config/nds32/nds32.c (nds32_option_override): Consider ARCH_V3J.
12943 * config/nds32/nds32.h (TARGET_ISA_V3): Add ARCH_V3J checking.
12944 * config/nds32/nds32.opt (march): Add enum value Value(ARCH_V3J).
12946 2018-06-03 Chung-Ju Wu <jasonwucj@gmail.com>
12948 * common/config/nds32/nds32-common.c (nds32_option_optimization_table):
12949 Disable -fdelete-null-pointer-checks for ELF toolchain.
12951 2018-06-02 Chung-Ju Wu <jasonwucj@gmail.com>
12952 Kito Cheng <kito.cheng@gmail.com>
12954 * config.gcc (nds32*): Use nds32-linux.opt and nds32-elf.opt.
12955 (nds32le-*-*, nds32be-*-*): Integrate checking process.
12956 (nds32*-*-*): Add glibc and uclibc conditions.
12957 * common/config/nds32/nds32-common.c (nds32_except_unwind_info): New.
12958 (TARGET_EXCEPT_UNWIND_INFO): Define.
12959 * config/nds32/elf.h: New file.
12960 * config/nds32/linux.h: New file.
12961 * config/nds32/nds32-elf.opt: New file.
12962 * config/nds32/nds32-linux.opt: New file.
12963 * config/nds32/nds32-fp-as-gp.c
12964 (pass_nds32_fp_as_gp::gate): Consider TARGET_LINUX_ABI.
12965 * config/nds32/nds32.c (nds32_conditional_register_usage): Consider
12967 (nds32_asm_file_end): Ditto.
12968 (nds32_print_operand): Ditto.
12969 (nds32_insert_attributes): Ditto.
12970 (nds32_init_libfuncs): New function.
12971 (TARGET_HAVE_TLS): Define.
12972 (TARGET_INIT_LIBFUNCS): Define.
12973 * config/nds32/nds32.h (TARGET_DEFAULT_RELAX): Apply different relax
12975 (TARGET_ELF): Apply different mcmodel setting.
12976 (LINK_SPEC, LIB_SPEC, STARTFILE_SPEC, ENDFILE_SPEC): The content has
12977 been migrated into elf.h and linux.h files.
12978 * config/nds32/nds32.md (add_pc): Consider TARGET_LINUX_ABI.
12979 * config/nds32/nds32.opt (mvh): Consider TARGET_LINUX_ABI.
12980 (mcmodel): The content has been migrated into nds32-elf.opt and
12981 nds32-linux.opt files.
12982 * config/nds32/t-elf: New file.
12983 * config/nds32/t-linux: New file.
12985 2018-06-02 Chung-Ju Wu <jasonwucj@gmail.com>
12986 Shiva Chen <shiva0217@gmail.com>
12988 * config/nds32/constants.md (unspec_volatile_element): Add
12989 UNSPEC_VOLATILE_OMIT_FP_BEGIN and UNSPEC_VOLATILE_OMIT_FP_END.
12990 * config/nds32/nds32-fp-as-gp.c: New implementation of fp_as_gp
12992 * config/nds32/nds32-protos.h (nds32_naked_function_p): Declare.
12993 (make_pass_nds32_fp_as_gp): Declare.
12994 * config/nds32/nds32.c (nds32_register_passes): Add fp_as_gp as one
12996 (nds32_asm_function_end_prologue): Remove unused asm output.
12997 (nds32_asm_function_begin_epilogue): Remove unused asm output.
12998 (nds32_asm_file_start): Output necessary fp_as_gp information.
12999 (nds32_option_override): Adjust register usage.
13000 (nds32_expand_prologue): Consider fp_as_gp situation.
13001 (nds32_expand_prologue_v3push): Consider fp_as_gp situation.
13002 * config/nds32/nds32.md (prologue): Check fp_as_gp_p and naked_p.
13005 (simple_return): Ditto.
13006 (omit_fp_begin): Output special directive for fp_as_gp.
13007 (omit_fp_end): Output special directive for fp_as_gp.
13008 * config/nds32/nds32.opt (mfp-as-gp, mno-fp-as-gp, mforce-fp-as-gp,
13009 mforbid-fp-as-gp): New options.
13011 2018-06-01 Mark Wielaard <mark@klomp.org>
13013 * dwarf2out.c (dwarf2out_finish): Remove generation of
13014 DW_AT_loclists_base.
13016 2018-06-01 Eric Botcazou <ebotcazou@adacore.com>
13018 * gimple-ssa-store-merging.c: Include gimple-fold.h.
13019 (struct store_immediate_info): Document BIT_INSERT_EXPR stores.
13020 (struct merged_store_group): Add bit_insertion field.
13021 (dump_char_array): Use standard hexadecimal format.
13022 (merged_store_group::merged_store_group): Set bit_insertion to false.
13023 (merged_store_group::apply_stores): Use optimal buffer size. Deal
13024 with BIT_INSERT_EXPR stores. Move up code updating the mask and
13025 also print the mask in the dump file.
13026 (pass_store_merging::gate): Minor tweak.
13027 (imm_store_chain_info::coalesce_immediate): Fix wrong association
13028 of stores with groups in dump. Allow coalescing of BIT_INSERT_EXPR
13029 stores with INTEGER_CST stores.
13030 (count_multiple_uses) <BIT_INSERT_EXPR>: New case.
13031 (imm_store_chain_info::output_merged_store): Add try_bitpos variable
13032 and use it throughout. Generate bit insertion sequences if need be.
13033 (pass_store_merging::process_store): Remove redundant condition.
13034 Record stores from a SSA name to a bit-field with BIT_INSERT_EXPR.
13036 2018-06-01 Segher Boessenkool <segher@kernel.crashing.org>
13038 * config/rs6000/rs6000.c (rs6000_mangle_type): Change the mangling of
13039 the 128-bit floating point types. Fix function comment.
13041 2018-06-01 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13043 * config/aarch64/aarch64-simd.md
13044 (aarch64_simd_vec_unpack<su>_lo_<mode>): Use UXTL and SXTL assembler
13046 (aarch64_simd_vec_unpack<su>_hi_<mode>): Use UXTL2 and SXTL2 assembler
13049 2018-06-01 Richard Sandiford <richard.sandiford@linaro.org>
13051 PR tree-optimization/85989
13052 * gimple-ssa-backprop.c (backprop::m_visited_phis): New member
13054 (backprop::intersect_uses): Check it when deciding whether this
13055 is a backedge reference.
13056 (backprop::process_block): Add each phi to m_visited_phis
13057 after visiting it, then clear it at the end.
13059 2018-06-01 Richard Biener <rguenther@suse.de>
13061 * tree-vectorizer.h (vect_dr_stmt): New function.
13062 (vect_get_load_cost): Adjust.
13063 (vect_get_store_cost): Likewise.
13064 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence):
13065 Use vect_dr_stmt instead of DR_SMTT.
13066 (vect_record_base_alignments): Likewise.
13067 (vect_calculate_target_alignment): Likewise.
13068 (vect_compute_data_ref_alignment): Likewise and make static.
13069 (vect_update_misalignment_for_peel): Likewise.
13070 (vect_verify_datarefs_alignment): Likewise.
13071 (vector_alignment_reachable_p): Likewise.
13072 (vect_get_data_access_cost): Likewise. Pass down
13073 vinfo to vect_get_load_cost/vect_get_store_cost instead of DR.
13074 (vect_get_peeling_costs_all_drs): Likewise.
13075 (vect_peeling_hash_get_lowest_cost): Likewise.
13076 (vect_enhance_data_refs_alignment): Likewise.
13077 (vect_find_same_alignment_drs): Likewise.
13078 (vect_analyze_data_refs_alignment): Likewise.
13079 (vect_analyze_group_access_1): Likewise.
13080 (vect_analyze_group_access): Likewise.
13081 (vect_analyze_data_ref_access): Likewise.
13082 (vect_analyze_data_ref_accesses): Likewise.
13083 (vect_vfa_segment_size): Likewise.
13084 (vect_small_gap_p): Likewise.
13085 (vectorizable_with_step_bound_p): Likewise.
13086 (vect_prune_runtime_alias_test_list): Likewise.
13087 (vect_analyze_data_refs): Likewise.
13088 (vect_supportable_dr_alignment): Likewise.
13089 * tree-vect-loop-manip.c (get_misalign_in_elems): Likewise.
13090 (vect_gen_prolog_loop_niters): Likewise.
13091 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
13092 * tree-vect-patterns.c (vect_recog_bool_pattern): Do not
13094 (vect_recog_mask_conversion_pattern): Likewise.
13095 (vect_try_gather_scatter_pattern): Likewise.
13096 * tree-vect-stmts.c (vect_model_store_cost): Pass stmt_info
13097 to vect_get_store_cost.
13098 (vect_get_store_cost): Get stmt_info instead of DR.
13099 (vect_model_load_cost): Pass stmt_info to vect_get_load_cost.
13100 (vect_get_load_cost): Get stmt_info instead of DR.
13102 2018-06-01 Richard Biener <rguenther@suse.de>
13104 PR middle-end/86017
13105 * gimple-fold.c (var_decl_component_p): Also allow offsetted
13106 vars wrapped in MEM_REFs.
13108 2018-06-01 Richard Sandiford <richard.sandiford@linaro.org>
13110 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
13111 Fix subreg tests so that we only return a choice between
13112 GENERAL_REGS and FP_REGS if the original classes included both.
13114 2018-06-01 Richard Biener <rguenther@suse.de>
13117 * tree-ssa-structalias.c (get_function_part_constraint):
13118 Handle NULL fi->decl.
13119 (find_func_aliases_for_call): Properly handle indirect
13120 fi from direct call.
13121 (find_func_clobbers): Likewise.
13122 (ipa_pta_execute): Likewise.
13123 (create_variable_info_for): For functions that are ifunc_resolver
13124 resolve to a varinfo that contains the result of the resolver call.
13125 (associate_varinfo_to_alias): Do not treat ifunc resolvers as
13128 2018-05-31 Michael Collison <michael.collison@arm.com>
13130 * config/aarch64/aarch64.md:
13131 (*fix_to_zero_extenddfdi2): New pattern.
13132 * gcc.target/aarch64/fix_extend1.c: New testcase.
13134 2018-05-31 Qing Zhao <qing.zhao@oracle.com>
13136 PR middle-end/78809
13137 PR middle-end/83026
13138 * builtins.c (expand_builtin): Add the handling of BUILT_IN_STRCMP_EQ
13139 and BUILT_IN_STRNCMP_EQ.
13140 * builtins.def: Add new builtins BUILT_IN_STRCMP_EQ and
13141 BUILT_IN_STRNCMP_EQ.
13142 * gimple-fold.c (gimple_fold_builtin_string_compare): Add the
13143 handling of BUILTIN_IN_STRCMP_EQ and BUILT_IN_STRNCMP_EQ.
13144 (gimple_fold_builtin): Likewise.
13145 * tree-ssa-strlen.c (compute_string_length): New function.
13146 (determine_min_obsize): New function.
13147 (handle_builtin_string_cmp): New function to handle calls to
13148 string compare functions.
13149 (strlen_optimize_stmt): Add handling to builtin string compare
13151 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
13152 Add the handling of BUILT_IN_STRCMP_EQ and BUILT_IN_STRNCMP_EQ.
13153 * tree.c (build_common_builtin_nodes): Add new defines of
13154 BUILT_IN_STRNCMP_EQ and BUILT_IN_STRCMP_EQ.
13156 2018-05-31 Jakub Jelinek <jakub@redhat.com>
13159 * bb-reorder.c (pass_partition_blocks::gate): Return false for
13160 functions with naked attribute.
13162 2018-05-31 Uros Bizjak <ubizjak@gmail.com>
13164 * config/i386/sse.md (avx_vec_concat<mode>):
13165 Substitute concat_tg_mode mode attribute with xtg_mode.
13166 (<mask_codefor>avx512dq_broadcast<mode><mask_name>_1): Ditto.
13167 (concat_tg_mode): Remove mode attribute.
13169 2018-05-31 Martin Sebor <msebor@redhat.com>
13172 * calls.c (alloc_max_size): Correct a logic error/typo.
13173 Treat excessive arguments as infinite. Warn for invalid arguments.
13174 * doc/invoke.texi (-Walloc-size-larger-than): Update.
13176 2018-05-31 H.J. Lu <hongjiu.lu@intel.com>
13179 * config/i386/x86-tune.def: Re-enable partial_reg_dependency
13180 and movx for Haswell.
13182 2018-05-31 Chung-Lin Tang <cltang@codesourcery.com>
13183 Cesar Philippidis <cesar@codesourcery.com>
13185 PR middle-end/85879
13186 * gimplify.c (gimplify_adjust_omp_clauses): Add 'remove = true'
13187 when emitting error on private/firstprivate reductions.
13188 * omp-low.c (lower_omp_target): Avoid reference-type processing
13189 on pointers for firstprivate clause.
13191 2018-05-31 Sameera Deshpande <sameera.deshpande@linaro.org>
13193 * config/aarch64/aarch64-simd-builtins.def (ld1x3): New.
13196 * config/aarch64/aarch64-simd.md
13197 (aarch64_ld1x3<VALLDIF:mode>): New pattern.
13198 (aarch64_ld1_x3_<mode>): Likewise
13199 (aarch64_st1x2<VALLDIF:mode>): Likewise
13200 (aarch64_st1_x2_<mode>): Likewise
13201 (aarch64_st1x3<VALLDIF:mode>): Likewise
13202 (aarch64_st1_x3_<mode>): Likewise
13203 * config/aarch64/arm_neon.h (vld1_u8_x3): New function.
13204 (vld1_s8_x3): Likewise.
13205 (vld1_u16_x3): Likewise.
13206 (vld1_s16_x3): Likewise.
13207 (vld1_u32_x3): Likewise.
13208 (vld1_s32_x3): Likewise.
13209 (vld1_u64_x3): Likewise.
13210 (vld1_s64_x3): Likewise.
13211 (vld1_f16_x3): Likewise.
13212 (vld1_f32_x3): Likewise.
13213 (vld1_f64_x3): Likewise.
13214 (vld1_p8_x3): Likewise.
13215 (vld1_p16_x3): Likewise.
13216 (vld1_p64_x3): Likewise.
13217 (vld1q_u8_x3): Likewise.
13218 (vld1q_s8_x3): Likewise.
13219 (vld1q_u16_x3): Likewise.
13220 (vld1q_s16_x3): Likewise.
13221 (vld1q_u32_x3): Likewise.
13222 (vld1q_s32_x3): Likewise.
13223 (vld1q_u64_x3): Likewise.
13224 (vld1q_s64_x3): Likewise.
13225 (vld1q_f16_x3): Likewise.
13226 (vld1q_f32_x3): Likewise.
13227 (vld1q_f64_x3): Likewise.
13228 (vld1q_p8_x3): Likewise.
13229 (vld1q_p16_x3): Likewise.
13230 (vld1q_p64_x3): Likewise.
13231 (vst1_s64_x2): Likewise.
13232 (vst1_u64_x2): Likewise.
13233 (vst1_f64_x2): Likewise.
13234 (vst1_s8_x2): Likewise.
13235 (vst1_p8_x2): Likewise.
13236 (vst1_s16_x2): Likewise.
13237 (vst1_p16_x2): Likewise.
13238 (vst1_s32_x2): Likewise.
13239 (vst1_u8_x2): Likewise.
13240 (vst1_u16_x2): Likewise.
13241 (vst1_u32_x2): Likewise.
13242 (vst1_f16_x2): Likewise.
13243 (vst1_f32_x2): Likewise.
13244 (vst1_p64_x2): Likewise.
13245 (vst1q_s8_x2): Likewise.
13246 (vst1q_p8_x2): Likewise.
13247 (vst1q_s16_x2): Likewise.
13248 (vst1q_p16_x2): Likewise.
13249 (vst1q_s32_x2): Likewise.
13250 (vst1q_s64_x2): Likewise.
13251 (vst1q_u8_x2): Likewise.
13252 (vst1q_u16_x2): Likewise.
13253 (vst1q_u32_x2): Likewise.
13254 (vst1q_u64_x2): Likewise.
13255 (vst1q_f16_x2): Likewise.
13256 (vst1q_f32_x2): Likewise.
13257 (vst1q_f64_x2): Likewise.
13258 (vst1q_p64_x2): Likewise.
13259 (vst1_s64_x3): Likewise.
13260 (vst1_u64_x3): Likewise.
13261 (vst1_f64_x3): Likewise.
13262 (vst1_s8_x3): Likewise.
13263 (vst1_p8_x3): Likewise.
13264 (vst1_s16_x3): Likewise.
13265 (vst1_p16_x3): Likewise.
13266 (vst1_s32_x3): Likewise.
13267 (vst1_u8_x3): Likewise.
13268 (vst1_u16_x3): Likewise.
13269 (vst1_u32_x3): Likewise.
13270 (vst1_f16_x3): Likewise.
13271 (vst1_f32_x3): Likewise.
13272 (vst1_p64_x3): Likewise.
13273 (vst1q_s8_x3): Likewise.
13274 (vst1q_p8_x3): Likewise.
13275 (vst1q_s16_x3): Likewise.
13276 (vst1q_p16_x3): Likewise.
13277 (vst1q_s32_x3): Likewise.
13278 (vst1q_s64_x3): Likewise.
13279 (vst1q_u8_x3): Likewise.
13280 (vst1q_u16_x3): Likewise.
13281 (vst1q_u32_x3): Likewise.
13282 (vst1q_u64_x3): Likewise.
13283 (vst1q_f16_x3): Likewise.
13284 (vst1q_f32_x3): Likewise.
13285 (vst1q_f64_x3): Likewise.
13286 (vst1q_p64_x3): Likewise.
13288 2018-05-30 Jozef Lawrynowicz <jozef.l@mittosystems.com>
13290 * config/msp430/msp430.c (msp430_output_labelref): Prepend
13291 user_label_prefix to name.
13293 * tree-core.h: Update comment about the format of NAME string
13294 passed to handler in attribute_spec.
13296 * config/msp430/msp430.md: Remove erroneous subreg expression from
13297 zero_extendqisi2 insn pattern. Remove msp430x ISA restriction on
13298 zero_extend{q,h}isi2.
13300 2018-05-30 Borislav Petkov <bp@suse.de>
13302 * doc/extend.texi: Document some architecture specific
13303 constraints and sort entries.
13305 2018-05-30 Martin Sebor <msebor@redhat.com>
13307 PR middle-end/85369
13308 * builtins.c (expand_builtin_stpcpy_1): New function.
13309 (expand_builtin_stpcpy): Call it, and call maybe_warn_nonstring_arg
13310 only if the former succeeds.
13312 2018-05-31 Sameera Deshpande <sameera.deshpande@linaro.org>
13314 * config/aarch64/aarch64-cores.def (saphira) : Add support for ARMv8.4
13317 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
13319 * doc/invoke.texi (-flinker-output): Document
13321 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
13323 * passes.c (ipa_write_summaries): Only modify statements if body
13325 * cgraphunit.c (ipa_passes): Also produce intermeidate code when
13326 incrementally linking.
13327 (ipa_passes): Likewise.
13328 * lto-cgraph.c (lto_output_node): When incrementally linking do not
13329 pass down resolution info.
13330 * common.opt (flag_incremental_link): Update info.
13331 * gcc.c (plugin specs): Turn flinker-output=* to
13332 -plugin-opt=-linker-output-known
13333 * toplev.c (compile_file): Also cut compilation when doing incremental
13335 * flag-types. (enum lto_partition_model): Add
13336 LTO_LINKER_OUTPUT_NOLTOREL.
13337 (invoke.texi): Add -flinker-output docs.
13338 * ipa.c (symbol_table::remove_unreachable_nodes): Handle LTO incremental
13339 link same way as WPA; do not stream in dead initializers.
13341 * dwarf2out.c (dwarf2out_die_ref_for_decl,
13342 darf2out_register_external_decl): Support incremental link.
13344 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
13346 * lto-opts.c (lto_write_options): Skip OPT_dumpdir, OPT_fresolution_.
13348 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
13350 * lto-wrapper.c (debug_objcopy): Add rename parameter; pass
13351 it down to simple_object_copy_lto_debug_sections.
13352 (run_gcc): Determine incremental LTO link time and configure
13353 lto1 into non-wpa mode, disable renaming of debug sections.
13355 2018-05-30 Kelvin Nilsen <kelvin@gcc.gnu.org>
13357 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Remove
13358 descriptions of various incorrectly documented functions.
13360 2018-05-30 Andre Vieira <andre.simoesdiasvieira@arm.com>
13363 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Make
13364 address check not strict.
13366 2018-05-30 Richard Biener <rguenther@suse.de>
13368 PR tree-optimization/85964
13369 * tracer.c (better_p): Drop initialized count check, we only
13370 call the function with initialized counts now.
13371 (find_best_successor): Do find a best edge if one
13372 has uninitialized count.
13373 (find_best_predecessor): Likewise. Do BB frequency check only
13374 if count is initialized.
13376 2017-05-30 Jackson Woodruff <jackson.woodruff@arm.com>
13378 * config/aarch64/aarch64.c (aarch64_host_wide_int_compare): New.
13379 (aarch64_ldrstr_offset_compare): New.
13380 (aarch64_operands_adjust_ok_for_ldpstp): Update to consider all
13381 load/store orderings.
13382 (aarch64_gen_adjusted_ldpstp): Likewise.
13384 2018-05-30 Wilco Dijkstra <wdijkstr@arm.com>
13386 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
13387 Check for subset of GENERAL_REGS and FP_REGS.
13388 * config/aarch64/aarch64-simd.md (aarch64_get_lane): Increase cost of
13391 2018-05-30 Richard Sandiford <richard.sandiford@linaro.org>
13393 * alias.c (adjust_offset_for_component_ref): Use poly_int_tree_p
13394 and wi::to_poly_offset. Add the current offset and then check
13395 whether the sum fits, rather than using an unchecked addition of
13396 a checked term. Check for a shwi rather than a uhwi.
13397 * expr.c (get_bit_range): Use tree_to_poly_uint64.
13398 (store_constructor): Use poly_int_tree_p.
13399 (expand_expr_real_1): Likewise.
13400 * function.c (assign_temp): Likewise.
13401 * fold-const.c (const_binop): Use poly_int_tree_p and
13402 wi::to_poly_offset.
13403 (fold_indirect_ref_1): Likewise. Use multiple_p to attempt an exact
13405 * ipa-icf-gimple.c (func_checker::compare_operand): Use
13406 to_poly_offset for MEM offsets.
13407 * ipa-icf.c (sem_variable::equals): Likewise.
13408 * stor-layout.c (compute_record_mode): Use poly_int_tree_p.
13409 * tree-ssa-sccvn.c (ao_ref_init_from_vn_reference): Use
13410 wi::to_poly_offset for BIT_FIELD_REF offsets.
13411 (vn_reference_maybe_forwprop_address): Use poly_int_tree_p and
13412 wi::to_poly_offset.
13413 * var-tracking.c (emit_note_insn_var_location): Use
13414 tree_to_poly_uint64.
13416 2018-05-29 Jim Wilson <jimw@sifive.com>
13418 * config/riscv/riscv.c (riscv_interrupt_type): Fix comment typo.
13420 2018-05-29 Uros Bizjak <ubizjak@gmail.com>
13423 * config/i386/i386.md (l<rounding_insn><MODEF:mode><SWI48:mode>2):
13424 Enable for TARGET_SSE4_1 and generate rounds{s,d} and cvtts{s,d}2si{,q}
13426 (sse4_1_round<mode>2): Use nonimmediate_operand
13427 for operand 1 predicate.
13429 2018-05-29 Martin Sebor <msebor@redhat.com>
13430 Richard Biener <rguenther@suse.de>
13433 * calls.c (get_size_range): Call determine_value_range instead
13434 of get_value_range..
13435 * tree-vrp.h (determine_value_range): Declared new function.
13436 * tree-vrp.c (determine_value_range_1, determine_value_range): New.
13438 2018-05-29 Richard Biener <rguenther@suse.de>
13440 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Make
13441 sure to use non-pattern stmts for get_earlier_stmt arguments.
13442 * tree-vectorizer.h (get_earlier_stmt): Assert we do not get
13443 called on pattern stmts.
13444 (get_later_stmt): Likewise.
13446 2018-05-29 Martin Liska <mliska@suse.cz>
13448 PR gcov-profile/85759
13449 * doc/gcov.texi: Document GCOV_ERROR_FILE and GCOV_EXIT_AT_ERROR
13452 2018-05-29 Jakub Jelinek <jakub@redhat.com>
13454 * tree-cfg.c (verify_gimple_assign_unary): Add checking for
13456 (verify_gimple_assign_binary): Check TYPE_VECTOR_SUBPARTS for
13460 * tree.def (VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR,
13461 VEC_PACK_FLOAT_EXPR): New tree codes.
13462 * tree-pretty-print.c (op_code_prio): Handle
13463 VEC_UNPACK_FIX_TRUNC_HI_EXPR and VEC_UNPACK_FIX_TRUNC_LO_EXPR.
13464 (dump_generic_node): Handle VEC_UNPACK_FIX_TRUNC_HI_EXPR,
13465 VEC_UNPACK_FIX_TRUNC_LO_EXPR and VEC_PACK_FLOAT_EXPR.
13466 * tree-inline.c (estimate_operator_cost): Likewise.
13467 * gimple-pretty-print.c (dump_binary_rhs): Handle VEC_PACK_FLOAT_EXPR.
13468 * fold-const.c (const_binop): Likewise.
13469 (const_unop): Handle VEC_UNPACK_FIX_TRUNC_HI_EXPR and
13470 VEC_UNPACK_FIX_TRUNC_LO_EXPR.
13471 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
13472 (verify_gimple_assign_binary): Handle VEC_PACK_FLOAT_EXPR.
13473 * cfgexpand.c (expand_debug_expr): Handle VEC_UNPACK_FIX_TRUNC_HI_EXPR,
13474 VEC_UNPACK_FIX_TRUNC_LO_EXPR and VEC_PACK_FLOAT_EXPR.
13475 * expr.c (expand_expr_real_2): Likewise.
13476 * optabs.def (vec_packs_float_optab, vec_packu_float_optab,
13477 vec_unpack_sfix_trunc_hi_optab, vec_unpack_sfix_trunc_lo_optab,
13478 vec_unpack_ufix_trunc_hi_optab, vec_unpack_ufix_trunc_lo_optab): New
13480 * optabs.c (expand_widen_pattern_expr): For
13481 VEC_UNPACK_FIX_TRUNC_HI_EXPR and VEC_UNPACK_FIX_TRUNC_LO_EXPR use
13482 sign from result type rather than operand's type.
13483 (expand_binop_directly): For vec_packu_float_optab and
13484 vec_packs_float_optab allow result type to be different from operand's
13486 * optabs-tree.c (optab_for_tree_code): Handle
13487 VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR and
13488 VEC_PACK_FLOAT_EXPR. Formatting fixes.
13489 * tree-vect-generic.c (expand_vector_operations_1): Handle
13490 VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR and
13491 VEC_PACK_FLOAT_EXPR.
13492 * tree-vect-stmts.c (supportable_widening_operation): Handle
13494 (supportable_narrowing_operation): Handle FLOAT_EXPR.
13495 * config/i386/i386.md (fixprefix, floatprefix): New code attributes.
13496 * config/i386/sse.md (*float<floatunssuffix>v2div2sf2): Rename to ...
13497 (float<floatunssuffix>v2div2sf2): ... this. Formatting fix.
13498 (vpckfloat_concat_mode, vpckfloat_temp_mode, vpckfloat_op_mode): New
13500 (vec_pack<floatprefix>_float_<mode>): New expander.
13501 (vunpckfixt_mode, vunpckfixt_model, vunpckfixt_extract_mode): New mode
13503 (vec_unpack_<fixprefix>fix_trunc_lo_<mode>,
13504 vec_unpack_<fixprefix>fix_trunc_hi_<mode>): New expanders.
13505 * doc/md.texi (vec_packs_float_@var{m}, vec_packu_float_@var{m},
13506 vec_unpack_sfix_trunc_hi_@var{m}, vec_unpack_sfix_trunc_lo_@var{m},
13507 vec_unpack_ufix_trunc_hi_@var{m}, vec_unpack_ufix_trunc_lo_@var{m}):
13509 * doc/generic.texi (VEC_UNPACK_FLOAT_HI_EXPR,
13510 VEC_UNPACK_FLOAT_LO_EXPR): Fix pasto in description.
13511 (VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR,
13512 VEC_PACK_FLOAT_EXPR): Document.
13514 2018-05-29 Richard Biener <rguenther@suse.de>
13516 * tree-vectorizer.h (struct vec_info): Add stmt_vec_infos
13518 (stmt_vec_info_vec): Make pointer.
13519 (init_stmt_vec_info_vec): Remove.
13520 (free_stmt_vec_info_vec): Likewise.
13521 (set_stmt_vec_info_vec): New function.
13522 (free_stmt_vec_infos): Likewise.
13523 (vinfo_for_stmt): Adjust for stmt_vec_info_vec indirection.
13524 (set_vinfo_for_stmt): Likewise.
13525 (get_earlier_stmt): Likewise.
13526 (get_later_stmt): Likewise.
13527 * tree-vectorizer.c (stmt_vec_info_vec): Make pointer.
13528 (vec_info::vec_info): Allocate stmt_vec_infos and set the global.
13529 (vec_info::~vec_info): Free stmt_vec_infos.
13530 (vectorize_loops): Set the global stmt_vec_info_vec to NULL.
13531 Remove old init_stmt_vec_info_vec/free_stmt_vec_info_vec calls.
13532 (pass_slp_vectorize::execute): Likewise.
13533 * tree-vect-stmts.c (init_stmt_vec_info_vec): Remove.
13534 (free_stmt_vec_info_vec): Likewise.
13535 (set_stmt_vec_info_vec): New function.
13536 (free_stmt_vec_infos): Likewise.
13537 * tree-vect-loop.c (_loop_vec_info::~_loop_vec_info): Set
13538 the global stmt_vec_info_vec.
13539 * tree-parloops.c (gather_scalar_reductions): Use
13540 set_stmt_vec_info_vec/free_stmt_vec_infos and maintain a local
13543 2018-05-29 Richard Biener <rguenther@suse.de>
13545 * dominance.c (iterate_fix_dominators): Push/pop TV_DOMINANCE.
13547 2018-05-29 Martin Liska <mliska@suse.cz>
13548 David Malcolm <dmalcolm@redhat.com>
13550 * vec.c (test_reverse): New.
13551 (vec_c_tests): Add new test.
13552 * vec.h (vl_ptr>::reverse): New function.
13554 2018-05-29 Gerald Pfeifer <gerald@pfeifer.com>
13556 * config.gcc: Identify FreeBSD 3.x and 4.x as unsupported.
13558 * config/freebsd-spec.h (FBSD_LIB_SPEC): Only consider FreeBSD 5
13561 2018-05-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
13563 * tree-dump.c (dump_node): Use splay_tree_delete_pointers.
13565 2018-05-28 Richard Biener <rguenther@suse.de>
13567 PR tree-optimization/85933
13568 * tree-vect-data-refs.c (vect_record_base_alignments): Only
13569 look at stmts marked as vectorizable.
13571 2018-05-28 Richard Biener <rguenther@suse.de>
13573 PR tree-optimization/85934
13574 * tree-vect-generic.c (expand_vector_operations_1): Hoist
13575 vector boolean check before scalar optimization.
13577 2018-05-28 Jakub Jelinek <jakub@redhat.com>
13579 * doc/invoke.texi (ARM Options): Use @item instead of @itemx
13582 2018-05-28 Mark Wielaard <mark@klomp.org>
13584 * dwarf2asm.c (dw2_asm_output_delta_uleb128): Add brackets around lab2
13585 if it is an expression containing a minus sign.
13587 2018-05-27 John David Anglin <danglin@gcc.gnu.org>
13589 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Define to 0.
13591 2018-05-27 Paul Koning <ni1d@arrl.net>
13593 * config/pdp11/pdp11.md (truncsihi2): Remove.
13595 2018-05-27 Monk Chiang <sh.chiang04@gmail.com>
13596 Chung-Ju Wu <jasonwucj@gmail.com>
13598 * config/nds32/nds32-intrinsic.md (unaligned_storedi): Modify patterns
13600 (unaligned_store_dw): Ditto.
13601 * config/nds32/nds32-memory-manipulation.c
13602 (nds32_expand_movmemsi_loop_known_size): Refactoring implementation.
13603 (nds32_gen_dup_4_byte_to_word_value): Rename to ...
13604 (nds32_gen_dup_4_byte_to_word_value_aux): ... this.
13605 (emit_setmem_word_loop): Rename to ...
13606 (emit_setmem_doubleword_loop): ... this.
13607 (nds32_gen_dup_4_byte_to_word_value): New function.
13608 (nds32_gen_dup_8_byte_to_double_word_value): New function.
13609 (nds32_expand_setmem_loop): Refine implementation.
13610 (nds32_expand_setmem_loop_v3m): Ditto.
13611 * config/nds32/nds32-multiple.md (unaligned_store_update_base_dw): New
13614 2018-05-27 Chung-Ju Wu <jasonwucj@gmail.com>
13616 * config/nds32/nds32.md (bswapsi2, bswaphi2): New patterns.
13618 2018-05-27 Chung-Ju Wu <jasonwucj@gmail.com>
13620 * config/nds32/nds32.c (nds32_attribute_table): Add "no_prologue".
13621 (nds32_init_machine_status): Initialize machine->attr_naked_p and
13622 machine->attr_no_prologue_p.
13623 (nds32_compute_stack_frame): Check "naked" and "no_prologue" attributes.
13624 (nds32_naked_function_p): Handle "naked" and "no_prologue" attributes.
13625 (nds32_expand_epilogue): Consider attr_naked_p.
13626 (nds32_expand_epilogue_v3pop): Likewise.
13627 (nds32_can_use_return_insn): Likewise.
13628 * config/nds32/nds32.h (machine_function): Add attr_naked_p and
13629 attr_no_prologue_p fields.
13630 * config/nds32/nds32.opt (mret-in-naked-func): New option.
13632 2018-05-27 Jakub Jelinek <jakub@redhat.com>
13635 * config/i386/i386.md (fixunssuffix, floatunssuffix): New code
13637 * config/i386/sse.md
13638 (<floatsuffix>float<sseintvecmodelower><mode>2<mask_name><round_name>):
13640 (float<floatunssuffix><sseintvecmodelower><mode>2<mask_name><round_name>):
13642 (<floatsuffix>float<sselongvecmodelower><mode>2<mask_name><round_name>):
13644 (float<floatunssuffix><sselongvecmodelower><mode>2<mask_name><round_name>):
13646 (*<floatsuffix>floatv2div2sf2): Rename to ...
13647 (*float<floatunssuffix>v2div2sf2): ... this.
13648 (<floatsuffix>floatv2div2sf2_mask): Rename to ...
13649 (float<floatunssuffix>v2div2sf2_mask): ... this.
13650 (*<floatsuffix>floatv2div2sf2_mask_1): Rename to ...
13651 (*float<floatunssuffix>v2div2sf2_mask_1): ... this.
13652 (<fixsuffix>fix_truncv8dfv8si2<mask_name><round_saeonly_name>): Rename
13654 (fix<fixunssuffix>_truncv8dfv8si2<mask_name><round_saeonly_name>):
13656 (<fixsuffix>fix_trunc<mode><sseintvecmodelower>2<mask_name><round_saeonly_name>):
13658 (fix<fixunssuffix>_trunc<mode><sseintvecmodelower>2<mask_name><round_saeonly_name>):
13660 (<fixsuffix>fix_trunc<mode><sselongvecmodelower>2<mask_name><round_saeonly_name>):
13662 (fix<fixunssuffix>_trunc<mode><sselongvecmodelower>2<mask_name><round_saeonly_name>):
13664 (<fixsuffix>fix_truncv2sfv2di2<mask_name>): Rename to ...
13665 (fix<fixunssuffix>_truncv2sfv2di2<mask_name>): ... this.
13666 (vec_pack_ufix_trunc_<mode>): Use gen_fixuns_truncv8dfv8si2 instead of
13667 gen_ufix_truncv8dfv8si2.
13668 * config/i386/i386-builtin.def (__builtin_ia32_cvttpd2uqq256_mask,
13669 __builtin_ia32_cvttpd2uqq128_mask, __builtin_ia32_cvttps2uqq256_mask,
13670 __builtin_ia32_cvttps2uqq128_mask, __builtin_ia32_cvtuqq2ps256_mask,
13671 __builtin_ia32_cvtuqq2ps128_mask, __builtin_ia32_cvtuqq2pd256_mask,
13672 __builtin_ia32_cvtuqq2pd128_mask, __builtin_ia32_cvttpd2udq512_mask,
13673 __builtin_ia32_cvtuqq2ps512_mask, __builtin_ia32_cvtuqq2pd512_mask,
13674 __builtin_ia32_cvttps2uqq512_mask, __builtin_ia32_cvttpd2uqq512_mask):
13675 Use fixuns instead ufix or floatuns instead ufloat in CODE_FOR_ names.
13677 2018-05-24 H.J. Lu <hongjiu.lu@intel.com>
13681 * varasm.c (assemble_alias): Lookup ifunc attribute on error.
13683 2018-05-25 Jim Wilson <jimw@sifive.com>
13685 * config/riscv/riscv-protos.h (riscv_epilogue_uses): New.
13686 * config/riscv/riscv.c (struct machine_function): Add
13687 interrupt_handler_p and attribute_checked_p fields.
13688 (riscv_attribute_table): Add interrupt.
13689 (riscv_interrupt_type_p): New.
13690 (riscv_save_reg_p): Save extra regs for interrupt handler.
13691 (riscv_use_save_libcall): Return false for interrupt handler.
13692 (riscv_first_stack_step): Add forward declaration.
13693 (riscv_compute_frame_info): New local interrupt_save_t1. Set it
13694 for interrupt handler with large frame. Use it for saved reg list.
13695 (riscv_expand_prologue): Move flag_stack_usage_info support to
13696 eliminate duplication.
13697 (riscv_expand_epilogue): Generate mret for interrupt handler.
13698 (riscv_epilogue_uses): New.
13699 (riscv_can_use_return_insn): Return false for interrupt handler.
13700 (riscv_function_ok_for_sibcall): Likewise.
13701 (riscv_set_current_function): Add interrupt handler support.
13702 * config/riscv/riscv.h (EPILOGUE_USES): Call riscv_epilogue_uses.
13703 * config/riscv/riscv.md (UNSPECV_MRET): New.
13705 (riscv_frflags, riscv_fsflags): Use tab after opcode.
13707 * doc/extend.texi (RISC-V Function Attributes) <interrupt>: New.
13709 2018-05-25 Bill Schmidt <wschmidt@linux.ibm.com>
13711 PR tree-optimization/85712
13712 * gimple-ssa-strength-reduction.c (replace_one_candidate): Skip if
13713 this candidate has already been replaced in-situ by a copy.
13715 2018-05-25 Jason Merrill <jason@redhat.com>
13717 PR c++/80485 - inline function non-zero address.
13718 * symtab.c (nonzero_address): Check DECL_COMDAT.
13720 2018-05-25 Uros Bizjak <ubizjak@gmail.com>
13723 * config/alpha/alpha.md (ashlsi3): New insn pattern.
13724 (*ashlsi_se): Rename from *ashldi_se. Define as sign
13725 extension of SImode operation. Use const123_operand predicate.
13726 (*saddsi_1): Remove.
13727 (*saddl_se_1): Ditto.
13728 (*ssubsi_1): Ditto.
13729 (*ssubl_se_1): Ditto.
13730 * config/alpha/predicates.md (const123_operand): New predicate.
13731 * config/alpha/constraints.md (P): Use IN_RANGE.
13733 2018-05-25 Richard Biener <rguenther@suse.de>
13735 * tree-ssa-alias.h (refs_may_alias_p): Add tbaa_p bool parameter,
13737 (ref_maybe_used_by_stmt_p): Likewise.
13738 (stmt_may_clobber_ref_p): Likewise.
13739 (stmt_may_clobber_ref_p_1): Likewise.
13740 * tree-ssa-alias.c (refs_may_alias_p): Add tbaa_p bool parameter
13742 (ref_maybe_used_by_stmt_p): Likewise.
13743 (stmt_may_clobber_ref_p): Likewise.
13744 (stmt_may_clobber_ref_p_1): Likewise.
13745 * tree-vect-data-refs.c (vect_slp_analyze_node_dependences): Use
13746 the alias oracle to disambiguate DRs with stmts DR analysis
13748 (vect_analyze_data_refs): Do not give up on not analyzable
13749 DRs for BB vectorization. Remove code truncating the dataref
13752 2018-05-25 Jakub Jelinek <jakub@redhat.com>
13755 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>_1):
13756 Add (=Yk,v,C) variant using vptestm insn. Use TARGET_AVX512BW
13757 in test instead of TARGET_AVX512F for VI12_AVX512VL iterator.
13759 2018-05-25 Richard Biener <rguenther@suse.de>
13761 * tree-vect-data-refs.c (vect_find_stmt_data_reference): New
13762 function, combining stmt data ref gathering and fatal analysis
13764 (vect_analyze_data_refs): Remove now redudnant code and simplify.
13765 * tree-vect-loop.c (vect_get_datarefs_in_loop): Factor out from
13766 vect_analyze_loop_2 and use vect_find_stmt_data_reference.
13767 * tree-vect-slp.c (vect_slp_bb): Use vect_find_stmt_data_reference.
13768 * tree-vectorizer.h (vect_find_stmt_data_reference): Declare.
13770 2018-05-25 Bin Cheng <bin.cheng@arm.com>
13772 PR tree-optimization/85720
13773 * tree-loop-distribution.c (break_alias_scc_partitions): Don't merge
13774 SCC if all partitions are builtins.
13775 (version_loop_by_alias_check): New parameter. Generate cancelable
13776 runtime alias check if all partitions are builtins.
13777 (distribute_loop): Update call to above function.
13779 2018-05-25 Bin Cheng <bin.cheng@arm.com>
13781 * tree-outof-ssa.c (tree-ssa.h, tree-dfa.h): Include header files.
13782 (create_default_def, for_all_parms): Moved from tree-ssa-coalesce.c.
13783 (parm_default_def_partition_arg): Ditto.
13784 (set_parm_default_def_partition): Ditto.
13785 (get_parm_default_def_partitions): Ditto and make it static.
13786 (get_undefined_value_partitions): Ditto and make it static.
13787 (remove_ssa_form): Refactor call to init_var_map here.
13788 * tree-ssa-coalesce.c (build_ssa_conflict_graph): Support live range
13789 computation for loop region.
13790 (coalesce_partitions, compute_optimized_partition_bases): Ditto.
13791 (register_default_def): Delete.
13792 (for_all_parms, create_default_def): Move to tree-outof-ssa.c.
13793 (parm_default_def_partition_arg): Ditto.
13794 (set_parm_default_def_partition): Ditto.
13795 (get_parm_default_def_partitions): Ditto and make it static.
13796 (get_undefined_value_partitions): Ditto and make it static.
13797 (coalesce_with_default, coalesce_with_default): Update comment.
13798 (create_coalesce_list_for_region): New func factored out from
13799 create_outofssa_var_map.
13800 (populate_coalesce_list_for_outofssa): New func factored out from
13801 create_outofssa_var_map and coalesce_ssa_name.
13802 (create_outofssa_var_map): Delete.
13803 (coalesce_ssa_name): Refactor to support live range computation.
13804 * tree-ssa-coalesce.h (coalesce_ssa_name): Change decl.
13805 (get_parm_default_def_partitions): Delete.
13806 (get_undefined_value_partitions): Ditto.
13807 * tree-ssa-live.c (init_var_map, delete_var_map): Support live range
13808 computation for loop region.
13809 (new_tree_live_info, loe_visit_block): Ditto.
13810 (live_worklist, set_var_live_on_entry): Ditto.
13811 (calculate_live_on_exit, verify_live_on_entry): Ditto.
13812 * tree-ssa-live.h (struct _var_map): New fields.
13813 (init_var_map): Change decl.
13814 (region_contains_p): New.
13816 2018-05-25 Bin Cheng <bin.cheng@arm.com>
13818 * tree-ssa-live.h (live_merge_and_clear): Delete.
13820 2018-05-25 Richard Biener <rguenther@suse.de>
13823 * tree-dump.c (dequeue_and_dump): Remove access to removed
13824 operand 2 of a SWITCH_EXPR.
13826 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
13828 * doc/sourcebuild.texi (vect_double_cond_arith): Include
13829 multiplication and division.
13830 * doc/md.texi (cond_mul@var{m}, cond_div@var{m}, cond_mod@var{m})
13831 (cond_udiv@var{m}, cond_umod@var{m}): Document.
13832 * optabs.def (cond_smul_optab, cond_sdiv_optab, cond_smod_optab)
13833 (cond_udiv_optab, cond_umod_optab): New optabs.
13834 * internal-fn.def (IFN_COND_MUL, IFN_COND_DIV, IFN_COND_MOD)
13835 (IFN_COND_RDIV): New internal functions.
13836 * internal-fn.c (get_conditional_internal_fn): Handle TRUNC_DIV_EXPR,
13837 TRUNC_MOD_EXPR and RDIV_EXPR.
13838 * match.pd (UNCOND_BINARY, COND_BINARY): Handle them.
13839 * config/aarch64/iterators.md (UNSPEC_COND_MUL, UNSPEC_COND_DIV):
13841 (SVE_INT_BINARY): Include mult.
13842 (SVE_COND_FP_BINARY): Include UNSPEC_MUL and UNSPEC_DIV.
13843 (optab, sve_int_op): Handle mult.
13844 (optab, sve_fp_op, commutative): Handle UNSPEC_COND_MUL and
13846 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): New pattern
13847 for SVE_INT_BINARY_SD.
13849 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
13851 * config/aarch64/iterators.md (SVE_INT_BINARY_SD): New code iterator.
13852 (optab, sve_int_op): Handle div and udiv.
13853 * config/aarch64/aarch64-sve.md (<optab><mode>3): New expander
13854 for SVE_INT_BINARY_SD.
13855 (*<optab><mode>3): New insn for the same.
13857 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
13859 * tree-vect-patterns.c: Include predict.h.
13860 (vect_recog_divmod_pattern): Restrict check for division support
13861 to when optimizing for size.
13863 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
13865 * doc/sourcebuild.texi (vect_double_cond_arith: Document.
13866 * gimple-match.h (gimple_match_op::MAX_NUM_OPS): Bump to 4.
13867 (gimple_match_op::gimple_match_op): Add an overload for 4 operands.
13868 (gimple_match_op::set_op): Likewise.
13869 (gimple_resimplify4): Declare.
13870 * genmatch.c (get_operand_type): Handle CFN_COND_* functions.
13871 (expr::gen_transform): Likewise.
13872 (decision_tree::gen): Generate a simplification routine for 4 operands.
13873 * gimple-match-head.c (gimple_simplify): Add an overload for
13874 4 operands. In the top-level function, handle up to 4 call
13875 arguments and call gimple_resimplify4.
13876 (gimple_resimplify4): New function.
13877 (build_call_internal): Pass a fourth operand.
13878 (maybe_push_to_seq): Likewise.
13879 * match.pd (UNCOND_BINARY, COND_BINARY): New operator lists.
13880 Fold VEC_COND_EXPRs of an operation and a default value into
13881 an IFN_COND_* function if possible.
13882 * config/aarch64/iterators.md (UNSPEC_COND_MAX, UNSPEC_COND_MIN):
13884 (SVE_COND_FP_BINARY): Include them.
13885 (optab, sve_fp_op): Handle them.
13886 (SVE_INT_BINARY_REV): New code iterator.
13887 (SVE_COND_FP_BINARY_REV): New int iterator.
13888 (commutative): New int attribute.
13889 * config/aarch64/aarch64-protos.h (aarch64_sve_prepare_conditional_op):
13891 * config/aarch64/aarch64.c (aarch64_sve_prepare_conditional_op): New
13893 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): Use it.
13894 (*cond_<optab><mode>): New patterns for reversed operands.
13896 2018-05-25 Richard Biener <rguenther@suse.de>
13898 * tree-vectorizer.h (STMT_VINFO_GROUP_*, GROUP_*): Remove.
13899 (DR_GROUP_*): New, assert we have non-NULL ->data_ref_info.
13900 (REDUC_GROUP_*): New, assert we have NULL ->data_ref_info.
13901 (STMT_VINFO_GROUPED_ACCESS): Adjust.
13902 * tree-vect-data-refs.c (everywhere): Adjust users.
13903 * tree-vect-loop.c (everywhere): Likewise.
13904 * tree-vect-slp.c (everywhere): Likewise.
13905 * tree-vect-stmts.c (everywhere): Likewise.
13906 * tree-vect-patterns.c (vect_reassociating_reduction_p): Likewise.
13908 2018-05-25 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
13910 * configure.ac (gcc_cv_as_section_has_e): Move to common section.
13912 (gcc_cv_as_section_exclude): ... this.
13913 Try Solaris as #exclude syntax.
13914 * configure: Regenerate.
13915 * config.in: Regenerate.
13916 * config/i386/i386.c (i386_solaris_elf_named_section): Handle
13918 * config/sparc/sparc.c (sparc_solaris_elf_asm_named_section)
13919 [HAVE_GAS_SECTION_EXCLUDE]: Handle SECTION_EXCLUDE.
13921 * varasm.c (default_elf_asm_named_section): Don't check if
13922 HAVE_GAS_SECTION_EXCLUDE is defined.
13924 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
13926 * doc/md.texi: Update the documentation of the cond_* optabs
13927 to mention the new final operand. Fix GET_MODE_NUNITS call.
13928 Describe the scalar case too.
13929 * internal-fn.def (IFN_EXTRACT_LAST): Change type to fold_left.
13930 * internal-fn.c (expand_cond_unary_optab_fn): Expect 3 operands
13932 (expand_cond_binary_optab_fn): Expect 4 operands instead of 3.
13933 (get_conditional_internal_fn): Update comment.
13934 * tree-vect-loop.c (vectorizable_reduction): Pass the original
13935 accumulator value as a final argument to conditional functions.
13936 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): Turn into
13937 a define_expand and add an "else" operand. Assert for now that
13938 the else operand is equal to operand 2. Use SVE_INT_BINARY and
13939 SVE_COND_FP_BINARY instead of SVE_COND_INT_OP and SVE_COND_FP_OP.
13940 (*cond_<optab><mode>): New patterns.
13941 * config/aarch64/iterators.md (UNSPEC_COND_SMAX, UNSPEC_COND_UMAX)
13942 (UNSPEC_COND_SMIN, UNSPEC_COND_UMIN, UNSPEC_COND_AND, UNSPEC_COND_ORR)
13943 (UNSPEC_COND_EOR): Delete.
13944 (optab): Remove associated mappings.
13945 (SVE_INT_BINARY): New code iterator.
13946 (sve_int_op): Remove int attribute and add "minus" to the code
13948 (SVE_COND_INT_OP): Delete.
13949 (SVE_COND_FP_OP): Rename to...
13950 (SVE_COND_FP_BINARY): ...this.
13952 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
13954 * optabs.c (can_reuse_operands_p): New function.
13955 (maybe_legitimize_operands): Try to reuse the results for
13958 2018-05-24 Uros Bizjak <ubizjak@gmail.com>
13960 * config/i386/sse.md (cvtusi2<ssescalarmodesuffix>64<round_name>):
13961 Add {q} suffix to insn mnemonic.
13963 2018-05-23 Jozef Lawrynowicz <jozef.l@mittosystems.com>
13965 * config/msp430/msp430.c (TARGET_WARN_FUNC_RETURN): Define.
13966 (msp430_warn_func_return): New.
13968 2018-05-24 Roger Sayle <roger@nextmovesoftware.com>
13970 * fold-const.c (tree_nonzero_bits): New function.
13971 * fold-const.h (tree_nonzero_bits): Likewise.
13972 * match.pd (POPCOUNT): New patterns to fold BUILTIN_POPCOUNT and
13973 friends. POPCOUNT(x&1) => x&1, POPCOUNT(x)==0 => x==0, etc.
13975 2018-05-24 H.J. Lu <hongjiu.lu@intel.com>
13979 * varasm.c (assemble_alias): Check ifunc_resolver only on
13982 2018-05-24 Uros Bizjak <ubizjak@gmail.com>
13985 * config/i386/sse.md (movdi_to_sse): Do not generate pseudo
13986 when memory input operand is handled.
13988 2018-05-24 Luis Machado <luis.machado@linaro.org>
13990 * config/aarch64/aarch64.c (qdf24xx_addrcost_table): New static
13992 (qdf24xx_tunings) <addr_costs>: Set to qdf24xx_addrcost_table.
13994 2018-05-24 Richard Sandiford <richard.sandiford@linaro.org>
13996 * match.pd: Delay FMA folds until after vectorization.
13998 2018-05-24 Andre Vieira <andre.simoesdiasvieira@arm.com>
14001 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Make
14002 address check not strict.
14004 2018-05-24 Richard Sandiford <richard.sandiford@linaro.org>
14006 * gimple-match.h (gimple_match_op): New class.
14007 (mprts_hook): Replace parameters with a gimple_match_op *.
14008 (maybe_build_generic_op): Likewise.
14009 (gimple_simplified_result_is_gimple_val): Replace parameters with
14010 a const gimple_match_op *.
14011 (gimple_simplify): Replace code_helper * and tree * parameters with
14012 a gimple_match_op * parameter.
14013 (gimple_resimplify1): Replace code_helper *, tree and tree *
14014 parameters with a gimple_match_op * parameter.
14015 (gimple_resimplify2): Likewise.
14016 (gimple_resimplify3): Likewise.
14017 (maybe_push_res_to_seq): Replace code_helper, tree and tree *
14018 parameters with a gimple_match_op * parameter.
14019 * gimple-match-head.c (gimple_simplify): Change prototypes of
14020 auto-generated functions to take a gimple_match_op * instead of
14021 separate code_helper * and tree * parameters. Make the same
14022 change in the top-level overload and update calls to the
14023 gimple_resimplify routines. Update calls to the auto-generated
14024 functions and to maybe_push_res_to_seq in the publicly-facing
14025 operation-specific gimple_simplify overloads.
14026 (gimple_match_op::MAX_NUM_OPS): Define.
14027 (gimple_resimplify1): Replace rcode and ops with a single res_op
14028 parameter. Update call to gimple_simplify.
14029 (gimple_resimplify2): Likewise.
14030 (gimple_resimplify3): Likewise.
14031 (mprts_hook): Replace parameters with a gimple_match_op *.
14032 (maybe_build_generic_op): Likewise.
14033 (build_call_internal): Replace type, nargs and ops with
14034 a gimple_match_op *.
14035 (maybe_push_res_to_seq): Replace res_code, type and ops parameters
14036 with a single gimple_match_op *. Update calls to mprts_hook,
14037 build_call_internal and gimple_simplified_result_is_gimple_val.
14038 Factor out code that is common to the tree_code and combined_fn cases.
14039 * genmatch.c (expr::gen_transform): Replace tem_code and
14040 tem_ops with a gimple_match_op called tem_op. Update calls
14041 to the gimple_resimplify functions and maybe_push_res_to_seq.
14042 (dt_simplify::gen_1): Manipulate res_op instead of res_code and
14043 res_ops. Update call to the gimple_resimplify functions.
14044 (dt_simplify::gen): Pass res_op instead of res_code and res_ops.
14045 (decision_tree::gen): Make the functions take a gimple_match_op *
14046 called res_op instead of separate res_code and res_ops parameters.
14047 Update call accordingly.
14048 * gimple-fold.c (replace_stmt_with_simplification): Replace rcode
14049 and ops with a single res_op parameter. Update calls to
14050 maybe_build_generic_op and maybe_push_res_to_seq.
14051 (fold_stmt_1): Update calls to gimple_simplify and
14052 replace_stmt_with_simplification.
14053 (gimple_fold_stmt_to_constant_1): Update calls to gimple_simplify
14054 and gimple_simplified_result_is_gimple_val.
14055 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update call to
14057 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Replace parameters
14058 with a gimple_match_op *.
14059 (vn_nary_build_or_lookup): Likewise. Update call to
14060 vn_nary_build_or_lookup_1.
14061 (vn_nary_build_or_lookup_1): Replace rcode, type and ops with a
14062 gimple_match_op *. Update calls to the gimple_resimplify routines
14063 and to gimple_simplified_result_is_gimple_val.
14064 (vn_nary_simplify): Update call to vn_nary_build_or_lookup_1.
14065 Use gimple_match_op::MAX_NUM_OPS instead of a hard-coded 3.
14066 (vn_reference_lookup_3): Update call to vn_nary_build_or_lookup.
14067 (visit_nary_op): Likewise.
14068 (visit_reference_op_load): Likewise.
14070 2018-05-23 Luis Machado <luis.machado@linaro.org>
14072 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Use correct type
14073 modifier for printing the step amount.
14075 2018-05-23 Jozef Lawrynowicz <jozef.l@somniumtech.com>
14078 * tree.c (build_common_tree_nodes): Dont set TYPE_SIZE for __intN
14081 2018-05-23 Segher Boessenkool <segher@kernel.crashing.org>
14083 * doc/sourcebuild.texi (Endianness): New subsubsection.
14085 2018-05-23 Luis Machado <luis.machado@linaro.org>
14087 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
14088 <prefetch_dynamic_strides>: New const bool field.
14089 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
14090 prefetch_dynamic_strides.
14091 (exynosm1_prefetch_tune): Likewise.
14092 (thunderxt88_prefetch_tune): Likewise.
14093 (thunderx_prefetch_tune): Likewise.
14094 (thunderx2t99_prefetch_tune): Likewise.
14095 (qdf24xx_prefetch_tune): Likewise. Set prefetch_dynamic_strides to
14097 (aarch64_override_options_internal): Update to set
14098 PARAM_PREFETCH_DYNAMIC_STRIDES.
14099 * doc/invoke.texi (prefetch-dynamic-strides): Document new option.
14100 * params.def (PARAM_PREFETCH_DYNAMIC_STRIDES): New.
14101 * params.h (PARAM_PREFETCH_DYNAMIC_STRIDES): Define.
14102 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Account for
14103 prefetch-dynamic-strides setting.
14105 2018-05-23 Luis Machado <luis.machado@linaro.org>
14107 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
14108 <minimum_stride>: New const int field.
14109 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
14110 minimum_stride field defaulting to -1.
14111 (exynosm1_prefetch_tune): Likewise.
14112 (thunderxt88_prefetch_tune): Likewise.
14113 (thunderx_prefetch_tune): Likewise.
14114 (thunderx2t99_prefetch_tune): Likewise.
14115 (qdf24xx_prefetch_tune) <minimum_stride>: Set to 2048.
14116 <default_opt_level>: Set to 3.
14117 (aarch64_override_options_internal): Update to set
14118 PARAM_PREFETCH_MINIMUM_STRIDE.
14119 * doc/invoke.texi (prefetch-minimum-stride): Document new option.
14120 * params.def (PARAM_PREFETCH_MINIMUM_STRIDE): New.
14121 * params.h (PARAM_PREFETCH_MINIMUM_STRIDE): Define.
14122 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Return false if
14123 stride is constant and is below the minimum stride threshold.
14125 2018-05-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14127 * config/arm/arm-cpus.in (mode26): Delete.
14128 (armv4): Delete mode26 reference.
14129 * config/arm/arm.c (arm_configure_build_target): Delete use of
14132 2018-05-23 Uros Bizjak <ubizjak@gmail.com>
14134 * config/i386/i386.md (*floatuns<SWI48:mode><MODEF:mode>2_avx512):
14136 (floatunssi<mode>2): Also enable for AVX512F and TARGET_SSE_MATH.
14137 Rewrite expander pattern. Emit gen_floatunssi<mode>2_i387_with_xmm
14139 (floatunsdisf2): Rewrite expander pattern. Hanlde TARGET_AVX512F.
14140 (floatunsdidf2): Ditto.
14142 2018-05-23 Uros Bizjak <ubizjak@gmail.com>
14144 * config/i386/i386.md (fixuns_trunc<mode>di2): New insn pattern.
14145 (fixuns_trunc<mode>si2_avx512f): Ditto.
14146 (*fixuns_trunc<mode>si2_avx512f_zext): Ditto.
14147 (fixuns_trunc<mode>si2): Also enable for AVX512F and TARGET_SSE_MATH.
14148 Emit fixuns_trunc<mode>si2_avx512f for AVX512F targets.
14150 2018-05-23 Alexander Monakov <amonakov@ispras.ru>
14152 PR rtl-optimization/79985
14153 * df-scan.c (df_insn_refs_collect): Remove special case for
14154 global registers and asm statements.
14156 2018-05-23 Alexander Monakov <amonakov@ispras.ru>
14158 * extend.texi (Global Register Variables): Rewrite the bullet list.
14159 Note that the register is available for allocation. Note that access
14160 via inline asm must use constraints. Add note about async-signal
14161 handlers. Remove paragraph about automagic register selection.
14163 2018-05-23 Richard Biener <rguenther@suse.de>
14165 * tree-ssa-sccvn.c (vn_reference_lookup_3): Remove restriction
14166 of fixed offset from memset VN.
14168 2018-05-23 Bill Schmidt <wschmidt@linux.ibm.com>
14170 * gimple-ssa-strength-reduction.c (struct slsr_cand_d): Add
14171 first_interp field.
14172 (alloc_cand_and_find_basis): Initialize first_interp field.
14173 (slsr_process_mul): Modify first_interp field.
14174 (slsr_process_add): Likewise.
14175 (slsr_process_cast): Modify first_interp field for each new
14177 (slsr_process_copy): Likewise.
14178 (dump_candidate): Dump first_interp field.
14179 (replace_mult_candidate): Process all interpretations, not just
14181 (replace_rhs_if_not_dup): Likewise.
14182 (replace_one_candidate): Likewise.
14184 2018-05-23 Wilco Dijkstra <wdijkstr@arm.com>
14186 * config/aarch64/aarch64.c (aarch64_use_frame_pointer):
14188 (aarch64_needs_frame_chain): New function.
14189 (aarch64_parse_override_string): Set aarch64_use_frame_pointer.
14191 2018-05-23 Sudakshina Das <sudi.das@arm.com>
14194 * common/config/aarch64/aarch64-common.c (aarch64_handle_option):
14195 Check val before adding MASK_STRICT_ALIGN to opts->x_target_flags.
14196 * config/aarch64/aarch64.opt (mstrict-align): Remove RejectNegative.
14197 * config/aarch64/aarch64.c (aarch64_attributes): Mark allow_neg
14198 as true for strict-align.
14199 (aarch64_can_inline_p): Perform checks even when callee has no
14200 attributes to check for strict alignment.
14201 * doc/extend.texi (AArch64 Function Attributes): Document
14203 * doc/invoke.texi: (AArch64 Options): Likewise.
14205 2018-05-23 Richard Sandiford <richard.sandiford@linaro.org>
14207 PR tree-optimization/85853
14208 * tree-vect-slp.c (vect_slp_analyze_node_operations): Split out
14209 the handling of the root of the node to...
14210 (vect_slp_analyze_node_operations_1): ...this new function,
14211 and run the whole thing with the child nodes' def types
14212 set according to their SLP node's def type.
14214 2018-05-23 Richard Biener <rguenther@suse.de>
14216 PR middle-end/85874
14217 * tree-data-ref.c (create_runtime_alias_checks): Defer
14218 and ignore overflow warnings.
14220 2018-05-23 Yury Gribov <tetra2005@gmail.com>
14222 PR tree-optimization/85822
14223 * tree-vrp.c (is_masked_range_test): Fix handling of negative
14226 2018-05-23 Richard Biener <rguenther@suse.de>
14228 * tree-ssa-sccvn.c (vn_reference_lookup_3): Handle arbitrary
14229 memset constants via native_interpret_expr.
14231 2018-05-22 H.J. Lu <hongjiu.lu@intel.com>
14234 * cgraph.h (cgraph_node::create): Set ifunc_resolver for ifunc
14236 (cgraph_node::create_alias): Likewise.
14237 (cgraph_node::get_availability): Check ifunc_resolver instead
14238 of looking up ifunc attribute.
14239 * cgraphunit.c (maybe_diag_incompatible_alias): Likewise.
14240 * varasm.c (do_assemble_alias): Likewise.
14241 (assemble_alias): Likewise.
14242 (default_binds_local_p_3): Likewise.
14243 * cgraph.h (cgraph_node): Add ifunc_resolver.
14244 (cgraph_node::only_called_directly_or_aliased_p): Return false
14245 for IFUNC resolver.
14246 * lto-cgraph.c (input_node): Set ifunc_resolver for ifunc
14248 * symtab.c (symtab_node::verify_base): Verify that ifunc_resolver
14249 is equivalent to lookup_attribute ("ifunc", DECL_ATTRIBUTES (decl)).
14250 (symtab_node::binds_to_current_def_p): Check ifunc_resolver
14251 instead of looking up ifunc attribute.
14253 2018-05-22 Luis Machado <luis.machado@linaro.org>
14255 * config/aarch64/aarch64.md (*ashift<mode>_extv_bfiz): New pattern.
14257 2018-05-22 Martin Sebor <msebor@redhat.com>
14259 PR middle-end/85359
14260 * builtins.c (expand_builtin_strcpy): Call maybe_warn_nonstring_arg
14261 only when expasion succeeds.
14262 (expand_builtin_strcmp): Same.
14263 (expand_builtin_strncmp): Same.
14265 2018-05-22 Martin Sebor <msebor@redhat.com>
14267 * calls.c (maybe_warn_nonstring_arg): Fix a typo in a comment.
14269 2018-05-22 Jackson Woodruff <jackson.woodruff@arm.com>
14270 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14272 * config/aarch64/aarch64-ldpstp.md: Replace uses of
14273 aarch64_mem_pair_operand with memory_operand and delete operand swapping
14275 * config/aarch64/aarch64.c (aarch64_operands_ok_for_ldpstp):
14276 Add check for legitimate_address.
14277 (aarch64_gen_adjusted_ldpstp): Swap operands where appropriate.
14278 (aarch64_swap_ldrstr_operands): New.
14279 * config/aarch64/aarch64-protos.h (aarch64_swap_ldrstr_operands):
14282 2018-05-22 Jackson Woodruff <jackson.woodruff@arm.com>
14283 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14285 * config/aarch64/aarch64.md: New patterns to generate stp
14287 (store_pair_sw, store_pair_dw): New patterns to generate stp for
14288 single words and double words.
14289 (load_pair_sw, load_pair_dw): Likewise.
14290 (store_pair_sf, store_pair_df, store_pair_si, store_pair_di):
14292 (load_pair_sf, load_pair_df, load_pair_si, load_pair_di):
14294 * config/aarch64/aarch64-ldpstp.md: Modify peephole
14295 for different mode ldpstp and add peephole for merged zero stores.
14296 Likewise for loads.
14297 * config/aarch64/aarch64.c (aarch64_operands_ok_for_ldpstp):
14299 (aarch64_gen_store_pair): Rename calls to match new patterns.
14300 (aarch64_gen_load_pair): Rename calls to match new patterns.
14301 * config/aarch64/aarch64-simd.md (load_pair<mode>): Rename to...
14302 (load_pair<DREG:mode><DREG2:mode>): ... This.
14303 (store_pair<mode>): Rename to...
14304 (vec_store_pair<DREG:mode><DREG2:mode>): ... This.
14305 * config/aarch64/iterators.md (DREG, DREG2, DX2, SX, SX2, DSX):
14306 New mode iterators.
14307 (V_INT_EQUIV): Handle SImode.
14308 * config/aarch64/predicates.md (aarch64_reg_zero_or_fp_zero):
14311 2018-05-22 Martin Sebor <msebor@redhat.com>
14314 * calls.c (maybe_warn_nonstring_arg): Use string length to set
14315 or ajust the presumed bound on an operation to avoid unnecessary
14318 2018-05-22 Martin Sebor <msebor@redhat.com>
14320 PR tree-optimization/85826
14321 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Avoid
14322 assuming that a DECL necesarily has a constant size.
14324 2018-05-22 Richard Sandiford <richard.sandiford@linaro.org>
14326 PR middle-end/85862
14327 * internal-fn.c (expand_direct_optab_fn): Cope with a null lhs.
14329 2018-05-22 Richard Biener <rguenther@suse.de>
14331 PR tree-optimization/85834
14332 * tree-ssa-sccvn.c (vn_reference_lookup_3): Properly handle
14333 non-constant and non-zero memset arguments.
14335 2018-05-22 Martin Liska <mliska@suse.cz>
14338 * ipa-icf.c (sem_item::add_type): Do not ICE for incomplete types.
14340 2018-05-22 Richard Biener <rguenther@suse.de>
14342 PR tree-optimization/85863
14343 * tree-vect-stmts.c (vect_is_simple_cond): Only widen invariant
14344 comparisons when vectype is specified.
14345 (vectorizable_condition): Do not specify vectype for
14346 vect_is_simple_cond when SLP vectorizing.
14348 2018-05-21 Michael Meissner <meissner@linux.ibm.com>
14351 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Do not
14352 define __ibm128 as long double.
14353 * config/rs6000/rs6000.c (rs6000_init_builtins): Create __ibm128
14354 as a distinct type when IEEE 128-bit support is enabled.
14355 (init_float128_ieee): Fix up conversions between IFmode and IEEE
14356 128-bit types to use the correct functions.
14357 (rs6000_expand_float128_convert): Use explicit FLOAT_EXTEND to
14358 convert between 128-bit floating point types that have different
14359 modes but the same representation, instead of using gen_lowpart to
14361 * config/rs6000/rs6000.md (IFKF): New iterator for IFmode and
14363 (IFKF_reg): New attributes to give the register constraints for
14365 (extend<mode>tf2_internal): New insns to mark an explicit
14366 conversion between 128-bit floating point types that have a
14367 different mode but share the same representation.
14369 2018-05-21 Richard Sandiford <richard.sandiford@linaro.org>
14371 PR tree-optimization/85814
14372 * tree-ssa-strlen.c (get_stridx_plus_constant): Cope with
14373 a null return from get_strinfo when unsharing the next
14374 strinfo in the chain.
14376 2018-05-21 Vladimir Mezentsev <vladimir.mezentsev@oracle.com>
14379 * varasm.c (weak_finish): Clean up weak_decls.
14381 2018-05-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14383 * config/aarch64/aarch64.md ("unspec"): Define UNSPEC_SABAL,
14384 UNSPEC_SABDL2, UNSPEC_SADALP, UNSPEC_UABAL, UNSPEC_UABDL2,
14385 UNSPEC_UADALP values.
14386 * config/aarch64/iterators.md (ABAL): New int iterator.
14389 (sur): Add mappings for the above.
14390 * config/aarch64/aarch64-simd.md (aarch64_<sur>abdl2<mode>_3):
14392 (aarch64_<sur>abal<mode>_4): Likewise.
14393 (aarch64_<sur>adalp<mode>_3): Likewise.
14394 (<sur>sadv16qi): New define_expand.
14396 2018-05-21 Alexander Nesterovskiy <alexander.nesterovskiy@intel.com>
14398 * config/i386/i386.md (*movsf_internal): AVX falsedep fix.
14399 (*movdf_internal): Ditto.
14400 (*rcpsf2_sse): Ditto.
14401 (*rsqrtsf2_sse): Ditto.
14402 (*sqrt<mode>2_sse): Ditto.
14404 2018-05-21 Tamar Christina <tamar.christina@arm.com>
14406 * config/aarch64/aarch64-simd.md (aarch64_eor3qv8hi): Change to
14408 (aarch64_bcaxqv8hi): Change to bcaxq<mode>4.
14409 * config/aarch64/aarch64-simd-builtins.def (veor3q_u8, veor3q_u32,
14410 veor3q_u64, veor3q_s8, veor3q_s16, veor3q_s32, veor3q_s64, vbcaxq_u8,
14411 vbcaxq_u32, vbcaxq_u64, vbcaxq_s8, vbcaxq_s16, vbcaxq_s32,
14413 * config/aarch64/arm_neon.h: Likewise.
14414 * config/aarch64/iterators.md (VQ_I): New.
14416 2018-05-21 Alexey Brodkin <abrodkin@synopsys.com>
14418 * config.gcc: Add arc/t-multilib-linux to tmake_file for
14420 * config/arc/t-multilib-linux: Specify MULTILIB_OPTIONS and
14423 2018-05-20 Chung-Ju Wu <jasonwucj@gmail.com>
14425 * config/nds32/constraints.md (S): New constraint.
14426 * config/nds32/nds32.md (call_internal): Use constraint S.
14427 (call_value_internal): Likewise.
14428 (sibcall_internal): Likewise.
14429 (sibcall_value_internal): Likewise.
14431 2018-05-20 Kito Cheng <kito.cheng@gmail.com>
14432 Chung-Ju Wu <jasonwucj@gmail.com>
14434 * config/nds32/nds32.c (nds32_register_move_cost): Take garywolf cpu
14435 into consideration.
14437 2018-05-20 Kito Cheng <kito.cheng@gmail.com>
14438 Chung-Ju Wu <jasonwucj@gmail.com>
14440 * config/nds32/nds32-cost.c (rtx_cost_model_t): New structure.
14441 (insn_size_16bit, insn_size_32bit): New variables for cost evaluation.
14442 (nds32_rtx_costs_impl): Simplify.
14443 (nds32_address_cost_impl): Simplify.
14444 (nds32_init_rtx_costs): New function.
14445 (nds32_rtx_costs_speed_prefer): Likewise.
14446 (nds32_rtx_costs_size_prefer): Likewise.
14447 (nds32_address_cost_speed_prefer): Likewise.
14448 (nds32_address_cost_speed_fwprop): Likewise.
14449 (nds32_address_cost_size_prefer): Likewise.
14450 * config/nds32/nds32-protos.h (nds32_init_rtx_costs): Declare.
14451 * config/nds32/nds32.c (nds32_option_override): Use
14452 nds32_init_rtx_costs function.
14454 2018-05-20 Chung-Ju Wu <jasonwucj@gmail.com>
14456 * config/nds32/nds32.c (nds32_asm_file_start): Output pipeline model.
14457 * config/nds32/nds32.h (TARGET_PIPELINE_N7): Define.
14458 (TARGET_PIPELINE_N8): Likewise.
14459 (TARGET_PIPELINE_N10): Likewise.
14460 (TARGET_PIPELINE_N13): Likewise.
14461 (TARGET_PIPELINE_GRAYWOLF): Likewise.
14463 2018-05-19 Monk Chiang <sh.chiang04@gmail.com>
14465 * config/nds32/nds32-fpu.md: Update copyright year.
14467 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
14469 * config/nds32/nds32.h (ASM_SPEC): Adjust spec rule.
14471 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
14473 * config/nds32/nds32.c
14474 (nds32_md_asm_adjust): Consider flag_inline_asm_r15 variable.
14475 * config/nds32/nds32.opt (minline-asm-r15): New option.
14477 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
14479 * common/config/nds32/nds32-common.c (TARGET_DEFAULT_TARGET_FLAGS): Add
14481 * config/nds32/nds32.md (abssi2): New pattern.
14483 2018-05-19 Uros Bizjak <ubizjak@gmail.com>
14485 * config/i386/i386.md (rex64namesuffix): New mode attribute.
14486 * config/i386/sse.md (sse_cvtsi2ss<rex64namesuffix><round_name>):
14487 Merge insn pattern from sse_cvtsi2ss<round_name> and
14488 sse_cvtsi2ssq<round_name> using SWI48 mode iterator.
14489 (sse_cvtss2si<rex64namesuffix><round_name>): Merge insn pattern
14490 from sse_cvtss2si<round_name> and sse_cvtss2siq<round_name>
14491 using SWI48 mode iterator.
14492 (sse_cvtss2si<rex64namesuffix>_2): Merge insn pattern from
14493 sse_cvtss2si_2 and sse_cvtss2siq_2 using SWI48 mode iterator.
14494 (sse_cvttss2si<rex64namesuffix><round_saeonly_name>): Merge insn
14495 pattern from sse_cvttss2si<round_saeonly_name>
14496 and sse_cvttss2siq<round_saeonly_name> using SWI48 mode iterator.
14497 (avx512f_vcvtss2usi<rex64namesuffix><round_name>): Merge insn pattern
14498 from avx512f_vcvtss2usi<round_name> and avx512f_vcvtss2usiq<round_name>
14499 using SWI48 mode iterator.
14500 (avx512f_vcvttss2usi<rex64namesuffix><round_saeonly_name>): Merge
14501 insn pattern from avx512f_vcvttss2usi<round_saeonly_name> and
14502 avx512f_vcvttss2usiq<round_saeonly_name> using SWI48 mode iterator.
14503 (avx512f_vcvtsd2usi<rex64namesuffix><round_name>): Merge insn pattern
14504 from avx512f_vcvtsd2usi<round_name> and avx512f_vcvtsd2usiq<round_name>
14505 using SWI48 mode iterator.
14506 (avx512f_vcvttsd2usi<rex64namesuffix><round_saeonly_name>): Merge
14507 insn pattern from avx512f_vcvttsd2usi<round_saeonly_name> and
14508 avx512f_vcvttsd2usiq<round_saeonly_name> using SWI48 mode iterator.
14509 (sse2_cvtsd2si<rex64namesuffix><round_name>): Merge insn pattern from
14510 sse2_cvtsd2si<round_name> and sse2_cvtsd2siq<round_name> using
14511 SWI48 mode iterator.
14512 (sse2_cvtsd2si<rex64namesuffix>_2): Merge insn pattern from
14513 sse2_cvtsd2si_2 and sse2_cvtsd2siq_2 using SWI48 mode iterator.
14514 (sse_cvttsd2si<rex64namesuffix><round_saeonly_name>): Merge insn
14515 pattern from sse_cvttsd2si<round_saeonly_name>
14516 and sse_cvttsd2siq<round_saeonly_name> using SWI48 mode iterator.
14518 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
14520 * config/nds32/nds32-md-auxiliary.c
14521 (nds32_valid_smw_lwm_base_p): Refine.
14522 (nds32_output_smw_single_word): Refine.
14523 (nds32_output_smw_double_word): New.
14524 * config/nds32/nds32-protos.h (nds32_output_smw_double_word): New.
14526 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
14528 * config/nds32/nds32-md-auxiliary.c (nds32_output_stack_push): Refine.
14529 (nds32_output_stack_pop): Refine.
14530 (nds32_expand_unaligned_load): Refine.
14531 (nds32_expand_unaligned_store): Refine.
14533 2018-05-19 Kuan-Lin Chen <kuanlinchentw@gmail.com>
14534 Chung-Ju Wu <jasonwucj@gmail.com>
14536 * config/nds32/constants.md: Add TP_REGNUM constant.
14537 (unspec_element): Add UNSPEC_GOTINIT, UNSPEC_GOT, UNSPEC_GOTOFF,
14538 UNSPEC_PLT, UNSPEC_TLSGD, UNSPEC_TLSLD, UNSPEC_TLSIE, UNSPEC_TLSLE and
14540 * config/nds32/nds32-doubleword.md: Consider flag_pic.
14541 * config/nds32/nds32-dspext.md (mov<mode>): Expand TLS and PIC cases.
14542 * config/nds32/nds32-predicates.c (nds32_const_unspec_p): New.
14543 * config/nds32/nds32-md-auxiliary.c: Implementation that support TLS
14544 and PIC code generation.
14545 * config/nds32/nds32-protos.h: Declarations that support TLS and PIC
14547 * config/nds32/nds32-relax-opt.c: Consider TLS and PIC for relax
14549 * config/nds32/nds32.md: Support TLS and PIC.
14550 * config/nds32/nds32.c: Support TLS and PIC.
14551 * config/nds32/nds32.h (nds32_relax_insn_type): New enum type.
14552 * config/nds32/predicates.md (nds32_nonunspec_symbolic_operand): New
14555 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
14557 * config/nds32/nds32-predicates.c (const_vector_to_hwint): Use machine
14558 mode with E_ prefix.
14560 2018-05-19 Kuan-Lin Chen <kuanlinchentw@gmail.com>
14561 Chung-Ju Wu <jasonwucj@gmail.com>
14563 * config/nds32/constants.md (unspec_element): Add UNSPEC_ICT.
14564 * config/nds32/nds32-md-auxiliary.c
14565 (symbolic_reference_mentioned_p): New.
14566 (nds32_legitimize_ict_address): New.
14567 (nds32_expand_ict_move): New.
14568 (nds32_indirect_call_referenced_p): New.
14569 (nds32_symbol_binds_local_p): Delete.
14570 (nds32_long_call_p): Modify.
14571 * config/nds32/nds32-opts.h (nds32_ict_model_type): New enum type.
14572 * config/nds32/nds32-protos.h
14573 (symbolic_reference_mentioned_p): Declare.
14574 (nds32_legitimize_ict_address): Declare.
14575 (nds32_expand_ict_move): Declare.
14576 (nds32_indirect_call_referenced_p): Declare.
14577 * config/nds32/nds32-relax-opt.c (nds32_ict_const_p): New.
14578 (nds32_relax_group): Use nds32_ict_const_p as condition.
14579 * config/nds32/nds32.c (nds32_attribute_table): Add "indirect_call".
14580 (nds32_asm_file_start): Output ict_model directive in asm code.
14581 (nds32_legitimate_address_p): Consider indirect call.
14582 (nds32_print_operand): Consider indirect call.
14583 (nds32_print_operand_address): Consider indirect call.
14584 (nds32_insert_attributes): Handle "indirect_call" attribute.
14585 (TARGET_LEGITIMATE_ADDRESS_P): Define.
14586 (TARGET_LEGITIMATE_CONSTANT_P): Define.
14587 (TARGET_CANNOT_FORCE_CONST_MEM): Define.
14588 (TARGET_DELEGITIMIZE_ADDRESS): Define.
14589 (TARGET_ASM_OUTPUT_ADDR_CONST_EXTRA): Define.
14590 * config/nds32/nds32.h (SYMBOLIC_CONST_P): Define.
14591 (TARGET_ICT_MODEL_SMALL): Define.
14592 (TARGET_ICT_MODEL_LARGE): Define.
14593 * config/nds32/nds32.md (movsi): Consider ict model.
14594 (call, call_value): Consider ict model.
14595 (sibcall, sibcall_value): Consider ict model.
14596 * config/nds32/nds32.opt (mict-model): New option.
14597 * config/nds32/predicates.md (nds32_symbolic_operand): Consider ict
14600 2018-05-18 Kito Cheng <kito.cheng@gmail.com>
14601 Monk Chiang <sh.chiang04@gmail.com>
14602 Jim Wilson <jimw@sifive.com>
14604 * common/config/riscv/riscv-common.c (riscv_parse_arch_string):
14605 Add support to parse rv32e*. Clear MASK_RVE for rv32i and rv64i.
14606 * config.gcc (riscv*-*-*): Add support for rv32e* and ilp32e.
14607 * config/riscv/riscv-c.c (riscv_cpu_cpp_builtins): Define
14608 __riscv_32e when TARGET_RVE. Handle ABI_ILP32E as soft-float ABI.
14609 * config/riscv/riscv-opts.h (riscv_abi_type): Add ABI_ILP32E.
14610 * config/riscv/riscv.c (riscv_compute_frame_info): When TARGET_RVE,
14611 compute save_libcall_adjustment properly.
14612 (riscv_option_override): Call error if TARGET_RVE and not ABI_ILP32E.
14613 (riscv_conditional_register_usage): Handle TARGET_RVE and ABI_ILP32E.
14614 * config/riscv/riscv.h (UNITS_PER_FP_ARG): Handle ABI_ILP32E.
14615 (STACK_BOUNDARY, ABI_STACK_BOUNDARY): Handle TARGET_RVE.
14616 (GP_REG_LAST, MAX_ARGS_IN_REGISTERS): Likewise.
14617 (ABI_SPEC): Handle mabi=ilp32e.
14618 * config/riscv/riscv.opt (abi_type): Add ABI_ILP32E.
14619 (RVE): Add RVE mask.
14620 * doc/invoke.texi (RISC-V options) <-mabi>: Add ilp32e info.
14621 <-march>: Add rv32e as an example.
14623 2018-05-18 Marc Glisse <marc.glisse@inria.fr>
14626 * tree-ssa-structalias.c (create_variable_info_for_1): Extra argument.
14627 (intra_create_variable_infos): Handle C++ constructors.
14629 2018-05-18 Martin Liska <mliska@suse.cz>
14631 * passes.def: Remove a redundant pass.
14633 2018-05-18 Eric Botcazou <ebotcazou@adacore.com>
14636 * config/sparc/sparc.c (sparc_expand_builtin): Always initialize op[0].
14638 2018-05-18 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14640 * config/arm/arm-cpus.in (armv3m, mode32): Delete features.
14642 (ARMv2, ARMv3, ARMv3m): Delete fgroups.
14644 (armv2, armv2a, armv3, armv3m): Delete architectures.
14645 (arm2, arm250, arm3, arm6, arm60, arm600, arm610, arm620,
14646 arm7, arm7d, arm7di, arm70, arm700, arm700i, arm710, arm720,
14647 arm710c, arm7100, arm7500, arm7500fe, arm7m, arm7dm, arm7dmi):
14649 * config/arm/arm.md (maddsidi4): Remove check for arm_arch3m.
14650 (*mulsidi3adddi): Likewise.
14651 (mulsidi3): Likewise.
14652 (*mulsidi3_nov6): Likewise.
14653 (umulsidi3): Likewise.
14654 (umulsidi3_nov6): Likewise.
14655 (umaddsidi4): Likewise.
14656 (*umulsidi3adddi): Likewise.
14657 (smulsi3_highpart): Likewise.
14658 (*smulsi3_highpart_nov6): Likewise.
14659 (umulsi3_highpart): Likewise.
14660 (*umulsi3_highpart_nov6): Likewise.
14661 * config/arm/arm.h (arm_arch3m): Delete.
14662 * config/arm/arm.c (arm_arch3m): Delete.
14663 (arm_option_override_internal): Update armv3-related comment.
14664 (arm_configure_build_target): Delete use of isa_bit_mode32.
14665 (arm_option_reconfigure_globals): Delete set of arm_ach3m.
14666 (arm_rtx_costs_internal): Delete check of arm_arch3m.
14667 * config/arm/arm-fixed.md (mulsq3): Delete check for arm_arch3m.
14668 (mulsa3): Likewise.
14669 (mulusa3): Likewise.
14670 * config/arm/arm-protos.h (arm_arch3m): Delete.
14671 * config/arm/arm-tables.opt: Regenerate.
14672 * config/arm/arm-tune.md: Likewise.
14673 * config/arm/t-arm-elf (all_early_nofp): Delete mentions of
14674 deleted architectures.
14676 2018-05-18 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14678 * config/arm/arm-cpus.in (armv5, armv5e): Delete features.
14679 (armv5t, armv5te): New features.
14680 (ARMv5, ARMv5e): Delete fgroups.
14681 (ARMv5t, ARMv5te): Adjust for above changes.
14682 (ARMv6m): Likewise.
14683 (armv5, armv5e): Delete arches.
14684 * config/arm/arm.md (*call_reg_armv5): Use arm_arch5t instead of
14686 (*call_reg_arm): Likewise.
14687 (*call_value_reg_armv5): Likewise.
14688 (*call_value_reg_arm): Likewise.
14689 (*call_symbol): Likewise.
14690 (*call_value_symbol): Likewise.
14691 (*sibcall_insn): Likewise.
14692 (*sibcall_value_insn): Likewise.
14693 (clzsi2): Likewise.
14694 (prefetch): Likewise.
14695 (define_split and define_peephole2 dependent on arm_arch5):
14697 * config/arm/arm.h (TARGET_LDRD): Use arm_arch5te instead of
14699 (TARGET_ARM_QBIT): Likewise.
14700 (TARGET_DSP_MULTIPLY): Likewise.
14701 (enum base_architecture): Delete BASE_ARCH_5, BASE_ARCH_5E.
14702 (arm_arch5, arm_arch5e): Delete.
14703 (arm_arch5t, arm_arch5te): Declare.
14704 * config/arm/arm.c (arm_arch5, arm_arch5e): Delete.
14705 (arm_arch5t): Declare.
14706 (arm_option_reconfigure_globals): Update for the above.
14707 (arm_options_perform_arch_sanity_checks): Update comment, replace
14708 use of arm_arch5 with arm_arch5t.
14709 (use_return_insn): Likewise.
14710 (arm_emit_call_insn): Likewise.
14711 (output_return_instruction): Likewise.
14712 (arm_final_prescan_insn): Likewise.
14713 (arm_coproc_builtin_available): Likewise.
14714 * config/arm/arm-c.c (arm_cpu_builtins): Replace arm_arch5 and
14715 arm_arch5e with arm_arch5t and arm_arch5te.
14716 * config/arm/arm-protos.h (arm_arch5, arm_arch5e): Delete.
14717 (arm_arch5t, arm_arch5te): Declare.
14718 * config/arm/arm-tables.opt: Regenerate.
14719 * config/arm/t-arm-elf: Remove references to armv5, armv5e.
14720 * config/arm/t-multilib: Likewise.
14721 * config/arm/thumb1.md (*call_reg_thumb1_v5): Check arm_arch5t
14722 instead of arm_arch5.
14723 (*call_reg_thumb1): Likewise.
14724 (*call_value_reg_thumb1_v5): Likewise.
14725 (*call_value_reg_thumb1): Likewise.
14726 * config/arm/vxworks.h (TARGET_OS_CPP_BUILTINS): Remove now
14728 * doc/invoke.texi (ARM Options): Remove references to armv5, armv5e.
14730 2018-05-18 Martin Liska <mliska@suse.cz>
14732 PR gcov-profile/84846
14733 * doc/gcov.texi: Document -t option of gcov tool.
14735 2018-05-18 Martin Liska <mliska@suse.cz>
14737 PR gcov-profile/84846
14738 * gcov.c (print_usage): Add new -t option.
14739 (process_args): Handle the option.
14740 (generate_results): Use stdout as output when requested by
14743 2018-05-18 Martin Liska <mliska@suse.cz>
14745 PR gcov-profile/84846
14746 * coverage.c (coverage_init): Write PWD to .gcno file.
14747 * doc/gcov.texi: Document how working directory is printed.
14748 * gcov-dump.c (dump_gcov_file): Print PWD.
14749 * gcov.c (output_intermediate_file): Likewise.
14750 (read_graph_file): Read PWD string.
14751 (output_lines): Print PWD.
14753 2018-05-18 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
14755 PR middle-end/85817
14756 * ipa-pure-const.c (malloc_candidate_p): Remove the check integer_zerop
14757 for retval and return false if all args to phi are zero.
