1 2018-11-21 Renlin Li <renlin.li@arm.com>
4 * explow.h (get_dynamic_stack_size): Declare it as external.
5 * explow.c (record_new_stack_level): Remove function static attribute.
6 * function.c (assign_stack_local_1): Dynamically align the stack slot
7 addr for parameter copy on the stack.
9 2018-11-21 Richard Biener <rguenther@suse.de>
12 * bitmap.c (bitmap_last_set_bit): Refactor to avoid warning.
13 * Makefile.in (bitmap.o-warn): Remove again.
15 2018-11-20 Jeff Law <law@redhat.com>
17 PR tree-optimization/88069
18 * tree-ssa-dom.c (record_equivalences_from_phis): Propagate away
19 degenerate virtual PHIs.
21 2018-11-21 H.J. Lu <hongjiu.lu@intel.com>
24 * config/i386/sse.md (sse4_1_<code>v8qiv8hi2<mask_name>): Replace
25 nonimmediate_operand with register_operand.
26 (avx2_<code>v8qiv8si2<mask_name>): Likewise.
27 (sse4_1_<code>v4qiv4si2<mask_name>): Likewise.
28 (sse4_1_<code>v4hiv4si2<mask_name>): Likewise.
29 (sse4_1_<code>v2qiv2di2<mask_name>): Likewise.
30 (avx512f_<code>v8qiv8di2<mask_name>): Likewise.
31 (avx2_<code>v4qiv4di2<mask_name>): Likewise.
32 (avx2_<code>v4hiv4di2<mask_name>): Likewise.
33 (sse4_1_<code>v2hiv2di2<mask_name>): Likewise.
34 (sse4_1_<code>v2siv2di2<mask_name>): Likewise.
35 (*sse4_1_<code>v8qiv8hi2<mask_name>_1): New pattern.
36 (*sse4_1_<code>v8qiv8hi2<mask_name>_2): Likewise.
37 (*avx2_<code>v8qiv8si2<mask_name>_1): Likewise.
38 (*avx2_<code>v8qiv8si2<mask_name>_2): Likewise.
39 (*sse4_1_<code>v4qiv4si2<mask_name>_1): Likewise.
40 (*sse4_1_<code>v4qiv4si2<mask_name>_2): Likewise.
41 (*sse4_1_<code>v4hiv4si2<mask_name>_1): Likewise.
42 (*sse4_1_<code>v4hiv4si2<mask_name>_2): Likewise.
43 (*avx512f_<code>v8qiv8di2<mask_name>_1): Likewise.
44 (*avx512f_<code>v8qiv8di2<mask_name>_2): Likewise.
45 (*avx2_<code>v4qiv4di2<mask_name>_1): Likewise.
46 (*avx2_<code>v4qiv4di2<mask_name>_2): Likewise.
47 (*avx2_<code>v4hiv4di2<mask_name>_1): Likewise.
48 (*avx2_<code>v4hiv4di2<mask_name>_2): Likewise.
49 (*sse4_1_<code>v2hiv2di2<mask_name>_1): Likewise.
50 (*sse4_1_<code>v2hiv2di2<mask_name>_2): Likewise.
51 (*sse4_1_<code>v2siv2di2<mask_name>_1): Likewise.
52 (*sse4_1_<code>v2siv2di2<mask_name>_2): Likewise.
54 2018-11-21 H.J. Lu <hongjiu.lu@intel.com>
56 * read-rtl.c (apply_subst_iterator): Handle define_split and
57 define_insn_and_split.
59 2018-11-21 Jakub Jelinek <jakub@redhat.com>
61 PR rtl-optimization/87817
62 * config/i386/i386.md (bmi2_bzhi_<mode>3, *bmi2_bzhi_<mode>3,
63 *bmi2_bzhi_<mode>3_1, *bmi2_bzhi_<mode>3_1_ccz): Use IF_THEN_ELSE
64 in the pattern to avoid triggering UB when operands[2] is zero.
65 (tbm_bextri_<mode>): New expander. Renamed the old define_insn to ...
66 (*tbm_bextri_<mode>): ... this.
68 2018-11-21 Tom de Vries <tdevries@suse.de>
71 * params.def (HOT_BB_COUNT_FRACTION): Terminate help message with
74 2018-11-21 Andreas Krebbel <krebbel@linux.ibm.com>
76 * configure.ac: Add check for Binutils to determine whether vector
77 load/store alignments hints are being supported.
78 * config.in: Regenerate.
79 * configure: Regenerate.
80 * config/s390/s390.c (print_operand): Support new output
82 * config/s390/s390.md ("movti"): Append alignment hint output
83 using the new output modifier 'A'.
84 * config/s390/vector.md ("mov<mode>", "*vec_tf_to_v1tf")
85 ("*vec_ti_to_v1ti"): Likewise.
87 2018-11-20 Martin Sebor <msebor@redhat.com>
89 * doc/extend.texi (Other Builtins): Add __builtin_has_attribute.
91 2018-11-20 Jan Hubicka <hubicka@ucw.cz>
94 * ipa-devirt.c (odr_types_equivalent_p): Use operand_equal_p to
97 2018-11-20 Jakub Jelinek <jakub@redhat.com>
99 PR tree-optimization/87895
100 * omp-simd-clone.c (ipa_simd_modify_function_body): When removing
101 or replacing GIMPLE_RETURN, set EDGE_FALLTHRU on the edge to EXIT.
102 (simd_clone_adjust): Don't set EDGE_FALLTHRU here. In a loop that
103 redirects edges to EXIT to edges to incr_bb, iterate while EXIT
104 has any preds and always use EDGE_PRED (, 0).
106 2018-11-20 Uros Bizjak <ubizjak@gmail.com>
109 * mode-switching.c (create_pre_exit): After reload, always split the
110 fallthrough edge to the exit block.
112 2018-11-20 Jan Hubicka <hubicka@ucw.cz>
114 * ipa-devirt.c (add_type_duplicate): Do not ICE on incomplete enums.
115 * tree.c (build_array_type_1): Forward declare.
116 (fld_type_variant_equal_p): Add INNER_TYPE parameter.
117 (fld_type_variant): Likewise.
118 (fld_simplified_types): New hash.
119 (fld_process_array_type): New function.
120 (fld_incomplete_type_of): Handle array and enumeration types.
121 (fld_simplified_type): Handle simplification of arrays.
122 (free_lang_data): Allocate and free simplified types hash.
124 2018-11-20 Jan Hubicka <hubicka@ucw.cz>
127 * ipa-devirt.c (odr_subtypes_equivalent_p): Report ODR violation
128 when sybtype already violates ODR.
129 (get_odr_type): Do not ICE when insert is false and type duplicate
130 is not registered yet.
131 (register_odr_type): Be sure to register subtypes first.
133 2018-11-20 Andreas Krebbel <krebbel@linux.ibm.com>
135 * config/s390/s390.md ("clztidi2"): Swap the RTX's written to the
136 DImode parts of the target operand.
138 2018-11-20 Nathan Sidwell <nathan@acm.org>
141 * Makefile.in (bitmap.o-warn): Use -Wno-error=array-bounds.
143 2018-11-20 Jan Hubicka <hubicka@ucw.cz>
146 * tree.c (free_lang_data_in_cgraph): Add argument fld; break out
148 (free_lang_data) ... here; update call of free_lang_data_in_cgraph.
150 2018-11-20 Jan Hubicka <hubicka@ucw.cz>
153 * ipa-fnsummary.c (pass_ipa_fnsummary): Do not remove functions
154 * ipa.c (possible_inline_candidate_p): Break out from ..
155 (process_references): ... here ; drop before_inlining_p;
156 cleanup handling of alises.
157 (walk_polymorphic_call_targets): Likewise.
158 (symbol_table::remove_unreachable_nodes): Likewise.
159 * passes.c (pass_data_ipa_remove_symbols): New structure.
160 (pass_ipa_remove_symbols): New pass.
161 (make_pass_ipa_remove_symbols): New function.
162 * tree-pass.h (make_pass_ipa_remove_symbols): Declare.
163 * passes.def (pass_ipa_remove_symbols): Schedule after early passes.
165 2018-11-20 Richard Biener <rguenther@suse.de>
167 * tree-vect-stmts.c (vectorizable_condition): Do not get
168 at else_clause vect def for EXTRACT_LAST_REDUCTION. Remove
169 pointless vect_is_simple_use calls.
171 2018-11-20 Richard Biener <rguenther@suse.de>
173 PR tree-optimization/88074
174 * tree-ssa-dom.c (pass_dominator::execute): Do not walk
177 2018-11-20 Richard Biener <rguenther@suse.de>
179 PR tree-optimization/88069
180 * tree-ssa-sccvn.c (visit_phi): Do not value-number to unvisited
181 virtual PHI arguments.
183 2018-11-20 Ilya Leoshkevich <iii@linux.ibm.com>
186 * config/s390/s390.md: Skip LT(G) peephole when literal pool is
188 * rtl.h (contains_constant_pool_address_p): New function.
189 * rtlanal.c (contains_constant_pool_address_p): Likewise.
191 2018-11-20 Richard Biener <rguenther@suse.de>
194 * alias.c (component_uses_parent_alias_set_from): Remove
195 alias-set zero and TYPE_TYPELESS_STORAGE case both already
196 handled in other ways.
198 2018-11-20 Richard Biener <rguenther@suse.de>
200 PR tree-optimization/88087
201 * tree-ssa-pre.c (create_expression_by_pieces): Re-materialize
203 * tree-ssa-sccvn.c (copy_reference_ops_from_call): Remember
206 2018-11-20 Richard Biener <rguenther@suse.de>
209 * tree-data-ref.c (lambda_matrix_right_hermite): Use abs_hwi.
211 2018-11-19 Jan Hubicka <hubicka@ucw.cz>
214 * ipa-devirt.c (free_enum_values): Do not ICE on ODR vilations.
216 2018-11-19 Sandra Loosemore <sandra@codesourcery.com>
219 * doc/invoke.texi (Link Options): Mention shared libraries
220 in documentation for the -l option. Simplify discussion and
221 point to the system linker documentation for details.
223 2018-11-19 Peter Bergner <bergner@linux.ibm.com>
225 PR rtl-optimization/88033
226 * ira-lives.c (non_conflicting_reg_copy_p): Skip copies from a register
227 to itself. Use HARD_REGISTER_NUM_P.
229 2018-11-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
231 * tree-vect-loop.c (vect_transform_loop): Disable further unrolling
232 of the loop if vf is non-constant.
234 2018-11-19 David Malcolm <dmalcolm@redhat.com>
236 PR tree-optimization/87025
237 * dumpfile.c (dump_context::begin_scope): Call end_any_optinfo
238 immediately after creating the scope optinfo.
239 (selftest::test_pr87025): New function.
240 (selftest::dumpfile_c_tests): Call it.
241 * optinfo-emit-json.cc (optrecord_json_writer::pop_scope): Assert
242 that we're not popping the top-level records array.
243 * optinfo.cc (optinfo::handle_dump_file_kind): Assert that we're
244 not changing the kind of a "scope" optinfo.
246 2018-11-19 David Malcolm <dmalcolm@redhat.com>
248 PR tree-optimization/87025
249 * dump-context.h: Include "optinfo.h".
250 (class optrecord_json_writer): New forward decl.
251 (dump_context::forcibly_enable_optinfo_p): Delete.
252 (dump_context::optinfo_enabled_p): New member function.
253 (dump_context::optimization_records_enabled_p): New member
255 (dump_context::set_json_writer): New member function.
256 (dump_context::emit_optinfo): New member function.
257 (dump_context::m_forcibly_enable_optinfo): Delete.
258 (dump_context::m_json_writer): New member data.
259 * dumpfile.c (dump_context::set_json_writer): New member function.
260 (dump_context::finish_any_json_writer): New member function.
261 (dump_context::end_scope): Replace call to
262 optimization_records_maybe_pop_dump_scope with call to
263 m_json_writer->pop_scope.
264 (dump_context::optinfo_enabled_p): New member function.
265 (dump_context::end_any_optinfo): Replace call to optinfo::emit with call
266 to dump_context::emit_optinfo.
267 (dump_context::emit_optinfo): New member function.
268 (temp_dump_context::temp_dump_context): Replace
269 m_forcibly_enable_optinfo with call to set_json_writer.
270 (temp_dump_context::~temp_dump_context): Clean up any json writer.
271 * optinfo-emit-json.cc (class optrecord_json_writer): Move to
273 (the_json_writer): Delete.
274 (optimization_records_start): Delete.
275 (optimization_records_finish): Delete.
276 (optimization_records_enabled_p): Delete, in favor of
277 dump_context::optimization_records_enabled_p.
278 (optimization_records_maybe_record_optinfo): Delete.
279 (optimization_records_maybe_pop_dump_scope): Delete.
280 * optinfo-emit-json.h: Include "json.h". Delete forward
282 (optimization_records_start): Delete.
283 (optimization_records_finish): Delete.
284 (optimization_records_enabled_p): Delete.
285 (optimization_records_maybe_record_optinfo): Delete.
286 (optimization_records_maybe_pop_dump_scope): Delete.
287 (class optrecord_json_writer): Move here from
288 optinfo-emit-json.cc.
289 * optinfo.cc (optinfo::emit_for_opt_problem): Replace call
290 to optinfo::emit with call to dump_context::emit_optinfo.
291 (optinfo::emit): Delete, in favor of dump_context::emit_optinfo.
292 (optinfo_enabled_p): Delete, in favor of
293 dump_context::optinfo_enabled_p.
294 (optinfo_wants_inlining_info_p): Update for conversion o
295 optimization_records_enabled_p to a member function of
297 * optinfo.h (optinfo_enabled_p): Delete, in favor of
298 dump_context::optinfo_enabled_p.
299 (optinfo::emit): Delete, in favor of dump_context::emit_optinfo.
300 * toplev.c: Include "dump-context.h".
301 (compile_file): Replace call to optimization_records_finish with
302 dump_context::finish_any_json_writer.
303 (do_compile): Replace call to optimization_records_start with
304 conditionally creating a optrecord_json_writer for the
307 2018-11-19 Jakub Jelinek <jakub@redhat.com>
309 PR tree-optimization/88071
310 * tree-vect-loop.c (vectorize_fold_left_reduction): Pass true instead
311 of false as last argument to gsi_remove.
312 * tree-vect-stmts.c (vect_finish_replace_stmt): Pass true instead of
313 false as last argument to gsi_replace.
316 * omp-expand.c: Don't include debug.h.
317 (adjust_context_and_scope): Add REGION argument. Find DECL_CONTEXT
318 from innermost outer parallel, task, teams or target that has a
319 child_fn set, or, if there is no such outer region, use
320 current_function_decl. Do the DECL_CONTEXT adjustment regardless of
321 whether a suitable BLOCK is found or not.
322 (expand_parallel_call, expand_teams_call): Don't call
323 adjust_context_and_scope here.
324 (grid_expand_target_grid_body): Revert 2017-01-25 changes.
325 (expand_omp_taskreg, expand_omp_target): Likewise. Call
326 adjust_context_and_scope.
327 * dwarf2out.c (dwarf2out_early_global_decl): For
328 decl_function_context recurse instead of calling dwarf2out_decl.
330 2018-11-20 Eric Botcazou <ebotcazou@adacore.com>
332 PR rtl-optimization/85925
333 * rtl.h (word_register_operation_p): New predicate.
334 * combine.c (record_dead_and_set_regs_1): Only apply specific handling
335 for WORD_REGISTER_OPERATIONS targets to word_register_operation_p RTX.
336 * rtlanal.c (nonzero_bits1): Likewise. Adjust couple of comments.
337 (num_sign_bit_copies1): Likewise.
339 2018-11-19 Richard Biener <rguenther@suse.de>
342 * tree.c (free_lang_data_in_one_sizepos): Free non-gimple-val
345 2018-11-19 Eric Botcazou <ebotcazou@adacore.com>
347 * compare-elim.c (struct comparison): Add not_in_a field.
348 (is_not): New static function.
349 (strip_not): Likewise.
350 (conforming_compare): Handle a NOT in the first operand.
351 (can_eliminate_compare): Likewise.
352 (find_comparison_dom_walker::before_dom_children): Likewise.
353 (try_eliminate_compare): Likewise.
354 * config/visium/visium.md (negsi2_insn_set_carry): Turn into...
355 (neg<mode>2_insn_set_carry): ...this and add missing NEG operation.
357 2018-11-19 Jonathan Wakely <jwakely@redhat.com>
359 * doc/extend.texi (Common Type Attributes): Fix typos.
361 2018-11-19 Martin Liska <mliska@suse.cz>
363 PR gcov-profile/88045
364 * coverage.c (coverage_begin_function): Add assert.
366 2018-11-18 Sandra Loosemore <sandra@codesourcery.com>
369 * doc/extend.texi (Common Function Attributes): Document that
370 no_instrument_function applies to -p and -pg, too.
371 * doc/invoke.texi (Instrumentation Options): Add cross-references
372 to docs for -p, -pg, and -finstrument-functions.
374 2018-11-18 Alan Modra <amodra@gmail.com>
376 * config/gnu-user.h (GNU_USER_TARGET_CRTI): Define.
377 (GNU_USER_TARGET_STARTFILE_SPEC): Use it here.
378 (GNU_USER_TARGET_CRTN): Define.
379 (GNU_USER_TARGET_ENDFILE_SPEC): Use it here.
380 (GNU_USER_TARGET_LINK_GCC_C_SEQUENCE_SPEC): Define.
382 2018-11-18 Alan Modra <amodra@gmail.com>
384 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Delete
385 !HAVE_LD_PIE variant.
386 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
388 2018-11-17 Nick Clifton <nickc@redhat.com>
389 Sandra Loosemore <sandra@codesourcery.com>
392 * doc/invoke.texi (Overall Options): Document that --help and
393 --help=<value> options cannot be combined.
395 2018-11-17 Jim Wilson <jimw@sifive.com>
397 * config/riscv/riscv.c (epilogue_cfa_sp_offset): New.
398 (riscv_restore_reg): If restoring HARD_FRAME_POINTER_REGNUM, and
399 epilogue_cfa_sp_offset set, then add REG_CFA_DEF_CFA regnote.
400 (riscv_expand_epilogue): Initialize epilogue_cfa_sp_offset. Set it
401 to step2 if frame_pointer_needed and step1 is 0.
403 2018-11-17 Sandra Loosemore <sandra@codesourcery.com>
406 * doc/cpp.texi (System Headers): Add note about implicit
407 extern "C" block on targets that define SYSTEM_IMPLICIT_EXTERN_C.
409 2018-11-17 Sandra Loosemore <sandra@codesourcery.com>
412 * doc/invoke.texi (C++ Dialect Options): Clarify usage of
413 -fno-implicit-templates.
415 2018-11-17 Jakub Jelinek <jakub@redhat.com>
417 PR tree-optimization/87546
418 * tree-vect-patterns.c (vect_look_through_possible_promotion): Add
419 min_precision variable, initially set it to orig_precision, only does
420 something if op_type's precision is <= min_precision and update
421 min_precision whenever calling set_op.
423 2018-11-16 Jan Hubicka <hubicka@ucw.cz>
426 * ipa-devirt.c (warn_odr): Look for main variant to get TYPE_DECL.
428 2018-11-16 Sandra Loosemore <sandra@codesourcery.com>
430 * doc/invoke.texi (Option Summary): Fix whitespace and line
431 breaks in @gccoptlist environments.
432 (Warning Options): Likewise.
433 (Optimize Options): Likewise.
434 (PowerPC SPE Options): Likewise.
435 (RS/6000 and PowerPC Options): Likewise.
437 2018-11-16 Jeff Law <law@redhat.com>
439 * config/mn10300/mn10300.md (adddi3_degenerate): Remove bogus
442 2018-11-16 Sandra Loosemore <sandra@codesourcery.com>
445 * doc/invoke.texi (Optimize Options): Update options enabled by
446 fprofile-generate, -fprofile-use, and -fauto-profile.
448 2018-11-16 Jozef Lawrynowicz <jozef.l@mittosystems.com>
451 * target-def.h: Initialize TARGET_ASM_{,UN}ALIGNED_P{S,D,T}I_OP.
452 Add them to the TARGET_ASM_{,UN}ALIGNED_INT_OP structs.
453 * target.def: Enumerate TARGET_ASM_{,UN}ALIGNED_P{S,D,T}I_OP in
455 * target.h: Add psi, pdi, pti to struct asm_int_op definition.
456 * targhooks.c (default_print_patchable_function_entry): Assert
457 asm_int_op does not return a NULL string.
458 * varasm.c (integer_asm_op): Return the op for a partial int type
459 when the requested size does not correspond to an integer type.
460 * config/msp430/msp430.c: Initialize TARGET_ASM_{,UN}ALIGNED_PSI_OP.
461 * doc/tm.texi: Regenerate.
463 2018-11-12 Jason Merrill <jason@redhat.com>
465 * gimplify.c (gimplify_case_label_expr): Handle hot/cold attributes.
467 2018-11-16 Michael Meissner <meissner@linux.ibm.com>
469 * config/rs6000/constraints.md (wF constraint): Remove power9
470 fusion documentation. Just document wF's use for power8 fusion.
471 * config/rs6000/predicates.md (p9_fusion_reg_operand): Delete.
472 (fusion_gpr_addis): Delete power9 fusion support. Change power8
473 fusion support to require the upper 12 bits to be all 0's or all
475 (fusion_gpr_mem_load): Add comment.
476 (fusion_addis_mem_combo_load): Remove power9 fusion support.
477 (fusion_addis_mem_combo_store): Delete.
478 (fusion_offsettable_mem_operand): Delete.
479 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Do not set
481 (ISA_3_0_MASKS_SERVER): Delete power9 fusion.
482 (POWERPC_MASKS): Delete power9 fusion.
483 * config/rs6000/rs6000-protos.h (emit_fusion_load_store): Delete.
484 (fusion_p9_p): Delete.
485 (expand_fusion_p9_load): Delete.
486 (expand_fusion_p9_store): Delete.
487 (emit_fusion_p9_load): Delete.
488 (emit_fusion_p9_store): Delete.
489 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Delete power9
491 (rs6000_option_override_internal): Set power8 fusion based on
492 whether we are tuning for power8. Delete power9 fusion support.
493 (rs6000_opt_masks): Delete -mpower9-fusion switch.
494 (emit_fusion_load): Rename emit_fusion_load_store to
495 emit_fusion_load, and drop fusion store support. Update callers.
496 (emit_fusion_load_store): Likewise.
497 (emit_fusion_gpr_load): Likewise.
498 (fusion_p9_p): Delete.
499 (expand_fusion_p9_load): Delete.
500 (expand_fusion_p9_store): Delete.
501 (emit_fusion_p9_load): Delete.
502 (emit_fusion_p9_store): Delete.
503 * config/rs6000/rs6000.md (UNSPEC_FUSION_P9): Delete.
504 (GPR_FUSION): Delete.
505 (FPR_FUSION): Delete.
506 (power9 fusion peephole2s): Delete.
507 (fusion_gpr_<P:mode>_<GPR_FUSION:mode>_load): Delete.
508 (fusion_gpr_<P:mode>_<GPR_FUSION:mode>_store): Delete.
509 (fusion_vsx_<P:mode>_<FPR_FUSION:mode>_load): Delete.
510 (fusion_vsx_<P:mode>_<FPR_FUSION:mode>_store): Delete.
511 (fusion_p9_<mode>_constant): Delete.
512 * config/rs6000/rs6000.opt (-mpower9-fusion): Delete undocumented
513 power9 fusion switch.
514 * doc/md.texi (PowerPC constraints): Update wF constraint
515 documentation for power8 fusion only.
517 2018-11-16 Jakub Jelinek <jakub@redhat.com>
519 PR rtl-optimization/87475
520 * cfgrtl.c (patch_jump_insn): Allow redirection failure for
521 CROSSING_JUMP_P insns.
522 (cfg_layout_redirect_edge_and_branch): Don't ICE if ret is NULL.
524 2018-11-16 Uros Bizjak <ubizjak@gmail.com>
527 * config/i386/i386.md (floatunsdidf2): Allow only 64bit AVX512F targets.
528 * config/i386/sse.md (UNSPEC_MOVDI_TO_SSE): New UNSPEC.
529 (movdi_to_sse): Rewrite using UNSPEC_MOVDI_TO_SSE unspec.
531 2018-11-16 Jakub Jelinek <jakub@redhat.com>
534 * optabs.c (expand_binop): For op0_mode use GET_MODE (op0), unless it
535 is VOIDmode, in which case use int_mode. Similarly for op1_mode.
537 2018-11-16 Eric Botcazou <ebotcazou@adacore.com>
539 * tree-switch-conversion.h (switch_decision_tree::emit_case_nodes): Add
540 location_t parameter.
541 (switch_decision_tree::emit_cmp_and_jump_insns): Likewise.
542 (switch_decision_tree::do_jump_if_equal): Likewise.
543 * tree-switch-conversion.c (switch_decision_tree::emit): Pass location
544 of switch statement to emit_case_nodes.
545 (switch_decision_tree::emit_cmp_and_jump_insns): Add LOC parameter and
546 set it on the newly built GIMPLE comparison statement.
547 (switch_decision_tree::do_jump_if_equal): Likewise.
548 (switch_decision_tree::emit_case_nodes): Add LOC parameter and pass it
549 in calls to do_jump_if_equal as well as recursive calls.
551 2018-11-16 Andi Kleen <ak@linux.intel.com>
553 * config/i386/i386.md: Allow memory operands to ptwrite.
555 2018-11-16 Michael Matz <matz@suse.de>
557 * system.h (PRsa): New macro.
558 (SIZE_AMOUNT): Cast number to uint64_t.
559 * alloc-pool.h (pool_usage::dump): Don't use %zu but PRsa.
560 (pool_usage::dump_footer): Likewise and also use PRIu64.
561 * bitmap.h (bitmap_usage::dump): Likewise.
562 * ggc-common.c (ggc_usage::dump): Likewise.
563 * ggc-page.c (ggc_print_statistics): Likewise.
564 * input.c (dump_line_table_statistics): Likewise.
565 * mem-stats.h (mem_usage::dump): Likewise.
566 (mem_usage::dump_footer): Likewise.
567 * rtl.c (dump_rtx_statistics): Likewise.
568 * tree-cfg.c (dump_cfg_stats): Likewise.
569 * tree-dfa.c (dump_dfa_stats): Likewise.
570 * tree-phinodes.c (phinodes_print_statistics): Likewise.
571 * tree-ssanames (ssanames_print_statistics): Likewise.
572 * vec.c (vec_usage::dump): Likewise.
573 (vec_usage::dump_footer): Likewise.
575 2018-11-16 Richard Biener <rguenther@suse.de>
577 PR tree-optimization/88011
578 * tree-vrp.c (extract_range_from_binary_expr): Fix error in
579 replacing set_value_range_to_undefined and
580 set_value_range_to_varying with method calls.
582 2018-11-16 Ilya Leoshkevich <iii@linux.ibm.com>
584 * config/s390/s390.md
585 (*r<noxa>sbg_<mode>_srl_bitmask): Do not delegate arithmetic to
587 (*r<noxa>sbg_<mode>_sll): Likewise.
588 (*r<noxa>sbg_<mode>_srl): Likewise.
589 (*r<noxa>sbg_sidi_srl): New pattern.
591 2018-11-16 Jerome Lambourg <lambourg@adacore.com>
593 * config/vxworks.h (TARGET_VXWORKS_HAVE_CTORS_DTORS): New macro.
594 Default to TARGET_VXWORKS_RTP.
595 (SUPPORTS_INIT_PRIORITY): Use TARGET_VXWORKS_HAVE_CTORS_DTORS instead
596 of TARGET_VXWORKS_RTP.
597 * config/vxworksae.h: Also define TARGET_VXWORKS_HAVE_CTORS_DTORS.
598 * config/vxworks.c: Use TARGET_VXWORKS_HAVE_CTORS_DTORS instead
599 of TARGET_VXWORKS_RTP to set targetm.have_ctors_dtors.
601 2018-11-15 Sandra Loosemore <sandra@codesourcery.com>
604 * doc/extend.texi (Common Type Attributes): Make it explicit
605 that attribute "packed" can apply to C++ classes.
607 2018-11-15 Martin Sebor <msebor@redhat.com>
611 * tree.c (type_argument_type): New function.
612 * tree.h (type_argument_type): Declare it.
613 * gcc/doc/extend.texi (alloc_align): Update and clarify.
614 (alloc_size, nonnull, sentinel): Same.
616 2018-11-15 Andrew Stubbs <ams@codesourcery.com>
617 Kwok Cheung Yeung <kcy@codesourcery.com>
619 * tree-vect-stmts.c (vectorizable_store): Don't ICE when
620 int_mode_for_size fails.
621 (vectorizable_load): Likewise.
623 2018-11-15 David Malcolm <dmalcolm@redhat.com>
625 * doc/ux.texi (Group logically-related diagnostics): Move
626 discussion of auto_diagnostic_group into this new subsection.
627 Give an example of where this grouping is used.
629 2018-11-15 David Malcolm <dmalcolm@redhat.com>
632 * Makefile.in (OBJS): Move json.o to...
633 (OBJS-libcommon): ...here and add diagnostic-format-json.o.
634 * common.opt (fdiagnostics-format=): New option.
635 (diagnostics_output_format): New enum.
636 * diagnostic-format-json.cc: New file.
637 * diagnostic.c (default_diagnostic_final_cb): New function, taken
638 from start of diagnostic_finish.
639 (diagnostic_initialize): Initialize final_cb to
640 default_diagnostic_final_cb.
641 (diagnostic_finish): Move "being treated as errors" messages to
642 default_diagnostic_final_cb. Call any final_cb.
643 (default_diagnostic_finalizer): Add diagnostic_t param.
644 (diagnostic_report_diagnostic): Pass "orig_diag_kind" to
645 diagnostic_finalizer callback.
646 * diagnostic.h (enum diagnostics_output_format): New enum.
647 (diagnostic_finalizer_fn): Reimplement, adding diagnostic_t param.
648 (struct diagnostic_context): Add "final_cb".
649 (default_diagnostic_finalizer): Add diagnostic_t param.
650 (diagnostic_output_format_init): New decl.
651 * doc/invoke.texi (-fdiagnostics-format): New option.
652 * dwarf2out.c (gen_producer_string): Ignore
653 OPT_fdiagnostics_format_.
654 * gcc.c (driver_handle_option): Handle OPT_fdiagnostics_format_.
655 * lto-wrapper.c (append_diag_options): Ignore it.
656 * opts.c (common_handle_option): Handle it.
658 2018-11-15 David Malcolm <dmalcolm@redhat.com>
660 PR tree-optimization/88015
661 * graphite-isl-ast-to-gimple.c
662 (translate_isl_ast_to_gimple::scop_to_isl_ast): Add missing check
664 * graphite-sese-to-poly.c (build_poly_scop): Likewise.
666 2018-11-15 Richard Biener <rguenther@suse.de>
669 * gimple.c (gimple_call_flags): Union flags from decl, type
671 * trans-mem.c (is_tm_pure_call): Simplify.
673 2018-11-15 Richard Biener <rguenther@suse.de>
675 PR tree-optimization/88031
676 * tree-vect-loop.c (vectorizable_reduction): Move check
677 for multiple types earlier so we get the expected dump.
678 Simplify calls to vectorizable_condition.
679 * tree-vect-stmts.h (vectorizable_condition): Update prototype.
680 * tree-vect-stmts.c (vectorizable_condition): Instead of
681 reduc_def and reduc_index take just a flag. Simplify
682 code-generation now that we can rely on the defs being set up.
683 (vectorizable_comparison): Remove unused argument.
685 2018-11-15 Richard Biener <rguenther@suse.de>
687 PR tree-optimization/88030
688 * tree-complex.c (need_eh_cleanup): New global.
689 (update_complex_assignment): Mark blocks that need EH update.
690 (expand_complex_comparison): Likewise.
691 (tree_lower_complex): Allocate and deallocate need_eh_cleanup,
692 perform EH cleanup and schedule CFG cleanup if that did anything.
694 2018-11-15 Jakub Jelinek <jakub@redhat.com>
696 PR rtl-optimization/88018
697 * cfgrtl.c (fixup_abnormal_edges): Guard moving insns to fallthru edge
698 on the presence of fallthru edge, rather than if it is a USE or not.
700 2018-11-15 Richard Biener <rguenther@suse.de>
703 * tree-data-ref.c (analyze_miv_subscript): Guard calls to
704 analyze_subscript_affine_affine properly.
706 2018-11-15 Xianmiao Qu <xianmiao_qu@c-sky.com>
708 * config/csky/csky-linux-elf.h (LINUX_DYNAMIC_LINKER): Remove.
709 (GLIBC_DYNAMIC_LINKER): Define.
710 (LINUX_TARGET_LINK_SPEC): Update the dynamic linker's name.
712 2018-11-15 Bin Cheng <bin.cheng@linux.alibaba.com>
714 PR tree-optimization/84648
715 * tree-ssa-loop-niter.c (adjust_cond_for_loop_until_wrap): New.
716 (number_of_iterations_cond): Adjust exit cond for loop-until-wrap case
717 by calling adjust_cond_for_loop_until_wrap.
719 2018-11-15 Sandra Loosemore <sandra@codesourcery.com>
722 * doc/extend.texi (Common Function Attributes): Clarify linker
723 restrictions on "aligned" attribute.
724 (Common Variable Attributes): Likewise. Mention that linker
725 restrictions don't apply to stack-allocated variables.
727 2018-11-15 Joern Rennecke <joern.rennecke@riscy-ip.com>
729 * gensupport.c (add_predicate_code): Properly handle ZERO_EXTRACT
732 2018-11-14 Sandra Loosemore <sandra@codesourcery.com>
736 * doc/invoke.texi (Optimize Options): Remove bad example about
737 interaction between -flto and -O. Replace it with a note that
738 you need to compile with -O and not just link. Copy-edit -flto
739 discussion to reduce verbiage and improve flow.
741 2018-11-14 Sandra Loosemore <sandra@codesourcery.com>
744 * doc/invoke.texi (Optimize Options): Clarify that -O0 and -Og
745 also suppress many optimizations. Alphabetize option lists for
746 -O1, -O2, and -Os. Add list of options disabled with -Og, and
747 correct documentation for those options to say that.
748 * opts.c (default_options_table): Sort table by level and option
749 name, to make it easier to correlate to the manual.
751 2018-11-14 Uros Bizjak <ubizjak@gmail.com>
753 * config/i386/i386.c (ix86_print_operand_address_as): Simplify
754 printing of the names of segment registers.
756 2018-11-14 Nathan Sidwell <nathan@acm.org>
760 * dwarf2out.c (dwarf2out_finish): Apply resolve_addr to comdat
763 2018-11-14 David Malcolm <dmalcolm@redhat.com>
765 * Makefile.in (CFLAGS-optinfo-emit-json.o): Add $(ZLIBINC).
767 2018-11-14 Jakub Jelinek <jakub@redhat.com>
769 P1236R1 - Signed integers are two's complement
770 * doc/invoke.texi (Wshift-overflow): Adjust documentation for c++2a.
773 * hash-map.h (hash_map::iterator::reference_pair): New class.
774 (hash_map::iterator::operator*): Return it rather than std::pair.
776 2018-11-14 Jeff Law <law@redhat.com>
778 * optabs.c (expand_binop): Pass INT_MODE to operand_subword_force
779 iff the operand is a constant.
781 2018-11-14 Aldy Hernandez <aldyh@redhat.com>
783 * gimple-ssa-evrp-analyze.c
784 (evrp_range_analyzer::record_ranges_from_incoming_edge): Rename
785 ignore_equivs_equal_p to equal_p.
786 * ipa-cp.c (meet_with_1): Use equal_p instead of
787 ignore_equivs_equal_p.
788 * ipa-prop.c (ipa_vr_ggc_hash_traits::equal): Same.
789 * tree-vrp.c (value_range::ignore_equivs_equal_p): Remove.
790 (value_range::operator==): Remove.
791 (value_range::operator!=): Remove.
792 (vrp_prop::visit_stmt): Use equal_p.
793 * tree-vrp.h (value_range): Remove operator==, operator!=,
794 ignore_equivs_equal_p.
795 * vr-values.c (update_value_range): Use equal_p.
797 2018-11-14 Michael Matz <matz@suse.de>
800 * gimplify.c (collect_fallthrough_labels): Add new argument,
801 return location via that, don't modify statements.
802 (warn_implicit_fallthrough_r): Adjust call, don't use
803 statement location directly.
805 2018-11-14 Richard Biener <rguenther@suse.de>
808 * tree-data-ref.c (split_constant_offset): Add wrapper
809 allocating a cache hash-map.
810 (split_constant_offset_1): Cache results of expanding
811 expressions from SSA def stmts.
813 2018-11-14 Richard Biener <rguenther@suse.de>
816 * tree-data-ref.c (lambda_matrix_row_add): Change const1 argument
818 (lambda_vector_mult_const): Likewise.
819 (lambda_matrix_right_hermite): Use lambda_int temporaries.
821 2018-11-14 Wilco Dijkstra <wdijkstr@arm.com>
822 Jackson Woodruff <jackson.woodruff@arm.com>
824 PR 71026/tree-optimization
825 * match.pd: Simplify floating point comparisons.
827 2018-11-14 Jakub Jelinek <jakub@redhat.com>
829 PR rtl-optimization/87817
830 * config/i386/i386.c (ix86_fold_builtin): For _bzhi_u{32,64} if
831 last argument has low 8 bits clear, fold to 0.
833 2018-11-14 Iain Sandoe <iain@sandoe.co.uk>
835 * tree-vect-data-refs.c (vect_can_force_dr_alignment_p): Cast
836 MAX_OFILE_ALIGNMENT as needed.
838 2018-11-14 Richard Biener <rguenther@suse.de>
840 PR tree-optimization/88019
841 * graphite-isl-ast-to-gimple.c (ternary_op_to_tree): Rewrite
842 COND_EXPR arguments to non-trapping overflow.
844 2018-11-14 Claudiu Zissulescu <claziss@synopsys.com>
846 * config/arc/arc.h (reg_class): Reorder registers classes, remove
847 unused register classes.
848 (REG_CLASS_NAMES): Likewise.
849 (REG_CLASS_CONTENTS): Likewise.
850 (FIXED_REGISTERS): Make lp_count fixed.
851 (BASE_REG_CLASS): Remove ACC16_BASE_REGS reference.
852 (PROGRAM_COUNTER_REGNO): Remove.
853 * config/arc/arc.c (arc_conditional_register_usage): Remove unused
854 register classes, use constants for register numbers, remove
855 reg_class_contents references.
856 (arc_process_double_reg_moves): Add asserts.
857 (arc_secondary_reload): Remove LPCOUNT_REG reference, use
858 lra_in_progress predicate.
859 (arc_init_reg_tables): Remove unused register classes.
860 (arc_register_move_cost): Likewise.
861 (arc_preferred_reload_class): Likewise.
862 (hwloop_optimize): Update rtx patterns involving lp_count register.
863 (arc_return_address_register): Rename ILINK1, INLINK2 regnums macros.
864 * config/arc/constraints.md ("c"): Choose between GENERAL_REGS and
865 CHEAP_CORE_REGS. Former one will be used for LRA.
866 ("Rac"): Choose between GENERAL_REGS and ALL_CORE_REGS. Former
867 one will be used for LRA.
868 ("w"): Choose between GENERAL_REGS and WRITABLE_CORE_REGS. Former
869 one will be used for LRA.
870 ("W"): Choose between GENERAL_REGS and MPY_WRITABLE_CORE_REGS.
871 Former one will be used for LRA.
872 ("f"): Delete constraint.
875 ("l"): Change it from register constraint to constraint.
876 * config/arc/arc.md (movqi_insn): Remove unsed lp_count constraints.
877 (movhi_insn): Likewise.
878 (movsi_insn): Update pattern.
881 (stack_tie): Remove 'b' constraint letter.
883 (R9_REG, R15_REG, R16_REG, R25_REG): Likewise.
884 (R32_REG, R40_REG, R41_REG, R42_REG, R43_REG, R44_REG): Likewise.
885 (R57_REG, R59_REG, PCL_REG): Likewise.
886 (ILINK1_REGNUM): Renamed to ILINK1_REG.
887 (ILINK2_REGNUM): Renamed to ILINK2_REG.
890 (Rcw): Remove unused reg classes.
891 * config/arc/predicates.md (dest_reg_operand): Just default on
892 register_operand predicate.
893 (mpy_dest_reg_operand): Likewise.
894 (move_dest_operand): Use macros instead of constants.
895 * config/arc/arc.opt (mlra): Switch to lra as default.
897 2018-11-14 Richard Biener <rguenther@suse.de>
899 PR tree-optimization/87974
900 * tree-vect-loop.c (vectorizable_reduction): When computing
901 the vectorized reduction PHI vector type ignore constant
904 2018-11-14 Jakub Jelinek <jakub@redhat.com>
906 PR tree-optimization/87977
907 * tree-ssa-math-opts.c (optimize_recip_sqrt): Don't reuse division
908 stmt, build a new one and replace the old one with it. Formatting fix.
909 Call release_ssa_name (x) if !has_other_use and !delete_div.
910 (pass_cse_reciprocals::execute): Before calling optimize_recip_sqrt
911 verify lhs of stmt is still def.
913 2018-11-13 Peter Bergner <bergner@linux.ibm.com>
915 PR rtl-optimization/87507
916 * lower-subreg.c (operand_for_swap_move_operator): New function.
917 (simple_move): Strip simple operators.
918 (find_pseudo_copy): Likewise.
919 (resolve_operand_for_swap_move_operator): New function.
920 (resolve_simple_move): Strip simple operators and swap operands.
922 2018-11-13 Jakub Jelinek <jakub@redhat.com>
924 PR tree-optimization/87898
925 * omp-simd-clone.c (ipa_simd_modify_stmt_ops): Formatting fix.
926 (ipa_simd_modify_function_body): Remove debug stmts where the first
927 argument was changed into a non-decl.
929 2018-11-13 Peter Bergner <bergner@linux.ibm.com>
931 PR rtl-optimization/87899
932 * lra-lives.c (start_living): Update white space in comment.
933 (enum point_type): New.
934 (sparseset_contains_pseudos_p): New function.
935 (update_pseudo_point): Likewise.
936 (make_hard_regno_live): Use HARD_REGISTER_NUM_P macro.
937 (make_hard_regno_dead): Likewise. Remove ignore_reg_for_conflicts
938 handling. Move early exit after adding conflicts.
939 (mark_pseudo_live): Use HARD_REGISTER_NUM_P macro. Add early exit
940 if regno is already live. Remove all handling of program points.
941 (mark_pseudo_dead): Use HARD_REGISTER_NUM_P macro. Add early exit
942 after adding conflicts. Remove all handling of program points and
943 ignore_reg_for_conflicts.
944 (mark_regno_live): Use HARD_REGISTER_NUM_P macro. Remove return value
945 and do not guard call to mark_pseudo_live.
946 (mark_regno_dead): Use HARD_REGISTER_NUM_P macro. Remove return value
947 and do not guard call to mark_pseudo_dead.
948 (check_pseudos_live_through_calls): Use HARD_REGISTER_NUM_P macro.
949 (process_bb_lives): Use HARD_REGISTER_NUM_P and HARD_REGISTER_P macros.
950 Use new function update_pseudo_point. Handle register copies by
951 removing the source register from the live set. Handle INOUT operands.
952 Update to the next program point using the unused_set, dead_set and
954 (lra_create_live_ranges_1): Use HARD_REGISTER_NUM_P macro.
956 2018-11-13 David Malcolm <dmalcolm@redhat.com>
958 * builtins.c: Replace "source_location" with "location_t".
959 * diagnostic-show-locus.c: Likewise.
960 * diagnostic.c: Likewise.
961 * dumpfile.c: Likewise.
962 * gcc-rich-location.h: Likewise.
963 * genmatch.c: Likewise.
964 * gimple.h: Likewise.
965 * gimplify.c: Likewise.
967 * input.h: Likewise. Eliminate the typedef.
968 * omp-expand.c: Likewise.
969 * selftest.h: Likewise.
970 * substring-locations.h (get_source_location_for_substring):
972 (get_location_within_string): ...this.
973 * tree-cfg.c: Replace "source_location" with "location_t".
974 * tree-cfgcleanup.c: Likewise.
975 * tree-diagnostic.c: Likewise.
976 * tree-into-ssa.c: Likewise.
977 * tree-outof-ssa.c: Likewise.
978 * tree-parloops.c: Likewise.
979 * tree-phinodes.c: Likewise.
980 * tree-phinodes.h: Likewise.
981 * tree-ssa-loop-ivopts.c: Likewise.
982 * tree-ssa-loop-manip.c: Likewise.
983 * tree-ssa-phiopt.c: Likewise.
984 * tree-ssa-phiprop.c: Likewise.
985 * tree-ssa-threadupdate.c: Likewise.
986 * tree-ssa.c: Likewise.
987 * tree-ssa.h: Likewise.
988 * tree-vect-loop-manip.c: Likewise.
990 2018-11-13 Michael Ploujnikov <michael.ploujnikov@oracle.com>
992 * doc/extend.texi: Fix typo in the weakref description.
994 2018-11-13 Richard Biener <rguenther@suse.de>
996 PR tree-optimization/86991
997 * tree-vect-loop.c (vect_is_slp_reduction): Delay reduction
998 group building until we have successfully detected the SLP
1000 (vect_is_simple_reduction): Remove fixup code here.
1002 2018-11-13 David Malcolm <dmalcolm@redhat.com>
1004 * dumpfile.c (VERIFY_DUMP_ENABLED_P): New macro.
1005 (dump_gimple_stmt): Use it.
1006 (dump_gimple_stmt_loc): Likewise.
1007 (dump_gimple_expr): Likewise.
1008 (dump_gimple_expr_loc): Likewise.
1009 (dump_generic_expr): Likewise.
1010 (dump_generic_expr_loc): Likewise.
1011 (dump_printf): Likewise.
1012 (dump_printf_loc): Likewise.
1013 (dump_dec): Likewise.
1014 (dump_dec): Likewise.
1015 (dump_hex): Likewise.
1016 (dump_symtab_node): Likewise.
1017 * gimple-loop-interchange.cc (tree_loop_interchange::interchange):
1018 Guard dump call with dump_enabled_p.
1019 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl): Likewise.
1020 * graphite-optimize-isl.c (optimize_isl): Likewise.
1021 * graphite.c (graphite_transform_loops): Likewise.
1022 * tree-loop-distribution.c (pass_loop_distribution::execute): Likewise.
1023 * tree-parloops.c (parallelize_loops): Likewise.
1024 * tree-ssa-loop-niter.c (number_of_iterations_exit): Likewise.
1025 * tree-vect-data-refs.c (vect_analyze_group_access_1): Likewise.
1026 (vect_prune_runtime_alias_test_list): Likewise.
1027 * tree-vect-loop.c (vect_update_vf_for_slp): Likewise.
1028 (vect_estimate_min_profitable_iters): Likewise.
1029 * tree-vect-slp.c (vect_record_max_nunits): Likewise.
1030 (vect_build_slp_tree_2): Likewise.
1031 (vect_supported_load_permutation_p): Likewise.
1032 (vect_slp_analyze_operations): Likewise.
1033 (vect_slp_analyze_bb_1): Likewise.
1034 (vect_slp_bb): Likewise.
1035 * tree-vect-stmts.c (vect_analyze_stmt): Likewise.
1036 * tree-vectorizer.c (try_vectorize_loop_1): Likewise.
1037 (pass_slp_vectorize::execute): Likewise.
1038 (increase_alignment): Likewise.
1040 2018-11-13 David Malcolm <dmalcolm@redhat.com>
1043 * ipa-inline.c (report_inline_failed_reason): Guard calls to
1044 cl_target_option_print_diff and cl_optimization_print_diff with
1047 2018-11-13 David Malcolm <dmalcolm@redhat.com>
1049 * doc/invoke.texi (-fsave-optimization-record): Note that the
1050 output is compressed.
1051 * optinfo-emit-json.cc: Include <zlib.h>.
1052 (optrecord_json_writer::write): Compress the output.
1054 2018-11-13 Aldy Hernandez <aldyh@redhat.com>
1056 * tree-vrp.c (value_range_base::dump): Dump type.
1057 Do not use INF nomenclature for 1-bit types.
1058 (dump_value_range): Group all variants to common dumping code.
1059 (debug): New overloaded functions for value_ranges.
1060 (value_range_base::dump): Remove no argument version.
1061 (value_range::dump): Same.
1063 2018-11-13 Richard Biener <rguenther@suse.de>
1065 PR tree-optimization/87931
1066 * tree-vect-loop.c (vect_is_simple_reduction): Restrict
1067 nested cycles we support to latch computations vectorizable_reduction
1070 2018-11-13 Martin Liska <mliska@suse.cz>
1072 PR tree-optimization/87885
1073 * cfghooks.c (account_profile_record): Rename
1075 (profile_record_check_consistency): ... this.
1076 Calculate missing num_mismatched_freq_in.
1077 (profile_record_account_profile): New function
1078 that calculates time and size of a function.
1079 * cfghooks.h (struct profile_record): Remove
1081 (struct cfg_hooks): Remove after_pass flag.
1082 (account_profile_record): Rename to ...
1083 (profile_record_check_consistency): ... this.
1084 (profile_record_account_profile): New.
1085 * cfgrtl.c (rtl_account_profile_record): Remove
1087 * passes.c (check_profile_consistency): Do only
1089 (account_profile): Calculate size and time of
1091 (pass_manager::dump_profile_report): Reformat
1093 (execute_one_ipa_transform_pass): Call
1094 consistency check before clean upand call account_profile
1095 after a clean up is done.
1096 (execute_one_pass): Call check_profile_consistency and
1097 account_profile instead of using after_pass flag..
1098 * tree-cfg.c (gimple_account_profile_record): Likewise.
1100 2018-11-13 Martin Liska <mliska@suse.cz>
1103 * config/i386/i386.c (ix86_option_override_internal): Error
1104 about usage -mabi=ms and -fsanitize={,kernel-}address.
1106 2018-11-13 Andre Vieira <andre.simoesdiasvieira@arm.com>
1108 * config/aarch64/aarch64.c
1109 (aarch64_vectorize_preferred_vector_alignment): Change return type to
1111 (aarch64_simd_vector_alignment_reachable): Adapt to preferred vector
1112 alignment being a poly int.
1113 * doc/tm.texi (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): Change
1114 return type to poly_uint64.
1115 * target.def (default_preferred_vector_alignment): Likewise.
1116 * targhooks.c (default_preferred_vector_alignment): Likewise.
1117 * targhooks.h (default_preferred_vector_alignment): Likewise.
1118 * tree-vect-data-refs.c (vect_calculate_target_alignment): Likewise.
1119 (vect_compute_data_ref_alignment): Adapt to vector alignment being a
1121 (vect_update_misalignment_for_peel): Likewise.
1122 (vect_enhance_data_refs_alignment): Likewise.
1123 (vect_find_same_alignment_drs): Likewise.
1124 (vect_duplicate_ssa_name_ptr_info): Likewise.
1125 (vect_setup_realignment): Likewise.
1126 (vect_can_force_dr_alignment_p): Change alignment parameter type to
1128 * tree-vect-loop-manip.c (get_misalign_in_elems): Learn to construct a
1129 mask with a compile time variable vector alignment.
1130 (vect_gen_prolog_loop_niters): Adapt to vector alignment being a poly
1132 (vect_do_peeling): Exit early if vector alignment is not constant.
1133 * tree-vect-stmts.c (ensure_base_align): Adapt to vector alignment being
1135 (vectorizable_store): Likewise.
1136 (vectorizable_load): Likweise.
1137 * tree-vectorizer.h (struct dr_vec_info): Make target_alignment field a
1139 (vect_known_alignment_in_bytes): Adapt to vector alignment being a
1141 (vect_can_force_dr_alignment_p): Change alignment parameter type to
1144 2018-11-13 Richard Biener <rguenther@suse.de>
1146 PR tree-optimization/87962
1147 * tree-vect-loop.c (vect_is_simple_reduction): More reliably
1148 detect outer reduction for disqualifying in-loop uses.
1150 2018-11-13 Richard Biener <rguenther@suse.de>
1152 PR tree-optimization/87967
1153 * tree-vect-loop.c (vect_transform_loop): Also copy PHIs
1154 for constants for the scalar loop.
1156 2018-11-13 Alan Modra <amodra@gmail.com>
1158 * config/rs6000/linux64.h (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Exclude
1159 integer constants when -mcmodel=medium.
1161 2018-11-13 Claudiu Zissulescu <claziss@synopsys.com>
1163 * config/arc/arc.h (FUNCTION_PROFILER): Redefine to empty.
1164 * config/arc/elf.h (PROFILE_HOOK): Define.
1165 * config/arc/linux.h (PROFILE_HOOK): Likewise.
1167 2018-11-13 Claudiu Zissulescu <claziss@synopsys.com>
1169 * config/arc/arc.c (hwloop_optimize): Bailout when detecting a
1170 jump table data in the text section.
1172 2018-11-13 Claudiu Zissulescu <claziss@synopsys.com>
1174 * config/arc/arc.c (arc_eh_return_address_location): Repurpose it
1175 to fit the eh_return pattern.
1176 * config/arc/arc.md (eh_return): Define.
1177 (VUNSPEC_ARC_EH_RETURN): Likewise.
1178 * config/arc/arc-protos.h (arc_eh_return_address_location): Match
1180 * config/arc/arc.h (EH_RETURN_HANDLER_RTX): Remove it.
1183 2018-11-13 Claudiu Zissulescu <claziss@synopsys.com>
1185 * common/config/arc/arc-common.c (arc_option_optimization_table):
1186 Millicode optimization is default on for size optimizations.
1187 * config/arc/arc-protos.h (arc_check_multi): New function.
1188 * config/arc/arc.c (RTX_OK_FOR_OFFSET_P): Rearange.
1189 (ENTER_LEAVE_START_REG): Define.
1190 (ENTER_LEAVE_END_REG): Likewise.
1191 (arc_override_options): Disable millicode when long calls option
1193 (arc_frame_info): Change it from int to bool.
1194 (arc_compute_frame_size): Clean up.
1195 (arc_save_restore): Remove.
1196 (frame_save_reg): New function.
1197 (frame_restore_reg): Likewise.
1198 (arc_enter_leave_p): Likewise.
1199 (arc_save_callee_saves): Likewise.
1200 (arc_restore_callee_saves): Likewise.
1201 (arc_save_callee_enter): Likewise.
1202 (arc_restore_callee_leave): Likewise.
1203 (arc_save_callee_milli): Likewise.
1204 (arc_restore_callee_milli): Likewise.
1205 (arc_expand_prologue): Reimplement to emit enter/leave
1207 (arc_expand_epilogue): Likewise.
1208 (arc_check_multi): New function.
1209 * config/arc/arc.md (push_multi_fp): New pattern.
1210 (push_multi_fp_blink): Likewise.
1211 (pop_multi_fp): Likewise.
1212 (pop_multi_fp_blink): Likewise.
1213 (pop_multi_fp_ret): Likewise.
1214 (pop_multi_fp_blink_ret): Likewise.
1215 * config/arc/arc.opt (mmillicode): Update option.
1216 (mcode-density-frame): New option.
1217 * config/arc/predicates.md (push_multi_operand): New predicate.
1218 (pop_multi_operand): Likewise.
1219 * doc/invoke.texi (ARC): Update ARC options information.
1222 2018-11-13 Claudiu Zissulescu <claziss@synopsys.com>
1224 * config/arc/arc-protos.h (gen_operands_ldd_std): Add.
1225 * config/arc/arc.c (operands_ok_ldd_std): New function.
1226 (mem_ok_for_ldd_std): Likewise.
1227 (gen_operands_ldd_std): Likewise.
1228 * config/arc/arc.md: Add peephole2 rules for std/ldd.
1230 2018-11-13 Eric Botcazou <ebotcazou@adacore.com>
1232 * toplev.c (output_stack_usage): Turn test on flag_stack_usage into
1233 test on stack_usage_file.
1234 (lang_dependent_init): Do not open the .su file if generating LTO.
1236 2018-11-13 Jakub Jelinek <jakub@redhat.com>
1238 PR rtl-optimization/87918
1239 * simplify-rtx.c (simplify_merge_mask): For COMPARISON_P, use
1240 simplify_gen_relational rather than simplify_gen_binary.
1242 2018-11-13 Richard Biener <rguenther@suse.de>
1244 * tree-ssanames.h (set_range_info): Use value_range_base.
1245 (get_range_info): Likewise.
1246 * tree-ssanames.c (set_range_info): Likewise.
1247 (get_range_info): Likewise.
1248 * tree-vrp.c (value_range_base::union_helper): Split
1249 out common parts of value_range[_base]::union_.
1250 (value_range_base::union_): Update.
1251 (value_range::union_): Likewise.
1252 (determine_value_range_1): Use value_range_base.
1253 (determine_value_range): Likewise.
1254 * tree-vrp.h (value_range_base::union_helper): Move ...
1255 (value_range::union_helper): ... from here.
1257 2018-11-13 Alan Modra <amodra@gmail.com>
1259 * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Negate
1261 (rs6000_secondary_reload_gpr): Don't call find_replacement.
1263 2018-11-13 Sandra Loosemore <sandra@codesourcery.com>
1266 * doc/invoke.texi (Optimize Options): Clarify that the
1267 l1-cache-line-size, l1-cache-size, and l2-cache-size parameters
1268 apply to data cache size.
1270 2018-11-13 Alan Modra <amodra@gmail.com>
1272 * config/rs6000/predicates.md (logical_const_operand),
1273 (logical_operand): Correct comment.
1274 * config/rs6000/rs6000.c (rs6000_emit_prologue): Comment fix.
1276 2018-11-13 Alan Modra <amodra@gmail.com>
1278 * config/rs6000/rs6000.md (addsi3_high): Prefix with '*'.
1280 2018-11-13 Alan Modra <amodra@gmail.com>
1282 * gcc/doc/md.texi (Machine Constraints): Correct rs6000 h constraint
1284 * config/rs6000/rs6000.md (movsi_internal1): Delete MT%0 case
1285 covered by alternative.
1286 (movcc_internal1): Ignore h for register preference.
1287 (mov<mode>_hardfloat64): Likewise.
1288 (mov<mode>_softfloat): Ignore c, l, h for register preference.
1290 2018-11-12 Sandra Loosemore <sandra@codesourcery.com>
1292 PR preprocessor/47823
1293 * doc/cpp.texi (Alternatives to Wrapper #ifndef): Move #pragma once
1296 * doc/extend.texi (Pragmas): Note additional pragmas documented
1299 2018-11-13 Kugan Vivekanandarajah <kuganv@linaro.org>
1303 * tree-scalar-evolution.c (expression_expensive_p): Make
1304 BUILTIN POPCOUNT as expensive when backend does not define it.
1306 2018-11-12 Fredrik Noring <noring@nocrew.org>
1308 * config.gcc: Update with-llsc defaults for MIPS r5900.
1310 2018-11-12 Martin Liska <mliska@suse.cz>
1312 PR gcov-profile/87442
1313 * common.opt: Add -fprofile-filter-files and -fprofile-exclude-files
1315 * doc/invoke.texi: Document them.
1316 * tree-profile.c (parse_profile_filter): New.
1317 (parse_profile_file_filtering): Likewise.
1318 (release_profile_file_filtering): Likewise.
1319 (include_source_file_for_profile): Likewise.
1320 (tree_profiling): Filter source files based on the
1321 newly added options.
1323 2018-11-12 Martin Liska <mliska@suse.cz>
1326 * doc/extend.texi: Add missing values for __builtin_cpu_is and
1327 __builtin_cpu_supports for x86 target.
1329 2018-11-12 Sandra Loosemore <sandra@codesourcery.com>
1332 * doc/rtl.texi (Constants): Clarify that mode of "high" doesn't
1334 (Arithmetic): Likewise for "lo_sum".
1336 2018-11-12 Renlin Li <renlin.li@arm.com>
1339 * dse.c (get_stored_val): Add check for compile-time constantness
1342 2018-11-12 Sudakshina Das <sudi.das@arm.com>
1344 * config/arm/arm-cpus.in (armv8_5, sb, predres): New features.
1345 (ARMv8_5a): New fgroup.
1346 (armv8.5-a): New arch.
1347 (armv8-a, armv8.1-a, armv8.2-a, armv8.3-a, armv8.4-a): New options
1349 * config/arm/arm-tables.opt: Regenerate.
1350 * config/arm/t-aprofile: Add matching rules for -march=armv8.5-a
1351 * config/arm/t-arm-elf (all_v8_archs): Add armv8.5-a.
1352 * config/arm/t-multilib (v8_5_a_simd_variants): New variable.
1353 Add matching rules for -march=armv8.5-a and extensions.
1354 * doc/invoke.texi (ARM options): Document -march=armv8.5-a.
1355 Add sb and predres to all armv8-a except armv8.5-a.
1357 2018-11-12 Richard Biener <rguenther@suse.de>
1359 * tree-vrp.h (value_range[_base]::set): Make public. Provide
1360 overload for single value.
1361 (value_range[_base]::set_nonnull): New.
1362 (value_range[_base]::set_null): Likewise.
1363 (value_range): Document bitmap copying behavior, mark
1364 copy constructor and assignment operator deleted.
1365 (value_range::move): New.
1366 (value_range::set_and_canonicalize): Default bitmap to zero.
1367 (set_value_range_to_nonnull): Remove.
1368 (set_value_range_to_null): Likewise.
1369 (set_value_range): Likewise.
1370 (set_value_range_to_value): Likewise.
1371 (extract_range_from_unary_expr): Work on value_range_base.
1372 (extract_range_from_binary_expr_1): Likewise. Rename to...
1373 (extract_range_from_binary_expr): ... this.
1374 * tree-vrp.c (value_range::update): Clear equiv bitmap
1376 (value_range::move): New, move equiv bitmap.
1377 (value_range_base::set_undefined): Avoid assignment.
1378 (value_range::set_undefined): Likewise.
1379 (value_range_base::set_varying): Likewise.
1380 (value_range::set_varying): Likewise.
1381 (set_value_range): Remove.
1382 (value_range_base::set): New overload for value.
1383 (value_range::set): Likewise.
1384 (set_value_range_to_nonnull): Remove.
1385 (value_range_base::set_nonnull): New.
1386 (value_range::set_nonnull): Likewise.
1387 (set_value_range_to_null): Remove.
1388 (value_range_base::set_null): New.
1389 (value_range::set_null): Likewise.
1390 (range_is_null): Work on value_range_base.
1391 (range_is_nonnull): Likewise.
1392 (ranges_from_anti_range): Likewise.
1393 (extract_range_into_wide_ints): Likewise.
1394 (extract_range_from_multiplicative_op): Likewise.
1395 (extract_range_from_binary_expr): Likewise. Update for API changes.
1396 (extract_range_from_unary_expr): Likewise. Remove OBJ_TYPE_REF
1398 (value_range::intersect_helper): Avoid copy and assignment.
1399 (value_range::union_helper): Likewise.
1400 (determine_value_range_1): Adjust.
1401 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::try_find_new_range):
1402 Avoid assignment by using move.
1403 (evrp_range_analyzer::record_ranges_from_stmt): Avoid assignment.
1404 * tree-ssa-threadedge.c (record_temporary_equivalences_from_phis):
1406 * tree-ssanames.c (get_range_info): Likewise.
1407 * vr-values.h (vr_values::get_vr_for_comparison): Adjust API.
1408 * vr-values.c (vr_values::get_value_range): Adjust.
1409 (vr_values::update_value_range): Likewise.
1410 (symbolic_range_based_on_p): Work on value_range_base.
1411 (vr_values::extract_range_from_binary_expr): Use value_range_base.
1412 (vr_values::extract_range_from_unary_expr): Likewise.
1413 (vr_values::extract_range_from_cond_expr): Avoid assignment.
1414 (vr_values::extract_range_from_comparison): Adjust.
1415 (vr_values::check_for_binary_op_overflow): Use value_range_base.
1416 (vr_values::extract_range_basic): Adjust.
1417 (vr_values::adjust_range_with_scev): Likewise.
1418 (vr_values::vrp_visit_assignment_or_call): Likewise.
1419 (vr_values::get_vr_for_comparison): Change API to avoid
1420 assignment and copy construction.
1421 (vr_values::compare_name_with_value): Adjust accordingly.
1422 (vr_values::compare_names): Likewise.
1423 (vr_values::extract_range_from_phi_node): Avoid assignment and
1424 bogus in-place modify of equiv bitmap.
1425 (vr_values::simplify_bit_ops_using_ranges): Use value_range_base.
1426 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Adjust
1427 for extract_range_from_unary_expr API change.
1428 * ipa-cp.c (ipa_vr_operation_and_type_effects): Likewise.
1430 2018-11-12 Eric Botcazou <ebotcazou@adacore.com>
1432 * config/mcore/mcore.h (WORD_REGISTER_OPERATIONS): Remove duplicate.
1434 2018-11-12 Richard Biener <rguenther@suse.de>
1436 * tree-vrp.h (value_range_base::symbolic_p,
1437 value_range_base::constant_p, value_range_base::zero_p,
1438 value_range_base::singleton_p): Move from value_range.
1439 (value_range::dump): Add.
1440 * gimple-ssa-evrp-analyze.c
1441 (evrp_range_analyzer::record_ranges_from_phis): Use set_varying.
1442 * ipa-cp.c (ipcp_vr_lattice::print): Use dump_value_range.
1443 * tree-ssa-threadedge.c (record_temporary_equivalences_from_phis):
1445 * tree-vrp.c (value_range::symbolic_p): Move to value_range_base.
1446 (value_range::constant_p): Likewise.
1447 (value_range::singleton_p): Likewise.
1448 (value_range_base::dump): Add.
1449 (set_value_range_to_undefined): Remove.
1450 (set_value_range_to_varying): Likewise.
1451 (range_int_cst_p): Take value_range_base argument.
1452 (range_int_cst_singleton_p): Likewise.
1453 (value_range_constant_singleton): Likewise.
1454 (vrp_set_zero_nonzero_bits): Likewise.
1455 (extract_range_from_multiplicative_op): Use set_varying.
1456 (extract_range_from_binary_expr_1): Likewise. Use set_undefined.
1457 (extract_range_from_unary_expr): Likewise.
1458 (dump_value_range_base): Change to overload of dump_value_range.
1459 (vrp_prop::vrp_initialize): Use set_varying and set_undefined.
1460 (vrp_prop::visit_stmt): Likewise.
1461 (value_range::intersect_helper): Likewise.
1462 (value_range::union_helper): Likewise.
1463 (determine_value_range_1): Likewise.
1465 2018-11-12 Richard Biener <rguenther@suse.de>
1467 * tree-vrp.c (set_value_range_to_nonnull): Clear equiv.
1468 (set_value_range_to_null): Likewise.
1469 * vr-values.c (vr_values::extract_range_from_comparison):
1470 Clear equiv for constant singleton ranges.
1472 2018-11-12 Wei Xiao <wei3.xiao@intel.com>
1474 * config/i386/sse.md: Combine VFIXUPIMM* patterns
1475 (<avx512>_fixupimm<mode>_maskz<round_saeonly_expand_name>): Update.
1476 (<avx512>_fixupimm<mode><sd_maskz_name><round_saeonly_name>): Update.
1477 (<avx512>_fixupimm<mode>_mask<round_saeonly_name>): Remove.
1478 (avx512f_sfixupimm<mode>_maskz<round_saeonly_expand_name>): Update.
1479 (avx512f_sfixupimm<mode><sd_maskz_name><round_saeonly_name>): Update.
1480 (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Remove.
1482 2018-11-11 Sandra Loosemore <sandra@codesourcery.com>
1485 * doc/extend.texi (Common Type Attributes): For the align type
1486 attribute, copy language about decreasing alignment from the
1487 corresponding variable attribute.
1489 2018-11-11 Xianmiao Qu <xianmiao_qu@c-sky.com>
1491 * config/csky/csky.md (*fpuv2_nmulsf3_1, *fpuv2_nmuldf3_1): Handle
1494 2018-11-11 Sandra Loosemore <sandra@codesourcery.com>
1497 * doc/invoke.texi (C++ Dialect Options): Add warning about mixing
1498 -frtti and -fno-rtti code.
1500 2018-11-11 Sandra Loosemore <sandra@codesourcery.com>
1503 * doc/extend.texi (Other Builtins): Document probability associated
1504 with __builtin_expect.
1506 2018-11-11 Uros Bizjak <ubizjak@gmail.com>
1509 * config/i386/i386.h (STACK_BOUNDARY): Use TARGET_64BIT_MS_ABI
1510 instead of (TARGET_64BIT && ix86_abi == MS_ABI).
1511 * config/i386/darwin.h (STACK_BOUNDARY): Ditto.
1512 * config/i386/cygming.h (STACK_BOUNDARY): Remove.
1514 2018-11-11 Xianmiao Qu <xianmiao_qu@c-sky.com>
1516 * config/csky/csky-linux-elf.h (CC1_SPEC): Support -profile.
1518 2018-11-11 Xianmiao Qu <xianmiao_qu@c-sky.com>
1520 * config/csky/csky.h (ASM_PREFERRED_EH_DATA_FORMAT): Define.
1522 2018-11-11 Richard Biener <rguenther@suse.de>
1524 * tree-vrp.h (class value_range_base): New base class for
1525 value_range containing all but the m_equiv member.
1526 (dump_value_range_base): Add.
1527 (range_includes_zero_p): Work on value_range_base.
1528 * tree-vrp.c (value_range_base::set): Split out base handling
1530 (value_range::set): this.
1531 (value_range::set_equiv): New.
1532 (value_range_base::value_range_base): New constructors.
1533 (value_range_base::check): Split out base handling from...
1534 (value_range::check): this.
1535 (value_range::equal_p): Refactor in terms of
1536 ignore_equivs_equal_p which is now member of the base.
1537 (value_range_base::set_undefined): New.
1538 (value_range_base::set_varying): Likewise.
1539 (value_range_base::dump):Split out base handling from...
1540 (value_range::dump): this.
1541 (value_range_base::set_and_canonicalize): Split out base handling
1543 (value_range::set_and_canonicalize): this.
1544 (value_range_base::union_): New.
1545 * ipa-prop.h (struct ipa_jump_func): Use value_range_base *
1547 * ipa-cp.c (class ipcp_vr_lattice): Use value_range_base
1548 instead of value_range everywhere.
1549 (ipcp_vr_lattice::print): Use dump_value_range_base.
1550 (ipcp_vr_lattice::meet_with): Adjust.
1551 (ipcp_vr_lattice::meet_with_1): Likewise.
1552 (ipa_vr_operation_and_type_effects): Likewise.
1553 (propagate_vr_across_jump_function): Likewise.
1554 * ipa-prop.c (struct ipa_vr_ggc_hash_traits): Likewise.
1555 (ipa_get_value_range): Likewise.
1556 (ipa_set_jfunc_vr): Likewise.
1557 (ipa_compute_jump_functions_for_edge): Likewise.
1559 2018-11-10 Sandra Loosemore <sandra@codesourcery.com>
1562 * doc/invoke.texi (Optimize Options): Add @opindex entries
1563 for the positive forms of -fno-xxx and -mno-xxx options
1564 that were lacking them.
1566 2018-11-10 Segher Boessenkool <segher@kernel.crashing.org>
1568 * combine.c (make_more_copies): Only make an intermediate copy if the
1569 dest of a move is a pseudo.
1571 2018-11-09 Maya Rashish <coypu@sdf.org>
1574 * config/netbsd-elf.h (NETBSD_STARTFILE_SPEC): Use crtbeginS.o for PIE.
1575 (NETBSD_ENDFILE_SPEC): Use crtendS.o for PIE.
1577 2018-11-09 Sandra Loosemore <sandra@codesourcery.com>
1581 * doc/invoke.texi (Optimize Options): Clarify default behavior
1582 for -fno-toplevel-reorder, -fno-defer-pop, and -fno-branch-count-reg.
1584 2018-11-09 Bernd Edlinger <bernd.edlinger@hotmail.de>
1586 PR tree-optimization/87940
1587 * expr.c (string_constant): Don't strip NOPS in subexpressions.
1588 Fold PLUS_EXPR correctly.
1590 2018-11-09 Ilya Leoshkevich <iii@linux.ibm.com>
1593 * config/s390/s390.c (s390_safe_relative_long_p): New function.
1594 (annotate_constant_pool_refs): Skip insns which support
1595 relative addressing.
1596 (annotate_constant_pool_refs_1): New helper function.
1597 (find_constant_pool_ref): Skip insns which support relative
1599 (find_constant_pool_ref_1): New helper function.
1600 (replace_constant_pool_ref): Skip insns which support
1601 relative addressing.
1602 (replace_constant_pool_ref_1): New helper function.
1603 (s390_mainpool_start): Adapt to the new signature.
1604 (s390_mainpool_finish): Likewise.
1605 (s390_chunkify_start): Likewise.
1606 (s390_chunkify_finish): Likewise.
1607 (pass_s390_early_mach::execute): Likewise.
1608 (s390_prologue_plus_offset): Likewise.
1609 (s390_emit_prologue): Likewise.
1610 (s390_emit_epilogue): Likewise.
1612 2018-11-09 Jakub Jelinek <jakub@redhat.com>
1614 * gimplify.c (gimplify_scan_omp_clauses): Call sorry_at for valid
1615 but unsupported lastprivate with conditional modifier.
1617 2018-11-09 Jeff Law <law@redhat.com>
1619 * config/mips/mips.c (mips_loongson_ext2_prefetch_cookie): Handle
1620 unused argument better. Add gcc_unreachable to silence warning.
1622 2018-11-09 Martin Sebor <msebor@redhat.com>
1625 * attribs.c (has_attribute): New helper function.
1626 (decls_mismatched_attributes, maybe_diag_alias_attributes): Same.
1627 * attribs.h (decls_mismatched_attributes): Declare.
1628 * cgraphunit.c (handle_alias_pairs): Call maybe_diag_alias_attributes.
1629 (maybe_diag_incompatible_alias): Use OPT_Wattribute_alias_.
1630 * common.opt (-Wattribute-alias): Take an argument.
1631 (-Wno-attribute-alias): New option.
1632 * doc/extend.texi (Common Function Attributes): Document copy.
1633 (Common Variable Attributes): Same.
1634 * doc/invoke.texi (-Wmissing-attributes): Document enhancement.
1635 (-Wattribute-alias): Document new option argument.
1637 2018-11-09 Richard Earnshaw <rearnsha@arm.com>
1639 * config/arm/parsecpu.awk (/alias/): Tighten invisible alias
1640 matching criteria. Remove unused array initializer.
1642 2018-11-09 Bill Schmidt <wschmidt@linux.ibm.com>
1643 Jinsong Ji <jji@us.ibm.com>
1645 * config/rs6000/xmmintrin.h (_mm_cvtss_si32): Fix incorrect
1646 constraints by introducing a new temporary.
1647 (_mm_cvtss_si64): Likewise.
1649 2018-11-09 Martin Liska <mliska@suse.cz>
1651 * common.opt: Add -fipa-stack-alignment flag.
1652 * doc/invoke.texi: Document it.
1653 * final.c (rest_of_clean_state): Guard stack
1654 shrinking with flag.
1656 2018-11-09 Martin Liska <mliska@suse.cz>
1658 * cgraph.h (ipa_discover_readonly_nonaddressable_vars): Rename
1660 (ipa_discover_variable_flags): ... this.
1661 * common.opt: Come up with new flag -fipa-reference-addressable.
1662 * doc/invoke.texi: Document it.
1663 * ipa-reference.c (propagate): Call the renamed fn.
1664 * ipa-visibility.c (whole_program_function_and_variable_visibility):
1666 * ipa.c (ipa_discover_readonly_nonaddressable_vars): Renamed to
1668 (ipa_discover_variable_flags): ... this. Discover
1669 non-addressable variables only with the newly added flag.
1670 * opts.c: Enable the newly added flag with -O1 and higher
1673 2018-11-09 David Malcolm <dmalcolm@redhat.com>
1675 * json.cc (selftest::test_writing_literals): Fix comment.
1677 2018-11-09 Martin Liska <mliska@suse.cz>
1679 * config/aarch64/aarch64.c (aarch64_parse_arch): Do not copy
1680 string to a stack buffer.
1681 (aarch64_parse_cpu): Likewise.
1682 (aarch64_parse_tune): Likewise.
1684 2018-11-09 Richard Biener <rguenther@suse.de>
1686 PR tree-optimization/87953
1687 * tree-vect-loop.c (vectorizable_reduction): For analysis
1688 always pass ops[0] to vectorizable_condition.
1690 2018-11-09 Stafford Horne <shorne@gmail.com>
1691 Richard Henderson <rth@twiddle.net>
1692 Joel Sherrill <joel@rtems.org>
1694 * common/config/or1k/or1k-common.c: New file.
1695 * config/or1k/*: New.
1696 * config.gcc (or1k*-*-*): New.
1697 * configure.ac (or1k*-*-*): New test for openrisc tls.
1698 * configure: Regenerated.
1699 * doc/install.texi: Document OpenRISC triplets.
1700 * doc/invoke.texi: Document OpenRISC arguments.
1701 * doc/md.texi: Document OpenRISC.
1703 2018-11-09 Richard Earnshaw <rearnsha@arm.com>
1705 * config/arm/arm-cpus.in (arm7tdmi): Add an alias for arm7tdmi-s.
1706 (arm7tdmi-s): Delete CPU.
1707 (arm710t): Add aliases for arm720t and arm740t.
1708 (arm720t, arm740t): Delete CPUs.
1709 (arm920t): Add aliases for arm920, arm922t and arm940t.
1710 (arm920, arm922t, arm940t): Delete CPUs.
1711 (arm10tdmi): Add alias for arm1020t.
1712 (arm1020t): Delete CPU.
1713 (arm9e): Add aliases for arm946e-s, arm966e-s and arm968e-s.
1714 (arm946e-s, arm966e-s, arm968e-s): Delete CPUs.
1715 (arm10e): Add aliases for arm1020e and arm1022e.
1716 (arm1020e, arm1022e): Delete CPU.
1717 * config/arm/arm.md (generic_sched): Remove entries that are now
1719 (generic_vfp): Likewise.
1720 * config/arm/arm1020e.md: Simplify tuning selection based on alias
1722 * config/arm/arm-tune.md: Regenerated.
1723 * config/arm/arm-tables.opt: Regenerated.
1725 2018-11-09 Richard Biener <rguenther@suse.de>
1727 PR tree-optimization/87621
1728 * tree-vect-loop.c (vectorizable_reduction): Handle reduction
1729 op with only phi inputs.
1730 * tree-ssa-loop-ch.c: Include tree-ssa-sccvn.h.
1731 (ch_base::copy_headers): Run CSE on copied loop headers.
1732 (pass_ch_vect::process_loop_p): Simplify.
1734 2018-11-09 Alexandre Oliva <oliva@adacore.com>
1736 * config/i386/mingw32.h (LINK_SPEC_LARGE_ADDR_AWARE): Adjust
1737 for non-w64 x86_64 biarch.
1739 2018-11-09 Alexandre Oliva <aoliva@redhat.com>
1741 PR rtl-optimization/86438
1742 * compare-elim.c (try_eliminate_compare): Use SET_SRC instead
1743 of in_b for the compare if in_b is SET_DEST.
1746 * config/i386/i386.c (ix86_const_not_ok_for_debug_p): Reject
1747 non-toplevel UNSPEC.
1749 2018-11-08 Aldy Hernandez <aldyh@redhat.com>
1751 * tree-vrp.c (value_range::check): Do not access internals
1753 (value_range::singleton_p): Same.
1754 (value_range::type): Same.
1755 (vrp_finalize): Use value_range API.
1757 2018-11-08 Aldy Hernandez <aldyh@redhat.com>
1759 * tree-vrp.c (may_contain_p): Do not access m_min/m_max directly.
1761 2018-11-08 Aldy Hernandez <aldyh@redhat.com>
1763 * vr-values.c (vr_values::get_value_range): Use value_range API
1764 instead of piecing together ranges.
1765 (vr_values::update_value_range): Same.
1767 2018-11-08 Aldy Hernandez <aldyh@redhat.com>
1769 * gimple-fold.c (size_must_be_zero_p): Use value_range API instead
1770 of performing ad-hoc calculations.
1771 * tree-ssanames.c (set_range_info): New overloaded function
1772 accepting value_range &.
1773 (get_range_info): Same.
1774 * tree-ssanames.h (set_range_info_raw): Remove.
1775 (set_range_info): New prototype.
1776 (get_range_info): Same.
1777 * tree-vrp.h (value_range::null_p): Rename to zero_p.
1778 * tree-vrp.c (value_range::null_p): Same.
1780 2018-11-09 Jan Hubicka <jh@suse.cz>
1782 * tree.c (fld_type_variant_equal_p): Test user align flag.
1783 (flt_type_variant): Copy user align flag.
1784 (fld_incomplete_type_of): Clear it.
1786 2018-11-09 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
1788 * config/arm/neon.md (div<mode>3): New pattern.
1790 2018-11-08 Andi Kleen <ak@linux.intel.com>
1792 * common/config/i386/i386-common.c (OPTION_MASK_ISA_PTWRITE_SET): New.
1793 (OPTION_MASK_ISA_PTWRITE_UNSET): New.
1794 (ix86_handle_option): Handle OPT_mptwrite.
1795 * config/i386/cpuid.h (bit_PTWRITE): Add.
1796 * config/i386/driver-i386.c (host_detect_local_cpu): Detect ptwrite.
1797 * config/i386/i386-builtin.def (BDESC): Add ptwrite32/64.
1798 * config/i386/i386-c.c (ix86_target_macros_internal): Define __PTWRITE__.
1799 * config/i386/i386.c (ix86_target_string): Handle ptwrite.
1800 (ix86_option_override_internal): Handle PTA_PTWRITE.
1801 (ix86_valid_target_attribute_inner_p): Define ptwrite.
1802 (def_builtin2): Force UINT64 to be 64bit only.
1803 * config/i386/i386.h (TARGET_PTWRITE): Add.
1804 (TARGET_PTWRITE_P): Add.
1806 * config/i386/i386.md: Define ptwrite.
1807 * config/i386/i386.opt: Add -mptwrite.
1808 * config/i386/immintrin.h (_ptwrite64): Add.
1810 * doc/extend.texi: Document __builtin_ia32_ptwrite*.
1811 * doc/invoke.texi: Document -mptwrite.
1813 2018-11-08 Peter Bergner <bergner@linux.ibm.com>
1815 PR rtl-optimization/87600
1816 * cfgexpand.c (expand_asm_stmt): Catch illegal asm constraint usage.
1817 * lra-constraints.c (process_alt_operands): Skip illegal hard
1818 register usage. Prefer reloading non hard register operands.
1820 2018-11-08 Sandra Loosemore <sandra@codesourcery.com>
1823 * doc/invoke.texi (Optimize Options): Clarify default behavior
1824 for -fno-sched-interblock and -fno-sched-spec.
1826 2018-11-08 Roman Geissler <roman.geissler@amadeus.com>
1828 * collect2.c (linker_select): Add USE_LLD_LD.
1829 (ld_suffixes): Add ld.lld.
1830 (main): Handle -fuse-ld=lld.
1831 * common.opt (-fuse-ld=lld): New option.
1832 * doc/invoke.texi (-fuse-ld=lld): Document.
1833 * opts.c (common_handle_option): Handle OPT_fuse_ld_lld.
1835 2018-11-08 Paul Koning <ni1d@arrl.net>
1837 * config/pdp11/constraints.md: Add "Z" series constraints for use
1838 with pre-dec and post-inc addressing.
1839 * config/pdp11/pdp11-protos.m (expand_block_move): Delete.
1840 (pdp11_expand_operands): Add int argument (word count).
1841 (pdp11_sp_frame_offset): Delete.
1842 (pdp11_cmp_length): New function.
1843 (pushpop_regeq): New function.
1844 * config/pdp11/pdp11.c (TARGET_STACK_PROTECT_RUNTIME_ENABLED_P):
1846 (pdp11_expand_prologue, pdp11_expand_epilogue): Rewrite for new
1848 (pdp11_initial_elimination_offset): Ditto.
1849 (pdp11_expand_operands): Add word count argument. Bugfixes.
1850 (output_move_multiple): Change how pointer adjustment is done.
1851 (pdp11_gen_int_label): Correct format.
1852 (output_ascii): Ditto.
1853 (pdp11_asm_output_var): Add code for DEC assembler case.
1854 (pdp11_asm_print_operand): Bugfix for CONST_DOUBLE holding integer
1856 (legitimate_const_double_p): Ditto.
1857 (pdp11_register_move_cost): Adjust for new register classes.
1858 (pdp11_regno_reg_class): Ditto.
1859 (expand_block_move): Delete.
1860 (pushpop_regeq): New function.
1861 (pdp11_legitimate_address_p): Bugfix in check for constant
1863 (pdp11_sp_frame_offset): Delete.
1864 (pdp11_reg_save_size): New helper function for new frame layout.
1865 (output_addr_const_pdp11): Remove CONST_DOUBLE case.
1866 (pdp11_expand_shift): Bugfix in check for constant shift count.
1867 (pdp11_shift_length): Ditto.
1868 (pdp11_assemble_shift): Copy input to pdp11_expand_operands.
1869 (pdp11_cmp_length): New function.
1870 * config/pdp11/pdp11.h (TARGET_CPU_CPP_BUILTINS): Add macros for
1871 some compile options.
1872 (FIXED_REGISTERS): Remove HARD_FRAME_POINTER_REGNUM.
1873 (CALL_USED_REGISTERS): Ditto.
1874 (ELIMINABLE_REGS): Ditto.
1875 (REGISTER_NAMES): Ditto.
1876 (reg_class): Add classes NOTR0_REG through NOTSP_REG for use by Z
1878 (REG_CLASS_NAMES): Ditto.
1879 (REG_CLASS_CONTENTS): Ditto. Also remove
1880 HARD_FRAME_POINTER_REGNUM.
1881 (CPU_REG_CLASS): New macro.
1882 (CLASS_MAX_NREGS): Adjust for new register classes.
1883 (FUNCTION_PROFILER): Make no-op.
1884 (may_call_alloca): Remove unused declaration.
1885 (ASM_OUTPUT_ALIGN): Add workaround for PR87795.
1886 (ASM_OUTPUT_SKIP): Fix format.
1887 * config/pdp11/pdp11.md (unspecv): Add UNSPECV_MOVMEM.
1888 (HARD_FRAME_POINTER_REGNUM): Remove.
1890 (*rts): Rename. Remove epilogue related checks.
1891 (cmpsi, cmpdi): New insn.
1892 (cbranch<mode>4): Change to apply to SI and DI modes as well.
1893 (mov<mode>): Change constraints to enforce that push/pop
1894 destination cannot use the same register as source.
1895 (*mov<mode><cc_cc>): Ditto.
1896 (movmemhi, movmemhi1, movmemhi_nocc): Change to expand block move
1897 at assembly output rather than as RTL expander.
1898 (zero_extendqihi2): Bugfix in check for same registers.
1899 (adddi3_nocc): Bugfix in check for constant operand.
1900 (addsi3_nocc): Ditto.
1901 (subdi3_nocc): Ditto.
1902 (subsi3_nocc): Ditto.
1903 (negdi2_nocc): Copy input to pdp11_expand_operands.
1904 (negsi2_nocc): Ditto.
1905 (bswap2_nocc): Ditto.
1906 * config/pdp11/pdp11.opt (mlra): Fix documentation.
1907 * config/pdp11/t-pdp11: Use -Os.
1909 2018-11-08 Richard Earnshaw <rearnsha@arm.com>
1911 * config/arm/parsecpu.awk (/alias/): New parsing rule.
1912 (/begin cpu/): Check that the cpu name hasn't been previously defined.
1913 (gen_comm_data): Print out CPU alias tables.
1914 (check_cpu): Match aliases when checking the CPU name.
1915 * config/arm/arm-protos.h (cpu_alias): New structure.
1916 (cpu_option): Add entry for aliases.
1917 * config/arm/arm-cpus.in (strongarm): Add aliases for strongarm110
1918 strongarm1100 and strongarm1110.
1919 (strongarm110, strongarm1100, strongarm1110): Delete CPU entries.
1920 (config/arm/arm-generic.md): Remove redundant references to
1921 strongarm110, strongarm1100 and strongarm1110.
1922 * common/config/arm/arm-common.c (arm_print_hint_for_cpu_option):
1923 Scan aliases for additional hints.
1924 (arm_parse_cpu_option_name): Also match a cpu name against the list
1926 * config/arm/arm-tables.opt: Regenerated.
1927 * config/arm/arm-tune.md: Regenerated.
1929 2018-11-08 Jakub Jelinek <jakub@redhat.com>
1931 * builtin-types.def (BT_FN_VOID_BOOL, BT_FN_VOID_SIZE_SIZE_PTR,
1932 BT_FN_UINT_UINT_PTR_PTR, BT_FN_UINT_OMPFN_PTR_UINT_UINT,
1933 BT_FN_BOOL_UINT_LONGPTR_LONG_LONG_LONGPTR_LONGPTR_PTR_PTR,
1934 BT_FN_BOOL_UINT_ULLPTR_LONG_ULL_ULLPTR_ULLPTR_PTR_PTR,
1935 BT_FN_BOOL_LONG_LONG_LONG_LONG_LONG_LONGPTR_LONGPTR_PTR_PTR,
1936 BT_FN_BOOL_BOOL_ULL_ULL_ULL_LONG_ULL_ULLPTR_ULLPTR_PTR_PTR): New.
1937 * gengtype.c (open_base_files): Add omp-general.h.
1938 * gimple.c (gimple_build_omp_critical):
1939 (gimple_build_omp_taskgroup): Add CLAUSES argument. Call
1940 gimple_omp_taskgroup_set_clauses.
1941 (gimple_build_omp_atomic_load): Add mo argument, call
1942 gimple_omp_atomic_set_memory_order.
1943 (gimple_build_omp_atomic_store): Likewise.
1944 (gimple_copy): Adjust handling of GIMPLE_OMP_TASKGROUP.
1945 * gimple.def (GIMPLE_OMP_TASKGROUP): Use GSS_OMP_SINGLE_LAYOUT
1947 (GIMPLE_OMP_TEAMS): Use GSS_OMP_PARALLEL_LAYOUT instead
1948 of GSS_OMP_SINGLE_LAYOUT, adjust comments.
1949 * gimple.h (enum gf_mask): Add GF_OMP_TEAMS_HOST, GF_OMP_TASK_TASKWAIT
1950 and GF_OMP_ATOMIC_MEMORY_ORDER. Remove GF_OMP_ATOMIC_SEQ_CST, use
1951 different value for GF_OMP_ATOMIC_NEED_VALUE.