14759 2018-05-18 Richard Biener <rguenther@suse.de>
14761 * gimple-ssa-evrp.c (class evrp_folder): Add simplify_stmt_using_ranges
14763 (evrp_dom_walker::before_dom_children): Call it.
14765 2018-05-18 Richard Biener <rguenther@suse.de>
14767 * tree-dfa.c (get_ref_base_and_extent): Use range-info to refine
14768 results when processing array refs with variable index.
14770 2018-05-18 Toon Moene <toon@moene.org>
14772 * doc/invoke.texi: Move -floop-unroll-and-jam documentation
14773 directly after that of -floop-interchange. Indicate that both
14774 options are enabled by default when specifying -O3.
14776 2018-05-18 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14778 * config/aarch64/aarch64-simd.md (vec_set<mode>): Use VALL_F16 mode
14779 iterator. Delete separate integer-mode vec_set<mode> expander.
14780 (aarch64_simd_vec_setv2di): Delete.
14781 (vec_setv2di): Delete.
14782 (aarch64_simd_vec_set<mode>): Delete all other patterns with that name.
14783 Use VALL_F16 mode iterator. Add LD1 alternative and use vwcore for
14784 the "w, r" alternative.
14786 2018-05-18 Martin Liska <mliska@suse.cz>
14788 * passes.def: Add pass_lower_switch and pass_lower_switch_O0.
14789 * tree-pass.h (make_pass_lower_switch_O0): New function.
14790 * tree-switch-conversion.c (node_has_low_bound): Remove.
14791 (node_has_high_bound): Likewise.
14792 (node_is_bounded): Likewise.
14793 (class pass_lower_switch): Make it a template type and create
14795 (pass_lower_switch::execute): Add template argument.
14796 (make_pass_lower_switch): New function.
14797 (make_pass_lower_switch_O0): New function.
14798 (do_jump_if_equal): Remove.
14799 (emit_case_nodes): Simplify to just handle all 3 cases and leave
14800 all the hard work to tree optimization passes.
14802 2018-05-18 Martin Liska <mliska@suse.cz>
14804 * dbgcnt.c (limit_low): Renamed from limit.
14805 (limit_high): New variable.
14806 (dbg_cnt_is_enabled): Check for upper limit.
14807 (dbg_cnt): Adjust dumping.
14808 (dbg_cnt_set_limit_by_index): Add new argument for high
14810 (dbg_cnt_set_limit_by_name): Likewise.
14811 (dbg_cnt_process_single_pair): Parse new format.
14812 (dbg_cnt_process_opt): Use strtok.
14813 (dbg_cnt_list_all_counters): Remove 'value' and add
14815 * doc/invoke.texi: Document changes.
14817 2018-05-18 Richard Sandiford <richard.sandiford@linaro.org>
14819 * doc/sourcebuild.texi (scalar_all_fma): Document.
14820 * tree.def (FMA_EXPR): Delete.
14821 * internal-fn.def (FMA, FMS, FNMA, FNMS): New internal functions.
14822 * internal-fn.c (ternary_direct): New macro.
14823 (expand_ternary_optab_fn): Likewise.
14824 (direct_ternary_optab_supported_p): Likewise.
14825 * Makefile.in (build/genmatch.o): Depend on case-fn-macros.h.
14826 * builtins.c (fold_builtin_fma): Delete.
14827 (fold_builtin_3): Don't call it.
14828 * cfgexpand.c (expand_debug_expr): Remove FMA_EXPR handling.
14829 * expr.c (expand_expr_real_2): Likewise.
14830 * fold-const.c (operand_equal_p): Likewise.
14831 (fold_ternary_loc): Likewise.
14832 * gimple-pretty-print.c (dump_ternary_rhs): Likewise.
14833 * gimple.c (DEFTREECODE): Likewise.
14834 * gimplify.c (gimplify_expr): Likewise.
14835 * optabs-tree.c (optab_for_tree_code): Likewise.
14836 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
14837 * tree-eh.c (operation_could_trap_p): Likewise.
14838 (stmt_could_throw_1_p): Likewise.
14839 * tree-inline.c (estimate_operator_cost): Likewise.
14840 * tree-pretty-print.c (dump_generic_node): Likewise.
14841 (op_code_prio): Likewise.
14842 * tree-ssa-loop-im.c (stmt_cost): Likewise.
14843 * tree-ssa-operands.c (get_expr_operands): Likewise.
14844 * tree.c (commutative_ternary_tree_code, add_expr): Likewise.
14845 * fold-const-call.h (fold_fma): Delete.
14846 * fold-const-call.c (fold_const_call_ssss): Handle CFN_FMS,
14847 CFN_FNMA and CFN_FNMS.
14848 (fold_fma): Delete.
14849 * genmatch.c (combined_fn): New enum.
14850 (commutative_ternary_tree_code): Remove FMA_EXPR handling.
14851 (commutative_op): New function.
14852 (commutate): Use it. Handle more than 2 operands.
14853 (dt_operand::gen_gimple_expr): Use commutative_op.
14854 (parser::parse_expr): Allow :c to be used with non-binary
14855 operators if the commutative operand is known.
14856 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Handle
14857 CFN_FMS, CFN_FNMA and CFN_FNMS.
14858 (backprop::process_assign_use): Remove FMA_EXPR handling.
14859 * hsa-gen.c (gen_hsa_insns_for_operation_assignment): Likewise.
14860 (gen_hsa_fma): New function.
14861 (gen_hsa_insn_for_internal_fn_call): Use it for IFN_FMA, IFN_FMS,
14862 IFN_FNMA and IFN_FNMS.
14863 * match.pd: Add folds for IFN_FMS, IFN_FNMA and IFN_FNMS.
14864 * gimple-fold.h (follow_all_ssa_edges): Declare.
14865 * gimple-fold.c (follow_all_ssa_edges): New function.
14866 * tree-ssa-math-opts.c (convert_mult_to_fma_1): Use the
14867 gimple_build interface and use follow_all_ssa_edges to fold the result.
14868 (convert_mult_to_fma): Use direct_internal_fn_suppoerted_p
14869 instead of checking for optabs directly.
14870 * config/i386/i386.c (ix86_add_stmt_cost): Recognize FMAs as calls
14871 rather than FMA_EXPRs.
14872 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Create a
14873 call to IFN_FMA instead of an FMA_EXPR.
14875 2018-05-17 Jim Wilson <jimw@sifive.com>
14877 * expr.c (do_tablejump): When converting index to Pmode, if we have a
14878 sign extended promoted subreg, and the range does not have the sign bit
14879 set, then do a sign extend.
14881 * config/riscv/riscv.c (riscv_extend_comparands): In unsigned QImode
14882 test, check for sign extended subreg and/or constant operands, and
14883 do a sign extend in that case.
14885 2018-05-17 Steve Ellcey <sellcey@cavium.com>
14887 * config/aarch64/thunderx2t99.md (thunderx2t99_ls_both): Delete.
14888 (thunderx2t99_multiple): Delete psuedo-units from used cpus.
14890 (thunderx2t99_alu_shift): Remove alu_shift_reg, alus_shift_reg.
14891 Change logics_shift_reg to logics_shift_imm.
14892 (thunderx2t99_fp_loadpair_basic): Delete.
14893 (thunderx2t99_fp_storepair_basic): Delete.
14894 (thunderx2t99_asimd_int): Add neon_sub and neon_sub_q types.
14895 (thunderx2t99_asimd_polynomial): Delete.
14896 (thunderx2t99_asimd_fp_simple): Add neon_fp_mul_s_scalar_q
14897 and neon_fp_mul_d_scalar_q.
14898 (thunderx2t99_asimd_fp_conv): Add *int_to_fp* types.
14899 (thunderx2t99_asimd_misc): Delete neon_dup and neon_dup_q.
14900 (thunderx2t99_asimd_recip_step): Add missing *sqrt* types.
14901 (thunderx2t99_asimd_lut): Add missing tbl types.
14902 (thunderx2t99_asimd_ext): Delete.
14903 (thunderx2t99_asimd_load1_1_mult): Delete.
14904 (thunderx2t99_asimd_load1_2_mult): Delete.
14905 (thunderx2t99_asimd_load1_ldp): New.
14906 (thunderx2t99_asimd_load1): New.
14907 (thunderx2t99_asimd_load2): Add missing *load2* types.
14908 (thunderx2t99_asimd_load3): New.
14909 (thunderx2t99_asimd_load4): New.
14910 (thunderx2t99_asimd_store1_1_mult): Delete.
14911 (thunderx2t99_asimd_store1_2_mult): Delete.
14912 (thunderx2t99_asimd_store2_mult): Delete.
14913 (thunderx2t99_asimd_store2_onelane): Delete.
14914 (thunderx2t99_asimd_store_stp): New.
14915 (thunderx2t99_asimd_store1): New.
14916 (thunderx2t99_asimd_store2): New.
14917 (thunderx2t99_asimd_store3): New.
14918 (thunderx2t99_asimd_store4): New.
14920 2018-05-17 Jerome Lambourg <lambourg@adacore.com>
14922 * config/arm/arm_cmse.h (cmse_nsfptr_create, cmse_is_nsfptr): Remove
14923 #include <stdint.h>. Replace intptr_t with __INTPTR_TYPE__.
14925 2018-05-17 Pat Haugen <pthaugen@us.ibm.com>
14926 Segher Boessenkool <segher@kernel.crashing.org>
14929 * config/rs6000/rs6000.c (rs6000_output_move_128bit): Check dest
14932 2018-05-17 Richard Biener <rguenther@suse.de>
14934 * tree-ssa-dse.c (dse_classify_store): Fix iterator increment
14935 for pruning loop and prune defs feeding only already visited PHIs.
14937 2018-05-17 Richard Biener <rguenther@suse.de>
14939 * tree-ssa-sccvn.c (vn_reference_lookup_3): Improve memset handling.
14941 2018-05-17 Bin Cheng <bin.cheng@arm.com>
14942 Richard Biener <rguenther@suse.de>
14944 PR tree-optimization/85793
14945 * tree-vect-stmts.c (vectorizable_load): Handle 1 element-wise load
14946 for VMAT_ELEMENTWISE.
14948 2018-05-17 Richard Sandiford <richard.sandiford@linaro.org>
14950 * internal-fn.h (lookup_internal_fn): Declare
14951 * internal-fn.c (lookup_internal_fn): New function.
14952 * gimple.c (gimple_build_call_from_tree): Handle calls to
14953 internal functions.
14954 * gimple-pretty-print.c (dump_gimple_call): Print "." before
14955 internal function names.
14956 * tree-pretty-print.c (dump_generic_node): Likewise.
14957 * tree-ssa-scopedtables.c (expr_hash_elt::print): Likewise.
14959 2018-05-17 Richard Sandiford <richard.sandiford@linaro.org>
14961 * gimple-fold.h (gimple_build): Make the function forms take
14962 combined_fn rather than built_in_function.
14963 (gimple_simplify): Likewise.
14964 * gimple-match-head.c (gimple_simplify): Likewise.
14965 * gimple-fold.c (gimple_build): Likewise.
14966 * tree-vect-loop.c (get_initial_def_for_reduction): Use gimple_build
14967 rather than gimple_build_call_internal.
14968 (get_initial_defs_for_reduction): Likewise.
14969 (vect_create_epilog_for_reduction): Likewise.
14970 (vectorizable_live_operation): Likewise.
14972 2018-05-17 Martin Liska <mliska@suse.cz>
14974 * gimple-ssa-sprintf.c (format_directive): Do not use
14975 space in between 'G_' and '('.
14977 2018-05-17 Jakub Jelinek <jakub@redhat.com>
14980 * config/i386/i386.c (ix86_fold_builtin): Handle masked shifts
14981 even if the mask is not all ones.
14984 * config/i386/i386.c (ix86_fold_builtin): Fold shift builtins by
14986 (ix86_gimple_fold_builtin): Likewise.
14989 * config/i386/i386.c: Include tree-vector-builder.h.
14990 (ix86_vector_shift_count): New function.
14991 (ix86_fold_builtin): Fold shift builtins by scalar count.
14992 (ix86_gimple_fold_builtin): Likewise.
14994 * config/i386/avx512fintrin.h (_mm512_set_epi16, _mm512_set_epi8,
14995 _mm512_setzero): New intrinsics.
14997 2018-05-17 James Greenhalgh <james.greenhalgh@arm.com>
14998 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15000 * config/aarch64/aarch64.c (aarch64_expand_vector_init): Modify
15001 code generation for cases where splatting a value is not useful.
15002 * simplify-rtx.c (simplify_ternary_operation): Simplify
15003 vec_merge across a vec_duplicate and a paradoxical subreg forming
15004 a vector mode to a vec_concat.
15006 2018-05-17 Olga Makhotina <olga.makhotina@intel.com>
15008 * config.gcc: Support "goldmont-plus".
15009 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
15011 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
15012 PROCESSOR_GOLDMONT_PLUS.
15013 * config/i386/i386.c (m_GOLDMONT_PLUS): Define.
15014 (processor_target_table): Add "goldmont-plus".
15015 (PTA_GOLDMONT_PLUS): Define.
15016 (ix86_lea_outperforms): Add TARGET_GOLDMONT_PLUS.
15017 (get_builtin_code_for_version): Handle PROCESSOR_GOLDMONT_PLUS.
15018 (fold_builtin_cpu): Add M_INTEL_GOLDMONT_PLUS.
15019 (fold_builtin_cpu): Add "goldmont-plus".
15020 (ix86_add_stmt_cost): Add TARGET_GOLDMONT_PLUS.
15021 (ix86_option_override_internal): Add "goldmont-plus".
15022 * config/i386/i386.h (processor_costs): Define TARGET_GOLDMONT_PLUS.
15023 (processor_type): Add PROCESSOR_GOLDMONT_PLUS.
15024 * config/i386/x86-tune.def: Add m_GOLDMONT_PLUS.
15025 * doc/invoke.texi: Add goldmont-plus as x86 -march=/-mtune= CPU type.
15027 2018-05-17 Richard Biener <rguenther@suse.de>
15029 PR tree-optimization/85757
15030 * tree-ssa-dse.c (dse_classify_store): Record a PHI def and
15031 remove defs that only feed that PHI from further processing.
15033 2018-05-16 Jim Wilson <jimw@sifive.com>
15035 * config/riscv/riscv.md (<optab>si3_mask, <optab>si3_mask_1): Prepend
15037 (<optab>di3_mask, <optab>di3_mask_1): Likewise.
15039 2018-05-16 Mark Wielaard <mark@klomp.org>
15041 * dwarf2out.c (count_index_strings): New function.
15042 (output_indirect_strings): Call count_index_strings and generate
15043 header for dwarf_version >= 5.
15045 2018-05-16 Mark Wielaard <mark@klomp.org>
15047 * dwarf2out.c (dwarf_FORM): New function.
15048 (set_indirect_string): Use dwarf_FORM.
15049 (reset_indirect_string): Likewise.
15050 (size_of_die): Likewise.
15051 (value_format): Likewise.
15052 (output_die): Likewise.
15053 (add_skeleton_AT_string): Likewise.
15054 (output_macinfo_op): Likewise.
15055 (index_string): Likewise.
15056 (output_index_string_offset): Likewise.
15057 (output_index_string): Likewise.
15058 (count_index_strings): Likewise.
15060 2018-05-16 Carl Love <cel@us.ibm.com>
15062 * config/rs6000/rs6000.md (prefetch): Generate ISA 2.06 instructions
15063 dcbt and dcbtstt with TH=16 if operands[2] is 0 and Power 8 or newer.
15065 2018-05-16 Martin Jambor <mjambor@suse.cz>
15067 * ipa-prop.c (ipa_free_all_edge_args): Remove.
15068 * ipa-prop.h (ipa_free_all_edge_args): Likewise.
15070 2018-05-16 Wilco Dijkstra <wdijkstr@arm.com>
15072 * config/aarch64/aarch64.md (fma<mode>4): Change into expand pattern.
15073 (fnma<mode>4): Likewise.
15074 (fms<mode>4): Likewise.
15075 (fnms<mode>4): Likewise.
15076 (aarch64_fma<mode>4): Rename insn, reorder accumulator operand.
15077 (aarch64_fnma<mode>4): Likewise.
15078 (aarch64_fms<mode>4): Likewise.
15079 (aarch64_fnms<mode>4): Likewise.
15080 (aarch64_fnmadd<mode>4): Likewise.
15082 2018-05-16 Jason Merrill <jason@redhat.com>
15084 * tree.c (warn_deprecated_use): Return bool. Simplify logic.
15086 2018-05-16 Richard Biener <rguenther@suse.de>
15088 * tree-vectorizer.h (struct stmt_info_for_cost): Add where member.
15089 (dump_stmt_cost): Declare.
15090 (add_stmt_cost): Dump cost we add.
15091 (add_stmt_costs): New function.
15092 (vect_model_simple_cost, vect_model_store_cost, vect_model_load_cost):
15093 No longer exported.
15094 (vect_analyze_stmt): Adjust prototype.
15095 (vectorizable_condition): Likewise.
15096 (vectorizable_live_operation): Likewise.
15097 (vectorizable_reduction): Likewise.
15098 (vectorizable_induction): Likewise.
15099 * tree-vect-loop.c (vect_analyze_loop_operations): Create local
15100 cost vector to pass to vectorizable_ and record afterwards.
15101 (vect_model_reduction_cost): Take cost vector argument and adjust.
15102 (vect_model_induction_cost): Likewise.
15103 (vectorizable_reduction): Likewise.
15104 (vectorizable_induction): Likewise.
15105 (vectorizable_live_operation): Likewise.
15106 * tree-vect-slp.c (vect_create_new_slp_node): Initialize
15107 SLP_TREE_NUMBER_OF_VEC_STMTS.
15108 (vect_analyze_slp_cost_1): Remove.
15109 (vect_analyze_slp_cost): Likewise.
15110 (vect_slp_analyze_node_operations): Take visited args and
15111 a target cost vector. Avoid processing already visited stmt sets.
15112 (vect_slp_analyze_operations): Use a local cost vector to gather
15113 costs and register those of non-discarded instances.
15114 (vect_bb_vectorization_profitable_p): Use add_stmt_costs.
15115 (vect_schedule_slp_instance): Remove copying of
15116 SLP_TREE_NUMBER_OF_VEC_STMTS. Instead assert that it is not
15118 * tree-vect-stmts.c (record_stmt_cost): Remove path directly
15119 adding cost. Record cost entry location.
15120 (vect_prologue_cost_for_slp_op): Function to compute cost of
15121 a constant or invariant generated for SLP vect in the prologue,
15122 split out from vect_analyze_slp_cost_1.
15123 (vect_model_simple_cost): Make static. Adjust for SLP costing.
15124 (vect_model_promotion_demotion_cost): Likewise.
15125 (vect_model_store_cost): Likewise, make static.
15126 (vect_model_load_cost): Likewise.
15127 (vectorizable_bswap): Add cost vector arg and adjust.
15128 (vectorizable_call): Likewise.
15129 (vectorizable_simd_clone_call): Likewise.
15130 (vectorizable_conversion): Likewise.
15131 (vectorizable_assignment): Likewise.
15132 (vectorizable_shift): Likewise.
15133 (vectorizable_operation): Likewise.
15134 (vectorizable_store): Likewise.
15135 (vectorizable_load): Likewise.
15136 (vectorizable_condition): Likewise.
15137 (vectorizable_comparison): Likewise.
15138 (can_vectorize_live_stmts): Likewise.
15139 (vect_analyze_stmt): Likewise.
15140 (vect_transform_stmt): Adjust calls to vectorizable_*.
15141 * tree-vectorizer.c: Include gimple-pretty-print.h.
15142 (dump_stmt_cost): New function.
15144 2018-05-16 Richard Biener <rguenther@suse.de>
15146 * params.def (PARAM_DSE_MAX_ALIAS_QUERIES_PER_STORE): New param.
15147 * doc/invoke.texi (dse-max-alias-queries-per-store): Document.
15148 * tree-ssa-dse.c: Include tree-ssa-loop.h.
15149 (check_name): New callback.
15150 (dse_classify_store): Track cycles via a visited bitmap of PHI
15151 defs and simplify handling of in-loop and across loop dead stores
15152 and properly fail for loop-variant refs. Handle byte-tracking with
15153 multiple defs. Use PARAM_DSE_MAX_ALIAS_QUERIES_PER_STORE for
15156 2018-05-16 Richard Sandiford <richard.sandiford@linaro.org>
15158 * tree-vectorizer.h (vect_get_vector_types_for_stmt): Declare.
15159 (vect_get_mask_type_for_stmt): Likewise.
15160 * tree-vect-slp.c (vect_two_operations_perm_ok_p): New function,
15162 (vect_build_slp_tree_1): ...here. Use vect_get_vector_types_for_stmt
15163 to determine the statement's vector type and the vector type that
15164 should be used for calculating nunits. Deal with cases in which
15165 the type has to be deferred.
15166 (vect_slp_analyze_node_operations): Use vect_get_vector_types_for_stmt
15167 and vect_get_mask_type_for_stmt to calculate STMT_VINFO_VECTYPE.
15168 * tree-vect-loop.c (vect_determine_vf_for_stmt_1)
15169 (vect_determine_vf_for_stmt): New functions, split out from...
15170 (vect_determine_vectorization_factor): ...here.
15171 * tree-vect-stmts.c (vect_get_vector_types_for_stmt)
15172 (vect_get_mask_type_for_stmt): New functions, split out from
15173 vect_determine_vectorization_factor.
15175 2018-05-16 Richard Biener <rguenther@suse.de>
15177 * tree-cfg.c (verify_gimple_assign_ternary): Properly
15178 verify the [VEC_]COND_EXPR embedded comparison.
15180 2018-05-15 Martin Sebor <msebor@redhat.com>
15182 PR tree-optimization/85753
15183 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Handle
15184 RECORD_TYPE in addition to ARRAY_TYPE.
15186 2018-05-15 Martin Sebor <msebor@redhat.com>
15188 PR middle-end/85643
15189 * calls.c (get_attr_nonstring_decl): Handle MEM_REF.
15191 2018-05-15 Richard Biener <rguenther@suse.de>
15193 * tree-ssa-dse.c (dse_classify_store): Remove use_stmt parameter,
15194 add by_clobber_p one. Change algorithm to collect all defs
15195 representing uses we need to walk and try reducing them to
15196 a single one before failing.
15197 (dse_dom_walker::dse_optimize_stmt): Adjust.
15199 2018-05-13 Mark Wielaard <mark@klomp.org>
15201 * dwarf2out.c (dwarf_OP): Handle DW_OP_addrx and DW_OP_constx.
15202 (size_of_loc_descr): Likewise.
15203 (output_loc_operands): Likewise.
15204 (output_loc_operands_raw): Likewise.
15205 (dw_addr_op): Use dwarf_OP () for DW_OP_constx and DW_OP_addrx.
15206 (resolve_addr_in_expr): Handle DW_OP_addrx and DW_OP_constx.
15207 (hash_loc_operands): Likewise.
15208 (compare_loc_operands): Likewise.
15210 2018-05-14 Mark Wielaard <mark@klomp.org>
15212 * dwarf2out.c (count_index_addrs): New function.
15213 (dwarf2out_finish): Use count_index_addrs to calculate addrs_length.
15215 2018-05-15 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
15217 PR tree-optimization/83648
15218 * ipa-pure-const.c (malloc_candidate_p): Allow function with NULL
15219 return value as malloc candidate.
15221 2018-05-15 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
15224 * ipa-pure-const.c (warn_function_malloc): Pass value of known_finite
15225 param as true in call to suggest_attribute.
15227 2018-05-14 Segher Boessenkool <segher@kernel.crashing.org>
15229 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete @itemx for
15230 -mreadonly-in-sdata.
15232 2018-05-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15234 * config/aarch64/aarch64-simd.md (*aarch64_crypto_aese_fused):
15236 (aarch64_crypto_aesd_fused): Likewise.
15238 2018-05-14 Wilco Dijkstra <wdijkstr@arm.com>
15240 * config/aarch64/aarch64.md (mov<mode>): Remove '*' in alternatives.
15241 (movsi_aarch64): Likewise.
15242 (load_pairsi): Likewise.
15243 (load_pairdi): Likewise.
15244 (store_pairsi): Likewise.
15245 (store_pairdi): Likewise.
15246 (load_pairsf): Likewise.
15247 (load_pairdf): Likewise.
15248 (store_pairsf): Likewise.
15249 (store_pairdf): Likewise.
15250 (zero_extend): Likewise.
15251 (trunc): Swap alternatives.
15252 (fcvt_target): Add '?' to prefer w over r.
15254 2018-05-14 Jakub Jelinek <jakub@redhat.com>
15257 * config/i386/i386.md: Disallow non-commutative arithmetics in
15258 last twpeephole for mem {+,-,&,|,^}= x; mem != 0 after cmpelim
15259 optimization. Use COMMUTATIVE_ARITH_P test rather than != MINUS
15260 in the peephole2 before it.
15262 2018-05-14 Sebastian Peryt <sebastian.peryt@intel.com>
15264 * common/config/i386/i386-common.c (OPTION_MASK_ISA_CLDEMOTE_SET,
15265 OPTION_MASK_ISA_CLDEMOTE_UNSET): New defines.
15266 (ix86_handle_option): Handle -mcldemote.
15267 * config.gcc: New header.
15268 * config/i386/cldemoteintrin.h: New file.
15269 * config/i386/cpuid.h (bit_CLDEMOTE): New bit.
15270 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
15272 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
15273 OPTION_MASK_ISA_CLDEMOTE.
15274 * config/i386/i386.c (ix86_target_string): Add -mcldemote.
15275 (ix86_valid_target_attribute_inner_p): Ditto.
15276 (enum ix86_builtins): Add IX86_BUILTIN_CLDEMOTE.
15277 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_cldemote.
15278 (ix86_expand_builtin): Expand IX86_BUILTIN_CLDEMOTE.
15279 * config/i386/i386.h (TARGET_CLDEMOTE, TARGET_CLDEMOTE_P): New.
15280 * config/i386/i386.md (UNSPECV_CLDEMOTE): New.
15282 * config/i386/i386.opt: Add -mcldemote.
15283 * config/i386/x86intrin.h: New header.
15284 * doc/invoke.texi: Add -mcldemote.
15286 2018-05-14 Richard Biener <rguenther@suse.de>
15288 * doc/match-and-simplify.texi: Adjust :s documentation.
15290 2018-05-14 Alexander Monakov <amonakov@ispras.ru>
15292 * sort.cc (REORDER_23): Pass the type for the temporaries instead of
15293 intended memcpy size.
15294 (REORDER_45): Likewise.
15296 2018-05-13 Alexander Monakov <amonakov@ispras.ru>
15298 * sort.cc: New file.
15299 * system.h [!CHECKING_P] (qsort): Redirect to gcc_qsort.
15300 * vec.c (qsort_chk): Use gcc_qsort.
15301 * Makefile.in (OBJS-libcommon): Add sort.o.
15302 (build/sort.o): New target. Use it...
15303 (BUILD_RTL): ... here, and...
15304 (build/gencfn-macros): ... here, and...
15305 (build/genmatch): ... here.
15307 2018-05-13 Kito Cheng <kito.cheng@gmail.com>
15308 Chung-Ju Wu <jasonwucj@gmail.com>
15310 * config.gcc (nds32*-*-*): Check that n15 is valid to --with-cpu.
15311 * config/nds32/nds32-graywolf.md: New file.
15312 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_GRAYWOLF.
15313 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n15
15315 * config/nds32/nds32-protos.h: More declarations for n15 pipeline.
15316 * config/nds32/nds32-utils.c: More implementations for n15 pipeline.
15317 * config/nds32/nds32.md (pipeline_model): Add graywolf.
15318 * config/nds32/nds32.opt (mcpu): Support n15 pipeline cpus.
15319 * config/nds32/pipelines.md: Include n15 settings.
15321 2018-05-13 Kito Cheng <kito.cheng@gmail.com>
15322 Chung-Ju Wu <jasonwucj@gmail.com>
15324 * config.gcc (nds32*-*-*): Check that n12/n13 are valid to --with-cpu.
15325 * config/nds32/nds32-n13.md: New file.
15326 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N12 and CPU_N13.
15327 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n12/n13
15329 * config/nds32/nds32-protos.h: More declarations for n12/n13 pipeline.
15330 * config/nds32/nds32.md (pipeline_model): Add n13.
15331 * config/nds32/nds32.opt (mcpu): Support n12/n13 pipeline cpus.
15332 * config/nds32/pipelines.md: Include n13 settings.
15334 2018-05-13 Kito Cheng <kito.cheng@gmail.com>
15335 Chung-Ju Wu <jasonwucj@gmail.com>
15337 * config.gcc (nds32*-*-*): Check that n10/d10 are valid to --with-cpu.
15338 * config/nds32/nds32-n10.md: New file.
15339 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N10.
15340 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n10
15342 * config/nds32/nds32-protos.h: More declarations for n10 pipeline.
15343 * config/nds32/nds32-utils.c: More implementations for n10 pipeline.
15344 * config/nds32/nds32.md (pipeline_model): Add n10.
15345 * config/nds32/nds32.opt (mcpu): Support n10 pipeline cpus.
15346 * config/nds32/pipelines.md: Include n10 settings.
15348 2018-05-13 Monk Chiang <sh.chiang04@gmail.com>
15349 Kito Cheng <kito.cheng@gmail.com>
15350 Chung-Ju Wu <jasonwucj@gmail.com>
15352 * config.gcc (nds32be-*-*): Handle --with-ext-dsp.
15353 * config/nds32/constants.md (unspec_element, unspec_volatile_element):
15354 Add enum values for DSP extension instructions.
15355 * config/nds32/constraints.md (Iu06, IU06, CVp5, CVs5, CVs2, CVhi):
15357 * config/nds32/iterators.md (shifts, shiftrt, sat_plus, all_plus,
15358 sat_minus, all_minus, plus_minus, extend, sumax, sumin, sumin_max):
15359 New code iterators.
15360 (su, zs, uk, opcode, add_rsub, add_sub): New code attributes.
15361 * config/nds32/nds32-dspext.md: New file for DSP implementation.
15362 * config/nds32/nds32-intrinsic.c: Implementation of DSP extension.
15363 * config/nds32/nds32-intrinsic.md: Likewise.
15364 * config/nds32/nds32_intrinsic.h: Likewise.
15365 * config/nds32/nds32-md-auxiliary.c: Likewise.
15366 * config/nds32/nds32-memory-manipulation.c: Consider DSP extension.
15367 * config/nds32/nds32-predicates.c (const_vector_to_hwint): New.
15368 (nds32_valid_CVp5_p, nds32_valid_CVs5_p): New.
15369 (nds32_valid_CVs2_p, nds32_valid_CVhi_p): New.
15370 * config/nds32/nds32-protos.h: New declarations for DSP extension.
15371 * config/nds32/nds32-utils.c (extract_mac_non_acc_rtx): New case
15372 TYPE_DMAC in switch statement.
15373 * config/nds32/nds32.c: New checking and implementation for DSP
15374 extension instructions.
15375 * config/nds32/nds32.h: Likewise.
15376 * config/nds32/nds32.md: Likewise.
15377 * config/nds32/nds32.opt (mhw-abs, mext-dsp): New options.
15378 * config/nds32/predicates.md: Implement new predicates for DSP
15381 2018-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
15383 * config/rs6000/rs6000.md (mov<mode>_softfloat, FMOVE32):
15384 Reformat alternatives and attributes so it is easier to identify
15385 which constraints/attributes go with which instruction.
15386 (mov<mode>_hardfloat32, FMOVE64): Likewise.
15387 (mov<mode>_softfloat32, FMOVE64): Likewise.
15388 (mov<mode>_hardfloat64, FMOVE64): Likewise.
15389 (mov<mode>_softfloat64, FMOVE64): Likewise.
15391 2018-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
15393 * doc/extend.texi (PowerPC Built-in Functions): Rename this
15395 (Basic PowerPC Built-in Functions): The new name of the
15396 subsection previously known as "PowerPC Built-in Functions".
15397 (Basic PowerPC Built-in Functions Available on all Configurations):
15399 (Basic PowerPC Built-in Functions Available on ISA 2.05): Likewise.
15400 (Basic PowerPC Built-in Functions Available on ISA 2.06): Likewise.
15401 (Basic PowerPC Built-in Functions Available on ISA 2.07): Likewise.
15402 (Basic PowerPC Built-in Functions Available on ISA 3.0): Likewise.
15404 2018-05-11 Martin Jambor <mjambor@suse.cz>
15407 * ipa-cp.c (intersect_with_plats): Check that the lattice contains
15410 2018-05-11 Richard Earnshaw <rearnsha@arm.com>
15413 * config/arm/arm-cpus.in (fgroup ARMv6m): Add be8 feature.
15415 2018-05-11 Sebastian Peryt <sebastian.peryt@intel.com>
15417 * common/config/i386/i386-common.c (OPTION_MASK_ISA_WAITPKG_SET,
15418 OPTION_MASK_ISA_WAITPKG_UNSET): New defines.
15419 (ix86_handle_option): Handle -mwaitpkg.
15420 * config.gcc: New header.
15421 * config/i386/cpuid.h (bit_WAITPKG): New bit.
15422 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mwaitpkg.
15423 * config/i386/i386-builtin-types.def ((UINT8, UNSIGNED, UINT64)): New
15425 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
15426 OPTION_MASK_ISA_WAITPKG.
15427 * config/i386/i386.c (ix86_target_string): Add -mwaitpkg.
15428 (ix86_option_override_internal): Add PTA_WAITPKG.
15429 (ix86_valid_target_attribute_inner_p): Add -mwaitpkg.
15430 (enum ix86_builtins): Add IX86_BUILTIN_UMONITOR, IX86_BUILTIN_UMWAIT,
15431 IX86_BUILTIN_TPAUSE.
15432 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_umonitor,
15433 __builtin_ia32_umwait and __builtin_ia32_tpause.
15434 (ix86_expand_builtin): Expand IX86_BUILTIN_UMONITOR,
15435 IX86_BUILTIN_UMWAIT, IX86_BUILTIN_TPAUSE.
15436 * config/i386/i386.h (TARGET_WAITPKG, TARGET_WAITPKG_P): New.
15437 * config/i386/i386.md (UNSPECV_UMWAIT, UNSPECV_UMONITOR,
15438 UNSPECV_TPAUSE): New.
15439 (umwait, umwait_rex64, umonitor_<mode>, tpause, tpause_rex64): New.
15440 * config/i386/i386.opt: Add -mwaitpkg.
15441 * config/i386/waitpkgintrin.h: New file.
15442 * config/i386/x86intrin.h: New header.
15443 * doc/invoke.texi: Add -mwaitpkg.
15445 2018-05-11 Richard Earnshaw <rearnsha@arm.com>
15448 * config/arm/arm-cpus.in: Add comment that ARMv6-m and ARMv6S-m are now
15450 (cortex-m0): Use armv6s-m isa.
15451 (cortex-m0plus): Likewise.
15452 (cortex-m1): Likewise.
15453 (cortex-m0.small-multiply): Likewise.
15454 (cortex-m0plus.small-multiply): Likewise.
15455 (cortex-m1.small-multiply): Likewise.
15457 2018-05-11 Allan Sandfeld Jensen <allan.jensen@qt.io>
15458 Jakub Jelinek <jakub@redhat.com>
15460 PR tree-optimization/85692
15461 * tree-ssa-forwprop.c (simplify_vector_constructor): Try two
15462 source permute as well.
15464 2018-05-11 Martin Liska <mliska@suse.cz>
15467 * doc/extend.texi: Document LLVM style format for no_sanitize
15470 2018-05-10 Michael Meissner <meissner@linux.vnet.ibm.com>
15472 * config/rs6000/rs6000.c (mode_supports_dq_form): Rename
15473 mode_supports_vsx_dform_quad to mode_supports_dq_form.
15474 (mode_supports_vsx_dform_quad): Likewise.
15475 (mode_supports_vmx_dform): Move these functions to be next to the
15476 other mode_supports functions.
15477 (mode_supports_dq_form): Likewise.
15478 (quad_address_p): Change calls of mode_supports_vsx_dform_quad to
15479 mode_supports_dq_form.
15480 (reg_offset_addressing_ok_p): Likewise.
15481 (offsettable_ok_by_alignment): Likewise.
15482 (rs6000_legitimate_offset_address_p): Likewise.
15483 (legitimate_lo_sum_address_p): Likewise.
15484 (rs6000_legitimize_address): Likewise.
15485 (rs6000_legitimize_reload_address): Likewise.
15486 (rs6000_secondary_reload_inner): Likewise.
15487 (rs6000_preferred_reload_class): Likewise.
15488 (rs6000_output_move_128bit): Likewise.
15490 2018-05-10 Uros Bizjak <ubizjak@gmail.com>
15492 * config/i386/i386.c (ix86_expand_builtin) <case IX86_BUILTIN_RDPID>:
15493 Generate SImode target register for null target.
15494 <case IX86_BUILTIN_XGETBV>: Ditto.
15495 <case IX86_BUILTIN_XSETBV>: Optimize LSHIFTRT generation.
15496 * config/i386/xsaveintrin.h (_xgetbv): Add missing return.
15498 2018-05-10 Carl Love <cel@us.ibm.com>
15500 * config/rs6000/rs6000.md (prefetch): Generate ISA 2.06 instructions
15501 dcbtt and dcbtstt if operands[2] is 0.
15503 2018-05-10 Uros Bizjak <ubizjak@gmail.com>
15506 * config/i386/sse.md (usadv64qi): New expander.
15508 2018-05-10 Segher Boessenkool <segher@kernel.crashing.org>
15510 * config/rs6000/altivec.md (altivec_vmrghb, altivec_vmrghh,
15511 altivec_vmrghw, altivec_vmrglb, altivec_vmrglh, altivec_vmrglw): Remove
15512 -maltivec=be support.
15513 (vec_widen_umult_even_v16qi, vec_widen_smult_even_v16qi,
15514 vec_widen_umult_even_v8hi, vec_widen_smult_even_v8hi,
15515 vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
15516 vec_widen_umult_odd_v16qi, vec_widen_smult_odd_v16qi,
15517 vec_widen_umult_odd_v8hi, vec_widen_smult_odd_v8hi,
15518 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si, altivec_vpkpx,
15519 altivec_vpks<VI_char>ss, altivec_vpks<VI_char>us,
15520 altivec_vpku<VI_char>us, altivec_vpku<VI_char>um, altivec_vsum2sws,
15521 altivec_vsumsws): Adjust.
15522 (altivec_vspltb *altivec_vspltb_internal, altivec_vsplth,
15523 *altivec_vsplth_internal, altivec_vspltw, *altivec_vspltw_internal,
15524 altivec_vspltsf, *altivec_vspltsf_internal): Remove -maltivec=be
15526 (altivec_vperm_<mode>, altivec_vperm_<mode>_uns,
15527 altivec_vupkhs<VU_char>, altivec_vupkls<VU_char>, altivec_vupkhpx,
15528 altivec_vupklpx, altivec_lvsl, altivec_lvsr): Adjust.
15529 (altivec_lve<VI_char>x): Delete expand.
15530 (*altivec_lve<VI_char>x_internal): Rename to...
15531 (altivec_lve<VI_char>x): ... this.
15532 (altivec_lvxl_<mode>): Delete expand.
15533 (*altivec_lvxl_<mode>_internal): Rename to ...
15534 (altivec_lvxl_<mode>): ... this.
15535 (altivec_stvxl_<mode>): Delete expand.
15536 (*altivec_stvxl_<mode>_internal): Rename to ...
15537 (altivec_stvxl_<mode>): ... this.
15538 (altivec_stve<VI_char>x): Delete expand.
15539 (*altivec_stve<VI_char>x_internal): Rename to ...
15540 (altivec_stve<VI_char>x): ... this.
15541 (doublee<mode>2, unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2,
15542 doubleh<mode>2, unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2,
15543 reduc_plus_scal_<mode>): Adjust.
15544 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Adjust
15546 (rs6000_cpu_cpp_builtins): Adjust.
15547 (altivec_resolve_overloaded_builtin): Remove -maltivec=be support.
15548 * config/rs6000/rs6000-protos.h (altivec_expand_lvx_be,
15549 altivec_expand_stvx_be, altivec_expand_stvex_be): Delete.
15550 * config/rs6000/rs6000.c (rs6000_option_override_internal): Remove
15551 -maltivec=be support.
15552 (rs6000_split_vec_extract_var): Adjust.
15553 (rs6000_split_v4si_init): Adjust.
15554 (swap_selector_for_mode): Delete.
15555 (altivec_expand_lvx_be, altivec_expand_stvx_be,
15556 altivec_expand_stvex_be): Delete.
15557 (altivec_expand_lv_builtin, altivec_expand_stv_builtin): Remove
15558 -maltivec=be support.
15559 (rs6000_gimple_fold_builtin): Ditto.
15560 (rs6000_generate_float2_double_code, rs6000_generate_float2_code):
15562 * config/rs6000/rs6000.h (VECTOR_ELT_ORDER_BIG): Delete.
15563 (TARGET_DIRECT_MOVE_64BIT): Adjust.
15564 * config/rs6000/rs6000.md (split for extendsidi2 for vectors): Adjust.
15565 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Delete.
15566 * config/rs6000/vsx.md (floate<mode>, unsfloatev2di, floato<mode>,
15567 unsfloatov2di, vsignedo_v2df, vsignede_v2df, vunsignedo_v2df,
15568 vunsignede_v2df, vsx_extract_<mode>_p9, *vsx_extract_si,
15569 *vsx_extract_<mode>_p8, *vsx_extract_si_<uns>float_df,
15570 *vsx_extract_si_<uns>float_<mode>, vsx_set_<mode>_p9, vsx_set_v4sf_p9,
15571 *vsx_insert_extract_v4sf_p9, *vsx_insert_extract_v4sf_p9_2, and an
15572 anonymous split): Adjust.
15573 (vsx_mergel_<mode>, vsx_mergeh_<mode>): Remove -maltivec=be support.
15574 (vsx_xxspltd_<mode>, extract4b, insert4b): Adjust.
15576 2018-05-10 Eric Botcazou <ebotcazou@adacore.com>
15578 * configure.ac (gcc_gxx_include_dir_add_sysroot): Set it to 1 only
15579 when --with-gxx-include-dir is also specified.
15580 * configure: Regenerate.
15582 2018-05-09 Jim Wilson <jimw@sifive.com>
15585 * config.gcc (riscv*-*-*): Handle --with-multilib-list.
15586 * config/riscv/t-withmultilib: New.
15587 * config/riscv/withmultilib.h: New.
15588 * doc/install.texi: Document RISC-V --with-multilib-list support.
15590 2018-05-09 Richard Biener <rguenther@suse.de>
15592 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Fill a cost
15594 (vect_bb_vectorization_profitable_p): Adjust. Compute
15595 actual scalar cost using the cost vector and the add_stmt_cost
15598 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
15600 PR rtl-optimization/85645
15601 * config/rs6000/rs6000.c (rs6000_emit_prologue_components): Put a SET
15602 in the REG_CFA_REGISTER note for LR, don't leave it empty.
15604 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
15606 PR rtl-optimization/85645
15607 * shrink-wrap.c (spread_components): Return a boolean saying if
15608 anything was changed.
15609 (try_shrink_wrapping_separate): Iterate spread_components until
15610 nothing changes anymore.
15612 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
15614 PR rtl-optimization/85645
15615 * regrename.c (build_def_use): Also kill the chains that include the
15616 destination of a REG_CFA_REGISTER note.
15618 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
15620 PR rtl-optimization/85645
15621 * regcprop.c (copyprop_hardreg_forward_1): Don't propagate into an
15622 insn that has a REG_CFA_REGISTER note.
15624 2018-05-09 Richard Sandiford <richard.sandiford@linaro.org>
15626 * cfgexpand.c (expand_clobber): New function.
15627 (expand_gimple_stmt_1): Use it.
15628 * tree-vect-stmts.c (vect_clobber_variable): New function,
15630 (vectorizable_simd_clone_call): ...here.
15631 (vectorizable_store): Emit a clobber either side of an
15632 IFN_STORE_LANES sequence.
15633 (vectorizable_load): Emit a clobber after an IFN_LOAD_LANES sequence.
15635 2018-05-09 Tom de Vries <tom@codesourcery.com>
15638 * config/nvptx/nvptx.md (define_insn "trap", define_insn "trap_if_true")
15639 (define_insn "trap_if_false"): Add exit after trap.
15641 2018-05-09 Eric Botcazou <ebotcazou@adacore.com>
15643 PR rtl-optimization/85638
15644 * bb-reorder.c: Include common/common-target.h.
15645 (create_forwarder_block): New function extracted from...
15646 (fix_up_crossing_landing_pad): ...here. Rename into...
15647 (dw2_fix_up_crossing_landing_pad): ...this.
15648 (sjlj_fix_up_crossing_landing_pad): New function.
15649 (find_rarely_executed_basic_blocks_and_crossing_edges): In SJLJ mode,
15650 call sjlj_fix_up_crossing_landing_pad if there are incoming EH edges
15651 from both partitions and exit the loop after one iteration.
15653 2018-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
15656 * doc/extend.texi (PowerPC Built-in Functions): Rename this
15658 (Basic PowerPC Built-in Functions): The new name of the
15659 subsection previously known as "PowerPC Built-in Functions".
15660 (Basic PowerPC Built-in Functions Available on all Configurations):
15662 (Basic PowerPC Built-in Functions Available on ISA 2.05): New
15664 (Basic PowerPC Built-in Functions Available on ISA 2.06): New
15666 (Basic PowerPC Built-in Functions Available on ISA 2.07): New
15668 (Basic PowerPC Built-in Functions Available on ISA 3.0): New
15671 2018-05-08 Jim Wilson <jimw@sifive.com>
15673 * config/riscv/linux.h (MUSL_ABI_SUFFIX): Delete unnecessary backslash.
15674 (LD_EMUL_SUFFIX): New.
15675 (LINK_SPEC): Use it.
15677 2018-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
15679 * doc/extend.texi (PowerPC Built-in Functions): Rename this
15681 (Basic PowerPC Built-in Functions): The new name of the
15682 subsection previously known as "PowerPC Built-in Functions".
15683 (Basic PowerPC Built-in Functions Available on all Configurations):
15685 (Basic PowerPC Built-in Functions Available on ISA 2.05): New
15687 (Basic PowerPC Built-in Functions Available on ISA 2.06): New
15689 (Basic PowerPC Built-in Functions Available on ISA 2.07): New
15691 (Basic PowerPC Built-in Functions Available on ISA 3.0): New
15694 2018-05-08 Jakub Jelinek <jakub@redhat.com>
15697 * config/i386/i386.md: Add peepholes for mem {+,-,&,|,^}= x; mem != 0
15698 after cmpelim optimization.
15700 2018-05-08 Olga Makhotina <olga.makhotina@intel.com>
15702 * config.gcc: Support "goldmont".
15703 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "goldmont".
15704 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
15705 PROCESSOR_GOLDMONT.
15706 * config/i386/i386.c (m_GOLDMONT): Define.
15707 (processor_target_table): Add "goldmont".
15708 (PTA_GOLDMONT): Define.
15709 (ix86_lea_outperforms): Add TARGET_GOLDMONT.
15710 (get_builtin_code_for_version): Handle PROCESSOR_GOLDMONT.
15711 (fold_builtin_cpu): Add M_INTEL_GOLDMONT.
15712 (fold_builtin_cpu): Add "goldmont".
15713 (ix86_add_stmt_cost): Add TARGET_GOLDMONT.
15714 (ix86_option_override_internal): Add "goldmont".
15715 * config/i386/i386.h (processor_costs): Define TARGET_GOLDMONT.
15716 (processor_type): Add PROCESSOR_GOLDMONT.
15717 * config/i386/i386.md: Add CPU "glm".
15718 * config/i386/glm.md: New file.
15719 * config/i386/x86-tune.def: Add m_GOLDMONT.
15720 * doc/invoke.texi: Add goldmont as x86 -march=/-mtune= CPU type.
15722 2018-05-08 Jakub Jelinek <jakub@redhat.com>
15725 * config/i386/i386.c (ix86_expand_sse2_abs): Handle E_V2DImode and
15727 * config/i386/sse.md (abs<mode>2): Use VI_AVX2 iterator instead of
15728 VI1248_AVX512VL_AVX512BW. Handle V2DImode and V4DImode if not
15729 TARGET_AVX512VL using ix86_expand_sse2_abs. Formatting fixes.
15732 * config/i386/i386.c (ix86_fold_builtin): Handle
15733 IX86_BUILTIN_{,P}MOVMSK{PS,PD,B}{,128,256}.
15736 * config/i386/sse.md (ssequaterinsnmode): New mode attribute.
15737 (*<extract_type>_vinsert<shuffletype><extract_suf>_0): New pattern.
15739 2018-05-08 Richard Earnshaw <rearnsha@arm.com>
15742 * config/arm/parsecpu.awk (check_cpu): Fix operator precedence.
15743 (check_arch): Likewise.
15744 (check_fpu): Return the result rather than printing it.
15745 (end arch): Fix operator precedence.
15746 (end cpu): Likewise.
15747 (END): Print the result from check_fpu.
15749 2018-05-08 Richard Sandiford <richard.sandiford@linaro.org>
15750 Alan Hayward <alan.hayward@arm.com>
15751 David Sherwood <david.sherwood@arm.com>
15753 * config/aarch64/aarch64-sve.md (*pred_cmp<cmp_op><mode>_combine)
15754 (*pred_cmp<cmp_op><mode>, *fcm<cmp_op><mode>_and_combine)
15755 (*fcmuo<mode>_and_combine, *fcm<cmp_op><mode>_and)
15756 (*fcmuo<mode>_and): New patterns.
15758 2018-05-08 Richard Sandiford <richard.sandiford@linaro.org>
15760 * config/aarch64/iterators.md (UNSPEC_COND_LO, UNSPEC_COND_LS)
15761 (UNSPEC_COND_HI, UNSPEC_COND_HS, UNSPEC_COND_UO): Delete.
15762 (SVE_INT_CMP, SVE_FP_CMP): New code iterators.
15763 (cmp_op, sve_imm_con): New code attributes.
15764 (SVE_COND_INT_CMP, imm_con): Delete.
15765 (cmp_op): Remove above unspecs from int attribute.
15766 * config/aarch64/aarch64-sve.md (*vec_cmp<cmp_op>_<mode>): Rename
15768 (*cmp<cmp_op><mode>): ...this. Use UNSPEC_MERGE_PTRUE instead of
15769 comparison-specific unspecs.
15770 (*vec_cmp<cmp_op>_<mode>_ptest): Rename to...
15771 (*cmp<cmp_op><mode>_ptest): ...this and adjust likewise.
15772 (*vec_cmp<cmp_op>_<mode>_cc): Rename to...
15773 (*cmp<cmp_op><mode>_cc): ...this and adjust likewise.
15774 (*vec_fcm<cmp_op><mode>): Rename to...
15775 (*fcm<cmp_op><mode>): ...this and adjust likewise.
15776 (*vec_fcmuo<mode>): Rename to...
15777 (*fcmuo<mode>): ...this and adjust likewise.
15778 (*pred_fcm<cmp_op><mode>): New pattern.
15779 * config/aarch64/aarch64.c (aarch64_emit_unop, aarch64_emit_binop)
15780 (aarch64_emit_sve_ptrue_op, aarch64_emit_sve_ptrue_op_cc): New
15782 (aarch64_unspec_cond_code): Remove handling of LTU, GTU, LEU, GEU
15784 (aarch64_gen_unspec_cond, aarch64_emit_unspec_cond): Delete.
15785 (aarch64_emit_sve_predicated_cond): New function.
15786 (aarch64_expand_sve_vec_cmp_int): Use aarch64_emit_sve_ptrue_op_cc.
15787 (aarch64_emit_unspec_cond_or): Replace with...
15788 (aarch64_emit_sve_or_conds): ...this new function. Use
15789 aarch64_emit_sve_ptrue_op for the individual comparisons and
15790 aarch64_emit_binop to OR them together.
15791 (aarch64_emit_inverted_unspec_cond): Replace with...
15792 (aarch64_emit_sve_inverted_cond): ...this new function. Use
15793 aarch64_emit_sve_ptrue_op for the comparison and
15794 aarch64_emit_unop to invert the result.
15795 (aarch64_expand_sve_vec_cmp_float): Update after the above
15796 changes. Use aarch64_emit_sve_ptrue_op for native comparisons.
15798 2018-05-07 Nathan Sidwell <nathan@acm.org>
15800 * doc/invoke.texi (C++ Dialect Options): Remove -ffor-scope.
15801 * doc/extend.texi (Deprecated Features): Remove -fno-for-scope
15802 (Backwards Compatibility): Likewise.
15804 2018-05-07 Luis Machado <luis.machado@linaro.org>
15808 2018-05-07 Luis Machado <luis.machado@linaro.org>
15810 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
15811 <prefetch_dynamic_strides>: New const bool field.
15812 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
15813 prefetch_dynamic_strides.
15814 (exynosm1_prefetch_tune): Likewise.
15815 (thunderxt88_prefetch_tune): Likewise.
15816 (thunderx_prefetch_tune): Likewise.
15817 (thunderx2t99_prefetch_tune): Likewise.
15818 (qdf24xx_prefetch_tune): Likewise. Set prefetch_dynamic_strides
15820 (aarch64_override_options_internal): Update to set
15821 PARAM_PREFETCH_DYNAMIC_STRIDES.
15822 * doc/invoke.texi (prefetch-dynamic-strides): Document new option.
15823 * params.def (PARAM_PREFETCH_DYNAMIC_STRIDES): New.
15824 * params.h (PARAM_PREFETCH_DYNAMIC_STRIDES): Define.
15825 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Account for
15826 prefetch-dynamic-strides setting.
15828 2018-05-07 Luis Machado <luis.machado@linaro.org>
15830 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
15831 <minimum_stride>: New const int field.
15832 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
15833 minimum_stride field.
15834 (exynosm1_prefetch_tune): Likewise.
15835 (thunderxt88_prefetch_tune): Likewise.
15836 (thunderx_prefetch_tune): Likewise.
15837 (thunderx2t99_prefetch_tune): Likewise.
15838 (qdf24xx_prefetch_tune): Likewise. Set minimum_stride to 2048.
15839 (aarch64_override_options_internal): Update to set
15840 PARAM_PREFETCH_MINIMUM_STRIDE.
15841 * doc/invoke.texi (prefetch-minimum-stride): Document new option.
15842 * params.def (PARAM_PREFETCH_MINIMUM_STRIDE): New.
15843 * params.h (PARAM_PREFETCH_MINIMUM_STRIDE): Define.
15844 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Return false if
15845 stride is constant and is below the minimum stride threshold.
15847 2018-05-07 Luis Machado <luis.machado@linaro.org>
15849 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune) <l2_cache_size>: Set
15852 2018-05-07 Luis Machado <luis.machado@linaro.org>
15854 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
15855 <prefetch_dynamic_strides>: New const bool field.
15856 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
15857 prefetch_dynamic_strides.
15858 (exynosm1_prefetch_tune): Likewise.
15859 (thunderxt88_prefetch_tune): Likewise.
15860 (thunderx_prefetch_tune): Likewise.
15861 (thunderx2t99_prefetch_tune): Likewise.
15862 (qdf24xx_prefetch_tune): Likewise. Set prefetch_dynamic_strides
15864 (aarch64_override_options_internal): Update to set
15865 PARAM_PREFETCH_DYNAMIC_STRIDES.
15866 * doc/invoke.texi (prefetch-dynamic-strides): Document new option.
15867 * params.def (PARAM_PREFETCH_DYNAMIC_STRIDES): New.
15868 * params.h (PARAM_PREFETCH_DYNAMIC_STRIDES): Define.
15869 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Account for
15870 prefetch-dynamic-strides setting.
15872 2018-05-07 Luis Machado <luis.machado@linaro.org>
15874 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
15875 <minimum_stride>: New const int field.
15876 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
15877 minimum_stride field.
15878 (exynosm1_prefetch_tune): Likewise.
15879 (thunderxt88_prefetch_tune): Likewise.
15880 (thunderx_prefetch_tune): Likewise.
15881 (thunderx2t99_prefetch_tune): Likewise.
15882 (qdf24xx_prefetch_tune): Likewise. Set minimum_stride to 2048.
15883 (aarch64_override_options_internal): Update to set
15884 PARAM_PREFETCH_MINIMUM_STRIDE.
15885 * doc/invoke.texi (prefetch-minimum-stride): Document new option.
15886 * params.def (PARAM_PREFETCH_MINIMUM_STRIDE): New.
15887 * params.h (PARAM_PREFETCH_MINIMUM_STRIDE): Define.
15888 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Return false if
15889 stride is constant and is below the minimum stride threshold.
15891 2018-05-06 Jakub Jelinek <jakub@redhat.com>
15894 * cfgexpand.c (expand_asm_stmt): Don't create a temporary if
15895 the type is addressable. Don't force op into register if it has
15898 2018-05-05 Roland McGrath <mcgrathr@google.com>
15901 * varasm.c (default_section_type_flags): Set SECTION_NOTYPE for
15902 any section for which we don't know a specific type it should have,
15903 regardless of name. Previously this was done only for the exact
15904 names ".init_array", ".fini_array", and ".preinit_array".
15905 (default_elf_asm_named_section): Add comment about
15906 relationship with default_section_type_flags and SECTION_NOTYPE.
15907 (get_section): Don't consider it a type conflict if one side has
15908 SECTION_NOTYPE and the other doesn't, as long as neither has the
15909 SECTION_BSS et al used in the default_section_type_flags logic.
15911 2018-05-05 Tom de Vries <tom@codesourcery.com>
15914 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG_3): Define.
15915 (workaround_barsyncs): New function.
15916 (nvptx_reorg): Use workaround_barsyncs.
15917 * config/nvptx/nvptx.md (define_c_enum "unspecv"): Add UNSPECV_MEMBAR.
15918 (define_expand "nvptx_membar_cta"): New define_expand.
15919 (define_insn "*nvptx_membar_cta"): New insn.
15921 2018-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
15923 * brig-builtins.def: Add consts to ptrs etc. in BRIG builtin defs.
15924 To improve optimization opportunities.
15925 * builtin-types.def: The new needed builtin types for the above.
15927 2018-05-04 Richard Biener <rguenther@suse.de>
15929 * bb-reorder.c (sanitize_hot_paths): Release hot_bbs_to_check.
15930 * gimple-ssa-store-merging.c
15931 (imm_store_chain_info::output_merged_store): Remove redundant create,
15932 release split_store vector contents on failure.
15933 * tree-vect-slp.c (vect_schedule_slp_instance): Avoid leaking
15934 scalar stmt vector on cache hit.
15936 2018-05-04 Segher Boessenkool <segher@kernel.crashing.org>
15938 * common/config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
15940 * config.gcc (powerpc-xilinx-eabi*): Remove.
15941 * config/rs6000/predicates.md (easy_fp_constant): Remove Xilinx FP
15943 (fusion_addis_mem_combo_load): Ditto.
15944 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove Xilinx
15946 (rs6000_cpu_cpp_builtins): Ditto.
15947 * config/rs6000/rs6000-linux.c
15948 (rs6000_linux_float_exceptions_rounding_supported_p): Ditto.
15949 * config/rs6000/rs6000-opts.h (enum fpu_type_t): Delete.
15950 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove Xilinx FP
15952 (rs6000_setup_reg_addr_masks): Ditto.
15953 (rs6000_init_hard_regno_mode_ok): Ditto.
15954 (rs6000_option_override_internal): Ditto.
15955 (legitimate_lo_sum_address_p): Ditto.
15956 (rs6000_legitimize_address): Ditto.
15957 (rs6000_legitimize_reload_address): Ditto.
15958 (rs6000_legitimate_address_p): Ditto.
15959 (abi_v4_pass_in_fpr): Ditto.
15960 (setup_incoming_varargs): Ditto.
15961 (rs6000_gimplify_va_arg): Ditto.
15962 (rs6000_split_multireg_move): Ditto.
15963 (rs6000_savres_strategy): Ditto.
15964 (rs6000_emit_prologue_components): Ditto.
15965 (rs6000_emit_epilogue_components): Ditto.
15966 (rs6000_emit_prologue): Ditto.
15967 (rs6000_emit_epilogue): Ditto.
15968 (rs6000_elf_file_end): Ditto.
15969 (rs6000_function_value): Ditto.
15970 (rs6000_libcall_value): Ditto.
15971 * config/rs6000/rs6000.h: Ditto.
15972 (TARGET_MINMAX_SF, TARGET_MINMAX_DF): Delete, merge to ...
15973 (TARGET_MINMAX): ... this. New.
15974 (TARGET_SF_FPR, TARGET_DF_FPR, TARGET_SF_INSN, TARGET_DF_INSN): Delete.
15975 * config/rs6000/rs6000.md: Remove Xilinx FP support.
15976 (*movsi_internal1_single): Delete.
15977 * config/rs6000/rs6000.opt (msingle-float, mdouble-float, msimple-fpu,
15978 mfpu=, mxilinx-fpu): Delete.
15979 * config/rs6000/singlefp.h: Delete.
15980 * config/rs6000/sysv4.h: Remove Xilinx FP support.
15981 * config/rs6000/t-rs6000: Ditto.
15982 * config/rs6000/t-xilinx: Delete.
15983 * config/rs6000/titan.md: Adjust for fp_type removal.
15984 * config/rs6000/vsx.md: Remove Xilinx FP support.
15985 (VStype_simple): Delete.
15986 (VSfptype_simple, VSfptype_mul, VSfptype_div, VSfptype_sqrt): Delete.
15987 * config/rs6000/xfpu.h: Delete.
15988 * config/rs6000/xfpu.md: Delete.
15989 * config/rs6000/xilinx.h: Delete.
15990 * config/rs6000/xilinx.opt: Delete.
15991 * doc/invoke.texi (RS/6000 and PowerPC Options): Remove
15992 -msingle-float, -mdouble-float, -msimple-fpu, -mfpu=, and -mxilinx-fpu.
15994 2018-05-04 Tom de Vries <tom@codesourcery.com>
15997 * builtins.c (expand_builtin_goacc_parlevel_id_size): Handle null target
16000 2018-05-04 Richard Biener <rguenther@suse.de>
16002 PR middle-end/85627
16003 * tree-complex.c (update_complex_assignment): We are always in SSA form.
16004 (expand_complex_div_wide): Likewise.
16005 (expand_complex_operations_1): Likewise.
16006 (expand_complex_libcall): Preserve EH info of the original stmt.
16007 (tree_lower_complex): Handle removed blocks.
16008 * tree.c (build_common_builtin_nodes): Do not set ECF_NOTRHOW
16009 on complex multiplication and division libcall builtins.
16011 2018-05-04 Richard Biener <rguenther@suse.de>
16013 PR middle-end/85574
16014 * fold-const.c (negate_expr_p): Restrict negation of operand
16015 zero of a division to when we know that can happen without
16017 (fold_negate_expr_1): Likewise.
16019 2018-05-04 Jakub Jelinek <jakub@redhat.com>
16022 * real.h (real_nextafter): Declare.
16023 * real.c (real_nextafter): New function.
16024 * fold-const-call.c (fold_const_nextafter): New function.
16025 (fold_const_call_sss): Call it for CASE_CFN_NEXTAFTER and
16026 CASE_CFN_NEXTTOWARD.
16027 (fold_const_call_1): For CASE_CFN_NEXTTOWARD call fold_const_call_sss
16028 even when arg1_mode is different from arg0_mode.
16030 2018-05-03 Nathan Sidwell <nathan@acm.org>
16032 * doc/extend.texi (Deprecated Features): Remove
16033 -ffriend-injection.
16034 (Backwards Compatibility): Likewise.
16035 * doc/invoke.texi (C++ Language Options): Likewise.
16036 (C++ Dialect Options): Likewise.
16038 2018-05-03 Jakub Jelinek <jakub@redhat.com>
16041 * config/i386/avx512fintrin.h (_mm512_mullox_epi64,
16042 _mm512_mask_mullox_epi64): New intrinsics.
16044 2018-05-03 Tom de Vries <tom@codesourcery.com>
16047 * doc/sourcebuild.texi (Commands for use in dg-final, Scan optimization
16048 dump files): Add offload-tree.
16050 2018-05-03 Richard Biener <rguenther@suse.de>
16052 PR tree-optimization/85615
16053 * tree-ssa-threadupdate.c (thread_block_1): Only allow exits
16054 to loops not nested in BBs loop father to avoid creating multi-entry
16057 2018-05-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16059 PR tree-optimization/70291
16060 * tree-complex.c (expand_complex_libcall): Add type, inplace_p
16061 arguments. Change return type to tree. Emit libcall as a new
16062 statement rather than replacing existing one when inplace_p is true.
16063 (expand_complex_multiplication_components): New function.
16064 (expand_complex_multiplication): Expand floating-point complex
16065 multiplication using the above.
16066 (expand_complex_division): Rename inner_type parameter to type.
16067 Update expand_complex_libcall call-site.
16068 (expand_complex_operations_1): Update expand_complex_multiplication
16069 and expand_complex_division call-sites.
16071 2018-05-02 Jakub Jelinek <jakub@redhat.com>
16074 * config/i386/i386.md (*ashl<dwi>3_doubleword_mask,
16075 *ashl<dwi>3_doubleword_mask_1, *<shift_insn><dwi>3_doubleword_mask,
16076 *<shift_insn><dwi>3_doubleword_mask_1): In condition require that
16077 the highest significant bit of the shift count mask is clear. In
16078 check whether and[sq]i3 is needed verify that all significant bits
16079 of the shift count other than the highest are set.
16081 2018-05-02 Tom de Vries <tom@codesourcery.com>
16084 * builtins.def (DEF_GOACC_BUILTIN_ONLY): Define.
16085 * omp-builtins.def (BUILT_IN_GOACC_PARLEVEL_ID)
16086 (BUILT_IN_GOACC_PARLEVEL_SIZE): New builtin.
16087 * builtins.c (expand_builtin_goacc_parlevel_id_size): New function.
16088 (expand_builtin): Call expand_builtin_goacc_parlevel_id_size.
16089 * doc/extend.texi (Other Builtins): Add __builtin_goacc_parlevel_id and
16090 __builtin_goacc_parlevel_size.
16092 2018-05-02 Richard Biener <rguenther@suse.de>
16094 PR tree-optimization/85597
16095 * tree-vect-stmts.c (vectorizable_operation): For ternary SLP
16096 do not use split vect_get_vec_defs call but call vect_get_slp_defs
16099 2018-05-02 Tom de Vries <tom@codesourcery.com>
16102 * doc/sourcebuild.texi (Commands for use in dg-final, Scan optimization
16103 dump files): Add ltrans-tree.
16105 2018-05-02 Tom de Vries <tom@codesourcery.com>
16108 * doc/sourcebuild.texi (Commands for use in dg-final, Scan optimization
16109 dump files): Add wpa-ipa.
16111 2018-05-02 Segher Boessenkool <segher@kernel.crashing.org>
16113 * config.gcc (powerpc*-*-*): Remove paired.h. Unsupport the
16114 powerpc*-*-linux*paired* target.
16115 * config/rs6000/750cl.h: Delete.
16116 * config/rs6000/paired.h: Delete.
16117 * config/rs6000/paired.md: Delete.
16118 * config/rs6000/predicates.md (easy_vector_constant): Remove paired
16120 * config/rs6000/rs6000-builtin.def: Remove paired float support.
16121 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
16122 comment. Remove paired float support.
16123 * config/rs6000/rs6000-modes.def: Remove V2SF and V2SI.
16124 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete
16126 * config/rs6000/rs6000-protos.h (paired_expand_vector_init,
16127 paired_emit_vector_cond_expr, paired_expand_vector_move): Delete
16129 * config/rs6000/rs6000.c: Remove paired float support.
16130 (paired_expand_vector_init, paired_expand_vector_move,
16131 paired_emit_vector_compare, paired_emit_vector_cond_expr,
16132 (paired_expand_lv_builtin, paired_expand_stv_builtin,
16133 paired_expand_builtin, paired_expand_predicate_builtin,
16134 paired_init_builtins): Delete.
16135 * config/rs6000/rs6000.h: Remove paired float support.
16136 * config/rs6000/rs6000.md: Remove paired float support.
16137 (move_from_CR_ov_bit): Delete.
16138 * config/rs6000/rs6000.opt (mpaired): Delete.
16139 * config/rs6000/t-rs6000: Remove paired.md from MD_INCLUDES.
16140 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mpaired.
16142 2018-05-02 Richard Biener <rguenther@suse.de>
16144 PR middle-end/85567
16145 * gimplify.c (gimplify_save_expr): When in SSA form allow
16146 SAVE_EXPRs to compute to SSA vars.
16148 2018-05-02 Jakub Jelinek <jakub@redhat.com>
16151 * config/i386/i386.md (*ashl<dwi>3_doubleword_mask,
16152 *ashl<dwi>3_doubleword_mask_1, *<shift_insn><dwi>3_doubleword_mask,
16153 *<shift_insn><dwi>3_doubleword_mask_1): If and[sq]i3 is needed, don't
16154 clobber operands[2], instead use a new pseudo. Formatting fixes.
16156 2018-05-02 Richard Sandiford <richard.sandiford@linaro.org>
16158 PR tree-optimization/85586
16159 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Only
16160 exit early for statements in the same group if the accesses are
16163 2018-05-02 Tom de Vries <tom@codesourcery.com>
16166 * lto-wrapper.c (compile_offload_image): Add "could not find mkoffload"
16169 2018-05-01 Marc Glisse <marc.glisse@inria.fr>
16171 PR tree-optimization/85143
16172 * match.pd (A<B&A<C): Extend to BIT_IOR_EXPR.
16174 2018-05-01 Tom de Vries <tom@codesourcery.com>
16177 * config/nvptx/mkoffload.c (main): Suggest using -B in "offload compiler
16178 not found" error message.
16180 2018-05-01 Tom de Vries <tom@codesourcery.com>
16183 * vec.h (VEC_ORDERED_REMOVE_IF, VEC_ORDERED_REMOVE_IF_FROM_TO): Define.
16184 * vec.c (test_ordered_remove_if): New function.
16185 (vec_c_tests): Call test_ordered_remove_if.
16186 * dwarf2cfi.c (connect_traces): Use VEC_ORDERED_REMOVE_IF_FROM_TO.
16187 * lto-streamer-out.c (prune_offload_funcs): Use VEC_ORDERED_REMOVE_IF.
16188 * tree-vect-patterns.c (vect_pattern_recog_1): Use
16189 VEC_ORDERED_REMOVE_IF.
16191 2018-05-01 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
16193 PR tree-optimization/82665
16194 * vr-values.c (vr_values::extract_range_from_binary_expr): Handle
16195 pointer subtraction where arguments come from a memchr call.
16197 2018-05-01 Jakub Jelinek <jakub@redhat.com>
16199 * configure.ac (LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION): Use
16200 --push-state --as-needed and --pop-state instead of --as-needed and
16201 --no-as-needed if ld supports it.
16202 * configure: Regenerated.
16205 * doc/install.texi2html: Replace _002d with - and _002a with * in
16206 generated html files using sed.
16208 2018-04-30 David Malcolm <dmalcolm@redhat.com>
16211 * gcc-rich-location.c (blank_line_before_p): New function.
16212 (use_new_line): New function.
16213 (gcc_rich_location::add_fixit_insert_formatted): New function.
16214 * gcc-rich-location.h
16215 (gcc_rich_location::add_fixit_insert_formatted): New function.
16217 2018-04-30 David Malcolm <dmalcolm@redhat.com>
16219 * selftest.c (assert_streq): Rename "expected" and "actual" to
16220 "val1" and "val2". Extend NULL-handling to cover both inputs
16221 symmetrically, while still requiring both to be non-NULL for a pass.
16222 * selftest.h (assert_streq): Rename "expected" and "actual" to
16224 (ASSERT_EQ): Likewise.
16225 (ASSERT_EQ_AT): Likewise.
16226 (ASSERT_KNOWN_EQ): Likewise.
16227 (ASSERT_KNOWN_EQ_AT): Likewise.
16228 (ASSERT_NE): Likewise.
16229 (ASSERT_MAYBE_NE): Likewise.
16230 (ASSERT_MAYBE_NE_AT): Likewise.
16231 (ASSERT_STREQ): Likewise. Clarify that both must be non-NULL for
16232 the assertion to pass.
16233 (ASSERT_STREQ_AT): Likewise.
16235 2018-04-30 Jonathan Wakely <jwakely@redhat.com>
16237 * doc/invoke.texi (-fpie, -fPIE): Fix grammar and clarify
16238 interaction with -pie.
16240 2018-04-30 David Malcolm <dmalcolm@redhat.com>
16242 * selftest.h: Fix alphabetization of per-source-file selftest
16245 2018-04-30 Jason Merrill <jason@redhat.com>
16247 PR c++/61982 - dead stores to destroyed objects.
16248 * gimplify.c (gimplify_modify_expr): Simplify complex lvalue on LHS
16251 2018-04-30 Jason Merrill <jason@redhat.com>
16253 * tree.c (build_clobber): New.
16254 * tree.h: Declare it.
16255 * gimplify.c (gimplify_bind_expr, gimplify_target_expr): Use it.
16257 2018-04-30 David Malcolm <dmalcolm@redhat.com>
16259 * diagnostic-show-locus.c (layout::layout): Update for
16260 location_get_source_line returning a char_span.
16261 (struct char_span): Move to input.h.
16262 (struct correction): Update for fields in char_span becoming
16264 (struct source_line): Update for location_get_source_line
16265 returning a char_span.
16266 (layout::print_line): Likewise.
16267 * edit-context.c (edited_file::print_content): Likewise.
16268 (edited_file::print_diff_hunk): Likewise.
16269 (edited_file::print_run_of_changed_lines): Likewise.
16270 (edited_file::get_num_lines): Likewise.
16271 (edited_line::edited_line): Likewise.
16272 * final.c (asm_show_source): Likewise.
16273 * input.c (location_get_source_line): Convert return type
16274 from const char * to char_span, losing the final "line_len"
16276 (dump_location_info): Update for the above.
16277 (get_substring_ranges_for_loc): Likewise. Use a char_span
16278 when handling the literal within the line.
16279 (test_reading_source_line): Update for location_get_source_line
16280 returning a char_span.
16281 * input.h (class char_span): Move here from
16282 diagnostic-show-locus.c, converting from a struct to a class.
16283 Make data members private.
16284 (char_span::operator bool): New.
16285 (char_span::length): New.
16286 (char_span::get_buffer): New.
16287 (char_span::operator[]): New.
16288 (char_span::subspan): Make const.
16289 (char_span::xstrdup): New.
16290 (location_get_source_line): Convert return type from const char *
16291 to char_span, losing the final "line_size" param.
16293 2018-04-30 Jan Hubicka <jh@suse.cz>
16295 * lto-wrapper.c (ltrans_priorities): New static var.
16296 (cmp_priority): New.
16297 (run_gcc): Read priorities and if doing parallel build order
16298 the Makefile by them.
16300 2018-04-30 David Malcolm <dmalcolm@redhat.com>
16302 * input.h (builtins_location_check): Convert to a STATIC_ASSERT.
16304 2018-04-30 Richard Biener <rguenther@suse.de>
16306 * tree-cfg.c (verify_address): Remove base argument, add
16307 flag whether to check TREE_ADDRESSABLE and do that.
16308 (verify_expr): Remove.
16309 (verify_types_in_gimple_reference): Add pieces from verify_expr.
16310 (verify_gimple_assign_single): Likewise.
16311 (verify_gimple_switch): Likewise.
16312 (verify_expr_location_1): Dereference tp once. Add (disabled)
16313 piece from verify_expr.
16314 (verify_gimple_in_cfg): Do not call verify_expr on all ops.
16316 2018-04-30 Claudiu Zissulescu <claziss@synopsys.com>
16318 * config/arc/linux.h (CLEAR_INSN_CACHE): Define.
16320 2018-04-30 Claudiu Zissulescu <claziss@synopsys.com>
16322 * config/arc/arc-protos.h (prepare_extend_operands): Remove.
16323 (small_data_pattern): Likewise.
16324 (arc_rewrite_small_data): Likewise.
16325 * config/arc/arc.c (LEGITIMATE_SMALL_DATA_OFFSET_P): Remove.
16326 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Likewise.
16327 (get_symbol_alignment): New function.
16328 (legitimate_small_data_address_p): Likewise.
16329 (legitimate_scaled_address): Update, call
16330 legitimate_small_data_address_p.
16331 (output_sdata): New static variable.
16332 (arc_print_operand): Update how we handle small data operands.
16333 (arc_print_operand_address): Likewise.
16334 (arc_legitimate_address_p): Update, use
16335 legitimate_small_data_address_p.
16336 (arc_rewrite_small_data_p): Remove.
16337 (arc_rewrite_small_data_1): Likewise.
16338 (arc_rewrite_small_data): Likewise.
16339 (small_data_pattern): Likewise.
16340 (compact_sda_memory_operand): Update to use
16341 legitimate_small_data_address_p and get_symbol_alignment.
16342 (prepare_move_operands): Don't rewite sdata pattern.
16343 (prepare_extend_operands): Remove.
16344 * config/arc/arc.md (zero_extendqihi2): Don't rewrite sdata
16346 (zero_extendqisi2): Likewise.
16347 (zero_extendhisi2): Likewise.
16348 (extendqihi2): Likewise.
16349 (extendqisi2): Likewise.
16350 (extendhisi2): Likewise.
16351 (addsi3): Likewise.
16352 (subsi3): Likewise.
16353 (andsi3): Likewise.
16354 * config/arc/constraints.md (Usd): Change it to memory constraint.
16356 2018-04-30 Claudiu Zissulescu <claziss@synopsys.com>
16358 * config/arc/arc.c (arc_split_move): Allow signed 6-bit constants
16359 as source of std instructions.
16360 * config/arc/arc.md (movsi_insn): Update pattern predicate to
16361 allow 6-bit constants as source for store instructions.
16362 (movdi_insn): Update instruction pattern to allow 6-bit constants
16363 as source for store instructions.
16365 2018-04-30 Jonathan Wakely <jwakely@redhat.com>
16367 * doc/invoke.texi (-fdebug-types-section): Fix grammar.
16369 2018-04-30 Nathan Sidwell <nathan@acm.org>
16370 Sandra Loosemore <sandra@codesourcery.com>
16372 * dumpfile.c (dump_open): Allow '-' for stdout.
16373 * doc/invoke.texi (Developer Options): Document dump filename
16374 determination early. Document stdin/stdout selection.
16376 2018-04-30 Andrew Sadek <andrew.sadek.se@gmail.com>
16378 Microblaze Target: PIC data text relative
16380 * config/microblaze/microblaze.opt: add new option -mpic-data-text-rel.
16381 * config/microblaze/microblaze-protos.h (microblaze_constant_address_p):
16383 * config/microblaze/microblaze.h (microblaze_constant_address_p):
16384 CONSTANT_ADDRESS_P definition to microblaze_constant_address_p.
16385 * config/microblaze/microblaze.c (TARGET_PIC_DATA_TEXT_REL):
16386 New addressing mode for data-text relative position indepenedent code.
16387 (microblaze_classify_unspec): add 'UNSPEC_TEXT' case ->
16388 'ADDRESS_SYMBOLIC_TXT_REL'.
16389 (microblaze_classify_address): Add handling for UNSPEC + CONST_INT.
16390 (microblaze_legitimate_pic_operand): Exclude function calls from
16391 pic operands in case of TARGET_PIC_DATA_TEXT_REL option.
16392 (microblaze_legitimize_address): Generate 'UNSPEC_TEXT' for all possible
16394 (microblaze_address_insns): Add 'ADDRESS_SYMBOLIC_TXT_REL' case.
16395 (print_operand): Add 'ADDRESS_SYMBOLIC_TXT_REL' case.
16396 (print_operand_address): Add 'ADDRESS_SYMBOLIC_TXT_REL' case + handling
16397 for 'address + offset'.
16398 (microblaze_expand_prologue): Add new function prologue call for
16400 (microblaze_asm_generate_pic_addr_dif_vec): Override new target hook
16401 'TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC' to disable address diff vector
16402 table in case of TARGET_PIC_DATA_TEXT_REL.
16403 (expand_pic_symbol_ref): Add handling for 'UNSPEC_TEXT'.
16404 * config/microblaze/microblaze.md (TARGET_PIC_DATA_TEXT_REL):
16405 Add new macros 'UNSPEC_TEXT',
16406 'UNSPEC_SET_TEXT' + add rule for setting r20 in function prologue
16407 + exclude function calls from 'UNSPEC_PLT' in case of data text
16409 * doc/tm.texi.in (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC): Add
16410 new target hook for generating address diff vector tables in case of
16412 * doc/tm.texi : Regenerate.
16413 * stmt.c (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC): Append new condition
16414 'targetm.asm_out.generate_pic_addr_diff_vec' to flag_pic in case
16415 of addr diff vector generation.
16416 * target.def (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC): Add
16417 target hook definition.
16418 * targhooks.h, targhooks.c (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC):
16419 Add default function for generate_pic_addr_diff_vec -> flag_pic.
16420 * doc/invoke.texi (Add new pic option): Add new microblaze pic
16421 option for data text relative.
16423 2018-04-30 Richard Biener <rguenther@suse.de>
16425 * tree-chrec.h (evolution_function_is_constant_p): Remove
16427 * tree-cfg.c (tree_node_can_be_shared): Re-order checks.
16429 2018-04-30 Richard Biener <rguenther@suse.de>
16432 * dwarf2out.c (gen_producer_string): Ignore -fchecking[=].
16434 2018-04-30 Richard Biener <rguenther@suse.de>
16436 PR tree-optimization/28364
16437 PR tree-optimization/85275
16438 * tree-ssa-loop-ch.c (ch_base::copy_headers): Stop after
16439 copying first exit test.
16441 2018-04-28 Mark Wielaard <mark@klomp.org>
16443 * dwarf2out.c (dwarf2out_finish): Add .debug_addr table header for
16444 dwarf_version >= 5.
16445 (dwarf_AT): Handle DW_AT_addr_base.
16446 (add_top_level_skeleton_die_attrs): Use dwarf_AT for DW_AT_addr_base.
16448 2018-04-28 Uros Bizjak <ubizjak@gmail.com>
16451 * config/i386/i386.md (*ashl<dwi>3_doubleword_mask): New pattern.
16452 (*ashl<dwi>3_doubleword_mask_1): Ditto.
16453 (*<shift_insn><dwi>3_doubleword_mask): Ditto.
16454 (*<shift_insn><dwi>3_doubleword_mask_1): Ditto.
16456 2018-04-28 Richard Biener <rguenther@suse.de>
16458 * tree-cfg.c (verify_gimple_phi): Take a gphi * argument.
16459 (verify_gimple_in_cfg): Rename visited_stmts to visited_throwing_stmts
16460 to reflect use. Only add interesting stmts.
16462 2018-04-27 Martin Jambor <mjambor@suse.cz>
16465 * ipa-cp.c (find_aggregate_values_for_callers_subset): Make sure
16466 the jump function allows for passing through aggregate values.
16468 2018-04-27 David Malcolm <dmalcolm@redhat.com>
16470 * input.h (in_system_header_at): Convert from macro to inline
16472 (from_macro_expansion_at): Likewise.
16473 (from_macro_definition_at): Likewise.
16475 2018-04-27 Jeff Law <law@redhat.com>
16477 * config.gcc: Mark tile* targets as deprecated/obsolete.
16479 2018-04-27 Richard Biener <rguenther@suse.de>
16481 * config/aarch64/aarch64.c: Simplify ap.__stack advance and
16484 2018-04-27 Richard Biener <rguenther@suse.de>
16486 * tree-cfg.c (verify_expr): Make dead code hit gcc_unreachable.
16488 2018-04-27 Uros Bizjak <ubizjak@gmail.com>
16490 * config/i386/i386.md (*movti_internal): Substitute Ye constraint
16491 with Yd constraint. Set "preferred_for_speed" attribute from
16492 TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC for alternatives
16493 with Yd constraint.
16494 (*movdi_internal): Ditto.
16495 (movti_interunit splitters): Remove
16496 TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC from insn condition.
16497 (movdi_interunit splitters): Ditto.
16498 * config/i386/constraints.md (Ye): Remove.
16499 (Yd): Do not depend on TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC.
16501 2018-04-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16504 * config/aarch64/constraints.md (Usg): Limit to 31.
16505 (Usj): Limit to 63.
16507 2018-04-27 Jakub Jelinek <jakub@redhat.com>
16509 PR tree-optimization/85529
16510 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Add FIRST_BB
16511 argument. Don't call get_nonzero_bits if opcode is ERROR_MARK_NODE,
16512 rhs2 def stmt's bb is dominated by first_bb and it isn't an obvious
16513 zero extension or masking of the MSB bit.
16514 (optimize_range_tests): Add FIRST_BB argument, pass it through
16515 to optimize_range_tests_var_bound.
16516 (maybe_optimize_range_tests, reassociate_bb): Adjust
16517 optimize_range_tests callers.
16519 2018-04-26 Richard Biener <rguenther@suse.de>
16520 Jakub Jelinek <jakub@redhat.com>
16522 * cgraph.h (symbol_table): Just declare debug method here.
16523 * symtab.c (symbol_table::debug): Define.
16525 2018-04-26 Eric Botcazou <ebotcazou@adacore.com>
16527 * loop-invariant.c (may_assign_reg_p): Return false for frame pointer.
16529 2018-04-26 Uros Bizjak <ubizjak@gmail.com>
16531 * config/i386/i386.md ("isa" attribute): Add x64_sse2.
16532 ("enabled" attribute): Handle x64_sse2 "isa" attribute.
16533 (*movdi_internal): Substitute Yi and Yj constraint with x
16534 and Ym and Yn constraint with y constraint. Update "isa"
16535 attribute and set "preferred_for_speed" attribute from
16536 TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC for updated alternatives.
16537 (*movsi_internal): Ditto.
16538 (*movdf_internal): Ditto.
16539 (*movsf_internal): Ditto.
16540 (*zero_extendsidi2): Ditto.
16541 * config/i386/sse.md (vec_set<mode>_0): Ditto.
16542 (sse2_loadld): Ditto.
16543 (*vec_extract<ssevecmodelower>_0): Ditto.
16544 (*vec_extractv4si_0_zext_sse4): Ditto.
16545 (vec_concatv2di): Ditto.
16546 (*vec_dup<mode>): Ditto.
16547 * config/i386/mmx.md (*mov<mode>_internal): Ditto.
16548 * config/i386/constraints.md (Yi): Remove.
16553 2018-04-26 Nathan Sidwell <nathan@acm.org>
16555 * dumpfile.c (dump_open): New.
16556 (dump_open_alternate_stream, dump_start, dump_begin): Call it.
16557 (dump_finish): Detect stdio/stderr by value not name.
16559 2018-04-26 Jonathan Wakely <jwakely@redhat.com>
16561 * doc/invoke.texi (-Wreturn-type): Document default status for C++.
16563 2018-04-26 Tom de Vries <tom@codesourcery.com>
16566 * config/nvptx/nvptx.c (verify_neutering_jumps)
16567 (verify_neutering_labels): New function
16568 (nvptx_single): Use verify_neutering_jumps and verify_neutering_labels.
16570 2018-04-26 Tom de Vries <tom@codesourcery.com>
16573 * config/nvptx/nvptx.c (needs_neutering_p): New function.
16574 (nvptx_single): Use needs_neutering_p to skip over insns that do not
16577 2018-04-26 Richard Biener <rguenther@suse.de>
16578 Tom de Vries <tom@codesourcery.com>
16581 * lto-streamer-out.c (output_function): Fixup loops if required to match
16582 discovery done in the reader.
16584 2018-04-26 Richard Biener <rguenther@suse.de>
16586 PR tree-optimization/85116
16587 * tree-ssa-loop-ch.c (do_while_loop_p): A do-while loop should
16588 have a loop exit from the single latch predecessor. Remove
16589 case of header with just condition.
16590 (ch_base::copy_headers): Exclude infinite loops from any
16592 (pass_ch::execute): Record exits.
16594 2018-04-26 Richard Biener <rguenther@suse.de>
16596 * tree-vect-data-refs.c (vect_get_data_access_cost): Get
16597 prologue cost vector and pass it to vect_get_load_cost.
16598 (vect_get_peeling_costs_all_drs): Likewise.
16599 (vect_peeling_hash_get_lowest_cost): Likewise.
16600 (vect_enhance_data_refs_alignment): Likewise.
16602 2018-04-26 Richard Biener <rguenther@suse.de>
16604 PR middle-end/85450
16605 * tree-cfg.c (verify_gimple_assign_unary): Restore proper
16606 checking of integer<->pointer conversions.
16607 * omp-expand.c (expand_omp_for_static_nochunk): Avoid
16608 sign-/zero-extending pointer types.
16609 (expand_omp_for_static_chunk): Likewise.
16611 2018-03-22 Hans-Peter Nilsson <hp@axis.com>
16612 Jean Lee <xiaoyur347@gmail.com>
16614 * config/mips/mips.c (mips_asan_shadow_offset): New function.
16615 (TARGET_ASAN_SHADOW_OFFSET): Define.
16616 * config/mips/mips.h (FRAME_GROWS_DOWNWARD): Augment to also be
16617 true for -fsanitize=address.
16619 2018-04-25 Mark Wielaard <mark@klomp.org>
16621 * dwarf2out.c (file_info_cmp): Sort longer dir prefixes before
16624 2018-04-25 Jakub Jelinek <jakub@redhat.com>
16626 * config/i386/i386.md (*x86_mov<mode>cc_0_m1): Use type "alu1" rather
16627 than "alu", remove explicit "memory" and "imm_disp" attributes.
16628 (*x86_mov<mode>cc_0_m1_se, *x86_mov<mode>cc_0_m1_neg): Likewise.
16630 PR middle-end/85414
16631 * simplify-rtx.c (simplify_unary_operation_1) <case SIGN_EXTEND,
16632 case ZERO_EXTEND>: Pass SUBREG_REG (op) rather than op to
16633 gen_lowpart_no_emit.
16635 2018-04-25 Sebastian Peryt <sebastian.peryt@intel.com>
16638 * config/i386/i386.c (ix86_expand_builtin): Change memory
16639 operand to XI, extend p0 to Pmode.
16640 * config/i386/i386.md: Change unspec volatile and operand
16641 1 mode to XI, change operand 0 mode to P.
16643 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
16645 * config/nds32/nds32-predicates.c (nds32_can_use_bclr_p): Mask with
16646 GET_MODE_MASK before any checking.
16647 (nds32_can_use_bset_p): Likewise.
16648 (nds32_can_use_btgl_p): Likewise.
16650 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
16652 * config/nds32/nds32-doubleword.md: New define_split pattern for
16653 illegal register number.
16655 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
16657 * config/nds32/nds32.c (nds32_print_operand): Set op_value ealier.
16659 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
16661 * config/nds32/nds32.h (ASM_APP_ON): Add missing newline character.
16663 2018-04-25 Richard Biener <rguenther@suse.de>
16665 * lto-streamer.h (LTO_major_version): Bump to 8.
16667 2018-04-25 Jakub Jelinek <jakub@redhat.com>
16669 * BASE-VER: Set to 9.0.0.
16671 2018-04-24 Segher Boessenkool <segher@kernel.crashing.org>
16673 * config/rs6000/rs6000.c (init_float128_ieee): Fix spelling mistakes
16674 in __abskf2 and __powikf2.
16676 2018-04-24 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16679 * config/aarch64/constraints.md (Usg, Usj): New constraints.
16680 * config/aarch64/iterators.md (cmode_simd): New mode attribute.
16681 * config/aarch64/aarch64.md (*aarch64_ashr_sisd_or_int_<mode>3):
16682 Use the above on operand 2. Reindent.
16683 (*aarch64_lshr_sisd_or_int_<mode>3): Likewise.
16685 2018-04-24 H.J. Lu <hongjiu.lu@intel.com>
16688 * common/config/i386/i386-common.c (ix86_handle_option): Don't
16690 * config/i386/i386.opt (mcet): Removed.
16691 * doc/install.texi: Remove -mcet documentation.
16692 * doc/invoke.texi: Likewise.
16694 2018-04-24 H.J. Lu <hongjiu.lu@intel.com>
16697 * doc/install.texi: Remove -mcet from bootstrap-cet.
16699 2018-04-24 Jakub Jelinek <jakub@redhat.com>
16702 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Don't define
16703 __builtin_ia32_readeflags_u32 and __builtin_ia32_writeeflags_u32
16707 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Punt if
16708 const_vector is not CONST_VECTOR or SYMBOL_REF for a constant pool
16709 containing a CONST_VECTOR.
16711 2018-04-24 Cesar Philippidis <cesar@codesourcery.com>
16713 * doc/install.texi: Update newlib dependency for nvptx.
16715 2018-04-24 Jakub Jelinek <jakub@redhat.com>
16718 * config/i386/i386.c (ix86_expand_vector_init_one_var): Use UINTVAL
16719 instead of INTVAL when shifting x left.
16721 2018-04-24 Andreas Krebbel <krebbel@linux.ibm.com>
16723 PR tree-optimization/85478
16724 * tree-vect-loop.c (vect_analyze_loop_2): Do not call
16725 vect_grouped_store_supported for single element vectors.
16727 2018-04-24 Richard Biener <rguenther@suse.de>
16730 * config/i386/i386.c (ix86_add_stmt_cost): Restrict strided
16731 load cost increase to the case of non-constant step.
16733 2018-04-24 Jakub Jelinek <jakub@redhat.com>
16736 * reg-stack.c (move_for_stack_reg): Don't fail assertion about dead
16737 destination if any_malformed_asm.
16739 2018-04-23 Eric Botcazou <ebotcazou@adacore.com>
16741 PR middle-end/85496
16742 * expr.c (store_field): In the bitfield case, if the value comes from
16743 a function call and is returned in registers by means of a PARALLEL,
16744 do not change the mode of the temporary unless BLKmode and VOIDmode.
16746 2018-04-23 Andrey Belevantsev <abel@ispras.ru>
16748 PR rtl-optimization/85423
16749 * sel-sched-ir.c (has_dependence_note_mem_dep): Only discard
16750 dependencies to debug insns when the previous insn is non-debug.
16752 2018-04-23 Claudiu Zissulescu <claziss@synopsys.com>
16754 * config/arc/arc.md ("vunspec"): Delete it, unify all the unspec
16755 enums into a single definition.
16756 (fls): Fix predicates and printing.
16759 2018-04-23 Claudiu Zissulescu <claziss@synopsys.com>
16761 * config/arc/arc-protos.h (check_if_valid_sleep_operand): Remove.
16762 * config/arc/arc.c (arc_expand_builtin): Sleep accepts registers
16763 and short u6 immediate.
16764 (check_if_valid_sleep_operand): Remove.
16765 * config/arc/arc.md (Sleep): Accepts registers and u6 immediates.
16767 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
16769 * config/nds32/nds32.c (nds32_compute_stack_frame): Consider
16770 flag_always_save_lp condition.
16771 * config/nds32/nds32.opt (malways-save-lp): New option.
16773 2018-04-22 Shiva Chen <shiva0217@gmail.com>
16775 * config/nds32/nds32-protos.h (nds32_use_load_post_increment): Declare.
16776 * config/nds32/nds32.c (nds32_use_load_post_increment): New.
16777 * config/nds32/nds32.h
16778 (USE_LOAD_POST_INCREMENT, USE_LOAD_POST_DECREMENT): Define.
16779 (USE_STORE_POST_INCREMENT, USE_STORE_POST_DECREMENT): Define.
16781 2018-04-22 Shiva Chen <shiva0217@gmail.com>
16783 * config/nds32/nds32-protos.h (nds32_ls_333_p): Remove.
16784 * config/nds32/nds32.c (nds32_ls_333_p): Remove.
16786 2018-04-22 Shiva Chen <shiva0217@gmail.com>
16787 Chung-Ju Wu <jasonwucj@gmail.com>
16789 * config/nds32/nds32-protos.h (nds32_case_vector_shorten_mode):
16791 * config/nds32/nds32.c (nds32_case_vector_shorten_mode): New function.
16792 * config/nds32/nds32.h (CASE_VECTOR_SHORTEN_MODE): Modify.
16794 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
16796 * config/nds32/nds32.c (nds32_compute_stack_frame): Fix wrong value.
16798 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
16800 * config/nds32/nds32-protos.h (nds32_data_alignment,
16801 nds32_local_alignment): Declare.
16802 * config/nds32/nds32.c (nds32_data_alignment, nds32_constant_alignment,
16803 nds32_local_alignment): New functions.
16804 (TARGET_CONSTANT_ALIGNMENT): Define.
16805 * config/nds32/nds32.h (DATA_ALIGNMENT, LOCAL_ALIGNMENT): Define.
16807 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
16809 * config/nds32/nds32.c
16810 (TARGET_HARD_REGNO_MODE_OK): Move to the bottom of file.
16811 (TARGET_MODES_TIEABLE_P): Likewise.
16813 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
16815 * config/nds32/nds32.c (nds32_asm_file_start): Display optimization
16816 level Ofast and Og.
16818 2018-04-22 Monk Chiang <sh.chiang04@gmail.com>
16819 Chung-Ju Wu <jasonwucj@gmail.com>
16821 * config/nds32/constants.md (unspec_volatile_element): Add enum values
16822 for unaligned access.
16823 * config/nds32/nds32-intrinsic.c: Implementation of expanding
16825 * config/nds32/nds32-intrinsic.md: Likewise.
16826 * config/nds32/nds32_intrinsic.h: Likewise.
16827 * config/nds32/nds32.h (nds32_builtins): Likewise.
16828 * config/nds32/nds32.opt (munaligned-access): New option.
16829 * config/nds32/nds32.c (nds32_asm_file_start): Display
16830 flag_unaligned_access status.
16832 2018-04-20 Kito Cheng <kito.cheng@gmail.com>
16834 * config/riscv/elf.h (LINK_SPEC): Pass --no-relax if
16835 -mno-relax is present.
16836 * config/riscv/linux.h (LINK_SPEC): Ditto.
16838 2018-04-20 Martin Sebor <msebor@redhat.com>
16841 * gimple-fold.c (gimple_fold_builtin_strcpy): Suppress -Wrestrict
16843 (gimple_fold_builtin_stxcpy_chk): Same.
16844 * gimple-ssa-warn-restrict.c (check_bounds_or_overlap): Same.
16846 2018-04-20 Michael Meissner <meissner@linux.ibm.com>
16849 * config/rs6000/rs6000.c (init_float128_ieee): Add support to call
16850 __powikf2 when long double is IEEE 128-bit.
16852 2018-04-20 Kito Cheng <kito.cheng@gmail.com>
16854 * config/riscv/riscv.c (riscv_first_stack_step): Round up min
16855 step to make sure stack always aligned.
16857 2018-04-20 Carl Love <cel@us.ibm.com>
16860 * config/rs6000/rs6000-c.c (rs6000_gimple_fold_builtin): Add
16861 size check for arg0.
16863 2018-04-20 Nathan Sidwell <nathan@codesourcery.com>
16864 Tom de Vries <tom@codesourcery.com>
16867 * config/nvptx/nvptx.c (nvptx_emit_forking, nvptx_emit_joining):
16868 Emit insns for calls too.
16869 (nvptx_find_par): Always look for worker-level predecessor insn.
16870 (nvptx_propagate): Add is_call parm, return bool. Copy frame for
16872 (nvptx_vpropagate, nvptx_wpropagate): Adjust.
16873 (nvptx_process_pars): Propagate frames for calls.
16875 2018-04-20 H.J. Lu <hongjiu.lu@intel.com>
16878 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET):
16880 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
16881 (ix86_handle_option): Don't handle OPT_mibt.
16882 * config/i386/cet.h: Check __CET__ instead of __IBT__ and
16884 * config/i386/driver-i386.c (host_detect_local_cpu): Remove
16886 * config/i386/i386-c.c (ix86_target_macros_internal): Don't
16887 check OPTION_MASK_ISA_IBT nor flag_cf_protection.
16888 (ix86_target_macros): Define __CET__ with flag_cf_protection
16889 for -fcf-protection.
16890 * config/i386/i386.c (isa2_opts): Remove -mibt.
16891 * config/i386/i386.h (TARGET_IBT): Removed.
16892 (TARGET_IBT_P): Likewise.
16893 (ix86_valid_target_attribute_inner_p): Don't check OPT_mibt.
16894 * config/i386/i386.md (nop_endbr): Don't check TARGET_IBT.
16895 * config/i386/i386.opt (mcet): Update help message.
16896 (mshstk): Likewise.
16898 * doc/invoke.texi: Remove -mibt. Document __CET__. Document
16899 -mcet as an alias for -mshstk.
16901 2018-04-20 Richard Biener <rguenther@suse.de>
16903 PR middle-end/85475
16904 * match.pd ((X * CST) * Y -> (X * Y) * CST): Avoid exponential
16905 complexity by forcing a single use of the multiply operand.
16907 2018-04-20 Martin Jambor <mjambor@suse.cz>
16910 * ipa-cp.c (cgraph_edge_brings_value_p): Move check for self-feeding
16911 recursion dependency to only apply to non-clones.
16913 2018-04-20 Martin Jambor <mjambor@suse.cz>
16916 * ipa-cp.c (create_specialized_node): Check that clones of
16917 self-recursive edges exist during IPA-CP.
16919 2018-04-19 Toon Moene <toon@moene.org>
16921 * doc/invoke.texi: Add -floop-unroll-and-jam to options enabled
16924 2018-04-19 Jakub Jelinek <jakub@redhat.com>
16926 PR tree-optimization/85467
16927 * fold-const.c (fold_ternary_loc) <case BIT_FIELD_REF>: Use
16928 VECTOR_TYPE_P macro. If type is vector type, VIEW_CONVERT_EXPR the
16929 VECTOR_CST element to type.
16931 2018-04-19 H.J. Lu <hongjiu.lu@intel.com>
16934 * config/i386/i386.h (STACK_SAVEAREA_MODE): New.
16935 * config/i386/i386.md (builtin_setjmp_setup): Removed.
16936 (builtin_longjmp): Likewise.
16937 (save_stack_nonlocal): New pattern.
16938 (restore_stack_nonlocal): Likewise.
16940 2018-04-19 H.J. Lu <hongjiu.lu@intel.com>
16943 * config/i386/cet.c (file_end_indicate_exec_stack_and_cet):
16944 Replace ASM_OUTPUT_LABEL with fprintf.
16946 2018-04-19 H.J. Lu <hongjiu.lu@intel.com>
16949 * config/i386/cet.c (file_end_indicate_exec_stack_and_cet):
16950 Check flag_cf_protection instead of TARGET_IBT and TARGET_SHSTK.
16951 * config/i386/i386-c.c (ix86_target_macros_internal): Also
16952 define __IBT__ and __SHSTK__ for -fcf-protection.
16953 * config/i386/i386.c (pass_insert_endbranch::gate): Don't check
16955 (ix86_trampoline_init): Likewise.
16956 (x86_output_mi_thunk): Likewise.
16957 (ix86_notrack_prefixed_insn_p): Likewise.
16958 (ix86_option_override_internal): Don't disallow -fcf-protection.
16959 * config/i386/i386.md (rdssp<mode>): Also enable for
16961 (incssp<mode>): Likewise.
16962 (nop_endbr): Likewise.
16963 * config/i386/i386.opt (mcet): Change help message to built-in
16966 (mshstk): Likewise.
16967 * doc/invoke.texi: Remove -mcet, -mibt and -mshstk condition
16968 on -fcf-protection. Change -mcet, -mibt and -mshstk to only
16969 enable CET built-in functions.
16971 2018-04-19 Sebastian Peryt <sebastian.peryt@intel.com>
16973 * common/config/i386/i386-common.c
16974 (OPTION_MASK_ISA_MOVDIRI_SET, OPTION_MASK_ISA_MOVDIR64B_SET,
16975 OPTION_MASK_ISA_MOVDIRI_UNSET,
16976 OPTION_MASK_ISA_MOVDIR64B_UNSET): New defines.
16977 (ix86_handle_option): Handle -mmovdiri and -mmovdir64b.
16978 * config.gcc (movdirintrin.h): New header.
16979 * config/i386/cpuid.h (bit_MOVDIRI,
16980 bit_MOVDIR64B): New bits.
16981 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mmovdiri
16983 * config/i386/i386-builtin-types.def ((VOID, PUNSIGNED, UNSIGNED),
16984 (VOID, PVOID, PCVOID)): New function types.
16985 * config/i386/i386-builtin.def (__builtin_ia32_directstoreu_u32,
16986 __builtin_ia32_directstoreu_u64,
16987 __builtin_ia32_movdir64b): New builtins.
16988 * config/i386/i386-c.c (__MOVDIRI__, __MOVDIR64B__): New.
16989 * config/i386/i386.c (ix86_target_string): Added -mmovdir64b
16991 (ix86_valid_target_attribute_inner_p): Ditto.
16992 (ix86_expand_special_args_builtin): Added VOID_FTYPE_PUNSIGNED_UNSIGNED
16993 and VOID_FTYPE_PUNSIGNED_UNSIGNED.
16994 (ix86_expand_builtin): Expand IX86_BUILTIN_MOVDIR64B.
16995 * config/i386/i386.h (TARGET_MOVDIRI, TARGET_MOVDIRI_P,
16996 TARGET_MOVDIR64B, TARGET_MOVDIR64B_P): New.
16997 * config/i386/i386.md (UNSPECV_MOVDIRI, UNSPECV_MOVDIR64B): New.
16998 (movdiri<mode>, movdir64b_<mode>): New.
16999 * config/i386/i386.opt: Add -mmovdiri and -mmovdir64b.
17000 * config/i386/immintrin.h: Include movdirintrin.h.
17001 * config/i386/movdirintrin.h: New file.
17002 * doc/invoke.texi: Added -mmovdiri and -mmovdir64b.
17004 2018-04-19 Richard Biener <rguenther@suse.de>
17006 PR middle-end/85455
17007 * cfg.c (clear_bb_flags): When loop state says we have
17008 marked irreducible regions also preserve BB_IRREDUCIBLE_LOOP.
17010 2018-04-19 Richard Biener <rguenther@suse.de>
17012 PR tree-optimization/84737
17013 * tree-vect-data-refs.c (vect_copy_ref_info): New function
17014 copying restrict info.
17015 (vect_setup_realignment): Use it.
17016 * tree-vectorizer.h (vect_copy_ref_info): Declare.
17017 * tree-vect-stmts.c (vectorizable_store): Copy ref info from
17018 the first DR to all generated stores.
17019 (vectorizable_load): Likewise for loads.
17021 2018-04-19 Jakub Jelinek <jakub@redhat.com>
17023 PR tree-optimization/85446
17024 * match.pd ((intptr_t) x eq/ne CST to x eq/ne (typeof x) cst): Require
17025 the integral and pointer types to have the same precision.
17027 * doc/install.texi: Document --disable-cet being the default and
17030 2018-04-18 Martin Liska <mliska@suse.cz>
17032 * ipa-devirt.c (odr_subtypes_equivalent_p): Fix GNU coding
17035 2018-04-18 Martin Liska <mliska@suse.cz>
17038 2018-03-02 Eric Botcazou <ebotcazou@adacore.com>
17041 * ipa-devirt.c (odr_subtypes_equivalent_p): Get the ODR type of both
17042 arguments if they are comparable.
17044 2018-04-18 Martin Liska <mliska@suse.cz>
17047 2018-03-13 Eric Botcazou <ebotcazou@adacore.com>
17050 * ipa-devirt.c (odr_subtypes_equivalent_p): Do not get the ODR type of
17053 2018-04-18 H.J. Lu <hongjiu.lu@intel.com>
17056 * config/i386/i386.c (ix86_expand_split_stack_prologue): Insert
17057 ENDBR after calling __morestack.
17059 2018-04-18 David Malcolm <dmalcolm@redhat.com>
17062 * configure.ac (gcc-driver-name.h): Honor --with-gcc-major-version
17063 by using gcc_base_ver to generate a gcc_driver_version, and use
17064 it when generating GCC_DRIVER_NAME.
17065 * configure: Regenerate.
17067 2018-04-18 Jakub Jelinek <jakub@redhat.com>
17070 * config.gcc: Obsolete powerpc*-*-*spe*.
17072 2018-04-17 Jakub Jelinek <jakub@redhat.com>
17075 * dbxout.c (dbxout_int): Perform negation in unsigned int type.
17076 (stabstr_D): Change type of unum from unsigned int to
17077 unsigned HOST_WIDE_INT. Perform negation in unsigned HOST_WIDE_INT
17080 2018-04-17 Jim Wilson <jimw@sifive.com>
17083 * config/riscv/riscv.c (riscv_compute_frame_info): Add calls to
17084 RISCV_STACK_ALIGN when using outgoing_args_size and pretend_args_size.
17085 Set arg_pointer_offset after using pretend_args_size.
17087 2018-04-17 Jakub Jelinek <jakub@redhat.com>
17089 PR rtl-optimization/85431
17090 * dse.c (record_store): Ignore zero width stores.
17093 * asan.c (handle_builtin_stack_restore): Adjust comment. Emit
17094 __asan_allocas_unpoison call and last_alloca_addr = new_sp before
17095 __builtin_stack_restore rather than after it.
17096 * builtins.c (expand_asan_emit_allocas_unpoison): Pass
17097 arg1 + (virtual_dynamic_stack_rtx - stack_pointer_rtx) as second
17098 argument instead of virtual_dynamic_stack_rtx.
17100 2018-04-17 Kelvin Nilsen <kelvin@gcc.gnu.org>
17102 * config/rs6000/rs6000-protos.h (rs6000_builtin_is_supported_p):
17104 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
17105 Add note to error message to explain internal mapping of overloaded
17106 built-in function name to non-overloaded built-in function name.
17107 * config/rs6000/rs6000.c (rs6000_builtin_is_supported_p): New
17110 2018-04-17 Michael Meissner <meissner@linux.vnet.ibm.com>
17113 * config/rs6000/rs6000.md (pack<mode>): Do not try handle a pack
17114 where the inputs overlap with the output.
17116 2018-04-17 Jakub Jelinek <jakub@redhat.com>
17118 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): Add
17119 (=v, v) alternative and explicit "memory" attribute.
17120 (vec_extract_lo_<mode><mask_name>): Likewise. Also add
17121 "type", "prefix", "prefix_extra", "length_immediate" and "mode"
17123 (vec_extract_lo_<mode><mask_name>): Add (=v, v) alternative and use
17124 "sselog1" type instead of "sselog".
17125 (vec_extract_hi_<mode><mask_name>): Use "sselog1" type instead of
17126 "sselog". Remove explicit "memory" attribute.
17127 (vec_extract_lo_v32hi): Add (=v, v) alternative and explicit "memory",
17128 "type", "prefix", "prefix_extra", "length_immediate" and "mode"
17130 (vec_extract_hi_v32hi): Merge all alternatives into one, use
17131 "sselog1" type instead of "sselog". Remove explicit "memory"
17133 (vec_extract_hi_v16hi): Merge each pair of alternatives into one,
17134 use "sselog1" type instead of "sselog". Remove explicit "memory"
17136 (vec_extract_lo_v64qi): Add (=v, v) alternative and explicit "memory",
17137 "type", "prefix", "prefix_extra", "length_immediate" and "mode"
17139 (vec_extract_hi_v64qi): Merge all alternatives into one, use
17140 "sselog1" type instead of "sselog". Remove explicit "memory"
17142 (vec_extract_hi_v32qi): Merge each pair of alternatives into one,
17143 use "sselog1" type instead of "sselog". Remove explicit "memory"
17147 * config/i386/i386.md (*ashlqi3_1_slp): Use alu1 type instead of alu.
17149 PR middle-end/85414
17150 * rtlhooks.c (gen_lowpart_if_possible): Don't call gen_lowpart_SUBREG
17153 2018-04-17 Martin Jambor <mjambor@suse.cz>
17156 * ipa-cp.c (create_specialized_node): Call
17157 expand_all_artificial_thunks if necessary.
17159 2018-04-17 Martin Liska <mliska@suse.cz>
17162 * ipa-devirt.c (odr_types_equivalent_p): Remove trailing
17163 in message, remote space in between '_G' and '('.
17165 2018-04-17 Jakub Jelinek <jakub@redhat.com>
17168 * config/i386/sse.md (reduces<mode><mask_scalar_name>,
17169 avx512f_vmcmp<mode>3<round_saeonly_name>,
17170 avx512f_vmcmp<mode>3_mask<round_saeonly_name>,
17171 avx512f_sgetexp<mode><mask_scalar_name><round_saeonly_scalar_name>,
17172 avx512f_rndscale<mode><round_saeonly_name>,
17173 avx512dq_ranges<mode><mask_scalar_name><round_saeonly_scalar_name>,
17174 avx512f_vgetmant<mode><mask_scalar_name><round_saeonly_scalar_name>):
17175 Use %<iptr>2 instead of %2 for -masm=intel.
17176 (avx512f_vcvtss2usi<round_name>, avx512f_vcvtss2usiq<round_name>,
17177 avx512f_vcvttss2usi<round_saeonly_name>,
17178 avx512f_vcvttss2usiq<round_saeonly_name>): Use %k1 instead of %1 for
17180 (avx512f_vcvtsd2usi<round_name>, avx512f_vcvtsd2usiq<round_name>,
17181 avx512f_vcvttsd2usi<round_saeonly_name>,
17182 avx512f_vcvttsd2usiq<round_saeonly_name>, ufloatv2siv2df2<mask_name>):
17183 Use %q1 instead of %1 for -masm=intel.
17184 (avx512f_sfixupimm<mode><sd_maskz_name><round_saeonly_name>,
17185 avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Use %<iptr>3 instead
17186 of %3 for -masm=intel.
17187 (sse2_shufpd_v2df_mask): Fix a typo, change %{6%} to %{%6%} for
17189 (*avx512vl_<code>v2div2qi2_store): Use %w0 instead of %0 for
17191 (*avx512vl_<code><mode>v4qi2_store): Use %k0 instead of %0 for
17193 (avx512vl_<code><mode>v4qi2_mask_store): Use a single pattern with
17194 %k0 and %1 for -masm=intel rather than two patterns, one with %0 and
17196 (*avx512vl_<code><mode>v8qi2_store): Use %q0 instead of %0 for
17198 (avx512vl_<code><mode>v8qi2_mask_store): Use a single pattern with
17199 %q0 and %1 for -masm=intel rather than two patterns, one with %0 and
17200 %g1 and one with %0 and %1.
17201 (avx512er_vmrcp28<mode><round_saeonly_name>,
17202 avx512er_vmrsqrt28<mode><round_saeonly_name>): Use %<iptr>1 instead of
17203 %1 for -masm=intel.
17204 (avx5124fmaddps_4fmaddps_mask, avx5124fmaddps_4fmaddss_mask,
17205 avx5124fmaddps_4fnmaddps_mask, avx5124fmaddps_4fnmaddss_mask,
17206 avx5124vnniw_vp4dpwssd_mask, avx5124vnniw_vp4dpwssds_mask): Swap order
17207 of %0 and %{%4%} for -masm=intel.
17208 (avx5124fmaddps_4fmaddps_maskz, avx5124fmaddps_4fmaddss_maskz,
17209 avx5124fmaddps_4fnmaddps_maskz, avx5124fmaddps_4fnmaddss_maskz,
17210 avx5124vnniw_vp4dpwssd_maskz, avx5124vnniw_vp4dpwssds_maskz): Swap
17211 order of %0 and %{%5%}%{z%} for -masm=intel.
17213 2018-04-17 Jan Hubicka <jh@suse.cz>
17216 * ipa-devirt.c (odr_types_equivalent_p): Handle bit fields.
17218 2018-04-17 Martin Liska <mliska@suse.cz>
17221 * multiple_target.c (create_dispatcher_calls): Set apostrophes
17222 for target_clone error message. Make default implementation
17223 clone to be a local declaration.
17224 (separate_attrs): Add new argument and check for an empty
17226 (expand_target_clones): Handle it.
17227 (ipa_target_clone): Make redirection just for target_clones
17230 2018-04-16 Cesar Philippidis <cesar@codesourcery.com>
17231 Tom de Vries <tom@codesourcery.com>
17233 PR middle-end/84955
17234 * omp-expand.c (expand_oacc_for): Add dummy false branch for
17235 tiled basic blocks without omp continue statements.
17237 2018-04-16 Aaron Sawdey <acsawdey@linux.ibm.com>
17240 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Mark
17241 vec_extract expression as having side effects to make sure it gets
17244 2018-04-16 H.J. Lu <hongjiu.lu@intel.com>
17247 * config/i386/i386.c (get_builtin_code_for_version): Check
17250 2018-04-16 Olga Makhotina <olga.makhotina@intel.com>
17253 * config.gcc: Support "skylake".
17254 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
17256 * config/i386/i386.c (m_SKYLAKE): Define.
17257 (processor_target_table): Add "skylake".
17258 (ix86_option_override_internal): Add "skylake".
17259 (get_builtin_code_for_version): Handle PROCESSOR_SKYLAKE,
17260 PROCESSOR_CANNONLAKE.
17261 (get_builtin_code_for_version): Fix priority for
17262 PROCESSOR_ICELAKE_CLIENT, PROCESSOR_ICELAKE_SERVER,
17263 PROCESSOR_SKYLAKE-AVX512.
17264 * config/i386/i386.h (processor_costs): Define TARGET_SKYLAKE.
17265 (processor_type): Add PROCESSOR_SKYLAKE.
17267 2018-04-16 Paolo Carlini <paolo.carlini@oracle.com>
17268 Jason Merrill <jason@redhat.com>
17271 * convert.c (convert_to_integer_1): Use direct recursion for
17272 enumeral types and types with a precision less than the number
17273 of bits in their mode.
17275 2018-04-16 Julia Koval <julia.koval@intel.com>
17278 * config/i386/x86-tune.def (X86_TUNE_SSE_UNALIGNED_LOAD_OPTIMAL,
17279 X86_TUNE_SSE_UNALIGNED_STORE_OPTIMAL): Add m_SKYLAKE_AVX512
17281 2018-04-14 Segher Boessenkool <segher@kernel.crashing.org>
17284 * config/rs6000/rs6000.opt (mdirect-move): Make deprecated.
17285 * doc/invoke.texi (RS/6000 and PowerPC Options): Remove -mdirect-move
17286 and -mno-direct-move.
17288 2018-04-13 Paul A. Clarke <pc@us.ibm.com>
17291 * config/rs6000/emmintrin.h (_mm_slli_epi{16,32,64}):
17292 Ensure that vec_splat_s32 is only called with 0 <= shift < 16.
17293 Ensure negative shifts result in {0}.
17295 2018-04-13 Vladimir Makarov <vmakarov@redhat.com>
17297 PR rtl-optimization/79916
17298 * config/rs6000/rs6000.c (rs6000_emit_move): Use assigned hard
17299 regs (if any) to define how to gnerate SD moves when LRA is in
17302 2018-04-13 Jakub Jelinek <jakub@redhat.com>
17304 PR rtl-optimization/85393
17305 * except.h (expand_dw2_landing_pad_for_region): Remove declaration.
17306 * except.c (expand_dw2_landing_pad_for_region): Make static.
17307 * bb-reorder.c (fix_up_crossing_landing_pad): In new_bb emit just
17308 a label and unconditional jump to old_bb, rather than
17309 expand_dw2_landing_pad_for_region insn(s) and jump to single_succ
17312 PR rtl-optimization/85376
17313 * simplify-rtx.c (simplify_const_unary_operation): For CLZ and CTZ and
17314 zero op0, if C?Z_DEFINED_VALUE_AT_ZERO is false, return NULL_RTX
17315 instead of a specific value.
17317 2018-04-13 Jan Hubicka <hubicka@ucw.cz>
17318 Bin Cheng <bin.cheng@arm.com>
17320 PR tree-optimization/82965
17321 PR tree-optimization/83991
17322 * cfgloopanal.c (expected_loop_iterations_unbounded): Add
17323 by_profile_only parameter.
17324 * cfgloopmanip.c (scale_loop_profile): Further scale loop's profile
17325 information if the loop was predicted to iterate too many times.
17326 * cfgloop.h (expected_loop_iterations_unbounded): Update prototype
17328 2018-04-13 Jan Hubicka <hubicka@ucw.cz>
17331 * config/i386/i386.c (ix86_can_inline_p): Allow safe transitions for
17334 2018-04-13 Martin Liska <mliska@suse.cz>
17335 Jakub Jelinek <jakub@redhat.com>
17337 PR middle-end/81657
17338 * expr.h (enum block_op_methods): Add BLOCK_OP_NO_LIBCALL_RET.
17339 * expr.c (emit_block_move_hints): Handle BLOCK_OP_NO_LIBCALL_RET.
17340 * builtins.c (expand_builtin_memory_copy_args): Use
17341 BLOCK_OP_NO_LIBCALL_RET method for mempcpy with non-ignored target,
17342 handle dest_addr == pc_rtx.
17344 2018-04-12 Segher Boessenkool <segher@kernel.crashing.org>
17347 * config/rs6000/rs6000.md (fix_trunc<mode>si2): Use legacy code if
17348 asked to not generate direct moves.
17349 (fix_trunc<mode>si2_stfiwx): Similar.
17350 (fix_trunc<mode>si2_internal): Similar.
17352 2018-04-12 Jakub Jelinek <jakub@redhat.com>
17355 * var-tracking.c (add_stores): Handle STRICT_LOW_PART SET_DEST.
17356 * cselib.c (cselib_record_sets): For STRICT_LOW_PART dest,
17357 lookup if dest in some wider mode is known to be const0_rtx and
17358 if so, record permanent equivalence for it to be ZERO_EXTEND of
17359 the narrower mode destination.
17361 2018-04-12 Cesar Philippidis <cesar@codesourcery.com>
17363 * lto-streamer-out.c (output_function): Revert 259346.
17364 * omp-expand.c (expand_oacc_for): Likewise.
17366 2018-04-12 Alexander Monakov <amonakov@ispras.ru>
17368 PR rtl-optimization/85354
17369 * sel-sched-ir.c (sel_init_pipelining): Move cfg_cleanup call...
17370 * sel-sched.c (sel_global_init): ... here.
17372 2018-04-12 Eric Botcazou <ebotcazou@adacore.com>
17375 * lto-wrapper.c (debug_objcopy): Open the files in binary mode.
17376 * dwarf2out.c (dwarf2out_early_finish): Do not generate assembly in LTO
17377 mode for PE-COFF targets.
17378 * config/i386/i386-protos.h (i386_pe_asm_lto_start): Declare.
17379 (i386_pe_asm_lto_end): Likewise.
17380 * config/i386/cygming.h (TARGET_ASM_LTO_START): Define.
17381 (TARGET_ASM_LTO_END): Likewise.
17382 * config/i386/winnt.c (saved_debug_info_level): New static variable.
17383 (i386_pe_asm_lto_start): New function.
17384 (i386_pe_asm_lto_end): Likewise.
17386 2018-04-12 Cesar Philippidis <cesar@codesourcery.com>
17387 Richard Biener <rguenther@suse.de>
17389 PR middle-end/84955
17390 * lto-streamer-out.c (output_function): Fix CFG loop state before
17392 * omp-expand.c (expand_oacc_for): Handle calls to internal
17393 functions like regular functions.
17395 2018-04-12 Richard Biener <rguenther@suse.de>
17398 * dwarf2out.c (init_sections_and_labels): Use debug_line_section[_label]
17399 for the early LTO debug to properly generate references to it
17400 during DIE emission. Do not re-use that for the skeleton for
17402 (dwarf2out_early_finish): Likewise.
17404 2018-04-12 Jakub Jelinek <jakub@redhat.com>
17407 * config/i386/sse.md
17408 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name> split,
17409 <mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name> split,
17410 vec_extract_lo_<mode><mask_name> split, vec_extract_lo_v32hi,
17411 vec_extract_lo_v64qi): For non-AVX512VL if input is xmm16+ reg
17412 and output is a reg, avoid creating invalid lowpart subreg, but
17413 instead split into a 512-bit move. Don't split if not AVX512VL,
17414 input is xmm16+ reg and output is a mem.
17415 (vec_extract_lo_<mode><mask_name>, vec_extract_lo_v32hi,
17416 vec_extract_lo_v64qi): Don't require split if not AVX512VL, input is
17417 xmm16+ reg and output is a mem.
17419 2018-04-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17421 * config/s390/s390.c (s390_output_indirect_thunk_function): Check
17422 also for flag_dwarf2_cfi_asm.
17424 2018-04-12 Jakub Jelinek <jakub@redhat.com>
17426 PR rtl-optimization/85342
17427 * regcprop.c (copyprop_hardreg_forward_1): Remove replaced array, use
17428 a bool scalar var inside of the loop instead. Don't try to update
17429 recog_data.operand after failed apply_change_group.
17431 2018-04-12 Tom de Vries <tom@codesourcery.com>
17434 * config/nvptx/nvptx.c (flexible_array_member_type_p): New function.
17435 (nvptx_assemble_decl_begin): Add undefined param. Declare undefined
17436 array with flexible array member as array without given dimension.
17437 (nvptx_assemble_undefined_decl): Set nvptx_assemble_decl_begin call
17438 argument for undefined param to true.
17440 2018-04-11 Aaron Sawdey <acsawdey@linux.ibm.com>
17443 * doc/invoke.texi (RS/6000 and PowerPC Options): Document options
17444 -mcall- and -mtraceback=. Remove options -mabi=spe and -mabi=no-spe
17445 from PowerPC section.
17446 * config/rs6000/sysv4.opt (mcall-): Improve help text.
17447 * config/rs6000/rs6000.opt (mblock-compare-inline-limit=): Trim
17448 help text that is too long.
17449 * config/rs6000/rs6000.opt (mblock-compare-inline-loop-limit=): Trim
17450 help text that is too long.
17451 * config/rs6000/rs6000.opt (mstring-compare-inline-limit=): Trim
17452 help text that is too long.
17454 2018-04-11 Uros Bizjak <ubizjak@gmail.com>
17456 * config/alpha/alpha.md (stack_probe_internal): Rename
17457 from "probe_stack". Update all callers.
17459 2018-04-11 Alexander Monakov <amonakov@ispras.ru>
17461 PR rtl-optimization/84566
17462 * sched-deps.c (sched_analyze_insn): Check deps->readonly when invoking
17463 sched_macro_fuse_insns.
17465 2018-04-11 Alexander Monakov <amonakov@ispras.ru>
17468 * sched-rgn.c (add_branch_dependences): Move sel_sched_p check here...
17469 (compute_block_dependences): ... from here.
17471 2018-04-11 Jakub Jelinek <jakub@redhat.com>
17473 PR tree-optimization/85331
17474 * vec-perm-indices.h (vec_perm_indices::clamp): Change input type
17475 from int to HOST_WIDE_INT.
17477 2018-04-11 Martin Jambor <mjambor@suse.cz>
17480 * ipa-cp.c (propagate_vals_across_pass_through): Expand comment.
17481 (cgraph_edge_brings_value_p): New parameter dest_val, check if it is
17482 not the same as the source val.
17483 (cgraph_edge_brings_value_p): New parameter.
17484 (gather_edges_for_value): Pass destination value to
17485 cgraph_edge_brings_value_p.
17486 (perhaps_add_new_callers): Likewise.
17487 (get_info_about_necessary_edges): Likewise and exclude values brought
17488 only by self-recursive edges.
17489 (create_specialized_node): Redirect only clones of self-calling edges.
17490 (+self_recursive_pass_through_p): New function.
17491 (find_more_scalar_values_for_callers_subset): Use it.
17492 (find_aggregate_values_for_callers_subset): Likewise.
17493 (known_aggs_to_agg_replacement_list): Removed.
17494 (decide_whether_version_node): Re-calculate known constants for all
17495 remaining context clones.
17497 2018-04-11 Richard Biener <rguenther@suse.de>
17500 * dwarf2out.c (dwarf2out_finish): Remove DW_AT_stmt_list attribute
17501 from early DWARF output.
17502 (dwarf2out_early_finish): Output line info unconditionally into
17503 early DWARF and add reference to it.
17505 2018-04-11 Jakub Jelinek <jakub@redhat.com>
17508 * config/i386/sse.md (iptr): Add V16SFmode and V8DFmode cases.
17509 (<avx512>_vec_dup<mode><mask_name>): Use a single pattern for modes
17510 other than V2DFmode using iptr mode attribute.
17511 (<avx512>_vec_dup<mode><mask_name>): Use iptr mode attribute.
17513 2018-04-11 Alexander Monakov <amonakov@ispras.ru>
17515 PR rtl-optimization/84659
17516 * sel-sched-ir.c (sel_init_pipelining): Invoke cleanup_cfg.
17518 2018-04-11 Jakub Jelinek <jakub@redhat.com>
17521 * dwarf2out.c (skip_loc_list_entry): Don't call size_of_locs if
17523 (output_loc_list): Pass address of a dummy size variable even in the
17524 locview handling loop.
17525 (index_location_lists): Add comment on why skip_loc_list_entry can't
17528 2018-04-11 Thomas Preud'homme <thomas.preudhomme@arm.com>
17531 * config/arm/arm-builtins.c (arm_expand_builtin): Force input operand
17534 2018-04-10 Aaron Sawdey <acsawdey@linux.ibm.com>
17537 * doc/invoke.texi (RS/6000 and PowerPC Options): Document options
17538 -mblock-compare-inline-limit, -mblock-compare-inline-loop-limit,
17539 and -mstring-compare-inline-limit.
17541 2018-04-10 Segher Boessenkool <segher@kernel.crashing.org>
17544 * config/rs6000/rs6000.md (allocate_stack): Put the residual size
17545 for stack clash protection in a register whenever we need it to be in
17548 2018-04-10 Segher Boessenkool <segher@kernel.crashing.org>
17550 * common/config/rs6000/rs6000-common.c (rs6000_option_init_struct):
17551 Enable -fasynchronous-unwind-tables by default if OBJECT_FORMAT_ELF.
17553 2018-04-10 Segher Boessenkool <segher@kernel.crashing.org>
17556 * config/rs6000/rs6000.opt (mtraceback=): Show the allowed values in
17558 (mlong-double-): Ditto.
17559 * config/rs6000/sysv4.opt (msdata=): Ditto.
17560 (mtls-size=): Ditto.
17562 2018-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
17564 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
17565 erroneous entries for
17566 "vector int vec_ldl (int, long int *)", and
17567 "vector unsigned int vec_ldl (int, unsigned long int *)".
17568 Add comments and entries for
17569 "vector bool char vec_ldl (int, bool char *)",
17570 "vector bool short vec_ldl (int, bool short *)",
17571 "vector bool int vec_ldl (int, bool int *)",
17572 "vector bool long long vec_ldl (int, bool long long *)",
17573 "vector pixel vec_ldl (int, pixel *)",
17574 "vector long long vec_ldl (int, long long *)",
17575 "vector unsigned long long vec_ldl (int, unsigned long long *)".
17576 * config/rs6000/rs6000.c (rs6000_init_builtins): Initialize new
17577 type tree bool_long_long_type_node and correct definition of
17578 bool_V2DI_type_node to make reference to this new type tree.
17579 (rs6000_mangle_type): Replace erroneous reference to
17580 bool_long_type_node with bool_long_long_type_node.
17581 * config/rs6000/rs6000.h (enum rs6000_builtin_type_index): Add
17582 comments to emphasize sign distinctions for char and int types and
17583 replace RS6000_BTI_bool_long constant with
17584 RS6000_BTI_bool_long_long constant. Also add comment to restrict
17585 use of RS6000_BTI_pixel.
17586 (bool_long_type_node): Remove this macro definition.
17587 (bool_long_long_type_node): New macro definition
17589 2018-04-10 Jakub Jelinek <jakub@redhat.com>
17591 PR rtl-optimization/85300
17592 * combine.c (subst): Handle subst of CONST_SCALAR_INT_P new_rtx also
17593 into FLOAT and UNSIGNED_FLOAT like ZERO_EXTEND, return a CLOBBER if
17594 simplify_unary_operation fails.
17596 2018-04-10 Martin Liska <mliska@suse.cz>
17598 * gdbhooks.py: Add pretty-printers for varpool_node, symtab_node,
17599 cgraph_edge and ipa_ref.
17601 2018-04-10 Jakub Jelinek <jakub@redhat.com>
17605 * config/i386/sse.md
17606 (<extract_type>_vinsert<shuffletype><extract_suf>_mask): Fix
17607 computation of the VEC_MERGE selector from mask.
17608 (<extract_type>_vinsert<shuffletype><extract_suf>_1<mask_name>):
17609 Fix decoding of the VEC_MERGE selector into mask.
17611 2018-04-10 Richard Sandiford <richard.sandiford@linaro.org>
17613 PR tree-optimization/85286
17614 * tree-vect-data-refs.c (vect_get_smallest_scalar_type):
17616 2018-04-10 Richard Sandiford <richard.sandiford@linaro.org>
17618 * final.c (final_1): Set insn_last_address as well as
17619 insn_current_address.
17621 2018-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17624 * explow.c (emit_stack_probe): Call validize_mem on memory location
17625 before passing it to gen_probe_stack. Create address operand and
17626 legitimize it for the probe_stack_address case.
17628 2018-04-09 Jan Hubicka <jh@suse.cz>
17631 * ipa-devirt.c (rebuild_type_inheritance-hash): New.
17632 * ipa-utils.h (rebuild_type_inheritance-hash): Declare.
17633 * tree.c (free_lang_data_in_type): Fix handling of binfos;
17635 (free_lang_data): Rebuild type inheritance graph.
17637 2018-04-09 Martin Sebor <msebor@redhat.com>
17639 * invoke.texi (-finline-small-functions): Mention other optimization
17641 (-findirect-inlining, -fpartial-inlining): Same.
17642 (-finline-functions-called-once): Same.
17643 (-freorder-blocks-and-partition): Same.
17645 2018-04-09 Jan Hubicka <jh@suse.cz>
17648 * cfgcleanup.c (try_forward_edges): Do not give up on crossing
17649 jumps; choose last target that matches the criteria (i.e.
17650 no partition changes for non-crossing jumps).
17651 * cfgrtl.c (cfg_layout_redirect_edge_and_branch): Add basic
17652 support for redirecting crossing jumps to non-crossing.
17654 2018-04-09 Alexey Brodkin <abrodkin@synopsys.com>
17656 * config/arc/arc.c (arc_expand_prologue): Set stack usage info
17657 also for naked functions.
17659 2018-04-09 Claudiu Zissulescu <claziss@synopsys.com>
17661 * config/arc/arc.md (add_shift): New pattern.
17662 (add_shift2): Likewise.
17663 (sub_shift): Likewise.
17664 (sub_shift_cmp0_noout): Likewise.
17665 (compare_si_ashiftsi): Likewise.
17666 (xbfu_cmp0_noout): New combine pattern.
17667 (xbfu_cmp0"): Likewise.
17668 (movsi_set_cc_insn): Place the predicable variant first.
17669 (commutative_binary_cmp0_noout): Remove clobber.
17670 (commutative_binary_cmp0): New pattern.
17671 (noncommutative_binary_cmp0): Likewise.
17672 (noncommutative_binary_cmp0_noout): Likewise.
17673 (noncommutative_binary_comparison_result_used): Removed.
17674 (rsub_cmp0): New pattern.
17675 (rsub_cmp0_noout): Likewise.
17676 (extzvsi): Changed, keep only meaningful variants.
17677 (SQH, SEZ): New iterators.
17678 (SQH_postfix): New mode attribute.
17679 (SEZ_prefix): New code attribute.
17680 (<SEZ_prefix>xt<SQH_postfix>_cmp0_noout): New instruction pattern.
17681 (<SEZ_prefix>xt<SQH_postfix>_cmp0): Likewise.
17682 * config/arc/predicates.md (cc_set_register): Use CC_REG instead
17683 of numerical value.
17684 (noncommutative_operator): Check the availability of barrel
17687 2018-04-09 Richard Biener <rguenther@suse.de>
17689 PR tree-optimization/85284
17690 * tree-ssa-loop-niter.c (number_of_iterations_exit_assumptions):
17691 Only use the niter constraining form of simple_iv when the exit
17692 is always executed.
17694 2018-04-09 Tom de Vries <tom@codesourcery.com>
17697 * config/nvptx/nvptx.md (define_c_enum "unspecv"): Add UNSPECV_MEMBAR.
17698 (define_expand "*memory_barrier"): New define_expand.
17699 (define_insn "memory_barrier"): New insn.
17701 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
17703 PR rtl-optimization/80463
17704 PR rtl-optimization/83972
17705 PR rtl-optimization/83480
17707 * sel-sched-ir.c (has_dependence_note_mem_dep): Take into account the
17708 correct producer for the insn.
17709 (tidy_control_flow): Fixup seqnos in case of debug insns.
17711 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
17713 PR rtl-optimization/83913
17715 * sel-sched-ir.c (merge_expr_data): Choose the middle between two
17716 different sched-times when merging exprs.
17718 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
17720 PR rtl-optimization/83962
17722 * sel-sched-ir.c (tidy_control_flow): Correct the order in which we call
17723 tidy_fallthru_edge and tidy_control_flow.
17725 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
17727 PR rtl-optimization/83530
17729 * sel-sched.c (force_next_insn): New global variable.
17730 (remove_insn_for_debug): When force_next_insn is true, also leave only
17731 next insn in the ready list.
17732 (sel_sched_region): When the region wasn't scheduled, make another pass
17733 over it with force_next_insn set to 1.
17735 2018-04-08 Monk Chiang <sh.chiang04@gmail.com>
17737 * config.gcc (nds32le-*-*, nds32be-*-*): Add nds32/nds32_intrinsic.h
17739 * config/nds32/constants.md (unspec_volatile_element): Add enum values
17740 for interrupt control.
17741 * config/nds32/nds32-intrinsic.c: Implementation of intrinsic
17742 functions for interrupt control.
17743 * config/nds32/nds32-intrinsic.md: Likewise.
17744 * config/nds32/nds32_intrinsic.h: Likewise.
17745 * config/nds32/nds32.h (nds32_builtins): Likewise.
17747 2018-04-08 Chung-Ju Wu <jasonwucj@gmail.com>
17749 * config/nds32/nds32.c (nds32_init_machine_status,
17750 nds32_legitimate_index_p, nds32_legitimate_address_p): Consider
17751 strict_aligned_p field.
17752 (nds32_expand_to_rtl_hook): New function.
17753 (TARGET_EXPAND_TO_RTL_HOOK): Define.
17754 * config/nds32/nds32.h (machine_function): Add strict_aligned_p field.
17756 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
17757 Chung-Ju Wu <jasonwucj@gmail.com>
17759 * config.gcc (nds32*-*-*): Check that n7 is valid to --with-cpu.
17760 * config/nds32/nds32-n7.md: New file.
17761 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N7.
17762 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n7
17764 * config/nds32/nds32-protos.h: More declarations for n7 pipeline.
17765 * config/nds32/nds32.md (pipeline_model): Add n7.
17766 * config/nds32/nds32.opt (mcpu): Support n7 pipeline cpus.
17767 * config/nds32/pipelines.md: Include n7 settings.
17769 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
17770 Chung-Ju Wu <jasonwucj@gmail.com>
17772 * config.gcc (nds32*-*-*): Check that e8 is valid to --with-cpu.
17773 * config/nds32/nds32-e8.md: New file.
17774 * config/nds32/nds32-opts.h (nds32-cpu_type): Add CPU_E8.
17775 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for e8
17777 * config/nds32/nds32-protos.h: More declarations for e8 pipeline.
17778 * config/nds32/nds32.md (pipeline_model): Add e8.
17779 * config/nds32/nds32.opt (mcpu): Support e8 pipeline cpus.
17780 * config/nds32/pipelines.md: Include e8 settings.
17782 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
17783 Chung-Ju Wu <jasonwucj@gmail.com>
17785 * config.gcc (nds32*-*-*): Check that n6/n8/s8 are valid to --with-cpu.
17786 * config/nds32/nds32-n8.md: New file.
17787 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N6 and CPU_N8.
17788 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n8
17790 * config/nds32/nds32-protos.h: More declarations for n8 pipeline.
17791 * config/nds32/nds32-utils.c: More implementations for n8 pipeline.
17792 * config/nds32/nds32.md (pipeline_model): Add n8.
17793 * config/nds32/nds32.opt (mcpu): Support n8 pipeline cpus.
17794 * config/nds32/pipelines.md: Include n8 settings.
17796 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
17797 Chung-Ju Wu <jasonwucj@gmail.com>
17799 * config.gcc (nds32*): Add nds32-utils.o into extra_objs.
17800 * config/nds32/nds32-n9-2r1w.md: New file.
17801 * config/nds32/nds32-n9-3r2w.md: New file.
17802 * config/nds32/nds32-opts.h (nds32_cpu_type, nds32_mul_type,
17803 nds32_register_ports): New or modify for cpu n9.
17804 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n9
17806 * config/nds32/nds32-protos.h: More declarations for n9 pipeline.
17807 * config/nds32/nds32-utils.c: New file.
17808 * config/nds32/nds32.h (TARGET_PIPELINE_N9, TARGET_PIPELINE_SIMPLE,
17809 TARGET_MUL_SLOW): Define.
17810 * config/nds32/nds32.md (pipeline_model): New attribute.
17811 * config/nds32/nds32.opt (mcpu, mconfig-mul, mconfig-register-ports):
17812 New options that support cpu n9.
17813 * config/nds32/pipelines.md: Include n9 settings.
17814 * config/nds32/t-nds32 (nds32-utils.o): Add dependency.
17816 2018-04-08 Chung-Ju Wu <jasonwucj@gmail.com>
17818 * config/nds32/nds32-md-auxiliary.c (output_cond_branch): Output align
17819 information if necessary.
17820 (output_cond_branch_compare_zero): Likewise.
17821 * config/nds32/nds32.c (nds32_adjust_insn_length): Consider align case.
17822 (nds32_target_alignment): Refine for alignment.
17823 * config/nds32/nds32.h (NDS32_ALIGN_P): Define.
17824 (FUNCTION_BOUNDARY): Modify.
17825 * config/nds32/nds32.md (call_internal, call_value_internal): Consider
17827 * config/nds32/nds32.opt (malways-align, malign-functions): New.
17829 2018-04-08 Monk Chiang <sh.chiang04@gmail.com>
17831 * config/nds32/constants.md (unspec_volatile_element): Add values for
17832 TLB operation and data prefetch.
17833 * config/nds32/nds32-intrinsic.c: Implementation of intrinsic
17834 functions for TLB operation and data prefetch.
17835 * config/nds32/nds32-intrinsic.md: Likewise.
17836 * config/nds32/nds32_intrinsic.h: Likewise.
17837 * config/nds32/nds32.c (nds32_dpref_names): Likewise.
17838 (nds32_print_operand): Likewise.
17839 * config/nds32/nds32.h (nds32_builtins): Likewise.
17841 2018-04-07 Thomas Koenig <tkoenig@gcc.gnu.org>
17842 Andrew Pinski <pinsika@gcc.gnu.org>
17844 PR middle-end/82976
17845 * match.pd: Use constant_boolean_node of correct type instead of
17846 boolean_true_node or boolean_false_node for simplifying
17847 pointer comparisons to zero.
17849 2018-04-07 Jakub Jelinek <jakub@redhat.com>
17851 PR tree-optimization/80021
17852 * tree.c (verify_type_variant): Make error call in verify_variant_match
17853 translatable and remove final full stop.
17855 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
17857 * config/nds32/constants.md (unspec_volatile_element): Add
17858 UNSPEC_VOLATILE_EH_RETURN.
17859 * config/nds32/nds32-md-auxiliary.c (nds32_output_stack_push,
17860 nds32_output_stack_pop): Support dwarf exception handling process.
17861 * config/nds32/nds32-protos.h (nds32_dynamic_chain_address): Declare.
17862 * config/nds32/nds32.c (nds32_init_machine_status): Support dwarf
17863 exception handling process.
17864 (nds32_compute_stack_frame): Likewise.
17865 (nds32_return_addr_rtx): Likewise.
17866 (nds32_initial_elimination_offset): Likewise.
17867 (nds32_expand_prologue): Likewise.
17868 (nds32_expand_epilogue): Likewise.
17869 (nds32_dynamic_chain_address): New function.
17870 * config/nds32/nds32.h (machine_function): Add fields for dwarf
17871 exception handling.
17872 (DYNAMIC_CHAIN_ADDRESS): Define.
17873 (EH_RETURN_DATA_REGNO): Define.
17874 (EH_RETURN_STACKADJ_RTX): Define.
17875 * config/nds32/nds32.md (eh_return, nds32_eh_return): Implement
17876 patterns for dwarf exception handling.
17878 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
17880 * config/nds32/nds32.h: Clean up obsolete macros.
17882 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
17884 * config/nds32/constants.md (unspec_element, unspec_volatile_element):
17885 Add enum values for particular instructions.
17886 * config/nds32/nds32-intrinsic.c: Implementation of expanding
17887 particular intrinsic functions.
17888 * config/nds32/nds32-intrinsic.md: Likewise.
17889 * config/nds32/nds32_intrinsic.h: Likewise.
17890 * config/nds32/nds32.h (nds32_builtins): Likewise.
17891 * config/nds32/nds32.md (type): Add pbsad and pbsada.
17892 (btst, ave): New patterns for particular instructions.
17894 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
17896 * config/nds32/constants.md (unspec_element, unspec_volatile_element):
17897 Add enum values for atomic load/store and memory sync.
17898 * config/nds32/nds32-intrinsic.c: Implementation for atomic load/store
17900 * config/nds32/nds32-intrinsic.md: Likewise.
17901 * config/nds32/nds32_intrinsic.h: Likewise.
17902 * config/nds32/nds32.h (nds32_builtins): Likewise.
17904 2018-04-07 Jakub Jelinek <jakub@redhat.com>
17906 PR tree-optimization/85257
17907 * fold-const.c (native_encode_vector): If not all elts could fit
17908 and off is -1, return 0 rather than offset.
17909 * tree-ssa-sccvn.c (vn_reference_lookup_3): Pass
17910 (offseti - offset2) / BITS_PER_UNIT as 4th argument to
17911 native_encode_expr. Verify len * BITS_PER_UNIT >= maxsizei. Don't
17912 adjust buffer in native_interpret_expr call.
17914 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
17916 * config/nds32/constants.md (unspec_volatile_element): Add cache
17917 control enum values.
17918 * config/nds32/nds32-intrinsic.c: Add cache control expand functions.
17919 * config/nds32/nds32-intrinsic.md: Add cache control patterns.
17920 * config/nds32/nds32.c (nds32_cctl_names): New.
17921 (nds32_print_operand): Handle cache control register names.
17922 * config/nds32/nds32.h (nds32_builtins): New enum values.
17923 * config/nds32/nds32_intrinsic.h: Add cache control enum types and
17925 * config/nds32/nds32.md (type): Add mmu.
17926 * config/nds32/pipelines.md (simple_insn): Add mmu.
17928 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
17930 * config/nds32/nds32.md (type): Remove call.
17931 * config/nds32/pipelines.md (simple_insn): Likewise.
17933 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
17935 * config/nds32/constants.md (unspec_volatile_element): Add
17936 UNSPEC_VOLATILE_FMFCSR, UNSPEC_VOLATILE_FMTCSR and
17937 UNSPEC_VOLATILE_FMFCFG.
17938 * config/nds32/nds32-intrinsic.c (bdesc_noarg): New builtin
17939 description for fmfcfg and fmfcsr.
17940 (bdesc_1arg): Add fmtcsr.
17941 (bdesc_2arg): Add fcpynss, fcpyss, fcpynsd and fcpysd.
17942 (nds32_expand_builtin_impl): Deal with FPU intrinsic functions.
17943 * config/nds32/nds32-intrinsic.md (unspec_fcpynsd, unspec_fcpysd,
17944 unspec_fcpynss, unspec_fcpysd, unspec_fcpyss, unspec_fmfcsr,
17945 unspec_fmfcfg): New patterns.
17946 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_FMFCFG,
17947 NDS32_BUILTIN_FMFCSR, NDS32_BUILTIN_FMTCSR, NDS32_BUILTIN_FCPYNSS,
17948 NDS32_BUILTIN_FCPYSS,NDS32_BUILTIN_FCPYNSD and NDS32_BUILTIN_FCPYSD.
17949 * config/nds32/nds32_intrinsic.h (__nds32__fcpynsd, __nds32__fcpynss,
17950 __nds32__fcpysd, __nds32__fcpyss, __nds32__fmfcsr, __nds32__fmtcsr,
17951 __nds32__fmfcfg): Define.
17953 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
17955 * config/nds32/nds32.c (nds32_intrinsic_register_names): Add more
17956 intrinsic register names.
17957 * config/nds32/nds32_intrinsic.h (nds32_intrinsic_registers): Add more
17958 intrinsic register enum values and macros.
17960 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
17962 * config/nds32/nds32.c (nds32_legitimate_index_p): Modify condition
17963 for load/store addressing form.
17964 (nds32_print_operand_address): Likewise.
17966 2018-04-06 Eric Botcazou <ebotcazou@adacore.com>
17969 * config/sparc/sparc.c (sparc_expand_move): Deal with symbolic operands
17970 based on LABEL_REF. Remove useless assertion.
17971 (pic_address_needs_scratch): Fix formatting.
17972 (sparc_legitimize_pic_address): Minor tweaks.
17973 (sparc_delegitimize_address): Adjust assertion accordingly.
17974 * config/sparc/sparc.md (movsi_pic_label_ref): Change label_ref_operand
17975 into symbolic_operand.
17976 (movsi_high_pic_label_ref): Likewise.
17977 (movsi_lo_sum_pic_label_ref): Likewise.
17978 (movdi_pic_label_ref): Likewise.
17979 (movdi_high_pic_label_ref): Likewise.
17980 (movdi_lo_sum_pic_label_ref): Likewise.
17982 2018-04-06 Amaan Cheval <amaan.cheval@gmail.com>
17984 * config.gcc (x86_64-*-rtems*): Add rtems.h to tm_file for
17985 custom LIB_SPEC setup.
17987 2018-04-06 Ruslan Bukin <br@bsdpad.com>
17988 Kito Cheng <kito.cheng@gmail.com>
17990 * config.gcc (riscv*-*-freebsd*): Add RISC-V FreeBSD support.
17991 * config/riscv/freebsd.h: New.
17993 2018-04-06 Chung-Ju Wu <jasonwucj@gmail.com>
17995 * config/nds32/nds32.c (nds32_adjust_insn_length): Refine.
17996 * config/nds32/nds32.h (ADJUST_INSN_LENGTH): Change the location in
17999 2018-04-06 Chung-Ju Wu <jasonwucj@gmail.com>
18000 Kito Cheng <kito.cheng@gmail.com>
18002 * config/nds32/nds32-md-auxiliary.c (nds32_output_return,
18003 nds32_output_call, nds32_symbol_binds_local_p): New functions.
18004 * config/nds32/nds32-protos.h (nds32_output_call,
18005 nds32_output_return): Declare.
18006 * config/nds32/nds32.md: Refine all the call and return patterns.
18008 2018-04-06 Jakub Jelinek <jakub@redhat.com>
18011 * dwarf2out.c (rtl_for_decl_init): For STRING_CST initializer only
18012 build CONST_STRING if TYPE_MAX_VALUE is non-NULL and is INTEGER_CST.
18014 PR rtl-optimization/84872
18015 * cfgloopmanip.c (create_preheader): Use make_forwarder_block even if
18016 nentry == 1 when CP_FALLTHRU_PREHEADERS and single_entry is
18017 EDGE_CROSSING edge.
18019 2018-04-06 Tamar Christina <tamar.christina@arm.com>
18021 * expr.c (copy_blkmode_to_reg): Revert 254862.
18022 * doc/sourcebuild.texi (word_mode_no_slow_unalign): Likewise.
18024 2018-04-06 Richard Biener <rguenther@suse.de>
18026 PR middle-end/85244
18027 * tree-dfa.c (get_ref_base_and_extent): Reset seen_variable_array_ref
18028 after seeing a component reference with an adjacent field. Treat
18029 refs to arrays at struct end of external decls similar to
18030 refs to unconstrained commons.
18032 2018-04-06 Jakub Jelinek <jakub@redhat.com>
18035 * fold-const.c (twoval_comparison_p): Remove SAVE_P argument and don't
18036 look through SAVE_EXPRs with non-side-effects argument. Adjust
18038 (fold_comparison): Adjust twoval_comparison_p caller, don't handle
18041 2018-04-06 Richard Biener <rguenther@suse.de>
18043 PR middle-end/85180
18044 * alias.c (find_base_term): New wrapper around find_base_term
18045 unwinding CSELIB_VAL_PTR changes.
18046 (find_base_term): Do not restore CSELIB_VAL_PTR during the
18049 2018-04-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18051 * config/s390/s390.c (s390_z10_optimize_cmp): Expand dedicated NOP
18053 * config/s390/s390.md (UNSPECV_NOP_LR_0, UNSPECV_NOP_LR_1): New
18054 constant definitions.
18055 ("nop"): lr 0,0 -> nopr r0
18056 ("nop_lr0", "nop_lr1"): New insn definitions.
18058 2018-04-06 Chung-Ju Wu <jasonwucj@gmail.com>
18060 * config/nds32/nds32.md (*stack_push, *stack_pop): Use
18061 NDS32_V3PUSH_AVAILABLE_P macro.
18063 2018-04-06 Monk Chiang <sh.chiang04@gmail.com>
18064 Chung-Ju Wu <jasonwucj@gmail.com>
18066 * config.gcc (nds32*-*-*): Add v2j v3f v3s checking.
18067 (nds32*-*-*): Add float and fpu_config into supported_defaults.
18068 * common/config/nds32/nds32-common.c (TARGET_DEFAULT_TARGET_FLAGS):
18069 Include TARGET_DEFAULT_FPU_ISA and TARGET_DEFAULT_FPU_FMA.
18070 * config/nds32/constants.md (unspec_element): Add UNSPEC_COPYSIGN,
18071 UNSPEC_FCPYNSD, UNSPEC_FCPYNSS, UNSPEC_FCPYSD and UNSPEC_FCPYSS.
18072 * config/nds32/constraints.md: New constraints and checking for hard
18073 float configuration.
18074 * config/nds32/iterators.md: New mode iterator and attribute for hard
18075 float configuration.
18076 * config/nds32/nds32-doubleword.md: Use hard float alternatives and
18078 * config/nds32/nds32-fpu.md: New file.
18079 * config/nds32/nds32-md-auxiliary.c: New functions and checkings to
18080 deal with hard float code generation.
18081 * config/nds32/nds32-opts.h (nds32_arch_type): Add ARCH_V3F and
18083 (abi_type, float_reg_number): New enum type.
18084 * config/nds32/nds32-predicates.c: New predicates for hard float.
18085 * config/nds32/nds32-protos.h: Declare functions for hard float.
18086 * config/nds32/nds32.c: Implementation for hard float configuration.
18087 * config/nds32/nds32.h: Definitions for hard float configuration.
18088 * config/nds32/nds32.md: Include hard float machine description and
18089 modify patterns for hard float configuration.
18090 * config/nds32/nds32.opt: New options for hard float configuration.
18091 * config/nds32/predicates.md: New predicates for hard float
18094 2018-04-06 Kuan-Lin Chen <kuanlinchentw@gmail.com>
18096 * common/config/nds32/nds32-common.c
18097 (nds32_option_optimization_table): Enable -mreleax-hint by default.
18099 2018-04-05 Jakub Jelinek <jakub@redhat.com>
18101 PR middle-end/85195
18102 * match.pd (BIT_FIELD_REF CONSTRUCTOR@0 @1 @2): Use view_convert around
18103 CONSTRUCTOR_ELT (ctor, ...)->value.
18105 2018-04-05 Uros Bizjak <ubizjak@gmail.com>
18108 * config/i386/i386.md (define_attr "memory"): Handle rotate1 type.
18110 2018-04-05 Tom de Vries <tom@codesourcery.com>
18113 * config/nvptx/nvptx.c (nvptx_single): Fix neutering of bb with only
18116 2018-04-05 Shiva Chen <shiva0217@gmail.com>
18117 Kito Cheng <kito.cheng@gmail.com>
18119 * config/nds32/constraints.md (U33): Fine-tune checking condition.
18120 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Ditto.
18121 * config/nds32/nds32.h (nds32_16bit_address_type): Add
18122 ADDRESS_POST_MODIFY_LO_REG_IMM3U.
18124 2018-04-05 Shiva Chen <shiva0217@gmail.com>
18125 Kito Cheng <kito.cheng@gmail.com>
18127 * config/nds32/constraints.md (Ufe): New memory constraint.
18128 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format,
18129 nds32_output_16bit_load): Consider r8 register for lwi45.fe format.
18130 * config/nds32/nds32.c (nds32_print_operand): Output lwi45.fe
18132 * config/nds32/nds32.h (nds32_16bit_address_type): Add ADDRESS_R8_IMM7U.
18133 * config/nds32/nds32.md (*mov<mode>): Adjust pattern.
18135 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
18137 * config/nds32/nds32.md: Use optimize_size in the condition for
18138 alu-shift instructions.
18140 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
18142 * config/nds32/nds32.md (divsi4, udivsi4): New patterns.
18144 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
18146 * config/nds32/nds32.md (negsi2): Refine pattern.
18148 2018-04-05 Kito Cheng <kito.cheng@gmail.com>
18149 Chung-Ju Wu <jasonwucj@gmail.com>
18151 * config/nds32/iterators.md (shift_rotate): New code iterator.
18152 (shift): New code attribute.
18153 * config/nds32/nds32-md-auxiliary.c (nds32_expand_constant): New.
18154 * config/nds32/nds32-protos.h (nds32_expand_constant): Declare.
18155 * config/nds32/nds32.c (nds32_print_operand): Deal with more cases.
18156 * config/nds32/nds32.md (addsi3, *add_srli): Refine implementation for
18157 bit-wise operations.
18158 (andsi3, *andsi3): Ditto.
18159 (iorsi3, *iorsi3, *or_slli, *or_srli): Ditto.
18160 (xorsi3, *xorsi3, *xor_slli, *xor_srli): Ditto.
18161 (<shift>si3, *ashlsi3, *ashrsi3, *lshrsi3, *rotrsi3): Ditto.
18162 * config/nds32/predicates.md (nds32_rimm5u_operand, nds32_and_operand,
18163 nds32_ior_operand, nds32_xor_operand): New predicates.
18165 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
18167 * config/nds32/nds32.md (add<mode>3, sub<mode>3): Rename to ...
18168 (addsi3, subsi3): ... this.
18170 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
18172 * config/nds32/nds32.md (*sub_srli, *and_slli): Fine-tune predicator.
18174 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
18176 * config/nds32/nds32.md: Adjust indention.
18178 2018-04-05 Kito Cheng <kito.cheng@gmail.com>
18180 * config/nds32/nds32.md (feature): New attribute.
18182 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
18184 * config/nds32/nds32.md (subtype): New attribute.
18186 2018-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
18189 * config/arm/arm-builtins.c (arm_expand_builtin): Change
18190 expansion to perform a bitwise AND of the argument followed by a
18191 boolean negation of the result.
18193 2018-04-04 Peter Bergner <bergner@vnet.ibm.com>
18195 PR rtl-optimization/84878
18196 * ddg.c (add_cross_iteration_register_deps): Use DF_REF_BB to determine
18197 the basic block. Assert the use reference is not artificial and that
18198 it has an associated insn.
18200 2018-04-04 Michael Matz <matz@suse.de>
18202 * builtins.c (compute_objsize): Pass correct operand
18203 to array_at_struct_end_p.
18205 2018-04-04 Richard Biener <rguenther@suse.de>
18208 * dwarf2out.c (dwarf2out_register_external_die): Peel namespaces
18209 from contexts for DINFO_LEVEL_TERSE and below.
18211 2018-04-04 Kito Cheng <kito.cheng@gmail.com>
18213 * config/nds32/nds32-doubleword.md (move_<mode>): Require
18214 resiter_operand condition.
18215 * config/nds32/nds32.md (*move<mode>): Ditto.
18217 2018-04-04 Kito Cheng <kito.cheng@gmail.com>
18218 Monk Chiang <sh.chiang04@gmail.com>
18220 * config/nds32/nds32.md (movmisalign<mode>): New pattern.
18222 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
18224 * config/nds32/nds32.md (movqi, movhi): Merge into mov<mode>.
18226 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
18227 Kito Cheng <kito.cheng@gmail.com>
18229 * config/nds32/nds32-md-auxiliary.c (nds32_inverse_cond_code,
18230 nds32_cond_code_str, output_cond_branch,
18231 output_cond_branch_compare_zero, nds32_expand_cbranch,
18232 nds32_expand_cstore, nds32_expand_movcc,
18233 nds32_output_cbranchsi4_equality_zero,
18234 nds32_output_cbranchsi4_equality_reg,
18235 nds32_output_cbranchsi4_equality_reg_or_const_int,
18236 nds32_output_cbranchsi4_greater_less_zero: New functions.
18237 * config/nds32/nds32-protos.h (nds32_expand_cbranch,
18238 nds32_expand_cstore, nds32_expand_movcc,
18239 nds32_output_cbranchsi4_equality_zero,
18240 nds32_output_cbranchsi4_equality_reg,
18241 nds32_output_cbranchsi4_equality_reg_or_const_int,
18242 nds32_output_cbranchsi4_greater_less_zero): Declare.
18243 * config/nds32/predicates.md (nds32_movecc_comparison_operator,
18244 nds32_rimm11s_operand): New predicates.
18245 * config/nds32/nds32.h (nds32_expand_result_type): New enum type.
18246 * config/nds32/nds32.md: Rewrite all the branch and conditional move
18249 2018-04-04 Kito Cheng <kito.cheng@gmail.com>
18251 * config/nds32/nds32-doubleword.md: Refine all the instruction type.
18252 * config/nds32/nds32.md: Ditto.
18253 * config/nds32/pipelines.md: Ditto.
18255 2018-04-04 Richard Biener <rguenther@suse.de>
18257 PR tree-optimization/85168
18258 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address): Avoid
18259 propagating abnormals.
18261 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
18263 * config/nds32/nds32.md (enabled): Use yes/no for this attribute.
18265 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
18266 Kito Cheng <kito.cheng@gmail.com>
18268 * config/nds32/nds32-md-auxiliary.c (nds32_long_call_p): New function.
18269 * config/nds32/nds32-protos.h (nds32_long_call_p): Declare.
18270 * config/nds32/nds32.c (nds32_function_ok_for_sibcall): New function.
18271 (TARGET_FUNCTION_OK_FOR_SIBCALL): Define.
18272 * config/nds32/nds32.md (sibcall_internal): New.
18273 (sibcall_register): Remove.
18274 (sibcall_immediate): Remove.
18275 (sibcall_value_internal): New.
18276 (sibcall_value_register): Remove.
18277 (sibcall_value_immediate): Remove.
18278 * config/nds32/predicates.md (nds32_general_register_operand): New.
18279 (nds32_call_address_operand): New.
18281 2018-04-03 Jakub Jelinek <jakub@redhat.com>
18283 PR rtl-optimization/85167
18284 * shrink-wrap.c (move_insn_for_shrink_wrap): Don't set bb_uses and
18285 bb_defs if *split_p, instead preinitialize it to NULL.
18287 PR tree-optimization/85156
18288 * builtins.c (fold_builtin_expect): Use save_expr on arg1 to avoid
18289 evaluating the argument multiple times.
18291 2018-04-03 Bill Schmidt <wschmidt@linux.ibm.com>
18293 * config/rs6000/emmintrin.h (_mm_cvtpd_epi32): Use __vector rather
18295 (_mm_cvtpd_ps): Likewise.
18296 (_mm_cvttpd_epi32): Likewise.
18297 * config/rs6000/mmintrin.h (_mm_unpacklo_pi8): Likewise.
18298 * config/rs6000/xmmintrin.h: For strict-ANSI C++ or C11, undefine
18299 vector, pixel, and bool following altivec.h include.
18301 2018-04-03 Martin Sebor <msebor@redhat.com>
18303 * doc/extend.texi (Common Function Attributes): Clarify.
18304 (const attribute): Likewise.
18305 (pure attribute): Likewise.
18307 2018-04-03 Jakub Jelinek <jakub@redhat.com>
18310 * config/i386/i386.c (ix86_expand_vector_set): Use
18311 HOST_WIDE_INT_1U << elt instead of 1 << elt. Formatting fix.
18313 2018-04-03 Uros Bizjak <ubizjak@gmail.com>
18315 * config/i386/i386.c (emit_i387_cw_initialization): Always use logic
18316 instructions when changing rounding bits to preserve precision bits
18317 in the x87 control word.
18319 2018-04-03 Martin Liska <mliska@suse.cz>
18321 PR tree-optimization/82491
18322 * rtl.h (strip_offset_and_add): Replace += suboffset with
18323 poly_uint64 () + suboffset.
18325 2018-03-29 Martin Liska <mliska@suse.cz>
18326 Martin Jambor <mjambor@suse.cz>
18329 * ipa-cp.c (propagate_bits_across_jump_function): Bail out if
18330 param_type is not an integral or pointer type.
18332 2018-04-03 Richard Biener <rguenther@suse.de>
18334 * sese.h (recompute_all_dominators): Remove.
18336 2018-04-02 Martin Sebor <msebor@redhat.com>
18338 * doc/invoke.texi (-Wrestrict): Fix typos.
18340 2018-04-02 Jim Wilson <jimw@sifive.com>
18342 * config/riscv/riscv.h (SHIFT_COUNT_TRUNCATED): Set to zero.
18343 * config/riscv/riscv.md (<optab>si3): Use QImode shift count.
18344 (<optab>di3, <optab>si3_extend): Likewise.
18345 (<optab>si3_mask, <optab>si3_mask_1): New.
18346 (<optab>di3_mask, <optab>di3_mask_1): New.
18347 (<optab>si3_extend_mask, <optab>si3_extend_mask_1): New.
18348 (lshrsi3_zero_extend_1): Use VOIDmode shift count.
18349 * config/riscv/sync.md (atomic_test_and_set): Emit QImode shift count.
18351 2018-04-02 Gerald Pfeifer <gerald@pfeifer.com>
18353 * doc/cpp.texi (Variadic Macros): Fix line continuation in an
18356 2018-04-02 Chung-Ju Wu <jasonwucj@gmail.com>
18358 * config/nds32/nds32.c (TARGET_CANONICALIZE_COMPARISON): Define.
18359 (nds32_canonicalize_comparison): New function.
18361 2018-04-01 Chung-Ju Wu <jasonwucj@gmail.com>
18362 Kito Cheng <kito.cheng@gmail.com>
18363 Kuan-Lin Chen <kuanlinchentw@gmail.com>
18365 * config.gcc (nds32): Add nds32-relax-opt.o into extra_objs.
18366 * config/nds32/constants.md (unspec_volatile_element): Add
18367 UNSPEC_VOLATILE_RELAX_GROUP.
18368 * config/nds32/nds32-relax-opt.c: New file.
18369 * config/nds32/nds32-predicates.c
18370 (nds32_symbol_load_store_p): New function.
18371 * config/nds32/nds32-protos.h
18372 (nds32_symbol_load_store_p): Declare function.
18373 (make_pass_nds32_relax_opt): Declare new rtl pass function.
18374 * config/nds32/nds32.c
18375 (nds32_register_pass): New function to register pass.
18376 (nds32_register_passes): New function to register passes.
18377 * config/nds32/nds32.md (relax_group): New pattern.
18378 * config/nds32/nds32.opt (mrelax-hint): New option.
18379 * config/nds32/t-nds32 (nds32-relax-opt.o): New dependency.
18381 2018-04-01 Kito Cheng <kito.cheng@gmail.com>
18383 * config/nds32/t-nds32: Modify files dependency.
18385 2018-04-01 Chung-Ju Wu <jasonwucj@gmail.com>
18387 * config/nds32/nds32.h (FUNCTION_PROFILER): Output newline character.
18388 (PROFILE_HOOK): Define its implementation.
18390 2018-04-01 Chung-Ju Wu <jasonwucj@gmail.com>
18392 * config/nds32/nds32.h (WCHAR_TYPE, WCHAR_TYPE_SIZE): Use unsigned int
18393 type and 32-bit size.
18395 2018-04-01 Jakub Jelinek <jakub@redhat.com>
18397 PR middle-end/85090
18398 * config/i386/sse.md (V): Add V64QI and V32HI for TARGET_AVX512F.
18399 (V_128_256): New mode iterator.
18400 (*avx512dq_vextract<shuffletype>64x2_1 splitter): New define_split.
18401 (*avx512f_vextract<shuffletype>32x4_1 splitter): Likewise.
18402 (xop_pcmov_<mode><avxsizesuffix>): Use V_128_256 mode iterator instead
18404 * config/i386/i386.c (ix86_expand_vector_set): Improve V32HImode and
18405 V64QImode expansion for !TARGET_AVX512BW && TARGET_AVX512F.
18407 2018-03-31 Segher Boessenkool <segher@kernel.crashing.org>
18410 * config/rs6000/xmmintrin.h (_mm_set_ps, _mm_max_ps): Handle (quiet)
18411 NaN inputs correctly.
18413 2018-03-30 Peter Bergner <bergner@vnet.ibm.com>
18416 * config/rs6000/vsx.md (??r): New mode attribute.
18417 (*vsx_mov<mode>_64bit): Use it.
18418 (*vsx_mov<mode>_32bit): Likewise.
18420 2018-03-30 Martin Sebor <msebor@redhat.com>
18422 PR tree-optimization/84818
18423 * builtins.c (check_access): Use warning_n.
18425 2018-03-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
18428 * config/rs6000/rs6000-string.c (expand_compare_loop): Fix redundant
18430 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Fix redundant
18433 2018-03-30 Julia Koval <julia.koval@intel.com>
18436 * x86-tune.def (movx, partial_reg_dependency): Enable for
18439 2018-03-29 Vladimir Makarov <vmakarov@redhat.com>
18441 PR inline-asm/84985
18442 * lra-constraints.c (process_alt_operands): Move setting
18443 this_alternative_matches below.
18445 2018-03-29 Martin Liska <mliska@suse.cz>
18448 * doc/invoke.texi: Document how LTO works with debug info.
18449 Describe auto-load support of binutils. Mention 'x86-64'
18450 as valid option value of -march option.
18452 2018-03-29 Jakub Jelinek <jakub@redhat.com>
18454 * config/i386/sse.md (<avx512>_blendm<mode>): Use <sseintprefix>.
18457 * fold-const.c (operand_equal_p): Handle DEBUG_BEGIN_STMT.
18458 For STATEMENT_LIST, pass down OEP_LEXICOGRAPHIC and maybe
18459 OEP_NO_HASH_CHECK for recursive call, to avoid exponential
18462 2018-03-28 Peter Bergner <bergner@vnet.ibm.com>
18465 * config/rs6000/rs6000.h (RS6000_BTM_POWERPC64): New define.
18466 (RS6000_BTM_COMMON): Add RS6000_BTM_POWERPC64.
18467 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add support
18468 for RS6000_BTM_POWERPC64.
18469 (rs6000_invalid_builtin): Add handling for RS6000_BTM_POWERPC64
18470 (rs6000_builtin_mask_names): Add RS6000_BTM_POWERPC64.
18471 * config/rs6000/rs6000-builtin.def (BU_P7_POWERPC64_MISC_2): New macro
18474 (DIVDEU): Likewise.
18476 2018-03-28 Carl Love <cel@us.ibm.com>
18479 2017-09-27 Carl Love <cel@us.ibm.com>
18481 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
18482 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
18483 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
18486 2018-03-28 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18488 * config/rs6000/xmmintrin.h (_mm_max_pi16): Use __vector __bool
18489 instead of __vector bool.
18490 (_mm_max_pu8): Likewise.
18491 (_mm_min_pi16): Likewise.
18493 2018-03-28 Peter Bergner <bergner@vnet.ibm.com>
18496 * config/rs6000/rs6000-builtin.def (DIVWEO): Delete macro expansion.
18497 (DIVWEUO): Likewise.
18498 (DIVDEO): Likewise.
18499 (DIVDEUO): Likewise.
18500 * config/rs6000/rs6000.c (builtin_function_type): Remove support for
18501 DIVWEUO and DIVDEUO.
18502 * config/rs6000/rs6000.md (UNSPEC_DIVEO, UNSPEC_DIVEUO): Delete unspecs.
18503 (UNSPEC_DIV_EXTEND): Remove deleted unspecs.
18504 (div_extend): Likewise.
18505 * doc/extend.texi (__builtin_divweo): Remove documentation for deleted
18507 (__builtin_divweuo): Likewise.
18508 (__builtin_divdeo): Likewise.
18509 (__builtin_divdeuo): Likewise.
18511 2018-03-28 Jakub Jelinek <jakub@redhat.com>
18514 * config/i386/i386.md (*add<mode>3_carry_0, *addsi3_carry_zext_0,
18515 *sub<mode>3_carry_0, *subsi3_carry_zext_0): New patterns.
18517 PR tree-optimization/82004
18518 * gimple-match-head.c (optimize_pow_to_exp): New function.
18519 * match.pd (pow(C,x) -> exp(log(C)*x)): Wrap with #if GIMPLE.
18520 Don't fold to exp if optimize_pow_to_exp is false.
18522 2018-03-28 Martin Liska <mliska@suse.cz>
18525 * calls.c (initialize_argument_information): Fix trailing space.
18526 * common.opt: Fix typo and provide better explanation for
18527 -fsanitize-coverage option.
18528 * config/i386/i386.opt: Fix typo.
18530 2018-03-28 Jakub Jelinek <jakub@redhat.com>
18531 Martin Liska <mliska@suse.cz>
18534 * gimplify.c (asan_poison_variable): Don't do the check for
18535 gimplify_omp_ctxp here.
18536 (gimplify_decl_expr): Do it here.
18537 (gimplify_target_expr): Likewise.
18539 2018-03-28 Martin Liska <mliska@suse.cz>
18542 * config/i386/i386.c (ix86_function_arg_advance): Do not call
18543 chkp_type_bounds_count if MPX is not enabled.
18545 2018-03-27 Chung-Ju Wu <jasonwucj@gmail.com>
18547 * config/nds32/nds32.h (BRANCH_COST): Adjust cost.
18549 2018-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
18552 * config/rs6000/rs6000.c (create_complex_muldiv): New helper
18553 function to create the function decl for complex long double
18554 multiply and divide for -mabi=ieeelongdouble.
18555 (init_float128_ieee): Call it.
18557 2018-03-27 H.J. Lu <hongjiu.lu@intel.com>
18560 * config/i386/i386.c (ix86_trampoline_init): Insert ENDBR for
18561 -fcf-protection=branch -mibt.
18562 * config/i386/i386.h (TRAMPOLINE_SIZE): Increased by 4 bytes.
18564 2018-03-27 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
18567 * config/arm/arm.c (arm_valid_symbolic_address): Handle
18568 arm_word_relocations.
18570 2018-03-27 Cesar Philippidis <cesar@codesourcery.com>
18573 * config/nvptx/nvptx.c (nvptx_assemble_decl_begin): Add '[]' to
18574 extern array declarations.
18576 2018-03-27 Richard Biener <rguenther@suse.de>
18578 PR middle-end/84067
18579 * match.pd ((A * C) +- (B * C) -> (A+-B) * C): Guard with
18580 explicit single_use checks.
18582 2018-03-27 Richard Biener <rguenther@suse.de>
18584 PR tree-optimization/85082
18585 * tree-ssa-sccvn.c (vn_reference_lookup_or_insert_for_pieces):
18588 2018-03-27 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
18590 * config.gcc (aarch64*-*-linux*): New TARGET_DEFAULT_ASYNC_UNWIND_TABLES
18591 * common/config/aarch64/aarch64-common.c (aarch64_optimization_table[]):
18592 Turn on fasynchronous-unwind-tables and funwind-tables.
18594 2018-03-26 Uros Bizjak <ubizjak@gmail.com>
18597 * config/i386/i386.md (*bmi_blsr_<mode>_cmp): New insn pattern.
18598 (*bmi_blsr_<mode>_ccz): Ditto.
18600 2018-03-26 Tom de Vries <tom@codesourcery.com>
18602 PR tree-optimization/85063
18603 * omp-general.c (offloading_function_p): New function. Factor out
18605 * omp-offload.c (pass_omp_target_link::gate): ... here.
18606 * omp-general.h (offloading_function_p): Declare.
18607 * tree-switch-conversion.c (build_one_array): Mark CSWTCH.x variable
18608 with attribute omp declare target for offloading functions.
18610 2018-03-24 Richard Sandiford <richard.sandiford@linaro.org>
18612 PR tree-optimization/84005
18613 * tree-data-ref.h (get_base_for_alignment): Declare.
18614 * tree-data-ref.c (get_base_for_alignment_1): New function.
18615 (get_base_for_alignment): Likewise.
18616 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
18617 get_base_for_alignment to find a suitable base object, instead
18618 of always using drb->base_address.
18620 2018-03-23 Jakub Jelinek <jakub@redhat.com>
18622 PR inline-asm/85022
18623 * emit-rtl.c (init_emit_regs): Indicate that VOIDmode MEMs don't have
18624 known size by default.
18626 2018-03-23 Vladimir Makarov <vmakarov@redhat.com>
18628 PR inline-asm/85030
18629 * lra-constraints.c (process_alt_operands): Don't match BLKmode
18630 and non BLKmode operands.
18632 2018-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18635 * config/arm/arm.md (unaligned_loadhis): Remove first alternative.
18636 Clean up attributes.
18638 2018-03-23 Richard Biener <rguenther@suse.de>
18641 * dwarf2out.c (rtl_for_decl_location): Do not generate RTL early when
18642 we are going to emit early debug for LTO.
18644 2018-03-23 Jakub Jelinek <jakub@redhat.com>
18646 PR inline-asm/85034
18647 * function.c (match_asm_constraints_1): Don't optimize if input
18648 doesn't satisfy general_operand predicate for output's mode.
18650 PR inline-asm/85022
18651 * alias.c (write_dependence_p): Don't require for x_canonicalized
18652 non-VOIDmode if x has VOIDmode.
18655 * sanopt.c (maybe_optimize_ubsan_ptr_ifn): If DECL_REGISTER (base),
18656 just don't try to optimize it rather than assert it never happens.
18658 2018-03-22 Kelvin Nilsen <kelvin@gcc.gnu.org>
18660 * config/rs6000/rs6000-builtin.def: Remove various BU_ALTIVEC_X
18661 macro expansions for definition of ST_INTERNAL_<mode> and
18662 LD_INTERNAL_<mode> builtins.
18663 * config/rs6000/rs6000-protos.h (rs6000_address_for_altivec):
18665 * config/rs6000/rs6000.c (altivec_expand_ld_builtin): Delete this
18667 (altivec_expand_st_builtin): Likewise.
18668 (altivec_expand_builtin): Remove calls to deleted functions.
18669 (rs6000_address_for_altivec): Delete this function.
18670 * config/rs6000/vector.md: Remove expands for
18671 vector_altivec_load_<mode> and vector_altivec_store_<mode>.
18673 2018-03-22 Sudakshina Das <sudi.das@arm.com>
18676 * config/arm/arm.h (machine_function): Add static_chain_stack_bytes.
18677 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Avoid
18678 re-computing once computed.
18679 (arm_expand_prologue): Compute machine->static_chain_stack_bytes.
18680 (arm_init_machine_status): Initialize
18681 machine->static_chain_stack_bytes.
18683 2018-03-22 Kelvin Nilsen <kelvin@gcc.gnu.org>
18686 * doc/extend.texi: Add four new prototypes for vec_ld.
18687 * config/rs6000/rs6000-builtin.def (LVX_V1TI): Reorder symbol
18688 definitions for more logical presentation.
18689 * config/rs6000/rs6000-c.c: (altivec_overloaded_builtins): Add
18690 entries for V1TI variants of __builtin_altivec_ld builtin.
18691 * config/rs6000/rs6000.c: (altivec_expand_lv_builtin): Add test and
18692 handling of V1TI variant of LVX icode pattern.
18693 (altivec_expand_builtin): Add case for ALTIVEC_BUILTIN_LVX_V1TI.
18694 (rs6000_gimple_fold_builtin): Likewise.
18695 (altivec_init_builtins): Add code to define
18696 __builtin_altivec_lvx_v1ti function.
18698 2018-03-22 Jakub Jelinek <jakub@redhat.com>
18700 PR inline-asm/84941
18701 * function.c (match_asm_constraints_1): Don't do the optimization
18702 if input isn't a REG, SUBREG, MEM or constant.
18704 2018-03-22 Tom de Vries <tom@codesourcery.com>
18706 PR tree-optimization/84956
18707 * tree-ssa-tail-merge.c (find_clusters_1): Skip bbs with
18708 bb_has_abnormal_pred.