1952 (struct gimple_statement_omp_taskreg): Add GIMPLE_OMP_TEAMS to
1954 (struct gimple_statement_omp_single_layout): And remove here.
1955 (struct gomp_teams): Inherit from gimple_statement_omp_taskreg rather
1956 than gimple_statement_omp_single_layout.
1957 (is_a_helper <gimple_statement_omp_taskreg *>::test): Allow
1959 (is_a_helper <const gimple_statement_omp_taskreg *>::test): Likewise.
1960 (gimple_omp_subcode): Formatting fix.
1961 (gimple_omp_teams_child_fn, gimple_omp_teams_child_fn_ptr,
1962 gimple_omp_teams_set_child_fn, gimple_omp_teams_data_arg,
1963 gimple_omp_teams_data_arg_ptr, gimple_omp_teams_set_data_arg,
1964 gimple_omp_teams_host, gimple_omp_teams_set_host,
1965 gimple_omp_task_taskwait_p, gimple_omp_task_set_taskwait_p,
1966 gimple_omp_taskgroup_clauses, gimple_omp_taskgroup_clauses_ptr,
1967 gimple_omp_taskgroup_set_clauses): New inline functions.
1968 (gimple_build_omp_atomic_load): Add enum omp_memory_order argument.
1969 (gimple_build_omp_atomic_store): Likewise.
1970 (gimple_omp_atomic_seq_cst_p): Remove.
1971 (gimple_omp_atomic_memory_order): New function.
1972 (gimple_omp_atomic_set_seq_cst): Remove.
1973 (gimple_omp_atomic_set_memory_order): New function.
1974 (gimple_build_omp_taskgroup): Add clauses argument.
1975 * gimple-pretty-print.c (dump_gimple_omp_taskgroup): New function.
1976 (dump_gimple_omp_task): Print taskwait with depend clauses.
1977 (dump_gimple_omp_atomic_load, dump_gimple_omp_atomic_store): Use
1978 dump_omp_atomic_memory_order.
1979 (pp_gimple_stmt_1): Handle GIMPLE_OMP_TASKGROUP.
1980 * gimplify.c (enum gimplify_omp_var_data): Add GOVD_MAP_ALLOC_ONLY,
1981 GOVD_MAP_FROM_ONLY and GOVD_NONTEMPORAL.
1982 (enum omp_region_type): Reserve bits 1 and 2 for auxiliary flags,
1983 renumber values of most of ORT_* enumerators, add ORT_HOST_TEAMS,
1984 ORT_COMBINED_HOST_TEAMS, ORT_TASKGROUP, ORT_TASKLOOP and
1985 ORT_UNTIED_TASKLOOP enumerators.
1986 (enum gimplify_defaultmap_kind): New.
1987 (struct gimplify_omp_ctx): Remove target_map_scalars_firstprivate and
1988 target_map_pointers_as_0len_arrays members, add defaultmap.
1989 (new_omp_context): Initialize defaultmap member.
1990 (gimple_add_tmp_var): Handle ORT_TASKGROUP like ORT_WORKSHARE.
1991 (maybe_fold_stmt): Don't fold even in host teams regions.
1992 (omp_firstprivatize_variable): Handle ORT_TASKGROUP like
1993 ORT_WORKSHARE. Test ctx->defaultmap[GDMK_SCALAR] instead of
1994 ctx->omp_firstprivatize_variable.
1995 (omp_add_variable): Don't add private/firstprivate for VLAs in
1997 (omp_default_clause): Print "taskloop" rather than "task" if
1999 (omp_notice_variable): Handle ORT_TASKGROUP like ORT_WORKSHARE.
2000 Handle new defaultmap clause kinds.
2001 (omp_is_private): Handle ORT_TASKGROUP like ORT_WORKSHARE. Allow simd
2002 iterator to be lastprivate or private. Fix up diagnostics if linear
2003 is used on collapse>1 simd iterator.
2004 (omp_check_private): Handle ORT_TASKGROUP like ORT_WORKSHARE.
2005 (gimplify_omp_depend): New function.
2006 (gimplify_scan_omp_clauses): Add shared clause on parallel for
2007 combined parallel master taskloop{, simd} if taskloop has
2008 firstprivate, lastprivate or reduction clause. Handle
2009 OMP_CLAUSE_REDUCTION_TASK diagnostics. Adjust tests for
2010 ORT_COMBINED_TEAMS. Gimplify depend clauses with iterators. Handle
2011 cancel and simd OMP_CLAUSE_IF_MODIFIERs. Handle
2012 OMP_CLAUSE_NONTEMPORAL. Handle new defaultmap clause kinds. Handle
2013 OMP_CLAUSE_{TASK,IN}_REDUCTION. Diagnose invalid conditional
2015 (gimplify_adjust_omp_clauses_1): Ignore GOVD_NONTEMPORAL. Handle
2016 GOVD_MAP_ALLOC_ONLY and GOVD_MAP_FROM_ONLY.
2017 (gimplify_adjust_omp_clauses): Handle OMP_CLAUSE_NONTEMPORAL. Handle
2018 OMP_CLAUSE_{TASK,IN}_REDUCTION.
2019 (gimplify_omp_task): Handle taskwait with depend clauses.
2020 (gimplify_omp_for): Add shared clause on parallel for combined
2021 parallel master taskloop{, simd} if taskloop has firstprivate,
2022 lastprivate or reduction clause. Use ORT_TASKLOOP or
2023 ORT_UNTIED_TASKLOOP instead of ORT_TASK or ORT_UNTIED_TASK. Adjust
2024 tests for ORT_COMBINED_TEAMS. Handle C++ range for loops with
2025 NULL TREE_PURPOSE in OMP_FOR_ORIG_DECLS. Firstprivatize
2026 __for_end and __for_range temporaries on OMP_PARALLEL for
2027 distribute parallel for{, simd}. Move OMP_CLAUSE_REDUCTION
2028 and OMP_CLAUSE_IN_REDUCTION from taskloop to the task construct
2029 sandwiched in between two taskloops.
2030 (computable_teams_clause): Test ctx->defaultmap[GDMK_SCALAR]
2031 instead of ctx->omp_firstprivatize_variable.
2032 (gimplify_omp_workshare): Set ort to ORT_HOST_TEAMS or
2033 ORT_COMBINED_HOST_TEAMS if not inside of target construct. If
2034 host teams, use gimplify_and_return_first etc. for body like
2035 for target or target data constructs, and at the end call
2036 gimple_omp_teams_set_host on the GIMPLE_OMP_TEAMS object.
2037 (gimplify_omp_atomic): Use OMP_ATOMIC_MEMORY_ORDER instead
2038 of OMP_ATOMIC_SEQ_CST, pass it as new argument to
2039 gimple_build_omp_atomic_load and gimple_build_omp_atomic_store, remove
2040 gimple_omp_atomic_set_seq_cst calls.
2041 (gimplify_expr) <case OMP_TASKGROUP>: Move handling into a separate
2042 case, handle taskgroup clauses.
2043 * lto-streamer-out.c (hash_tree): Handle
2044 OMP_CLAUSE_{TASK,IN}_REDUCTION.
2045 * Makefile.in (GTFILES): Add omp-general.h.
2046 * omp-builtins.def (BUILT_IN_GOMP_TASKWAIT_DEPEND,
2047 BUILT_IN_GOMP_LOOP_NONMONOTONIC_RUNTIME_START,
2048 BUILT_IN_GOMP_LOOP_MAYBE_NONMONOTONIC_RUNTIME_START,
2049 BUILT_IN_GOMP_LOOP_START, BUILT_IN_GOMP_LOOP_ORDERED_START,
2050 BUILT_IN_GOMP_LOOP_DOACROSS_START,
2051 BUILT_IN_GOMP_LOOP_NONMONOTONIC_RUNTIME_NEXT,
2052 BUILT_IN_GOMP_LOOP_MAYBE_NONMONOTONIC_RUNTIME_NEXT,
2053 BUILT_IN_GOMP_LOOP_ULL_NONMONOTONIC_RUNTIME_START,
2054 BUILT_IN_GOMP_LOOP_ULL_MAYBE_NONMONOTONIC_RUNTIME_START,
2055 BUILT_IN_GOMP_LOOP_ULL_START, BUILT_IN_GOMP_LOOP_ULL_ORDERED_START,
2056 BUILT_IN_GOMP_LOOP_ULL_DOACROSS_START,
2057 BUILT_IN_GOMP_LOOP_ULL_NONMONOTONIC_RUNTIME_NEXT,
2058 BUILT_IN_GOMP_LOOP_ULL_MAYBE_NONMONOTONIC_RUNTIME_NEXT,
2059 BUILT_IN_GOMP_PARALLEL_LOOP_NONMONOTONIC_RUNTIME,
2060 BUILT_IN_GOMP_PARALLEL_LOOP_MAYBE_NONMONOTONIC_RUNTIME,
2061 BUILT_IN_GOMP_PARALLEL_REDUCTIONS, BUILT_IN_GOMP_SECTIONS2_START,
2062 BUILT_IN_GOMP_TEAMS_REG, BUILT_IN_GOMP_TASKGROUP_REDUCTION_REGISTER,
2063 BUILT_IN_GOMP_TASKGROUP_REDUCTION_UNREGISTER,
2064 BUILT_IN_GOMP_TASK_REDUCTION_REMAP,
2065 BUILT_IN_GOMP_WORKSHARE_TASK_REDUCTION_UNREGISTER): New builtins.
2066 * omp-expand.c (workshare_safe_to_combine_p): Return false for
2067 non-worksharing loops.
2068 (omp_adjust_chunk_size): Don't adjust anything if chunk_size is zero.
2069 (determine_parallel_type): Don't combine parallel with worksharing
2070 which has _reductemp_ clause.
2071 (expand_parallel_call): Emit the GOMP_*nonmonotonic_runtime* or
2072 GOMP_*maybe_nonmonotonic_runtime* builtins instead of GOMP_*runtime*
2073 if there is nonmonotonic modifier or if there is no modifier and no
2074 ordered clause. For dynamic and guided schedule without monotonic
2075 and nonmonotonic modifier, default to nonmonotonic.
2076 (expand_omp_for): Likewise. Adjust expand_omp_for_generic caller, use
2077 GOMP_loop{,_ull}{,_ordered,_doacross}_start builtins if there are
2079 (expand_task_call): Add GOMP_TASK_FLAG_REDUCTION flag to flags if
2080 there are any reduction clauses.
2081 (expand_taskwait_call): New function.
2082 (expand_teams_call): New function.
2083 (expand_omp_taskreg): Allow GIMPLE_OMP_TEAMS and call
2084 expand_teams_call for it. Formatting fix. Handle taskwait with
2086 (expand_omp_for_generic): Add SCHED_ARG argument. Handle expansion
2087 of worksharing loops with task reductions.
2088 (expand_omp_for_static_nochunk, expand_omp_for_static_chunk): Handle
2089 expansion of worksharing loops with task reductions.
2090 (expand_omp_sections): Handle expansion of sections with task
2092 (expand_omp_synch): For host teams call expand_omp_taskreg.
2093 (omp_memory_order_to_memmodel): New function.
2094 (expand_omp_atomic_load, expand_omp_atomic_store,
2095 expand_omp_atomic_fetch_op): Use it and gimple_omp_atomic_memory_order
2096 instead of gimple_omp_atomic_seq_cst_p.
2097 (build_omp_regions_1, omp_make_gimple_edges): Treat taskwait with
2098 depend clauses as a standalone directive.
2099 * omp-general.c (enum omp_requires): New variable.
2100 (omp_extract_for_data): Initialize have_reductemp member. Allow
2101 NE_EXPR even in OpenMP loops, transform them into LT_EXPR or
2102 GT_EXPR loops depending on incr sign. Formatting fixes.
2103 * omp-general.h (struct omp_for_data): Add have_reductemp member.
2104 (enum omp_requires): New enum.
2105 (omp_requires_mask): Declare.
2106 * omp-grid.c (grid_eliminate_combined_simd_part): Formatting fix.
2108 * omp-low.c (struct omp_context): Add task_reductions and
2109 task_reduction_map fields.
2110 (is_host_teams_ctx): New function.
2111 (is_taskreg_ctx): Return true also if is_host_teams_ctx.
2112 (use_pointer_for_field): Use is_global_var instead of
2113 TREE_STATIC || DECL_EXTERNAL, and apply only if not privatized
2115 (build_outer_var_ref): Ignore taskgroup outer contexts.
2116 (delete_omp_context): Release task_reductions and task_reduction_map.
2117 (scan_sharing_clauses): Don't add any fields for reduction clause on
2118 taskloop. Handle OMP_CLAUSE__REDUCTEMP_. Handle
2119 OMP_CLAUSE_{IN,TASK}_REDUCTION and OMP_CLAUSE_REDUCTION with task
2120 modifier. Don't ignore shared clauses in is_host_teams_ctx contexts.
2121 Handle OMP_CLAUSE_NONTEMPORAL.
2122 (add_taskreg_looptemp_clauses): Add OMP_CLAUSE__REDUCTEMP_ clause if
2124 (scan_omp_parallel): Add _reductemp_ clause if there are any reduction
2125 clauses with task modifier.
2126 (scan_omp_task): Handle taskwait with depend clauses.
2127 (finish_taskreg_scan): Move field corresponding to _reductemp_ clause
2128 first. Move also OMP_CLAUSE__REDUCTEMP_ clause in front if present.
2129 Handle GIMPLE_OMP_TEAMS like GIMPLE_OMP_PARALLEL.
2130 (scan_omp_for): Fix comment formatting.
2131 (scan_omp_teams): Handle host teams constructs.
2132 (check_omp_nesting_restrictions): Allow teams with no outer
2133 OpenMP context. Adjust diagnostics for teams strictly nested into
2134 some explicit OpenMP construct other than target. Allow OpenMP atomics
2135 inside of simd regions.
2136 (scan_omp_1_stmt): Call scan_sharing_clauses for taskgroups.
2137 (scan_omp_1_stmt) <case GIMPLE_OMP_TEAMS>: Temporarily bump
2138 taskreg_nesting_level while scanning host teams construct.
2139 (task_reduction_read): New function.
2140 (lower_rec_input_clauses): Handle OMP_CLAUSE_REDUCTION on taskloop
2141 construct. Handle OMP_CLAUSE_IN_REDUCTION and OMP_CLAUSE__REDUCTEMP_
2142 clauses. Handle OMP_CLAUSE_REDUCTION with task modifier. Remove
2143 second argument create_tmp_var if it is NULL. Don't ignore shared
2144 clauses in is_host_teams_ctx contexts. Handle
2145 OMP_CLAUSE_FIRSTPRIVATE_NO_REFERENCE on OMP_CLAUSE_FIRSTPRIVATE
2147 (lower_reduction_clauses): Ignore reduction clauses with task
2148 modifier. Remove second argument create_tmp_var if it is NULL.
2149 Initialize OMP_ATOMIC_MEMORY_ORDER to relaxed.
2150 (lower_send_clauses): Ignore reduction clauses with task modifier.
2151 Handle OMP_CLAUSE__REDUCTEMP_. Don't send anything for
2152 OMP_CLAUSE_REDUCTION on taskloop. Handle OMP_CLAUSE_IN_REDUCTION.
2153 (maybe_add_implicit_barrier_cancel): Add OMP_RETURN argument, don't
2154 rely that it is the last stmt in body so far. Ignore outer taskgroup
2156 (omp_task_reductions_find_first, omp_task_reduction_iterate,
2157 lower_omp_task_reductions): New functions.
2158 (lower_omp_sections): Handle reduction clauses with taskgroup
2159 modifiers. Adjust maybe_add_implicit_barrier_cancel caller.
2160 (lower_omp_single): Adjust maybe_add_implicit_barrier_cancel caller.
2161 (lower_omp_for): Likewise. Handle reduction clauses with taskgroup
2163 (lower_omp_taskgroup): Handle taskgroup reductions.
2164 (create_task_copyfn): Copy over OMP_CLAUSE__REDUCTEMP_ pointer.
2165 Handle OMP_CLAUSE_IN_REDUCTION and OMP_CLAUSE_REDUCTION clauses.
2166 (lower_depend_clauses): If there are any
2167 OMP_CLAUSE_DEPEND_DEPOBJ or OMP_CLAUSE_DEPEND_MUTEXINOUTSET
2168 depend clauses, use a new array format. If OMP_CLAUSE_DEPEND_LAST is
2169 seen, assume lowering is done already and return early. Set kind
2170 on artificial depend clause to OMP_CLAUSE_DEPEND_LAST.
2171 (lower_omp_taskreg): Handle reduction clauses with task modifier on
2172 parallel construct. Handle reduction clause on taskloop construct.
2173 Handle taskwait with depend clauses.
2174 (lower_omp_1): Use lower_omp_taskreg instead of lower_omp_teams
2175 for host teams constructs.
2176 * tree.c (omp_clause_num_ops): Add in_reduction, task_reduction,
2177 nontemporal and _reductemp_ clause entries.
2178 (omp_clause_code_name): Likewise.
2179 (walk_tree_1): Handle OMP_CLAUSE_{IN,TASK}_REDUCTION,
2180 OMP_CLAUSE_NONTEMPORAL and OMP_CLAUSE__REDUCTEMP_.
2181 * tree-core.h (enum omp_clause_code): Add
2182 OMP_CLAUSE_{{IN,TASK}_REDUCTION,NONTEMPORAL,_REDUCTEMP_}.
2183 (enum omp_clause_defaultmap_kind, enum omp_memory_order): New.
2184 (struct tree_base): Add omp_atomic_memory_order field into union.
2185 Remove OMP_ATOMIC_SEQ_CST comment.
2186 (enum omp_clause_depend_kind): Add OMP_CLAUSE_DEPEND_MUTEXINOUTSET
2187 and OMP_CLAUSE_DEPEND_DEPOBJ.
2188 (struct tree_omp_clause): Add subcode.defaultmap_kind.
2189 * tree.def (OMP_TASKGROUP): Add another operand, move next to other
2190 OpenMP constructs with body and clauses operands.
2191 * tree.h (OMP_BODY): Use OMP_MASTER instead of OMP_TASKGROUP.
2192 (OMP_CLAUSES): Use OMP_TASKGROUP instead of OMP_SINGLE.
2193 (OMP_TASKGROUP_CLAUSES): Define.
2194 (OMP_CLAUSE_DECL): Use OMP_CLAUSE__REDUCTEMP_ instead of
2195 OMP_CLAUSE__LOOPTEMP_.
2196 (OMP_ATOMIC_SEQ_CST): Remove.
2197 (OMP_ATOMIC_MEMORY_ORDER, OMP_CLAUSE_FIRSTPRIVATE_NO_REFERENCE,
2198 OMP_CLAUSE_LASTPRIVATE_CONDITIONAL): Define.
2199 (OMP_CLAUSE_REDUCTION_CODE, OMP_CLAUSE_REDUCTION_INIT,
2200 OMP_CLAUSE_REDUCTION_MERGE, OMP_CLAUSE_REDUCTION_PLACEHOLDER,
2201 OMP_CLAUSE_REDUCTION_DECL_PLACEHOLDER,
2202 OMP_CLAUSE_REDUCTION_OMP_ORIG_REF): Handle
2203 OMP_CLAUSE_{,IN_,TASK_}REDUCTION.
2204 (OMP_CLAUSE_REDUCTION_TASK, OMP_CLAUSE_REDUCTION_INSCAN,
2205 OMP_CLAUSE_DEFAULTMAP_KIND, OMP_CLAUSE_DEFAULTMAP_CATEGORY,
2206 OMP_CLAUSE_DEFAULTMAP_BEHAVIOR, OMP_CLAUSE_DEFAULTMAP_SET_KIND):
2208 * tree-inline.c (remap_gimple_stmt): Remap taskgroup clauses.
2209 * tree-nested.c (convert_nonlocal_omp_clauses): Handle
2210 OMP_CLAUSE__REDUCTEMP_, OMP_CLAUSE_NONTEMPORAL.
2211 (convert_local_omp_clauses): Likewise. Remove useless test.
2212 * tree-parloops.c (create_call_for_reduction_1): Pass
2213 OMP_MEMORY_ORDER_RELAXED as new argument to
2214 dump_gimple_omp_atomic_load and dump_gimple_omp_atomic_store.
2215 * tree-pretty-print.c (dump_omp_iterators): New function.
2216 (dump_omp_clause): Handle OMP_CLAUSE__REDUCTEMP_,
2217 OMP_CLAUSE_NONTEMPORAL, OMP_CLAUSE_{TASK,IN}_REDUCTION. Print
2218 reduction modifiers. Handle OMP_CLAUSE_DEPEND_DEPOBJ and
2219 OMP_CLAUSE_DEPEND_MUTEXINOUTSET. Print iterators in depend clauses.
2220 Print __internal__ for OMP_CLAUSE_DEPEND_LAST. Handle cancel and
2221 simd OMP_CLAUSE_IF_MODIFIERs. Handle new kinds of
2222 OMP_CLAUSE_DEFAULTMAP. Print conditional: for
2223 OMP_CLAUSE_LASTPRIVATE_CONDITIONAL.
2224 (dump_omp_atomic_memory_order): New function.
2225 (dump_generic_node): Use it. Print taskgroup clauses. Print
2226 taskwait with depend clauses.
2227 * tree-pretty-print.h (dump_omp_atomic_memory_order): Declare.
2228 * tree-streamer-in.c (unpack_ts_omp_clause_value_fields):
2229 Handle OMP_CLAUSE_{TASK,IN}_REDUCTION.
2230 * tree-streamer-out.c (pack_ts_omp_clause_value_fields,
2231 write_ts_omp_clause_tree_pointers): Likewise.
2233 2018-11-08 David Malcolm <dmalcolm@redhat.com>
2236 * doc/invoke.texi (-fdump-ipa-): Document the "-optimized",
2237 "-missed", "-note", and "-all" sub-options.
2238 * ipa-inline.c (caller_growth_limits): Port from fprintf to dump
2240 (can_early_inline_edge_p): Likewise.
2241 (want_early_inline_function_p): Likewise.
2242 (want_inline_self_recursive_call_p): Likewise.
2243 (recursive_inlining): Likewise.
2244 (inline_small_functions): Likewise.
2245 (flatten_function): Likewise.
2246 (ipa_inline): Likewise.
2247 (inline_always_inline_functions): Likewise.
2248 (early_inline_small_functions): Likewise.
2249 (early_inliner): Likewise.
2250 * tree-inline.c (expand_call_inline): Likewise.
2252 2018-11-08 David Malcolm <dmalcolm@redhat.com>
2254 * pretty-print.c (pp_format): Handle %f.
2255 (selftest::test_pp_format): Add test of %f.
2256 * pretty-print.h (pp_double): New macro.
2258 2018-11-08 David Malcolm <dmalcolm@redhat.com>
2260 * dump-context.h (ASSERT_IS_CGRAPH_NODE): New macro.
2261 * dumpfile.c (make_item_for_dump_cgraph_node): Move to before...
2262 (dump_pretty_printer::decode_format): Implement "%C" for
2264 (selftest::test_capture_of_dump_calls): Rename "where" to
2265 "stmt_loc". Convert test_decl to a function decl and set its
2266 location. Add a symbol_table_test RAII instance and a
2267 cgraph_node, using it to test "%C" and dump_symtab_node.
2269 2018-11-08 Eric Botcazou <ebotcazou@adacore.com>
2272 * cgraphclones.c (duplicate_thunk_for_node): Also set DECL_IGNORED_P.
2274 2018-11-08 David Malcolm <dmalcolm@redhat.com>
2276 * cgraph.c: Include "selftest.h".
2277 (saved_symtab): New variable.
2278 (selftest::symbol_table_test::symbol_table_test): New ctor.
2279 (selftest::symbol_table_test::~symbol_table_test): New dtor.
2280 (selftest::test_symbol_table_test): New test.
2281 (selftest::cgraph_c_tests): New.
2282 * cgraph.h (saved_symtab): New decl.
2283 (selftest::symbol_table_test): New class.
2284 * selftest-run-tests.c (selftest::run_tests): Call
2285 selftest::cgraph_c_tests.
2286 * selftest.h (selftest::cgraph_c_tests): New decl.
2288 2018-11-08 Richard Biener <rguenther@suse.de>
2290 * tree-data-ref.h (lambda_int): New typedef.
2291 (lambda_vector_gcd): Adjust.
2292 (lambda_vector_new): Likewise.
2293 (lambda_matrix_new): Likewise.
2294 * tree-data-ref.c (print_lambda_vector): Adjust.
2296 2018-11-08 Richard Biener <rguenther@suse.de>
2298 PR tree-optimization/87929
2299 * tree-complex.c (expand_complex_comparison): Clean EH.
2301 2018-11-08 Martin Liska <mliska@suse.cz>
2303 * doc/extend.texi: Reword.
2304 * predict.c (expr_expected_value_1): Likewise.
2306 2018-11-08 Richard Biener <rguenther@suse.de>
2308 PR tree-optimization/87913
2309 * tree-ssa-phiopt.c (minmax_replacement): Turn EQ/NE compares
2310 of extreme values to ordered comparisons.
2312 2018-11-07 Sandra Loosemore <sandra@codesourcery.com>
2315 * doc/invoke.texi (Code Gen Options): Clarify -fno-common behavior.
2317 2018-11-07 Sandra Loosemore <sandra@codesourcery.com>
2319 * doc/invoke.texi: Remove leading dash from @opindex entries
2320 throughout the file.
2322 2018-11-07 Sandra Loosemore <sandra@codesourcery.com>
2325 * doc/invoke.texi (Option Summary): Add -e and --entry.
2326 (Link Options): Likewise.
2328 2018-11-07 Nathan Sidwell <nathan@acm.org>
2331 * Makefile.in (bitmap.o-warn): Add -Wno-error to unbreak
2332 --disable-checking bootstrap.
2334 2018-11-07 Hafiz Abid Qadeer <abidh@codesourcery.com>
2336 * configure: Regenerated.
2338 2018-11-07 Jozef Lawrynowicz <jozef.l@mittosystems.com>
2341 * stor-layout.c (compute_record_mode): Set TYPE_MODE of UNION_TYPE
2342 to the mode of the widest field iff the widest field has mode class
2343 MODE_INT, or MODE_PARTIAL_INT and the union would be passed by
2346 2018-11-07 Nikolai Merinov <n.merinov@inango-systems.com>
2348 * common.opt: Add -Wattribute-warning.
2349 * doc/invoke.texi: Add documentation for -Wno-attribute-warning.
2350 * expr.c (expand_expr_real_1): Add new attribute to warning_at
2351 call to allow user configure behavior of "warning" attribute.
2353 2018-11-07 Segher Boessenkool <segher@kernel.crashing.org>
2355 * target.def: Put @: after every vs., e.g., and i.e. where it is
2356 followed by whitespace.
2357 * doc/extend.texi: Ditto.
2358 * doc/fragments.texi: Ditto.
2359 * doc/gimple.texi: Ditto.
2360 * doc/implement-c.texi: Ditto.
2361 * doc/install.texi: Ditto.
2362 * doc/invoke.texi: Ditto.
2363 * doc/md.texi: Ditto.
2364 * doc/plugins.texi: Ditto.
2365 * doc/rtl.texi: Ditto.
2366 * doc/sourcebuild.texi: Ditto.
2367 * doc/tm.texi.in: Ditto.
2368 * doc/ux.texi: Ditto.
2369 * doc/tm.texi: Regenerate.
2371 2018-11-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2373 * config/arm/arm-cpus.in (ares): New entry.
2374 * config/arm/arm-tables.opt: Regenerate.
2375 * config/arm/arm-tune.md: Likewise.
2376 * doc/invoke.texi (ARM Options): Document ares.
2378 2018-11-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2380 * config/aarch64/aarch64-cores.def (ares): Define.
2381 * config/aarch64/aarch64-tune.md: Regenerate.
2382 * doc/invoke.texi (AArch64 Options): Document ares value for mtune.
2384 2018-11-07 Jan Hubicka <jh@suse.cz>
2386 * tree.c (fld_incomplete_type_of): Clear TREE_ADDRESSABLE.
2387 (free_lang_data_in_decl): Set TREE_ADDRESSABLE for public vars and
2388 functions; clear TYPE_DECL_SUPPRESS_DEBUG and DECL_MODE for
2391 2018-11-07 Richard Biener <rguenther@suse.de>
2393 PR tree-optimization/87914
2394 * tree-vect-loop.c (vect_is_simple_reduction): Improve detection
2396 (vectorizable_reduction): Handle shifts and rotates by dispatching
2397 to vectorizable_shift.
2398 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Handle
2399 in-loop uses of vect_nested_cycle defs. Merge cycle and internal
2401 (vectorizable_shift): Export and handle being called as
2403 (vect_analyze_stmt): Call vectorizable_shift after
2404 vectorizable_reduction.
2405 * tree-vectorizer.h (vectorizable_shift): Declare.
2407 2018-11-07 Jan Hubicka <jh@suse.cz>
2409 * ipa-devirt.c (odr_types_equivalent_p): Expect constants
2410 than const decls in TREE_VALUE of enum.
2411 (dump_type_inheritance_graph): Improve duplicate dumping.
2412 (free_enum_values): New.
2413 (build_type_inheritance_graph): Use it.
2414 * tree.c (free_lang_data_in_type): Free TYPE_VALUES of enums
2415 which are not main variants or not ODR types.
2416 (verify_type_variant): Expect variants to have no TYPE_VALUES.
2418 2018-11-07 Richard Biener <rguenther@suse.de>
2420 * ipa-inline.c (want_inline_small_function_p): Compute
2421 big_speedup_p lazily and last.
2423 2018-11-07 Jan Hubicka <jh@suse.cz>
2425 * tree.c (fld_type_variant_equal_p): Skip TYPE_ALIGN check when
2426 building incomplete variant of complete type.
2427 (fld_type_variant): Do not copy TYPE_ALIGN when building incomplete
2428 variant of complete type.
2430 2018-11-07 Chenghua Xu <paul.hua.gm@gmail.com>
2432 * config/mips/mips.c: Fix typo in documentation of
2433 mips_loongson_ext2_prefetch_cookie.
2434 (mips_option_override): fix brain twister logical.
2435 * config/mips/mips.h: Fix typo in documentation of
2436 ISA_HAS_CTZ_CTO and define pattern.
2437 * config/mips/mips.md (prefetch): Hoist EXT2 above
2439 (prefetch_indexed): Hoist EXT2 above the EXT block.
2441 2018-11-07 Jan Hubicka <jh@suse.cz>
2443 * tree.c (free_lang_data_in_type): Add fld parameter; simplify
2444 return and parameter types of function and method types.
2445 (free_lang_data_in_cgraph): Update.
2447 2018-11-07 Martin Liska <mliska@suse.cz>
2449 PR rtl-optimization/87868
2450 * postreload-gcse.c (eliminate_partially_redundant_load): Set
2451 threshold to max_count if we would overflow.
2452 * profile-count.h: Make max_count a public constant.
2454 2018-11-07 Martin Liska <mliska@suse.cz>
2456 * mem-stats.h: Fix GNU coding style.
2458 2018-11-07 Chenghua Xu <paul.hua.gm@gmail.com>
2460 * config/mips/gs264e.md: New.
2461 * config/mips/mips-cpus.def: Define gs264e.
2462 * config/mips/mips-tables.opt: Regenerate.
2463 * config/mips/mips.c (mips_rtx_cost_data): Add DEFAULT_COSTS for
2465 (mips_issue_rate): Add support for gs264e.
2466 (mips_multipass_dfa_lookahead): Likewise.
2467 * config/mips/mips.h: Define TARGET_GS264E and TUNE_GS264E.
2468 (MIPS_ISA_LEVEL_SPEC): Infer mips64r2 from gs264e.
2469 (MIPS_ASE_MSA_SPEC): New.
2470 (BASE_DRIVER_SELF_SPECS): march=gs264e implies -mmsa.
2471 (ISA_HAS_FUSED_MADD4): Enable for TARGET_GS264E.
2472 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_GS264E.
2473 * config/mips/mips.md: Include gs264e.md.
2474 (processor): Add gs264e.
2475 * config/mips/mips.opt (MSA): Use Mask instead of Var.
2476 * doc/invoke.texi: Add gs264e to supported architectures.
2478 2018-11-07 Chenghua Xu <paul.hua.gm@gmail.com>
2480 * config/mips/gs464e.md: New.
2481 * config/mips/mips-cpus.def: Define gs464e.
2482 * config/mips/mips-tables.opt: Regenerate.
2483 * config/mips/mips.c (mips_rtx_cost_data): Add DEFAULT_COSTS for
2485 (mips_issue_rate): Add support for gs464e.
2486 (mips_multipass_dfa_lookahead): Likewise.
2487 (mips_option_override): Enable MMI, EXT and EXT2 for gs464e.
2488 * config/mips/mips.h: Define TARGET_GS464E and TUNE_GS464E.
2489 (MIPS_ISA_LEVEL_SPEC): Infer mips64r2 from gs464e.
2490 (ISA_HAS_FUSED_MADD4): Enable for TARGET_GS464E.
2491 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_GS464E.
2492 * config/mips/mips.md: Include gs464e.md.
2493 (processor): Add gs464e.
2494 * doc/invoke.texi: Add gs464e to supported architectures.
2496 2018-11-07 Chenghua Xu <paul.hua.gm@gmail.com>
2498 * config/mips/loongson3a.md: Rename to ...
2499 * config/mips/gs464.md: ... here.
2500 * config/mips/mips-cpus.def: Define gs464; Add loongson3a
2501 as an alias of gs464 processor.
2502 * config/mips/mips-tables.opt: Regenerate.
2503 * config/mips/mips.c (mips_issue_rate): Use PROCESSOR_GS464
2504 instead of PROCESSOR_LOONGSON_3A.
2505 (mips_multipass_dfa_lookahead): Use TUNE_GS464 instead of
2507 (mips_option_override): Enable MMI and EXT for gs464.
2508 * config/mips/mips.h: Rename TARGET_LOONGSON_3A to TARGET_GS464;
2509 Rename TUNE_LOONGSON_3A to TUNE_GS464.
2510 (MIPS_ISA_LEVEL_SPEC): Infer mips64r2 from gs464.
2511 (ISA_HAS_ODD_SPREG, ISA_AVOID_DIV_HILO, ISA_HAS_FUSED_MADD4,
2512 ISA_HAS_UNFUSED_MADD4): Use TARGET_GS464 instead of
2514 * config/mips/mips.md: Include gs464.md instead of loongson3a.md.
2515 (processor): Add gs464;
2516 * doc/invoke.texi: Add gs464 to supported architectures.
2518 2018-11-07 Chenghua Xu <paul.hua.gm@gmail.com>
2520 * config/mips/mips-protos.h
2521 (mips_loongson_ext2_prefetch_cookie): New prototype.
2522 * config/mips/mips.c (mips_loongson_ext2_prefetch_cookie): New.
2523 (mips_option_override): Enable TARGET_LOONGSON_EXT when
2524 TARGET_LOONGSON_EXT2 is true.
2525 * config/mips/mips.h (TARGET_CPU_CPP_BUILTINS): Define
2526 __mips_loongson_ext2, __mips_loongson_ext_rev=2.
2527 (ISA_HAS_CTZ_CTO): New, true if TARGET_LOONGSON_EXT2.
2528 (ISA_HAS_PREFETCH): Include TARGET_LOONGSON_EXT and
2529 TARGET_LOONGSON_EXT2.
2530 (ASM_SPEC): Add mloongson-ext2 and mno-loongson-ext2.
2531 (define_insn "ctz<mode>2"): New insn pattern.
2532 (define_insn "prefetch"): Include TARGET_LOONGSON_EXT2.
2533 (define_insn "prefetch_indexed_<mode>"): Include
2534 TARGET_LOONGSON_EXT and TARGET_LOONGSON_EXT2.
2535 * config/mips/mips.opt (-mloongson-ext2): Add option.
2536 * gcc/doc/invoke.texi (-mloongson-ext2): Document.
2538 2018-11-07 Chenghua Xu <paul.hua.gm@gmail.com>
2540 * config/mips/mips.h (TARGET_CPU_CPP_BUILTINS): Add
2541 __mips_loongson_ext.
2542 (MIPS_ASE_LOONGSON_EXT_SPEC): New.
2543 (BASE_DRIVER_SELF_SPECS): march=loongson3a implies
2545 (ASM_SPEC): Add mloongson-ext and mno-loongson-ext.
2546 * config/mips/mips.md (mul<mode>3, mul<mode>3_mul3_nohilo,
2547 <u>div<mode>3, <u>mod<mode>3, prefetch): Use TARGET_LOONGSON_EXT
2548 instead of TARGET_LOONGSON_3A.
2549 * config/mips/mips.opt (-mloongson-ext): Add option.
2550 * gcc/doc/invoke.texi (-mloongson-ext): Document.
2552 2018-11-07 Chenghua Xu <paul.hua.gm@gmail.com>
2554 * config.gcc (extra_headers): Add loongson-mmiintrin.h.
2555 * config/mips/loongson.md: Move to ...
2556 * config/mips/loongson-mmi.md: here; Adjustment.
2557 * config/mips/loongson.h: Move to ...
2558 State as deprecated. Include loongson-mmiintrin.h for back
2559 compatibility and warning.
2560 * config/mips/loongson-mmiintrin.h: ... here.
2561 * config/mips/mips.c (mips_hard_regno_mode_ok_uncached,
2562 mips_vector_mode_supported_p, AVAIL_NON_MIPS16): Use
2563 TARGET_LOONGSON_MMI instead of TARGET_LOONGSON_VECTORS.
2564 (mips_option_override): Make sure MMI use hard float;
2565 (mips_shift_truncation_mask, mips_expand_vpc_loongson_even_odd,
2566 mips_expand_vpc_loongson_pshufh, mips_expand_vpc_loongson_bcast,
2567 mips_expand_vector_init): Use TARGET_LOONGSON_MMI instead of
2568 TARGET_LOONGSON_VECTORS.
2569 * gcc/config/mips/mips.h (TARGET_LOONGSON_VECTORS): Delete.
2570 (TARGET_CPU_CPP_BUILTINS): Add __mips_loongson_mmi.
2571 (MIPS_ASE_DSP_SPEC, MIPS_ASE_LOONGSON_MMI_SPEC): New.
2572 (BASE_DRIVER_SELF_SPECS): march=loongson2e/2f/3a implies
2574 (SHIFT_COUNT_TRUNCATED): Use TARGET_LOONGSON_MMI instead of
2575 TARGET_LOONGSON_VECTORS.
2576 * gcc/config/mips/mips.md (MOVE64, MOVE128): Use
2577 TARGET_LOONGSON_MMI instead of TARGET_LOONGSON_VECTORS.
2578 (Loongson MMI patterns): Include loongson-mmi.md instead of
2580 * gcc/config/mips/mips.opt (-mloongson-mmi): New option.
2581 * gcc/doc/invoke.texi (-mloongson-mmi): Document.
2583 2018-11-07 Richard Biener <rguenther@suse.de>
2586 * tree-streamer-in.c (lto_input_ts_block_tree_pointers): Fixup
2587 BLOCK_ABSTRACT_ORIGIN to be the ultimate origin.
2589 2018-11-07 Alexandre Oliva <aoliva@redhat.com>
2591 PR rtl-optimization/87874
2592 * lra.c (lra_substitute_pseudo): Do not create a subreg for
2595 2018-11-06 Aaron Sawdey <acsawdey@linux.ibm.com>
2597 * config/rs6000/rs6000.md (bswap<mode>2): Force address into register
2598 if not in indexed or indirect form.
2599 (bswap<mode>2_load): Change predicate to indexed_or_indirect_operand.
2600 (bswap<mode>2_store): Ditto.
2602 2018-11-06 Richard Earnshaw <rearnsha@arm.com>
2604 * config/aarch64/aarch64.md (speculation_tracker): Set the mode for
2607 2018-11-06 Richard Biener <rguenther@suse.de>
2609 PR tree-optimization/86850
2610 * vec.h (vec<T, va_heap, vl_ptr>::splice): Check src.length ()
2611 instead of src.m_vec.
2613 2018-11-06 Jan Hubicka <jh@suse.cz>
2615 * tree.c (fld_simplified_type_name): Break out form ...
2616 (free_lang_data_in_type): ... here.
2617 (fld_type_variant_equal_p): Use it.
2619 2018-11-06 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2621 * config/default-d.c: Include memmodel.h.
2623 * config/sol2-d.c: New file.
2624 * config/t-sol2 (sol2-d.o): New rule.
2625 * config.gcc <*-*-solaris2*>: Set d_target_objs,
2626 target_has_targetdm.
2628 2018-11-06 Jan Hubicka <jh@suse.cz>
2630 * tree.c (fld_type_variant): Also copy alignment; be sure that
2631 new variant is equal.
2633 2018-11-06 Ilya Leoshkevich <iii@linux.ibm.com>
2636 * config/s390/s390.md: Add relative_long attribute.
2638 2018-11-06 Jan Hubicka <jh@suse.cz>
2640 * ipa-pure-const.c (check_decl): Do not test TYPE_NEEDS_CONSTRUCTING.
2641 * lto-streamer-out.c (hash_tree): Do not hash TYPE_NEEDS_CONSTRUCTING.
2642 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Do not
2643 stream TYPE_NEEDS_CONSTRUCTING.
2644 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
2645 * tree.c (free_lang_data_in_type): Clear TYPE_NEEDS_CONSTRUCTING.
2647 2018-11-06 Richard Biener <rguenther@suse.de>
2649 * tree-vect-slp.c (vect_slp_bb): Move opening of vect_slp_analyze_bb
2651 (vect_slp_analyze_bb_1): ... here to avoid hiding optimized locations.
2653 2018-11-06 Jozef Lawrynowicz <jozef.l@mittosystems.com>
2655 * gcc/config/msp430/msp430.h (REG_CLASS_CONTENTS): Add R0 to
2656 REG_CLASS_CONTENTS[GEN_REGS].
2657 (REGNO_REG_CLASS): Return NO_REGS for R2 and R3.
2659 2018-11-06 Jan Hubicka <jh@suse.cz>
2661 * tree.c (fld_simplified_type_of): Clear TYPELESS_STORAGE flag.
2663 2018-11-06 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2666 * config/sol2.h (ASAN_CC1_SPEC): Define.
2667 (LD_WHOLE_ARCHIVE_OPTION): Define.
2668 (LD_NO_WHOLE_ARCHIVE_OPTION): Define.
2669 (ASAN_REJECT_SPEC): Provide default.
2670 (LIBASAN_EARLY_SPEC): Define.
2671 (LIBTSAN_EARLY_SPEC): Define.
2672 (LIBLSAN_EARLY_SPEC): Define.
2673 * config/i386/sol2.h (CC1_SPEC): Redefine.
2674 (ASAN_REJECT_SPEC): Define.
2676 * config/sparc/sparc.c (sparc_asan_shadow_offset): Declare.
2677 (TARGET_ASAN_SHADOW_OFFSET): Define.
2678 (sparc_asan_shadow_offset): New function.
2679 * config/sparc/sol2.h (CC1_SPEC): Append ASAN_CC1_SPEC.
2680 (ASAN_REJECT_SPEC): Define.
2682 2018-11-06 Jan Hubicka <jh@suse.cz>
2684 * tree.c (fld_type_variant): Copy canonical type.
2685 (fld_incomplete_type_of): Check that canonical types looks sane;
2686 copy canonical type.
2687 (verify_type): Accept when incomplete type has complete canonical type.
2689 2018-11-06 Jan Hubicka <jh@suse.cz>
2691 * tree.c (free_lang_data): Reset overwite_assembler_name,
2692 print_xnode, print_decl, print_type and print_identifier of
2695 2018-11-06 Richard Biener <rguenther@suse.de>
2697 PR tree-optimization/87889
2698 * tree-vect-loop-manip.c (slpeel_duplicate_current_defs_from_edges):
2699 Do nothing if old and new arg are the same
2701 2018-11-06 Andreas Krebbel <krebbel@linux.ibm.com>
2704 * config/s390/s390.md ("*r<noxa>sbg_di_rotl"): Remove mode
2705 attributes for operands 3 and 4.
2707 2018-11-06 Richard Biener <rguenther@suse.de>
2710 * simplify-rtx.c (simplify_binary_operation_1): Add pattern
2711 matching bitfield insertion.
2713 2018-11-06 Alexandre Oliva <aoliva@redhat.com>
2715 * auto-inc-dec.c: Include valtrack.h. Improve comments.
2716 (reg_next_debug_use): New.
2717 (attempt_change): Propagate adjusted expression into affected
2719 (merge_in_block): Track uses in debug insns.
2720 (pass_inc_dec::execute): Allocate and release
2723 2018-11-06 Wei Xiao <wei3.xiao@intel.com>
2725 * config/i386/avx512fintrin.h: Update VFIXUPIMM* intrinsics.
2726 (_mm512_fixupimm_round_pd): Update parameters and builtin.
2727 (_mm512_maskz_fixupimm_round_pd): Ditto.
2728 (_mm512_fixupimm_round_ps): Ditto.
2729 (_mm512_maskz_fixupimm_round_ps): Ditto.
2730 (_mm_fixupimm_round_sd): Ditto.
2731 (_mm_maskz_fixupimm_round_sd): Ditto.
2732 (_mm_fixupimm_round_ss): Ditto.
2733 (_mm_maskz_fixupimm_round_ss): Ditto.
2734 (_mm512_fixupimm_pd): Ditto.
2735 (_mm512_maskz_fixupimm_pd): Ditto.
2736 (_mm512_fixupimm_ps): Ditto.
2737 (_mm512_maskz_fixupimm_ps): Ditto.
2738 (_mm_fixupimm_sd): Ditto.
2739 (_mm_maskz_fixupimm_sd): Ditto.
2740 (_mm_fixupimm_ss): Ditto.
2741 (_mm_maskz_fixupimm_ss): Ditto.
2742 (_mm512_mask_fixupimm_round_pd): Update builtin.
2743 (_mm512_mask_fixupimm_round_ps): Ditto.
2744 (_mm_mask_fixupimm_round_sd): Ditto.
2745 (_mm_mask_fixupimm_round_ss): Ditto.
2746 (_mm512_mask_fixupimm_pd): Ditto.
2747 (_mm512_mask_fixupimm_ps): Ditto.
2748 (_mm_mask_fixupimm_sd): Ditto.
2749 (_mm_mask_fixupimm_ss): Ditto.
2750 * config/i386/avx512vlintrin.h:
2751 (_mm256_fixupimm_pd): Update parameters and builtin.
2752 (_mm256_maskz_fixupimm_pd): Ditto.
2753 (_mm256_fixupimm_ps): Ditto.
2754 (_mm256_maskz_fixupimm_ps): Ditto.
2755 (_mm_fixupimm_pd): Ditto.
2756 (_mm_maskz_fixupimm_pd): Ditto.
2757 (_mm_fixupimm_ps): Ditto.
2758 (_mm_maskz_fixupimm_ps): Ditto.
2759 (_mm256_mask_fixupimm_pd): Update builtin.
2760 (_mm256_mask_fixupimm_ps): Ditto.
2761 (_mm_mask_fixupimm_pd): Ditto.
2762 (_mm_mask_fixupimm_ps): Ditto.
2763 * config/i386/i386-builtin-types.def: Add new types and remove useless ones.
2764 * config/i386/i386-builtin.def: Update builtin definitions.
2765 * config/i386/i386.c: Handle new builtin types and remove useless ones.
2766 * config/i386/sse.md: Update VFIXUPIMM* patterns.
2767 (<avx512>_fixupimm<mode>_maskz<round_saeonly_expand_name>): Update.
2768 (<avx512>_fixupimm<mode><sd_maskz_name><round_saeonly_name>): Update.
2769 (<avx512>_fixupimm<mode>_mask<round_saeonly_name>): Update.
2770 (avx512f_sfixupimm<mode>_maskz<round_saeonly_expand_name>): Update.
2771 (avx512f_sfixupimm<mode><sd_maskz_name><round_saeonly_name>): Update.
2772 (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Update.
2773 * config/i386/subst.md:
2774 (round_saeonly_sd_mask_operand4): Add new subst_attr.
2775 (round_saeonly_sd_mask_op4): Ditto.
2776 (round_saeonly_expand_operand5): Ditto.
2777 (round_saeonly_expand): Update.
2779 2018-11-05 Max Filippov <jcmvbkbc@gmail.com>
2781 * config/xtensa/uclinux.h (XTENSA_ALWAYS_PIC): Change to 0.
2783 2018-11-05 Segher Boessenkool <segher@kernel.crashing.org>
2785 PR rtl-optimization/87871
2786 * combine.c (make_more_copies): Skip if dest is frame_pointer_rtx.
2788 2018-11-05 Paul Koning <ni1d@arrl.net>
2790 * doc/sourcebuild.texi (target attributes): Document new "inf"
2791 effective target keyword.
2793 2018-11-05 Robin Dapp <rdapp@linux.ibm.com>
2795 * config/s390/s390.c (s390_register_move_cost): Increase costs for
2796 moves involving the CC reg.
2798 2018-11-05 Richard Biener <rguenther@suse.de>
2800 PR tree-optimization/87873
2801 * tree-ssa-loop-manip.h (split_loop_exit_edge): Add copy_constants_p
2803 * tree-ssa-loop-manip.c (split_loop_exit_edge): Likewise.
2804 * tree-vect-loop.c (vect_transform_loop): When splitting the
2805 loop exit also create forwarder PHIs for constants.
2806 * tree-vect-loop-manip.c (slpeel_duplicate_current_defs_from_edges):
2807 Handle constant to_arg, add extra checking we match up the correct
2810 2018-11-05 Robin Dapp <rdapp@linux.ibm.com>
2812 * config/s390/s390.md: QImode and HImode for load on condition.
2814 2018-11-05 Robin Dapp <rdapp@linux.ibm.com>
2816 * config/s390/predicates.md: Fix typo.
2817 * config/s390/s390.md: Allow immediates for load on condition.
2819 2018-11-05 Martin Liska <mliska@suse.cz>
2821 * alloc-pool.h (struct pool_usage): Use SIZE_AMOUNT.
2822 * bitmap.h (struct bitmap_usage): Likewise.
2823 * ggc-common.c (SCALE): Remove.
2825 (struct ggc_usage): Use SIZE_AMOUNT. And update
2827 * ggc-page.c (SCALE): Remove.
2828 (STAT_LABEL): Remove.
2829 (ggc_print_statistics): Use SIZE_AMOUNT.
2830 * gimple.h (SCALE): Remove.
2832 * input.c (ONE_K): Remove.
2835 (STAT_LABEL): Likewise.
2836 (FORMAT_AMOUNT): Likewise.
2837 (dump_line_table_statistics): Use SIZE_AMOUNT.
2838 * mem-stats.h (struct mem_usage): Likewise.
2839 * rtl.c (dump_rtx_statistics): Likewise.
2840 (rtx_alloc_counts): Change type to size_t.
2841 (rtx_alloc_sizes): Likewise.
2842 (rtx_count_cmp): New.
2843 (dump_rtx_statistics): Sort first based on counts.
2844 * tree.c (tree_nodes_cmp): New.
2845 (tree_codes_cmp): New.
2846 (dump_tree_statistics): Sort first based on counts.
2847 * system.h (ONE_K): New.
2849 (SIZE_SCALE): Likewise.
2850 (SIZE_LABEL): Likewise.
2851 (SIZE_AMOUNT): Likewise.
2852 * tree-cfg.c (dump_cfg_stats): Use SIZE_AMOUNT.
2853 * tree-dfa.c (dump_dfa_stats): Likewise.
2854 * tree-phinodes.c (phinodes_print_statistics): Likewise.
2855 * tree-ssanames.c (ssanames_print_statistics): Likewise.
2856 * tree.c (dump_tree_statistics): Likewise.
2857 * vec.c (struct vec_usage): Likewise.
2858 * trans-mem.c (tm_mangle): Enlarge buffer in order to not
2859 trigger a -Werror=format-overflow with
2860 --enable-gather-detailed-stats.
2862 2018-11-05 Martin Liska <mliska@suse.cz>
2864 * mem-stats.h (mem_alloc_description::release_instance_overhead):
2866 * vec.c (struct vec_usage): Register m_element_size.
2867 (vec_prefix::register_overhead): New arguments: elements and
2869 (vec_prefix::release_overhead): Subtract elements.
2870 * vec.h (struct vec_prefix): Change signature.
2871 (va_heap::reserve): Pass proper arguments.
2872 (va_heap::release): Likewise.
2874 2018-11-05 Martin Liska <mliska@suse.cz>
2876 * mem-stats.h (mem_alloc_description::get_list): Fix GNU coding
2880 2018-11-05 Richard Biener <rguenther@suse.de>
2882 * tree-scalar-evolution.h (final_value_replacement_loop): Update
2884 * tree-scalar-evolution.c (final_value_replacement_loop): Return
2885 whether anything was done.
2886 (scev_const_prop): Remove constant propagation part, fold
2888 * tree-ssa-loop.c (pass_scev_cprop::execute): ... here.
2889 (pass_data_scev_cprop): TODO_cleanup_cfg is now done
2892 2018-11-05 Jakub Jelinek <jakub@redhat.com>
2894 PR tree-optimization/87859
2895 * gimple-ssa-store-merging.c (struct merged_store_group): Add
2896 only_constants and first_nonmergeable_order members.
2897 (merged_store_group::merged_store_group): Initialize them.
2898 (merged_store_group::do_merge): Clear only_constants member if
2899 adding something other than INTEGER_CST store.
2900 (imm_store_chain_info::coalesce_immediate_stores): Don't merge
2901 stores with order >= first_nonmergeable_order. Use
2902 merged_store->only_constants instead of always recomputing it.
2903 Set merged_store->first_nonmergeable_order if we've skipped any
2904 stores. Attempt to merge overlapping INTEGER_CST stores that
2905 we would otherwise skip.
2908 * match.pd (X + Y < X): Don't optimize if TYPE_OVERFLOW_SANITIZED.
2910 2018-11-05 Xuepeng Guo <xuepeng.guo@intel.com>
2913 * config/i386/emmintrin.h (__v16qs): New to cope with option
2915 (_mm_cmpeq_epi8): Replace __v16qi with __v16qs.
2916 (_mm_cmplt_epi8): Likewise.
2917 (_mm_cmpgt_epi8): Likewise.
2919 2018-11-05 Richard Biener <rguenther@suse.de>
2921 PR rtl-optimization/87852
2922 * fwprop.c (use_killed_between): Only consider single-defs of the
2923 use whose definition statement dominates the use.
2925 2018-11-05 Martin Liska <mliska@suse.cz>
2928 * doc/invoke.texi: Remove options that are
2929 not disabled with -Os.
2931 2018-11-05 Martin Liska <mliska@suse.cz>
2934 * doc/extend.texi: Update constrain about the last argument
2935 of __builtin_expect_with_probability.
2937 2018-11-05 Martin Liska <mliska@suse.cz>
2940 * predict.c (expr_expected_value_1): Verify
2941 that last argument is a real constants and emit
2944 2018-11-05 Martin Liska <mliska@suse.cz>
2946 PR gcov-profile/77698
2947 * ipa-profile.c (ipa_profile): Adjust hotness threshold
2950 2018-11-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
2952 PR tree-optimization/86572
2953 * builtins.c (c_strlen): Handle negative offsets in a safe way.
2955 2018-11-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
2957 PR tree-optimization/87672
2958 * gimple-fold.c (gimple_fold_builtin_stxcpy_chk): Gimplify.
2959 * tree-ssa-strlen.c (handle_builtin_strcat): Adjust object size.
2961 2018-11-04 Uros Bizjak <ubizjak@gmail.com>
2964 * cfgexpand.c (pass_expand::execute): Move the call to
2965 finish_eh_generation in front of the call to expand_stack_alignment.
2967 2018-11-04 Venkataramanan Kumar <venkataramanan.kumar@amd.com>
2969 * common/config/i386/i386-common.c (processor_alias_table): Add
2971 * config.gcc (i[34567]86-*-linux* | ...): Add znver2.
2972 (case ${target}): Add znver2.
2973 * config/i386/driver-i386.c: (host_detect_local_cpu): Let
2974 -march=native recognize znver2 processors.
2975 * config/i386/i386-c.c (ix86_target_macros_internal): Add znver2.
2976 * config/i386/i386.c (m_znver2): New definition.
2977 (m_ZNVER): New definition.
2978 (m_AMD_MULTIPLE): Includes m_znver2.
2979 (processor_cost_table): Add znver2 entry.
2980 (processor_target_table): Add znver2 entry.
2981 (get_builtin_code_for_version): Set priority for
2983 (processor_model): Add M_AMDFAM17H_ZNVER2.
2984 (arch_names_table): Ditto.
2985 (ix86_reassociation_width): Include znver2.
2986 * config/i386/i386.h (TARGET_znver2): New definition.
2987 (struct ix86_size_cost): Add TARGET_ZNVER2.
2988 (enum processor_type): Add PROCESSOR_ZNVER2.
2989 * config/i386/i386.md (define_attr "cpu"): Add znver2.
2990 * config/i386/x86-tune-costs.h: (processor_costs) Add znver2 costs.
2991 * config/i386/x86-tune-sched.c: (ix86_issue_rate): Add znver2.
2992 (ix86_adjust_cost): Add znver2.
2993 * config/i386/x86-tune.def: Replace m_ZNVER1 by m_ZNVER.
2994 * gcc/doc/extend.texi: Add details about znver2.
2995 * gcc/doc/invoke.texi: Add details about znver2.
2997 2018-11-03 Sandra Loosemore <sandra@codesourcery.com>
3001 * config/nios2/nios2.c (nios2_rtx_costs): Recognize <mul>sidi3
3004 2018-11-02 Aaron Sawdey <acsawdey@linux.ibm.com>
3006 * config/rs6000/rs6000-string.c (expand_strncmp_gpr_sequence): Pay
3007 attention to TARGET_AVOID_XFORM and BYTES_BIG_ENDIAN.
3009 2018-11-02 Richard Earnshaw <rearnsha@arm.com>
3011 * config/aarch64/aarch64.c ((aarch64_override_options): Disable
3012 shrink-wrapping when -mtrack-speculation.
3014 2018-11-02 Richard Biener <rguenther@suse.de>
3016 * tree-ssa-coalesce.c (struct coalesce_list): Add obstack member.
3017 (pop_cost_one_pair): Do not free pair.
3018 (pop_best_coalesce): Likewise.
3019 (create_coalesce_list): Initialize obstack.
3020 (delete_coalesce_list): Free obstack.
3021 (find_coalesce_pair): Obstack-allocate coalesce pairs.
3022 (add_cost_one_coalesce): Likewise.
3023 (struct live_track): Remove bitmap pointer indirections.
3024 (new_live_track): Adjust.
3025 (delete_live_track): Likewise.
3026 (live_track_remove_partition): Likewise.
3027 (live_track_add_partition): Likewise.
3028 (live_track_live_p): Likewise.
3029 (live_track_process_def): Likewise.
3030 (live_track_clear_base_vars): Likewise.
3032 2018-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3034 * configure.ac (gcc_cv_as_sparc_register_op): Remove.
3035 * configure: Regenerate.
3036 * config.in: Regenerate.
3037 * config/sparc/sparc.c (sparc_output_scratch_registers): Remove
3038 HAVE_AS_REGISTER_PSEUDO_OP guard.
3039 * config/sparc/sparc.h (ASM_ARCH64_SPEC): Remove
3040 !HAVE_AS_REGISTER_PSEUDO_OP support.
3041 (ASM_DECLARE_REGISTER_GLOBAL): Remove HAVE_AS_REGISTER_PSEUDO_OP
3044 2018-11-02 Richard Biener <rguenther@suse.de>
3046 * dwarf2out.c (build_abbrev_table): Guard lookup_external_ref call
3049 2018-11-02 Richard Biener <rguenther@suse.de>
3051 PR tree-optimization/87776
3052 * tree-ssa-sccvn.c (do_rpo_vn): Do not mark backedges
3053 executable when iterating but running into rpo-vn-max-loop-depth
3054 and not eliding the iteration.
3056 2018-11-30 Jan Hubicka <jh@suse.cz>
3058 * tree.c (free_lang_data_in_decl): Free TREE_TYPE and alignment of
3061 2018-11-01 Aaron Sawdey <acsawdey@linux.ibm.com>
3063 * config/rs6000/rs6000-protos.h (rs6000_address_for_fpconvert): Remove
3065 * config/rs6000/rs6000.c (rs6000_force_indexed_or_indirect_mem):
3066 Combine with rs6000_address_for_fpconvert.
3067 (rs6000_address_for_fpconvert): Combine with
3068 rs6000_force_indexed_or_indirect_mem.
3069 (rs6000_expand_vector_init): Change function call from
3070 rs6000_address_for_fpconvert to rs6000_force_indexed_or_indirect_mem.
3071 * config/rs6000/rs6000.md (floatsi<mode>2_lfiwax): Change call from
3072 rs6000_address_for_fpconvert to rs6000_force_indexed_or_indirect_mem.
3073 (floatsi<mode>2_lfiwax_mem): Ditto.
3074 (floatunssi<mode>2_lfiwzx): Ditto.
3075 (floatunssi<mode>2_lfiwzx_mem): Ditto.
3076 (float<QHI:mode><FP_ISA3:mode>2): Ditto.
3077 (floatuns<QHI:mode><FP_ISA3:mode>2): Ditto.
3078 (fix_trunc<mode>si2_stfiwx): Ditto.
3079 (fixuns_trunc<mode>si2_stfiwx): Ditto.
3080 (float_<mode>si2_hw): Ditto.
3081 (floatuns_<mode>si2_hw): Ditto.
3082 * config/rs6000/vsx.md (*vsx_extract_si): Ditto.
3083 (vsx_splat_<mode>): Ditto.
3085 2018-11-01 Joseph Myers <joseph@codesourcery.com>
3087 * configure.ac (gcc_cv_as_ia64_ltoffx_ldxmov_relocs): Add newline
3088 at end of assembler input text.
3089 * configure: Regenerate.
3091 2018-11-01 Jakub Jelinek <jakub@redhat.com>
3093 PR tree-optimization/87826
3094 * gimple-ssa-store-merging.c (do_shift_rotate): Punt if count is
3095 negative or larger or equal to type's precision.
3097 2018-10-31 Alexandre Oliva <aoliva@redhat.com>
3099 * opts.c (default_options_table): Do not enable
3100 OPT_fdelayed_branch at -Og.
3101 * doc/invoke.texi (-fdelayed-branch): Document it.
3103 2018-10-31 Richard Henderson <richard.henderson@linaro.org>
3105 * optabs-libfuncs.c (build_libfunc_function_visibility):
3106 New, split out from...
3107 (build_libfunc_function): ... here.
3108 (init_one_libfunc_visibility): New, split out from ...
3109 (init_one_libfunc): ... here.
3111 * config/aarch64/atomics.md (aarch64_atomic_<ATOMIC_LDOP><ALLI>_lse):
3112 scratch register need not be early-clobber. Document the reason
3113 why we cannot use ST<OP>.
3115 2018-10-31 Joseph Myers <joseph@codesourcery.com>
3118 * configure.ac: Remove AC_PREREQ. Use AC_LANG_SOURCE. Use single
3119 line for second argument of AC_DEFINE_UNQUOTED.
3120 * doc/install.texi (Tools/packages necessary for modifying GCC):
3121 Update to autoconf 2.69 and automake 1.15.1.
3122 * aclocal.m4, config.in, configure: Regenerate.
3124 2018-10-31 Pat Haugen <pthaugen@us.ibm.com>
3126 * tree-ssa-loop-ivcanon.c (try_peel_loop): Fix typo and remove dead
3129 2018-10-31 Martin Liska <mliska@suse.cz>
3132 * common/config/aarch64/aarch64-common.c (aarch64_parse_extension):
3133 Add new argument invalid_extension.
3134 (aarch64_get_all_extension_candidates): New function.
3135 (aarch64_rewrite_selected_cpu): Add NULL to function call.
3136 * config/aarch64/aarch64-protos.h (aarch64_parse_extension): Add
3138 (aarch64_get_all_extension_candidates): New function.
3139 * config/aarch64/aarch64.c (aarch64_parse_arch): Add new
3140 argument invalid_extension.
3141 (aarch64_parse_cpu): Likewise.
3142 (aarch64_print_hint_for_extensions): New function.
3143 (aarch64_validate_mcpu): Provide hint about invalid extension.
3144 (aarch64_validate_march): Likewise.
3145 (aarch64_handle_attr_arch): Pass new argument.
3146 (aarch64_handle_attr_cpu): Provide hint about invalid extension.
3147 (aarch64_handle_attr_isa_flags): Likewise.
3149 2018-10-31 Richard Biener <rguenther@suse.de>
3153 * tree-outof-ssa.c (insert_backedge_copies): Restrict
3154 copy generation to useful cases. Place the copy before
3155 the definition of the backedge value when possible.
3157 2018-10-31 Claudiu Zissulescu <claziss@synopsys.com>
3159 * config/arc/arc-arch.h (ARC_TUNE_ARC7XX): New tune value.
3160 * config/arc/arc.c (arc_active_insn): New function.
3161 (check_store_cacheline_hazard): Likewise.
3162 (workaround_arc_anomaly): Use check_store_cacheline_hazard.
3163 (arc_override_options): Disable delay slot scheduler for older
3165 (arc_store_addr_hazard_p): New implementation, old one renamed to
3167 (arc_store_addr_hazard_internal_p): Renamed.
3168 (arc_reorg): Don't combine into brcc instructions which are part
3169 of hardware hazard solution.
3170 * config/arc/arc.md (attr tune): Consider new arc7xx tune value.
3171 (tune_arc700): Likewise.
3172 * config/arc/arc.opt (arc7xx): New tune value.
3173 * config/arc/arc700.md: Improve A7 scheduler.
3175 2018-10-31 Claudiu Zissulescu <claziss@synopsys.com>
3177 * config/arc/arc.c (arc_override_options): Remove
3178 TARGET_COMPACT_CASESI.
3179 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Update.
3180 (CASE_VECTOR_MODE): Likewise.
3181 (CASE_VECTOR_PC_RELATIVE): Likewise.
3182 (CASE_VECTOR_SHORTEN_MODE): Likewise.
3183 (CASE_VECTOR_SHORTEN_MODE1): Delete.
3184 (ADDR_VEC_ALIGN): Update.
3185 (ASM_OUTPUT_CASE_LABEL): Undefine.
3186 (ASM_OUTPUT_BEFORE_CASE_LABEL): Undefine.
3187 (TARGET_BI_BIH): Define.
3188 (DEFAULT_BRANCH_INDEX): Likewise.
3189 * config/arc/arc.md (casesi): Rework to accept BI/BIH
3190 instructions, remove compact_casesi use case.
3191 (casesi_compact_jump): Remove.
3192 (casesi_dispatch): New pattern.
3193 * config/arc/arc.opt: Add mbranch-index option. Deprecate
3194 compact_casesi option.
3195 * doc/invoke.texi: Document mbranch-index option.
3197 2018-10-31 Claudiu Zissulescu <claziss@synopsys.com>
3199 * config/arc/arc.c (arc_get_tp): Remove function.
3200 (arc_emit_call_tls_get_addr): Likewise.
3201 (arc_call_tls_get_addr): New function.
3202 (arc_legitimize_tls_address): Make use of arc_call_tls_get_addr.
3203 * config/arc/arc.md (tls_load_tp_soft): Remove.
3204 (tls_gd_get_addr): Likewise.
3206 2018-10-31 Claudiu Zissulescu <claziss@synopsys.com>
3208 * config/arc/arc.md (mulsi3): Remove call to mulsi_600_lib.
3209 (mulsi3_600_lib): Remove pattern.
3210 (umulsi3_highpart_600_lib_le): Likewise.
3211 (umulsi3_highpart): Remove call to umulsi3_highpart_600_lib_le.
3212 (umulsidi3): Remove call to umulsidi3_600_lib.
3213 (umulsidi3_600_lib): Remove pattern.
3214 (peephole2): Remove peephole using the above deprecated patterns.
3216 2018-10-31 Thomas Preud'homme <thomas.preudhomme@linaro.org>
3219 * config/arm/arm.c (arm_option_check_internal): Disable the combined
3220 use of -mslow-flash-data and -mword-relocations.
3221 (arm_option_override): Enable -mword-relocations if -fpic or -fPIC.
3222 * config/arm/arm.md (SYMBOL_REF MOVT splitter): Stop checking for
3224 * doc/invoke.texi (-mword-relocations): Mention conflict with
3226 (-mslow-flash-data): Reciprocally.
3228 2018-10-31 Richard Henderson <richard.henderson@linaro.org>
3230 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Force
3231 16-byte modes held in GP registers to use an even regno.
3233 * config/aarch64/aarch64.c (aarch64_emit_bic): Remove.
3234 (aarch64_atomic_ldop_supported_p): Remove.
3235 (aarch64_gen_atomic_ldop): Remove.
3236 * config/aarch64/atomic.md (atomic_<atomic_optab><ALLI>):
3237 Fully expand LSE operations here.
3238 (atomic_fetch_<atomic_optab><ALLI>): Likewise.
3239 (atomic_<atomic_optab>_fetch<ALLI>): Likewise.
3240 (aarch64_atomic_<ATOMIC_LDOP><ALLI>_lse): Drop atomic_op iterator
3241 and use ATOMIC_LDOP instead; use register_operand for the input;
3242 drop the split and emit insns directly.
3243 (aarch64_atomic_fetch_<ATOMIC_LDOP><ALLI>_lse): Likewise.
3244 (aarch64_atomic_<atomic_op>_fetch<ALLI>_lse): Remove.
3245 (@aarch64_atomic_load<ATOMIC_LDOP><ALLI>): Remove.
3247 * config/aarch64/aarch64.c (aarch64_emit_atomic_swap): Remove.
3248 (aarch64_gen_atomic_ldop): Don't call it.
3249 * config/aarch64/atomics.md (atomic_exchange<ALLI>):
3250 Use aarch64_reg_or_zero.
3251 (aarch64_atomic_exchange<ALLI>): Likewise.
3252 (aarch64_atomic_exchange<ALLI>_lse): Remove split; remove & from
3253 operand 0; use aarch64_reg_or_zero for input; merge ...
3254 (@aarch64_atomic_swp<ALLI>): ... this and remove.
3256 * config/aarch64/aarch64.c (aarch64_gen_compare_reg_maybe_ze): New.
3257 (aarch64_split_compare_and_swap): Use it.
3258 (aarch64_expand_compare_and_swap): Likewise. Remove convert_modes;
3259 test oldval against the proper predicate.
3260 * config/aarch64/atomics.md (@atomic_compare_and_swap<ALLI>):
3261 Use nonmemory_operand for expected.
3262 (cas_short_expected_pred): New.
3263 (@aarch64_compare_and_swap<SHORT>): Use it; use "rn" not "rI" to match.
3264 (@aarch64_compare_and_swap<GPI>): Use "rn" not "rI" for expected.
3265 * config/aarch64/predicates.md (aarch64_plushi_immediate): New.
3266 (aarch64_plushi_operand): New.
3268 * config/aarch64/aarch64.c (aarch64_expand_compare_and_swap):
3269 Force oldval into the rval register for TARGET_LSE; emit the compare
3270 during initial expansion so that it may be deleted if unused.
3271 (aarch64_gen_atomic_cas): Remove.
3272 * config/aarch64/atomics.md (@aarch64_compare_and_swap<SHORT>_lse):
3273 Change =&r to +r for operand 0; use match_dup for operand 2;
3274 remove is_weak and mod_f operands as unused. Drop the split
3276 (@aarch64_atomic_cas<SHORT>): ... this pattern's output; remove.
3277 (@aarch64_compare_and_swap<GPI>_lse): Similarly.
3278 (@aarch64_atomic_cas<GPI>): Similarly.
3280 2018-10-31 Richard Biener <rguenther@suse.de>
3282 * tree-eh.c (replace_trapping_overflow): Simplify ABS_EXPR case
3285 2018-10-31 Sameera Deshpande <sameera.deshpande@linaro.org>
3287 * config/aarch64/aarch64-cores.def (saphira): Use saphira pipeline.
3288 * config/aarch64/aarch64.md: Include saphira.md
3289 * config/aarch64/saphira.md: New file for pipeline description.
3291 2018-10-30 Martin Sebor <msebor@redhat.com>
3294 * gimple-ssa-sprintf.c (format_directive): Use %G to include
3296 (sprintf_dom_walker::compute_format_length):
3297 Avoid setting POSUNDER4K here.
3298 (get_destination_size): Handle null argument values.
3299 (get_user_idx_format): New function.
3300 (sprintf_dom_walker::handle_gimple_call): Handle all printf-like
3301 functions, including user-defined with attribute format printf.
3302 Use %G to include inlining context.
3303 Set POSUNDER4K here.
3305 2018-10-30 Jan Hubicka <jh@suse.cz>
3307 * params.def (lto-partitions): Bump from 32 to 128.
3309 2018-10-30 Jan Hubicka <jh@suse.cz>
3312 (free_lang_data_d, add_tree_to_fld_list, fld_worklist_push): Move
3314 (free_lang_data_in_type): Forward declare.
3315 (fld_type_variant_equal_p): New function.
3316 (fld_type_variant): New function
3317 (fld_incomplete_types): New hash.
3318 (fld_incomplete_type_of): New function
3319 (fld_simplfied-type): New function.
3320 (free_lang_data_in_decl): Add fld parameter; simplify type of FIELD_DECL
3321 (free_lang_data): Allocate and free fld_incomplete_type; update call
3322 of free_lang_data_in_decl.
3324 2018-10-30 Eric Botcazou <ebotcazou@adacore.com>
3326 * gcov.c (output_lines): Remove duplicate line.
3328 2018-10-30 Aaron Sawdey <acsawdey@linux.ibm.com>
3330 * config/rs6000/rs6000.md (bswapdi2): Force address into register
3331 if not in indexed or indirect form.
3332 (bswapdi2_load): Change predicate to indexed_or_indirect_operand.
3333 (bswapdi2_store): Ditto.
3334 * config/rs6000/rs6000.c (rs6000_force_indexed_or_indirect_mem): New
3336 * config/rs6000/rs6000-protos.h (rs6000_force_indexed_or_indirect_mem):
3337 Prototype for helper function.
3339 2018-10-30 Martin Sebor <msebor@redhat.com>
3341 * doc/extend.texi (optimize): Clarify/expand attribute documentation.
3342 (target, pragma GCC optimize, pragma GCC target): Ditto.
3344 2018-10-30 Jonathan Wakely <jwakely@redhat.com>
3346 * doc/extend.texi: Fix prototype and description of
3347 __builtin_expect_with_probability.
3349 2018-10-30 Michael Ploujnikov <michael.ploujnikov@oracle.com>
3351 * cgraph.h (clone_function_name_1): Replaced by new
3352 clone_function_name_numbered that takes name as string; for
3353 privatize_symbol_name_1 use only.
3354 (clone_function_name): Renamed to
3355 clone_function_name_numbered to be explicit about numbering.
3356 (clone_function_name): New two-argument function that does
3357 not number its output.
3358 (clone_function_name): New three-argument function that
3359 takes a number to append to its output.
3360 * cgraphclones.c (duplicate_thunk_for_node):
3361 (clone_function_name_1): Renamed.
3362 (clone_function_name_numbered): Two new functions.
3363 (clone_function_name): Improved documentation.
3364 (cgraph_node::create_virtual_clone): Use clone_function_name_numbered.
3365 * config/rs6000/rs6000.c (make_resolver_func): Ditto.
3366 * final.c (final_scan_insn_1): Use the new clone_function_name
3368 * multiple_target.c (create_dispatcher_calls): Ditto.
3369 (create_target_clone): Ditto.
3370 * omp-expand.c (grid_expand_target_grid_body): Ditto.
3371 * omp-low.c (create_omp_child_function_name): Ditto.
3372 * omp-simd-clone.c (simd_clone_create): Ditto.
3373 * symtab.c (simd_symtab_node::noninterposable_alias): Use the
3374 new clone_function_name without numbering.
3376 2018-10-30 Richard Earnshaw <rearnsha@arm.com>
3378 * alloc-pool.h (base_pool_allocator <TBlockAllocator>::initialize):
3379 Assert that the allocation size is not zero.
3381 2018-10-30 Richard Biener <rguenther@suse.de>
3383 PR tree-optimization/87800
3384 * tree-vect-slp.c (vect_build_slp_tree_2): Reject any
3385 non-induction or reduction PHIs.
3387 2018-10-30 Sameera Deshpande <sameera.deshpande@linaro.org
3389 * config/aarch64/falkor-tag-collision-avoidance.c
3390 (execute_tag_collision_avoidance): Call df_note_add_problem.
3392 2018-10-30 Martin Liska <mliska@suse.cz>
3394 * doc/extend.texi: Fix typo in documentation
3395 of __builtin_expect_with_probability.
3397 2018-10-29 David Malcolm <dmalcolm@redhat.com>
3400 * input.c (get_substring_ranges_for_loc): Detect if
3401 linemap_resolve_location gives us a NULL map, and reject
3404 2018-10-29 Iain Buclaw <ibuclaw@gdcproject.org>
3406 * config.gcc (xstormy16-*-elf): Set tm_d_file.
3408 2018-10-29 David Malcolm <dmalcolm@redhat.com>
3409 Martin Sebor <msebor@redhat.com>
3410 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
3412 * doc/ux.texi (Quoting): New subsection, adapted from material at
3413 https://gcc.gnu.org/wiki/DiagnosticsGuidelines written by
3414 MartinSebor and ManuelLopezIbanez.
3415 (Fix-it hints): Note that fix-it hints shouldn't be marked for
3418 2018-10-29 Kugan Vivekanandarajah <kuganv@linaro.org>
3421 * tree-ssa-loop-niter.c (number_of_iterations_popcount): Fix niter
3424 2018-10-29 Paul A. Clarke <pc@us.ibm.com>
3426 * gcc.target/powerpc/ssse3-check.h: Remove inaccurate comment.
3428 2018-10-29 Paul Koning <ni1d@arrl.net>
3430 * doc/rtl.texi (CONST_WIDE_INT_ELT): Give correct macro name.
3432 2018-10-29 Paul A. Clarke <pc@us.ibm.com>
3434 * gcc/config/rs6000/mmintrin.h (_mm_packs_pi16, _mm_packs_pi32,
3435 _mm_packs_pu16, _mm_unpackhi_pi8, _mm_unpacklo_pi8, _mm_add_pi8,
3436 _mm_add_pi16, _mm_add_pi32, _mm_sub_pi8, _mm_sub_pi16, _mm_sub_pi32,
3437 _mm_cmpgt_pi8, _mm_cmpeq_pi16, _mm_cmpgt_pi16, _mm_cmpeq_pi32,
3438 _mm_cmpgt_pi32, _mm_adds_pi8, _mm_adds_pi16, _mm_adds_pu8,
3439 _mm_adds_pu16, _mm_subs_pi8, _mm_subs_pi16, _mm_subs_pu8,
3440 _mm_subs_pu16, _mm_madd_pi16, _mm_mulhi_pi16, _mm_mullo_pi16,
3441 _mm_sll_pi16, _mm_sra_pi16, _mm_srl_pi16, _mm_set1_pi16, _mm_set1_pi8):
3442 Change 'vector' to '__vector'.
3443 * gcc/config/rs6000/xmmintrin.h (_mm_cvtps_pi32, _mm_cvttps_pi32,
3444 _mm_cvtps_pi16, _mm_cvtps_pi8, _mm_max_pi16, _mm_max_pu8, _mm_min_pi16,
3445 _mm_min_pu8, _mm_mulhi_pu16, _mm_shuffle_pi16, _mm_avg_pu8,
3446 _mm_avg_pu16): Likewise. And, whitespace corrections.
3448 2018-10-29 Richard Biener <rguenther@suse.de>
3450 PR tree-optimization/87785
3451 * tree-vect-slp.c (vect_gather_slp_loads): Only gather
3454 2018-10-29 Olivier Hainque <hainque@adacore.com>
3456 * VXWORKS_NET_LIBS_RTP: New macro, network part of VXWORKS_LIBS_RTP.
3458 2018-10-29 Olivier Hainque <hainque@adacore.com>
3460 * config/vxworks.h (VXWORKS_PERSONALITY): New VxWorks
3461 ports configuration macro, defaults to "gnu".
3462 (VXWORKS_OS_CPP_BUILTINS): Use it. Feed TOOL and
3463 TOOL_FAMILY instead of _VX_TOOL/_VXTOOL_FAMILY.
3465 2018-10-29 Olivier Hainque <hainque@adacore.com>
3467 * config/vxworks.h (CLEAR_INSN_CACHE): #undef before #define.
3469 2018-10-29 Richard Biener <rguenther@suse.de>
3471 PR tree-optimization/87790
3472 * tree-vect-slp.c (vect_mark_slp_stmts): Simplify.
3473 (vect_make_slp_decision): Adjust.
3474 (vect_slp_analyze_bb_1): Likewise.
3475 (vect_detect_hybrid_slp_stmts): Properly union SLP type over
3478 2018-10-29 Richard Biener <rguenther@suse.de>
3480 PR tree-optimization/87785
3481 * tree-vect-slp.c (vect_build_slp_tree_2): Remove loads argument
3483 (vect_build_slp_tree): Likewise.
3484 (vect_gather_slp_loads): New function.
3485 (vect_analyze_slp_instance): Gather loads separately from the
3488 2018-10-29 Martin Liska <mliska@suse.cz>
3490 * Makefile.in: Make dependency to json.o.
3491 * doc/gcov.texi: Document new JSON format, remove
3492 old intermediate format documentation.
3493 * gcov.c (struct function_info): Come up with m_name and
3495 (function_info::function_info): Initialize it.
3496 (function_info::~function_info): Release it.
3497 (main): Rename flag_intermediate_format to flag_json_format.
3498 (print_usage): Describe --json-format.
3499 (process_args): Set flag_json_format.
3500 (output_intermediate_line): Remove.
3501 (output_intermediate_json_line): Likewise.
3502 (get_gcov_intermediate_filename): Return new extension
3504 (output_intermediate_file): Implement JSON emission.
3505 (output_json_intermediate_file): Implement JSON emission.
3506 (generate_results): Use ::get_name for function name.
3507 Handle JSON output file.
3508 (read_graph_file): Use ::get_name instead of cplus_demangle.
3509 (read_count_file): Likewise.
3510 (solve_flow_graph): Likewise.
3511 (add_line_counts): Likewise.
3512 (accumulate_line_counts): Use new flag_json_format.
3513 (output_function_details): Use ::get_name instead of cplus_demangle.
3514 (output_lines): Likewise.
3515 * json.cc (test_writing_literals): Add new tests.
3516 * json.h (class literal): Add new boolean constructor.
3518 2018-10-29 Segher Boessenkool <segher@kernel.crashing.org>
3520 PR rtl-optimization/87701
3521 PR rtl-optimization/87780
3522 * combine.c (make_more_copies): Rewrite.
3524 2018-10-28 Kugan Vivekanandarajah <kuganv@linaro.org>
3526 * doc/generic.texi (ABSU_EXPR): Document.
3527 * match.pd (absu(x)*absu(x) -> x*x): Handle.
3528 (absu(absu(X)) -> absu(X)): Likewise.
3529 (absu(-X) -> absu(X)): Likewise.
3530 (absu(X) where X is nonnegative -> X): Likewise.
3532 2018-10-28 Iain Buclaw <ibuclaw@gdcproject.org>
3534 * Makefile.in (tm_d_file_list, tm_d_include_list): New variables.
3535 (TM_D_H, D_TARGET_DEF, D_TARGET_H, D_TARGET_OBJS): New variables.
3536 (tm_d.h, cs-tm_d.h, default-d.o): New rules.
3537 (d/d-target-hooks-def.h, s-d-target-hooks-def-h): New rules.
3538 (s-tm-texi): Also check timestamp on d-target.def.
3539 (generated_files): Add TM_D_H and d-target-hooks-def.h.
3540 (build/genhooks.o): Also depend on D_TARGET_DEF.
3541 * config.gcc (tm_d_file, d_target_objs, target_has_targetdm): New
3543 * config/aarch64/aarch64-d.c: New file.
3544 * config/aarch64/aarch64-linux.h (GNU_USER_TARGET_D_CRITSEC_SIZE):
3546 * config/aarch64/aarch64-protos.h (aarch64_d_target_versions): New
3548 * config/aarch64/aarch64.h (TARGET_D_CPU_VERSIONS): Define.
3549 * config/aarch64/t-aarch64 (aarch64-d.o): New rule.
3550 * config/arm/arm-d.c: New file.
3551 * config/arm/arm-protos.h (arm_d_target_versions): New prototype.
3552 * config/arm/arm.h (TARGET_D_CPU_VERSIONS): Define.
3553 * config/arm/linux-eabi.h (EXTRA_TARGET_D_OS_VERSIONS): Define.
3554 * config/arm/t-arm (arm-d.o): New rule.
3555 * config/default-d.c: New file.
3556 * config/glibc-d.c: New file.
3557 * config/gnu.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
3558 * config/i386/i386-d.c: New file.
3559 * config/i386/i386-protos.h (ix86_d_target_versions): New prototype.
3560 * config/i386/i386.h (TARGET_D_CPU_VERSIONS): Define.
3561 * config/i386/linux-common.h (EXTRA_TARGET_D_OS_VERSIONS): Define.
3562 (GNU_USER_TARGET_D_CRITSEC_SIZE): Define.
3563 * config/i386/t-i386 (i386-d.o): New rule.
3564 * config/kfreebsd-gnu.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
3565 * config/kopensolaris-gnu.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
3566 * config/linux-android.h (ANDROID_TARGET_D_OS_VERSIONS): Define.
3567 * config/linux.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
3568 * config/mips/linux-common.h (EXTRA_TARGET_D_OS_VERSIONS): Define.
3569 * config/mips/mips-d.c: New file.
3570 * config/mips/mips-protos.h (mips_d_target_versions): New prototype.
3571 * config/mips/mips.h (TARGET_D_CPU_VERSIONS): Define.
3572 * config/mips/t-mips (mips-d.o): New rule.
3573 * config/powerpcspe/linux.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
3574 * config/powerpcspe/linux64.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
3575 * config/powerpcspe/powerpcspe-d.c: New file.
3576 * config/powerpcspe/powerpcspe-protos.h (rs6000_d_target_versions):
3578 * config/powerpcspe/powerpcspe.c (rs6000_output_function_epilogue):
3579 Support GNU D by using 0 as the language type.
3580 * config/powerpcspe/powerpcspe.h (TARGET_D_CPU_VERSIONS): Define.
3581 * config/powerpcspe/t-powerpcspe (powerpcspe-d.o): New rule.
3582 * config/riscv/riscv-d.c: New file.
3583 * config/riscv/riscv-protos.h (riscv_d_target_versions): New
3585 * config/riscv/riscv.h (TARGET_D_CPU_VERSIONS): Define.
3586 * config/riscv/t-riscv (riscv-d.o): New rule.
3587 * config/rs6000/linux.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
3588 * config/rs6000/linux64.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
3589 * config/rs6000/rs6000-d.c: New file.
3590 * config/rs6000/rs6000-protos.h (rs6000_d_target_versions): New
3592 * config/rs6000/rs6000.c (rs6000_output_function_epilogue):
3593 Support GNU D by using 0 as the language type.
3594 * config/rs6000/rs6000.h (TARGET_D_CPU_VERSIONS): Define.
3595 * config/rs6000/t-rs6000 (rs6000-d.o): New rule.
3596 * config/s390/s390-d.c: New file.
3597 * config/s390/s390-protos.h (s390_d_target_versions): New prototype.
3598 * config/s390/s390.h (TARGET_D_CPU_VERSIONS): Define.
3599 * config/s390/t-s390 (s390-d.o): New rule.
3600 * config/sparc/sparc-d.c: New file.
3601 * config/sparc/sparc-protos.h (sparc_d_target_versions): New
3603 * config/sparc/sparc.h (TARGET_D_CPU_VERSIONS): Define.
3604 * config/sparc/t-sparc (sparc-d.o): New rule.
3605 * config/t-glibc (glibc-d.o): New rule.
3606 * configure: Regenerated.
3607 * configure.ac (tm_d_file): New variable.
3608 (tm_d_file_list, tm_d_include_list, d_target_objs): Add substitutes.
3609 * doc/contrib.texi (Contributors): Add self for the D frontend.
3610 * doc/frontends.texi (G++ and GCC): Mention D as a supported language.
3611 * doc/install.texi (Configuration): Mention libphobos as an option for
3612 --enable-shared. Mention d as an option for --enable-languages.
3613 (Testing): Mention check-d as a target.
3614 * doc/invoke.texi (Overall Options): Mention .d, .dd, and .di as file
3615 name suffixes. Mention d as a -x option.
3616 * doc/sourcebuild.texi (Top Level): Mention libphobos.
3617 * doc/standards.texi (Standards): Add section on D language.
3618 * doc/tm.texi: Regenerated.
3619 * doc/tm.texi.in: Add @node for D language and ABI, and @hook for
3620 TARGET_CPU_VERSIONS, TARGET_D_OS_VERSIONS, and TARGET_D_CRITSEC_SIZE.
3621 * dwarf2out.c (is_dlang): New function.
3622 (gen_compile_unit_die): Use DW_LANG_D for D.
3623 (declare_in_namespace): Return module die for D, instead of adding
3624 extra declarations into the namespace.
3625 (gen_namespace_die): Generate DW_TAG_module for D.
3626 (gen_decl_die): Handle CONST_DECLSs for D.
3627 (dwarf2out_decl): Likewise.
3628 (prune_unused_types_walk_local_classes): Handle DW_tag_interface_type.
3629 (prune_unused_types_walk): Handle DW_tag_interface_type same as other
3630 kinds of aggregates.
3631 * gcc.c (default_compilers): Add entries for .d, .dd and .di.
3632 * genhooks.c: Include d/d-target.def.
3634 2018-10-28 Iain Sandoe <iain@sandoe.co.uk>
3637 * config/rs6000/darwin.h (STACK_BOUNDARY): New.
3638 (RS6000_STARTING_FRAME_OFFSET): Adjust to preserve 16byte alignment.
3639 (STACK_DYNAMIC_OFFSET): Likewise.
3641 2018-10-27 Sandra Loosemore <sandra@codesourcery.com>
3644 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Fix
3647 2018-10-26 Tulio Magno Quites Machado Filho <tuliom@linux.ibm.com>
3649 * doc/extend.texi (PowerPC builtins): Fix __builtin_unpack_ibm128
3650 return type and other typos.
3652 2018-10-26 Aaron Sawdey <acsawdey@linux.ibm.com>
3654 * config/rs6000/rs6000-string.c (expand_strncmp_gpr_sequence): Change to
3655 a shorter sequence with fewer branches.
3656 (emit_final_str_compare_gpr): Ditto.