18710 2018-03-22 Jakub Jelinek <jakub@redhat.com>
18713 * dwarf2asm.c (dw2_output_indirect_constant_1): Set
18714 DECL_INITIAL (decl) to decl at the end.
18715 * varasm.c (use_blocks_for_decl_p): Revert the 2018-03-20 change,
18716 adjust the comment.
18718 2018-03-21 Joseph Myers <joseph@codesourcery.com>
18720 * doc/extend.texi (__builtin_tgmath): Document when complex
18721 integer types are treated as _Complex _Float64.
18723 2018-03-21 Tom de Vries <tom@codesourcery.com>
18725 * doc/extend.texi (__builtin_extend_pointer): Remove pasto.
18727 2018-03-21 Jakub Jelinek <jakub@redhat.com>
18729 PR tree-optimization/84960
18730 * tree-cfg.c (remove_bb): Don't move forced labels into bb->prev_bb
18731 if it is ENTRY block, move them into single succ of ENTRY in that case.
18733 2018-03-21 Richard Sandiford <richard.sandiford@linaro.org>
18735 PR tree-optimization/84811
18736 * poly-int.h (poly_span_traits): Remove the T3 parameter and
18737 promote HOST_WIDE_INT T2 - T1 results to unsigned HOST_WIDE_INT.
18738 (maybe_in_range_p, known_in_range_p, ranges_known_overlap_p):
18739 (known_subrange_p): Update accordingly. Cast each value involved
18740 in the size comparison, rather than casting the result of the
18743 2018-03-21 Jakub Jelinek <jakub@redhat.com>
18745 PR tree-optimization/84982
18746 * gimple-ssa-store-merging.c (invert_op): Handle boolean inversion
18747 by flipping the least significant bit rather than all bits from
18748 bitpos to bitpos + bitsize - 1.
18750 2018-03-21 Nathan Sidwell <nathan@acm.org>
18752 * doc/extend.texi (Deprecated Features): Remove mention of
18753 long-deleted deprecations.
18755 2018-03-21 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
18758 * configure.ac (gcc_cv_ld_soname) <*-*-solaris2*>: Set.
18759 * configure: Regenerate.
18761 2018-03-21 Tom de Vries <tom@codesourcery.com>
18763 PR tree-optimization/83126
18764 * tree-parloops.c (num_phis): New function.
18765 (gen_parallel_loop): Detect and handle canonicalize_loop_ivs failure.
18767 2018-03-21 Nathan Sidwell <nathan@acm.org>
18769 * doc/extend.texi (Deprecated Features): Update deprecated flags,
18770 mention anon-struct/union members and trailing attributes.
18772 2018-03-21 Bin Cheng <bin.cheng@arm.com>
18774 PR tree-optimization/84969
18775 * tree-loop-distribution.c (fuse_memset_builtins): Don't reorder
18776 builtin memset partitions if they set different rhs values.
18778 2018-03-21 Jakub Jelinek <jakub@redhat.com>
18780 PR rtl-optimization/84989
18781 * simplify-rtx.c (simplify_unary_operation_1): Don't try to simplify
18782 VEC_DUPLICATE with scalar result mode.
18784 2018-03-21 Martin Liska <mliska@suse.cz>
18787 * ipa-icf.c (sem_item_optimizer::fixup_points_to_sets): Remove
18788 not intended return statement.
18790 2018-03-21 Martin Liska <mliska@suse.cz>
18793 * tree-chkp.c (CHKP_ARRAY_MAX_CHECK_STEPS): Define a new macro.
18794 (chkp_find_bound_slots_1): Limit number of iterations.
18796 2018-03-20 David H. Gutteridge <dhgutteridge@sympatico.ca>
18799 * Minor grammar fixes for x86 options.
18801 2018-03-20 Jakub Jelinek <jakub@redhat.com>
18804 * dce.c (delete_unmarked_insns): Don't remove frame related noop moves
18805 holding REG_CFA_RESTORE notes, instead turn them into a USE.
18807 2018-03-20 Peter Bergner <bergner@vnet.ibm.com>
18810 * config/rs6000/altivec.md (altivec_lvx_<mode>_2op): Delete define_insn.
18811 (altivec_lvx_<mode>_1op): Likewise.
18812 (altivec_stvx_<mode>_2op): Likewise.
18813 (altivec_stvx_<mode>_1op): Likewise.
18814 (altivec_lvx_<VM2:mode>): New define_expand.
18815 (altivec_stvx_<VM2:mode>): Likewise.
18816 (altivec_lvx_<VM2:mode>_2op_<P:mptrsize>): New define_insn.
18817 (altivec_lvx_<VM2:mode>_1op_<P:mptrsize>): Likewise.
18818 (altivec_stvx_<VM2:mode>_2op_<P:mptrsize>): Likewise.
18819 (altivec_stvx_<VM2:mode>_1op_<P:mptrsize>): Likewise.
18820 * config/rs6000/rs6000-p8swap.c (rs6000_gen_stvx): Use new expanders.
18821 (rs6000_gen_lvx): Likewise.
18822 * config/rs6000/rs6000.c (altivec_expand_lv_builtin): Likewise.
18823 (altivec_expand_stv_builtin): Likewise.
18824 (altivec_expand_builtin): Likewise.
18825 * config/rs6000/vector.md: Likewise.
18827 2018-03-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18830 * config/arm/arm.c (arm_array_mode_supported_p): Return false for
18833 2018-03-20 Richard Biener <rguenther@suse.de>
18836 * config/i386/i386.c (ix86_add_stmt_cost): Only cost
18837 sign-conversions as zero, fall back to standard scalar_stmt
18840 2018-03-20 Martin Liska <mliska@suse.cz>
18843 * predict.c (rebuild_frequencies): Handle case when we have
18844 PROFILE_ABSENT, but flag_guess_branch_prob is false.
18846 2018-03-20 Jakub Jelinek <jakub@redhat.com>
18849 * dwarf2asm.c (dw2_output_indirect_constant_1): Temporarily turn off
18850 flag_section_anchors.
18851 * varasm.c (use_blocks_for_decl_p): Remove hack for
18852 dw2_force_const_mem.
18855 * config/aarch64/aarch64.md (*aarch64_reg_<mode>3_neg_mask2): Rename
18857 (*aarch64_<optab>_reg_<mode>3_neg_mask2): ... this. If pseudos can't
18858 be created, use lowpart_subreg of operands[0] rather than operands[0]
18860 (*aarch64_reg_<mode>3_minus_mask): Rename to ...
18861 (*aarch64_ashl_reg_<mode>3_minus_mask): ... this.
18862 (*aarch64_<optab>_reg_di3_mask2): Use const_int_operand predicate
18863 and n constraint instead of aarch64_shift_imm_di and Usd.
18864 (*aarch64_reg_<optab>_minus<mode>3): Rename to ...
18865 (*aarch64_<optab>_reg_minus<mode>3): ... this.
18867 2018-03-20 Sudakshina Das <sudi.das@arm.com>
18870 * config/arm/neon.md (ashldi3_neon): Update ?s for constraints
18871 to favor GPR over NEON registers.
18872 (<shift>di3_neon): Likewise.
18874 2018-03-20 Tom de Vries <tom@codesourcery.com>
18877 * config/nvptx/nvptx.c (nvptx_single): Don't neuter bar.sync.
18878 (nvptx_process_pars): Emit bar.sync asap and alap.
18880 2018-03-20 Tom de Vries <tom@codesourcery.com>
18883 * config/nvptx/nvptx.c (prevent_branch_around_nothing): Also update
18884 seen_label if seen_label is already set.
18886 2018-03-20 Jakub Jelinek <jakub@redhat.com>
18889 * config/i386/i386.c (fold_builtin_cpu): For features above 31
18890 use __cpu_features2 variable instead of __cpu_model.__cpu_features[0].
18891 Use 1U instead of 1. Formatting fixes.
18894 * builtins.c (fold_builtin_strpbrk): For strpbrk(x, "") use type
18895 instead of TREE_TYPE (s1) for the return value.
18897 2018-03-19 Jakub Jelinek <jakub@redhat.com>
18899 PR tree-optimization/84946
18900 * gimple-ssa-store-merging.c (mem_valid_for_store_merging): Compute
18901 bitsize + bitsize in poly_uint64 rather than poly_int64.
18904 * dwarf2asm.c: Include fold-const.c.
18905 (dw2_output_indirect_constant_1): Set DECL_INITIAL (decl) to ADDR_EXPR
18906 of decl rather than decl itself.
18908 PR rtl-optimization/84643
18909 * memmodel.h (enum memmodel): Add MEMMODEL_MAX enumerator.
18911 2018-03-19 Maxim Ostapenko <m.ostapenko@samsung.com>
18914 * dwarf2asm.c (dw2_output_indirect_constant_1): Disable ASan before
18915 calling assemble_variable.
18917 2018-03-19 Sudakshina Das <sudi.das@arm.com>
18920 * config/aarch64/aarch64-simd.md (vec_cmp<mode><v_int_equiv>): Modify
18921 instructions for UNLT, UNLE, UNGT, UNGE, UNEQ, UNORDERED and ORDERED.
18923 2018-03-19 Jim Wilson <jimw@sifive.com>
18926 * config/riscv/riscv.c (riscv_function_arg_boundary): Use
18927 PREFERRED_STACK_BOUNDARY instead of STACK_BOUNDARY.
18928 (riscv_first_stack_step): Likewise.
18929 (riscv_option_override): Use STACK_BOUNDARY instead of
18930 MIN_STACK_BOUNDARY.
18931 * config/riscv/riscv.h (STACK_BOUNDARY): Renamed from
18932 MIN_STACK_BOUNDARY.
18933 (BIGGEST_ALIGNMENT): Set to 128.
18934 (PREFERRED_STACK_BOUNDARY): Renamed from STACK_BOUNDARY.
18935 (RISCV_STACK_ALIGN): Use PREFERRED_STACK_BOUNDARY instead of
18938 2018-03-19 Richard Biener <rguenther@suse.de>
18940 PR tree-optimization/84933
18941 * tree-vrp.c (set_and_canonicalize_value_range): Treat out-of-bound
18942 values as -INF/INF when canonicalizing an ANTI_RANGE to a RANGE.
18944 2018-03-19 Richard Biener <rguenther@suse.de>
18946 PR tree-optimization/84859
18947 * tree-ssa-phiopt.c (single_trailing_store_in_bb): New function.
18948 (cond_if_else_store_replacement): Perform sinking operation on
18949 single-store BBs regardless of MAX_STORES_TO_SINK setting.
18950 Generalize what a BB with a single eligible store is.
18952 2018-03-19 Richard Biener <rguenther@suse.de>
18954 PR tree-optimization/84929
18955 * tree-data-ref.c (analyze_siv_subscript_cst_affine): Guard
18956 chrec_is_positive against non-chrec arg.
18958 2018-03-19 Tamar Christina <tamar.christina@arm.com>
18961 * config/arm/arm.c (arm_can_change_mode_class): revert r258554.
18963 2018-03-18 Martin Liska <mliska@suse.cz>
18965 PR rtl-optimization/84635
18966 * regrename.c (build_def_use): Use matches_mode only when
18969 2018-03-18 Richard Sandiford <richard.sandiford@linaro.org>
18971 PR tree-optimization/84913
18972 * tree-vect-loop.c (vectorizable_reduction): Don't try to
18973 vectorize chains of COND_EXPRs.
18975 2018-03-18 Chung-Ju Wu <jasonwucj@gmail.com>
18977 * config/nds32/nds32.h (MAX_REGS_PER_ADDRESS): Fix the value.
18979 2018-03-18 Chung-Ju Wu <jasonwucj@gmail.com>
18981 * config/nds32/nds32.h (LOGICAL_OP_NON_SHORT_CIRCUIT): Define.
18983 2018-03-18 Chung-Ju Wu <jasonwucj@gmail.com>
18985 * config/nds32/nds32.h (CLZ_DEFINED_VALUE_AT_ZERO): Define.
18987 2018-03-17 Chung-Ju Wu <jasonwucj@gmail.com>
18988 Kito Cheng <kito.cheng@gmail.com>
18990 * config/nds32/nds32-protos.h (nds32_adjust_reg_alloc_order): Declare.
18991 * config/nds32/nds32.c (nds32_reg_alloc_order_for_speed): New array.
18992 (nds32_adjust_reg_alloc_order): New function.
18993 * config/nds32/nds32.h (ADJUST_REG_ALLOC_ORDER): Define.
18995 2018-03-17 Kito Cheng <kito.cheng@gmail.com>
18997 * config/nds32/nds32.c (nds32_asm_output_mi_thunk,
18998 nds32_print_operand, nds32_print_operand_address): Use
18999 HOST_WIDE_INT_PRINT_DEC instead.
19001 2018-03-17 Chung-Ju Wu <jasonwucj@gmail.com>
19003 * config/nds32/nds32.c (nds32_register_priority): Modify cost.
19005 2018-03-17 Jakub Jelinek <jakub@redhat.com>
19008 * config/i386/i386.c (initial_ix86_tune_features,
19009 initial_ix86_arch_features): Use unsigned HOST_WIDE_INT rather than
19010 unsigned long long.
19011 (set_ix86_tune_features): Change ix86_tune_mask from unsigned int
19012 to unsigned HOST_WIDE_INT, initialize to HOST_WIDE_INT_1U << ix86_tune
19013 rather than 1u << ix86_tune. Formatting fix.
19014 (ix86_option_override_internal): Change ix86_arch_mask from
19015 unsigned int to unsigned HOST_WIDE_INT, initialize to
19016 HOST_WIDE_INT_1U << ix86_arch rather than 1u << ix86_arch.
19017 (ix86_function_specific_restore): Likewise.
19019 2018-03-16 Jakub Jelinek <jakub@redhat.com>
19022 * postreload.c (reload_combine_recognize_pattern): Perform
19023 INTVAL addition in unsigned HOST_WIDE_INT type to avoid UB and
19024 truncate_int_for_mode the result for the destination's mode.
19027 * hsa-gen.c (mem_type_for_type): Fix comment typo.
19028 * tree-vect-loop-manip.c (vect_create_cond_for_niters_checks):
19030 * gimple-ssa-warn-restrict.c (builtin_memref::set_base_and_offset):
19033 2018-03-16 Vladimir Makarov <vmakarov@redhat.com>
19036 * lra-assigns.c (lra_split_hard_reg_for): Don't use
19037 regno_allocno_class_array and sorted_pseudos.
19038 * lra-constraints.c (spill_hard_reg_in_range): Ignore hard regs in
19039 insns where regno is used.
19041 2018-03-16 Martin Liska <mliska@suse.cz>
19044 * multiple_target.c (create_dispatcher_calls): Redirect
19045 reference in the symbol table.
19047 2018-03-16 Martin Liska <mliska@suse.cz>
19050 * multiple_target.c (create_dispatcher_calls): Redirect also
19053 2018-03-16 Jakub Jelinek <jakub@redhat.com>
19057 * tree.h (TARGET_EXPR_NO_ELIDE): Define.
19058 * gimplify.c (gimplify_modify_expr_rhs): Don't elide TARGET_EXPRs with
19059 TARGET_EXPR_NO_ELIDE flag set unless *expr_p is INIT_EXPR.
19061 2018-03-16 Julia Koval <julia.koval@intel.com>
19063 * doc/invoke.texi (Skylake Server): Add CLWB.
19064 Cannonlake): Remove CLWB.
19066 2018-03-16 Jakub Jelinek <jakub@redhat.com>
19068 PR tree-optimization/84841
19069 * tree-ssa-reassoc.c (INTEGER_CONST_TYPE): Change to 1 << 4 from
19071 (FLOAT_ONE_CONST_TYPE): Define.
19072 (constant_type): Return FLOAT_ONE_CONST_TYPE for -1.0 and 1.0.
19073 (sort_by_operand_rank): Put entries with higher constant_type last
19074 rather than first to match comments.
19076 2018-03-15 Sandra Loosemore <sandra@codesourcery.com>
19078 * config/nios2/nios2.md (movsi_internal): Fix thinko in
19081 2018-03-15 Jakub Jelinek <jakub@redhat.com>
19084 * calls.c (expand_call): For TREE_ADDRESSABLE rettype ignore alignment
19085 check and use address of target always.
19087 2018-03-15 H.J. Lu <hongjiu.lu@intel.com>
19090 * config/i386/i386.c (indirect_thunk_needed): Update comments.
19091 (indirect_thunk_bnd_needed): Likewise.
19092 (indirect_thunks_used): Likewise.
19093 (indirect_thunks_bnd_used): Likewise.
19094 (indirect_return_needed): New.
19095 (indirect_return_bnd_needed): Likewise.
19096 (output_indirect_thunk_function): Add a bool argument for
19098 (output_indirect_thunk_function): Don't generate alias for
19099 function return thunk.
19100 (ix86_code_end): Call output_indirect_thunk_function to generate
19101 function return thunks.
19102 (ix86_output_function_return): Set indirect_return_bnd_needed
19103 and indirect_return_needed instead of indirect_thunk_bnd_needed
19104 and indirect_thunk_needed.
19106 2018-03-15 Olga Makhotina <olga.makhotina@intel.com>
19108 * config/i386/sgxintrin.h (_enclv_u32): New intrinsic.
19109 (__enclv_bc, __enclv_cd, __enclv_generic): New definitions.
19110 (ERDINFO, ETRACKC, ELDBC, ELDUC): New leaves.
19112 2018-03-15 David Malcolm <dmalcolm@redhat.com>
19113 Paul Hua <paul.hua.gm@gmail.com>
19116 * gcc.dg/fixits-pr84852-1.c: Fix filename in dg-regexp.
19118 2018-03-15 Segher Boessenkool <segher@kernel.crashing.org>
19120 * config/rs6000/rs6000.c (abi_v4_pass_in_fpr): Add back the
19121 TARGET_DOUBLE_FLOAT and TARGET_SINGLE_FLOAT conditions on the DFmode
19122 resp. SFmode cases.
19124 2018-03-15 Tamar Christina <tamar.christina@arm.com>
19127 * config/arm/arm.c (arm_can_change_mode_class): Use GET_MODE_UNIT_SIZE
19128 instead of GET_MODE_SIZE when comparing Units.
19130 2018-03-15 Vladimir Mezentsev <vladimir.mezentsev@oracle.com>
19133 * varasm.c (hash_section): Return an unchangeble hash value
19134 * config/aarch64/aarch64.c (aarch64_use_blocks_for_constant_p):
19135 Return !aarch64_can_use_per_function_literal_pools_p ().
19137 2018-03-15 Jakub Jelinek <jakub@redhat.com>
19140 * optabs.c (emit_conditional_move): Pass address of cmode's copy
19141 rather than address of cmode as last argument to prepare_cmp_insn.
19143 2018-03-15 Julia Koval <julia.koval@intel.com>
19145 * config/i386/i386.c (F_AVX512VBMI2, F_GFNI, F_VPCLMULQDQ,
19146 F_AVX512VNNI, F_AVX512BITALG): New.
19148 2018-03-14 John David Anglin <danglin@gcc.gnu.org>
19151 * config/pa/pa.c (pa_emit_move_sequence): Always emit secondary reload
19152 insn for floating-point loads and stores.
19154 2018-03-14 Carl Love <cel@us.ibm.com>
19156 * config/rs6000/rs6000-c.c: Add macro definitions for
19157 ALTIVEC_BUILTIN_VEC_PERMXOR.
19158 * config/rs6000/rs6000.h: Add #define for vec_permxor builtin.
19159 * config/rs6000/rs6000-builtin.def: Add macro expansions for VPERMXOR.
19160 * config/rs6000/altivec.md (altivec_vpermxor): New define expand.
19161 * config/rs6000/rs6000-p8swap.c (rtx_is_swappable_p): Add case
19163 * config/doc/extend.texi: Add prototypes for vec_permxor.
19165 2018-03-14 David Malcolm <dmalcolm@redhat.com>
19168 * diagnostic-show-locus.c (class layout_point): Convert m_line
19169 from int to linenum_type.
19170 (line_span::comparator): Use linenum "compare" function when
19171 comparing line numbers.
19172 (test_line_span): New function.
19173 (layout_range::contains_point): Convert param "row" from int to
19175 (layout_range::intersects_line_p): Likewise.
19176 (layout::will_show_line_p): Likewise.
19177 (layout::print_source_line): Likewise.
19178 (layout::should_print_annotation_line_p): Likewise.
19179 (layout::print_annotation_line): Likewise.
19180 (layout::print_leading_fixits): Likewise.
19181 (layout::annotation_line_showed_range_p): Likewise.
19182 (struct line_corrections): Likewise for field m_row.
19183 (line_corrections::line_corrections): Likewise for param "row".
19184 (layout::print_trailing_fixits): Likewise.
19185 (layout::get_state_at_point): Likewise.
19186 (layout::get_x_bound_for_row): Likewise.
19187 (layout::print_line): Likewise.
19188 (diagnostic_show_locus): Likewise for locals "last_line" and "row".
19189 (selftest::diagnostic_show_locus_c_tests): Call test_line_span.
19190 * input.c (selftest::test_linenum_comparisons): New function.
19191 (selftest::input_c_tests): Call it.
19192 * selftest.c (selftest::test_assertions): Test ASSERT_GT,
19193 ASSERT_GT_AT, ASSERT_LT, and ASSERT_LT_AT.
19194 * selftest.h (ASSERT_GT): New macro.
19195 (ASSERT_GT_AT): New macro.
19196 (ASSERT_LT): New macro.
19197 (ASSERT_LT_AT): New macro.
19199 2018-03-14 Segher Boessenkool <segher@kernel.crashing.org>
19201 PR rtl-optimization/84780
19202 * combine.c (distribute_links): Don't make a link based on pc_rtx.
19204 2018-03-14 Martin Liska <mliska@suse.cz>
19206 * tree.c (record_node_allocation_statistics): Use
19207 get_stats_node_kind.
19208 (get_stats_node_kind): New function extracted from
19209 record_node_allocation_statistics.
19210 (free_node): Use get_stats_node_kind.
19212 2018-03-14 Richard Biener <rguenther@suse.de>
19214 * tree-ssa-pre.c (compute_antic_aux): Remove code that asserts
19215 that the value-set of ANTIC_IN doesn't grow.
19218 * tree-ssa-pre.c (struct bb_bitmap_sets): Add visited_with_visited_succs
19220 (BB_VISITED_WITH_VISITED_SUCCS): New define.
19221 (compute_antic): Initialize BB_VISITED_WITH_VISITED_SUCCS.
19223 2018-03-14 Julia Koval <julia.koval@intel.com>
19225 * config.gcc (icelake-client, icelake-server): New.
19227 * config/i386/i386.c (initial_ix86_tune_features): Extend to 64 bit.
19228 (initial_ix86_arch_features): Ditto.
19229 (PTA_SKYLAKE): Add SGX.
19230 (PTA_ICELAKE): Remove.
19231 (PTA_ICELAKE_CLIENT): New.
19232 (PTA_ICELAKE_SERVER): New.
19233 (ix86_option_override_internal): Split up icelake on icelake client and
19235 (get_builtin_code_for_version): Ditto.
19236 (fold_builtin_cpu): Ditto.
19237 * config/i386/driver-i386.c (config/i386/driver-i386.c): Ditto.
19238 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto
19239 * config/i386/i386.h (processor_type): Ditto.
19240 * doc/invoke.texi: Ditto.
19242 2018-03-14 Jakub Jelinek <jakub@redhat.com>
19245 * sanopt.c (maybe_optimize_ubsan_ptr_ifn): Handle also
19246 INTEGER_CST offset, add it together with bitpos / 8 and
19247 sign extend based on POINTER_SIZE.
19251 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
19254 * config/i386/constraints.md (Yc): New register constraint.
19255 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
19256 Use Yc constraint for alternative 2 of operand 0. Remove
19257 preferred_for_speed attribute.
19259 2018-03-14 Richard Biener <rguenther@suse.de>
19261 PR tree-optimization/84830
19262 * tree-ssa-pre.c (compute_antic_aux): Intersect the new ANTIC_IN
19263 with the old one to avoid oscillations.
19265 2018-03-13 Vladimir Makarov <vmakarov@redhat.com>
19268 * lra-assigns.c (find_all_spills_for): Ignore uninteresting
19270 (assign_by_spills): Return a flag of reload assignment failure.
19271 Do not process the reload assignment failures. Do not spill other
19272 reload pseudos if they has the same reg class. Update n if
19274 (lra_assign): Add a return arg. Set up from the result of
19275 assign_by_spills call.
19276 (find_reload_regno_insns, lra_split_hard_reg_for): New functions.
19277 * lra-constraints.c (split_reg): Add a new arg. Use it instead of
19278 usage_insns if it is not NULL.
19279 (spill_hard_reg_in_range): New function.
19280 (split_if_necessary, inherit_in_ebb): Pass a new arg to split_reg.
19281 * lra-int.h (spill_hard_reg_in_range, lra_split_hard_reg_for): New
19282 function prototypes.
19283 (lra_assign): Change prototype.
19284 * lra.c (lra): Add code to deal with fails by splitting hard reg
19287 2018-03-01 Palmer Dabbelt <palmer@sifive.com>
19289 * config/riscv/riscv.opt (mrelax): New option.
19290 * config/riscv/riscv.c (riscv_file_start): Emit ".option
19291 "norelax" when riscv_mrelax is disabled.
19292 * doc/invoke.texi (RISC-V): Document "-mrelax" and "-mno-relax".
19294 2018-03-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
19297 * config/rs6000/rs6000.c (rs6000_reassociation_width): Disable parallel
19298 reassociation for int modes.
19300 2018-03-13 Richard Sandiford <richard.sandiford@linaro.org>
19302 * tree-vect-loop-manip.c (vect_maybe_permute_loop_masks):
19303 Reverse the choice between VEC_UNPACK_LO_EXPR and VEC_UNPACK_HI_EXPR
19305 * config/aarch64/iterators.md (hi_lanes_optab): New int attribute.
19306 * config/aarch64/aarch64-sve.md
19307 (*aarch64_sve_<perm_insn><perm_hilo><mode>): Rename to...
19308 (aarch64_sve_<perm_insn><perm_hilo><mode>): ...this.
19309 (*extend<mode><Vwide>2): Rename to...
19310 (aarch64_sve_extend<mode><Vwide>2): ...this.
19311 (vec_unpack<su>_<perm_hilo>_<mode>): Turn into a define_expand,
19312 renaming the old pattern to...
19313 (aarch64_sve_punpk<perm_hilo>_<mode>): ...this. Only define
19315 (vec_unpack<su>_<perm_hilo>_<SVE_BHSI:mode>): Turn into a
19316 define_expand, renaming the old pattern to...
19317 (aarch64_sve_<su>unpk<perm_hilo>_<SVE_BHSI:mode>): ...this.
19318 (*vec_unpacku_<perm_hilo>_<mode>_no_convert): Delete.
19319 (vec_unpacks_<perm_hilo>_<mode>): Take BYTES_BIG_ENDIAN into
19320 account when deciding which SVE instruction the optab should use.
19321 (vec_unpack<su_optab>_float_<perm_hilo>_vnx4si): Likewise.
19323 2018-03-13 Richard Sandiford <richard.sandiford@linaro.org>
19325 * config/aarch64/aarch64.md (V4_REGNUM, V8_REGNUM, V12_REGNUM)
19326 (V20_REGNUM, V24_REGNUM, V28_REGNUM, P1_REGNUM, P2_REGNUM, P3_REGNUM)
19327 (P4_REGNUM, P5_REGNUM, P6_REGNUM, P8_REGNUM, P9_REGNUM, P10_REGNUM)
19328 (P11_REGNUM, P12_REGNUM, P13_REGNUM, P14_REGNUM): New define_constants.
19329 (tlsdesc_small_<mode>): Turn a define_expand and use
19330 tlsdesc_small_sve_<mode> for SVE. Rename original define_insn to...
19331 (tlsdesc_small_advsimd_<mode>): ...this.
19332 (tlsdesc_small_sve_<mode>): New pattern.
19334 2018-03-13 Richard Sandiford <richard.sandiford@linaro.org>
19336 * config/aarch64/iterators.md (UNSPEC_SMUL_HIGHPART)
19337 (UNSPEC_UMUL_HIGHPART): New constants.
19338 (MUL_HIGHPART): New int iteraor.
19339 (su): Handle UNSPEC_SMUL_HIGHPART and UNSPEC_UMUL_HIGHPART.
19340 * config/aarch64/aarch64-sve.md (<su>mul<mode>3_highpart): New
19342 (*<su>mul<mode>3_highpart): New define_insn.
19344 2018-03-13 Eric Botcazou <ebotcazou@adacore.com>
19347 * ipa-devirt.c (odr_subtypes_equivalent_p): Do not get the ODR type of
19350 2018-03-13 Martin Liska <mliska@suse.cz>
19353 * (sem_item_optimizer::sem_item_optimizer): Initialize new
19355 (sem_item_optimizer::~sem_item_optimizer): Release it.
19356 (sem_item_optimizer::merge_classes): Register variable aliases.
19357 (sem_item_optimizer::fixup_pt_set): New function.
19358 (sem_item_optimizer::fixup_points_to_sets): Likewise.
19359 * ipa-icf.h: Declare new variables and functions.
19361 2018-03-13 Jakub Jelinek <jakub@redhat.com>
19363 PR middle-end/84834
19364 * match.pd ((A & C) != 0 ? D : 0): Use INTEGER_CST@2 instead of
19365 integer_pow2p@2 and test integer_pow2p in condition.
19366 (A < 0 ? C : 0): Similarly for @1.
19368 PR middle-end/84831
19369 * stmt.c (parse_output_constraint): If the CONSTRAINT_LEN (*p, p)
19370 characters starting at p contain '\0' character, don't look beyond
19374 * config/i386/i386.md (round<mode>2): For 387 fancy math, disable
19375 pattern if -ftrapping-math -fno-fp-int-builtin-inexact.
19378 * reg-stack.c (change_stack): Change update_end var from int to
19379 rtx_insn *, if non-NULL don't update just BB_END (current_block), but
19380 also call set_block_for_insn on the newly added insns and rescan.
19383 * config/i386/sse.md (sse2_loadhpd): Use Yv constraint rather than v
19384 on the last operand.
19387 * tree.c (stabilize_reference_1): Return save_expr (e) for
19388 STATEMENT_LIST even if it doesn't have side-effects.
19390 2018-03-12 Jonathan Wakely <jwakely@redhat.com>
19392 * doc/invoke.texi (-mclflushopt): Fix spelling of option.
19394 2018-03-12 Renlin Li <renlin.li@arm.com>
19396 * config/aarch64/aarch64.md (movhf_aarch64): Fix mode argument to
19397 aarch64_output_scalar_simd_mov_immediate.
19399 2018-03-12 Martin Sebor <msebor@redhat.com>
19401 PR tree-optimization/83456
19402 * gimple-fold.c (gimple_fold_builtin_memory_op): Avoid warning
19403 for perfectly overlapping calls to memcpy.
19404 (gimple_fold_builtin_memory_chk): Same.
19405 (gimple_fold_builtin_strcpy): Handle no-warning.
19406 (gimple_fold_builtin_stxcpy_chk): Same.
19407 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Handle no-warning.
19409 2018-03-12 Segher Boessenkool <segher@kernel.crashing.org>
19411 * config/rs6000/rs6000.c (abi_v4_pass_in_fpr): Add bool "named"
19412 parameter. Use it for SFmode.
19413 (rs6000_function_arg_advance_1): Adjust.
19414 (rs6000_function_arg): Adjust.
19415 (rs6000_gimplify_va_arg): Pass false for that new parameter.
19417 2018-03-12 Segher Boessenkool <segher@kernel.crashing.org>
19419 PR rtl-optimization/84169
19420 PR rtl-optimization/84780
19421 * combine.c (can_combine_p): Check for a 2-insn combination whether
19422 the destination register is used between the two insns, too.
19424 2018-03-12 Richard Biener <rguenther@suse.de>
19426 PR tree-optimization/84803
19427 * tree-if-conv.c (ifcvt_memrefs_wont_trap): Don't do anything
19428 for refs DR analysis didn't process.
19430 2018-03-12 Richard Biener <rguenther@suse.de>
19432 PR tree-optimization/84777
19433 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): For
19434 force-vectorize loops ignore whether we are optimizing for size.
19436 2018-03-12 Chung-Ju Wu <jasonwucj@gmail.com>
19438 * config/nds32/nds32.c (nds32_md_asm_adjust): New function.
19439 (TARGET_MD_ASM_ADJUST): Define.
19441 2018-03-12 Monk Chiang <sh.chiang04@gmail.com>
19442 Kito Cheng <kito.cheng@gmail.com>
19443 Chung-Ju Wu <jasonwucj@gmail.com>
19445 * config/nds32/nds32.c (nds32_compute_stack_frame,
19446 nds32_emit_stack_push_multiple, nds32_emit_stack_pop_multiple,
19447 nds32_emit_stack_v3push, nds32_emit_stack_v3pop,
19448 nds32_emit_adjust_frame, nds32_expand_prologue, nds32_expand_epilogue,
19449 nds32_expand_prologue_v3push, nds32_expand_epilogue_v3pop): Refine.
19450 * config/nds32/nds32.h (NDS32_FIRST_CALLEE_SAVE_GPR_REGNUM,
19451 NDS32_LAST_CALLEE_SAVE_GPR_REGNUM, NDS32_V3PUSH_AVAILABLE_P): New.
19452 * config/nds32/nds32.md (prologue, epilogue): Use macro
19453 NDS32_V3PUSH_AVAILABLE_P to do checking.
19455 2018-03-11 Jakub Jelinek <jakub@redhat.com>
19458 * dwarf2out.c (gen_enumeration_type_die): Don't guard adding
19459 DW_AT_declaration for ENUM_IS_OPAQUE on -gdwarf-4 or -gno-strict-dwarf,
19460 but on TYPE_SIZE. Don't do anything for ENUM_IS_OPAQUE if not creating
19461 a new die. Don't set TREE_ASM_WRITTEN if ENUM_IS_OPAQUE. Guard
19462 addition of most attributes on !orig_type_die or the attribute not
19463 being present already. Assert TYPE_VALUES is NULL for ENUM_IS_OPAQUE.
19465 2018-03-11 Kito Cheng <kito.cheng@gmail.com>
19466 Chung-Ju Wu <jasonwucj@gmail.com>
19468 * config/nds32/nds32.c (nds32_cpu_cpp_builtins): Modify to define
19469 __NDS32_VH__ macro.
19470 * config/nds32/nds32.opt (mvh): New option.
19472 2018-03-11 Kito Cheng <kito.cheng@gmail.com>
19473 Chung-Ju Wu <jasonwucj@gmail.com>
19475 * config/nds32/nds32-protos.h (nds32_cpu_cpp_builtins): Declare
19477 * config/nds32/nds32.c (nds32_cpu_cpp_builtins): New function.
19478 * config/nds32/nds32.h (TARGET_CPU_CPP_BUILTINS): Modify its
19481 2018-03-11 Kito Cheng <kito.cheng@gmail.com>
19482 Chung-Ju Wu <jasonwucj@gmail.com>
19484 * config/nds32/nds32-memory-manipulation.c (nds32_expand_strlen): New
19486 * config/nds32/nds32-multiple.md (strlensi): New pattern.
19487 * config/nds32/nds32-protos.h (nds32_expand_strlen): Declare function.
19489 2018-03-11 Monk Chiang <sh.chiang04@gmail.com>
19490 Kito Cheng <kito.cheng@gmail.com>
19491 Chung-Ju Wu <jasonwucj@gmail.com>
19493 * config/nds32/constants.md (unspec_element): Add UNSPEC_FFB,
19494 UNSPEC_FFMISM and UNSPEC_FLMISM.
19495 * config/nds32/nds32-intrinsic.c (bdesc_2arg): Add builtin description
19496 for ffb, ffmism and flmism.
19497 * config/nds32/nds32-intrinsic.md (unspec_ffb): Define new pattern.
19498 (unspec_ffmism): Ditto.
19499 (unspec_flmism): Ditto.
19500 (nds32_expand_builtin_impl): Check if string extension is available.
19501 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_FFB,
19502 NDS32_BUILTIN_FFMISM and NDS32_BUILTIN_FLMISM.
19504 2018-03-10 Vladimir Makarov <vmakarov@redhat.com>
19507 2018-03-09 Vladimir Makarov <vmakarov@redhat.com>
19510 * lra-assigns.c (assign_by_spills): Return a flag of reload
19511 assignment failure. Do not process the reload assignment
19512 failures. Do not spill other reload pseudos if they has the same
19514 (lra_assign): Add a return arg. Set up from the result of
19515 assign_by_spills call.
19516 (find_reload_regno_insns, lra_split_hard_reg_for): New functions.
19517 * lra-constraints.c (split_reg): Add a new arg. Use it instead of
19518 usage_insns if it is not NULL.
19519 (spill_hard_reg_in_range): New function.
19520 (split_if_necessary, inherit_in_ebb): Pass a new arg to split_reg.
19521 * lra-int.h (spill_hard_reg_in_range, lra_split_hard_reg_for): New
19522 function prototypes.
19523 (lra_assign): Change prototype.
19524 * lra.c (lra): Add code to deal with fails by splitting hard reg
19527 2018-03-10 H.J. Lu <hongjiu.lu@intel.com>
19530 * config/i386/i386.opt: Replace Enforcment with Enforcement.
19532 2018-03-10 Alexandre Oliva <aoliva@redhat.com>
19535 * dwarf2out.h (dw_val_class): Add dw_val_class_symview.
19536 (dw_val_node): Add val_symbolic_view.
19537 * dwarf2out.c (dw_line_info_table): Add symviews_since_reset.
19538 (symview_upper_bound): New.
19539 (new_line_info_table): Initialize symviews_since_reset.
19540 (dwarf2out_source_line): Count symviews_since_reset and set
19541 symview_upper_bound.
19542 (dw_val_equal_p): Handle symview.
19543 (add_AT_symview): New.
19544 (print_dw_val): Handle symview.
19545 (attr_checksum, attr_checksum_ordered): Likewise.
19546 (same_dw_val_p, size_of_die): Likewise.
19547 (value_format, output_die): Likewise.
19548 (add_high_low_attributes): Use add_AT_symview for entry_view.
19549 (dwarf2out_finish): Reset symview_upper_bound, clear
19552 2018-03-09 Peter Bergner <bergner@vnet.ibm.com>
19555 * config/rs6000/rs6000.c (rs6000_offsettable_memref_p): New prototype.
19556 Add strict argument and use it.
19557 (rs6000_split_multireg_move): Update for new strict argument.
19558 (mem_operand_gpr): Disallow all non-offsettable addresses.
19559 * config/rs6000/rs6000.md (*movdi_internal64): Use YZ constraint.
19561 2018-03-09 Jakub Jelinek <jakub@redhat.com>
19564 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Mark va_arg_tmp
19565 temporary TREE_ADDRESSABLE before gimplification of BUILT_IN_MEMCPY.
19566 * config/powerpcspe/powerpcspe.c (rs6000_gimplify_va_arg): Likewise.
19569 * tree-inline.c (copy_tree_body_r): For INDIRECT_REF of a remapped
19570 decl, use remap_type if we want to use the type.
19572 2018-03-09 Martin Sebor <msebor@redhat.com>
19574 PR tree-optimization/84526
19575 * gimple-ssa-warn-restrict.c (builtin_memref::set_base_and_offset):
19577 (builtin_access::generic_overlap): Be prepared to handle non-array
19580 2018-03-09 Alexandre Oliva <aoliva@redhat.com>
19582 PR rtl-optimization/84682
19583 * lra-constraints.c (process_address_1): Check is_address flag
19584 for address constraints.
19585 (process_alt_operands): Likewise.
19586 * lra.c (lra_set_insn_recog_data): Pass asm operand locs to
19587 preprocess_constraints.
19588 * recog.h (preprocess_constraints): Add oploc parameter.
19590 * recog.c (preprocess_constraints): Test address_operand for
19591 CT_ADDRESS constraints.
19593 2018-03-09 Vladimir Makarov <vmakarov@redhat.com>
19596 * lra-assigns.c (assign_by_spills): Return a flag of reload
19597 assignment failure. Do not process the reload assignment
19598 failures. Do not spill other reload pseudos if they has the same
19600 (lra_assign): Add a return arg. Set up from the result of
19601 assign_by_spills call.
19602 (find_reload_regno_insns, lra_split_hard_reg_for): New functions.
19603 * lra-constraints.c (split_reg): Add a new arg. Use it instead of
19604 usage_insns if it is not NULL.
19605 (spill_hard_reg_in_range): New function.
19606 (split_if_necessary, inherit_in_ebb): Pass a new arg to split_reg.
19607 * lra-int.h (spill_hard_reg_in_range, lra_split_hard_reg_for): New
19608 function prototypes.
19609 (lra_assign): Change prototype.
19610 * lra.c (lra): Add code to deal with fails by splitting hard reg
19613 2018-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19616 * common/config/arm/arm-common.c (arm_parse_arch_option_name):
19617 Accept complain bool parameter. Only emit errors if it is true.
19618 (arm_parse_cpu_option_name): Likewise.
19619 (arm_target_thumb_only): Adjust callers of the above.
19620 * config/arm/arm-protos.h (arm_parse_cpu_option_name): Adjust
19621 prototype to take a default true bool parameter.
19622 (arm_parse_arch_option_name): Likewise.
19624 2018-03-09 David Malcolm <dmalcolm@redhat.com>
19625 Francois-Xavier Coudert <fxcoudert@gcc.gnu.org>
19629 * Makefile.in (LD_VERSION_SCRIPT_OPTION, LD_SONAME_OPTION): New.
19630 * configure: Regenerate.
19631 * configure.ac ("linker --version-script option"): New.
19632 ("linker soname option"): New.
19634 2018-03-09 Richard Biener <rguenther@suse.de>
19636 PR tree-optimization/84775
19637 * tree-if-conv.c (add_bb_predicate_gimplified_stmts): Delink
19638 immediate uses of predicate stmts and mark them modified.
19641 PR tree-optimization/84178
19642 * tree-if-conv.c (combine_blocks): Move insert_gimplified_predicates
19644 (version_loop_for_if_conversion): Delay update_ssa call.
19645 (tree_if_conversion): Delay update_ssa until after predicate
19648 2018-03-09 Eric Botcazou <ebotcazou@adacore.com>
19651 * config/i386/winnt.c (i386_pe_seh_cold_init): Use small pre-allocation
19652 when the function accesses prior frames.
19654 2018-03-08 Jakub Jelinek <jakub@redhat.com>
19657 * dwarf2out.c (dw_loc_list): If list && loc_list->first->next, call
19658 gen_llsym, otherwise call maybe_gen_llsym.
19660 PR inline-asm/84742
19661 * recog.c (asm_operand_ok): Return 0 if multi-character constraint
19662 has ',' character inside of it.
19664 2018-03-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19667 * config/aarch64/aarch64.md (*compare_cstore<mode>_insn): Mark pattern
19668 as clobbering CC_REGNUM.
19670 2018-03-08 Richard Biener <rguenther@suse.de>
19672 PR middle-end/84552
19673 * tree-scalar-evolution.c: Include tree-into-ssa.h.
19674 (follow_copies_to_constant): Do not follow SSA names registered
19677 2018-03-08 Richard Biener <rguenther@suse.de>
19679 PR tree-optimization/84178
19680 * tree-if-conv.c (combine_blocks): Move insert_gimplified_predicates
19682 (version_loop_for_if_conversion): Delay update_ssa call.
19683 (tree_if_conversion): Delay update_ssa until after predicate
19686 2018-03-08 David Malcolm <dmalcolm@redhat.com>
19688 PR tree-optimization/84178
19689 * tree-if-conv.c (release_bb_predicate): Remove the
19690 the assertion that the stmts have NULL use_ops.
19691 Discard the statements, asserting that they haven't
19692 yet been added to a BB.
19694 2018-03-08 Richard Biener <rguenther@suse.de>
19696 PR tree-optimization/84746
19697 * tree-ssa-pre.c (find_leader_in_sets): Deal with SET1 being NULL.
19698 (phi_translate): Pass in destination ANTIC_OUT set.
19699 (phi_translate_1): Likewise. For a simplified result lookup
19700 a leader in ANTIC_OUT and AVAIL_OUT, not the ANTIC_IN sets.
19701 (phi_translate_set): Adjust.
19702 (do_pre_regular_insertion): Likewise.
19703 (do_pre_partial_partial_insertion): Likewise.
19705 2018-03-08 Martin Liska <mliska@suse.cz>
19707 PR gcov-profile/84735
19708 * doc/gcov.texi: Document usage of profile files.
19709 * gcov-io.h: Document changes in the format.
19711 2018-03-08 Alexandre Oliva <aoliva@redhat.com>
19715 * dwarf2out.c (struct dw_line_info_table): Update comments for
19717 (FORCE_RESET_NEXT_VIEW): New.
19718 (FORCE_RESETTING_VIEW_P): New.
19719 (RESETTING_VIEW_P): Check for -1 too.
19720 (ZERO_VIEW_P): Likewise.
19721 (new_line_info_table): Force-reset next view.
19722 (dwarf2out_begin_function): Likewise.
19723 (dwarf2out_source_line): Simplify zero_view_p initialization.
19724 Test FORCE_RESETTING_VIEW_P and RESETTING_VIEW_P instead of
19725 view directly. Omit view when omitting .loc at line 0.
19727 2018-03-08 Jakub Jelinek <jakub@redhat.com>
19729 PR tree-optimization/84740
19730 * tree-switch-conversion.c (process_switch): Call build_constructors
19731 only if info.phi_count is non-zero.
19733 PR tree-optimization/84739
19734 * tree-tailcall.c (find_tail_calls): Check call arguments against
19735 DECL_ARGUMENTS (current_function_decl) rather than
19736 DECL_ARGUMENTS (func) when checking for tail recursion.
19738 2018-03-07 Jakub Jelinek <jakub@redhat.com>
19740 * doc/contrib.texi: Add entries for Martin Liska, David Malcolm,
19741 Marek Polacek, extend Vladimir Makarov's, Jonathan Wakely's and
19742 Volker Reichelt's entry and add entries for people that perform
19743 GCC fuzzy testing and report numerous bugs.
19745 2018-03-07 Segher Boessenkool <segher@kernel.crashing.org>
19748 * config/rs6000/rs6000.c (rs6000_elf_in_small_data_p): Don't put
19749 readonly data in sdata, if that is disabled.
19750 * config/rs6000/sysv4.opt (mreadonly-in-sdata): New option.
19751 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
19752 -mreadonly-in-sdata option.
19754 2018-03-07 Martin Sebor <msebor@redhat.com>
19756 PR tree-optimization/84468
19757 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Consider successor
19758 basic block when looking for nul assignment.
19760 2018-03-07 Eric Botcazou <ebotcazou@adacore.com>
19763 * except.h (output_function_exception_table): Adjust prototype.
19764 * except.c (output_function_exception_table): Remove FNNAME parameter
19765 and add SECTION parameter. Ouput one part of the table at a time.
19766 * final.c (final_scan_insn_1) <NOTE_INSN_SWITCH_TEXT_SECTIONS>: Output
19767 the first part of the exception table and emit unwind directives.
19768 * config/i386/i386-protos.h (i386_pe_end_cold_function): Declare.
19769 (i386_pe_seh_cold_init): Likewise.
19770 * config/i386/cygming.h (ASM_DECLARE_COLD_FUNCTION_NAME): New macro.
19771 (ASM_DECLARE_COLD_FUNCTION_SIZE): Likewise.
19772 * config/i386/i386.c (x86_expand_epilogue): Fix wording in comment.
19773 (ix86_output_call_insn): Emit a nop in one more case for SEH.
19774 * config/i386/winnt.c: Include except.h.
19775 (struct seh_frame_state): Add reg_offset, after_prologue and
19776 in_cold_section fields.
19777 (i386_pe_seh_end_prologue): Set seh->after_prologue.
19778 (i386_pe_seh_cold_init): New function.
19779 (i386_pe_seh_fini): Add COLD parameter and bail out if it is not equal
19780 to seh->in_cold_section.
19781 (seh_emit_push): Record the offset of the push.
19782 (seh_emit_save): Record the offet of the save.
19783 (i386_pe_seh_unwind_emit): Deal with NOTE_INSN_SWITCH_TEXT_SECTIONS.
19784 Test seh->after_prologue to disregard the epilogue.
19785 (i386_pe_end_function): Pass FALSE to i386_pe_seh_fini.
19786 (i386_pe_end_cold_function): New function.
19788 2018-03-07 Jakub Jelinek <jakub@redhat.com>
19791 * config/aarch64/predicates.md (aarch64_simd_reg_or_zero): Use
19792 aarch64_simd_or_scalar_imm_zero rather than aarch64_simd_imm_zero.
19795 * gimple-expr.c (create_tmp_var_raw): Set DECL_NAMELESS flag
19797 * tree-pretty-print.c (dump_decl_name): For TDF_COMPARE_DEBUG,
19798 don't print names of DECL_NAMELESS DECL_IGNORED_P decls.
19800 PR middle-end/84723
19801 * multiple_target.c: Include tree-inline.h and intl.h.
19802 (expand_target_clones): Diagnose and fail if node->definition and
19803 !tree_versionable_function_p (node->decl).
19805 2018-03-06 John David Anglin <danglin@gcc.gnu.org>
19807 * config/pa/pa.h (ASM_GENERATE_INTERNAL_LABEL): Revise to use
19809 (ASM_OUTPUT_ADDR_VEC_ELT): Revise for above change.
19810 (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
19811 * config/pa/pa64-hpux.h (ASM_GENERATE_INTERNAL_LABEL): Revise as above.
19813 2018-03-06 Jakub Jelinek <jakub@redhat.com>
19816 * combine.c (try_combine): Use reg_or_subregno instead of handling
19817 just paradoxical SUBREGs and REGs.
19819 2018-03-06 Claudiu Zissulescu <claziss@synopsys.com>
19821 * config/arc/arc.c (arc_finalize_pic): Remove function.
19822 (arc_must_save_register): We use single base PIC register, remove
19823 checks to save/restore the PIC register.
19824 (arc_expand_prologue): Likewise.
19825 * config/arc/arc-protos.h (arc_set_default_type_attributes):
19827 (arc_verify_short): Likewise.
19828 (arc_attr_type): Likewise.
19829 * config/arc/arc.c (arc_set_default_type_attributes): Remove.
19830 (walk_stores): Likewise.
19831 (arc_address_cost): Make it static.
19832 (arc_verify_short): Likewise.
19833 (branch_dest): Likewise.
19834 (arc_attr_type): Likewise.
19835 * config/arc/arc.c (TARGET_ADJUST_INSN_LENGTH): Remove.
19836 (TARGET_INSN_LENGTH_PARAMETERS): Likewise.
19837 (arc_final_prescan_insn): Remove inserting the nops due to
19838 hardware hazards. It is done in reorg step.
19839 (insn_length_variant_t): Remove.
19840 (insn_length_parameters_t): Likewise.
19841 (arc_insn_length_parameters): Likewise.
19842 (arc_get_insn_variants): Likewise.
19843 * config/arc/arc.h (TARGET_UPSIZE_DBR): Remove.
19845 2018-03-06 Jakub Jelinek <jakub@redhat.com>
19847 PR inline-asm/84683
19848 * reg-stack.c (move_for_stack_reg): If any_malformed_asm, avoid
19851 PR tree-optimization/84687
19852 * omp-simd-clone.c (simd_clone_create): Clear DECL_BUILT_IN_CLASS
19854 * match.pd (pow(C,x)*expN(y) -> expN(logN(C)*x+y)): New optimization.
19856 2018-03-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19858 * config/rs6000/rs6000-builtin.def (rs6000_speculation_barrier):
19859 Rename to ppc_speculation_barrier.
19860 * config/rs6000/rs6000.c (rs6000_init_builtins): Rename builtin to
19861 __builtin_ppc_speculation_barrier.
19863 2018-03-05 Jakub Jelinek <jakub@redhat.com>
19866 * combine.c (combine_simplify_rtx): Don't try to simplify if
19867 if_then_else_cond returned non-NULL, but either true_rtx or false_rtx
19870 2018-03-05 Segher Boessenkool <segher@kernel.crashing.org>
19872 * config/rs6000/rs6000.c (rs6000_loop_align): Don't align tiny loops
19873 to 32 bytes when compiling for POWER9.
19875 2018-03-05 Jakub Jelinek <jakub@redhat.com>
19878 * config/i386/i386.c (ix86_function_ok_for_sibcall): Check for
19879 regparm >= 3 with no arg reg available also for calls with
19880 flag_force_indirect_call. Pass decl to ix86_function_regparm.
19883 * config/i386/sse.md (*<code><mode>3): Replace <mask_prefix3> with
19885 (*<plusminus_insn><mode>3): Likewise. Remove <mask_operand3> uses.
19887 2018-03-05 Peter Bergner <bergner@vnet.ibm.com>
19890 * config/rs6000/vector.md (mov<mode>): Disallow altivec memory operands.
19892 2018-03-05 Richard Biener <rguenther@suse.de>
19894 PR tree-optimization/84486
19895 * tree-ssa-pre.c (create_expression_by_pieces): Remove dead code.
19896 When inserting a __builtin_assume_aligned call set the LHS
19897 SSA name alignment info accordingly.
19899 2018-03-05 Wilco Dijkstra <wdijkstr@arm.com>
19901 PR tree-optimization/84114
19902 * config/aarch64/aarch64.c (aarch64_reassociation_width)
19903 Avoid reassociation of FLOAT_MODE addition.
19905 2018-03-05 Olga Makhotina <olga.makhotina@intel.com>
19907 * common/config/i386/i386-common.c (OPTION_MASK_ISA_PCONFIG_SET,
19908 OPTION_MASK_ISA_PCONFIG_UNSET, OPTION_MASK_ISA_WBNOINVD_SET,
19909 OPTION_MASK_ISA_WBNOINVD_UNSET): New definitions.
19910 (ix86_handle_option): Handle -mpconfig and -mwbnoinvd.
19911 * config.gcc (pconfigintrin.h, wbnoinvdintrin.h) : Add headers.
19912 * config/i386/cpuid.h (bit_PCONFIG, bit_WBNOINVD): New.
19913 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mpconfig
19915 * config/i386/i386-builtin.def (__builtin_ia32_wbnoinvd,
19916 __builtin_ia32_wbinvd): New builtins.
19917 (SPECIAL_ARGS2): New.
19918 * config/i386/i386-c.c (__WBNOINVD__, __PCONFIG__): New.
19919 (SPECIAL_ARGS2): New.
19920 * config/i386/i386.c (ix86_target_string): Add -mpconfig and -mwbnoinvd.
19921 (ix86_valid_target_attribute_inner_p): Ditto.
19922 (ix86_init_mmx_sse_builtins): Add special_args2.
19923 * config/i386/i386.h (TARGET_PCONFIG, TARGET_PCONFIG_P, TARGET_WBNOINVD,
19924 TARGET_WBNOINVD_P): New.
19925 * config/i386/i386.md (UNSPECV_WBINVD, UNSPECV_WBNOINVD): New.
19926 (define_insn "wbinvd", define_insn "wbnoinvd"): New.
19927 * config/i386/i386.opt: Add -mpconfig and -mwbnoinvd.
19928 * config/i386/immintrin.h (_wbinvd): New intrinsic.
19929 * config/i386/pconfigintrin.h: New file.
19930 * config/i386/wbnoinvdintrin.h: Ditto.
19931 * config/i386/x86intrin.h: Add headers pconfigintrin.h and
19933 * doc/invoke.texi (-mpconfig, -mwbnoinvd): New.
19935 2018-03-05 Richard Biener <rguenther@suse.de>
19937 PR tree-optimization/84670
19938 * tree-ssa-pre.c (struct bb_bitmap_sets): Add visited_with_visited_succs
19940 (BB_VISITED_WITH_VISITED_SUCCS): New define.
19941 (compute_antic): Initialize BB_VISITED_WITH_VISITED_SUCCS.
19942 (compute_antic_aux): Only assert the number of values in ANTIC_IN
19943 doesn't grow if all successors (recursively) were visited at least
19946 2018-03-05 Richard Biener <rguenther@suse.de>
19948 PR tree-optimization/84650
19949 * tree-ssa-loop-im.c (pass_lim::execute): Reset the SCEV cache
19950 if executed in the loop pipeline.
19952 2018-03-05 Sandra Loosemore <sandra@codesourcery.com>
19954 * doc/configfiles.texi (Configuration Files): Move info about
19955 conditionalizing $target-protos.h to...
19956 * doc/sourcebuild.texi (Back End): Here. Explain how $target.h
19957 differs from $target-protos.h.
19959 2018-03-05 Kito Cheng <kito.cheng@gmail.com>
19960 Chung-Ju Wu <jasonwucj@gmail.com>
19962 * config/nds32/nds32-protos.h (nds32_expand_setmem): Declare.
19963 * config/nds32/nds32-multiple.md (setmemsi): Define.
19964 * config/nds32/nds32-memory-manipulation.c
19965 (nds32_gen_dup_4_byte_to_word_value): New.
19966 (emit_setmem_word_loop): New.
19967 (emit_setmem_byte_loop): New.
19968 (nds32_expand_setmem_loop): New.
19969 (nds32_expand_setmem_loop_v3m): New.
19970 (nds32_expand_setmem_unroll): New.
19971 (nds32_expand_setmem): New.
19973 2018-03-04 Kito Cheng <kito.cheng@gmail.com>
19974 Chung-Ju Wu <jasonwucj@gmail.com>
19976 * config/nds32/nds32-memory-manipulation.c
19977 (nds32_emit_load_store): New.
19978 (nds32_emit_post_inc_load_store): New.
19979 (nds32_emit_mem_move): New.
19980 (nds32_emit_mem_move_block): New.
19981 (nds32_expand_movmemsi_loop_unknown_size): New.
19982 (nds32_expand_movmemsi_loop_known_size): New.
19983 (nds32_expand_movmemsi_loop): New.
19984 (nds32_expand_movmemsi_unroll): New.
19985 (nds32_expand_movmemqi): Rename ...
19986 (nds32_expand_movmemsi): ... to this.
19987 * config/nds32/nds32-multiple.md (movmemqi): Rename ...
19988 (movmemsi): ... to this.
19989 * config/nds32/nds32-protos.h (nds32_expand_movmemqi): Rename ...
19990 (nds32_expand_movmemsi): ... to this.
19992 2018-03-04 Kito Cheng <kito.cheng@gmail.com>
19993 Monk Chiang <sh.chiang04@gmail.com>
19994 Chung-Ju Wu <jasonwucj@gmail.com>
19996 * config/nds32/nds32-protos.h
19997 (nds32_expand_load_multiple): New arguments.
19998 (nds32_expand_store_multiple): Ditto.
19999 (nds32_valid_multiple_load_store): Rename ...
20000 (nds32_valid_multiple_load_store_p): ... to this.
20001 * config/nds32/nds32-memory-manipulation.c
20002 (nds32_expand_load_multiple): Refine implementation.
20003 (nds32_expand_store_multiple): Ditto.
20004 * config/nds32/nds32-multiple.md
20005 (load_multiple): Update nds32_expand_load_multiple interface.
20006 (store_multiple): Update nds32_expand_store_multiple interface.
20007 * config/nds32/nds32-predicates.c
20008 (nds32_valid_multiple_load_store): Rename ...
20009 (nds32_valid_multiple_load_store_p): ... to this and refine
20011 * config/nds32/predicates.md
20012 (nds32_load_multiple_and_update_address_operation): New predicate.
20013 (nds32_store_multiple_and_update_address_operation): New predicate.
20015 2018-03-04 Kito Cheng <kito.cheng@gmail.com>
20016 Chung-Ju Wu <jasonwucj@gmail.com>
20018 * config/nds32/nds32.md (type): Add load_multiple and store_multiple.
20019 (combo): New attribute.
20020 * config/nds32/nds32-multiple.md: Refine patterns with new attributes.
20022 2018-03-03 Chung-Ju Wu <jasonwucj@gmail.com>
20024 * config/nds32/nds32.opt: Change -mcmodel= default value.
20026 2018-03-03 Kito Cheng <kito.cheng@gmail.com>
20027 Monk Chiang <sh.chiang04@gmail.com>
20028 Chung-Ju Wu <jasonwucj@gmail.com>
20030 * config/nds32/constants.md (unspec_element): New enum.
20031 * config/nds32/constraints.md (Umw): New constraint.
20032 * config/nds32/nds32-intrinsic.c: Add more builtin functions.
20033 * config/nds32/nds32-intrinsic.md: Likewise.
20034 * config/nds32/nds32-md-auxiliary.c (nds32_regno_to_enable4): New.
20035 (nds32_valid_smw_lwm_base_p): New.
20036 (nds32_output_smw_single_word): New.
20037 (nds32_output_lmw_single_word): New.
20038 (nds32_expand_unaligned_load): New.
20039 (nds32_expand_unaligned_store): New.
20040 * config/nds32/nds32-protos.h (nds32_valid_smw_lwm_base_p): Declare.
20041 (nds32_output_smw_single_word): Declare.
20042 (nds32_output_lmw_single_word): Declare.
20043 (nds32_expand_unaligned_load): Declare.
20044 (nds32_expand_unaligned_store): Declare.
20045 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_UALOAD_HW,
20046 NDS32_BUILTIN_UALOAD_W, NDS32_BUILTIN_UALOAD_DW,
20047 NDS32_BUILTIN_UASTORE_HW, NDS32_BUILTIN_UASTORE_W,
20048 NDS32_BUILTIN_UASTORE_DW.
20049 * config/nds32/predicates.md (nds32_lmw_smw_base_operand): New
20052 2018-03-03 Monk Chiang <sh.chiang04@gmail.com>
20053 Kito Cheng <kito.cheng@gmail.com>
20054 Chung-Ju Wu <jasonwucj@gmail.com>
20056 * config/nds32/nds32-intrinsic.c
20057 (nds32_expand_builtin_null_ftype_reg): Delete.
20058 (nds32_expand_builtin_reg_ftype_imm): Ditto.
20059 (nds32_expand_builtin_null_ftype_reg_imm): Ditto.
20060 (nds32_read_argument): New.
20061 (nds32_legitimize_target): Ditto.
20062 (nds32_legitimize_argument): Ditto.
20063 (nds32_check_constant_argument): Ditto.
20064 (nds32_expand_unop_builtin): Ditto.
20065 (nds32_expand_unopimm_builtin): Ditto.
20066 (nds32_expand_binop_builtin): Ditto.
20067 (nds32_builtin_decl_impl): Ditto.
20068 (builtin_description): Ditto.
20069 (nds32_expand_builtin_impl): Rewrite with new infrastructure.
20070 (nds32_init_builtins_impl): Ditto.
20071 * config/nds32/nds32.c (TARGET_BUILTIN_DECL): Define.
20072 (nds32_builtin_decl): New.
20073 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_COUNT.
20074 * config/nds32/nds32-protos.h (nds32_builtin_decl_impl): Declare.
20076 2018-03-02 Jeff Law <law@redhat.com>
20078 * reorg.c (stop_search_p): Handle DEBUG_INSN.
20079 (redundant_insn, fill_simple_delay_slots): Likewise.
20080 (fill_slots_from_thread): Likewise.
20081 * resource.c (mark_referenced_resources): Likewise.
20082 (mark_set_resources, find_dead_or_set_registers): Likewise.
20084 2018-03-02 Jakub Jelinek <jakub@redhat.com>
20086 * substring-locations.h (format_warning_va): Formatting fix for
20087 ATTRIBUTE_GCC_DIAG.
20088 (format_warning_at_substring): Fix up ATTRIBUTE_GCC_DIAG second
20090 (format_warning_n_va, format_warning_at_substring_n): New prototypes.
20091 * substring-locations.c: Include intl.h.
20092 (format_warning_va): Turned into small wrapper around
20093 format_warning_n_va, renamed to ...
20094 (format_warning_n_va): ... this, add N and PLURAL_GMSGID arguments,
20095 rename GMSGID to SINGULAR_GMSGID, if SINGULAR_GMSGID != PLURAL_GMSGID,
20097 (format_warning_at_substring_n): New function.
20098 * gimple-ssa-sprintf.c: Remove GCC diagnostic ignored pragma.
20099 (fmtwarn): Add ATTRIBUTE_GCC_DIAG. Turn into a copy of
20100 format_warning_at_substring with just a shorter name instead of
20101 const function pointer.
20102 (fmtwarn_n): New function.
20103 (maybe_warn, format_directive, parse_directive): Use fmtwarn_n where
20104 appropriate, get rid of all the fmtstr temporaries, move conditionals
20105 with G_() wrapped string literals directly into fmtwarn arguments,
20106 cast dir.len to (int), formatting fixes.
20108 2018-03-02 Thomas Schwinge <thomas@codesourcery.com>
20110 * doc/invoke.texi: Remove "Cilk Plus" references.
20112 2018-03-02 Jakub Jelinek <jakub@redhat.com>
20113 Richard Biener <rguenther@suse.de>
20116 * expr.c (expand_expr_real_1) <case CALL_EXPR>: Don't emit diagnostics
20117 for error or warning attributes if CALL_FROM_THUNK_P is set.
20120 2018-03-02 Jakub Jelinek <jakub@redhat.com>
20123 * config/pa/pa.h (TARGET_CPU_CPP_BUILTINS): Predefine
20124 __SIZEOF_128__ macro if HPUX_LONG_DOUBLE_LIBRARY.
20127 * config/ia64/ia64.h (TARGET_CPU_CPP_BUILTINS): Predefine
20128 __SIZEOF_{FPREG,FLOAT{80,128}}__ macros.
20130 * predict.c (test_prediction_value_range): Use PROB_UNINITIALIZED
20131 instead of -1U in last predictors element's probability member.
20133 2018-03-02 Eric Botcazou <ebotcazou@adacore.com>
20136 * ipa-devirt.c (odr_subtypes_equivalent_p): Get the ODR type of both
20137 arguments if they are comparable.
20139 2018-03-02 Richard Sandiford <richard.sandiford@linaro.org>
20141 PR tree-optimization/84634
20142 * tree-vect-stmts.c (vectorizable_store, vectorizable_load): Replace
20143 masks and masked_loop_p with a single loop_masks, making sure it's
20144 null for bb vectorization.
20146 2018-03-02 Richard Sandiford <richard.sandiford@linaro.org>
20148 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence)
20149 (vect_analyze_data_ref_access): Use loop->safe_len rather than
20150 loop->force_vectorize to check whether there is no alias.
20152 2018-03-02 Jakub Jelinek <jakub@redhat.com>
20155 * rtl.h (prev_real_nondebug_insn, next_real_nondebug_insn): New
20157 * emit-rtl.c (next_real_insn, prev_real_insn): Fix up function
20159 (next_real_nondebug_insn, prev_real_nondebug_insn): New functions.
20160 * cfgcleanup.c (try_head_merge_bb): Use prev_real_nondebug_insn
20161 instead of a loop around prev_real_insn.
20162 * combine.c (move_deaths): Use prev_real_nondebug_insn instead of
20165 PR inline-asm/84625
20166 * config/i386/i386.c (ix86_print_operand): Use conditional
20167 output_operand_lossage instead of gcc_assert if CONST_VECTOR is not
20170 2018-03-02 Richard Biener <rguenther@suse.de>
20172 PR tree-optimization/84427
20173 * tree-ssa-pre.c (bitmap_remove_expr_from_set): Remove.
20174 (bitmap_set_subtract_values): Rewrite to handle multiple
20177 (prune_clobbered_mems): Likewise.
20178 (phi_translate): Take edge instead of pred/phiblock.
20179 (phi_translate_1): Likewise.
20180 (phi_translate_set): Likewise. Insert all translated
20181 exprs for a value into the set, keeping possibly multiple
20182 expressions per value.
20183 (compute_antic_aux): Adjust for phi_translate changes.
20184 When intersecting union the expressions and prune those
20185 not in the final value set, keeping possibly multiple
20186 expressions per value. Do not use value-insertion
20187 for unioning ANTIC_OUT U EXP_GEN - TMP_GEN but merge
20188 all expressions. Add verification that the value-sets
20189 only shrink during iteration.
20190 (compute_partial_antic_aux): Adjust for the phi_translate changes.
20191 (do_pre_regular_insertion): Likewise.
20192 (do_pre_partial_partial_insertion): Likewise.
20194 2018-03-02 Richard Biener <rguenther@suse.de>
20197 * config/darwin.c (saved_debug_info_level): New static global.
20198 (darwin_asm_lto_start): Disable debug info generation for LTO out.
20199 (darwin_asm_lto_end): Restore debug info generation settings.
20201 2018-03-01 Martin Liska <mliska@suse.cz>
20204 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
20205 volatile arguments.
20207 2018-03-01 Richard Biener <rguenther@suse.de>
20210 * dwarf2out.c (gen_variable_die): Properly handle late VLA
20211 type annotation with LTO when debug was disabled at compile-time.
20213 2018-03-01 Matthew Fortune <mfortune@gmail.com>
20215 * config/mips/mips.c (mips_final_prescan_insn): Fix incorrect
20217 (mips_final_postscan_insn): Likewise.
20219 2018-03-01 Richard Sandiford <richard.sandiford@linaro.org>
20221 PR rtl-optimization/84528
20222 * alias.c (init_alias_target): Add commentary.
20223 (init_alias_analysis): Only give HARD_FRAME_POINTER_REGNUM
20224 a unique base value if the frame pointer is not eliminated
20225 to the stack pointer.
20227 2018-03-01 Tom de Vries <tom@codesourcery.com>
20229 PR rtl-optimization/83327
20230 * lra-int.h (hard_regs_spilled_into): Declare.
20231 * lra.c (hard_regs_spilled_into): Define.
20232 (init_reg_info): Init hard_regs_spilled_into.
20233 * lra-spills.c (assign_spill_hard_regs): Update hard_regs_spilled_into.
20234 * lra-lives.c (make_hard_regno_born, make_hard_regno_dead)
20235 (process_bb_lives): Handle hard_regs_spilled_into.
20236 (lra_create_live_ranges_1): Before doing liveness propagation, clear
20237 regs in all_hard_regs_bitmap if set in hard_regs_spilled_into.
20239 2018-02-28 David Edelsohn <dje.gcc@gmail.com>
20241 * config.gcc (powerpc-ibm-aix7.1.*): New stanza.
20242 (powerpc-ibm-aix[789]*): Default to AIX 7.2.
20243 * config/rs6000/aix71.h (TARGET_DEFAULT): Revert to Power4 ISA.
20244 * config/rs6000/aix72.h: New file.
20246 2018-02-28 Jakub Jelinek <jakub@redhat.com>
20248 * gimple-ssa-warn-restrict.c (maybe_diag_overlap): Use warning_n
20249 instead of warning_at with conditional singular and plural messages
20253 * stor-layout.c (update_alignment_for_field): For
20254 targetm.ms_bitfield_layout_p (rli->t), if !is_bitfield
20255 && !DECL_PACKED (field), do the alignment update, just use
20256 only desired_align instead of MAX (type_align, desired_align)
20258 (place_field): Don't do known_align < desired_align handling
20259 early if targetm.ms_bitfield_layout_p (rli->t) and rli->prev_field
20260 is non-NULL, instead do it after rli->prev_field handling and
20261 only if not within a bitfield word. For DECL_PACKED (field)
20262 use type_align of BITS_PER_UNIT.
20264 2018-02-28 Eric Botcazou <ebotcazou@adacore.com>
20266 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Remove
20267 superfluous parentheses and trailing spaces.
20269 2018-02-28 Richard Biener <rguenther@suse.de>
20271 PR tree-optimization/84584
20272 * graphite-scop-detection.c (scop_detection::add_scop): Discard
20273 SCoPs with fake exit edge.
20275 2018-02-28 Martin Liska <mliska@suse.cz>
20278 * timevar.c (timer::print): Fix format to properly print 100%
20281 2018-02-28 Richard Biener <rguenther@suse.de>
20283 PR middle-end/84607
20284 * genmatch.c (capture_info::walk_match): Do not mark
20285 captured expressions without operands as expr_p given
20286 they act more like predicates and should be subject to
20287 "lost tail" side-effect preserving.
20289 2018-02-28 Alexandre Oliva <aoliva@redhat.com>
20291 PR rtl-optimization/81611
20292 * auto-inc-dec.c (attempt_change): Move dead note from
20293 mem_insn if it's the next use of regno
20294 (find_address): Take address use of reg holding
20295 non-incremented value. Add parm to limit search to the named
20297 (merge_in_block): Attempt to use a mem insn that is the next
20298 use of the original regno.
20300 2018-02-27 Martin Sebor <msebor@redhat.com>
20303 * doc/invoke.texi (-Wmissing-attributes): New option.
20304 * print-tree.c (print_node): Handle DECL_UNINLINABLE.
20306 2018-02-27 Martin Sebor <msebor@redhat.com>
20308 PR translation/84207
20309 * diagnostic-core.h (warning_n, error_n, inform_n): Change
20310 n argument to unsigned HOST_WIDE_INT.
20311 * diagnostic.c (warning_n, error_n, inform_n): Ditto.
20312 (diagnostic_n_impl): Ditto. Handle arguments in excess of LONG_MAX.
20313 * gimple-ssa-sprintf.c (format_directive): Simplify inform_n call.
20314 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Use warning_n.
20316 2018-02-27 Richard Biener <rguenther@suse.de>
20318 PR tree-optimization/84512
20319 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost):
20320 Do not use the estimate returned from record_stmt_cost for
20321 the scalar iteration cost but sum properly using add_stmt_cost.
20323 2018-02-27 Richard Biener <rguenther@suse.de>
20325 PR tree-optimization/84466
20326 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
20327 Adjust last change to less strictly validate use operands.
20329 2018-02-27 Martin Liska <mliska@suse.cz>
20331 PR gcov-profile/84548
20332 * gcov.c (process_file): Allow partial overlap and consider it
20333 also as group functions.
20334 (output_lines): Properly calculate range of lines for a group.
20336 2018-02-27 Martin Liska <mliska@suse.cz>
20338 * timevar.c (timer::print_row): Remove 'usr', 'sys', 'wall' and
20339 'ggc' suffixes. Change first column width.
20340 (timer::print): Fix formatting of the column.
20342 2018-02-27 Alexandre Oliva <aoliva@redhat.com>
20344 * tree-ssa-live.c (remove_unused_scope_block_p): Do not
20345 preserve inline entry blocks for the sake of debug inline
20346 entry point markers alone.
20347 (remove_unused_locals): Suggest in comments a better place to
20348 force the preservation of inline entry blocks that are
20349 otherwise unused, but do not preserve them.
20351 2018-02-26 H.J. Lu <hongjiu.lu@intel.com>
20353 * config/i386/i386.c (ix86_output_indirect_jmp): Update comments.
20355 2018-02-26 H.J. Lu <hongjiu.lu@intel.com>
20358 * config/i386/constraints.md (Bs): Replace
20359 ix86_indirect_branch_register with
20360 TARGET_INDIRECT_BRANCH_REGISTER.
20362 * config/i386/i386.md (indirect_jump): Likewise.
20363 (tablejump): Likewise.
20364 (*sibcall_memory): Likewise.
20365 (*sibcall_value_memory): Likewise.
20366 Peepholes of indirect call and jump via memory: Likewise.
20367 (*sibcall_GOT_32): Disallowed for TARGET_INDIRECT_BRANCH_REGISTER.
20368 (*sibcall_value_GOT_32): Likewise.
20369 * config/i386/predicates.md (indirect_branch_operand): Likewise.
20370 (GOT_memory_operand): Likewise.
20371 (call_insn_operand): Likewise.
20372 (sibcall_insn_operand): Likewise.
20373 (GOT32_symbol_operand): Likewise.
20374 * config/i386/i386.h (TARGET_INDIRECT_BRANCH_REGISTER): New.
20376 2018-02-26 Eric Botcazou <ebotcazou@adacore.com>
20378 PR rtl-optimization/83496
20379 * reorg.c (steal_delay_list_from_target): Change REDUNDANT array from
20380 booleans to RTXes. Call fix_reg_dead_note on every non-null element.
20381 (steal_delay_list_from_fallthrough): Call fix_reg_dead_note on a
20382 redundant insn, if any.
20383 (relax_delay_slots): Likewise.
20384 (update_reg_unused_notes): Rename REDUNDANT_INSN to OTHER_INSN.
20386 2018-02-26 Richard Sandiford <richard.sandiford@linaro.org>
20388 PR tree-optimization/83965
20389 * tree-vect-patterns.c (vect_reassociating_reduction_p): Assume
20390 that grouped statements are part of a reduction chain. Return
20391 true if the statement is not marked as a reduction itself but
20392 is part of a group.
20393 (vect_recog_dot_prod_pattern): Don't check whether the statement
20394 is part of a group here.
20395 (vect_recog_sad_pattern): Likewise.
20396 (vect_recog_widen_sum_pattern): Likewise.
20398 2018-02-26 Eric Botcazou <ebotcazou@adacore.com>
20401 * final.c (rest_of_clean_state): Also look for calls inside sequences.
20403 2018-02-26 H.J. Lu <hongjiu.lu@intel.com>
20406 * config/i386/i386-protos.h (ix86_output_indirect_jmp): Remove
20408 (ix86_output_indirect_function_return): New prototype.
20409 (ix86_split_simple_return_pop_internal): Likewise.
20410 * config/i386/i386.c (indirect_return_via_cx): New.
20411 (indirect_return_via_cx_bnd): Likewise.
20412 (indirect_thunk_name): Handle return va CX_REG.
20413 (output_indirect_thunk_function): Create alias for
20414 __x86_return_thunk_[re]cx and __x86_return_thunk_[re]cx_bnd.
20415 (ix86_output_indirect_jmp): Remove the bool argument.
20416 (ix86_output_indirect_function_return): New function.
20417 (ix86_split_simple_return_pop_internal): Likewise.
20418 * config/i386/i386.md (*indirect_jump): Don't pass false
20419 to ix86_output_indirect_jmp.
20420 (*tablejump_1): Likewise.
20421 (simple_return_pop_internal): Change it to define_insn_and_split.
20422 Call ix86_split_simple_return_pop_internal to split it for
20423 -mfunction-return=.
20424 (simple_return_indirect_internal): Call
20425 ix86_output_indirect_function_return instead of
20426 ix86_output_indirect_jmp.
20428 2018-02-26 Jakub Jelinek <jakub@redhat.com>
20431 * vec.h (vec_default_construct): For BROKEN_VALUE_INITIALIZATION use
20432 memset and value initialization afterwards.
20434 2018-02-26 Christophe Lyon <christophe.lyon@linaro.org>
20436 * Makefile.in (lto-wrapper): Use ALL_LINKERFLAGS.
20438 2018-02-26 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
20441 * common/config/aarch64/aarch64-common.c
20442 (aarch_option_optimization_table[]): Switch
20443 off fomit-frame-pointer
20445 2018-02-26 Kito Cheng <kito.cheng@gmail.com>
20446 Chung-Ju Wu <jasonwucj@gmail.com>
20448 * config/nds32/nds32-multiple.md (load_multiple): Disallow
20450 (store_multiple): Ditto.
20452 2018-02-26 Kito Cheng <kito.cheng@gmail.com>
20454 * config.gcc: Add --with-cpu support for nds32 target.
20455 * config/nds32/nds32-opts.h (nds32_cpu_type): New.
20456 * config/nds32/nds32.opt: Add -mcpu= option.
20458 2018-02-25 Segher Boessenkool <segher@kernel.crashing.org>
20460 * config/rs6000/rs6000.opt (mvrsave=no, mvrsave=yes, isel=no,
20461 isel=yes): Warn for these deprecated options.
20463 2018-02-23 David Edelsohn <dje.gcc@gmail.com>
20465 * config/rs6000/aix71.h (TARGET_DEFAULT): Change to
20466 ISA_2_5_MASKS_EMBEDDED.
20468 2018-02-23 Jakub Jelinek <jakub@redhat.com>
20470 * ipa-prop.c (ipa_vr_ggc_hash_traits::hash): Hash p->min and
20471 p->max as pointers rather than using iterative_hash_expr.
20473 2018-02-23 Carl Love <cel@us.ibm.com>
20475 * config/rs6000/rs6000-builtin.def: Change VSIGNED2 and VUNSIGNED2
20476 macro expansions from BU_VSX_2 to BU_P8V_VSX_2 and BU_VSX_OVERLOAD_2 to
20478 * config/rs6000/rs6000-c.c: Change VSX_BUILTIN_VEC_VSIGNED2 to
20479 P8V_BUILTIN_VEC_VSIGNED2. Change VSX_BUILTIN_VEC_VUNSIGNED2 to
20480 P8V_BUILTIN_VEC_VUNSIGNED2.
20482 2018-02-22 Vladimir Makarov <vmakarov@redhat.com>
20485 * lra-int.h (LRA_UNKNOWN_ALT, LRA_NON_CLOBBERED_ALT): New macros.
20486 * lra.c (lra_set_insn_recog_data, lra_update_insn_recog_data): Use
20488 * lra-constraints.c (curr_insn_transform): Set up
20489 LRA_NON_CLOBBERED_ALT for moves processed on the fast path. Use
20491 (remove_inheritance_pseudos): Use LRA_UNKNOWN_ALT.
20492 * lra-eliminations.c (spill_pseudos): Ditto.
20493 (process_insn_for_elimination): Ditto.
20494 * lra-lives.c (reg_early_clobber_p): Use the new macros.
20495 * lra-spills.c (spill_pseudos): Use LRA_UNKNOWN_ALT and
20496 LRA_NON_CLOBBERED_ALT.
20498 2018-02-22 Martin Sebor <msebor@redhat.com>
20500 PR tree-optimization/84480
20501 * gimple-fold.c (gimple_fold_builtin_strcpy): Move warnings
20502 to maybe_diag_stxncpy_trunc. Call it.
20503 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Integrate warnings
20504 from gimple_fold_builtin_strcpy. Print inlining stack.
20505 (handle_builtin_stxncpy): Print inlining stack.
20506 * tree-ssa-strlen.h (maybe_diag_stxncpy_trunc): Declare.
20508 2018-02-22 H.J. Lu <hongjiu.lu@intel.com>
20511 * config/i386/i386.c (ix86_set_indirect_branch_type): Issue an
20512 error when -mindirect-branch=thunk-extern, -fcf-protection=branch
20513 and -fcheck-pointer-bounds are used together.
20514 (indirect_thunk_prefix): New enum.
20515 (indirect_thunk_need_prefix): New function.
20516 (indirect_thunk_name): Replace need_bnd_p with need_prefix. Use
20517 "_nt" instead of "_bnd" for NOTRACK prefix.
20518 (output_indirect_thunk): Replace need_bnd_p with need_prefix.
20519 (output_indirect_thunk_function): Likewise.
20521 (ix86_code_end): Update output_indirect_thunk_function calls.
20522 (ix86_output_indirect_branch_via_reg): Replace
20523 ix86_bnd_prefixed_insn_p with indirect_thunk_need_prefix.
20524 (ix86_output_indirect_branch_via_push): Likewise.
20525 (ix86_output_function_return): Likewise.
20526 * doc/invoke.texi: Document -mindirect-branch=thunk-extern is
20527 incompatible with -fcf-protection=branch and
20528 -fcheck-pointer-bounds.
20530 2018-02-22 Steve Ellcey <sellcey@cavium.com>
20533 * config/aarch64/aarch64.c (aarch64_print_address_internal):
20534 Change gcc_assert call to output_operand_lossage.
20536 2018-02-22 Steve Ellcey <sellcey@cavium.com>
20538 * doc/extend.texi (__builtin_extend_pointer): Document builtin.
20540 2018-02-22 DJ Delorie <dj@redhat.com>
20541 Sebastian Perta <sebastian.perta@renesas.com>
20542 Oleg Endo <olegendo@gcc.gnu.org>
20544 * config/rx/rx.c (rx_rtx_costs): New function.
20545 (TARGET_RTX_COSTS): Override to use rx_rtx_costs.
20547 2018-02-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
20549 * config/arm/t-multilib: Map Armv8-R to Armv7 multilibs.
20551 2018-02-22 Martin Liska <mliska@suse.cz>
20554 * common/config/arm/arm-common.c (arm_print_hint_for_cpu_option):
20555 Add "native" as a possible value.
20557 2018-02-22 Martin Liska <mliska@suse.cz>
20560 * config/i386/i386.c (ix86_option_override_internal):
20561 Add "native" as a possible value for -march and -mtune.
20563 2018-02-22 Jakub Jelinek <jakub@redhat.com>
20566 * stor-layout.c (finalize_type_size): Propagate TYPE_EMPTY_P flag
20567 to all type variants.
20569 PR tree-optimization/84503
20570 * gimple-ssa-store-merging.c (merged_store_group::merge_into): Compute
20571 width as info->bitpos + info->bitsize - start.
20572 (merged_store_group::merge_overlapping): Simplify width computation.
20573 (check_no_overlap): New function.
20574 (imm_store_chain_info::try_coalesce_bswap): Compute expected
20575 start + width and last_order of the group, fail if check_no_overlap
20577 (imm_store_chain_info::coalesce_immediate_stores): Don't merge info
20578 to group if check_no_overlap fails.
20580 2018-02-21 Segher Boessenkool <segher@kernel.crashing.org>
20582 * config/rs6000/altivec.md: Delete contraint arguments to
20583 define_expand, define_split, and define_peephole2, and in
20584 define_insn_and_split if always unused.
20585 * config/rs6000/darwin.md: Ditto.
20586 * config/rs6000/dfp.md: Ditto.
20587 * config/rs6000/rs6000.md: Ditto.
20588 * config/rs6000/sync.md: Ditto.
20589 * config/rs6000/vector.md: Ditto.
20590 * config/rs6000/vsx.md: Ditto.
20592 2018-02-21 Segher Boessenkool <segher@kernel.crashing.org>
20594 * config/rs6000/altivec.md: Write output control strings as braced
20595 blocks instead of double-quoted strings.
20596 * config/rs6000/darwin.md: Ditto.
20597 * config/rs6000/rs6000.md: Ditto.
20598 * config/rs6000/vector.md: Ditto.
20599 * config/rs6000/vsx.md: Ditto.
20601 2018-02-21 Jason Merrill <jason@redhat.com>
20603 PR c++/84314 - ICE with templates and fastcall attribute.
20604 * attribs.c (build_type_attribute_qual_variant): Remove assert.
20606 2018-02-21 Jan Hubicka <hubicka@ucw.cz>
20608 * ipa-cp.c (determine_versionability): Fix comment typos.
20610 2018-02-21 Jan Hubicka <hubicka@ucw.cz>
20613 * ipa-cp.c (determine_versionability): Do not version functions caling
20616 2018-02-21 Martin Liska <mliska@suse.cz>
20619 * config/aarch64/aarch64.c (aarch64_print_hint_for_core_or_arch):
20620 Add "native" as a possible value.
20621 * config/aarch64/aarch64.h (HAVE_LOCAL_CPU_DETECT): Define
20622 the macro when native cpu detection is available.
20624 2018-02-21 Martin Liska <mliska@suse.cz>
20627 * common/config/arm/arm-common.c (arm_print_hint_for_arch_option):
20628 Add "native" as a possible value.
20629 * config/arm/arm.h (HAVE_LOCAL_CPU_DETECT): Define the macro
20630 when native cpu detection is available.
20632 2018-02-21 Jakub Jelinek <jakub@redhat.com>
20633 Martin Sebor <msebor@redhat.com>
20635 PR tree-optimization/84478
20636 * gimple-fold.h (get_range_strlen): Add a bool argument defaulted to
20638 * gimple-fold.c (get_range_strlen): Make minlen const and assume it
20639 can't be NULL. Change FUZZY from bool to int, for 1 add PHI/COND_EXPR
20640 support which is conservatively correct, for 2 only stay conservative
20641 for maxlen. Formatting and comment capitalization fixes. Add STRICT
20642 argument to the 2 argument get_range_strlen, adjust 6 arg
20643 get_range_strlen caller and clear minmaxlen[0] and [1] if it returned
20645 (get_maxval_strlen): Adjust 6 arg get_range_strlen caller.
20646 (gimple_fold_builtin_strlen): Pass true as last argument to
20649 2018-02-20 Martin Sebor <msebor@redhat.com>
20651 PR middle-end/84095
20652 * gimple-ssa-warn-restrict.c (builtin_memref::extend_offset_range): New.
20653 (builtin_memref::set_base_and_offset): Same. Handle inner references.
20654 (builtin_memref::builtin_memref): Factor out parts into
20655 set_base_and_offset and call it.
20657 2018-02-20 Richard Sandiford <richard.sandiford@linaro.org>
20659 PR middle-end/84406
20660 * optabs-query.c (find_widening_optab_handler_and_mode): If from_mode
20661 is a scalar_int_mode, assert that to_mode is a scalar_int_mode with
20662 greater precision. If to_mode is a MODE_PARTIAL_INT, stop the
20663 search at the associated MODE_INT.
20665 2018-02-20 Jeff Law <law@redhat.com>
20667 PR middle-end/82123
20668 PR tree-optimization/81592
20669 PR middle-end/79257
20670 * gimple-ssa-sprintf.c (format_integer): Query EVRP range analyzer
20671 for range data rather than using global data.
20672 * gimple-ssa-sprintf.c (get_int_range): Query EVRP range analyzer for
20673 range data rather than using global data.
20674 * gimple-ssa-sprintf.c (get_int_range): Accept vr_values parameter
20675 pass it to children as needed.
20676 (struct directive::fmtresult): Similarly.
20677 (struct directive::set_width): Similarly.
20678 (struct directive::set_precision): Similarly.
20679 (format_integer, format_directive, parse_directive): Similarly.
20680 (format_none): Accept unnamed vr_values parameter.
20681 (format_percent, format_floating, format_character): Similarly.
20682 (format_string, format_plain): Similarly.
20683 * gimple-ssa-sprintf.c (sprintf_dom_walker::handle_gimple_call): Query
20684 the EVRP range analyzer for range data rather than using global data.
20685 * gimple-ssa-sprintf.c: Include alloc-pool.h, vr-values.h and
20686 gimple-ssa-evrp-analyze.h
20687 (class sprintf_dom_walker): Add after_dom_children member function.
20688 Add evrp_range_analyzer member.
20689 (sprintf_dom_walker::before_dom_children): Call into the EVRP
20690 range analyzer as needed.
20691 (sprintf_dom_walker::after_dom_children): New member function.
20692 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::enter): Do nothing
20694 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
20695 (evrp_range_analyzer::pop_to_marker): Likewise.
20697 2018-02-20 Richard Sandiford <richard.sandiford@linaro.org>
20699 PR tree-optimization/84419
20700 * internal-fn.c (expand_call_mem_ref): Create a TARGET_MEM_REF
20701 with the required type if its current type is compatible but
20704 2018-02-20 Jakub Jelinek <jakub@redhat.com>
20706 PR middle-end/82004
20707 * match.pd (pow(C,x) -> exp(log(C)*x)): Delay all folding until
20708 after vectorization.
20710 2018-02-20 Martin Liska <mliska@suse.cz>
20713 * config/aarch64/aarch64.c (aarch64_print_hint_for_core_or_arch): Print
20714 possible values if we don't have a hint.
20716 2018-02-20 Martin Liska <mliska@suse.cz>
20720 * final.c (shorten_branches): Build align_tab array with one
20722 * opts.c (finish_options): Add alignment option limit check.
20723 (MAX_CODE_ALIGN): Likewise.
20724 (MAX_CODE_ALIGN_VALUE): Likewise.
20725 * doc/invoke.texi: Document maximum allowed option value for
20726 all -falign-* options.
20728 2018-02-19 Jakub Jelinek <jakub@redhat.com>
20731 * reg-notes.def (REG_CALL_ARG_LOCATION): New reg note.
20732 * insn-notes.def (NOTE_INSN_CALL_ARG_LOCATION): Remove.
20733 * var-tracking.c (emit_note_insn_var_location): Remove all references
20734 to NOTE_INSN_CALL_ARG_LOCATION.
20735 (emit_notes_in_bb): Emit arguments as REG_CALL_ARG_LOCATION note on
20736 the CALL_INSN rather than separate NOTE_INSN_CALL_ARG_LOCATION note.
20737 Use copy_rtx_if_shared.
20738 * dwarf2out.c (gen_subprogram_die): Use XEXP with 0 instead of
20739 NOTE_VAR_LOCATION on ca_loc->call_arg_loc_note.
20740 (dwarf2out_var_location): Remove handling of
20741 NOTE_INSN_CALL_ARG_LOCATION, instead handle REG_CALL_ARG_LOCATION note
20743 * final.c (final_scan_insn): Remove all references to
20744 NOTE_INSN_CALL_ARG_LOCATION.
20745 (rest_of_clean_state): Likewise. Remove REG_CALL_ARG_LOCATION notes
20746 before dumping final insns.
20747 * except.c (emit_note_eh_region_end): Remove all references to
20748 NOTE_INSN_CALL_ARG_LOCATION.
20749 * config/alpha/alpha.c (alpha_pad_function_end): Likewise.
20750 * config/c6x/c6x.c (c6x_gen_bundles): Likewise.
20751 * config/arc/arc.c (hwloop_optimize): Likewise.
20752 * config/arm/arm.c (create_fix_barrier): Likewise.
20753 * config/s390/s390.c (s390_chunkify_start): Likewise.
20754 * config/sh/sh.c (find_barrier): Likewise.
20755 * config/i386/i386.c (rest_of_insert_endbranch,
20756 ix86_seh_fixup_eh_fallthru): Likewise.
20757 * config/xtensa/xtensa.c (hwloop_optimize): Likewise.
20758 * config/iq2000/iq2000.c (final_prescan_insn): Likewise.
20759 * config/frv/frv.c (frv_function_prologue): Likewise.
20760 * emit-rtl.c (try_split): Likewise. Copy over REG_CALL_ARG_LOCATION
20762 (note_outside_basic_block_p): Remove all references to
20763 NOTE_INSN_CALL_ARG_LOCATION.
20764 * gengtype.c (adjust_field_rtx_def): Likewise.
20765 * print-rtl.c (rtx_writer::print_rtx_operand_code_0, print_insn):
20767 * jump.c (cleanup_barriers, delete_related_insns): Likewise.
20768 * cfgrtl.c (force_nonfallthru_and_redirect): Likewise.
20771 * builtins.c (builtin_mathfn_code): Don't check if CALL_EXPR_FN (t)
20774 PR tree-optimization/84452
20775 * tree-vect-patterns.c (vect_recog_pow_pattern): Don't call
20776 expand_simd_clones if targetm.simd_clone.compute_vecsize_and_simdlen
20779 2018-02-19 Martin Liska <mliska@suse.cz>
20782 * passes.def: Put pass_sancov_O0 before pass_lower_switch with -O0.
20784 2018-02-19 Martin Liska <mliska@suse.cz>
20785 Richard Sandiford <richard.sandiford@linaro.org>
20787 PR tree-optimization/82491
20788 * gimple-fold.c (get_base_constructor): Make earlier bail out
20791 2018-02-19 Carl Love <cel@us.ibm.com>
20793 * config/rs6000/rs6000-builtin.def: Change NEG macro expansions from
20794 BU_ALTIVEC_A to BU_P8V_AV_1 and BU_ALTIVEC_OVERLOAD_1 to
20796 * config/rs6000/rs6000-c.c: Change ALTIVEC_BUILTIN_VEC_NEG to
20797 P8V_BUILTIN_VEC_NEG.
20799 2018-02-19 Sebastian Perta <sebastian.perta@renesas.com>
20801 * config/rl78/rl78.md (movdf): New define expand.
20803 2018-02-19 Martin Liska <mliska@suse.cz>
20806 * doc/invoke.texi: Fix typo.
20807 * params.def (PARAM_MAX_LOOP_HEADER_INSNS): Likewise.
20809 2018-02-18 Segher Boessenkool <segher@kernel.crashing.org>
20811 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
20812 handle rs6000_single_float and rs6000_double_float specially for
20815 2018-02-16 Jeff Law <law@redhat.com>
20817 * config/rx/rx.c (add_pop_cfi_notes): New function.;
20818 (pop_regs): Use it.
20820 2018-02-16 Jakub Jelinek <jakub@redhat.com>
20823 * ipa-inline.c (inline_small_functions): Fix a typo.
20825 2018-02-16 Nathan Sidwell <nathan@acm.org>
20827 * doc/extend.texi (Backwards Compatibility): Americanize 'behaviour'.
20829 2018-02-16 Carl Love <cel@us.ibm.com>
20831 * config/rs6000/rs6000-builtin.def: Add BU_P8V_VSX_2 macro definition.
20832 Change FLOAT2 expansions from BU_VSX_2 to BU_P8V_VSX_2 and
20833 from BU_VSX_OVERLOAD_2 to BU_P8V_OVERLOAD_2.
20834 * config/rs6000/rs6000-c.c: Changed macro VSX_BUILTIN_VEC_FLOAT2
20835 expansion to P8V_BUILTIN_VEC_FLOAT2.
20837 2018-02-16 Vladimir Makarov <vmakarov@redhat.com>
20839 PR rtl-optimization/70023
20840 * lra-constraints.c (inherit_in_ebb): Take hard reg mode of
20841 src_regno into account.
20843 2018-02-16 Carl Love <cel@us.ibm.com>
20845 * config/rs6000/altivec.h: Remove vec_vextract4b and vec_vinsert4b.
20846 * config/rs6000/rs6000-builtin.def: Remove macro expansion for
20847 VEXTRACT4B, VINSERT4B, VINSERT4B_DI and VEXTRACT4B.
20848 * config/rs6000/rs6000.c: Remove case statements for
20849 P9V_BUILTIN_VEXTRACT4B, P9V_BUILTIN_VEC_VEXTRACT4B,
20850 P9V_BUILTIN_VINSERT4B, P9V_BUILTIN_VINSERT4B_DI,
20851 and P9V_BUILTIN_VEC_VINSERT4B.
20852 * config/rs6000/rs6000-c.c (altivec_expand_builtin): Remove entries for
20853 P9V_BUILTIN_VEC_VEXTRACT4B and P9V_BUILTIN_VEC_VINSERT4B.
20854 * config/rs6000/vsx.md:
20855 * doc/extend.texi: Remove vec_vextract4b, non ABI definitions for
20858 2018-02-16 Carl Love <cel@us.ibm.com>
20860 * config/rs6000/altivec.h: Add builtin names vec_extract4b
20862 * config/rs6000/rs6000-builtin.def: Add INSERT4B and EXTRACT4B
20864 * config/rs6000/rs6000-c.c: Add the definitions for
20865 P9V_BUILTIN_VEC_EXTRACT4B and P9V_BUILTIN_VEC_INSERT4B.
20866 * config/rs6000/rs6000.c (altivec_expand_builtin): Add
20867 P9V_BUILTIN_EXTRACT4B and P9V_BUILTIN_INSERT4B case statements.
20868 * config/rs6000/vsx.md: Add define_insn extract4b. Add define_expand
20869 definition for insert4b and define insn *insert3b_internal.
20870 * doc/extend.texi: Add documentation for vec_extract4b.
20872 2018-02-16 Nathan Sidwell <nathan@acm.org>
20874 * doc/extend.texi (Backwards Compatibility): Mention friend
20875 injection. Note for-scope is deprecated.
20876 * doc/invoke.texi (-ffriend-injection): Deprecate.
20878 2018-02-16 Segher Boessenkool <segher@kernel.crashing.org>
20880 * combine.c (try_combine): When adjusting LOG_LINKS for the destination
20881 that moved to I2, also allow destinations that are a paradoxical
20882 subreg (instead of a normal reg).
20884 2018-02-16 Oleg Endo <olegendo@gcc.gnu.org>
20887 * config/rx/rx.c (rx_fuse_in_memory_bitop): Convert shift operand
20890 2018-02-16 Richard Biener <rguenther@suse.de>
20892 PR tree-optimization/84037
20893 PR tree-optimization/84016
20895 * config/i386/i386.c (ix86_builtin_vectorization_cost):
20896 Adjust vec_construct for the fact we need additional higher latency
20897 128bit inserts for AVX256 and AVX512 vector builds.
20898 (ix86_add_stmt_cost): Scale vector construction cost for
20901 2018-02-16 Richard Biener <rguenther@suse.de>
20903 PR tree-optimization/84417
20904 * tree-ssa.c (non_rewritable_mem_ref_base): Properly constrain
20905 the MEM_REF offset when conversion to BIT_FIELD_REF is desired.
20906 (non_rewritable_lvalue_p): Likewise, use poly-ints.
20908 2018-02-16 Martin Liska <mliska@suse.cz>
20911 * internal-fn.def (ASAN_CHECK): Set proper flags.
20912 (ASAN_MARK): Likewise.
20914 2018-02-16 Julia Koval <julia.koval@intel.com>
20916 * config/i386/i386.c (ix86_option_override_internal): Remove PTA_CLWB
20917 from PTA_CANNONLAKE.
20919 2018-02-16 Jakub Jelinek <jakub@redhat.com>
20922 * config/aarch64/cortex-a57-fma-steering.c (fma_forest::merge_forest):
20923 Use ++iter rather than iter++ for std::list iterators.
20924 (func_fma_steering::dfs): Likewise. Don't delete nodes right away,
20925 defer deleting them until all nodes in the forest are processed. Do
20926 free even leaf nodes. Change to_process into auto_vec.
20929 * system.h (BROKEN_VALUE_INITIALIZATION): Define for GCC < 4.3.
20930 * vec.h (vec_default_construct): Use memset instead of placement new
20931 if BROKEN_VALUE_INITIALIZATION is defined.
20932 * hash-table.h (hash_table<Descriptor, Allocator>::empty_slow): Use
20933 memset instead of value initialization if BROKEN_VALUE_INITIALIZATION
20936 PR rtl-optimization/83723
20937 * lra-int.h (lra_substitute_pseudo): Add DEBUG_P argument.
20938 * lra.c (lra_substitute_pseudo): Likewise. If true, use
20939 gen_rtx_raw_SUBREG instead of gen_rtx_SUBREG. Pass DEBUG_P to
20941 (lra_substitute_pseudo_within_insn): Adjust lra_substitute_pseudo
20943 * lra-constraints.c (inherit_reload_reg, split_reg): Likewise.
20945 2018-02-16 Eric Botcazou <ebotcazou@adacore.com>
20947 PR rtl-optimization/81443
20948 * rtlanal.c (num_sign_bit_copies1) <SUBREG>: Do not propagate results
20949 from inner REGs to paradoxical SUBREGs.
20951 2018-02-16 Richard Biener <rguenther@suse.de>
20953 PR tree-optimization/84399
20954 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
20955 For operands we can analyze at their definition make sure we can
20956 analyze them at each use as well.
20958 2018-02-16 Richard Biener <rguenther@suse.de>
20960 PR tree-optimization/84190
20961 * tree-ssa.c (non_rewritable_mem_ref_base): Do not touch
20962 volatile accesses if the decl isn't volatile.
20964 2018-02-15 Jason Merrill <jason@redhat.com>
20966 PR c++/84314 - ICE with templates and fastcall attribute.
20967 * attribs.c (build_type_attribute_qual_variant): Don't clobber
20968 TYPE_CANONICAL on an existing type.
20970 2018-02-15 Jakub Jelinek <jakub@redhat.com>
20972 PR tree-optimization/84383
20973 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Don't look at
20974 dstoff nor call operand_equal_p if dstbase is NULL.
20976 PR tree-optimization/84334
20977 * match.pd ((A +- CST1) +- CST2 -> A + CST3): If A is
20978 also a CONSTANT_CLASS_P, punt.
20980 2018-02-14 Jim Wilson <jimw@sifive.com>
20982 * config/riscv/riscv.c (riscv_first_stack_step): Move locals after
20983 first SMALL_OPERAND check. New local min_second_step. Move assert
20984 to where locals are set. Add TARGET_RVC support.
20985 * config/riscv/riscv.h (C_SxSP_BITS, SWSP_REACH, SDSP_REACH): New.
20987 2018-02-14 Indu Bhagat <indu.bhagat@oracle.com>
20989 * doc/invoke.texi: Correct -Wformat-overflow code sample.
20991 2018-02-14 Martin Sebor <msebor@redhat.com>
20993 PR tree-optimization/83698
20994 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): For
20995 arrays constrain the offset range to their bounds.
20996 (builtin_access::strcat_overlap): Adjust the bounds of overlap offset.
20997 (builtin_access::overlap): Avoid setting the size of overlap if it's
20999 (maybe_diag_overlap): Also consider arrays when deciding what values
21000 of offsets to include in diagnostics.
21002 2018-02-14 Martin Sebor <msebor@redhat.com>
21005 * attribs.c (diag_attr_exclusions): Consider the exclusion(s)
21006 that correspond to the kind of a declaration.
21008 2018-02-14 John David Anglin <danglin@gcc.gnu.org>
21011 * config/pa/pa.md: Load address of PIC label using the linkage table
21012 if the label is nonlocal.
21014 2018-02-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
21016 * config/rs6000/rs6000.c (rs6000_option_override_internal): Issue
21017 warning message if user requests -maltivec=be.
21018 * doc/invoke.texi: Document deprecation of -maltivec=be.
21020 2018-02-14 Will Schmidt <will_schmidt@vnet.ibm.com>
21023 * config/rs6000/rs6000-c.c: Update definitions for
21024 ALTIVEC_BUILTIN_VEC_SLD, ALTIVEC_BUILTIN_VEC_SLDW,
21025 VEC_XXSLDWI and ALTIVEC_BUILTIN_VEC_XXPERMDI builtins.
21027 2018-02-14 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
21030 * config/i386/cetintrin.h: Remove _rdssp[d|q] and
21031 add _get_ssp intrinsics. Remove argument from
21032 __builtin_ia32_rdssp[d|q].
21033 * config/i386/i386-builtin-types.def: Add UINT_FTYPE_VOID.
21034 * config/i386/i386-builtin.def: Remove argument from
21035 __builtin_ia32_rdssp[d|q].
21036 * config/i386/i386.c: Use UINT_FTYPE_VOID. Use
21037 ix86_expand_special_args_builtin for _rdssp[d|q].
21038 * config/i386/i386.md: Remove argument from rdssp[si|di] insn.
21039 Clear register before usage.
21040 * doc/extend.texi: Remove argument from __builtin_ia32_rdssp[d|q].
21041 Add documentation for new _get_ssp and _inc_ssp intrinsics.
21043 2018-02-14 Richard Sandiford <richard.sandiford@linaro.org>
21045 PR tree-optimization/84357
21046 * tree-data-ref.c (object_address_invariant_in_loop_p): Check
21047 operand 1 of an ARRAY_REF too.
21049 2018-02-14 Oleg Endo <olegendo@gcc.gnu.org>
21052 * config/rx/rx-protos.h (rx_reg_dead_or_unused_after_insn,
21053 rx_copy_reg_dead_or_unused_notes, rx_fuse_in_memory_bitop): New
21055 (set_of_reg): New struct.
21056 (rx_find_set_of_reg, rx_find_use_of_reg): New functions.
21057 * config/rx/rx.c (rx_reg_dead_or_unused_after_insn,
21058 rx_copy_reg_dead_or_unused_notes, rx_fuse_in_memory_bitop): New
21060 * config/rx/rx.md (andsi3, iorsi3, xorsi3): Convert to insn_and_split.
21061 Split into bitclr, bitset, bitinvert patterns if appropriate.
21062 (*bitset, *bitinvert, *bitclr): Convert to named insn_and_split and
21063 use rx_fuse_in_memory_bitop.
21064 (*bitset_in_memory, *bitinvert_in_memory, *bitclr_in_memory): Convert
21065 to named insn, correct maximum insn length.
21067 2018-02-14 Jozef Lawrynowicz <jozefl.gcc@gmail.com>
21070 * machmode.def: Define a complex mode for PARTIAL_INT.
21071 * genmodes.c (complex_class): Return MODE_COMPLEX_INT for
21073 * doc/rtl.texi: Document CSPImode.
21074 * config/msp430/msp430.c (msp430_hard_regno_nregs): Add CPSImode
21076 (msp430_hard_regno_nregs_with_padding): Likewise.
21078 2018-02-13 Peter Bergner <bergner@vnet.ibm.com>
21081 * config/rs6000/rs6000.c (mem_operand_gpr): Disallow altivec addresses.
21083 2018-02-13 Segher Boessenkool <segher@kernel.crashing.org>
21085 PR rtl-optimization/84169
21086 * combine.c (try_combine): New variable split_i2i3. Set it to true if
21087 we generated a parallel as new i3 and we split that to new i2 and i3
21088 instructions. Handle split_i2i3 similar to swap_i2i3: scan the
21089 LOG_LINKs of i3 to see which of those need to link to i2 now. Link
21090 those to i2, not i1. Partially rewrite this scan code.
21092 2018-02-13 Jakub Jelinek <jakub@redhat.com>
21095 * stor-layout.c (place_field): For variable length fields, adjust
21096 offset_align afterwards not just based on the field's alignment,
21097 but also on the size.
21099 PR middle-end/84309
21100 * match.pd (pow(C,x) -> exp(log(C)*x)): Use exp2s and log2s instead
21101 of exps and logs in the use_exp2 case.
21103 2018-02-13 Jeff Law <law@redhat.com>
21105 * config/rl/rl78.c (rl78_attribute_table): Fix terminator and
21106 entry for "vector".
21108 * config/rl78/rl78.c (rl78_handle_func_attribute): Mark
21111 2018-02-13 Alexandre Oliva <aoliva@redhat.com>
21115 * common.opt (gas-loc-support, gas-locview-support): New.
21116 (ginline-points, ginternal-reset-location-views): New.
21117 * doc/invoke.texi: Document them. Use @itemx where intended.
21118 (gvariable-location-views): Adjust.
21119 * target.def (reset_location_view): New.
21120 * doc/tm.texi.in (DWARF2_ASM_VIEW_DEBUG_INFO): New.
21121 (TARGET_RESET_LOCATION_VIEW): New.
21122 * doc/tm.texi: Rebuilt.
21123 * dwarf2out.c (dwarf2out_default_as_loc_support): New.
21124 (dwarf2out_default_as_locview_support): New.
21125 (output_asm_line_debug_info): Use option variables.
21126 (dwarf2out_maybe_output_loclist_view_pair): Likewise.
21127 (output_loc_list): Likewise.
21128 (add_high_low_attributes): Check option variables.
21129 Don't output entry view attribute in strict mode.
21130 (gen_inlined_subroutine_die): Check option variables.
21131 (dwarf2out_inline_entry): Likewise.
21132 (init_sections_and_labels): Likewise.
21133 (dwarf2out_early_finish): Likewise.
21134 (maybe_reset_location_view): New, from...
21135 (dwarf2out_var_location): ... here. Call it.
21136 * debug.h (dwarf2out_default_as_loc_support): Declare.
21137 (dwarf2out_default_as_locview_support): Declare.
21138 * hooks.c (hook_int_rtx_insn_0): New.
21139 * hooks.h (hook_int_rtx_insn_0): Declare.
21140 * toplev.c (process_options): Take -gas-loc-support and
21141 -gas-locview-support from dwarf2out. Enable
21142 -gvariable-location-views by default only with locview
21143 assembler support. Enable -ginternal-reset-location-views by
21144 default only if the target defines the corresponding hook.
21145 Enable -ginline-points by default if location views are
21146 enabled; force it disabled if statement frontiers are
21148 * tree-inline.c (expand_call_inline): Check option variables.
21149 * tree-ssa-live.c (remove_unused_scope_block_p): Likewise.
21151 2018-02-13 Richard Sandiford <richard.sandiford@linaro.org>
21153 PR tree-optimization/84321
21154 * tree-vrp.c (intersect_range_with_nonzero_bits): Fix VR_ANTI_RANGE
21155 handling. Also check whether the anti-range contains any values
21156 that satisfy the mask; switch to a VR_RANGE if not.
21158 2018-02-13 Paolo Bonzini <bonzini@gnu.org>
21161 * internal-fn.def (ASAN_CHECK, ASAN_MARK): Revert changes to fnspec.
21163 2018-02-13 Martin Jambor <mjambor@suse.cz>
21166 * ipa-param-manipulation.c (ipa_modify_call_arguments): Use location
21167 of call statements, also set location of a load to a temporary.
21169 2018-02-13 Sebastian Perta <sebastian.perta@renesas.com>
21171 * config/rl78/rl78.c (add_vector_labels): New function.
21172 * config/rl78/rl78.c (rl78_handle_vector_attribute): New function.
21173 * config/rl78/rl78.c (rl78_start_function): Call add_vector_labels.
21174 * config/rl78/rl78.c (rl78_handle_func_attribute): Removed the assert
21175 which checks that no arguments are passed.
21176 * config/rl78/rl78.c (rl78_attribute_table): Add "vector" attribute.
21177 * doc/extend.texi: Documentation for the new attribute.
21179 2018-02-13 Andreas Schwab <schwab@suse.de>
21181 * config/riscv/linux.h (CPP_SPEC): Define.
21183 2018-02-13 Jakub Jelinek <jakub@redhat.com>
21186 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Pass
21187 OPTION_MASK_ISA_AES | OPTION_MASK_ISA_SSE2 instead of
21188 OPTION_MASK_ISA_AES as first argument to def_builtin_const
21189 for AES builtins. Pass OPTION_MASK_ISA_PCLMUL | OPTION_MASK_ISA_SSE2
21190 instead of OPTION_MASK_ISA_PCLMUL as first argument to
21191 def_builtin_const for __builtin_ia32_pclmulqdq128 builtin.
21192 * config/i386/wmmintrin.h: If __SSE2__ is not defined, enable it
21193 temporarily for AES and PCLMUL builtins.
21195 PR tree-optimization/84339
21196 * gimple-fold.c (get_range_strlen): Set *FLEXP to true when handling
21197 ARRAY_REF where first operand is array_at_struct_end_p COMPONENT_REF.
21200 PR middle-end/84309
21201 * match.pd (pow(C,x) -> exp(log(C)*x)): Optimize instead into
21202 exp2(log2(C)*x) if C is a power of 2 and c99 runtime is available.
21203 * generic-match-head.c (canonicalize_math_after_vectorization_p): New
21205 * gimple-match-head.c (canonicalize_math_after_vectorization_p): New
21207 * omp-simd-clone.h: New file.
21208 * omp-simd-clone.c: Include omp-simd-clone.h.
21209 (expand_simd_clones): No longer static.
21210 * tree-vect-patterns.c: Include fold-const-call.h, attribs.h,
21211 cgraph.h and omp-simd-clone.h.
21212 (vect_recog_pow_pattern): Optimize pow(C,x) to exp(log(C)*x).
21213 (vect_recog_widen_shift_pattern): Formatting fix.
21214 (vect_pattern_recog_1): Don't check optab for calls.
21217 * config/i386/sse.md (<avx512>_vpermi2var<mode>3_mask): Force
21218 operands[2] into a REG before using gen_lowpart on it.
21220 2018-02-12 Jeff Law <law@redhat.com>
21223 * config/sh/sh.c (find_barrier): Consider a sibling call
21226 * cse.c (try_back_substitute_reg): Move any REG_ARGS_SIZE note when
21227 successfully back substituting a reg.
21229 2018-02-12 Richard Biener <rguenther@suse.de>
21231 PR tree-optimization/84037
21232 * tree-vect-slp.c (vect_analyze_slp_cost): Add visited
21233 parameter, move visited init to caller.
21234 (vect_slp_analyze_operations): Separate cost from validity
21235 check, initialize visited once for all instances.
21236 (vect_schedule_slp): Analyze map to CSE vectorized nodes once
21238 * tree-vect-stmts.c (vect_model_simple_cost): Make early
21240 (vect_model_promotion_demotion_cost): Likewise.
21241 (vectorizable_bswap): Guard cost modeling with !slp_node
21242 instead of !PURE_SLP_STMT to avoid double-counting on hybrid
21244 (vectorizable_call): Likewise.
21245 (vectorizable_conversion): Likewise.
21246 (vectorizable_assignment): Likewise.
21247 (vectorizable_shift): Likewise.
21248 (vectorizable_operation): Likewise.
21249 (vectorizable_store): Likewise.
21250 (vectorizable_load): Likewise.
21251 (vectorizable_condition): Likewise.
21252 (vectorizable_comparison): Likewise.
21254 2018-02-12 Paolo Bonzini <bonzini@gnu.org>
21257 * internal-fn.def (ASAN_CHECK): Fix fnspec to account for return value.
21258 (ASAN_MARK): Fix fnspec to account for return value, change pointer
21259 argument from 'R' to 'W' so that the pointed-to datum is clobbered.
21261 2018-02-08 Jan Hubicka <hubicka@ucw.cz>
21263 PR middle-end/83665
21264 * params.def (inline-min-speedup): Increase from 8 to 15.
21265 (max-inline-insns-auto): Decrease from 40 to 30.
21266 * ipa-split.c (consider_split): Add some buffer for function to
21267 be considered inlining candidate.
21268 * invoke.texi (max-inline-insns-auto, inline-min-speedup): UPdate
21271 2018-02-12 Richard Biener <rguenther@suse.de>
21273 PR tree-optimization/84037
21274 * tree-vect-slp.c (vect_build_slp_tree_2): Try swapping the
21275 matched stmts if we cannot swap the non-matched ones.
21277 2018-02-12 Olga Makhotina <olga.makhotina@intel.com>
21279 * config/i386/avx512fintrin.h (_mm_mask_scalef_round_sd,
21280 _mm_maskz_scalef_round_sd, _mm_mask_scalef_round_ss,
21281 _mm_maskz_scalef_round_ss): New intrinsics.
21282 (__builtin_ia32_scalefsd_round, __builtin_ia32_scalefss_round): Fix.
21283 * config/i386/i386-builtin.def (__builtin_ia32_scalefsd_round,
21284 __builtin_ia32_scalefss_round): Remove.
21285 (__builtin_ia32_scalefsd_mask_round,
21286 __builtin_ia32_scalefss_mask_round): New intrinsics.
21287 * config/i386/sse.md (vmscalef<mode><round_name>): Renamed to ...
21288 (vmscalef<mode><mask_scalar_name><round_scalar_name>): ... this.
21289 ((match_operand:VF_128 2 "<round_nimm_predicate>"
21290 "<round_constraint>")): Changed to ...
21291 ((match_operand:VF_128 2 "<round_scalar_nimm_predicate>"
21292 "<round_scalar_constraint>")): ... this.
21293 ("vscalef<ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|
21294 %0, %1, %2<round_op3>}"): Changed to ...
21295 ("vscalef<ssescalarmodesuffix>\t{<round_scalar_mask_op3>%2, %1,
21296 %0<mask_scalar_operand3>|%0<mask_scalar_operand3>, %1,
21297 %2<round_scalar_mask_op3>}"): ... this.
21298 * config/i386/subst.md (round_scalar_nimm_predicate): New.
21300 2018-02-12 Olga Makhotina <olga.makhotina@intel.com>
21302 * config/i386/avx512fintrin.h (_mm_mask_sqrt_round_sd)
21303 (_mm_maskz_sqrt_round_sd, _mm_mask_sqrt_round_ss)
21304 (_mm_maskz_sqrt_round_ss): New intrinsics.
21305 (__builtin_ia32_sqrtsd_round, __builtin_ia32_sqrtss_round): Remove.
21306 (__builtin_ia32_sqrtsd_mask_round)
21307 (__builtin_ia32_sqrtss_mask_round): New builtins.
21308 * config/i386/i386-builtin.def (__builtin_ia32_sqrtsd_round)
21309 (__builtin_ia32_sqrtss_round): Remove.
21310 (__builtin_ia32_sqrtsd_mask_round)
21311 (__builtin_ia32_sqrtss_mask_round): New builtins.
21312 * config/i386/sse.md (vmsqrt<mode>2<round_name>): Renamed to ...
21313 (vmsqrt<mode>2<mask_scalar_name><round_scalar_name>): ... this.
21314 ((match_operand:VF_128 1 "vector_operand"
21315 "xBm,<round_constraint>")): Changed to ...
21316 ((match_operand:VF_128 1 "vector_operand"
21317 "xBm,<round_scalar_constraint>")): ... this.
21318 (vsqrt<ssescalarmodesuffix>\t{<round_op3>%1, %2, %0|
21319 %0, %2, %<iptr>1<round_op3>}): Changed to ...
21320 (vsqrt<ssescalarmodesuffix>\t{<round_scalar_mask_op3>%1, %2,
21321 %0<mask_scalar_operand3>|%0<mask_scalar_operand3>, %2,
21322 %<iptr>1<round_scalar_mask_op3>}): ... this.
21323 ((set_attr "prefix" "<round_prefix>")): Changed to ...
21324 ((set_attr "prefix" "<round_scalar_prefix>")): ... this.
21326 2018-02-11 Steven Munroe <munroesj@gcc.gnu.org>
21329 * config/rs6000/mmintrin.h (_mm_cmpeq_pi32 [_ARCH_PWR9]):
21330 Cast vec_cmpeq result to correct type.
21331 * config/rs6000/mmintrin.h (_mm_cmpgt_pi32 [_ARCH_PWR9]):
21332 Cast vec_cmpgt result to correct type.
21334 2018-02-11 Alexandre Oliva <aoliva@redhat.com>
21336 * final.c (final_scan_insn_1): Renamed from...
21337 (final_scan_insn): ... this. New wrapper, to recover
21338 seen from the outermost call in recursive ones.
21339 * config/sparc/sparc.c (output_return): Drop seen from call.
21340 (output_sibcall): Likewise.
21341 * config/visium/visium.c (output_branch): Likewise.
21343 2018-02-10 John David Anglin <danglin@gcc.gnu.org>
21345 * config/pa/pa.c (hppa_profile_hook): Mark SYMBOL_REF for _mcount as
21348 2018-02-10 Alan Modra <amodra@gmail.com>
21351 * config/rs6000/rs6000.md (split_stack_return): Remove (use ..).
21352 Specify LR as an input.
21354 2018-02-10 Jakub Jelinek <jakub@redhat.com>
21357 * omp-low.c (maybe_remove_omp_member_access_dummy_vars,
21358 remove_member_access_dummy_vars): New functions.
21359 (lower_omp_for, lower_omp_taskreg, lower_omp_target,
21360 lower_omp_1, execute_lower_omp): Use them.
21362 PR rtl-optimization/84308
21363 * shrink-wrap.c (spread_components): Release todo vector.
21365 2018-02-09 Vladimir Makarov <vmakarov@redhat.com>
21367 PR rtl-optimization/57193
21368 * ira-color.c (struct allocno_color_data): Add member
21369 conflict_allocno_hard_prefs.
21370 (update_conflict_allocno_hard_prefs): New.
21371 (bucket_allocno_compare_func): Add a preference based on
21372 conflict_allocno_hard_prefs.
21373 (push_allocno_to_stack): Update conflict_allocno_hard_prefs.
21374 (color_allocnos): Remove a dead code. Initiate
21375 conflict_allocno_hard_prefs. Call update_costs_from_prefs.
21377 2018-02-09 Jakub Jelinek <jakub@redhat.com>
21380 * config/rs6000/vsx.md (p9_xxbrq_v16qi): Change input operand
21381 constraint from =wa to wa. Avoid a subreg on the output operand,
21382 instead use a pseudo and subreg it in a move.
21383 (p9_xxbrd_<mode>): Changed to ...
21384 (p9_xxbrd_v2di): ... this insn, without VSX_D iterator.
21385 (p9_xxbrd_v2df): New expander.
21386 (p9_xxbrw_<mode>): Changed to ...
21387 (p9_xxbrw_v4si): ... this insn, without VSX_W iterator.
21388 (p9_xxbrw_v4sf): New expander.
21390 2018-02-09 Sebastian Perta <sebastian.perta@renesas.com>
21392 * config/rx/rx.md (movsicc): Update expander to be matched by GCC.
21394 2018-02-09 Peter Bergner <bergner@vnet.ibm.com>
21397 * config/rs6000/vsx.md (vsx_mul_v2di): Handle generating a 64-bit
21398 multiply in 32-bit mode.
21399 (vsx_div_v2di): Handle generating a 64-bit signed divide in 32-bit mode.
21400 (vsx_udiv_v2di): Handle generating a 64-bit unsigned divide in 32-bit
21403 2018-02-09 Sebastian Perta <sebastian.perta@renesas.com>
21405 * config/rx/constraints.md (CALL_OP_SYMBOL_REF): Added new constraint
21406 to allow or block "symbol_ref" depending on the value of TARGET_JSR.
21407 * config/rx/rx.md (call_internal): Use CALL_OP_SYMBOL_REF.
21408 * config/rx/rx.md (call_value_internal): Use CALL_OP_SYMBOL_REF.
21410 2018-02-09 Pierre-Marie de Rodat <derodat@adacore.com>
21413 * dwarf2out.c (is_trivial_indirect_ref): New function.
21414 (dwarf2out_late_global_decl): Do not generate a location
21415 attribute for variables that have a non-trivial DECL_VALUE_EXPR
21416 and that are not defined in the current unit.
21418 2018-02-09 Eric Botcazou <ebotcazou@adacore.com>
21420 * optabs.c (prepare_cmp_insn): Try harder to emit a direct comparison
21421 instead of a libcall for UNORDERED.
21423 2018-02-09 Tamar Christina <tamar.christina@arm.com>
21426 * config/arm/arm-c.c (arm_cpu_builtins): Un-define __ARM_FEATURE_LDREX,
21427 __ARM_ARCH_PROFILE, __ARM_ARCH_ISA_THUMB, __ARM_FP and __ARM_NEON_FP.
21429 2018-02-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21432 * config/s390/s390.c (s390_set_current_function): Invoke
21433 s390_indirect_branch_settings also if fndecl didn't change.
21435 2018-02-09 Alexandre Oliva <aoliva@redhat.com>
21437 * config/rs6000/rs6000.md (blockage): Set length to zero.
21439 2018-02-09 Eric Botcazou <ebotcazou@adacore.com>
21441 * expr.c (optimize_bitfield_assignment_op): Remove obsolete assertion.
21443 2018-02-09 Jakub Jelinek <jakub@redhat.com>
21446 * gcc.c (STATIC_LIBASAN_LIBS, STATIC_LIBTSAN_LIBS,
21447 STATIC_LIBLSAN_LIBS, STATIC_LIBUBSAN_LIBS): Handle -static like
21451 * var-tracking.c (vt_add_function_parameter): Punt for non-onepart
21452 PARALLEL incoming that failed vt_get_decl_and_offset check.
21454 PR middle-end/84237
21455 * output.h (bss_initializer_p): Add NAMED argument, defaulted to false.
21456 * varasm.c (bss_initializer_p): Add NAMED argument, if true, ignore
21458 (get_variable_section): For decls in named .bss* sections pass true as
21459 second argument to bss_initializer_p.
21461 2018-02-09 Marek Polacek <polacek@redhat.com>
21462 Jakub Jelinek <jakub@redhat.com>
21465 * fold-const.c (fold_indirect_ref_1): Use VECTOR_TYPE_P macro.
21466 Formatting fixes. Verify first that tree_fits_poly_int64_p (op01).
21467 Sync some changes from cxx_fold_indirect_ref.
21469 2018-02-09 Alexandre Oliva <aoliva@redhat.com>
21471 * cfgexpand.c (expand_gimple_basic_block): Handle inline entry
21473 * dwarf2out.c (dwarf2_debug_hooks): Enable inline_entry hook.
21474 (BLOCK_INLINE_ENTRY_LABEL): New.
21475 (dwarf2out_var_location): Disregard inline entry markers.
21476 (inline_entry_data): New struct.
21477 (inline_entry_data_hasher): New hashtable type.
21478 (inline_entry_data_hasher::hash): New.
21479 (inline_entry_data_hasher::equal): New.
21480 (inline_entry_data_table): New variable.
21481 (add_high_low_attributes): Add DW_AT_entry_pc and
21482 DW_AT_GNU_entry_view attributes if a pending entry is found
21483 in inline_entry_data_table. Add old entry_pc attribute only
21484 if debug nonbinding markers are disabled.
21485 (gen_inlined_subroutine_die): Set BLOCK_DIE if nonbinding
21486 markers are enabled.
21487 (block_within_block_p, dwarf2out_inline_entry): New.
21488 (dwarf2out_finish): Check that no entries remained in
21489 inline_entry_data_table.
21490 * final.c (reemit_insn_block_notes): Handle inline entry notes.
21491 (final_scan_insn, notice_source_line): Likewise.
21492 (rest_of_clean_state): Skip inline entry markers.
21493 * gimple-pretty-print.c (dump_gimple_debug): Handle inline entry
21495 * gimple.c (gimple_build_debug_inline_entry): New.
21496 * gimple.h (enum gimple_debug_subcode): Add
21497 GIMPLE_DEBUG_INLINE_ENTRY.
21498 (gimple_build_debug_inline_entry): Declare.
21499 (gimple_debug_inline_entry_p): New.
21500 (gimple_debug_nonbind_marker_p): Adjust.
21501 * insn-notes.def (INLINE_ENTRY): New.
21502 * print-rtl.c (rtx_writer::print_rtx_operand_code_0): Handle
21503 inline entry marker notes.
21504 (print_insn): Likewise.
21505 * rtl.h (NOTE_MARKER_P): Add INLINE_ENTRY support.
21506 (INSN_DEBUG_MARKER_KIND): Likewise.
21507 (GEN_RTX_DEBUG_MARKER_INLINE_ENTRY_PAT): New.
21508 * tree-inline.c (expand_call_inline): Build and insert
21509 debug_inline_entry stmt.
21510 * tree-ssa-live.c (remove_unused_scope_block_p): Preserve
21511 inline entry blocks early, if nonbind markers are enabled.
21512 (dump_scope_block): Dump fragment info.
21513 * var-tracking.c (reemit_marker_as_note): Handle inline entry note.
21514 * doc/gimple.texi (gimple_debug_inline_entry_p): New.
21515 (gimple_build_debug_inline_entry): New.
21516 * doc/invoke.texi (gstatement-frontiers, gno-statement-frontiers):
21517 Enable/disable inline entry points too.
21518 * doc/rtl.texi (NOTE_INSN_INLINE_ENTRY): New.
21519 (DEBUG_INSN): Describe inline entry markers.
21521 * common.opt (gvariable-location-views): New.
21522 (gvariable-location-views=incompat5): New.
21523 * config.in: Rebuilt.
21524 * configure: Rebuilt.
21525 * configure.ac: Test assembler for view support.
21526 * dwarf2asm.c (dw2_asm_output_symname_uleb128): New.
21527 * dwarf2asm.h (dw2_asm_output_symname_uleb128): Declare.
21528 * dwarf2out.c (var_loc_view): New typedef.
21529 (struct dw_loc_list_struct): Add vl_symbol, vbegin, vend.
21530 (dwarf2out_locviews_in_attribute): New.
21531 (dwarf2out_locviews_in_loclist): New.
21532 (dw_val_equal_p): Compare val_view_list of dw_val_class_view_lists.
21533 (enum dw_line_info_opcode): Add LI_adv_address.
21534 (struct dw_line_info_table): Add view.
21535 (RESET_NEXT_VIEW, RESETTING_VIEW_P): New macros.
21536 (DWARF2_ASM_VIEW_DEBUG_INFO): Define default.
21537 (zero_view_p): New variable.
21538 (ZERO_VIEW_P): New macro.
21539 (output_asm_line_debug_info): New.
21540 (struct var_loc_node): Add view.
21541 (add_AT_view_list, AT_loc_list): New.
21542 (add_var_loc_to_decl): Add view param. Test it against last.
21543 (new_loc_list): Add view params. Record them.
21544 (AT_loc_list_ptr): Handle loc and view lists.
21545 (view_list_to_loc_list_val_node): New.
21546 (print_dw_val): Handle dw_val_class_view_list.
21547 (size_of_die): Likewise.
21548 (value_format): Likewise.
21549 (loc_list_has_views): New.
21550 (gen_llsym): Set vl_symbol too.
21551 (maybe_gen_llsym, skip_loc_list_entry): New.
21552 (dwarf2out_maybe_output_loclist_view_pair): New.
21553 (output_loc_list): Output view list or entries too.
21554 (output_view_list_offset): New.
21555 (output_die): Handle dw_val_class_view_list.
21556 (output_dwarf_version): New.
21557 (output_compilation_unit_header): Use it.
21558 (output_skeleton_debug_sections): Likewise.
21559 (output_rnglists, output_line_info): Likewise.
21560 (output_pubnames, output_aranges): Update version comments.
21561 (output_one_line_info_table): Output view numbers in asm comments.
21562 (dw_loc_list): Determine current endview, pass it to new_loc_list.
21563 Call maybe_gen_llsym.
21564 (loc_list_from_tree_1): Adjust.
21565 (add_AT_location_description): Create view list attribute if
21566 needed, check it's absent otherwise.
21567 (convert_cfa_to_fb_loc_list): Adjust.
21568 (maybe_emit_file): Call output_asm_line_debug_info for test.
21569 (dwarf2out_var_location): Reset views as needed. Precompute
21570 add_var_loc_to_decl args. Call get_attr_min_length only if we have the
21571 attribute. Set view.
21572 (new_line_info_table): Reset next view.
21573 (set_cur_line_info_table): Call output_asm_line_debug_info for test.
21574 (dwarf2out_source_line): Likewise. Output view resets and labels to
21575 the assembler, or select appropriate line info opcodes.
21576 (prune_unused_types_walk_attribs): Handle dw_val_class_view_list.
21577 (optimize_string_length): Catch it. Adjust.
21578 (resolve_addr): Copy vl_symbol along with ll_symbol. Handle
21579 dw_val_class_view_list, and remove it if no longer needed.
21580 (hash_loc_list): Hash view numbers.
21581 (loc_list_hasher::equal): Compare them.
21582 (optimize_location_lists): Check whether a view list symbol is
21583 needed, and whether the locview attribute is present, and
21584 whether they match. Remove the locview attribute if no longer
21586 (index_location_lists): Call skip_loc_list_entry for test.
21587 (dwarf2out_finish): Call output_asm_line_debug_info for test.
21588 Use output_dwarf_version.
21589 * dwarf2out.h (enum dw_val_class): Add dw_val_class_view_list.
21590 (struct dw_val_node): Add val_view_list.
21591 * final.c (SEEN_NEXT_VIEW): New.
21592 (set_next_view_needed): New.
21593 (clear_next_view_needed): New.
21594 (maybe_output_next_view): New.
21595 (final_start_function): Rename to...
21596 (final_start_function_1): ... this. Take pointer to FIRST,
21597 add SEEN parameter. Emit param bindings in the initial view.
21598 (final_start_function): Reintroduce SEEN-less interface.
21599 (final): Rename to...
21600 (final_1): ... this. Take SEEN parameter. Output final pending
21601 next view at the end.
21602 (final): Reintroduce seen-less interface.
21603 (final_scan_insn): Output pending next view before switching
21604 sections or ending a block. Mark the next view as needed when
21605 outputting variable locations. Notify debug backend of section
21606 changes, and of location view changes.
21607 (rest_of_handle_final): Adjust.
21608 * toplev.c (process_options): Autodetect value for debug variable
21609 location views option. Warn on incompat5 without -gdwarf-5.
21610 * doc/invoke.texi (gvariable-location-views): New.
21611 (gvariable-location-views=incompat5): New.
21612 (gno-variable-location-views): New.
21614 2018-02-08 David Malcolm <dmalcolm@redhat.com>
21616 PR tree-optimization/84136
21617 * tree-cfg.c (find_taken_edge_computed_goto): Remove assertion
21618 that the result of find_edge is non-NULL.
21620 2018-02-08 Sergey Shalnov <sergey.shalnov@intel.com>
21623 * config/i386/x86-tune-costs.h (skylake_cost): Fix cost of
21624 storing integer register in SImode. Fix cost of 256 and 512
21625 byte aligned SSE register store.
21627 2018-02-08 Sergey Shalnov <sergey.shalnov@intel.com>
21629 * config/i386/i386.c (ix86_multiplication_cost): Fix
21630 multiplication cost for TARGET_AVX512DQ.
21632 2018-02-08 Marek Polacek <polacek@redhat.com>
21634 PR tree-optimization/84238
21635 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Verify the result of
21638 2018-02-08 Richard Sandiford <richard.sandiford@linaro.org>
21640 PR tree-optimization/84265
21641 * tree-vect-stmts.c (vectorizable_store): Don't treat
21642 VMAT_CONTIGUOUS accesses as grouped.
21643 (vectorizable_load): Likewise.
21645 2018-02-08 Richard Sandiford <richard.sandiford@linaro.org>
21647 PR tree-optimization/81635
21648 * wide-int.h (wi::round_down_for_mask, wi::round_up_for_mask): Declare.
21649 * wide-int.cc (wi::round_down_for_mask, wi::round_up_for_mask)
21650 (test_round_for_mask): New functions.
21651 (wide_int_cc_tests): Call test_round_for_mask.
21652 * tree-vrp.h (intersect_range_with_nonzero_bits): Declare.
21653 * tree-vrp.c (intersect_range_with_nonzero_bits): New function.
21654 * tree-data-ref.c (split_constant_offset_1): Use it to refine the
21655 range returned by get_range_info.
21657 2018-02-08 Jan Hubicka <hubicka@ucw.cz>
21660 * cgraph.h (symtab_node::output_to_lto_symbol_table_p): Declare
21661 * symtab.c: Include builtins.h
21662 (symtab_node::output_to_lto_symbol_table_p): Move here
21663 from lto-streamer-out.c:output_symbol_p.
21664 * lto-streamer-out.c (write_symbol): Turn early exit to assert.
21665 (output_symbol_p): Move all logic to symtab.c
21666 (produce_symtab): Update.
21668 2018-02-08 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21670 * config/s390/s390-opts.h (enum indirect_branch): Define.
21671 * config/s390/s390-protos.h (s390_return_addr_from_memory)
21672 (s390_indirect_branch_via_thunk)
21673 (s390_indirect_branch_via_inline_thunk): Add function prototypes.
21674 (enum s390_indirect_branch_type): Define.
21675 * config/s390/s390.c (struct s390_frame_layout, struct
21676 machine_function): Remove.
21677 (indirect_branch_prez10thunk_mask, indirect_branch_z10thunk_mask)
21678 (indirect_branch_table_label_no, indirect_branch_table_name):
21680 (INDIRECT_BRANCH_NUM_OPTIONS): Define macro.
21681 (enum s390_indirect_branch_option): Define.
21682 (s390_return_addr_from_memory): New function.
21683 (s390_handle_string_attribute): New function.
21684 (s390_attribute_table): Add new attribute handler.
21685 (s390_execute_label): Handle UNSPEC_EXECUTE_JUMP patterns.
21686 (s390_indirect_branch_via_thunk): New function.
21687 (s390_indirect_branch_via_inline_thunk): New function.
21688 (s390_function_ok_for_sibcall): When jumping via thunk disallow
21689 sibling call optimization for non z10 compiles.
21690 (s390_emit_call): Force indirect branch target to be a single
21691 register. Add r1 clobber for non-z10 compiles.
21692 (s390_emit_epilogue): Emit return jump via return_use expander.
21693 (s390_reorg): Handle JUMP_INSNs as execute targets.
21694 (s390_option_override_internal): Perform validity checks for the
21695 new command line options.
21696 (s390_indirect_branch_attrvalue): New function.
21697 (s390_indirect_branch_settings): New function.
21698 (s390_set_current_function): Invoke s390_indirect_branch_settings.
21699 (s390_output_indirect_thunk_function): New function.
21700 (s390_code_end): Implement target hook.
21701 (s390_case_values_threshold): Implement target hook.
21702 (TARGET_ASM_CODE_END, TARGET_CASE_VALUES_THRESHOLD): Define target
21704 * config/s390/s390.h (struct s390_frame_layout)
21705 (struct machine_function): Move here from s390.c.
21706 (TARGET_INDIRECT_BRANCH_NOBP_RET)
21707 (TARGET_INDIRECT_BRANCH_NOBP_JUMP)
21708 (TARGET_INDIRECT_BRANCH_NOBP_JUMP_THUNK)
21709 (TARGET_INDIRECT_BRANCH_NOBP_JUMP_INLINE_THUNK)
21710 (TARGET_INDIRECT_BRANCH_NOBP_CALL)
21711 (TARGET_DEFAULT_INDIRECT_BRANCH_TABLE)
21712 (TARGET_INDIRECT_BRANCH_THUNK_NAME_EXRL)
21713 (TARGET_INDIRECT_BRANCH_THUNK_NAME_EX)
21714 (TARGET_INDIRECT_BRANCH_TABLE): Define macros.
21715 * config/s390/s390.md (UNSPEC_EXECUTE_JUMP)
21716 (INDIRECT_BRANCH_THUNK_REGNUM): Define constants.
21717 (mnemonic attribute): Add values which aren't recognized
21719 ("*cjump_long", "*icjump_long", "*basr", "*basr_r"): Disable
21720 pattern for branch conversion. Fix mnemonic attribute.
21721 ("*c<code>", "*sibcall_br", "*sibcall_value_br", "*return"): Emit
21722 indirect branch via thunk if requested.
21723 ("indirect_jump", "<code>"): Expand patterns for branch conversion.
21724 ("*indirect_jump"): Disable for branch conversion using out of
21726 ("indirect_jump_via_thunk<mode>_z10")
21727 ("indirect_jump_via_thunk<mode>")
21728 ("indirect_jump_via_inlinethunk<mode>_z10")
21729 ("indirect_jump_via_inlinethunk<mode>", "*casesi_jump")
21730 ("casesi_jump_via_thunk<mode>_z10", "casesi_jump_via_thunk<mode>")
21731 ("casesi_jump_via_inlinethunk<mode>_z10")
21732 ("casesi_jump_via_inlinethunk<mode>", "*basr_via_thunk<mode>_z10")
21733 ("*basr_via_thunk<mode>", "*basr_r_via_thunk_z10")
21734 ("*basr_r_via_thunk", "return<mode>_prez10"): New pattern.
21735 ("*indirect2_jump"): Disable for branch conversion.
21736 ("casesi_jump"): Turn into expander and expand patterns for branch
21738 ("return_use"): New expander.
21739 ("*return"): Emit return via thunk and rename it to ...
21740 ("*return<mode>"): ... this one.
21741 * config/s390/s390.opt: Add new options and and enum for the
21744 2018-02-08 Richard Sandiford <richard.sandiford@linaro.org>
21746 * lra-constraints.c (match_reload): Unconditionally use
21747 gen_lowpart_SUBREG, rather than selecting between that
21748 and equivalent gen_rtx_SUBREG code.
21750 2018-02-08 Richard Biener <rguenther@suse.de>
21752 PR tree-optimization/84233
21753 * tree-ssa-phiprop.c (propagate_with_phi): Use separate
21754 changed flag instead of boguously re-using phi_inserted.
21756 2018-02-08 Martin Jambor <mjambor@suse.cz>
21758 * hsa-gen.c (get_symbol_for_decl): Set program allocation for
21759 static local variables.
21761 2018-02-08 Richard Biener <rguenther@suse.de>
21763 PR tree-optimization/84278
21764 * tree-vect-stmts.c (vectorizable_store): When looking for
21765 smaller vector types to perform grouped strided loads/stores
21766 make sure the mode is supported by the target.
21767 (vectorizable_load): Likewise.
21769 2018-02-08 Wilco Dijkstra <wdijkstr@arm.com>
21771 * config/aarch64/aarch64.c (aarch64_components_for_bb):
21772 Increase LDP/STP opportunities by adding adjacent callee-saves.
21774 2018-02-08 Wilco Dijkstra <wdijkstr@arm.com>
21776 PR rtl-optimization/84068
21777 PR rtl-optimization/83459
21778 * haifa-sched.c (rank_for_schedule): Fix SCHED_PRESSURE_MODEL sorting.
21780 2018-02-08 Aldy Hernandez <aldyh@redhat.com>
21782 PR tree-optimization/84224
21783 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Remove assert.
21784 * calls.c (gimple_alloca_call_p): Only return TRUE when we have
21785 non-zero arguments.
21787 2018-02-07 Iain Sandoe <iain@codesourcery.com>
21790 * config/rs6000/altivec.md (*restore_world): Remove LR use.
21791 * config/rs6000/predicates.md (restore_world_operation): Adjust op
21792 count, remove one USE.
21794 2018-02-07 Michael Meissner <meissner@linux.vnet.ibm.com>
21796 * doc/install.texi (Configuration): Document the
21797 --with-long-double-format={ibm,ieee} PowerPC configuration
21801 * config/rs6000/rs6000.md (fix_trunc<SFDF:mode><QHI:mode>2):
21802 Convert from define_expand to be define_insn_and_split. Rework
21803 float/double/_Float128 conversions to QI/HI/SImode to work with
21804 both ISA 2.07 (power8) or ISA 3.0 (power9). Fix regression where
21805 conversions to QI/HImode types did a store and then a load to
21806 truncate the value. For conversions to VSX registers, don't split
21807 the insn, instead emit the code directly. Use the code iterator
21808 any_fix to combine signed and unsigned conversions.
21809 (fix<uns>_trunc<SFDF:mode>si2_p8): Likewise.
21810 (fixuns_trunc<SFDF:mode><QHI:mode>2): Likewise.
21811 (fix_trunc<IEEE128:mode><QHI:mode>2): Likewise.
21812 (fix<uns>_trunc<SFDF:mode><QHI:mode>2): Likewise.
21813 (fix_<mode>di2_hw): Likewise.
21814 (fixuns_<mode>di2_hw): Likewise.
21815 (fix_<mode>si2_hw): Likewise.
21816 (fixuns_<mode>si2_hw): Likewise.
21817 (fix<uns>_<IEEE128:mode><SDI:mode>2_hw): Likewise.
21818 (fix<uns>_trunc<IEEE128:mode><QHI:mode>2): Likewise.
21819 (fctiw<u>z_<mode>_smallint): Rename fctiw<u>z_<mode>_smallint to
21820 fix<uns>_trunc<SFDF:mode>si2_p8.
21821 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Delete, no longer used.
21822 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
21823 (fix<uns>_<mode>_mem): Likewise.
21824 (fctiw<u>z_<mode>_mem): Likewise.
21825 (fix<uns>_<mode>_mem): Likewise.
21826 (fix<uns>_trunc<SFDF:mode><QHSI:mode>2_mem): On ISA 3.0, prevent
21827 the register allocator from doing a direct move to the GPRs to do
21828 a store, and instead use the ISA 3.0 store byte/half-word from
21829 vector register instruction. For IEEE 128-bit floating point,
21830 also optimize stores of 32-bit ints.
21831 (fix<uns>_trunc<IEEE128:mode><QHSI:mode>2_mem): Likewise.
21833 2018-02-07 Alan Hayward <alan.hayward@arm.com>
21835 * genextract.c (push_pathstr_operand): New function to support
21837 (walk_rtx): Call push_pathstr_operand.
21838 (print_path): Support [a-zA-Z].
21840 2018-02-07 Richard Biener <rguenther@suse.de>
21842 PR tree-optimization/84037
21843 * tree-vectorizer.h (struct _loop_vec_info): Add ivexpr_map member.
21844 (cse_and_gimplify_to_preheader): Declare.
21845 (vect_get_place_in_interleaving_chain): Likewise.
21846 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
21848 (_loop_vec_info::~_loop_vec_info): Delete it.
21849 (cse_and_gimplify_to_preheader): New function.
21850 * tree-vect-slp.c (vect_get_place_in_interleaving_chain): Export.
21851 * tree-vect-stmts.c (vectorizable_store): CSE base and steps.
21852 (vectorizable_load): Likewise. For grouped stores always base
21853 the IV on the first element.
21854 * tree-vect-loop-manip.c (vect_loop_versioning): Unshare versioning
21855 condition before gimplifying.
21857 2018-02-07 Jakub Jelinek <jakub@redhat.com>
21859 * tree-eh.c (operation_could_trap_helper_p): Ignore honor_trapv for
21860 *DIV_EXPR and *MOD_EXPR.
21862 2018-02-07 H.J. Lu <hongjiu.lu@intel.com>
21865 * config/i386/i386.c (ix86_option_override_internal): Mask out
21866 the CF_SET bit when checking -fcf-protection.
21868 2018-02-07 Tom de Vries <tom@codesourcery.com>
21871 * omp-expand.c (expand_oacc_collapse_init): Ensure diff_type is large
21874 2018-02-07 Richard Biener <rguenther@suse.de>
21876 PR tree-optimization/84204
21877 * tree-chrec.c (chrec_fold_plus_1): Remove size limiting in
21880 PR tree-optimization/84205
21881 * graphite-isl-ast-to-gimple.c (binary_op_to_tree): Also
21882 special-case isl_ast_op_zdiv_r.
21884 PR tree-optimization/84223
21885 * graphite-scop-detection.c (gather_bbs::before_dom_children):
21886 Only add conditions from within the region.
21887 (gather_bbs::after_dom_children): Adjust.
21889 2018-02-07 Georg-Johann Lay <avr@gjlay.de>
21892 * config/avr/avr.h (GENERAL_REGNO_P, GENERAL_REG_P): New macros.
21893 * config/avr/avr.md: Only post-reload split REG-REG moves if
21894 either register is GENERAL_REG_P.
21896 2018-02-07 Jakub Jelinek <jakub@redhat.com>
21898 PR tree-optimization/84235
21899 * tree-ssa-scopedtables.c
21900 (avail_exprs_stack::simplify_binary_operation): Fir MINUS_EXPR, punt
21901 if the subtraction is performed in floating point type where NaNs are
21902 honored. For *DIV_EXPR, punt for ALL_FRACT_MODE_Ps where we can't
21903 build 1. Formatting fix.
21905 2018-02-06 Jakub Jelinek <jakub@redhat.com>
21908 * config/i386/i386.c (rest_of_insert_endbranch): Only skip
21909 NOTE_INSN_CALL_ARG_LOCATION after a call, not anything else,
21910 and skip it regardless of bb boundaries. Use CALL_P macro,
21911 don't test INSN_P (insn) together with CALL_P or JUMP_P check
21912 unnecessarily, formatting fix.
21914 2018-02-06 Michael Collison <michael.collison@arm.com>
21916 * config/arm/thumb2.md:
21917 (*thumb2_mov_negscc): Split only if TARGET_THUMB2 && !arm_restrict_it.
21918 (*thumb_mov_notscc): Ditto.
21920 2018-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
21923 * config/rs6000/rs6000.md (su code attribute): Use "u" for
21924 unsigned_fix, not "s".
21926 2018-02-06 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
21928 * configure.ac (gcc_fn_eh_frame_ro): New function.
21929 (gcc_cv_as_cfi_directive): Check both 32 and 64-bit assembler for
21930 correct .eh_frame permissions.
21931 * configure: Regenerate.
21933 2018-02-06 Andrew Jenner <andrew@codeourcery.com>
21935 * doc/invoke.texi: Add section for the PowerPC SPE backend. Remove
21936 irrelevant options.
21938 2018-02-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21940 * config/rs6000/rs6000.c (rs6000_option_override_internal):
21941 Display warning message for -mno-speculate-indirect-jumps.
21943 2018-02-06 Andrew Jenner <andrew@codesourcery.com>
21945 * config/powerpcspe/powerpcspe.opt: (msimple-fpu, mfpu) Add
21947 * config/powerpcspe/sysv4.opt (mbit-align): Likewise.
21949 2018-02-06 Aldy Hernandez <aldyh@redhat.com>
21951 PR tree-optimization/84225
21952 * tree-eh.c (find_trapping_overflow): Only call
21953 operation_no_trapping_overflow when ANY_INTEGRAL_TYPE_P.
21955 2018-02-06 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
21958 * config/i386/i386.c: Reimplement the check of possible options
21959 -mibt/-mshstk conbination. Change error messages.
21960 * doc/invoke.texi: Fix a typo: remove extra '='.
21962 2018-02-06 Marek Polacek <polacek@redhat.com>
21964 PR tree-optimization/84228
21965 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Skip debug statements.
21967 2018-02-06 Tamar Christina <tamar.christina@arm.com>
21970 * config/arm/arm.c (arm_print_asm_arch_directives): Record already
21971 emitted arch directives.
21972 * config/arm/arm-c.c (arm_cpu_builtins): Undefine __ARM_ARCH and
21973 __ARM_FEATURE_COPROC before changing architectures.
21975 2018-02-06 Richard Biener <rguenther@suse.de>
21977 * config/i386/i386.c (print_reg): Fix typo.
21978 (ix86_loop_unroll_adjust): Do not unroll beyond the original nunroll.
21980 2018-02-06 Eric Botcazou <ebotcazou@adacore.com>
21982 * configure: Regenerate.
21984 2018-02-05 Martin Sebor <msebor@redhat.com>
21986 PR tree-optimization/83369
21987 * tree-ssa-ccp.c (pass_post_ipa_warn::execute): Use %G to print
21990 2018-02-05 Martin Liska <mliska@suse.cz>
21992 * doc/invoke.texi: Cherry-pick upstream r323995.
21994 2018-02-05 Richard Sandiford <richard.sandiford@linaro.org>
21996 * ira.c (ira_init_register_move_cost): Adjust comment.
21998 2018-02-05 Martin Liska <mliska@suse.cz>
22000 PR gcov-profile/84137
22001 * doc/gcov.texi: Fix typo in documentation.
22003 2018-02-05 Martin Liska <mliska@suse.cz>
22005 PR gcov-profile/83879
22006 * doc/gcov.texi: Document necessity of --dynamic-list-data when
22007 using dlopen functionality.
22009 2018-02-05 Olga Makhotina <olga.makhotina@intel.com>
22011 * config/i386/avx512dqintrin.h (_mm_mask_range_sd, _mm_maskz_range_sd,
22012 _mm_mask_range_round_sd, _mm_maskz_range_round_sd, _mm_mask_range_ss,
22013 _mm_maskz_range_ss, _mm_mask_range_round_ss,
22014 _mm_maskz_range_round_ss): New intrinsics.
22015 (__builtin_ia32_rangesd128_round)
22016 (__builtin_ia32_rangess128_round): Remove.
22017 (__builtin_ia32_rangesd128_mask_round,
22018 __builtin_ia32_rangess128_mask_round): New builtins.
22019 * config/i386/i386-builtin.def (__builtin_ia32_rangesd128_round,
22020 __builtin_ia32_rangess128_round): Remove.
22021 (__builtin_ia32_rangesd128_mask_round,
22022 __builtin_ia32_rangess128_mask_round): New builtins.
22023 * config/i386/sse.md (ranges<mode><round_saeonly_name>): Renamed to ...
22024 (ranges<mode><mask_scalar_name><round_saeonly_scalar_name>): ... this.
22025 ((match_operand:VF_128 2 "<round_saeonly_nimm_predicate>"
22026 "<round_saeonly_constraint>")): Changed to ...
22027 ((match_operand:VF_128 2 "<round_saeonly_scalar_nimm_predicate>"
22028 "<round_saeonly_scalar_constraint>")): ... this.
22029 ("vrange<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
22030 %0, %1, %2<round_saeonly_op4>, %3}"): Changed to ...
22031 ("vrange<ssescalarmodesuffix>\t{%3, <round_saeonly_scalar_mask_op4>%2,
22032 %1, %0<mask_scalar_operand4>|%0<mask_scalar_operand4>, %1,
22033 %2<round_saeonly_scalar_mask_op4>, %3}"): ... this.
22035 2018-02-02 Andrew Jenner <andrew@codesourcery.com>
22037 * config/powerpcspe/powerpcspe.opt: Add Undocumented to irrelevant
22039 * config/powerpcspe/powerpcspe-tables.opt (rs6000_cpu_opt_value):
22040 Remove all values except native, 8540 and 8548.
22042 2018-02-02 H.J. Lu <hongjiu.lu@intel.com>
22044 * config/i386/i386.c (ix86_output_function_return): Pass
22045 INVALID_REGNUM, instead of -1, as invalid register number to
22046 indirect_thunk_name and output_indirect_thunk.
22048 2018-02-02 Julia Koval <julia.koval@intel.com>
22050 * config.gcc: Add -march=icelake.
22051 * config/i386/driver-i386.c (host_detect_local_cpu): Detect icelake.
22052 * config/i386/i386-c.c (ix86_target_macros_internal): Handle icelake.
22053 * config/i386/i386.c (processor_costs): Add m_ICELAKE.
22054 (PTA_ICELAKE, PTA_AVX512VNNI, PTA_GFNI, PTA_VAES, PTA_AVX512VBMI2,
22055 PTA_VPCLMULQDQ, PTA_RDPID, PTA_AVX512BITALG): New.
22056 (processor_target_table): Add icelake.
22057 (ix86_option_override_internal): Handle new PTAs.
22058 (get_builtin_code_for_version): Handle icelake.
22059 (M_INTEL_COREI7_ICELAKE): New.
22060 (fold_builtin_cpu): Handle icelake.
22061 * config/i386/i386.h (TARGET_ICELAKE, PROCESSOR_ICELAKE): New.
22062 * doc/invoke.texi: Add -march=icelake.
22064 2018-02-02 Julia Koval <julia.koval@intel.com>
22066 * config/i386/i386.c (ix86_option_override_internal): Change flags type
22067 to wide_int_bitmask.
22068 * wide-int-bitmask.h: New.
22070 2018-02-02 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
22073 * config/i386/i386.md: Replace Pmode with word_mode in
22074 builtin_setjmp_setup and builtin_longjmp to support x32.
22076 2018-02-01 Peter Bergner <bergner@vnet.ibm.com>
22080 * config/rs6000/driver-rs6000.c: #include "diagnostic.h".
22082 (rs6000_supported_cpu_names): New static variable.
22083 (linux_cpu_translation_table): Likewise.
22084 (elf_platform) <cpu>: Define new static variable and use it.
22085 Translate kernel AT_PLATFORM name to canonical name if needed.
22086 Error if platform name is unknown.
22088 2018-02-01 Aldy Hernandez <aldyh@redhat.com>
22091 * config/pa/predicates.md (base14_operand): Handle E_VOIDmode.
22093 2018-02-01 Jeff Law <law@redhat.com>
22096 * config/i386/i386.c (release_scratch_register_on_entry): Add new
22097 OFFSET and RELEASE_VIA_POP arguments. Use SP+OFFSET to restore
22098 the scratch if RELEASE_VIA_POP is false.
22099 (ix86_adjust_stack_and_probe_stack_clash): Un-constify SIZE.
22100 If we have to save a temporary register, decrement SIZE appropriately.
22101 Pass new arguments to release_scratch_register_on_entry.
22102 (ix86_adjust_stack_and_probe): Likewise.
22103 (ix86_emit_probe_stack_range): Pass new arguments to
22104 release_scratch_register_on_entry.
22106 2018-02-01 Uros Bizjak <ubizjak@gmail.com>
22108 PR rtl-optimization/84157
22109 * combine.c (change_zero_ext): Use REG_P predicate in
22110 front of HARD_REGISTER_P predicate.
22112 2018-02-01 Georg-Johann Lay <avr@gjlay.de>
22114 * config/avr/avr.c (avr_option_override): Move disabling of
22115 -fdelete-null-pointer-checks to...
22116 * common/config/avr/avr-common.c (avr_option_optimization_table):
22119 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
22121 PR tree-optimization/81635
22122 * tree-data-ref.c (split_constant_offset_1): For types that
22123 wrap on overflow, try to use range info to prove that wrapping
22126 2018-02-01 Renlin Li <renlin.li@arm.com>
22129 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
22130 TAILCALL_ADDR_REGS.
22131 (aarch64_register_move_cost): Likewise.
22132 * config/aarch64/aarch64.h (reg_class): Rename CALLER_SAVE_REGS to
22133 TAILCALL_ADDR_REGS.
22134 (REG_CLASS_NAMES): Likewise.
22135 (REG_CLASS_CONTENTS): Rename CALLER_SAVE_REGS to
22136 TAILCALL_ADDR_REGS. Remove IP registers.
22137 * config/aarch64/aarch64.md (Ucs): Update register constraint.
22139 2018-02-01 Richard Biener <rguenther@suse.de>
22141 * domwalk.h (dom_walker::dom_walker): Add additional constructor
22142 for specifying RPO order and allow NULL for that.
22143 * domwalk.c (dom_walker::dom_walker): Likewise.
22144 (dom_walker::walk): Handle NULL RPO order.
22145 * tree-into-ssa.c (rewrite_dom_walker): Do not walk dom children
22147 (rewrite_update_dom_walker): Likewise.
22148 (mark_def_dom_walker): Likewise.
22150 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
22152 * config/aarch64/aarch64-protos.h (aarch64_split_sve_subreg_move)
22153 (aarch64_maybe_expand_sve_subreg_move): Declare.
22154 * config/aarch64/aarch64.md (UNSPEC_REV_SUBREG): New unspec.
22155 * config/aarch64/predicates.md (aarch64_any_register_operand): New
22157 * config/aarch64/aarch64-sve.md (mov<mode>): Optimize subreg moves
22158 that are semantically a reverse operation.
22159 (*aarch64_sve_mov<mode>_subreg_be): New pattern.
22160 * config/aarch64/aarch64.c (aarch64_maybe_expand_sve_subreg_move):
22161 (aarch64_replace_reg_mode, aarch64_split_sve_subreg_move): New
22163 (aarch64_can_change_mode_class): For big-endian, forbid changes
22164 between two SVE modes if they have different element sizes.
22166 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
22168 * config/aarch64/aarch64.c (aarch64_expand_sve_const_vector): Prefer
22169 the TImode handling for big-endian targets.
22171 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
22173 * config/aarch64/aarch64-sve.md (sve_ld1rq): Replace with...
22174 (*sve_ld1rq<Vesize>): ... this new pattern. Handle all element sizes,
22176 * config/aarch64/aarch64.c (aarch64_expand_sve_widened_duplicate):
22177 Remove BSWAP handing for big-endian targets and use the form of
22178 LD1RQ appropariate for the mode.
22180 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
22182 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Handle
22183 all CONST_VECTOR_DUPLICATE_P vectors, not just those with a single
22184 duplicated element.
22186 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
22189 * config/aarch64/aarch64.c (aarch64_secondary_reload): Tighten
22190 check for operands that need to go through aarch64_sve_reload_be.
22192 2018-02-01 Jakub Jelinek <jakub@redhat.com>
22194 PR tree-optimization/81661
22195 PR tree-optimization/84117
22196 * tree-eh.h (rewrite_to_non_trapping_overflow): Declare.
22197 * tree-eh.c: Include gimplify.h.
22198 (find_trapping_overflow, replace_trapping_overflow,
22199 rewrite_to_non_trapping_overflow): New functions.
22200 * tree-vect-loop.c: Include tree-eh.h.
22201 (vect_get_loop_niters): Use rewrite_to_non_trapping_overflow.
22202 * tree-data-ref.c: Include tree-eh.h.
22203 (get_segment_min_max): Use rewrite_to_non_trapping_overflow.
22205 2018-01-31 Uros Bizjak <ubizjak@gmail.com>
22207 PR rtl-optimization/84123
22208 * combine.c (change_zero_ext): Check if hard register satisfies
22209 can_change_dest_mode before calling gen_lowpart_SUBREG.
22211 2018-01-31 Vladimir Makarov <vmakarov@redhat.com>
22214 * ira.c (ira_init_register_move_cost): Remove assert.
22216 2018-01-31 Eric Botcazou <ebotcazou@adacore.com>
22218 PR rtl-optimization/84071
22219 * doc/tm.texi.in (WORD_REGISTER_OPERATIONS): Add explicit case.
22220 * doc/tm.texi: Regenerate.
22222 2018-01-31 Richard Biener <rguenther@suse.de>
22224 PR tree-optimization/84132
22225 * tree-data-ref.c (analyze_miv_subscript): Properly
22226 check whether evolution_function_is_affine_multivariate_p
22227 before calling gcd_of_steps_may_divide_p.
22229 2018-01-31 Julia Koval <julia.koval@intel.com>
22232 * config/i386/i386.c (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
22233 * config/i386/i386.md (rdpid_rex64) New.
22234 (rdpid): Make 32bit only.
22236 2018-01-29 Aldy Hernandez <aldyh@redhat.com>
22239 * tree-pretty-print.c (dump_generic_node): Handle a TYPE_NAME with
22240 an IDENTIFIER_NODE for FUNCTION_TYPE's.
22242 2018-01-31 Eric Botcazou <ebotcazou@adacore.com>
22245 2018-01-12 Eric Botcazou <ebotcazou@adacore.com>
22247 * config/sparc/sparc.md (vxworks_load_got): Set the GOT register.
22249 2018-01-31 Eric Botcazou <ebotcazou@adacore.com>
22251 PR rtl-optimization/84071
22252 * combine.c (record_dead_and_set_regs_1): Record the source unmodified
22253 for a paradoxical SUBREG on a WORD_REGISTER_OPERATIONS target.
22255 2018-01-31 Claudiu Zissulescu <claziss@synopsys.com>
22257 * config/arc/arc.c (arc_handle_aux_attribute): New function.
22258 (arc_attribute_table): Add 'aux' attribute.
22259 (arc_in_small_data_p): Consider aux like variables.
22260 (arc_is_aux_reg_p): New function.
22261 (arc_asm_output_aligned_decl_local): Ignore 'aux' like variables.
22262 (arc_get_aux_arg): New function.
22263 (prepare_move_operands): Handle aux-register access.
22264 (arc_handle_aux_attribute): New function.
22265 * doc/extend.texi (ARC Variable attributes): Add subsection.
22267 2018-01-31 Claudiu Zissulescu <claziss@synopsys.com>
22269 * config/arc/arc-protos.h (arc_is_uncached_mem_p): Function proto.
22270 * config/arc/arc.c (arc_handle_uncached_attribute): New function.
22271 (arc_attribute_table): Add 'uncached' attribute.
22272 (arc_print_operand): Print '.di' flag for uncached memory
22274 (arc_in_small_data_p): Do not consider for small data the uncached
22276 (arc_is_uncached_mem_p): New function.
22277 * config/arc/predicates.md (compact_store_memory_operand): Check
22278 for uncached memory accesses.
22279 (nonvol_nonimm_operand): Likewise.
22280 * doc/extend.texi (ARC Type Attribute): New subsection.
22282 2018-01-31 Jakub Jelinek <jakub@redhat.com>
22285 * common.opt (falign-functions=, falign-jumps=, falign-labels=,
22286 falign-loops=): Add Optimization flag.
22288 2018-01-30 Jeff Law <law@redhat.com>
22291 * i386.c (ix86_adjust_stack_and_probe_stack_clash): New argument
22292 INT_REGISTERS_SAVED. Check it prior to calling
22293 get_scratch_register_on_entry.
22294 (ix86_adjust_stack_and_probe): Similarly.
22295 (ix86_emit_probe_stack_range): Similarly.
22296 (ix86_expand_prologue): Corresponding changes.
22298 2018-01-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
22301 * config/sol2.h (STARTFILE_ARCH_SPEC): Use -std=c*,
22302 -std=iso9899:199409 instead of -pedantic to select values-Xc.o.
22304 2018-01-30 Vladimir Makarov <vmakarov@redhat.com>
22307 * lra-constraints.c (curr_insn_transform): Process AND in the
22310 2018-01-30 Jakub Jelinek <jakub@redhat.com>
22312 PR rtl-optimization/83986
22313 * sched-deps.c (sched_analyze_insn): For frame related insns, add anti
22314 dependence against last_pending_memory_flush in addition to
22315 pending_jump_insns.
22317 2018-01-30 Alexandre Oliva <aoliva@redhat.com>
22319 PR tree-optimization/81611
22320 * tree-ssa-dom.c (simple_iv_increment_p): Skip intervening
22323 2018-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
22326 * config/rs6000/rs6000.c (rs6000_internal_arg_pointer): Only return
22329 2018-01-30 Richard Biener <rguenther@suse.de>
22330 Jakub Jelinek <jakub@redhat.com>
22332 PR tree-optimization/84111
22333 * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely_1): Skip
22334 inner loops added during recursion, as they don't have up-to-date
22337 2018-01-30 Jan Hubicka <hubicka@ucw.cz>
22340 * ipa-inline.c (can_inline_edge_p): Break out late tests to...
22341 (can_inline_edge_by_limits_p): ... here.
22342 (can_early_inline_edge_p, check_callers,
22343 update_caller_keys, update_callee_keys, recursive_inlining,
22344 add_new_edges_to_heap, speculation_useful_p,
22345 inline_small_functions,
22346 inline_small_functions, flatten_function,
22347 inline_to_all_callers_1): Update.
22349 2018-01-30 Jan Hubicka <hubicka@ucw.cz>
22351 * profile-count.c (profile_count::combine_with_ipa_count): Handle
22354 2018-01-30 Richard Biener <rguenther@suse.de>
22356 PR tree-optimization/83008
22357 * tree-vect-slp.c (vect_analyze_slp_cost_1): Properly cost
22358 invariant and constant vector uses in stmts when they need
22359 more than one stmt.
22361 2018-01-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
22364 * configure.ac (gcc_cv_as_shf_merge): Disable on Solaris 10/x86.
22365 * configure: Regenerate.
22367 2018-01-30 Richard Sandiford <richard.sandiford@linaro.org>
22369 * config/aarch64/aarch64-sve.md (*vec_extract<mode><Vel>_0): New
22371 (*vec_extract<mode><Vel>_v128): Require a nonzero lane number.
22372 Use gen_rtx_REG rather than gen_lowpart.
22374 2018-01-30 Richard Sandiford <richard.sandiford@linaro.org>
22376 * lra-constraints.c (match_reload): Use subreg_lowpart_offset
22377 rather than 0 when creating partial subregs.
22379 2018-01-30 Richard Sandiford <richard.sandiford@linaro.org>
22381 * vec-perm-indices.c (vec_perm_indices::series_p): Give examples
22384 2018-01-29 Michael Meissner <meissner@linux.vnet.ibm.com>
22387 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): If DFmode
22388 and SFmode can go in Altivec registers (-mcpu=power7 for DFmode,
22389 -mcpu=power8 for SFmode) don't set the PRE_INCDEC or PRE_MODIFY
22390 flags. This restores the settings used before the 2017-07-24.
22391 Turning off pre increment/decrement/modify allows IVOPTS to
22392 optimize DF/SF loops where the index is an int.
22394 2018-01-29 Richard Biener <rguenther@suse.de>
22395 Kelvin Nilsen <kelvin@gcc.gnu.org>
22398 * tree-vect-stmts.c (vectorizable_call): Don't call
22399 targetm.vectorize_builtin_md_vectorized_function if callee is
22402 2018-01-22 Carl Love <cel@us.ibm.com>
22404 * doc/extend.tex: Fix typo in second arg in
22405 __builtin_bcdadd_{lt|eq|gt|ov} and __builtin_bcdsub_{lt|eq|gt|ov}.
22407 2018-01-29 Richard Biener <rguenther@suse.de>
22409 PR tree-optimization/84086
22410 * tree-ssanames.c: Include cfgloop.h and tree-scalar-evolution.h.
22411 (flush_ssaname_freelist): When SSA names were released reset
22412 the SCEV hash table.
22414 2018-01-29 Richard Biener <rguenther@suse.de>
22416 PR tree-optimization/84057
22417 * tree-ssa-loop-ivcanon.c (unloop_loops): Deal with already
22418 removed paths when removing edges.
22420 2018-01-27 H.J. Lu <hongjiu.lu@intel.com>
22422 * doc/invoke.texi: Replace -mfunction-return==@var{choice} with
22423 -mfunction-return=@var{choice}.
22425 2018-01-27 Bernd Edlinger <bernd.edlinger@hotmail.de>
22427 PR diagnostic/84034
22428 * diagnostic-show-locus.c (get_line_width_without_trailing_whitespace):
22429 Handle CR like TAB.
22430 (layout::print_source_line): Likewise.
22431 (test_get_line_width_without_trailing_whitespace): Add test cases.
22433 2018-01-27 Jakub Jelinek <jakub@redhat.com>
22435 PR middle-end/84040
22436 * sched-deps.c (sched_macro_fuse_insns): Return immediately for
22439 2018-01-26 Jim Wilson <jimw@sifive.com>
22441 * config/riscv/riscv.h (MAX_FIXED_MODE_SIZE): New.
22443 * config/riscv/elf.h (LIB_SPEC): Don't include -lgloss when nosys.specs
22446 2018-01-26 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22448 * config/aarch64/aarch64.md: Add peepholes for CMP + SUB -> SUBS
22449 and CMP + SUB-immediate -> SUBS.
22451 2018-01-26 Martin Sebor <msebor@redhat.com>
22453 PR tree-optimization/83896
22454 * tree-ssa-strlen.c (get_string_len): Rename...
22455 (get_string_cst_length): ...to this. Return HOST_WIDE_INT.
22456 Avoid assuming length is constant.
22457 (handle_char_store): Use HOST_WIDE_INT for string length.
22459 2018-01-26 Uros Bizjak <ubizjak@gmail.com>
22462 * config/i386/i386.md (*andndi3_doubleword): Add earlyclobber
22463 to (=&r,r,rm) alternative. Add (=r,0,rm) and (=r,r,0) alternatives.
22465 2018-01-26 Richard Biener <rguenther@suse.de>
22467 PR rtl-optimization/84003
22468 * dse.c (record_store): Only record redundant stores when
22469 the earlier store aliases at least all accesses the later one does.
22471 2018-01-26 Jakub Jelinek <jakub@redhat.com>
22473 PR rtl-optimization/83985
22474 * dce.c (deletable_insn_p): Return false for separate shrink wrapping
22475 REG_CFA_RESTORE insns.
22476 (delete_unmarked_insns): Don't ignore separate shrink wrapping
22477 REG_CFA_RESTORE insns here.
22480 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Don't
22481 use SSA_NAME_VAR as base for SSA_NAMEs with non-NULL SSA_NAME_VAR.
22483 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
22485 * config/arc/arc-arch.h (arc_tune_attr): Add ARC_TUNE_CORE_3.
22486 * config/arc/arc.c (arc_sched_issue_rate): Use ARC_TUNE_... .
22487 (arc_init): Likewise.
22488 (arc_override_options): Likewise.
22489 (arc_file_start): Choose Tag_ARC_CPU_variation based on arc_tune
22491 (hwloop_fail): Use TARGET_DBNZ when we want to check for dbnz insn
22493 * config/arc/arc.h (TARGET_DBNZ): Define.
22494 * config/arc/arc.md (attr tune): Add core_3, use ARC_TUNE_... to
22495 properly set the tune attribute.
22496 (dbnz): Use TARGET_DBNZ guard.
22497 * config/arc/arc.opt (mtune): Add core3 option.
22499 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
22501 * config/arc/arc.c (arc_delegitimize_address_0): Refactored to
22502 recognize new pic like addresses.
22503 (arc_delegitimize_address): Clean up.
22505 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
22507 * config/arc/arc-arches.def: Option mrf16 valid for all
22509 * config/arc/arc-c.def (__ARC_RF16__): New predefined macro.
22510 * config/arc/arc-cpus.def (em_mini): New cpu with rf16 on.
22511 * config/arc/arc-options.def (FL_RF16): Add mrf16 option.
22512 * config/arc/arc-tables.opt: Regenerate.
22513 * config/arc/arc.c (arc_conditional_register_usage): Handle
22514 reduced register file case.
22515 (arc_file_start): Set must have build attributes.
22516 * config/arc/arc.h (MAX_ARC_PARM_REGS): Conditional define using
22517 mrf16 option value.
22518 * config/arc/arc.opt (mrf16): Add new option.
22519 * config/arc/elf.h (ATTRIBUTE_PCS): Define.
22520 * config/arc/genmultilib.awk: Handle new mrf16 option.
22521 * config/arc/linux.h (ATTRIBUTE_PCS): Define.
22522 * config/arc/t-multilib: Regenerate.
22523 * doc/invoke.texi (ARC Options): Document mrf16 option.
22525 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
22527 * config/arc/arc-protos.h: Add arc_is_secure_call_p proto.
22528 * config/arc/arc.c (arc_handle_secure_attribute): New function.
22529 (arc_attribute_table): Add 'secure_call' attribute.
22530 (arc_print_operand): Print secure call operand.
22531 (arc_function_ok_for_sibcall): Don't optimize tail calls when
22533 (arc_is_secure_call_p): New function. * config/arc/arc.md
22534 (call_i): Add support for sjli instruction.
22535 (call_value_i): Likewise.
22536 * config/arc/constraints.md (Csc): New constraint.
22538 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
22539 John Eric Martin <John.Martin@emmicro-us.com>
22541 * config/arc/arc-protos.h: Add arc_is_jli_call_p proto.
22542 * config/arc/arc.c (_arc_jli_section): New struct.
22543 (arc_jli_section): New type.
22544 (rc_jli_sections): New static variable.
22545 (arc_handle_jli_attribute): New function.
22546 (arc_attribute_table): Add jli_always and jli_fixed attribute.
22547 (arc_file_end): New function.
22548 (TARGET_ASM_FILE_END): Define.
22549 (arc_print_operand): Reuse 'S' letter for JLI output instruction.
22550 (arc_add_jli_section): New function.
22551 (jli_call_scan): Likewise.
22552 (arc_reorg): Call jli_call_scan.
22553 (arc_output_addsi): Remove 'S' from printing asm operand.
22554 (arc_is_jli_call_p): New function.
22555 * config/arc/arc.md (movqi_insn): Remove 'S' from printing asm
22557 (movhi_insn): Likewise.
22558 (movsi_insn): Likewise.
22559 (movsi_set_cc_insn): Likewise.
22560 (loadqi_update): Likewise.
22561 (load_zeroextendqisi_update): Likewise.
22562 (load_signextendqisi_update): Likewise.
22563 (loadhi_update): Likewise.
22564 (load_zeroextendhisi_update): Likewise.
22565 (load_signextendhisi_update): Likewise.
22566 (loadsi_update): Likewise.
22567 (loadsf_update): Likewise.
22568 (movsicc_insn): Likewise.
22569 (bset_insn): Likewise.
22570 (bxor_insn): Likewise.
22571 (bclr_insn): Likewise.
22572 (bmsk_insn): Likewise.
22573 (bicsi3_insn): Likewise.
22574 (cmpsi_cc_c_insn): Likewise.
22575 (movsi_ne): Likewise.
22576 (movsi_cond_exec): Likewise.
22577 (clrsbsi2): Likewise.
22578 (norm_f): Likewise.
22588 (call_i): Remove 'S' asm letter, add jli instruction.
22589 (call_value_i): Likewise.
22590 * config/arc/arc.op (mjli-always): New option.
22591 * config/arc/constraints.md (Cji): New constraint.
22592 * config/arc/fpx.md (addsf3_fpx): Remove 'S' from printing asm
22594 (subsf3_fpx): Likewise.
22595 (mulsf3_fpx): Likewise.
22596 * config/arc/simdext.md (vendrec_insn): Remove 'S' from printing
22598 * doc/extend.texi (ARC): Document 'jli-always' and 'jli-fixed'
22599 function attrbutes.
22600 * doc/invoke.texi (ARC): Document mjli-always option.
22602 2018-01-26 Sebastian Perta <sebastian.perta@renesas.com>
22604 * config/rl78/rl78.c (rl78_addsi3_internal): If operand 2 is const
22605 avoid addition with 0 and use incw and decw where possible.
22607 2018-01-26 Richard Biener <rguenther@suse.de>
22609 PR tree-optimization/81082
22610 * fold-const.c (fold_plusminus_mult_expr): Do not perform the
22611 association if it requires casting to unsigned.
22612 * match.pd ((A * C) +- (B * C) -> (A+-B)): New patterns derived
22613 from fold_plusminus_mult_expr to catch important cases late when
22614 range info is available.
22616 2018-01-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
22618 * config/i386/sol2.h (USE_HIDDEN_LINKONCE): Remove.
22619 * configure.ac (hidden_linkonce): New test.
22620 * configure: Regenerate.
22621 * config.in: Regenerate.
22623 2018-01-26 Julia Koval <julia.koval@intel.com>
22625 * config/i386/avx512bitalgintrin.h (_mm512_bitshuffle_epi64_mask,
22626 _mm512_mask_bitshuffle_epi64_mask, _mm256_bitshuffle_epi64_mask,
22627 _mm256_mask_bitshuffle_epi64_mask, _mm_bitshuffle_epi64_mask,
22628 _mm_mask_bitshuffle_epi64_mask): Fix type.
22629 * config/i386/i386-builtin-types.def (UHI_FTYPE_V2DI_V2DI_UHI,
22630 USI_FTYPE_V4DI_V4DI_USI): Remove.
22631 * config/i386/i386-builtin.def (__builtin_ia32_vpshufbitqmb512_mask,
22632 __builtin_ia32_vpshufbitqmb256_mask,
22633 __builtin_ia32_vpshufbitqmb128_mask): Fix types.
22634 * config/i386/i386.c (ix86_expand_args_builtin): Remove old types.
22635 * config/i386/sse.md (VI1_AVX512VLBW): Change types.