3658 2018-10-26 Paul A. Clarke <pc@us.ibm.com>
3660 * config/rs6000/tmmintrin.h: New file.
3661 * config.gcc (powerpc*-*-*): Add tmmintrin.h to extra_headers.
3663 2018-10-26 Paul A. Clarke <pc@us.ibm.com>
3665 * config/rs6000/mmintrin.h: Enable 32bit compilation.
3666 * config/rs6000/xmmintrin.h: Likewise.
3668 2018-10-26 Paul A. Clarke <pc@us.ibm.com>
3670 * config/rs6000/xmmintrin.h (_mm_extract_pi16): Fix for big-endian.
3672 2018-10-26 Richard Biener <rguenther@suse.de>
3674 * tree-vect-slp.c (vect_mark_slp_stmts): Add visited hash_set
3676 (vect_mark_slp_stmts_relevant): Likewise.
3677 (vect_detect_hybrid_slp_stmts): Likewise.
3678 (vect_bb_slp_scalar_cost): Likewise.
3679 (vect_remove_slp_scalar_calls): Likewise.
3681 2018-10-26 Jan Hubicka <jh@suse.cz>
3683 * ipa-devirt.c (odr_subtypes_equivalent_p): Fix recursion.
3684 (warn_types_mismatch): Fix walk of DECL_NAME.
3685 (odr_types_equivalent_p): Fix overactive assert.
3687 2018-10-26 Richard Biener <rguenther@suse.de>
3689 PR tree-optimization/87105
3690 * tree-vectorizer.h (_slp_tree::refcnt): New member.
3691 * tree-vect-slp.c (vect_free_slp_tree): Decrement and honor
3693 (vect_create_new_slp_node): Initialize refcnt to one.
3695 (scalar_stmts_set_t, bst_fail): Remove.
3696 (vect_build_slp_tree_2): Add bst_map argument and adjust calls.
3697 (vect_build_slp_tree): Add bst_map argument and lookup
3698 already created SLP nodes.
3699 (vect_print_slp_tree): Handle a SLP graph, print SLP node
3701 (vect_slp_rearrange_stmts): Handle a SLP graph.
3702 (vect_analyze_slp_instance): Adjust and free SLP nodes from
3703 the CSE map. Fix indenting.
3704 (vect_schedule_slp_instance): Add short-cut.
3706 2018-10-26 Martin Liska <mliska@suse.cz>
3709 * ipa-prop.c (struct ipa_vr_ggc_hash_traits): Hash with
3710 addr_expr and not with pointers.
3712 2018-10-26 Jan Hubicka <jh@suse.cz>
3714 * tree.c (free_lang_data_in_type): Only check main variants.
3715 * ipa-devirt.c (warn_odr): Make static.
3716 (types_same_for_odr): Drop strict variant.
3717 (types_odr_comparable): Likewise.
3718 (odr_or_derived_type_p): Look for main variants.
3719 (odr_name_hasher::equal): Cleanup comment.
3720 (odr_subtypes_equivalent): Add warn and warned arguments; check main
3722 (type_variants_equivalent_p): break out from ...
3723 (odr_types_equivalent): ... here; go for main variants where needed.
3724 (warn_odr): ... here; turn static.
3725 (warn_types_mismatch): Compare mangled names of main variants.
3726 * ipa-utils.h (types_odr_comparable): Drop strict parameter.
3727 (type_with_linkage_p): Sanity check that we look at main variant.
3728 * lto.c (lto_read_decls): Only consider main variant to be ODR type.
3729 * tree.h (types_same_for_odr): Drop strict argument.
3731 2018-10-26 Richard Biener <rguenther@suse.de>
3733 PR tree-optimization/87746
3734 * tree-vect-data-refs.c (vect_update_misalignment_for_peel):
3735 Simplify and fix WRT strided store groups with size not
3736 equal to step in element count.
3737 (vect_analyze_group_access_1): Dump the whole group.
3739 2018-10-25 Carl Love <cel@us.ibm.com>
3741 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_VSCEDPGT,
3742 P9V_BUILTIN_VEC_VSCEDPLT, P9V_BUILTIN_VEC_VSCEDPEQ,
3743 P9V_BUILTIN_VEC_VSCEDPUO): Rename base overloaded name. Add quad
3744 precicion entry for each overloaded builtin.
3745 * config/rs6000/rs6000-builtin.def (VSCEDPGT, VSCEDPLT, VSCEDPEQ,
3746 VSCEDPUO): Rename overloaded name.
3747 (VSCEDPGT, VSCEQPGT, VSCEDPLT, VSCEQPLT, VSCEDPEQ, VSCEQPEQ,
3748 VSCEDPUO, VSCEQPUO): Add defitions for overloaded builtins.
3749 * config/rs6000/vsx.md (xscmpexpqp_<code>_<mode>): Add
3750 define_expand for xscmpexqp instruction.
3751 (*xscmpexpqp): Add define_insn for the xscmpexqp instruction.
3753 2018-10-25 Bill Schmidt <wschmidt@linux.ibm.com>
3754 Jinsong Ji <jji@us.ibm.com>
3756 * config/rs6000/emmintrin.h (_mm_slli_epi16): Replace deprecated
3757 function with vec_sl.
3758 (_mm_slli_epi32): Likewise.
3759 (_mm_slli_epi64): Likewise.
3760 (_mm_srai_epi16): Replace deprecated function with vec_sra.
3761 (_mm_srai_epi32): Likewise.
3762 (_mm_srli_epi16): Replace deprecated function with vec_sr.
3763 (_mm_srli_epi32): Likewise.
3764 (_mm_srli_epi64): Likewise.
3765 (_mm_sll_epi16): Replace deprecated function with vec_sl.
3766 (_mm_sll_epi32): Likewise.
3767 (_mm_sll_epi64): Likewise.
3768 (_mm_sra_epi16): Replace deprecated function with vec_sra.
3769 (_mm_sra_epi32): Likewise.
3770 (_mm_srl_epi16): Replace deprecated function with vec_sr.
3771 (_mm_srl_epi32): Likewise.
3772 (_mm_srl_epi64): Likewise.
3774 2018-10-25 Bill Schmidt <wschmidt@linux.ibm.com>
3775 Jinsong Ji <jji@us.ibm.com>
3777 * gcc/config/rs6000/emmintrin.h (_mm_sll_epi16): Replace
3778 comparison operators with vec_cmp* for compatibility due to
3779 unfortunate history; clean up formatting and use types more
3781 (_mm_sll_epi32): Likewise.
3782 (_mm_sll_epi64): Likewise.
3783 (_mm_srl_epi16): Likewise.
3784 (_mm_srl_epi32): Likewise.
3785 (_mm_srl_epi64): Likewise.
3787 2018-10-25 Bill Schmidt <wschmidt@linux.ibm.com>
3788 Jinsong Ji <jji@us.ibm.com>
3790 * config/rs6000/emmintrin.h (_mm_sll_epi64): Remove wrong cast.
3791 * config/rs6000/xmmintrin.h (_mm_min_ps): Change m's type to
3792 __vector __bool int. Use vec_cmpgt in preference to deprecated
3793 function vec_vcmpgtfp.
3794 (_mm_max_ps): Likewise.
3796 2018-10-25 Jeff Law <law@redhat.com>
3798 * config/rl78/rl78.c (insn_ok_now): Always re-recognize the insn
3801 2018-10-25 Martin Sebor <msebor@redhat.com>
3803 * doc/extend.texi (aligned): Expand attribute description.
3804 (Alignment): Rename section. Discuss function arguments.
3806 2018-10-25 Jan Hubicka <jh@suse.cz>
3808 * ipa-devirt.c (main_odr_variant): Remove.
3809 (hash_odr_name, types_same_for_odr, types_odr_comparable,
3810 odr_name_hasher::equal, odr_subtypes_equivalent_p):
3811 Drop use of main_odr_variant.
3812 (add_type_duplicate): Silence confused warnings on integer types.
3813 (get_odr_type): Always look for main variant.
3814 (register_odr_type): Simplify.
3816 2018-10-25 Richard Biener <rguenther@suse.de>
3818 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses):
3819 Initialize ng to silence error with release checking bootstrap.
3821 2018-10-25 Richard Biener <rguenther@suse.de>
3823 * tree-if-conv.c: Include tree-ssa-sccvn.h.
3824 (tree_if_conversion): Run CSE on the if-converted loop body.
3826 2018-10-25 Ilya Leoshkevich <iii@linux.ibm.com>
3828 * config/s390/constraints.md (ZL): New constraint.
3829 * config/s390/s390.c (legitimate_pic_operand_p): Accept LARL
3831 * config/s390/s390.md (movdi_larl): Remove.
3832 (movdi_64): Add the LARL alternative.
3834 2018-10-25 Ilya Leoshkevich <iii@linux.ibm.com>
3837 * rtl.c (RTX_CODE_HWINT_P_1): New helper macro.
3838 (RTX_CODE_HWINT_P): New macro.
3839 (rtx_code_size): Use RTX_CODE_HWINT_P ().
3841 2018-10-25 Jan Hubicka <jh@suse.cz>
3843 * ipa-devirt.c (odr_types_equivalent_p): Do not ICE if one of types
3846 2018-10-25 Richard Biener <rguenther@suse.de>
3848 PR tree-optimization/87665
3849 PR tree-optimization/87745
3850 * tree-vectorizer.h (get_earlier_stmt): Remove.
3851 (get_later_stmt): Pick up UID from the original non-pattern stmt.
3853 2018-10-25 Sam Tebbs <sam.tebbs@arm.com>
3855 * options.texi (Deprecated): Move list to Var section.
3857 2018-10-24 Bill Schmidt <wschmidt@linux.ibm.com>
3858 Jinsong Ji <jji@us.ibm.com>
3860 * config/rs6000/emmintrin.h (_mm_cvtpd_epi32): Change deprecated
3861 __vector long to __vector long long.
3862 (_mm_cvtpd_ps): Likewise.
3863 (_mm_cvttpd_epi32): Likewise.
3864 (_mm_cvtpi32_pd): Likewise.
3865 (_mm_unpackhi_epi64): Likewise.
3866 (_mm_unpacklo_epi64): Likewise.
3868 2018-10-24 Segher Boessenkool <segher@kernel.crashing.org>
3870 PR rtl-optimization/87720
3871 * combine.c (make_more_copies): Skip if the dest is pc_rtx.
3873 2018-10-24 Alexandre Oliva <aoliva@redhat.com>
3875 * gimple-ssa-isolate-paths.c
3876 (find_implicit_erroneous_behavior): Do not change code if the
3877 pass is running for warnings only.
3878 (find_explicit_erroneous_behavior): Likewise.
3880 2018-10-24 Michael Meissner <meissner@linux.ibm.com>
3882 * config/rs6000/rs6000.c (TARGET_MANGLE_DECL_ASSEMBLER_NAME):
3883 Define as rs6000_mangle_decl_assembler_name.
3884 (rs6000_mangle_decl_assembler_name): If the user switched from IBM
3885 long double to IEEE long double, switch the names of the long
3886 double built-in functions to be <func>f128 instead of <func>l.
3888 2018-10-24 Martin Sebor <msebor@redhat.com>
3890 * doc/extend.texi (nonnull): List no-argument form. Reference
3891 -fno-delete-null-pointer-checks and -fisolate-erroneous-paths-attribute.
3893 2018-10-24 Richard Biener <rguenther@suse.de>
3895 * tree-ssa-sccvn.c (do_rpo_vn): Free rpo_state.
3897 2018-10-24 Martin Liska <mliska@suse.cz>
3899 PR tree-optimization/84436
3900 * tree-switch-conversion.c (switch_conversion::contains_same_values_p):
3902 (switch_conversion::contains_linear_function_p): New.
3903 (switch_conversion::build_one_array): Support linear
3904 transformation on input.
3905 * tree-switch-conversion.h (struct switch_conversion): Add
3906 contains_linear_function_p declaration.
3908 2018-10-24 Richard Biener <rguenther@suse.de>
3910 * varasm.c (const_hash_1): Return hash of ADDR_EXPR
3911 if its argument is CONSTANT_CLASS_P.
3913 2018-10-11 Jan Hubicka <hubicka@ucw.cz>
3915 * ipa-utils.h (type_with_linkage_p): No longer check for TYPE_STUB_DECL;
3916 it is wrong for forward declarations.
3918 2018-10-24 Ilya Leoshkevich <iii@linux.ibm.com>
3920 * config/s390/s390.c (s390_check_qrst_address): Add the missing
3921 SYMBOL_REF_P () check.
3923 2018-10-24 Richard Biener <rguenther@suse.de>
3925 PR tree-optimization/87105
3926 * tree-vect-data-refs.c (vect_analyze_group_access_1): Adjust
3927 dump classification.
3928 (vect_analyze_data_ref_accesses): Handle duplicate loads and
3929 stores by splitting the affected group after the fact.
3930 * tree-vect-slp.c (vect_build_slp_tree_2): Dump when we
3931 fail the SLP build because of size constraints.
3933 2018-10-24 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3935 * configure.ac (gcc_cv_ld_aligned_shf_merge): New test.
3936 * configure: Regenerate.
3937 * config.in: Regenerate.
3938 * varasm.c (mergeable_string_section): Use readonly_data_section
3939 if linker doesn't support SHF_MERGE with alignment > 8.
3940 (mergeable_constant_section): Likewise.
3942 2018-10-24 Richard Biener <rguenther@suse.de>
3944 PR tree-optimization/84013
3945 * tree-ssa-structalias.c (struct msdi_data): New struct for
3946 marshalling data to walk_stmt_load_store_ops.
3947 (maybe_set_dependence_info): Refactor as callback for
3948 walk_stmt_load_store_ops.
3949 (compute_dependence_clique): Set restrict info on all stmt kinds.
3951 2018-10-24 Martin Liska <mliska@suse.cz>
3953 * cgraph.c (cgraph_node::dump):
3954 Remove reduntant dumps and make tp_first_run dump more compact.
3956 2018-10-24 Richard Biener <rguenther@suse.de>
3958 PR tree-optimization/87665
3959 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Adjust
3962 2018-10-12 Jeff Law <law@redhat.com>
3964 * config/h8300/h8300.c (h8300_expand_prologue): Fix stm generation
3967 2018-10-23 Richard Biener <rguenther@suse.de>
3969 * tree-vrp.c (add_assert_info): Guard dump_printf with
3971 * gimple-ssa-evrp-analyze.c
3972 (evrp_range_analyzer::record_ranges_from_incoming_edge):
3973 Use value_range::ignore_equivs_equal_p.
3975 2018-10-23 Richard Biener <rguenther@suse.de>
3977 PR tree-optimization/87105
3978 PR tree-optimization/87608
3979 * passes.def (pass_all_early_optimizations): Add early phi-opt
3981 * tree-ssa-phiopt.c (value_replacement): Ignore NOPs and predicts in
3982 addition to debug stmts.
3983 (tree_ssa_phiopt_worker): Add early_p argument, do only min/max
3984 and abs replacement early.
3985 * tree-cfg.c (gimple_empty_block_p): Likewise.
3987 2018-10-23 Richard Earnshaw <rearnsha@arm.com>
3990 * config.gcc (arm*-*-netbsdelf*): Default to StrongARM if no CPU
3991 specified to configure.
3992 (arm*-*-*): Use ARM7TDMI as the target CPU if no default provided.
3994 2018-10-23 Richard Biener <rguenther@suse.de>
3996 PR tree-optimization/87700
3997 * tree-ssa-copy.c (set_copy_of_val): Fix change detection logic.
3999 2018-10-23 Jakub Jelinek <jakub@redhat.com>
4002 * config/i386/avx512vlintrin.h (_mm_mask_mullo_epi32): Change type of
4003 second argument from __mmask16 to __mmask8.
4004 * config/i386/avx512vlbwintrin.h (_mm_mask_packus_epi32,
4005 _mm_mask_packs_epi32): Likewise.
4006 * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i64scatter_ps):
4008 (_mm512_mask_prefetch_i64scatter_pd): Likewise. Formatting fix.
4010 2018-10-23 Richard Biener <rguenther@suse.de>
4012 * tree-vect-stmts.c (vect_analyze_stmt): Fix typo in comment.
4014 2018-10-23 Richard Biener <rguenther@suse.de>
4016 PR tree-optimization/86144
4017 * tree-vect-stmts.c (vect_analyze_stmt): Prefer -mveclibabi
4018 over simd attribute.
4020 2018-10-23 Richard Biener <rguenther@suse.de>
4022 PR tree-optimization/87693
4023 * tree-ssa-threadedge.c (thread_around_empty_blocks): Handle
4024 the case we do not find the taken edge.
4026 2018-10-22 Bill Schmidt <wschmidt@linux.ibm.com>
4027 Jinsong Ji <jji@us.ibm.com>
4029 * config/rs6000/emmintrin.h (_MM_SHUFFLE2): Comment cleanup.
4030 (_mm_store_pd): Use unaligned vector type for pointer cast.
4031 (_mm_maskmoveu_si128): Likewise.
4032 * config/rs6000/xmmintrin.h (__m128_u): New typedef.
4033 (_mm_store_ps): Use unaligned vector type for pointer cast.
4035 2018-10-22 Paul Koning <ni1d@arrl.net>
4037 * symtab.c (symtab_node::increase_alignment): Correct max
4040 2018-10-22 Yury Gribov <tetra2005@gmail.com>
4042 PR tree-optimization/87633
4043 * match.pd: Do not generate unordered integer comparisons.
4045 2018-10-22 Segher Boessenkool <segher@kernel.crashing.org>
4047 PR rtl-optimization/87600
4048 * combine.c: Add include of expr.h.
4049 (cant_combine_insn_p): Do not combine moves from any hard non-fixed
4050 register to a pseudo.
4051 (make_more_copies): New function, add a copy to a new pseudo after
4052 the moves from hard registers into pseudos.
4053 (rest_of_handle_combine): Declare rebuild_jump_labels_after_combine
4054 later. Call make_more_copies.
4056 2018-10-22 Andrew Stubbs <ams@codesourcery.com>
4058 * lra-constraints.c (process_alt_operands): New local array,
4059 matching_early_clobber. Check matching_early_clobber before
4060 decrementing reject, and set matching_early_clobber after.
4062 2018-10-22 Segher Boessenkool <segher@kernel.crashing.org>
4065 * config/rs6000/rs6000.c (print_operand_address): For unexpected RTL
4066 call output_addr_const and hope for the best.
4068 2018-10-22 Richard Biener <rguenther@suse.de>
4070 * gimple-ssa-evrp-analyze.c
4071 (evrp_range_analyzer::record_ranges_from_incoming_edge): Be
4072 smarter about what ranges to use.
4073 * tree-vrp.c (add_assert_info): Dump here.
4074 (register_edge_assert_for_2): Instead of here at multiple but
4077 * gcc.dg/tree-ssa/evrp12.c: New testcase.
4078 * gcc.dg/predict-6.c: Adjust.
4079 * gcc.dg/tree-ssa/vrp33.c: Disable EVRP.
4080 * gcc.dg/tree-ssa/vrp02.c: Likewise.
4081 * gcc.dg/tree-ssa/cunroll-9.c: Likewise.
4083 2018-10-22 Steven Bosscher <steven@gcc.gnu.org>
4084 Richard Biener <rguenther@suse.de>
4086 * bitmap.h: Update data structure documentation, including a
4087 description of bitmap views as either linked-lists or splay trees.
4088 (struct bitmap_element_def): Update comments for splay tree bitmaps.
4089 (struct bitmap_head_def): Likewise.
4090 (bitmap_list_view, bitmap_tree_view): New prototypes.
4091 (bitmap_initialize_stat): Initialize a bitmap_head's indx and
4093 (bmp_iter_set_init): Assert the iterated bitmaps are in list form.
4094 (bmp_iter_and_init, bmp_iter_and_compl_init): Likewise.
4095 * bitmap.c (bitmap_elem_to_freelist): Unregister overhead of a
4096 released bitmap element here.
4097 (bitmap_element_free): Remove.
4098 (bitmap_elt_clear_from): Work on splay tree bitmaps.
4099 (bitmap_list_link_element): Renamed from bitmap_element_link. Move
4100 this function similar ones such that linked-list bitmap implementation
4101 functions are grouped.
4102 (bitmap_list_unlink_element): Renamed from bitmap_element_unlink,
4103 and moved for grouping.
4104 (bitmap_list_insert_element_after): Renamed from
4105 bitmap_elt_insert_after, and moved for grouping.
4106 (bitmap_list_find_element): New function spliced from bitmap_find_bit.
4107 (bitmap_tree_link_left, bitmap_tree_link_right,
4108 bitmap_tree_rotate_left, bitmap_tree_rotate_right, bitmap_tree_splay,
4109 bitmap_tree_link_element, bitmap_tree_unlink_element,
4110 bitmap_tree_find_element): New functions for splay-tree bitmap
4112 (bitmap_element_link, bitmap_element_unlink, bitmap_elt_insert_after):
4113 Renamed and moved, see above entries.
4114 (bitmap_tree_listify_from): New function to convert part of a splay
4115 tree bitmap to a linked-list bitmap.
4116 (bitmap_list_view): Convert a splay tree bitmap to linked-list form.
4117 (bitmap_tree_view): Convert a linked-list bitmap to splay tree form.
4118 (bitmap_find_bit): Remove.
4119 (bitmap_clear, bitmap_clear_bit, bitmap_set_bit,
4120 bitmap_single_bit_set_p, bitmap_first_set_bit, bitmap_last_set_bit):
4121 Handle splay tree bitmaps.
4122 (bitmap_copy, bitmap_count_bits, bitmap_and, bitmap_and_into,
4123 bitmap_elt_copy, bitmap_and_compl, bitmap_and_compl_into,
4124 bitmap_compl_and_into, bitmap_elt_ior, bitmap_ior, bitmap_ior_into,
4125 bitmap_xor, bitmap_xor_into, bitmap_equal_p, bitmap_intersect_p,
4126 bitmap_intersect_compl_p, bitmap_ior_and_compl,
4127 bitmap_ior_and_compl_into, bitmap_set_range, bitmap_clear_range,
4128 bitmap_hash): Reject trying to act on splay tree bitmaps. Make
4129 corresponding changes to use linked-list specific bitmap_element
4130 manipulation functions as applicable for efficiency.
4131 (bitmap_tree_to_vec): New function.
4132 (debug_bitmap_elt_file): New function split out from ...
4133 (debug_bitmap_file): ... here. Handle splay tree bitmaps.
4134 (bitmap_print): Likewise.
4136 PR tree-optimization/63155
4137 * tree-ssa-propagate.c (ssa_prop_init): Use tree-view for the
4139 * tree-ssa-coalesce.c (coalesce_ssa_name): Populate used_in_copies
4142 2018-10-22 Martin Liska <mliska@suse.cz>
4144 PR tree-optimization/87686
4146 2018-08-29 Martin Liska <mliska@suse.cz>
4148 * tree-switch-conversion.c (switch_conversion::expand):
4149 Strenghten assumption about gswitch statements.
4151 2018-10-22 Martin Liska <mliska@suse.cz>
4153 * ipa-icf.c (sem_item::compare_attributes): Remove.
4154 (sem_item::compare_referenced_symbol_properties): Use
4155 attribute_list_equal instead.
4156 (sem_function::equals_wpa): Likewise.
4157 * ipa-icf.h: Remove compare_attributes.
4159 2018-10-22 Richard Biener <rguenther@suse.de>
4162 * mem-stats.h (mem_usage::operator==): Fix pasto.
4164 2018-10-22 Richard Biener <rguenther@suse.de>
4166 PR tree-optimization/87640
4167 * tree-vrp.c (set_value_range_with_overflow): Decompose
4169 (extract_range_from_binary_expr_1): Adjust.
4171 2018-10-22 Martin Jambor <mjambor@suse.cz>
4173 * tree-eh.h (stmt_could_throw_p): Add function parameter.
4174 (stmt_can_throw_external): Likewise.
4175 (stmt_can_throw_internal): Likewise.
4176 * tree-eh.c (lower_eh_constructs_2): Pass cfun to stmt_could_throw_p.
4177 (lower_eh_constructs_2): Likewise.
4178 (stmt_could_throw_p): Add fun parameter, use it instead of cfun.
4179 (stmt_can_throw_external): Likewise.
4180 (stmt_can_throw_internal): Likewise.
4181 (maybe_clean_eh_stmt_fn): Pass cfun to stmt_could_throw_p.
4182 (maybe_clean_or_replace_eh_stmt): Pass cfun to stmt_could_throw_p.
4183 (maybe_duplicate_eh_stmt_fn): Pass new_fun to stmt_could_throw_p.
4184 (maybe_duplicate_eh_stmt): Pass cfun to stmt_could_throw_p.
4185 (pass_lower_eh_dispatch::execute): Pass cfun to
4186 stmt_can_throw_external.
4187 (cleanup_empty_eh): Likewise.
4188 (verify_eh_edges): Pass cfun to stmt_could_throw_p.
4189 * cgraph.c (cgraph_edge::set_call_stmt): Pass a function to
4190 stmt_can_throw_external instead of pushing it to cfun.
4191 (symbol_table::create_edge): Likewise.
4192 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Pass cfun to
4193 stmt_can_throw_internal.
4194 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Pass cfun
4195 to stmt_could_throw_p.
4196 * gimple-ssa-store-merging.c (handled_load): Pass cfun to
4197 stmt_can_throw_internal.
4198 (pass_store_merging::execute): Likewise.
4199 * gimple-ssa-strength-reduction.c
4200 (find_candidates_dom_walker::before_dom_children): Pass cfun to
4202 * gimplify-me.c (gimple_regimplify_operands): Pass cfun to
4203 stmt_can_throw_internal.
4204 * ipa-pure-const.c (check_call): Pass cfun to stmt_could_throw_p and
4205 to stmt_can_throw_external.
4206 (check_stmt): Pass cfun to stmt_could_throw_p.
4207 (check_stmt): Pass cfun to stmt_can_throw_external.
4208 (pass_nothrow::execute): Likewise.
4209 * trans-mem.c (expand_call_tm): Pass cfun to stmt_can_throw_internal.
4210 * tree-cfg.c (is_ctrl_altering_stmt): Pass cfun to
4211 stmt_can_throw_internal.
4212 (verify_gimple_in_cfg): Pass cfun to stmt_could_throw_p.
4213 (stmt_can_terminate_bb_p): Pass cfun to stmt_can_throw_external.
4214 (gimple_purge_dead_eh_edges): Pass cfun to stmt_can_throw_internal.
4215 * tree-complex.c (expand_complex_libcall): Pass cfun to
4216 stmt_could_throw_p and to stmt_can_throw_internal.
4217 (expand_complex_multiplication): Pass cfun to stmt_can_throw_internal.
4218 * tree-inline.c (copy_edges_for_bb): Likewise.
4219 (maybe_move_debug_stmts_to_successors): Likewise.
4220 * tree-outof-ssa.c (ssa_is_replaceable_p): Pass cfun to
4222 * tree-parloops.c (oacc_entry_exit_ok_1): Likewise.
4223 * tree-sra.c (scan_function): Pass cfun to stmt_can_throw_external.
4224 * tree-ssa-alias.c (stmt_kills_ref_p): Pass cfun to
4225 stmt_can_throw_internal.
4226 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Likewise.
4227 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Pass cfun to
4229 (mark_aliased_reaching_defs_necessary_1): Pass cfun to
4230 stmt_can_throw_internal.
4231 * tree-ssa-forwprop.c (pass_forwprop::execute): Likewise.
4232 * tree-ssa-loop-im.c (movement_possibility): Pass cfun to
4234 * tree-ssa-loop-ivopts.c (find_givs_in_stmt_scev): Likewise.
4235 (add_autoinc_candidates): Pass cfun to stmt_can_throw_internal.
4236 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
4237 (convert_mult_to_fma_1): Likewise.
4238 (convert_to_divmod): Likewise.
4239 * tree-ssa-phiprop.c (propagate_with_phi): Likewise.
4240 * tree-ssa-pre.c (compute_avail): Pass cfun to stmt_could_throw_p.
4241 * tree-ssa-propagate.c
4242 (substitute_and_fold_dom_walker::before_dom_children): Likewise.
4243 * tree-ssa-reassoc.c (suitable_cond_bb): Likewise.
4244 (maybe_optimize_range_tests): Likewise.
4245 (linearize_expr_tree): Likewise.
4246 (reassociate_bb): Likewise.
4247 * tree-ssa-sccvn.c (copy_reference_ops_from_call): Likewise.
4248 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Likewise.
4249 * tree-ssa-strlen.c (adjust_last_stmt): Likewise.
4250 (handle_char_store): Likewise.
4251 * tree-vect-data-refs.c (vect_find_stmt_data_reference): Pass cfun to
4252 stmt_can_throw_internal.
4253 * tree-vect-patterns.c (check_bool_pattern): Pass cfun to
4255 * tree-vect-stmts.c (vect_finish_stmt_generation_1): Likewise.
4256 (vectorizable_call): Pass cfun to stmt_can_throw_internal.
4257 (vectorizable_simd_clone_call): Likewise.
4258 * value-prof.c (gimple_ic): Pass cfun to stmt_could_throw_p.
4259 (gimple_stringop_fixed_value): Likewise.
4261 2018-10-22 Ilya Leoshkevich <iii@linux.ibm.com>
4263 * config/s390/s390.c (s390_loadrelative_operand_p): Accept
4264 literal pool references.
4265 (s390_check_qrst_address): Adapt to the new behavior of
4266 s390_loadrelative_operand_p ().
4268 2018-10-22 H.J. Lu <hongjiu.lu@intel.com>
4271 * config/i386/sse.md (*andnot<mode>3_bcst): New.
4273 2018-10-22 H.J. Lu <hongjiu.lu@intel.com>
4276 * config/i386/sse.md (*<code><mode>3_bcst): New.
4278 2018-10-22 H.J. Lu <hongjiu.lu@intel.com>
4281 * config/i386/sse.md (avx512bcst): Updated for V4SI, V2DI, V8SI,
4282 V4DI, V16SI and V8DI.
4283 (*sub<mode>3<mask_name>_bcst): New.
4284 (*add<mode>3<mask_name>_bcst): Likewise.
4286 2018-10-21 Bill Schmidt <wschmidt@linux.ibm.com>
4287 Jinsong Ji <jji@us.ibm.com>
4289 * config/rs6000/emmintrin.h (_mm_movemask_pd): Replace __vector
4290 __m64 with __vector unsigned long long for compatibility.
4291 (_mm_movemask_epi8): Likewise.
4292 * config/rs6000/xmmintrin.h (_mm_cvtps_pi32): Likewise.
4293 (_mm_cvttps_pi32): Likewise.
4294 (_mm_cvtpi32_ps): Likewise.
4295 (_mm_cvtps_pi16): Likewise.
4296 (_mm_loadh_pi): Likewise.
4297 (_mm_storeh_pi): Likewise.
4298 (_mm_movehl_ps): Likewise.
4299 (_mm_movelh_ps): Likewise.
4300 (_mm_loadl_pi): Likewise.
4301 (_mm_storel_pi): Likewise.
4302 (_mm_movemask_ps): Likewise.
4303 (_mm_shuffle_pi16): Likewise.
4305 2018-10-21 H.J. Lu <hongjiu.lu@intel.com>
4308 * config/i386/avx512fintrin.h (_mm512_fnmsub_round_pd): Use
4309 __builtin_ia32_vfnmsubpd512_mask.
4310 (_mm512_mask_fnmsub_round_pd): Likewise.
4311 (_mm512_fnmsub_pd): Likewise.
4312 (_mm512_mask_fnmsub_pd): Likewise.
4313 (_mm512_maskz_fnmsub_round_pd): Use
4314 __builtin_ia32_vfnmsubpd512_maskz.
4315 (_mm512_maskz_fnmsub_pd): Likewise.
4316 (_mm512_fnmsub_round_ps): Use __builtin_ia32_vfnmsubps512_mask.
4317 (_mm512_mask_fnmsub_round_ps): Likewise.
4318 (_mm512_fnmsub_ps): Likewise.
4319 (_mm512_mask_fnmsub_ps): Likewise.
4320 (_mm512_maskz_fnmsub_round_ps): Use
4321 __builtin_ia32_vfnmsubps512_maskz.
4322 (_mm512_maskz_fnmsub_ps): Likewise.
4323 * config/i386/avx512vlintrin.h (_mm256_mask_fnmsub_pd): Use
4324 __builtin_ia32_vfnmsubpd256_mask.
4325 (_mm256_maskz_fnmsub_pd): Use __builtin_ia32_vfnmsubpd256_maskz.
4326 (_mm_mask_fnmsub_pd): Use __builtin_ia32_vfmaddpd128_mask
4327 (_mm_maskz_fnmsub_pd): Use __builtin_ia32_vfnmsubpd128_maskz.
4328 (_mm256_mask_fnmsub_ps): Use __builtin_ia32_vfnmsubps256_mask.
4329 (_mm256_mask_fnmsub_ps): Use __builtin_ia32_vfnmsubps256_mask.
4330 (_mm256_maskz_fnmsub_ps): Use __builtin_ia32_vfnmsubps256_maskz.
4331 (_mm_mask_fnmsub_ps): Use __builtin_ia32_vfnmsubps128_mask.
4332 (_mm_maskz_fnmsub_ps): Use __builtin_ia32_vfnmsubps128_maskz.
4333 * config/i386/fmaintrin.h (_mm_fnmsub_pd): Use
4334 __builtin_ia32_vfnmsubpd.
4335 (_mm256_fnmsub_pd): Use __builtin_ia32_vfnmsubpd256.
4336 (_mm_fnmsub_ps): Use __builtin_ia32_vfnmsubps.
4337 (_mm256_fnmsub_ps): Use __builtin_ia32_vfnmsubps256.
4338 (_mm_fnmsub_sd): Use __builtin_ia32_vfnmsubsd3.
4339 (_mm_fnmsub_ss): Use __builtin_ia32_vfnmsubss3.
4340 * config/i386/i386-builtin.def: Add
4341 __builtin_ia32_vfnmsubpd256_mask,
4342 __builtin_ia32_vfnmsubpd256_maskz,
4343 __builtin_ia32_vfnmsubpd128_mask,
4344 __builtin_ia32_vfnmsubpd128_maskz,
4345 __builtin_ia32_vfnmsubps256_mask,
4346 __builtin_ia32_vfnmsubps256_maskz,
4347 __builtin_ia32_vfnmsubps128_mask,
4348 __builtin_ia32_vfnmsubps128_maskz,
4349 __builtin_ia32_vfnmsubpd512_mask,
4350 __builtin_ia32_vfnmsubpd512_maskz,
4351 __builtin_ia32_vfnmsubps512_mask,
4352 __builtin_ia32_vfnmsubps512_maskz, __builtin_ia32_vfnmsubss3,
4353 __builtin_ia32_vfnmsubsd3, __builtin_ia32_vfnmsubps,
4354 __builtin_ia32_vfnmsubpd, __builtin_ia32_vfnmsubps256 and.
4355 __builtin_ia32_vfnmsubpd256.
4356 * config/i386/sse.md (fma4i_fnmsub_<mode>): New.
4357 (<avx512>_fnmsub_<mode>_maskz<round_expand_name>): Likewise.
4358 (*<sd_mask_codefor>fma_fnmsub_<mode><sd_maskz_name>_bcst_1):
4360 (*<sd_mask_codefor>fma_fnmsub_<mode><sd_maskz_name>_bcst_2):
4362 (*<sd_mask_codefor>fma_fnmsub_<mode><sd_maskz_name>_bcst_3):
4364 (fmai_vmfnmsub_<mode><round_name>): Likewise.
4366 2018-10-21 H.J. Lu <hongjiu.lu@intel.com>
4369 * config/i386/avx512fintrin.h (_mm512_fnmadd_round_pd): Use
4370 __builtin_ia32_vfnmaddpd512_mask.
4371 (_mm512_mask_fnmadd_round_pd): Likewise.
4372 (_mm512_fnmadd_pd): Likewise.
4373 (_mm512_mask_fnmadd_pd): Likewise.
4374 (_mm512_maskz_fnmadd_round_pd): Use
4375 __builtin_ia32_vfnmaddpd512_maskz.
4376 (_mm512_maskz_fnmadd_pd): Likewise.
4377 (_mm512_fnmadd_round_ps): Use __builtin_ia32_vfnmaddps512_mask.
4378 (_mm512_mask_fnmadd_round_ps): Likewise.
4379 (_mm512_fnmadd_ps): Likewise.
4380 (_mm512_mask_fnmadd_ps): Likewise.
4381 (_mm512_maskz_fnmadd_round_ps): Use
4382 __builtin_ia32_vfnmaddps512_maskz.
4383 (_mm512_maskz_fnmadd_ps): Likewise.
4384 * config/i386/avx512vlintrin.h (_mm256_mask_fnmadd_pd): Use
4385 __builtin_ia32_vfnmaddpd256_mask.
4386 (_mm256_maskz_fnmadd_pd): Use __builtin_ia32_vfnmaddpd256_maskz.
4387 (_mm_mask_fnmadd_pd): Use __builtin_ia32_vfmaddpd128_mask
4388 (_mm_maskz_fnmadd_pd): Use __builtin_ia32_vfnmaddpd128_maskz.
4389 (_mm256_mask_fnmadd_ps): Use __builtin_ia32_vfnmaddps256_mask.
4390 (_mm256_mask_fnmadd_ps): Use __builtin_ia32_vfnmaddps256_mask.
4391 (_mm256_maskz_fnmadd_ps): Use __builtin_ia32_vfnmaddps256_maskz.
4392 (_mm_mask_fnmadd_ps): Use __builtin_ia32_vfnmaddps128_mask.
4393 (_mm_maskz_fnmadd_ps): Use __builtin_ia32_vfnmaddps128_maskz.
4394 * config/i386/fmaintrin.h (_mm_fnmadd_pd): Use
4395 __builtin_ia32_vfnmaddpd.
4396 (_mm256_fnmadd_pd): Use __builtin_ia32_vfnmaddpd256.
4397 (_mm_fnmadd_ps): Use __builtin_ia32_vfnmaddps.
4398 (_mm256_fnmadd_ps): Use __builtin_ia32_vfnmaddps256.
4399 (_mm_fnmadd_sd): Use __builtin_ia32_vfnmaddsd3.
4400 (_mm_fnmadd_ss): Use __builtin_ia32_vfnmaddss3.
4401 * config/i386/i386-builtin.def: Add
4402 __builtin_ia32_vfnmaddpd256_mask,
4403 __builtin_ia32_vfnmaddpd256_maskz,
4404 __builtin_ia32_vfnmaddpd128_mask,
4405 __builtin_ia32_vfnmaddpd128_maskz,
4406 __builtin_ia32_vfnmaddps256_mask,
4407 __builtin_ia32_vfnmaddps256_maskz,
4408 __builtin_ia32_vfnmaddps128_mask,
4409 __builtin_ia32_vfnmaddps128_maskz,
4410 __builtin_ia32_vfnmaddpd512_mask,
4411 __builtin_ia32_vfnmaddpd512_maskz,
4412 __builtin_ia32_vfnmaddps512_mask,
4413 __builtin_ia32_vfnmaddps512_maskz, __builtin_ia32_vfnmaddss3,
4414 __builtin_ia32_vfnmaddsd3, __builtin_ia32_vfnmaddps,
4415 __builtin_ia32_vfnmaddpd, __builtin_ia32_vfnmaddps256 and.
4416 __builtin_ia32_vfnmaddpd256.
4417 * config/i386/sse.md (fma4i_fnmadd_<mode>): New.
4418 (<avx512>_fnmadd_<mode>_maskz<round_expand_name>): Likewise.
4419 (*<sd_mask_codefor>fma_fnmadd_<mode><sd_maskz_name>_bcst_1):
4421 (*<sd_mask_codefor>fma_fnmadd_<mode><sd_maskz_name>_bcst_2):
4423 (*<sd_mask_codefor>fma_fnmadd_<mode><sd_maskz_name>_bcst_3):
4425 (fmai_vmfnmadd_<mode><round_name>): Likewise.
4427 2018-10-21 H.J. Lu <hongjiu.lu@intel.com>
4430 * config/i386/avx512fintrin.h (_mm512_fmsub_round_pd): Use
4431 __builtin_ia32_vfmsubpd512_mask.
4432 (_mm512_mask_fmsub_round_pd): Likewise.
4433 (_mm512_fmsub_pd): Likewise.
4434 (_mm512_mask_fmsub_pd): Likewise.
4435 (_mm512_maskz_fmsub_round_pd): Use
4436 __builtin_ia32_vfmsubpd512_maskz.
4437 (_mm512_maskz_fmsub_pd): Likewise.
4438 (_mm512_fmsub_round_ps): Use __builtin_ia32_vfmsubps512_mask.
4439 (_mm512_mask_fmsub_round_ps): Likewise.
4440 (_mm512_fmsub_ps): Likewise.
4441 (_mm512_mask_fmsub_ps): Likewise.
4442 (_mm512_maskz_fmsub_round_ps): Use
4443 __builtin_ia32_vfmsubps512_maskz.
4444 (_mm512_maskz_fmsub_ps): Likewise.
4445 * config/i386/avx512vlintrin.h (_mm256_mask_fmsub_pd): Use
4446 __builtin_ia32_vfmsubpd256_mask.
4447 (_mm256_maskz_fmsub_pd): Use __builtin_ia32_vfmsubpd256_maskz.
4448 (_mm_mask_fmsub_pd): Use __builtin_ia32_vfmaddpd128_mask
4449 (_mm_maskz_fmsub_pd): Use __builtin_ia32_vfmsubpd128_maskz.
4450 (_mm256_mask_fmsub_ps): Use __builtin_ia32_vfmsubps256_mask.
4451 (_mm256_mask_fmsub_ps): Use __builtin_ia32_vfmsubps256_mask.
4452 (_mm256_maskz_fmsub_ps): Use __builtin_ia32_vfmsubps256_maskz.
4453 (_mm_mask_fmsub_ps): Use __builtin_ia32_vfmsubps128_mask.
4454 (_mm_maskz_fmsub_ps): Use __builtin_ia32_vfmsubps128_maskz.
4455 * config/i386/fmaintrin.h (_mm_fmsub_pd): Use
4456 __builtin_ia32_vfmsubpd.
4457 (_mm256_fmsub_pd): Use __builtin_ia32_vfmsubpd256.
4458 (_mm_fmsub_ps): Use __builtin_ia32_vfmsubps.
4459 (_mm256_fmsub_ps): Use __builtin_ia32_vfmsubps256.
4460 (_mm_fmsub_sd): Use __builtin_ia32_vfmsubsd3.
4461 (_mm_fmsub_ss): Use __builtin_ia32_vfmsubss3.
4462 * config/i386/i386-builtin.def: Add
4463 __builtin_ia32_vfmsubpd256_mask,
4464 __builtin_ia32_vfmsubpd256_maskz,
4465 __builtin_ia32_vfmsubpd128_mask,
4466 __builtin_ia32_vfmsubpd128_maskz,
4467 __builtin_ia32_vfmsubps256_mask,
4468 __builtin_ia32_vfmsubps256_maskz,
4469 __builtin_ia32_vfmsubps128_mask,
4470 __builtin_ia32_vfmsubps128_maskz,
4471 __builtin_ia32_vfmsubpd512_mask,
4472 __builtin_ia32_vfmsubpd512_maskz,
4473 __builtin_ia32_vfmsubps512_mask,
4474 __builtin_ia32_vfmsubps512_maskz, __builtin_ia32_vfmsubss3,
4475 __builtin_ia32_vfmsubsd3, __builtin_ia32_vfmsubps,
4476 __builtin_ia32_vfmsubpd, __builtin_ia32_vfmsubps256 and.
4477 __builtin_ia32_vfmsubpd256.
4478 * config/i386/sse.md (fma4i_fmsub_<mode>): New.
4479 (<avx512>_fmsub_<mode>_maskz<round_expand_name>): Likewise.
4480 (*<sd_mask_codefor>fma_fmsub_<mode><sd_maskz_name>_bcst_1):
4482 (*<sd_mask_codefor>fma_fmsub_<mode><sd_maskz_name>_bcst_2):
4484 (*<sd_mask_codefor>fma_fmsub_<mode><sd_maskz_name>_bcst_3):
4486 (fmai_vmfmsub_<mode><round_name>): Likewise.
4488 2018-10-21 H.J. Lu <hongjiu.lu@intel.com>
4490 * config/i386/sse.md (*<plusminus_insn><mode>3<mask_name>_bcst_1):
4491 Remove plus. Renamed to ...
4492 (*sub<mode>3<mask_name>_bcst): This.
4493 (*add<mode>3<mask_name>_bcst_2): Renamede to ...
4494 (*add<mode>3<mask_name>_bcst): This.
4496 2018-10-21 H.J. Lu <hongjiu.lu@intel.com>
4499 * config/i386/sse.md (*mul<mode>3<mask_name>_bcst): New.
4501 2018-10-21 H.J. Lu <hongjiu.lu@intel.com>
4504 * i386/avx512vlintrin.h (_mm256_or_epi32): New.
4505 (_mm_or_epi32): Likewise.
4506 (_mm256_xor_epi32): Likewise.
4507 (_mm_xor_epi32): Likewise.
4508 (_mm256_or_epi64): Likewise.
4509 (_mm_or_epi64): Likewise.
4510 (_mm256_xor_epi64): Likewise.
4511 (_mm_xor_epi64): Likewise.
4513 2018-10-20 H.J. Lu <hongjiu.lu@intel.com>
4516 * config/i386/sse.md (*<avx512>_div<mode>3<mask_name>_bcst): New.
4518 2018-10-20 Jakub Jelinek <jakub@redhat.com>
4521 * varasm.c (decode_addr_const): Handle COMPOUND_LITERAL_EXPR.
4523 2018-10-20 Andreas Schwab <schwab@linux-m68k.org>
4525 * doc/ux.texi: Move @section directly after @node.
4527 2018-10-19 Jakub Jelinek <jakub@redhat.com>
4531 * omp-low.c (check_omp_nesting_restrictions): Diagnose ordered without
4532 depend closely nested inside of loop with ordered clause with
4535 2018-10-19 David Malcolm <dmalcolm@redhat.com>
4537 * Makefile.in (TEXI_GCCINT_FILES): Add ux.texi.
4538 * doc/gccint.texi: Include ux.texi and use it in top-level menu.
4539 * doc/ux.texi: New file.
4541 2018-10-19 Segher Boessenkool <segher@kernel.crashing.org>
4543 * config/rs6000/rs6000.h (REG_ALLOC_ORDER): Move 68 (that is, CR0) to
4544 be the first CR field allocated.
4546 2018-10-19 Richard Biener <rguenther@suse.de>
4549 * config/i386/i386.c (ix86_builtin_vectorization_cost): Use
4550 TYPE_VECTOR_SUBPARTS and avoid relying on vector mode.
4552 2018-10-19 H.J. Lu <hongjiu.lu@intel.com>
4555 * config/i386/sse.md
4556 (*<plusminus_insn><mode>3<mask_name>_bcst_1): New.
4557 (*add<mode>3<mask_name>_bcst_2): Likewise.
4559 2018-10-19 H.J. Lu <hongjiu.lu@intel.com>
4561 * config/i386/sse.md
4562 (*<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name>_bcst_1):
4563 Replace nonimmediate_operand with register_operand.
4564 (*<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name>_bcst_2):
4566 (*<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name>_bcst_3):
4569 2018-10-19 Ilya Leoshkevich <iii@linux.ibm.com>
4571 PR rtl-optimization/87596
4572 * lra-constraints.c (spill_hard_reg_in_range): Use INSN_P () +
4573 lra_get_insn_recog_data () instead of lra_insn_recog_data[]
4574 for instructions in FROM..TO range.
4576 2018-10-19 Eric Botcazou <ebotcazou@adacore.com>
4578 * cfgexpand.c (expand_one_var): Use specific wording in error message
4579 for non-local frame variables.
4580 * stor-layout.c (layout_decl): Do not issue a warning for them.
4582 2018-10-19 Robin Dapp <rdapp@linux.ibm.com>
4584 * haifa-sched.c (priority): Add force_recompute parameter.
4585 (apply_replacement): Call priority () with force_recompute = true.
4586 (restore_pattern): Likewise.
4588 2018-10-18 H.J. Lu <hongjiu.lu@intel.com>
4590 * simplify-rtx.c (simplify_subreg): Limit mask of vec_merge to
4591 HOST_BITS_PER_WIDE_INT.
4592 (test_vector_ops_duplicate): Likewise.
4594 2018-10-18 H.J. Lu <hongjiu.lu@intel.com>
4597 * config/i386/sse.md (VF_AVX512): New.
4598 (avx512bcst): Likewise.
4599 (*<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name>_bcst_1):
4601 (*<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name>_bcst_2):
4603 (*<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name>_bcst_3):
4606 2018-10-18 Jonathan Wakely <jwakely@redhat.com>
4608 * doc/invoke.texi (-dumpversion): Improve grammar.
4609 (-dumpfullversion): Make more consistent with -dumpversion.
4611 2018-10-18 Uros Bizjak <ubizjak@gmail.com>
4613 * config/i386/i386.c (ix86_emit_fp_unordered_jump):
4614 Set JUMP_LABEL to the jump insn.
4615 (ix86_emit_i387_log1p): Use ix86_expand_branch to expand branch.
4616 Predict emitted jump and add label to jump insn.
4618 2018-10-18 David Malcolm <dmalcolm@redhat.com>
4620 PR tree-optimization/87562
4621 * input.c (get_substring_ranges_for_loc): Use
4622 LRK_SPELLING_LOCATION rather than LRK_MACRO_EXPANSION_POINT when
4623 getting the linemap for the endpoint. Verify that it's either
4624 in the same linemap as the start point's spelling location, or
4625 at least in the same file.
4627 2018-10-18 Richard Biener <rguenther@suse.de>
4629 * config/i386/i386.c (ix86_builtin_vectorization_cost): Do not
4630 feed width-specific load/store costs through ix86_vec_cost.
4631 * config/i386/x86-tune-costs.h (athlon_cost): Adjust.
4632 (k8_cost): Likewise.
4633 (bdver_cost): Likewise.
4634 (znver1_cost): Likewise.
4635 (btver1_cost): Likewise.
4636 (btver2_cost): Likewise.
4638 2018-10-18 H.J. Lu <hongjiu.lu@intel.com>
4640 * simplify-rtx.c (simplify_subreg): Call simplify_gen_subreg
4641 to simplify subreg of vec_merge.
4643 2018-10-18 Richard Biener <rguenther@suse.de>
4645 * config/i386/i386.c: Fix costing of vector FMA.
4647 2018-10-18 Richard Biener <rguenther@suse.de>
4649 * config/i386/i386.c (ix86_vec_cost): Remove !parallel path
4651 (ix86_builtin_vectorization_cost): For vec_construct properly
4652 cost insertion into SSE regs.
4653 (...): Adjust calls to ix86_vec_cost.
4655 2018-10-18 Richard Biener <rguenther@suse.de>
4659 2018-02-07 Richard Biener <rguenther@suse.de>
4661 PR tree-optimization/84204
4662 * tree-chrec.c (chrec_fold_plus_1): Remove size limiting in
4665 2018-10-18 H.J. Lu <hongjiu.lu@intel.com>
4668 * simplify-rtx.c (simplify_subreg): Simplify subreg of vec_merge
4670 (test_vector_ops_duplicate): Add test for a scalar subreg of a
4671 VEC_MERGE of a VEC_DUPLICATE.
4673 2018-10-17 Joseph Myers <joseph@codesourcery.com>
4675 * doc/cpp.texi (__STDC_VERSION__): Document C2X handling.
4676 * doc/invoke.texi (-std=c2x, -std=gnu2x): Document new options.
4677 * doc/standards.texi (C Language): Document C2X.
4678 * dwarf2out.c (highest_c_language), config/rl78/rl78.c
4679 (rl78_option_override): Handle "GNU C2X" language name.
4681 2018-10-17 Joseph Myers <joseph@codesourcery.com>
4683 * doc/invoke.texi (-std=c17), doc/standards.texi (C Language):
4684 Document C17 as published in 2018.
4686 2018-10-17 Eric Botcazou <ebotcazou@adacore.com>
4689 * fold-const.c (fold_truth_andor_1): If the right side is not constant,
4690 bail out if both sides do not have the same storage order.
4692 2018-10-17 Aldy Hernandez <aldyh@redhat.com>
4694 * bitmap.c (bitmap_head::dump): New.
4695 * bitmap.h (bitmap_head): Add dump().
4696 * gimple-ssa-evrp-analyze.c
4697 (evrp_range_analyzer::try_find_new_range): Adjust for value_range API.
4698 (evrp_range_analyzer::set_ssa_range_info): Same.
4699 (evrp_range_analyzer::record_ranges_from_phis): Same.
4700 (evrp_range_analyzer::record_ranges_from_stmt): Same.
4701 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Same.
4702 * gimple-ssa-sprintf.c (get_int_range): Same.
4703 (format_integer): Same.
4704 (sprintf_dom_walker::handle_gimple_call): Same.
4705 * ipa-cp.c (ipcp_vr_lattice::meet_with_1): Same.
4706 (ipcp_vr_lattice::top_p): Same.
4707 (ipcp_vr_lattice::bottom_p): Same.
4708 (ipcp_vr_lattice::set_to_bottom): Same.
4709 (ipa_vr_operation_and_type_effects): Same.
4710 (propagate_vr_across_jump_function): Same.
4711 (ipcp_store_vr_results): Same.
4712 * ipa-prop.c (struct ipa_vr_ggc_hash_traits): Same.
4713 (ipa_print_node_jump_functions_for_edge): Same.
4714 (ipa_get_value_range): Same.
4715 (ipa_compute_jump_functions_for_edge): Same.
4716 (ipa_write_jump_function): Same.
4717 * tree-ssa-dom.c (simplify_stmt_for_jump_threading): Same.
4718 * tree-ssa-threadedge.c (record_temporary_equivalences_from_phis):
4720 * vr-values.c (set_value_range_to_nonnegative): Same.
4721 (set_value_range_to_truthvalue): Same.
4722 (vr_values::get_value_range): Same.
4723 (vr_values::set_defs_to_varying): Same.
4724 (vr_values::update_value_range): Same.
4725 (symbolic_range_based_on_p): Same.
4726 (vr_values::op_with_boolean_value_range_p): Same.
4727 (vr_values::extract_range_for_var_from_comparison_expr): Same.
4728 (vr_values::extract_range_from_ssa_name): Same.
4729 (vr_values::extract_range_from_binary_expr): Same.
4730 (vr_values::extract_range_from_unary_expr): Same.
4731 (vr_values::extract_range_from_cond_expr): Same.
4732 (vr_values::extract_range_from_comparison): Same.
4733 (vr_values::check_for_binary_op_overflow): Same.
4734 (vr_values::extract_range_basic): Same.
4735 (vr_values::extract_range_from_assignment): Same.
4736 (compare_ranges): Same.
4737 (compare_range_with_value): Same.
4738 (vr_values::adjust_range_with_scev): Same.
4739 (vrp_valueize): Same.
4740 (vrp_valueize_1): Same.
4741 (vr_values::get_vr_for_comparison): Same.
4742 (vr_values::compare_name_with_value): Same.
4743 (vr_values::compare_names): Same.
4744 (vr_values::vrp_evaluate_conditional): Same.
4745 (find_case_label_ranges): Same.
4746 (vr_values::vrp_visit_switch_stmt): Same.
4747 (vr_values::extract_range_from_phi_node): Same.
4748 (vr_values::simplify_div_or_mod_using_ranges): Same.
4749 (vr_values::simplify_bit_ops_using_ranges): Same.
4750 (test_for_singularity): Same.
4751 (range_fits_type_p): Same.
4752 (vr_values::simplify_cond_using_ranges_1): Same.
4753 (vr_values::simplify_switch_using_ranges): Same.
4754 (vr_values::simplify_float_conversion_using_ranges): Same.
4755 (vr_values::two_valued_val_range_p): Same.
4756 (vr_values::add_equivalence): Move to value_range::equiv_add.
4757 * vr-values.h (vr_values::add_equivalence): Remove.
4758 (VR_INITIALIZER): Remove.
4759 * tree-vrp.c (value_range::set): New.
4760 (value_range::equiv_add): New.
4761 (value_range::value_range): New.
4762 (value_range::deep_copy): New.
4763 (value_range::check): New.
4764 (value_range::equal_p): New.
4765 (value_range::ignore_equivs_equal_p): New.
4766 (value_range::operator==): New.
4767 (value_range::operator!=): New.
4768 (value_range::symbolic_p): New.
4769 (value_range::numeric_p): New.
4770 (value_range::set_undefined): New.
4771 (value_range::set_varying): New.
4772 (value_range::may_contain_p): New.
4773 (value_range::equiv_clear): New.
4774 (value_range::singleton_p): New.
4775 (value_range::intersect): New.
4776 (value_range::dump): New.
4777 (value_range::set_and_canonicalize): New.
4778 (set_value_range): Adjust for value_range API.
4779 (set_value_range_to_undefined): Same.
4780 (set_value_range_to_varying): Same.
4781 (set_and_canonicalize_value_range): Same.
4782 (set_value_range_to_nonnull): Same.
4783 (set_value_range_to_null): Same.
4784 (range_is_null): Same.
4785 (range_is_nonnull): Same.
4786 (range_int_cst_p): Same.
4787 (range_int_cst_singleton_p): Same.
4788 (symbolic_range_p): Same.
4789 (range_includes_zero_p): Same.
4790 (value_range_constant_singleton): Same.
4791 (vrp_set_zero_nonzero_bits): Same.
4792 (ranges_from_anti_range): Same.
4793 (extract_range_into_wide_ints): Same.
4794 (extract_range_from_multiplicative_op): Same.
4795 (set_value_range_with_overflow): Same.
4796 (extract_range_from_binary_expr_1): Same.
4797 (extract_range_from_unary_expr): Same.
4798 (dump_value_range): Same.
4799 (debug_value_range): Same.
4800 (vrp_prop::check_array_ref): Same.
4801 (vrp_prop::check_mem_ref): Same.
4802 (vrp_prop::vrp_initialize): Same.
4803 (vrp_prop::visit_stmt): Same.
4804 (intersect_ranges): Same.
4805 (vrp_prop::visit_phi): Same.
4806 (vrp_prop::vrp_finalize): Same.
4807 (determine_value_range_1): Same.
4808 (determine_value_range): Same.
4809 (vrp_intersect_ranges_1): Rename to...
4810 (vrp_intersect_1): this.
4811 (vrp_intersect_ranges): Rename to...
4812 (value_range::intersect_helper): ...this.
4813 (vrp_meet_1): Rename to...
4814 (value_range::union_helper): ...this.
4815 (vrp_meet): Rename to...
4816 (value_range::union_): ...this.
4817 (copy_value_range): Remove.
4818 * tree-vrp.h (struct value_range): Rewrite into a proper class.
4819 (value_range::vrtype): New.
4820 (value_range::type): New.
4821 (value_range::equiv): New.
4822 (value_range::min): New.
4823 (value_range::max): New.
4824 (value_range::varying_p): New.
4825 (value_range::undefined_p): New.
4826 (value_range::null_p): New.
4827 (value_range::equiv_add): New.
4828 (copy_value_range): Remove.
4830 2018-10-17 David Malcolm <dmalcolm@redhat.com>
4832 * Makefile.in (SELFTEST_TARGETS): New.
4833 (selftest) Change from s-selftest-c to $(SELFTEST_TARGETS).
4834 (C_SELFTEST_FLAGS, C_SELFTEST_DEPS, s-selftest-c, selftest-c-gdb)
4835 (selftest-gdb, selftest-c-valgrind, selftest-valgrind): Move to
4837 (CPP_SELFTEST_FLAGS, CPP_SELFTEST_DEPS, s-selftest-c++)
4838 (selftest-c++-gdb, selftest-c++-valgrind): Move to
4840 * configure: Regenerate.
4841 * configure.ac (selftest_languages): New.
4843 2018-10-17 Aldy Hernandez <aldyh@redhat.com>
4845 * tree-vrp.c (extract_range_from_multiplicative_op): Remove
4846 overflow wraps argument.
4847 (extract_range_from_binary_expr_1): Do not pass overflow wraps to
4848 wide_int_range_multiplicative_op.
4849 * wide-int-range.cc (wide_int_range_mult_wrapping): Remove
4850 overflow wraps argument.
4851 (wide_int_range_multiplicative_op): Same.
4852 (wide_int_range_lshift): Same.
4853 (wide_int_range_div): Same.
4854 * wide-int-range.h (wide_int_range_multiplicative_op): Same.
4855 (wide_int_range_lshift): Same.
4856 (wide_int_range_div): Same.
4858 2018-10-17 Aldy Hernandez <aldyh@redhat.com>
4860 * wide-int-range.h (wide_int_range_shift_undefined_p): Adjust to
4861 use sign as argument.
4862 * tree-vrp.c (extract_range_from_binary_expr_1): Pass sign to
4863 wide_int_range_shift_undefined_p.
4865 2018-10-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4867 * config/aarch64/aarch64.md (despeculate_copy<ALLI_TI:mode>):
4869 (@despeculate_copy<ALLI_TI:mode>): ... This.
4870 * config/aarch64/aarch64.c (aarch64_speculation_safe_value): Remove
4873 2018-10-16 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4875 * config.gcc: Obsolete *-*-solaris2.10*.
4876 * doc/install.texi (Specific, *-*-solaris2*): Document it.
4878 2018-10-12 Jeff Law <law@redhat.com>
4880 * config/ft32/ft32.md (ft32_general_movsrc_operand): Disable
4881 reg + sym +- const_int addressing modes.
4883 2018-10-15 David Malcolm <dmalcolm@redhat.com>
4885 * common.opt (fdiagnostics-minimum-margin-width=): New option.
4886 * diagnostic-show-locus.c (layout::layout): Apply the minimum
4888 (layout::start_annotation_line): Only print up to 3 of the
4889 margin character, to avoid touching the left-hand side.
4890 (selftest::test_diagnostic_show_locus_fixit_lines): Update for
4891 minimum margin width, as set by test_diagnostic_context's ctor.
4892 (selftest::test_fixit_insert_containing_newline): Likewise.
4893 (selftest::test_fixit_insert_containing_newline_2): Likewise.
4894 (selftest::test_line_numbers_multiline_range): Clear
4895 dc.min_margin_width.
4896 * diagnostic.c (diagnostic_initialize): Initialize
4898 * diagnostic.h (struct diagnostic_context): Add field
4900 * doc/invoke.texi: Add -fdiagnostics-minimum-margin-width=.
4901 * opts.c (common_handle_option): Handle
4902 OPT_fdiagnostics_minimum_margin_width_.
4903 * selftest-diagnostic.c
4904 (selftest::test_diagnostic_context::test_diagnostic_context):
4905 Initialize min_margin_width to 6.
4906 * toplev.c (general_init): Initialize global_dc->min_margin_width.
4908 2018-10-15 David Malcolm <dmalcolm@redhat.com>
4910 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
4911 Fix usage of "error_at_rich_loc" in the comment.
4913 2018-10-15 Renlin Li <renlin.li@arm.com>
4916 * tree-vectorizer.c (try_vectorize_loop_1): Don't use
4917 if-conversioned loop when it contains ifn with types not
4918 supported by backend.
4919 * internal-fn.c (expand_direct_optab_fn): Add an assert.
4920 (direct_internal_fn_supported_p): New helper function.
4921 * internal-fn.h (direct_internal_fn_supported_p): Declare.
4923 2018-10-15 Jakub Jelinek <jakub@redhat.com>
4926 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512F_UNSET):
4927 Remove OPTION_MASK_ISA_AVX5124FMAPS_UNSET and
4928 OPTION_MASK_ISA_AVX5124VNNIW_UNSET.
4930 2018-10-15 Bin Cheng <bin.cheng@linux.alibaba.com>
4932 PR tree-optimization/87022
4933 * tree-loop-distribution.c (pg_add_dependence_edges): Check all
4934 bits in dist vector rather than the first one.
4936 2018-10-15 Richard Biener <rguenther@suse.de>
4939 * tree-ssa-structalias.c (struct vls_data): Add escaped_p member.
4940 (visit_loadstore): When a used restrict tag escaped verify that
4941 the points-to solution of "other" pointers do not include
4943 (compute_dependence_clique): If a used restrict tag escaped
4944 communicated that down to visit_loadstore.
4946 2018-10-15 Andreas Krebbel <krebbel@linux.ibm.com>
4948 * config/s390/s390.c (s390_expand_vec_init): Force vector element
4949 into reg if it isn't a general operand.
4951 2018-10-14 H.J. Lu <hongjiu.lu@intel.com>
4954 * config/i386/sse.md (*vec_dupv2di): Add register source to
4957 2018-10-14 H.J. Lu <hongjiu.lu@intel.com>
4960 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512F_UNSET):
4961 Add OPTION_MASK_ISA_AVX512IFMA_UNSET,
4962 OPTION_MASK_ISA_AVX5124FMAPS_UNSET and
4963 OPTION_MASK_ISA_AVX5124VNNIW_UNSET.
4965 2018-10-13 Eric Botcazou <ebotcazou@adacore.com>
4967 * dwarf2cfi.c (struct dw_trace_info): Add args_size_defined_for_eh.
4968 (notice_args_size): Set it in the current trace if no insn that can
4969 throw internally has been seen yet.
4970 (connect_traces): When connecting args_size between traces, allow the
4971 incoming values not to match if there is an insn setting it before the
4972 first insn that can throw internally; in that case, force the creation
4973 of a CFI note on this latter insn.
4975 2018-10-13 Jonathan Wakely <jwakely@redhat.com>
4977 * opt-problem.h (opt_wrapper): Use template-argument-list when naming
4978 the base class, because using the injected-class-name was not clearly
4979 specified until DR 176.
4981 2018-10-12 Paul Koning <ni1d@arrl.net>
4983 * config/pdp11/pdp11.md (doloop_end): New expander.
4984 (doloop_end_insn): renamed from "doloop_end".
4985 (addqi3): New pattern.
4986 (subqi3): New pattern.
4987 * config/pdp11/predicates.md (incdec_operand): New predicate.
4989 2018-10-12 Yury Gribov <tetra2005@gmail.com>
4992 * real.c (format_helper::can_represent_integral_type_p): New function
4993 * real.h (format_helper::can_represent_integral_type_p): Ditto.
4994 * match.pd: New pattern.
4996 2018-10-12 Alexandre Oliva <oliva@adacore.com>
4998 * configure.ac: Introduce --enable-large-address-aware
4999 to define MINGW_DEFAULT_LARGE_ADDR_AWARE.
5000 * doc/install.texi: Document it.
5001 * configure, config.in: Rebuilt.
5002 * config/i386/mingw32.h (LINK_SPEC_LARGE_ADDR_AWARE): Define,
5003 based on MINGW_DEFAULT_LARGE_ADDR_AWARE.
5004 (LINK_SPEC): Insert it.
5005 * config/i386/mingw-w64.h: Likewise.
5007 * cygming.h (HAVE_GAS_ALIGNED_COMM): Fallback-define.
5009 2018-10-12 Peter Bergner <bergner@linux.ibm.com>
5011 PR rtl-optimization/87600
5012 * ira-lives (non_conflicting_reg_copy_p): Disable for non LRA targets.
5014 2018-10-12 Paul Koning <ni1d@arrl.net>
5016 * doc/md.texi (doloop_end): Document that the pattern code may
5017 need to check operand mode.
5019 2018-10-12 Wilco Dijkstra <wdijkstr@arm.com>
5021 * config/aarch64/aarch64.md (zero_extendsidi2_aarch64): Add alternatives
5022 to zero-extend between int and floating-point registers.
5023 (load_pair_zero_extendsidi2_aarch64): Add alternative for zero-extended
5024 ldp into floating-point registers. Add type and arch attributes.
5025 (zero_extend<SHORT:mode><GPI:mode>2_aarch64): Add arch attribute.
5026 Use f_loads for type attribute.
5028 2018-10-11 Martin Sebor <msebor@redhat.com>
5030 * doc/extend.texi (attribute packed): Correct typos.
5032 2018-10-11 Martin Sebor <msebor@redhat.com>
5034 * doc/extend.texi (attribute flatten): Mention interaction with
5037 2018-10-11 Jan Hubicka <hubicka@ucw.cz>
5040 * cgraphclones.c (cgraph_node::create_version_clone_with_body):
5041 Set new_decl virtual flag to zero.
5043 2018-10-11 Martin Sebor <msebor@redhat.com>
5046 * doc/extend.texi (attribute format_arg): Discuss using multiple
5047 attributes on a single function.
5049 2018-10-11 Giuliano Belinassi <giuliano.belinassi@usp.br>
5051 PR tree-optimization/86829
5052 * match.pd (sin (atan (x))): New simplification rules.
5053 (cos (atan (x))): Likewise.
5054 * real.c (build_sinatan_real): New function.
5055 * real.h (build_sinatan_real): Prototype.
5057 2018-10-11 Will Schmidt <will_schmidt@vnet.ibm.com>
5059 * config/rs6000/rs6000.c (map_to_integral_tree_type): New helper
5061 (fold_mergeeo_helper): New helper function.
5062 (rs6000_gimple_fold_builtin): Add hooks for vec_mergee and vec_mergeo
5063 intrinsics. Correct some whitespace indentation issues.
5065 2018-10-11 Wilco Dijkstra <wdijkstr@arm.com>
5068 * config/aarch64/aarch64.c (aarch64_mask_and_shift_for_ubfiz_p):
5069 Use HOST_WIDE_INT_1U for shift.
5071 2018-10-11 Doug Rupp <rupp@adacore.com>
5072 Olivier Hainque <hainque@adacore.com>
5074 * config/rs6000/vxworks.h (VXWORKS_RELAX_LINK_SPEC): New macro.
5075 Pass --relax to the linker for RTPs.
5076 (LINK_SPEC): Append VXWORKS_RELAX_LINK_SPEC.
5078 2018-10-11 Andrew Stubbs <ams@codesourcery.com>
5079 Jan Hubicka <jh@suse.cz>
5080 Martin Jambor <mjambor@suse.cz>
5082 * print-rtl.c (print_rtx_operand_codes_E_and_V): Print how many times
5083 the same elements are repeated rather than printing all of them.
5084 * read-rtl.c (rtx_reader::read_rtx_operand): Recognize and expand
5085 "repeated" elements.
5086 * read-rtl-function.c (test_loading_repeat): New function.
5087 (read_rtl_function_c_tests): Call test_loading_repeat.
5088 * rtl-tests.c (test_dumping_repeat): New function.
5089 (rtl_tests_c_tests): Call test_dumping_repeat.
5091 2018-10-11 Richard Biener <rguenther@suse.de>
5093 * config/i386/x86-tune-costs.h (bdver?_memcpy, bdver?_memset,
5094 bdver?_cost): Unify to ...
5095 (bdver_memcpy, bdver_memset, bdver_cost): ... this.
5096 * config/i386/i386.c (processor_cost_table): Adjust.
5098 2018-10-10 Eric Botcazou <ebotcazou@adacore.com>
5101 * cgraphunit.c (cgraph_node::expand_thunk): Force DECL_IGNORED_P on
5102 the thunk when expanding to GIMPLE.
5104 2018-10-10 Bernd Edlinger <bernd.edlinger@hotmail.de>
5106 * varasm.c (mergeable_string_section): Don't try to move zero-length
5107 strings to the merge section.
5109 2018-10-10 Uros Bizjak <ubizjak@gmail.com>
5112 * config/i386/mmx.md (const_vector 0 -> mem splitter): New splitter.
5114 2018-10-10 Jakub Jelinek <jakub@redhat.com>
5117 * config/i386/i386-builtin.def (IX86_BUILTIN_RDPMC): Move from args set
5118 to special_args set.
5120 2018-10-10 Richard Biener <rguenther@suse.de>
5122 * config/i386/sse.md (reduc_plus_scal_v8df, reduc_plus_scal_v4df,
5123 reduc_plus_scal_v2df, reduc_plus_scal_v16sf, reduc_plus_scal_v8sf,
5124 reduc_plus_scal_v4sf): Merge into pattern reducing to half width
5125 and recursing and pattern terminating the recursion on SSE
5126 vector width using ix86_expand_reduc.
5127 (reduc_sminmax_scal_<mode>): Split into part reducing to half
5128 width and recursing and SSE2 vector variant doing the final
5129 reduction with ix86_expand_reduc.
5130 (reduc_uminmax_scal_<mode>): Likewise for the AVX512 variants
5131 with terminating the recursion at AVX level, splitting that
5134 2018-10-09 David Malcolm <dmalcolm@redhat.com>
5136 * genmatch.c (error_cb): Rename to...
5137 (diagnostic_cb): ...this, converting int params to enums.
5138 (fatal_at): Update for renaming.
5139 (warning_at): Likewise.
5141 * input.c (selftest::ebcdic_execution_charset::apply):
5142 Update for renaming of...
5143 (selftest::ebcdic_execution_charset::on_error): ...this, renaming
5145 (selftest::ebcdic_execution_charset::on_diagnostic): ...this,
5146 converting level and reason to enums.
5147 (class selftest::lexer_error_sink): Rename to...
5148 (class selftest::lexer_test_options): ...this, renaming field
5149 "m_errors" to "m_diagnostics".
5150 (selftest::lexer_test_options::apply): Update for renaming of...
5151 (selftest::lexer_test_options::on_error): ...this, renaming to...
5152 (selftest::lexer_test_options::on_diagnostic): ...this
5153 converting level and reason to enums.
5154 (selftest::test_lexer_string_locations_raw_string_unterminated):
5155 Update for renamings.
5156 * opth-gen.awk (struct cpp_reason_option_codes_t): Use enum for
5159 2018-10-09 Paul A. Clarke <pc@us.ibm.com>
5161 * config.gcc (powerpc*-*-*): Add pmmintrin.h to extra_headers.
5162 * config/rs6000/pmmintrin.h: New file.
5164 2018-10-09 Eric Botcazou <ebotcazou@adacore.com>
5166 PR tree-optimization/86659
5167 * gimple-match.h (gimple_match_op constructors): Initialize reverse.
5169 2018-10-09 Richard Biener <rguenther@suse.de>
5171 PR tree-optimization/63155
5172 * tree-ssa-structalias.c: Include tree-ssa.h.
5173 (get_constraint_for_ssa_var): For undefs return nothing_id.
5174 (find_func_aliases): Cleanup PHI handling.
5176 2018-10-09 Richard Biener <rguenther@suse.de>
5178 * tree-vectorizer.c (dump_stmt_cost): Fix cut&paste missing
5181 2018-10-09 Martin Liska <mliska@suse.cz>
5183 * asan.c (asan_emit_stack_protection): If a stack variable
5184 is located in a same file as current function, then emit
5185 line info into variable definition string.
5187 2018-10-08 Eric Botcazou <ebotcazou@adacore.com>
5189 * print-rtl.c (rtx_writer::print_rtx_operand_code_i): Print column
5192 2018-10-08 Eric Botcazou <ebotcazou@adacore.com>
5194 * cgraphunit.c (cgraph_node::expand_thunk): Do not force DECL_IGNORED_P
5197 2018-10-08 Eric Botcazou <ebotcazou@adacore.com>
5199 PR tree-optimization/86659
5200 * gimple-match.h (struct gimple_match_op): Add reverse field.
5201 (gimple_match_op::set_op): New overloaded method.
5202 * gimple-match-head.c (maybe_build_generic_op) <BIT_FIELD_REF>: Set
5203 the REF_REVERSE_STORAGE_ORDER flag on the value.
5204 (gimple_simplify) <GIMPLE_ASSIGN>: For BIT_FIELD_REF, propagate the
5205 REF_REVERSE_STORAGE_ORDER flag and avoid simplifying if it is set.
5207 2018-10-08 Richard Sandiford <richard.sandiford@arm.com>
5210 * gimple-ssa-backprop.c (backprop::intersect_uses): Use
5211 FOR_EACH_IMM_USE_FAST instead of FOR_EACH_IMM_USE_STMT.
5213 2018-10-08 H.J. Lu <hongjiu.lu@intel.com>
5216 * config/i386/avx512fintrin.h (_mm512_mask_fmaddsub_round_pd):
5217 Defined with __builtin_ia32_vfmaddsubpd512_mask.
5219 2018-10-08 Richard Biener <rguenther@suse.de>
5221 * config/i386/x86-tune-costs.h (znver1_cost): Make AVX256 vector loads
5222 cost the same as AVX128 ones.
5224 2018-10-08 Paul Koning <ni1d@arrl.net>
5226 * config/pdp11/pdp11-protos.h (output_block_move): Remove.
5227 (expand_block_move): New function.
5228 * config/pdp11/pdp11.c (output_block_move): Remove.
5229 (expand_block_move): New function.
5230 * config/pdp11/pdp11.h (MOVE_RATIO): New definition.
5231 * config/pdp11/pdp11.md (movmemhi): Use expand_block_move.
5232 (*movmemhi1): Remove.
5234 2018-10-08 Robin Dapp <rdapp@linux.ibm.com>
5236 * config/s390/2827.md: Increase latencies for some FP instructions.
5238 2018-10-08 Richard Biener <rguenther@suse.de>
5240 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost):
5242 * tree-vectorizer.c (dump_stmt_cost): Add cost param and dump it.
5243 * tree-vectorizer.h (dump_stmt_cost): Adjust.
5244 (add_stmt_cost): Dump return value of the hook.
5246 2018-10-08 Richard Biener <rguenther@suse.de>
5248 PR tree-optimization/63155
5249 * tree-ssa-propagate.c (add_ssa_edge): Do cheap check first.
5250 (ssa_propagation_engine::ssa_propagate): Remove redundant
5251 bitmap bit clearing.
5253 2018-10-05 Peter Bergner <bergner@linux.ibm.com>
5255 PR rtl-optimization/86939
5256 PR rtl-optimization/87479
5257 * ira.h (non_conflicting_reg_copy_p): New prototype.
5258 * ira-lives.c (ignore_reg_for_conflicts): New static variable.
5259 (make_hard_regno_dead): Don't add conflicts for register
5260 ignore_reg_for_conflicts.
5261 (make_object_dead): Likewise.
5262 (non_conflicting_reg_copy_p): New function.
5263 (process_bb_node_lives): Set ignore_reg_for_conflicts for copies.
5264 Remove special conflict handling of REAL_PIC_OFFSET_TABLE_REGNUM.
5265 * lra-lives.c (ignore_reg_for_conflicts): New static variable.
5266 (make_hard_regno_dead): Don't add conflicts for register
5267 ignore_reg_for_conflicts. Remove special conflict handling of
5268 REAL_PIC_OFFSET_TABLE_REGNUM. Remove now unused argument
5269 check_pic_pseudo_p and update callers.
5270 (mark_pseudo_dead): Don't add conflicts for register
5271 ignore_reg_for_conflicts.
5272 (process_bb_lives): Set ignore_reg_for_conflicts for copies.
5274 2018-10-05 Andrew Waterman <andrew@sifive.com>
5275 Jim Wilson <jimw@sifive.com>
5277 * config/riscv/riscv.md (f<quiet_pattern>_quiet<ANYF:mode><X:mode>4):
5278 Add define_expand. Add ! HONOR_SNANS check to current pattern. Add
5279 new pattern using HONOR_SNANS that emits one extra instruction.
5281 2018-10-05 Segher Boessenkool <segher@kernel.crashing.org>
5283 * config/rs6000/rs6000.md (unnamed mfcr scc_comparison_operator
5284 patterns): Merge SI and DI patterns to a GPR pattern.
5285 (unnamed define_insn and define_split for record form of that): Merge
5286 to a single define_insn_and_split pattern.
5288 2018-10-05 David Malcolm <dmalcolm@redhat.com>
5291 * input.c (expand_location_to_spelling_point): Add param "aspect"
5292 and use rather than hardcoding LOCATION_ASPECT_CARET.
5293 (get_substring_ranges_for_loc): Handle the case of a single token
5294 within a macro expansion.
5295 * input.h (expand_location_to_spelling_point): Add "aspect" param,
5296 defaulting to LOCATION_ASPECT_CARET.
5298 2018-10-05 Paul Koning <ni1d@arrl.net>
5300 * config/pdp11/pdp11.c (TARGET_CXX_GUARD_TYPE): Define.
5301 (TARGET_CXX_CLASS_DATA_ALWAYS_COMDAT): Define.
5302 (TARGET_CXX_LIBRARY_RTTI_COMDAT): Define.
5303 (pdp11_guard_type): New function.
5305 2018-10-05 Paul Koning <ni1d@arrl.net>
5307 * config/pdp11/pdp11.h (FLOAT_TYPE_SIZE): Always 32.
5308 * config/pdp11/pdp11.opt (mfloat32): Remove.
5310 * doc/invoke.texi (pdp11 -mfloat32): Remove:
5311 (pdp11 -mfloat64): Remove.
5313 2018-10-05 Uros Bizjak <ubizjak@gmail.com>
5315 * config/i386/i386.md (*cmpxf_cc_i387): Remove pattern.
5316 (*cmp<mode>_cc_i387): Ditto.
5317 (*cmpu<mode>_cc_i387): Ditto.
5318 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
5319 * config/i386/i386.c (ix86_expand_fp_compare): Remove
5321 <case IX86_FPCMP_SAHF>: Do not generate pattern with HImode clobber.
5322 Emit x86_sahf_1 pattern.
5323 (ix86_expand_compare): Update call to ix86_expand_fp_compare.
5324 (ix86_expand_carry_flag_compare): Ditto.
5326 2018-10-05 Uros Bizjak <ubizjak@gmail.com>
5328 * config/i386/i386.md (*cmpxf_i387): Change operand 2 predicate
5329 to reg_or_0_operand. Add "C" constraint.
5330 (*cmpxf_cc_i387): Ditto.
5331 (*cmp<mode>_i387): Change operand 2 predicate
5332 to nonimm_or_0_operand. Add "C" constraint.
5333 (*cmp<mode>_cc_i387): Ditto.
5334 (*cmp<mode>_0_i387): Remove insn pattern.
5335 (*cmp<mode>_0_cc_i387): Ditto.
5337 2018-10-05 Uros Bizjak <ubizjak@gmail.com>
5339 * config/i386/constraints.md ("C"): Do not depend on TARGET_SSE.
5340 * config/i386/predicates.md (nonimm_or_0_operand): Rename
5341 from vector_move_operand. Update all uses.
5343 2018-10-05 Martin Sebor <msebor@redhat.com>
5345 PR tree-optimization/87490
5346 * builtins.c (expand_builtin_strnlen): Handle a null data.decl
5349 2018-10-05 Richard Biener <rguenther@suse.de>
5351 PR tree-optimization/63155
5352 * tree-ssa-ccp.c (ccp_propagate::visit_phi): Avoid excess
5353 vertical space in dumpfiles.
5354 * tree-ssa-propagate.h
5355 (ssa_propagation_engine::process_ssa_edge_worklist): Remove.
5356 * tree-ssa-propagate.c (cfg_blocks_back): New global.
5357 (ssa_edge_worklist_back): Likewise.
5358 (curr_order): Likewise.
5359 (cfg_blocks_get): Remove abstraction.
5360 (cfg_blocks_add): Likewise.
5361 (cfg_blocks_empty_p): Likewise.
5362 (add_ssa_edge): Add to current or next worklist based on
5364 (add_control_edge): Likewise.
5365 (ssa_propagation_engine::process_ssa_edge_worklist): Fold
5367 (ssa_propagation_engine::ssa_propagate): ... here. Unify
5368 iteration from CFG and SSA edge worklist so we process
5369 everything in RPO order, prioritizing forward progress
5371 (ssa_prop_init): Allocate new worklists, do not dump
5373 (ssa_prop_fini): Free new worklists.
5375 2018-10-05 Richard Biener <rguenther@suse.de>
5377 * tree-core.h (tree_block::abstract_flag): Remove.
5378 (tree_block::block_num): Make full 32bits.
5379 * tree.def (BLOCK): Remove docs about BLOCK_ABSTRACT.
5380 * tree.h (BLOCK_ABSTRACT): Remove.
5381 * dwarf2out.c (gen_lexical_block_die): Remove dead code
5382 resulting from BLOCK_ABSTRACT being always false.
5383 (gen_inlined_subroutine_die): Likewise.
5384 (gen_block_die): Likewise.
5385 * tree.c (block_ultimate_origin): Likewise.
5386 * tree-pretty-print.c (dump_block_node): Remove code dealing
5387 with BLOCK_ABSTRACT.
5388 * tree-ssa-live.c (dump_scope_block): Likewise.
5389 * tree-streamer-in.c (unpack_ts_block_value_fields): Likewise.
5390 * tree-streamer-out.c (pack_ts_block_value_fields): Likewise.
5392 2018-10-05 Richard Biener <rguenther@suse.de>
5394 * config/i386/i386.c (ix86_add_stmt_cost): When scalar cost
5395 is asked for initialize mode to the component mode of the
5398 2018-10-05 H.J. Lu <hongjiu.lu@intel.com>
5401 * config/i386/gnu-user.h (ASM_SPEC): Don't pass -msse2avx to
5402 assembler for -mavx.
5403 * config/i386/gnu-user64.h (ASM_SPEC): Likewise.
5405 2018-10-05 Segher Boessenkool <segher@kernel.crashing.org>
5408 * config/rs6000/rs6000-builtin.def (RS6000_BUILTIN_SET_FPSCR_DRN): Use
5410 * config/rs6000/rs6000.md (rs6000_set_fpscr_rn): Require the operand
5411 to be DImode. When using mffscrn, force the operand to a register.
5413 2018-10-04 Uros Bizjak <ubizjak@gmail.com>
5415 * config/i386/i386.md (*fop_<X87MODEF:mode>_2_i387): Macroize insn
5416 from *fop_<MODEF:mode>_2_i387 and *fop_xf_2_i387 using
5417 X87MODEF mode iterator.
5418 (*fop_<X87MODEF:mode>_3_i387): Macroize insn from
5419 *fop_<MODEF:mode>_3_i387 and *fop_xf_3_i387 using
5420 X87MODEF mode iterator.
5422 2018-10-04 Vinay Kumar <vinay.kumar@blackfigtech.com>
5424 * doc/invoke.texi (-Wno-prio-ctor-dtor): Document new warning
5425 -Wno-prio-ctor-dtor.
5427 2018-10-04 David Malcolm <dmalcolm@redhat.com>
5429 * Makefile.in (OBJS): Add opt-problem.o.
5430 * dump-context.h: Include "selftest.h.
5431 (selftest::temp_dump_context): New forward decl.
5432 (class dump_context): Make friend of class
5433 selftest::temp_dump_context.
5434 (dump_context::dump_loc_immediate): New decl.
5435 (class dump_pretty_printer): Move here from dumpfile.c.
5436 (class temp_dump_context): Move to namespace selftest.
5437 (temp_dump_context::temp_dump_context): Add param
5438 "forcibly_enable_dumping".
5439 (selftest::verify_dumped_text):
5440 (ASSERT_DUMPED_TEXT_EQ): Move here from dumpfile.c.
5441 (selftest::verify_item):
5442 (ASSERT_IS_TEXT): Move here from dumpfile.c.
5443 (ASSERT_IS_TREE): Likewise.
5444 (ASSERT_IS_GIMPLE): Likewise.
5445 * dumpfile.c (dump_context::dump_loc): Move immediate dumping
5447 (dump_context::dump_loc_immediate): ...this new function.
5448 (class dump_pretty_printer): Move to dump-context.h.
5449 (dump_switch_p_1): Don't enable MSG_PRIORITY_REEMITTED.
5450 (opt_info_switch_p_1): Enable MSG_PRIORITY_REEMITTED.
5451 (temp_dump_context::temp_dump_context): Move to "selftest"
5452 namespace. Add param "forcibly_enable_dumping", and use it to
5453 conditionalize the use of m_pp;
5454 (selftest::verify_dumped_text): Make non-static.
5455 (ASSERT_DUMPED_TEXT_EQ): Move to dump-context.h.
5456 (selftest::verify_item): Make non-static.
5457 (ASSERT_IS_TEXT): Move to dump-context.h.
5458 (ASSERT_IS_TREE): Likewise.
5459 (ASSERT_IS_GIMPLE): Likewise.
5460 (selftest::test_capture_of_dump_calls): Pass "true" for new
5461 param of temp_dump_context.
5462 * dumpfile.h (enum dump_flag): Add MSG_PRIORITY_REEMITTED, adding
5463 it to MSG_ALL_PRIORITIES. Update values of TDF_COMPARE_DEBUG and
5465 * opt-problem.cc: New file.
5466 * opt-problem.h: New file.
5467 * optinfo-emit-json.cc
5468 (selftest::test_building_json_from_dump_calls): Pass "true" for
5469 new param of temp_dump_context.
5470 * optinfo.cc (optinfo_kind_to_dump_flag): New function.
5471 (optinfo::emit_for_opt_problem): New function.
5472 (optinfo::emit): Clarity which emit_item is used.
5473 * optinfo.h (optinfo::get_dump_location): New accessor.
5474 (optinfo::emit_for_opt_problem): New decl.
5475 (optinfo::emit): Make const.
5476 * selftest-run-tests.c (selftest::run_tests): Call
5477 selftest::opt_problem_cc_tests.
5478 * selftest.h (selftest::opt_problem_cc_tests): New decl.
5479 * tree-data-ref.c (dr_analyze_innermost): Convert return type from
5480 bool to opt_result, converting fprintf messages to
5481 opt_result::failure_at calls. Add "stmt" param for use by the
5483 (create_data_ref): Pass "stmt" to the dr_analyze_innermost call.
5484 (runtime_alias_check_p): Convert return type from bool to
5485 opt_result, converting dump_printf calls to
5486 opt_result::failure_at, using the statement DDR_A for their
5488 (find_data_references_in_stmt): Convert return type from bool to
5489 opt_result, converting "return false" to opt_result::failure_at
5491 * tree-data-ref.h: Include "opt-problem.h".
5492 (dr_analyze_innermost): Convert return type from bool to opt_result,
5493 and add a const gimple * param.
5494 (find_data_references_in_stmt): Convert return type from bool to
5496 (runtime_alias_check_p): Likewise.
5497 * tree-predcom.c (find_looparound_phi): Pass "init_stmt" to
5498 dr_analyze_innermost.
5499 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test):
5500 Convert return type from bool to opt_result, adding a message for
5501 the PARAM_VECT_MAX_VERSION_FOR_ALIAS_CHECKS zero case.
5502 (vect_analyze_data_ref_dependence): Convert return type from bool
5503 to opt_result. Change sense of return type from "false"
5504 effectively meaning "no problems" to "false" meaning a problem,
5505 so that "return false" becomes "return opt_result::success".
5506 Convert "return true" calls to opt_result::failure_at, using
5507 the location of statement A rather than vect_location.