22637 2018-01-26 Alan Modra <amodra@gmail.com>
22640 * config/rs6000/rs6000-p8swap.c (rtx_is_swappable_p): Exclude
22641 UNSPEC_VBPERMQ. Sort other unspecs.
22643 2018-01-25 David Edelsohn <dje.gcc@gmail.com>
22645 * doc/invoke.texi (PowerPC Options): Document 'native' cpu type.
22647 2018-01-25 Jan Hubicka <hubicka@ucw.cz>
22649 PR middle-end/83055
22650 * predict.c (drop_profile): Do not push/pop cfun; update also
22652 (handle_missing_profiles): Fix logic looking for zero profiles.
22654 2018-01-25 Jakub Jelinek <jakub@redhat.com>
22656 PR middle-end/83977
22657 * ipa-fnsummary.c (compute_fn_summary): Clear can_change_signature
22658 on functions with #pragma omp declare simd or functions with simd
22660 * omp-simd-clone.c (expand_simd_clones): Revert 2018-01-24 change.
22661 * config/i386/i386.c (ix86_simd_clone_compute_vecsize_and_simdlen):
22662 Remove trailing \n from warning_at calls.
22664 2018-01-25 Tom de Vries <tom@codesourcery.com>
22667 * config/nvptx/nvptx.c (nvptx_single): Add exit insn after noreturn call
22668 for neutered workers.
22670 2018-01-24 Joseph Myers <joseph@codesourcery.com>
22673 * config/m68k/m68k.c (m68k_promote_function_mode): New function.
22674 (TARGET_PROMOTE_FUNCTION_MODE): New macro.
22676 2018-01-24 Jeff Law <law@redhat.com>
22679 * i386.c (get_probe_interval): Move to earlier point.
22680 (ix86_compute_frame_layout): If -fstack-clash-protection and
22681 the frame is larger than the probe interval, then use pushes
22682 to save registers rather than reg->mem moves.
22683 (ix86_expand_prologue): Remove conditional for int_registers_saved
22686 2018-01-24 Vladimir Makarov <vmakarov@redhat.com>
22689 * ira-build.c (setup_min_max_allocno_live_range_point): Set up
22690 min/max for never referenced object.
22692 2018-01-24 Jakub Jelinek <jakub@redhat.com>
22694 PR middle-end/83977
22695 * tree.c (free_lang_data_in_decl): Don't clear DECL_ABSTRACT_ORIGIN
22697 * omp-low.c (create_omp_child_function): Remove "omp declare simd"
22698 attributes from DECL_ATTRIBUTES (decl) without affecting
22699 DECL_ATTRIBUTES (current_function_decl).
22700 * omp-simd-clone.c (expand_simd_clones): Ignore DECL_ARTIFICIAL
22701 functions with non-NULL DECL_ABSTRACT_ORIGIN.
22703 2018-01-24 Richard Sandiford <richard.sandiford@linaro.org>
22705 PR tree-optimization/83979
22706 * fold-const.c (fold_comparison): Use constant_boolean_node
22707 instead of boolean_{true,false}_node.
22709 2018-01-24 Jan Hubicka <hubicka@ucw.cz>
22711 * ipa-profile.c (ipa_propagate_frequency_1): Fix logic skipping calls
22714 2018-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22716 * config/rs6000/rs6000.md (*call_indirect_nonlocal_sysv<mode>):
22717 Simplify the clause that sets the length attribute.
22718 (*call_value_indirect_nonlocal_sysv<mode>): Likewise.
22719 (*sibcall_nonlocal_sysv<mode>): Clean up code block; simplify the
22720 clause that sets the length attribute.
22721 (*sibcall_value_nonlocal_sysv<mode>): Likewise.
22723 2018-01-24 Tom de Vries <tom@codesourcery.com>
22726 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG_2): Define to 1.
22727 (nvptx_pc_set, nvptx_condjump_label): New function. Copy from jump.c.
22728 Add strict parameter.
22729 (prevent_branch_around_nothing): Insert dummy insn between branch to
22730 label and label with no ptx insn inbetween.
22731 * config/nvptx/nvptx.md (define_insn "fake_nop"): New insn.
22733 2018-01-24 Tom de Vries <tom@codesourcery.com>
22736 * config/nvptx/nvptx.c (nvptx_single): Add exit insn after noreturn call
22737 for neutered threads in warp.
22738 * config/nvptx/nvptx.md (define_insn "exit"): New insn.
22740 2018-01-24 Richard Biener <rguenther@suse.de>
22742 PR tree-optimization/83176
22743 * tree-chrec.c (chrec_fold_plus_1): Handle (signed T){(T) .. }
22746 2018-01-24 Richard Biener <rguenther@suse.de>
22748 PR tree-optimization/82819
22749 * graphite-isl-ast-to-gimple.c (binary_op_to_tree): Avoid
22750 code generating pluses that are no-ops in the target precision.
22752 2018-01-24 Richard Biener <rguenther@suse.de>
22754 PR middle-end/84000
22755 * tree-cfg.c (replace_loop_annotate): Handle annot_expr_parallel_kind.
22757 2018-01-23 Jan Hubicka <hubicka@ucw.cz>
22759 * cfgcleanup.c (try_crossjump_to_edge): Use combine_with_count
22760 to merge probabilities.
22761 * predict.c (probably_never_executed): Also mark as cold functions
22762 with global 0 profile and guessed local profile.
22763 * profile-count.c (profile_probability::combine_with_count): New
22765 * profile-count.h (profile_probability::operator*,
22766 profile_probability::operator*=, profile_probability::operator/,
22767 profile_probability::operator/=): Reduce precision to adjusted
22768 and set value to guessed on contradictory divisions.
22769 (profile_probability::combine_with_freq): Remove.
22770 (profile_probability::combine_wiht_count): Declare.
22771 (profile_count::force_nonzero):: Set to adjusted.
22772 (profile_count::probability_in):: Set quality to adjusted.
22773 * tree-ssa-tail-merge.c (replace_block_by): Use
22774 combine_with_count.
22776 2018-01-23 Andrew Waterman <andrew@sifive.com>
22777 Jim Wilson <jimw@sifive.com>
22779 * config/riscv/riscv.c (riscv_stack_boundary): New.
22780 (riscv_option_override): Set riscv_stack_boundary. Handle
22781 riscv_preferred_stack_boundary_arg.
22782 * config/riscv/riscv.h (MIN_STACK_BOUNDARY, ABI_STACK_BOUNDARY): New.
22783 (BIGGEST_ALIGNMENT): Set to STACK_BOUNDARY.
22784 (STACK_BOUNDARY): Set to riscv_stack_boundary.
22785 (RISCV_STACK_ALIGN): Use STACK_BOUNDARY.
22786 * config/riscv/riscv.opt (mpreferred-stack-boundary): New.
22787 * doc/invoke.tex (RISC-V Options): Add -mpreferred-stack-boundary.
22789 2018-01-23 H.J. Lu <hongjiu.lu@intel.com>
22792 * config/i386/i386.c (ix86_expand_prologue): Use cost reference
22793 of struct ix86_frame.
22794 (ix86_expand_epilogue): Likewise. Add a local variable for
22795 the reg_save_offset field in struct ix86_frame.
22797 2018-01-23 Bin Cheng <bin.cheng@arm.com>
22799 PR tree-optimization/82604
22800 * tree-loop-distribution.c (enum partition_kind): New enum item
22801 PKIND_PARTIAL_MEMSET.
22802 (partition_builtin_p): Support above new enum item.
22803 (generate_code_for_partition): Ditto.
22804 (compute_access_range): Differentiate cases that equality can be
22805 proven at all loops, the innermost loops or no loops.
22806 (classify_builtin_st, classify_builtin_ldst): Adjust call to above
22807 function. Set PKIND_PARTIAL_MEMSET for partition appropriately.
22808 (finalize_partitions, distribute_loop): Don't fuse partition of
22809 PKIND_PARTIAL_MEMSET kind when distributing 3-level loop nest.
22810 (prepare_perfect_loop_nest): Distribute 3-level loop nest only if
22811 parloop is enabled.
22813 2018-01-23 Martin Liska <mliska@suse.cz>
22815 * predict.def (PRED_INDIR_CALL): Set probability to PROB_EVEN in
22816 order to ignore the predictor.
22817 (PRED_POLYMORPHIC_CALL): Likewise.
22818 (PRED_RECURSIVE_CALL): Likewise.
22820 2018-01-23 Martin Liska <mliska@suse.cz>
22822 * tree-profile.c (tree_profiling): Print function header to
22823 aware reader which function we are working on.
22824 * value-prof.c (gimple_find_values_to_profile): Do not print
22825 not interesting value histograms.
22827 2018-01-23 Martin Liska <mliska@suse.cz>
22829 * profile-count.h (enum profile_quality): Add
22830 profile_uninitialized as the first value. Do not number values
22831 as they are zero based.
22832 (profile_count::verify): Update sanity check.
22833 (profile_probability::verify): Likewise.
22835 2018-01-23 Nathan Sidwell <nathan@acm.org>
22837 * doc/invoke.texi (ffor-scope): Deprecate.
22839 2018-01-23 David Malcolm <dmalcolm@redhat.com>
22841 PR tree-optimization/83510
22842 * domwalk.c (set_all_edges_as_executable): New function.
22843 (dom_walker::dom_walker): Convert bool param
22844 "skip_unreachable_blocks" to enum reachability. Move setup of
22845 edge flags to set_all_edges_as_executable and only do it when
22846 reachability is REACHABLE_BLOCKS.
22847 * domwalk.h (enum dom_walker::reachability): New enum.
22848 (dom_walker::dom_walker): Convert bool param
22849 "skip_unreachable_blocks" to enum reachability.
22850 (set_all_edges_as_executable): New decl.
22851 * graphite-scop-detection.c (gather_bbs::gather_bbs): Convert
22852 from false for "skip_unreachable_blocks" to ALL_BLOCKS for
22854 * tree-ssa-dom.c (dom_opt_dom_walker::dom_opt_dom_walker): Likewise,
22855 but converting true to REACHABLE_BLOCKS.
22856 * tree-ssa-sccvn.c (sccvn_dom_walker::sccvn_dom_walker): Likewise.
22858 (check_array_bounds_dom_walker::check_array_bounds_dom_walker):
22859 Likewise, but converting it to REACHABLE_BLOCKS_PRESERVING_FLAGS.
22860 (vrp_dom_walker::vrp_dom_walker): Likewise, but converting it to
22862 (vrp_prop::vrp_finalize): Call set_all_edges_as_executable
22863 if check_all_array_refs will be called.
22865 2018-01-23 David Malcolm <dmalcolm@redhat.com>
22867 * tree.c (selftest::test_location_wrappers): Add more test
22870 2018-01-23 David Malcolm <dmalcolm@redhat.com>
22872 * sbitmap.c (selftest::test_set_range): Fix memory leaks.
22873 (selftest::test_bit_in_range): Likewise.
22875 2018-01-23 Richard Sandiford <richard.sandiford@linaro.org>
22878 * doc/sourcebuild.texi (vect_float): Say that the selector
22879 only describes the situation when -funsafe-math-optimizations is on.
22880 (vect_float_strict): Document.
22882 2018-01-23 Richard Sandiford <richard.sandiford@linaro.org>
22884 PR tree-optimization/83965
22885 * tree-vect-patterns.c (vect_reassociating_reduction_p): New function.
22886 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern): Use it
22887 instead of checking only for a reduction.
22888 (vect_recog_widen_sum_pattern): Likewise.
22890 2018-01-23 Jan Hubicka <hubicka@ucw.cz>
22892 * predict.c (probably_never_executed): Only use precise profile info.
22893 (compute_function_frequency): Skip after inlining hack since we now
22894 have quality checking.
22896 2018-01-23 Jan Hubicka <hubicka@ucw.cz>
22898 * profile-count.h (profile_probability::very_unlikely,
22899 profile_probability::unlikely, profile_probability::even): Set
22900 precision to guessed.
22902 2018-01-23 Richard Biener <rguenther@suse.de>
22904 PR tree-optimization/83963
22905 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
22906 Properly terminate dominator walk when crossing the exit edge not
22907 when visiting its source block.
22909 2018-01-23 Jakub Jelinek <jakub@redhat.com>
22912 * tree.c (maybe_wrap_with_location): Use NON_LVALUE_EXPR rather than
22913 VIEW_CONVERT_EXPR to wrap CONST_DECLs.
22915 2018-01-22 Jakub Jelinek <jakub@redhat.com>
22917 PR tree-optimization/83957
22918 * omp-expand.c (expand_omp_for_generic): Ignore virtual PHIs. Remove
22919 semicolon after for body surrounded by braces.
22921 PR tree-optimization/83081
22922 * profile-count.h (profile_probability::split): New method.
22923 * dojump.c (do_jump_1) <case TRUTH_ANDIF_EXPR, case TRUTH_ORIF_EXPR>:
22924 Use profile_probability::split.
22925 (do_compare_rtx_and_jump): Fix adjustment of probabilities
22926 when splitting a single conditional jump into 2.
22928 2018-01-22 David Malcolm <dmalcolm@redhat.com>
22930 PR tree-optimization/69452
22931 * tree-ssa-loop-im.c (class move_computations_dom_walker): Remove
22934 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
22936 * config/rl78/rl78-expand.md (bswaphi2): New define_expand.
22937 * config/rl78/rl78-virt.md (*bswaphi2_virt): New define_insn.
22938 * config/rl78/rl78-real.md (*bswaphi2_real): New define_insn.
22940 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
22942 * config/rl78/rl78-protos.h (rl78_split_movdi): New function
22944 * config/rl78/rl78.md (movdi): New define_expand.
22945 * config/rl78/rl78.c (rl78_split_movdi): New function.
22947 2018-01-22 Michael Meissner <meissner@linux.vnet.ibm.com>
22950 * config/rs6000/rs6000-protos.h (rs6000_split_signbit): Delete,
22952 * config/rs6000/rs6000.c (rs6000_split_signbit): Likewise.
22953 * config/rs6000/rs6000.md (signbit<mode>2): Change code for IEEE
22954 128-bit to produce an UNSPEC move to get the double word with the
22955 signbit and then a shift directly to do signbit.
22956 (signbit<mode>2_dm): Replace old IEEE 128-bit signbit
22957 implementation with a new version that just does either a direct
22958 move or a regular move. Move memory interface to separate insns.
22959 Move insns so they are next to the expander.
22960 (signbit<mode>2_dm_mem_be): New combiner insns to combine load
22961 with signbit move. Split big and little endian case.
22962 (signbit<mode>2_dm_mem_le): Likewise.
22963 (signbit<mode>2_dm_<su>ext): Delete, no longer used.
22964 (signbit<mode>2_dm2): Likewise.
22966 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
22968 * config/rl78/rl78.md (anddi3): New define_expand.
22970 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
22972 * config/rl78/rl78.md (umindi3): New define_expand.
22974 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
22976 * config/rl78/rl78.md (smindi3): New define_expand.
22978 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
22980 * config/rl78/rl78.md (smaxdi3): New define_expand.
22982 2018-01-22 Carl Love <cel@us.ibm.com>
22984 * config/rs6000/rs6000-builtin.def (ST_ELEMREV_V1TI, LD_ELEMREV_V1TI,
22985 LVX_V1TI): Add macro expansion.
22986 * config/rs6000/rs6000-c.c (altivec_builtin_types): Add argument
22987 definitions for VSX_BUILTIN_VEC_XST_BE, VSX_BUILTIN_VEC_ST,
22988 VSX_BUILTIN_VEC_XL, LD_ELEMREV_V1TI builtins.
22989 * config/rs6000/rs6000-p8swap.c (insn_is_swappable_p);
22990 Change check to determine if the instruction is a byte reversing
22991 entry. Fix typo in comment.
22992 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case entry
22993 for VSX_BUILTIN_ST_ELEMREV_V1TI and VSX_BUILTIN_LD_ELEMREV_V1TI.
22994 Add def_builtin calls for new builtins.
22995 * config/rs6000/vsx.md (vsx_st_elemrev_v1ti, vsx_ld_elemrev_v1ti):
22996 Add define_insn expansion.
22998 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
23000 * config/rl78/rl78.md (umaxdi3): New define_expand.
23002 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
23004 * config/rl78/rl78.c (rl78_note_reg_set): Fixed dead reg check
23005 for non-QImode registers.
23007 2018-01-22 Richard Biener <rguenther@suse.de>
23009 PR tree-optimization/83963
23010 * graphite-scop-detection.c (scop_detection::get_sese): Delay
23011 including the loop exit block.
23012 (scop_detection::merge_sese): Likewise.
23013 (scop_detection::add_scop): Do it here instead.
23015 2018-01-22 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23017 * doc/sourcebuild.texi (arm_softfloat): Document.
23019 2018-01-21 John David Anglin <danglin@gcc.gnu.org>
23022 * config/pa/pa.c (pa_function_ok_for_sibcall): Use
23023 targetm.binds_local_p instead of TREE_PUBLIC to check local binding.
23024 Move TARGET_PORTABLE_RUNTIME check after TARGET_64BIT check.
23026 2018-01-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23027 David Edelsohn <dje.gcc@gmail.com>
23030 * config/rs6000/rs6000.md (*call_indirect_nonlocal_sysv<mode>):
23031 Change "crset eq" to "crset 2".
23032 (*call_value_indirect_nonlocal_sysv<mode>): Likewise.
23033 (*call_indirect_aix<mode>_nospec): Likewise.
23034 (*call_value_indirect_aix<mode>_nospec): Likewise.
23035 (*call_indirect_elfv2<mode>_nospec): Likewise.
23036 (*call_value_indirect_elfv2<mode>_nospec): Likewise.
23037 (*sibcall_nonlocal_sysv<mode>): Change "crset eq" to "crset 2";
23038 change assembly output from . to $.
23039 (*sibcall_value_nonlocal_sysv<mode>): Likewise.
23040 (indirect_jump<mode>_nospec): Change assembly output from . to $.
23041 (*tablejump<mode>_internal1_nospec): Likewise.
23043 2018-01-21 Oleg Endo <olegendo@gcc.gnu.org>
23046 * config/sh/sh_optimize_sett_clrt.cc:
23047 Use INCLUDE_ALGORITHM and INCLUDE_VECTOR instead of direct includes.
23049 2018-01-20 Richard Sandiford <richard.sandiford@linaro.org>
23051 PR tree-optimization/83940
23052 * tree-vect-stmts.c (vect_truncate_gather_scatter_offset): Set
23053 offset_dt to vect_constant_def rather than vect_unknown_def_type.
23054 (vect_check_load_store_mask): Add a mask_dt_out parameter and
23055 use it to pass back the definition type.
23056 (vect_check_store_rhs): Likewise rhs_dt_out.
23057 (vect_build_gather_load_calls): Add a mask_dt argument and use
23058 it instead of a call to vect_is_simple_use.
23059 (vectorizable_store): Update calls to vect_check_load_store_mask
23060 and vect_check_store_rhs. Use the dt returned by the latter instead
23061 of scatter_src_dt. Use the cached mask_dt and gs_info.offset_dt
23062 instead of calls to vect_is_simple_use. Pass the scalar rather
23063 than the vector operand to vect_is_simple_use when handling
23064 second and subsequent copies of an rhs value.
23065 (vectorizable_load): Update calls to vect_check_load_store_mask
23066 and vect_build_gather_load_calls. Use the cached mask_dt and
23067 gs_info.offset_dt instead of calls to vect_is_simple_use.
23069 2018-01-20 Jakub Jelinek <jakub@redhat.com>
23071 PR middle-end/83945
23072 * tree-emutls.c: Include gimplify.h.
23073 (lower_emutls_2): New function.
23074 (lower_emutls_1): If ADDR_EXPR is a gimple invariant and walk_tree
23075 with lower_emutls_2 callback finds some TLS decl in it, unshare_expr
23076 it before further processing.
23079 * simplify-rtx.c (simplify_binary_operation_1) <case UMOD>: Use
23080 UINTVAL (trueop1) instead of INTVAL (op1).
23082 2018-01-19 Jakub Jelinek <jakub@redhat.com>
23086 * dwarf2cfi.c (DEFAULT_INCOMING_FRAME_SP_OFFSET): Define to
23087 INCOMING_FRAME_SP_OFFSET if not defined.
23088 (scan_trace): Add ENTRY argument. If true and
23089 DEFAULT_INCOMING_FRAME_SP_OFFSET != INCOMING_FRAME_SP_OFFSET,
23090 emit a note to adjust the CFA offset.
23091 (create_cfi_notes): Adjust scan_trace callers.
23092 (create_cie_data): Use DEFAULT_INCOMING_FRAME_SP_OFFSET rather than
23093 INCOMING_FRAME_SP_OFFSET in the CIE.
23094 * config/i386/i386.h (DEFAULT_INCOMING_FRAME_SP_OFFSET): Define.
23095 * config/stormy16/stormy16.h (DEFAULT_INCOMING_FRAME_SP_OFFSET):
23097 * doc/tm.texi.in (DEFAULT_INCOMING_FRAME_SP_OFFSET): Document.
23098 * doc/tm.texi: Regenerated.
23100 2018-01-19 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23102 PR rtl-optimization/83147
23103 * lra-constraints.c (remove_inheritance_pseudos): Use
23104 lra_substitute_pseudo_within_insn.
23106 2018-01-19 Tom de Vries <tom@codesourcery.com>
23107 Cesar Philippidis <cesar@codesourcery.com>
23110 * config/nvptx/nvptx.c (nvptx_single): Fix jit workaround.
23112 2018-01-19 Cesar Philippidis <cesar@codesourcery.com>
23115 * config/nvptx/nvptx.c (output_init_frag): Don't use generic address
23116 spaces for function labels.
23118 2018-01-19 Martin Liska <mliska@suse.cz>
23120 * predict.def (PRED_LOOP_EXIT): Change from 85 to 89.
23121 (PRED_LOOP_EXIT_WITH_RECURSION): Change from 72 to 78.
23122 (PRED_LOOP_EXTRA_EXIT): Change from 83 to 67.
23123 (PRED_OPCODE_POSITIVE): Change from 64 to 59.
23124 (PRED_TREE_OPCODE_POSITIVE): Change from 64 to 59.
23125 (PRED_CONST_RETURN): Change from 69 to 65.
23126 (PRED_NULL_RETURN): Change from 91 to 71.
23127 (PRED_LOOP_IV_COMPARE_GUESS): Change from 98 to 64.
23128 (PRED_LOOP_GUARD): Change from 66 to 73.
23130 2018-01-19 Martin Liska <mliska@suse.cz>
23132 * predict.c (predict_insn_def): Add new assert.
23133 (struct branch_predictor): Change type to signed integer.
23134 (test_prediction_value_range): Amend test to cover
23135 PROB_UNINITIALIZED.
23136 * predict.def (PRED_LOOP_ITERATIONS): Use the new constant.
23137 (PRED_LOOP_ITERATIONS_GUESSED): Likewise.
23138 (PRED_LOOP_ITERATIONS_MAX): Likewise.
23139 (PRED_LOOP_IV_COMPARE): Likewise.
23140 * predict.h (PROB_UNINITIALIZED): Define new constant.
23142 2018-01-19 Martin Liska <mliska@suse.cz>
23144 * predict.c (dump_prediction): Add new format for
23145 analyze_brprob.py script which is enabled with -details
23147 * profile-count.h (precise_p): New function.
23149 2018-01-19 Richard Sandiford <richard.sandiford@linaro.org>
23151 PR tree-optimization/83922
23152 * tree-vect-loop.c (vect_verify_full_masking): Return false if
23153 there are no statements that need masking.
23154 (vect_active_double_reduction_p): New function.
23155 (vect_analyze_loop_operations): Use it when handling phis that
23156 are not in the loop header.
23158 2018-01-19 Richard Sandiford <richard.sandiford@linaro.org>
23160 PR tree-optimization/83914
23161 * tree-vect-loop.c (vectorizable_induction): Don't convert
23162 init_expr or apply the peeling adjustment for inductions
23163 that are nested within the vectorized loop.
23165 2018-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23167 * config/arm/thumb2.md (*thumb2_negsi2_short): Use RSB mnemonic
23170 2018-01-18 Jakub Jelinek <jakub@redhat.com>
23174 * function.h (gimplify_parameters): Add gimple_seq * argument.
23175 * function.c: Include gimple.h and options.h.
23176 (gimplify_parameters): Add cleanup argument, add CLOBBER stmts
23177 for the added local temporaries if needed.
23178 * gimplify.c (gimplify_body): Adjust gimplify_parameters caller,
23179 if there are any parameter cleanups, wrap whole body into a
23180 try/finally with the cleanups.
23182 2018-01-18 Wilco Dijkstra <wdijkstr@arm.com>
23185 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
23186 Use GET_MODE_CLASS for scalar floating point.
23188 2018-01-18 Jan Hubicka <hubicka@ucw.cz>
23192 * cgraphclones.c (cgraph_node::create_version_clone_with_body):
23193 Fix call of call_cgraph_insertion_hooks.
23195 2018-01-18 Martin Sebor <msebor@redhat.com>
23197 * doc/invoke.texi (-Wclass-memaccess): Tweak text.
23199 2018-01-18 Jan Hubicka <hubicka@ucw.cz>
23202 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Update edge
23205 2018-01-18 Boris Kolpackov <boris@codesynthesis.com>
23208 * common.opt: (-ffile-prefix-map): New option.
23209 * opts.c (common_handle_option): Defer it.
23210 * opts-global.c (handle_common_deferred_options): Handle it.
23211 * debug.h (remap_debug_filename, add_debug_prefix_map): Move to...
23212 * file-prefix-map.h: New file.
23213 (remap_debug_filename, add_debug_prefix_map): ...here.
23214 (add_macro_prefix_map, add_file_prefix_map, remap_macro_filename): New.
23215 * final.c (debug_prefix_map, add_debug_prefix_map
23216 remap_debug_filename): Move to...
23217 * file-prefix-map.c: New file.
23218 (file_prefix_map, add_prefix_map, remap_filename) ...here and rename,
23219 generalize, get rid of alloca(), use strrchr() instead of strchr().
23220 (add_macro_prefix_map, add_debug_prefix_map, add_file_prefix_map):
23221 Implement in terms of add_prefix_map().
23222 (remap_macro_filename, remap_debug_filename): Implement in term of
23224 * Makefile.in (OBJS, PLUGIN_HEADERS): Add new files.
23225 * builtins.c (fold_builtin_FILE): Call remap_macro_filename().
23226 * dbxout.c: Include file-prefix-map.h.
23227 * varasm.c: Likewise.
23228 * vmsdbgout.c: Likewise.
23229 * xcoffout.c: Likewise.
23230 * dwarf2out.c: Likewise plus omit new options from DW_AT_producer.
23231 * doc/cppopts.texi (-fmacro-prefix-map): Document.
23232 * doc/invoke.texi (-ffile-prefix-map): Document.
23233 (-fdebug-prefix-map): Update description.
23235 2018-01-18 Martin Liska <mliska@suse.cz>
23237 * config/i386/i386.c (indirect_thunk_name): Document that also
23239 (output_indirect_thunk): Document why both instructions
23240 (pause and lfence) are generated.
23242 2018-01-18 Richard Biener <rguenther@suse.de>
23244 PR tree-optimization/83887
23245 * graphite-scop-detection.c
23246 (scop_detection::get_nearest_dom_with_single_entry): Remove.
23247 (scop_detection::get_nearest_pdom_with_single_exit): Likewise.
23248 (scop_detection::merge_sese): Re-implement with a flood-fill
23249 algorithm that properly finds a SESE region if it exists.
23251 2018-01-18 Jakub Jelinek <jakub@redhat.com>
23254 * match.pd ((P + A) - P, P - (P + A), (P + A) - (P + B)): For
23255 pointer_diff optimizations use view_convert instead of convert.
23257 2018-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23259 * config/rs6000/rs6000.md (*call_indirect_nonlocal_sysv<mode>):
23260 Generate different code for -mno-speculate-indirect-jumps.
23261 (*call_value_indirect_nonlocal_sysv<mode>): Likewise.
23262 (*call_indirect_aix<mode>): Disable for
23263 -mno-speculate-indirect-jumps.
23264 (*call_indirect_aix<mode>_nospec): New define_insn.
23265 (*call_value_indirect_aix<mode>): Disable for
23266 -mno-speculate-indirect-jumps.
23267 (*call_value_indirect_aix<mode>_nospec): New define_insn.
23268 (*sibcall_nonlocal_sysv<mode>): Generate different code for
23269 -mno-speculate-indirect-jumps.
23270 (*sibcall_value_nonlocal_sysv<mode>): Likewise.
23272 2018-01-17 Michael Meissner <meissner@linux.vnet.ibm.com>
23274 * config/rs6000/rs6000.c (rs6000_emit_move): If we load or store a
23275 long double type, set the flags for noting the default long double
23276 type, even if we don't pass or return a long double type.
23278 2018-01-17 Jan Hubicka <hubicka@ucw.cz>
23281 * ipa-inline.c (flatten_function): Do not overwrite final inlining
23284 2018-01-17 Will Schmidt <will_schmidt@vnet.ibm.com>
23286 * config/rs6000/rs6000.c (rs6000_gimple_builtin): Add gimple folding
23287 support for merge[hl].
23288 (fold_mergehl_helper): New helper function.
23289 (tree-vector-builder.h): New #include for tree_vector_builder usage.
23290 * config/rs6000/altivec.md (altivec_vmrghw_direct): Add xxmrghw insn.
23291 (altivec_vmrglw_direct): Add xxmrglw insn.
23293 2018-01-17 Andrew Waterman <andrew@sifive.com>
23295 * config/riscv/riscv.c (riscv_conditional_register_usage): If
23296 UNITS_PER_FP_ARG is 0, set call_used_regs to 1 for all FP regs.
23298 2018-01-17 David Malcolm <dmalcolm@redhat.com>
23301 * ipa-devirt.c (add_type_duplicate): When comparing memory layout,
23302 call the lto_location_cache before reading the
23303 DECL_SOURCE_LOCATION of the types.
23305 2018-01-17 Wilco Dijkstra <wdijkstr@arm.com>
23306 Richard Sandiford <richard.sandiford@linaro.org>
23308 * config/aarch64/aarch64.md (movti_aarch64): Use Uti constraint.
23309 * config/aarch64/aarch64.c (aarch64_mov128_immediate): New function.
23310 (aarch64_legitimate_constant_p): Just support CONST_DOUBLE
23311 SF/DF/TF mode to avoid creating illegal CONST_WIDE_INT immediates.
23312 * config/aarch64/aarch64-protos.h (aarch64_mov128_immediate):
23314 * config/aarch64/constraints.md (aarch64_movti_operand):
23316 * config/aarch64/predicates.md (Uti): Add new constraint.
23318 2018-01-17 Carl Love <cel@us.ibm.com>
23320 * config/rs6000/vsx.md (define_expand xl_len_r,
23321 define_expand stxvl, define_expand *stxvl): Add match_dup argument.
23322 (define_insn): Add, match_dup 1 argument to define_insn stxvll and
23324 (define_expand, define_insn): Move the shift left from the
23325 define_insn to the define_expand for lxvl and stxvl instructions.
23326 * config/rs6000/rs6000-builtin.def (BU_P9V_64BIT_VSX_2): Change LXVL
23327 and XL_LEN_R definitions to PURE.
23329 2018-01-17 Uros Bizjak <ubizjak@gmail.com>
23331 * config/i386/i386.c (indirect_thunk_name): Declare regno
23332 as unsigned int. Compare regno with INVALID_REGNUM.
23333 (output_indirect_thunk): Ditto.
23334 (output_indirect_thunk_function): Ditto.
23335 (ix86_code_end): Declare regno as unsigned int. Use INVALID_REGNUM
23336 in the call to output_indirect_thunk_function.
23338 2018-01-17 Richard Sandiford <richard.sandiford@linaro.org>
23340 PR middle-end/83884
23341 * expr.c (expand_expr_real_1): Use the size of GET_MODE (op0)
23342 rather than the size of inner_type to determine the stack slot size
23343 when handling VIEW_CONVERT_EXPRs on strict-alignment targets.
23345 2018-01-16 Sebastian Peryt <sebastian.peryt@intel.com>
23348 * config/i386/i386.c (ix86_option_override_internal): Add PTA_RDRND
23351 2018-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
23353 * config.gcc (powerpc*-linux*-*): Add support for 64-bit little
23354 endian Linux systems to optionally enable multilibs for selecting
23355 the long double type if the user configured an explicit type.
23356 * config/rs6000/rs6000.h (TARGET_IEEEQUAD_MULTILIB): Indicate we
23357 have no long double multilibs if not defined.
23358 * config/rs6000/rs6000.c (rs6000_option_override_internal): Do not
23359 warn if the user used -mabi={ieee,ibm}longdouble and we built
23360 multilibs for long double.
23361 * config/rs6000/linux64.h (MULTILIB_DEFAULTS_IEEE): Define as the
23362 appropriate multilib option.
23363 (MULTILIB_DEFAULTS): Add MULTILIB_DEFAULTS_IEEE to the default
23365 * config/rs6000/t-ldouble-linux64le-ibm: New configuration files
23366 for building long double multilibs.
23367 * config/rs6000/t-ldouble-linux64le-ieee: Likewise.
23369 2018-01-16 John David Anglin <danglin@gcc.gnu.org>
23371 * config.gcc (hppa*-*-linux*): Change callee copies ABI to caller
23374 * config/pa.h (MALLOC_ABI_ALIGNMENT): Set 32-bit alignment default to
23376 * config/pa/pa32-linux.h (MALLOC_ABI_ALIGNMENT): Set alignment to
23379 * config/pa/som.h (ASM_DECLARE_FUNCTION_NAME): Cleanup type and mode
23382 * config/pa/pa.c (pa_function_arg_size): Apply CEIL to GET_MODE_SIZE
23385 2018-01-16 Eric Botcazou <ebotcazou@adacore.com>
23387 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): For an
23388 ADDR_EXPR, do not count the offset of a COMPONENT_REF twice.
23390 2018-01-16 Kelvin Nilsen <kelvin@gcc.gnu.org>
23392 * config/rs6000/rs6000-p8swap.c (rs6000_gen_stvx): Generate
23393 different rtl trees depending on TARGET_64BIT.
23394 (rs6000_gen_lvx): Likewise.
23396 2018-01-16 Eric Botcazou <ebotcazou@adacore.com>
23398 * config/visium/visium.md (nop): Tweak comment.
23399 (hazard_nop): Likewise.
23401 2018-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23403 * config/rs6000/rs6000.c (rs6000_opt_vars): Add entry for
23404 -mspeculate-indirect-jumps.
23405 * config/rs6000/rs6000.md (*call_indirect_elfv2<mode>): Disable
23406 for -mno-speculate-indirect-jumps.
23407 (*call_indirect_elfv2<mode>_nospec): New define_insn.
23408 (*call_value_indirect_elfv2<mode>): Disable for
23409 -mno-speculate-indirect-jumps.
23410 (*call_value_indirect_elfv2<mode>_nospec): New define_insn.
23411 (indirect_jump): Emit different RTL for
23412 -mno-speculate-indirect-jumps.
23413 (*indirect_jump<mode>): Disable for
23414 -mno-speculate-indirect-jumps.
23415 (*indirect_jump<mode>_nospec): New define_insn.
23416 (tablejump): Emit different RTL for
23417 -mno-speculate-indirect-jumps.
23418 (tablejumpsi): Disable for -mno-speculate-indirect-jumps.
23419 (tablejumpsi_nospec): New define_expand.
23420 (tablejumpdi): Disable for -mno-speculate-indirect-jumps.
23421 (tablejumpdi_nospec): New define_expand.
23422 (*tablejump<mode>_internal1): Disable for
23423 -mno-speculate-indirect-jumps.
23424 (*tablejump<mode>_internal1_nospec): New define_insn.
23425 * config/rs6000/rs6000.opt (mspeculate-indirect-jumps): New
23428 2018-01-16 Artyom Skrobov tyomitch@gmail.com
23430 * caller-save.c (insert_save): Drop unnecessary parameter. All
23433 2018-01-16 Jakub Jelinek <jakub@redhat.com>
23434 Richard Biener <rguenth@suse.de>
23437 * gimplify.c (gimplify_one_sizepos): For is_gimple_constant (expr)
23438 return early, inline manually is_gimple_sizepos. Make sure if we
23439 call gimplify_expr we don't end up with a gimple constant.
23440 * tree.c (variably_modified_type_p): Don't return true for
23441 is_gimple_constant (_t). Inline manually is_gimple_sizepos.
23442 * gimplify.h (is_gimple_sizepos): Remove.
23444 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
23446 PR tree-optimization/83857
23447 * tree-vect-loop.c (vect_analyze_loop_operations): Don't call
23448 vectorizable_live_operation for pure SLP statements.
23449 (vectorizable_live_operation): Handle PHIs.
23451 2018-01-16 Richard Biener <rguenther@suse.de>
23453 PR tree-optimization/83867
23454 * tree-vect-stmts.c (vect_transform_stmt): Precompute
23455 nested_in_vect_loop_p since the scalar stmt may get invalidated.
23457 2018-01-16 Jakub Jelinek <jakub@redhat.com>
23460 * stor-layout.c (handle_warn_if_not_align): Use byte_position and
23461 multiple_of_p instead of unchecked tree_to_uhwi and UHWI check.
23462 If off is not INTEGER_CST, issue a may not be aligned warning
23463 rather than isn't aligned. Use isn%'t rather than isn't.
23464 * fold-const.c (multiple_of_p) <case BIT_AND_EXPR>: Don't fall through
23466 <case MULT_EXPR>: Improve the case when bottom and one of the
23467 MULT_EXPR operands are INTEGER_CSTs and bottom is multiple of that
23468 operand, in that case check if the other operand is multiple of
23469 bottom divided by the INTEGER_CST operand.
23471 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
23474 * config/pa/pa.h (FUNCTION_ARG_SIZE): Delete.
23475 * config/pa/pa-protos.h (pa_function_arg_size): Declare.
23476 * config/pa/som.h (ASM_DECLARE_FUNCTION_NAME): Use
23477 pa_function_arg_size instead of FUNCTION_ARG_SIZE.
23478 * config/pa/pa.c (pa_function_arg_advance): Likewise.
23479 (pa_function_arg, pa_arg_partial_bytes): Likewise.
23480 (pa_function_arg_size): New function.
23482 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
23484 * fold-const.c (fold_ternary_loc): Construct the vec_perm_indices
23485 in a separate statement.
23487 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
23489 PR tree-optimization/83847
23490 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Don't
23491 group gathers and scatters.
23493 2018-01-16 Jakub Jelinek <jakub@redhat.com>
23495 PR rtl-optimization/86620
23496 * params.def (max-sched-ready-insns): Bump minimum value to 1.
23498 PR rtl-optimization/83213
23499 * recog.c (peep2_attempt): Copy over CROSSING_JUMP_P from peepinsn
23500 to last if both are JUMP_INSNs.
23502 PR tree-optimization/83843
23503 * gimple-ssa-store-merging.c
23504 (imm_store_chain_info::output_merged_store): Handle bit_not_p on
23505 store_immediate_info for bswap/nop orig_stores.
23507 2018-01-15 Andrew Waterman <andrew@sifive.com>
23509 * config/riscv/riscv.c (riscv_rtx_costs) <MULT>: Increase cost if
23511 <UDIV>: Increase cost if !TARGET_DIV.
23513 2018-01-15 Segher Boessenkool <segher@kernel.crashing.org>
23515 * config/rs6000/rs6000.md (define_attr "type"): Remove delayed_cr.
23516 (define_attr "cr_logical_3op"): New.
23517 (cceq_ior_compare): Adjust.
23518 (cceq_ior_compare_complement): Adjust.
23519 (*cceq_rev_compare): Adjust.
23520 * config/rs6000/rs6000.c (rs6000_adjust_cost): Adjust.
23521 (is_cracked_insn): Adjust.
23522 (insn_must_be_first_in_group): Adjust.
23523 * config/rs6000/40x.md: Adjust.
23524 * config/rs6000/440.md: Adjust.
23525 * config/rs6000/476.md: Adjust.
23526 * config/rs6000/601.md: Adjust.
23527 * config/rs6000/603.md: Adjust.
23528 * config/rs6000/6xx.md: Adjust.
23529 * config/rs6000/7450.md: Adjust.
23530 * config/rs6000/7xx.md: Adjust.
23531 * config/rs6000/8540.md: Adjust.
23532 * config/rs6000/cell.md: Adjust.
23533 * config/rs6000/e300c2c3.md: Adjust.
23534 * config/rs6000/e500mc.md: Adjust.
23535 * config/rs6000/e500mc64.md: Adjust.
23536 * config/rs6000/e5500.md: Adjust.
23537 * config/rs6000/e6500.md: Adjust.
23538 * config/rs6000/mpc.md: Adjust.
23539 * config/rs6000/power4.md: Adjust.
23540 * config/rs6000/power5.md: Adjust.
23541 * config/rs6000/power6.md: Adjust.
23542 * config/rs6000/power7.md: Adjust.
23543 * config/rs6000/power8.md: Adjust.
23544 * config/rs6000/power9.md: Adjust.
23545 * config/rs6000/rs64.md: Adjust.
23546 * config/rs6000/titan.md: Adjust.
23548 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
23550 * config/i386/predicates.md (indirect_branch_operand): Rewrite
23551 ix86_indirect_branch_register logic.
23553 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
23555 * config/i386/constraints.md (Bs): Update
23556 ix86_indirect_branch_register check. Don't check
23557 ix86_indirect_branch_register with GOT_memory_operand.
23559 * config/i386/predicates.md (GOT_memory_operand): Don't check
23560 ix86_indirect_branch_register here.
23561 (GOT32_symbol_operand): Likewise.
23563 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
23565 * config/i386/predicates.md (constant_call_address_operand):
23566 Rewrite ix86_indirect_branch_register logic.
23567 (sibcall_insn_operand): Likewise.
23569 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
23571 * config/i386/constraints.md (Bs): Replace
23572 ix86_indirect_branch_thunk_register with
23573 ix86_indirect_branch_register.
23575 * config/i386/i386.md (indirect_jump): Likewise.
23576 (tablejump): Likewise.
23577 (*sibcall_memory): Likewise.
23578 (*sibcall_value_memory): Likewise.
23579 Peepholes of indirect call and jump via memory: Likewise.
23580 * config/i386/i386.opt: Likewise.
23581 * config/i386/predicates.md (indirect_branch_operand): Likewise.
23582 (GOT_memory_operand): Likewise.
23583 (call_insn_operand): Likewise.
23584 (sibcall_insn_operand): Likewise.
23585 (GOT32_symbol_operand): Likewise.
23587 2018-01-15 Jakub Jelinek <jakub@redhat.com>
23589 PR middle-end/83837
23590 * omp-expand.c (expand_omp_atomic_pipeline): Use loaded_val
23591 type rather than type addr's type points to.
23592 (expand_omp_atomic_mutex): Likewise.
23593 (expand_omp_atomic): Likewise.
23595 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
23598 * config/i386/i386.c (output_indirect_thunk_function): Use
23599 ASM_OUTPUT_LABEL, instead of ASM_OUTPUT_DEF, for TARGET_MACHO
23600 for __x86_return_thunk.
23602 2018-01-15 Richard Biener <rguenther@suse.de>
23604 PR middle-end/83850
23605 * expmed.c (extract_bit_field_1): Fix typo.
23607 2018-01-15 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23610 * config/arm/iterators.md (VF): New mode iterator.
23611 * config/arm/neon.md (neon_vabd<mode>_2): Use the above.
23612 Remove integer-related logic from pattern.
23613 (neon_vabd<mode>_3): Likewise.
23615 2018-01-15 Jakub Jelinek <jakub@redhat.com>
23617 PR middle-end/82694
23618 * common.opt (fstrict-overflow): No longer an alias.
23619 (fwrapv-pointer): New option.
23620 * tree.h (TYPE_OVERFLOW_WRAPS, TYPE_OVERFLOW_UNDEFINED): Define
23621 also for pointer types based on flag_wrapv_pointer.
23622 * opts.c (common_handle_option) <case OPT_fstrict_overflow>: Set
23623 opts->x_flag_wrap[pv] to !value, clear opts->x_flag_trapv if
23624 opts->x_flag_wrapv got set.
23625 * fold-const.c (fold_comparison, fold_binary_loc): Revert 2017-08-01
23626 changes, just use TYPE_OVERFLOW_UNDEFINED on pointer type instead of
23627 POINTER_TYPE_OVERFLOW_UNDEFINED.
23628 * match.pd: Likewise in address comparison pattern.
23629 * doc/invoke.texi: Document -fwrapv and -fstrict-overflow.
23631 2018-01-15 Richard Biener <rguenther@suse.de>
23634 * tree.c (free_lang_data_in_type): Always unlink TYPE_DECLs
23635 from TYPE_FIELDS. Free TYPE_BINFO if not used by devirtualization.
23636 Reset type names to their identifier if their TYPE_DECL doesn't
23637 have linkage (and thus is used for ODR and devirt).
23638 (save_debug_info_for_decl): Remove.
23639 (save_debug_info_for_type): Likewise.
23640 (add_tree_to_fld_list): Adjust.
23641 * tree-pretty-print.c (dump_generic_node): Make dumping of
23642 type names more robust.
23644 2018-01-15 Richard Biener <rguenther@suse.de>
23646 * BASE-VER: Bump to 8.0.1.
23648 2018-01-14 Martin Sebor <msebor@redhat.com>
23651 * builtins.c (check_access): Avoid warning when the no-warning bit
23654 2018-01-14 Cory Fields <cory-nospam-@coryfields.com>
23656 * tree-ssa-loop-im.c (sort_bbs_in_loop_postorder_cmp): Stabilize sort.
23657 * ira-color (allocno_hard_regs_compare): Likewise.
23659 2018-01-14 Nathan Rossi <nathan@nathanrossi.com>
23662 * config/microblaze/microblaze.c (microblaze_asm_output_ident):
23663 Use .pushsection/.popsection.
23665 2018-01-14 Martin Sebor <msebor@redhat.com>
23668 * doc/invoke.texi (-Wlass-memaccess): Document suppression by casting.
23670 2018-01-14 Jakub Jelinek <jakub@redhat.com>
23672 * config.gcc (i[34567]86-*-*): Remove one duplicate gfniintrin.h
23673 entry from extra_headers.
23674 (x86_64-*-*): Remove two duplicate gfniintrin.h entries from
23675 extra_headers, make the list bitwise identical to the i?86-*-* one.
23677 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
23679 * config/i386/i386.c (ix86_set_indirect_branch_type): Disallow
23680 -mcmodel=large with -mindirect-branch=thunk,
23681 -mindirect-branch=thunk-extern, -mfunction-return=thunk and
23682 -mfunction-return=thunk-extern.
23683 * doc/invoke.texi: Document -mcmodel=large is incompatible with
23684 -mindirect-branch=thunk, -mindirect-branch=thunk-extern,
23685 -mfunction-return=thunk and -mfunction-return=thunk-extern.
23687 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
23689 * config/i386/i386.c (print_reg): Print the name of the full
23690 integer register without '%'.
23691 (ix86_print_operand): Handle 'V'.
23692 * doc/extend.texi: Document 'V' modifier.
23694 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
23696 * config/i386/constraints.md (Bs): Disallow memory operand for
23697 -mindirect-branch-register.
23699 * config/i386/predicates.md (indirect_branch_operand): Likewise.
23700 (GOT_memory_operand): Likewise.
23701 (call_insn_operand): Likewise.
23702 (sibcall_insn_operand): Likewise.
23703 (GOT32_symbol_operand): Likewise.
23704 * config/i386/i386.md (indirect_jump): Call convert_memory_address
23705 for -mindirect-branch-register.
23706 (tablejump): Likewise.
23707 (*sibcall_memory): Likewise.
23708 (*sibcall_value_memory): Likewise.
23709 Disallow peepholes of indirect call and jump via memory for
23710 -mindirect-branch-register.
23711 (*call_pop): Replace m with Bw.
23712 (*call_value_pop): Likewise.
23713 (*sibcall_pop_memory): Replace m with Bs.
23714 * config/i386/i386.opt (mindirect-branch-register): New option.
23715 * doc/invoke.texi: Document -mindirect-branch-register option.
23717 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
23719 * config/i386/i386-protos.h (ix86_output_function_return): New.
23720 * config/i386/i386.c (ix86_set_indirect_branch_type): Also
23721 set function_return_type.
23722 (indirect_thunk_name): Add ret_p to indicate thunk for function
23724 (output_indirect_thunk_function): Pass false to
23725 indirect_thunk_name.
23726 (ix86_output_indirect_branch_via_reg): Likewise.
23727 (ix86_output_indirect_branch_via_push): Likewise.
23728 (output_indirect_thunk_function): Create alias for function
23729 return thunk if regno < 0.
23730 (ix86_output_function_return): New function.
23731 (ix86_handle_fndecl_attribute): Handle function_return.
23732 (ix86_attribute_table): Add function_return.
23733 * config/i386/i386.h (machine_function): Add
23734 function_return_type.
23735 * config/i386/i386.md (simple_return_internal): Use
23736 ix86_output_function_return.
23737 (simple_return_internal_long): Likewise.
23738 * config/i386/i386.opt (mfunction-return=): New option.
23739 (indirect_branch): Mention -mfunction-return=.
23740 * doc/extend.texi: Document function_return function attribute.
23741 * doc/invoke.texi: Document -mfunction-return= option.
23743 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
23745 * config/i386/i386-opts.h (indirect_branch): New.
23746 * config/i386/i386-protos.h (ix86_output_indirect_jmp): Likewise.
23747 * config/i386/i386.c (ix86_using_red_zone): Disallow red-zone
23748 with local indirect jump when converting indirect call and jump.
23749 (ix86_set_indirect_branch_type): New.
23750 (ix86_set_current_function): Call ix86_set_indirect_branch_type.
23751 (indirectlabelno): New.
23752 (indirect_thunk_needed): Likewise.
23753 (indirect_thunk_bnd_needed): Likewise.
23754 (indirect_thunks_used): Likewise.
23755 (indirect_thunks_bnd_used): Likewise.
23756 (INDIRECT_LABEL): Likewise.
23757 (indirect_thunk_name): Likewise.
23758 (output_indirect_thunk): Likewise.
23759 (output_indirect_thunk_function): Likewise.
23760 (ix86_output_indirect_branch_via_reg): Likewise.
23761 (ix86_output_indirect_branch_via_push): Likewise.
23762 (ix86_output_indirect_branch): Likewise.
23763 (ix86_output_indirect_jmp): Likewise.
23764 (ix86_code_end): Call output_indirect_thunk_function if needed.
23765 (ix86_output_call_insn): Call ix86_output_indirect_branch if
23767 (ix86_handle_fndecl_attribute): Handle indirect_branch.
23768 (ix86_attribute_table): Add indirect_branch.
23769 * config/i386/i386.h (machine_function): Add indirect_branch_type
23770 and has_local_indirect_jump.
23771 * config/i386/i386.md (indirect_jump): Set has_local_indirect_jump
23773 (tablejump): Likewise.
23774 (*indirect_jump): Use ix86_output_indirect_jmp.
23775 (*tablejump_1): Likewise.
23776 (simple_return_indirect_internal): Likewise.
23777 * config/i386/i386.opt (mindirect-branch=): New option.
23778 (indirect_branch): New.
23781 (thunk-inline): Likewise.
23782 (thunk-extern): Likewise.
23783 * doc/extend.texi: Document indirect_branch function attribute.
23784 * doc/invoke.texi: Document -mindirect-branch= option.
23786 2018-01-14 Jan Hubicka <hubicka@ucw.cz>
23789 * ipa-inline.c (edge_badness): Tolerate roundoff errors.
23791 2018-01-14 Richard Sandiford <richard.sandiford@linaro.org>
23793 * ipa-inline.c (want_inline_small_function_p): Return false if
23794 inlining has already failed with CIF_FINAL_ERROR.
23795 (update_caller_keys): Call want_inline_small_function_p before
23797 (update_callee_keys): Likewise.
23799 2018-01-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
23801 * config/rs6000/rs6000-p8swap.c (rs6000_sum_of_two_registers_p):
23803 (rs6000_quadword_masked_address_p): Likewise.
23804 (quad_aligned_load_p): Likewise.
23805 (quad_aligned_store_p): Likewise.
23806 (const_load_sequence_p): Add comment to describe the outer-most loop.
23807 (mimic_memory_attributes_and_flags): New function.
23808 (rs6000_gen_stvx): Likewise.
23809 (replace_swapped_aligned_store): Likewise.
23810 (rs6000_gen_lvx): Likewise.
23811 (replace_swapped_aligned_load): Likewise.
23812 (replace_swapped_load_constant): Capitalize argument name in
23813 comment describing this function.
23814 (rs6000_analyze_swaps): Add a third pass to search for vector loads
23815 and stores that access quad-word aligned addresses and replace
23816 with stvx or lvx instructions when appropriate.
23817 * config/rs6000/rs6000-protos.h (rs6000_sum_of_two_registers_p):
23818 New function prototype.
23819 (rs6000_quadword_masked_address_p): Likewise.
23820 (rs6000_gen_lvx): Likewise.
23821 (rs6000_gen_stvx): Likewise.
23822 * config/rs6000/vsx.md (*vsx_le_perm_load_<mode>): For modes
23823 VSX_D (V2DF, V2DI), modify this split to select lvx instruction
23824 when memory address is aligned.
23825 (*vsx_le_perm_load_<mode>): For modes VSX_W (V4SF, V4SI), modify
23826 this split to select lvx instruction when memory address is aligned.
23827 (*vsx_le_perm_load_v8hi): Modify this split to select lvx
23828 instruction when memory address is aligned.
23829 (*vsx_le_perm_load_v16qi): Likewise.
23830 (four unnamed splitters): Modify to select the stvx instruction
23831 when memory is aligned.
23833 2018-01-13 Jan Hubicka <hubicka@ucw.cz>
23835 * predict.c (determine_unlikely_bbs): Handle correctly BBs
23836 which appears in the queue multiple times.
23838 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
23839 Alan Hayward <alan.hayward@arm.com>
23840 David Sherwood <david.sherwood@arm.com>
23842 * tree-vectorizer.h (vec_lower_bound): New structure.
23843 (_loop_vec_info): Add check_nonzero and lower_bounds.
23844 (LOOP_VINFO_CHECK_NONZERO): New macro.
23845 (LOOP_VINFO_LOWER_BOUNDS): Likewise.
23846 (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Check lower_bounds too.
23847 * tree-data-ref.h (dr_with_seg_len): Add access_size and align
23848 fields. Make seg_len the distance travelled, not including the
23850 (dr_direction_indicator): Declare.
23851 (dr_zero_step_indicator): Likewise.
23852 (dr_known_forward_stride_p): Likewise.
23853 * tree-data-ref.c: Include stringpool.h, tree-vrp.h and
23855 (runtime_alias_check_p): Allow runtime alias checks with
23857 (operator ==): Compare access_size and align.
23858 (prune_runtime_alias_test_list): Rework for new distinction between
23859 the access_size and seg_len.
23860 (create_intersect_range_checks_index): Likewise. Cope with polynomial
23862 (get_segment_min_max): New function.
23863 (create_intersect_range_checks): Use it.
23864 (dr_step_indicator): New function.
23865 (dr_direction_indicator): Likewise.
23866 (dr_zero_step_indicator): Likewise.
23867 (dr_known_forward_stride_p): Likewise.
23868 * tree-loop-distribution.c (data_ref_segment_size): Return
23869 DR_STEP * (niters - 1).
23870 (compute_alias_check_pairs): Update call to the dr_with_seg_len
23872 * tree-vect-data-refs.c (vect_check_nonzero_value): New function.
23873 (vect_preserves_scalar_order_p): New function, split out from...
23874 (vect_analyze_data_ref_dependence): ...here. Check for zero steps.
23875 (vect_vfa_segment_size): Return DR_STEP * (length_factor - 1).
23876 (vect_vfa_access_size): New function.
23877 (vect_vfa_align): Likewise.
23878 (vect_compile_time_alias): Take access_size_a and access_b arguments.
23879 (dump_lower_bound): New function.
23880 (vect_check_lower_bound): Likewise.
23881 (vect_small_gap_p): Likewise.
23882 (vectorizable_with_step_bound_p): Likewise.
23883 (vect_prune_runtime_alias_test_list): Ignore cross-iteration
23884 depencies if the vectorization factor is 1. Convert the checks
23885 for nonzero steps into checks on the bounds of DR_STEP. Try using
23886 a bunds check for variable steps if the minimum required step is
23887 relatively small. Update calls to the dr_with_seg_len
23888 constructor and to vect_compile_time_alias.
23889 * tree-vect-loop-manip.c (vect_create_cond_for_lower_bounds): New
23891 (vect_loop_versioning): Call it.
23892 * tree-vect-loop.c (vect_analyze_loop_2): Clear LOOP_VINFO_LOWER_BOUNDS
23894 (vect_estimate_min_profitable_iters): Account for any bounds checks.
23896 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
23897 Alan Hayward <alan.hayward@arm.com>
23898 David Sherwood <david.sherwood@arm.com>
23900 * doc/sourcebuild.texi (vect_scatter_store): Document.
23901 * optabs.def (scatter_store_optab, mask_scatter_store_optab): New
23903 * doc/md.texi (scatter_store@var{m}, mask_scatter_store@var{m}):
23905 * genopinit.c (main): Add supports_vec_scatter_store and
23906 supports_vec_scatter_store_cached to target_optabs.
23907 * gimple.h (gimple_expr_type): Handle IFN_SCATTER_STORE and
23908 IFN_MASK_SCATTER_STORE.
23909 * internal-fn.def (SCATTER_STORE, MASK_SCATTER_STORE): New internal
23911 * internal-fn.h (internal_store_fn_p): Declare.
23912 (internal_fn_stored_value_index): Likewise.
23913 * internal-fn.c (scatter_store_direct): New macro.
23914 (expand_scatter_store_optab_fn): New function.
23915 (direct_scatter_store_optab_supported_p): New macro.
23916 (internal_store_fn_p): New function.
23917 (internal_gather_scatter_fn_p): Handle IFN_SCATTER_STORE and
23918 IFN_MASK_SCATTER_STORE.
23919 (internal_fn_mask_index): Likewise.
23920 (internal_fn_stored_value_index): New function.
23921 (internal_gather_scatter_fn_supported_p): Adjust operand numbers
23922 for scatter stores.
23923 * optabs-query.h (supports_vec_scatter_store_p): Declare.
23924 * optabs-query.c (supports_vec_scatter_store_p): New function.
23925 * tree-vectorizer.h (vect_get_store_rhs): Declare.
23926 * tree-vect-data-refs.c (vect_analyze_data_ref_access): Return
23927 true for scatter stores.
23928 (vect_gather_scatter_fn_p): Handle scatter stores too.
23929 (vect_check_gather_scatter): Consider using scatter stores if
23930 supports_vec_scatter_store_p.
23931 * tree-vect-patterns.c (vect_try_gather_scatter_pattern): Handle
23932 scatter stores too.
23933 * tree-vect-stmts.c (exist_non_indexing_operands_for_use_p): Use
23934 internal_fn_stored_value_index.
23935 (check_load_store_masking): Handle scatter stores too.
23936 (vect_get_store_rhs): Make public.
23937 (vectorizable_call): Use internal_store_fn_p.
23938 (vectorizable_store): Handle scatter store internal functions.
23939 (vect_transform_stmt): Compare GROUP_STORE_COUNT with GROUP_SIZE
23940 when deciding whether the end of the group has been reached.
23941 * config/aarch64/aarch64.md (UNSPEC_ST1_SCATTER): New unspec.
23942 * config/aarch64/aarch64-sve.md (scatter_store<mode>): New expander.
23943 (mask_scatter_store<mode>): New insns.
23945 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
23946 Alan Hayward <alan.hayward@arm.com>
23947 David Sherwood <david.sherwood@arm.com>
23949 * tree-vectorizer.h (vect_gather_scatter_fn_p): Declare.
23950 * tree-vect-data-refs.c (vect_gather_scatter_fn_p): Make public.
23951 * tree-vect-stmts.c (vect_truncate_gather_scatter_offset): New
23953 (vect_use_strided_gather_scatters_p): Take a masked_p argument.
23954 Use vect_truncate_gather_scatter_offset if we can't treat the
23955 operation as a normal gather load or scatter store.
23956 (get_group_load_store_type): Take the gather_scatter_info
23957 as argument. Try using a gather load or scatter store for
23958 single-element groups.
23959 (get_load_store_type): Update calls to get_group_load_store_type
23960 and vect_use_strided_gather_scatters_p.
23962 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
23963 Alan Hayward <alan.hayward@arm.com>
23964 David Sherwood <david.sherwood@arm.com>
23966 * tree-vectorizer.h (vect_create_data_ref_ptr): Take an extra
23967 optional tree argument.
23968 * tree-vect-data-refs.c (vect_check_gather_scatter): Check for
23970 (vect_create_data_ref_ptr): Take the iv_step as an optional argument,
23971 but continue to use the current value as a fallback.
23972 (bump_vector_ptr): Use operand_equal_p rather than tree_int_cst_compare
23973 to compare the updates.
23974 * tree-vect-stmts.c (vect_use_strided_gather_scatters_p): New function.
23975 (get_load_store_type): Use it when handling a strided access.
23976 (vect_get_strided_load_store_ops): New function.
23977 (vect_get_data_ptr_increment): Likewise.
23978 (vectorizable_load): Handle strided gather loads. Always pass
23979 a step to vect_create_data_ref_ptr and bump_vector_ptr.
23981 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
23982 Alan Hayward <alan.hayward@arm.com>
23983 David Sherwood <david.sherwood@arm.com>
23985 * doc/md.texi (gather_load@var{m}): Document.
23986 (mask_gather_load@var{m}): Likewise.
23987 * genopinit.c (main): Add supports_vec_gather_load and
23988 supports_vec_gather_load_cached to target_optabs.
23989 * optabs-tree.c (init_tree_optimization_optabs): Use
23990 ggc_cleared_alloc to allocate target_optabs.
23991 * optabs.def (gather_load_optab, mask_gather_laod_optab): New optabs.
23992 * internal-fn.def (GATHER_LOAD, MASK_GATHER_LOAD): New internal
23994 * internal-fn.h (internal_load_fn_p): Declare.
23995 (internal_gather_scatter_fn_p): Likewise.
23996 (internal_fn_mask_index): Likewise.
23997 (internal_gather_scatter_fn_supported_p): Likewise.
23998 * internal-fn.c (gather_load_direct): New macro.
23999 (expand_gather_load_optab_fn): New function.
24000 (direct_gather_load_optab_supported_p): New macro.
24001 (direct_internal_fn_optab): New function.
24002 (internal_load_fn_p): Likewise.
24003 (internal_gather_scatter_fn_p): Likewise.
24004 (internal_fn_mask_index): Likewise.
24005 (internal_gather_scatter_fn_supported_p): Likewise.
24006 * optabs-query.c (supports_at_least_one_mode_p): New function.
24007 (supports_vec_gather_load_p): Likewise.
24008 * optabs-query.h (supports_vec_gather_load_p): Declare.
24009 * tree-vectorizer.h (gather_scatter_info): Add ifn, element_type
24010 and memory_type field.
24011 (NUM_PATTERNS): Bump to 15.
24012 * tree-vect-data-refs.c: Include internal-fn.h.
24013 (vect_gather_scatter_fn_p): New function.
24014 (vect_describe_gather_scatter_call): Likewise.
24015 (vect_check_gather_scatter): Try using internal functions for
24016 gather loads. Recognize existing calls to a gather load function.
24017 (vect_analyze_data_refs): Consider using gather loads if
24018 supports_vec_gather_load_p.
24019 * tree-vect-patterns.c (vect_get_load_store_mask): New function.
24020 (vect_get_gather_scatter_offset_type): Likewise.
24021 (vect_convert_mask_for_vectype): Likewise.
24022 (vect_add_conversion_to_patterm): Likewise.
24023 (vect_try_gather_scatter_pattern): Likewise.
24024 (vect_recog_gather_scatter_pattern): New pattern recognizer.
24025 (vect_vect_recog_func_ptrs): Add it.
24026 * tree-vect-stmts.c (exist_non_indexing_operands_for_use_p): Use
24027 internal_fn_mask_index and internal_gather_scatter_fn_p.
24028 (check_load_store_masking): Take the gather_scatter_info as an
24029 argument and handle gather loads.
24030 (vect_get_gather_scatter_ops): New function.
24031 (vectorizable_call): Check internal_load_fn_p.
24032 (vectorizable_load): Likewise. Handle gather load internal
24034 (vectorizable_store): Update call to check_load_store_masking.
24035 * config/aarch64/aarch64.md (UNSPEC_LD1_GATHER): New unspec.
24036 * config/aarch64/iterators.md (SVE_S, SVE_D): New mode iterators.
24037 * config/aarch64/predicates.md (aarch64_gather_scale_operand_w)
24038 (aarch64_gather_scale_operand_d): New predicates.
24039 * config/aarch64/aarch64-sve.md (gather_load<mode>): New expander.
24040 (mask_gather_load<mode>): New insns.
24042 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
24043 Alan Hayward <alan.hayward@arm.com>
24044 David Sherwood <david.sherwood@arm.com>
24046 * optabs.def (fold_left_plus_optab): New optab.
24047 * doc/md.texi (fold_left_plus_@var{m}): Document.
24048 * internal-fn.def (IFN_FOLD_LEFT_PLUS): New internal function.
24049 * internal-fn.c (fold_left_direct): Define.
24050 (expand_fold_left_optab_fn): Likewise.
24051 (direct_fold_left_optab_supported_p): Likewise.
24052 * fold-const-call.c (fold_const_fold_left): New function.
24053 (fold_const_call): Use it to fold CFN_FOLD_LEFT_PLUS.
24054 * tree-parloops.c (valid_reduction_p): New function.
24055 (gather_scalar_reductions): Use it.
24056 * tree-vectorizer.h (FOLD_LEFT_REDUCTION): New vect_reduction_type.
24057 (vect_finish_replace_stmt): Declare.
24058 * tree-vect-loop.c (fold_left_reduction_fn): New function.
24059 (needs_fold_left_reduction_p): New function, split out from...
24060 (vect_is_simple_reduction): ...here. Accept reductions that
24061 forbid reassociation, but give them type FOLD_LEFT_REDUCTION.
24062 (vect_force_simple_reduction): Also store the reduction type in
24063 the assignment's STMT_VINFO_REDUC_TYPE.
24064 (vect_model_reduction_cost): Handle FOLD_LEFT_REDUCTION.
24065 (merge_with_identity): New function.
24066 (vect_expand_fold_left): Likewise.
24067 (vectorize_fold_left_reduction): Likewise.
24068 (vectorizable_reduction): Handle FOLD_LEFT_REDUCTION. Leave the
24069 scalar phi in place for it. Check for target support and reject
24070 cases that would reassociate the operation. Defer the transform
24071 phase to vectorize_fold_left_reduction.
24072 * config/aarch64/aarch64.md (UNSPEC_FADDA): New unspec.
24073 * config/aarch64/aarch64-sve.md (fold_left_plus_<mode>): New expander.
24074 (*fold_left_plus_<mode>, *pred_fold_left_plus_<mode>): New insns.
24076 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
24078 * tree-if-conv.c (predicate_mem_writes): Remove redundant
24079 call to ifc_temp_var.
24081 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
24082 Alan Hayward <alan.hayward@arm.com>
24083 David Sherwood <david.sherwood@arm.com>
24085 * target.def (legitimize_address_displacement): Take the original
24086 offset as a poly_int.
24087 * targhooks.h (default_legitimize_address_displacement): Update
24089 * targhooks.c (default_legitimize_address_displacement): Likewise.
24090 * doc/tm.texi: Regenerate.
24091 * lra-constraints.c (base_plus_disp_to_reg): Take the displacement
24092 as an argument, moving assert of ad->disp == ad->disp_term to...
24093 (process_address_1): ...here. Update calls to base_plus_disp_to_reg.
24094 Try calling targetm.legitimize_address_displacement before expanding
24095 the address rather than afterwards, and adjust for the new interface.
24096 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
24097 Match the new hook interface. Handle SVE addresses.
24098 * config/sh/sh.c (sh_legitimize_address_displacement): Make the
24099 new hook interface.
24101 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
24103 * Makefile.in (OBJS): Add early-remat.o.
24104 * target.def (select_early_remat_modes): New hook.
24105 * doc/tm.texi.in (TARGET_SELECT_EARLY_REMAT_MODES): New hook.
24106 * doc/tm.texi: Regenerate.
24107 * targhooks.h (default_select_early_remat_modes): Declare.
24108 * targhooks.c (default_select_early_remat_modes): New function.
24109 * timevar.def (TV_EARLY_REMAT): New timevar.
24110 * passes.def (pass_early_remat): New pass.
24111 * tree-pass.h (make_pass_early_remat): Declare.
24112 * early-remat.c: New file.
24113 * config/aarch64/aarch64.c (aarch64_select_early_remat_modes): New
24115 (TARGET_SELECT_EARLY_REMAT_MODES): Define.
24117 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
24118 Alan Hayward <alan.hayward@arm.com>
24119 David Sherwood <david.sherwood@arm.com>
24121 * tree-vect-loop-manip.c (vect_gen_scalar_loop_niters): Replace
24122 vfm1 with a bound_epilog parameter.
24123 (vect_do_peeling): Update calls accordingly, and move the prologue
24124 call earlier in the function. Treat the base bound_epilog as 0 for
24125 fully-masked loops and retain vf - 1 for other loops. Add 1 to
24126 this base when peeling for gaps.
24127 * tree-vect-loop.c (vect_analyze_loop_2): Allow peeling for gaps
24128 with fully-masked loops.
24129 (vect_estimate_min_profitable_iters): Handle the single peeled
24130 iteration in that case.
24132 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
24133 Alan Hayward <alan.hayward@arm.com>
24134 David Sherwood <david.sherwood@arm.com>
24136 * tree-vect-data-refs.c (vect_analyze_group_access_1): Allow
24137 single-element interleaving even if the size is not a power of 2.
24138 * tree-vect-stmts.c (get_load_store_type): Disallow elementwise
24139 accesses for single-element interleaving if the group size is
24142 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
24143 Alan Hayward <alan.hayward@arm.com>
24144 David Sherwood <david.sherwood@arm.com>
24146 * doc/md.texi (fold_extract_last_@var{m}): Document.
24147 * doc/sourcebuild.texi (vect_fold_extract_last): Likewise.
24148 * optabs.def (fold_extract_last_optab): New optab.
24149 * internal-fn.def (FOLD_EXTRACT_LAST): New internal function.
24150 * internal-fn.c (fold_extract_direct): New macro.
24151 (expand_fold_extract_optab_fn): Likewise.
24152 (direct_fold_extract_optab_supported_p): Likewise.
24153 * tree-vectorizer.h (EXTRACT_LAST_REDUCTION): New vect_reduction_type.
24154 * tree-vect-loop.c (vect_model_reduction_cost): Handle
24155 EXTRACT_LAST_REDUCTION.
24156 (get_initial_def_for_reduction): Do not create an initial vector
24157 for EXTRACT_LAST_REDUCTION reductions.
24158 (vectorizable_reduction): Leave the scalar phi in place for
24159 EXTRACT_LAST_REDUCTIONs. Try using EXTRACT_LAST_REDUCTION
24160 ahead of INTEGER_INDUC_COND_REDUCTION. Do not check for an
24161 epilogue code for EXTRACT_LAST_REDUCTION and defer the
24162 transform phase to vectorizable_condition.
24163 * tree-vect-stmts.c (vect_finish_stmt_generation_1): New function,
24165 (vect_finish_stmt_generation): ...here.
24166 (vect_finish_replace_stmt): New function.
24167 (vectorizable_condition): Handle EXTRACT_LAST_REDUCTION.
24168 * config/aarch64/aarch64-sve.md (fold_extract_last_<mode>): New
24170 * config/aarch64/aarch64.md (UNSPEC_CLASTB): New unspec.
24172 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
24173 Alan Hayward <alan.hayward@arm.com>
24174 David Sherwood <david.sherwood@arm.com>
24176 * doc/md.texi (extract_last_@var{m}): Document.
24177 * optabs.def (extract_last_optab): New optab.
24178 * internal-fn.def (EXTRACT_LAST): New internal function.
24179 * internal-fn.c (cond_unary_direct): New macro.
24180 (expand_cond_unary_optab_fn): Likewise.
24181 (direct_cond_unary_optab_supported_p): Likewise.
24182 * tree-vect-loop.c (vectorizable_live_operation): Allow fully-masked
24183 loops using EXTRACT_LAST.
24184 * config/aarch64/aarch64-sve.md (aarch64_sve_lastb<mode>): Rename to...
24185 (extract_last_<mode>): ...this optab.
24186 (vec_extract<mode><Vel>): Update accordingly.
24188 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
24189 Alan Hayward <alan.hayward@arm.com>
24190 David Sherwood <david.sherwood@arm.com>
24192 * target.def (empty_mask_is_expensive): New hook.
24193 * doc/tm.texi.in (TARGET_VECTORIZE_EMPTY_MASK_IS_EXPENSIVE): New hook.
24194 * doc/tm.texi: Regenerate.
24195 * targhooks.h (default_empty_mask_is_expensive): Declare.
24196 * targhooks.c (default_empty_mask_is_expensive): New function.
24197 * tree-vectorizer.c (vectorize_loops): Only call optimize_mask_stores
24198 if the target says that empty masks are expensive.
24199 * config/aarch64/aarch64.c (aarch64_empty_mask_is_expensive):
24201 (TARGET_VECTORIZE_EMPTY_MASK_IS_EXPENSIVE): Redefine.
24203 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
24204 Alan Hayward <alan.hayward@arm.com>
24205 David Sherwood <david.sherwood@arm.com>
24207 * tree-vectorizer.h (_loop_vec_info::mask_skip_niters): New field.
24208 (LOOP_VINFO_MASK_SKIP_NITERS): New macro.
24209 (vect_use_loop_mask_for_alignment_p): New function.
24210 (vect_prepare_for_masked_peels, vect_gen_while_not): Declare.
24211 * tree-vect-loop-manip.c (vect_set_loop_masks_directly): Add an
24212 niters_skip argument. Make sure that the first niters_skip elements
24213 of the first iteration are inactive.
24214 (vect_set_loop_condition_masked): Handle LOOP_VINFO_MASK_SKIP_NITERS.
24215 Update call to vect_set_loop_masks_directly.
24216 (get_misalign_in_elems): New function, split out from...
24217 (vect_gen_prolog_loop_niters): ...here.
24218 (vect_update_init_of_dr): Take a code argument that specifies whether
24219 the adjustment should be added or subtracted.
24220 (vect_update_init_of_drs): Likewise.
24221 (vect_prepare_for_masked_peels): New function.
24222 (vect_do_peeling): Skip prologue peeling if we're using a mask
24223 instead. Update call to vect_update_inits_of_drs.
24224 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
24226 (vect_analyze_loop_2): Allow fully-masked loops with peeling for
24227 alignment. Do not include the number of peeled iterations in
24228 the minimum threshold in that case.
24229 (vectorizable_induction): Adjust the start value down by
24230 LOOP_VINFO_MASK_SKIP_NITERS iterations.
24231 (vect_transform_loop): Call vect_prepare_for_masked_peels.
24232 Take the number of skipped iterations into account when calculating
24234 * tree-vect-stmts.c (vect_gen_while_not): New function.
24236 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
24237 Alan Hayward <alan.hayward@arm.com>
24238 David Sherwood <david.sherwood@arm.com>
24240 * doc/sourcebuild.texi (vect_fully_masked): Document.
24241 * params.def (PARAM_MIN_VECT_LOOP_BOUND): Change minimum and
24242 default value to 0.
24243 * tree-vect-loop.c (vect_analyze_loop_costing): New function,
24245 (vect_analyze_loop_2): ...here. Don't check the vectorization
24246 factor against the number of loop iterations if the loop is
24249 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
24250 Alan Hayward <alan.hayward@arm.com>
24251 David Sherwood <david.sherwood@arm.com>
24253 * tree-ssa-loop-ivopts.c (USE_ADDRESS): Split into...
24254 (USE_REF_ADDRESS, USE_PTR_ADDRESS): ...these new use types.
24255 (dump_groups): Update accordingly.
24256 (iv_use::mem_type): New member variable.
24257 (address_p): New function.
24258 (record_use): Add a mem_type argument and initialize the new
24260 (record_group_use): Add a mem_type argument. Use address_p.
24261 Remove obsolete null checks of base_object. Update call to record_use.
24262 (find_interesting_uses_op): Update call to record_group_use.
24263 (find_interesting_uses_cond): Likewise.
24264 (find_interesting_uses_address): Likewise.
24265 (get_mem_type_for_internal_fn): New function.
24266 (find_address_like_use): Likewise.
24267 (find_interesting_uses_stmt): Try find_address_like_use before
24268 calling find_interesting_uses_op.
24269 (addr_offset_valid_p): Use the iv mem_type field as the type
24270 of the addressed memory.
24271 (add_autoinc_candidates): Likewise.
24272 (get_address_cost): Likewise.
24273 (split_small_address_groups_p): Use address_p.
24274 (split_address_groups): Likewise.
24275 (add_iv_candidate_for_use): Likewise.
24276 (autoinc_possible_for_pair): Likewise.
24277 (rewrite_groups): Likewise.
24278 (get_use_type): Check for USE_REF_ADDRESS instead of USE_ADDRESS.
24279 (determine_group_iv_cost): Update after split of USE_ADDRESS.
24280 (get_alias_ptr_type_for_ptr_address): New function.
24281 (rewrite_use_address): Rewrite address uses in calls that were
24282 identified by find_address_like_use.
24284 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
24285 Alan Hayward <alan.hayward@arm.com>
24286 David Sherwood <david.sherwood@arm.com>
24288 * expr.c (expand_expr_addr_expr_1): Handle ADDR_EXPRs of
24290 * gimple-expr.h (is_gimple_addressable: Likewise.
24291 * gimple-expr.c (is_gimple_address): Likewise.
24292 * internal-fn.c (expand_call_mem_ref): New function.
24293 (expand_mask_load_optab_fn): Use it.
24294 (expand_mask_store_optab_fn): Likewise.
24296 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
24297 Alan Hayward <alan.hayward@arm.com>
24298 David Sherwood <david.sherwood@arm.com>
24300 * doc/md.texi (cond_add@var{mode}, cond_sub@var{mode})
24301 (cond_and@var{mode}, cond_ior@var{mode}, cond_xor@var{mode})
24302 (cond_smin@var{mode}, cond_smax@var{mode}, cond_umin@var{mode})
24303 (cond_umax@var{mode}): Document.
24304 * optabs.def (cond_add_optab, cond_sub_optab, cond_and_optab)
24305 (cond_ior_optab, cond_xor_optab, cond_smin_optab, cond_smax_optab)
24306 (cond_umin_optab, cond_umax_optab): New optabs.
24307 * internal-fn.def (COND_ADD, COND_SUB, COND_MIN, COND_MAX, COND_AND)
24308 (COND_IOR, COND_XOR): New internal functions.
24309 * internal-fn.h (get_conditional_internal_fn): Declare.
24310 * internal-fn.c (cond_binary_direct): New macro.
24311 (expand_cond_binary_optab_fn): Likewise.
24312 (direct_cond_binary_optab_supported_p): Likewise.
24313 (get_conditional_internal_fn): New function.
24314 * tree-vect-loop.c (vectorizable_reduction): Handle fully-masked loops.
24315 Cope with reduction statements that are vectorized as calls rather
24317 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): New insns.
24318 * config/aarch64/iterators.md (UNSPEC_COND_ADD, UNSPEC_COND_SUB)
24319 (UNSPEC_COND_SMAX, UNSPEC_COND_UMAX, UNSPEC_COND_SMIN)
24320 (UNSPEC_COND_UMIN, UNSPEC_COND_AND, UNSPEC_COND_ORR)
24321 (UNSPEC_COND_EOR): New unspecs.
24322 (optab): Add mappings for them.
24323 (SVE_COND_INT_OP, SVE_COND_FP_OP): New int iterators.
24324 (sve_int_op, sve_fp_op): New int attributes.
24326 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
24327 Alan Hayward <alan.hayward@arm.com>
24328 David Sherwood <david.sherwood@arm.com>
24330 * optabs.def (while_ult_optab): New optab.
24331 * doc/md.texi (while_ult@var{m}@var{n}): Document.
24332 * internal-fn.def (WHILE_ULT): New internal function.
24333 * internal-fn.h (direct_internal_fn_supported_p): New override
24334 that takes two types as argument.
24335 * internal-fn.c (while_direct): New macro.
24336 (expand_while_optab_fn): New function.
24337 (convert_optab_supported_p): Likewise.
24338 (direct_while_optab_supported_p): New macro.
24339 * wide-int.h (wi::udiv_ceil): New function.
24340 * tree-vectorizer.h (rgroup_masks): New structure.
24341 (vec_loop_masks): New typedef.
24342 (_loop_vec_info): Add masks, mask_compare_type, can_fully_mask_p
24343 and fully_masked_p.
24344 (LOOP_VINFO_CAN_FULLY_MASK_P, LOOP_VINFO_FULLY_MASKED_P)
24345 (LOOP_VINFO_MASKS, LOOP_VINFO_MASK_COMPARE_TYPE): New macros.
24346 (vect_max_vf): New function.
24347 (slpeel_make_loop_iterate_ntimes): Delete.
24348 (vect_set_loop_condition, vect_get_loop_mask_type, vect_gen_while)
24349 (vect_halve_mask_nunits, vect_double_mask_nunits): Declare.
24350 (vect_record_loop_mask, vect_get_loop_mask): Likewise.
24351 * tree-vect-loop-manip.c: Include tree-ssa-loop-niter.h,
24352 internal-fn.h, stor-layout.h and optabs-query.h.
24353 (vect_set_loop_mask): New function.
24354 (add_preheader_seq): Likewise.
24355 (add_header_seq): Likewise.
24356 (interleave_supported_p): Likewise.
24357 (vect_maybe_permute_loop_masks): Likewise.
24358 (vect_set_loop_masks_directly): Likewise.
24359 (vect_set_loop_condition_masked): Likewise.
24360 (vect_set_loop_condition_unmasked): New function, split out from
24361 slpeel_make_loop_iterate_ntimes.
24362 (slpeel_make_loop_iterate_ntimes): Rename to..
24363 (vect_set_loop_condition): ...this. Use vect_set_loop_condition_masked
24364 for fully-masked loops and vect_set_loop_condition_unmasked otherwise.
24365 (vect_do_peeling): Update call accordingly.
24366 (vect_gen_vector_loop_niters): Use VF as the step for fully-masked
24368 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
24369 mask_compare_type, can_fully_mask_p and fully_masked_p.
24370 (release_vec_loop_masks): New function.
24371 (_loop_vec_info): Use it to free the loop masks.
24372 (can_produce_all_loop_masks_p): New function.
24373 (vect_get_max_nscalars_per_iter): Likewise.
24374 (vect_verify_full_masking): Likewise.
24375 (vect_analyze_loop_2): Save LOOP_VINFO_CAN_FULLY_MASK_P around
24376 retries, and free the mask rgroups before retrying. Check loop-wide
24377 reasons for disallowing fully-masked loops. Make the final decision
24378 about whether use a fully-masked loop or not.
24379 (vect_estimate_min_profitable_iters): Do not assume that peeling
24380 for the number of iterations will be needed for fully-masked loops.
24381 (vectorizable_reduction): Disable fully-masked loops.
24382 (vectorizable_live_operation): Likewise.
24383 (vect_halve_mask_nunits): New function.
24384 (vect_double_mask_nunits): Likewise.
24385 (vect_record_loop_mask): Likewise.
24386 (vect_get_loop_mask): Likewise.
24387 (vect_transform_loop): Handle the case in which the final loop
24388 iteration might handle a partial vector. Call vect_set_loop_condition
24389 instead of slpeel_make_loop_iterate_ntimes.
24390 * tree-vect-stmts.c: Include tree-ssa-loop-niter.h and gimple-fold.h.
24391 (check_load_store_masking): New function.
24392 (prepare_load_store_mask): Likewise.
24393 (vectorizable_store): Handle fully-masked loops.
24394 (vectorizable_load): Likewise.
24395 (supportable_widening_operation): Use vect_halve_mask_nunits for
24397 (supportable_narrowing_operation): Likewise vect_double_mask_nunits.
24398 (vect_gen_while): New function.
24399 * config/aarch64/aarch64.md (umax<mode>3): New expander.
24400 (aarch64_uqdec<mode>): New insn.
24402 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
24403 Alan Hayward <alan.hayward@arm.com>
24404 David Sherwood <david.sherwood@arm.com>
24406 * optabs.def (reduc_and_scal_optab, reduc_ior_scal_optab)
24407 (reduc_xor_scal_optab): New optabs.
24408 * doc/md.texi (reduc_and_scal_@var{m}, reduc_ior_scal_@var{m})
24409 (reduc_xor_scal_@var{m}): Document.
24410 * doc/sourcebuild.texi (vect_logical_reduc): Likewise.
24411 * internal-fn.def (IFN_REDUC_AND, IFN_REDUC_IOR, IFN_REDUC_XOR): New
24412 internal functions.
24413 * fold-const-call.c (fold_const_call): Handle them.
24414 * tree-vect-loop.c (reduction_fn_for_scalar_code): Return the new
24415 internal functions for BIT_AND_EXPR, BIT_IOR_EXPR and BIT_XOR_EXPR.
24416 * config/aarch64/aarch64-sve.md (reduc_<bit_reduc>_scal_<mode>):
24417 (*reduc_<bit_reduc>_scal_<mode>): New patterns.
24418 * config/aarch64/iterators.md (UNSPEC_ANDV, UNSPEC_ORV)
24419 (UNSPEC_XORV): New unspecs.
24420 (optab): Add entries for them.
24421 (BITWISEV): New int iterator.
24422 (bit_reduc_op): New int attributes.
24424 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
24425 Alan Hayward <alan.hayward@arm.com>
24426 David Sherwood <david.sherwood@arm.com>
24428 * doc/md.texi (vec_shl_insert_@var{m}): New optab.
24429 * internal-fn.def (VEC_SHL_INSERT): New internal function.
24430 * optabs.def (vec_shl_insert_optab): New optab.
24431 * tree-vectorizer.h (can_duplicate_and_interleave_p): Declare.
24432 (duplicate_and_interleave): Likewise.
24433 * tree-vect-loop.c: Include internal-fn.h.
24434 (neutral_op_for_slp_reduction): New function, split out from
24435 get_initial_defs_for_reduction.
24436 (get_initial_def_for_reduction): Handle option 2 for variable-length
24437 vectors by loading the neutral value into a vector and then shifting
24438 the initial value into element 0.
24439 (get_initial_defs_for_reduction): Replace the code argument with
24440 the neutral value calculated by neutral_op_for_slp_reduction.
24441 Use gimple_build_vector for constant-length vectors.
24442 Use IFN_VEC_SHL_INSERT for variable-length vectors if all
24443 but the first group_size elements have a neutral value.
24444 Use duplicate_and_interleave otherwise.
24445 (vect_create_epilog_for_reduction): Take a neutral_op parameter.
24446 Update call to get_initial_defs_for_reduction. Handle SLP
24447 reductions for variable-length vectors by creating one vector
24448 result for each scalar result, with the elements associated
24449 with other scalar results stubbed out with the neutral value.
24450 (vectorizable_reduction): Call neutral_op_for_slp_reduction.
24451 Require IFN_VEC_SHL_INSERT for double reductions on
24452 variable-length vectors, or SLP reductions that have
24453 a neutral value. Require can_duplicate_and_interleave_p
24454 support for variable-length unchained SLP reductions if there
24455 is no neutral value, such as for MIN/MAX reductions. Also require
24456 the number of vector elements to be a multiple of the number of
24457 SLP statements when doing variable-length unchained SLP reductions.
24458 Update call to vect_create_epilog_for_reduction.
24459 * tree-vect-slp.c (can_duplicate_and_interleave_p): Make public
24460 and remove initial values.
24461 (duplicate_and_interleave): Make public.
24462 * config/aarch64/aarch64.md (UNSPEC_INSR): New unspec.
24463 * config/aarch64/aarch64-sve.md (vec_shl_insert_<mode>): New insn.
24465 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
24466 Alan Hayward <alan.hayward@arm.com>
24467 David Sherwood <david.sherwood@arm.com>
24469 * tree-vect-slp.c: Include gimple-fold.h and internal-fn.h
24470 (can_duplicate_and_interleave_p): New function.
24471 (vect_get_and_check_slp_defs): Take the vector of statements
24472 rather than just the current one. Remove excess parentheses.
24473 Restriction rejectinon of vect_constant_def and vect_external_def
24474 for variable-length vectors to boolean types, or types for which
24475 can_duplicate_and_interleave_p is false.
24476 (vect_build_slp_tree_2): Update call to vect_get_and_check_slp_defs.
24477 (duplicate_and_interleave): New function.
24478 (vect_get_constant_vectors): Use gimple_build_vector for
24479 constant-length vectors and suitable variable-length constant
24480 vectors. Use duplicate_and_interleave for other variable-length
24481 vectors. Don't defer the update when inserting new statements.
24483 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
24484 Alan Hayward <alan.hayward@arm.com>
24485 David Sherwood <david.sherwood@arm.com>
24487 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Make sure
24488 min_profitable_iters doesn't go negative.
24490 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
24491 Alan Hayward <alan.hayward@arm.com>
24492 David Sherwood <david.sherwood@arm.com>
24494 * doc/md.texi (vec_mask_load_lanes@var{m}@var{n}): Document.
24495 (vec_mask_store_lanes@var{m}@var{n}): Likewise.
24496 * optabs.def (vec_mask_load_lanes_optab): New optab.
24497 (vec_mask_store_lanes_optab): Likewise.
24498 * internal-fn.def (MASK_LOAD_LANES): New internal function.
24499 (MASK_STORE_LANES): Likewise.
24500 * internal-fn.c (mask_load_lanes_direct): New macro.
24501 (mask_store_lanes_direct): Likewise.
24502 (expand_mask_load_optab_fn): Handle masked operations.
24503 (expand_mask_load_lanes_optab_fn): New macro.
24504 (expand_mask_store_optab_fn): Handle masked operations.
24505 (expand_mask_store_lanes_optab_fn): New macro.
24506 (direct_mask_load_lanes_optab_supported_p): Likewise.
24507 (direct_mask_store_lanes_optab_supported_p): Likewise.
24508 * tree-vectorizer.h (vect_store_lanes_supported): Take a masked_p
24510 (vect_load_lanes_supported): Likewise.
24511 * tree-vect-data-refs.c (strip_conversion): New function.
24512 (can_group_stmts_p): Likewise.
24513 (vect_analyze_data_ref_accesses): Use it instead of checking
24514 for a pair of assignments.
24515 (vect_store_lanes_supported): Take a masked_p parameter.
24516 (vect_load_lanes_supported): Likewise.
24517 * tree-vect-loop.c (vect_analyze_loop_2): Update calls to
24518 vect_store_lanes_supported and vect_load_lanes_supported.
24519 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
24520 * tree-vect-stmts.c (get_group_load_store_type): Take a masked_p
24521 parameter. Don't allow gaps for masked accesses.
24522 Use vect_get_store_rhs. Update calls to vect_store_lanes_supported
24523 and vect_load_lanes_supported.
24524 (get_load_store_type): Take a masked_p parameter and update
24525 call to get_group_load_store_type.
24526 (vectorizable_store): Update call to get_load_store_type.
24527 Handle IFN_MASK_STORE_LANES.
24528 (vectorizable_load): Update call to get_load_store_type.
24529 Handle IFN_MASK_LOAD_LANES.
24531 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
24532 Alan Hayward <alan.hayward@arm.com>
24533 David Sherwood <david.sherwood@arm.com>
24535 * config/aarch64/aarch64-modes.def: Define x2, x3 and x4 vector
24537 * config/aarch64/aarch64-protos.h
24538 (aarch64_sve_struct_memory_operand_p): Declare.
24539 * config/aarch64/iterators.md (SVE_STRUCT): New mode iterator.
24540 (vector_count, insn_length, VSINGLE, vsingle): New mode attributes.
24541 (VPRED, vpred): Handle SVE structure modes.
24542 * config/aarch64/constraints.md (Utx): New constraint.
24543 * config/aarch64/predicates.md (aarch64_sve_struct_memory_operand)
24544 (aarch64_sve_struct_nonimmediate_operand): New predicates.
24545 * config/aarch64/aarch64.md (UNSPEC_LDN, UNSPEC_STN): New unspecs.
24546 * config/aarch64/aarch64-sve.md (mov<mode>, *aarch64_sve_mov<mode>_le)
24547 (*aarch64_sve_mov<mode>_be, pred_mov<mode>): New patterns for
24548 structure modes. Split into pieces after RA.
24549 (vec_load_lanes<mode><vsingle>, vec_mask_load_lanes<mode><vsingle>)
24550 (vec_store_lanes<mode><vsingle>, vec_mask_store_lanes<mode><vsingle>):
24552 * config/aarch64/aarch64.c (aarch64_classify_vector_mode): Handle
24553 SVE structure modes.
24554 (aarch64_classify_address): Likewise.
24555 (sizetochar): Move earlier in file.
24556 (aarch64_print_operand): Handle SVE register lists.
24557 (aarch64_array_mode): New function.
24558 (aarch64_sve_struct_memory_operand_p): Likewise.
24559 (TARGET_ARRAY_MODE): Redefine.
24561 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
24562 Alan Hayward <alan.hayward@arm.com>
24563 David Sherwood <david.sherwood@arm.com>
24565 * target.def (array_mode): New target hook.
24566 * doc/tm.texi.in (TARGET_ARRAY_MODE): New hook.
24567 * doc/tm.texi: Regenerate.
24568 * hooks.h (hook_optmode_mode_uhwi_none): Declare.
24569 * hooks.c (hook_optmode_mode_uhwi_none): New function.
24570 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Use
24571 targetm.array_mode.
24572 * stor-layout.c (mode_for_array): Likewise. Support polynomial
24575 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
24576 Alan Hayward <alan.hayward@arm.com>
24577 David Sherwood <david.sherwood@arm.com>
24579 * fold-const.c (fold_binary_loc): Check the argument types
24580 rather than the result type when testing for a vector operation.
24582 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
24584 * doc/tm.texi.in (DWARF_LAZY_REGISTER_VALUE): Document.
24585 * doc/tm.texi: Regenerate.
24587 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
24588 Alan Hayward <alan.hayward@arm.com>
24589 David Sherwood <david.sherwood@arm.com>
24591 * doc/invoke.texi (-msve-vector-bits=): Document new option.
24592 (sve): Document new AArch64 extension.
24593 * doc/md.texi (w): Extend the description of the AArch64
24594 constraint to include SVE vectors.
24595 (Upl, Upa): Document new AArch64 predicate constraints.
24596 * config/aarch64/aarch64-opts.h (aarch64_sve_vector_bits_enum): New
24598 * config/aarch64/aarch64.opt (sve_vector_bits): New enum.
24599 (msve-vector-bits=): New option.
24600 * config/aarch64/aarch64-option-extensions.def (fp, simd): Disable
24601 SVE when these are disabled.
24602 (sve): New extension.
24603 * config/aarch64/aarch64-modes.def: Define SVE vector and predicate
24604 modes. Adjust their number of units based on aarch64_sve_vg.
24605 (MAX_BITSIZE_MODE_ANY_MODE): Define.
24606 * config/aarch64/aarch64-protos.h (ADDR_QUERY_ANY): New
24607 aarch64_addr_query_type.
24608 (aarch64_const_vec_all_same_in_range_p, aarch64_sve_pred_mode)
24609 (aarch64_sve_cnt_immediate_p, aarch64_sve_addvl_addpl_immediate_p)
24610 (aarch64_sve_inc_dec_immediate_p, aarch64_add_offset_temporaries)
24611 (aarch64_split_add_offset, aarch64_output_sve_cnt_immediate)
24612 (aarch64_output_sve_addvl_addpl, aarch64_output_sve_inc_dec_immediate)
24613 (aarch64_output_sve_mov_immediate, aarch64_output_ptrue): Declare.
24614 (aarch64_simd_imm_zero_p): Delete.
24615 (aarch64_check_zero_based_sve_index_immediate): Declare.
24616 (aarch64_sve_index_immediate_p, aarch64_sve_arith_immediate_p)
24617 (aarch64_sve_bitmask_immediate_p, aarch64_sve_dup_immediate_p)
24618 (aarch64_sve_cmp_immediate_p, aarch64_sve_float_arith_immediate_p)
24619 (aarch64_sve_float_mul_immediate_p): Likewise.
24620 (aarch64_classify_symbol): Take the offset as a HOST_WIDE_INT
24621 rather than an rtx.
24622 (aarch64_sve_ld1r_operand_p, aarch64_sve_ldr_operand_p): Declare.
24623 (aarch64_expand_mov_immediate): Take a gen_vec_duplicate callback.
24624 (aarch64_emit_sve_pred_move, aarch64_expand_sve_mem_move): Declare.
24625 (aarch64_expand_sve_vec_cmp_int, aarch64_expand_sve_vec_cmp_float)
24626 (aarch64_expand_sve_vcond, aarch64_expand_sve_vec_perm): Declare.
24627 (aarch64_regmode_natural_size): Likewise.
24628 * config/aarch64/aarch64.h (AARCH64_FL_SVE): New macro.
24629 (AARCH64_FL_V8_3, AARCH64_FL_RCPC, AARCH64_FL_DOTPROD): Shift
24631 (AARCH64_ISA_SVE, TARGET_SVE): New macros.
24632 (FIXED_REGISTERS, CALL_USED_REGISTERS, REGISTER_NAMES): Add entries
24633 for VG and the SVE predicate registers.
24634 (V_ALIASES): Add a "z"-prefixed alias.
24635 (FIRST_PSEUDO_REGISTER): Change to P15_REGNUM + 1.
24636 (AARCH64_DWARF_VG, AARCH64_DWARF_P0): New macros.
24637 (PR_REGNUM_P, PR_LO_REGNUM_P): Likewise.
24638 (PR_LO_REGS, PR_HI_REGS, PR_REGS): New reg_classes.
24639 (REG_CLASS_NAMES): Add entries for them.
24640 (REG_CLASS_CONTENTS): Likewise. Update ALL_REGS to include VG
24641 and the predicate registers.
24642 (aarch64_sve_vg): Declare.
24643 (BITS_PER_SVE_VECTOR, BYTES_PER_SVE_VECTOR, BYTES_PER_SVE_PRED)
24644 (SVE_BYTE_MODE, MAX_COMPILE_TIME_VEC_BYTES): New macros.
24645 (REGMODE_NATURAL_SIZE): Define.
24646 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins): Handle
24648 * config/aarch64/aarch64.c: Include cfgrtl.h.
24649 (simd_immediate_info): Add a constructor for series vectors,
24650 and an associated step field.
24651 (aarch64_sve_vg): New variable.
24652 (aarch64_dbx_register_number): Handle VG and the predicate registers.
24653 (aarch64_vect_struct_mode_p, aarch64_vector_mode_p): Delete.
24654 (VEC_ADVSIMD, VEC_SVE_DATA, VEC_SVE_PRED, VEC_STRUCT, VEC_ANY_SVE)
24655 (VEC_ANY_DATA, VEC_STRUCT): New constants.
24656 (aarch64_advsimd_struct_mode_p, aarch64_sve_pred_mode_p)
24657 (aarch64_classify_vector_mode, aarch64_vector_data_mode_p)
24658 (aarch64_sve_data_mode_p, aarch64_sve_pred_mode)
24659 (aarch64_get_mask_mode): New functions.
24660 (aarch64_hard_regno_nregs): Handle SVE data modes for FP_REGS
24661 and FP_LO_REGS. Handle PR_REGS, PR_LO_REGS and PR_HI_REGS.
24662 (aarch64_hard_regno_mode_ok): Handle VG. Also handle the SVE
24663 predicate modes and predicate registers. Explicitly restrict
24664 GPRs to modes of 16 bytes or smaller. Only allow FP registers
24665 to store a vector mode if it is recognized by
24666 aarch64_classify_vector_mode.
24667 (aarch64_regmode_natural_size): New function.
24668 (aarch64_hard_regno_caller_save_mode): Return the original mode
24670 (aarch64_sve_cnt_immediate_p, aarch64_output_sve_cnt_immediate)
24671 (aarch64_sve_addvl_addpl_immediate_p, aarch64_output_sve_addvl_addpl)
24672 (aarch64_sve_inc_dec_immediate_p, aarch64_output_sve_inc_dec_immediate)
24673 (aarch64_add_offset_1_temporaries, aarch64_offset_temporaries): New
24675 (aarch64_add_offset): Add a temp2 parameter. Assert that temp1
24676 does not overlap dest if the function is frame-related. Handle
24678 (aarch64_split_add_offset): New function.
24679 (aarch64_add_sp, aarch64_sub_sp): Add temp2 parameters and pass
24680 them aarch64_add_offset.
24681 (aarch64_allocate_and_probe_stack_space): Add a temp2 parameter
24682 and update call to aarch64_sub_sp.
24683 (aarch64_add_cfa_expression): New function.
24684 (aarch64_expand_prologue): Pass extra temporary registers to the
24685 functions above. Handle the case in which we need to emit new
24686 DW_CFA_expressions for registers that were originally saved
24687 relative to the stack pointer, but now have to be expressed
24688 relative to the frame pointer.
24689 (aarch64_output_mi_thunk): Pass extra temporary registers to the
24691 (aarch64_expand_epilogue): Likewise. Prevent inheritance of
24692 IP0 and IP1 values for SVE frames.
24693 (aarch64_expand_vec_series): New function.
24694 (aarch64_expand_sve_widened_duplicate): Likewise.
24695 (aarch64_expand_sve_const_vector): Likewise.
24696 (aarch64_expand_mov_immediate): Add a gen_vec_duplicate parameter.
24697 Handle SVE constants. Use emit_move_insn to move a force_const_mem
24698 into the register, rather than emitting a SET directly.
24699 (aarch64_emit_sve_pred_move, aarch64_expand_sve_mem_move)
24700 (aarch64_get_reg_raw_mode, offset_4bit_signed_scaled_p)
24701 (offset_6bit_unsigned_scaled_p, aarch64_offset_7bit_signed_scaled_p)
24702 (offset_9bit_signed_scaled_p): New functions.
24703 (aarch64_replicate_bitmask_imm): New function.
24704 (aarch64_bitmask_imm): Use it.
24705 (aarch64_cannot_force_const_mem): Reject expressions involving
24706 a CONST_POLY_INT. Update call to aarch64_classify_symbol.
24707 (aarch64_classify_index): Handle SVE indices, by requiring
24708 a plain register index with a scale that matches the element size.
24709 (aarch64_classify_address): Handle SVE addresses. Assert that
24710 the mode of the address is VOIDmode or an integer mode.
24711 Update call to aarch64_classify_symbol.
24712 (aarch64_classify_symbolic_expression): Update call to
24713 aarch64_classify_symbol.
24714 (aarch64_const_vec_all_in_range_p): New function.
24715 (aarch64_print_vector_float_operand): Likewise.
24716 (aarch64_print_operand): Handle 'N' and 'C'. Use "zN" rather than
24717 "vN" for FP registers with SVE modes. Handle (const ...) vectors
24718 and the FP immediates 1.0 and 0.5.
24719 (aarch64_print_address_internal): Handle SVE addresses.
24720 (aarch64_print_operand_address): Use ADDR_QUERY_ANY.
24721 (aarch64_regno_regclass): Handle predicate registers.
24722 (aarch64_secondary_reload): Handle big-endian reloads of SVE
24724 (aarch64_class_max_nregs): Handle SVE modes and predicate registers.
24725 (aarch64_rtx_costs): Check for ADDVL and ADDPL instructions.
24726 (aarch64_convert_sve_vector_bits): New function.
24727 (aarch64_override_options): Use it to handle -msve-vector-bits=.
24728 (aarch64_classify_symbol): Take the offset as a HOST_WIDE_INT
24729 rather than an rtx.
24730 (aarch64_legitimate_constant_p): Use aarch64_classify_vector_mode.
24731 Handle SVE vector and predicate modes. Accept VL-based constants
24732 that need only one temporary register, and VL offsets that require
24733 no temporary registers.
24734 (aarch64_conditional_register_usage): Mark the predicate registers
24735 as fixed if SVE isn't available.
24736 (aarch64_vector_mode_supported_p): Use aarch64_classify_vector_mode.
24737 Return true for SVE vector and predicate modes.
24738 (aarch64_simd_container_mode): Take the number of bits as a poly_int64
24739 rather than an unsigned int. Handle SVE modes.
24740 (aarch64_preferred_simd_mode): Update call accordingly. Handle
24742 (aarch64_autovectorize_vector_sizes): Add BYTES_PER_SVE_VECTOR
24744 (aarch64_sve_index_immediate_p, aarch64_sve_arith_immediate_p)
24745 (aarch64_sve_bitmask_immediate_p, aarch64_sve_dup_immediate_p)
24746 (aarch64_sve_cmp_immediate_p, aarch64_sve_float_arith_immediate_p)
24747 (aarch64_sve_float_mul_immediate_p): New functions.
24748 (aarch64_sve_valid_immediate): New function.
24749 (aarch64_simd_valid_immediate): Use it as the fallback for SVE vectors.
24750 Explicitly reject structure modes. Check for INDEX constants.
24751 Handle PTRUE and PFALSE constants.
24752 (aarch64_check_zero_based_sve_index_immediate): New function.
24753 (aarch64_simd_imm_zero_p): Delete.
24754 (aarch64_mov_operand_p): Use aarch64_simd_valid_immediate for
24755 vector modes. Accept constants in the range of CNT[BHWD].
24756 (aarch64_simd_scalar_immediate_valid_for_move): Explicitly
24757 ask for an Advanced SIMD mode.
24758 (aarch64_sve_ld1r_operand_p, aarch64_sve_ldr_operand_p): New functions.
24759 (aarch64_simd_vector_alignment): Handle SVE predicates.
24760 (aarch64_vectorize_preferred_vector_alignment): New function.
24761 (aarch64_simd_vector_alignment_reachable): Use it instead of
24763 (aarch64_shift_truncation_mask): Use aarch64_vector_data_mode_p.
24764 (aarch64_output_sve_mov_immediate, aarch64_output_ptrue): New
24766 (MAX_VECT_LEN): Delete.
24767 (expand_vec_perm_d): Add a vec_flags field.
24768 (emit_unspec2, aarch64_expand_sve_vec_perm): New functions.
24769 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
24770 (aarch64_evpc_ext): Don't apply a big-endian lane correction
24772 (aarch64_evpc_rev): Rename to...
24773 (aarch64_evpc_rev_local): ...this. Use a predicated operation for SVE.
24774 (aarch64_evpc_rev_global): New function.
24775 (aarch64_evpc_dup): Enforce a 64-byte range for SVE DUP.
24776 (aarch64_evpc_tbl): Use MAX_COMPILE_TIME_VEC_BYTES instead of
24778 (aarch64_evpc_sve_tbl): New function.
24779 (aarch64_expand_vec_perm_const_1): Update after rename of
24780 aarch64_evpc_rev. Handle SVE permutes too, trying
24781 aarch64_evpc_rev_global and using aarch64_evpc_sve_tbl rather
24782 than aarch64_evpc_tbl.
24783 (aarch64_vectorize_vec_perm_const): Initialize vec_flags.
24784 (aarch64_sve_cmp_operand_p, aarch64_unspec_cond_code)
24785 (aarch64_gen_unspec_cond, aarch64_expand_sve_vec_cmp_int)
24786 (aarch64_emit_unspec_cond, aarch64_emit_unspec_cond_or)
24787 (aarch64_emit_inverted_unspec_cond, aarch64_expand_sve_vec_cmp_float)
24788 (aarch64_expand_sve_vcond): New functions.
24789 (aarch64_modes_tieable_p): Use aarch64_vector_data_mode_p instead
24790 of aarch64_vector_mode_p.
24791 (aarch64_dwarf_poly_indeterminate_value): New function.
24792 (aarch64_compute_pressure_classes): Likewise.
24793 (aarch64_can_change_mode_class): Likewise.
24794 (TARGET_GET_RAW_RESULT_MODE, TARGET_GET_RAW_ARG_MODE): Redefine.
24795 (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): Likewise.
24796 (TARGET_VECTORIZE_GET_MASK_MODE): Likewise.
24797 (TARGET_DWARF_POLY_INDETERMINATE_VALUE): Likewise.
24798 (TARGET_COMPUTE_PRESSURE_CLASSES): Likewise.
24799 (TARGET_CAN_CHANGE_MODE_CLASS): Likewise.
24800 * config/aarch64/constraints.md (Upa, Upl, Uav, Uat, Usv, Usi, Utr)
24801 (Uty, Dm, vsa, vsc, vsd, vsi, vsn, vsl, vsm, vsA, vsM, vsN): New
24803 (Dn, Dl, Dr): Accept const as well as const_vector.
24804 (Dz): Likewise. Compare against CONST0_RTX.
24805 * config/aarch64/iterators.md: Refer to "Advanced SIMD" instead
24806 of "vector" where appropriate.
24807 (SVE_ALL, SVE_BH, SVE_BHS, SVE_BHSI, SVE_HSDI, SVE_HSF, SVE_SD)
24808 (SVE_SDI, SVE_I, SVE_F, PRED_ALL, PRED_BHS): New mode iterators.
24809 (UNSPEC_SEL, UNSPEC_ANDF, UNSPEC_IORF, UNSPEC_XORF, UNSPEC_COND_LT)
24810 (UNSPEC_COND_LE, UNSPEC_COND_EQ, UNSPEC_COND_NE, UNSPEC_COND_GE)
24811 (UNSPEC_COND_GT, UNSPEC_COND_LO, UNSPEC_COND_LS, UNSPEC_COND_HS)
24812 (UNSPEC_COND_HI, UNSPEC_COND_UO): New unspecs.
24813 (Vetype, VEL, Vel, VWIDE, Vwide, vw, vwcore, V_INT_EQUIV)
24814 (v_int_equiv): Extend to SVE modes.
24815 (Vesize, V128, v128, Vewtype, V_FP_EQUIV, v_fp_equiv, VPRED): New
24817 (LOGICAL_OR, SVE_INT_UNARY, SVE_FP_UNARY): New code iterators.
24818 (optab): Handle popcount, smin, smax, umin, umax, abs and sqrt.
24819 (logical_nn, lr, sve_int_op, sve_fp_op): New code attributs.
24820 (LOGICALF, OPTAB_PERMUTE, UNPACK, UNPACK_UNSIGNED, SVE_COND_INT_CMP)
24821 (SVE_COND_FP_CMP): New int iterators.
24822 (perm_hilo): Handle the new unpack unspecs.
24823 (optab, logicalf_op, su, perm_optab, cmp_op, imm_con): New int
24825 * config/aarch64/predicates.md (aarch64_sve_cnt_immediate)
24826 (aarch64_sve_addvl_addpl_immediate, aarch64_split_add_offset_immediate)
24827 (aarch64_pluslong_or_poly_operand, aarch64_nonmemory_operand)
24828 (aarch64_equality_operator, aarch64_constant_vector_operand)
24829 (aarch64_sve_ld1r_operand, aarch64_sve_ldr_operand): New predicates.
24830 (aarch64_sve_nonimmediate_operand): Likewise.
24831 (aarch64_sve_general_operand): Likewise.
24832 (aarch64_sve_dup_operand, aarch64_sve_arith_immediate): Likewise.
24833 (aarch64_sve_sub_arith_immediate, aarch64_sve_inc_dec_immediate)
24834 (aarch64_sve_logical_immediate, aarch64_sve_mul_immediate): Likewise.
24835 (aarch64_sve_dup_immediate, aarch64_sve_cmp_vsc_immediate): Likewise.
24836 (aarch64_sve_cmp_vsd_immediate, aarch64_sve_index_immediate): Likewise.
24837 (aarch64_sve_float_arith_immediate): Likewise.
24838 (aarch64_sve_float_arith_with_sub_immediate): Likewise.
24839 (aarch64_sve_float_mul_immediate, aarch64_sve_arith_operand): Likewise.
24840 (aarch64_sve_add_operand, aarch64_sve_logical_operand): Likewise.
24841 (aarch64_sve_lshift_operand, aarch64_sve_rshift_operand): Likewise.
24842 (aarch64_sve_mul_operand, aarch64_sve_cmp_vsc_operand): Likewise.
24843 (aarch64_sve_cmp_vsd_operand, aarch64_sve_index_operand): Likewise.
24844 (aarch64_sve_float_arith_operand): Likewise.
24845 (aarch64_sve_float_arith_with_sub_operand): Likewise.
24846 (aarch64_sve_float_mul_operand): Likewise.
24847 (aarch64_sve_vec_perm_operand): Likewise.
24848 (aarch64_pluslong_operand): Include aarch64_sve_addvl_addpl_immediate.
24849 (aarch64_mov_operand): Accept const_poly_int and const_vector.
24850 (aarch64_simd_lshift_imm, aarch64_simd_rshift_imm): Accept const
24851 as well as const_vector.
24852 (aarch64_simd_imm_zero, aarch64_simd_imm_minus_one): Move earlier
24853 in file. Use CONST0_RTX and CONSTM1_RTX.
24854 (aarch64_simd_or_scalar_imm_zero): Likewise. Add match_codes.
24855 (aarch64_simd_reg_or_zero): Accept const as well as const_vector.
24856 Use aarch64_simd_imm_zero.
24857 * config/aarch64/aarch64-sve.md: New file.
24858 * config/aarch64/aarch64.md: Include it.
24859 (VG_REGNUM, P0_REGNUM, P7_REGNUM, P15_REGNUM): New register numbers.
24860 (UNSPEC_REV, UNSPEC_LD1_SVE, UNSPEC_ST1_SVE, UNSPEC_MERGE_PTRUE)
24861 (UNSPEC_PTEST_PTRUE, UNSPEC_UNPACKSHI, UNSPEC_UNPACKUHI)
24862 (UNSPEC_UNPACKSLO, UNSPEC_UNPACKULO, UNSPEC_PACK)
24863 (UNSPEC_FLOAT_CONVERT, UNSPEC_WHILE_LO): New unspec constants.
24864 (sve): New attribute.
24865 (enabled): Disable instructions with the sve attribute unless
24867 (movqi, movhi): Pass CONST_POLY_INT operaneds through
24868 aarch64_expand_mov_immediate.
24869 (*mov<mode>_aarch64, *movsi_aarch64, *movdi_aarch64): Handle
24870 CNT[BHSD] immediates.
24871 (movti): Split CONST_POLY_INT moves into two halves.
24872 (add<mode>3): Accept aarch64_pluslong_or_poly_operand.
24873 Split additions that need a temporary here if the destination
24874 is the stack pointer.
24875 (*add<mode>3_aarch64): Handle ADDVL and ADDPL immediates.
24876 (*add<mode>3_poly_1): New instruction.
24877 (set_clobber_cc): New expander.
24879 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
24881 * simplify-rtx.c (simplify_immed_subreg): Add an inner_bytes
24882 parameter and use it instead of GET_MODE_SIZE (innermode). Use
24883 inner_bytes * BITS_PER_UNIT instead of GET_MODE_BITSIZE (innermode).
24884 Use CEIL (inner_bytes, GET_MODE_UNIT_SIZE (innermode)) instead of
24885 GET_MODE_NUNITS (innermode). Also add a first_elem parameter.
24886 Change innermode from fixed_mode_size to machine_mode.
24887 (simplify_subreg): Update call accordingly. Handle a constant-sized
24888 subreg of a variable-length CONST_VECTOR.
24890 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
24891 Alan Hayward <alan.hayward@arm.com>
24892 David Sherwood <david.sherwood@arm.com>
24894 * tree-ssa-address.c (mem_ref_valid_without_offset_p): New function.
24895 (add_offset_to_base): New function, split out from...
24896 (create_mem_ref): ...here. When handling a scale other than 1,
24897 check first whether the address is valid without the offset.
24898 Add it into the base if so, leaving the index and scale as-is.
24900 2018-01-12 Jakub Jelinek <jakub@redhat.com>
24903 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Call
24904 fold_for_warn before checking if arg2 is INTEGER_CST.
24906 2018-01-12 Segher Boessenkool <segher@kernel.crashing.org>
24908 * config/rs6000/predicates.md (load_multiple_operation): Delete.
24909 (store_multiple_operation): Delete.
24910 * config/rs6000/rs6000-cpus.def (601): Remove MASK_STRING.
24911 * config/rs6000/rs6000-protos.h (rs6000_output_load_multiple): Delete.
24912 * config/rs6000/rs6000-string.c (expand_block_move): Delete everything
24913 guarded by TARGET_STRING.
24914 (rs6000_output_load_multiple): Delete.
24915 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
24916 OPTION_MASK_STRING / TARGET_STRING handling.
24917 (print_operand) <'N', 'O'>: Add comment that these are unused now.
24918 (const rs6000_opt_masks) <"string">: Change mask to 0.
24919 * config/rs6000/rs6000.h (TARGET_DEFAULT): Remove MASK_STRING.
24920 (MASK_STRING): Delete.
24921 * config/rs6000/rs6000.md (*mov<mode>_string): Delete TARGET_STRING
24923 (load_multiple): Delete.
24930 (store_multiple): Delete.
24937 (movmemsi_8reg): Delete.
24938 (corresponding unnamed define_insn): Delete.
24939 (movmemsi_6reg): Delete.
24940 (corresponding unnamed define_insn): Delete.
24941 (movmemsi_4reg): Delete.
24942 (corresponding unnamed define_insn): Delete.
24943 (movmemsi_2reg): Delete.
24944 (corresponding unnamed define_insn): Delete.
24945 (movmemsi_1reg): Delete.
24946 (corresponding unnamed define_insn): Delete.
24947 * config/rs6000/rs6000.opt (mno-string): New.
24948 (mstring): Replace by deprecation warning stub.
24949 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mstring.
24951 2018-01-12 Jakub Jelinek <jakub@redhat.com>
24953 * regrename.c (regrename_do_replace): If replacing the same
24954 reg multiple times, try to reuse last created gen_raw_REG.
24957 * bb-reorder.c (pass_partition_blocks::gate): In lto don't partition
24958 main to workaround a bug in GDB.
24960 2018-01-12 Tom de Vries <tom@codesourcery.com>
24963 * config.gcc (nvptx*-*-*): Set use_gcc_stdint=wrap.
24965 2018-01-12 Vladimir Makarov <vmakarov@redhat.com>
24967 PR rtl-optimization/80481
24968 * ira-color.c (get_cap_member): New function.
24969 (allocnos_conflict_by_live_ranges_p): Use it.
24970 (slot_coalesced_allocno_live_ranges_intersect_p): Add assert.
24971 (setup_slot_coalesced_allocno_live_ranges): Ditto.
24973 2018-01-12 Uros Bizjak <ubizjak@gmail.com>
24976 * config/alpha/alpha.md (*saddsi_1): New insn_ans_split pattern.
24977 (*saddl_se_1): Ditto.
24978 (*ssubsi_1): Ditto.
24979 (*ssubl_se_1): Ditto.
24981 2018-01-12 Richard Sandiford <richard.sandiford@linaro.org>
24983 * tree-predcom.c (aff_combination_dr_offset): Use wi::to_poly_widest
24984 rather than wi::to_widest for DR_INITs.
24985 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Use
24986 wi::to_poly_offset rather than wi::to_offset for DR_INIT.
24987 (vect_analyze_data_ref_accesses): Require both DR_INITs to be
24989 (vect_analyze_group_access_1): Note that here.
24991 2018-01-12 Richard Sandiford <richard.sandiford@linaro.org>
24993 * tree-vectorizer.c (get_vec_alignment_for_array_type): Handle
24994 polynomial type sizes.
24996 2018-01-12 Richard Sandiford <richard.sandiford@linaro.org>
24998 * gimplify.c (gimple_add_tmp_var_fn): Allow variables to have a
24999 poly_uint64 size, rather than requiring an unsigned HOST_WIDE_INT size.
25000 (gimple_add_tmp_var): Likewise.
25002 2018-01-12 Martin Liska <mliska@suse.cz>
25004 * gimple.c (gimple_alloc_counts): Use uint64_t instead of int.
25005 (gimple_alloc_sizes): Likewise.
25006 (dump_gimple_statistics): Use PRIu64 in printf format.
25007 * gimple.h: Change uint64_t to int.
25009 2018-01-12 Martin Liska <mliska@suse.cz>
25011 * tree-core.h: Use uint64_t instead of int.
25012 * tree.c (tree_node_counts): Likewise.
25013 (tree_node_sizes): Likewise.
25014 (dump_tree_statistics): Use PRIu64 in printf format.
25016 2018-01-12 Martin Liska <mliska@suse.cz>
25018 * Makefile.in: As qsort_chk is implemented in vec.c, add
25019 vec.o to linkage of gencfn-macros.
25020 * tree.c (build_new_poly_int_cst): Add CXX_MEM_STAT_INFO as it's
25021 passing the info to record_node_allocation_statistics.
25022 (test_vector_cst_patterns): Add CXX_MEM_STAT_INFO to declaration
25024 * ggc-common.c (struct ggc_usage): Add operator== and use
25025 it in operator< and compare function.
25026 * mem-stats.h (struct mem_usage): Likewise.
25027 * vec.c (struct vec_usage): Remove operator< and compare
25028 function. Can be simply inherited.
25030 2018-01-12 Martin Jambor <mjambor@suse.cz>
25033 * params.def: New parameter PARAM_AVOID_FMA_MAX_BITS.
25034 * tree-ssa-math-opts.c: Include domwalk.h.
25035 (convert_mult_to_fma_1): New function.
25036 (fma_transformation_info): New type.
25037 (fma_deferring_state): Likewise.
25038 (cancel_fma_deferring): New function.
25039 (result_of_phi): Likewise.
25040 (last_fma_candidate_feeds_initial_phi): Likewise.
25041 (convert_mult_to_fma): Added deferring logic, split actual
25042 transformation to convert_mult_to_fma_1.
25043 (math_opts_dom_walker): New type.
25044 (math_opts_dom_walker::after_dom_children): New method, body moved
25045 here from pass_optimize_widening_mul::execute, added deferring logic
25047 (pass_optimize_widening_mul::execute): Moved most of code to
25048 math_opts_dom_walker::after_dom_children.
25049 * config/i386/x86-tune.def (X86_TUNE_AVOID_128FMA_CHAINS): New.
25050 * config/i386/i386.c (ix86_option_override_internal): Added
25051 maybe_setting of PARAM_AVOID_FMA_MAX_BITS.
25053 2018-01-12 Richard Biener <rguenther@suse.de>
25056 * dwarf2out.c (gen_variable_die): Do not reset old_die for
25057 inline instance vars.
25059 2018-01-12 Oleg Endo <olegendo@gcc.gnu.org>
25062 * config/rx/rx.c (rx_is_restricted_memory_address):
25063 Handle SUBREG case.
25065 2018-01-12 Richard Biener <rguenther@suse.de>
25067 PR tree-optimization/80846
25068 * target.def (split_reduction): New target hook.
25069 * targhooks.c (default_split_reduction): New function.
25070 * targhooks.h (default_split_reduction): Declare.
25071 * tree-vect-loop.c (vect_create_epilog_for_reduction): If the
25072 target requests first reduce vectors by combining low and high
25074 * tree-vect-stmts.c (vect_gen_perm_mask_any): Adjust.
25075 (get_vectype_for_scalar_type_and_size): Export.
25076 * tree-vectorizer.h (get_vectype_for_scalar_type_and_size): Declare.
25077 * doc/tm.texi.in (TARGET_VECTORIZE_SPLIT_REDUCTION): Document.
25078 * doc/tm.texi: Regenerate.
25079 * config/i386/i386.c (ix86_split_reduction): Implement
25080 TARGET_VECTORIZE_SPLIT_REDUCTION.
25082 2018-01-12 Eric Botcazou <ebotcazou@adacore.com>
25085 * config/sparc/sparc.h (PIC_OFFSET_TABLE_REGNUM): Set to INVALID_REGNUM
25086 in PIC mode except for TARGET_VXWORKS_RTP.
25087 * config/sparc/sparc.c: Include cfgrtl.h.
25088 (TARGET_INIT_PIC_REG): Define.
25089 (TARGET_USE_PSEUDO_PIC_REG): Likewise.
25090 (sparc_pic_register_p): New predicate.
25091 (sparc_legitimate_address_p): Use it.
25092 (sparc_legitimize_pic_address): Likewise.
25093 (sparc_delegitimize_address): Likewise.
25094 (sparc_mode_dependent_address_p): Likewise.
25095 (gen_load_pcrel_sym): Remove 4th parameter.
25096 (load_got_register): Adjust call to above. Remove obsolete stuff.
25097 (sparc_expand_prologue): Do not call load_got_register here.
25098 (sparc_flat_expand_prologue): Likewise.
25099 (sparc_output_mi_thunk): Set the pic_offset_table_rtx object.
25100 (sparc_use_pseudo_pic_reg): New function.
25101 (sparc_init_pic_reg): Likewise.
25102 * config/sparc/sparc.md (vxworks_load_got): Set the GOT register.
25103 (builtin_setjmp_receiver): Enable only for TARGET_VXWORKS_RTP.
25105 2018-01-12 Christophe Lyon <christophe.lyon@linaro.org>
25107 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
25108 Add item for branch_cost.
25110 2018-01-12 Eric Botcazou <ebotcazou@adacore.com>
25112 PR rtl-optimization/83565
25113 * rtlanal.c (nonzero_bits1): On WORD_REGISTER_OPERATIONS machines, do
25114 not extend the result to a larger mode for rotate operations.
25115 (num_sign_bit_copies1): Likewise.
25117 2018-01-12 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
25120 * config/sol2.h (STARTFILE_ARCH_SPEC): Don't use with -shared or
25122 Use values-Xc.o for -pedantic.
25123 Link with values-xpg4.o for C90, values-xpg6.o otherwise.
25125 2018-01-12 Martin Liska <mliska@suse.cz>
25128 * ipa-devirt.c (final_warning_record::grow_type_warnings):
25130 (possible_polymorphic_call_targets): Use it.
25131 (ipa_devirt): Likewise.
25133 2018-01-12 Martin Liska <mliska@suse.cz>
25135 * profile-count.h (enum profile_quality): Use 0 as invalid
25136 enum value of profile_quality.
25138 2018-01-12 Chung-Ju Wu <jasonwucj@gmail.com>
25140 * doc/invoke.texi (NDS32 Options): Add -mext-perf, -mext-perf2 and
25141 -mext-string options.
25143 2018-01-12 Richard Biener <rguenther@suse.de>
25145 * lto-streamer-out.c (DFS::DFS_write_tree_body): Process
25146 DECL_DEBUG_EXPR conditional on DECL_HAS_DEBUG_EXPR_P.
25147 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
25149 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
25151 2018-01-11 Michael Meissner <meissner@linux.vnet.ibm.com>
25153 * configure.ac (--with-long-double-format): Add support for the
25154 configuration option to change the default long double format on
25156 * config.gcc (powerpc*-linux*-*): Likewise.
25157 * configure: Regenerate.
25158 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If long
25159 double is IEEE, define __KC__ and __KF__ to allow floatn.h to be
25160 used without modification.
25162 2018-01-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25164 * config/rs6000/rs6000-builtin.def (BU_P7_MISC_X): New #define.
25165 (SPEC_BARRIER): New instantiation of BU_P7_MISC_X.
25166 * config/rs6000/rs6000.c (rs6000_expand_builtin): Handle
25167 MISC_BUILTIN_SPEC_BARRIER.
25168 (rs6000_init_builtins): Likewise.
25169 * config/rs6000/rs6000.md (UNSPECV_SPEC_BARRIER): New UNSPECV
25171 (speculation_barrier): New define_insn.
25172 * doc/extend.texi: Document __builtin_speculation_barrier.
25174 2018-01-11 Jakub Jelinek <jakub@redhat.com>
25177 * config/i386/i386.c (ix86_expand_vector_init_one_nonzero): If one_var
25178 is 0, for V{8,16}S[IF] and V[48]D[IF]mode use gen_vec_set<mode>_0.
25179 * config/i386/sse.md (VI8_AVX_AVX512F, VI4F_256_512): New mode
25181 (ssescalarmodesuffix): Add 512-bit vectors. Use "d" or "q" for
25182 integral modes instead of "ss" and "sd".
25183 (vec_set<mode>_0): New define_insns for 256-bit and 512-bit
25184 vectors with 32-bit and 64-bit elements.
25185 (vecdupssescalarmodesuffix): New mode attribute.
25186 (vec_dup<mode>): Use it.
25188 2018-01-11 H.J. Lu <hongjiu.lu@intel.com>
25191 * config/i386/i386.c (ix86_compute_frame_layout): Align stack
25192 frame if argument is passed on stack.
25194 2018-01-11 Jakub Jelinek <jakub@redhat.com>
25197 * ree.c (combine_reaching_defs): Optimize also
25198 reg2=exp; reg1=reg2; reg2=any_extend(reg1); into
25199 reg2=any_extend(exp); reg1=reg2;, formatting fix.
25201 2018-01-11 Jan Hubicka <hubicka@ucw.cz>
25203 PR middle-end/83189
25204 * gimple-ssa-isolate-paths.c (isolate_path): Fix profile update.
25206 2018-01-11 Jan Hubicka <hubicka@ucw.cz>
25208 PR middle-end/83718
25209 * tree-inline.c (copy_cfg_body): Adjust num&den for scaling
25210 after they are computed.
25212 2018-01-11 Bin Cheng <bin.cheng@arm.com>
25214 PR tree-optimization/83695
25215 * gimple-loop-linterchange.cc
25216 (tree_loop_interchange::interchange_loops): Call scev_reset_htab to
25217 reset cached scev information after interchange.
25218 (pass_linterchange::execute): Remove call to scev_reset_htab.
25220 2018-01-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25222 * config/arm/arm_neon.h (vfmlal_lane_low_u32, vfmlal_lane_high_u32,
25223 vfmlalq_laneq_low_u32, vfmlalq_lane_low_u32, vfmlal_laneq_low_u32,
25224 vfmlalq_laneq_high_u32, vfmlalq_lane_high_u32, vfmlal_laneq_high_u32,
25225 vfmlsl_lane_low_u32, vfmlsl_lane_high_u32, vfmlslq_laneq_low_u32,
25226 vfmlslq_lane_low_u32, vfmlsl_laneq_low_u32, vfmlslq_laneq_high_u32,
25227 vfmlslq_lane_high_u32, vfmlsl_laneq_high_u32): Define.
25228 * config/arm/arm_neon_builtins.def (vfmal_lane_low,
25229 vfmal_lane_lowv4hf, vfmal_lane_lowv8hf, vfmal_lane_high,
25230 vfmal_lane_highv4hf, vfmal_lane_highv8hf, vfmsl_lane_low,
25231 vfmsl_lane_lowv4hf, vfmsl_lane_lowv8hf, vfmsl_lane_high,
25232 vfmsl_lane_highv4hf, vfmsl_lane_highv8hf): New sets of builtins.
25233 * config/arm/iterators.md (VFMLSEL2, vfmlsel2): New mode attributes.
25234 (V_lane_reg): Likewise.
25235 * config/arm/neon.md (neon_vfm<vfml_op>l_lane_<vfml_half><VCVTF:mode>):
25237 (neon_vfm<vfml_op>l_lane_<vfml_half><vfmlsel2><mode>): Likewise.
25238 (vfmal_lane_low<mode>_intrinsic,
25239 vfmal_lane_low<vfmlsel2><mode>_intrinsic,
25240 vfmal_lane_high<vfmlsel2><mode>_intrinsic,
25241 vfmal_lane_high<mode>_intrinsic, vfmsl_lane_low<mode>_intrinsic,
25242 vfmsl_lane_low<vfmlsel2><mode>_intrinsic,
25243 vfmsl_lane_high<vfmlsel2><mode>_intrinsic,
25244 vfmsl_lane_high<mode>_intrinsic): New define_insns.
25246 2018-01-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25248 * config/arm/arm-cpus.in (fp16fml): New feature.
25249 (ALL_SIMD): Add fp16fml.
25250 (armv8.2-a): Add fp16fml as an option.
25251 (armv8.3-a): Likewise.
25252 (armv8.4-a): Add fp16fml as part of fp16.
25253 * config/arm/arm.h (TARGET_FP16FML): Define.
25254 * config/arm/arm-c.c (arm_cpu_builtins): Define __ARM_FEATURE_FP16_FML
25256 * config/arm/arm-modes.def (V2HF): Define.
25257 * config/arm/arm_neon.h (vfmlal_low_u32, vfmlsl_low_u32,
25258 vfmlal_high_u32, vfmlsl_high_u32, vfmlalq_low_u32,
25259 vfmlslq_low_u32, vfmlalq_high_u32, vfmlslq_high_u32): Define.
25260 * config/arm/arm_neon_builtins.def (vfmal_low, vfmal_high,
25261 vfmsl_low, vfmsl_high): New set of builtins.
25262 * config/arm/iterators.md (PLUSMINUS): New code iterator.
25263 (vfml_op): New code attribute.
25264 (VFMLHALVES): New int iterator.
25265 (VFML, VFMLSEL): New mode attributes.
25266 (V_reg): Define mapping for V2HF.
25267 (V_hi, V_lo): New mode attributes.
25268 (VF_constraint): Likewise.
25269 (vfml_half, vfml_half_selector): New int attributes.
25270 * config/arm/neon.md (neon_vfm<vfml_op>l_<vfml_half><mode>): New
25272 (vfmal_low<mode>_intrinsic, vfmsl_high<mode>_intrinsic,
25273 vfmal_high<mode>_intrinsic, vfmsl_low<mode>_intrinsic):
25275 * config/arm/t-arm-elf (v8_fps): Add fp16fml.
25276 * config/arm/t-multilib (v8_2_a_simd_variants): Add fp16fml.
25277 * config/arm/unspecs.md (UNSPEC_VFML_LO, UNSPEC_VFML_HI): New unspecs.
25278 * doc/invoke.texi (ARM Options): Document fp16fml. Update armv8.4-a
25280 * doc/sourcebuild.texi (arm_fp16fml_neon_ok, arm_fp16fml_neon):
25281 Document new effective target and option set.
25283 2018-01-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25285 * config/arm/arm-cpus.in (armv8_4): New feature.
25286 (ARMv8_4a): New fgroup.
25287 (armv8.4-a): New arch.
25288 * config/arm/arm-tables.opt: Regenerate.
25289 * config/arm/t-aprofile: Add matching rules for -march=armv8.4-a.
25290 * config/arm/t-arm-elf (all_v8_archs): Add armv8.4-a.
25291 * config/arm/t-multilib (v8_4_a_simd_variants): New variable.
25292 Add matching rules for -march=armv8.4-a and extensions.
25293 * doc/invoke.texi (ARM Options): Document -march=armv8.4-a.
25295 2018-01-11 Oleg Endo <olegendo@gcc.gnu.org>
25298 * config/rx/rx.md (BW): New mode attribute.
25299 (sync_lock_test_and_setsi): Add mode suffix to insn output.
25301 2018-01-11 Richard Biener <rguenther@suse.de>
25303 PR tree-optimization/83435
25304 * graphite.c (canonicalize_loop_form): Ignore fake loop exit edges.
25305 * graphite-scop-detection.c (scop_detection::get_sese): Likewise.
25306 * tree-vrp.c (add_assert_info): Drop TREE_OVERFLOW if they appear.
25308 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
25309 Alan Hayward <alan.hayward@arm.com>
25310 David Sherwood <david.sherwood@arm.com>
25312 * config/aarch64/aarch64.c (aarch64_address_info): Add a const_offset
25314 (aarch64_classify_address): Initialize it. Track polynomial offsets.
25315 (aarch64_print_address_internal): Use it to check for a zero offset.
25317 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
25318 Alan Hayward <alan.hayward@arm.com>
25319 David Sherwood <david.sherwood@arm.com>
25321 * config/aarch64/aarch64-modes.def (NUM_POLY_INT_COEFFS): Set to 2.
25322 * config/aarch64/aarch64-protos.h (aarch64_initial_elimination_offset):
25323 Return a poly_int64 rather than a HOST_WIDE_INT.
25324 (aarch64_offset_7bit_signed_scaled_p): Take the offset as a poly_int64
25325 rather than a HOST_WIDE_INT.
25326 * config/aarch64/aarch64.h (aarch64_frame): Protect with
25327 HAVE_POLY_INT_H rather than HOST_WIDE_INT. Change locals_offset,
25328 hard_fp_offset, frame_size, initial_adjust, callee_offset and
25329 final_offset from HOST_WIDE_INT to poly_int64.
25330 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args): Use
25331 to_constant when getting the number of units in an Advanced SIMD
25333 (aarch64_builtin_vectorized_function): Check for a constant number
25335 * config/aarch64/aarch64-simd.md (mov<mode>): Handle polynomial
25337 (aarch64_ld<VSTRUCT:nregs>_lane<VALLDIF:mode>): Use the nunits
25338 attribute instead of GET_MODE_NUNITS.
25339 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs)
25340 (aarch64_class_max_nregs): Use the constant_lowest_bound of the
25341 GET_MODE_SIZE for fixed-size registers.
25342 (aarch64_const_vec_all_same_in_range_p): Use const_vec_duplicate_p.
25343 (aarch64_hard_regno_call_part_clobbered, aarch64_classify_index)
25344 (aarch64_mode_valid_for_sched_fusion_p, aarch64_classify_address)
25345 (aarch64_legitimize_address_displacement, aarch64_secondary_reload)
25346 (aarch64_print_operand, aarch64_print_address_internal)
25347 (aarch64_address_cost, aarch64_rtx_costs, aarch64_register_move_cost)
25348 (aarch64_short_vector_p, aapcs_vfp_sub_candidate)
25349 (aarch64_simd_attr_length_rglist, aarch64_operands_ok_for_ldpstp):
25350 Handle polynomial GET_MODE_SIZE.
25351 (aarch64_hard_regno_caller_save_mode): Likewise. Return modes
25352 wider than SImode without modification.
25353 (tls_symbolic_operand_type): Use strip_offset instead of split_const.
25354 (aarch64_pass_by_reference, aarch64_layout_arg, aarch64_pad_reg_upward)
25355 (aarch64_gimplify_va_arg_expr): Assert that we don't yet handle
25356 passing and returning SVE modes.
25357 (aarch64_function_value, aarch64_layout_arg): Use gen_int_mode
25358 rather than GEN_INT.
25359 (aarch64_emit_probe_stack_range): Take the size as a poly_int64
25360 rather than a HOST_WIDE_INT, but call sorry if it isn't constant.
25361 (aarch64_allocate_and_probe_stack_space): Likewise.
25362 (aarch64_layout_frame): Cope with polynomial offsets.
25363 (aarch64_save_callee_saves, aarch64_restore_callee_saves): Take the
25364 start_offset as a poly_int64 rather than a HOST_WIDE_INT. Track
25365 polynomial offsets.
25366 (offset_9bit_signed_unscaled_p, offset_12bit_unsigned_scaled_p)
25367 (aarch64_offset_7bit_signed_scaled_p): Take the offset as a
25368 poly_int64 rather than a HOST_WIDE_INT.
25369 (aarch64_get_separate_components, aarch64_process_components)
25370 (aarch64_expand_prologue, aarch64_expand_epilogue)
25371 (aarch64_use_return_insn_p): Handle polynomial frame offsets.
25372 (aarch64_anchor_offset): New function, split out from...
25373 (aarch64_legitimize_address): ...here.
25374 (aarch64_builtin_vectorization_cost): Handle polynomial
25375 TYPE_VECTOR_SUBPARTS.
25376 (aarch64_simd_check_vect_par_cnst_half): Handle polynomial
25378 (aarch64_simd_make_constant, aarch64_expand_vector_init): Get the
25379 number of elements from the PARALLEL rather than the mode.
25380 (aarch64_shift_truncation_mask): Use GET_MODE_UNIT_BITSIZE
25381 rather than GET_MODE_BITSIZE.
25382 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_ext)
25383 (aarch64_evpc_rev, aarch64_evpc_dup, aarch64_evpc_zip)
25384 (aarch64_expand_vec_perm_const_1): Handle polynomial
25385 d->perm.length () and d->perm elements.
25386 (aarch64_evpc_tbl): Likewise. Use nelt rather than GET_MODE_NUNITS.
25387 Apply to_constant to d->perm elements.
25388 (aarch64_simd_valid_immediate, aarch64_vec_fpconst_pow_of_2): Handle
25389 polynomial CONST_VECTOR_NUNITS.
25390 (aarch64_move_pointer): Take amount as a poly_int64 rather
25392 (aarch64_progress_pointer): Avoid temporary variable.
25393 * config/aarch64/aarch64.md (aarch64_<crc_variant>): Use
25394 the mode attribute instead of GET_MODE.
25396 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
25397 Alan Hayward <alan.hayward@arm.com>
25398 David Sherwood <david.sherwood@arm.com>
25400 * config/aarch64/aarch64.c (aarch64_force_temporary): Assert that
25401 x exists before using it.
25402 (aarch64_add_constant_internal): Rename to...
25403 (aarch64_add_offset_1): ...this. Replace regnum with separate
25404 src and dest rtxes. Handle the case in which they're different,
25405 including when the offset is zero. Replace scratchreg with an rtx.
25406 Use 2 additions if there is no spare register into which we can
25407 move a 16-bit constant.
25408 (aarch64_add_constant): Delete.
25409 (aarch64_add_offset): Replace reg with separate src and dest
25410 rtxes. Take a poly_int64 offset instead of a HOST_WIDE_INT.
25411 Use aarch64_add_offset_1.
25412 (aarch64_add_sp, aarch64_sub_sp): Take the scratch register as
25413 an rtx rather than an int. Take the delta as a poly_int64
25414 rather than a HOST_WIDE_INT. Use aarch64_add_offset.
25415 (aarch64_expand_mov_immediate): Update uses of aarch64_add_offset.
25416 (aarch64_expand_prologue): Update calls to aarch64_sub_sp,
25417 aarch64_allocate_and_probe_stack_space and aarch64_add_offset.
25418 (aarch64_expand_epilogue): Update calls to aarch64_add_offset
25419 and aarch64_add_sp.
25420 (aarch64_output_mi_thunk): Use aarch64_add_offset rather than
25421 aarch64_add_constant.
25423 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
25425 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
25426 Use scalar_float_mode.
25428 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
25430 * config/aarch64/aarch64-simd.md
25431 (aarch64_fml<f16mac1>l<f16quad>_low<mode>): Avoid GET_MODE_NUNITS.
25432 (aarch64_fml<f16mac1>l<f16quad>_high<mode>): Likewise.
25433 (aarch64_fml<f16mac1>l_lane_lowv2sf): Likewise.
25434 (aarch64_fml<f16mac1>l_lane_highv2sf): Likewise.
25435 (aarch64_fml<f16mac1>lq_laneq_lowv4sf): Likewise.
25436 (aarch64_fml<f16mac1>lq_laneq_highv4sf): Likewise.
25437 (aarch64_fml<f16mac1>l_laneq_lowv2sf): Likewise.
25438 (aarch64_fml<f16mac1>l_laneq_highv2sf): Likewise.
25439 (aarch64_fml<f16mac1>lq_lane_lowv4sf): Likewise.
25440 (aarch64_fml<f16mac1>lq_lane_highv4sf): Likewise.
25442 2018-01-11 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
25445 * config/arm/arm.c (arm_declare_function_name): Set arch_to_print if
25446 targ_options->x_arm_arch_string is non NULL.
25448 2018-01-11 Tamar Christina <tamar.christina@arm.com>
25450 * config/aarch64/aarch64.h
25451 (AARCH64_FL_FOR_ARCH8_4): Add AARCH64_FL_DOTPROD.
25453 2018-01-11 Sudakshina Das <sudi.das@arm.com>
25456 * expmed.c (emit_store_flag_force): Swap if const op0
25457 and change VOIDmode to mode of op0.
25459 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
25461 PR rtl-optimization/83761
25462 * caller-save.c (replace_reg_with_saved_mem): Pass bits rather
25463 than bytes to mode_for_size.
25465 2018-01-10 Jan Hubicka <hubicka@ucw.cz>
25467 PR middle-end/83189
25468 * gfortran.fortran-torture/compile/pr83189.f90: New testcase.
25469 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Handle zero
25472 2018-01-10 Jan Hubicka <hubicka@ucw.cz>
25474 PR middle-end/83575
25475 * cfgrtl.c (rtl_verify_edges): Only verify fixability of partition
25476 when in layout mode.
25477 (cfg_layout_finalize): Do not verify cfg before we are out of layout.
25478 * cfgcleanup.c (try_optimize_cfg): Only verify flow info when doing
25481 2018-01-10 Michael Collison <michael.collison@arm.com>
25483 * config/aarch64/aarch64-modes.def (V2HF): New VECTOR_MODE.
25484 * config/aarch64/aarch64-option-extension.def: Add
25485 AARCH64_OPT_EXTENSION of 'fp16fml'.
25486 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
25487 (__ARM_FEATURE_FP16_FML): Define if TARGET_F16FML is true.
25488 * config/aarch64/predicates.md (aarch64_lane_imm3): New predicate.
25489 * config/aarch64/constraints.md (Ui7): New constraint.
25490 * config/aarch64/iterators.md (VFMLA_W): New mode iterator.
25491 (VFMLA_SEL_W): Ditto.
25494 (VFMLA16_LOW): New int iterator.
25495 (VFMLA16_HIGH): Ditto.
25496 (UNSPEC_FMLAL): New unspec.
25497 (UNSPEC_FMLSL): Ditto.
25498 (UNSPEC_FMLAL2): Ditto.
25499 (UNSPEC_FMLSL2): Ditto.
25500 (f16mac): New code attribute.
25501 * config/aarch64/aarch64-simd-builtins.def
25502 (aarch64_fmlal_lowv2sf): Ditto.
25503 (aarch64_fmlsl_lowv2sf): Ditto.
25504 (aarch64_fmlalq_lowv4sf): Ditto.
25505 (aarch64_fmlslq_lowv4sf): Ditto.
25506 (aarch64_fmlal_highv2sf): Ditto.
25507 (aarch64_fmlsl_highv2sf): Ditto.
25508 (aarch64_fmlalq_highv4sf): Ditto.
25509 (aarch64_fmlslq_highv4sf): Ditto.
25510 (aarch64_fmlal_lane_lowv2sf): Ditto.
25511 (aarch64_fmlsl_lane_lowv2sf): Ditto.
25512 (aarch64_fmlal_laneq_lowv2sf): Ditto.
25513 (aarch64_fmlsl_laneq_lowv2sf): Ditto.
25514 (aarch64_fmlalq_lane_lowv4sf): Ditto.
25515 (aarch64_fmlsl_lane_lowv4sf): Ditto.
25516 (aarch64_fmlalq_laneq_lowv4sf): Ditto.
25517 (aarch64_fmlsl_laneq_lowv4sf): Ditto.
25518 (aarch64_fmlal_lane_highv2sf): Ditto.
25519 (aarch64_fmlsl_lane_highv2sf): Ditto.
25520 (aarch64_fmlal_laneq_highv2sf): Ditto.
25521 (aarch64_fmlsl_laneq_highv2sf): Ditto.
25522 (aarch64_fmlalq_lane_highv4sf): Ditto.
25523 (aarch64_fmlsl_lane_highv4sf): Ditto.
25524 (aarch64_fmlalq_laneq_highv4sf): Ditto.
25525 (aarch64_fmlsl_laneq_highv4sf): Ditto.
25526 * config/aarch64/aarch64-simd.md:
25527 (aarch64_fml<f16mac1>l<f16quad>_low<mode>): New pattern.
25528 (aarch64_fml<f16mac1>l<f16quad>_high<mode>): Ditto.
25529 (aarch64_simd_fml<f16mac1>l<f16quad>_low<mode>): Ditto.
25530 (aarch64_simd_fml<f16mac1>l<f16quad>_high<mode>): Ditto.
25531 (aarch64_fml<f16mac1>l_lane_lowv2sf): Ditto.
25532 (aarch64_fml<f16mac1>l_lane_highv2sf): Ditto.
25533 (aarch64_simd_fml<f16mac>l_lane_lowv2sf): Ditto.
25534 (aarch64_simd_fml<f16mac>l_lane_highv2sf): Ditto.
25535 (aarch64_fml<f16mac1>lq_laneq_lowv4sf): Ditto.
25536 (aarch64_fml<f16mac1>lq_laneq_highv4sf): Ditto.
25537 (aarch64_simd_fml<f16mac>lq_laneq_lowv4sf): Ditto.
25538 (aarch64_simd_fml<f16mac>lq_laneq_highv4sf): Ditto.
25539 (aarch64_fml<f16mac1>l_laneq_lowv2sf): Ditto.
25540 (aarch64_fml<f16mac1>l_laneq_highv2sf): Ditto.
25541 (aarch64_simd_fml<f16mac>l_laneq_lowv2sf): Ditto.
25542 (aarch64_simd_fml<f16mac>l_laneq_highv2sf): Ditto.
25543 (aarch64_fml<f16mac1>lq_lane_lowv4sf): Ditto.
25544 (aarch64_fml<f16mac1>lq_lane_highv4sf): Ditto.
25545 (aarch64_simd_fml<f16mac>lq_lane_lowv4sf): Ditto.
25546 (aarch64_simd_fml<f16mac>lq_lane_highv4sf): Ditto.
25547 * config/aarch64/arm_neon.h (vfmlal_low_u32): New intrinsic.
25548 (vfmlsl_low_u32): Ditto.
25549 (vfmlalq_low_u32): Ditto.
25550 (vfmlslq_low_u32): Ditto.
25551 (vfmlal_high_u32): Ditto.
25552 (vfmlsl_high_u32): Ditto.
25553 (vfmlalq_high_u32): Ditto.
25554 (vfmlslq_high_u32): Ditto.
25555 (vfmlal_lane_low_u32): Ditto.
25556 (vfmlsl_lane_low_u32): Ditto.
25557 (vfmlal_laneq_low_u32): Ditto.
25558 (vfmlsl_laneq_low_u32): Ditto.
25559 (vfmlalq_lane_low_u32): Ditto.
25560 (vfmlslq_lane_low_u32): Ditto.
25561 (vfmlalq_laneq_low_u32): Ditto.
25562 (vfmlslq_laneq_low_u32): Ditto.
25563 (vfmlal_lane_high_u32): Ditto.
25564 (vfmlsl_lane_high_u32): Ditto.
25565 (vfmlal_laneq_high_u32): Ditto.
25566 (vfmlsl_laneq_high_u32): Ditto.
25567 (vfmlalq_lane_high_u32): Ditto.
25568 (vfmlslq_lane_high_u32): Ditto.
25569 (vfmlalq_laneq_high_u32): Ditto.
25570 (vfmlslq_laneq_high_u32): Ditto.
25571 * config/aarch64/aarch64.h (AARCH64_FL_F16SML): New flag.
25572 (AARCH64_FL_FOR_ARCH8_4): New.
25573 (AARCH64_ISA_F16FML): New ISA flag.
25574 (TARGET_F16FML): New feature flag for fp16fml.
25575 (doc/invoke.texi): Document new fp16fml option.
25577 2018-01-10 Michael Collison <michael.collison@arm.com>
25579 * config/aarch64/aarch64-builtins.c:
25580 (aarch64_types_ternopu_imm_qualifiers, TYPES_TERNOPUI): New.
25581 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
25582 (__ARM_FEATURE_SHA3): Define if TARGET_SHA3 is true.
25583 * config/aarch64/aarch64.h (AARCH64_FL_SHA3): New flags.
25584 (AARCH64_ISA_SHA3): New ISA flag.
25585 (TARGET_SHA3): New feature flag for sha3.
25586 * config/aarch64/iterators.md (sha512_op): New int attribute.
25587 (CRYPTO_SHA512): New int iterator.
25588 (UNSPEC_SHA512H): New unspec.
25589 (UNSPEC_SHA512H2): Ditto.
25590 (UNSPEC_SHA512SU0): Ditto.
25591 (UNSPEC_SHA512SU1): Ditto.
25592 * config/aarch64/aarch64-simd-builtins.def
25593 (aarch64_crypto_sha512hqv2di): New builtin.
25594 (aarch64_crypto_sha512h2qv2di): Ditto.
25595 (aarch64_crypto_sha512su0qv2di): Ditto.
25596 (aarch64_crypto_sha512su1qv2di): Ditto.
25597 (aarch64_eor3qv8hi): Ditto.
25598 (aarch64_rax1qv2di): Ditto.
25599 (aarch64_xarqv2di): Ditto.
25600 (aarch64_bcaxqv8hi): Ditto.
25601 * config/aarch64/aarch64-simd.md:
25602 (aarch64_crypto_sha512h<sha512_op>qv2di): New pattern.
25603 (aarch64_crypto_sha512su0qv2di): Ditto.
25604 (aarch64_crypto_sha512su1qv2di): Ditto.
25605 (aarch64_eor3qv8hi): Ditto.
25606 (aarch64_rax1qv2di): Ditto.
25607 (aarch64_xarqv2di): Ditto.
25608 (aarch64_bcaxqv8hi): Ditto.
25609 * config/aarch64/arm_neon.h (vsha512hq_u64): New intrinsic.
25610 (vsha512h2q_u64): Ditto.
25611 (vsha512su0q_u64): Ditto.
25612 (vsha512su1q_u64): Ditto.
25613 (veor3q_u16): Ditto.
25614 (vrax1q_u64): Ditto.
25615 (vxarq_u64): Ditto.
25616 (vbcaxq_u16): Ditto.
25617 * config/arm/types.md (crypto_sha512): New type attribute.
25618 (crypto_sha3): Ditto.
25619 (doc/invoke.texi): Document new sha3 option.
25621 2018-01-10 Michael Collison <michael.collison@arm.com>
25623 * config/aarch64/aarch64-builtins.c:
25624 (aarch64_types_quadopu_imm_qualifiers, TYPES_QUADOPUI): New.
25625 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
25626 (__ARM_FEATURE_SM3): Define if TARGET_SM4 is true.
25627 (__ARM_FEATURE_SM4): Define if TARGET_SM4 is true.
25628 * config/aarch64/aarch64.h (AARCH64_FL_SM4): New flags.
25629 (AARCH64_ISA_SM4): New ISA flag.
25630 (TARGET_SM4): New feature flag for sm4.
25631 * config/aarch64/aarch64-simd-builtins.def
25632 (aarch64_sm3ss1qv4si): Ditto.
25633 (aarch64_sm3tt1aq4si): Ditto.
25634 (aarch64_sm3tt1bq4si): Ditto.
25635 (aarch64_sm3tt2aq4si): Ditto.
25636 (aarch64_sm3tt2bq4si): Ditto.
25637 (aarch64_sm3partw1qv4si): Ditto.
25638 (aarch64_sm3partw2qv4si): Ditto.
25639 (aarch64_sm4eqv4si): Ditto.
25640 (aarch64_sm4ekeyqv4si): Ditto.
25641 * config/aarch64/aarch64-simd.md:
25642 (aarch64_sm3ss1qv4si): Ditto.
25643 (aarch64_sm3tt<sm3tt_op>qv4si): Ditto.
25644 (aarch64_sm3partw<sm3part_op>qv4si): Ditto.
25645 (aarch64_sm4eqv4si): Ditto.
25646 (aarch64_sm4ekeyqv4si): Ditto.
25647 * config/aarch64/iterators.md (sm3tt_op): New int iterator.
25648 (sm3part_op): Ditto.
25649 (CRYPTO_SM3TT): Ditto.
25650 (CRYPTO_SM3PART): Ditto.
25651 (UNSPEC_SM3SS1): New unspec.
25652 (UNSPEC_SM3TT1A): Ditto.
25653 (UNSPEC_SM3TT1B): Ditto.
25654 (UNSPEC_SM3TT2A): Ditto.
25655 (UNSPEC_SM3TT2B): Ditto.
25656 (UNSPEC_SM3PARTW1): Ditto.
25657 (UNSPEC_SM3PARTW2): Ditto.
25658 (UNSPEC_SM4E): Ditto.
25659 (UNSPEC_SM4EKEY): Ditto.
25660 * config/aarch64/constraints.md (Ui2): New constraint.
25661 * config/aarch64/predicates.md (aarch64_imm2): New predicate.
25662 * config/arm/types.md (crypto_sm3): New type attribute.
25663 (crypto_sm4): Ditto.
25664 * config/aarch64/arm_neon.h (vsm3ss1q_u32): New intrinsic.
25665 (vsm3tt1aq_u32): Ditto.
25666 (vsm3tt1bq_u32): Ditto.
25667 (vsm3tt2aq_u32): Ditto.
25668 (vsm3tt2bq_u32): Ditto.
25669 (vsm3partw1q_u32): Ditto.
25670 (vsm3partw2q_u32): Ditto.
25671 (vsm4eq_u32): Ditto.
25672 (vsm4ekeyq_u32): Ditto.
25673 (doc/invoke.texi): Document new sm4 option.
25675 2018-01-10 Michael Collison <michael.collison@arm.com>
25677 * config/aarch64/aarch64-arches.def (armv8.4-a): New architecture.
25678 * config/aarch64/aarch64.h (AARCH64_ISA_V8_4): New ISA flag.
25679 (AARCH64_FL_FOR_ARCH8_4): New.
25680 (AARCH64_FL_V8_4): New flag.
25681 (doc/invoke.texi): Document new armv8.4-a option.
25683 2018-01-10 Michael Collison <michael.collison@arm.com>
25685 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
25686 (__ARM_FEATURE_AES): Define if TARGET_AES is true.
25687 (__ARM_FEATURE_SHA2): Define if TARGET_SHA2 is true.
25688 * config/aarch64/aarch64-option-extension.def: Add
25689 AARCH64_OPT_EXTENSION of 'sha2'.
25690 (aes): Add AARCH64_OPT_EXTENSION of 'aes'.
25691 (crypto): Disable sha2 and aes if crypto disabled.
25692 (crypto): Enable aes and sha2 if enabled.
25693 (simd): Disable sha2 and aes if simd disabled.
25694 * config/aarch64/aarch64.h (AARCH64_FL_AES, AARCH64_FL_SHA2):
25696 (AARCH64_ISA_AES, AARCH64_ISA_SHA2): New ISA flags.
25697 (TARGET_SHA2): New feature flag for sha2.
25698 (TARGET_AES): New feature flag for aes.
25699 * config/aarch64/aarch64-simd.md:
25700 (aarch64_crypto_aes<aes_op>v16qi): Make pattern
25701 conditional on TARGET_AES.
25702 (aarch64_crypto_aes<aesmc_op>v16qi): Ditto.
25703 (aarch64_crypto_sha1hsi): Make pattern conditional
25705 (aarch64_crypto_sha1hv4si): Ditto.
25706 (aarch64_be_crypto_sha1hv4si): Ditto.
25707 (aarch64_crypto_sha1su1v4si): Ditto.
25708 (aarch64_crypto_sha1<sha1_op>v4si): Ditto.
25709 (aarch64_crypto_sha1su0v4si): Ditto.
25710 (aarch64_crypto_sha256h<sha256_op>v4si): Ditto.
25711 (aarch64_crypto_sha256su0v4si): Ditto.
25712 (aarch64_crypto_sha256su1v4si): Ditto.
25713 (doc/invoke.texi): Document new aes and sha2 options.
25715 2018-01-10 Martin Sebor <msebor@redhat.com>
25717 PR tree-optimization/83781
25718 * gimple-fold.c (get_range_strlen): Avoid treating arrays of pointers
25721 2018-01-11 Martin Sebor <msebor@gmail.com>
25722 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
25724 PR tree-optimization/83501
25725 PR tree-optimization/81703
25727 * tree-ssa-strlen.c (get_string_cst): Rename...
25728 (get_string_len): ...to this. Handle global constants.
25729 (handle_char_store): Adjust.
25731 2018-01-10 Kito Cheng <kito.cheng@gmail.com>
25732 Jim Wilson <jimw@sifive.com>
25734 * config/riscv/riscv-protos.h (riscv_output_return): New.
25735 * config/riscv/riscv.c (struct machine_function): New naked_p field.
25736 (riscv_attribute_table, riscv_output_return),
25737 (riscv_handle_fndecl_attribute, riscv_naked_function_p),
25738 (riscv_allocate_stack_slots_for_args, riscv_warn_func_return): New.
25739 (riscv_compute_frame_info): Only compute frame->mask if not a naked
25741 (riscv_expand_prologue): Add early return for naked function.
25742 (riscv_expand_epilogue): Likewise.
25743 (riscv_function_ok_for_sibcall): Return false for naked function.
25744 (riscv_set_current_function): New.
25745 (TARGET_SET_CURRENT_FUNCTION, TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS),
25746 (TARGET_ATTRIBUTE_TABLE, TARGET_WARN_FUNC_RETURN): New.
25747 * config/riscv/riscv.md (simple_return): Call riscv_output_return.
25748 * doc/extend.texi (RISC-V Function Attributes): New.
25750 2018-01-10 Michael Meissner <meissner@linux.vnet.ibm.com>
25752 * config/rs6000/rs6000.c (is_complex_IBM_long_double): Explicitly
25753 check for 128-bit long double before checking TCmode.
25754 * config/rs6000/rs6000.h (FLOAT128_IEEE_P): Explicitly check for
25755 128-bit long doubles before checking TFmode or TCmode.
25756 (FLOAT128_IBM_P): Likewise.
25758 2018-01-10 Martin Sebor <msebor@redhat.com>
25760 PR tree-optimization/83671
25761 * builtins.c (c_strlen): Unconditionally return zero for the empty
25763 Use -Warray-bounds for warnings.
25764 * gimple-fold.c (get_range_strlen): Handle non-constant lengths
25765 for non-constant array indices with COMPONENT_REF, arrays of
25766 arrays, and pointers to arrays.
25767 (gimple_fold_builtin_strlen): Determine and set length range for
25768 non-constant character arrays.
25770 2018-01-10 Aldy Hernandez <aldyh@redhat.com>
25772 PR middle-end/81897
25773 * tree-ssa-uninit.c (convert_control_dep_chain_into_preds): Skip
25776 2018-01-10 Eric Botcazou <ebotcazou@adacore.com>
25778 * dwarf2out.c (dwarf2out_var_location): Do not pass NULL to fprintf.
25780 2018-01-10 Peter Bergner <bergner@vnet.ibm.com>
25783 * config/rs6000/rs6000.c (print_operand) <'y'>: Use
25784 VECTOR_MEM_ALTIVEC_OR_VSX_P.
25785 * config/rs6000/vsx.md (*vsx_le_perm_load_<mode> for VSX_D): Use
25786 indexed_or_indirect_operand predicate.
25787 (*vsx_le_perm_load_<mode> for VSX_W): Likewise.
25788 (*vsx_le_perm_load_v8hi): Likewise.
25789 (*vsx_le_perm_load_v16qi): Likewise.
25790 (*vsx_le_perm_store_<mode> for VSX_D): Likewise.
25791 (*vsx_le_perm_store_<mode> for VSX_W): Likewise.
25792 (*vsx_le_perm_store_v8hi): Likewise.
25793 (*vsx_le_perm_store_v16qi): Likewise.
25794 (eight unnamed splitters): Likewise.
25796 2018-01-10 Peter Bergner <bergner@vnet.ibm.com>
25798 * config/rs6000/x86intrin.h: Change #warning to #error. Update message.
25799 * config/rs6000/emmintrin.h: Likewise.
25800 * config/rs6000/mmintrin.h: Likewise.
25801 * config/rs6000/xmmintrin.h: Likewise.
25803 2018-01-10 David Malcolm <dmalcolm@redhat.com>
25806 * tree-core.h: Document EXPR_LOCATION_WRAPPER_P's usage of
25808 * tree.c (tree_nop_conversion): Return true for location wrapper
25810 (maybe_wrap_with_location): New function.
25811 (selftest::check_strip_nops): New function.
25812 (selftest::test_location_wrappers): New function.
25813 (selftest::tree_c_tests): Call it.
25814 * tree.h (STRIP_ANY_LOCATION_WRAPPER): New macro.
25815 (maybe_wrap_with_location): New decl.
25816 (EXPR_LOCATION_WRAPPER_P): New macro.
25817 (location_wrapper_p): New inline function.
25818 (tree_strip_any_location_wrapper): New inline function.
25820 2018-01-10 H.J. Lu <hongjiu.lu@intel.com>
25823 * config/i386/i386.c (ix86_compute_frame_layout): Always adjust
25824 stack_realign_offset for the largest alignment of stack slot
25826 (ix86_find_max_used_stack_alignment): New function.
25827 (ix86_finalize_stack_frame_flags): Use it. Set
25828 max_used_stack_alignment if we don't realign stack.
25829 * config/i386/i386.h (machine_function): Add
25830 max_used_stack_alignment.
25832 2018-01-10 Christophe Lyon <christophe.lyon@linaro.org>
25834 * config/arm/arm.opt (-mbranch-cost): New option.
25835 * config/arm/arm.h (BRANCH_COST): Take arm_branch_cost into
25838 2018-01-10 Segher Boessenkool <segher@kernel.crashing.org>
25841 * config/rs6000/rs6000.md (load_toc_v4_PIC_2, load_toc_v4_PIC_3b,
25842 load_toc_v4_PIC_3c): Wrap const term in CONST RTL.
25844 2018-01-10 Richard Biener <rguenther@suse.de>
25847 * dwarf2out.c (gen_subprogram_die): Hoist old_die && declaration
25848 early out so it also covers the case where we have a non-NULL
25851 2018-01-10 Richard Sandiford <richard.sandiford@linaro.org>
25853 PR tree-optimization/83753
25854 * tree-vect-stmts.c (get_group_load_store_type): Use VMAT_CONTIGUOUS
25855 for non-strided grouped accesses if the number of elements is 1.
25857 2018-01-10 Jan Hubicka <hubicka@ucw.cz>
25860 * i386.c (ix86_vectorize_builtin_gather): Check TARGET_USE_GATHER.
25861 * i386.h (TARGET_USE_GATHER): Define.
25862 * x86-tune.def (X86_TUNE_USE_GATHER): New.
25864 2018-01-10 Martin Liska <mliska@suse.cz>
25867 * basic-block.h (CLEANUP_NO_PARTITIONING): New define.
25868 * bb-reorder.c (pass_reorder_blocks::execute): Do not clean up
25870 * cfgcleanup.c (try_optimize_cfg): Fix up partitioning if
25871 CLEANUP_NO_PARTITIONING is not set.
25873 2018-01-10 Richard Sandiford <richard.sandiford@linaro.org>
25875 * doc/rtl.texi: Remove documentation of (const ...) wrappers
25876 for vectors, as a partial revert of r254296.
25877 * rtl.h (const_vec_p): Delete.
25878 (const_vec_duplicate_p): Don't test for vector CONSTs.
25879 (unwrap_const_vec_duplicate, const_vec_series_p): Likewise.
25880 * expmed.c (make_tree): Likewise.
25883 * common.md (E, F): Use CONSTANT_P instead of checking for
25885 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
25886 checking for CONST_VECTOR.
25888 2018-01-09 Jan Hubicka <hubicka@ucw.cz>
25890 PR middle-end/83575
25891 * predict.c (force_edge_cold): Handle in more sane way edges
25892 with no prediction.
25894 2018-01-09 Carl Love <cel@us.ibm.com>
25896 * config/rs6002/altivec.md (p8_vmrgow): Add support for V2DI, V2DF,
25898 (p8_vmrgew): Add support for V2DI, V2DF, V4SF types.
25899 * config/rs6000/rs6000-builtin.def: Add definitions for FLOAT2_V2DF,
25900 VMRGEW_V2DI, VMRGEW_V2DF, VMRGEW_V4SF, VMRGOW_V4SI, VMRGOW_V4SF,
25901 VMRGOW_V2DI, VMRGOW_V2DF. Remove definition for VMRGOW.
25902 * config/rs6000/rs6000-c.c (VSX_BUILTIN_VEC_FLOAT2,
25903 P8V_BUILTIN_VEC_VMRGEW, P8V_BUILTIN_VEC_VMRGOW): Add definitions.
25904 * config/rs6000/rs6000-protos.h: Add extern defition for
25905 rs6000_generate_float2_double_code.
25906 * config/rs6000/rs6000.c (rs6000_generate_float2_double_code): Add
25908 * config/rs6000/vsx.md (vsx_xvcdpsp): Add define_insn.
25909 (float2_v2df): Add define_expand.
25911 2018-01-09 Uros Bizjak <ubizjak@gmail.com>
25914 * combine.c (force_int_to_mode) <case ASHIFT>: Use mode instead of
25915 op_mode in the force_to_mode call.
25917 2018-01-09 Richard Sandiford <richard.sandiford@linaro.org>
25919 * config/aarch64/aarch64.c (aarch64_evpc_trn): Use d.perm.series_p
25920 instead of checking each element individually.
25921 (aarch64_evpc_uzp): Likewise.
25922 (aarch64_evpc_zip): Likewise.
25923 (aarch64_evpc_ext): Likewise.
25924 (aarch64_evpc_rev): Likewise.
25925 (aarch64_evpc_dup): Test the encoding for a single duplicated element,
25926 instead of checking each element individually. Return true without
25928 (aarch64_vectorize_vec_perm_const): Use all_from_input_p to test
25929 whether all selected elements come from the same input, instead of
25930 checking each element individually. Remove calls to gen_rtx_REG,
25931 start_sequence and end_sequence and instead assert that no rtl is
25934 2018-01-09 Richard Sandiford <richard.sandiford@linaro.org>
25936 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p): Fix
25937 order of HIGH and CONST checks.
25939 2018-01-09 Richard Sandiford <richard.sandiford@linaro.org>
25941 * tree-vect-stmts.c (permute_vec_elements): Create a fresh variable
25942 if the destination isn't an SSA_NAME.
25944 2018-01-09 Richard Biener <rguenther@suse.de>
25946 PR tree-optimization/83668
25947 * graphite.c (canonicalize_loop_closed_ssa): Add edge argument,
25949 (canonicalize_loop_form): ... here, renamed from ...
25950 (canonicalize_loop_closed_ssa_form): ... this and amended to
25951 swap successor edges for loop exit blocks to make us use
25952 the RPO order we need for initial schedule generation.
25954 2018-01-09 Joseph Myers <joseph@codesourcery.com>
25956 PR tree-optimization/64811
25957 * match.pd: When optimizing comparisons with Inf, avoid
25958 introducing or losing exceptions from comparisons with NaN.
25960 2018-01-09 Martin Liska <mliska@suse.cz>
25963 * asan.c (shadow_mem_size): Add gcc_assert.
25965 2018-01-09 Georg-Johann Lay <avr@gjlay.de>
25967 Don't save registers in main().
25970 * doc/invoke.texi (AVR Options) [-mmain-is-OS_task]: Document it.
25971 * config/avr/avr.opt (-mmain-is-OS_task): New target option.
25972 * config/avr/avr.c (avr_set_current_function): Don't error if
25973 naked, OS_task or OS_main are specified at the same time.
25974 (avr_function_ok_for_sibcall): Don't disable sibcalls for OS_task,
25976 (avr_insert_attributes) [-mmain-is-OS_task] <main>: Add OS_task
25978 * common/config/avr/avr-common.c (avr_option_optimization_table):
25979 Switch on -mmain-is-OS_task for optimizing compilations.
25981 2018-01-09 Richard Biener <rguenther@suse.de>
25983 PR tree-optimization/83572
25984 * graphite.c: Include cfganal.h.
25985 (graphite_transform_loops): Connect infinite loops to exit
25986 and remove fake edges at the end.
25988 2018-01-09 Jan Hubicka <hubicka@ucw.cz>
25990 * ipa-inline.c (edge_badness): Revert accidental checkin.
25992 2018-01-09 Jan Hubicka <hubicka@ucw.cz>
25995 * ipa-comdats.c (set_comdat_group): Only set comdat group of real
25996 symbols; not inline clones.
25998 2018-01-09 Jakub Jelinek <jakub@redhat.com>
26001 * modulo-sched.c (schedule_reg_moves): Punt if we'd need to move
26002 hard registers. Formatting fixes.
26004 PR preprocessor/83722
26005 * gcc.c (try_generate_repro): Pass
26006 &temp_stderr_files[RETRY_ICE_ATTEMPTS - 1] rather than
26007 &temp_stdout_files[RETRY_ICE_ATTEMPTS - 1] as last argument to
26010 2018-01-08 Monk Chiang <sh.chiang04@gmail.com>
26011 Kito Cheng <kito.cheng@gmail.com>
26013 * config/riscv/riscv.c (machine_function::is_leaf): Remove field.
26014 (riscv_leaf_function_p): Delete.
26015 (riscv_function_ok_for_sibcall): Return false when TARGET_SAVE_RESTORE.
26017 2018-01-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
26019 * config/rs6000/rs6000-string.c (do_load_for_compare_from_addr): New
26021 (do_ifelse): New function.
26022 (do_isel): New function.
26023 (do_sub3): New function.
26024 (do_add3): New function.
26025 (do_load_mask_compare): New function.
26026 (do_overlap_load_compare): New function.
26027 (expand_compare_loop): New function.
26028 (expand_block_compare): Call expand_compare_loop() when appropriate.
26029 * config/rs6000/rs6000.opt (-mblock-compare-inline-limit): Change
26030 option description.
26031 (-mblock-compare-inline-loop-limit): New option.
26033 2018-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26036 * config/rs6000/altivec.md (*altivec_vpermr_<mode>_internal):
26037 Reverse order of second and third operands in first alternative.
26038 * config/rs6000/rs6000.c (rs6000_expand_vector_set): Reverse order
26039 of first and second elements in UNSPEC_VPERMR vector.
26040 (altivec_expand_vec_perm_le): Likewise.
26042 2018-01-08 Jeff Law <law@redhat.com>
26044 PR rtl-optimizatin/81308
26045 * tree-switch-conversion.c (cfg_altered): New file scoped static.
26046 (process_switch): If group_case_labels makes a change, then set
26048 (pass_convert_switch::execute): If a switch is converted, then
26049 set cfg_altered. Return TODO_cfg_cleanup if cfg_altered is true.
26051 PR rtl-optimization/81308
26052 * recog.c (split_all_insns): Conditionally cleanup the CFG after
26055 2018-01-08 Vidya Praveen <vidyapraveen@arm.com>
26057 PR target/83663 - Revert r255946
26058 * config/aarch64/aarch64.c (aarch64_expand_vector_init): Modify code
26059 generation for cases where splatting a value is not useful.
26060 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
26061 across a vec_duplicate and a paradoxical subreg forming a vector
26062 mode to a vec_concat.
26064 2018-01-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26066 * config/arm/t-aprofile (MULTILIB_MATCHES): Add mapping rules for
26067 -march=armv8.3-a variants.
26068 * config/arm/t-multilib: Likewise.
26069 * config/arm/t-arm-elf: Likewise. Handle dotprod extension.
26071 2018-01-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
26073 * config/rs6000/rs6000.md (cceq_ior_compare): Remove * so I can use it
26075 (cceq_ior_compare_complement): Give it a name so I can use it, and
26076 change boolean_or_operator predicate to boolean_operator so it can
26077 be used to generate a crand.
26078 (eqne): New code iterator.
26079 (bd/bd_neg): New code_attrs.
26080 (<bd>_<mode>): New name for ctr<mode>_internal[12] now combined into
26081 a single define_insn.
26082 (<bd>tf_<mode>): A new insn pattern for the conditional form branch
26083 decrement (bdnzt/bdnzf/bdzt/bdzf).
26084 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Updated
26085 with the new names of the branch decrement patterns, and added the
26086 names of the branch decrement conditional patterns.
26088 2018-01-08 Richard Biener <rguenther@suse.de>
26090 PR tree-optimization/83563
26091 * graphite.c (canonicalize_loop_closed_ssa_form): Reset the SCEV
26094 2018-01-08 Richard Biener <rguenther@suse.de>
26096 PR middle-end/83713
26097 * convert.c (do_narrow): Properly guard TYPE_OVERFLOW_WRAPS checks.
26099 2018-01-08 Richard Biener <rguenther@suse.de>
26101 PR tree-optimization/83685
26102 * tree-ssa-pre.c (create_expression_by_pieces): Do not insert
26103 references to abnormals.
26105 2018-01-08 Richard Biener <rguenther@suse.de>
26108 * dwarf2out.c (output_indirect_strings): Handle empty
26109 skeleton_debug_str_hash.
26110 (dwarf2out_early_finish): Index strings for -gsplit-dwarf.
26112 2018-01-08 Claudiu Zissulescu <claziss@synopsys.com>
26114 * config/arc/arc.c (TARGET_TRAMPOLINE_ADJUST_ADDRESS): Delete.
26115 (emit_store_direct): Likewise.
26116 (arc_trampoline_adjust_address): Likewise.
26117 (arc_asm_trampoline_template): New function.
26118 (arc_initialize_trampoline): Use asm_trampoline_template.
26119 (TARGET_ASM_TRAMPOLINE_TEMPLATE): Define.
26120 * config/arc/arc.h (TRAMPOLINE_SIZE): Adjust to 16.
26121 * config/arc/arc.md (flush_icache): Delete pattern.
26123 2018-01-08 Claudiu Zissulescu <claziss@synopsys.com>
26125 * config/arc/arc-c.def (__ARC_UNALIGNED__): New define.
26126 * config/arc/arc.h (STRICT_ALIGNMENT): Control this macro using
26129 2018-01-08 Sebastian Huber <sebastian.huber@embedded-brains.de>
26132 * config/epiphany/epiphany.h (make_pass_mode_switch_use): Guard
26133 by not USED_FOR_TARGET.
26134 (make_pass_resolve_sw_modes): Likewise.
26136 2018-01-08 Sebastian Huber <sebastian.huber@embedded-brains.de>
26138 * config/nios2/nios2.h (nios2_section_threshold): Guard by not
26141 2018-01-08 Richard Biener <rguenther@suse.de>
26143 PR middle-end/83580
26144 * tree-data-ref.c (split_constant_offset): Remove STRIP_NOPS.
26146 2018-01-08 Richard Biener <rguenther@suse.de>
26148 PR middle-end/83517
26149 * match.pd ((t * 2) / 2) -> t): Add missing :c.
26151 2018-01-06 Aldy Hernandez <aldyh@redhat.com>
26153 PR middle-end/81897
26154 * tree-ssa-uninit.c (compute_control_dep_chain): Do not bail on
26155 basic blocks with a small number of successors.
26156 (convert_control_dep_chain_into_preds): Improve handling of
26158 (dump_predicates): Split apart into...
26159 (dump_pred_chain): ...here...
26160 (dump_pred_info): ...and here.
26161 (can_one_predicate_be_invalidated_p): Add debugging printfs.
26162 (can_chain_union_be_invalidated_p): Improve check for invalidation
26164 (uninit_uses_cannot_happen): Avoid unnecessary if
26165 convert_control_dep_chain_into_preds yielded nothing.
26167 2018-01-06 Martin Sebor <msebor@redhat.com>
26169 PR tree-optimization/83640
26170 * gimple-ssa-warn-restrict.c (builtin_access::builtin_access): Avoid
26171 subtracting negative offset from size.
26172 (builtin_access::overlap): Adjust offset bounds of the access to fall
26173 within the size of the object if possible.
26175 2018-01-06 Richard Sandiford <richard.sandiford@linaro.org>
26177 PR rtl-optimization/83699
26178 * expmed.c (extract_bit_field_1): Restrict the vector usage of
26179 extract_bit_field_as_subreg to cases in which the extracted
26180 value is also a vector.
26182 * lra-constraints.c (process_alt_operands): Test for the equivalence
26183 substitutions when detecting a possible reload cycle.
26185 2018-01-06 Jakub Jelinek <jakub@redhat.com>
26188 * toplev.c (process_options): Don't enable debug_nonbind_markers_p
26189 by default if flag_selective_schedling{,2}. Formatting fixes.
26191 PR rtl-optimization/83682
26192 * rtl.h (const_vec_duplicate_p): Only return true for VEC_DUPLICATE
26193 if it has non-VECTOR_MODE element mode.
26194 (vec_duplicate_p): Likewise.
26196 PR middle-end/83694
26197 * cfgexpand.c (expand_debug_expr): Punt if mode1 is VOIDmode
26198 and bitsize might be greater than MAX_BITSIZE_MODE_ANY_INT.
26200 2018-01-05 Jakub Jelinek <jakub@redhat.com>
26203 * config/i386/i386-builtin.def
26204 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
26205 __builtin_ia32_vgf2p8affineqb_v64qi, __builtin_ia32_vgf2p8mulb_v64qi):
26206 Require also OPTION_MASK_ISA_AVX512F in addition to
26207 OPTION_MASK_ISA_GFNI.
26208 (__builtin_ia32_vgf2p8affineinvqb_v16qi_mask,
26209 __builtin_ia32_vgf2p8affineqb_v16qi_mask): Require
26210 OPTION_MASK_ISA_AVX512VL instead of OPTION_MASK_ISA_SSE in addition
26211 to OPTION_MASK_ISA_GFNI.