5508 (vect_analyze_data_ref_dependences): Convert return type from bool
5510 (verify_data_ref_alignment): Likewise, converting dump_printf_loc
5511 calls to opt_result::failure_at, using the stmt location rather
5513 (vect_verify_datarefs_alignment): Convert return type from bool
5515 (vect_enhance_data_refs_alignment): Likewise. Split local "stat"
5516 into multiple more-tightly-scoped copies.
5517 (vect_analyze_data_refs_alignment): Convert return type from bool
5519 (vect_analyze_data_ref_accesses): Likewise, converting a
5520 "return false" to a "return opt_result::failure_at", adding a
5522 (vect_prune_runtime_alias_test_list): Convert return type from
5523 bool to opt_result, converting dump_printf_loc to
5524 opt_result::failure_at. Add a %G to show the pertinent statement,
5525 and use the stmt's location rather than vect_location.
5526 (vect_find_stmt_data_reference): Convert return type from
5527 bool to opt_result, converting dump_printf_loc to
5528 opt_result::failure_at, using stmt's location.
5529 (vect_analyze_data_refs): Convert return type from bool to
5530 opt_result. Convert "return false" to "return
5531 opt_result::failure_at", adding messages as needed.
5532 * tree-vect-loop.c (vect_determine_vf_for_stmt_1): Convert return
5533 type from bool to opt_result.
5534 (vect_determine_vf_for_stmt): Likewise.
5535 (vect_determine_vectorization_factor): Likewise, converting
5536 dump_printf_loc to opt_result::failure_at, using location of phi
5537 rather than vect_location.
5538 (vect_analyze_loop_form_1): Convert return type from bool to
5539 opt_result, converting dump_printf_loc calls, retaining the use of
5541 (vect_analyze_loop_form): Convert return type from loop_vec_info
5542 to opt_loop_vec_info.
5543 (vect_analyze_loop_operations): Convert return type from bool to
5544 opt_result, converting dump_printf_loc calls, using the location
5545 of phi/stmt rather than vect_location where available. Convert
5546 various "return false" to "return opt_result::failure_at" with
5547 "unsupported phi" messages.
5548 (vect_get_datarefs_in_loop): Convert return type from bool to
5549 opt_result. Add a message for the
5550 PARAM_LOOP_MAX_DATAREFS_FOR_DATADEPS failure.
5551 (vect_analyze_loop_2): Convert return type from bool to
5552 opt_result. Ensure "ok" is set to a opt_result::failure_at before
5553 each "goto again;", adding new messages where needed.
5554 Add "unsupported grouped {store|load}" messages.
5555 (vect_analyze_loop): Convert return type from loop_vec_info to
5557 * tree-vect-slp.c (vect_analyze_slp): Convert return type from
5559 * tree-vect-stmts.c (process_use): Likewise, converting
5560 dump_printf_loc call and using stmt location, rather than
5562 (vect_mark_stmts_to_be_vectorized): Likeise.
5563 (vect_analyze_stmt): Likewise, adding a %G.
5564 (vect_get_vector_types_for_stmt): Convert return type from bool to
5565 opt_result, converting dump_printf_loc calls and using stmt
5566 location, rather than vect_location.
5567 (vect_get_mask_type_for_stmt): Convert return type from tree to
5568 opt_tree, converting dump_printf_loc calls and using stmt location.
5569 * tree-vectorizer.c: Include "opt-problem.h.
5570 (try_vectorize_loop_1): Flag "Analyzing loop at" dump message as
5571 MSG_PRIORITY_INTERNALS. Convert local "loop_vinfo" from
5572 loop_vec_info to opt_loop_vec_info. If if fails, and dumping is
5573 enabled, use it to report at the top level "couldn't vectorize
5574 loop" followed by the problem.
5575 * tree-vectorizer.h (opt_loop_vec_info): New typedef.
5576 (vect_mark_stmts_to_be_vectorized): Convert return type from bool
5578 (vect_analyze_stmt): Likewise.
5579 (vect_get_vector_types_for_stmt): Likewise.
5580 (tree vect_get_mask_type_for_stmt): Likewise.
5581 (vect_analyze_data_ref_dependences): Likewise.
5582 (vect_enhance_data_refs_alignment): Likewise.
5583 (vect_analyze_data_refs_alignment): Likewise.
5584 (vect_verify_datarefs_alignment): Likewise.
5585 (vect_analyze_data_ref_accesses): Likewise.
5586 (vect_prune_runtime_alias_test_list): Likewise.
5587 (vect_find_stmt_data_reference): Likewise.
5588 (vect_analyze_data_refs): Likewise.
5589 (vect_analyze_loop): Convert return type from loop_vec_info to
5591 (vect_analyze_loop_form): Likewise.
5592 (vect_analyze_slp): Convert return type from bool to opt_result.
5594 2018-10-04 David Malcolm <dmalcolm@redhat.com>
5596 * doc/invoke.texi (-fopt-info): Document new "internals"
5598 * dump-context.h (dump_context::apply_dump_filter_p): New decl.
5599 * dumpfile.c (dump_options): Update for renaming of MSG_ALL to
5601 (optinfo_verbosity_options): Add "internals".
5602 (kind_as_string): Update for renaming of MSG_ALL to MSG_ALL_KINDS.
5603 (dump_context::apply_dump_filter_p): New member function.
5604 (dump_context::dump_loc): Use apply_dump_filter_p rather than
5605 explicitly masking the dump_kind.
5606 (dump_context::begin_scope): Increment the scope depth first. Use
5607 apply_dump_filter_p rather than explicitly masking the dump_kind.
5608 (dump_context::emit_item): Use apply_dump_filter_p rather than
5609 explicitly masking the dump_kind.
5610 (dump_dec): Likewise.
5611 (dump_hex): Likewise.
5612 (dump_switch_p_1): Default to MSG_ALL_PRIORITIES.
5613 (opt_info_switch_p_1): Default to MSG_PRIORITY_USER_FACING.
5614 (opt_info_switch_p): Update handling of default
5615 MSG_OPTIMIZED_LOCATIONS to cope with default of
5616 MSG_PRIORITY_USER_FACING.
5617 (dump_basic_block): Use apply_dump_filter_p rather than explicitly
5618 masking the dump_kind.
5619 (selftest::test_capture_of_dump_calls): Update test_dump_context
5620 instances to use MSG_ALL_KINDS | MSG_PRIORITY_USER_FACING rather
5621 than MSG_ALL. Generalize scope test to be run at all four
5622 combinations of with/without MSG_PRIORITY_USER_FACING and
5623 MSG_PRIORITY_INTERNALS, adding examples of explicit priority
5624 for each of the two values.
5625 * dumpfile.h (enum dump_flag): Add comment about the MSG_* flags.
5626 Rename MSG_ALL to MSG_ALL_KINDS. Add MSG_PRIORITY_USER_FACING,
5627 MSG_PRIORITY_INTERNALS, and MSG_ALL_PRIORITIES, updating the
5628 values for TDF_COMPARE_DEBUG and TDF_ALL_VALUES.
5629 (AUTO_DUMP_SCOPE): Add a note to the comment about the interaction
5630 with MSG_PRIORITY_*.
5631 * tree-vect-loop-manip.c (vect_loop_versioning): Mark versioning
5632 dump messages as MSG_PRIORITY_USER_FACING.
5633 * tree-vectorizer.h (DUMP_VECT_SCOPE): Add a note to the comment
5634 about the interaction with MSG_PRIORITY_*.
5636 2018-10-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
5638 * varasm.c (output_constant): Add new parameter merge_strings.
5639 Make strings properly zero terminated in merge string sections.
5640 (mergeable_string_section): Don't fail if the last char is non-zero.
5641 (assemble_variable_contents): Handle merge string sections.
5642 (assemble_variable): Likewise.
5643 (assemble_constant_contents): Likewise.
5644 (output_constant_def_contents): Likewise.
5645 (output_constructor_array_range,
5646 output_constructor_regular_field): Adjust call to output_constant.
5647 (output_object_block): Adjust call to assemble_constant_contents
5648 and assemble_variable_contents.
5650 2018-10-04 Martin Liska <mliska@suse.cz>
5653 * cgraphunit.c (process_function_and_variable_attributes):
5654 Warn about a function with alias attribute and a body.
5656 2018-10-04 Martin Liska <mliska@suse.cz>
5659 * multiple_target.c (redirect_to_specific_clone): New function.
5660 (ipa_target_clone): Use it.
5661 * tree-inline.c: Fix comment.
5663 2018-10-04 David Malcolm <dmalcolm@redhat.com>
5665 * dumpfile.c (gcc::dump_manager::dump_manager): Initialize new
5667 (gcc::dump_manager::~dump_manager): Free m_optinfo_filename.
5668 (gcc::dump_manager::register_pass): New member function, adapted
5669 from loop body in gcc::pass_manager::register_pass, adding a
5670 call to update_dfi_for_opt_info.
5671 (gcc::dump_manager::opt_info_enable_passes): Store the
5672 -fopt-info options into the new fields. Move the loop
5674 (gcc::dump_manager::update_dfi_for_opt_info): ...this new member
5676 * dumpfile.h (struct opt_pass): New forward decl.
5677 (gcc::dump_manager::register_pass): New decl.
5678 (gcc::dump_manager::update_dfi_for_opt_info): New decl.
5679 (class gcc::dump_manager): Add fields "m_optgroup_flags",
5680 "m_optinfo_flags", and "m_optinfo_filename".
5681 * passes.c (gcc::pass_manager::register_pass): Move all of the
5682 dump-handling code to gcc::dump_manager::register_pass.
5684 2018-10-04 Peter Bergner <bergner@linux.ibm.com>
5686 PR rtl-optimization/87466
5687 * target.def (setjmp_preserves_nonvolatile_regs_p): New target hook.
5688 * doc/tm.texi.in (TARGET_SETJMP_PRESERVES_NONVOLATILE_REGS_P): New hook.
5689 * doc/tm.texi: Regenerate.
5690 * ira-lives.c (process_bb_node_lives): Use the new target hook.
5691 * lra-lives.c (process_bb_lives): Likewise.
5692 * config/rs6000/rs6000.c (TARGET_SETJMP_PRESERVES_NONVOLATILE_REGS_P):
5695 2018-10-04 Tamar Christina <tamar.christina@arm.com>
5697 * params.c (add_params): Fix initialization.
5699 2018-10-04 Martin Liska <mliska@suse.cz>
5701 PR gcov-profile/84107
5702 * tree-profile.c (init_ic_make_global_vars):
5703 Remove ic_void_ptr_var and ic_gcov_type_ptr_var.
5704 Come up with new ic_tuple* variables. Emit
5705 __gcov_indirect_call{,_topn} variables.
5706 (gimple_gen_ic_profiler): Access the variable
5708 (gimple_gen_ic_func_profiler): Access
5709 __gcov_indirect_call.callee field.
5710 (gimple_init_gcov_profiler): Use ptr_type_node.
5711 * value-prof.c (gimple_ic): Use ptr_type_node.
5713 2018-10-04 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
5715 PR tree-optimization/85787
5716 * ipa-pure-const.c (malloc_candidate_p_1): Move most of
5717 malloc_candidate_p into this function and add support for
5718 detecting multiple phis.
5719 (DUMP_AND_RETURN): Move from malloc_candidate_p into top-level macro.
5721 2018-10-04 Martin Liska <mliska@suse.cz>
5724 * ipa-inline.c (inline_to_all_callers_1):
5725 Call ultimate_alias_target for node being inlined.
5727 2018-10-03 Jeff Law <law@redhat.com>
5729 * gimple-ssa-sprintf.c (format_string): Do not hardcode size of
5731 * tree.c (get_typenode_from_name): Moved from fortran/trans-types.c.
5732 * tree.h (get_typenode_from_name): Prototype.
5734 2018-10-03 Uros Bizjak <ubizjak@gmail.com>
5736 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
5737 Change operand 2 predicate to nonimmediate_operand.
5738 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
5740 2018-10-03 Martin Sebor <msebor@redhat.com>
5741 Jeff Law <law@redhat.com>
5743 * gimple-ssa-sprintf.c (struct fmtresult): Add new member and
5745 (get_string_length): Detect unterminated arrays.
5746 (format_string): Same.
5747 (format_directive): Warn about unterminated arrays.
5748 (handle_gimple_call): Mark statements with no_warning as needed.
5750 2018-10-03 Jim Wilson <jimw@sifive.com>
5752 * config/riscv/riscv-c.c (riscv_cpu_cpp_builtins): For ABI_ILP32E,
5753 also define __riscv_abi_rve. Delete trailing white space.
5755 2018-10-03 Paul Koning <ni1d@arrl.net>
5757 Enable LRA register allocator for PDP11.
5758 * config/pdp11/constraints.md (Q): Use define_memory_constraint.
5761 * config/pdp11/pdp11.c (pdp11_lra_p): New function.
5762 * config/pdp11/pdp11.opt (-mlra): New option.
5763 * doc/invoke.texi (PDP-11 Options): Document -mlra.
5765 2018-10-03 Uros Bizjak <ubizjak@gmail.com>
5767 * config/i386/i386.md (*<absneg:code>extendsfdf2): Remove.
5768 (*<absneg:code>extend<mode>xf2): Ditto.
5770 2018-10-03 Aldy Hernandez <aldyh@redhat.com>
5772 PR tree-optimization/87415
5773 * tree-vrp.c (set_value_range_with_overflow): Special case one bit
5776 2018-10-02 Jeff Law <law@redhat.com>
5778 * gimple-fold.c (get_range_strlen): Only set *nonstr when
5779 an unterminated string is discovered. Bubble up range
5780 even for unterminated strings.
5781 (gimple_fold_builtin_strlen): Do not fold if get_range_strlen
5782 indicates the string was not terminated via NONSTR.
5784 2018-10-03 Aldy Hernandez <aldyh@redhat.com>
5786 * tree-vrp.c (extract_range_from_unary_expr): Special case all
5787 pointer conversions.
5788 Do not do anything special for anti-ranges.
5790 2018-10-03 Jérôme Lambourg <lambourg@adacore.com>
5792 * config/arm/vxworks.h (ARM_TARGET2_DWARF_FORMAT): Adjust to
5793 DW_EH_PE_pcrel | DW_EH_PE_indirect for RTPs.
5795 2018-10-03 Martin Liska <mliska@suse.cz>
5797 PR gcov-profile/86109
5798 * coverage.c (coverage_begin_function): Do not
5799 mark lambdas as artificial.
5800 * tree-core.h (struct GTY): Remove tm_clone_flag
5801 and introduce new lambda_function.
5802 * tree.h (DECL_LAMBDA_FUNCTION): New macro.
5804 2018-10-02 Aaron Sawdey <acsawdey@linux.ibm.com>
5807 * config/rs6000/rs6000-string.c (expand_strn_compare): Check that both
5808 P8_VECTOR and VSX are enabled.
5810 2018-10-02 Andreas Krebbel <krebbel@linux.ibm.com>
5812 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
5813 0x3907 as CPU model number.
5815 2018-10-02 Andreas Krebbel <krebbel@linux.ibm.com>
5817 * common/config/s390/s390-common.c: Rename PF_ARCH12 to PF_Z14.
5818 * config/s390/s390.h (enum processor_flags): Rename PF_ARCH12 to
5819 PF_Z14. Rename TARGET_CPU_ARCH12 to TARGET_CPU_Z14,
5820 TARGET_CPU_ARCH12_P to TARGET_CPU_Z14_P, TARGET_ARCH12 to
5821 TARGET_Z14, and TARGET_ARCH12_P to TARGET_Z14_P.
5822 * config/s390/s390.md: Likewise. Rename also the cpu attribute
5823 value from arch12 to z14.
5825 2018-10-02 Uros Bizjak <ubizjak@gmail.com>
5827 * config/i386/i386.md (fxam<mode>2_i387_with_temp): Remove.
5829 (isinf<mode>2): Ditto.
5831 2018-10-02 Uros Bizjak <ubizjak@gmail.com>
5833 * config/i386/i386.c (ix86_emit_i387_round): Extend op1 to XFmode
5834 before emitting fxam. Perform calculations in XFmode.
5836 2018-10-02 Marc Glisse <marc.glisse@inria.fr>
5838 * match.pd (((X /[ex] A) +- B) * A): New transformation.
5840 2018-10-02 Marc Glisse <marc.glisse@inria.fr>
5843 * fold-const.c (fold_plusminus_mult_expr): Handle complex and vectors.
5844 * tree.c (signed_or_unsigned_type_for): Handle complex.
5846 2018-10-02 Jeff Law <law@redhat.com>
5848 * gimple-fold.c (get_range_strlen): Remove dead code.
5850 2018-10-02 Martin Sebor <msebor@redhat.com>
5851 Jeff Law <law@redhat.com>
5853 * builtins.c (unterminated_array): Add new arguments.
5854 If argument is not terminated, bubble up size and exact
5856 (expand_builtin_strnlen): Detect, avoid expanding
5857 and diagnose unterminated arrays.
5858 (c_strlen): Fill in offset of start of unterminated strings.
5859 * builtins.h (unterminated_array): Update prototype.
5861 2018-10-02 Richard Biener <rguenther@suse.de>
5863 * config/i386/sse.md (reduc_plus_scal_v4df): Avoid the use
5864 of haddv4df, first reduce to SSE width and exploit the fact
5865 that we only need element zero with the reduction result.
5866 (reduc_plus_scal_v2df): Likewise.
5868 2018-10-02 Eric Botcazou <ebotcazou@adacore.com>
5870 * dojump.h (do_jump): Delete.
5871 (do_jump_1): Likewise.
5872 (split_comparison): Move around.
5873 * dojump.c (do_jump): Make static.
5874 (do_jump_1): Likewise.
5875 (jumpifnot): Move around.
5876 (jumpifnot_1): Likewise.
5878 (jumpif_1): Likewise.
5879 * expr.c (expand_expr_real_1): Call jumpif[not] instead of do_jump.
5881 2018-10-02 Eric Botcazou <ebotcazou@adacore.com>
5883 * reorg.c (make_return_insns): Use emit_copy_of_insn_after for the
5884 insns in the delay slot and add_insn_after for the jump insn.
5886 2018-10-02 Richard Biener <rguenther@suse.de>
5888 * tree-inline.c (expand_call_inline): Use the location of
5889 the callee declaration for the inline-entry marker.
5890 * final.c (notice_source_line): Remove special-casing of
5891 NOTE_INSN_INLINE_ENTRY.
5893 2018-10-01 Carl Love <cel@us.ibm.com>
5896 * config/rs6000/rs6000-builtin.def (__builtin_mffsl): New.
5897 (__builtin_mtfsb0): New.
5898 (__builtin_mtfsb1): New.
5899 ( __builtin_set_fpscr_rn): New.
5900 (__builtin_set_fpscr_drn): New.
5901 * config/rs6000/rs6000.c (rs6000_expand_mtfsb_builtin): Add.
5902 (rs6000_expand_set_fpscr_rn_builtin): Add.
5903 (rs6000_expand_set_fpscr_drn_builtin): Add.
5904 (rs6000_expand_builtin): Add case statement entries for
5905 RS6000_BUILTIN_MTFSB0, RS6000_BUILTIN_MTFSB1,
5906 RS6000_BUILTIN_SET_FPSCR_RN, RS6000_BUILTIN_SET_FPSCR_DRN,
5907 RS6000_BUILTIN_MFFSL.
5908 (rs6000_init_builtins): Add ftype initialization and def_builtin
5909 calls for __builtin_mffsl, __builtin_mtfsb0, __builtin_mtfsb1,
5910 __builtin_set_fpscr_rn, __builtin_set_fpscr_drn.
5911 * config/rs6000.md (rs6000_mtfsb0, rs6000_mtfsb1, rs6000_mffscrn,
5912 rs6000_mffscdrn): Add define_insn.
5913 (rs6000_set_fpscr_rn, rs6000_set_fpscr_drn): Add define_expand.
5914 * doc/extend.texi: Add documentation for the builtins.
5916 2018-10-01 Richard Biener <rguenther@suse.de>
5918 PR tree-optimization/87465
5919 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Fix typo
5920 causing branch miscounts.
5922 2018-10-01 Tamar Christina <tamar.christina@arm.com>
5924 * common/config/aarch64/aarch64-common.c (TARGET_OPTION_DEFAULT_PARAM,
5925 aarch64_option_default_param): New.
5926 (params.h): Include.
5927 (TARGET_OPTION_VALIDATE_PARAM, aarch64_option_validate_param): New.
5928 * config/aarch64/aarch64.c (aarch64_override_options_internal): Simplify
5929 stack-clash protection validation code.
5931 2018-10-01 Tamar Christina <tamar.christina@arm.com>
5933 * params.c (validate_param): New.
5934 (add_params): Use it.
5935 (set_param_value): Refactor param validation into validate_param.
5936 (diagnostic.h): Include.
5937 * diagnostic.h (diagnostic_ready_p): New.
5939 2018-10-01 Tamar Christina <tamar.christina@arm.com>
5941 * params.c (set_param_value):
5942 Add index of parameter being validated.
5943 * common/common-target.def (option_validate_param): New.
5944 * common/common-targhooks.h (default_option_validate_param): New.
5945 * common/common-targhooks.c (default_option_validate_param): New.
5946 * doc/tm.texi.in (TARGET_OPTION_VALIDATE_PARAM): New.
5947 * doc/tm.texi: Regenerate.
5949 2018-10-01 Tamar Christina <tamar.christina@arm.com>
5952 * config/aarch64/aarch64.c (aarch64_override_options_internal):
5953 Add validation for stack-clash parameters and set defaults.
5955 2018-10-01 Tamar Christina <tamar.christina@arm.com>
5958 * configure.ac: Add stack-clash-protection-guard-size.
5959 * doc/install.texi: Document it.
5960 * config.in (DEFAULT_STK_CLASH_GUARD_SIZE): New.
5961 * params.def: Update comment for guard-size.
5962 (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE,
5963 PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Update description.
5964 * configure: Regenerate.
5966 2018-10-01 Tamar Christina <tamar.christina@arm.com>
5969 * config/aarch64/aarch64.h (STACK_CLASH_MIN_BYTES_OUTGOING_ARGS,
5970 STACK_DYNAMIC_OFFSET): New.
5971 * config/aarch64/aarch64.c (aarch64_layout_frame):
5972 Update outgoing args size.
5973 (aarch64_stack_clash_protection_alloca_probe_range,
5974 TARGET_STACK_CLASH_PROTECTION_ALLOCA_PROBE_RANGE): New.
5976 2018-10-01 Tamar Christina <tamar.christina@arm.com>
5979 * explow.c (anti_adjust_stack_and_probe_stack_clash): Support custom
5981 * target.def (stack_clash_protection_alloca_probe_range): New.
5982 (stack_clash_protection_final_dynamic_probe): Remove.
5983 * targhooks.h (default_stack_clash_protection_alloca_probe_range) New.
5984 (default_stack_clash_protection_final_dynamic_probe): Remove.
5985 * targhooks.c: Likewise.
5986 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_ALLOCA_PROBE_RANGE): New.
5987 (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE): Remove.
5988 * doc/tm.texi: Regenerate.
5990 2018-10-01 Tamar Christina <tamar.christina@arm.com>
5993 * config/aarch64/aarch64-protos.h (aarch64_output_probe_sve_stack_clash): New.
5994 * config/aarch64/aarch64.c (aarch64_output_probe_sve_stack_clash,
5995 aarch64_clamp_to_uimm12_shift): New.
5996 (aarch64_allocate_and_probe_stack_space): Add SVE specific section.
5997 * config/aarch64/aarch64.md (probe_sve_stack_clash): New.
5999 2018-10-01 Tamar Christina <tamar.christina@arm.com>
6002 * config/aarch64/aarch64.c (aarch64_layout_frame): Add assert.
6004 2018-10-01 Jeff Law <law@redhat.com>
6005 Richard Sandiford <richard.sandiford@linaro.org>
6006 Tamar Christina <tamar.christina@arm.com>
6009 * config/aarch64/aarch64.md
6010 (probe_stack_range): Add k (SP) constraint.
6011 * config/aarch64/aarch64.h (STACK_CLASH_CALLER_GUARD,
6012 STACK_CLASH_MAX_UNROLL_PAGES): New.
6013 * config/aarch64/aarch64.c (aarch64_output_probe_stack_range): Emit
6014 stack probes for stack clash.
6015 (aarch64_allocate_and_probe_stack_space): New.
6016 (aarch64_expand_prologue): Use it.
6017 (aarch64_expand_epilogue): Likewise and update IP regs re-use criteria.
6018 (aarch64_sub_sp): Add emit_move_imm optional param.
6020 2018-10-01 MCC CS <deswurstes@users.noreply.github.com>
6022 PR tree-optimization/87261
6023 * match.pd: Remove trailing whitespace.
6024 Add (x & y) | ~(x | y) -> ~(x ^ y),
6025 (~x | y) ^ (x ^ y) -> x | ~y and (x ^ y) | ~(x | y) -> ~(x & y)
6027 2018-10-01 Claudiu Zissulescu <claziss@synopsys.com>
6029 * config/arc/arc.md (*add_n): Clean up pattern, update instruction
6031 (ashlsi3_insn): Update instruction constraints.
6032 (ashrsi3_insn): Likewise.
6033 (rotrsi3): Likewise.
6034 (add_shift): Likewise.
6035 * config/arc/constraints.md (Csz): New 32 bit constraint. It
6036 avoids placing in the limm field small constants which, otherwise,
6037 could end into a small instruction.
6039 2018-10-01 Claudiu Zissulescu <claziss@synopsys.com>
6041 * config/arc/arc.md (maddsidi4_split): Don't use dmac if the
6042 destination register is not odd-even.
6043 (umaddsidi4_split): Likewise.
6045 2018-10-01 Richard Biener <rguenther@suse.de>
6047 * tree-inline.c (expand_call_inline): Store origin of fn
6048 in BLOCK_ABSTRACT_ORIGIN for the inline BLOCK.
6049 * tree.c (block_ultimate_origin): Simplify and do some
6052 2018-09-30 Uros Bizjak <ubizjak@gmail.com>
6054 * config/i386/mmx.md (EMMS): New int iterator.
6055 (emms): New int attribute.
6056 (mmx_<emms>): Macroize insn from *mmx_emms and *mmx_femms using
6057 EMMS int iterator. Explicitly declare clobbers.
6058 (mmx_emms): Remove expander.
6060 * config/i386/predicates.md (emms_operation): Remove predicate.
6061 (vzeroall_pattern): New predicate.
6062 (vzeroupper_pattern): Rename from vzeroupper_operation.
6063 * config/i386/i386.c (ix86_avx_u128_mode_after): Use
6064 vzeroupper_pattern and vzeroall_pattern predicates.
6066 2018-09-30 Peter Bergner <bergner@linux.ibm.com>
6068 PR rtl-optimization/86939
6069 * ira-lives.c (make_hard_regno_born): Rename from this...
6070 (make_hard_regno_live): ... to this. Remove update to conflict
6071 information. Update function comment.
6072 (make_hard_regno_dead): Add conflict information update. Update
6074 (make_object_born): Rename from this...
6075 (make_object_live): ... to this. Remove update to conflict information.
6076 Update function comment.
6077 (make_object_dead): Add conflict information update. Update function
6079 (mark_pseudo_regno_live): Call make_object_live.
6080 (mark_pseudo_regno_subword_live): Likewise.
6081 (mark_hard_reg_dead): Update function comment.
6082 (mark_hard_reg_live): Call make_hard_regno_live.
6083 (process_bb_node_lives): Likewise.
6084 * lra-lives.c (make_hard_regno_born): Rename from this...
6085 (make_hard_regno_live): ... to this. Remove update to conflict
6086 information. Remove now uneeded check_pic_pseudo_p argument.
6087 Update function comment.
6088 (make_hard_regno_dead): Add check_pic_pseudo_p argument and add update
6089 to conflict information. Update function comment.
6090 (mark_pseudo_live): Remove update to conflict information. Update
6092 (mark_pseudo_dead): Add conflict information update.
6093 (mark_regno_live): Call make_hard_regno_live.
6094 (mark_regno_dead): Call make_hard_regno_dead with new arguement.
6095 (process_bb_lives): Call make_hard_regno_live and make_hard_regno_dead.
6097 2018-09-29 H.J. Lu <hongjiu.lu@intel.com>
6100 * config/i386/i386.c (construct_container): Use TImode for
6101 BLKmode values in 2 integer registers.
6103 2018-09-29 Jeff Law <law@redhat.com>
6105 * builtins.c (unterminated_array): Pass in c_strlen_data * to
6106 c_strlen rather than just a tree *.
6107 (c_strlen): Change NONSTR argument to a c_strlen_data pointer.
6108 Update recursive calls appropriately. If caller did not provide a
6109 suitable data pointer, create a local one. When a non-terminated
6110 string is discovered, bubble up information about the string via the
6111 c_strlen_data object.
6112 * builtins.h (c_strlen): Update prototype.
6113 (c_strlen_data): New structure.
6114 * gimple-fold.c (get_range_strlen): Update calls to c_strlen.
6115 For a type 2 call, if c_strlen indicates a non-terminated string
6116 use the length of the non-terminated string.
6117 (gimple_fold_builtin_stpcpy): Update calls to c_strlen.
6119 2018-09-29 Jakub Jelinek <jakub@redhat.com>
6122 * config/i386/avx512fintrin.h (_mm512_abs_pd, _mm512_mask_abs_pd): Use
6123 __m512d type for __A argument rather than __m512.
6125 2018-09-28 John David Anglin <danglin@gcc.gnu.org>
6127 * match.pd (simple_comparison): Don't optimize if either operand is
6128 a function pointer when target needs function pointer canonicalization.
6130 2018-09-28 Segher Boessenkool <segher@kernel.crashing.org>
6132 * config/rs6000/driver-rs6000.c (asm_names): Adjust the entries for
6133 power5 .. power9 to remove indirection.
6134 * config/rs6000/rs6000.h (ASM_CPU_POWER5_SPEC, ASM_CPU_POWER6_SPEC,
6135 ASM_CPU_POWER7_SPEC, ASM_CPU_POWER8_SPEC, ASM_CPU_POWER9_SPEC,
6136 ASM_CPU_476_SPEC): Delete.
6137 (ASM_CPU_SPEC): Adjust.
6138 (EXTRA_SPECS): Delete asm_cpu_power5, asm_cpu_power6, asm_cpu_power7,
6139 asm_cpu_power8, asm_cpu_power9, asm_cpu_476.
6141 2018-09-28 Segher Boessenkool <segher@kernel.crashing.org>
6143 * config.in: Delete HAVE_AS_DCI.
6144 * config/powerpcspe/powerpcspe.h: Treat HAVE_AS_DCI as always true.
6145 * config/rs6000/rs6000.h: Ditto.
6146 * configure.ac: Delete HAVE_AS_DCI.
6147 * configure: Regenerate.
6149 2018-09-28 Segher Boessenkool <segher@kernel.crashing.org>
6151 * config.in (HAVE_AS_LWSYNC): Delete.
6152 * config/powerpcspe/powerpcspe.h (TARGET_LWSYNC_INSTRUCTION): Delete.
6153 * config/powerpcspe/sync.md (*lwsync): Always generate lwsync, never
6155 * config/rs6000/rs6000.h (TARGET_LWSYNC_INSTRUCTION): Delete.
6156 * config/rs6000/sync.md (*lwsync): Always generate lwsync, never do it
6158 * configure.ac: Delete HAVE_AS_LWSYNC.
6159 * configure: Regenerate.
6161 2018-09-28 Eric Botcazou <ebotcazou@adacore.com>
6162 Pierre-Marie de Rodat <derodat@adacore.com>
6164 * calls.c (expand_call): Try to do a tail call for thunks at -O0 too.
6165 * cgraph.h (struct cgraph_thunk_info): Add indirect_offset.
6166 (cgraph_node::create_thunk): Add indirect_offset parameter.
6167 (thunk_adjust): Likewise.
6168 * cgraph.c (cgraph_node::create_thunk): Add indirect_offset parameter
6169 and initialize the corresponding field with it.
6170 (cgraph_node::dump): Dump indirect_offset field.
6171 * cgraphclones.c (duplicate_thunk_for_node): Deal with indirect_offset.
6172 * cgraphunit.c (cgraph_node::analyze): Be prepared for external thunks.
6173 (thunk_adjust): Add indirect_offset parameter and deal with it.
6174 (cgraph_node::expand_thunk): Deal with the indirect_offset field and
6175 pass it to thunk_adjust. Do not call the target hook if it's non-zero
6176 or if the thunk is external or local. Fix formatting. Do not chain
6177 the RESULT_DECL to BLOCK_VARS. Pass the static chain to the target,
6178 if any, in the GIMPLE representation.
6179 * ipa-icf.c (sem_function::equals_wpa): Deal with indirect_offset.
6180 * lto-cgraph.c (lto_output_node): Write indirect_offset field.
6181 (input_node): Read indirect_offset field.
6182 * tree-inline.c (expand_call_inline): Pass indirect_offset field in the
6183 call to thunk_adjust.
6184 * tree-nested.c (struct nesting_info): Add thunk_p field.
6185 (create_nesting_tree): Set it.
6186 (convert_all_function_calls): Copy static chain from targets to thunks.
6187 (finalize_nesting_tree_1): Return early for thunks.
6188 (unnest_nesting_tree_1): Do not finalize thunks.
6189 (gimplify_all_functions): Do not gimplify thunks.
6191 2018-09-28 David Malcolm <dmalcolm@redhat.com>
6193 * opt-suggestions.c (option_proposer::build_option_suggestions):
6194 Release "option_values".
6196 2018-09-28 David Malcolm <dmalcolm@redhat.com>
6198 * coverage.c (get_coverage_counts): Convert problem-reporting dump
6199 messages from MSG_OPTIMIZED_LOCATIONS to MSG_MISSED_OPTIMIZATION.
6200 * dumpfile.c (kind_as_string): New function.
6201 (dump_loc): Rather than a hardcoded prefix of "note: ", use
6202 kind_as_string to vary the prefix based on dump_kind.
6203 (selftest::test_capture_of_dump_calls): Update for above.
6205 2018-09-28 Uros Bizjak <ubizjak@gmail.com>
6207 * config/i386/i386.h (SSE_REGNO): Fix check for FIRST_REX_SSE_REG.
6208 (GET_SSE_REGNO): Rename from SSE_REGNO. Update all uses for rename.
6210 2018-09-28 Uros Bizjak <ubizjak@gmail.com>
6212 * config/i386/i386.h (CC_REGNO): Remove FPSR_REG.
6213 * config/i386/i386.c (ix86_fixed_condition_code_regs): Use
6214 INVALID_REGNUM instead of FPSR_REG.
6215 (ix86_md_asm_adjust): Do not clobber FPSR_REG.
6216 * config/i386/i386.md: Update comment of FP compares.
6217 (fldenv): Do not clobber FPSR_REG.
6219 2018-09-28 Richard Biener <rguenther@suse.de>
6221 * tree.h (BLOCK_ORIGIN): New.
6222 * omp-expand.c (grid_expand_target_grid_body): Assign
6223 BLOCK_ORIGIN to BLOCK_ABSTRACT_ORIGIN.
6224 * tree-inline.c (remap_block): Likewise.
6225 * auto-profile.c (get_function_decl_from_block): Simplify
6226 by eliding the BLOCK_ABSTRACT_ORIGIN chasing.
6227 * langhooks.c (lhd_print_error_function): Likewise.
6228 * optinfo-emit-json.cc (optrecord_json_writer::inlining_chain_to):
6230 * tree-ssa-live.c (remove_unused_scope_block_p): Likewise.
6231 * tree.c (block_nonartificial_location): Likewise.
6232 (block_ultimate_origin): Likewise.
6233 * tree-pretty-print.c (percent_K_format): Likewise. Remove
6234 no longer needed LTO case.
6236 2018-09-28 Andrew Stubbs <ams@codesourcery.com>
6237 Jan Hubicka <jh@suse.cz>
6238 Martin Jambor <mjambor@suse.cz>
6240 * simplify-rtx.c (simplify_merge_mask): New function.
6241 (simplify_ternary_operation): Use it, also see if VEC_MERGEs with the
6242 same masks are used in op1 or op2.
6243 (test_vec_merge): New function.
6244 (test_vector_ops): Call test_vec_merge.
6246 2018-09-28 Eric Botcazou <ebotcazou@adacore.com>
6248 * config/sparc/sparc-protos.h (sparc_branch_cost): Declare.
6249 * config/sparc/sparc.h (BRANCH_COST): Call sparc_branch_cost.
6250 * config/sparc/sparc.c (struct processor_costs): Add branch_cost field.
6251 (cypress_costs): Set it.
6252 (supersparc_costs): Likewise.
6253 (hypersparc_costs): Likewise.
6254 (leon_cost): Likewise.
6255 (leon3_costs): Likewise.
6256 (sparclet_costs): Likewise.
6257 (ultrasparc_costs): Likewise.
6258 (ultrasparc_costs): Likewise.
6259 (niagara_costs): Likewise.
6260 (niagara2_costs): Likewise.
6261 (niagara3_costs): Likewise.
6262 (niagara4_costs): Likewise.
6263 (niagara7_costs): Likewise.
6264 (m8_costs): Likewise.
6265 (TARGET_CAN_FOLLOW_JUMP): Define.
6266 (pass_work_around_errata::gate): Minor tweak.
6267 (sparc_option_override): Remove MASK_FSMULD mask for V7 processors.
6268 Do not set both MASK_VIS4 and MASK_VIS4B for M8 processor.
6269 Automaitcally clear MASK_FSMULD mask for V7 processors.
6270 (sparc_can_follow_jump): New static function.
6271 (output_ubranch): Deal with CROSSING_JUMP_P.
6272 (sparc_use_sched_lookahead): Rewrite using switch statement.
6273 (sparc_issue_rate): Reorder.
6274 (sparc_branch_cost): New function.
6276 2018-09-27 Martin Sebor <msebor@redhat.com>
6278 * tree.h (tree_to_shwi): Add attribute nonnull and pure.
6279 (tree_to_poly_int64, tree_to_uhwi, tree_to_poly_uint64): Same.
6280 (int_fits_type_p): Same.
6282 2018-09-27 Uros Bizjak <ubizjak@gmail.com>
6284 * config/i386/i386.md (FPCR_REG): Remove.
6285 (UNSPEC_FLDCW): Remove.
6286 (x86_fnstcw_1): Use (const_int 0) instead of FPCR_REG.
6287 (x86_fldcw_1): Remove insn pattern.
6288 (fnstenv): Do not clobber FPCR_REG.
6290 * config/i386/i386.h (FIXED_REGISTERS) Remove fpsr register.
6291 (CALL_USED_REGISTERS): Ditto.
6292 (REG_ALLOC_ORDER): Ditto.
6293 (REG_CLASS_CONTENTS): Ditto.
6294 (HI_REGISTER_NAMES): Ditto.
6295 (ADDITIONAL_REGISTER_NAMES): Use defines instead
6296 of numerical constants.
6297 * config/i386/i386.c (regclass_map): Remove fpsr register.
6298 (dbx_register_map): Ditto.
6299 (dbx64_register_map): Ditto.
6300 (svr4_dbx_register_map): Ditto.
6301 (print_reg): Do not handle FPCR_REG.
6303 2018-09-27 Segher Boessenkool <segher@kernel.crashing.org>
6306 * config.in (HAVE_AS_CMPB, HAVE_AS_DFP, HAVE_AS_FPRND, HAVE_AS_MFPGPR,
6307 HAVE_AS_POPCNTB, HAVE_AS_POPCNTD, HAVE_AS_POWER8, HAVE_AS_POWER9):
6308 Delete, always treat as true.
6309 * config/powerpcspe/powerpcspe.c (rs6000_option_override_internal):
6310 Ditto. Simplify remaining code.
6311 * config/powerpcspe/powerpcspe.h: Ditto.
6312 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ditto.
6313 Simplify remaining code.
6314 (rs6000_expand_builtin): Ditto.
6315 * config/rs6000/rs6000.h: Ditto.
6316 * configure.ac: Ditto.
6317 * configure: Regenerate.
6319 2018-09-27 Martin Liska <mliska@suse.cz>
6321 * coverage.c (get_coverage_counts): Revert the formatting
6322 of missing profile opt info.
6324 2018-09-27 Richard Biener <rguenther@suse.de>
6328 * dwarf2out.c (set_block_origin_self): Do not mark outermost
6329 block as we do not output that.
6330 (gen_inlined_subroutine_die): Elide the originally outermost
6331 block, matching what we do for concrete instances.
6332 (decls_for_scope): Add parameter specifying whether to recurse
6335 2018-09-27 Andrew Stubbs <ams@codesourcery.com>
6336 Tom de Vries <tom@codesourcery.com>
6340 * expmed.c (emit_cstore): Fix handling of result_mode == BImode and
6341 STORE_FLAG_VALUE == 1.
6343 2018-09-27 Andreas Krebbel <krebbel@linux.ibm.com>
6345 * config/s390/s390.md (PPA_TX_ABORT, PPA_OOO_BARRIER): New
6346 constant definitions.
6347 ("tx_assist"): Replace magic number with PPA_TX_ABORT.
6348 ("*ppa"): Enable pattern also for -march=zEC12 -mno-htm.
6349 ("speculation_barrier"): New expander definition.
6351 2018-09-26 Indu Bhagat <indu.bhagat@oracle.com>
6353 PR gcov-profile/86957
6354 * common.opt: New warning option -Wmissing-profile.
6355 * coverage.c (get_coverage_counts): Add warning for missing .gcda file.
6356 * doc/invoke.texi: Document -Wmissing-profile.
6358 2018-09-26 Jim Wilson <jimw@sifive.com>
6360 * config/riscv/riscv.md (subsi3_extended2): Add J constraint.
6361 (negdi2, negsi2, negsi2_extended, negsi2_extended2): New.
6363 2018-09-26 Martin Sebor <msebor@redhat.com>
6365 * tree.c (zerop): Change return type to bool.
6366 (integer_zerop, integer_onep, integer_each_onep): Same.
6367 (integer_all_onesp, integer_minus_onep, integer_pow2p): Same.
6368 (integer_nonzerop, integer_truep, tree_ctz, real_zerop): Same.
6369 (real_onep, real_minus_onep, chain_index): Same.
6370 (print_type_hash_statistics, type_list_equal): Same.
6371 * tree.h (zerop): Same.
6372 (zerop, integer_zerop, integer_onep, integer_each_onep): Same.
6373 (integer_all_onesp, integer_minus_onep, integer_pow2p): Same.
6374 (integer_nonzerop, integer_truep, tree_ctz, real_zerop): Same.
6375 (real_onep, real_minus_onep, chain_index): Same.
6376 (print_type_hash_statistics, type_list_equal): Same.
6378 2018-09-26 Jim Wilson <jimw@sifive.com>
6380 * config/riscv/riscv.h (FUNCTION_ARG_REGNO_P): Fix comment.
6382 2018-09-26 Jakub Jelinek <jakub@redhat.com>
6385 * config/i386/i386.c: Include debug.h and dwarf2out.h.
6386 (output_indirect_thunk): Emit DW_CFA_def_cfa_offset after the
6389 2018-09-25 Andrew Stubbs <ams@codesourcery.com>
6391 * builtins.c (get_builtin_sync_mem): Force address mode conversion.
6393 2018-09-26 Uros Bizjak <ubizjak@gmail.com>
6395 * config/i386/i386.h (enum reg_class): Remove FP_TOP_SSE_REGS
6396 and FP_SECOND_SSE_REGS.
6397 (REG_CLASS_NAMES): Ditto.
6398 (REG_CLASS_CONTENTS): Ditto.
6399 * config/i386/i386.c (ix86_preferred_reload_class) Do not handle
6400 FP_TOP_SSE_REGS and FP_SECOND_SSE_REGS classes.
6401 (ix86_preferred_output_reload_class): Ditto.
6402 * config/i386/i386.md (fix_trunc<mode>_i387_fisttp): Change "=&1f"
6403 clobber constraint to "=&f".
6404 (fix_truncdi_i387): Ditto.
6405 (lrintxfdi2): Ditto.
6406 (fistdi2_<rounding>): Ditto.
6407 (fpremxf4_i387): Change "=u" constraint to "=f".
6408 (fprem1xf4_i387): Ditto.
6410 (fptanxf4_i387): Ditto.
6411 (fxtractxf3_i387): Ditto.
6412 (fscalexf4_i387): Ditto.
6413 (atan2xf3): Change "u" constraint to "f".
6414 (fyl2xxf3_i387): Ditto.
6415 (fyl2xp1xf3_i387): Ditto.
6417 2018-09-26 Uros Bizjak <ubizjak@gmail.com>
6420 * config/i386/i386.h (NUM_MODES_FOR_MODE_SWITCHING): Update
6421 for removed I387_MASK_PM entity.
6424 2018-09-26 Jeff Law <law@redhat.com>
6426 2018-09-26 Alexey Neyman <stilor@att.net>
6428 * graphite.h: Include <isl/id.h> and <isl/space.h>; these
6429 headers are no longer pulled in by <isl/val.h>.
6431 2018-09-26 Richard Biener <rguenther@suse.de>
6434 * dwarf2out.c (gen_lexical_block_die): Do not equate inline
6435 or concrete instance DIE to the tree. Create abstract origin
6436 attributes also for concrete instances.
6438 2018-09-26 Alexey Neyman <stilor@att.net>
6440 * graphite.h: Include <isl/id.h> and <isl/space.h>; these
6441 headers are no longer pulled in by <isl/val.h>.
6443 2018-09-26 Matthew Malcomson <matthew.malcomson@arm.com>
6445 * config/arm/arm.c (arm_split_compare_and_swap, arm_split_atomic_op):
6446 Use new helper functions.
6447 * config/arm/sync.md (atomic_load<mode>, atomic_store<mode>):
6448 Use new helper functions.
6449 * config/arm/aarch-common-protos.h (aarch_mm_needs_acquire,
6450 aarch_mm_needs_release): New declarations.
6451 * config/arm/aarch-common.c (aarch_mm_needs_acquire,
6452 aarch_mm_needs_release): New.
6454 2018-09-26 Eric Botcazou <ebotcazou@adacore.com>
6456 * config/arm/arm.c (arm_reorg): Skip Thumb reorg pass for thunks.
6457 (arm32_output_mi_thunk): Deal with long calls.
6459 2018-09-26 Richard Biener <rguenther@suse.de>
6463 * tree-inline.c (expand_call_inline): When the location
6464 of the call is UNKNOWN_LOCATION use DECL_SOURCE_LOCATION
6465 or BUILTINS_LOCATION for the BLOCK_SOURCE_LOCATION of
6466 the inserted BLOCK to make inlined_function_outer_scope_p
6468 * dwarf2out.c (add_call_src_coords_attributes): Do not add
6469 coords for reserved locations.
6471 2018-09-25 Segher Boessenkool <segher@kernel.crashing.org>
6473 * config/rs6000/rs6000.md (*movcc_internal1): Use set_attr_alternative.
6474 (*call_indirect_nonlocal_sysv<mode>): Ditto.
6475 (*call_value_indirect_nonlocal_sysv<mode>): Ditto.
6476 (*sibcall_nonlocal_sysv<mode>): Ditto.
6477 (*sibcall_value_nonlocal_sysv<mode>): Ditto.
6478 (<bd>_<mode>): Ditto.
6479 (<bd>tf_<mode>): Ditto.
6481 2018-09-25 Segher Boessenkool <segher@kernel.crashing.org>
6483 * config/rs6000/altivec.md (*altivec_mov<mode>): Write the output
6484 control string as a list of templates instead of as C code.
6485 (*altivec_movti): Ditto.
6486 * config/rs6000/darwin.md (movdf_low_di): Ditto.
6488 2018-09-25 Jim Wilson <jimw@sifive.com>
6490 * config/riscv/riscv.c (riscv_split_symbol): Mark auipc label as weak
6491 when target symbol is weak.
6493 2018-09-25 Bernd Edlinger <bernd.edlinger@hotmail.de>
6496 * builtins.c (unterminated_array): Simplify.
6497 * expr.c (string_constant): Handle SSA_NAME. Add more exceptions
6498 where pointer arithmetic is safe.
6500 2018-09-25 Segher Boessenkool <segher@kernel.crashing.org>
6503 * config/rs6000/altivec.md (altivec_vspltb): Use
6504 const_0_to_15_operand instead of u5bit_cint_operand.
6505 (*altivec_vspltb_internal): Ditto.
6506 (altivec_vspltb_direct): Ditto.
6507 (altivec_vsplth): Use const_0_to_7_operand instead of
6509 (*altivec_vsplth_internal): Ditto.
6510 (altivec_vsplth_direct): Ditto.
6511 (altivec_vspltw): Use const_0_to_3_operand instead of
6513 (*altivec_vspltw_internal): Ditto.
6514 (altivec_vspltw_direct): Ditto.
6515 (altivec_vspltsf): Ditto.
6516 (*altivec_vspltsf_internal): Ditto.
6517 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Handle the
6518 various splats with the proper size immediate. Reorder the various
6519 cases by ascending size of immediate, and put all such together.
6521 2018-09-25 Richard Biener <rguenther@suse.de>
6524 * dwarf2out.c (add_AT_external_die_ref): Remove now redundant
6526 (maybe_create_die_with_external_ref): Do not create
6527 DW_TAG_imported_unit here.
6528 (add_abstract_origin_attribute): Handle external BLOCK refs.
6529 (dwarf2out_abstract_function): Simplify LTO case.
6530 (dwarf2out_early_finish): Create DW_TAG_imported_unit explicitely
6531 rather than using maybe_create_die_with_external_ref.
6533 2018-09-25 Uros Bizjak <ubizjak@gmail.com>
6536 * config/i386/i386.md (frndintxf2_mask_pm): Remove.
6537 (frndintxf2_mask_pm_i387): Ditto.
6538 (nearbyintxf2): Rewrite expander pattern to match rintxf2.
6539 Enable for !flag_trapping_math.
6540 (nearbyint<mode>2): Enable x87 modes for !flag_trapping_math.
6541 Enable SSE modes for TARGET_SSE4_1 and expand them with round insn.
6542 Change operand 1 predicate to nonimmediate_operand.
6543 (attr "i387_cw"): Remove mask_pm.
6544 * config/i386/i386.h (enum ix86_stack_slot): Remove SLOT_CW_MASK_PM.
6545 (enum ix86_entity): Remove I387_MASK_PM.
6546 * config/i386/i386.c (ix86_i387_mode_needed): Do not
6547 handle I387_MASK_PM.
6548 (ix86_mode_needed): Ditto.
6549 (ix86_mode_after): Ditto.
6550 (ix86_mode_entry): Ditto.
6551 (ix86_mode_exit): Ditto.
6552 (emit_i387_cw_initialization): Do not handle I387_CW_MASK_PM.
6554 2018-09-25 Jakub Jelinek <jakub@redhat.com>
6556 * vr-values.c (vr_values::vr_values): Initialize to_remove_edges and
6557 to_update_switch_stmts to vNULL instead of calling create on them
6560 2018-09-25 Richard Biener <rguenther@suse.de>
6562 PR tree-optimization/87402
6563 * tree-ssa-sccvn.c (SSA_VISITED): Remove unused function.
6564 (visit_phi): Re-instantiate handling of supposed to be VARYING
6565 but non-VARYING backedge value.
6567 2018-09-25 Richard Biener <rguenther@suse.de>
6570 * dwarf2out.c (struct sym_off_pair): New.
6571 (external_die_map): New global.
6572 (lookup_decl_die): When in LTO create DIEs lazily from the
6574 (lookup_block_die): New function, create DIEs lazily in LTO.
6575 (equate_block_to_die): New function.
6576 (dwarf2out_die_ref_for_decl): During WPA get the association
6577 from the external DIE map.
6578 (dwarf2out_register_external_die): Record mapping into the
6580 (maybe_create_die_with_external_ref): New function split out from
6581 DIE generation part of old dwarf2out_register_external_die.
6582 (add_abstract_origin_attribute): Do not return the DIE. When
6583 in LTO reference externals directly.
6584 (dwarf2out_abstract_function): When in LTO ignore calls for
6585 decls with external DIEs (already present abstract instances).
6586 (gen_call_site_die): Adjust.
6587 (add_high_low_attributes): Likewise.
6588 (gen_lexical_block_die): Likewise.
6589 (gen_inlined_subroutine_die): Likewie.
6590 (gen_block_die): Likewise.
6591 (dwarf2out_inline_entry): Likewise.
6592 (dwarf2out_early_finish): In LTRANS phase create DW_TAG_imported_unit
6595 2018-09-25 Martin Liska <mliska@suse.cz>
6597 * ipa-fnsummary.c (estimate_node_size_and_time): Scale by two
6598 integers and not by a float value.
6600 2018-09-25 Martin Liska <mliska@suse.cz>
6603 * dbgcnt.c (dbg_cnt_process_single_pair): Return false
6605 * dumpfile.c (dump_enable_all): Remove extra parenthesis.
6606 * gcov-tool.c: Declare the function with ATTRIBUTE_NORETURN.
6607 * godump.c (go_format_type): Remove extra parenthesis.
6609 2018-09-25 Martin Liska <mliska@suse.cz>
6611 * alias.c (set_dest_equal_p): Remove unused function.
6612 * config/i386/i386.c (def_builtin_pure2): Likewise.
6613 * diagnostic-show-locus.c (class layout): Remove
6615 (layout::layout): Likewise here.
6616 * dump-context.h (class temp_dump_context): Likewise.
6617 * dwarf2out.c (add_AT_fde_ref): Remove unused function.
6618 (add_AT_loclistsptr): Likewise.
6619 (add_AT_offset): Likewise.
6620 (get_AT_hi_pc): Likewise.
6621 (is_comdat_die): Likewise.
6622 (type_is_enum): Likewise.
6623 (ceiling): Likewise.
6624 (add_AT_vms_delta): Likewise.
6625 (is_class_die): Likewise.
6626 * edit-context.c (class line_event): Remove unused field.
6627 * graphite-sese-to-poly.c (tree_int_to_gmp): Remove
6629 * ipa-cp.c (ipa_get_vr_lat): Likewise.
6630 * lra-constraints.c (ok_for_index_p_nonstrict): Likewise.
6631 (ok_for_base_p_nonstrict): Likewise.
6632 * tree-chrec.c (is_not_constant_evolution): Likewise.
6633 (chrec_fold_poly_cst): Likewise.
6634 * tree-if-conv.c (has_pred_critical_p): Likewise.
6635 * tree-ssa-coalesce.c (print_exprs): Likewise.
6636 * tree-ssa-pre.c (bitmap_set_contains_expr): Likewise.
6637 * tree-ssa-uninit.c (is_and_or_or_p): Likewise.
6638 * tree-vrp.c (value_ranges_intersect_p): Likewise.
6639 (value_range_nonnegative_p): Likewise.
6641 2018-09-25 Martin Liska <mliska@suse.cz>
6643 * config/powerpcspe/powerpcspe.c (rs6000_output_function_epilogue):
6644 Do not handle "GNU Pascal".
6645 * config/rs6000/rs6000.c (rs6000_output_function_epilogue):
6647 * config/sparc/sparc.c (sparc_pass_by_reference): Remove Pascal
6648 from documentation. Likewise.
6649 * dbxout.c (dbxout_range_type): Likewise.
6650 * doc/cpp.texi: Likewise.
6651 * doc/extend.texi: Likewise.
6652 * doc/frontends.texi: Likewise.
6653 * doc/invoke.texi: Remove Pascal entry.
6654 * tree.def (CLEANUP_POINT_EXPR): Likewise.
6655 * doc/rtl.texi (MODE_FUNCTION): Remove not used entry.
6657 2018-09-25 Martin Liska <mliska@suse.cz>
6660 * doc/invoke.texi: Document all parameters and remove default
6663 2018-09-25 Ilya Leoshkevich <iii@linux.ibm.com>
6666 * rtl.c (rtx_code_size): Take into account that EQ_ATTR_ALT
6667 contains HOST_WIDE_INTs when computing its size.
6669 2018-09-24 Jim Wilson <jimw@sifive.com>
6672 * config/riscv/riscv.h (STACK_BOUNDARY): Test riscv_abi == ABI_ILP32E
6674 (ABI_STACK_BOUNDARY, MAX_ARGS_IN_REGISTERS): Likewise.
6676 2018-09-24 Andrew Pinski <apinski@marvell.com>
6678 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Don't
6679 access prev before checking it for NULLness in the
6680 AARCH64_FUSE_CMP_BRANCH case.
6682 2018-09-24 H.J. Lu <hongjiu.lu@intel.com>
6685 * config/i386/i386.c (rest_of_insert_endbranch): Set
6686 endbr_queued_at_entrance to true and don't insert ENDBR if
6687 x86_function_profiler will be called.
6688 (x86_function_profiler): Insert ENDBR if endbr_queued_at_entrance
6690 * config/i386/i386.h (machine_function): Add
6691 endbr_queued_at_entrance.
6693 2018-09-24 Ilya Leoshkevich <iii@linux.ibm.com>
6695 * genattrtab.c (mk_attr_alt): Use alternative_mask.
6696 (attr_rtx_1): Adjust caching to match the new EQ_ATTR_ALT field
6698 (check_attr_test): Use alternative_mask.
6699 (get_attr_value): Likewise.
6700 (compute_alternative_mask): Use alternative_mask and XWINT.
6701 (make_alternative_compare): Use alternative_mask.
6702 (attr_alt_subset_p): Use XWINT.
6703 (attr_alt_subset_of_compl_p): Likewise.
6704 (attr_alt_intersection): Use alternative_mask and XWINT.
6705 (attr_alt_union): Likewise.
6706 (attr_alt_complement): Use HOST_WIDE_INT and XWINT.
6707 (mk_attr_alt): Use alternative_mask and HOST_WIDE_INT.
6708 (simplify_test_exp): Use alternative_mask and XWINT.
6709 (write_test_expr): Use alternative_mask and XWINT, adjust bit
6710 number calculation to support 64 bits. Generate code that
6711 checks 64-bit masks.
6712 (main): Use alternative_mask.
6713 * rtl.def (EQ_ATTR_ALT): Change field types from ii to ww.
6715 2018-09-24 Ilya Leoshkevich <iii@linux.ibm.com>
6718 * config/s390/s390.c (s390_emit_epilogue): Do not use PARALLEL
6719 RETURN+USE when returning via %r14.
6721 2018-09-24 Martin Liska <mliska@suse.cz>
6723 * gcov.c (output_lines): Print colorization legend
6724 for both flag_use_colors and flag_use_hotness_colors.
6727 2018-09-24 Martin Liska <mliska@suse.cz>
6729 * coverage.c (get_coverage_counts): Use warning_at
6730 with current_function_decl location. Use %qD in warning
6733 2018-09-24 Martin Liska <mliska@suse.cz>
6735 * memory-block.h (memory_block_pool::release): Annotate with
6736 valgrind that the memory is not accessible.
6738 2018-09-24 Martin Liska <mliska@suse.cz>
6741 * asan.c: Make asan_handled_variables extern.
6743 * cfgexpand.c (expand_stack_vars): Make sure
6744 a representative is unpoison if another
6745 variable in the partition is handled by
6746 use-after-scope sanitization.
6748 2018-09-24 Richard Biener <rguenther@suse.de>
6750 PR tree-optimization/63155
6751 * tree-ssa-propagate.c (add_ssa_edge): Avoid adding PHIs to
6752 the worklist when the edge of the respective argument isn't
6755 2018-09-23 Uros Bizjak <ubizjak@gmail.com>
6757 * config/i386/i386.h (enum reg_class): Rename MASK_REGS to
6758 ALL_MASK_REGS and MASK_EVEX_REGS to MASK_REGS.
6759 (MASK_CLASS_P): Update for rename.
6760 (MAYBE_MASK_CLASS_P): Ditto.
6761 (REG_CLASS_NAMES): Update.
6762 (REG_CLASS_CONTENT): Update.
6763 * config/i386/i386.c (regclass_map): Update for MASK_REG
6764 and ALL_MASK_REGS rename.
6765 * config/i386/constraints.md (Yk): Update for rename.
6768 2018-09-23 Uros Bizjak <ubizjak@gmail.com>
6770 * config/i386/i386.h (enum reg_class): Remove
6771 EVEX_SSE_REGS and MOD4_SSE_REGS.
6772 (REG_CLASS_NAMES): Update.
6773 (REG_CLASS_CONTENT): Update.
6774 * config/i386/i386.c (regclass_map): Declare AVX-512 SSE
6775 registers as ALL_SSE_REGS.
6776 (ix86_additional_allocno_class_p): Remove.
6777 (TARGET_ADDITIONAL_ALLOCNO_CLASS_P): Remove.
6778 (ix86_register_priority): Lower priority of EVEX SSE registers.
6779 Use IN_RANGE macro where appropriate.
6780 (ix86_hard_regno_mode_ok): Merge AVX-5124FMAPS and
6781 AVX-5124VNNIW checks.
6782 (ix86_modes_tieable_p): Tie 512-bit SSE modes.
6783 * config/i386/sse.md (avx5124fmaddps_4fmaddps)
6784 (avx5124fmaddps_4fmaddps_mask, avx5124fmaddps_4fmaddps_maskz)
6785 (avx5124fmaddps_4fmaddss, avx5124fmaddps_4fmaddss_mask)
6786 (avx5124fmaddps_4fmaddss_maskz, avx5124fmaddps_4fnmaddps)
6787 (avx5124fmaddps_4fnmaddps_mask, avx5124fmaddps_4fnmaddps_maskz)
6788 (avx5124fmaddps_4fnmaddss, avx5124fmaddps_4fnmaddss_mask)
6789 (avx5124fmaddps_4fnmaddss_maskz, avx5124vnniw_vp4dpwssd)
6790 (avx5124vnniw_vp4dpwssd_mask, avx5124vnniw_vp4dpwssd_maskz)
6791 (avx5124vnniw_vp4dpwssds, avx5124vnniw_vp4dpwssds_mask)
6792 (avx5124vnniw_vp4dpwssds_maskz): Use "v" instead of "Yh" constraint.
6793 * config/i386/constraints.md (Yh): Remove.
6795 2018-09-23 Uros Bizjak <ubizjak@gmail.com>
6797 * config/i386/i386.c (regclass_map): Declare integer REX registers
6800 2018-09-23 Gerald Pfeifer <gerald@pfeifer.com>
6802 * doc/service.texi (Service): Switch the fsf.org link to https.
6804 2018-09-22 Chung-Ju Wu <jasonwucj@gmail.com>
6807 * config/nds32/nds32.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
6808 Define to speculation_safe_value_not_needed.
6810 2018-09-21 Florian Weimer <fweimer@redhat.com>
6813 * doc/extend.texi (Common Function Attributes): Mention that
6814 noreturn suppresses tail call optimization.
6816 2018-09-21 Jeff Law <law@redhat.com>
6818 * gimple-ssa-evrp.c (evrp_dom_walker::cleanup): Call
6819 vr_values::cleanup_edges_and_switches.
6820 * tree-vrp.c (to_remove_edges, to_update_switch_stmts): Moved into
6822 (identify_jump_threads): Remove EDGE_IGNORE handling.
6823 (execute_vrp): Move handling of to_remove_edges and
6824 to_update_switch_stmts into vr_values class member functions.
6825 * tree-vrp.h (switch_update, to_remove_edges): Remove declarations.
6826 (to_update_switch_stmts): Likewise.
6827 * vr-values.c: Include cfghooks.h.
6828 (vr_values::vr_values): Initialize to_remove_edges and
6829 to_update_switch_stmts.
6830 (vr_values::~vr_values): Verify to_remove_edges and
6831 to_update_switch_stmts are empty.
6832 (vr_values::simplify_switch_using_ranges): Set EDGE_IGNORE as needed.
6833 (vr_values::cleanup_edges_and_switches): New member function.
6834 * vr-values.h (vr_values): Add cleanup_edges_and_switches member
6835 function. Add new data members.
6837 2018-09-21 David Malcolm <dmalcolm@redhat.com>
6839 PR tree-optimization/87309
6840 * dumpfile.c (dump_context::begin_scope): Filter the dump_loc
6841 calls with pflags and alt_flags.
6842 (selftest::test_capture_of_dump_calls): Add test of interaction of
6843 MSG_OPTIMIZED_LOCATIONS with AUTO_DUMP_SCOPE.
6845 2018-09-21 Olivier Hainque <hainque@adacore.com>
6847 * config.gcc: Factorize and comment inclusion of vxworks-dummy.h.
6849 2018-09-21 Olivier Hainque <hainque@adacore.com>
6851 * config/vxworks.h (CLEAR_INSN_CACHE): #define to 1.
6853 2018-09-21 Olivier Hainque <hainque@adacore.com>
6855 * config/vxworks.h (VXWORKS_LIBGCC_SPEC): Remove -lc_internal.
6856 Merge block comment with the one ahead of VXWORKS_LIBS_RTP. Then:
6857 (VXWORKS_LIBS_RTP): Minor reordering.
6859 2018-09-21 Olivier Hainque <hainque@adacore.com>
6861 * config/vxworks.h (STARTFILE_PREFIX_SPEC): Define.
6862 (VXWORKS_LIBS_DIR_RTP): Remove definition and use.
6864 2018-09-21 Olivier Hainque <hainque@adacore.com>
6866 * config/vxworks.h (SIZE_TYPE): Account for TARGET_VXWORKS64.
6867 (PTRDIFF_TYPE): Likewise.
6869 2018-09-21 Olivier Hainque <hainque@adacore.com>
6871 * config.gcc: Enforce def of TARGET_VXWORKS64 to 1 from
6872 triplet, similar to support for VxWorks7.
6873 * config/vxworks-dummy.h: Provide a default definition
6874 of TARGET_VXWORKS64 to 0.
6876 2018-09-21 Olivier Hainque <hainque@adacore.com>
6878 * config/vxworks.h (TARGET_VXWORKS7): Move default definition ...
6879 * config/vxworks-dummy.h: here.
6881 2018-09-21 Olivier Hainque <hainque@adacore.com>
6883 * config.gcc: Prepend vxworks-dummy.h to tm_file for powerpc*
6885 2018-09-21 Shaokun Zhang <zhangshaokun@hisilicon.com>
6886 Bo Zhou <zbo.zhou@hisilicon.com>
6888 * config/aarch64/aarch64-cores.def (tsv110): New CPU.
6889 * config/aarch64/aarch64-tune.md: Regenerated.
6890 * doc/invoke.texi (AArch64 Options/-mtune): Add "tsv110".
6891 * config/aarch64/aarch64.c (tsv110_tunings): New tuning table.
6892 * config/aarch64/aarch64-cost-tables.h: Add "tsv110" extra costs.
6894 2018-09-21 Andrew Stubbs <ams@codesourcery.com>
6895 Julian Brown <julian@codesourcery.com>
6897 * builtins.c (get_builtin_sync_mem): Handle address spaces.
6899 2018-09-21 Eric Botcazou <ebotcazou@adacore.com>
6901 * config/rs6000/rs6000.c (rs6000_function_ok_for_sibcall): Return false
6902 if the call takes a static chain.
6904 2018-09-21 Martin Liska <mliska@suse.cz>
6906 * auto-profile.c (autofdo_source_profile::read): Do not
6908 (read_profile): Do not add working sets.
6909 (read_autofdo_file): Remove sum_all.
6910 (afdo_callsite_hot_enough_for_early_inline): Remove const
6912 * coverage.c (struct counts_entry): Remove gcov_summary.
6913 (read_counts_file): Read new GCOV_TAG_OBJECT_SUMMARY,
6914 do not support GCOV_TAG_PROGRAM_SUMMARY.
6915 (get_coverage_counts): Remove summary and expected
6917 * coverage.h (get_coverage_counts): Likewise.
6918 * doc/gcov-dump.texi: Remove -w option.
6919 * gcov-dump.c (dump_working_sets): Remove.
6920 (main): Do not support '-w' option.
6921 (print_usage): Likewise.
6922 (tag_summary): Likewise.
6923 * gcov-io.c (gcov_write_summary): Do not dump
6925 (gcov_read_summary): Likewise.
6926 (gcov_histo_index): Remove.
6927 (gcov_histogram_merge): Likewise.
6928 (compute_working_sets): Likewise.
6929 * gcov-io.h (GCOV_TAG_OBJECT_SUMMARY): Mark
6931 (GCOV_TAG_PROGRAM_SUMMARY): Mark it obsolete.
6932 (GCOV_TAG_SUMMARY_LENGTH): Adjust.
6933 (GCOV_HISTOGRAM_SIZE): Remove.
6934 (GCOV_HISTOGRAM_BITVECTOR_SIZE): Likewise.
6935 (struct gcov_summary): Simplify rapidly just
6936 to runs and sum_max fields.
6937 (gcov_histo_index): Remove.
6938 (NUM_GCOV_WORKING_SETS): Likewise.
6939 (compute_working_sets): Likewise.
6940 * gcov-tool.c (print_overlap_usage_message): Remove
6941 trailing empty line.
6942 * gcov.c (read_count_file): Read GCOV_TAG_OBJECT_SUMMARY.
6943 (output_lines): Remove program related line.
6944 * ipa-profile.c (ipa_profile): Do not consider GCOV histogram.
6945 * lto-cgraph.c (output_profile_summary): Do not stream GCOV
6947 (input_profile_summary): Do not read it.
6948 (merge_profile_summaries): And do not merge it.
6949 (input_symtab): Do not call removed function.
6950 * modulo-sched.c (sms_schedule): Do not print sum_max.
6951 * params.def (HOT_BB_COUNT_FRACTION): Reincarnate param that was
6952 removed when histogram method was invented.
6953 (HOT_BB_COUNT_WS_PERMILLE): Mention that it's used only in LTO
6955 * postreload-gcse.c (eliminate_partially_redundant_load): Fix
6957 * predict.c (get_hot_bb_threshold): Use HOT_BB_COUNT_FRACTION
6958 and dump selected value.
6959 * profile.c (add_working_set): Remove.
6960 (get_working_sets): Likewise.
6961 (find_working_set): Likewise.
6962 (get_exec_counts): Do not work with working sets.
6963 (read_profile_edge_counts): Do not inform as sum_max is removed.
6964 (compute_branch_probabilities): Likewise.
6965 (compute_value_histograms): Remove argument for call of
6966 get_coverage_counts.
6967 * profile.h: Do not make gcov_summary const.
6969 2018-09-21 Monk Chiang <sh.chiang04@gmail.com>
6971 * config.gcc (nds32*-*-*): Set TARGET_DEFAULT_TLSDESC_TRAMPOLINE=0.
6973 2018-09-21 Eric Botcazou <ebotcazou@adacore.com>
6975 PR tree-optimization/86990
6976 * gimple-ssa-store-merging.c (imm_store_chain_info:coalesce_immediate):
6977 Check that the entire merged store group is made of constants only for
6980 2018-09-20 Allan Sandfeld Jensen <allan.jensen@qt.io>
6982 * gcc.c (LINK_COMMAND_SPEC): Handle -r like -nostdlib.
6983 (VTABLE_VERIFICATION_SPEC): Likewise.
6984 (SANITIZER_EARLY_SPEC): Likewise.
6985 (SANITIZER_SPEC): Likewise.
6986 * config/darwin.h (LINK_COMMAND_SPEC): Likewise.
6987 * doc/invoke.texi (Link Options): Document -r.
6989 2018-09-20 Richard Biener <rguenther@suse.de>
6992 * gimplify.c (gimplify_expr): Retain alignment of
6993 addressable lvalue in dereference.
6995 2018-09-20 Alexandre Oliva <aoliva@redhat.com>
6998 * configure.ac: Check for .loc is_stmt support.
6999 * configure, config.in: Rebuilt.
7000 * dwarf2out.c (dwarf2out_source_line): Skip is_stmt
7003 2018-09-20 Segher Boessenkool <segher@kernel.crashing.org>
7005 * config/rs6000/rs6000.opt (misel=no, misel=yes): Delete.
7006 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -misel=yes and
7009 2018-09-20 Segher Boessenkool <segher@kernel.crashing.org>
7011 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete
7013 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete
7016 2018-09-20 Marek Polacek <polacek@redhat.com>
7018 * doc/invoke.texi: Add -Wno-init-list-lifetime to C++ Language Options.
7020 2018-09-20 Richard Sandiford <richard.sandiford@arm.com>
7022 PR tree-optimization/87288
7023 * tree-vect-loop.c (vect_analyze_loop_2): Take PEELING_FOR_GAPS
7024 into account when determining PEELING_FOR_NITERS.
7026 2018-09-20 Richard Sandiford <richard.sandiford@arm.com>
7028 PR tree-optimization/86877
7029 * tree-vect-loop.c (vect_analyze_loop_2): Call
7030 vect_verify_datarefs_alignment.
7032 2018-09-19 Marek Polacek <polacek@redhat.com>
7034 * doc/invoke.texi: Document -Wclass-conversion.
7036 2018-09-19 John David Anglin <danglin@gcc.gnu.org>
7038 * config/pa/pa.c (pa_adjust_priority): Delete.
7039 (TARGET_SCHED_ADJUST_PRIORITY): Delete define.
7041 * config/pa/pa.md (atomic_storeqi): Restore deleted expander.
7042 (atomic_storehi): Likewise.
7043 (atomic_storesi): Likewise.
7044 (atomic_loaddi): Restore compare and swap exchange loop code.
7046 2018-09-19 Segher Boessenkool <segher@kernel.crashing.org>
7048 PR rtl-optimization/86902
7049 * combine.c (try_combine): When changing the CC mode used, don't change
7050 an unrelated mode in other_insn to that new CC mode.
7052 2018-09-19 David Malcolm <dmalcolm@redhat.com>
7054 * tree-data-ref.c (runtime_alias_check_p): Use formatted printing
7055 with %T in place of calls to dump_generic_expr.
7056 (prune_runtime_alias_test_list): Likewise.
7057 (create_runtime_alias_checks): Likewise.
7058 * tree-vect-data-refs.c (vect_check_nonzero_value): Likewise.
7059 (vect_analyze_data_ref_dependence): Likewise.
7060 (vect_slp_analyze_data_ref_dependence): Likewise.
7061 (vect_record_base_alignment): Likewise. Use %G in place of call
7062 to dump_gimple_stmt.
7063 (vect_compute_data_ref_alignment): Likewise.
7064 (verify_data_ref_alignment): Likewise.
7065 (vect_find_same_alignment_drs): Likewise.
7066 (vect_analyze_group_access_1): Likewise.
7067 (vect_analyze_data_ref_accesses): Likewise.
7068 (dependence_distance_ge_vf): Likewise.
7069 (dump_lower_bound): Likewise.
7070 (vect_prune_runtime_alias_test_list): Likewise.
7071 (vect_find_stmt_data_reference): Likewise.
7072 (vect_analyze_data_refs): Likewise.
7073 (vect_create_addr_base_for_vector_ref): Likewise.
7074 (vect_create_data_ref_ptr): Likewise.
7075 * tree-vect-loop-manip.c (vect_set_loop_condition): Likewise.
7076 (vect_can_advance_ivs_p): Likewise.
7077 (vect_update_ivs_after_vectorizer): Likewise.
7078 (vect_gen_prolog_loop_niters): Likewise.
7079 (vect_prepare_for_masked_peels): Likewise.
7080 * tree-vect-loop.c (vect_determine_vf_for_stmt): Likewise.
7081 (vect_determine_vectorization_factor): Likewise.
7082 (vect_is_simple_iv_evolution): Likewise.
7083 (vect_analyze_scalar_cycles_1): Likewise.
7084 (vect_analyze_loop_operations): Likewise.
7085 (report_vect_op): Likewise.
7086 (vect_is_slp_reduction): Likewise.
7087 (check_reduction_path): Likewise.
7088 (vect_is_simple_reduction): Likewise.
7089 (vect_create_epilog_for_reduction): Likewise.
7090 (vect_finalize_reduction:): Likewise.
7091 (vectorizable_induction): Likewise.
7092 (vect_transform_loop_stmt): Likewise.
7093 (vect_transform_loop): Likewise.
7094 (optimize_mask_stores): Likewise.
7095 * tree-vect-patterns.c (vect_pattern_detected): Likewise.
7096 (vect_split_statement): Likewise.
7097 (vect_recog_over_widening_pattern): Likewise.
7098 (vect_recog_average_pattern): Likewise.
7099 (vect_determine_min_output_precision_1): Likewise.
7100 (vect_determine_precisions_from_range): Likewise.
7101 (vect_determine_precisions_from_users): Likewise.
7102 (vect_mark_pattern_stmts): Likewise.
7103 (vect_pattern_recog_1): Likewise.
7104 * tree-vect-slp.c (vect_get_and_check_slp_defs): Likewise.
7105 (vect_record_max_nunits): Likewise.
7106 (vect_build_slp_tree_1): Likewise.
7107 (vect_build_slp_tree_2): Likewise.
7108 (vect_print_slp_tree): Likewise.
7109 (vect_analyze_slp_instance): Likewise.
7110 (vect_detect_hybrid_slp_stmts): Likewise.
7111 (vect_detect_hybrid_slp_1): Likewise.
7112 (vect_slp_analyze_operations): Likewise.
7113 (vect_slp_analyze_bb_1): Likewise.
7114 (vect_transform_slp_perm_load): Likewise.
7115 (vect_schedule_slp_instance): Likewise.
7116 * tree-vect-stmts.c (vect_mark_relevant): Likewise.
7117 (vect_mark_stmts_to_be_vectorized): Likewise.
7118 (vect_init_vector_1): Likewise.
7119 (vect_get_vec_def_for_operand): Likewise.
7120 (vect_finish_stmt_generation_1): Likewise.
7121 (vect_check_load_store_mask): Likewise.
7122 (vectorizable_call): Likewise.
7123 (vectorizable_conversion): Likewise.
7124 (vectorizable_operation): Likewise.
7125 (vectorizable_load): Likewise.
7126 (vect_analyze_stmt): Likewise.
7127 (vect_is_simple_use): Likewise.
7128 (vect_get_vector_types_for_stmt): Likewise.
7129 (vect_get_mask_type_for_stmt): Likewise.
7130 * tree-vectorizer.c (increase_alignment): Likewise.
7132 2018-09-19 Andrew Stubbs <ams@codesourcery.com>
7134 * doc/rtl.texi: Adjust vec_select description.
7135 * simplify-rtx.c (simplify_binary_operation_1): Allow VEC_SELECT to use
7136 non-constant selectors.
7138 2018-09-19 Matthew Malcomson <matthew.malcomson@arm.com>
7140 * config/aarch64/aarch64-protos.h
7141 (aarch64_offset_9bit_signed_unscaled_p): New declaration.
7142 * config/aarch64/aarch64.md (arches): New "rcpc8_4" attribute value.
7143 (arch_enabled): Add check for "rcpc8_4" attribute value of "arch".
7144 * config/aarch64/aarch64.h (AARCH64_FL_RCPC8_4): New bitfield.
7145 (AARCH64_FL_FOR_ARCH8_4): Include AARCH64_FL_RCPC8_4.
7146 (AARCH64_FL_PROFILE): Move index so flags are ordered.
7147 (AARCH64_ISA_RCPC8_4): New flag.
7148 * config/aarch64/aarch64.c (offset_9bit_signed_unscaled_p): Renamed
7149 to aarch64_offset_9bit_signed_unscaled_p.
7150 * config/aarch64/atomics.md (atomic_store<mode>): Allow offset
7152 * config/aarch64/constraints.md (Ust): New constraint.
7153 * config/aarch64/predicates.md.
7154 (aarch64_9bit_offset_memory_operand): New predicate.
7155 (aarch64_rcpc_memory_operand): New predicate.
7157 2018-09-19 Eric Botcazou <ebotcazou@adacore.com>
7159 PR rtl-optimization/87361
7160 * rtlanal.c (nonzero_bits1): Revert accidental change.
7162 2018-09-19 Richard Biener <rguenther@suse.de>
7164 PR tree-optimization/87349
7165 PR tree-optimization/87342
7166 * tree-ssa-sccvn.c (do_rpo_vn): Iterate max_rpo computation.
7168 2018-09-18 Marek Polacek <polacek@redhat.com>
7170 P1064R0 - Allowing Virtual Function Calls in Constant Expressions
7171 * gimple-fold.c (gimple_get_virt_method_for_vtable): Adjust assert.
7173 2018-09-18 Segher Boessenkool <segher@kernel.crashing.org>
7175 * config/rs6000/rs6000.md: Remove old "Cygnus sibcall" comment.
7177 2018-09-18 Segher Boessenkool <segher@kernel.crashing.org>
7179 PR rtl-optimization/86882
7180 * rtlanal.c (reg_overlap_mentioned_p): Handle CLOBBER.
7182 2018-09-18 Uros Bizjak <ubizjak@gmail.com>
7184 * config/i386/i386.md (*<code>extend<mode>xf2): Macroize insn from
7185 *<code>extendsfxf2 and *<code>extenddfxf2 using MODEF mode iterator.
7187 2018-09-18 Jonathan Wakely <jwakely@redhat.com>
7190 * doc/invoke.texi (Link Options): Fix formatting and grammar.
7192 2018-09-18 Richard Biener <rguenther@suse.de>
7195 * tree-ssa-coalesce.c (tree_int_map_hasher): Remove.
7196 (compute_samebase_partition_bases): Likewise.
7197 (coalesce_ssa_name): Always use compute_optimized_partition_bases.
7198 (gimple_can_coalesce_p): Simplify.
7200 2018-09-18 Hans-Peter Nilsson <hp@bitrange.com>
7202 Handle a library implementation of ffs calling __builtin_ffs.
7203 * config/mmix/mmix.c (TARGET_INIT_LIBFUNCS): Override with...
7204 (mmix_init_libfuncs): New function: make __builtin_ffs expand
7207 2018-09-17 David Malcolm <dmalcolm@redhat.com>
7209 * diagnostic-show-locus.c (class layout_range): Add field
7211 (layout_range::layout_range): Add "original_idx" param and use it
7212 to initialize new field.
7213 (make_range): Use 0 for original_idx.
7214 (layout::layout): Pass in index to calls to
7215 maybe_add_location_range.
7216 (layout::maybe_add_location_range): Add param "original_idx" and
7217 pass it on to layout_range.
7218 (layout::print_any_labels): Pass on range->m_original_idx to
7220 (gcc_rich_location::add_location_if_nearby): Use 0 for
7222 * gcc-rich-location.h (text_range_label::get_text): Update for new
7224 (range_label_for_type_mismatch::get_text): Likewise.
7226 2018-09-17 Uros Bizjak <ubizjak@gmail.com>
7228 * config/i386/i386.c (ix86_emit_i387_log1p): Emit fldln2 earlier.
7230 2018-09-17 David Malcolm <dmalcolm@redhat.com>
7232 * gimple-ssa-sprintf.c (fmtwarn): Update for introduction of
7233 format_string_diagnostic_t.
7234 (fmtwarn_n): Likewise.
7235 * substring-locations.c
7236 (format_string_diagnostic_t::format_string_diagnostic_t) New ctor.
7237 (format_warning_n_va): Convert to...
7238 (format_string_diagnostic_t::emit_warning_n_va): ...this.
7239 (format_warning_va): Convert to...
7240 (format_string_diagnostic_t::emit_warning_va): ...this.
7241 (format_warning_at_substring): Convert to...
7242 (format_string_diagnostic_t::emit_warning): ...this.
7243 (format_warning_at_substring_n): Convert to...
7244 (format_string_diagnostic_t::emit_warning_n): ...this.
7245 * substring-locations.h (class format_string_diagnostic_t): New
7247 (format_warning_va): Convert to
7248 format_string_diagnostic_t::emit_warning_va.
7249 (format_warning_n_va): Convert to
7250 format_string_diagnostic_t::emit_warning_n_va.
7251 (format_warning_at_substring): Convert to
7252 format_string_diagnostic_t::emit_warning.
7253 (format_warning_at_substring_n): Convert to
7254 format_string_diagnostic_t::emit_warning_n.
7256 2018-09-17 Cesar Philippidis <cesar@codesourcery.com>
7257 Bernd Schmidt <bernds_cb1@t-online.de>
7259 * config/nvptx/nvptx.md (atomic_fetch_<logic><mode>): Enable with
7262 2018-09-17 Uros Bizjak <ubizjak@gmail.com>
7264 * config/i386/i386.md (truncxf<mode>2_i387_noop_unspec): Change
7265 operand 0 predicate to nonimmediate operand.
7266 (rint<mode>2_frndint): Remove insn pattern.
7267 (rint<mode>2): Change operand 1 predicate to general_operand.
7268 Extend operand 1 to XFmode and generate rintxf2 insn.
7269 (frndintxf2_<rounding>): Rename from frndint<mode>2_<rounding>.
7270 Do not use X87MODEF mode macro.
7271 (frndintxf2_<rounding>_i387): Rename from
7272 frndint<mode>2_<rounding>_i387. Do not use X87MODEF mode macro.
7273 (<rounding_insn><mode>2): For non-SSE modes, extend operand 1
7274 to XFmode and generate significandxf3 insn.
7276 2018-09-17 Richard Biener <rguenther@suse.de>
7278 PR tree-optimization/87328
7279 * tree-ssa-sccvn.c (process_bb): Remove assertion about not
7280 visiting unexecutable backedges when not iterating.
7281 (do_rpo_vn): Mark all edges not executable even when not
7284 2018-09-17 Martin Jambor <mjambor@suse.cz>
7287 * doc/invoke.texi (Warning Options): Likewise.
7289 2018-09-17 Richard Biener <rguenther@suse.de>
7291 PR tree-optimization/87301
7292 * tree-ssa-sccvn.c (eliminate_dom_walker::eliminate_cleanup): Properly
7293 clean EH info from leftover copy assignments.
7295 2018-09-17 Martin Liska <mliska@suse.cz>
7297 PR gcov-profile/85871
7298 * gcov.c (output_intermediate_file): Fix out of bounds
7301 2018-09-17 Vineet Gupta <vgupta@synopsys.com>
7303 * config/arc/arc.c: Object attributes for core4 not reflected
7305 * config/arc/arc.h: Don't restrict DBNZ to core3 (core4 includes
7308 2018-09-17 Alexey Brodkin <abrodkin@synopsys.com>
7310 * config/arc/linux.h (LINK_EH_SPEC): Add missing space.
7312 2018-09-17 Martin Liska <mliska@suse.cz>
7314 * doc/gcov.texi: Document new option --use-hotness-colors.
7315 * gcov.c (struct source_info): Declare new field.
7316 (source_info::source_info): Set default for maximum_count.
7317 (print_usage): Add new -q option.
7318 (process_args): Process it.
7319 (accumulate_line_info): Save src->maximum_count.
7320 (output_line_beginning): Make color line number if
7321 flag_use_hotness_colors is set.
7322 (output_line_details): Pass default argument value.
7323 (output_lines): Pass src->maximum_count.
7325 2018-09-17 Martin Liska <mliska@suse.cz>
7327 * common/config/i386/i386-common.c (ix86_get_valid_option_values):
7328 Use processor_names table.
7329 * config/i386/i386.c (ix86_default_align): Use
7330 processor_cost_table for alignment values.
7331 (ix86_option_override_internal): Use processor_names.
7332 (ix86_function_specific_print): Likewise.
7333 * config/i386/i386.h (struct processor_costs):
7334 Add alignment values.
7335 (struct ptt): Remove and replace with const char *.
7336 * config/i386/x86-tune-costs.h (struct processor_costs):
7337 Declare default alignments for all costs.
7339 2018-09-17 Aldy Hernandez <aldyh@redhat.com>
7341 * tree-vrp.c (extract_range_from_unary_expr): Do not special case
7342 symbolics or VR_VARYING ranges for ABS_EXPR.
7343 * wide-int-range.cc (wide_int_range_abs): Return positive numbers
7344 when range will wrap.
7346 2018-09-15 Eric Botcazou <ebotcazou@adacore.com>
7349 * cfgexpand.c (expand_gimple_basic_block): Be prepared for a BARRIER
7350 before and after a JUMP_TABLE_DATA.
7352 2018-09-14 John David Anglin <danglin@gcc.gnu.org>
7355 * dojump.c (do_compare_and_jump): Canonicalize function pointers
7356 when one operand is a function pointer. Use POINTER_TYPE_P and
7357 FUNC_OR_METHOD_TYPE_P.
7358 * expr.c (do_store_flag): Use POINTER_TYPE_P and FUNC_OR_METHOD_TYPE_P.
7359 * fold-const.c (build_range_check): Likewise.
7360 * match.pd (simple_comparison): Likewise.
7362 2018-09-14 David Malcolm <dmalcolm@redhat.com>
7365 * spellcheck.c (get_edit_distance_cutoff): New function.
7366 (selftest::test_edit_distance_unit_test_oneway): Rename to...
7367 (selftest::test_get_edit_distance_one_way): ...this.
7368 (selftest::test_get_edit_distance_unit): Rename to...
7369 (selftest::test_get_edit_distance_both_ways): ...this.
7370 (selftest::test_edit_distances): Move tests to this new function,
7371 and test some more pairs of strings. Update for above renaming.
7372 (selftest::get_old_cutoff): New function.
7373 (selftest::test_get_edit_distance_cutoff): New function.
7374 (selftest::assert_suggested_for): New function.
7375 (ASSERT_SUGGESTED_FOR): New macro.
7376 (selftest::assert_not_suggested_for): New function.
7377 (ASSERT_NOT_SUGGESTED_FOR): New macro.
7378 (selftest::test_suggestions): New function.
7379 (selftest::spellcheck_c_tests): Move test_get_edit_distance_unit
7380 tests to selftest::test_edit_distances and call it. Add calls to
7381 selftest::test_get_edit_distance_cutoff and
7382 selftest::test_suggestions.
7383 * spellcheck.h (get_edit_distance_cutoff): New function declaration.
7384 (best_match::consider): Replace hard-coded cutoff calculation with
7386 (best_match::get_cutoff): New declaration.
7387 (best_match::get_best_meaningful_candidate): Likewise.
7389 2018-09-14 Bernd Edlinger <bernd.edlinger@hotmail.de>
7391 * builtins.c (fold_builtin_strlen): Remove TODO comment.
7393 2018-09-14 Bernd Edlinger <bernd.edlinger@hotmail.de>
7396 2018-07-30 Bernd Edlinger <bernd.edlinger@hotmail.de>
7398 * tree-ssa-forwprop.c (simplify_builtin_call): Don't create a not NUL
7399 terminated string literal.
7401 2018-09-14 Martin Sebor <msebor@redhat.com>
7403 * builtins.c (unterminated_array): Handle ARRAY_REF.
7404 (expand_builtin_stpcpy_1): Detect unterminated char arrays.
7405 * builtins.h (unterminated_array): Declare extern.
7406 * gimple-fold.c (gimple_fold_builtin_stpcpy): Detect unterminated
7408 (gimple_fold_builtin_sprintf): Propagate NO_WARNING to transformed
7411 2018-09-14 Martin Sebor <msebor@redhat.com>
7412 Jeff Law <law@redhat.com>
7414 * builtins.c (unterminated_array): New.