26212 (__builtin_ia32_vgf2p8mulb_v32qi_mask): Require
26213 OPTION_MASK_ISA_AVX512VL in addition to OPTION_MASK_ISA_GFNI and
26214 OPTION_MASK_ISA_AVX512BW.
26215 (__builtin_ia32_vgf2p8mulb_v16qi_mask): Require
26216 OPTION_MASK_ISA_AVX512VL instead of OPTION_MASK_ISA_AVX512BW in
26217 addition to OPTION_MASK_ISA_GFNI.
26218 (__builtin_ia32_vgf2p8affineinvqb_v16qi,
26219 __builtin_ia32_vgf2p8affineqb_v16qi, __builtin_ia32_vgf2p8mulb_v16qi):
26220 Require OPTION_MASK_ISA_SSE2 instead of OPTION_MASK_ISA_SSE in addition
26221 to OPTION_MASK_ISA_GFNI.
26222 * config/i386/i386.c (def_builtin): Change to builtin isa/isa2 being
26223 a requirement for all ISAs rather than any of them with a few
26225 (ix86_add_new_builtins): Clear OPTION_MASK_ISA_64BIT from isa before
26227 (ix86_expand_builtin): Require all ISAs from builtin's isa and isa2
26228 bitmasks to be enabled with 3 exceptions, instead of requiring any
26229 enabled ISA with lots of exceptions.
26230 * config/i386/sse.md (vgf2p8affineinvqb_<mode><mask_name>,
26231 vgf2p8affineqb_<mode><mask_name>, vgf2p8mulb_<mode><mask_name>):
26232 Change avx512bw in isa attribute to avx512f.
26233 * config/i386/sgxintrin.h: Add license boilerplate.
26234 * config/i386/vaesintrin.h: Likewise. Fix macro spelling __AVX512F
26235 to __AVX512F__ and __AVX512VL to __AVX512VL__.
26236 (_mm256_aesdec_epi128, _mm256_aesdeclast_epi128, _mm256_aesenc_epi128,
26237 _mm256_aesenclast_epi128): Enable temporarily avx if __AVX__ is not
26239 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8,
26240 _mm_gf2p8affineinv_epi64_epi8, _mm_gf2p8affine_epi64_epi8): Enable
26241 temporarily sse2 rather than sse if not enabled already.
26244 * config/i386/sse.md (VI248_VLBW): Rename to ...
26245 (VI248_AVX512VL): ... this. Don't guard V32HI with TARGET_AVX512BW.
26246 (vpshrd_<mode><mask_name>, vpshld_<mode><mask_name>,
26247 vpshrdv_<mode>, vpshrdv_<mode>_mask, vpshrdv_<mode>_maskz,
26248 vpshrdv_<mode>_maskz_1, vpshldv_<mode>, vpshldv_<mode>_mask,
26249 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1): Use VI248_AVX512VL
26250 mode iterator instead of VI248_VLBW.
26252 2018-01-05 Jan Hubicka <hubicka@ucw.cz>
26254 * ipa-fnsummary.c (record_modified_bb_info): Add OP.
26255 (record_modified): Skip clobbers; add debug output.
26256 (param_change_prob): Use sreal frequencies.
26258 2018-01-05 Richard Sandiford <richard.sandiford@linaro.org>
26260 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
26261 punt for user-aligned variables.
26263 2018-01-05 Richard Sandiford <richard.sandiford@linaro.org>
26265 * tree-chrec.c (chrec_contains_symbols): Return true for
26268 2018-01-05 Sudakshina Das <sudi.das@arm.com>
26271 * simplify-rtx.c (simplify_relational_operation_1): Add simplifications
26272 of (x|y) == x for BICS pattern.
26274 2018-01-05 Jakub Jelinek <jakub@redhat.com>
26276 PR tree-optimization/83605
26277 * gimple-ssa-strength-reduction.c: Include tree-eh.h.
26278 (find_candidates_dom_walker::before_dom_children): Ignore stmts that
26281 2018-01-05 Sebastian Huber <sebastian.huber@embedded-brains.de>
26283 * config.gcc (epiphany-*-elf*): Add (epiphany-*-rtems*) configuration.
26284 * config/epiphany/rtems.h: New file.
26286 2018-01-04 Jakub Jelinek <jakub@redhat.com>
26287 Uros Bizjak <ubizjak@gmail.com>
26290 * config/i386/i386.md (*<rotate_insn>hi3_1 splitter): Use
26291 QIreg_operand instead of register_operand predicate.
26292 * config/i386/i386.c (ix86_rop_should_change_byte_p,
26293 set_rop_modrm_reg_bits, ix86_mitigate_rop): Use -mmitigate-rop in
26294 comments instead of -fmitigate[-_]rop.
26296 2018-01-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
26299 * cgraphunit.c (symbol_table::compile): Switch to text_section
26300 before calling assembly_start debug hook.
26301 * run-rtl-passes.c (run_rtl_passes): Likewise.
26304 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
26306 * tree-vrp.c (extract_range_from_binary_expr_1): Check
26307 range_int_cst_p rather than !symbolic_range_p before calling
26308 extract_range_from_multiplicative_op_1.
26310 2018-01-04 Jeff Law <law@redhat.com>
26312 * tree-ssa-math-opts.c (execute_cse_reciprocals_1): Remove
26313 redundant test in assertion.
26315 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
26317 * doc/rtl.texi: Document machine_mode wrapper classes.
26319 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
26321 * fold-const.c (fold_ternary_loc): Check tree_fits_uhwi_p before
26322 using tree_to_uhwi.
26324 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
26326 * tree-ssa-forwprop.c (is_combined_permutation_identity): Allow
26327 the VEC_PERM_EXPR fold to fail.
26329 2018-01-04 Jakub Jelinek <jakub@redhat.com>
26332 * bb-reorder.c (insert_section_boundary_note): Set has_bb_partition
26333 to switched_sections.
26335 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
26338 * config/arm/arm.c (arm_vectorize_vec_perm_const): Fix inverted
26339 test for d.testing.
26341 2018-01-04 Peter Bergner <bergner@vnet.ibm.com>
26344 * config/rs6000/rs6000.c (rs6000_discover_homogeneous_aggregate): Do not
26345 allow arguments in FP registers if TARGET_HARD_FLOAT is false.
26347 2018-01-04 Jakub Jelinek <jakub@redhat.com>
26350 * cfgexpand.c (expand_debug_expr) <case BIT_FIELD_REF>: Punt if mode
26351 is BLKmode and bitpos not zero or mode change is needed.
26353 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
26356 * config/sparc/sparc.c (sparc_vectorize_vec_perm_const): Require
26359 2018-01-04 Uros Bizjak <ubizjak@gmail.com>
26362 * config/alpha/alpha.md (*sadd<modesuffix>): Use ASHIFT
26363 instead of MULT rtx. Update all corresponding splitters.
26364 (*saddl_se): Ditto.
26365 (*ssub<modesuffix>): Ditto.
26366 (*ssubl_se): Ditto.
26367 (*cmp_sadd_di): Update split patterns.
26368 (*cmp_sadd_si): Ditto.
26369 (*cmp_sadd_sidi): Ditto.
26370 (*cmp_ssub_di): Ditto.
26371 (*cmp_ssub_si): Ditto.
26372 (*cmp_ssub_sidi): Ditto.
26373 * config/alpha/predicates.md (const23_operand): New predicate.
26374 * config/alpha/alpha.c (alpha_rtx_costs) [PLUS, MINUS]:
26375 Look for ASHIFT, not MULT inner operand.
26376 (alpha_split_conditional_move): Update for *sadd<modesuffix> change.
26378 2018-01-04 Martin Liska <mliska@suse.cz>
26380 PR gcov-profile/83669
26381 * gcov.c (output_intermediate_file): Add version to intermediate
26383 * doc/gcov.texi: Document new field 'version' in intermediate
26384 file format. Fix location of '-k' option of gcov command.
26386 2018-01-04 Martin Liska <mliska@suse.cz>
26389 * ipa-icf.c (sem_function::merge): Do not cross comdat boundary.
26391 2018-01-04 Jakub Jelinek <jakub@redhat.com>
26393 * gimple-ssa-sprintf.c (parse_directive): Cast second dir.len to uhwi.
26395 2018-01-03 Martin Sebor <msebor@redhat.com>
26397 PR tree-optimization/83655
26398 * gimple-ssa-warn-restrict.c (wrestrict_dom_walker::check_call): Avoid
26399 checking calls with invalid arguments.
26401 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
26403 * tree-vect-stmts.c (vect_get_store_rhs): New function.
26404 (vectorizable_mask_load_store): Delete.
26405 (vectorizable_call): Return false for masked loads and stores.
26406 (vectorizable_store): Handle IFN_MASK_STORE. Use vect_get_store_rhs
26407 instead of gimple_assign_rhs1.
26408 (vectorizable_load): Handle IFN_MASK_LOAD.
26409 (vect_transform_stmt): Don't set is_store for call_vec_info_type.
26411 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
26413 * tree-vect-stmts.c (vect_build_gather_load_calls): New function,
26415 (vectorizable_mask_load_store): ...here.
26416 (vectorizable_load): ...and here.
26418 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
26420 * tree-vect-stmts.c (vect_build_all_ones_mask)
26421 (vect_build_zero_merge_argument): New functions, split out from...
26422 (vectorizable_load): ...here.
26424 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
26426 * tree-vect-stmts.c (vect_check_store_rhs): New function,
26428 (vectorizable_mask_load_store): ...here.
26429 (vectorizable_store): ...and here.
26431 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
26433 * tree-vect-stmts.c (vect_check_load_store_mask): New function,
26435 (vectorizable_mask_load_store): ...here.
26437 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
26439 * tree-vectorizer.h (vec_load_store_type): Moved from tree-vec-stmts.c
26440 (vect_model_store_cost): Take a vec_load_store_type instead of a
26442 * tree-vect-stmts.c (vec_load_store_type): Move to tree-vectorizer.h.
26443 (vect_model_store_cost): Take a vec_load_store_type instead of a
26445 (vectorizable_mask_load_store): Update accordingly.
26446 (vectorizable_store): Likewise.
26447 * tree-vect-slp.c (vect_analyze_slp_cost_1): Update accordingly.
26449 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
26451 * tree-vect-loop.c (vect_transform_loop): Stub out scalar
26452 IFN_MASK_LOAD calls here rather than...
26453 * tree-vect-stmts.c (vectorizable_mask_load_store): ...here.
26455 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
26456 Alan Hayward <alan.hayward@arm.com>
26457 David Sherwood <david.sherwood@arm.com>
26459 * expmed.c (extract_bit_field_1): For vector extracts,
26460 fall back to extract_bit_field_as_subreg if vec_extract
26463 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
26464 Alan Hayward <alan.hayward@arm.com>
26465 David Sherwood <david.sherwood@arm.com>
26467 * lra-spills.c (pseudo_reg_slot_compare): Sort slots by whether
26468 they are variable or constant sized.
26469 (assign_stack_slot_num_and_sort_pseudos): Don't reuse variable-sized
26470 slots for constant-sized data.
26472 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
26473 Alan Hayward <alan.hayward@arm.com>
26474 David Sherwood <david.sherwood@arm.com>
26476 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern): When
26477 handling COND_EXPRs with boolean comparisons, try to find a better
26478 basis for the mask type than the boolean itself.
26480 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
26482 * doc/rtl.texi (MAX_BITSIZE_MODE_ANY_MODE): Describe how the default
26483 is calculated and how it can be overridden.
26484 * genmodes.c (max_bitsize_mode_any_mode): New variable.
26485 (create_modes): Initialize it from MAX_BITSIZE_MODE_ANY_MODE,
26487 (emit_max_int): Use it to set the output MAX_BITSIZE_MODE_ANY_MODE,
26490 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
26491 Alan Hayward <alan.hayward@arm.com>
26492 David Sherwood <david.sherwood@arm.com>
26494 * config/aarch64/aarch64-protos.h (aarch64_output_simd_mov_immediate):
26495 Remove the mode argument.
26496 (aarch64_simd_valid_immediate): Remove the mode and inverse
26498 * config/aarch64/iterators.md (bitsize): New iterator.
26499 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>, and<mode>3)
26500 (ior<mode>3): Update calls to aarch64_output_simd_mov_immediate.
26501 * config/aarch64/constraints.md (Do, Db, Dn): Update calls to
26502 aarch64_simd_valid_immediate.
26503 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Likewise.
26504 (aarch64_reg_or_bic_imm): Likewise.
26505 * config/aarch64/aarch64.c (simd_immediate_info): Replace mvn
26506 with an insn_type enum and msl with a modifier_type enum.
26507 Replace element_width with a scalar_mode. Change the shift
26508 to unsigned int. Add constructors for scalar_float_mode and
26509 scalar_int_mode elements.
26510 (aarch64_vect_float_const_representable_p): Delete.
26511 (aarch64_can_const_movi_rtx_p)
26512 (aarch64_simd_scalar_immediate_valid_for_move)
26513 (aarch64_simd_make_constant): Update call to
26514 aarch64_simd_valid_immediate.
26515 (aarch64_advsimd_valid_immediate_hs): New function.
26516 (aarch64_advsimd_valid_immediate): Likewise.
26517 (aarch64_simd_valid_immediate): Remove mode and inverse
26518 arguments. Rewrite to use the above. Use const_vec_duplicate_p
26519 to detect duplicated constants and use aarch64_float_const_zero_rtx_p
26520 and aarch64_float_const_representable_p on the result.
26521 (aarch64_output_simd_mov_immediate): Remove mode argument.
26522 Update call to aarch64_simd_valid_immediate and use of
26523 simd_immediate_info.
26524 (aarch64_output_scalar_simd_mov_immediate): Update call
26527 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
26528 Alan Hayward <alan.hayward@arm.com>
26529 David Sherwood <david.sherwood@arm.com>
26531 * machmode.h (mode_precision): Prefix with CONST_MODE_PRECISION.
26532 (mode_nunits): Likewise CONST_MODE_NUNITS.
26533 * machmode.def (ADJUST_NUNITS): Document.
26534 * genmodes.c (mode_data::need_nunits_adj): New field.
26535 (blank_mode): Update accordingly.
26536 (adj_nunits): New variable.
26537 (print_maybe_const_decl): Replace CATEGORY with a NEEDS_ADJ
26539 (emit_mode_size_inline): Set need_bytesize_adj for all modes
26540 listed in adj_nunits.
26541 (emit_mode_nunits_inline): Set need_nunits_adj for all modes
26542 listed in adj_nunits. Don't emit case statements for such modes.
26543 (emit_insn_modes_h): Emit definitions of CONST_MODE_NUNITS
26544 and CONST_MODE_PRECISION. Make CONST_MODE_SIZE expand to
26545 nothing if adj_nunits is nonnull.
26546 (emit_mode_precision, emit_mode_nunits): Use print_maybe_const_decl.
26547 (emit_mode_unit_size, emit_mode_base_align, emit_mode_ibit)
26548 (emit_mode_fbit): Update use of print_maybe_const_decl.
26549 (emit_move_size): Likewise. Treat the array as non-const
26551 (emit_mode_adjustments): Handle adj_nunits.
26553 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
26555 * machmode.def (VECTOR_MODES_WITH_PREFIX): Document.
26556 * genmodes.c (VECTOR_MODES_WITH_PREFIX): New macro.
26557 (VECTOR_MODES): Use it.
26558 (make_vector_modes): Take the prefix as an argument.
26560 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
26561 Alan Hayward <alan.hayward@arm.com>
26562 David Sherwood <david.sherwood@arm.com>
26564 * mode-classes.def (MODE_VECTOR_BOOL): New mode class.
26565 * machmode.h (INTEGRAL_MODE_P, VECTOR_MODE_P): Return true
26566 for MODE_VECTOR_BOOL.
26567 * machmode.def (VECTOR_BOOL_MODE): Document.
26568 * genmodes.c (VECTOR_BOOL_MODE): New macro.
26569 (make_vector_bool_mode): New function.
26570 (complete_mode, emit_mode_wider, emit_mode_adjustments): Handle
26572 * lto-streamer-in.c (lto_input_mode_table): Likewise.
26573 * rtx-vector-builder.c (rtx_vector_builder::find_cached_value):
26575 * stor-layout.c (int_mode_for_mode): Likewise.
26576 * tree.c (build_vector_type_for_mode): Likewise.
26577 * varasm.c (output_constant_pool_2): Likewise.
26578 * emit-rtl.c (init_emit_once): Make sure that CONST1_RTX (BImode) and
26579 CONSTM1_RTX (BImode) are the same thing. Initialize const_tiny_rtx
26580 for MODE_VECTOR_BOOL.
26581 * expr.c (expand_expr_real_1): Use VECTOR_MODE_P instead of a list
26582 of mode class checks.
26583 * tree-vect-generic.c (expand_vector_operation): Use VECTOR_MODE_P
26584 instead of a list of mode class checks.
26585 (expand_vector_scalar_condition): Likewise.
26586 (type_for_widest_vector_mode): Handle BImode as an inner mode.
26588 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
26589 Alan Hayward <alan.hayward@arm.com>
26590 David Sherwood <david.sherwood@arm.com>
26592 * machmode.h (mode_size): Change from unsigned short to
26594 (mode_to_bytes): Return a poly_uint16 rather than an unsigned short.
26595 (GET_MODE_SIZE): Return a constant if ONLY_FIXED_SIZE_MODES,
26596 or if measurement_type is not polynomial.
26597 (fixed_size_mode::includes_p): Check for constant-sized modes.
26598 * genmodes.c (emit_mode_size_inline): Make mode_size_inline
26599 return a poly_uint16 rather than an unsigned short.
26600 (emit_mode_size): Change the type of mode_size from unsigned short
26601 to poly_uint16_pod. Use ZERO_COEFFS for the initializer.
26602 (emit_mode_adjustments): Cope with polynomial vector sizes.
26603 * lto-streamer-in.c (lto_input_mode_table): Use bp_unpack_poly_value
26605 * lto-streamer-out.c (lto_write_mode_table): Use bp_pack_poly_value
26607 * auto-inc-dec.c (try_merge): Treat GET_MODE_SIZE as polynomial.
26608 * builtins.c (expand_ifn_atomic_compare_exchange_into_call): Likewise.
26609 * caller-save.c (setup_save_areas): Likewise.
26610 (replace_reg_with_saved_mem): Likewise.
26611 * calls.c (emit_library_call_value_1): Likewise.
26612 * combine-stack-adj.c (combine_stack_adjustments_for_block): Likewise.
26613 * combine.c (simplify_set, make_extraction, simplify_shift_const_1)
26614 (gen_lowpart_for_combine): Likewise.
26615 * convert.c (convert_to_integer_1): Likewise.
26616 * cse.c (equiv_constant, cse_insn): Likewise.
26617 * cselib.c (autoinc_split, cselib_hash_rtx): Likewise.
26618 (cselib_subst_to_values): Likewise.
26619 * dce.c (word_dce_process_block): Likewise.
26620 * df-problems.c (df_word_lr_mark_ref): Likewise.
26621 * dwarf2cfi.c (init_one_dwarf_reg_size): Likewise.
26622 * dwarf2out.c (multiple_reg_loc_descriptor, mem_loc_descriptor)
26623 (concat_loc_descriptor, concatn_loc_descriptor, loc_descriptor)
26624 (rtl_for_decl_location): Likewise.
26625 * emit-rtl.c (gen_highpart, widen_memory_access): Likewise.
26626 * expmed.c (extract_bit_field_1, extract_integral_bit_field): Likewise.
26627 * expr.c (emit_group_load_1, clear_storage_hints): Likewise.
26628 (emit_move_complex, emit_move_multi_word, emit_push_insn): Likewise.
26629 (expand_expr_real_1): Likewise.
26630 * function.c (assign_parm_setup_block_p, assign_parm_setup_block)
26631 (pad_below): Likewise.
26632 * gimple-fold.c (optimize_atomic_compare_exchange_p): Likewise.
26633 * gimple-ssa-store-merging.c (rhs_valid_for_store_merging_p): Likewise.
26634 * ira.c (get_subreg_tracking_sizes): Likewise.
26635 * ira-build.c (ira_create_allocno_objects): Likewise.
26636 * ira-color.c (coalesced_pseudo_reg_slot_compare): Likewise.
26637 (ira_sort_regnos_for_alter_reg): Likewise.
26638 * ira-costs.c (record_operand_costs): Likewise.
26639 * lower-subreg.c (interesting_mode_p, simplify_gen_subreg_concatn)
26640 (resolve_simple_move): Likewise.
26641 * lra-constraints.c (get_reload_reg, operands_match_p): Likewise.
26642 (process_addr_reg, simplify_operand_subreg, curr_insn_transform)
26643 (lra_constraints): Likewise.
26644 (CONST_POOL_OK_P): Reject variable-sized modes.
26645 * lra-spills.c (slot, assign_mem_slot, pseudo_reg_slot_compare)
26646 (add_pseudo_to_slot, lra_spill): Likewise.
26647 * omp-low.c (omp_clause_aligned_alignment): Likewise.
26648 * optabs-query.c (get_best_extraction_insn): Likewise.
26649 * optabs-tree.c (expand_vec_cond_expr_p): Likewise.
26650 * optabs.c (expand_vec_perm_var, expand_vec_cond_expr): Likewise.
26651 (expand_mult_highpart, valid_multiword_target_p): Likewise.
26652 * recog.c (offsettable_address_addr_space_p): Likewise.
26653 * regcprop.c (maybe_mode_change): Likewise.
26654 * reginfo.c (choose_hard_reg_mode, record_subregs_of_mode): Likewise.
26655 * regrename.c (build_def_use): Likewise.
26656 * regstat.c (dump_reg_info): Likewise.
26657 * reload.c (complex_word_subreg_p, push_reload, find_dummy_reload)
26658 (find_reloads, find_reloads_subreg_address): Likewise.
26659 * reload1.c (eliminate_regs_1): Likewise.
26660 * rtlanal.c (for_each_inc_dec_find_inc_dec, rtx_cost): Likewise.
26661 * simplify-rtx.c (avoid_constant_pool_reference): Likewise.
26662 (simplify_binary_operation_1, simplify_subreg): Likewise.
26663 * targhooks.c (default_function_arg_padding): Likewise.
26664 (default_hard_regno_nregs, default_class_max_nregs): Likewise.
26665 * tree-cfg.c (verify_gimple_assign_binary): Likewise.
26666 (verify_gimple_assign_ternary): Likewise.
26667 * tree-inline.c (estimate_move_cost): Likewise.
26668 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
26669 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Likewise.
26670 (get_address_cost_ainc): Likewise.
26671 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
26672 (vect_supportable_dr_alignment): Likewise.
26673 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
26674 (vectorizable_reduction): Likewise.
26675 * tree-vect-stmts.c (vectorizable_assignment, vectorizable_shift)
26676 (vectorizable_operation, vectorizable_load): Likewise.
26677 * tree.c (build_same_sized_truth_vector_type): Likewise.
26678 * valtrack.c (cleanup_auto_inc_dec): Likewise.
26679 * var-tracking.c (emit_note_insn_var_location): Likewise.
26680 * config/arc/arc.h (ASM_OUTPUT_CASE_END): Use as_a <scalar_int_mode>.
26681 (ADDR_VEC_ALIGN): Likewise.
26683 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
26684 Alan Hayward <alan.hayward@arm.com>
26685 David Sherwood <david.sherwood@arm.com>
26687 * machmode.h (mode_to_bits): Return a poly_uint16 rather than an
26689 (GET_MODE_BITSIZE): Return a constant if ONLY_FIXED_SIZE_MODES,
26690 or if measurement_type is polynomial.
26691 * calls.c (shift_return_value): Treat GET_MODE_BITSIZE as polynomial.
26692 * combine.c (make_extraction): Likewise.
26693 * dse.c (find_shift_sequence): Likewise.
26694 * dwarf2out.c (mem_loc_descriptor): Likewise.
26695 * expmed.c (store_integral_bit_field, extract_bit_field_1): Likewise.
26696 (extract_bit_field, extract_low_bits): Likewise.
26697 * expr.c (convert_move, convert_modes, emit_move_insn_1): Likewise.
26698 (optimize_bitfield_assignment_op, expand_assignment): Likewise.
26699 (store_expr_with_bounds, store_field, expand_expr_real_1): Likewise.
26700 * fold-const.c (optimize_bit_field_compare, merge_ranges): Likewise.
26701 * gimple-fold.c (optimize_atomic_compare_exchange_p): Likewise.
26702 * reload.c (find_reloads): Likewise.
26703 * reload1.c (alter_reg): Likewise.
26704 * stor-layout.c (bitwise_mode_for_mode, compute_record_mode): Likewise.
26705 * targhooks.c (default_secondary_memory_needed_mode): Likewise.
26706 * tree-if-conv.c (predicate_mem_writes): Likewise.
26707 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
26708 * tree-vect-patterns.c (adjust_bool_pattern): Likewise.
26709 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
26710 * valtrack.c (dead_debug_insert_temp): Likewise.
26711 * varasm.c (mergeable_constant_section): Likewise.
26712 * config/sh/sh.h (LOCAL_ALIGNMENT): Use as_a <fixed_size_mode>.
26714 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
26715 Alan Hayward <alan.hayward@arm.com>
26716 David Sherwood <david.sherwood@arm.com>
26718 * expr.c (expand_assignment): Cope with polynomial mode sizes
26719 when assigning to a CONCAT.
26721 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
26722 Alan Hayward <alan.hayward@arm.com>
26723 David Sherwood <david.sherwood@arm.com>
26725 * machmode.h (mode_precision): Change from unsigned short to
26727 (mode_to_precision): Return a poly_uint16 rather than an unsigned
26729 (GET_MODE_PRECISION): Return a constant if ONLY_FIXED_SIZE_MODES,
26730 or if measurement_type is not polynomial.
26731 (HWI_COMPUTABLE_MODE_P): Turn into a function. Optimize the case
26732 in which the mode is already known to be a scalar_int_mode.
26733 * genmodes.c (emit_mode_precision): Change the type of mode_precision
26734 from unsigned short to poly_uint16_pod. Use ZERO_COEFFS for the
26736 * lto-streamer-in.c (lto_input_mode_table): Use bp_unpack_poly_value
26737 for GET_MODE_PRECISION.
26738 * lto-streamer-out.c (lto_write_mode_table): Use bp_pack_poly_value
26739 for GET_MODE_PRECISION.
26740 * combine.c (update_rsp_from_reg_equal): Treat GET_MODE_PRECISION
26742 (try_combine, find_split_point, combine_simplify_rtx): Likewise.
26743 (expand_field_assignment, make_extraction): Likewise.
26744 (make_compound_operation_int, record_dead_and_set_regs_1): Likewise.
26745 (get_last_value): Likewise.
26746 * convert.c (convert_to_integer_1): Likewise.
26747 * cse.c (cse_insn): Likewise.
26748 * expr.c (expand_expr_real_1): Likewise.
26749 * lra-constraints.c (simplify_operand_subreg): Likewise.
26750 * optabs-query.c (can_atomic_load_p): Likewise.
26751 * optabs.c (expand_atomic_load): Likewise.
26752 (expand_atomic_store): Likewise.
26753 * ree.c (combine_reaching_defs): Likewise.
26754 * rtl.h (partial_subreg_p, paradoxical_subreg_p): Likewise.
26755 * rtlanal.c (nonzero_bits1, lsb_bitfield_op_p): Likewise.
26756 * tree.h (type_has_mode_precision_p): Likewise.
26757 * ubsan.c (instrument_si_overflow): Likewise.
26759 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
26760 Alan Hayward <alan.hayward@arm.com>
26761 David Sherwood <david.sherwood@arm.com>
26763 * tree.h (TYPE_VECTOR_SUBPARTS): Turn into a function and handle
26764 polynomial numbers of units.
26765 (SET_TYPE_VECTOR_SUBPARTS): Likewise.
26766 (valid_vector_subparts_p): New function.
26767 (build_vector_type): Remove temporary shim and take the number
26768 of units as a poly_uint64 rather than an int.
26769 (build_opaque_vector_type): Take the number of units as a
26770 poly_uint64 rather than an int.
26771 * tree.c (build_vector_from_ctor): Handle polynomial
26772 TYPE_VECTOR_SUBPARTS.
26773 (type_hash_canon_hash, type_cache_hasher::equal): Likewise.
26774 (uniform_vector_p, vector_type_mode, build_vector): Likewise.
26775 (build_vector_from_val): If the number of units is variable,
26776 use build_vec_duplicate_cst for constant operands and
26777 VEC_DUPLICATE_EXPR otherwise.
26778 (make_vector_type): Remove temporary is_constant ().
26779 (build_vector_type, build_opaque_vector_type): Take the number of
26780 units as a poly_uint64 rather than an int.
26781 (check_vector_cst): Handle polynomial TYPE_VECTOR_SUBPARTS and
26783 * cfgexpand.c (expand_debug_expr): Likewise.
26784 * expr.c (count_type_elements, categorize_ctor_elements_1): Likewise.
26785 (store_constructor, expand_expr_real_1): Likewise.
26786 (const_scalar_mask_from_tree): Likewise.
26787 * fold-const-call.c (fold_const_reduction): Likewise.
26788 * fold-const.c (const_binop, const_unop, fold_convert_const): Likewise.
26789 (operand_equal_p, fold_vec_perm, fold_ternary_loc): Likewise.
26790 (native_encode_vector, vec_cst_ctor_to_array): Likewise.
26791 (fold_relational_const): Likewise.
26792 (native_interpret_vector): Likewise. Change the size from an
26793 int to an unsigned int.
26794 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Handle polynomial
26795 TYPE_VECTOR_SUBPARTS.
26796 (gimple_fold_indirect_ref, gimple_build_vector): Likewise.
26797 (gimple_build_vector_from_val): Use VEC_DUPLICATE_EXPR when
26798 duplicating a non-constant operand into a variable-length vector.
26799 * hsa-brig.c (hsa_op_immed::emit_to_buffer): Handle polynomial
26800 TYPE_VECTOR_SUBPARTS and VECTOR_CST_NELTS.
26801 * ipa-icf.c (sem_variable::equals): Likewise.
26802 * match.pd: Likewise.
26803 * omp-simd-clone.c (simd_clone_subparts): Likewise.
26804 * print-tree.c (print_node): Likewise.
26805 * stor-layout.c (layout_type): Likewise.
26806 * targhooks.c (default_builtin_vectorization_cost): Likewise.
26807 * tree-cfg.c (verify_gimple_comparison): Likewise.
26808 (verify_gimple_assign_binary): Likewise.
26809 (verify_gimple_assign_ternary): Likewise.
26810 (verify_gimple_assign_single): Likewise.
26811 * tree-pretty-print.c (dump_generic_node): Likewise.
26812 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
26813 (simplify_bitfield_ref, is_combined_permutation_identity): Likewise.
26814 * tree-vect-data-refs.c (vect_permute_store_chain): Likewise.
26815 (vect_grouped_load_supported, vect_permute_load_chain): Likewise.
26816 (vect_shift_permute_load_chain): Likewise.
26817 * tree-vect-generic.c (nunits_for_known_piecewise_op): Likewise.
26818 (expand_vector_condition, optimize_vector_constructor): Likewise.
26819 (lower_vec_perm, get_compute_type): Likewise.
26820 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
26821 (get_initial_defs_for_reduction, vect_transform_loop): Likewise.
26822 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
26823 (vect_recog_mask_conversion_pattern): Likewise.
26824 * tree-vect-slp.c (vect_supported_load_permutation_p): Likewise.
26825 (vect_get_constant_vectors, vect_transform_slp_perm_load): Likewise.
26826 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
26827 (get_group_load_store_type, vectorizable_mask_load_store): Likewise.
26828 (vectorizable_bswap, simd_clone_subparts, vectorizable_assignment)
26829 (vectorizable_shift, vectorizable_operation, vectorizable_store)
26830 (vectorizable_load, vect_is_simple_cond, vectorizable_comparison)
26831 (supportable_widening_operation): Likewise.
26832 (supportable_narrowing_operation): Likewise.
26833 * tree-vector-builder.c (tree_vector_builder::binary_encoded_nelts):
26835 * varasm.c (output_constant): Likewise.
26837 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
26838 Alan Hayward <alan.hayward@arm.com>
26839 David Sherwood <david.sherwood@arm.com>
26841 * tree-vect-data-refs.c (vect_permute_store_chain): Reorganize
26842 so that both the length == 3 and length != 3 cases set up their
26843 own permute vectors. Add comments explaining why we know the
26844 number of elements is constant.
26845 (vect_permute_load_chain): Likewise.
26847 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
26848 Alan Hayward <alan.hayward@arm.com>
26849 David Sherwood <david.sherwood@arm.com>
26851 * machmode.h (mode_nunits): Change from unsigned char to
26853 (ONLY_FIXED_SIZE_MODES): New macro.
26854 (pod_mode::measurement_type, scalar_int_mode::measurement_type)
26855 (scalar_float_mode::measurement_type, scalar_mode::measurement_type)
26856 (complex_mode::measurement_type, fixed_size_mode::measurement_type):
26858 (mode_to_nunits): Return a poly_uint16 rather than an unsigned short.
26859 (GET_MODE_NUNITS): Return a constant if ONLY_FIXED_SIZE_MODES,
26860 or if measurement_type is not polynomial.
26861 * genmodes.c (ZERO_COEFFS): New macro.
26862 (emit_mode_nunits_inline): Make mode_nunits_inline return a
26864 (emit_mode_nunits): Change the type of mode_nunits to poly_uint16_pod.
26865 Use ZERO_COEFFS when emitting initializers.
26866 * data-streamer.h (bp_pack_poly_value): New function.
26867 (bp_unpack_poly_value): Likewise.
26868 * lto-streamer-in.c (lto_input_mode_table): Use bp_unpack_poly_value
26869 for GET_MODE_NUNITS.
26870 * lto-streamer-out.c (lto_write_mode_table): Use bp_pack_poly_value
26871 for GET_MODE_NUNITS.
26872 * tree.c (make_vector_type): Remove temporary shim and make
26873 the real function take the number of units as a poly_uint64
26874 rather than an int.
26875 (build_vector_type_for_mode): Handle polynomial nunits.
26876 * dwarf2out.c (loc_descriptor, add_const_value_attribute): Likewise.
26877 * emit-rtl.c (const_vec_series_p_1): Likewise.
26878 (gen_rtx_CONST_VECTOR): Likewise.
26879 * fold-const.c (test_vec_duplicate_folding): Likewise.
26880 * genrecog.c (validate_pattern): Likewise.
26881 * optabs-query.c (can_vec_perm_var_p, can_mult_highpart_p): Likewise.
26882 * optabs-tree.c (expand_vec_cond_expr_p): Likewise.
26883 * optabs.c (expand_vector_broadcast, expand_binop_directly): Likewise.
26884 (shift_amt_for_vec_perm_mask, expand_vec_perm_var): Likewise.
26885 (expand_vec_cond_expr, expand_mult_highpart): Likewise.
26886 * rtlanal.c (subreg_get_info): Likewise.
26887 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
26888 (vect_grouped_load_supported): Likewise.
26889 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
26890 * tree-vect-loop.c (have_whole_vector_shift): Likewise.
26891 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
26892 (simplify_const_unary_operation, simplify_binary_operation_1)
26893 (simplify_const_binary_operation, simplify_ternary_operation)
26894 (test_vector_ops_duplicate, test_vector_ops): Likewise.
26895 (simplify_immed_subreg): Use GET_MODE_NUNITS on a fixed_size_mode
26896 instead of CONST_VECTOR_NUNITS.
26897 * varasm.c (output_constant_pool_2): Likewise.
26898 * rtx-vector-builder.c (rtx_vector_builder::build): Only include the
26899 explicit-encoded elements in the XVEC for variable-length vectors.
26901 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
26903 * lra-constraints.c (curr_insn_transform): Use partial_subreg_p.
26905 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
26906 Alan Hayward <alan.hayward@arm.com>
26907 David Sherwood <david.sherwood@arm.com>
26909 * coretypes.h (fixed_size_mode): Declare.
26910 (fixed_size_mode_pod): New typedef.
26911 * builtins.h (target_builtins::x_apply_args_mode)
26912 (target_builtins::x_apply_result_mode): Change type to
26913 fixed_size_mode_pod.
26914 * builtins.c (apply_args_size, apply_result_size, result_vector)
26915 (expand_builtin_apply_args_1, expand_builtin_apply)
26916 (expand_builtin_return): Update accordingly.
26918 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
26920 * cse.c (hash_rtx_cb): Hash only the encoded elements.
26921 * cselib.c (cselib_hash_rtx): Likewise.
26922 * expmed.c (make_tree): Build VECTOR_CSTs directly from the
26923 CONST_VECTOR encoding.
26925 2018-01-03 Jakub Jelinek <jakub@redhat.com>
26926 Jeff Law <law@redhat.com>
26929 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): For
26930 noreturn probe, use gen_pop instead of ix86_emit_restore_reg_using_pop,
26931 only set RTX_FRAME_RELATED_P on both the push and pop if cfa_reg is sp
26932 and add REG_CFA_ADJUST_CFA notes in that case to both insns.
26935 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Do not
26936 explicitly probe *sp in a noreturn function if there were any callee
26937 register saves or frame pointer is needed.
26939 2018-01-03 Jakub Jelinek <jakub@redhat.com>
26942 * cfgexpand.c (expand_debug_expr): Return NULL if mode is
26943 BLKmode for ternary, binary or unary expressions.
26946 * var-tracking.c (delete_vta_debug_insn): New inline function.
26947 (delete_vta_debug_insns): Add USE_CFG argument, if true, walk just
26948 insns from get_insns () to NULL instead of each bb separately.
26949 Use delete_vta_debug_insn. No longer static.
26950 (vt_debug_insns_local, variable_tracking_main_1): Adjust
26951 delete_vta_debug_insns callers.
26952 * rtl.h (delete_vta_debug_insns): Declare.
26953 * final.c (rest_of_handle_final): Call delete_vta_debug_insns
26954 instead of variable_tracking_main.
26956 2018-01-03 Martin Sebor <msebor@redhat.com>
26958 PR tree-optimization/83603
26959 * calls.c (maybe_warn_nonstring_arg): Avoid accessing function
26960 arguments past the endof the argument list in functions declared
26961 without a prototype.
26962 * gimple-ssa-warn-restrict.c (wrestrict_dom_walker::check_call):
26963 Avoid checking when arguments are null.
26965 2018-01-03 Martin Sebor <msebor@redhat.com>
26968 * doc/extend.texi (attribute const): Fix a typo.
26969 * ipa-pure-const.c ((warn_function_const, warn_function_pure): Avoid
26970 issuing -Wsuggest-attribute for void functions.
26972 2018-01-03 Martin Sebor <msebor@redhat.com>
26974 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Use
26975 offset_int::from instead of wide_int::to_shwi.
26976 (maybe_diag_overlap): Remove assertion.
26977 Use HOST_WIDE_INT_PRINT_DEC instead of %lli.
26978 * gimple-ssa-sprintf.c (format_directive): Same.
26979 (parse_directive): Same.
26980 (sprintf_dom_walker::compute_format_length): Same.
26981 (try_substitute_return_value): Same.
26983 2018-01-03 Jeff Law <law@redhat.com>
26985 PR middle-end/83654
26986 * explow.c (anti_adjust_stack_and_probe_stack_clash): Test a
26987 non-constant residual for zero at runtime and avoid probing in
26988 that case. Reorganize code for trailing problem to mirror handling
26991 2018-01-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
26993 PR tree-optimization/83501
26994 * tree-ssa-strlen.c (get_string_cst): New.
26995 (handle_char_store): Call get_string_cst.
26997 2018-01-03 Martin Liska <mliska@suse.cz>
26999 PR tree-optimization/83593
27000 * tree-ssa-strlen.c: Include tree-cfg.h.
27001 (strlen_check_and_optimize_stmt): Add new argument cleanup_eh.
27002 (strlen_dom_walker): Add new member variable m_cleanup_cfg.
27003 (strlen_dom_walker::strlen_dom_walker): Initialize m_cleanup_cfg
27005 (strlen_dom_walker::before_dom_children): Call
27006 gimple_purge_dead_eh_edges. Dump tranformation with details
27008 (strlen_dom_walker::before_dom_children): Update call by adding
27009 new argument cleanup_eh.
27010 (pass_strlen::execute): Return TODO_cleanup_cfg if needed.
27012 2018-01-03 Martin Liska <mliska@suse.cz>
27015 * cif-code.def (VARIADIC_THUNK): New enum value.
27016 * ipa-fnsummary.c (compute_fn_summary): Do not inline variadic
27019 2018-01-03 Jan Beulich <jbeulich@suse.com>
27021 * sse.md (mov<mode>_internal): Tighten condition for when to use
27022 vmovdqu<ssescalarsize> for TI and OI modes.
27024 2018-01-03 Jakub Jelinek <jakub@redhat.com>
27026 Update copyright years.
27028 2018-01-03 Martin Liska <mliska@suse.cz>
27031 * ipa-visibility.c (function_and_variable_visibility): Skip
27032 functions with noipa attribure.
27034 2018-01-03 Jakub Jelinek <jakub@redhat.com>
27036 * gcc.c (process_command): Update copyright notice dates.
27037 * gcov-dump.c (print_version): Ditto.
27038 * gcov.c (print_version): Ditto.
27039 * gcov-tool.c (print_version): Ditto.
27040 * gengtype.c (create_file): Ditto.
27041 * doc/cpp.texi: Bump @copying's copyright year.
27042 * doc/cppinternals.texi: Ditto.
27043 * doc/gcc.texi: Ditto.
27044 * doc/gccint.texi: Ditto.
27045 * doc/gcov.texi: Ditto.
27046 * doc/install.texi: Ditto.
27047 * doc/invoke.texi: Ditto.
27049 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
27051 * vector-builder.h (vector_builder::m_full_nelts): Change from
27052 unsigned int to poly_uint64.
27053 (vector_builder::full_nelts): Update prototype accordingly.
27054 (vector_builder::new_vector): Likewise.
27055 (vector_builder::encoded_full_vector_p): Handle polynomial full_nelts.
27056 (vector_builder::operator ==): Likewise.
27057 (vector_builder::finalize): Likewise.
27058 * int-vector-builder.h (int_vector_builder::int_vector_builder):
27059 Take the number of elements as a poly_uint64 rather than an
27061 * vec-perm-indices.h (vec_perm_indices::m_nelts_per_input): Change
27062 from unsigned int to poly_uint64.
27063 (vec_perm_indices::vec_perm_indices): Update prototype accordingly.
27064 (vec_perm_indices::new_vector): Likewise.
27065 (vec_perm_indices::length): Likewise.
27066 (vec_perm_indices::nelts_per_input): Likewise.
27067 (vec_perm_indices::input_nelts): Likewise.
27068 * vec-perm-indices.c (vec_perm_indices::new_vector): Take the
27069 number of elements per input as a poly_uint64 rather than an
27070 unsigned int. Use the original encoding for variable-length
27071 vectors, rather than clamping each individual element.
27072 For the second and subsequent elements in each pattern,
27073 clamp the step and base before clamping their sum.
27074 (vec_perm_indices::series_p): Handle polynomial element counts.
27075 (vec_perm_indices::all_in_range_p): Likewise.
27076 (vec_perm_indices_to_tree): Likewise.
27077 (vec_perm_indices_to_rtx): Likewise.
27078 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
27079 * tree-vector-builder.c (tree_vector_builder::new_unary_operation)
27080 (tree_vector_builder::new_binary_operation): Handle polynomial
27081 element counts. Return false if we need to know the number
27082 of elements at compile time.
27083 * fold-const.c (fold_vec_perm): Punt if the number of elements
27084 isn't known at compile time.
27086 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
27088 * vec-perm-indices.h (vec_perm_builder): Change element type
27089 from HOST_WIDE_INT to poly_int64.
27090 (vec_perm_indices::element_type): Update accordingly.
27091 (vec_perm_indices::clamp): Handle polynomial element_types.
27092 * vec-perm-indices.c (vec_perm_indices::series_p): Likewise.
27093 (vec_perm_indices::all_in_range_p): Likewise.
27094 (tree_to_vec_perm_builder): Check for poly_int64 trees rather
27096 * vector-builder.h (vector_builder::stepped_sequence_p): Handle
27097 polynomial vec_perm_indices element types.
27098 * int-vector-builder.h (int_vector_builder::equal_p): Likewise.
27099 * fold-const.c (fold_vec_perm): Likewise.
27100 * optabs.c (shift_amt_for_vec_perm_mask): Likewise.
27101 * tree-vect-generic.c (lower_vec_perm): Likewise.
27102 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
27103 * config/aarch64/aarch64.c (aarch64_evpc_tbl): Cast d->perm
27104 element type to HOST_WIDE_INT.
27106 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
27107 Alan Hayward <alan.hayward@arm.com>
27108 David Sherwood <david.sherwood@arm.com>
27110 * alias.c (addr_side_effect_eval): Take the size as a poly_int64
27111 rather than an int. Use plus_constant.
27112 (memrefs_conflict_p): Take the sizes as poly_int64s rather than ints.
27113 Take the offset "c" as a poly_int64 rather than a HOST_WIDE_INT.
27115 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
27116 Alan Hayward <alan.hayward@arm.com>
27117 David Sherwood <david.sherwood@arm.com>
27119 * calls.c (emit_call_1, expand_call): Change struct_value_size from
27120 a HOST_WIDE_INT to a poly_int64.
27122 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
27123 Alan Hayward <alan.hayward@arm.com>
27124 David Sherwood <david.sherwood@arm.com>
27126 * calls.c (load_register_parameters): Cope with polynomial
27127 mode sizes. Require a constant size for BLKmode parameters
27128 that aren't described by a PARALLEL. If BLOCK_REG_PADDING
27129 forces a parameter to be padded at the lsb end in order to
27130 fill a complete number of words, require the parameter size
27131 to be ordered wrt UNITS_PER_WORD.
27133 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
27134 Alan Hayward <alan.hayward@arm.com>
27135 David Sherwood <david.sherwood@arm.com>
27137 * reload1.c (spill_stack_slot_width): Change element type
27138 from unsigned int to poly_uint64_pod.
27139 (alter_reg): Treat mode sizes as polynomial.
27141 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
27142 Alan Hayward <alan.hayward@arm.com>
27143 David Sherwood <david.sherwood@arm.com>
27145 * reload.c (complex_word_subreg_p): New function.
27146 (reload_inner_reg_of_subreg, push_reload): Use it.
27148 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
27149 Alan Hayward <alan.hayward@arm.com>
27150 David Sherwood <david.sherwood@arm.com>
27152 * lra-constraints.c (process_alt_operands): Reject matched
27153 operands whose sizes aren't ordered.
27154 (match_reload): Refer to this check here.
27156 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
27157 Alan Hayward <alan.hayward@arm.com>
27158 David Sherwood <david.sherwood@arm.com>
27160 * builtins.c (expand_ifn_atomic_compare_exchange_into_call): Assert
27161 that the mode size is in the set {1, 2, 4, 8, 16}.
27163 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
27164 Alan Hayward <alan.hayward@arm.com>
27165 David Sherwood <david.sherwood@arm.com>
27167 * var-tracking.c (adjust_mems): Treat mode sizes as polynomial.
27168 Use plus_constant instead of gen_rtx_PLUS.
27170 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
27171 Alan Hayward <alan.hayward@arm.com>
27172 David Sherwood <david.sherwood@arm.com>
27174 * config/cr16/cr16-protos.h (cr16_push_rounding): Declare.
27175 * config/cr16/cr16.h (PUSH_ROUNDING): Move implementation to...
27176 * config/cr16/cr16.c (cr16_push_rounding): ...this new function.
27177 * config/h8300/h8300-protos.h (h8300_push_rounding): Declare.
27178 * config/h8300/h8300.h (PUSH_ROUNDING): Move implementation to...
27179 * config/h8300/h8300.c (h8300_push_rounding): ...this new function.
27180 * config/i386/i386-protos.h (ix86_push_rounding): Declare.
27181 * config/i386/i386.h (PUSH_ROUNDING): Move implementation to...
27182 * config/i386/i386.c (ix86_push_rounding): ...this new function.
27183 * config/m32c/m32c-protos.h (m32c_push_rounding): Take and return
27185 * config/m32c/m32c.c (m32c_push_rounding): Likewise.
27186 * config/m68k/m68k-protos.h (m68k_push_rounding): Declare.
27187 * config/m68k/m68k.h (PUSH_ROUNDING): Move implementation to...
27188 * config/m68k/m68k.c (m68k_push_rounding): ...this new function.
27189 * config/pdp11/pdp11-protos.h (pdp11_push_rounding): Declare.
27190 * config/pdp11/pdp11.h (PUSH_ROUNDING): Move implementation to...
27191 * config/pdp11/pdp11.c (pdp11_push_rounding): ...this new function.
27192 * config/stormy16/stormy16-protos.h (xstormy16_push_rounding): Declare.
27193 * config/stormy16/stormy16.h (PUSH_ROUNDING): Move implementation to...
27194 * config/stormy16/stormy16.c (xstormy16_push_rounding): ...this new
27196 * expr.c (emit_move_resolve_push): Treat the input and result
27197 of PUSH_ROUNDING as a poly_int64.
27198 (emit_move_complex_push, emit_single_push_insn_1): Likewise.
27199 (emit_push_insn): Likewise.
27200 * lra-eliminations.c (mark_not_eliminable): Likewise.
27201 * recog.c (push_operand): Likewise.
27202 * reload1.c (elimination_effects): Likewise.
27203 * rtlanal.c (nonzero_bits1): Likewise.
27204 * calls.c (store_one_arg): Likewise. Require the padding to be
27205 known at compile time.
27207 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
27208 Alan Hayward <alan.hayward@arm.com>
27209 David Sherwood <david.sherwood@arm.com>
27211 * expr.c (emit_single_push_insn_1): Treat mode sizes as polynomial.
27212 Use plus_constant instead of gen_rtx_PLUS.
27214 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
27215 Alan Hayward <alan.hayward@arm.com>
27216 David Sherwood <david.sherwood@arm.com>
27218 * auto-inc-dec.c (set_inc_state): Take the mode size as a poly_int64
27219 rather than an int.
27221 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
27222 Alan Hayward <alan.hayward@arm.com>
27223 David Sherwood <david.sherwood@arm.com>
27225 * expr.c (expand_expr_real_1): Use tree_to_poly_uint64
27226 instead of int_size_in_bytes when handling VIEW_CONVERT_EXPRs
27227 via stack temporaries. Treat the mode size as polynomial too.
27229 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
27230 Alan Hayward <alan.hayward@arm.com>
27231 David Sherwood <david.sherwood@arm.com>
27233 * expr.c (expand_expr_real_2): When handling conversions involving
27234 unions, apply tree_to_poly_uint64 to the TYPE_SIZE rather than
27235 multiplying int_size_in_bytes by BITS_PER_UNIT. Treat GET_MODE_BISIZE
27236 as a poly_uint64 too.
27238 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
27239 Alan Hayward <alan.hayward@arm.com>
27240 David Sherwood <david.sherwood@arm.com>
27242 * rtlanal.c (subreg_get_info): Handle polynomial mode sizes.
27244 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
27245 Alan Hayward <alan.hayward@arm.com>
27246 David Sherwood <david.sherwood@arm.com>
27248 * combine.c (can_change_dest_mode): Handle polynomial
27249 REGMODE_NATURAL_SIZE.
27250 * expmed.c (store_bit_field_1): Likewise.
27251 * expr.c (store_constructor): Likewise.
27252 * emit-rtl.c (validate_subreg): Operate on polynomial mode sizes
27253 and polynomial REGMODE_NATURAL_SIZE.
27254 (gen_lowpart_common): Likewise.
27255 * reginfo.c (record_subregs_of_mode): Likewise.
27256 * rtlanal.c (read_modify_subreg_p): Likewise.
27258 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
27259 Alan Hayward <alan.hayward@arm.com>
27260 David Sherwood <david.sherwood@arm.com>
27262 * internal-fn.c (expand_vector_ubsan_overflow): Handle polynomial
27263 numbers of elements.
27265 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
27266 Alan Hayward <alan.hayward@arm.com>
27267 David Sherwood <david.sherwood@arm.com>
27269 * match.pd: Cope with polynomial numbers of vector elements.
27271 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
27272 Alan Hayward <alan.hayward@arm.com>
27273 David Sherwood <david.sherwood@arm.com>
27275 * fold-const.c (fold_indirect_ref_1): Handle polynomial offsets
27276 in a POINTER_PLUS_EXPR.
27278 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
27279 Alan Hayward <alan.hayward@arm.com>
27280 David Sherwood <david.sherwood@arm.com>
27282 * omp-simd-clone.c (simd_clone_subparts): New function.
27283 (simd_clone_init_simd_arrays): Use it instead of TYPE_VECTOR_SUBPARTS.
27284 (ipa_simd_modify_function_body): Likewise.
27286 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
27287 Alan Hayward <alan.hayward@arm.com>
27288 David Sherwood <david.sherwood@arm.com>
27290 * tree-vect-generic.c (nunits_for_known_piecewise_op): New function.
27291 (expand_vector_piecewise): Use it instead of TYPE_VECTOR_SUBPARTS.
27292 (expand_vector_addition, add_rshift, expand_vector_divmod): Likewise.
27293 (expand_vector_condition, vector_element): Likewise.
27294 (subparts_gt): New function.
27295 (get_compute_type): Use subparts_gt.
27296 (count_type_subparts): Delete.
27297 (expand_vector_operations_1): Use subparts_gt instead of
27298 count_type_subparts.
27300 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
27301 Alan Hayward <alan.hayward@arm.com>
27302 David Sherwood <david.sherwood@arm.com>
27304 * tree-vect-data-refs.c (vect_no_alias_p): Replace with...
27305 (vect_compile_time_alias): ...this new function. Do the calculation
27306 on poly_ints rather than trees.
27307 (vect_prune_runtime_alias_test_list): Update call accordingly.
27309 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
27310 Alan Hayward <alan.hayward@arm.com>
27311 David Sherwood <david.sherwood@arm.com>
27313 * tree-vect-slp.c (vect_build_slp_tree_1): Handle polynomial
27315 (vect_schedule_slp_instance): Likewise.
27317 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
27318 Alan Hayward <alan.hayward@arm.com>
27319 David Sherwood <david.sherwood@arm.com>
27321 * tree-vect-slp.c (vect_get_and_check_slp_defs): Reject
27322 constant and extern definitions for variable-length vectors.
27323 (vect_get_constant_vectors): Note that the number of units
27324 is known to be constant.
27326 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
27327 Alan Hayward <alan.hayward@arm.com>
27328 David Sherwood <david.sherwood@arm.com>
27330 * tree-vect-stmts.c (vectorizable_conversion): Treat the number
27331 of units as polynomial. Choose between WIDE and NARROW based
27334 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
27335 Alan Hayward <alan.hayward@arm.com>
27336 David Sherwood <david.sherwood@arm.com>
27338 * tree-vect-stmts.c (simd_clone_subparts): New function.
27339 (vectorizable_simd_clone_call): Use it instead of TYPE_VECTOR_SUBPARTS.
27341 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
27342 Alan Hayward <alan.hayward@arm.com>
27343 David Sherwood <david.sherwood@arm.com>
27345 * tree-vect-stmts.c (vectorizable_call): Treat the number of
27346 vectors as polynomial. Use build_index_vector for
27347 IFN_GOMP_SIMD_LANE.
27349 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
27350 Alan Hayward <alan.hayward@arm.com>
27351 David Sherwood <david.sherwood@arm.com>
27353 * tree-vect-stmts.c (get_load_store_type): Treat the number of
27354 units as polynomial. Reject VMAT_ELEMENTWISE and VMAT_STRIDED_SLP
27355 for variable-length vectors.
27356 (vectorizable_mask_load_store): Treat the number of units as
27357 polynomial, asserting that it is constant if the condition has
27358 already been enforced.
27359 (vectorizable_store, vectorizable_load): Likewise.
27361 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
27362 Alan Hayward <alan.hayward@arm.com>
27363 David Sherwood <david.sherwood@arm.com>
27365 * tree-vect-loop.c (vectorizable_live_operation): Treat the number
27366 of units as polynomial. Punt if we can't tell at compile time
27367 which vector contains the final result.
27369 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
27370 Alan Hayward <alan.hayward@arm.com>
27371 David Sherwood <david.sherwood@arm.com>
27373 * tree-vect-loop.c (vectorizable_induction): Treat the number
27374 of units as polynomial. Punt on SLP inductions. Use an integer
27375 VEC_SERIES_EXPR for variable-length integer reductions. Use a
27376 cast of such a series for variable-length floating-point
27379 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
27380 Alan Hayward <alan.hayward@arm.com>
27381 David Sherwood <david.sherwood@arm.com>
27383 * tree.h (build_index_vector): Declare.
27384 * tree.c (build_index_vector): New function.
27385 * tree-vect-loop.c (get_initial_defs_for_reduction): Treat the number
27386 of units as polynomial, forcibly converting it to a constant if
27387 vectorizable_reduction has already enforced the condition.
27388 (vect_create_epilog_for_reduction): Likewise. Use build_index_vector
27389 to create a {1,2,3,...} vector.
27390 (vectorizable_reduction): Treat the number of units as polynomial.
27391 Choose vectype_in based on the largest scalar element size rather
27392 than the smallest number of units. Enforce the restrictions
27395 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
27396 Alan Hayward <alan.hayward@arm.com>
27397 David Sherwood <david.sherwood@arm.com>
27399 * tree-vect-data-refs.c (vector_alignment_reachable_p): Treat the
27400 number of units as polynomial.
27402 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
27403 Alan Hayward <alan.hayward@arm.com>
27404 David Sherwood <david.sherwood@arm.com>
27406 * target.h (vector_sizes, auto_vector_sizes): New typedefs.
27407 * target.def (autovectorize_vector_sizes): Return the vector sizes
27408 by pointer, using vector_sizes rather than a bitmask.
27409 * targhooks.h (default_autovectorize_vector_sizes): Update accordingly.
27410 * targhooks.c (default_autovectorize_vector_sizes): Likewise.
27411 * config/aarch64/aarch64.c (aarch64_autovectorize_vector_sizes):
27413 * config/arc/arc.c (arc_autovectorize_vector_sizes): Likewise.
27414 * config/arm/arm.c (arm_autovectorize_vector_sizes): Likewise.
27415 * config/i386/i386.c (ix86_autovectorize_vector_sizes): Likewise.
27416 * config/mips/mips.c (mips_autovectorize_vector_sizes): Likewise.
27417 * omp-general.c (omp_max_vf): Likewise.
27418 * omp-low.c (omp_clause_aligned_alignment): Likewise.
27419 * optabs-query.c (can_vec_mask_load_store_p): Likewise.
27420 * tree-vect-loop.c (vect_analyze_loop): Likewise.
27421 * tree-vect-slp.c (vect_slp_bb): Likewise.
27422 * doc/tm.texi: Regenerate.
27423 * tree-vectorizer.h (current_vector_size): Change from an unsigned int
27425 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Take
27426 the vector size as a poly_uint64 rather than an unsigned int.
27427 (current_vector_size): Change from an unsigned int to a poly_uint64.
27428 (get_vectype_for_scalar_type): Update accordingly.
27429 * tree.h (build_truth_vector_type): Take the size and number of
27430 units as a poly_uint64 rather than an unsigned int.
27431 (build_vector_type): Add a temporary overload that takes
27432 the number of units as a poly_uint64 rather than an unsigned int.
27433 * tree.c (make_vector_type): Likewise.
27434 (build_truth_vector_type): Take the number of units as a poly_uint64
27435 rather than an unsigned int.
27437 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
27438 Alan Hayward <alan.hayward@arm.com>
27439 David Sherwood <david.sherwood@arm.com>
27441 * target.def (get_mask_mode): Take the number of units and length
27442 as poly_uint64s rather than unsigned ints.
27443 * targhooks.h (default_get_mask_mode): Update accordingly.
27444 * targhooks.c (default_get_mask_mode): Likewise.
27445 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
27446 * doc/tm.texi: Regenerate.
27448 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
27449 Alan Hayward <alan.hayward@arm.com>
27450 David Sherwood <david.sherwood@arm.com>
27452 * omp-general.h (omp_max_vf): Return a poly_uint64 instead of an int.
27453 * omp-general.c (omp_max_vf): Likewise.
27454 * omp-expand.c (omp_adjust_chunk_size): Update call to omp_max_vf.
27455 (expand_omp_simd): Handle polynomial safelen.
27456 * omp-low.c (omplow_simd_context): Add a default constructor.
27457 (omplow_simd_context::max_vf): Change from int to poly_uint64.
27458 (lower_rec_simd_input_clauses): Update accordingly.
27459 (lower_rec_input_clauses): Likewise.
27461 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
27462 Alan Hayward <alan.hayward@arm.com>
27463 David Sherwood <david.sherwood@arm.com>
27465 * tree-vectorizer.h (vect_nunits_for_cost): New function.
27466 * tree-vect-loop.c (vect_model_reduction_cost): Use it.
27467 * tree-vect-slp.c (vect_analyze_slp_cost_1): Likewise.
27468 (vect_analyze_slp_cost): Likewise.
27469 * tree-vect-stmts.c (vect_model_store_cost): Likewise.
27470 (vect_model_load_cost): Likewise.
27472 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
27473 Alan Hayward <alan.hayward@arm.com>
27474 David Sherwood <david.sherwood@arm.com>
27476 * tree-vect-slp.c (vect_record_max_nunits, vect_build_slp_tree_1)
27477 (vect_build_slp_tree_2, vect_build_slp_tree): Change max_nunits
27478 from an unsigned int * to a poly_uint64_pod *.
27479 (calculate_unrolling_factor): New function.
27480 (vect_analyze_slp_instance): Use it. Track polynomial max_nunits.
27482 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
27483 Alan Hayward <alan.hayward@arm.com>
27484 David Sherwood <david.sherwood@arm.com>
27486 * tree-vectorizer.h (_slp_instance::unrolling_factor): Change
27487 from an unsigned int to a poly_uint64.
27488 (_loop_vec_info::slp_unrolling_factor): Likewise.
27489 (_loop_vec_info::vectorization_factor): Change from an int
27491 (MAX_VECTORIZATION_FACTOR): Bump from 64 to INT_MAX.
27492 (vect_get_num_vectors): New function.
27493 (vect_update_max_nunits, vect_vf_for_cost): Likewise.
27494 (vect_get_num_copies): Use vect_get_num_vectors.
27495 (vect_analyze_data_ref_dependences): Change max_vf from an int *
27496 to an unsigned int *.
27497 (vect_analyze_data_refs): Change min_vf from an int * to a
27499 (vect_transform_slp_perm_load): Take the vf as a poly_uint64 rather
27500 than an unsigned HOST_WIDE_INT.
27501 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr)
27502 (vect_analyze_data_ref_dependence): Change max_vf from an int *
27503 to an unsigned int *.
27504 (vect_analyze_data_ref_dependences): Likewise.
27505 (vect_compute_data_ref_alignment): Handle polynomial vf.
27506 (vect_enhance_data_refs_alignment): Likewise.
27507 (vect_prune_runtime_alias_test_list): Likewise.
27508 (vect_shift_permute_load_chain): Likewise.
27509 (vect_supportable_dr_alignment): Likewise.
27510 (dependence_distance_ge_vf): Take the vectorization factor as a
27511 poly_uint64 rather than an unsigned HOST_WIDE_INT.
27512 (vect_analyze_data_refs): Change min_vf from an int * to a
27514 * tree-vect-loop-manip.c (vect_gen_scalar_loop_niters): Take
27515 vfm1 as a poly_uint64 rather than an int. Make the same change
27516 for the returned bound_scalar.
27517 (vect_gen_vector_loop_niters): Handle polynomial vf.
27518 (vect_do_peeling): Likewise. Update call to
27519 vect_gen_scalar_loop_niters and handle polynomial bound_scalars.
27520 (vect_gen_vector_loop_niters_mult_vf): Assert that the vf must
27522 * tree-vect-loop.c (vect_determine_vectorization_factor)
27523 (vect_update_vf_for_slp, vect_analyze_loop_2): Handle polynomial vf.
27524 (vect_get_known_peeling_cost): Likewise.
27525 (vect_estimate_min_profitable_iters, vectorizable_reduction): Likewise.
27526 (vect_worthwhile_without_simd_p, vectorizable_induction): Likewise.
27527 (vect_transform_loop): Likewise. Use the lowest possible VF when
27528 updating the upper bounds of the loop.
27529 (vect_min_worthwhile_factor): Make static. Return an unsigned int
27530 rather than an int.
27531 * tree-vect-slp.c (vect_attempt_slp_rearrange_stmts): Cope with
27532 polynomial unroll factors.
27533 (vect_analyze_slp_cost_1, vect_analyze_slp_instance): Likewise.
27534 (vect_make_slp_decision): Likewise.
27535 (vect_supported_load_permutation_p): Likewise, and polynomial
27537 (vect_analyze_slp_cost): Handle polynomial vf.
27538 (vect_slp_analyze_node_operations): Likewise.
27539 (vect_slp_analyze_bb_1): Likewise.
27540 (vect_transform_slp_perm_load): Take the vf as a poly_uint64 rather
27541 than an unsigned HOST_WIDE_INT.
27542 * tree-vect-stmts.c (vectorizable_simd_clone_call, vectorizable_store)
27543 (vectorizable_load): Handle polynomial vf.
27544 * tree-vectorizer.c (simduid_to_vf::vf): Change from an int to
27546 (adjust_simduid_builtins, shrink_simd_arrays): Update accordingly.
27548 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
27549 Alan Hayward <alan.hayward@arm.com>
27550 David Sherwood <david.sherwood@arm.com>
27552 * match.pd: Handle bit operations involving three constants
27553 and try to fold one pair.
27555 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
27557 * tree-vect-loop-manip.c: Include gimple-fold.h.
27558 (slpeel_make_loop_iterate_ntimes): Add step, final_iv and
27559 niters_maybe_zero parameters. Handle other cases besides a step of 1.
27560 (vect_gen_vector_loop_niters): Add a step_vector_ptr parameter.
27561 Add a path that uses a step of VF instead of 1, but disable it
27563 (vect_do_peeling): Add step_vector, niters_vector_mult_vf_var
27564 and niters_no_overflow parameters. Update calls to
27565 slpeel_make_loop_iterate_ntimes and vect_gen_vector_loop_niters.
27566 Create a new SSA name if the latter choses to use a ste other
27567 than zero, and return it via niters_vector_mult_vf_var.
27568 * tree-vect-loop.c (vect_transform_loop): Update calls to
27569 vect_do_peeling, vect_gen_vector_loop_niters and
27570 slpeel_make_loop_iterate_ntimes.
27571 * tree-vectorizer.h (slpeel_make_loop_iterate_ntimes, vect_do_peeling)
27572 (vect_gen_vector_loop_niters): Update declarations after above changes.
27574 2018-01-02 Michael Meissner <meissner@linux.vnet.ibm.com>
27576 * config/rs6000/rs6000.md (floor<mode>2): Add support for IEEE
27577 128-bit round to integer instructions.
27578 (ceil<mode>2): Likewise.
27579 (btrunc<mode>2): Likewise.
27580 (round<mode>2): Likewise.
27582 2018-01-02 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27584 * config/rs6000/rs6000-string.c (expand_block_move): Allow the use of
27585 unaligned VSX load/store on P8/P9.
27586 (expand_block_clear): Allow the use of unaligned VSX
27587 load/store on P8/P9.
27589 2018-01-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27591 * config/rs6000/rs6000-p8swap.c (swap_feeds_both_load_and_store):
27593 (rs6000_analyze_swaps): Mark a web unoptimizable if it contains a
27594 swap associated with both a load and a store.
27596 2018-01-02 Andrew Waterman <andrew@sifive.com>
27598 * config/riscv/linux.h (ICACHE_FLUSH_FUNC): New.
27599 * config/riscv/riscv.md (clear_cache): Use it.
27601 2018-01-02 Artyom Skrobov <tyomitch@gmail.com>
27603 * web.c: Remove out-of-date comment.
27605 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
27607 * expr.c (fixup_args_size_notes): Check that any existing
27608 REG_ARGS_SIZE notes are correct, and don't try to re-add them.
27609 (emit_single_push_insn_1): Move stack_pointer_delta adjustment to...
27610 (emit_single_push_insn): ...here.
27612 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
27614 * rtl.h (CONST_VECTOR_ELT): Redefine to const_vector_elt.
27615 (const_vector_encoded_nelts): New function.
27616 (CONST_VECTOR_NUNITS): Redefine to use GET_MODE_NUNITS.
27617 (const_vector_int_elt, const_vector_elt): Declare.
27618 * emit-rtl.c (const_vector_int_elt_1): New function.
27619 (const_vector_elt): Likewise.
27620 * simplify-rtx.c (simplify_immed_subreg): Avoid taking the address
27621 of CONST_VECTOR_ELT.
27623 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
27625 * expr.c: Include rtx-vector-builder.h.
27626 (const_vector_mask_from_tree): Use rtx_vector_builder and operate
27627 directly on the tree encoding.
27628 (const_vector_from_tree): Likewise.
27629 * optabs.c: Include rtx-vector-builder.h.
27630 (expand_vec_perm_var): Use rtx_vector_builder and create a repeating
27631 sequence of "u" values.
27632 * vec-perm-indices.c: Include rtx-vector-builder.h.
27633 (vec_perm_indices_to_rtx): Use rtx_vector_builder and operate
27634 directly on the vec_perm_indices encoding.
27636 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
27638 * doc/rtl.texi (const_vector): Describe new encoding scheme.
27639 * Makefile.in (OBJS): Add rtx-vector-builder.o.
27640 * rtx-vector-builder.h: New file.
27641 * rtx-vector-builder.c: Likewise.
27642 * rtl.h (rtx_def::u2): Add a const_vector field.
27643 (CONST_VECTOR_NPATTERNS): New macro.
27644 (CONST_VECTOR_NELTS_PER_PATTERN): Likewise.
27645 (CONST_VECTOR_DUPLICATE_P): Likewise.
27646 (CONST_VECTOR_STEPPED_P): Likewise.
27647 (CONST_VECTOR_ENCODED_ELT): Likewise.
27648 (const_vec_duplicate_p): Check for a duplicated vector encoding.
27649 (unwrap_const_vec_duplicate): Likewise.
27650 (const_vec_series_p): Check for a non-duplicated vector encoding.
27651 Say that the function only returns true for integer vectors.
27652 * emit-rtl.c: Include rtx-vector-builder.h.
27653 (gen_const_vec_duplicate_1): Delete.
27654 (gen_const_vector): Call gen_const_vec_duplicate instead of
27655 gen_const_vec_duplicate_1.
27656 (const_vec_series_p_1): Operate directly on the CONST_VECTOR encoding.
27657 (gen_const_vec_duplicate): Use rtx_vector_builder.
27658 (gen_const_vec_series): Likewise.
27659 (gen_rtx_CONST_VECTOR): Likewise.
27660 * config/powerpcspe/powerpcspe.c: Include rtx-vector-builder.h.
27661 (swap_const_vector_halves): Take an rtx pointer rather than rtx.
27662 Build a new vector rather than modifying a CONST_VECTOR in-place.
27663 (handle_special_swappables): Update call accordingly.
27664 * config/rs6000/rs6000-p8swap.c: Include rtx-vector-builder.h.
27665 (swap_const_vector_halves): Take an rtx pointer rather than rtx.
27666 Build a new vector rather than modifying a CONST_VECTOR in-place.
27667 (handle_special_swappables): Update call accordingly.
27669 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
27671 * simplify-rtx.c (simplify_const_binary_operation): Use
27672 CONST_VECTOR_ELT instead of XVECEXP.
27674 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
27676 * tree-cfg.c (verify_gimple_assign_ternary): Allow the size of
27677 the selector elements to be different from the data elements
27678 if the selector is a VECTOR_CST.
27679 * tree-vect-stmts.c (vect_gen_perm_mask_any): Use a vector of
27680 ssizetype for the selector.
27682 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
27684 * optabs.c (shift_amt_for_vec_perm_mask): Try using series_p
27685 before testing each element individually.
27686 * tree-vect-generic.c (lower_vec_perm): Likewise.
27688 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
27690 * selftest.h (selftest::vec_perm_indices_c_tests): Declare.
27691 * selftest-run-tests.c (selftest::run_tests): Call it.
27692 * vector-builder.h (vector_builder::operator ==): New function.
27693 (vector_builder::operator !=): Likewise.
27694 * vec-perm-indices.h (vec_perm_indices::series_p): Declare.
27695 (vec_perm_indices::all_from_input_p): New function.
27696 * vec-perm-indices.c (vec_perm_indices::series_p): Likewise.
27697 (test_vec_perm_12, selftest::vec_perm_indices_c_tests): Likewise.
27698 * fold-const.c (fold_ternary_loc): Use tree_to_vec_perm_builder
27699 instead of reading the VECTOR_CST directly. Detect whether both
27700 vector inputs are the same before constructing the vec_perm_indices,
27701 and update the number of inputs argument accordingly. Use the
27702 utility functions added above. Only construct sel2 if we need to.
27704 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
27706 * optabs.c (expand_vec_perm_var): Use an explicit encoding for
27707 the broadcast of the low byte.
27708 (expand_mult_highpart): Use an explicit encoding for the permutes.
27709 * optabs-query.c (can_mult_highpart_p): Likewise.
27710 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
27711 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
27712 (vectorizable_bswap): Likewise.
27713 * tree-vect-data-refs.c (vect_grouped_store_supported): Use an
27714 explicit encoding for the power-of-2 permutes.
27715 (vect_permute_store_chain): Likewise.
27716 (vect_grouped_load_supported): Likewise.
27717 (vect_permute_load_chain): Likewise.
27719 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
27721 * vec-perm-indices.h (vec_perm_indices_to_tree): Declare.
27722 * vec-perm-indices.c (vec_perm_indices_to_tree): New function.
27723 * tree-ssa-forwprop.c (simplify_vector_constructor): Use it.
27724 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
27725 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
27726 (vect_gen_perm_mask_any): Likewise.
27728 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
27730 * int-vector-builder.h: New file.
27731 * vec-perm-indices.h: Include int-vector-builder.h.
27732 (vec_perm_indices): Redefine as an int_vector_builder.
27733 (auto_vec_perm_indices): Delete.
27734 (vec_perm_builder): Redefine as a stand-alone class.
27735 (vec_perm_indices::vec_perm_indices): New function.
27736 (vec_perm_indices::clamp): Likewise.
27737 * vec-perm-indices.c: Include fold-const.h and tree-vector-builder.h.
27738 (vec_perm_indices::new_vector): New function.
27739 (vec_perm_indices::new_expanded_vector): Update for new
27740 vec_perm_indices class.
27741 (vec_perm_indices::rotate_inputs): New function.
27742 (vec_perm_indices::all_in_range_p): Operate directly on the
27743 encoded form, without computing elided elements.
27744 (tree_to_vec_perm_builder): Operate directly on the VECTOR_CST
27745 encoding. Update for new vec_perm_indices class.
27746 * optabs.c (expand_vec_perm_const): Create a vec_perm_indices for
27747 the given vec_perm_builder.
27748 (expand_vec_perm_var): Update vec_perm_builder constructor.
27749 (expand_mult_highpart): Use vec_perm_builder instead of
27750 auto_vec_perm_indices.
27751 * optabs-query.c (can_mult_highpart_p): Use vec_perm_builder and
27752 vec_perm_indices instead of auto_vec_perm_indices. Use a single
27753 or double series encoding as appropriate.
27754 * fold-const.c (fold_ternary_loc): Use vec_perm_builder and
27755 vec_perm_indices instead of auto_vec_perm_indices.
27756 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
27757 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
27758 (vect_permute_store_chain): Likewise.
27759 (vect_grouped_load_supported): Likewise.
27760 (vect_permute_load_chain): Likewise.
27761 (vect_shift_permute_load_chain): Likewise.
27762 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
27763 (vect_transform_slp_perm_load): Likewise.
27764 (vect_schedule_slp_instance): Likewise.
27765 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
27766 (vectorizable_mask_load_store): Likewise.
27767 (vectorizable_bswap): Likewise.
27768 (vectorizable_store): Likewise.
27769 (vectorizable_load): Likewise.
27770 * tree-vect-generic.c (lower_vec_perm): Use vec_perm_builder and
27771 vec_perm_indices instead of auto_vec_perm_indices. Use
27772 tree_to_vec_perm_builder to read the vector from a tree.
27773 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Take a
27774 vec_perm_builder instead of a vec_perm_indices.
27775 (have_whole_vector_shift): Use vec_perm_builder and
27776 vec_perm_indices instead of auto_vec_perm_indices. Leave the
27777 truncation to calc_vec_perm_mask_for_shift.
27778 (vect_create_epilog_for_reduction): Likewise.
27779 * config/aarch64/aarch64.c (expand_vec_perm_d::perm): Change
27780 from auto_vec_perm_indices to vec_perm_indices.
27781 (aarch64_expand_vec_perm_const_1): Use rotate_inputs on d.perm
27782 instead of changing individual elements.
27783 (aarch64_vectorize_vec_perm_const): Use new_vector to install
27784 the vector in d.perm.
27785 * config/arm/arm.c (expand_vec_perm_d::perm): Change
27786 from auto_vec_perm_indices to vec_perm_indices.
27787 (arm_expand_vec_perm_const_1): Use rotate_inputs on d.perm
27788 instead of changing individual elements.
27789 (arm_vectorize_vec_perm_const): Use new_vector to install
27790 the vector in d.perm.
27791 * config/powerpcspe/powerpcspe.c (rs6000_expand_extract_even):
27792 Update vec_perm_builder constructor.
27793 (rs6000_expand_interleave): Likewise.
27794 * config/rs6000/rs6000.c (rs6000_expand_extract_even): Likewise.
27795 (rs6000_expand_interleave): Likewise.
27797 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
27799 * optabs-query.c (can_vec_perm_var_p): Check whether lowering
27800 to qimode could truncate the indices.
27801 * optabs.c (expand_vec_perm_var): Likewise.
27803 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
27805 * Makefile.in (OBJS): Add vec-perm-indices.o.
27806 * vec-perm-indices.h: New file.
27807 * vec-perm-indices.c: Likewise.
27808 * target.h (vec_perm_indices): Replace with a forward class
27810 (auto_vec_perm_indices): Move to vec-perm-indices.h.
27811 * optabs.h: Include vec-perm-indices.h.
27812 (expand_vec_perm): Delete.
27813 (selector_fits_mode_p, expand_vec_perm_var): Declare.
27814 (expand_vec_perm_const): Declare.
27815 * target.def (vec_perm_const_ok): Replace with...
27816 (vec_perm_const): ...this new hook.
27817 * doc/tm.texi.in (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Replace with...
27818 (TARGET_VECTORIZE_VEC_PERM_CONST): ...this new hook.
27819 * doc/tm.texi: Regenerate.
27820 * optabs.def (vec_perm_const): Delete.
27821 * doc/md.texi (vec_perm_const): Likewise.
27822 (vec_perm): Refer to TARGET_VECTORIZE_VEC_PERM_CONST.
27823 * expr.c (expand_expr_real_2): Use expand_vec_perm_const rather than
27824 expand_vec_perm for constant permutation vectors. Assert that
27825 the mode of variable permutation vectors is the integer equivalent
27826 of the mode that is being permuted.
27827 * optabs-query.h (selector_fits_mode_p): Declare.
27828 * optabs-query.c: Include vec-perm-indices.h.
27829 (selector_fits_mode_p): New function.
27830 (can_vec_perm_const_p): Check whether targetm.vectorize.vec_perm_const
27831 is defined, instead of checking whether the vec_perm_const_optab
27832 exists. Use targetm.vectorize.vec_perm_const instead of
27833 targetm.vectorize.vec_perm_const_ok. Check whether the indices
27834 fit in the vector mode before using a variable permute.
27835 * optabs.c (shift_amt_for_vec_perm_mask): Take a mode and a
27836 vec_perm_indices instead of an rtx.
27837 (expand_vec_perm): Replace with...
27838 (expand_vec_perm_const): ...this new function. Take the selector
27839 as a vec_perm_indices rather than an rtx. Also take the mode of
27840 the selector. Update call to shift_amt_for_vec_perm_mask.
27841 Use targetm.vectorize.vec_perm_const instead of vec_perm_const_optab.
27842 Use vec_perm_indices::new_expanded_vector to expand the original
27843 selector into bytes. Check whether the indices fit in the vector
27844 mode before using a variable permute.
27845 (expand_vec_perm_var): Make global.
27846 (expand_mult_highpart): Use expand_vec_perm_const.
27847 * fold-const.c: Includes vec-perm-indices.h.
27848 * tree-ssa-forwprop.c: Likewise.
27849 * tree-vect-data-refs.c: Likewise.
27850 * tree-vect-generic.c: Likewise.
27851 * tree-vect-loop.c: Likewise.
27852 * tree-vect-slp.c: Likewise.
27853 * tree-vect-stmts.c: Likewise.
27854 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm_const):
27856 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>): Delete.
27857 * config/aarch64/aarch64.c (aarch64_expand_vec_perm_const)
27858 (aarch64_vectorize_vec_perm_const_ok): Fuse into...
27859 (aarch64_vectorize_vec_perm_const): ...this new function.
27860 (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
27861 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
27862 * config/arm/arm-protos.h (arm_expand_vec_perm_const): Delete.
27863 * config/arm/vec-common.md (vec_perm_const<mode>): Delete.
27864 * config/arm/arm.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
27865 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
27866 (arm_expand_vec_perm_const, arm_vectorize_vec_perm_const_ok): Merge
27868 (arm_vectorize_vec_perm_const): ...this new function. Explicitly
27869 check for NEON modes.
27870 * config/i386/i386-protos.h (ix86_expand_vec_perm_const): Delete.
27871 * config/i386/sse.md (VEC_PERM_CONST, vec_perm_const<mode>): Delete.
27872 * config/i386/i386.c (ix86_expand_vec_perm_const_1): Update comment.
27873 (ix86_expand_vec_perm_const, ix86_vectorize_vec_perm_const_ok): Merge
27875 (ix86_vectorize_vec_perm_const): ...this new function. Incorporate
27876 the old VEC_PERM_CONST conditions.
27877 * config/ia64/ia64-protos.h (ia64_expand_vec_perm_const): Delete.
27878 * config/ia64/vect.md (vec_perm_const<mode>): Delete.
27879 * config/ia64/ia64.c (ia64_expand_vec_perm_const)
27880 (ia64_vectorize_vec_perm_const_ok): Merge into...
27881 (ia64_vectorize_vec_perm_const): ...this new function.
27882 * config/mips/loongson.md (vec_perm_const<mode>): Delete.
27883 * config/mips/mips-msa.md (vec_perm_const<mode>): Delete.
27884 * config/mips/mips-ps-3d.md (vec_perm_constv2sf): Delete.
27885 * config/mips/mips-protos.h (mips_expand_vec_perm_const): Delete.
27886 * config/mips/mips.c (mips_expand_vec_perm_const)
27887 (mips_vectorize_vec_perm_const_ok): Merge into...
27888 (mips_vectorize_vec_perm_const): ...this new function.
27889 * config/powerpcspe/altivec.md (vec_perm_constv16qi): Delete.
27890 * config/powerpcspe/paired.md (vec_perm_constv2sf): Delete.
27891 * config/powerpcspe/spe.md (vec_perm_constv2si): Delete.
27892 * config/powerpcspe/vsx.md (vec_perm_const<mode>): Delete.
27893 * config/powerpcspe/powerpcspe-protos.h (altivec_expand_vec_perm_const)
27894 (rs6000_expand_vec_perm_const): Delete.
27895 * config/powerpcspe/powerpcspe.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK):
27897 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
27898 (altivec_expand_vec_perm_const_le): Take each operand individually.
27899 Operate on constant selectors rather than rtxes.
27900 (altivec_expand_vec_perm_const): Likewise. Update call to
27901 altivec_expand_vec_perm_const_le.
27902 (rs6000_expand_vec_perm_const): Delete.
27903 (rs6000_vectorize_vec_perm_const_ok): Delete.
27904 (rs6000_vectorize_vec_perm_const): New function.
27905 (rs6000_do_expand_vec_perm): Take a vec_perm_builder instead of
27906 an element count and rtx array.
27907 (rs6000_expand_extract_even): Update call accordingly.
27908 (rs6000_expand_interleave): Likewise.
27909 * config/rs6000/altivec.md (vec_perm_constv16qi): Delete.
27910 * config/rs6000/paired.md (vec_perm_constv2sf): Delete.
27911 * config/rs6000/vsx.md (vec_perm_const<mode>): Delete.
27912 * config/rs6000/rs6000-protos.h (altivec_expand_vec_perm_const)
27913 (rs6000_expand_vec_perm_const): Delete.
27914 * config/rs6000/rs6000.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
27915 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
27916 (altivec_expand_vec_perm_const_le): Take each operand individually.
27917 Operate on constant selectors rather than rtxes.
27918 (altivec_expand_vec_perm_const): Likewise. Update call to
27919 altivec_expand_vec_perm_const_le.
27920 (rs6000_expand_vec_perm_const): Delete.
27921 (rs6000_vectorize_vec_perm_const_ok): Delete.
27922 (rs6000_vectorize_vec_perm_const): New function. Remove stray
27923 reference to the SPE evmerge intructions.
27924 (rs6000_do_expand_vec_perm): Take a vec_perm_builder instead of
27925 an element count and rtx array.
27926 (rs6000_expand_extract_even): Update call accordingly.
27927 (rs6000_expand_interleave): Likewise.
27928 * config/sparc/sparc.md (vec_perm_constv8qi): Delete in favor of...
27929 * config/sparc/sparc.c (sparc_vectorize_vec_perm_const): ...this
27931 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
27933 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
27935 * optabs.c (expand_vec_perm_1): Assert that SEL has an integer
27936 vector mode and that that mode matches the mode of the data
27938 (expand_vec_perm): Split handling of non-CONST_VECTOR selectors
27939 out into expand_vec_perm_var. Do all CONST_VECTOR handling here,
27940 directly using expand_vec_perm_1 when forcing selectors into
27942 (expand_vec_perm_var): New function, split out from expand_vec_perm.
27944 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
27946 * optabs-query.h (can_vec_perm_p): Delete.
27947 (can_vec_perm_var_p, can_vec_perm_const_p): Declare.
27948 * optabs-query.c (can_vec_perm_p): Split into...
27949 (can_vec_perm_var_p, can_vec_perm_const_p): ...these two functions.
27950 (can_mult_highpart_p): Use can_vec_perm_const_p to test whether a
27951 particular selector is valid.
27952 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
27953 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
27954 (vect_grouped_load_supported): Likewise.
27955 (vect_shift_permute_load_chain): Likewise.
27956 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
27957 (vect_transform_slp_perm_load): Likewise.
27958 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
27959 (vectorizable_bswap): Likewise.
27960 (vect_gen_perm_mask_checked): Likewise.
27961 * fold-const.c (fold_ternary_loc): Likewise. Don't take
27962 implementations of variable permutation vectors into account
27963 when deciding which selector to use.
27964 * tree-vect-loop.c (have_whole_vector_shift): Don't check whether
27965 vec_perm_const_optab is supported; instead use can_vec_perm_const_p
27966 with a false third argument.
27967 * tree-vect-generic.c (lower_vec_perm): Use can_vec_perm_const_p
27968 to test whether the constant selector is valid and can_vec_perm_var_p
27969 to test whether a variable selector is valid.
27971 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
27973 * optabs-query.h (can_vec_perm_p): Take a const vec_perm_indices *.
27974 * optabs-query.c (can_vec_perm_p): Likewise.
27975 * fold-const.c (fold_vec_perm): Take a const vec_perm_indices &
27976 instead of vec_perm_indices.
27977 * tree-vectorizer.h (vect_gen_perm_mask_any): Likewise,
27978 (vect_gen_perm_mask_checked): Likewise,
27979 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise,
27980 (vect_gen_perm_mask_checked): Likewise,
27982 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
27984 * optabs-query.h (qimode_for_vec_perm): Declare.
27985 * optabs-query.c (can_vec_perm_p): Split out qimode search to...
27986 (qimode_for_vec_perm): ...this new function.
27987 * optabs.c (expand_vec_perm): Use qimode_for_vec_perm.
27989 2018-01-02 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27991 * rtlanal.c (canonicalize_condition): Return 0 if final rtx
27992 does not have a conditional at the top.
27994 2018-01-02 Richard Biener <rguenther@suse.de>
27996 * ipa-inline.c (big_speedup_p): Fix expression.
27998 2018-01-02 Jan Hubicka <hubicka@ucw.cz>
28001 * config/i386/x86-tune-costs.h: Increase cost of integer load costs
28004 2018-01-02 Jan Hubicka <hubicka@ucw.cz>
28008 * x86-tune-costs.h (generic_cost): Reduce cost of FDIV 20->17,
28009 cost of sqrt 20->14, DIVSS 18->13, DIVSD 32->17, SQRtSS 30->14
28010 and SQRTsD 58->18, cond_not_taken_branch_cost. 2->1. Increase
28011 cond_taken_branch_cost 3->4.
28013 2018-01-01 Jakub Jelinek <jakub@redhat.com>
28015 PR tree-optimization/83581
28016 * tree-loop-distribution.c (pass_loop_distribution::execute): Return
28017 TODO_cleanup_cfg if any changes have been made.
28019 PR middle-end/83608
28020 * expr.c (store_expr_with_bounds): Use simplify_gen_subreg instead of
28021 convert_modes if target mode has the right side, but different mode
28024 PR middle-end/83609
28025 * expr.c (expand_assignment): Fix up a typo in simplify_gen_subreg
28026 last argument when extracting from CONCAT. If either from_real or
28027 from_imag is NULL, use expansion through memory. If result is not
28028 a CONCAT and simplify_gen_subreg fails, try to simplify_gen_subreg
28029 the parts directly to inner mode, if even that fails, use expansion
28032 PR middle-end/83623
28033 * expmed.c (expand_shift_1): For 2-byte rotates by BITS_PER_UNIT,
28034 check for bswap in mode rather than HImode and use that in expand_unop
28037 Copyright (C) 2018 Free Software Foundation, Inc.
28039 Copying and distribution of this file, with or without modification,
28040 are permitted in any medium without royalty provided the copyright
28041 notice and this notice are preserved.