7415 (expand_builtin_strcpy): Adjust.
7416 (expand_builtin_strcpy_args): Detect unterminated arrays.
7417 * gimple-fold.c (get_maxval_strlen): Add argument. Detect
7418 unterminated arrays.
7419 * gimple-fold.h (get_maxval_strlen): Add argument.
7420 (gimple_fold_builtin_strcpy): Detec unterminated arrays.
7422 * gimple-fold.c (get_range_strlen): Add argument.
7423 (get_maxval_strlen): Adjust.
7424 * gimple-fold.h (get_range_strlen): Add argument.
7426 2018-09-14 Wei Xiao <wei3.xiao@intel.com>
7428 * config/i386/movdirintrin.h: Fix copyright year.
7430 2018-09-14 Uros Bizjak <ubizjak@gmail.com>
7432 * reg-stack.c: Include regs.h.
7433 (replace_reg): Assert that mode is MODE_FLOAT or MODE_COMPLEX_FLOAT.
7434 (emit_pop_insn): Default pop insn mode to the reg_raw_mode of
7435 FIRST_STACK_REG, not DFmode.
7436 (emit_swap_insn): Default swap insn mode to the reg_raw_mode of
7437 FIRST_STACK_REG, not XFmode. Explicitly construct swap RTX.
7438 (change stack): Default register mode to the reg_raw_mode of
7439 FIRST_STACK_REG, not DFmode.
7440 * config/i386/i386.md (*swap<mode>): Remove insn pattern.
7441 (*swapxf): Rename from swapxf.
7443 2018-09-14 Carl Love <cel@us.ibm.com>
7445 * config/rs6000/emmintrin.h: Add _MM_SHUFFLE2.
7446 * config/rs6000/xmmintrin.h: Add _MM_SHUFFLE.
7448 2018-09-14 Segher Boessenkool <segher@kernel.crashing.org>
7451 * config/rs6000/rs6000.md (*mov<mode>_hardfloat64): Add Z to the Y
7454 2018-09-14 Sam Tebbs <sam.tebbs@arm.com>
7457 * config/aarch64/aarch64.md (*aarch64_bfxilsi_uxtw): Define.
7459 2018-09-14 Jason Merrill <jason@redhat.com>
7461 Fix --enable-gather-detailed-mem-stats.
7462 * hash-table.c (hash_table_usage): Change from variable to function.
7463 * hash-table.h: Adjust.
7464 * Makefile.in: Add missing dependencies on hash-table.h.
7466 2018-09-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7468 PR tree-optimization/87259
7470 (pass_cse_reciprocals::execute): Run optimize_recip_sqrt after
7471 execute_cse_reciprocals_1 has tried transforming.
7473 2018-09-14 Aldy Hernandez <aldyh@redhat.com>
7475 * tree-vrp.c (extract_range_from_binary_expr_1): Normalize
7476 VR_VARYING for PLUS/MINUS_EXPR.
7478 2018-09-14 Ilya Leoshkevich <iii@linux.ibm.com>
7480 * config/s390/s390-passes.def (INSERT_PASS_BEFORE): Improve
7483 2018-09-14 Richard Biener <rguenther@suse.de>
7486 * tree-ssa-coalesce.c (ssa_conflicts_merge): Remove conflict
7487 bits for the merged partition.
7489 2018-09-13 Martin Sebor <msebor@redhat.com>
7490 Bernd Edlinger <bernd.edlinger@hotmail.de>
7492 * builtins.h (c_srlen): Add argument.
7493 * builtins.c (warn_string_no_nul): New function.
7494 (c_strlen): Add argument and use it. Update recursive calls.
7495 Pass DECL argument to string_constant to get info on non
7496 terminated strings. Update *NONSTR as needed.
7497 (fold_builtin_strlen): Add argument to calls to c_strlen.
7498 Warn for unterminated arrays.
7499 (warn_string_no_null): Add prototype.
7500 * expr.c (string_constant): Update arguments. Update recursive
7501 calls appropriately. Detect missing NUL terminator and outermost
7502 declaration its missing in.
7503 Improve checks for arrays with nonzero lower bound or elements
7504 that are not a single byte. Simplify offset computation.
7505 Simplify checks for non-NUL terminated strings.
7506 * gimple-fold.c (get_range_strlen): Add argument to c_strlen call.
7507 * gimple-ssa-sprintf.c (get_string_length): Remove unnecessary code.
7509 2018-09-13 Bernd Edlinger <bernd.edlinger@hotmail.de>
7511 * builtins.c (c_strlen): Handle not zero terminated STRING_CSTs
7513 * fold-const.c (c_getstr): Fix function comment. Remove unused third
7514 argument. Fix range checks.
7515 * fold-const.h (c_getstr): Adjust protoype.
7516 * gimple-fold.c (gimple_fold_builtin_memory_op): Avoid folding when
7517 string is constant but contains no NUL byte.
7519 * expr.c (string_constant): Adjust function comment.
7520 Remove bogus check for zero termination.
7522 * fold-const.c (c_getstr): Clamp STRING_LENGTH to STRING_SIZE.
7524 * varasm.c (compare_constant): Compare type size of STRING_CSTs.
7525 (get_constant_size): Don't make STRING_CSTs larger than they are.
7526 (check_string_literal): New check function for STRING_CSTs.
7527 (output_constant): Use it.
7529 2018-09-13 Eric Botcazou <ebotcazou@adacore.com>
7532 * config/visium/visium.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Define.
7534 2018-09-13 Richard Biener <rguenther@suse.de>
7536 PR tree-optimization/87263
7537 * tree-ssa-sccvn.c (visit_phi): Revert some earlier changes.
7538 (struct unwind_state): Add max_rpo field.
7539 (do_rpo_vn): Allow up-to-date loop state to be used when not iterating.
7540 Compute max_rpo, the max RPO number a block can be backwards reached
7541 from. Re-write non-iterating mode to a RPO ordered worklist approach,
7542 separating it from the iterating mode.
7544 2018-09-13 Vlad Lazar <vlad.lazar@arm.com>
7546 * haifa-sched.c (rank_for_schedule): Schedule by INSN_COST.
7547 (rfs_decision): New scheduling decision.
7549 2018-09-13 Richard Biener <rguenther@suse.de>
7552 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Fix assert.
7553 (vn_nary_op_insert_pieces_predicated): Do not write useless
7554 valid_dominated_by_p entry outside of the allocated storage.
7556 2018-09-13 Omar Sandoval <osandov@osandov.com>
7557 Tom de Vries <tdevries@suse.de>
7560 * dwarf2out.c (is_c): New function.
7561 (add_subscript_info): Add DW_AT_count of 0 for C zero-length arrays.
7563 2018-09-13 Sam Tebbs <sam.tebbs@arm.com>
7566 * config/aarch64/aarch64.md (*aarch64_bfxil):
7568 * config/aarch64/constraints.md (Ulc): Define.
7569 * config/aarch64/aarch64-protos.h (aarch64_high_bits_all_ones_p):
7571 * config/aarch64/aarch64.c (aarch64_high_bits_all_ones_p):
7574 2018-09-13 Vlad Lazar <vlad.lazar@arm.com>
7576 * config/aarch64/aarch64.h (TARGET_COMPUTE_FRAME_LAYOUT): Define.
7577 * config/aarch64/aarch64.c (aarch64_expand_prologue): Remove
7578 aarch64_layout_frame call.
7579 (aarch64_expand_epilogue): Likewise.
7580 (aarch64_initial_elimination_offset): Likewise.
7581 (aarch64_get_separate_components): Likewise.
7582 (aarch64_use_return_insn_p): Likewise.
7583 (aarch64_layout_frame): Remove unneeded check.
7585 2018-09-13 Jakub Jelinek <jakub@redhat.com>
7587 * configure.ac: Only append
7588 " : (reconfigured) $TOPLEVEL_CONFIGURE_ARGUMENTS" to
7589 gcc_config_arguments if it was never reconfigured or last reconfigure
7590 was with different arguments.
7591 * configure: Regenerated.
7593 2018-09-13 Jakub Jelinek <jakub@redhat.com>
7594 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7597 * expr.c (maybe_optimize_pow2p_mod_cmp): New function.
7598 (maybe_optimize_mod_cmp): Use it if integer_pow2p treeop1.
7600 2018-09-13 Jakub Jelinek <jakub@redhat.com>
7602 PR tree-optimization/87287
7603 * fold-const.c (fold_binary_loc) <case EQ_EXPR>: Move signed modulo
7604 X % C == 0 to X % (unsigned) C == 0 optimization to ...
7605 * match.pd (X % C == 0): ... here. New optimization.
7607 2018-09-12 Jakub Jelinek <jakub@redhat.com>
7610 * expr.h (maybe_optimize_mod_cmp): Declare.
7611 * expr.c (mod_inv): New function.
7612 (maybe_optimize_mod_cmp): New function.
7613 (do_store_flag): Use it.
7614 * cfgexpand.c (expand_gimple_cond): Likewise.
7616 2018-09-09 Cesar Philippidis <cesar@codesourcery.com>
7617 Julian Brown <julian@codesourcery.com>
7620 * gimplify.c (gimplify_scan_omp_clauses): Set
7621 target_firstprivatize_array_bases in OpenACC parallel and kernels
7622 region contexts. Remove GOMP_MAP_FIRSTPRIVATE_REFERENCE clauses from
7623 OpenACC data regions.
7625 2018-09-12 Uros Bizjak <ubizjak@gmail.com>
7627 * config/i386/i386.md (sqrt_extend<mode>xf3_i387): Remove.
7628 (sqrt<mode>2): Extend operand 1 to XFmode and generate sqrtxf3 insn.
7630 2018-09-12 Richard Biener <rguenther@suse.de>
7632 PR tree-optimization/87280
7633 * tree-ssa-sccvn.c (process_bb): Handle the case of executable
7634 edge but unreachable target.
7635 (do_rpo_vn): For conservatively handling a PHI only mark
7636 the backedge executable but not the block reachable.
7638 2018-09-12 Richard Biener <rguenther@suse.de>
7640 PR tree-optimization/87266
7641 * tree-ssa-sccvn.c (do_rpo_vn): Always iterate to not yet
7644 2018-09-12 Andreas Krebbel <krebbel@linux.ibm.com>
7646 * config/s390/s390.md (PFPO_RND_MODE_DFP, PFPO_RND_MODE_BFP): New
7648 ("trunc<BFP:mode><DFP_ALL:mode>2")
7649 ("trunc<DFP_ALL:mode><BFP:mode>2")
7650 ("extend<BFP:mode><DFP_ALL:mode>2")
7651 ("extend<DFP_ALL:mode><BFP:mode>2"): Set proper rounding mode
7652 according to the target operand type.
7654 2018-09-12 Jakub Jelinek <jakub@redhat.com>
7655 Andreas Krebbel <krebbel@linux.ibm.com>
7657 PR tree-optimization/86844
7658 * gimple-ssa-store-merging.c
7659 (imm_store_chain_info::coalesce_immediate): For overlapping stores, if
7660 there are any overlapping stores in between them, make sure they are
7661 also coalesced or we give up completely.
7663 2018-09-12 Jakub Jelinek <jakub@redhat.com>
7666 * fold-const.c (fold_ternary_loc) <case COND_EXPR>: Verify also that
7667 BIT_AND_EXPR's second operand is a power of two. Formatting fix.
7669 2018-09-12 Tom de Vries <tdevries@suse.de>
7671 * common.opt (gdescribe-dies): Add option.
7672 * dwarf2out.c (add_name_and_src_coords_attributes): Add description
7673 attribute for artifical and nameless decls.
7674 (dwarf2out_register_external_die): Add description attribute to
7675 external reference die.
7676 (add_desc_attribute): New functions.
7677 (gen_subprogram_die): Add description attribute to
7678 DW_TAG_call_site_parameter.
7679 * tree-pretty-print.c (print_generic_expr_to_str): New function.
7680 * tree-pretty-print.h (print_generic_expr_to_str): Declare.
7681 * doc/invoke.texi (@item Debugging Options): Add -gdescribe-dies and
7683 (@item -gdescribe-dies): Add.
7685 2018-09-12 Aldy Hernandez <aldyh@redhat.com>
7687 * tree-vrp.c (vrp_shift_undefined_p): Remove.
7688 (extract_range_from_binary_expr_1: Call
7689 wide_int_range_shift_undefined_p instead of vrp_shift_undefined_p.
7690 * wide-int-range.h (wide_int_range_shift_undefined_p): Do not
7693 2018-09-12 Aldy Hernandez <aldyh@redhat.com>
7695 * gimple-ssa-warn-alloca.c
7696 (alloca_type_and_limit::alloca_type_and_limit): Initialize limit
7697 field for ALLOCA_BOUND_*_LARGE.
7699 2018-09-11 Nathan Sidwell <nathan@acm.org>
7701 * gcc.c (load_specs, execute, run_attempt): Use %qs not '%s'.
7703 2018-09-11 Uros Bizjak <ubizjak@gmail.com>
7705 * reg-stack.c (subst_asm_stack_regs): Call replace_reg also
7706 for clobbers. Remove obsolete comment.
7708 2018-09-11 Uros Bizjak <ubizjak@gmail.com>
7710 * config/i386/i386.md (define_attr "type"): Remove mpxmov, mpxmk,
7711 mpxchk, mpxld and mpxst types.
7712 (define_attr length_immediate): Remove all processing of mpx types.
7713 (define_attr prefix_0f): Ditto.
7714 (define_attr memory): Ditto.
7716 2018-09-11 Uros Bizjak <ubizjak@gmail.com>
7718 * config/i386/i386.md (fyl2x_extend<mode>xf3_i387): Remove.
7719 (log<mode>2): Change operand 1 predicate to general_operand.
7720 Extend operand 1 to XFmode and generate logxf3 insn.
7721 (log10<mode>2): Change operand 1 predicate to general_operand.
7722 Extend operand 1 to XFmode and generate log10xf3 insn.
7723 (log2<mode>2): Change operand 1 predicate to general_operand.
7724 Extend operand 1 to XFmode and generate log2xf3 insn.
7725 (fyl2xp1_extend<mode>xf3_i387): Remove.
7726 (log1p<mode>2): Change operand 1 predicate to general_operand.
7727 Extend operand 1 to XFmode and generate log1pxf3 insn.
7728 (fxtract_extend<mode>xf3_i387): Remove.
7729 (logb<mode>2): Change operand 1 predicate to general_operand.
7730 Extend operand 1 to XFmode and generate logbxf3 insn.
7731 (ilogb<mode>2): Change operand 1 predicate to general_operand.
7732 Extend operand 1 to XFmode and generate fxtractxf3_i387 insn.
7733 (significand<mode>2): Change operand 1 predicate to general_operand.
7734 Extend operand 1 to XFmode and generate significandxf3 insn.
7736 2018-09-11 Nathan Sidwell <nathan@acm.org>
7738 * gcc.c (perror_with_name, pfatal_with_name): Delete.
7739 (load_specs): Use fatal_error.
7740 (DELETE_IF_ORDINARY, process_command): Use error.
7741 (execute, run_attempt): Use fatal_error.
7743 2018-09-11 Andrew Stubbs <ams@codesourcery.com>
7745 * diagnostic-core.h (sorry_at): New prototype.
7746 * diagnostic.c (sorry_at): New function.
7748 2018-09-11 Aldy Hernandez <aldyh@redhat.com>
7750 * tree-vrp.c (extract_range_from_binary_expr_1): Treat all divisions
7751 by zero as VR_UNDEFINED.
7753 2018-09-10 Uros Bizjak <ubizjak@gmail.com>
7755 * config/i386/i386.md (<sincos>xf2): Rename from *<sincos>xf2_i387.
7756 (*<sincos>_extend<mode>xf2_i387): Remove insn pattern.
7757 (<sincos>mode2): New expander.
7758 (sincos_extend<mode>xf3_i387): Remove insn pattern.
7759 (sincos -> sin, cos splitters): Remove splitter patterns.
7760 (sincos<mode>3): Change operand 2 predicate to general_operand.
7761 Extend operand 2 to XFmode and generate sincosxf3 insn.
7762 (fptanxf4_i387): Change mode of operands 0 and 3 to SFmode.
7763 Change operand 3 predicate to const1_operand.
7764 (fptan_extend<mode>xf4_i387): Remove insn pattern.
7765 (tanxf2): Update operands in the call to fptanxf4_i387.
7766 (tan<mode>2): Change operand 1 predicate to general_operand.
7767 Extend operand 1 to XFmode and generate tanxf3 insn.
7768 (atan2xf3): Rename from *fpatanxf3_i387.
7769 (fpatan_extend<mode>xf3_i387): Remove insn pattern.
7770 (atan2xf3): Remove expander.
7771 (atan2<mode<3): Change operand 1 and 2 predicates to general_operand.
7772 Extend operands 1 and 2 to XFmode and generate atan2xf3 insn.
7773 (atan<mode>2): Change operand 1 predicate to general_operand.
7774 Extend operand 1 to XFmode and generate atanxf3 insn.
7776 2018-09-10 Uros Bizjak <ubizjak@gmail.com>
7778 * config/i386/i386.md (x87/SSE constant load splitter): Use
7779 memory_operand instead of nonimmediate_operand for input operand
7782 2018-09-09 Uros Bizjak <ubizjak@gmail.com>
7784 * config/i386/i386.md (float partial SSE register stall splitter): Move
7785 splitter near its instruction pattern.
7786 (float_extend partial SSE register stall splitter): Ditto.
7787 (float_truncate partial SSE register stall splitter): Ditto.
7789 2018-09-09 Hans-Peter Nilsson <hp@bitrange.com>
7792 * config/mmix/mmix.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Redefine
7793 to speculation_safe_value_not_needed.
7796 * config/mmix/mmix.c (mmix_assemble_integer): Handle byte-size
7797 non-CONST_INT rtx:es using assemble_integer_with_op ".byte".
7798 (MMIX_CFUN_NEEDS_SAVED_EH_RETURN_ADDRESS): Don't call
7799 leaf_function_p, instead use has_hard_reg_initial_val.
7801 2018-09-09 Nathan Sidwell <nathan@acm.org>
7803 * gcc.h (pfatal_with_name): Don't declare here.
7804 * gcc.c (pfatal_with_name): Make static.
7806 2018-09-09 Xianmiao Qu <xianmiao_qu@c-sky.com>
7808 * config/csky/csky.md (*cskyv2_adddi3, *cskyv2_subdi3): Add
7811 2018-09-08 John David Anglin <danglin@gcc.gnu.org>
7813 PR rtl-optimization/85458
7814 * sel-sched.c (sel_target_adjust_priority): Allow backend adjust
7815 priority hook to reduce the priority of EXPR.
7817 2018-09-07 Uros Bizjak <ubizjak@gmail.com>
7819 * config/i386/i386.md (float<SWI48x:mode><MODEF:mode>2) Enable
7820 DImode for x87 on 32bit targets. Conditionally disable x87 modes
7821 with X87_ENABLE_FLOAT. Remove preparation code.
7822 (*float<SWI48:mode><MODEF:mode>2): Rename from
7823 *float<SWI48:mode><MODEF:mode>2_mixed. Handle x87, SSE and mixed
7824 math using "enabled" attribute.
7825 (*floatdi<MODEF:mode>2_i387): Rename from
7826 *float<SWI48x:mode><MODEF:mode>2_i387. Handle only DImode and
7827 enable for 32bit targets only.
7828 (floatdi<X87MODEF:mode>2_i387_with_xmm pre-reload splitter): New
7830 (floatdi<X87MODEF:mode>2_i387_with_xmm): Use register_operand
7831 as operand 1 predicate. Rewrite as define_insn_and_split.
7832 (floatdi<X87MODEF:mode>2_i387_with_xmm memory input splitter): Remove.
7834 2018-09-06 Uros Bizjak <ubizjak@gmail.com>
7836 * reg-stack.c (get_true_reg) <case FLOAT_TRUNCATE>: Reorder
7837 to fallthru to FLOAT case.
7839 2018-09-06 Will Schmidt <will_schmidt@vnet.ibm.com>
7842 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Update logic
7843 around folding of vec_sl to handle out of range shift values.
7845 2018-09-06 Uros Bizjak <ubizjak@gmail.com>
7847 * config/i386/i386.md (fix_trunc<mode>_fisttp_i387_1): Remove.
7848 Update callers to gen_fix_trunc<mode>_i387_fisttp
7849 (fix_trunc<mode>_i387_fisttp): Change operand 0 predicate to
7850 nonimmediate_operand.
7851 (fix_trunc<mode>_i387_fisttp_with_temp): Remove insn pattern
7852 and corresponding splitters.
7853 (*fix_trunc<mode>_i387_1): Always expand with fix_trunc<mode>_i387.
7854 (fix_truncdi_i387): Change operand 0 predicate to nonimmediate_operand.
7855 (fix_truncdi_i387_with_temp): Remove insn pattern
7856 and corresponding splitters.
7857 (fix_trunc<mode>_i387): Change operand 0 predicate to
7858 nonimmediate_operand.
7859 (fix_trunc<mode>_i387_with_temp): Remove insn pattern
7860 and corresponding splitters.
7861 (*fistdi2_1): Remove.
7863 (fistdi2_with_temp): Remove insn pattern and corresponding splitters.
7864 (lrintxfdi2): Remove expander. Reimplement as define_insn.
7865 (*fist<mode>2_1): Remove.
7866 (fist<mode>2): Ditto.
7867 (fist<mode>2_with_temp): Remove insn pattern and corresponding
7869 (lrintxf<mode>2): Remove expander. Reimplement as define_insn.
7870 (*fist<mode>2_<rounding>_1): Always expand with fist<mode>2_<rounding>.
7871 (fistdi2_<rounding>): Change operand 0 predicate to
7872 nonimmediate_operand.
7873 (fistdi2_<rounding>_with_temp): Remove insn pattern
7874 and corresponding splitters.
7875 (fist<mode>2_<rounding>): Change operand 0 predicate to
7876 nonimmediate_operand.
7877 (fist<mode>2_<rounding>_with_temp): Remove insn pattern
7878 and corresponding splitters.
7880 (*fixuns_trunc<mode>si2_avx512f_zext): Depend on TARGET_SSE_MATH.
7882 2018-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
7884 * varasm.c (output_constructor_regular_field): Check TYPE_SIZE_UNIT of
7887 2018-09-06 Will Schmidt <will_schmidt@vnet.ibm.com>
7889 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
7890 early gimple folding of vec_splat().
7891 * tree-vect-generic.c: Remove static from tree_vec_extract() definition.
7892 * gimple-fold.h: Add an extern define for tree_vec_extract().
7894 2018-09-06 Will Schmidt <will_schmidt@vnet.ibm.com>
7896 * config/rs6000/rs6000.c (fold_mergehl_helper): Add types_compatible_p
7897 wrappers around TREE_TYPE comparisons.
7899 2018-09-06 Ilya Leoshkevich <iii@linux.ibm.com>
7902 * config/s390/predicates.md: Add nonsym_memory_operand.
7903 * config/s390/s390.c (s390_legitimize_cs_operand): If operand
7904 contains a SYMBOL_REF, load it into an intermediate pseudo.
7905 (s390_emit_compare_and_swap): Legitimize operand.
7906 * config/s390/s390.md: Use the new nonsym_memory_operand
7907 with UNSPECV_CAS patterns.
7909 2018-09-06 Ilya Leoshkevich <iii@linux.ibm.com>
7912 * config/s390/s390-passes.def: New file.
7913 * config/s390/s390-protos.h (class rtl_opt_pass): Add forward
7915 (make_pass_s390_early_mach): Add declaration.
7916 * config/s390/s390.c (make_pass_s390_early_mach):
7917 (s390_option_override): Remove dynamic registration.
7918 * config/s390/t-s390: Add s390-passes.def.
7920 2018-09-06 Ilya Leoshkevich <iii@linux.ibm.com>
7922 * config/s390/s390.c (s390_decompose_constant_pool_ref):
7923 Remove UNSPEC_LTREL_BASE check.
7924 (annotate_constant_pool_refs): Likewise.
7925 (find_constant_pool_ref): Likewise.
7926 (find_ltrel_base): Removed.
7927 (replace_ltrel_base): Removed.
7928 (s390_mainpool_finish): Remove replace_ltrel_base call.
7929 (s390_chunkify_start): Remove pending LTREL_BASE logic.
7930 (s390_chunkify_finish): Remove replace_ltrel_base call.
7931 * config/s390/s390.md: Remove UNSPEC_LTREL_BASE.
7933 2018-09-06 Hans-Peter Nilsson <hp@axis.com>
7936 * config/cris/cris.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Redefine
7937 to speculation_safe_value_not_needed.
7939 2018-09-05 Cesar Philippidis <cesar@codesourcery.com>
7940 Bernd Schmidt <bernds_cb1@t-online.de>
7942 * config/nvptx/nvptx-opts.h: New file.
7943 * config/nvptx/nvptx.c (nvptx_file_start): Print the correct .target.
7944 * config/nvptx/nvptx.h: Include "nvptx-opts.h".
7946 (TARGET_SM35): New macro.
7947 * config/nvptx/nvptx.md (atomic_fetch_<logic><mode>): Enable with the
7949 * config/nvptx/nvptx.opt (ptx_isa, sm_30, sm_35): New enum and its
7951 (misa=): New option.
7952 * doc/invoke.texi (Nvidia PTX Options): Document -misa.
7954 2018-09-05 Uros Bizjak <ubizjak@gmail.com>
7956 * config/i386/i386.md (truncdfsf2): Remove expander.
7957 (truncdfsf2_with_temp): Ditto.
7958 (truncxf<mode>2): Ditto.
7959 (*truncdfsf_fast_mixed): Remove insn pattern.
7960 (*truncdfsf_fast_i387): Ditto.
7961 (*truncdfsf_mixed): Ditto.
7962 (*truncdfsf_i387): Ditto.
7963 (*truncdfsf2_i387_1): Ditto.
7964 (*truncxfsf2_mixed): Ditto.
7965 (*truncxfdf2_mixed): Ditto.
7966 (*truncxf<mode>2_i387_noop): Ditto. Update callers
7967 to call gen_truncxf<mode>2 instead.
7968 (*truncxf<mode>2_i387): Remove.
7969 (reg->reg splitters): Remove splitter pattern.
7970 (reg->mem splitters): Ditto.
7972 (truncdfsf2): New insn pattern.
7973 (truncxf<mode>2): Ditto.
7975 2018-09-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7977 * tree-ssa-math-opts.c (is_mult_by): New function.
7978 (is_square_of): Use the above.
7979 (optimize_recip_sqrt): New function.
7980 (pass_cse_reciprocals::execute): Use the above.
7982 2018-09-05 Richard Biener <rguenther@suse.de>
7985 * tree-ssa-sccvn.c (rpo_elim::eliminate_push_avail): Make sure
7986 to zero-init the emplaced vec.
7988 2018-09-05 Martin Liska <mliska@suse.cz>
7990 PR tree-optimization/87205
7991 * tree-switch-conversion.c (pass_lower_switch::execute):
7992 Group cases for switch statements.
7994 2018-09-05 Richard Biener <rguenther@suse.de>
7996 PR tree-optimization/87217
7997 * tree-ssa-sccvn.c (vuse_valueize): New.
7998 (vn_reference_lookup_pieces): Use it.
7999 (vn_reference_lookup): Likewise.
8001 2018-09-05 Nathan Sidwell <nathan@acm.org>
8004 * stor-layout.c (place_field): Scan forwards to check last
8005 bitfield when ms_bitfield_placement is in effect.
8007 2018-09-05 Richard Biener <rguenther@suse.de>
8010 * tree-vect-stmts.c (vectorizable_simd_clone_call): Fix bogus
8013 2018-09-05 Siddhesh Poyarekar <siddhesh@sourceware.org>
8014 Kugan Vivekanandarajah <kugan.vivekanandarajah@linaro.org>
8016 * config/aarch64/falkor-tag-collision-avoidance.c: New file.
8017 * config.gcc (extra_objs): Build it.
8018 * config/aarch64/t-aarch64 (falkor-tag-collision-avoidance.o):
8020 * config/aarch64/aarch64-passes.def
8021 (pass_tag_collision_avoidance): New pass.
8022 * config/aarch64/aarch64.c (qdf24xx_tunings): Add
8023 AARCH64_EXTRA_TUNE_RENAME_LOAD_REGS to tuning_flags.
8024 (aarch64_classify_address): Remove static qualifier.
8025 (aarch64_address_info, aarch64_address_type): Move to...
8026 * config/aarch64/aarch64-protos.h: ... here.
8027 (make_pass_tag_collision_avoidance): New function.
8028 * config/aarch64/aarch64-tuning-flags.def (rename_load_regs):
8031 2018-09-05 Martin Liska <mliska@suse.cz>
8033 * doc/gcov.texi: Update documentation of humar
8035 * gcov.c (format_count): Print one decimal place, it provides
8036 more fine number of situations like '1G' vs. '1.4G'.
8038 2018-09-05 Martin Liska <mliska@suse.cz>
8041 * config/rs6000/rs6000.opt: Mark the option as Deprecated.
8042 * optc-gen.awk: Allow 'Var' for Deprecated options in order
8043 to generate a MASK value.
8045 2018-09-04 H.J. Lu <hongjiu.lu@intel.com>
8048 * dwarf2out.c (based_loc_descr): Allow hard frame pointer even
8049 if frame pointer isn't used.
8050 (compute_frame_pointer_to_fb_displacement): Likewise.
8052 2018-09-04 Jakub Jelinek <jakub@redhat.com>
8055 * common/config/i386/i386-common.c (OPTION_MASK_ISA_XSAVEOPT_SET,
8056 OPTION_MASK_ISA_XSAVES_SET, OPTION_MASK_ISA_XSAVEC_SET): Use
8057 OPTION_MASK_ISA_XSAVE_SET instead of OPTION_MASK_ISA_XSAVE.
8058 (OPTION_MASK_ISA_XSAVE_UNSET): Add OPTION_MASK_ISA_XSAVES_UNSET
8059 and OPTION_MASK_ISA_XSAVEC_UNSET.
8061 2018-09-04 Max Filippov <jcmvbkbc@gmail.com>
8063 * config/xtensa/xtensa.c (xtensa_expand_atomic): Reorder AND and
8064 XOR operations in NAND case.
8066 2018-09-04 Aldy Hernandez <aldyh@redhat.com>
8068 * wide-int-range.cc (wide_int_range_convert): New.
8069 * wide-int-range.h (wide_int_range_convert): New.
8070 * tree-vrp.c (extract_range_from_unary_expr): Abstract wide int
8071 code into wide_int_range_convert.
8072 (extract_range_into_wide_ints): Do not munge anti range constants
8073 into the entire domain. Just return the range back.
8075 2018-09-04 Martin Liska <mliska@suse.cz>
8077 * genmatch.c (output_line_directive): Add new argument
8079 (dt_simplify::gen_1): Encapsulate dump within __builtin_expect.
8081 2018-09-04 Jonathan Wakely <jwakely@redhat.com>
8083 * doc/invoke.texi (Option Summary): Add whitespace.
8085 * doc/invoke.texi (Option Summary): Add -Waligned-new.
8087 2018-09-04 Richard Biener <rguenther@suse.de>
8089 PR tree-optimization/87211
8090 * tree-ssa-sccvn.c (visit_phi): When value-numbering to a
8091 backedge value we're supposed to treat as VARYING also number
8092 the PHI to VARYING in case it got a different value-number already.
8094 2018-09-04 Aldy Hernandez <aldyh@redhat.com>
8096 * tree-vrp.c (vrp_can_optimize_bit_op): Remove.
8097 (extract_range_from_binary_expr_1): Do not call
8098 vrp_can_optimize_bit_op.
8099 * wide-int-range.cc (wide_int_range_can_optimize_bit_op): Make
8101 (wide_int_range_get_mask_and_bounds): New.
8102 (wide_int_range_optimize_bit_op): New.
8103 (wide_int_range_bit_ior): Call wide_int_range_optimize_bit_op.
8104 (wide_int_range_bit_and): Same.
8105 * wide-int-range.h (wide_int_range_can_optimize_bit_op): Remove.
8106 (wide_int_range_optimize_bit_op): New.
8107 (wide_int_range_get_mask_and_bounds): New.
8109 2018-09-04 Richard Biener <rguenther@suse.de>
8111 PR tree-optimization/87176
8112 * tree-ssa-sccvn.c (visit_phi): Remove redundant allsame
8113 variable. When value-numbering a virtual PHI node make sure
8114 to not value-number to the backedge value.
8116 2018-09-04 Jonathan Wakely <jwakely@redhat.com>
8118 * doc/extend.texi (Long Long, Hex Floats): Document support for
8119 long long and hex floats in more recent versions of ISO C++.
8121 2018-09-03 Richard Biener <rguenther@suse.de>
8123 PR tree-optimization/87177
8124 * tree-ssa-sccvn.c (vuse_ssa_val): Revert previous change, keep
8127 2018-09-03 Alexander Monakov <amonakov@ispras.ru>
8129 * bb-reorder.c (edge_order): Convert to C-qsort-style
8130 tri-state comparator.
8131 (reorder_basic_blocks_simple): Change std::stable_sort to
8134 2018-09-03 Alexander Monakov <amonakov@ispras.ru>
8136 * tree-loop-distribution.c (offset_cmp): Convert to C-qsort-style
8137 tri-state comparator.
8138 (fuse_memset_builtins): Change std::stable_sort to gcc_stablesort.
8140 2018-09-03 Alexander Monakov <amonakov@ispras.ru>
8142 * sort.cc (struct sort_ctx): New field 'nlim'. Use it...
8143 (mergesort): ... here as maximum count for using netsort.
8144 (gcc_qsort): Set nlim to 3 if stable sort is requested.
8145 (gcc_stablesort): New.
8146 * system.h (gcc_stablesort): Declare.
8148 2018-09-03 Alexander Monakov <amonakov@ispras.ru>
8150 * sort.cc (gcc_qsort) [CHECKING_P]: Call qsort_chk.
8151 * system.h (qsort): Always redirect to gcc_qsort. Update comment.
8152 * vec.c (qsort_chk): Do not call gcc_qsort. Update comment.
8154 2018-09-03 Segher Boessenkool <segher@kernel.crashing.org>
8156 * config/rs6000/rs6000.md (*mov<mode>_hardfloat32): Remove %U from the
8157 lxsdx and stxsdx alternatives.
8158 (*mov<mode>_hardfloat64): Ditto.
8159 * config/rs6000/vsx.md (*vsx_extract_<mode>_store): Ditto.
8161 2018-09-03 Richard Biener <rguenther@suse.de>
8163 PR tree-optimization/87200
8164 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Valueize a
8167 2018-09-03 Martin Liska <mliska@suse.cz>
8169 PR tree-optimization/87201
8170 * tree-switch-conversion.c (switch_decision_tree::balance_case_nodes):
8171 Fix parenthesis in an expression.
8173 2018-09-03 Richard Biener <rguenther@suse.de>
8175 PR tree-optimization/87197
8176 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Mark the new def
8177 visited. CSE the VN_INFO hashtable lookup.
8179 PR tree-optimization/87169
8180 * tree-ssa-sccvn.c (do_rpo_vn): When marking loops for not
8181 iterating make sure there's no extra backedges from irreducible
8182 regions feeding the header. Mark the destination block
8185 2018-09-03 Martin Liska <mliska@suse.cz>
8188 * common/common-target.def: Add TARGET_GET_VALID_OPTION_VALUES.
8189 * common/common-targhooks.c (default_get_valid_option_values):
8191 * common/common-targhooks.h (default_get_valid_option_values):
8193 * common/config/i386/i386-common.c: Move processor_target_table
8195 (ix86_get_valid_option_values): New function.
8196 (TARGET_GET_VALID_OPTION_VALUES): New macro.
8197 * config/i386/i386.c (struct ptt): Move to i386-common.c.
8198 (PTA_*): Move all defined masks into i386-common.c.
8199 (ix86_function_specific_restore): Use new processor_cost_table.
8200 * config/i386/i386.h (struct ptt): Moved from i386.c.
8201 (struct pta): Likewise.
8202 * doc/tm.texi: Document new TARGET_GET_VALID_OPTION_VALUES.
8203 * doc/tm.texi.in: Likewise.
8204 * opt-suggestions.c (option_proposer::suggest_option):
8205 Pass prefix to build_option_suggestions.
8206 (option_proposer::get_completions): Likewise.
8207 (option_proposer::build_option_suggestions): Use the new target
8209 * opts.c (struct option_help_tuple): New struct.
8210 (print_filtered_help): Use the new target hook.
8212 2018-09-03 Martin Liska <mliska@suse.cz>
8215 * predict.c (set_even_probabilities): Add likely_edges
8216 argument and handle cases where we have precisely one
8218 (combine_predictions_for_bb): Catch also likely_edges.
8219 (tree_predict_by_opcode): Handle gswitch statements.
8220 * tree-cfg.h (find_case_label_for_value): New declaration.
8221 (find_taken_edge_switch_expr): Likewise.
8222 * tree-switch-conversion.c (switch_decision_tree::balance_case_nodes):
8223 Find pivot in decision tree based on probabily, not by number of
8226 2018-09-02 Gerald Pfeifer <gerald@pfeifer.com>
8228 * doc/standards.texi (Standards): Update Objective-C reference.
8230 2018-09-01 Gerald Pfeifer <gerald@pfeifer.com>
8232 * doc/install.texi (Prerequisites): Update link for MPC.
8234 2018-09-01 Michael Matz <matz@suse.de>
8236 PR tree-optimization/87074
8237 * gimple-loop-jam.c (unroll_jam_possible_p): Check loop exit
8238 PHIs for outer-loop uses.
8240 2018-09-01 Gerald Pfeifer <gerald@pfeifer.com>
8242 * doc/generic.texi (OpenMP): Adjust link to openmp.org.
8243 * doc/invoke.texi (C Dialect Options): Ditto.
8245 2018-09-01 Gerald Pfeifer <gerald@pfeifer.com>
8247 * doc/install.texi (Prerequisites): Adjust link mpfr.org.
8249 2018-08-31 Richard Biener <rguenther@suse.de>
8251 PR tree-optimization/87168
8252 * tree-ssa-sccvn.c (SSA_VAL): Add visited output parameter.
8253 (rpo_elim::eliminate_avail): When OP was not visited it must
8256 2018-08-31 David Malcolm <dmalcolm@redhat.com>
8258 * tree-vrp.c (copy_value_range): Convert param "from" from
8259 "value_range *" to "const value_range *".
8260 (range_is_null): Likewise for param "vr".
8261 (range_int_cst_p): Likewise.
8262 (range_int_cst_singleton_p): Likewise.
8263 (symbolic_range_p): Likewise.
8264 (value_ranges_intersect_p): Likewise for both params.
8265 (value_range_nonnegative_p): Likewise for param "vr".
8266 (value_range_constant_singleton): Likewise.
8267 (vrp_set_zero_nonzero_bits): Likewise for param "ar".
8268 (extract_range_into_wide_ints): Likewise for param "vr".
8269 (extract_range_from_multiplicative_op): Likewise for params "vr0"
8271 (vrp_can_optimize_bit_op): Likewise.
8272 (extract_range_from_binary_expr_1): Likewise for params "vr0_" and
8274 (extract_range_from_unary_expr): Likewise.
8275 (debug_value_range): Likewise for param "vr".
8276 (value_range::dump): Add "const" qualifier.
8277 (vrp_prop::check_array_ref): Convert local "vr" from
8278 "value_range *" to "const value_range *".
8279 (vrp_prop::check_mem_ref): Likewise.
8280 (vrp_prop::visit_stmt): Likewise for local "old_vr".
8281 (vrp_intersect_ranges_1): Likewise for param "vr_1".
8282 (vrp_intersect_ranges): Likewise.
8283 (simplify_stmt_for_jump_threading): Likewise for local "vr".
8284 (vrp_prop::vrp_finalize): Likewise.
8285 * tree-vrp.h (value_range::dump): Add "const" qualifier.
8286 (vrp_intersect_ranges): Add "const" qualifier to params as above.
8287 (extract_range_from_unary_expr): Likewise.
8288 (value_range_constant_singleton): Likewise.
8289 (symbolic_range_p): Likewise.
8290 (copy_value_range): Likewise.
8291 (extract_range_from_binary_expr_1): Likewise.
8292 (range_int_cst_p): Likewise.
8293 (vrp_set_zero_nonzero_bits): Likewise.
8294 (range_int_cst_singleton_p): Likewise.
8296 2018-08-31 Vlad Lazar <vlad.lazar@arm.com>
8298 * config/aarch64/arm_neon.h (vabsd_s64): New.
8299 (vnegd_s64): Likewise.
8301 2018-08-31 Martin Jambor <mjambor@suse.cz>
8303 * ipa-cp.c (estimate_local_effects): Replace wrong MAX with MIN.
8305 2018-08-31 Martin Liska <mliska@suse.cz>
8307 * ipa-icf.c (sem_item::add_type): Use
8308 sem_item::m_type_hash_cache.
8309 * ipa-icf.h: Move the cache from sem_item_optimizer
8312 2018-08-31 Nathan Sidwell <nathan@acm.org>
8314 * doc/extend.texi (Backwards Compatibility): Remove implicit
8315 extern C leeway of () being (...).
8317 2018-08-31 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8319 * ipa-inline.c (can_inline_edge_by_limits_p): Fix typos in comment.
8321 2018-08-31 Segher Boessenkool <segher@kernel.crashing.org>
8325 * config/rs6000/rs6000.md (lround<mode>di2): Gate on TARGET_FPRND.
8327 2018-08-31 Jakub Jelinek <jakub@redhat.com>
8330 * expmed.c (expand_mult_const): Use immed_wide_int_const instead of
8331 gen_int_mode. Formatting fixes.
8333 2018-08-30 Sandra Loosemore <sandra@codesourcery.com>
8335 * target.def (custom_function_descriptors): Improve documentation.
8336 * doc/tm.texi.in (Trampolines): Expand discussion of function
8337 descriptors and move TARGET_CUSTOM_FUNCTION_DESCRIPTORS to the
8338 beginning of the section.
8339 * doc/tm.texi: Regenerated.
8341 2018-08-30 Jose E. Marchesi <jose.marchesi@oracle.com>
8343 * cfg.h (class auto_edge_flag): Spell out the template-id of the
8344 base class in the initializer list. This is a workaround for
8345 building with older GCC.
8346 (class auto_bb_flag): Likewise.
8348 2018-08-30 Aaron Sawdey <acsawdey@linux.ibm.com>
8350 * config/rs6000/altivec.md (altivec_eq<mode>): Remove star.
8351 (altivec_vcmpequ<VI_char>_p): Remove star.
8352 * config/rs6000/rs6000-string.c (do_load_for_compare): Support
8354 (expand_strncmp_vec_sequence): New function.
8355 (emit_final_str_compare_vec): New function.
8356 (expand_strn_compare): Add support for vector strncmp.
8357 * config/rs6000/rs6000.opt (-mstring-compare-inline-limit): Change
8358 length specification to bytes.
8359 * config/rs6000/vsx.md (vsx_ld_elemrev_v16qi_internal): Remove star.
8360 (vcmpnezb_p): New pattern.
8361 * doc/invoke.texi (RS/6000 and PowerPC Options): Update documentation
8362 for option -mstring-compare-inline-limit.
8364 2018-08-30 Thiago Macieira <thiago.macieira@intel.com>
8366 * config/i386/i386.c (PTA_WESTMERE): Remove PTA_AES.
8367 (PTA_SKYLAKE): Add PTA_AES.
8368 (PTA_GOLDMONT): Likewise.
8370 2018-08-29 Jan Hubicka <jh@suse.cz>
8373 * lto-opts.c (lto_write_options): Always stream PIC/PIE mode.
8374 * lto-wrapper.c (merge_and_complain): Fix merging of PIC/PIE.
8376 2018-08-29 Jan Hubicka <jh@suse.cz>
8378 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not follow
8380 (hash_tree): Do not visit TYPE_STUB_DECL.
8381 * tree-streamer-out.c (write_ts_type_common_tree_pointers): Do not
8382 stream TYPE_STUB_DECL.
8383 * tree-streamer-in.c (lto_input_ts_type_common_tree_pointers): Likewise.
8384 * ipa-utils.h (type_with_linkage_p): Do not rely on TYPE_STUB_DECL
8385 after free_lang_data.
8386 (type_in_anonymous_namespace_p): Likewise.
8388 2018-08-29 Jan Hubicka <jh@suse.cz>
8390 * sreal.h (SREAL_PART_BITS): Change to 31; remove seemingly unnecessary
8391 comment that it has to be even number.
8392 (class sreal): Change m_sig type to int32_t.
8393 * sreal.c (sreal::dump, sreal::to_int, opreator+, operator-): Use
8394 int64_t for temporary calculations.
8395 (sreal_verify_basics): Drop one bit from minimum and maximum.
8397 2018-08-30 Richard Biener <rguenther@suse.de>
8399 PR tree-optimization/87147
8400 * tree-ssa-sccvn.c (SSA_VISITED): New function.
8401 (visit_phi): When the degenerate result is from the backedge and
8402 we didn't visit its definition yet drop to VARYING.
8403 (do_rpo_vn): Properly mark blocks with incoming backedges as executable.
8405 2018-08-29 Jan Hubicka <jh@suse.cz>
8407 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not walk
8409 (hash_tree): Likewise.
8411 2018-08-29 Jan Hubicka <jh@suse.cz>
8413 * tree.c (find_decls_types_r): Walk also TYPE_NEXT_PTR_TO
8414 and TYPE_NEXT_REF_TO.
8416 2018-08-29 Jan Hubicka <jh@suse.cz>
8418 * sreal.h (SREAL_PART_BITS): Change to 31; remove seemingly unnecessary
8419 comment that it has to be even number.
8420 (class sreal): Change m_sig type to int32_t.
8421 * sreal.c (sreal::dump, sreal::to_int, opreator+, operator-): Use
8422 int64_t for temporary calculations.
8423 (sreal_verify_basics): Drop one bit from minimum and maximum.
8425 2018-08-30 Tamar Christina <tamar.christina@arm.com>
8427 * config/aarch64/aarch64.c (aarch64_expand_movmem): Set TImode max.
8429 2018-08-30 Vlad Lazar <vlad.lazar@arm.com>
8432 * expmed.c (canonicalize_comparison): Use wi::sub instead of wi::add
8433 if to_add is negative.
8435 2018-08-29 Bernd Edlinger <bernd.edlinger@hotmail.de>
8438 * builtins.c (c_strlen): Improve range checks.
8440 2018-08-29 Martin Sebor <msebor@redhat.com>
8441 Jeff Law <law@redhat.com>
8443 PR tree-optimization/86714
8444 PR tree-optimization/86711
8445 * builtins.c (c_strlen): Add arguments to call to string_constant.
8446 * expr.c (string_constant): Add argument. Detect missing nul
8447 terminator and outermost declaration it's missing in.
8448 * expr.h (string_constant): Add argument.
8449 * fold-const.c (read_from_constant_string): Add arguments to call to
8451 (c_getstr): Likewise.
8452 * tree-ssa-forwprop.c (simplify_builtin_call): Likewise.
8454 * tree-ssa-strlen.c (get_stridx): Likewise.
8456 2018-08-29 Jan Hubicka <jh@suse.cz>
8458 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers):
8459 Do not stream DECL_VINDEX.
8460 * tree-streamer-out.c (write_ts_function_decl_tree_pointers): Likewise.
8461 * tree.c (free_lang_data_in_decl): Clear DECL_VINDEX.
8462 (decl_function_context): Use DECL_VIRTUAL_P rather than DECL_VINDEX.
8464 2018-08-29 Richard Biener <rguenther@suse.de>
8466 * tree-ssa-sccvn.c (vuse_ssa_val): Return NULL for unvisited
8467 virtual operands that are not default defs to honor region
8469 (rpo_vn_valueize): Remove ineffective code here.
8471 2018-08-29 Richard Biener <rguenther@suse.de>
8473 PR tree-optimization/87132
8474 * tree-ssa-alias.c (get_continuation_for_phi): Do not translate
8475 when skipping defs reachable over backedges.
8477 2018-08-29 Richard Biener <rguenther@suse.de>
8479 * tree-core.h: Document use of deprecated_flag in SSA_NAME.
8480 * tree.h (SSA_NAME_POINTS_TO_READONLY_MEMORY): Define.
8481 * tree-into-ssa.c (pass_build_ssa::execute): Initialize
8482 function parameters SSA_NAME_POINTS_TO_READONLY_MEMORY from fnspec.
8483 * tree-ssa-sccvn.c (const_parms, init_const_parms): Remove.
8484 (vn_reference_lookup_3): Remove use of const_parms.
8485 (free_rpo_vn): Do not free const_parms.
8486 (do_rpo_vn): Do not call init_const_parms.
8487 * tree-ssa-alias.c (refs_may_alias_p_1): Honor
8488 SSA_NAME_POINTS_TO_READONLY_MEMORY.
8489 (call_may_clobber_ref_p_1): Likewise.
8491 2018-08-29 Alexander Monakov <amonakov@ispras.ru>
8494 * invoke.texi (Optimization Options): List -ftree-scev-cprop.
8496 (-ftree-scev-cprop): Document.
8498 2018-08-29 Jan Hubicka <jh@suse.cz>
8500 * sreal.h (normalize, normalize_up, normalize_down): Add new_sig/new_exp
8502 (sreal constructor): Update.
8503 * sreal.c (sreal:operator+, sreal:operator-, sreal:operator*,
8504 sreal:operator/): Update.
8506 2018-08-29 Martin Liska <mliska@suse.cz>
8508 * tree-switch-conversion.c (switch_conversion::expand):
8509 Strenghten assumption about gswitch statements.
8511 2018-08-29 Richard Biener <rguenther@suse.de>
8513 PR tree-optimization/87117
8514 * tree-ssa-sccvn.c (eliminate_dom_walker::eliminate_stmt): Only
8515 re-value-number released SSA VDEFs.
8517 2018-08-29 Richard Biener <rguenther@suse.de>
8519 PR tree-optimization/87126
8520 * tree-ssa-sccvn.c (vn_reference_insert): Remove assert.
8522 2018-08-28 Jim Wilson <jimw@sifive.com>
8524 * config/riscv/pic.md: Rewrite.
8525 * config/riscv/riscv.c (riscv_address_insns): Return cost of 3 for
8527 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Delete.
8528 (SOFTF, default_load, softload, softstore): New.
8530 2018-08-28 Jeff Law <law@redhat.com>
8532 * fold-const.c (fold_binary_loc): Remove recently added assert.
8534 2018-08-28 Joern Rennecke <joern.rennecke@riscy-ip.com>
8536 * genpreds.c (write_predicate_subfunction): Also add ATTRIBUTE_UNUSED
8537 to OP parmeter of generated function.
8539 2018-08-28 MCC CS <deswurstes@users.noreply.github.com>
8541 PR tree-optimization/87009
8542 * match.pd: Add boolean optimizations.
8544 2018-08-28 Martin Sebor <msebor@redhat.com>
8547 * calls.c (alloc_max_size): Treat HOST_WIDE_INT special.
8548 * gimple-ssa-warn-alloca.c (adjusted_warn_limit): New function.
8549 (pass_walloca::gate): Use it.
8550 (alloca_call_type): Same.
8551 (pass_walloca::execute): Same.
8552 * stor-layout.c (layout_decl): Treat HOST_WIDE_INT special.
8554 2018-08-28 David Malcolm <dmalcolm@redhat.com>
8556 * dumpfile.h (ATTRIBUTE_GCC_DUMP_PRINTF): Change version check on
8557 GCC_VERSION for usage of "__gcc_dump_printf__" format from
8560 2018-08-28 Richard Biener <rguenther@suse.de>
8562 PR tree-optimization/87124
8563 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Guard against
8564 constants before looking up avail.
8566 2018-08-28 Jakub Jelinek <jakub@redhat.com>
8569 * calls.c (maybe_warn_nonstring_arg): Punt early if
8570 warn_stringop_overflow is zero. Don't call get_range_strlen
8571 on 3rd argument, keep iterating until lenrng[1] is INTEGER_CST.
8572 Swap comparison operands to have constants on rhs. Only use
8573 lenrng[1] if non-NULL and INTEGER_CST. Don't uselessly
8574 increment lenrng[0].
8576 2018-08-28 Richard Sandiford <richard.sandiford@arm.com>
8578 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Fix unguarded
8579 use of tree_to_shwi. Remove duplicated test for the size being
8580 a whole number of bytes.
8582 2018-08-28 Richard Biener <rguenther@suse.de>
8584 PR tree-optimization/87117
8585 * tree-ssa-sccvn.c (eliminate_dom_walker::eliminate_cleanup):
8586 Handle removed stmt without LHS (GIMPLE_NOP).
8588 2018-08-28 Richard Biener <rguenther@suse.de>
8590 PR tree-optimization/87117
8591 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Exclude
8592 void which is is_gimple_reg_type by checking for COMPLETE_TYPE_P.
8594 2018-08-28 Richard Biener <rguenther@suse.de>
8596 PR tree-optimization/87117
8597 * tree-ssa-pre.c (compute_avail): Do not make expressions
8598 with predicated values available.
8599 (get_expr_value_id): Assert we do not run into predicated value
8602 2018-08-28 Richard Biener <rguenther@suse.de>
8604 PR tree-optimization/87117
8605 * tree-ssa-operands.c (add_stmt_operand): STRING_CST may
8606 get virtual operands.
8607 (get_expr_operands): Handle STRING_CST like other decls.
8609 2018-08-28 Martin Liska <mliska@suse.cz>
8611 * tree.h: Update documentation of fndecl_built_in_p
8615 2018-08-27 Jeff Law <law@redhat.com>
8616 PR tree-optimization/87110
8617 * tree-ssa-dse.c (compute_trims): Handle non-constant
8620 2018-08-27 Martin Sebor <msebor@redhat.com>
8622 PR tree-optimization/86914
8623 * tree-ssa-strlen.c (maybe_set_strlen_range): Avoid MEM_REF.
8625 2018-08-27 Martin Sebor <msebor@redhat.com>
8627 PR tree-optimization/87112
8628 * builtins.c (expand_builtin_strnlen): Convert c_strlen result to
8629 the type of the bound argument.
8631 2018-08-27 Jeff Law <law@redhat.com>
8633 * tree-ssa-dse.c (compute_trims): Handle case where the reference's
8634 type does not have a TYPE_SIZE_UNIT.
8636 2018-08-27 Steve Ellcey <sellcey@cavium.com>
8638 * config/aarch64/aarch64-speculation.cc: Replace include of cfg.h
8639 with include of backend.h.
8641 2018-08-27 Richard Biener <rguenther@suse.de>
8643 PR tree-optimization/86927
8644 * tree-vect-loop.c (vect_create_epilog_for_reduction): Properly
8645 use const cond reduction code.
8647 2018-08-27 Alexander Monakov <amonakov@ispras.ru>
8649 PR tree-optimization/85758
8650 * match.pd ((X & Y) ^ Y): Add :s qualifier to inner expression.
8652 2018-08-27 David Malcolm <dmalcolm@redhat.com>
8655 * diagnostic-show-locus.c (class layout_range): Update for
8656 conversion of show_caret_p to a tri-state.
8657 (layout_range::layout_range): Likewise.
8658 (make_range): Likewise.
8659 (layout::maybe_add_location_range): Likewise.
8660 (layout::should_print_annotation_line_p): Don't show annotation
8661 lines for ranges that are SHOW_LINES_WITHOUT_RANGE.
8662 (layout::get_state_at_point): Update for conversion of
8663 show_caret_p to a tri-state. Bail out early for
8664 SHOW_LINES_WITHOUT_RANGE, so that such ranges don't affect
8665 underlining or source colorization.
8666 (gcc_rich_location::add_location_if_nearby): Update for conversion
8667 of show_caret_p to a tri-state.
8668 (selftest::test_one_liner_multiple_carets_and_ranges): Likewise.
8669 (selftest::test_one_liner_fixit_replace_equal_secondary_range):
8671 (selftest::test_one_liner_labels): Likewise.
8672 * gcc-rich-location.c (gcc_rich_location::add_expr): Update for
8673 conversion of show_caret_p to a tri-state.
8674 * pretty-print.c (text_info::set_location): Likewise.
8675 * pretty-print.h (text_info::set_location): Likewise.
8676 * substring-locations.c (format_warning_n_va): Likewise.
8677 * tree-diagnostic.c (default_tree_printer): Likewise.
8678 * tree-pretty-print.c (newline_and_indent): Likewise.
8680 2018-08-27 David Malcolm <dmalcolm@redhat.com>
8683 * diagnostic-show-locus.c (get_line_span_for_fixit_hint): Show the
8684 line above for line-insertion fix-it hints.
8685 (selftest::test_fixit_insert_containing_newline): Update the
8686 expected results, and add a test with line-numbering enabled.
8688 2018-08-27 Martin Liska <mliska@suse.cz>
8691 * sanopt.c (sanitize_rewrite_addressable_params): Ignore
8692 params with DECL_HAS_VALUE_EXPR_P.
8694 2018-08-27 Martin Liska <mliska@suse.cz>
8696 * config/i386/i386.c (ix86_expand_set_or_movmem): Dump
8697 selected expansion strategy.
8699 2018-08-27 Martin Liska <mliska@suse.cz>
8701 * builtins.h (is_builtin_fn): Remove and fndecl_built_in_p.
8702 * builtins.c (is_builtin_fn): Likewise.
8703 * attribs.c (diag_attr_exclusions): Use new function
8704 fndecl_built_in_p and remove check for FUNCTION_DECL if
8706 (builtin_mathfn_code): Likewise.
8707 (fold_builtin_expect): Likewise.
8708 (fold_call_expr): Likewise.
8709 (fold_builtin_call_array): Likewise.
8710 (fold_call_stmt): Likewise.
8711 (set_builtin_user_assembler_name): Likewise.
8712 (is_simple_builtin): Likewise.
8713 * calls.c (gimple_alloca_call_p): Likewise.
8714 (maybe_warn_nonstring_arg): Likewise.
8715 * cfgexpand.c (expand_call_stmt): Likewise.
8716 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Likewise.
8717 (cgraph_edge::verify_corresponds_to_fndecl): Likewise.
8718 (cgraph_node::verify_node): Likewise.
8719 * cgraphclones.c (build_function_decl_skip_args): Likewise.
8720 (cgraph_node::create_clone): Likewise.
8721 * config/arm/arm.c (arm_insert_attributes): Likewise.
8722 * config/i386/i386.c (ix86_gimple_fold_builtin): Likewise.
8723 * dse.c (scan_insn): Likewise.
8724 * expr.c (expand_expr_real_1): Likewise.
8725 * fold-const.c (operand_equal_p): Likewise.
8726 (fold_binary_loc): Likewise.
8727 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
8728 * gimple-low.c (lower_stmt): Likewise.
8729 * gimple-pretty-print.c (dump_gimple_call): Likewise.
8730 * gimple-ssa-warn-restrict.c (wrestrict_dom_walker::check_call):
8732 * gimple.c (gimple_build_call_from_tree): Likewise.
8733 (gimple_call_builtin_p): Likewise.
8734 (gimple_call_combined_fn): Likewise.
8735 * gimplify.c (gimplify_call_expr): Likewise.
8736 (gimple_boolify): Likewise.
8737 (gimplify_modify_expr): Likewise.
8738 (gimplify_addr_expr): Likewise.
8739 * hsa-gen.c (gen_hsa_insns_for_call): Likewise.
8740 * ipa-cp.c (determine_versionability): Likewise.
8741 * ipa-fnsummary.c (compute_fn_summary): Likewise.
8742 * ipa-param-manipulation.c (ipa_modify_formal_parameters): Likewise.
8743 * ipa-split.c (visit_bb): Likewise.
8744 (split_function): Likewise.
8745 * ipa-visibility.c (cgraph_externally_visible_p): Likewise.
8746 * lto-cgraph.c (input_node): Likewise.
8747 * lto-streamer-out.c (write_symbol): Likewise.
8748 * omp-low.c (setjmp_or_longjmp_p): Likewise.
8749 (lower_omp_1): Likewise.
8750 * predict.c (strip_predict_hints): Likewise.
8751 * print-tree.c (print_node): Likewise.
8752 * symtab.c (symtab_node::output_to_lto_symbol_table_p): Likewise.
8753 * trans-mem.c (is_tm_irrevocable): Likewise.
8754 (is_tm_load): Likewise.
8755 (is_tm_simple_load): Likewise.
8756 (is_tm_store): Likewise.
8757 (is_tm_simple_store): Likewise.
8758 (is_tm_abort): Likewise.
8759 (tm_region_init_1): Likewise.
8760 * tree-call-cdce.c (gen_shrink_wrap_conditions): Likewise.
8761 * tree-cfg.c (verify_gimple_call): Likewise.
8762 (move_stmt_r): Likewise.
8763 (stmt_can_terminate_bb_p): Likewise.
8764 * tree-eh.c (lower_eh_constructs_2): Likewise.
8765 * tree-if-conv.c (if_convertible_stmt_p): Likewise.
8766 * tree-inline.c (remap_gimple_stmt): Likewise.
8767 (copy_bb): Likewise.
8768 (estimate_num_insns): Likewise.
8769 (fold_marked_statements): Likewise.
8770 * tree-sra.c (scan_function): Likewise.
8771 * tree-ssa-ccp.c (surely_varying_stmt_p): Likewise.
8772 (optimize_stack_restore): Likewise.
8773 (pass_fold_builtins::execute): Likewise.
8774 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
8775 (mark_all_reaching_defs_necessary_1): Likewise.
8776 * tree-ssa-dom.c (dom_opt_dom_walker::optimize_stmt): Likewise.
8777 * tree-ssa-forwprop.c (simplify_builtin_call): Likewise.
8778 (pass_forwprop::execute): Likewise.
8779 * tree-ssa-loop-im.c (stmt_cost): Likewise.
8780 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
8781 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Likewise.
8782 * tree-ssa-strlen.c (get_string_length): Likewise.
8783 * tree-ssa-structalias.c (handle_lhs_call): Likewise.
8784 (find_func_aliases_for_call): Likewise.
8785 * tree-ssa-ter.c (find_replaceable_in_bb): Likewise.
8786 * tree-stdarg.c (optimize_va_list_gpr_fpr_size): Likewise.
8787 * tree-tailcall.c (find_tail_calls): Likewise.
8788 * tree.c (need_assembler_name_p): Likewise.
8789 (free_lang_data_in_decl): Likewise.
8790 (get_call_combined_fn): Likewise.
8791 * ubsan.c (is_ubsan_builtin_p): Likewise.
8792 * varasm.c (incorporeal_function_p): Likewise.
8793 * tree.h (DECL_BUILT_IN): Remove and replace with
8795 (DECL_BUILT_IN_P): Transfort to fndecl_built_in_p.
8796 (fndecl_built_in_p): New.
8798 2018-08-27 Martin Liska <mliska@suse.cz>
8800 PR tree-optimization/86847
8801 * tree-switch-conversion.c (switch_decision_tree::dump_case_nodes):
8802 Dump also subtree probability.
8803 (switch_decision_tree::do_jump_if_equal): New function.
8804 (switch_decision_tree::emit_case_nodes): Handle special
8805 situations in balanced tree that can be emitted much simpler.
8806 Fix calculation of probabilities that happen in tree expansion.
8807 * tree-switch-conversion.h (struct cluster): Add
8809 (struct simple_cluster): Likewise.
8810 (struct case_tree_node): Add new function has_child.
8811 (do_jump_if_equal): New.
8813 2018-08-27 Martin Liska <mliska@suse.cz>
8815 * tree-switch-conversion.c (bit_test_cluster::find_bit_tests):
8816 Add new argument to bit_test_cluster constructor.
8817 (bit_test_cluster::emit): Set bits really number of values
8819 (bit_test_cluster::hoist_edge_and_branch_if_true): Add
8820 probability argument.
8821 * tree-switch-conversion.h (struct bit_test_cluster):
8822 Add m_handles_entire_switch.
8824 2018-08-27 Martin Liska <mliska@suse.cz>
8826 PR tree-optimization/86702
8827 * tree-switch-conversion.c (jump_table_cluster::emit):
8828 Make probabilities even for values in jump table
8829 according to number of cases handled.
8830 (switch_decision_tree::compute_cases_per_edge): Pass
8831 argument to reset_out_edges_aux function.
8832 (switch_decision_tree::analyze_switch_statement): Likewise.
8833 * tree-switch-conversion.h (switch_decision_tree::reset_out_edges_aux):
8836 2018-08-27 Martin Liska <mliska@suse.cz>
8838 * cfgexpand.c (expand_asm_stmt): Use label_to_block and pass
8839 cfun argument explicitly.
8840 * gimple-pretty-print.c (dump_gimple_switch): Likewise.
8841 * hsa-gen.c (gen_hsa_insns_for_switch_stmt): Use new
8842 function gimple_switch_default_bb.
8843 (convert_switch_statements):
8845 * ipa-fnsummary.c (set_switch_stmt_execution_predicate):
8846 * stmt.c (label_to_block_fn): Use label_to_block and pass
8847 cfun argument explicitly and use gimple_switch_label_bb.
8848 (expand_case): Likewise.
8849 * tree-cfg.c (lower_phi_internal_fn): Use label_to_block and pass
8850 cfun argument explicitly. Likewise.
8851 (make_edges_bb): Likewise.
8852 (make_cond_expr_edges): Likewise.
8853 (get_cases_for_edge): Likewise.
8854 (make_gimple_switch_edges): Likewise.
8855 (label_to_block_fn): Likewise.
8856 (label_to_block): Likewise.
8857 (make_goto_expr_edges): Likewise.
8858 (make_gimple_asm_edges): Likewise.
8859 (main_block_label): Likewise.
8860 (group_case_labels_stmt): Likewise.
8861 (find_taken_edge_computed_goto): Likewise.
8862 (find_taken_edge_switch_expr): Likewise.
8863 (gimple_verify_flow_info): Likewise.
8864 (gimple_redirect_edge_and_branch): Likewise.
8865 (gimple_switch_label_bb): New function.
8866 (gimple_switch_default_bb): Likewise.
8867 (gimple_switch_edge): Likewise.
8868 (gimple_switch_default_edge): Likewise.
8869 * tree-cfg.h (label_to_block_fn): Remove and replace ...
8870 (label_to_block): ... with this.
8871 (gimple_switch_label_bb): New.
8872 (gimple_switch_default_bb): Likewise.
8873 (gimple_switch_edge): Likewise.
8874 (gimple_switch_default_edge): Likewise.
8875 * tree-cfgcleanup.c (convert_single_case_switch): Use
8876 new gimple functions and pass new argument to label_to_block.
8877 (cleanup_control_flow_bb):
8878 * tree-eh.c (make_eh_dispatch_edges): Use label_to_block and pass
8879 cfun argument explicitly.
8880 (make_eh_edges): Likewise.
8881 (redirect_eh_dispatch_edge): Likewise.
8882 (lower_resx): Likewise.
8883 (lower_eh_dispatch): Likewise.
8884 (maybe_remove_unreachable_handlers): Likewise.
8885 (unsplit_eh): Likewise.
8886 (cleanup_empty_eh): Likewise.
8887 (verify_eh_edges): Likewise.
8888 (verify_eh_dispatch_edge): Likewise.
8889 * tree-ssa-dom.c (record_edge_info): Likewise.
8890 * tree-ssa-forwprop.c (simplify_gimple_switch_label_vec): Likewise.
8891 * tree-ssa-threadedge.c (thread_around_empty_blocks): Likewise.
8892 (thread_through_normal_block): Likewise.
8893 * tree-ssa-uncprop.c (associate_equivalences_with_edges): Likewise.
8894 * tree-ssa-uninit.c (convert_control_dep_chain_into_preds):
8895 * tree-switch-conversion.c (switch_conversion::collect): Use new
8897 (switch_conversion::check_final_bb): Likewise.
8898 (switch_conversion::gather_default_values): Pass new argument
8900 (switch_conversion::build_constructors): Likewise.
8901 (switch_decision_tree::compute_cases_per_edge): Use new
8902 gimple_switch_edge function.
8903 (switch_decision_tree::analyze_switch_statement): Pass new argument
8905 (switch_decision_tree::try_switch_expansion): Use
8906 gimple_switch_default_edge.
8907 * tree-vrp.c (find_switch_asserts): Pass new argument
8909 * vr-values.c (vr_values::vrp_visit_switch_stmt): Likewise.
8910 (vr_values::simplify_switch_using_ranges): Likewise.
8912 2018-08-27 Richard Biener <rguenther@suse.de>
8914 * cfganal.h (rev_post_order_and_mark_dfs_back_seme): Declare.
8915 * cfganal.c (rev_post_order_and_mark_dfs_back_seme): New function.
8917 * tree-ssa-sccvn.h (struct vn_pval): New structure.
8918 (struct vn_nary_op_s): Add unwind_to member. Add
8919 predicated_values flag and put result into a union together
8920 with a linked list of vn_pval.
8921 (struct vn_ssa_aux): Add name member to make maintaining
8922 a map of SSA name to vn_ssa_aux possible. Remove no longer
8923 needed info, dfsnum, low, visited, on_sccstack, use_processed
8924 and range_info_anti_range_p members.
8925 (run_scc_vn, vn_eliminate, free_scc_vn, vn_valueize): Remove.
8926 (do_rpo_vn, run_rpo_vn, eliminate_with_rpo_vn, free_rpo_vn):
8928 (vn_valueize): New global.
8929 (vn_context_bb): Likewise.
8930 (VN_INFO_RANGE_INFO, VN_INFO_ANTI_RANGE_P, VN_INFO_RANGE_TYPE,
8931 VN_INFO_PTR_INFO): Remove.
8932 * tree-ssa-sccvn.c: ... (rewrite)
8933 (pass_fre::execute): For -O2+ initialize loops and run
8934 RPO VN in optimistic mode (iterating). For -O1 and -Og
8935 run RPO VN in non-optimistic mode.
8936 * params.def (PARAM_SCCVN_MAX_SCC_SIZE): Remove.
8937 (PARAM_RPO_VN_MAX_LOOP_DEPTH): Add.
8938 * doc/invoke.texi (sccvn-max-scc-size): Remove.
8939 (rpo-vn-max-loop-depth): Document.
8940 * tree-ssa-alias.c (walk_non_aliased_vuses): Stop walking
8941 when valuezing the VUSE signals we walked out of the region.
8942 * tree-ssa-pre.c (phi_translate_1): Ignore predicated values.
8943 (phi_translate): Set VN context block to use for availability
8945 (compute_avail): Likewise.
8946 (pre_valueize): New function.
8947 (pass_pre::execute): Adjust to the RPO VN API.
8949 * tree-ssa-loop-ivcanon.c: Include tree-ssa-sccvn.h.
8950 (propagate_constants_for_unrolling): Remove.
8951 (tree_unroll_loops_completely): Perform value-numbering
8952 on the unrolled bodies loop parent.
8954 2018-08-27 Richard Biener <rguenther@suse.de>
8956 * tree-ssa-pre.c (compute_antic): Re-use inverted postorder
8957 for partial antic compute.
8959 2018-08-27 Jakub Jelinek <jakub@redhat.com>
8961 PR rtl-optimization/87065
8962 * combine.c (simplify_if_then_else): Formatting fix.
8963 (if_then_else_cond): Guard MULT optimization with SCALAR_INT_MODE_P
8965 (known_cond): Don't return const_true_rtx for vector modes. Use
8966 CONST0_RTX instead of const0_rtx. Formatting fixes.
8968 2018-08-27 Martin Liska <mliska@suse.cz>
8970 PR gcov-profile/87069
8971 * gcov.c (process_file): Record files already processed
8972 and warn about a file being processed multiple times.
8974 2018-08-27 Martin Liska <mliska@suse.cz>
8977 * config/aarch64/aarch64.c (aarch64_override_options_internal):
8978 Set default values for x_aarch64_*_string strings.
8979 * config/aarch64/aarch64.opt: Remove --{march,mcpu,mtune}==
8980 prefix. For -mabi do not print '=ABI' in help and use
8981 <option_value> format for -msve-vector-bits and -moverride
8984 2018-08-26 Jeff Law <law@redhat.com>
8986 * config/mips/frame-header-opt.c: Include "backend.h" rather than
8989 2018-08-26 Marek Polacek <polacek@redhat.com>
8991 PR c++/87029, Implement -Wredundant-move.
8992 * doc/invoke.texi: Document -Wredundant-move.
8994 2018-08-25 Martin Sebor <msebor@redhat.com>
8996 PR tree-optimization/87059
8997 * builtins.c (expand_builtin_strncmp): Convert MIN_EXPR operand
8998 to the same type as the other.
8999 * fold-const.c (fold_binary_loc): Assert expectation.
9001 2018-08-25 Iain Sandoe <iain@sandoe.co.uk>
9003 * config/darwin.c (machopic_legitimize_pic_address): Clean up
9004 extraneous parentheses, dead code section and formatting.
9006 2018-08-24 David Malcolm <dmalcolm@redhat.com>
9009 * diagnostic-show-locus.c (layout::layout): Ensure the margin is
9010 wide enough for jumps in the line-numbering to be visible.
9011 (layout::print_gap_in_line_numbering): New member function.
9012 (layout::calculate_line_spans): When using line numbering, merge
9013 line spans that are only 1 line apart.
9014 (diagnostic_show_locus): When printing line numbers, show gaps in
9015 line numbering directly, rather than printing headers.
9016 (selftest::test_diagnostic_show_locus_fixit_lines): Add test of
9017 line-numbering with multiple line spans.
9018 (selftest::test_fixit_insert_containing_newline_2): Add test of
9019 line-numbering, in which the spans are close enough to be merged.
9021 2018-08-24 Aldy Hernandez <aldyh@redhat.com>
9023 * gimple-ssa-evrp-analyze.c (set_ssa_range_info): Pass value_range
9024 to range_includes_zero_p. Do not special case VR_ANTI_RANGE.
9025 * tree-vrp.c (range_is_nonnull): Remove.
9026 (range_includes_zero_p): Accept value_range instead of min/max.
9027 (extract_range_from_binary_expr_1): Do not early bail on
9029 Use range_includes_zero_p instead of range_is_nonnull.
9030 (extract_range_from_unary_expr): Use range_includes_zero_p instead
9031 of range_is_nonnull.
9032 (vrp_meet_1): Pass value_range to range_includes_zero_p. Do not
9033 special case VR_ANTI_RANGE.
9034 (vrp_finalize): Same.
9035 * tree-vrp.h (range_includes_zero_p): Pass value_range as argument
9037 (range_is_nonnull): Remove.
9038 * vr-values.c (vrp_stmt_computes_nonzero): Use
9039 range_includes_zero_p instead of range_is_nonnull.
9040 (extract_range_basic): Pass value_range to range_includes_zero_p
9041 instead of range_is_nonnull.
9043 2018-08-24 Uros Bizjak <ubizjak@gmail.com>
9045 * emit-rtl.c (init_emit_once): Do not emit MODE_POINTER_BOUNDS RTXes.
9046 * emit-rtl.h (rtl_data): Remove return_bnd.
9047 * explow.c (trunc_int_for_mode): Do not handle POINTER_BOUNDS_MODE_P.
9048 * function.c (diddle_return_value): Do not handle crtl->return_bnd.
9049 * genmodes.c (complete_mode): Do not handle MODE_POINTER_BOUNDS.
9050 (POINTER_BOUNDS_MODE): Remove definition.
9051 (make_pointer_bounds_mode): Remove.
9052 (get_mode_class): Do not handle MODE_POINTER_BOUNDS.
9053 * machmode.h (POINTER_BOUNDS_MODE_P): Remove definition.
9054 (scalare_mode::includes_p): Do not handle MODE_POINTER_BOUNDS.
9055 * mode-classes.def: Do not define MODE_POINTER_BOUNDS.
9056 * stor-layout.c (int_mode_for_mode): Do not handle MODE_POINTER_BOUNDS.
9057 * tree-core.h (enum tree_index): Remove TI_POINTER_BOUNDS_TYPE.
9058 * varasm.c (output_constant_pool_2): Do not handle MODE_POINTER_BOUNDS.
9060 * config/i386/i386-modes.def (BND32, BND64): Remove.
9061 * config/i386/i386.c (dbx_register_map): Remove bound registers.
9062 (dbx64_register_map): Ditto.
9063 (svr4_dbx_register_map): Ditto.
9064 (indirect_thunk_bnd_needed): Remove.
9065 (indirect_thunks_bnd_used): Ditto.
9066 (indirect_return_bnd_needed): Ditto.
9067 (indirect_return_via_cx_bnd): Ditto.
9068 (enum indirect_thunk_prefix): Remove indirect_thunk_prefix_bnd.
9069 (indirect_thunk_name): Remove handling of indirect_thunk_prefix_bnd.
9070 (output_indirect_thunk): Ditto. Remove need_prefix argument.
9071 (output_indirect_thunk_function): Remove handling of
9072 indirect_return_bnd_needed, indirect_return_via_cx_bnd,
9073 indirect_thunk_bnd_needed and indirect_thunks_bnd_used variables.
9074 (ix86_save_reg): Remove handling of crtl->return_bnd.
9075 (ix86_legitimate_constant_p): Remove handling of POINTER_BOUNDS_MODE_P.
9076 (ix86_print_operand_address_as): Remove handling of UNSPEC_BNDMK_ADDR
9077 and UNSPEC_BNDLX_ADDR.
9078 (ix86_output_indirect_branch_via_reg): Remove handling of
9079 indirect_thunk_prefix_bnd.
9080 (ix86_output_indirect_branch_via_push): Ditto.
9081 (ix86_output_function_return): Ditto.
9082 (ix86_output_indirect_function_return): Ditto.
9083 (avoid_func_arg_motion): Do not handle UNSPEC_BNDSTX.
9084 * config/i386/i386.h (FIXED_REGISTERS): Remove bound registers.
9085 (CALL_USED_REGISTERS): Ditto.
9086 (REG_ALLOC_ORDER): Update for removal of bound registers.
9087 (HI_REGISTER_NAMES): Ditto.
9088 * config/i386/i386.md (UNSPEC_BNDMK, UNSPEC_BNDMK_ADDR, UNSPEC_BNDSTX)
9089 (UNSPEC_BNDLDX, UNSPEC_BNDLDX_ADDR, UNSPEC_BNDCL, UNSPEC_BNDCU)
9090 (UNSPEC_BNDCN, UNSPEC_MPX_FENCE): Remove.
9091 (BND0_REG, BND1_REG, BND2_REG, BND3_REG): Remove
9092 (FIRST_PSEUDO_REG): Update.
9093 (BND): Remove mode iterator.
9094 * config/i386/predicates.md (bnd_mem_operator): Remove.
9096 2018-08-24 Richard Sandiford <richard.sandiford@arm.com>
9098 * tree-vect-stmts.c (vectorizable_bswap): Handle variable-length
9101 2018-08-24 Richard Sandiford <richard.sandiford@arm.com>
9103 * tree-vect-slp.c (vect_transform_slp_perm_load): Separate out
9104 the case in which the permute needs only a single element and
9105 repeats for every vector of the result. Extend that case to
9106 handle variable-length vectors.
9107 * tree-vect-stmts.c (vectorizable_load): Update accordingly.
9109 2018-08-24 H.J. Lu <hongjiu.lu@intel.com>
9112 * dwarf2out.c (save_macinfo_strings): Call set_indirect_string
9113 on DW_MACINFO_start_file for -gsplit-dwarf -g3.
9115 2018-08-24 Richard Biener <rguenther@suse.de>
9117 * cfg.h (struct control_flow_graph): Add edge_flags_allocated and
9118 bb_flags_allocated members.
9119 (auto_flag): New RAII class for allocating flags.
9120 (auto_edge_flag): New RAII class for allocating edge flags.
9121 (auto_bb_flag): New RAII class for allocating bb flags.
9122 * cfgloop.c (verify_loop_structure): Allocate temporary edge
9124 * cfganal.c (dfs_enumerate_from): Remove use of visited sbitmap
9125 in favor of temporarily allocated BB flag.
9126 * hsa-brig.c: Re-order includes.
9127 * hsa-dump.c: Likewise.
9128 * hsa-regalloc.c: Likewise.
9129 * print-rtl.c: Likewise.
9130 * profile-count.c: Likewise.
9132 2018-08-24 Segher Boessenkool <segher@kernel.crashing.org>
9135 * config/rs6000/rs6000.c (toc_relative_expr_p): Check that the base is
9138 2018-08-24 Aldy Hernandez <aldyh@redhat.com>
9141 * wide-int-range.cc (wide_int_range_div): Do not ignore result
9142 from wide_int_range_multiplicative_op.
9144 2018-08-23 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
9146 * tree-vect-data-refs.c (vect_grouped_store_supported): Fix typo
9149 2018-08-23 Giuliano Belinassi <giuliano.belinassi@usp.br>
9151 * genmatch.c (parser::parse_operation): Fix typo 'exapnded'
9154 2018-08-23 Alexander Monakov <amonakov@ispras.ru>
9156 * tree-scalar-evolution.c (final_value_replacement_loop): Dump
9157 full GENERIC expression used for replacement.
9159 2018-08-23 Aldy Hernandez <aldyh@redhat.com>
9161 * tree-vrp.c (abs_extent_range): Remove.
9162 (extract_range_into_wide_ints): Pass wide ints by reference.
9163 (extract_range_from_binary_expr_1): Rewrite the *DIV_EXPR code.
9164 Pass wide ints by reference in all calls to
9165 extract_range_into_wide_ints.
9166 * wide-int-range.cc (wide_int_range_div): New.
9167 * wide-int-range.h (wide_int_range_div): New.
9168 (wide_int_range_includes_zero_p): New.
9169 (wide_int_range_zero_p): New.
9171 2018-08-23 Matthew Malcomson <matthew.malcomson@arm.com>
9173 * config/aarch64/aarch64.md (arches): New enum.
9174 (arch): New enum attr.
9175 (arch_enabled): New attr.
9176 (enabled): Now uses arch_enabled only.
9177 (simd, sve, fp16): Removed attribute.
9178 (fp): Attr now defined in terms of 'arch'.
9179 (*mov<mode>_aarch64, *movsi_aarch64, *movdi_aarch64, *movti_aarch64,
9180 *movhf_aarch64, <optab><fcvt_target><GPF:mode>2,
9181 <FCVT_F2FIXED:fcvt_fixed_insn><GPF:mode>3,
9182 <FCVT_FIXED2F:fcvt_fixed_insn><GPI:mode>3): Merge 'fp' and 'simd'
9183 attributes into 'arch'.
9184 (*movsf_aarch64, *movdf_aarch64, *movtf_aarch64, *add<mode>3_aarch64,
9185 subdi3, neg<mode>2, <optab><mode>3, one_cmpl<mode>2,
9186 *<NLOGICAL:optab>_one_cmpl<mode>3, *xor_one_cmpl<mode>3,
9187 *aarch64_ashl_sisd_or_int_<mode>3, *aarch64_lshr_sisd_or_int_<mode>3,
9188 *aarch64_ashr_sisd_or_int_<mode>3, *aarch64_sisd_ushl): Convert use of
9189 'simd' attribute into 'arch'.
9190 (load_pair_sw_<SX:mode><SX2:mode>, load_pair_dw_<DX:mode><DX2:mode>,
9191 store_pair_sw_<SX:mode><SX2:mode>, store_pair_dw_<DX:mode><DX2:mode>):
9192 Convert use of 'fp' attribute to 'arch'.
9193 * config/aarch64/aarch64-simd.md (move_lo_quad_internal_<mode>,
9194 move_lo_quad_internal_<mode>): (different modes) Merge 'fp' and 'simd'
9196 (move_lo_quad_internal_be_<mode>, move_lo_quad_internal_be_<mode>):
9197 (different modes) Merge 'fp' and 'simd' into 'arch'.
9198 (*aarch64_combinez<mode>, *aarch64_combinez_be<mode>): Merge 'fp' and
9201 2018-08-23 Segher Boessenkool <segher@kernel.crashing.org>
9203 PR rtl-optimization/87026
9204 * expmed.c (canonicalize_comparison): If we can no longer create
9205 pseudoregisters, don't.
9207 2018-08-23 Richard Earnshaw <rearnsha@arm.com>
9210 * config/arm/arm-protos.h (arm_emit_speculation_barrier): New
9212 * config/arm/arm.c (speculation_barrier_libfunc): New static
9214 (arm_init_libfuncs): Initialize it.
9215 (arm_emit_speculation_barrier): New function.
9216 * config/arm/arm.md (speculation_barrier): Call
9217 arm_emit_speculation_barrier for architectures that do not have
9219 (speculation_barrier_insn): Only match on Armv7 or later.
9221 2018-08-23 Richard Biener <rguenther@suse.de>
9224 * tree-inline.c (copy_bb): Drop unused __builtin_va_arg_pack_len
9227 2018-08-23 Richard Sandiford <richard.sandiford@arm.com>
9229 * config/aarch64/aarch64.c (aarch64_evpc_sve_tbl): Fix handling
9230 of single-vector TBLs.
9231 (aarch64_vectorize_vec_perm_const): Set one_vector_p when only
9234 2018-08-23 Richard Sandiford <richard.sandiford@arm.com>
9237 * config/aarch64/aarch64.c (aarch64_expand_vec_perm_const_1): Fix
9238 aarch64_evpc_tbl guard.
9240 2018-08-22 Bernd Edlinger <bernd.edlinger@hotmail.de>
9242 * tree-ssa-dse.c (compute_trims): Avoid folding away undefined
9245 2018-08-22 Martin Sebor <msebor@redhat.com>
9248 * tree-pretty-print.c (pretty_print_string): Add argument.
9249 (dump_generic_node): Call to pretty_print_string with string size.
9251 2018-08-22 Segher Boessenkool <segher@kernel.crashing.org>
9253 PR rtl-optimization/86771
9254 * combine.c (try_combine): Do not allow splitting a resulting PARALLEL
9255 of two SETs into those two SETs, one to be placed at i2, if that SETs
9256 destination is modified between i2 and i3.
9258 2018-08-22 Richard Sandiford <richard.sandiford@arm.com>
9260 PR tree-optimization/86725
9261 * tree-vect-loop.c (vect_inner_phi_in_double_reduction_p): New
9263 (vect_analyze_scalar_cycles_1): Check it.
9265 2018-08-22 Richard Sandiford <richard.sandiford@arm.com>
9267 PR tree-optimization/86725
9268 * tree-vect-loop.c (vect_is_simple_reduction): When treating
9269 an outer loop phi as a double reduction, make sure that the
9270 single user of the phi result is an inner loop phi.
9272 2018-08-22 Richard Sandiford <richard.sandiford@arm.com>
9274 * tree-vect-data-refs.c (vect_analyze_group_access_1): Convert
9275 grouped stores with gaps to a strided group.
9277 2018-08-22 Richard Sandiford <richard.sandiford@arm.com>
9279 * tree-vect-stmts.c (get_group_load_store_type)
9280 (get_load_store_type): Only test STMT_VINFO_STRIDED_P for the
9281 first statement in a group.
9283 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
9285 * config/darwin.h (LINK_COMMAND_SPEC_A): Sync LTO options with
9286 the sequence used in gcc/gcc.c.
9288 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
9291 * gcc-ar.c (main): Don’t try to invoke the plug-in if we’re not
9294 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
9296 * config/darwin10.h (LINK_GCC_C_SEQUENCE_SPEC): Adjust to use the
9297 Darwin10-specific unwinder-shim.
9298 * config/darwin12.h (LINK_GCC_C_SEQUENCE_SPEC): Remove.
9299 * config/rs6000/darwin.h (DARWIN_CRT1_SPEC, DARWIN_DYLIB1_SPEC):
9300 New to cater for Darwin10 Rosetta.
9302 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
9304 * config/i386/i386.c (ix86_output_addr_diff_elt): Move the MACH-O
9305 specific test before the one for HAVE_AS_GOTOFF_IN_DATA.
9307 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
9312 * gcc/dwarf2out.c (FUNC_SECOND_SECT_LABEL): New.
9313 (dwarf2out_switch_text_section): Generate a local label for the
9314 second function sub-section and apply it as the second FDE start
9316 * gcc/final.c (final_scan_insn_1): Emit second FDE label after the
9317 second sub-section start.
9319 2018-08-22 Richard Biener <rguenther@suse.de>
9321 PR tree-optimization/86988
9322 * tree-vrp.c (vrp_prop::check_mem_ref): Bail out on VLAs.
9324 2018-08-22 Richard Biener <rguenther@suse.de>
9326 PR tree-optimization/86945
9327 * tree-cfg.c (generate_range_test): Use unsigned arithmetic.
9329 2018-08-22 Alexandre Oliva <oliva@adacore.com>
9331 * config/rs6000/rs6000.c (SMALL_DATA_RELOC, SMALL_DATA_REG): Add
9332 a comment about how uses of r2 for .sdata2 come about.
9334 2018-08-22 Alexandre Oliva <aoliva@redhat.com>
9336 * tree-ssa-reassoc.c (is_reassociable_op): Fix cut&pasto.
9338 2018-08-21 Marek Polacek <polacek@redhat.com>
9340 PR c++/86981, Implement -Wpessimizing-move.
9341 * doc/invoke.texi: Document -Wpessimizing-move.
9343 2018-08-21 Jan Hubicka <jh@suse.cz>
9345 * tree.c (find_decls_types_r): Do not check for redundant typedefs.
9346 * tree.h (is_redundant_typedef): Remove.
9347 * dwarf2out.c (is_redundant_typedef): Turn into static function.
9349 2018-08-21 Jan Hubicka <jh@suse.cz>
9351 * tree.c (free_lang_data_in_decl): Remove types from DECL_CONTEXT
9354 2018-08-21 Tamar Christina <tamar.christina@arm.com>
9356 * expmed.c (extract_low_bits): Reject invalid subregs early.
9358 2018-08-21 Bernd Edlinger <bernd.edlinger@hotmail.de>
9361 * tree-ssa-strlen.c (adjust_last_stmt): Avoid folding away undefined
9364 2018-08-21 Rasmus Villemoes <rv@rasmusvillemoes.dk>
9366 * config/vxworks.h: Guard vxworks_asm_out_constructor and
9367 vxworks_asm_out_destructor by !HAVE_INITFINI_ARRAY_SUPPORT
9368 * config/vxworks.c: Likewise.
9370 2018-08-21 Rasmus Villemoes <rv@rasmusvillemoes.dk>
9372 * config/vxworks.c: Set targetm.have_ctors_dtors
9373 if HAVE_INITFINI_ARRAY_SUPPORT.
9374 * config/vxworks.h: Set SUPPORTS_INIT_PRIORITY
9375 if HAVE_INITFINI_ARRAY_SUPPORT.
9377 2018-08-21 Rasmus Villemoes <rv@rasmusvillemoes.dk>
9379 * config/vxworks.h: Add $(WIND_BASE)/target/h/wrn/coreip to
9380 default search path for VxWorks < 7.
9382 2018-08-21 Bernd Edlinger <bernd.edlinger@hotmail.de>
9384 * gimple-ssa-sprintf.c (decl_constant_value): Remove.
9385 (get_format_string): Refer to c_getstr.
9387 2018-08-21 Tom de Vries <tdevries@suse.de>
9389 * cgraph.h (debuginfo_early_init, debuginfo_init, debuginfo_fini)
9390 (debuginfo_start, debuginfo_stop, debuginfo_early_start)
9391 (debuginfo_early_stop): Declare.
9392 * cgraphunit.c (debuginfo_early_init, debuginfo_init, debuginfo_fini)
9393 (debuginfo_start, debuginfo_stop, debuginfo_early_start)
9394 (debuginfo_early_stop): New function.
9395 (symbol_table::finalize_compilation_unit): Call debuginfo_early_start
9396 and debuginfo_early_stop.
9397 * dwarf2out.c (dwarf2out_finish, dwarf2out_early_finish): Dump dwarf.
9398 * toplev.c (compile_file): Call debuginfo_start and debuginfo_stop.
9399 (general_init): Call debuginfo_early_init.
9400 (finalize): Call debuginfo_fini.
9401 (do_compile): Call debuginfo_init.
9402 * doc/invoke.texi (@gccoptlist): Add -fdump-debug and
9404 (@item -fdump-debug, @item -fdump-earlydebug): Add.
9406 2018-08-21 Tom de Vries <tdevries@suse.de>
9408 * dwarf2out.c (print_dw_val, print_loc_descr, print_die): Handle
9409 flag_dump_noaddr and flag_dump_unnumbered.
9411 2018-08-21 Aldy Hernandez <aldyh@redhat.com>
9413 * wide-int-range.cc (wide_int_range_abs): New.
9414 (wide_int_range_order_set): Rename from wide_int_range_min_max.
9415 * wide-int-range.h (wide_int_range_abs): New.
9416 (wide_int_range_min_max): New.
9417 * tree-vrp.c (extract_range_from_unary_expr): Rewrite ABS_EXPR
9418 case to call wide_int_range_abs.
9419 Rewrite MIN/MAX_EXPR to call wide_int_range_min_max.
9420 (extract_range_from_abs_expr): Delete.
9422 2018-08-20 Michael Meissner <meissner@linux.ibm.com>
9425 * config/rs6000/rs6000.md (extendsi<mode>2): Change constraints
9426 from 'Y' to 'YZ' to enable the LWAX instruction to be generated
9429 2018-08-20 Nathan Sidwell <nathan@acm.org>
9430 Jeff Law <law@redhat.com>
9432 * config/s390/s390-c (s390_macro_to_expand): Use cpp_macro_p.
9433 * config/spu/spu-c.c (spu_macro_to_expand): Likewise.
9435 2018-08-20 David Malcolm <dmalcolm@redhat.com>
9438 * attribs.c (diag_attr_exclusions): Add auto_diagnostic_group instance.
9439 (decl_attributes): Likewise.
9440 * calls.c (maybe_warn_nonstring_arg): Add auto_diagnostic_group
9442 * cgraphunit.c (maybe_diag_incompatible_alias): Likewise.
9443 * diagnostic-core.h (class auto_diagnostic_group): New class.
9444 * diagnostic.c (diagnostic_initialize): Initialize the new fields.
9445 (diagnostic_report_diagnostic): Handle the first diagnostics within
9447 (emit_diagnostic): Add auto_diagnostic_group instance.
9449 (inform_n): Likewise.
9450 (warning): Likewise.
9451 (warning_at): Likewise.
9452 (warning_n): Likewise.
9453 (pedwarn): Likewise.
9454 (permerror): Likewise.
9456 (error_n): Likewise.
9457 (error_at): Likewise.
9459 (fatal_error): Likewise.
9460 (internal_error): Likewise.
9461 (internal_error_no_backtrace): Likewise.
9462 (auto_diagnostic_group::auto_diagnostic_group): New ctor.
9463 (auto_diagnostic_group::~auto_diagnostic_group): New dtor.
9464 * diagnostic.h (struct diagnostic_context): Add fields
9465 "diagnostic_group_nesting_depth",
9466 "diagnostic_group_emission_count", "begin_group_cb",
9468 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior):
9469 Add auto_diagnostic_group instance(s).
9470 (find_explicit_erroneous_behavior): Likewise.
9471 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Likewise.
9472 * gimple-ssa-warn-restrict.c (maybe_diag_offset_bounds): Likewise.
9473 * gimplify.c (warn_implicit_fallthrough_r): Likewise.
9474 (gimplify_va_arg_expr): Likewise.
9475 * hsa-gen.c (HSA_SORRY_ATV): Likewise.
9476 (HSA_SORRY_AT): Likewise.
9477 * ipa-devirt.c (compare_virtual_tables): Likewise.
9478 (warn_odr): Likewise.
9479 * multiple_target.c (expand_target_clones): Likewise.
9480 * opts-common.c (cmdline_handle_error): Likewise.
9481 * reginfo.c (globalize_reg): Likewise.
9482 * substring-locations.c (format_warning_n_va): Likewise.
9483 * tree-inline.c (expand_call_inline): Likewise.
9484 * tree-ssa-ccp.c (pass_post_ipa_warn::execute): Likewise.
9485 * tree-ssa-loop-niter.c
9486 (do_warn_aggressive_loop_optimizations): Likewise.
9487 * tree-ssa-uninit.c (warn_uninit): Likewise.
9488 * tree.c (warn_deprecated_use): Likewise.
9490 2018-08-20 H.J. Lu <hongjiu.lu@intel.com>
9493 * config/i386/i386.md (eh_return): Always update EH return
9494 address in word_mode.
9496 2018-08-20 Chung-Lin Tang <cltang@codesourcery.com>
9498 * targhooks.c (std_gimplify_va_arg_expr): Properly handle case of when
9499 TARGET_SPLIT_COMPLEX_ARG is defined.
9501 2018-08-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
9503 * expr.c (store_field): Change gcc_assert to gcc_checking_assert.
9505 2018-08-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
9508 * expr.c (expand_assignment): Assert that bitpos is positive.
9509 (store_field): Likewise
9510 (expand_expr_real_1): Make sure that bitpos is positive.
9511 * config/alpha/alpha.h (CONSTANT_ADDRESS_P): Avoid signed
9514 2018-08-20 Nathan Sidwell <nathan@acm.org>
9516 * Makefile.in (CPP_ID_DATA_H): Delete.
9517 (CPP_INTERNAL_H): Don't add it.
9518 (GTFILES): Replace CPP_ID_DATA_H with CPPLIB_H.
9519 * gengtype.c (open_base_files): Replace cpp-id-data.h with cpplib.h
9521 2018-08-20 Richard Biener <rguenther@suse.de>
9523 PR tree-optimization/78655
9524 * tree-vrp.c (extract_range_from_binary_expr_1): Make
9525 pointer + offset nonnull if either operand is nonnull work.
9527 2018-08-20 Tom de Vries <tdevries@suse.de>
9529 * dwarf2out.c (add_scalar_info): Don't add reference to existing die
9530 unless the referenced die describes the added property using
9531 DW_AT_location or DW_AT_const_value. Fall back to exprloc case.
9532 Otherwise, add a DW_AT_location to the referenced die.
9534 2018-08-19 Uros Bizjak <ubizjak@gmail.com>
9537 * config/i386/i386.c (ix86_rtx_costs) [case SET]: Check source for
9538 register_operand when calling ix86_set_reg_reg_cost.
9539 [case CONST_INT, case CONST, case LABEL_REF, case SYMBOL_REF]:
9540 Set *total to 0 for operands that satisfy x86_64_immediate_operand
9541 predicate and to 1 otherwise.
9543 2018-08-18 Iain Sandoe <iain@sandoe.co.uk>
9545 * config/darwin.c (darwin_override_options): If -gsplit-dwarf is set,
9546 emit a diagnostic that it is not supported and reset the option.
9547 * config/darwin.h (DRIVER_SELF_SPECS): Note that gsplit-dwarf is not
9548 supported and consume the option. (ASM_FINAL_SPEC): New.
9550 2018-08-17 Segher Boessenkool <segher@kernel.crashing.org>
9552 * doc/md.texi (Patterns): Use @ref instead of @xref in the middle of
9555 2018-08-17 Sandra Loosemore <sandra@codesourcery.com>
9557 C-SKY port: Documentation
9559 * doc/extend.texi (C-SKY Function Attributes): New section.
9560 * doc/invoke.texi (Option Summary): Add C-SKY options.
9561 (C-SKY Options): New section.
9562 * doc/md.texi (Machine Constraints): Document C-SKY constraints.
9564 2018-08-17 Jojo <jijie_rong@c-sky.com>
9565 Huibin Wang <huibin_wang@c-sky.com>
9566 Sandra Loosemore <sandra@codesourcery.com>
9567 Chung-Lin Tang <cltang@codesourcery.com>
9569 C-SKY port: Backend implementation
9571 * config/csky/*: New.
9572 * common/config/csky/*: New.
9574 2018-08-17 Jojo <jijie_rong@c-sky.com>
9575 Huibin Wang <huibin_wang@c-sky.com>
9576 Sandra Loosemore <sandra@codesourcery.com>
9577 Chung-Lin Tang <cltang@codesourcery.com>
9578 Andrew Jenner <andrew@codesourcery.com>
9580 C-SKY port: Configury
9582 * config.gcc (csky-*-*): New.
9583 * configure.ac: Add csky to targets for dwarf2 debug_line support.
9584 * configure: Regenerated.
9586 2018-08-17 David Malcolm <dmalcolm@redhat.com>
9588 * dump-context.h: Include "dumpfile.h".
9589 (dump_context::dump_printf_va): Convert final param from va_list
9590 to va_list *. Convert from ATTRIBUTE_PRINTF to
9591 ATTRIBUTE_GCC_DUMP_PRINTF.
9592 (dump_context::dump_printf_loc_va): Likewise.
9593 * dumpfile.c: Include "stringpool.h".
9594 (make_item_for_dump_printf_va): Delete.
9595 (make_item_for_dump_printf): Delete.
9596 (class dump_pretty_printer): New class.
9597 (dump_pretty_printer::dump_pretty_printer): New ctor.
9598 (dump_pretty_printer::emit_items): New member function.
9599 (dump_pretty_printer::emit_any_pending_textual_chunks): New member
9601 (dump_pretty_printer::emit_item): New member function.
9602 (dump_pretty_printer::stash_item): New member function.
9603 (dump_pretty_printer::format_decoder_cb): New member function.
9604 (dump_pretty_printer::decode_format): New member function.
9605 (dump_context::dump_printf_va): Reimplement in terms of
9606 dump_pretty_printer.
9607 (dump_context::dump_printf_loc_va): Convert final param from va_list
9609 (dump_context::begin_scope): Reimplement call to
9610 make_item_for_dump_printf.
9611 (dump_printf): Update for change to dump_printf_va.
9612 (dump_printf_loc): Likewise.
9613 (selftest::test_capture_of_dump_calls): Convert "stmt" from
9614 greturn * to gimple *. Add a test_decl. Add tests of dump_printf
9615 with %T, %E, and %G.
9616 * dumpfile.h (ATTRIBUTE_GCC_DUMP_PRINTF): New macro.
9617 (dump_printf): Replace ATTRIBUTE_PRINTF_2 with
9618 ATTRIBUTE_GCC_DUMP_PRINTF (2, 3).
9619 (dump_printf_loc): Replace ATTRIBUTE_PRINTF_3 with
9620 ATTRIBUTE_GCC_DUMP_PRINTF (3, 0).
9621 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Convert
9622 use of HOST_WIDE_INT_PRINT_DEC on unsigned HOST_WIDE_INT "count"
9623 within a dump_printf_loc call to "%wu".
9624 (vector_alignment_reachable_p): Merge two dump_printf[_loc] calls,
9625 converting a use of HOST_WIDE_INT_PRINT_DEC to "%wd". Add a
9626 missing space after "=".
9627 * tree-vect-loop.c (vect_analyze_loop_2) Within a dump_printf
9628 call, convert use of HOST_WIDE_INT_PRINT_DEC to "%wd".
9629 * tree-vect-slp.c (vect_slp_bb): Within a dump_printf_loc call,
9630 convert use of HOST_WIDE_INT_PRINT_UNSIGNED to "%wu".
9631 * tree-vectorizer.c (try_vectorize_loop_1): Likewise. Remove
9632 duplicate "vectorized" from message.
9634 2018-08-17 Szabolcs Nagy <szabolcs.nagy@arm.com>
9636 * config/arm/arm-builtins.c (arm_init_simd_builtin_types): Clear
9637 polyNxK_t element's TYPE_STRING_FLAG.
9639 2018-08-17 Segher Boessenkool <segher@kernel.crashing.org>
9641 * config/rs6000/rs6000.md (*cbranch, *creturn): Name these patterns
9642 (they were unnamed before). Fix comments.
9644 2018-08-17 Nathan Sidwell <nathan@acm.org>
9646 * cppbuiltin.c: Include "cpplib.h", not "cpp-id-data.h".
9648 2018-08-17 Richard Biener <rguenther@suse.de>
9650 PR tree-optimization/86841
9651 * wide-int-range.cc (wide_int_range_lshift): Use to_uhwi.
9653 2018-08-17 Martin Liska <mliska@suse.cz>
9655 * common.opt: Remove Warn, Init and Report for options with
9656 Ignore/Deprecated flag. Warning is done automatically for
9658 * config/i386/i386.opt: Likewise.
9659 * config/ia64/ia64.opt: Likewise.
9660 * config/rs6000/rs6000.opt: Likewise.
9661 * cppbuiltin.c (define_builtin_macros_for_compilation_flags):
9662 Remove usage of flag_check_pointer_bounds.
9663 * lto-wrapper.c (merge_and_complain): Do not handle
9664 OPT_fcheck_pointer_bounds.
9665 (append_compiler_options): Likewise.
9666 * opt-functions.awk: Do not handle Deprecated.
9667 * optc-gen.awk: Check that Var, Report and Init are not
9668 used for an option with Ignore/Deprecated flag.
9669 * opts-common.c (decode_cmdline_option): Do not report
9671 (read_cmdline_option): Report warning for OPT_SPECIAL_deprecated
9673 * opts.h (struct cl_option): Remove cl_deprecated flag.
9674 (CL_ERR_DEPRECATED): Remove error enum value.
9676 2018-08-17 Richard Biener <rguenther@suse.de>
9679 * tree-inline.c (copy_bb): When inlining __builtin_va_arg_pack_len ()
9680 across a va-arg-pack using call adjust its return value accordingly.
9682 2018-08-16 Martin Sebor <msebor@redhat.com>
9684 PR tree-optimization/86853
9685 * gimple-ssa-sprintf.c (struct format_result): Rename member.
9686 (struct fmtresult): Add member and initialize it in ctors.
9687 (format_character): Handle %C. Extend range to NUL. Set MAYFAIL.
9688 (format_string): Handle %S the same as %ls. Set MAYFAIL.
9689 (format_directive): Set POSUNDER4K when MAYFAIL is set.
9690 (parse_directive): Handle %C same as %c.
9691 (sprintf_dom_walker::compute_format_length): Adjust.
9692 (is_call_safe): Adjust.
9694 2018-08-16 Bernd Edlinger <bernd.edlinger@hotmail.de>
9696 * builtins.c (c_strlen): Add new parameter eltsize. Use it
9697 for determining how to count the elements.
9698 * builtins.h (c_strlen): Adjust prototype.
9699 * expr.c (string_constant): Add new parameter mem_size.
9700 Set *mem_size appropriately.
9701 * expr.h (string_constant): Adjust protoype.
9702 * gimple-fold.c (get_range_strlen): Add new parameter eltsize.
9703 * gimple-fold.h (get_range_strlen): Adjust prototype.
9704 * gimple-ssa-sprintf.c (get_string_length): Add new parameter eltsize.
9705 (format_string): Call get_string_length with eltsize.
9707 2018-08-16 David Malcolm <dmalcolm@redhat.com>
9709 * diagnostic.c (default_diagnostic_start_span_fn): Call pp_string
9710 to emit the span, rather than setting it as the prefix.
9712 2018-08-16 David Malcolm <dmalcolm@redhat.com>
9714 * diagnostic-show-locus.c (layout::start_annotation_line): Add
9715 "margin_char" parameter, defaulting to space. Use it in place
9716 of pp_space for the initial part of the margin.
9717 (layout::print_leading_fixits): Use '+' when filling the margin
9718 of line-insertion fix-it hints.
9720 2018-08-16 Segher Boessenkool <segher@kernel.crashing.org>
9722 * config/rs6000/rs6000.md (two unnamed define_insn and define_split):
9725 2018-08-16 Segher Boessenkool <segher@kernel.crashing.org>
9727 * config/rs6000/altivec.md: Don't set length attribute to the default
9729 * config/rs6000/darwin.md: Ditto.
9730 * config/rs6000/dfp.md: Ditto.
9731 * config/rs6000/htm.md: Ditto.
9732 * config/rs6000/rs6000.md: Ditto.
9733 * config/rs6000/sync.md: Ditto.
9734 * config/rs6000/vsx.md: Ditto.
9736 2018-08-16 Segher Boessenkool <segher@kernel.crashing.org>
9738 * config/rs6000/altivec.md: Don't set length attribute to the default
9739 value, for branch instructions.
9740 * config/rs6000/darwin.md: Ditto.
9741 * config/rs6000/rs6000.md: Ditto.
9743 2018-08-16 Segher Boessenkool <segher@kernel.crashing.org>
9745 * config/rs6000/rs6000.md (length): Always define as const_int 4.
9746 (unnamed conditional branch define_insn): Set length to 4 or 8
9747 depending on offset.
9748 (<bd>_<mode>): Similar, for alternative 0.
9749 (<bd>tf_<mode>): Ditto.
9751 2018-08-16 Tamar Christina <tamar.christina@arm.com>
9753 * expr.c (copy_blkmode_to_reg): Perform larger copies when safe.
9755 2018-08-16 Matthew Malcomson <matthew.malcomson@arm.com>
9757 * doc/rtl.texi: Replace old RTX class names with new names.
9760 2018-08-16 Vlad Lazar <vlad.lazar@arm.com>
9762 * expmed.h (canonicalize_comparison): New declaration.
9763 * expmed.c (canonicalize_comparison, equivalent_cmp_code): New function.
9764 * expmed.c (emit_store_flag_1): Add call to canonicalize_comparison.
9765 * optabs.c (prepare_cmp_insn): Likewise.
9766 * rtl.h (unsigned_condition_p): New function which checks if a
9767 comparison operator is unsigned.
9769 2018-08-16 Nathan Sidwell <nathan@acm.org>
9771 * config/rs6000/rs6000-c.c (rs6000_macro_to_expend): Use cpp_macro_p.
9772 * config/powerpcspc/powerpcspe-c.c (rs6000_macro_to_expend): Likewise.
9774 2018-08-16 Tamar Christina <tamar.christina@arm.com>
9777 * config/arm/arm.c (arm_can_change_mode_class): Disallow subreg.
9778 * config/arm/neon.md (movv4hf, movv8hf): Refactored to..
9779 (mov<mov>): ..this and enable unconditionally.
9781 2018-08-16 Tamar Christina <tamar.christina@arm.com>
9783 * config/arm/neon.md (*neon_mov<mode>): Remove reg-to-reg alternative.
9785 2018-08-16 Sam Tebbs <sam.tebbs@arm.com>
9787 * config/aarch64/aarch64.opt (mlow-precision-recip-sqrt)
9788 (mlow-precision-sqrt, mlow-precision-div, mverbose-cost-dump): Replace
9789 "Common" with "Target".
9791 2018-08-15 Uros Bizjak <ubizjak@gmail.com>
9793 * config/i386/i386.opt (mmitigate-rop): Mark as deprecated.
9794 * doc/invoke.texi (mmitigate-rop): Remove.
9795 * config/i386/i386.c: Do not include "regrename.h".
9796 (ix86_rop_should_change_byte_p, reg_encoded_number)
9797 (ix86_get_modrm_for_rop, set_rop_modrm_reg_bits, ix86_mitigate_rop):
9799 (ix86_reorg): Remove call to ix86_mitigate_rop.
9800 * config/i386/i386.md (attr "modrm_class"): Remove.
9801 (cmp<mode>_ccno_1, mov<mode>_xor, movstrict<mode>_xor)
9802 (x86_mov<mode>cc_0_m1. x86_mov<mode>cc_0_m1_se)
9803 (x86_mov<mode>cc_0_m1_neg): Remove modrm_class attribute override.
9805 2018-08-15 Will Schmidt <will_schmidt@vnet.ibm.com>
9807 * config/rs6000/rs600.c (rs6000_gimple_fold_builtin): Add entries to
9808 allow folding of mergeh() and mergel() for the float and double types.
9809 (fold_mergehl_helper): Rework to handle building a permute tree
9812 2018-08-15 Uros Bizjak <ubizjak@gmail.com>
9814 * config/i386/i386.c (expand_vec_perm_movs): Enable V4SFmode
9817 2018-08-15 David Malcolm <dmalcolm@redhat.com>
9819 * common.opt (fdiagnostics-show-labels): New option.
9820 * diagnostic-show-locus.c (class layout_range): Add field
9822 (class layout): Add field "m_show_labels_p".
9823 (layout_range::layout_range): Add param "label" and use it to
9825 (make_range): Pass in NULL for new "label" param of layout_range's
9827 (layout::layout): Initialize m_show_labels_p.
9828 (layout::maybe_add_location_range): Pass in loc_range->m_label
9829 when constructing layout_range instances.
9830 (struct line_label): New struct.
9831 (layout::print_any_labels): New member function.
9832 (layout::print_line): Call it if label-printing is enabled.
9833 (selftest::test_one_liner_labels): New test.
9834 (selftest::test_diagnostic_show_locus_one_liner): Call it.
9835 * diagnostic.c (diagnostic_initialize): Initialize
9836 context->show_labels_p.
9837 * diagnostic.h (struct diagnostic_context): Add field
9839 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
9840 -fno-diagnostics-show-labels.
9841 * dwarf2out.c (gen_producer_string): Add
9842 OPT_fdiagnostics_show_labels to the ignored options.
9843 * gcc-rich-location.c (gcc_rich_location::add_expr): Add "label"
9845 (gcc_rich_location::maybe_add_expr): Likewise.
9846 * gcc-rich-location.h (gcc_rich_location::gcc_rich_location): Add
9847 label" param, defaulting to NULL.
9848 (gcc_rich_location::add_expr): Add "label" param.
9849 (gcc_rich_location::maybe_add_expr): Likewise.
9850 (class text_range_label): New class.
9851 (class range_label_for_type_mismatch): New class.
9852 * gimple-ssa-sprintf.c (fmtwarn): Pass NULL for new label params
9853 of format_warning_va.
9854 (fmtwarn_n): Likewise for new params of format_warning_n_va.
9855 * lto-wrapper.c (merge_and_complain): Add
9856 OPT_fdiagnostics_show_labels to the "pick one setting" options.
9857 (append_compiler_options): Likewise to the dropped options.
9858 (append_diag_options): Likewise to the passed-on options.
9859 * opts.c (common_handle_option): Handle the new option.
9860 * selftest-diagnostic.c
9861 (test_diagnostic_context::test_diagnostic_context): Enable
9863 * substring-locations.c: Include "gcc-rich-location.h".
9864 (format_warning_n_va): Add "fmt_label" and "param_label" params
9865 and use them as appropriate.
9866 (format_warning_va): Add "fmt_label" and "param_label" params,
9867 passing them on to format_warning_n_va.
9868 (format_warning_at_substring): Likewise.
9869 (format_warning_at_substring_n): Likewise.
9870 * substring-locations.h (format_warning_va): Add "fmt_label" and
9871 "param_label" params.
9872 (format_warning_n_va): Likewise.
9873 (format_warning_at_substring): Likewise.
9874 (format_warning_at_substring_n): Likewise.
9875 * toplev.c (general_init): Initialize global_dc->show_labels_p.
9877 2018-08-15 Qing Zhao <qing.zhao@oracle.com>
9880 * builtins.c (expand_builtin_memcmp): Do not expand the call
9881 when overflow is detected.
9883 2018-08-15 Martin Sebor <msebor@redhat.com>
9885 PR tree-optimization/71625
9886 * config/aarch64/aarch64-builtins.c
9887 (aarch64_init_simd_builtin_types): Clear Poly8_t's TYPE_STRING_FLAG.
9889 2018-08-15 Ilya Leoshkevich <iii@linux.ibm.com>
9891 * config/s390/s390.c (s390_reorg): Remove loop.
9893 2018-08-15 Iain Sandoe <iain@sandoe.co.uk>
9896 (darwin_function_switched_text_sections): Delete.
9897 * gcc/config/darwin.h
9898 (TARGET_ASM_FUNCTION_SWITCHED_TEXT_SECTIONS): Likewise.
9900 2018-08-15 Iain Sandoe <iain@sandoe.co.uk>
9903 * config/darwin.h: (DEBUG_STR_OFFSETS_SECTION, DEBUG_LOCLISTS_SECTION,
9904 DEBUG_RNGLISTS_SECTION) new macros. (DEBUG_PUBNAMES_SECTION,
9905 DEBUG_PUBTYPES_SECTION) update to include GNU variant.
9907 2018-08-15 Martin Liska <mliska@suse.cz>
9909 PR tree-optimization/86925
9910 * predict.c (expr_expected_value_1): When taking
9911 later predictor, assign also probability.
9912 Use fold_build2_initializer_loc in order to fold
9913 the expression in -frounding-math.
9915 2018-08-14 Allan Sandfeld Jensen <allan.jensen@qt.io>
9917 * config/i386/i386.c (expand_vec_perm_movs): New method matching movs
9919 (expand_vec_perm_1): Try the new method.
9921 2018-08-14 Ilya Leoshkevich <iii@linux.ibm.com>
9924 * lra-lives.c (remove_some_program_points_and_update_live_ranges):
9925 Check whether lra_live_max_point is 0 before dividing.
9927 2018-08-14 Martin Sebor <msebor@redhat.com>
9929 PR tree-optimization/86650
9930 * tree-vrp.c (vrp_prop::check_array_ref): Print an inform message.
9931 (vrp_prop::check_mem_ref): Same.
9933 2018-08-13 Liu Hao <lh_mouse@126.com>
9935 * pretty-print.c (eat_esc_sequence): Swap the foreground and
9936 background colors if the COMMON_LVB_REVERSE_VIDEO flag is set,
9937 and clear it thereafter, as it only works for DBCS.
9939 2018-08-13 Liu Hao <lh_mouse@126.com>
9941 * pretty-print.c (mingw_ansi_fputs): Do not call _close() on the
9942 handle returned by _get_osf_handle().
9944 2018-08-13 Will Schmidt <will_schmidt@vnet.ibm.com>
9946 * gcc/config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support
9947 for folding vec_perm.
9949 2018-08-13 Will Schmidt <will_schmidt@vnet.ibm.com>
9951 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin):
9952 Add support for gimple-folding of vec_pack() and vec_unpack()
9955 2018-08-13 Will Schmidt <will_schmidt@vnet.ibm.com>
9957 * config/rs6000/rs6000.c (rs6000_builtin_valid_without_lhs): Add
9958 vec_xst variants to the list.
9959 (rs6000_gimple_fold_builtin): Add support for folding unaligned
9960 vector loads and stores.
9962 2018-08-13 David Edelsohn <dje.gcc@gmail.com>
9964 * config.gcc (rs6000-ibm-aix4.x): Delete.
9965 (rs6000-ibm-aix5.1): Delete.
9966 (rs6000-ibm-aix5.2): Delete.
9967 (rs6000-ibm-aix5.3): Delete.
9968 * config/rs6000/aix43.h: Delete.
9969 * config/rs6000/aix51.h: Delete.
9970 * config/rs6000/aix52.h: Delete.
9971 * config/rs6000/t-aix43: Delete.
9973 2018-08-13 Ilya Leoshkevich <iii@linux.ibm.com>
9975 * config/s390/s390.c (s390_decompose_constant_pool_ref):
9977 (s390_decompose_address): Factor out constant pool ref
9980 2018-08-12 Chung-Ju Wu <jasonwucj@gmail.com>
9982 * config/nds32/nds32-predicates.c
9983 (nds32_can_use_bclr_p): Change return type as bool.
9984 (nds32_can_use_bset_p): Ditto.
9985 (nds32_can_use_btgl_p): Ditto.
9986 (nds32_can_use_bitci_p): Ditto.
9987 * config/nds32/nds32-protos.h
9988 (nds32_can_use_bclr_p): Change declaration.
9989 (nds32_can_use_bset_p): Ditto.
9990 (nds32_can_use_btgl_p): Ditto.
9991 (nds32_can_use_bitci_p): Ditto.
9993 2018-08-12 Chung-Ju Wu <jasonwucj@gmail.com>
9995 * config/nds32/nds32.c (nds32_expand_prologue, nds32_expand_epilogue):
9996 Support -msched-prolog-epilog option.
9997 * config/nds32/nds32.opt (msched-prolog-epilog): New option.
9999 2018-08-12 Chung-Ju Wu <jasonwucj@gmail.com>
10001 * common/config/nds32/nds32-common.c
10002 (nds32_option_optimization_table): Enalbe -malways-align.
10004 2018-08-12 Chung-Ju Wu <jasonwucj@gmail.com>
10006 * config.gcc (nds32*): Add nds32_isr.h and nds32_init.inc in
10008 * common/config/nds32/nds32-common.c (nds32_handle_option): Handle
10009 OPT_misr_secure_ case.
10010 * config/nds32/nds32-isr.c: Implementation of backward compatibility.
10011 * config/nds32/nds32-protos.h (nds32_isr_function_critical_p): New.
10012 * config/nds32/nds32.c (nds32_attribute_table): Add critical and
10014 * config/nds32/nds32.h (nds32_isr_nested_type): Add NDS32_CRITICAL.
10015 (nds32_isr_info): New field security_level.
10016 (TARGET_ISR_VECTOR_SIZE_4_BYTE): New macro.
10017 * config/nds32/nds32.md (return_internal): Consider critical attribute.
10018 * config/nds32/nds32.opt (misr-secure): New option.
10019 * config/nds32/nds32_init.inc: New file.
10020 * config/nds32/nds32_isr.h: New file.
10022 2018-08-11 John David Anglin <danglin@gcc.gnu.org>
10024 * config/pa/pa.md (UNSPEC_MEMORY_BARRIER): New unspec enum.
10025 Update comment for atomic instructions.
10026 (atomic_storeqi, atomic_storehi, atomic_storesi, atomic_storesf,
10027 atomic_loaddf, atomic_loaddf_1, atomic_storedf, atomic_storedf_1):
10029 (atomic_loaddi): Revise fence expansion to only emit fence prior to
10030 load for __ATOMIC_SEQ_CST model.
10031 (atomic_loaddi_1): Remove float register target.
10032 (atomic_storedi): Handle CONST_INT values.
10033 (atomic_storedi_1): Remove float register source. Add special case
10035 (memory_barrier): New expander and insn.
10037 2018-08-11 Jakub Jelinek <jakub@redhat.com>
10039 PR tree-optimization/86835
10040 * tree-ssa-math-opts.c (insert_reciprocals): Even when inserting
10041 new_stmt after def_gsi, make sure to insert new_square_stmt after
10042 that stmt, not 2 stmts before it.
10044 2018-08-10 Alexander Monakov <amonakov@ispras.ru>
10047 * config/i386/i386.md (<s>mul<mode>3_highpart): Use DWIH mode iterator
10050 2018-08-10 Martin Liska <mliska@suse.cz>
10053 * builtin-types.def (BT_FN_LONG_LONG_LONG_DOUBLE): Add new
10055 * builtins.c (expand_builtin_expect_with_probability):
10057 (expand_builtin_expect_with_probability): New function.
10058 (build_builtin_expect_predicate): Add new argumnet probability
10059 for BUILT_IN_EXPECT_WITH_PROBABILITY.
10060 (fold_builtin_expect):
10063 * builtins.def (BUILT_IN_EXPECT_WITH_PROBABILITY):
10064 * builtins.h (fold_builtin_expect): Set new argument.
10065 * doc/extend.texi: Document __builtin_expect_with_probability.
10066 * doc/invoke.texi: Likewise.
10067 * gimple-fold.c (gimple_fold_call): Pass new argument.
10068 * ipa-fnsummary.c (find_foldable_builtin_expect): Handle
10069 also BUILT_IN_EXPECT_WITH_PROBABILITY.
10070 * predict.c (get_predictor_value): New function.
10071 (expr_expected_value): Add new argument probability. Assume
10072 that predictor and probability are always non-null.
10073 (expr_expected_value_1): Likewise. For __builtin_expect and
10074 __builtin_expect_with_probability set probability. Handle
10075 combination in binary expressions.
10076 (tree_predict_by_opcode): Simplify code by simply calling
10077 get_predictor_value.
10078 (pass_strip_predict_hints::execute): Add handling of
10079 BUILT_IN_EXPECT_WITH_PROBABILITY.
10080 * predict.def (PRED_BUILTIN_EXPECT_WITH_PROBABILITY): Add
10082 * tree.h (DECL_BUILT_IN_P): New function.
10084 2018-08-10 Martin Liska <mliska@suse.cz>
10086 PR tree-optimization/85799
10087 * passes.def: Add argument for pass_strip_predict_hints.
10088 * predict.c (class pass_strip_predict_hints): Add new argument
10090 (strip_predictor_early): New function.
10091 (pass_strip_predict_hints::execute): Call the function to
10093 (strip_predict_hints): New function.
10094 * predict.def: Fix comment.
10096 2018-08-10 Thomas Preud'homme <thomas.preudhomme@linaro.org>
10098 * Makefile.in: Clarify which tm.texi to copy over to assert the
10099 right to grant a GFDL license for all.
10101 2018-08-09 Jeff Law <law@redhat.com>
10103 * config/m68k/m68k.c (m68k_adjust_decorated_operand): Remove
10106 2018-08-09 Andreas Schwab <schwab@linux-m68k.org>
10108 * config/m68k/m68k-protos.h (m68k_final_prescan_insn): Remove
10111 2018-08-09 Richard Sandiford <richard.sandiford@arm.com>
10113 * tree-vect-loop.c (vectorizable_reduction): Allow inner-loop
10114 reductions for variable-length vectors.
10116 2018-08-09 David Malcolm <dmalcolm@redhat.com>
10119 * common.opt (fdiagnostics-show-line-numbers): New option.
10120 * diagnostic-show-locus.c (class layout): Add fields
10121 "m_show_line_numbers_p" and "m_linenum_width";
10122 (num_digits): New function.
10123 (test_num_digits): New function.
10124 (layout::layout): Initialize new fields. Update m_x_offset
10125 logic to handle any left margin.
10126 (layout::print_source_line): Print line number when requested.
10127 (layout::start_annotation_line): New member function.
10128 (layout::print_annotation_line): Call it.
10129 (layout::print_leading_fixits): Likewise.
10130 (layout::print_trailing_fixits): Likewise. Update calls to
10131 move_to_column for new parameter.
10132 (layout::get_x_bound_for_row): Add "add_left_margin" param and use
10133 it to potentially call start_annotation_line.
10134 (layout::show_ruler): Call start_annotation_line.
10135 (selftest::test_line_numbers_multiline_range): New selftest.
10136 (selftest::diagnostic_show_locus_c_tests): Call test_num_digits
10137 and selftest::test_line_numbers_multiline_range.
10138 * diagnostic.c (diagnostic_initialize): Initialize
10139 show_line_numbers_p.
10140 * diagnostic.h (struct diagnostic_context): Add field
10141 "show_line_numbers_p".
10142 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
10143 -fno-diagnostics-show-line-numbers.
10144 * dwarf2out.c (gen_producer_string): Add
10145 OPT_fdiagnostics_show_line_numbers to the ignored options.
10146 * lto-wrapper.c (merge_and_complain): Likewise to the "pick
10147 one setting" options.
10148 (append_compiler_options): Likewise to the dropped options.
10149 (append_diag_options): Likewise to the passed-on options.
10150 * opts.c (common_handle_option): Handle the new option.
10151 * toplev.c (general_init): Set up global_dc->show_line_numbers_p.
10153 2018-08-09 Kelvin Nilsen <kelvin@gcc.gnu.org>
10155 * doc/extend.texi (PowerPC AltiVec Built-in Functions Available on
10156 ISA 2.07): Correct spelling of bcdsub to be __builtin_bcdsub. Add
10157 third argument of type "const signed char" to descriptions of
10158 __builtin_bcdadd, __builtin_bcdadd_lt, __builtin_bcdadd_eq,
10159 __builtin_bcdadd_gt, __builtin_bcdadd_ov, __builtin_bcdsub,
10160 __builtin_bcdsub_lt, __builtin_bcdsub_eq, __builtin_bcdsub_gt,
10161 __builtin_bcdsub_ov functions.
10163 2018-08-09 Richard Sandiford <richard.sandiford@arm.com>
10165 PR tree-optimization/86858
10166 * tree-vect-loop.c (vect_is_simple_reduction): Restore
10167 flow_bb_inside_loop_p calls.
10169 2018-08-09 Richard Sandiford <richard.sandiford@arm.com>
10171 PR tree-optimization/86871
10172 * tree-vect-stmts.c (vect_transform_stmt): Use gimple_get_lhs
10173 instead of gimple_assign_lhs.
10175 2018-08-09 Richard Earnshaw <rearnsha@arm.com>
10178 * config/aarch64/aarch64.md (add<mode>3_carryinC_zero): Add missing
10179 register constraint to operand 0.
10180 (add<mode>3_carryinC): Likewise.
10181 (add<mode>3_carryinV_zero, add<mode>3_carryinV): Likewise.
10183 2018-08-09 Martin Liska <mliska@suse.cz>
10186 * common.opt: Remove extra line.
10188 2018-08-09 Martin Liska <mliska@suse.cz>
10190 * params.def (PARAM_ALIGN_LOOP_ITERATIONS): Remove double dots
10191 at the end of a line, make first letter capital and end up
10192 a sentence with a dot.
10193 (PARAM_LOOP_INTERCHANGE_STRIDE_RATIO): Likewise.
10194 (PARAM_LOOP_BLOCK_TILE_SIZE): Likewise.
10195 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Likewise.
10196 (PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP): Likewise.
10197 (PARAM_MAX_ISL_OPERATIONS): Likewise.
10198 (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): Likewise.
10199 (PARAM_PROFILE_FUNC_INTERNAL_ID): Likewise.
10200 (PARAM_INDIR_CALL_TOPN_PROFILE): Likewise.
10201 (PARAM_SLP_MAX_INSNS_IN_BB): Likewise.
10202 (PARAM_IPA_CP_EVAL_THRESHOLD): Likewise.
10203 (PARAM_IPA_CP_RECURSION_PENALTY): Likewise.
10204 (PARAM_IPA_CP_SINGLE_CALL_PENALTY): Likewise.
10205 (PARAM_IPA_CP_LOOP_HINT_BONUS): Likewise.
10206 (PARAM_IPA_CP_ARRAY_INDEX_HINT_BONUS): Likewise.
10207 (PARAM_TREE_REASSOC_WIDTH): Likewise.
10208 (PARAM_HSA_GEN_DEBUG_STORES): Likewise.
10209 (PARAM_MAX_SPECULATIVE_DEVIRT_MAYDEFS): Likewise.
10210 (PARAM_MAX_VRP_SWITCH_ASSERTIONS): Likewise.
10212 2018-08-09 Andreas Krebbel <krebbel@linux.ibm.com>
10215 * config/s390/s390.c (s390_option_override_internal): Reduce the
10216 stack-clash-protection-probe-interval param if it would be too big
10219 2018-08-08 Andreas Schwab <schwab@linux-m68k.org>
10222 * config/m68k/m68k.h (FINAL_PRESCAN_INSN): Don't define.
10223 * config/m68k/m68k.c (handle_move_double): Don't call
10224 m68k_final_prescan_insn.
10225 (m68k_adjust_decorated_operand): Renamed from
10226 m68k_final_prescan_insn, remove first and third operand and
10228 (print_operand): Call it.
10229 (print_operand_address): Call it.
10231 2018-08-08 Nathan Sidwell <nathan@acm.org>
10233 * diagnostic.c (diagnostic_report_current_module): Use
10234 linemap_included_from & linemap_included_from_linemap.
10236 2018-08-08 Hongbo Zhang <hongbo.zhang@linaro.org>
10238 * config/aarch64/aarch64-cores.def: Add phecda core.
10239 * config/aarch64/aarch64-tune.md: Regenerate.
10240 * doc/invoke.texi: Add phecda core.
10242 2018-08-08 Andreas Krebbel <krebbel@linux.ibm.com>
10245 * config/s390/constraints.md ("NxHD0", "NxSD0"): New constraint
10247 * config/s390/s390.md ("movti"): Add more alternatives for
10248 constant to GPR copies.
10250 2018-08-08 Andreas Krebbel <krebbel@linux.ibm.com>
10252 * config/s390/s390.c: Fix whitespace damage throughout the file.
10253 * config/s390/s390.h: Likewise.
10254 * config/s390/tpf.h: Likewise.
10256 2018-08-08 Ilya Leoshkevich <iii@linux.ibm.com>
10258 * config/s390/s390.c (s390_loadrelative_operand_p):
10259 Remove TARGET_CPU_ZARCH usages.
10260 (s390_rtx_costs): Likewise.
10261 (s390_legitimate_constant_p): Likewise.
10262 (s390_cannot_force_const_mem): Likewise.
10263 (legitimate_reload_constant_p): Likewise.
10264 (s390_preferred_reload_class): Likewise.
10265 (legitimize_pic_address): Likewise.
10266 (legitimize_tls_address): Likewise.
10267 (s390_split_branches): Removed.
10268 (s390_add_execute): Removed.
10269 (s390_dump_pool): Remove TARGET_CPU_ZARCH usages.
10270 (s390_mainpool_start): Likewise.
10271 (s390_mainpool_finish): Likewise.
10272 (s390_mainpool_cancel): Removed.
10273 (s390_chunkify_start): Remove TARGET_CPU_ZARCH usages.
10274 (s390_chunkify_cancel): Likewise.
10275 (s390_return_addr_rtx): Likewise.
10276 (s390_register_info): Remove split_branches_pending_p uages.
10277 (s390_optimize_register_info): Likewise.
10278 (s390_init_frame_layout): Remove TARGET_CPU_ZARCH and
10279 split_branches_pending_p usages.
10280 (s390_can_eliminate): Remove TARGET_CPU_ZARCH usages.
10281 (s390_load_got): Likewise.
10282 (s390_expand_split_stack_prologue): Likewise.
10283 (output_asm_nops): Likewise.
10284 (s390_function_profiler): Likewise.
10285 (s390_emit_call): Likewise.
10286 (s390_conditional_register_usage): Likewise.
10287 (s390_optimize_prologue): Likewise.
10288 (s390_reorg): Remove TARGET_CPU_ZARCH and
10289 split_branches_pending_p usages.
10290 (s390_option_override_internal): Remove TARGET_CPU_ZARCH
10292 (s390_output_indirect_thunk_function): Likewise.
10293 * config/s390/s390.h (TARGET_CPU_ZARCH): Removed.
10294 (TARGET_CPU_ZARCH_P): Removed.
10295 (struct machine_function): Remove split_branches_pending_p.
10296 * config/s390/s390.md: Remove TARGET_CPU_ZARCH usages.
10298 2018-08-08 Ilya Leoshkevich <iii@linux.ibm.com>
10300 * common/config/s390/s390-common.c (processor_flags_table):
10302 * config.gcc: Remove with_arch/with_tune support.
10303 * config/s390/2064.md: Remove cpu attribute comparisons.
10304 * config/s390/driver-native.c (s390_host_detect_local_cpu):
10306 * config/s390/linux.h (ASM_SPEC):
10307 Remove -march support.
10308 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal):
10309 Use a table to get an arch level.
10310 * config/s390/s390-opts.h (enum processor_type):
10311 Remove enum values.
10312 * config/s390/s390.c
10313 (processor_table): Remove entries, add arch_level values.
10314 (s390_issue_rate): Remove cases.
10315 (s390_option_override): Adjust
10316 s390_option_override_internal() call.
10317 (s390_option_override_internal): Remove deprecation warning.
10318 (s390_valid_target_attribute_tree): Adjust
10319 s390_option_override_internal() call.
10320 * config/s390/s390.h (struct s390_processor):
10321 Share with s390-c.c, add arch_level field.
10322 * config/s390/s390.md:
10323 Remove occurrences in cpu attribute.
10324 * config/s390/s390.opt: Remove -march/-mtune support.
10325 * config/s390/tpf.h (ASM_SPEC): Remove -march support.
10326 * doc/invoke.texi: Remove deprecation warning.
10328 2018-08-08 Luis Machado <luis.machado@linaro.org>
10330 * config/aarch64/aarch64.c (qdf24xx_vector_cost): New static
10332 (qdf24xx_tunings): Set vector cost structure to
10333 qdf24xx_vector_cost.
10335 * config/aarch64/aarch64.c (qdf24xx_addrcost_table)
10336 <register_sextend>: Set to 3.
10338 2018-08-07 Richard Sandiford <richard.sandiford@arm.com>
10341 * config/aarch64/iterators.md (FRECP, frecp_suffix): Delete.
10342 * config/aarch64/aarch64-simd.md
10343 (aarch64_frecp<FRECP:frecp_suffix><mode>): Fold FRECPE into...
10344 (@aarch64_frecpe<mode>): ...here and the move FRECPX to...
10345 (aarch64_frecpx<mode>): ...this new pattern.
10346 * config/aarch64/aarch64-simd-builtins.def: Remove comment
10347 about aarch64_frecp<FRECP:frecp_suffix><mode>.
10349 2018-08-07 Martin Liska <mliska@suse.cz>
10351 PR middle-end/83023
10352 * predict.c (expr_expected_value_1): Handle DECL_IS_MALLOC,
10353 BUILT_IN_REALLOC and DECL_IS_OPERATOR_NEW.
10354 * predict.def (PRED_MALLOC_NONNULL): New predictor.
10355 * doc/extend.texi: Document that malloc attribute adds
10358 2018-08-06 John David Anglin <danglin@gcc.gnu.org>
10361 * config/pa/pa.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
10362 Define to speculation_safe_value_not_needed.
10364 2018-08-06 Jeff Law <law@redhat.com>
10366 * tree-ssa-dom.c (dom_opt_dom_walker::optimize_stmt): Pass down
10367 the vr_values instance to cprop_into_stmt.
10368 (cprop_into_stmt): Pass vr_values instance down to cprop_operand.
10369 (cprop_operand): Also query EVRP to determine if OP is a constant.
10371 2018-08-06 Nathan Sidwell <nathan@acm.org>
10373 * diagnostic.c (diagnostic_report_current_module): Reroll
10374 included-at loop. Translate text.
10376 2018-08-06 David Malcolm <dmalcolm@redhat.com>
10378 * function-tests.c (selftest::test_expansion_to_rtl): Call
10379 free_after_compilation.
10381 2018-08-06 Alan Hayward <alan.hayward@arm.com>
10383 * config/aarch64/aarch64.md: Add clobber highs to tls_desc.
10385 2018-08-06 Andreas Krebbel <krebbel@linux.ibm.com>
10387 * config/s390/s390.c (s390_loop_unroll_adjust): Prevent small
10388 loops with memory block operations from getting unrolled.
10390 2018-08-06 Ulrich Weigand <uweigand@de.ibm.com>
10393 * config/spu/spu.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
10394 Define to speculation_safe_value_not_needed.
10396 2018-08-06 Jeff Law <law@redhat.com>
10398 * reload1.c (forget_old_reloads_1): Adjust CLOBBER_HIGH
10401 2018-08-06 Jozef Lawrynowicz <jozef.l@mittosystems.com>
10404 * gcc/tree.c (build_common_tree_nodes): Initialize integer_types array
10405 with all enabled __intN types.
10407 * gcc/testsuite/gcc.target/msp430/pr86662.c: New test.
10409 2018-08-06 Alan Hayward <alan.hayward@arm.com>
10411 * alias.c (record_set): Check for clobber high.
10412 * cfgexpand.c (expand_gimple_stmt): Likewise.
10413 * combine-stack-adj.c (single_set_for_csa): Likewise.
10414 * combine.c (find_single_use_1): Likewise.
10415 (set_nonzero_bits_and_sign_copies): Likewise.
10416 (get_combine_src_dest): Likewise.
10417 (is_parallel_of_n_reg_sets): Likewise.
10418 (try_combine): Likewise.
10419 (record_dead_and_set_regs_1): Likewise.
10420 (reg_dead_at_p_1): Likewise.
10421 (reg_dead_at_p): Likewise.
10422 * dce.c (deletable_insn_p): Likewise.
10423 (mark_nonreg_stores_1): Likewise.
10424 (mark_nonreg_stores_2): Likewise.
10425 * df-scan.c (df_find_hard_reg_defs): Likewise.
10426 (df_uses_record): Likewise.
10427 (df_get_call_refs): Likewise.
10428 * dwarf2out.c (mem_loc_descriptor): Likewise.
10429 * haifa-sched.c (haifa_classify_rtx): Likewise.
10430 * ira-build.c (create_insn_allocnos): Likewise.
10431 * ira-costs.c (scan_one_insn): Likewise.
10432 * ira.c (equiv_init_movable_p): Likewise.
10433 (rtx_moveable_p): Likewise.
10434 (interesting_dest_for_shprep): Likewise.
10435 * jump.c (mark_jump_label_1): Likewise.
10436 * postreload-gcse.c (record_opr_changes): Likewise.
10437 * postreload.c (reload_cse_simplify): Likewise.
10438 (struct reg_use): Add source expr.
10439 (reload_combine): Check for clobber high.
10440 (reload_combine_note_use): Likewise.
10441 (reload_cse_move2add): Likewise.
10442 (move2add_note_store): Likewise.
10443 * print-rtl.c (print_pattern): Likewise.
10444 * recog.c (decode_asm_operands): Likewise.
10445 (store_data_bypass_p): Likewise.
10446 (if_test_bypass_p): Likewise.
10447 * regcprop.c (kill_clobbered_value): Likewise.
10448 (kill_set_value): Likewise.
10449 * reginfo.c (reg_scan_mark_refs): Likewise.
10450 * reload1.c (maybe_fix_stack_asms): Likewise.
10451 (eliminate_regs_1): Likewise.
10452 (elimination_effects): Likewise.
10453 (mark_not_eliminable): Likewise.
10454 (scan_paradoxical_subregs): Likewise.
10455 (forget_old_reloads_1): Likewise.
10456 * reorg.c (find_end_label): Likewise.
10457 (try_merge_delay_insns): Likewise.
10458 (redundant_insn): Likewise.
10459 (own_thread_p): Likewise.
10460 (fill_simple_delay_slots): Likewise.
10461 (fill_slots_from_thread): Likewise.
10462 (dbr_schedule): Likewise.
10463 * resource.c (update_live_status): Likewise.
10464 (mark_referenced_resources): Likewise.
10465 (mark_set_resources): Likewise.
10466 * rtl.c (copy_rtx): Likewise.
10467 * rtlanal.c (reg_referenced_p): Likewise.
10468 (single_set_2): Likewise.
10469 (noop_move_p): Likewise.
10470 (note_stores): Likewise.
10471 * sched-deps.c (sched_analyze_reg): Likewise.
10472 (sched_analyze_insn): Likewise.
10474 2018-08-06 Alan Hayward <alan.hayward@arm.com>
10476 * cse.c (invalidate_reg): New function extracted from...
10477 (invalidate): ...here.
10478 (canonicalize_insn): Check for clobber high.
10479 (invalidate_from_clobbers): invalidate clobber highs.
10480 (invalidate_from_sets_and_clobbers): Likewise.
10481 (count_reg_usage): Check for clobber high.
10482 (insn_live_p): Likewise.
10483 * cselib.c (cselib_expand_value_rtx_1):Likewise.
10484 (cselib_invalidate_regno): Check for clobber in setter.
10485 (cselib_invalidate_rtx): Pass through setter.
10486 (cselib_invalidate_rtx_note_stores):
10487 (cselib_process_insn): Check for clobber high.
10488 * cselib.h (cselib_invalidate_rtx): Add operand.
10490 2018-08-06 Alan Hayward <alan.hayward@arm.com>
10492 * lra-eliminations.c (lra_eliminate_regs_1): Check for clobber high.
10493 (mark_not_eliminable): Likewise.
10494 * lra-int.h (struct lra_insn_reg): Add clobber high marker.
10495 * lra-lives.c (process_bb_lives): Check for clobber high.
10496 * lra.c (new_insn_reg): Remember clobber highs.
10497 (collect_non_operand_hard_regs): Check for clobber high.
10498 (lra_set_insn_recog_data): Likewise.
10499 (add_regs_to_insn_regno_info): Likewise.
10500 (lra_update_insn_regno_info): Likewise.
10502 2018-08-06 Alan Hayward <alan.hayward@arm.com>
10504 * rtl.h (reg_is_clobbered_by_clobber_high): Add declarations.
10505 * rtlanal.c (reg_is_clobbered_by_clobber_high): Add function.
10507 2018-08-06 Alan Hayward <alan.hayward@arm.com>
10509 * emit-rtl.c (verify_rtx_sharing): Check for CLOBBER_HIGH.
10510 (copy_insn_1): Likewise.
10511 (gen_hard_reg_clobber_high): New gen function.
10512 * genconfig.c (walk_insn_part): Check for CLOBBER_HIGH.
10513 * genemit.c (gen_exp): Likewise.
10514 (gen_emit_seq): Pass through info.
10515 (gen_insn): Check for CLOBBER_HIGH.
10516 (gen_expand): Pass through info.
10517 (gen_split): Likewise.
10518 (output_add_clobbers): Likewise.
10519 * genrecog.c (validate_pattern): Check for CLOBBER_HIGH.
10520 (remove_clobbers): Likewise.
10521 * rtl.h (gen_hard_reg_clobber_high): New declaration.
10523 2018-08-06 Alan Hayward <alan.hayward@arm.com>
10525 * doc/rtl.texi (clobber_high): Add.
10526 (parallel): Add in clobber high
10527 * rtl.c (rtl_check_failed_code3): Add function.
10528 * rtl.def (CLOBBER_HIGH): Add expression.
10529 * rtl.h (RTL_CHECKC3): Add macro.
10530 (rtl_check_failed_code3): Add declaration.
10531 (XC3EXP): Add macro.
10533 2018-08-05 H.J. Lu <hongjiu.lu@intel.com>
10536 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Set
10537 cfun->machine->max_used_stack_alignment if needed.
10539 2018-08-04 Martin Sebor <msebor@redhat.com>
10541 PR tree-optimization/86571
10542 * gimple-ssa-sprintf.c (format_floating): Extend upper bound of
10545 2018-08-03 Sandra Loosemore <sandra@codesourcery.com>
10548 * config/nios2/nios2.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
10551 2018-08-03 Jeff Law <law@redhat.com>
10554 * config/mn10300/mn10300.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
10555 Define to speculation_safe_value_not_needed.
10557 2018-08-03 David Malcolm <dmalcolm@redhat.com>
10559 * doc/gcov.texi (-x): Remove duplicate "to".
10560 * doc/invoke.texi (-Wnoexcept-type): Remove duplicate "calls".
10561 (-Wif-not-aligned): Remove duplicate "is".
10562 (-flto): Remove duplicate "the".
10563 (MicroBlaze Options): In examples of "-mcpu=cpu-type", remove
10564 duplicate "v5.00.b".
10565 (MSP430 Options): Remove duplicate "and" from the description
10566 of "-mgprel-sec=regexp".
10567 (x86 Options): Remove duplicate copies of "vmldLog102" and
10568 vmlsLog104 from description of "-mveclibabi=type".
10570 2018-08-03 Richard Sandiford <richard.sandiford@arm.com>
10572 * internal-fn.h (first_commutative_argument): Declare.
10573 * internal-fn.c (first_commutative_argument): New function.
10574 * tree-vect-slp.c (vect_get_and_check_slp_defs): Remove extra
10575 restrictions for pattern statements. Use first_commutative_argument
10576 to look for commutative operands in calls to internal functions.
10578 2018-08-03 Aldy Hernandez <aldyh@redhat.com>
10580 * Makefile.in (wide-int-range.o): New.
10581 * tree-vrp.c: Move all the wide_int_* functions to...
10582 * wide-int-range.cc: ...here.
10583 * tree-vrp.h: Move all the wide_int_* prototypes to...
10584 * wide-int-range.h: ...here.
10586 2018-08-03 Tom de Vries <tdevries@suse.de>
10588 * common/config/nvptx/nvptx-common.c (nvptx_except_unwind_info): Return
10590 * config/nvptx/nvptx.c (TARGET_ASM_BYTE_OP): Remove define.
10591 * except.c (output_function_exception_table): Do early exit if
10592 targetm_common.except_unwind_info (&global_options) == UI_NONE.
10594 2018-08-03 Martin Liska <mliska@suse.cz>
10596 * predict.c (dump_prediction): Change to 2 digits
10599 2018-08-03 Siddhesh Poyarekar <siddhesh@sourceware.org>
10601 * config/aarch64/falkor.md (falkor_am_1_vxvy_vxvy): Move
10603 (falkor_am_1_gtov_gtov): ... a new insn reservation.
10605 2018-07-19 Ilya Leoshkevich <iii@linux.ibm.com>
10607 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Replace > with >=.
10608 * df-problems.c (df_remove_dead_eq_notes): Replace > with >=.
10609 * dwarf2out.c (mem_loc_descriptor): Replace > with >=.
10610 * lra-constraints.c (spill_hard_reg_in_range): Replace <= with <.
10611 * lra-remat.c (call_used_input_regno_present_p): Replace <= with <.
10613 2018-08-02 David Malcolm <dmalcolm@redhat.com>
10615 * diagnostic-show-locus.c (diagnostic_show_locus): Use
10616 pp_take_prefix when saving the existing prefix.
10617 * diagnostic.c (diagnostic_append_note): Likewise.
10618 * langhooks.c (lhd_print_error_function): Likewise.
10619 * pretty-print.c (pp_set_prefix): Drop the "const" from "prefix"
10620 param's type. Free the existing prefix.
10621 (pp_take_prefix): New function.
10622 (pretty_printer::pretty_printer): Drop the prefix parameter.
10623 Rename the length parameter to match the comment.
10624 (pretty_printer::~pretty_printer): Free the prefix.
10625 * pretty-print.h (pretty_printer::pretty_printer): Drop the prefix
10627 (struct pretty_printer): Drop the "const" from "prefix" field's
10628 type and clarify memory management.
10629 (pp_set_prefix): Drop the "const" from the 2nd param.
10630 (pp_take_prefix): New decl.
10632 2018-08-02 Aaron Sawdey <acsawdey@linux.ibm.com>
10634 * config/rs6000/rs6000-string.c (select_block_compare_mode): Move test
10635 for word_mode_ok here instead of passing as argument.
10636 (expand_block_compare): Change select_block_compare_mode() call.
10637 (expand_strncmp_gpr_sequence): New function.
10638 (expand_strn_compare): Make use of expand_strncmp_gpr_sequence.
10640 2018-08-02 Jeff Law <law@redhat.com>
10643 * config/m68k/m68k.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
10644 Define to speculation_safe_value_not_needed.
10647 * config/h8300/h8300.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
10648 Define to speculation_safe_value_not_needed.
10650 2018-08-02 Tom de Vries <tdevries@suse.de>
10653 * common/config/nvptx/nvptx-common.c (nvptx_except_unwind_info): New
10654 function. Return UI_TARGET unconditionally.
10655 (TARGET_EXCEPT_UNWIND_INFO): Redefine to nvptx_except_unwind_info.
10656 * config/nvptx/nvptx.c (TARGET_ASM_BYTE_OP): Emit commented out '.byte'.
10658 2018-08-02 Richard Sandiford <richard.sandiford@arm.com>
10660 * genemit.c (print_overload_test): Fix typo.
10662 2018-08-02 Richard Biener <rguenther@suse.de>
10664 PR tree-optimization/86816
10665 * tree-ssa-tail-merge.c (tail_merge_valueize): New function
10666 which checks for value availability before querying it.
10667 (gvn_uses_equal): Use it.
10668 (same_succ_hash): Likewise.
10669 (gimple_equal_p): Likewise.
10671 2018-08-02 Nick Clifton <nickc@redhat.com>
10674 * config/stormy16/stormy16.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
10675 Define to speculation_safe_value_not_needed.
10678 * config/v850/v850.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
10679 Define to speculation_safe_value_not_needed.
10682 * config/v850/v850.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
10683 Define to speculation_safe_value_not_needed.
10686 * config/rx/rx.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
10687 Define to speculation_safe_value_not_needed.
10690 * config/msp430/msp430.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
10691 Define to speculation_safe_value_not_needed.
10694 * config/mcore/mcore.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
10695 Define to speculation_safe_value_not_needed.
10698 * config/m32r/m32r.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
10699 Define to speculation_safe_value_not_needed.
10702 * config/iq2000/iq2000.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
10703 Define to speculation_safe_value_not_needed.
10706 * config/frv/frv.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Define to
10707 speculation_safe_value_not_needed.
10710 * config/fr30/fr30.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Define
10711 to speculation_safe_value_not_needed.
10713 2018-08-02 Richard Sandiford <richard.sandiford@arm.com>
10715 * doc/md.texi: Expand the documentation of instruction names
10716 to mention port-local uses. Document '@' in pattern names.
10717 * read-md.h (overloaded_instance, overloaded_name): New structs.
10718 (mapping): Declare.
10719 (md_reader::handle_overloaded_name): New member function.
10720 (md_reader::get_overloads): Likewise.
10721 (md_reader::m_first_overload): New member variable.
10722 (md_reader::m_next_overload_ptr): Likewise.
10723 (md_reader::m_overloads_htab): Likewise.
10724 * read-md.c (md_reader::md_reader): Initialize m_first_overload,
10725 m_next_overload_ptr and m_overloads_htab.
10726 * read-rtl.c (iterator_group): Add "type" and "get_c_token" fields.
10727 (get_mode_token, get_code_token, get_int_token): New functions.
10728 (map_attr_string): Add an optional argument that passes back
10729 the associated iterator.
10730 (overloaded_name_hash, overloaded_name_eq_p, named_rtx_p):
10731 (md_reader::handle_overloaded_name, add_overload_instance): New
10733 (apply_iterators): Handle '@' names. Report an error if '@'
10734 is used without iterators.
10735 (initialize_iterators): Initialize the new iterator_group fields.
10736 * genopinit.c (handle_overloaded_code_for)
10737 (handle_overloaded_gen): New functions.
10738 (main): Use them to print declarations of maybe_code_for_* and
10739 maybe_gen_* functions, and inline definitions of code_for_* and gen_*.
10740 * genemit.c (print_overload_arguments, print_overload_test)
10741 (handle_overloaded_code_for, handle_overloaded_gen): New functions.
10742 (main): Use it to print definitions of maybe_code_for_* and
10743 maybe_gen_* functions.
10744 * config/aarch64/aarch64.c (aarch64_split_128bit_move): Use
10745 gen_aarch64_mov{low,high}_di and gen_aarch64_movdi_{low,high}
10746 instead of explicit mode checks.
10747 (aarch64_split_simd_combine): Likewise gen_aarch64_simd_combine.
10748 (aarch64_split_simd_move): Likewise gen_aarch64_split_simd_mov.
10749 (aarch64_emit_load_exclusive): Likewise gen_aarch64_load_exclusive.
10750 (aarch64_emit_store_exclusive): Likewise gen_aarch64_store_exclusive.
10751 (aarch64_expand_compare_and_swap): Likewise
10752 gen_aarch64_compare_and_swap and gen_aarch64_compare_and_swap_lse
10753 (aarch64_gen_atomic_cas): Likewise gen_aarch64_atomic_cas.
10754 (aarch64_emit_atomic_swap): Likewise gen_aarch64_atomic_swp.
10755 (aarch64_constant_pool_reload_icode): Delete.
10756 (aarch64_secondary_reload): Use code_for_aarch64_reload_movcp
10757 instead of aarch64_constant_pool_reload_icode. Use
10758 code_for_aarch64_reload_mov instead of explicit mode checks.
10759 (rsqrte_type, get_rsqrte_type, rsqrts_type, get_rsqrts_type): Delete.
10760 (aarch64_emit_approx_sqrt): Use gen_aarch64_rsqrte instead of
10761 get_rsqrte_type and gen_aarch64_rsqrts instead of gen_rqrts_type.
10762 (recpe_type, get_recpe_type, recps_type, get_recps_type): Delete.
10763 (aarch64_emit_approx_div): Use gen_aarch64_frecpe instead of
10764 get_recpe_type and gen_aarch64_frecps instead of get_recps_type.
10765 (aarch64_atomic_load_op_code): Delete.
10766 (aarch64_emit_atomic_load_op): Likewise.
10767 (aarch64_gen_atomic_ldop): Use UNSPECV_ATOMIC_* instead of
10768 aarch64_atomic_load_op_code. Use gen_aarch64_atomic_load
10769 instead of aarch64_emit_atomic_load_op.
10770 * config/aarch64/aarch64.md (aarch64_reload_movcp<GPF_TF:mode><P:mode>)
10771 (aarch64_reload_movcp<VALL:mode><P:mode>, aarch64_reload_mov<mode>)
10772 (aarch64_movdi_<mode>low, aarch64_movdi_<mode>high)
10773 (aarch64_mov<mode>high_di, aarch64_mov<mode>low_di): Add a '@'
10774 character before the pattern name.
10775 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
10776 (aarch64_rsqrte<mode>, aarch64_rsqrts<mode>)
10777 (aarch64_simd_combine<mode>, aarch64_frecpe<mode>)
10778 (aarch64_frecps<mode>): Likewise.
10779 * config/aarch64/atomics.md (atomic_compare_and_swap<mode>)
10780 (aarch64_compare_and_swap<mode>, aarch64_compare_and_swap<mode>_lse)
10781 (aarch64_load_exclusive<mode>, aarch64_store_exclusive<mode>)
10782 (aarch64_atomic_swp<mode>, aarch64_atomic_cas<mode>)
10783 (aarch64_atomic_load<atomic_ldop><mode>): Likewise.
10785 2018-08-02 Richard Sandiford <richard.sandiford@arm.com>
10787 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
10788 Allow HFmode constants if TARGET_FP_F16INST.
10790 2018-08-02 Jackson Woodruff <jackson.woodruff@arm.com>
10793 * config/aarch64/aarch64.c (aarch64_operands_adjust_ok_for_ldpstp):
10794 No longer check last store for clobber of address register.
10796 2018-08-02 Martin Liska <mliska@suse.cz>
10798 PR gcov-profile/86817
10799 * gcov.c (process_all_functions): New function.
10801 (process_file): Move functions processing to
10802 process_all_functions.
10804 2018-08-02 David Malcolm <dmalcolm@redhat.com>
10806 * dumpfile.c (dump_user_location_t::dump_user_location_t): Add
10807 "const" to the "gimple *" and "rtx_insn *" parameters.
10808 * dumpfile.h (dump_user_location_t::dump_user_location_t):
10810 (dump_location_t::dump_location_t): Likewise.
10812 2018-08-01 Martin Sebor <msebor@redhat.com>
10814 PR tree-optimization/86650
10815 * gimple-pretty-print.c (percent_G_format): Accept a "gimple *"
10816 rather than a "gcall *". Directly pass the data of interest
10817 to percent_K_format, rather than building a temporary CALL_EXPR
10819 * gimple-fold.c (gimple_fold_builtin_strncpy): Adjust.
10820 (gimple_fold_builtin_strncat): Adjust.
10821 * gimple-ssa-warn-restrict.h (check_bounds_or_overlap): Replace
10822 gcall* argument with gimple*.
10823 * gimple-ssa-warn-restrict.c (check_call): Same.
10824 (wrestrict_dom_walker::before_dom_children): Same.
10825 (builtin_access::builtin_access): Same.
10826 (check_bounds_or_overlap): Same
10827 (maybe_diag_overlap): Same.
10828 (maybe_diag_offset_bounds): Same.
10829 * tree-diagnostic.c (default_tree_printer): Move usage of
10830 EXPR_LOCATION (t) and TREE_BLOCK (t) from within percent_K_format
10832 * tree-pretty-print.c (percent_K_format): Add argument.
10833 * tree-pretty-print.h: Add argument.
10834 * tree-ssa-ccp.c (pass_post_ipa_warn::execute): Adjust.
10835 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Adjust.
10836 (maybe_diag_stxncpy_trunc): Same.
10837 (handle_builtin_stxncpy): Same.
10838 (handle_builtin_strcat): Same.
10840 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
10842 * match.pd: Optimise pointer range checks.
10844 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
10846 PR tree-optimization/86758
10847 * tree-vect-stmts.c (vectorizable_simd_clone_call): Don't try
10848 to remove pattern statements.
10850 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
10852 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Use the
10853 result of dfs_enumerate_from when constructing stmt_vec_infos,
10854 instead of additionally calling get_loop_body.
10856 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
10858 * tree-vectorizer.h (vect_create_data_ref_ptr): Remove inv_p
10860 * tree-vect-data-refs.c (vect_create_data_ref_ptr): Likewise.
10861 When creating an iv, assert that the step is not known to be zero.
10862 (vect_setup_realignment): Update call accordingly.
10863 * tree-vect-stmts.c (vectorizable_store): Likewise.
10864 (vectorizable_load): Likewise. Handle VMAT_INVARIANT separately.
10866 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
10868 * tree-vectorizer.h (vect_stmt_to_vectorize): New function.
10869 * tree-vect-loop.c (vect_update_vf_for_slp): Use it.
10870 (vectorizable_reduction): Likewise.
10871 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
10872 (vect_detect_hybrid_slp_stmts): Likewise.
10873 * tree-vect-stmts.c (vect_is_simple_use): Likewise.
10875 2018-08-01 Aldy Hernandez <aldyh@redhat.com>
10877 * tree-vrp (zero_nonzero_bits_from_bounds): Rename to...
10878 (wide_int_set_zero_nonzero_bits): ...this.
10879 (zero_nonzero_bits_from_vr): Rename to...
10880 (vrp_set_zero_nonzero_bits): ...this.
10881 (extract_range_from_multiplicative_op_1): Abstract wide int
10883 (wide_int_range_multiplicative_op): ...here.
10884 (extract_range_from_binary_expr_1): Extract wide int binary
10885 operations into their own functions.
10886 (wide_int_range_lshift): New.
10887 (wide_int_range_can_optimize_bit_op): New.
10888 (wide_int_range_shift_undefined_p): New.
10889 (wide_int_range_bit_xor): New.
10890 (wide_int_range_bit_ior): New.
10891 (wide_int_range_bit_and): New.
10892 (wide_int_range_trunc_mod): New.
10893 (extract_range_into_wide_ints): New.
10894 (vrp_shift_undefined_p): New.
10895 (extract_range_from_multiplicative_op): New.
10896 (vrp_can_optimize_bit_op): New.
10897 * tree-vrp.h (value_range::dump): New.
10898 (wide_int_range_multiplicative_op): New.
10899 (wide_int_range_lshift):New.
10900 (wide_int_range_shift_undefined_p): New.
10901 (wide_int_range_bit_xor): New.
10902 (wide_int_range_bit_ior): New.
10903 (wide_int_range_bit_and): New.
10904 (wide_int_range_trunc_mod): New.
10905 (zero_nonzero_bits_from_bounds): Rename to...
10906 (wide_int_set_zero_nonzero_bits): ...this.
10907 (zero_nonzero_bits_from_vr): Rename to...
10908 (vrp_set_zero_nonzero_bits): ...this.
10909 (range_easy_mask_min_max): Rename to...
10910 (wide_int_range_can_optimize_bit_op): this.
10911 * vr-values.c (simplify_bit_ops_using_ranges): Rename
10912 zero_nonzero_bits_from_vr into vrp_set_zero_nonzero_bits.
10914 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
10916 * tree-vectorizer.h (vect_orig_stmt): New function.
10917 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Use it.
10918 * tree-vect-loop.c (vect_model_reduction_cost): Likewise.
10919 (vect_create_epilog_for_reduction): Likewise.
10920 (vectorizable_live_operation): Likewise.
10921 * tree-vect-slp.c (vect_find_last_scalar_stmt_in_slp): Likewise.
10922 (vect_detect_hybrid_slp_stmts, vect_schedule_slp): Likewise.
10923 * tree-vect-stmts.c (vectorizable_call): Likewise.
10924 (vectorizable_simd_clone_call, vect_remove_stores): Likewise.
10926 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
10928 * tree-vectorizer.h (vect_transform_stmt): Remove grouped_store
10930 * tree-vect-stmts.c (vect_transform_stmt): Likewise.
10931 * tree-vect-loop.c (vect_transform_loop_stmt): Update call accordingly.
10932 (vect_transform_loop): Likewise.
10933 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
10935 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
10937 * tree-vectorizer.h (vect_schedule_slp): Return void.
10938 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
10939 (vect_schedule_slp): Likewise.
10941 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
10943 * tree-vect-loop.c (vect_transform_loop_stmt): Remove slp_scheduled
10945 (vect_transform_loop): Update calls accordingly. Schedule SLP
10946 instances before the main loop, if any exist.
10948 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
10950 PR tree-optimization/86749
10951 * tree-vect-patterns.c (vect_determine_min_output_precision_1):
10952 If the lhs is used in a COND_EXPR, check that it is being used
10953 as the "then" or "else" value.
10955 2018-08-01 Tom de Vries <tdevries@suse.de>
10958 * config/nvptx/nvptx.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Define to
10959 speculation_safe_value_not_needed.
10961 2018-08-01 Richard Biener <rguenther@suse.de>
10963 * tree-ssa-sccvn.c (visit_phi): Compare invariant addresses
10964 as base and offset.
10966 2018-08-01 Martin Liska <mliska@suse.cz>
10968 * value-prof.c (gimple_divmod_fixed_value_transform): Unify
10969 format how successful transformation is dumped.
10970 (gimple_mod_pow2_value_transform): Likewise.
10971 (gimple_mod_subtract_transform): Likewise.
10972 (gimple_stringops_transform): Likewise.
10974 2018-08-01 Martin Liska <mliska@suse.cz>
10976 PR value-prof/35543
10977 * value-prof.c (interesting_stringop_to_profile_p):
10978 Simplify the code and add BUILT_IN_MEMMOVE.
10979 (gimple_stringops_transform): Likewise.
10981 2018-08-01 Sam Tebbs <sam.tebbs@arm.com>
10983 * config/aarch64/aarch64-simd.md
10984 (*aarch64_get_lane_zero_extendsi<mode>): Rename to...
10985 (*aarch64_get_lane_zero_extend<GPI:mode><VDQQH:mode>): ... This and
10986 use GPI iterator instead of SI mode.
10988 2018-08-01 Richard Earnshaw <rearnsha@arm.com>
10990 * config/rs6000/rs6000.md (speculation_barrier): Renamed from
10991 rs6000_speculation_barrier.
10992 * config/rs6000/rs6000.c (rs6000_expand_builtin): Adjust for
10993 new barrier pattern name.
10995 2018-08-01 Richard Earnshaw <rearnsha@arm.com>
10997 * config/i386/i386.md (unspecv): Add UNSPECV_SPECULATION_BARRIER.
10998 (speculation_barrier): New insn.
11000 2018-08-01 Richard Biener <rguenther@suse.de>
11003 * graphite.h: Include isl/id.h and isl/space.h to allow build
11006 2018-08-01 Jan Willem Jagersma <jwjagersma@gmail.com>
11009 * dwarf2out.c (dwarf2out_early_finish): Do not generate assembly in LTO
11010 mode for COFF targets.
11011 * defaults.h (TARGET_COFF): Define.
11012 * config/i386/djgpp.h (TARGET_ASM_LTO_START, TARGET_ASM_LTO_END,
11013 TARGET_COFF): Define.
11014 (i386_djgpp_asm_lto_start, i386_djgpp_asm_lto_end): Declare.
11015 * config/i386/djgpp.c (saved_debug_info_level): New static variable.
11016 (i386_djgpp_asm_lto_start, i386_djgpp_asm_lto_end): New functions.
11018 2018-07-31 Alexandre Oliva <oliva@adacore.com>
11020 * gimple-streamer-in.c (input_bb): Restore BB discriminator.
11021 * gimple-streamer-out.c (output_bb): Save it.
11022 * lto-streamer-in.c (input_struct_function_base): Restore
11023 instance discriminator if available. Create map on demand.
11024 * lto-streamer-out.c (output_struct_function_base): Save it if
11026 * final.c (decl_to_instance_map): Document LTO strategy.
11028 2018-07-31 Alexandre Oliva <oliva@adacore.com>
11029 Olivier Hainque <hainque@adacore.com>
11031 * debug.h (decl_to_instance_map_t): New type.
11032 (decl_to_instance_map): Declare.
11033 (maybe_create_decl_to_instance_map): New inline function.
11034 * final.c (bb_discriminator, last_bb_discriminator): New statics,
11035 to track basic block discriminators.
11036 (final_start_function_1): Initialize them.
11037 (final_scan_insn_1): On NOTE_INSN_BASIC_BLOCK, track
11039 (decl_to_instance_map): New variable.
11040 (map_decl_to_instance, maybe_set_discriminator): New functions.
11041 (notice_source_line): Set discriminator.
11043 2018-07-31 Ian Lance Taylor <iant@golang.org>
11045 * targhooks.c (default_have_speculation_safe_value): Add
11048 2018-07-31 David Malcolm <dmalcolm@redhat.com>
11050 * dump-context.h: Include "pretty-print.h".
11051 (dump_context::refresh_dumps_are_enabled): New decl.
11052 (dump_context::emit_item): New decl.
11053 (class dump_context): Add fields "m_test_pp" and
11055 (temp_dump_context::temp_dump_context): Add param "test_pp_flags".
11056 (temp_dump_context::get_dumped_text): New decl.
11057 (class temp_dump_context): Add field "m_pp".
11058 * dumpfile.c (refresh_dumps_are_enabled): Convert to...
11059 (dump_context::refresh_dumps_are_enabled): ...and add a test for
11061 (set_dump_file): Update for above change.
11062 (set_alt_dump_file): Likewise.
11063 (dump_loc): New overload, taking a pretty_printer *.
11064 (dump_context::dump_loc): Call end_any_optinfo. Dump the location
11065 to any test pretty-printer.
11066 (make_item_for_dump_gimple_stmt): New function, adapted from
11067 optinfo::add_gimple_stmt.
11068 (dump_context::dump_gimple_stmt): Call it, and use the result,
11069 eliminating the direct usage of dump_file and alt_dump_file in
11070 favor of indirectly using them via emit_item.
11071 (make_item_for_dump_gimple_expr): New function, adapted from
11072 optinfo::add_gimple_expr.
11073 (dump_context::dump_gimple_expr): Call it, and use the result,
11074 eliminating the direct usage of dump_file and alt_dump_file in
11075 favor of indirectly using them via emit_item.
11076 (make_item_for_dump_generic_expr): New function, adapted from
11078 (dump_context::dump_generic_expr): Call it, and use the result,
11079 eliminating the direct usage of dump_file and alt_dump_file in
11080 favor of indirectly using them via emit_item.
11081 (make_item_for_dump_printf_va): New function, adapted from
11082 optinfo::add_printf_va.
11083 (make_item_for_dump_printf): New function.
11084 (dump_context::dump_printf_va): Call make_item_for_dump_printf_va,
11085 and use the result, eliminating the direct usage of dump_file and
11086 alt_dump_file in favor of indirectly using them via emit_item.
11087 (make_item_for_dump_dec): New function.
11088 (dump_context::dump_dec): Call it, and use the result,
11089 eliminating the direct usage of dump_file and alt_dump_file in
11090 favor of indirectly using them via emit_item.
11091 (make_item_for_dump_symtab_node): New function, adapted from
11092 optinfo::add_symtab_node.
11093 (dump_context::dump_symtab_node): Call it, and use the result,
11094 eliminating the direct usage of dump_file and alt_dump_file in
11095 favor of indirectly using them via emit_item.
11096 (dump_context::begin_scope): Reimplement, avoiding direct usage
11097 of dump_file and alt_dump_file in favor of indirectly using them
11099 (dump_context::emit_item): New member function.
11100 (temp_dump_context::temp_dump_context): Add param "test_pp_flags".
11101 Set up test pretty-printer on the underlying context. Call
11102 refresh_dumps_are_enabled.
11103 (temp_dump_context::~temp_dump_context): Call
11104 refresh_dumps_are_enabled.
11105 (temp_dump_context::get_dumped_text): New member function.
11106 (selftest::verify_dumped_text): New function.
11107 (ASSERT_DUMPED_TEXT_EQ): New macro.
11108 (selftest::test_capture_of_dump_calls): Run all tests twice, with
11109 and then without optinfo enabled. Add uses of
11110 ASSERT_DUMPED_TEXT_EQ to all tests. Add test of nested scopes.
11111 * dumpfile.h: Update comment for the dump_* API.
11112 * optinfo-emit-json.cc
11113 (selftest::test_building_json_from_dump_calls): Update for new
11114 param for temp_dump_context ctor.
11115 * optinfo.cc (optinfo_item::optinfo_item): Remove "owned" param
11116 and "m_owned" field.
11117 (optinfo_item::~optinfo_item): Likewise.
11118 (optinfo::add_item): New member function.
11119 (optinfo::emit): Update comment.
11120 (optinfo::add_string): Delete.
11121 (optinfo::add_printf): Delete.
11122 (optinfo::add_printf_va): Delete.
11123 (optinfo::add_gimple_stmt): Delete.
11124 (optinfo::add_gimple_expr): Delete.
11125 (optinfo::add_tree): Delete.
11126 (optinfo::add_symtab_node): Delete.
11127 (optinfo::add_dec): Delete.
11128 * optinfo.h (class dump_context): New forward decl.
11129 (optinfo::add_item): New decl.
11130 (optinfo::add_string): Delete.
11131 (optinfo::add_printf): Delete.
11132 (optinfo::add_printf_va): Delete.
11133 (optinfo::add_gimple_stmt): Delete.
11134 (optinfo::add_gimple_expr): Delete.
11135 (optinfo::add_tree): Delete.
11136 (optinfo::add_symtab_node): Delete.
11137 (optinfo::add_dec): Delete.
11138 (optinfo::add_poly_int): Delete.
11139 (optinfo_item::optinfo_item): Remove "owned" param.
11140 (class optinfo_item): Remove field "m_owned".
11142 2018-07-31 Jozef Lawrynowicz <jozef.l@mittosystems.com>
11144 PR middle-end/86705
11145 * gcc/cfgexpand.c (set_parm_rtl): Use the alignment of Pmode when
11146 MAX_SUPPORTED_STACK_ALIGNMENT would otherwise be exceeded by the
11147 requested variable alignment.
11148 (expand_one_ssa_partition): Likewise.
11149 (expand_one_var): Likewise.
11151 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
11153 * config/pdp11/pdp11.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Redefine
11154 to speculation_safe_value_not_needed.
11156 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
11158 * targhooks.h (speculation_safe_value_not_needed): New prototype.
11159 * targhooks.c (speculation_safe_value_not_needed): New function.
11160 * target.def (have_speculation_safe_value): Update documentation.
11161 * doc/tm.texi: Regenerated.
11163 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
11165 * config/aarch64/iterators.md (ALLI_TI): New iterator.
11166 * config/aarch64/aarch64.md (despeculate_copy<ALLI_TI:mode>): New
11168 (despeculate_copy<ALLI:mode>_insn): New insn.
11169 (despeculate_copyti_insn): New insn.
11170 (despeculate_simple<ALLI:mode>): New insn
11171 (despeculate_simpleti): New insn.
11172 * config/aarch64/aarch64.c (aarch64_speculation_safe_value): New
11174 (TARGET_SPECULATION_SAFE_VALUE): Redefine to
11175 aarch64_speculation_safe_value.
11176 (aarch64_print_operand): Handle const0_rtx in modifier 'H'.
11178 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
11180 * config/aarch64/aarch64-speculation.cc: New file.
11181 * config/aarch64/aarch64-passes.def (pass_track_speculation): Add
11182 before pass_reorder_blocks.
11183 * config/aarch64/aarch64-protos.h (make_pass_track_speculation): Add
11185 * config/aarch64/aarch64.c (aarch64_conditional_register_usage): Fix
11186 X14 and X15 when tracking speculation.
11187 * config/aarch64/aarch64.md (register name constants): Add
11188 SPECULATION_TRACKER_REGNUM and SPECULATION_SCRATCH_REGNUM.
11189 (unspec): Add UNSPEC_SPECULATION_TRACKER.
11190 (speculation_barrier): New insn attribute.
11191 (cmp<mode>): Allow SP in comparisons.
11192 (speculation_tracker): New insn.
11193 (speculation_barrier): Add speculation_barrier attribute.
11194 * config/aarch64/t-aarch64: Add make rule for aarch64-speculation.o.
11195 * config.gcc (aarch64*-*-*): Add aarch64-speculation.o to extra_objs.
11196 * doc/invoke.texi (AArch64 Options): Document -mtrack-speculation.
11198 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
11200 * config/aarch64/aarch64.md (cb<optab><mode>1): Disable when
11201 aarch64_track_speculation is true.
11202 (tb<optab><mode>1): Likewise.
11203 * config/aarch64/aarch64.c (aarch64_split_compare_regs): Do not
11204 generate CB[N]Z when tracking speculation.
11205 (aarch64_split_compare_and_swap): Likewise.
11206 (aarch64_split_atomic_op): Likewise.
11208 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
11210 * config/aarch64/aarch64.opt (mtrack-speculation): New target option.
11212 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
11214 * config/aarch64.md (unspecv): Add UNSPECV_SPECULAION_BARRIER.
11215 (speculation_barrier): New insn.
11217 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
11219 * config/arm/unspecs.md (unspecv): Add VUNSPEC_SPECULATION_BARRIER.
11220 * config/arm/arm.md (speculation_barrier): New expand.
11221 (speculation_barrier_insn): New pattern.
11223 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
11225 * builtin-types.def (BT_FN_PTR_PTR_VAR): New function type.
11226 (BT_FN_I1_I1_VAR, BT_FN_I2_I2_VAR, BT_FN_I4_I4_VAR): Likewise.
11227 (BT_FN_I8_I8_VAR, BT_FN_I16_I16_VAR): Likewise.
11228 * builtin-attrs.def (ATTR_NOVOPS_NOTHROW_LEAF_LIST): New attribute
11230 * builtins.def (BUILT_IN_SPECULATION_SAFE_VALUE_N): New builtin.
11231 (BUILT_IN_SPECULATION_SAFE_VALUE_PTR): New internal builtin.
11232 (BUILT_IN_SPECULATION_SAFE_VALUE_1): Likewise.
11233 (BUILT_IN_SPECULATION_SAFE_VALUE_2): Likewise.
11234 (BUILT_IN_SPECULATION_SAFE_VALUE_4): Likewise.
11235 (BUILT_IN_SPECULATION_SAFE_VALUE_8): Likewise.
11236 (BUILT_IN_SPECULATION_SAFE_VALUE_16): Likewise.
11237 * builtins.c (expand_speculation_safe_value): New function.
11238 (expand_builtin): Call it.
11239 * doc/cpp.texi: Document predefine __HAVE_SPECULATION_SAFE_VALUE.
11240 * doc/extend.texi: Document __builtin_speculation_safe_value.
11241 * doc/md.texi: Document "speculation_barrier" pattern.
11242 * doc/tm.texi.in: Pull in TARGET_SPECULATION_SAFE_VALUE and
11243 TARGET_HAVE_SPECULATION_SAFE_VALUE.
11244 * doc/tm.texi: Regenerated.
11245 * target.def (have_speculation_safe_value, speculation_safe_value): New
11247 * targhooks.c (default_have_speculation_safe_value): New function.
11248 (default_speculation_safe_value): New function.
11249 * targhooks.h (default_have_speculation_safe_value): Add prototype.
11250 (default_speculation_safe_value): Add prototype.
11252 2018-07-31 David Malcolm <dmalcolm@redhat.com>
11254 * dump-context.h (dump_context::dump_loc): New decl.
11255 * dumpfile.c (dump_context::dump_loc): New member function.
11256 (dump_context::dump_gimple_stmt_loc): Reimplement using dump_loc
11257 and dump_gimple_stmt.
11258 (dump_context::dump_gimple_expr_loc): Likewise, using
11260 (dump_context::dump_generic_expr_loc): Likewise, using
11262 (dump_context::dump_printf_loc_va): Likewise, using
11264 (dump_context::begin_scope): Explicitly using the global function
11265 "dump_loc", rather than the member function.
11267 2018-07-31 Martin Sebor <msebor@redhat.com>
11269 PR tree-optimization/86741
11270 * tree-vrp.c (vrp_prop::check_mem_ref): Avoid incomplete types.
11272 2018-07-31 Andreas Krebbel <krebbel@linux.ibm.com>
11274 * config/s390/s390.c (s390_expand_setmem): Make the unrolling to
11275 depend on whether prefetch instructions will be emitted or not.
11276 Use TARGET_SETMEM_PFD for checking whether prefetch instructions
11277 will be emitted or not.
11278 * config/s390/s390.h (TARGET_SETMEM_PREFETCH_DISTANCE)
11279 (TARGET_SETMEM_PFD): New macros.
11281 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
11283 * tree-vectorizer.h (stmt_vec_info): Turn back into a typedef.
11284 (NULL_STMT_VEC_INFO): Delete.
11285 (stmt_vec_info::operator*): Likewise.
11286 (stmt_vec_info::operator gimple *): Likewise.
11287 * tree-vect-loop.c (vectorizable_reduction): Use NULL instead
11288 of NULL_STMT_VEC_INFO.
11289 * tree-vect-patterns.c (vect_init_pattern_stmt): Likewise.
11290 (vect_reassociating_reduction_p): Likewise.
11291 * tree-vect-stmts.c (vect_build_gather_load_calls): Likewise.
11292 (vectorizable_store): Likewise.
11293 * tree-vectorizer.c (vec_info::set_vinfo_for_stmt): Likewise.
11294 (vec_info::free_stmt_vec_infos): Likewise.
11296 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
11298 * tree-vectorizer.h (vect_stmt_in_region_p): Delete.
11299 * tree-vectorizer.c (vect_stmt_in_region_p): Likewise.
11301 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
11303 * tree-vectorizer.h (vec_info::new_vinfo_for_stmt)
11304 (vec_info::set_vinfo_for_stmt, vec_info::free_stmt_vec_infos)
11305 (vec_info::free_stmt_vec_info): New private member functions.
11306 (set_stmt_vec_info_vec, free_stmt_vec_infos, vinfo_for_stmt)
11307 (set_vinfo_for_stmt, new_stmt_vec_info, free_stmt_vec_info): Delete.
11308 * tree-parloops.c (gather_scalar_reductions): Remove calls to
11309 set_stmt_vec_info_vec and free_stmt_vec_infos.
11310 * tree-vect-loop.c (_loop_vec_info): Remove call to
11311 set_stmt_vec_info_vec.
11312 * tree-vect-stmts.c (new_stmt_vec_info, set_stmt_vec_info_vec)
11313 (free_stmt_vec_infos, free_stmt_vec_info): Delete in favor of...
11314 * tree-vectorizer.c (vec_info::new_stmt_vec_info)
11315 (vec_info::set_vinfo_for_stmt, vec_info::free_stmt_vec_infos)
11316 (vec_info::free_stmt_vec_info): ...these new functions. Remove
11317 assignments in {vec_info::,}new_stmt_vec_info that are redundant
11318 with the clearing in the xcalloc.
11319 (stmt_vec_info_vec): Delete.
11320 (vec_info::vec_info): Don't call set_stmt_vec_info_vec.
11321 (vectorize_loops): Likewise.
11322 (vec_info::~vec_info): Remove argument from call to
11323 free_stmt_vec_infos.
11324 (vec_info::add_stmt): Remove vinfo argument from call to
11327 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
11329 * tree-vectorizer.h (free_stmt_vec_info): Take a stmt_vec_info
11330 rather than a gimple stmt.
11331 * tree-vect-stmts.c (free_stmt_vec_info): Likewise. Don't free
11332 information for pattern statements when passed the original
11333 statement; instead wait to be passed the pattern statement itself.
11334 Don't call set_vinfo_for_stmt here.
11335 (free_stmt_vec_infos): Update call to free_stmt_vec_info.
11336 * tree-vect-loop.c (_loop_vec_info::~loop_vec_info): Don't free
11337 stmt_vec_infos here.
11338 * tree-vect-slp.c (_bb_vec_info::~bb_vec_info): Likewise.
11339 * tree-vectorizer.c (vec_info::remove_stmt): Nullify the statement's
11340 stmt_vec_infos entry.
11342 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
11344 * tree-vectorizer.h (vec_info::replace_stmt): Declare.
11345 * tree-vectorizer.c (vec_info::replace_stmt): New function.
11346 * tree-vect-slp.c (vect_remove_slp_scalar_calls): Use it.
11347 * tree-vect-stmts.c (vectorizable_call): Likewise.
11348 (vectorizable_simd_clone_call): Likewise.
11350 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
11352 * tree-vectorizer.h (vec_info::remove_stmt): Declare.
11353 * tree-vectorizer.c (vec_info::remove_stmt): New function.
11354 * tree-vect-loop-manip.c (vect_set_loop_condition): Use it.
11355 * tree-vect-loop.c (vect_transform_loop): Likewise.
11356 * tree-vect-slp.c (vect_schedule_slp): Likewise.
11357 * tree-vect-stmts.c (vect_remove_stores): Likewise.
11359 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
11361 * tree-vectorizer.h (vec_info::lookup_dr): New member function.
11362 (vect_dr_stmt): Delete.
11363 * tree-vectorizer.c (vec_info::lookup_dr): New function.
11364 * tree-vect-loop-manip.c (vect_update_inits_of_drs): Use it instead
11366 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr)
11367 (vect_analyze_data_ref_dependence, vect_record_base_alignments)
11368 (vect_verify_datarefs_alignment, vect_peeling_supportable)
11369 (vect_analyze_data_ref_accesses, vect_prune_runtime_alias_test_list)
11370 (vect_analyze_data_refs): Likewise.
11371 (vect_slp_analyze_data_ref_dependence): Likewise. Take a vec_info
11373 (vect_find_same_alignment_drs): Likewise.
11374 (vect_slp_analyze_node_dependences): Update calls accordingly.
11375 (vect_analyze_data_refs_alignment): Likewise. Use vec_info::lookup_dr
11376 instead of DR_VECT_AUX.
11377 (vect_get_peeling_costs_all_drs): Take a loop_vec_info instead
11378 of a vector data references. Use vec_info::lookup_dr instead of
11380 (vect_peeling_hash_get_lowest_cost): Update calls accordingly.
11381 (vect_enhance_data_refs_alignment): Likewise. Use vec_info::lookup_dr
11382 instead of DR_VECT_AUX.
11384 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
11386 * tree-vectorizer.h (_loop_vec_info::unaligned_dr): Change to
11388 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Update
11390 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
11391 * tree-vect-loop-manip.c (get_misalign_in_elems): Likewise.
11392 (vect_gen_prolog_loop_niters): Likewise.
11394 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
11396 * tree-vectorizer.h (set_dr_misalignment, dr_misalignment)
11397 (DR_TARGET_ALIGNMENT, aligned_access_p, known_alignment_for_access_p)
11398 (vect_known_alignment_in_bytes, vect_dr_behavior)
11399 (vect_get_scalar_dr_size): Take references as dr_vec_infos
11400 instead of data_references. Update calls to other routines for
11401 which the same change has been made.
11402 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Take
11403 dr_vec_infos instead of stmt_vec_infos.
11404 (vect_analyze_data_ref_dependence): Update call accordingly.
11405 (vect_slp_analyze_data_ref_dependence)
11406 (vect_record_base_alignments): Use DR_VECT_AUX.
11407 (vect_calculate_target_alignment, vect_compute_data_ref_alignment)
11408 (vect_update_misalignment_for_peel, verify_data_ref_alignment)
11409 (vector_alignment_reachable_p, vect_get_data_access_cost)
11410 (vect_peeling_supportable, vect_analyze_group_access_1)
11411 (vect_analyze_group_access, vect_analyze_data_ref_access)
11412 (vect_vfa_segment_size, vect_vfa_access_size, vect_vfa_align)
11413 (vect_compile_time_alias, vect_small_gap_p)
11414 (vectorizable_with_step_bound_p, vect_duplicate_ssa_name_ptr_info):
11415 (vect_supportable_dr_alignment): Take references as dr_vec_infos
11416 instead of data_references. Update calls to other routines for
11417 which the same change has been made.
11418 (vect_verify_datarefs_alignment, vect_get_peeling_costs_all_drs)
11419 (vect_find_same_alignment_drs, vect_analyze_data_refs_alignment)
11420 (vect_slp_analyze_and_verify_node_alignment)
11421 (vect_analyze_data_ref_accesses, vect_prune_runtime_alias_test_list)
11422 (vect_create_addr_base_for_vector_ref, vect_create_data_ref_ptr)
11423 (vect_setup_realignment): Use dr_vec_infos. Update calls after
11425 (_vect_peel_info::dr): Replace with...
11426 (_vect_peel_info::dr_info): ...this new field.
11427 (vect_peeling_hash_get_most_frequent)
11428 (vect_peeling_hash_choose_best_peeling): Update accordingly.
11429 (vect_peeling_hash_get_lowest_cost):
11430 (vect_enhance_data_refs_alignment): Likewise. Update calls to other
11431 routines for which the same change has been made.
11432 (vect_peeling_hash_insert): Likewise. Take a dr_vec_info instead of a
11434 * tree-vect-loop-manip.c (get_misalign_in_elems)
11435 (vect_gen_prolog_loop_niters): Use dr_vec_infos. Update calls after
11437 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
11438 * tree-vect-stmts.c (vect_get_store_cost, vect_get_load_cost)
11439 (vect_truncate_gather_scatter_offset, compare_step_with_zero)
11440 (get_group_load_store_type, get_negative_load_store_type)
11441 (vect_get_data_ptr_increment, vectorizable_store)
11442 (vectorizable_load): Likewise.
11443 (ensure_base_align): Take a dr_vec_info instead of a data_reference.
11444 Update calls to other routines for which the same change has been made.
11446 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
11448 * tree-vectorizer.h (vec_info::move_dr): New member function.
11449 (dataref_aux): Rename to...
11450 (dr_vec_info): ...this and add "dr" and "stmt" fields.
11451 (_stmt_vec_info::dr_aux): Update accordingly.
11452 (_stmt_vec_info::data_ref_info): Delete.
11453 (STMT_VINFO_GROUPED_ACCESS, DR_GROUP_FIRST_ELEMENT)
11454 (DR_GROUP_NEXT_ELEMENT, DR_GROUP_SIZE, DR_GROUP_STORE_COUNT)
11455 (DR_GROUP_GAP, DR_GROUP_SAME_DR_STMT, REDUC_GROUP_FIRST_ELEMENT):
11456 (REDUC_GROUP_NEXT_ELEMENT, REDUC_GROUP_SIZE): Use dr_aux.dr instead
11458 (STMT_VINFO_DATA_REF): Likewise. Turn into an lvalue.
11459 (STMT_VINFO_DR_INFO): New macro.
11460 (DR_VECT_AUX): Use STMT_VINFO_DR_INKFO and vect_dr_stmt.
11461 (set_dr_misalignment): Update after rename of dataref_aux.
11462 (vect_dr_stmt): Move earlier in file. Return dr_aux.stmt.
11463 * tree-vect-stmts.c (new_stmt_vec_info): Remove redundant
11464 initialization of STMT_VINFO_DATA_REF.
11465 * tree-vectorizer.c (vec_info::move_dr): New function.
11466 * tree-vect-patterns.c (vect_recog_bool_pattern)
11467 (vect_recog_mask_conversion_pattern)
11468 (vect_recog_gather_scatter_pattern): Use it.
11469 * tree-vect-data-refs.c (vect_analyze_data_refs): Initialize
11470 the "dr" and "stmt" fields of dr_vec_info instead of
11471 STMT_VINFO_DATA_REF.
11473 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
11475 * tree-vectorizer.h (_stmt_vec_info::pattern_stmt_p): New field.
11476 (is_pattern_stmt_p): Use it.
11477 * tree-vect-patterns.c (vect_init_pattern_stmt): Set pattern_stmt_p
11478 on pattern statements.
11480 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
11482 * tree-vect-patterns.c (vect_mark_pattern_stmts): Take the
11483 original stmt as a stmt_vec_info rather than a gimple stmt.
11484 (vect_pattern_recog_1): Take the statement directly as a
11485 stmt_vec_info, rather than via a gimple_stmt_iterator.
11486 Update call to vect_mark_pattern_stmts.
11487 (vect_pattern_recog): Update calls accordingly.
11489 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
11491 * tree-vectorizer.h (vect_get_vec_defs_for_stmt_copy)
11492 (vect_get_vec_def_for_stmt_copy): Take a vec_info rather than
11493 a vect_def_type for the first argument.
11494 * tree-vect-stmts.c (vect_get_vec_defs_for_stmt_copy): Likewise.
11495 (vect_get_vec_def_for_stmt_copy): Likewise. Return the original
11496 operand if it isn't defined by a vectorized statement.
11497 (vect_build_gather_load_calls): Remove the mask_dt argument and
11498 update calls to vect_get_vec_def_for_stmt_copy.
11499 (vectorizable_bswap): Likewise the dt argument.
11500 (vectorizable_call): Update calls to vectorizable_bswap and
11501 vect_get_vec_def_for_stmt_copy.
11502 (vectorizable_simd_clone_call, vectorizable_assignment)
11503 (vectorizable_shift, vectorizable_operation, vectorizable_condition)
11504 (vectorizable_comparison): Update calls to
11505 vect_get_vec_def_for_stmt_copy.
11506 (vectorizable_store): Likewise. Remove now-unnecessary calls to
11507 vect_is_simple_use.
11508 (vect_get_loop_based_defs): Remove dt argument and update call
11509 to vect_get_vec_def_for_stmt_copy.
11510 (vectorizable_conversion): Update calls to vect_get_loop_based_defs
11511 and vect_get_vec_def_for_stmt_copy.
11512 (vectorizable_load): Update calls to vect_build_gather_load_calls
11513 and vect_get_vec_def_for_stmt_copy.
11514 * tree-vect-loop.c (vect_create_epilog_for_reduction)
11515 (vectorizable_reduction, vectorizable_live_operation): Update calls
11516 to vect_get_vec_def_for_stmt_copy.
11518 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
11520 * tree-vect-data-refs.c (vect_record_base_alignment): Replace vec_info
11521 and gimple stmt arguments with a stmt_vec_info.
11522 (vect_record_base_alignments): Update calls accordingly.
11523 * tree-vect-slp.c (vect_record_max_nunits): Replace vec_info
11524 and gimple stmt arguments with a stmt_vec_info.
11525 (vect_build_slp_tree_1): Remove vinfo argument and update call
11526 to vect_record_max_nunits.
11527 (vect_build_slp_tree_2): Update calls to vect_build_slp_tree_1
11528 and vect_record_max_nunits.
11530 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
11532 * tree-vectorizer.h (nested_in_vect_loop_p): Move further down
11533 file and take a stmt_vec_info instead of a gimple stmt.
11534 (supportable_widening_operation, vect_finish_replace_stmt)
11535 (vect_finish_stmt_generation, vect_get_store_rhs)
11536 (vect_get_vec_def_for_operand_1, vect_get_vec_def_for_operand)
11537 (vect_get_vec_defs, vect_init_vector, vect_transform_stmt)
11538 (vect_remove_stores, vect_analyze_stmt, vectorizable_condition)
11539 (vect_get_smallest_scalar_type, vect_check_gather_scatter)
11540 (vect_create_data_ref_ptr, bump_vector_ptr)
11541 (vect_permute_store_chain, vect_setup_realignment)
11542 (vect_transform_grouped_load, vect_record_grouped_load_vectors)
11543 (vect_create_addr_base_for_vector_ref, vectorizable_live_operation)
11544 (vectorizable_reduction, vectorizable_induction)
11545 (get_initial_def_for_reduction, is_simple_and_all_uses_invariant)
11546 (vect_get_place_in_interleaving_chain): Take stmt_vec_infos rather
11547 than gimple stmts as arguments.
11548 * tree-vect-data-refs.c (vect_get_smallest_scalar_type)
11549 (vect_preserves_scalar_order_p, vect_slp_analyze_node_dependences)
11550 (can_group_stmts_p, vect_check_gather_scatter)
11551 (vect_create_addr_base_for_vector_ref, vect_create_data_ref_ptr)
11552 (bump_vector_ptr, vect_permute_store_chain, vect_setup_realignment)
11553 (vect_permute_load_chain, vect_shift_permute_load_chain)
11554 (vect_transform_grouped_load)
11555 (vect_record_grouped_load_vectors): Likewise.
11556 * tree-vect-loop.c (vect_fixup_reduc_chain)
11557 (get_initial_def_for_reduction, vect_create_epilog_for_reduction)
11558 (vectorize_fold_left_reduction, is_nonwrapping_integer_induction)
11559 (vectorizable_reduction, vectorizable_induction)
11560 (vectorizable_live_operation, vect_loop_kill_debug_uses): Likewise.
11561 * tree-vect-patterns.c (type_conversion_p, adjust_bool_stmts)
11562 (vect_get_load_store_mask): Likewise.
11563 * tree-vect-slp.c (vect_get_place_in_interleaving_chain)
11564 (vect_analyze_slp_instance, vect_mask_constant_operand_p): Likewise.
11565 * tree-vect-stmts.c (vect_mark_relevant)
11566 (is_simple_and_all_uses_invariant)
11567 (exist_non_indexing_operands_for_use_p, process_use)
11568 (vect_init_vector_1, vect_init_vector, vect_get_vec_def_for_operand_1)
11569 (vect_get_vec_def_for_operand, vect_get_vec_defs)
11570 (vect_finish_stmt_generation_1, vect_finish_replace_stmt)
11571 (vect_finish_stmt_generation, vect_truncate_gather_scatter_offset)
11572 (compare_step_with_zero, vect_get_store_rhs, get_group_load_store_type)
11573 (get_negative_load_store_type, get_load_store_type)
11574 (vect_check_load_store_mask, vect_check_store_rhs)
11575 (vect_build_gather_load_calls, vect_get_strided_load_store_ops)
11576 (vectorizable_bswap, vectorizable_call, vectorizable_simd_clone_call)
11577 (vect_create_vectorized_demotion_stmts, vectorizable_conversion)
11578 (vectorizable_assignment, vectorizable_shift, vectorizable_operation)
11579 (get_group_alias_ptr_type, vectorizable_store, hoist_defs_of_uses)
11580 (vectorizable_load, vectorizable_condition, vectorizable_comparison)
11581 (vect_analyze_stmt, vect_transform_stmt, vect_remove_stores)
11582 (supportable_widening_operation): Likewise.
11584 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
11586 * tree-vect-data-refs.c (vect_describe_gather_scatter_call): Take
11587 a stmt_vec_info instead of a gcall.
11588 (vect_check_gather_scatter): Update call accordingly.
11589 * tree-vect-loop-manip.c (iv_phi_p): Take a stmt_vec_info instead
11591 (vect_can_advance_ivs_p, vect_update_ivs_after_vectorizer)
11592 (slpeel_update_phi_nodes_for_loops):): Update calls accordingly.
11593 * tree-vect-loop.c (vect_transform_loop_stmt): Take a stmt_vec_info
11594 instead of a gimple stmt.
11595 (vect_transform_loop): Update calls accordingly.
11596 * tree-vect-slp.c (vect_split_slp_store_group): Take and return
11597 stmt_vec_infos instead of gimple stmts.
11598 (vect_analyze_slp_instance): Update use accordingly.
11599 * tree-vect-stmts.c (read_vector_array, write_vector_array)
11600 (vect_clobber_variable, vect_stmt_relevant_p, permute_vec_elements)
11601 (vect_use_strided_gather_scatters_p, vect_build_all_ones_mask)
11602 (vect_build_zero_merge_argument, vect_get_gather_scatter_ops)
11603 (vect_gen_widened_results_half, vect_get_loop_based_defs)
11604 (vect_create_vectorized_promotion_stmts, can_vectorize_live_stmts):
11605 Take a stmt_vec_info instead of a gimple stmt and pass stmt_vec_infos
11606 down to subroutines.
11608 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
11610 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Change the type
11611 of the worklist from a vector of gimple stmts to a vector of
11613 * tree-vect-stmts.c (vect_mark_relevant, process_use)
11614 (vect_mark_stmts_to_be_vectorized): Likewise
11616 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
11618 * tree-vect-loop.c (vect_analyze_loop_operations): Look up the
11619 statement before passing it to vect_analyze_stmt.
11620 (vect_create_epilog_for_reduction): Use a stmt_vec_info to walk
11621 the chain of phi vector definitions. Track the exit phi via its
11623 (vectorizable_reduction): Set cond_stmt_vinfo directly from the
11624 STMT_VINFO_REDUC_DEF.
11625 * tree-vect-slp.c (vect_get_place_in_interleaving_chain): Use
11626 stmt_vec_infos to handle the statement chains.
11627 (vect_get_slp_defs): Record the first statement in the node
11628 using a stmt_vec_info.
11629 * tree-vect-stmts.c (vect_mark_stmts_to_be_vectorized): Look up
11630 statements here and pass their stmt_vec_info down to subroutines.
11631 (vect_init_vector_1): Hoist call to vinfo_for_stmt and pass it
11632 down to vect_finish_stmt_generation.
11633 (vect_init_vector, vect_get_vec_defs, vect_finish_replace_stmt)
11634 (vect_finish_stmt_generation): Call vinfo_for_stmt and pass
11635 stmt_vec_infos to subroutines.
11636 (vect_remove_stores): Use stmt_vec_infos to handle the statement
11639 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
11641 * tree-vect-data-refs.c (vect_slp_analyze_node_dependences):
11642 (vect_check_gather_scatter, vect_create_data_ref_ptr, bump_vector_ptr)
11643 (vect_permute_store_chain, vect_setup_realignment)
11644 (vect_permute_load_chain, vect_shift_permute_load_chain)
11645 (vect_transform_grouped_load): Use stmt_vec_info rather than gimple
11646 stmts internally, and when passing values to other vectorizer routines.
11647 * tree-vect-loop-manip.c (vect_can_advance_ivs_p): Likewise.
11648 * tree-vect-loop.c (vect_analyze_scalar_cycles_1)
11649 (vect_analyze_loop_operations, get_initial_def_for_reduction)
11650 (vect_create_epilog_for_reduction, vectorize_fold_left_reduction)
11651 (vectorizable_reduction, vectorizable_induction)
11652 (vectorizable_live_operation, vect_transform_loop_stmt)
11653 (vect_transform_loop): Likewise.
11654 * tree-vect-patterns.c (vect_reassociating_reduction_p)
11655 (vect_recog_widen_op_pattern, vect_recog_mixed_size_cond_pattern)
11656 (vect_recog_bool_pattern, vect_recog_gather_scatter_pattern): Likewise.
11657 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
11658 (vect_slp_analyze_node_operations_1): Likewise.
11659 * tree-vect-stmts.c (vect_mark_relevant, process_use)
11660 (exist_non_indexing_operands_for_use_p, vect_init_vector_1)
11661 (vect_mark_stmts_to_be_vectorized, vect_get_vec_def_for_operand)
11662 (vect_finish_stmt_generation_1, get_group_load_store_type)
11663 (get_load_store_type, vect_build_gather_load_calls)
11664 (vectorizable_bswap, vectorizable_call, vectorizable_simd_clone_call)
11665 (vect_create_vectorized_demotion_stmts, vectorizable_conversion)
11666 (vectorizable_assignment, vectorizable_shift, vectorizable_operation)
11667 (vectorizable_store, vectorizable_load, vectorizable_condition)
11668 (vectorizable_comparison, vect_analyze_stmt, vect_transform_stmt)
11669 (supportable_widening_operation): Likewise.
11670 (vect_get_vector_types_for_stmt): Likewise.
11671 * tree-vectorizer.h (vect_dr_behavior): Likewise.
11673 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
11675 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence)
11676 (vect_slp_analyze_node_dependences, vect_analyze_data_ref_accesses)
11677 (vect_permute_store_chain, vect_permute_load_chain)
11678 (vect_shift_permute_load_chain, vect_transform_grouped_load): Avoid
11679 repeated stmt_vec_info lookups.
11680 * tree-vect-loop-manip.c (vect_can_advance_ivs_p): Likewise.
11681 (vect_update_ivs_after_vectorizer): Likewise.
11682 * tree-vect-loop.c (vect_is_simple_reduction): Likewise.
11683 (vect_create_epilog_for_reduction, vectorizable_reduction): Likewise.
11684 * tree-vect-patterns.c (adjust_bool_stmts): Likewise.
11685 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
11686 (vect_bb_slp_scalar_cost): Likewise.
11687 * tree-vect-stmts.c (get_group_alias_ptr_type): Likewise.
11689 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
11691 * tree-vect-data-refs.c (vect_check_gather_scatter): Pass the
11692 gcall rather than the generic gimple stmt to gimple_call_internal_fn.
11693 (vect_get_smallest_scalar_type, can_group_stmts_p): Use dyn_cast
11694 to get gassigns and gcalls, rather than operating on generc gimple
11696 * tree-vect-stmts.c (exist_non_indexing_operands_for_use_p)
11697 (vect_mark_stmts_to_be_vectorized, vectorizable_store)
11698 (vectorizable_load, vect_analyze_stmt): Likewise.
11699 * tree-vect-loop.c (vectorizable_reduction): Likewise gphi.
11701 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
11703 * tree-vectorizer.h (get_earlier_stmt, get_later_stmt): Take and
11704 return stmt_vec_infos rather than gimple stmts. Do not accept
11706 (vect_find_last_scalar_stmt_in_slp): Return a stmt_vec_info instead
11708 * tree-vect-slp.c (vect_find_last_scalar_stmt_in_slp): Likewise.
11709 Update use of get_later_stmt.
11710 (vect_get_constant_vectors): Update call accordingly.
11711 (vect_schedule_slp_instance): Likewise
11712 * tree-vect-data-refs.c (vect_slp_analyze_node_dependences): Likewise.
11713 (vect_slp_analyze_instance_dependence): Likewise.
11714 (vect_preserves_scalar_order_p): Update use of get_earlier_stmt.
11716 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
11718 * tree-vectorizer.h (stmt_info_for_cost::stmt): Replace with...
11719 (stmt_info_for_cost::stmt_info): ...this new field.
11720 (add_stmt_costs): Update accordingly.
11721 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost)
11722 (vect_get_known_peeling_cost): Likewise.
11723 (vect_estimate_min_profitable_iters): Likewise.
11724 * tree-vect-stmts.c (record_stmt_cost): Likewise.
11726 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
11728 * tree-vectorizer.h (_loop_vec_info::may_misalign_stmts): Change
11729 from an auto_vec<gimple *> to an auto_vec<stmt_vec_info>.
11730 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Update
11732 * tree-vect-loop-manip.c (vect_create_cond_for_align_checks): Likewise.
11734 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
11736 * tree-vectorizer.h (_stmt_vec_info::same_dr_stmt): Change from
11737 a gimple stmt to a stmt_vec_info.
11738 * tree-vect-stmts.c (vectorizable_load): Update accordingly.
11740 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
11742 * tree-vectorizer.h (vec_info::grouped_stores): Change from
11743 an auto_vec<gimple *> to an auto_vec<stmt_vec_info>.
11744 (_loop_vec_info::reduction_chains): Likewise.
11745 * tree-vect-loop.c (vect_fixup_scalar_cycles_with_patterns): Update
11747 * tree-vect-slp.c (vect_analyze_slp): Likewise.
11749 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
11751 * tree-vectorizer.h (_stmt_vec_info::first_element): Change from
11752 a gimple stmt to a stmt_vec_info.
11753 (_stmt_vec_info::next_element): Likewise.
11754 * tree-vect-data-refs.c (vect_update_misalignment_for_peel)
11755 (vect_slp_analyze_and_verify_node_alignment)
11756 (vect_analyze_group_access_1, vect_analyze_group_access)
11757 (vect_small_gap_p, vect_prune_runtime_alias_test_list)
11758 (vect_create_data_ref_ptr, vect_record_grouped_load_vectors)
11759 (vect_supportable_dr_alignment): Update accordingly.
11760 * tree-vect-loop.c (vect_fixup_reduc_chain): Likewise.
11761 (vect_fixup_scalar_cycles_with_patterns, vect_is_slp_reduction)
11762 (vect_is_simple_reduction, vectorizable_reduction): Likewise.
11763 * tree-vect-patterns.c (vect_reassociating_reduction_p): Likewise.
11764 * tree-vect-slp.c (vect_build_slp_tree_1)
11765 (vect_attempt_slp_rearrange_stmts, vect_supported_load_permutation_p)
11766 (vect_split_slp_store_group, vect_analyze_slp_instance)
11767 (vect_analyze_slp, vect_transform_slp_perm_load): Likewise.
11768 * tree-vect-stmts.c (vect_model_store_cost, vect_model_load_cost)
11769 (get_group_load_store_type, get_load_store_type)
11770 (get_group_alias_ptr_type, vectorizable_store, vectorizable_load)
11771 (vect_transform_stmt, vect_remove_stores): Likewise.
11773 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
11775 * tree-vectorizer.h (vect_dr_stmt): Return a stmt_vec_info rather
11776 than a gimple stmt.
11777 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence)
11778 (vect_slp_analyze_data_ref_dependence, vect_record_base_alignments)
11779 (vect_calculate_target_alignmentm, vect_compute_data_ref_alignment)
11780 (vect_update_misalignment_for_peel, vect_verify_datarefs_alignment)
11781 (vector_alignment_reachable_p, vect_get_data_access_cost)
11782 (vect_get_peeling_costs_all_drs, vect_peeling_hash_get_lowest_cost)
11783 (vect_peeling_supportable, vect_enhance_data_refs_alignment)
11784 (vect_find_same_alignment_drs, vect_analyze_data_refs_alignment)
11785 (vect_analyze_group_access_1, vect_analyze_group_access)
11786 (vect_analyze_data_ref_access, vect_analyze_data_ref_accesses)
11787 (vect_vfa_access_size, vect_small_gap_p, vect_analyze_data_refs)
11788 (vect_supportable_dr_alignment): Remove vinfo_for_stmt from the
11789 result of vect_dr_stmt and use the stmt_vec_info instead of
11790 the associated gimple stmt.
11791 * tree-vect-loop-manip.c (get_misalign_in_elems): Likewise.
11792 (vect_gen_prolog_loop_niters): Likewise.
11793 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
11795 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
11797 * tree-vectorizer.h (_slp_tree::stmts): Change from a vec<gimple *>
11798 to a vec<stmt_vec_info>.
11799 * tree-vect-slp.c (vect_free_slp_tree): Update accordingly.
11800 (vect_create_new_slp_node): Take a vec<gimple *> instead of a
11801 vec<stmt_vec_info>.
11802 (_slp_oprnd_info::def_stmts): Change from a vec<gimple *>
11803 to a vec<stmt_vec_info>.
11804 (bst_traits::value_type, bst_traits::value_type): Likewise.
11805 (bst_traits::hash): Update accordingly.
11806 (vect_get_and_check_slp_defs): Change the stmts parameter from
11807 a vec<gimple *> to a vec<stmt_vec_info>.
11808 (vect_two_operations_perm_ok_p, vect_build_slp_tree_1): Likewise.
11809 (vect_build_slp_tree): Likewise.
11810 (vect_build_slp_tree_2): Likewise. Update uses of
11811 SLP_TREE_SCALAR_STMTS.
11812 (vect_print_slp_tree): Update uses of SLP_TREE_SCALAR_STMTS.
11813 (vect_mark_slp_stmts, vect_mark_slp_stmts_relevant)
11814 (vect_slp_rearrange_stmts, vect_attempt_slp_rearrange_stmts)
11815 (vect_supported_load_permutation_p, vect_find_last_scalar_stmt_in_slp)
11816 (vect_detect_hybrid_slp_stmts, vect_slp_analyze_node_operations_1)
11817 (vect_slp_analyze_node_operations, vect_slp_analyze_operations)
11818 (vect_bb_slp_scalar_cost, vect_slp_analyze_bb_1)
11819 (vect_get_constant_vectors, vect_get_slp_defs)
11820 (vect_transform_slp_perm_load, vect_schedule_slp_instance)
11821 (vect_remove_slp_scalar_calls, vect_schedule_slp): Likewise.
11822 (vect_analyze_slp_instance): Build up a vec of stmt_vec_infos
11823 instead of gimple stmts.
11824 * tree-vect-data-refs.c (vect_slp_analyze_node_dependences): Change
11825 the stores parameter for a vec<gimple *> to a vec<stmt_vec_info>.
11826 (vect_slp_analyze_instance_dependence): Update uses of
11827 SLP_TREE_SCALAR_STMTS.
11828 (vect_slp_analyze_and_verify_node_alignment): Likewise.
11829 (vect_slp_analyze_and_verify_instance_alignment): Likewise.
11830 * tree-vect-loop.c (neutral_op_for_slp_reduction): Likewise.
11831 (get_initial_defs_for_reduction): Likewise.
11832 (vect_create_epilog_for_reduction): Likewise.
11833 (vectorize_fold_left_reduction): Likewise.
11834 * tree-vect-stmts.c (vect_prologue_cost_for_slp_op): Likewise.
11835 (vect_model_simple_cost, vectorizable_shift, vectorizable_load)
11836 (can_vectorize_live_stmts): Likewise.
11838 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
11840 * tree-vectorizer.h (_loop_vec_info::reductions): Change from an
11841 auto_vec<gimple *> to an auto_vec<stmt_vec_info>.
11842 (vect_force_simple_reduction): Take and return stmt_vec_infos rather
11844 * tree-parloops.c (valid_reduction_p): Take a stmt_vec_info instead
11846 (gather_scalar_reductions): Update after above interface changes.
11847 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Likewise.
11848 (vect_is_simple_reduction): Take and return stmt_vec_infos rather
11850 (vect_force_simple_reduction): Likewise.
11851 * tree-vect-patterns.c (vect_pattern_recog_1): Update use of
11852 LOOP_VINFO_REDUCTIONS.
11853 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
11855 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
11857 * tree-vectorizer.h (_stmt_vec_info::reduc_def): Change from
11858 a gimple stmt to a stmt_vec_info.
11859 * tree-vect-loop.c (vect_active_double_reduction_p)
11860 (vect_force_simple_reduction, vectorizable_reduction): Update
11863 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
11865 * tree-vectorizer.h (_slp_tree::vec_stmts): Change from a
11866 vec<gimple *> to a vec<stmt_vec_info>.
11867 * tree-vect-loop.c (vect_create_epilog_for_reduction): Change
11868 the reduction_phis argument from a vec<gimple *> to a
11869 vec<stmt_vec_info>.
11870 (vectorizable_reduction): Likewise the phis local variable that
11871 is passed to vect_create_epilog_for_reduction. Update for new type
11872 of SLP_TREE_VEC_STMTS.
11873 (vectorizable_induction): Update for new type of SLP_TREE_VEC_STMTS.
11874 (vectorizable_live_operation): Likewise.
11875 * tree-vect-slp.c (vect_get_slp_vect_defs): Likewise.
11876 (vect_transform_slp_perm_load, vect_schedule_slp_instance): Likewise.
11878 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
11880 * tree-vectorizer.h (_stmt_vec_info::vectorized_stmt): Change from
11881 a gimple stmt to a stmt_vec_info.
11882 (vectorizable_condition, vectorizable_live_operation)
11883 (vectorizable_reduction, vectorizable_induction): Pass back the
11884 vectorized statement as a stmt_vec_info.
11885 * tree-vect-data-refs.c (vect_record_grouped_load_vectors): Update
11886 use of STMT_VINFO_VEC_STMT.
11887 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise,
11888 accumulating the inner phis that feed the STMT_VINFO_VEC_STMT
11889 as stmt_vec_infos rather than gimple stmts.
11890 (vectorize_fold_left_reduction): Change vec_stmt from a gimple stmt
11891 to a stmt_vec_info.
11892 (vectorizable_live_operation): Likewise.
11893 (vectorizable_reduction, vectorizable_induction): Likewise,
11894 updating use of STMT_VINFO_VEC_STMT.
11895 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Update use
11896 of STMT_VINFO_VEC_STMT.
11897 (vect_build_gather_load_calls, vectorizable_bswap, vectorizable_call)
11898 (vectorizable_simd_clone_call, vectorizable_conversion)
11899 (vectorizable_assignment, vectorizable_shift, vectorizable_operation)
11900 (vectorizable_store, vectorizable_load, vectorizable_condition)
11901 (vectorizable_comparison, can_vectorize_live_stmts): Change vec_stmt
11902 from a gimple stmt to a stmt_vec_info.
11903 (vect_transform_stmt): Update use of STMT_VINFO_VEC_STMT. Pass a
11904 pointer to a stmt_vec_info to the vectorizable_* routines.
11906 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
11908 * tree-vectorizer.h (_stmt_vec_info::related_stmt): Change from
11909 a gimple stmt to a stmt_vec_info.
11910 (is_pattern_stmt_p): Update accordingly.
11911 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Likewise.
11912 (vect_record_grouped_load_vectors): Likewise.
11913 * tree-vect-loop.c (vect_determine_vf_for_stmt): Likewise.
11914 (vect_fixup_reduc_chain, vect_update_vf_for_slp): Likewise.
11915 (vect_model_reduction_cost): Likewise.
11916 (vect_create_epilog_for_reduction): Likewise.
11917 (vectorizable_reduction, vectorizable_induction): Likewise.
11918 * tree-vect-patterns.c (vect_init_pattern_stmt): Likewise.
11919 Return the stmt_vec_info for the pattern statement.
11920 (vect_set_pattern_stmt): Update use of STMT_VINFO_RELATED_STMT.
11921 (vect_split_statement, vect_mark_pattern_stmts): Likewise.
11922 * tree-vect-slp.c (vect_detect_hybrid_slp_stmts): Likewise.
11923 (vect_detect_hybrid_slp, vect_get_slp_defs): Likewise.
11924 * tree-vect-stmts.c (vect_mark_relevant): Likewise.
11925 (vect_get_vec_def_for_operand_1, vectorizable_call): Likewise.
11926 (vectorizable_simd_clone_call, vect_analyze_stmt, new_stmt_vec_info)
11927 (free_stmt_vec_info, vect_is_simple_use): Likewise.
11929 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
11931 * tree-vectorizer.h (vect_finish_replace_stmt): Return a stmt_vec_info
11932 (vect_finish_stmt_generation): Likewise.
11933 * tree-vect-stmts.c (vect_finish_stmt_generation_1): Likewise.
11934 (vect_finish_replace_stmt, vect_finish_stmt_generation): Likewise.
11935 (vect_build_gather_load_calls): Use the return value of the above
11936 functions instead of a separate call to vinfo_for_stmt. Use narrow
11937 scopes for the input gimple stmt and wider scopes for the associated
11938 stmt_vec_info. Use vec_info::lookup_def when setting these
11939 stmt_vec_infos from an SSA_NAME definition.
11940 (vectorizable_bswap, vectorizable_call, vectorizable_simd_clone_call)
11941 (vect_create_vectorized_demotion_stmts, vectorizable_conversion)
11942 (vectorizable_assignment, vectorizable_shift, vectorizable_operation)
11943 (vectorizable_store, vectorizable_load, vectorizable_condition)
11944 (vectorizable_comparison): Likewise.
11945 * tree-vect-loop.c (vectorize_fold_left_reduction): Likewise.
11946 (vectorizable_reduction): Likewise.
11948 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
11950 * tree-vectorizer.h (vect_is_simple_use): Add an optional
11951 stmt_vec_info * parameter before the optional gimple **.
11952 * tree-vect-stmts.c (vect_is_simple_use): Likewise.
11953 (process_use, vect_get_vec_def_for_operand_1): Update callers.
11954 (vect_get_vec_def_for_operand, vectorizable_shift): Likewise.
11955 * tree-vect-loop.c (vectorizable_reduction): Likewise.
11956 (vectorizable_live_operation): Likewise.
11957 * tree-vect-patterns.c (type_conversion_p): Likewise.
11958 (vect_look_through_possible_promotion): Likewise.
11959 (vect_recog_rotate_pattern): Likewise.
11960 * tree-vect-slp.c (vect_get_and_check_slp_defs): Likewise.
11962 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
11964 * tree-vectorizer.h (stmt_vec_info): Temporarily change from
11965 a typedef to a wrapper class.
11966 (NULL_STMT_VEC_INFO): New macro.
11967 (vec_info::stmt_infos): Change to vec<stmt_vec_info>.
11968 (stmt_vec_info::operator*): New function.
11969 (stmt_vec_info::operator gimple *): Likewise.
11970 (set_vinfo_for_stmt): Use NULL_STMT_VEC_INFO.
11971 (add_stmt_costs): Likewise.
11972 * tree-vect-loop-manip.c (iv_phi_p): Likewise.
11973 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost)
11974 (vect_get_known_peeling_cost): Likewise.
11975 (vect_estimate_min_profitable_iters): Likewise.
11976 * tree-vect-patterns.c (vect_init_pattern_stmt): Likewise.
11977 * tree-vect-slp.c (vect_remove_slp_scalar_calls): Likewise.
11978 * tree-vect-stmts.c (vect_build_gather_load_calls): Likewise.
11979 (vectorizable_store, free_stmt_vec_infos): Likewise.
11980 (new_stmt_vec_info): Change return type of xcalloc to
11983 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
11985 * tree-vectorizer.h (vec_info::lookup_single_use): Declare.
11986 * tree-vectorizer.c (vec_info::lookup_single_use): New function.
11987 * tree-vect-loop.c (vectorizable_reduction): Use it instead of
11988 a single_imm_use-based sequence.
11989 * tree-vect-stmts.c (supportable_widening_operation): Likewise.
11991 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
11993 * tree-vectorizer.h (vec_info::lookup_def): Declare.
11994 * tree-vectorizer.c (vec_info::lookup_def): New function.
11995 * tree-vect-patterns.c (vect_get_internal_def): Use it.
11996 (vect_widened_op_tree): Likewise.
11997 * tree-vect-stmts.c (vect_is_simple_use): Likewise.
11998 * tree-vect-loop.c (vect_analyze_loop_operations): Likewise.
11999 (vectorizable_reduction): Likewise.
12000 (vect_valid_reduction_input_p): Take a stmt_vec_info instead
12002 (vect_is_slp_reduction): Update calls accordingly. Use
12003 vec_info::lookup_def.
12004 (vect_is_simple_reduction): Likewise
12005 * tree-vect-slp.c (vect_detect_hybrid_slp_1): Use vec_info::lookup_def.
12007 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
12009 * tree-vectorizer.h (vec_info::lookup_stmt): Declare.
12010 * tree-vectorizer.c (vec_info::lookup_stmt): New function.
12011 * tree-vect-loop.c (vect_determine_vf_for_stmt): Use it instead
12013 (vect_determine_vectorization_factor, vect_analyze_scalar_cycles_1)
12014 (vect_compute_single_scalar_iteration_cost, vect_analyze_loop_form)
12015 (vect_update_vf_for_slp, vect_analyze_loop_operations)
12016 (vect_is_slp_reduction, vectorizable_induction)
12017 (vect_transform_loop_stmt, vect_transform_loop): Likewise.
12018 * tree-vect-patterns.c (vect_init_pattern_stmt):
12019 (vect_determine_min_output_precision_1, vect_determine_precisions)
12020 (vect_pattern_recog): Likewise.
12021 * tree-vect-stmts.c (vect_analyze_stmt, vect_transform_stmt): Likewise.
12022 * config/powerpcspe/powerpcspe.c (rs6000_density_test): Likewise.
12023 * config/rs6000/rs6000.c (rs6000_density_test): Likewise.
12024 * tree-vect-slp.c (vect_detect_hybrid_slp_stmts): Likewise.
12025 (vect_detect_hybrid_slp_1, vect_detect_hybrid_slp_2)
12026 (vect_detect_hybrid_slp): Likewise. Change the walk_stmt_info
12027 info field from a loop to a loop_vec_info.
12029 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
12031 * tree-vectorizer.h (stmt_vec_info): Move typedef earlier in file.
12032 (vec_info::add_stmt): Declare.
12033 * tree-vectorizer.c (vec_info::add_stmt): New function.
12034 * tree-vect-data-refs.c (vect_create_data_ref_ptr): Use it.
12035 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Likewise.
12036 (vect_create_epilog_for_reduction, vectorizable_reduction): Likewise.
12037 (vectorizable_induction): Likewise.
12038 * tree-vect-slp.c (_bb_vec_info::_bb_vec_info): Likewise.
12039 * tree-vect-stmts.c (vect_finish_stmt_generation_1): Likewise.
12040 (vectorizable_simd_clone_call, vectorizable_store): Likewise.
12041 (vectorizable_load): Likewise.
12042 * tree-vect-patterns.c (vect_init_pattern_stmt): Likewise.
12043 (vect_recog_bool_pattern, vect_recog_mask_conversion_pattern)
12044 (vect_recog_gather_scatter_pattern): Likewise.
12045 (append_pattern_def_seq): Likewise. Remove a check that is
12046 performed by add_stmt itself.
12048 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
12050 * tree-vect-loop.c (vectorizable_reduction): Fix an instance in
12051 which make_ssa_name was called with new_stmt before new_stmt
12054 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
12056 * tree-vect-loop.c (vect_valid_reduction_input_p): New function,
12058 (vect_is_slp_reduction): ...here...
12059 (vect_is_simple_reduction): ...and here. Remove repetition of tests
12060 that are already known to be false.
12062 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
12064 * tree-vectorizer.h (vect_free_slp_instance): Add a final_p parameter.
12065 * tree-vect-slp.c (vect_free_slp_tree): Likewise. Don't update
12066 STMT_VINFO_NUM_SLP_USES when it's true.
12067 (vect_free_slp_instance): Add a final_p parameter and pass it to
12068 vect_free_slp_tree.
12069 (vect_build_slp_tree_2): Update call to vect_free_slp_instance.
12070 (vect_analyze_slp_instance): Likewise.
12071 (vect_slp_analyze_operations): Likewise.
12072 (vect_slp_analyze_bb_1): Likewise.
12073 * tree-vectorizer.c (vec_info): Likewise.
12074 * tree-vect-loop.c (vect_transform_loop): Likewise.
12076 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
12078 * tree-vect-loop.c (vectorizable_reduction): Assert that the
12079 function is not called for second and subsequent members of
12082 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
12084 * tree-vect-loop.c (get_initial_def_for_reduction): Move special
12085 cases for nested loops from here to ...
12086 (vect_create_epilog_for_reduction): ...here. Only call
12087 vect_is_simple_use for inner-loop reductions.
12089 2018-07-31 Martin Liska <mliska@suse.cz>
12091 PR gcov-profile/85338
12092 PR gcov-profile/85350
12093 PR gcov-profile/85372
12094 * profile.c (struct location_triplet): New.
12095 (struct location_triplet_hash): Likewise.
12096 (output_location): Do not output a BB that
12097 is already recorded for a line.
12098 (branch_prob): Use streamed_locations.
12100 2018-07-31 Martin Liska <mliska@suse.cz>
12102 PR gcov-profile/85370
12103 * coverage.c (coverage_begin_function): Do not mark target
12104 clones as artificial functions.
12106 2018-07-31 Martin Liska <mliska@suse.cz>
12108 PR gcov-profile/83813
12109 PR gcov-profile/84758
12110 PR gcov-profile/85217
12111 PR gcov-profile/85332
12112 * profile.c (branch_prob): Do not record GOTO expressions
12113 for GIMPLE statements which locations are already streamed.
12115 2018-07-31 Olivier Hainque <hainque@adacore.com>
12117 * gcc.c (handle_spec_function): Accept a soft_matched_part
12118 argument, as do_spec_1. Pass it down to ...
12119 (eval_spec_function): Accept a soft_matched_part argument,
12120 and pass it down to ...
12121 (do_spec_2): Accept a soft_matched_part argument, and pass
12122 it down to do_spec_1.
12123 (do_spec_1): Pass soft_matched_part to handle_spec_function.
12124 (handle_braces): Update call to handle_spec_function.
12125 (driver::set_up_specs): Update calls to do_spec_2.
12126 (compare_debug_dump_opt_spec_function): Likewise.
12127 (compare_debug_self_opt_spec_function): Likewise.
12129 2018-07-31 Olivier Hainque <hainque@adacore.com>
12131 * common.opt (nolibc): New option.
12132 * doc/invoke.texi (Link Options): Document it.
12133 * gcc.c (LINK_GCC_C_SEQUENCE_SPEC): Honor nolibc.
12134 * config/alpha/linux.h: Likewise.
12135 * config/arc/elf.h: Likewise.
12136 * config/arm/uclinux-elf.h: Likewise.
12137 * config/arm/unknown-elf.h: Likewise.
12138 * config/avr/avrlibc.h: Likewise.
12139 * config/bfin/bfin.h: Likewise.
12140 * config/bfin/linux.h: Likewise.
12141 * config/bfin/uclinux.h: Likewise.
12142 * config/darwin.h: Likewise.
12143 * config/darwin10.h: Likewise.
12144 * config/darwin12.h: Likewise.
12145 * config/gnu-user.h: Likewise.
12146 * config/lm32/uclinux-elf.h: Likewise.
12147 * config/pa/pa-hpux11.h: Likewise.
12148 * config/pa/pa64-hpux.h: Likewise.
12149 * config/sparc/sparc.h: Likewise.
12151 2018-07-31 Olivier Hainque <hainque@adacore.com>
12153 * gcc.c (getenv_spec_function): Prepend '/' to value for allowed
12154 undefined variables.
12156 2018-07-30 Segher Boessenkool <segher@kernel.crashing.org>
12159 * config/arm/arm.c (arm_block_set_aligned_vect): Use gen_int_mode
12160 instead of GEN_INT.
12162 2018-07-30 Bernd Edlinger <bernd.edlinger@hotmail.de>
12164 * tree-ssa-forwprop.c (simplify_builtin_call): Don't create a not NUL
12165 terminated string literal.
12167 2018-07-30 Segher Boessenkool <segher@kernel.crashing.org>
12169 PR rtl-optimization/85160
12170 * combine.c (is_just_move): New function.
12171 (try_combine): Allow combining two instructions into two if neither of
12172 the original instructions was a move.
12174 2018-07-30 Alexander Monakov <amonakov@ispras.ru>
12177 * doc/extend.texi (Global Register Variables): Discourage use of type
12179 (Local Register Variables): Likewise.
12181 2018-07-30 Richard Sandiford <richard.sandiford@arm.com>
12183 PR tree-optimization/86506
12184 * hwint.h (ceil_log2): Resync with hwint.c implementation.
12186 2018-07-30 Ilya Leoshkevich <iii@linux.ibm.com>
12189 * lra-constraints.c (spill_hard_reg_in_range): When selecting the
12190 hard_regno, make sure no insn between `from` and `to` clobbers it.
12192 2018-07-30 Cesar Philippidis <cesar@codesourcery.com>
12193 Tom de Vries <tdevries@suse.de>
12195 * config/nvptx/nvptx.c (PTX_GANG_DEFAULT): Rename to ...
12196 (PTX_DEFAULT_RUNTIME_DIM): ... this.
12197 (nvptx_goacc_validate_dims): Set default worker and gang dims to
12198 PTX_DEFAULT_RUNTIME_DIM.
12199 (nvptx_dim_limit): Ignore GOMP_DIM_WORKER.
12201 2018-07-29 John David Anglin <danglin@gcc.gnu.org>
12203 * config/pa/pa.c (pa_output_addr_vec): Align address table.
12204 * config/pa/pa.h (JUMP_TABLES_IN_TEXT_SECTION): Revise comment.
12205 * config/pa/pa32-linux.h (JUMP_TABLES_IN_TEXT_SECTION): Define.
12207 2018-07-27 Michael Meissner <meissner@linux.ibm.com>
12209 * config/rs6000/constraints.md (wG constraint): Delete, no longer
12211 * config/rs6000/predicates.md (p9_fusion_reg_operand): Rename
12212 predicate to reflect toc fusion has been deleted.
12213 (toc_fusion_mem_raw): Delete, no longer used.
12214 (toc_fusion_mem_wrapped): Likewise.
12215 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete toc
12217 * config/rs6000/rs6000-protos.h (fusion_wrap_memory_address):
12218 Delete, no longer used.
12219 * config/rs6000/rs6000.c (struct rs6000_reg_addr): Delete fields
12220 meant to be used for toc fusion.
12221 (rs6000_debug_print_mode): Delete toc fusion debugging.
12222 (rs6000_debug_reg_global): Likewise.
12223 (rs6000_init_hard_regno_mode_ok): Delete setting up fields for toc
12224 fusion and secondary reload support that were never used.
12225 (rs6000_option_override_internal): Delete TOC fusion, that was only
12226 partially defined, and it did not work unless you also used the
12228 (rs6000_legitimate_address_p): Delete TOC fusion support.
12229 (rs6000_opt_masks): Likewise.
12230 (fusion_wrap_memory_address): Delete function, no longer used.
12231 (fusion_split_address); Delete TOC fusion support.
12232 * config/rs6000/rs6000.h (TARGET_TOC_FUSION_INT): Delete, no
12233 longer used with toc fusion being deleted.
12234 (TARGET_TOC_FUSION_FP): Likewise.
12235 * config/rs6000/rs6000.md (UNSPEC_FUSION_ADDIS): Delete TOC fusion
12237 (toc fusion spliter): Delete TOC fusion support.
12238 (toc_fusionload_<mode>): Likewise.
12239 (toc_fusionload_di): Likewise.
12240 (fusion_gpr_load_<mode>): Delete generator function, this insn no
12241 longer needs to be named. Rename predicate to delete TOC fusion.
12242 (fusion_gpr_<P:mode>_<GPR_FUSION:mode>_load): Likewise.
12243 (fusion_gpr_<P:mode>_<GPR_FUSION:mode>_store): Likewise.
12244 (fusion_vsx_<P:mode>_<GPR_FUSION:mode>_load): Likewise.
12245 (fusion_vsx_<P:mode>_<GPR_FUSION:mode>_store): Likewise.
12246 (p9 fusion peephole2s): Rename predicate to delete TOC fusion.
12248 2018-07-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
12250 * doc/extend.texi (Basic PowerPC Built-in Functions Available on
12251 ISA 2.05): Replace __uint128_t with __uint128 and __int128_t with
12252 __int128 in built-in function prototypes.
12253 (PowerPC AltiVec Built-in Functions on ISA 2.07): Likewise.
12254 (PowerPC AltiVec Built-in Functions on ISA 3.0): Likewise.
12256 2018-07-27 Martin Sebor <msebor@redhat.com>
12258 PR tree-optimization/86696
12259 * tree-ssa-strlen.c (get_min_string_length): Handle all integer
12260 types, including enums.
12261 (handle_char_store): Be prepared for the above function to fail.
12263 2018-07-26 Qing Zhao <qing.zhao@oracle.com>
12265 * builtins.c (inline_expand_builtin_string_cmp): Disable inlining
12266 when optimization level is lower than 2 or optimize for size.
12268 2018-07-26 Martin Sebor <msebor@redhat.com>
12270 PR tree-optimization/86043
12271 PR tree-optimization/86042
12272 * tree-ssa-strlen.c (handle_builtin_memcpy): Handle strict overlaps.
12273 (get_string_cst_length): Rename...
12274 (get_min_string_length): ...to this. Add argument.
12275 (handle_char_store): Extend to handle multi-character stores by
12277 * tree.c (initializer_zerop): Use new argument. Handle MEM_REF.
12278 * tree.h (initializer_zerop): Add argument.
12280 2018-07-26 Jakub Jelinek <jakub@redhat.com>
12282 PR middle-end/86660
12283 * omp-low.c (scan_sharing_clauses): Don't ignore map clauses for
12284 declare target to variables if they have always,{to,from,tofrom} map
12287 2018-07-26 Martin Liska <mliska@suse.cz>
12290 * lto-wrapper.c: Add linker_output as prefix
12291 for ltrans_output_file.
12293 2018-07-26 Segher Boessenkool <segher@kernel.crashing.org>
12295 PR rtl-optimization/85805
12296 * combine.c (reg_nonzero_bits_for_combine): Only use the last set
12297 value for hard registers if that was written in the same mode.
12299 2018-07-26 Martin Liska <mliska@suse.cz>
12301 PR gcov-profile/86536
12302 * gcov.c (format_gcov): Use printf format %.*f directly
12303 and do not handle special values.
12305 2018-07-25 Claudiu Zissulescu <claziss@synopsys.com>
12307 * common/config/arc/arc-common.c (arc_option_optimization_table):
12308 Update default optimizations for size.
12310 2018-07-25 Claudiu Zissulescu <claziss@synopsys.com>
12312 * config/arc/arc.md (movsf_insn): Add short instruction selection.
12313 * config/arc/constraints.md (CfZ): New constraint.
12314 * config/arc/fpu.md (addssf3_fpu): Use CfZ constraint.
12315 (subsf3_fpu): Likewise.
12316 (cmpsf_fpu): Likewise.
12317 (cmpsf_fpu_uneq): Likewise.
12319 2018-07-25 Claudiu Zissulescu <claziss@synopsys.com>
12321 * config/arc/arc.c (compact_memory_operand_p): Check for uncached
12323 (arc_is_uncached_mem_p): uncached applies to both the variable and
12326 2018-07-25 Claudiu Zissulescu <claziss@synopsys.com>
12328 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Add additional
12331 2018-07-25 David Malcolm <dmalcolm@redhat.com>
12333 * optinfo-emit-json.cc (class optrecord_json_writer): Convert
12334 field "m_scopes" from vec to auto_vec.
12336 2018-07-25 Martin Liska <mliska@suse.cz>
12338 * config/powerpcspe/powerpcspe-protos.h (rs6000_loop_align): Fix
12341 2018-07-25 Richard Biener <rguenther@suse.de>
12344 * dwarf2out.c (dwarf2out_decl): Do not handle nested functions
12345 special wrt context_die late.
12346 (gen_subprogram_die): Re-use DIEs in local scope.
12348 2018-07-25 Richard Sandiford <richard.sandiford@arm.com>
12350 PR tree-optimization/86644
12351 * hwint.c (ceil_log2): Fix comment. Return 0 for 0.
12353 2018-07-25 Martin Liska <mliska@suse.cz>
12355 PR middle-end/86645
12356 * dumpfile.c: And excluded values with TDF_ALL_VALUES.
12357 * dumpfile.h (enum dump_flag): Defince TDF_ALL_VALUES.
12359 2018-07-25 Martin Liska <mliska@suse.cz>
12362 * params.def: Explain ASan abbreviation and provide
12363 a documentation link.
12365 2018-07-24 Martin Sebor <msebor@redhat.com>
12367 PR tree-optimization/86622
12368 PR tree-optimization/86532
12369 * builtins.h (string_length): Declare.
12370 * builtins.c (c_strlen): Correct handling of non-constant offsets.
12371 (check_access): Be prepared for non-constant length ranges.
12372 (string_length): Make extern.
12373 * expr.c (string_constant): Only handle the minor non-constant
12374 array index. Use string_constant to compute the length of
12375 a generic string constant.
12377 2018-07-24 Richard Sandiford <richard.sandiford@arm.com>
12379 PR tree-optimization/86618
12380 * tree-vect-stmts.c (vectorizable_call): Don't take the address
12381 of LOOP_VINFO_MASKS (loop_vinfo) when loop_vinfo is null.
12383 2018-07-24 David Malcolm <dmalcolm@redhat.com>
12385 PR tree-optimization/86636
12386 * json.cc (json::object::set): Fix comment. Add assertions.
12387 (json::array::append): Move here from json.h. Add comment and an
12389 (json::string::string): Likewise.
12390 * json.h (json::array::append): Move to json.cc.
12391 (json::string::string): Likewise.
12392 * optinfo-emit-json.cc
12393 (optrecord_json_writer::impl_location_to_json): Assert that we
12394 aren't attempting to write out UNKNOWN_LOCATION, or an ad-hoc
12395 wrapper around it. Expand the location once, rather than three
12397 (optrecord_json_writer::inlining_chain_to_json): Fix the check for
12398 UNKNOWN_LOCATION, to use LOCATION_LOCUS to look through ad-hoc
12400 (optrecord_json_writer::optinfo_to_json): Likewise, in four
12401 places. Fix some overlong lines.
12403 2018-07-24 Matthew Malcomson <matthew.malcomson@arm.com>
12405 * config/aarch64/aarch64-simd.md
12406 (aarch64_<ANY_EXTEND:su><ADDSUB:optab>w<mode>): Split into...
12407 (aarch64_<ANY_EXTEND:su>subw<mode>): ... This...
12408 (aarch64_<ANY_EXTEND:su>addw<mode>): ... And this.
12409 (aarch64_<ANY_EXTEND:su><ADDSUB:optab>w<mode>_internal): Split into...
12410 (aarch64_<ANY_EXTEND:su>subw<mode>_internal): ... This...
12411 (aarch64_<ANY_EXTEND:su>addw<mode>_internal): ... And this.
12412 (aarch64_<ANY_EXTEND:su><ADDSUB:optab>w2<mode>_internal): Split into...
12413 (aarch64_<ANY_EXTEND:su>subw2<mode>_internal): ... This...
12414 (aarch64_<ANY_EXTEND:su>addw2<mode>_internal): ... And this.
12416 2018-07-24 Jakub Jelinek <jakub@redhat.com>
12418 PR middle-end/86627
12419 * expmed.c (expand_divmod): Punt if d == HOST_WIDE_INT_MIN
12420 and size > HOST_BITS_PER_WIDE_INT. For size > HOST_BITS_PER_WIDE_INT
12421 and abs_d == d, do the power of two handling if profitable.
12423 2018-07-24 Richard Biener <rguenther@suse.de>
12425 * match.pd: Add BIT_FIELD_REF canonicalizations.
12427 2018-07-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
12430 * genmatch.c (dt_operand::gen_match_op): Avoid folding volatile values.
12432 2018-07-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
12434 * gimple-fold.c (gimple_fold_builtin_printf): Don't create a not NUL
12435 terminated STRING_CST object.
12437 2018-07-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
12439 hsa-dump.c (dump_hsa_symbol): Avoid out of scope access to buf.
12441 2018-07-23 Segher Boessenkool <segher@kernel.crashing.org>
12443 * config/rs6000/rs6000-p8swap.c (rtx_is_swappable_p): Adjust.
12444 * config/rs6000/rs6000-protos.h (rs6000_split_v4si_init): Delete.
12445 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Always force
12446 the elements into a register.
12447 (rs6000_split_v4si_init_di_reg): Delete.
12448 (rs6000_split_v4si_init): Delete.
12449 * config/rs6000/vsx.md (unspec): Delete UNSPEC_VSX_VEC_INIT.
12450 (vsx_init_v4si): Rewrite as a define_expand.
12452 2018-07-23 Segher Boessenkool <segher@kernel.crashing.org>
12454 * config/rs6000/rs6000.md (splitters for rldimi and rlwimi with the
12455 zero_extend argument from memory): New.
12457 2018-07-22 Martin Sebor <msebor@redhat.com>
12460 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Avoid
12461 diagnosing calls with unknown arguments unless -Walloca-larger-than
12462 is restricted to less than PTRDIFF_MAX bytes.
12464 2018-07-22 Gerald Pfeifer <gerald@pfeifer.com>
12466 * doc/gcov.texi (Invoking Gcov): Editorial changes.
12468 2018-07-20 David Malcolm <dmalcolm@redhat.com>
12470 * pretty-print.c (text_info::set_location): Remove redundant
12471 "line_table" parameter from call to rich_location::set_range.
12473 2018-07-20 Martin Sebor <msebor@redhat.com>
12475 PR middle-end/82063
12476 * builtins.c (expand_builtin_alloca): Adjust.
12477 * calls.c (alloc_max_size): Simplify.
12478 * cgraphunit.c (cgraph_node::expand): Adjust.
12479 * common.opt (larger_than_size, warn_frame_larger_than): Remove
12481 (frame_larger_than_size): Same.
12482 (-Wframe-larger-than, -Wlarger-than, -Wstack-usage): Change options
12483 to take a HOST_WIDE_INT argument and accept a byte-size suffix.
12485 * doc/invoke.texi (GCC Command Options): Document option arguments.
12486 Explain byte-size arguments and suffixes.
12487 (-Wvla-larger-than, -Wno-alloc-size-larger-than): Update.
12488 (-Wno-alloca-larger-than, -Wno-vla-larger-than): Same.
12489 (-Wframe-larger-than, -Wlarger-than, -Wstack-usage): Same.
12490 * doc/options.texi (UInteger): Expand.
12491 (Host_Wide_Int, ByteSize): Document new properties.
12492 * final.c (final_start_function_1): Include sizes in an error message.
12493 * function.c (frame_offset_overflow): Same.
12494 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Adjust.
12495 (alloca_call_type_by_arg): Change function argument to HOST_WIDE_INT.
12496 Diagnose unbounded alloca calls only for limits of less than
12498 (alloca_call_type): Adjust. Diagnose possibly out-of-bounds alloca
12499 calls and VLA size only for limits of less than PTRDIFF_MAX. Same
12501 (pass_walloca::execute): Adjust. Diagnose alloca calls in loops
12502 only for limits of less than PTRDIFF_MAX.
12503 * langhooks-def.h (lhd_handle_option): Change function argument
12505 * langhooks.c (lhd_handle_option): Same.
12506 * langhooks.h (handle_option): Same.
12507 * opt-functions.awk (switch_bit_fields): Handle Host_Wide_Int and
12509 (var_type, var_type_struct): Same.
12510 (var_set): Handle ByteSize flag.
12511 * optc-gen.awk: Add comments to output to ease debugging. Make
12512 use of HOST_WIDE_INT where appropriate.
12513 * opts-gen-save.awk: Use %lx to format unsigned long.
12514 * opth-gen.awk: Change function argument to HOST_WIDE_INT.
12515 * opts-common.c (integral_argument): Return HOST_WIDE_INT and add
12516 arguments. Parse bytes-size suffixes.
12517 (enum_arg_to_value): Change function argument to HOST_WIDE_INT.
12518 (enum_value_to_arg): Same.
12519 (decode_cmdline_option): Handle cl_host_wide_int. Adjust.
12520 (handle_option): Adjust.
12521 (generate_option): Change function argument to HOST_WIDE_INT.
12522 (cmdline_handle_error): Adjust.
12523 (read_cmdline_option): Change function argument to HOST_WIDE_INT.
12524 (set_option): Change function argument to HOST_WIDE_INT.
12525 (option_enabled): Handle cl_host_wide_int.
12526 (get_option_state): Handle CLVC_SIZE.
12527 (control_warning_option): Same.
12528 * opts.c (common_handle_option): Change function argument to
12529 HOST_WIDE_INT. Remove handling of OPT_Walloca_larger_than_ and
12530 OPT_Wvla_larger_than_.
12531 * opts.h (enum cl_var_type): Add an enumerator.
12532 * stor-layout.c (layout_decl): Print a more meaningful warning.
12533 * toplev.c (output_stack_usage): Adjust.
12535 2018-07-20 Qing Zhao <qing.zhao@oracle.com>
12537 * builtins.c (expand_builtin_memcmp): Delete the last parameter for
12538 call to inline_expand_builtin_string_cmp.
12539 (expand_builtin_strcmp): Likewise.
12540 (expand_builtin_strncmp): Likewise.
12541 (inline_string_cmp): Delete the last parameter, change char_type_node
12542 to unsigned_char_type_node for strcmp/strncmp, add conversions to the
12544 (inline_expand_builtin_string_cmp): Delete the last parameter, give up
12545 the inlining expansion on target where the type of the call has same or
12546 narrower precision than unsigned char.
12548 2018-07-20 David Malcolm <dmalcolm@redhat.com>
12550 * Makefile.in (OBJS): Add json.o and optinfo-emit-json.o.
12551 (CFLAGS-optinfo-emit-json.o): Define TARGET_NAME.
12552 * common.opt (fsave-optimization-record): New option.
12553 * coretypes.h (struct kv_pair): Move here from dumpfile.c.
12554 * doc/invoke.texi (-fsave-optimization-record): New option.
12555 * dumpfile.c: Include "optinfo-emit-json.h".
12556 (struct kv_pair): Move to coretypes.h.
12557 (optgroup_options): Make non-static.
12558 (dump_context::end_scope): Call
12559 optimization_records_maybe_pop_dump_scope.
12560 * dumpfile.h (optgroup_options): New decl.
12561 * json.cc: New file.
12562 * json.h: New file.
12563 * optinfo-emit-json.cc: New file.
12564 * optinfo-emit-json.h: New file.
12565 * optinfo.cc: Include "optinfo-emit-json.h".
12566 (optinfo::emit): Call optimization_records_maybe_record_optinfo.
12567 (optinfo_enabled_p): Check optimization_records_enabled_p.
12568 (optinfo_wants_inlining_info_p): Likewise.
12569 * optinfo.h: Update comment.
12570 * profile-count.c (profile_quality_as_string): New function.
12571 * profile-count.h (profile_quality_as_string): New decl.
12572 (profile_count::quality): New accessor.
12573 * selftest-run-tests.c (selftest::run_tests): Call json_cc_tests
12574 and optinfo_emit_json_cc_tests.
12575 * selftest.h (selftest::json_cc_tests): New decl.
12576 (selftest::optinfo_emit_json_cc_tests): New decl.
12577 * toplev.c: Include "optinfo-emit-json.h".
12578 (compile_file): Call optimization_records_finish.
12579 (do_compile): Call optimization_records_start.
12580 * tree-ssa-live.c: Include optinfo.h.
12581 (remove_unused_scope_block_p): Retain inlining information if
12582 optinfo_wants_inlining_info_p returns true.
12584 2018-07-20 Richard Biener <rguenther@suse.de>
12587 * dwarf2out.c (dwarf2out_die_ref_for_decl): Test in_lto_p
12588 to cover -flto-partition=none.
12590 2018-07-20 Martin Liska <mliska@suse.cz>
12592 * tree.h (DECL_LOCATION_RANGE): Remove unused macro.
12593 (get_decl_source_range): Remove unused function.
12595 2018-07-20 Richard Biener <rguenther@suse.de>
12597 * tree-ssa-sccvn.h (struct vn_nary_op_s): Add next member.
12598 (struct vn_phi_s): Likewise.
12599 (struct vn_reference_s): Likewise.
12600 * tree-ssa-sccvn.c (vn_nary_op_hasher::equal): Add shortcut
12601 for searching the slot of an entry known to be in the hash itself.
12602 (vn_phi_hasher::equal): Likewise.
12603 (vn_reference_hasher::equal): Likewise.
12604 (last_inserted_ref, last_inserted_phi, last_inserted_nary): New
12606 (optimistic_info, current_info): Remove, keeping only valid_info.
12607 (vn_reference_lookup_1): Remove fallback lookup.
12608 (vn_reference_lookup_2): Likewise.
12609 (vn_nary_op_lookup_1): Likewise.
12610 (vn_phi_lookup): Likewise.
12611 (vn_nary_build_or_lookup_1): Make sure to not chain the built
12613 (vn_reference_insert): Adjust, chain the inserted hash element
12614 at last_inserted_ref.
12615 (vn_reference_insert_pieces): Likewise.
12616 (visit_reference_op_call): Likewise.
12617 (vn_nary_op_insert_into): Chain the inserted hash element at
12618 last_inserted_nary.
12619 (vn_nary_op_insert_pieces): Adjust.
12620 (vn_nary_op_insert): Likewise.
12621 (vn_nary_op_insert_stmt): Likewise.
12622 (vn_phi_insert): Adjust, chain the inserted hash element at
12624 (process_scc): Remove clearing and copying the optimistic
12625 table. Instead remove elements inserted during an optimistic
12626 iteration from the single table we maintain.
12627 (init_scc_vn): Adjust.
12628 (free_scc_vn): Likewise.
12629 (sccvn_dom_walker::record_cond): Likewise.
12630 (sccvn_dom_walker::after_dom_children): Likewise.
12632 2018-07-19 Martin Sebor <msebor@redhat.com>
12634 PR tree-optimization/84047
12635 PR tree-optimization/83776
12636 * tree-vrp.c (vrp_prop::check_mem_ref): New function.
12637 (check_array_bounds): Call it.
12639 2018-07-19 Martin Sebor <msebor@redhat.com>
12641 * align.h (align_flags): Use member initialization.
12643 2018-07-19 David Malcolm <dmalcolm@redhat.com>
12645 * Makefile.in (OBJS): Add optinfo.o.
12646 * coretypes.h (class symtab_node): New forward decl.
12647 (struct cgraph_node): New forward decl.
12648 (class varpool_node): New forward decl.
12649 * dump-context.h: New file.
12650 * dumpfile.c: Include "optinfo.h", "dump-context.h", "cgraph.h",
12652 (refresh_dumps_are_enabled): Use optinfo_enabled_p.
12653 (set_dump_file): Call dumpfile_ensure_any_optinfo_are_flushed.
12654 (set_alt_dump_file): Likewise.
12655 (dump_context::~dump_context): New dtor.
12656 (dump_gimple_stmt): Move implementation to...
12657 (dump_context::dump_gimple_stmt): ...this new member function.
12658 Add the stmt to any pending optinfo, creating one if need be.
12659 (dump_gimple_stmt_loc): Move implementation to...
12660 (dump_context::dump_gimple_stmt_loc): ...this new member function.
12661 Start a new optinfo and add the stmt to it.
12662 (dump_gimple_expr): Move implementation to...
12663 (dump_context::dump_gimple_expr): ...this new member function.
12664 Add the stmt to any pending optinfo, creating one if need be.
12665 (dump_gimple_expr_loc): Move implementation to...
12666 (dump_context::dump_gimple_expr_loc): ...this new member function.
12667 Start a new optinfo and add the stmt to it.
12668 (dump_generic_expr): Move implementation to...
12669 (dump_context::dump_generic_expr): ...this new member function.
12670 Add the tree to any pending optinfo, creating one if need be.
12671 (dump_generic_expr_loc): Move implementation to...
12672 (dump_context::dump_generic_expr_loc): ...this new member
12673 function. Add the tree to any pending optinfo, creating one if
12675 (dump_printf): Move implementation to...
12676 (dump_context::dump_printf_va): ...this new member function. Add
12677 the text to any pending optinfo, creating one if need be.
12678 (dump_printf_loc): Move implementation to...
12679 (dump_context::dump_printf_loc_va): ...this new member function.
12680 Start a new optinfo and add the stmt to it.
12681 (dump_dec): Move implementation to...
12682 (dump_context::dump_dec): ...this new member function. Add the
12683 value to any pending optinfo, creating one if need be.
12684 (dump_context::dump_symtab_node): New member function.
12685 (dump_context::get_scope_depth): New member function.
12686 (dump_context::begin_scope): New member function.
12687 (dump_context::end_scope): New member function.
12688 (dump_context::ensure_pending_optinfo): New member function.
12689 (dump_context::begin_next_optinfo): New member function.
12690 (dump_context::end_any_optinfo): New member function.
12691 (dump_context::s_current): New global.
12692 (dump_context::s_default): New global.
12693 (dump_scope_depth): Delete global.
12694 (dumpfile_ensure_any_optinfo_are_flushed): New function.
12695 (dump_symtab_node): New function.
12696 (get_dump_scope_depth): Reimplement in terms of dump_context.
12697 (dump_begin_scope): Likewise.
12698 (dump_end_scope): Likewise.
12699 (selftest::temp_dump_context::temp_dump_context): New ctor.
12700 (selftest::temp_dump_context::~temp_dump_context): New dtor.
12701 (selftest::verify_item): New function.
12702 (ASSERT_IS_TEXT): New macro.
12703 (ASSERT_IS_TREE): New macro.
12704 (ASSERT_IS_GIMPLE): New macro.
12705 (selftest::test_capture_of_dump_calls): New test.
12706 (selftest::dumpfile_c_tests): Call it.
12707 * dumpfile.h (dump_printf, dump_printf_loc, dump_basic_block)
12708 (dump_generic_expr_loc, dump_generic_expr, dump_gimple_stmt_loc)
12709 (dump_gimple_stmt, dump_dec): Gather these related decls and add a
12710 descriptive comment.
12711 (dump_function, print_combine_total_stats, enable_rtl_dump_file)
12712 (dump_node, dump_bb): Move these unrelated decls.
12713 (class dump_manager): Add leading comment.
12714 * optinfo.cc: New file.
12715 * optinfo.h: New file.
12717 2018-07-19 Michael Collison <michael.collison@arm.com>
12718 Richard Henderson <rth@redhat.com>
12720 * config/aarch64/aarch64.md (subv<GPI>4, usubv<GPI>4): New patterns.
12721 (subti): Handle op1 zero.
12722 (subvti4, usub4ti4): New.
12723 (*sub<GPI>3_compare1_imm): New.
12724 (sub<GPI>3_carryinCV): New.
12725 (*sub<GPI>3_carryinCV_z1_z2, *sub<GPI>3_carryinCV_z1): New.
12726 (*sub<GPI>3_carryinCV_z2, *sub<GPI>3_carryinCV): New.
12728 2018-07-19 Michael Collison <michael.collison@arm.com>
12729 Richard Henderson <rth@redhat.com>
12731 * config/aarch64/aarch64.md: (addv<GPI>4, uaddv<GPI>4): New.
12732 (addti3): Create simpler code if low part is already known to be 0.
12733 (addvti4, uaddvti4): New.
12734 (*add<GPI>3_compareC_cconly_imm): New.
12735 (*add<GPI>3_compareC_cconly): New.
12736 (*add<GPI>3_compareC_imm): New.
12737 (*add<GPI>3_compareC): Rename from add<GPI>3_compare1; do not
12738 handle constants within this pattern..
12739 (*add<GPI>3_compareV_cconly_imm): New.
12740 (*add<GPI>3_compareV_cconly): New.
12741 (*add<GPI>3_compareV_imm): New.
12742 (add<GPI>3_compareV): New.
12743 (add<GPI>3_carryinC, add<GPI>3_carryinV): New.
12744 (*add<GPI>3_carryinC_zero, *add<GPI>3_carryinV_zero): New.
12745 (*add<GPI>3_carryinC, *add<GPI>3_carryinV): New.
12746 ((*add<GPI>3_compareC_cconly_imm): Replace 'ne' operator
12747 with 'comparison' operator.
12748 (*add<GPI>3_compareV_cconly_imm): Ditto.
12749 (*add<GPI>3_compareV_cconly): Ditto.
12750 (*add<GPI>3_compareV_imm): Ditto.
12751 (add<GPI>3_compareV): Ditto.
12752 (add<mode>3_carryinC): Ditto.
12753 (*add<mode>3_carryinC_zero): Ditto.
12754 (*add<mode>3_carryinC): Ditto.
12755 (add<mode>3_carryinV): Ditto.
12756 (*add<mode>3_carryinV_zero): Ditto.
12757 (*add<mode>3_carryinV): Ditto.
12759 2018-07-19 Michael Collison <michael.collison@arm.com>
12760 Richard Henderson <rth@redhat.com>
12762 * config/aarch64/aarch64-modes.def (CC_V): New.
12763 * config/aarch64/aarch64-protos.h
12764 (aarch64_addti_scratch_regs): Declare
12765 (aarch64_subvti_scratch_regs): Declare.
12766 (aarch64_expand_subvti): Declare.
12767 (aarch64_gen_unlikely_cbranch): Declare
12768 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Test
12769 for signed overflow using CC_Vmode.
12770 (aarch64_get_condition_code_1): Handle CC_Vmode.
12771 (aarch64_gen_unlikely_cbranch): New function.
12772 (aarch64_addti_scratch_regs): New function.
12773 (aarch64_subvti_scratch_regs): New function.
12774 (aarch64_expand_subvti): New function.
12776 2018-07-19 Andre Vieira <andre.simoesdiasvieira@arm.com>
12778 * config/aarch64/aarch64-option-extensions.def: New entry for profile
12780 * config/aarch64/aarch64.h (AARCH64_FL_PROFILE): New.
12781 * doc/invoke.texi (aarch64-feature-modifiers): New entry for profile
12784 2018-07-19 Andre Vieira <andre.simoesdiasvieira@arm.com>
12787 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Make
12788 address check not strict.
12790 2018-07-19 Andre Vieira <andre.simoesdiasvieira@arm.com>
12792 * config/aarch64/aarch64-simd.md (aarch64_simd_mov<VQ:mode>): Replace
12794 (store_pair_lanes<mode>): Likewise.
12795 * config/aarch64/aarch64-protos.h (aarch64_addr_query_type): Add new
12796 enum value 'ADDR_QUERY_LDP_STP_N'.
12797 * config/aarch64/aarch64.c (aarch64_addr_query_type): Likewise.
12798 (aarch64_print_address_internal): Add declaration.
12799 (aarch64_print_ldpstp_address): Remove.
12800 (aarch64_classify_address): Adapt mode for 'ADDR_QUERY_LDP_STP_N'.
12801 (aarch64_print_operand): Change printing of 'y'.
12802 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Use
12803 new enum value 'ADDR_QUERY_LDP_STP_N', don't hardcode mode and use
12804 'true' rather than '1'.
12805 * config/aarch64/constraints.md (Uml): Likewise.
12806 (Uml): Rename to Umn.
12809 2018-07-19 Richard Biener <rguenther@suse.de>
12811 * tree-ssa-sccvn.h (struct vn_phi_s): Make phiargs member
12813 * tree-ssa-sccvn.c: Remove alloc-pool.h use.
12814 (vn_phi_hasher): Derive from nofree_ptr_hash and remove remove method.
12815 (vn_reference_hasher): Likewise.
12816 (struct vn_tables_s): Remove obstack and alloc-pool members.
12817 (vn_tables_obstack, vn_tables_insert_obstack): New global obstacks.
12818 (vn_nary_build_or_lookup_1): Manually build in vn_tables_insert_obstack.
12819 (vn_reference_insert): Allocate from obstack instead of from alloc-pool.
12820 (vn_reference_insert_pieces): Likewise.
12821 (alloc_vn_nary_op_noinit): Adjust.
12822 (vn_nary_op_insert_stmt): Allocate phiargs in-place.
12823 (vn_phi_eq): Adjust.
12824 (shared_lookup_phiargs): Remove.
12825 (vn_phi_lookup): Allocate temporary vn_phi_s on the stack.
12826 (vn_phi_insert): Allocate from obstack instead of from alloc-pool.
12827 (visit_reference_op_call): Likewise.
12828 (copy_nary, copy_phi, copy_reference): Remove.
12829 (process_scc): Rewind the obstack when iterating. Do not
12830 copy the elements to valid_info but just move them from one
12831 hashtable to the other.
12832 (allocate_vn_table): Adjust.
12833 (free_vn_table): Likewise.
12834 (init_scc_vn): Likewise.
12835 (free_scc_vn): Likewise.
12837 2018-07-19 H.J. Lu <hongjiu.lu@intel.com>
12840 * config/i386/i386.c (rest_of_insert_endbranch): Lookup
12841 indirect_return as function type attribute.
12842 (ix86_attribute_table): Change indirect_return to function
12844 * doc/extend.texi: Update indirect_return attribute.
12846 2018-07-19 Aldy Hernandez <aldyh@redhat.com>
12848 * wide-int.h (widest2_int): New.
12849 * gimple-fold.c (arith_overflowed_p): Use it.
12850 * tree.h (widest2_int_cst): New.
12851 * tree-vrp.c (wide_int_binop_overflow): Rename from
12852 vrp_int_const_binop.
12853 Rewrite to work on trees.
12854 (extract_range_from_multiplicative_op_1): Abstract code to...
12855 (wide_int_range_min_max): ...here.
12856 (wide_int_range_cross_product): ...and here.
12857 (extract_range_from_binary_expr_1): Abstract overflow code to...
12858 (wide_int_range_mult_wrapping): ...here.
12859 * tree-vrp.h (wide_int_range_cross_product): New.
12860 (wide_int_range_mult_wrapping): New.
12862 2018-07-19 Andrew Senkevich <andrew.senkevich@intel.com>
12863 Julia Koval <julia.koval@intel.com>
12865 * config/i386/x86-tune-costs.h (skylake_memcpy,
12866 skylake_memset): Replace rep_prefix with unrolling for size 512.
12868 2018-07-18 Kugan Vivekanandarajah <kuganv@linaro.org>
12870 PR middle-end/86544
12871 * tree-ssa-phiopt.c (cond_removal_in_popcount_pattern): Handle
12872 comparision with EQ_EXPR in last stmt.
12874 2018-07-18 Kelvin Nilsen <kelvin@gcc.gnu.org>
12876 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Rename
12877 this subsection to "PowerPC AltiVec/VSX Built-in Functions".
12878 (PowerPC AltiVec/VSX Built-in Functions): New name for subsection
12879 previously known as "PowerPC AltiVec Built-in Functions". Move
12880 some material to new subsubsections "PowerPC AltiVec Built-in
12881 Functions on ISA 2.06" and "PowerPC AltiVec Built-in Functions on
12883 (PowerPC Altivec Built-in Functions on ISA 2.05): New subsubsection.
12884 (PowerPC Altivec Built-in Functions on ISA 2.06): Likewise.
12885 (PowerPC Altivec Built-in Functions on ISA 2.07): Likewise.
12886 (PowerPC Altivec Built-in Functions on ISA 3.0): Likewise.
12888 2018-07-18 Richard Biener <rguenther@suse.de>
12890 PR tree-optimization/86557
12891 * tree-vect-patterns.c (vect_recog_divmod_pattern): Also handle
12894 2018-07-18 Ilya Leoshkevich <iii@linux.ibm.com>
12896 * config/s390/s390.c (s390_function_profiler): Generate CFI.
12898 2018-07-17 Jeff Law <law@redhat.com>
12900 * config/arm/arm.c (get_label_padding): Update for recent
12901 changes to label_to_alignment.
12903 PR tree-optimization/86010
12904 * tree-ssa-dse.c (compute_trims): Fix typo/thinko.
12906 * config/mips/mips.c (vr4130_align_insns): Update for recent
12907 changes to label_to_alignment.
12909 * config/frv/frv.c (frv_label_align): Update for recent changes
12910 to label_to_alignment.
12912 * config/nios2/nios2.c (nios2_label_align): Update for recent
12913 changes which dropped ALIGN_LABELS_LOG.
12915 2018-07-17 Andreas Schwab <schwab@linux-m68k.org>
12917 * config/m68k/m68k.md (umulsi3_highpart+1, const_umulsi3_highpart)
12918 (smulsi3_highpart+1, const_smulsi3_highpart): Add CC_STATUS_INIT.
12920 2018-07-17 Claudiu Zissulescu <claziss@synopsys.com>
12922 * config/arc/arc.c (arc_label_align): Use align_labels instead of
12923 deprecated align_labels_log.
12925 2018-07-17 Richard Biener <rguenther@suse.de>
12928 * dwarf2out.c (init_sections_and_labels): Always generate
12929 a debug_line_str_section for early LTO debug.
12930 (dwarf2out_finish): Reset debug_line_str_hash output early.
12931 Bump counter for extra dwarf5 .debug_loc labels to not conflict
12933 (dwarf2out_early_finish): Output debug_line_str.
12935 2018-07-17 Robin Dapp <rdapp@linux.ibm.com>
12937 * config/s390/s390.c (preferred_la_operand_p): Do not use LA with
12938 index register on z196 or later.
12940 2018-07-17 Robin Dapp <rdapp@linux.ibm.com>
12942 * config/s390/s390.c (s390_default_align): Set default function
12944 (s390_override_options_after_change): Call s390_default align.
12945 (s390_option_override_internal): Call s390_default_align.
12946 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define.
12948 2018-07-17 Jakub Jelinek <jakub@redhat.com>
12950 PR middle-end/86542
12951 * omp-low.c (create_task_copyfn): Copy over also fields corresponding
12952 to _looptemp_ clauses, other than the first two.
12954 2018-07-17 Martin Liska <mliska@suse.cz>
12956 * opts.c: Do not enable OPT_falign_* for -Os.
12958 2018-07-17 Martin Liska <mliska@suse.cz>
12960 * align.h (MAX_CODE_ALIGN): New.
12961 (MAX_CODE_ALIGN_VALUE): New.
12962 * common/config/i386/i386-common.c (ix86_handle_option):
12963 (MAX_CODE_ALIGN): Moved to align.h.
12964 * final.c (MAX_CODE_ALIGN): Likewise.
12965 * opts.c (parse_and_check_align_values):
12966 (MAX_CODE_ALIGN): Likewise.
12967 (MAX_CODE_ALIGN_VALUE): Likewise.
12969 2018-07-17 Martin Liska <mliska@suse.cz>
12971 * config/i386/att.h (ASM_OUTPUT_ALIGN): Fix spacing
12972 in order to fulfil coding style.
12973 * config/i386/cygming.h (ASM_OUTPUT_ALIGN): Likewise.
12974 * config/i386/gas.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
12975 * config/i386/x86-64.h (ASM_OUTPUT_MAX_SKIP_PAD): Likewise.
12976 * config/iq2000/iq2000.h (ASM_OUTPUT_ALIGN): Likewise.
12977 * config/pa/pa.h (ASM_OUTPUT_ALIGN): Likewise.
12978 * config/sparc/sol2.h (ASM_OUTPUT_ALIGN_WITH_NOP): Likewise.
12979 * config/sparc/sparc.h (ASM_OUTPUT_ALIGN): Likewise.
12980 * config/visium/visium.h (ASM_OUTPUT_ALIGN): Likewise.
12981 (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
12983 2018-07-17 Martin Liska <mliska@suse.cz>
12985 * align.h: New file.
12986 * config/alpha/alpha.c (alpha_align_insns_1): Use align_functions
12988 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Use new return type
12989 align_flags of label_to_alignment.
12990 * config/m32r/m32r.h (LOOP_ALIGN): Wrap returned values into
12992 * config/m68k/m68k.c: Do not use removed align_labels_value and
12994 * config/nds32/nds32.h (JUMP_ALIGN): Wrap result into align_flags class.
12995 (LOOP_ALIGN): Likewise.
12996 (LABEL_ALIGN): Likewise.
12997 * config/powerpcspe/powerpcspe.c (TARGET_ASM_LOOP_ALIGN_MAX_SKIP):
12998 Remove not used macro.
12999 (rs6000_loop_align): Change return type to align_flags.
13000 (rs6000_loop_align_max_skip): Remove.
13001 * config/rs6000/rs6000-protos.h (rs6000_loop_align):
13002 Change return type to align_flags.
13003 * config/rs6000/rs6000.c (TARGET_ASM_LOOP_ALIGN_MAX_SKIP):
13004 Remove not used macro.
13005 (rs6000_loop_align): Change return type to align_flags.
13006 (rs6000_loop_align_max_skip): Remove.
13007 * config/rx/rx.h (JUMP_ALIGN): Wrap integer values
13008 * config/rx/rx-protos.h (rx_align_for_label): Make it
13010 * config/rx/rx.c (rx_align_for_label): Change return type
13012 (rx_max_skip_for_label): Remove TARGET_ASM_*_ALIGN_MAX_SKIP
13014 into align_flags class.
13015 (LABEL_ALIGN): Likewise.
13016 (LOOP_ALIGN): Likewise.
13017 * config/s390/s390.c (s390_label_align): Use align_flags
13019 (s390_asm_output_function_label): Likewise.
13020 * config/sh/sh.c (sh_override_options_after_change):
13021 Use align_flags class directly without macros.
13022 (find_barrier): Likewise.
13023 (barrier_align): Likewise.
13024 (sh_loop_align): Likewise.
13025 * config/spu/spu.c (spu_option_override):
13026 Use align_flags_tuple::get_value instead of removed macros.
13027 (spu_sched_init): Likewise.
13028 * config/spu/spu.h (GTY): Likewise.
13029 * config/visium/visium.c (visium_option_override):
13030 Set "8" as default secondary alignment.
13031 * config/visium/visium.h (SUBALIGN_LOG): Define to 3
13032 in order to guarantee secondary alignment of 8.
13033 * coretypes.h: Include align.h header file.
13034 * doc/tm.texi: Remove TARGET_ASM_JUMP_ALIGN_MAX_SKIP,
13035 TARGET_ASM_LOOP_ALIGN_MAX_SKIP, TARGET_ASM_LABEL_ALIGN_MAX_SKIP
13036 and TARGET_ASM_LABEL_ALIGN_AFTER_BARRIER_MAX_SKIP macros.
13037 * doc/tm.texi.in: Likewise.
13038 * final.c (struct label_alignment): Remove not used structure.
13039 (LABEL_ALIGN): Change type to align_flags.
13040 (LOOP_ALIGN): Likewise.
13041 (JUMP_ALIGN): Likewise.
13042 (default_loop_align_max_skip): Remove.
13043 (default_label_align_max_skip): Likewise.
13044 (default_jump_align_max_skip): Likewise.
13045 (default_label_align_after_barrier_max_skip):
13046 (LABEL_TO_ALIGNMENT): Change to access label_align vector.
13047 (LABEL_TO_MAX_SKIP): Remove.
13048 (label_to_alignment): Return align_flags type instead of integer.
13049 (label_to_max_skip): Remove.
13050 (align_fuzz): Use align_flags type.
13051 (compute_alignments): Use align_flags type and use align_flags::max
13052 to combine multiple alignments.
13053 (grow_label_align): Grow vec instead of C array.
13054 (update_alignments): Assign just LABEL_TO_ALIGNMENT.
13055 (shorten_branches): Use align_flags type and use align_flags::max
13056 to combine multiple alignments.
13057 (final_scan_insn_1): Remove usage of secondary alignment that comes
13058 from label alignment, but instead use proper secondary alignment
13059 which is computed in grow_label_align.
13060 * flags.h (struct align_flags_tuple): Move to align.h.
13061 (struct align_flags): Likewise.
13062 (state_align_loops): Rename to align_loops.
13063 (state_align_jumps): Rename to align_jumps.
13064 (state_align_labels): Rename to align_labels.
13065 (state_align_functions): Rename to align_functions.
13066 (align_loops_log): Remove.
13067 (align_jumps_log): Remove.
13068 (align_labels_log): Remove.
13069 (align_functions_log): Remove.
13070 (align_loops_max_skip): Remove.
13071 (align_jumps_max_skip): Remove.
13072 (align_labels_max_skip): Remove.
13073 (align_functions_max_skip): Remove.
13074 (align_loops_value): Remove.
13075 (align_jumps_value): Remove.
13076 (align_labels_value): Remove.
13077 (align_functions_value): Remove.
13078 * output.h (label_to_alignment): Change return type to align_flags.
13079 (label_to_max_skip): Remove.
13080 * target.def: Remove loop_align_max_skip, label_align_max_skip,
13081 jump_align_max_skip macros.
13082 * targhooks.h (default_loop_align_max_skip): Remove.
13083 (default_label_align_max_skip): Likewise.
13084 (default_jump_align_max_skip): Likewise.
13085 (default_label_align_after_barrier_max_skip): Remove.
13086 * toplev.c (read_log_maxskip): Use ::normalize function.
13087 (parse_N_M): Remove not used argument and also call ::normalize.
13088 (parse_alignment_opts): Do not pass unused arguments.
13089 * varasm.c (assemble_start_function): Use directly align_functions
13090 instead of removed macros.
13091 * system.h: Do not poison removed macros.
13093 2018-07-17 Jakub Jelinek <jakub@redhat.com>
13095 PR middle-end/86539
13096 * gimplify.c (gimplify_omp_for): Ensure taskloop firstprivatized init
13097 and cond temporaries don't have reference type if iterator has
13098 pointer type. For init use &for_pre_body instead of pre_p if
13099 for_pre_body is non-empty.
13101 2018-07-16 Segher Boessenkool <segher@kernel.crashing.org>
13103 * config/rs6000/rs6000.md (trunc<mode>sf2): Expand truncates of
13104 double-double modes to SFmode directly directly.
13105 (trunc<mode>sf2_fprs): Delete.
13107 2018-07-16 Segher Boessenkool <segher@kernel.crashing.org>
13109 * config/rs6000/rs6000.c (init_float128_ibm): Use the correct names
13110 for conversions between IFmode and the decimal floating point modes.
13111 (init_float128_ieee): Use the correct names for conversions between
13112 KFmode and the decimal floating point modes.
13114 2018-07-16 Segher Boessenkool <segher@kernel.crashing.org>
13116 * config/rs6000/rs6000.c (init_float128_ibm): Use more correct names
13117 for the conversions between TDmode and IFmode.
13118 (init_float128_ieee): Use more correct names for the conversions
13119 between TDmode and KFmode.
13121 2018-07-16 Jakub Jelinek <jakub@redhat.com>
13123 PR tree-optimization/86526
13124 * builtins.c (expand_builtin_memcmp): Formatting fixes.
13125 (inline_expand_builtin_string_cmp): Likewise.
13126 (inline_string_cmp): Likewise. Use c_readstr instead of
13127 builtin_memcpy_read_str. Add unit_mode temporary.
13129 2018-07-16 Bernd Edlinger <bernd.edlinger@hotmail.de>
13131 PR middle-end/86528
13132 * builtins.c (check_access): Bail out if range[0] is no INTEGER_CST.
13133 * expr.c (string_constant): Fix the element size of ARRAY_TYPE.
13135 2018-07-16 Kelvin Nilsen <kelvin@gcc.gnu.org>
13137 * doc/extend.texi (PowerPC AltiVec Built-in Functions):
13138 Alphabetize prototypes of built-in functions, separating out
13139 built-in functions that are listed in this section but should be
13140 described elsewhere.
13142 2018-07-16 Uros Bizjak <ubizjak@gmail.com>
13145 * expmed.c (emit_store_flag): Do not emit setcc followed by a
13146 conditional move when trapping comparison was split to a
13147 non-trapping one (and vice versa).
13149 2018-07-16 Ilya Leoshkevich <iii@linux.ibm.com>
13151 * config/s390/s390.c (s390_function_profiler): Generate nops
13152 instead of profiler call sequences.
13153 * config/s390/s390.opt: Add the new option.
13155 2018-07-16 Ilya Leoshkevich <iii@linux.ibm.com>
13157 * config/s390/s390.c (s390_function_profiler): Generate
13158 __mcount_loc section.
13159 * config/s390/s390.opt: Add the new option.
13161 2018-07-16 Ilya Leoshkevich <iii@linux.ibm.com>
13163 * common.opt: Add the new warning.
13164 * config/s390/s390.c (s390_function_profiler): Emit "brasl
13165 %r0,__fentry__" when -mfentry is specified.
13166 (s390_option_override_internal): Disallow -mfentry for 31-bit
13168 * config/s390/s390.opt: Add the new option.
13170 2018-07-16 Richard Biener <rguenther@suse.de>
13173 * dwarf2out.c (dwarf2out_register_external_die): Assign DIE parents
13174 for function-local FUNCTION_DECL and RESULT_DECL immediately.
13176 2018-07-16 Martin Liska <mliska@suse.cz>
13179 * ipa-pure-const.c (malloc_candidate_p): Revert ::get
13182 2017-07-16 Claudiu Zissulescu <claziss@synopsys.com>
13184 * config/arc/arcHS.md: Update ARCHS scheduling rules.
13186 2017-07-16 Claudiu Zissulescu <claziss@synopsys.com>
13188 * config/arc/arc-arch.h (arc_tune_attr): Add new tune parameters
13190 * config/arc/arc-cpus.def (hs4x): New cpu.
13192 * config/arc/arc-tables.opt: Regenerate.
13193 * config/arc/arc.c (arc_sched_issue_rate): New function.
13194 (TARGET_SCHED_ISSUE_RATE): Define.
13195 (TARGET_SCHED_EXPOSED_PIPELINE): Likewise.
13196 * config/arc/arc.md (attr type): Add fpu_fuse, fpu_sdiv, fpu_ddiv,
13198 (attr tune): Add ARCHS4x tune values.
13199 (attr tune_dspmpy): Define.
13200 (*tst): Correct instruction type.
13201 * config/arc/arcHS.md: Don't use this automaton for ARCHS4x cpus.
13202 * config/arc/arcHS4x.md: New file.
13203 * config/arc/fpu.md: Update instruction type attributes.
13204 * config/arc/t-multilib: Regenerate.
13206 2018-07-16 Tom de Vries <tdevries@suse.de>
13209 * var-tracking.c (vt_initialize): Fix pre_dec handling.
13211 2018-07-16 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
13213 * config/aarch64/atomics.md (aarch64_store_execlusive<mode>): Add
13216 2018-07-16 Eric Botcazou <ebotcazou@adacore.com>
13218 PR tree-optimization/86514
13219 * tree-ssa-reassoc.c (init_range_entry) <CASE_CONVERT>: Return for a
13220 conversion to a boolean type from a type with greater precision.
13222 2018-07-16 Tom de Vries <tdevries@suse.de>
13224 * var-tracking.c (vt_initialize): Print adjusted insn slim if
13225 dump_flags request TDF_SLIM.
13227 2018-07-16 Aldy Hernandez <aldyh@redhat.com>
13229 * fold-const.c (int_const_binop_1): Abstract...
13230 (wide_int_binop): ...wide int code here.
13231 (poly_int_binop): ...poly int code here.
13232 Abstract the rest of int_const_binop_1 into int_const_binop.
13233 * fold-const.h (wide_int_binop): New.
13234 * tree-vrp.c (vrp_int_const_binop): Call wide_int_binop.
13235 Remove useless PLUS/MINUS_EXPR case.
13236 (zero_nonzero_bits_from_vr): Move wide int code...
13237 (zero_nonzero_bits_from_bounds): ...here.
13238 (extract_range_from_binary_expr_1): Move mask optimization code...
13239 (range_easy_mask_min_max): ...here.
13240 * tree-vrp.h (zero_nonzero_bits_from_bounds): New.
13241 (range_easy_mask_min_max): New.
13243 2018-07-15 Jeff Law <law@redhat.com>
13246 * config/sh/sh.c (output_mi_thunk): Remove dead conditional
13249 2018-07-14 Jim Wilson <jimw@sifive.com>
13251 * config/riscv/linux.h (TARGET_ASM_FILE_END): New.
13253 2018-07-14 Paul Koning <ni1d@arrl.net>
13255 * config/pdp11/pdp11.c (pdp11_rtx_costs): Bugfixes.
13257 2018-07-13 Jan Hubicka <hubicka@ucw.cz>
13259 * lto-streamer-out.c (copy_function_or_variable): Dump info about
13262 2018-07-13 Bill Schmidt <wschmidt@linux.ibm.com>
13263 Steve Munroe <munroesj52@gmail.com>
13265 * config/rs6000/emmintrin.h (_mm_and_si128): New function.
13266 (_mm_andnot_si128): Likewise.
13267 (_mm_or_si128): Likewise.
13268 (_mm_xor_si128): Likewise.
13270 2018-07-13 Qing Zhao <qing.zhao@oracle.com>
13272 PR middle-end/78809
13273 * builtins.c (expand_builtin_memcmp): Inline the calls first
13274 when result_eq is false.
13275 (expand_builtin_strcmp): Inline the calls first.
13276 (expand_builtin_strncmp): Likewise.
13277 (inline_string_cmp): New routine. Expand a string compare
13278 call by using a sequence of char comparison.
13279 (inline_expand_builtin_string_cmp): New routine. Inline expansion
13280 a call to str(n)cmp/memcmp.
13281 * doc/invoke.texi (--param builtin-string-cmp-inline-length):
13283 * params.def (BUILTIN_STRING_CMP_INLINE_LENGTH): New.
13285 2018-07-13 Richard Earnshaw <rearnsha@arm.com>
13287 * config/arm/driver-arm.c: Include arm-native.h.
13288 (host_detect_local_cpu): Use auto-generated data tables.
13289 (vendors, arm_cpu_table): Delete. Move part information to ...
13290 * config/arm/arm-cpus.in: ... here.
13291 * config/arm/parsecpu.awk (gen_native): New function.
13292 (vendor, part): New CPU fields.
13293 (END): Add support for building the native CPU detection tables.
13294 * config/arm/t-arm (arm-native.h): Add build rule.
13295 (driver-arm.o): Add dependency on arm-native.h.
13297 2018-07-13 Richard Biener <rguenther@suse.de>
13299 PR middle-end/85974
13300 * match.pd (addr1 - addr2): Allow either of the operand to
13303 2018-07-13 Tom de Vries <tdevries@suse.de>
13305 * tree-inline.c (remap_ssa_name): Save and reuse debug exprs generated
13308 2018-07-13 Jackson Woodruff <jackson.woodruff@arm.com>
13310 * config/aarch64/aarch64.c (aarch64_operands_adjust_ok_for_ldpstp): Use
13311 arrays instead of numbered variables.
13313 2018-07-13 Eric Botcazou <ebotcazou@adacore.com>
13315 * config/sparc/sparc-protos.h (sparc_compute_frame_size): Delete.
13316 * config/sparc/sparc.c (sparc_compute_frame_size): Make static.
13318 2018-07-13 Richard Biener <rguenther@suse.de>
13321 * dwarf2out.c (gen_type_die_with_usage): Use scope_die_for
13322 instead of get_context_die.
13324 2018-07-13 Kugan Vivekanandarajah <kuganv@linaro.org>
13325 Richard Biener <rguenther@suse.de>
13327 PR middle-end/86489
13328 * tree-ssa-loop-niter.c (number_of_iterations_popcount): Check
13329 that the loop latch destination where phi is defined.
13331 2018-07-12 Kito Cheng <kito.cheng@gmail.com>
13333 * config/riscv/riscv.c (enum riscv_privilege_levels): Add UNKNOWN_MODE.
13334 (riscv_expand_epilogue): Add assertion to check interrupt mode.
13335 (riscv_set_current_function): Extract getting interrupt type to new
13337 (riscv_get_interrupt_type): New function.
13338 (riscv_merge_decl_attributes): New function, checking interrupt type is
13340 (TARGET_MERGE_DECL_ATTRIBUTES): Define.
13342 2018-07-12 Paul Koning <ni1d@arrl.net>
13344 * config/pdp11/pdp11.c (pdp11_output_def): Fix typo in .set
13347 2018-07-12 Paul Koning <ni1d@arrl.net>
13349 * doc/rtl.texi (REG_NONNEG): Remove decrement and branch until
13350 zero reference, add doloop_end instead.
13351 * doc/md.texi (decrement_and_branch_until_zero): Remove.
13352 (Looping patterns): Remove decrement_and_branch_until_zero. Add
13353 detail for doloop_end.
13355 2018-07-12 Martin Sebor <msebor@redhat.com>
13358 * attribs.c (decl_attributes): Reject conflicting attributes before
13359 calling attribute handlers.
13361 2018-07-12 Jan Hubicka <hubicka@ucw.cz>
13363 * dumpfile.c (gcc::dump_manager::get_dump_file_name): Add PART
13365 (gcc::dump_manager::get_dump_file_name): likewise.
13366 (dump_begin): Likewise.
13367 * dumpfile.h (dump_begin): Update prototype.
13368 (gcc::dump_manager::get_dump_file_name,
13369 gcc::dump_manager::get_dump_file_name): Update prototype.
13371 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
13373 * internal-fn.h (vectorizable_internal_fn_p): New function.
13374 * tree-vect-slp.c (compatible_calls_p): Likewise.
13375 (vect_build_slp_tree_1): Remove nops argument. Handle calls
13376 to internal functions.
13377 (vect_build_slp_tree_2): Update call to vect_build_slp_tree_1.
13379 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
13381 * fold-const.h (inverse_conditions_p): Declare.
13382 * fold-const.c (inverse_conditions_p): New function.
13383 * match.pd: Use inverse_conditions_p. Add folds of view_converts
13384 that test the inverse condition of a conditional internal function.
13385 * internal-fn.h (vectorized_internal_fn_supported_p): Declare.
13386 * internal-fn.c (internal_fn_mask_index): Handle conditional
13387 internal functions.
13388 (vectorized_internal_fn_supported_p): New function.
13389 * tree-if-conv.c: Include internal-fn.h and fold-const.h.
13390 (any_pred_load_store): Replace with...
13391 (need_to_predicate): ...this new variable.
13392 (redundant_ssa_names): New variable.
13393 (ifcvt_can_use_mask_load_store): Move initial checks to...
13394 (ifcvt_can_predicate): ...this new function. Handle tree codes
13395 for which a conditional internal function exists.
13396 (if_convertible_gimple_assign_stmt_p): Use ifcvt_can_predicate
13397 instead of ifcvt_can_use_mask_load_store. Update after variable
13399 (predicate_load_or_store): New function, split out from
13400 predicate_mem_writes.
13401 (check_redundant_cond_expr): New function.
13402 (value_available_p): Likewise.
13403 (predicate_rhs_code): Likewise.
13404 (predicate_mem_writes): Rename to...
13405 (predicate_statements): ...this. Use predicate_load_or_store
13406 and predicate_rhs_code.
13407 (combine_blocks, tree_if_conversion): Update after above name changes.
13408 (ifcvt_local_dce): Handle redundant_ssa_names.
13409 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern): Handle
13410 general conditional functions.
13411 * tree-vect-stmts.c (vectorizable_call): Likewise.
13413 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
13414 Alan Hayward <alan.hayward@arm.com>
13415 David Sherwood <david.sherwood@arm.com>
13417 * internal-fn.h (can_interpret_as_conditional_op_p): Declare.
13418 * internal-fn.c (can_interpret_as_conditional_op_p): New function.
13419 * tree-ssa-math-opts.c (convert_mult_to_fma_1): Handle conditional
13420 plus and minus and convert them into IFN_COND_FMA-based sequences.
13421 (convert_mult_to_fma): Handle conditional plus and minus.
13423 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
13425 * doc/md.texi (cond_fma, cond_fms, cond_fnma, cond_fnms): Document.
13426 * optabs.def (cond_fma_optab, cond_fms_optab, cond_fnma_optab)
13427 (cond_fnms_optab): New optabs.
13428 * internal-fn.def (COND_FMA, COND_FMS, COND_FNMA, COND_FNMS): New
13429 internal functions.
13430 (FMA): Use DEF_INTERNAL_FLT_FN rather than DEF_INTERNAL_FLT_FLOATN_FN.
13431 * internal-fn.h (get_conditional_internal_fn): Declare.
13432 (get_unconditional_internal_fn): Likewise.
13433 * internal-fn.c (cond_ternary_direct): New macro.
13434 (expand_cond_ternary_optab_fn): Likewise.
13435 (direct_cond_ternary_optab_supported_p): Likewise.
13436 (FOR_EACH_COND_FN_PAIR): Likewise.
13437 (get_conditional_internal_fn): New function.
13438 (get_unconditional_internal_fn): Likewise.
13439 * gimple-match.h (gimple_match_op::MAX_NUM_OPS): Bump to 5.
13440 (gimple_match_op::gimple_match_op): Add a new overload for 5
13442 (gimple_match_op::set_op): Likewise.
13443 (gimple_resimplify5): Declare.
13444 * genmatch.c (decision_tree::gen): Generate simplifications for
13446 * gimple-match-head.c (gimple_simplify): Define an overload for
13447 5 operands. Handle calls with 5 arguments in the top-level overload.
13448 (convert_conditional_op): Handle conversions from unconditional
13449 internal functions to conditional ones.
13450 (gimple_resimplify5): New function.
13451 (build_call_internal): Pass a fifth operand.
13452 (maybe_push_res_to_seq): Likewise.
13453 (try_conditional_simplification): Try converting conditional
13454 internal functions to unconditional internal functions.
13455 Handle 3-operand unconditional forms.
13456 * match.pd (UNCOND_TERNARY, COND_TERNARY): Operator lists.
13457 Define ternary equivalents of the current rules for binary conditional
13458 internal functions.
13459 * config/aarch64/aarch64.c (aarch64_preferred_else_value): Handle
13460 ternary operations.
13461 * config/aarch64/iterators.md (UNSPEC_COND_FMLA, UNSPEC_COND_FMLS)
13462 (UNSPEC_COND_FNMLA, UNSPEC_COND_FNMLS): New unspecs.
13463 (optab): Handle them.
13464 (SVE_COND_FP_TERNARY): New int iterator.
13465 (sve_fmla_op, sve_fmad_op): New int attributes.
13466 * config/aarch64/aarch64-sve.md (cond_<optab><mode>)
13467 (*cond_<optab><mode>_2, *cond_<optab><mode_4)
13468 (*cond_<optab><mode>_any): New SVE_COND_FP_TERNARY patterns.
13470 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
13472 * target.def (preferred_else_value): New target hook.
13473 * doc/tm.texi.in (TARGET_PREFERRED_ELSE_VALUE): New hook.
13474 * doc/tm.texi: Regenerate.
13475 * targhooks.h (default_preferred_else_value): Declare.
13476 * targhooks.c (default_preferred_else_value): New function.
13477 * internal-fn.h (conditional_internal_fn_code): Declare.
13478 * internal-fn.c (FOR_EACH_CODE_MAPPING): New macro.
13479 (get_conditional_internal_fn): Use it.
13480 (conditional_internal_fn_code): New function.
13481 * gimple-match.h (gimple_match_cond): New struct.
13482 (gimple_match_op): Add a cond member function.
13483 (gimple_match_op::gimple_match_op): Update all forms to take a
13485 * genmatch.c (expr::gen_transform): Use the same condition as res_op
13486 for the suboperation, but don't specify a particular else_value.
13487 * tree-ssa-sccvn.c (vn_nary_simplify, vn_reference_lookup_3)
13488 (visit_nary_op, visit_reference_op_load): Pass
13489 gimple_match_cond::UNCOND to the gimple_match_op constructor.
13490 * gimple-match-head.c: Include tree-eh.h
13491 (convert_conditional_op): New function.
13492 (maybe_resimplify_conditional_op): Likewise.
13493 (gimple_resimplify1): Call maybe_resimplify_conditional_op.
13494 (gimple_resimplify2): Likewise.
13495 (gimple_resimplify3): Likewise.
13496 (gimple_resimplify4): Likewise.
13497 (maybe_push_res_to_seq): Return null for conditional operations.
13498 (try_conditional_simplification): New function.
13499 (gimple_simplify): Call it. Pass conditions to the gimple_match_op
13501 * match.pd: Fold VEC_COND_EXPRs of an IFN_COND_* call to a new
13503 * config/aarch64/aarch64.c (aarch64_preferred_else_value): New
13505 (TARGET_PREFERRED_ELSE_VALUE): Redefine.
13507 2018-07-12 Jan Hubicka <hubicka@ucw.cz>
13509 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not stream
13511 (hash_tree): Do not hash DECL_FCONTEXT
13512 * tree-streamer-in.c (lto_input_ts_field_decl_tree_pointers):
13513 Do not stream DECL_FCONTEXT.
13514 * tree-streamer-out.c (write_ts_field_decl_tree_pointers): Likewise.
13515 * tree.c (free_lang_data_in_decl): Free DECL_FCONTEXT.
13517 2018-07-12 Richard Biener <rguenther@suse.de>
13520 * dwarf2out.c (gen_block_die): Only output blocks when they have
13521 at least one !DECL_IGNORED_P variable.
13523 2018-07-12 Richard Biener <rguenther@suse.de>
13526 * config/gnu-user.h (GNU_USER_TARGET_NO_PTHREADS_LIB_SPEC):
13527 Remove -mieee-fp handling.
13529 2018-07-12 Richard Biener <rguenther@suse.de>
13531 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Remove bogus
13532 left-over from last patch.
13534 2018-07-12 Jakub Jelinek <jakub@redhat.com>
13536 PR tree-optimization/86492
13537 * gimple-ssa-store-merging.c
13538 (imm_store_chain_info::coalesce_immediate_stores): Call
13539 check_no_overlap even for the merge_overlapping case. Formatting fix.
13541 2018-07-12 Richard Biener <rguenther@suse.de>
13543 PR middle-end/86479
13544 * fold-const.c (fold_binary_op_with_conditional_arg): Do not
13545 move possibly trapping operations into the conditional.
13547 2018-07-12 Richard Biener <rguenther@suse.de>
13549 * tree-ssa-sccvn.c (mprts_hook_cnt): Remove.
13550 (vn_lookup_simplify_result): Remove recursion limit applied
13552 (vn_nary_build_or_lookup_1): Adjust.
13553 (try_to_simplify): Likewise.
13554 * gimple-match-head.c (gimple_resimplify1): Instead apply one
13556 (gimple_resimplify2): Likewise.
13557 (gimple_resimplify3): Likewise.
13558 (gimple_resimplify4): Likewise.
13560 2018-07-11 Jakub Jelinek <jakub@redhat.com>
13562 * config/i386/avx512bitalgintrin.h (_mm512_mask_bitshuffle_epi64_mask):
13563 Use __mmask64 type instead of __mmask8 for __M argument.
13564 * config/i386/avx512fintrin.h (_mm512_mask_xor_epi64,
13565 _mm512_maskz_xor_epi64): Use __mmask8 type instead of __mmask16 for
13567 (_mm512_mask_cmpneq_epi64_mask): Use __mmask8 type instead of
13568 __mmask16 for __M argument.
13569 (_mm512_maskz_insertf32x4, _mm512_maskz_inserti32x4,
13570 _mm512_mask_insertf32x4, _mm512_mask_inserti32x4): Cast last argument
13571 to __mmask16 instead of __mmask8.
13572 * config/i386/avx512vlintrin.h (_mm_mask_add_ps, _mm_maskz_add_ps,
13573 _mm256_mask_add_ps, _mm256_maskz_add_ps, _mm_mask_sub_ps,
13574 _mm_maskz_sub_ps, _mm256_mask_sub_ps, _mm256_maskz_sub_ps,
13575 _mm256_maskz_cvtepi32_ps, _mm_maskz_cvtepi32_ps): Use __mmask8 type
13576 instead of __mmask16 for __U argument.
13577 * config/i386/avx512vlbwintrin.h (_mm_mask_cmp_epi8_mask): Use
13578 __mmask16 instead of __mmask8 for __U argument.
13579 (_mm256_mask_cmp_epi8_mask): Use __mmask32 instead of __mmask16 for
13581 (_mm256_cmp_epi8_mask): Use __mmask32 return type instead of
13583 (_mm_mask_cmp_epu8_mask): Use __mmask16 instead of __mmask8 for __U
13585 (_mm256_mask_cmp_epu8_mask): Use __mmask32 instead of __mmask16 for
13587 (_mm256_cmp_epu8_mask): Use __mmask32 return type instead of
13589 (_mm_mask_cmp_epi16_mask): Cast last argument to __mmask8 instead
13591 (_mm256_mask_cvtepi8_epi16): Use __mmask16 instead of __mmask32 for
13593 (_mm_mask_cvtepi8_epi16): Use __mmask8 instead of __mmask32 for
13595 (_mm256_mask_cvtepu8_epi16): Use __mmask16 instead of __mmask32 for
13597 (_mm_mask_cvtepu8_epi16): Use __mmask8 instead of __mmask32 for
13599 (_mm256_mask_cmpneq_epu8_mask, _mm256_mask_cmplt_epu8_mask,
13600 _mm256_mask_cmpge_epu8_mask, _mm256_mask_cmple_epu8_mask): Change
13601 return type as well as __M argument type and all casts from __mmask8
13603 (_mm256_mask_cmpneq_epu16_mask, _mm256_mask_cmplt_epu16_mask,
13604 _mm256_mask_cmpge_epu16_mask, _mm256_mask_cmple_epu16_mask): Change
13605 return type as well as __M argument type and all casts from __mmask8
13607 (_mm256_mask_cmpneq_epi8_mask, _mm256_mask_cmplt_epi8_mask,
13608 _mm256_mask_cmpge_epi8_mask, _mm256_mask_cmple_epi8_mask): Change
13609 return type as well as __M argument type and all casts from __mmask8
13611 (_mm256_mask_cmpneq_epi16_mask, _mm256_mask_cmplt_epi16_mask,
13612 _mm256_mask_cmpge_epi16_mask, _mm256_mask_cmple_epi16_mask): Change
13613 return type as well as __M argument type and all casts from __mmask8
13615 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_shrdi_epi32,
13616 _mm_mask_shldi_epi32): Cast last argument to __mmask8 instead of
13619 2018-07-11 Grazvydas Ignotas <notasas@gmail.com>
13621 * config/i386/avx512bwintrin.h: (_mm512_mask_cmp_epi8_mask,
13622 _mm512_mask_cmp_epu8_mask): Use __mmask64 type instead of __mmask32
13625 2018-07-11 Paul Koning <ni1d@arrl.net>
13627 * doc/md.texi (define_subst): Document how multiple occurrences of
13628 the same argument in the replacement pattern are handled.
13630 2018-07-11 Paul Koning <ni1d@arrl.net>
13632 * doc/extend.texi (Common Variable Attributes): Move "mode" into
13633 alphabetical order.
13634 (Common Type Attributes): Add "mode" attribute.
13636 2018-07-11 Jan Hubicka <hubicka@ucw.cz>
13638 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not
13639 stream DECL_ORIGINAL_TYPE.
13640 (DFS::DFS_write_tree_body): Drop hack handling local external decls.
13641 (hash_tree): Do not walk DECL_ORIGINAL_TYPE.
13642 * tree-streamer-in.c (lto_input_ts_decl_non_common_tree_pointers):
13643 Do not walk original type.
13644 * tree-streamer-out.c (streamer_write_chain): Drop hack handling
13646 (write_ts_decl_non_common_tree_pointers): Do not stream
13648 * tree.c (free_lang_data_in_decl): Clear DECL_ORIGINAL_TYPE.
13649 (find_decls_types_r): Do not walk DEC_ORIGINAL_TYPE.
13651 2018-07-11 Aldy Hernandez <aldyh@redhat.com>
13653 * tree-ssa-threadupdate.c (thread_through_all_blocks): Do not jump
13654 thread twice from the same starting edge.
13656 2018-07-11 Aldy Hernandez <aldyh@redhat.com>
13658 * vr-values.c (gimple_stmt_nonzero_p): Abstract common code to...
13659 * gimple.c (gimple_call_nonnull_result_p): ...here...
13660 (gimple_call_nonnull_arg): ...and here.
13661 * gimple.h (gimple_call_nonnull_result_p): New.
13662 (gimple_call_nonnull_arg): New.
13664 2018-07-11 Richard Earnshaw <rearnsha@arm.com>
13666 * config/arm/arm-cpus.in: Move information from fpu field of each
13667 cpu definition to the isa field.
13668 * config/arm/parsecpu.awk (fpu): Delete match rule.
13669 (gen_comm_data): Don't add bits from the CPU's FPU entry.
13671 2018-07-11 Richard Biener <rguenther@suse.de>
13674 * dwarf2out.c (init_sections_and_labels): Use
13675 output_asm_line_debug_info consistently.
13676 (dwarf2out_early_finish): Likewise.
13677 (dwarf2out_finish): Remove DW_AT_stmt_list from early generated
13680 2018-07-11 Richard Biener <rguenther@suse.de>
13682 * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely_1):
13683 Rework father_bb setting in a way to avoid propagating constants
13684 multiple times on a loop body.
13686 2018-07-10 Mark Wielaard <mark@klomp.org>
13689 * dwarf2out.c (output_macinfo_op): Fix dwarf_FORM typo in gcc_assert.
13691 2018-07-10 Richard Biener <rguenther@suse.de>
13693 * hash-map.h (hash_map::iterator::operator*): Return
13694 references to key and value.
13696 2018-07-10 Jakub Jelinek <jakub@redhat.com>
13699 * gimplify.c (find_combined_omp_for): Add DATA argument, in addition
13700 to finding the inner OMP_FOR/OMP_SIMD stmt find non-trivial wrappers,
13701 BLOCKs with BLOCK_VARs, OMP_PARALLEL in between, OMP_FOR in between.
13702 (gimplify_omp_for): For composite loops, move outer
13703 OMP_{DISTRIBUTE,TASKLOOP,FOR,PARALLEL} right around innermost
13704 OMP_FOR/OMP_SIMD if there are any non-trivial wrappers. For class
13705 iterators add any needed clauses. Allow OMP_FOR_ORIG_DECLS to contain
13706 TREE_LIST for both the original class iterator and the "last" helper
13707 var. Gimplify OMP_FOR_PRE_BODY before the outermost composite
13708 loop, remember has_decl_expr from outer composite loops for the
13709 innermost OMP_SIMD in TREE_PRIVATE bit on OMP_FOR_INIT.
13711 2018-07-09 Martin Sebor <msebor@redhat.com>
13713 PR middle-end/77357
13714 PR middle-end/86428
13715 * builtins.c (c_strlen): Avoid out-of-bounds warnings when
13716 accessing implicitly initialized array elements.
13717 * expr.c (string_constant): Handle string initializers of
13718 character arrays within aggregates.
13719 * gimple-fold.c (fold_array_ctor_reference): Add argument.
13720 Store element offset. As a special case, handle zero size.
13721 (fold_nonarray_ctor_reference): Same.
13722 (fold_ctor_reference): Add argument. Store subobject offset.
13723 * gimple-fold.h (fold_ctor_reference): Add argument.
13725 2018-07-09 Paul Koning <ni1d@arrl.net>
13727 * config/pdp11/pdp11.c (pdp11_addr_cost): New function.
13728 (pdp11_insn_cost): New function.
13729 (pdp11_md_asm_adjust): New function.
13730 (TARGET_INVALID_WITHIN_DOLOOP): Define.
13731 (pdp11_rtx_costs): Update to match machine better.
13732 (output_addr_const_pdp11): Correct format mismatch warnings.
13733 * config/pdp11/pdp11.h (SLOW_BYTE_ACCESS): Correct definition.
13734 * config/pdp11/pdp11.md: General change to add base_cost and/or
13735 length attributes for use by new pdp11_insn_cost function.
13736 (MIN_BRANCH): Correct definition.
13738 (doloop_end): Use standard pattern name for looping pattern.
13739 (doloop_end_nocc): New.
13740 (movsf): Add another constraint alternative.
13741 (zero_extendqihi2): Add constraint alternatives for not in place
13743 (zero_extendhisi2): Remove.
13744 (shift patterns): Add CC handling variants.
13748 (define_peephole2): New peephole to recognize mov that sets CC for
13751 2018-07-09 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
13753 * config/sparc/sparc.c (sparc_fold_builtin) <SPARC_BUILTIN_PDIST,
13754 SPARC_BUILTIN_PDISTN>: Adapt for signature change of wi::neg,
13757 2018-07-09 Jakub Jelinek <jakub@redhat.com>
13760 * real.c (real_nextafter): Return true if result is denormal.
13762 2018-07-09 Martin Liska <mliska@suse.cz>
13764 * common.opt: Add back wrongly removed attribute.
13766 2018-07-09 Richard Biener <rguenther@suse.de>
13769 * dwarf2out.c (gen_block_die): For an early generated DIE
13770 always output high/low PC attributes.
13772 2018-07-09 Tom de Vries <tdevries@suse.de>
13774 * cfgexpand.c (expand_debug_source_expr): Handle VAR_DECL.
13775 * tree-inline.c (remap_ssa_name): Handle default def ssa_name mapping
13776 onto VAR_DECL with abstract origin.
13778 2018-07-07 Jim Wilson <jimw@sifive.com>
13780 * config/riscv/riscv.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): New.
13782 2018-07-07 Tom de Vries <tdevries@suse.de>
13784 * tree-dfa.c (dump_enumerated_decls): Handle cfun->cfg == NULL.
13786 2018-07-07 Aldy Hernandez <aldyh@redhat.com>
13788 * tree-vrp.c (vrp_int_const_binop): Change overflow type to
13790 (combine_bound): Use wide-int overflow calculation instead of
13792 * calls.c (maybe_warn_alloc_args_overflow): Change overflow type to
13794 * fold-const.c (int_const_binop_2): Same.
13795 (extract_muldiv_1): Same.
13796 (fold_div_compare): Same.
13797 (fold_abs_const): Same.
13799 * poly-int.h (add): Same.
13803 * predict.c (predict_iv_comparison): Same.
13804 * profile-count.c (slow_safe_scale_64bit): Same.
13805 * simplify-rtx.c (simplify_const_binary_operation): Same.
13806 * tree-chrec.c (tree_fold_binomial): Same.
13807 * tree-data-ref.c (split_constant_offset_1): Same.
13808 * tree-if-conv.c (idx_within_array_bound): Same.
13809 * tree-scalar-evolution.c (iv_can_overflow_p): Same.
13810 * tree-ssa-phiopt.c (minmax_replacement): Same.
13811 * tree-vect-loop.c (is_nonwrapping_integer_induction): Same.
13812 * tree-vect-stmts.c (vect_truncate_gather_scatter_offset): Same.
13813 * vr-values.c (vr_values::adjust_range_with_scev): Same.
13814 * wide-int.cc (wi::add_large): Same.
13815 (wi::mul_internal): Same.
13816 (wi::sub_large): Same.
13817 (wi::divmod_internal): Same.
13818 * wide-int.h: Change overflow type to overflow_type for neg, add,
13819 mul, smul, umul, div_trunc, div_floor, div_ceil, div_round,
13820 mod_trunc, mod_ceil, mod_round, add_large, sub_large,
13821 mul_internal, divmod_internal.
13822 (overflow_type): New enum.
13823 (accumulate_overflow): New.
13825 2018-07-06 Kugan Vivekanandarajah <kugan.vivekanandarajah@linaro.org>
13827 * tree-ssa-phiopt.c (cond_removal_in_popcount_pattern): New.
13828 (tree_ssa_phiopt_worker): Call cond_removal_in_popcount_pattern.
13830 2018-07-06 Kugan Vivekanandarajah <kuganv@linaro.org>
13832 * tree-ssa-loop-niter.c (number_of_iterations_popcount): If popcount
13833 argument is checked for zero before entering loop, avoid checking again.
13835 2018-07-06 Kugan Vivekanandarajah <kuganv@linaro.org>
13837 * gimplify.h (generic_expr_could_trap_p): Set as global function.
13838 * gimplify.h (generic_expr_could_trap_p): Likwise.
13839 * tree-scalar-evolution.c (expression_expensive_p): Handle COND_EXPR.
13841 2018-07-06 Jakub Jelinek <jakub@redhat.com>
13843 PR tree-optimization/86401
13844 * fold-const.c (fold_binary_loc) <case BIT_AND_EXPR>: Move the
13845 ((A & N) + B) & M -> (A + B) & M etc. optimization into ...
13846 (fold_bit_and_mask): ... here. New helper function for match.pd.
13847 * fold-const.h (fold_bit_and_mask): Declare.
13848 * match.pd (((A & N) + B) & M -> (A + B) & M): New optimization.
13850 2018-07-06 Peter Bergner <bergner@linux.ibm.com>
13853 * target.def (translate_mode_attribute): New hook.
13854 * targhooks.h (default_translate_mode_attribute): Declare.
13855 * targhooks.c (default_translate_mode_attribute): New function.
13856 * doc/tm.texi.in (TARGET_TRANSLATE_MODE_ATTRIBUTE): New hook.
13857 * doc/tm.texi: Regenerate.
13858 * config/rs6000/rs6000.c (TARGET_TRANSLATE_MODE_ATTRIBUTE): Define.
13859 (rs6000_translate_mode_attribute): New function.
13861 2018-07-06 Paul Koning <ni1d@arrl.net>
13863 * doc/md.texi (define_split): Document DONE and FAIL.
13864 (define_peephole2): Ditto.
13866 2018-07-05 Jeff Law <law@redhat.com>
13868 PR tree-optimization/86010
13869 * tree-ssa-dse.c (compute_trims): More aggressively trim at
13870 both the head and tail of mem* and str* calls.
13872 2018-07-05 Jim Wilson <jimw@sifive.com>
13874 * config.gcc (riscv*-*-*): When setting xlen, handle riscv-*.
13876 2018-07-05 Indu Bhagat <indu.bhagat@oracle.com>
13878 * config/aarch64/aarch64-simd.md: correct flags text for
13879 MIN_EXPR replacement.
13881 2018-07-05 James Clarke <jrtc27@jrtc27.com>
13883 * configure: Regenerated.
13885 2018-07-05 Carl Love <cel@us.ibm.com>
13887 * config/rs6000/rs6000-c.c: Map ALTIVEC_BUILTIN_VEC_UNPACKH for
13888 float argument to VSX_BUILTIN_DOUBLEH_V4SF.
13889 Map ALTIVEC_BUILTIN_VEC_UNPACKL for float argument to
13890 VSX_BUILTIN_DOUBLEL_V4SF.
13892 2018-07-05 Martin Sebor <msebor@redhat.com>
13895 * tree-ssa-strlen.c (maybe_set_strlen_range): Use type size rather
13896 than its domain to compute its the upper bound of a char array.
13898 2018-07-05 Nathan Sidwell <nathan@acm.org>
13900 Replace NO_IMPLICIT_EXTERN_C with SYSTEM_IMPLICIT_EXTERN_C.
13901 * doc/cpp.texi: Update comment.
13902 * doc/tm.texi: Rebuilt.
13903 * doc/tm.texi.in (NO_IMPLICIT_EXTERN_C): Replace with ...
13904 (SYSTEM_IMPLICIT_EXTERN_C): ... this, opposite sense.
13905 * doc/extend.texi (Backwards Compatibility): Clarify it is system
13906 headers affected by extern "C".
13907 * system.h: Poison NO_IMPLICIT_EXTERN_C.
13908 * config/alpha/alpha.h, config/arm/uclinux-elf.h,
13909 config/bfin/elf.h, config/cris/cris.h, config/darwin.h,
13910 config/dragonfly.h, config/freebsd.h, config/gnu-user.h,
13911 config/i386/cygming.h, config/i386/djgpp.h, config/i386/nto.h,
13912 config/ia64/hpux.h, config/lm32/lm32.h, config/lm32/uclinux-elf.h,
13913 config/lynx.h, config/mips/elf.h, config/mmix/mmix.h,
13914 config/netbsd.h, config/pa/pa-hpux.h, config/powerpcspe/sysv4.h,
13915 config/riscv/elf.h, config/rs6000/sysv4.h, config/rtems.h,
13916 config/s390/tpf.h, config/sh/newlib.h, config/sol2.h,
13917 config/sparc/openbsd64.h, config/sparc/sp-elf.h,
13918 config/sparc/sp64-elf.h, config/spu/spu.h,
13919 config/stormy16/stormy16.h, config/v850/v850.h,
13920 config/visium/visium.h, config/vx-common.h, config/xtensa/elf.h: Don't
13921 define NO_IMPLICIT_EXTERN_C.
13922 * config/rs6000/aix.h: Set SYSTEM_IMPLICIT_EXTERN_C.
13924 2018-07-05 Tamar Christina <tamar.christina@arm.com>
13927 * config/arm/arm.c (arm_can_change_mode_class): Use GET_MODE_UNIT_SIZE
13928 instead of GET_MODE_SIZE when comparing Units.
13930 2018-07-05 Tamar Christina <tamar.christina@arm.com>
13933 * rtlanal.c (set_noop_p): Constrain on mode change,
13934 include hard-reg-set.h
13936 2018-07-05 Tamar Christina <tamar.christina@arm.com>
13938 * config/aarch64/aarch64.c (aarch64_expand_movmem): Fix mode size.
13940 2018-07-05 Jakub Jelinek <jakub@redhat.com>
13943 2018-07-04 Maxim Ostapenko <m.ostapenko@samsung.com>
13946 * config/gnu-user.h (LIBASAN_EARLY_SPEC): Pass -lstdc++ for static
13948 * gcc.c: Do not pass LIBUBSAN_SPEC if ASan is enabled with UBSan.
13950 2018-07-04 Maxim Ostapenko <m.ostapenko@samsung.com>
13953 * config/gnu-user.h (LIBASAN_EARLY_SPEC): Pass -lstdc++ for static
13955 * gcc.c: Do not pass LIBUBSAN_SPEC if ASan is enabled with UBSan.
13957 2018-07-04 Eric Botcazou <ebotcazou@adacore.com>
13959 PR middle-end/86380
13960 * expmed.c (choose_multiplier): Fix incorrect comparison with mask.
13962 2018-07-04 Aldy Hernandez <aldyh@redhat.com>
13964 * tree-vrp.c (extract_range_from_binary_expr_1): Initialize
13965 neg_*_op* variables.
13967 2018-07-04 Martin Liska <mliska@suse.cz>
13969 * tree-switch-conversion.c: Define
13970 max_ratio_for_speed and max_ratio_for_size constants.
13972 2018-07-04 Denys Vlasenko <dvlasenk@redhat.com>
13973 Martin Liska <mliska@suse.cz>
13975 PR middle-end/66240
13978 * common.opt: Rename align options with 'str_' prefix.
13979 * common/config/i386/i386-common.c (set_malign_value): New
13981 (ix86_handle_option): Use it to set -falign-* options/
13982 * config/aarch64/aarch64-protos.h (struct tune_params): Change
13983 type from int to string.
13984 * config/aarch64/aarch64.c: Update default values from int
13986 * config/alpha/alpha.c (alpha_override_options_after_change):
13988 * config/arm/arm.c (arm_override_options_after_change_1): Likewise.
13989 * config/i386/dragonfly.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
13990 max skip conditionally.
13991 * config/i386/freebsd.h (SUBALIGN_LOG): New.
13992 (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
13993 max skip conditionally.
13994 * config/i386/gas.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
13995 max skip conditionally.
13996 * config/i386/gnu-user.h (SUBALIGN_LOG): New.
13997 (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
13998 max skip conditionally.
13999 * config/i386/i386.c (struct ptt): Change type from int to
14001 (ix86_default_align): Set default values.
14002 * config/i386/i386.h (ASM_OUTPUT_MAX_SKIP_PAD): Print
14003 max skip conditionally.
14004 * config/i386/iamcu.h (SUBALIGN_LOG): New.
14005 (ASM_OUTPUT_MAX_SKIP_ALIGN):
14006 * config/i386/lynx.h (ASM_OUTPUT_MAX_SKIP_ALIGN):
14007 * config/i386/netbsd-elf.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
14008 max skip conditionally.
14009 * config/i386/openbsdelf.h (SUBALIGN_LOG): New.
14010 (ASM_OUTPUT_MAX_SKIP_ALIGN) Print max skip conditionally.:
14011 * config/i386/x86-64.h (SUBALIGN_LOG): New.
14012 (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
14013 max skip conditionally.
14014 (ASM_OUTPUT_MAX_SKIP_PAD): Likewise.
14015 * config/ia64/ia64.c (ia64_option_override): Set default values
14016 for alignment options.
14017 * config/m68k/m68k.c: Handle new str_align_* options.
14018 * config/mips/mips.c (mips_set_compression_mode): Change
14020 (mips_option_override): Set default values for options.
14021 * config/powerpcspe/powerpcspe.c (rs6000_option_override_internal):
14023 * config/rs6000/rs6000.c (rs6000_option_override_internal):
14025 * config/rx/rx.c (rx_option_override): Likewise.
14026 * config/rx/rx.h (JUMP_ALIGN): Use align_jumps_log.
14027 (LABEL_ALIGN): Use align_labels_log.
14028 (LOOP_ALIGN): Use align_loops_align.
14029 * config/s390/s390.c (s390_asm_output_function_label): Use new
14031 * config/sh/sh.c (sh_override_options_after_change):
14032 Change type of constants.
14033 * config/spu/spu.c (spu_sched_init): Likewise.
14034 * config/sparc/sparc.c (sparc_option_override): Set default
14035 values for options.
14036 * config/visium/visium.c (visium_option_override): Likewise.
14037 * config/visium/visium.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Do not
14038 emit p2align format with last argument if it's not needed.
14039 * doc/invoke.texi: Document extended format of -falign-*.
14040 * final.c: Use align_labels alignment.
14041 * flags.h (struct target_flag_state): Change type to use
14043 (struct align_flags_tuple): New.
14044 (struct align_flags): Likewise.
14045 (align_loops_log): Redefine macro to use new types.
14046 (align_loops_max_skip): Redefine macro to use new types.
14047 (align_jumps_log): Redefine macro to use new types.
14048 (align_jumps_max_skip): Redefine macro to use new types.
14049 (align_labels_log): Redefine macro to use new types.
14050 (align_labels_max_skip): Redefine macro to use new types.
14051 (align_functions_log): Redefine macro to use new types.
14052 (align_loops): Redefine macro to use new types.
14053 (align_jumps): Redefine macro to use new types.
14054 (align_labels): Redefine macro to use new types.
14055 (align_functions): Redefine macro to use new types.
14056 (align_functions_max_skip): Redefine macro to use new types.
14057 (align_loops_value): New macro.
14058 (align_jumps_value): New macro.
14059 (align_labels_value): New macro.
14060 (align_functions_value): New macro.
14061 * function.c (invoke_set_current_function_hook): Propagate
14062 alignment values from flags to global variables default in
14064 * ipa-icf.c (sem_function::equals_wpa): Use
14065 cl_optimization_option_eq instead of memcmp.
14066 * lto-streamer.h (cl_optimization_stream_out): Support streaming
14068 (cl_optimization_stream_in): Likewise.
14069 * optc-save-gen.awk: Support strings in cl_optimization.
14070 * opth-gen.awk: Likewise.
14071 * opts.c (finish_options): Remove error checking of invalid
14073 (MAX_CODE_ALIGN): Remove.
14074 (MAX_CODE_ALIGN_VALUE): Likewise.
14075 (parse_and_check_align_values): New function.
14076 (check_alignment_argument): Likewise.
14077 (common_handle_option): Use check_alignment_argument.
14078 * opts.h (parse_and_check_align_values): Declare.
14079 * toplev.c (init_alignments): Remove.
14080 (read_log_maxskip): New.
14081 (parse_N_M): Likewise.
14082 (parse_alignment_opts): Likewise.
14083 (backend_init_target): Remove usage of init_alignments.
14084 * toplev.h (parse_alignment_opts): Declare.
14085 * tree-streamer-in.c (streamer_read_tree_bitfields): Add new
14087 * tree-streamer-out.c (streamer_write_tree_bitfields): Likewise.
14088 * tree.c (cl_option_hasher::equal): New.
14089 * varasm.c: Use new global macros.
14091 2018-07-04 Denys Vlasenko <dvlasenk@redhat.com>
14093 * config/i386/dragonfly.h: (ASM_OUTPUT_MAX_SKIP_ALIGN):
14094 Use a simpler align directive also if MAXSKIP = ALIGN-1.
14095 * config/i386/gas.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
14096 * config/i386/lynx.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
14097 * config/i386/netbsd-elf.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
14098 * config/i386/i386.h (ASM_OUTPUT_MAX_SKIP_PAD): Likewise.
14099 * config/i386/freebsd.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Remove "If N
14100 is large, do at least 8 byte alignment" code. Add SUBALIGN_LOG
14101 define. Use a simpler align directive also if MAXSKIP = ALIGN-1.
14102 * config/i386/gnu-user.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
14103 * config/i386/iamcu.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
14104 * config/i386/openbsdelf.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
14105 * config/i386/x86-64.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
14107 2018-07-04 Martin Liska <mliska@suse.cz>
14108 Jonathan Wakely <jwakely@redhat.com>
14110 * coverage.c: Use correct type.
14111 * doc/invoke.texi: Language correction.
14113 2018-07-03 H.J. Lu <hongjiu.lu@intel.com>
14116 * config/i386/i386.c (rest_of_insert_endbranch): Also generate
14117 ENDBRANCH for non-tail call which may return via indirect branch.
14118 * doc/extend.texi: Document indirect_return attribute.
14120 2018-07-03 Martin Sebor <msebor@redhat.com>
14122 PR tree-optimization/86274
14123 * gimple-ssa-sprintf.c (fmtresult::type_max_digits): Verify
14125 (format_floating): Correct handling of infinities and NaNs.
14127 2018-07-03 Martin Sebor <msebor@redhat.com>
14129 * print-tree.c (print_real_cst): New function.
14130 (print_node_brief): Call it.
14131 (print_node): Ditto.
14133 2018-07-03 Jeff Law <law@redhat.com>
14135 * config/h8300/h8300.md (logical<mode>3_sn, logical<mode>3): Merge
14136 into a single pattern.
14138 * config/h8300/h8300.md (ors code_iterator): New.
14139 (bsetqi_msx, bnotqi_msx patterns and splitters): Consolidate into
14140 a single pattern and single splitter.
14141 (bsethi_msx, bnothi_msx patterns): Consolidate into a single pattern.
14142 (iorqi3_1, xorqi3_1): Likewise.
14143 (iorqi3, xorqi3 expanders): Similarly.
14145 * config/h8300/h8300.md (movmd_internal_normal): Consolidated with
14146 (movmd_internal) into a single pattern using the P mode iterator.
14147 (movmd splitters): Similarly.
14148 (stpcpy_internal_normal, stpcpy_internal): Similarly for thes patterns.
14149 (movsd splitters): Similarly.
14151 * config/h8300/h8300.c (h8300_insn_length_from_table): Consolidate
14152 ADDB, ADDW and ADDL into a single ADD attribute which selects the
14153 right table based on the size of the operand.
14154 * config/h8300/h8300.md (length_table): Corresponding changes. All
14155 references to "addb", "addw" and "addl" changed to "add".
14156 (btst patterns): Merge two variants into a single pattern.
14157 (tstqi, tsthi): Likewise.
14158 (addhi3_incdec, addsi3_incdec): Likewise.
14159 (subhi3_h8300hs, subsi3_h8300hs): Likewise.
14160 (mulhi3, mulsi3): Likewise.
14161 (udivhi3, udivsi3): Likewise.
14162 (divhi3, divsi3): Likewise.
14163 (andorqi3, andorhi3, andorsi3): Likewise.
14165 2018-07-03 Uros Bizjak <ubizjak@gmail.com>
14168 * config/i386/sse.md (uavg<mode>3_ceil): New expander.
14169 (<sse2_avx2>_uavg<mode>3<mask_name>): Simplify expander.
14171 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
14173 PR tree-optimization/85694
14174 * config/aarch64/iterators.md (HADD, RHADD): New int iterators.
14175 (u): Handle UNSPEC_SHADD, UNSPEC_UHADD, UNSPEC_SRHADD and
14177 * config/aarch64/aarch64-simd.md (<u>avg<mode>3_floor)
14178 (<u>avg<mode>3_ceil): New patterns.
14180 2018-07-03 David Malcolm <dmalcolm@redhat.com>
14182 * gcc.dg/vect/slp-perm-1.c: Remove "note: " prefix from
14183 scan-tree-dump directive.
14184 * gcc.dg/vect/slp-perm-2.c: Likewise.
14185 * gcc.dg/vect/slp-perm-3.c: Likewise.
14186 * gcc.dg/vect/slp-perm-5.c: Likewise.
14187 * gcc.dg/vect/slp-perm-6.c: Likewise.
14188 * gcc.dg/vect/slp-perm-7.c: Likewise.
14189 * gcc.dg/vect/slp-perm-8.c: Likewise.
14191 2018-07-03 Marek Polacek <polacek@redhat.com>
14193 PR middle-end/86202
14194 * gimple-fold.c (size_must_be_zero_p): Check the type of the size.
14196 2018-07-03 Richard Biener <rguenther@suse.de>
14199 * tree-ssa-structalias.c (find_func_clobbers): Properly
14200 handle indirect calls.
14202 2018-07-03 Jeff Law <law@redhat.com>
14204 * config/h8300/h8300.md (HSI, QHSI, QHSIF): New mode iterators.
14205 (shifts): New code iterator.
14206 (movqi, movhi, movsi, movsf expanders): Consolidate into a single
14207 expander. Fix HImode handling on H8/SX.
14208 (addqi3, addhi3, addsi3 expanders): Consolidate into a single expander.
14209 (subqi3, subhi3, subsi3 expanders): Likewise.
14210 (andqi3, andhi3, andsi3 expanders): Likewise.
14211 (iorqi3, iorhi3, iorsi3 expanders): Likewise.
14212 (xorqi3, xorhi3, xorsi3 expanders): Likewise.
14213 (negqi2, neghi2, negsi2, negsf2 expanders): Likewise.
14214 (one_cmplqi2, one_cmplhi2, one_cmplsi2): Likewise.
14215 (zero_extendqihi2, zero_extendqisi2): Likewise.
14216 (extendqihi2, extendqisi2): Likewise.
14217 (rotlqi3, rotlhi3, rotlsi3): Likewise.
14218 (neghi2_h8300, negsi2_h8300): Likewise for these patterns.
14219 (rotlqi3_1, rotlhi3_1): Likewise.
14220 (logicalhi3_sn, logicalsi3_sn): Likewise.
14221 (logicalhi3, logicalsi3): Likewise.
14223 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
14225 * tree-vect-patterns.c (vect_recog_rotate_pattern)
14226 (vect_recog_vector_vector_shift_pattern, vect_recog_divmod_pattern)
14227 (vect_recog_mixed_size_cond_pattern, adjust_bool_pattern_cast)
14228 (adjust_bool_pattern, vect_recog_bool_pattern): Pass the vector
14229 type to append_pattern_def_seq instead of creating a stmt_vec_info
14231 (build_mask_conversion): Likewise. Remove vinfo argument.
14232 (vect_add_conversion_to_patterm): Likewise, renaming to...
14233 (vect_add_conversion_to_pattern): ...this.
14234 (vect_recog_mask_conversion_pattern): Update call to
14235 build_mask_conversion. Pass the vector type to
14236 append_pattern_def_seq here too.
14237 (vect_recog_gather_scatter_pattern): Update call to
14238 vect_add_conversion_to_pattern.
14240 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
14242 * tree-vect-patterns.c (new_pattern_def_seq): Delete.
14243 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern)
14244 (vect_recog_widen_op_pattern, vect_recog_over_widening_pattern)
14245 (vect_recog_rotate_pattern, vect_synth_mult_by_constant): Don't set
14246 STMT_VINFO_PATTERN_DEF_SEQ to null here.
14247 (vect_recog_pow_pattern, vect_recog_vector_vector_shift_pattern)
14248 (vect_recog_mixed_size_cond_pattern, vect_recog_bool_pattern): Use
14249 append_pattern_def_seq instead of new_pattern_def_seq.
14250 (vect_recog_divmod_pattern): Do both of the above.
14251 (vect_pattern_recog_1): Assert that STMT_VINO_PATTERN_DEF_SEQ
14254 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
14256 * tree-vect-patterns.c (vect_recog_dot_prod_pattern):
14257 (vect_recog_sad_pattern, vect_recog_widen_op_pattern)
14258 (vect_recog_widen_mult_pattern, vect_recog_pow_pattern):
14259 (vect_recog_widen_sum_pattern, vect_recog_over_widening_pattern)
14260 (vect_recog_average_pattern, vect_recog_cast_forwprop_pattern)
14261 (vect_recog_widen_shift_pattern, vect_recog_rotate_pattern)
14262 (vect_recog_vector_vector_shift_pattern, vect_synth_mult_by_constant)
14263 (vect_recog_mult_pattern, vect_recog_divmod_pattern)
14264 (vect_recog_mixed_size_cond_pattern, vect_recog_bool_pattern)
14265 (vect_recog_mask_conversion_pattern): Replace vec<gimple *>
14266 parameter with a single stmt_vec_info.
14267 (vect_recog_func_ptr): Likewise.
14268 (vect_recog_gather_scatter_pattern): Likewise, folding in...
14269 (vect_try_gather_scatter_pattern): ...this.
14270 (vect_pattern_recog_1): Remove stmts_to_replace and just pass
14271 the stmt_vec_info of the statement to be matched. Don't clear
14272 STMT_VINFO_RELATED_STMT.
14273 (vect_pattern_recog): Update call accordingly.
14275 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
14277 PR tree-optimization/85694
14278 * doc/md.texi (avgM3_floor, uavgM3_floor, avgM3_ceil)
14279 (uavgM3_ceil): Document new optabs.
14280 * doc/sourcebuild.texi (vect_avg_qi): Document new target selector.
14281 * internal-fn.def (IFN_AVG_FLOOR, IFN_AVG_CEIL): New internal
14283 * optabs.def (savg_floor_optab, uavg_floor_optab, savg_ceil_optab)
14284 (savg_ceil_optab): New optabs.
14285 * tree-vect-patterns.c (vect_recog_average_pattern): New function.
14286 (vect_vect_recog_func_ptrs): Add it.
14287 * tree-vect-stmts.c (vectorizable_call): Get the type of the zero
14288 constant directly from the associated lhs.
14290 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
14292 * tree-vect-patterns.c (vect_split_statement): New function.
14293 (vect_convert_input): Use it to try to split an existing cast.
14295 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
14297 * poly-int.h (print_hex): New function.
14298 * dumpfile.h (dump_dec, dump_hex): Declare.
14299 * dumpfile.c (dump_dec, dump_hex): New poly_wide_int functions.
14300 * tree-vectorizer.h (_stmt_vec_info): Add min_output_precision,
14301 min_input_precision, operation_precision and operation_sign.
14302 * tree-vect-patterns.c (vect_get_range_info): New function.
14303 (vect_same_loop_or_bb_p, vect_single_imm_use)
14304 (vect_operation_fits_smaller_type): Delete.
14305 (vect_look_through_possible_promotion): Add an optional
14306 single_use_p parameter.
14307 (vect_recog_over_widening_pattern): Rewrite to use new
14308 stmt_vec_info infomration. Handle one operation at a time.
14309 (vect_recog_cast_forwprop_pattern, vect_narrowable_type_p)
14310 (vect_truncatable_operation_p, vect_set_operation_type)
14311 (vect_set_min_input_precision): New functions.
14312 (vect_determine_min_output_precision_1): Likewise.
14313 (vect_determine_min_output_precision): Likewise.
14314 (vect_determine_precisions_from_range): Likewise.
14315 (vect_determine_precisions_from_users): Likewise.
14316 (vect_determine_stmt_precisions, vect_determine_precisions): Likewise.
14317 (vect_vect_recog_func_ptrs): Put over_widening first.
14319 (vect_pattern_recog): Call vect_determine_precisions.
14321 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
14323 * tree-vect-patterns.c (vect_mark_pattern_stmts): Remove pattern
14324 statements that have been replaced by further pattern statements.
14325 (vect_pattern_recog_1): Clear STMT_VINFO_PATTERN_DEF_SEQ on failure.
14327 2018-07-03 Richard Biener <rguenther@suse.de>
14329 * tree-vect-stmts.c (vect_is_simple_use): Consolidate dumping,
14330 always set *dt. Dump vectype in vectype overload.
14331 * dumpfile.h (dump_gimple_expr): New function.
14332 (dump_gimple_expr_loc): Likewise.
14333 * dumpfile.c (dump_gimple_expr): New function.
14334 (dump_gimple_expr_loc): Likewise.
14336 2018-07-02 Jeff Law <law@redhat.com>
14338 * config/h8300/h8300.md (movqi_h8300, movqi_h8300hs): Consolidate
14339 the H8/300, H8/300H and H8/S variants into a single pattern.
14340 (movhi_h8300, movqi_h8300hs): Similarly.
14341 (pushqi_h8300hs, pushhi_h8300hs): Consolidate into a single pattern.
14342 (QHI mode iterator): New.
14344 * config/h8300/h8300.md: Remove trailing whitespace.
14346 2018-07-02 Jim Wilson <jimw@sifive.com>
14348 * config/riscv/riscv.c (riscv_expand_epilogue): Use emit_jump_insn
14349 instead of emit_insn for interrupt returns.
14350 * config/riscv/riscv.md (riscv_met): Add (return) to rtl.
14351 (riscv_sret, riscv_uret): Likewise.
14353 2018-07-02 David Malcolm <dmalcolm@redhat.com>
14355 * pretty-print.c (selftest::test_pp_format): Move save and restore
14356 of quotes to class auto_fix_quotes, and add an instance.
14357 * selftest.c: Include "intl.h".
14358 (selftest::auto_fix_quotes::auto_fix_quotes): New ctor.
14359 (selftest::auto_fix_quotes::~auto_fix_quotes): New dtor.
14360 * selftest.h (selftest::auto_fix_quotes): New class.
14362 2018-07-02 Richard Henderson <richard.henderson@linaro.org>
14364 * config/aarch64/aarch64-protos.h, config/aarch64/aarch64.c
14365 (aarch64_sve_prepare_conditional_op): Remove.
14366 * config/aarch64/aarch64-sve.md (cond_<SVE_INT_BINARY><SVE_I>):
14367 Allow aarch64_simd_reg_or_zero as select operand; remove
14368 the aarch64_sve_prepare_conditional_op call.
14369 (cond_<SVE_INT_BINARY_SD><SVE_SDI>): Likewise.
14370 (cond_<SVE_COND_FP_BINARY><SVE_F>): Likewise.
14371 (*cond_<SVE_INT_BINARY><SVE_I>_z): New pattern.
14372 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_z): New pattern.
14373 (*cond_<SVE_COND_FP_BINARY><SVE_F>_z): New pattern.
14374 (*cond_<SVE_INT_BINARY><SVE_I>_any): New pattern.
14375 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_any): New pattern.
14376 (*cond_<SVE_COND_FP_BINARY><SVE_F>_any): New pattern
14377 and a splitters to match all of the *_any patterns.
14378 * config/aarch64/predicates.md (aarch64_sve_any_binary_operator): New.
14380 * config/aarch64/iterators.md (SVE_INT_BINARY_REV): Remove.
14381 (SVE_COND_FP_BINARY_REV): Remove.
14382 (sve_int_op_rev, sve_fp_op_rev): New.
14383 * config/aarch64/aarch64-sve.md (*cond_<SVE_INT_BINARY><SVE_I>_0): New.
14384 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_0): New.
14385 (*cond_<SVE_COND_FP_BINARY><SVE_F>_0): New.
14386 (*cond_<SVE_INT_BINARY><SVE_I>_2): Rename, add movprfx alternative.
14387 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_2): Similarly.
14388 (*cond_<SVE_COND_FP_BINARY><SVE_F>_2): Similarly.
14389 (*cond_<SVE_INT_BINARY><SVE_I>_3): Similarly; use sve_int_op_rev.
14390 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_3): Similarly.
14391 (*cond_<SVE_COND_FP_BINARY><SVE_F>_3): Similarly; use sve_fp_op_rev.
14393 * config/aarch64/aarch64-sve.md (cond_<SVE_COND_FP_BINARY><SVE_F>):
14394 Remove match_dup 1 from the inner unspec.
14395 (*cond_<SVE_COND_FP_BINARY><SVE_F>): Likewise.
14397 * config/aarch64/aarch64.md (movprfx): New attr.
14398 (length): Default movprfx to 8.
14399 * config/aarch64/aarch64-sve.md (*mul<SVE_I>3): Add movprfx alt.
14400 (*madd<SVE_I>, *msub<SVE_I): Likewise.
14401 (*<su>mul<SVE_I>3_highpart): Likewise.
14402 (*<SVE_INT_BINARY_SD><SVE_SDI>3): Likewise.
14403 (*v<ASHIFT><SVE_I>3): Likewise.
14404 (*<su><MAXMIN><SVE_I>3): Likewise.
14405 (*<su><MAXMIN><SVE_F>3): Likewise.
14406 (*fma<SVE_F>4, *fnma<SVE_F>4): Likewise.
14407 (*fms<SVE_F>4, *fnms<SVE_F>4): Likewise.
14408 (*div<SVE_F>4): Likewise.
14410 2018-07-02 Richard Sandiford <richard.sandiford@arm.com>
14412 * tree-vect-patterns.c (vect_recog_widen_shift_pattern): Fix typo
14415 2018-07-02 Richard Biener <rguenther@suse.de>
14417 PR tree-optimization/86363
14418 * tree-ssa-sccvn.c (vn_reference_lookup_3): Check the
14419 memset argument refers to a non-variable address.
14421 2018-07-02 Aldy Hernandez <aldyh@redhat.com>
14423 * tree-vrp.c (extract_range_from_binary_expr_1): Abstract a lot of the
14424 {PLUS,MINUS}_EXPR code to...
14425 (adjust_symbolic_bound): ...here,
14426 (combine_bound): ...here,
14427 (set_value_range_with_overflow): ...and here.
14429 2018-07-02 Aldy Hernandez <aldyh@redhat.com>
14431 * tree-vrp.c (extract_range_from_unary_expr): Abstract ABS_EXPR
14433 (extract_range_from_abs_expr): ...here.
14435 2018-07-02 Eric Botcazou <ebotcazou@adacore.com>
14437 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Do not overrule
14438 -fno-omit-frame-pointer when not optimizing.
14440 2018-07-02 Martin Liska <mliska@suse.cz>
14443 * ipa-pure-const.c (malloc_candidate_p): Revert usage of ::get.
14444 (propagate_nothrow): Likewise.
14446 2018-07-02 Martin Liska <mliska@suse.cz>
14449 * ipa-inline.c (early_inliner): Revert wrongly added ::get call.
14451 2018-07-02 David Malcolm <dmalcolm@redhat.com>
14453 * dumpfile.c (dump_generic_expr_loc): Undo removal of this
14454 function in r262149, changing "loc" param from source_location to
14455 const dump_location_t &.
14456 * dumpfile.h (dump_generic_expr_loc): Undo removal of this
14457 declaration, as above.
14459 2018-07-01 Paul Koning <ni1d@arrl.net>
14461 * common/config/pdp11/pdp11-common.c (pdp11_handle_option): Handle
14462 -munit-asm, -mgnu-asm, -mdec-asm.
14463 * config/pdp11/pdp11-protos.h (pdp11_gen_int_label): New.
14464 (pdp11_output_labelref): New.
14465 (pdp11_output_def): New.
14466 (pdp11_output_addr_vec_elt): New.
14467 * config/pdp11/pdp11.c: Use tab between opcode and operands. Use
14468 %# and %@ format codes.
14469 (pdp11_option_override): New.
14470 (TARGET_ASM_FILE_START_FILE_DIRECTIVE): Define.
14471 (pdp11_output_ident): New.
14472 (pdp11_asm_named_section): New.
14473 (pdp11_asm_init_sections): New.
14474 (pdp11_file_start): New.
14475 (pdp11_file_end): New.
14476 (output_ascii): Use .ascii/.asciz for -mdec-asm.
14477 (pdp11_asm_print_operand): Update %# and %$ for -mdec-asm. Add
14478 %o, like %c but octal.
14479 (pdp11_option_override): New.
14480 * config/pdp11/pdp11.h (TEXT_SECTION_ASM_OP): Update for
14482 (DATA_SECTION_ASM_OP): Ditto.
14483 (READONLY_DATA_SECTION_ASM_OP): New.
14484 (IS_ASM_LOGICAL_LINE_SEPARATOR): New.
14485 (ASM_GENERATE_INTERNAL_LABEL): Use new function.
14486 (ASM_OUTPUT_LABELREF): Ditto.
14487 (ASM_OUTPUT_DEF): Ditto.
14488 (ASM_OUTPUT_EXTERNAL): New.
14489 (ASM_OUTPUT_SOURCE_FILENAME): New.
14490 (ASM_OUTPUT_ADDR_VEC_ELT): Use new function.
14491 (ASM_OUTPUT_SKIP): Update for -mdec-asm.
14492 * config/pdp11/pdp11.md: Use tab between opcode and operands. Use
14493 %# and %@ format codes.
14494 * config/pdp11/pdp11.opt (mgnu-asm): New.
14495 (mdec-asm): Conflicts with -mgnu-asm and -munix-asm.
14496 (munix-asm): Conflicts with -mdec-asm and -mgnu-asm.
14497 * doc/invoke.txt (PDP-11 Options): Add -mgnu-asm.
14499 2018-07-01 Aldy Hernandez <aldyh@redhat.com>
14501 * tree-ssa-threadupdate.c (mark_threaded_blocks): Avoid
14502 dereferencing path[] beyond its length.
14504 (debug_all_paths): New.
14505 (rewire_first_differing_edge): New.
14506 (adjust_paths_after_duplication): New.
14507 (duplicate_thread_path): Call adjust_paths_after_duplication.
14509 (thread_through_all_blocks): Add new argument to
14510 duplicate_thread_path.
14512 2018-06-30 Jim Wilson <jimw@sifive.com>
14514 * config/riscv/predicates.md (p2m1_shift_operand): New.
14515 (high_mask_shift_operand): New.
14516 * config/riscv/riscv.md (lshrsi3_zero_extend_3+1): New combiner
14517 pattern using p2m1_shift_operand.
14518 (lshsi3_zero_extend_3+2): New combiner pattern using
14519 high_mask_shift_operand.
14521 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
14523 * tree-vect-patterns.c (vect_get_external_def_edge): New function,
14525 (vect_recog_rotate_pattern): ...here.
14526 (vect_convert_input): Try to insert casts of invariants in the
14528 * tree-vect-loop-manip.c (vect_loop_versioning): Don't require the
14529 preheader to be empty.
14531 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
14533 * tree-vect-patterns.c (append_pattern_def_seq): Take an optional
14534 vector type. If given, install it in the new statement's
14535 STMT_VINFO_VECTYPE.
14536 (vect_element_precision): New function.
14537 (vect_unpromoted_value): New struct.
14538 (vect_unpromoted_value::vect_unpromoted_value): New function.
14539 (vect_unpromoted_value::set_op): Likewise.
14540 (vect_look_through_possible_promotion): Likewise.
14541 (vect_joust_widened_integer, vect_joust_widened_type): Likewise.
14542 (vect_widened_op_tree, vect_convert_input): Likewise.
14543 (vect_convert_inputs, vect_convert_output): Likewise.
14544 (vect_recog_dot_prod_pattern): Use vect_look_through_possible_promotion
14545 to handle the optional cast of the multiplication result and
14546 vect_widened_op_tree to detect the widened multiplication itself.
14547 Do not require the input and output of promotion casts to have
14548 the same sign, but base the signedness of the operation on the
14549 input rather than the result. If the pattern includes two
14550 promotions, check that those promotions have the same sign.
14551 Do not restrict the MULT_EXPR handling to a double-width result;
14552 handle quadruple-width results and wider. Use vect_convert_inputs
14553 to convert the inputs to the common type.
14554 (vect_recog_sad_pattern): Use vect_look_through_possible_promotion
14555 to handle the optional cast of the ABS result. Also allow a sign
14556 change or a sign extension between the ABS and MINUS.
14557 Use vect_widened_op_tree to detect the widened subtraction and use
14558 vect_convert_inputs to convert the inputs to the common type.
14559 (vect_handle_widen_op_by_const): Delete.
14560 (vect_recog_widen_op_pattern): New function.
14561 (vect_recog_widen_mult_pattern): Use it.
14562 (vect_recog_widen_shift_pattern): Likewise.
14563 (vect_recog_widen_sum_pattern): Use
14564 vect_look_through_possible_promotion to handle the promoted
14567 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
14569 * gimple-iterator.c (gsi_for_stmt): Add a new overload that takes
14570 the containing gimple_seq *.
14571 * gimple-iterator.h (gsi_for_stmt): Declare it.
14572 * tree-vect-patterns.c (vect_recog_dot_prod_pattern)
14573 (vect_recog_sad_pattern, vect_recog_widen_sum_pattern)
14574 (vect_recog_widen_shift_pattern, vect_recog_rotate_pattern)
14575 (vect_recog_vector_vector_shift_pattern, vect_recog_divmod_pattern)
14576 (vect_recog_mask_conversion_pattern): Remove STMT_VINFO_IN_PATTERN_P
14578 (vect_init_pattern_stmt, vect_set_pattern_stmt): New functions,
14580 (vect_mark_pattern_stmts): ...here. Handle cases in which the
14581 statement being replaced is part of an existing pattern
14582 definition sequence, inserting the new pattern statements before
14584 (vect_pattern_recog_1): Don't return a bool. If the statement
14585 is already part of a pattern, instead apply pattern matching
14586 to the pattern definition statements. Don't clear the
14587 STMT_VINFO_RELATED_STMT if is_pattern_stmt_p.
14588 (vect_pattern_recog): Don't break after the first match;
14589 continue processing the pattern definition statements instead.
14590 Don't bail out for STMT_VINFO_IN_PATTERN_P here.
14592 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
14594 * tree-vect-patterns.c (vect_reassociating_reduction_p): New function.
14595 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern)
14596 (vect_recog_widen_sum_pattern): Use it.
14598 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
14600 * tree-vect-loop.c (vectorizable_reduction): Assert that the
14601 phi is not a pattern statement and has not been replaced by
14602 a pattern statement.
14603 * tree-vect-patterns.c (type_conversion_p): Don't check
14604 STMT_VINFO_IN_PATTERN_P.
14605 (vect_recog_vector_vector_shift_pattern): Likewise.
14606 (vect_recog_dot_prod_pattern): Expect vect_is_simple_use to return
14607 the pattern statement rather than the original statement; check
14608 directly for a WIDEN_MULT_EXPR here.
14609 * tree-vect-slp.c (vect_get_and_check_slp_defs): Expect
14610 vect_is_simple_use to return the pattern statement rather
14611 than the original statement; use is_pattern_stmt_p to check
14612 for such a pattern statement.
14613 * tree-vect-stmts.c (process_use): Expect vect_is_simple_use
14614 to return the pattern statement rather than the original statement;
14615 don't do the same transformation here.
14616 (vect_is_simple_use): If the defining statement has been replaced
14617 by a pattern statement, return the pattern statement instead.
14618 Remove the corresponding (local) transformation from the vectype
14621 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
14623 * tree-vectorizer.h (vect_is_simple_use): Move the gimple ** to the
14624 end and default to null.
14625 * tree-vect-loop.c (vect_create_epilog_for_reduction)
14626 (vectorizable_reduction): Update calls accordingly, dropping the
14627 gimple ** argument if the passed-back statement isn't needed.
14628 * tree-vect-patterns.c (vect_get_internal_def, type_conversion_p)
14629 (vect_recog_rotate_pattern): Likewise.
14630 (vect_recog_mask_conversion_pattern): Likewise.
14631 * tree-vect-slp.c (vect_get_and_check_slp_defs): Likewise.
14632 (vect_mask_constant_operand_p): Likewise.
14633 * tree-vect-stmts.c (is_simple_and_all_uses_invariant, process_use):
14634 (vect_model_simple_cost, vect_get_vec_def_for_operand): Likewise.
14635 (get_group_load_store_type, get_load_store_type): Likewise.
14636 (vect_check_load_store_mask, vect_check_store_rhs): Likewise.
14637 (vectorizable_call, vectorizable_simd_clone_call): Likewise.
14638 (vectorizable_conversion, vectorizable_assignment): Likewise.
14639 (vectorizable_shift, vectorizable_operation): Likewise.
14640 (vectorizable_store, vect_is_simple_cond): Likewise.
14641 (vectorizable_condition, vectorizable_comparison): Likewise.
14642 (get_same_sized_vectype, vect_get_mask_type_for_stmt): Likewise.
14643 (vect_is_simple_use): Rename the def_stmt argument to def_stmt_out
14644 and move it to the end. Cope with null def_stmt_outs.
14646 2018-06-30 Bernd Edlinger <bernd.edlinger@hotmail.de>
14648 * Makefile.in (FLAGS_TO_PASS): Add RANLIB_FOR_TARGET.
14650 2018-06-29 Jeff Law <law@redhat.com>
14652 * config/v850/v850.c (v850_legitimate_address_p): Handle large
14653 displacements for TARGET_V850E2V3 and newer.
14654 (TARGET_LRA_P): Remove. Defaults to LRA now.
14655 * config/v850/v850.md (sign23byte_load): Remove.
14656 (unsign23byte_load, sign23hword_load, unsign23hword_load): Likewise.
14657 (23word_load, 23byte_store, 23hword_store, 23word_store): Likewise.
14659 2018-06-29 Martin Liska <mliska@suse.cz>
14662 * coverage.c (coverage_init): Mangle full path name.
14663 * doc/invoke.texi: Document the change.
14664 * gcov-io.c (mangle_path): New.
14665 * gcov-io.h (mangle_path): Likewise.
14666 * gcov.c (mangle_name): Use mangle_path for path mangling.
14668 2018-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14670 * config/arm/arm.c (output_move_double): Don't allow STRD instructions
14671 if starting source register is not even.
14673 2018-06-29 Martin Liska <mliska@suse.cz>
14675 PR tree-optimization/86263
14676 * tree-switch-conversion.c (switch_decision_tree::try_switch_expansion):
14677 Make edge redirection.
14679 2018-06-29 David Malcolm <dmalcolm@redhat.com>
14681 * dumpfile.c (dump_loc): Add indentation based on scope depth.
14682 (dump_scope_depth): New variable.
14683 (get_dump_scope_depth): New function.
14684 (dump_begin_scope): New function.
14685 (dump_end_scope): New function.
14686 * dumpfile.h (get_dump_scope_depth): New declaration.
14687 (dump_begin_scope): New declaration.
14688 (dump_end_scope): New declaration.
14689 (class auto_dump_scope): New class.
14690 (AUTO_DUMP_SCOPE): New macro.
14691 * tree-vectorizer.h (DUMP_VECT_SCOPE): Reimplement in terms of
14694 2018-06-29 Richard Biener <rguenther@suse.de>
14696 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Assert
14697 compute_all_dependences succeeds.
14698 * tree-vect-loop.c (vect_get_datarefs_in_loop): Fail early if we
14699 exceed --param loop-max-datarefs-for-datadeps.
14701 2018-06-29 Jakub Jelinek <jakub@redhat.com>
14703 * config/rs6000/t-rs6000: Append rs6000-modes.h to TM_H.
14705 2018-06-28 Uros Bizjak <ubizjak@gmail.com>
14708 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): Use
14709 alternative 0 in preferred_for_speed attribute.
14711 2018-06-28 Paul Koning <ni1d@arrl.net>
14713 * config/pdp11/pdp11-protos.h (pdp11_shift_length): New function.
14714 * config/pdp11/pdp11.c (pdp11_shift_length): New function.
14715 * config/pdp11/pdp11.h (ADJUST_INSN_LENGTH): Remove.
14716 * config/pdp11/pdp11.md: Correct "length" attribute calculation
14717 for shift insn patterns.
14719 2018-06-28 David Malcolm <dmalcolm@redhat.com>
14721 * cgraph.c (cgraph_node::get_body): Replace assignments to
14722 "dump_file" with calls to set_dump_file.
14723 * dumpfile.c (alt_dump_file): Make static, and group with...
14724 (alt_flags): ...this definition.
14725 (dumps_are_enabled): New variable.
14726 (refresh_dumps_are_enabled): New function.
14727 (set_dump_file): New function.
14728 (set_alt_dump_file): New function.
14729 (gcc::dump_manager::dump_start): Replace assignments to
14730 "dump_file" and "alt_dump_file" with calls to set_dump_file and
14732 (gcc::dump_manager::dump_finish): Likewise.
14733 * dumpfile.h (alt_dump_file): Delete decl.
14734 (dumps_are_enabled): New variable decl.
14735 (set_dump_file): New function decl.
14736 (dump_enabled_p): Rewrite in terms of new "dumps_are_enabled"
14738 * tree-nested.c (lower_nested_functions): Replace assignments to
14739 "dump_file" with calls to set_dump_file.
14741 2018-06-28 Eric Botcazou <ebotcazou@adacore.com>
14743 * tree-cfg.c (verify_gimple_in_cfg): Call verify_location on the
14744 goto_locus of each outgoing edge of each basic block.
14746 2018-06-28 Richard Biener <rguenther@suse.de>
14748 * dwarf2out.c (decl_scope_table): Remove.
14749 (push_decl_scope): Likewise.
14750 (pop_decl_scope): Likewise.
14751 (gen_type_die_for_member): Do not call push/pop_decl_scope.
14752 (gen_struct_or_union_type_die): Likewise.
14753 (gen_tagged_type_die): Likewise.
14754 (dwarf2out_init): Do not initialize decl_scope_table.
14755 (dwarf2out_c_finalize): Do not free it.
14757 2018-06-28 Richard Biener <rguenther@suse.de>
14759 * dwarf2out.c (gen_subprogram_die): Use is_unit_die when
14760 deciding whether to not re-use a DIE.
14762 2018-06-28 Richard Biener <rguenther@suse.de>
14764 * dwarf2out.c (gen_subprogram_die): Always re-use DIEs with an
14765 DW_AT_abstract_origin attribute.
14767 2018-06-28 Martin Liska <mliska@suse.cz>
14769 * tree-switch-conversion.c (jump_table_cluster::can_be_handled):
14770 Use newly introduced constants.
14771 * tree-switch-conversion.h (struct jump_table_cluster):
14772 Define max_ratio_for_size and max_ratio_for_speed.
14774 2018-06-28 Martin Liska <mliska@suse.cz>
14776 * tree-switch-conversion.c (jump_table_cluster::find_jump_tables):
14777 Add new checking assert to catch invalid state.
14778 (jump_table_cluster::can_be_handled): Handle single case
14780 (jump_table_cluster::is_beneficial): Bail out for such case.
14781 (bit_test_cluster::find_bit_tests):
14782 Add new checking assert to catch invalid state.
14783 (bit_test_cluster::can_be_handled): Handle single case
14785 (bit_test_cluster::is_beneficial): Bail out for such case.
14786 (switch_decision_tree::analyze_switch_statement):
14789 2018-06-28 Martin Liska <mliska@suse.cz>
14791 * common.opt: Introduce -completion option.
14792 * gcc.c (driver_handle_option): Handle it.
14793 (driver::main): Print completions if completion
14795 * opt-suggestions.c (option_proposer::get_completions):
14797 (option_proposer::suggest_completion): Likewise.
14798 (option_proposer::find_param_completions): Likewise.
14799 (verify_autocompletions): Likewise.
14800 (test_completion_valid_options): Likewise.
14801 (test_completion_valid_params): Likewise.
14802 (in_completion_p): Likewise.
14803 (empty_completion_p): Likewise.
14804 (test_completion_partial_match): Likewise.
14805 (test_completion_garbage): Likewise.
14806 (opt_proposer_c_tests): Likewise.
14807 * opt-suggestions.h: Declare new functions.
14808 * opts.c (common_handle_option): Handle OPT__completion_.
14809 * selftest-run-tests.c (selftest::run_tests): Add
14810 opt_proposer_c_tests.
14811 * selftest.c (assert_str_startswith): New.
14812 * selftest.h (assert_str_startswith): Likewise.
14813 (opt_proposer_c_tests): New.
14814 (ASSERT_STR_STARTSWITH): Likewise.
14816 2018-06-28 Martin Liska <mliska@suse.cz>
14818 * Makefile.in: Add opt-suggestions.o.
14819 * gcc-main.c: Include opt-suggestions.h.
14820 * gcc.c (driver::driver): Likewise.
14821 (driver::~driver): Remove m_option_suggestions.
14822 (driver::build_option_suggestions): Moved to option_proposer.
14823 (driver::suggest_option): Likewise.
14824 (driver::handle_unrecognized_options): Use option_proposer.
14825 * gcc.h (class driver): Add new memver m_option_proposer.
14826 * opt-suggestions.c: New file.
14827 * opt-suggestions.h: New file.
14829 2018-06-28 Martin Liska <mliska@suse.cz>
14831 * vec.h (class auto_string_vec): New (moved from auto_argvec).
14832 (auto_string_vec::~auto_string_vec): Likewise.
14834 2018-06-28 Eric Botcazou <ebotcazou@adacore.com>
14836 * tree-inline.h (struct copy_body_data): Move remapping_type_depth and
14837 prevent_decl_creation_for_types fields up and add reset_location field.
14838 * tree-inline.c (remap_gimple_stmt): Force input_location on the new
14839 statement if id->reset_location is true.
14840 (copy_edges_for_bb): Do not set goto_locus on the new edges if
14841 id->reset_location is true.
14842 (copy_phis_for_bb): Force input_location on the arguments if
14843 id->reset_location is true.
14844 (expand_call_inline): Set id->reset_location if DECL_IGNORED_P
14845 is set on the function to be inlined.
14847 2018-06-27 Stephan Bergmann <sbergman@redhat.com>
14849 * doc/invoke.texi (Debugging Options): Add -gsplit-dwarf.
14851 2018-06-27 Dimitar Dimitrov <dimitar@dinux.eu>
14853 * lra-eliminations.c (update_reg_eliminate): Mark all spanning hard
14854 registers for Pmode.
14855 * lra-lives.c (check_pseudos_live_through_calls): Mark all spanning
14856 hard registers for the clobbered pseudo.
14858 2018-06-27 Paul Koning <ni1d@arrl.net>
14860 * common/config/pdp11/pdp11-common.c (pdp11_handle_option): Handle
14861 mutually exclusive options.
14862 * config/pdp11/constraints.md (h): New constraint.
14863 (O): Update definition to match shift code generation.
14864 (D): New constraint.
14865 * config/pdp11/pdp11-modes.def (CCNZ): Define mode.
14867 * config/pdp11/pdp11-protos.h (int_no_side_effect_operand): New
14869 (output_jump): Change arguments.
14870 (pdp11_fixed_cc_regs): New function.
14871 (pdp11_cc_mode): Ditto.
14872 (pdp11_expand_shift): Ditto.
14873 (pdp11_assemble_shift): Ditto.
14874 (pdp11_small_shift): Ditto.
14875 (pdp11_branch_cost): Remove.
14876 * config/pdp11/pdp11.c (pdp11_assemble_integer): Remove comments
14878 (pdp11_register_move_cost): Update for CC registers.
14879 (pdp11_rtx_costs): Add case for LSHIFTRT.
14880 (pdp11_output_jump): Add CCNZ mode conditional branches.
14881 (notice_update_cc_on_set): Remove.
14882 (pdp11_cc_mode): New function.
14883 (simple_memory_operand): Correct pre/post decrement case.
14884 (no_side_effect_operand): New function.
14885 (pdp11_regno_reg_class): Add CC_REGS class.
14886 (pdp11_fixed_cc_regs): New function.
14887 (pdp11_small_shift): New function.
14888 (pdp11_expand_shift): New function to expand shift insns.
14889 (pdp11_assemble_shift): New function to output shifts.
14890 (pdp11_branch_cost): Remove.
14891 (pdp11_modes_tieable_p): Make QI/HI modes tieable.
14892 * config/pdp11/pdp11.h (SIZE_TYPE): Ensure 16-bit type.
14893 (WCHAR_TYPE): Ditto.
14894 (PTRDIFF_TYPE): Ditto.
14895 (ADJUST_INSN_LENGTH): New macro.
14896 (FIXED_REGISTERS): Add CC registers.
14897 (CALL_USED_REGISTERS): Ditto.
14898 (reg_class): Ditto.
14899 (REG_CLASS_NAMES): Ditto.
14900 (REG_CLASS_CONTENTS): Ditto.
14901 (SELECT_CC_MODE): Use new function.
14902 (TARGET_FLAGS_REGNUM): New macro.
14903 (TARGET_FIXED_CONDITION_CODE_REGS): Ditto.
14904 (cc0_reg_rtx): Remove.
14905 (CC_STATUS_MDEP): Remove.
14906 (CC_STATUS_MDEFP_INIT): Remove.
14907 (CC_IN_FPU): Remove.
14908 (NOTICE_UPDATE_CC): Remove.
14909 (REGISTER_NAMES): Add CC registers.
14910 (BRANCH_COST): Change to constant 1.
14911 * config/pdp11/pdp11.md: Rewrite for CCmode condition code
14913 * config/pdp11/pdp11.opt (mbcopy): Remove.
14914 (mbcopy-builtin): Remove.
14915 (mbranch-cheap): Remove.
14916 (mbranch-expensive): Remove.
14917 * config/pdp11/predicates.md (expand_shift_operand): Update to
14918 match shift code generation.
14919 (ccnz_operator): New predicate.
14920 * doc/invoke.texi (PDP-11 Options): Remove deleted options
14921 -mbcopy, -mbcopy-builtin, -mbranch-cheap, -mbranch-expensive.
14922 Remove non-existent option -mabshi, -mno-abshi. Document mutually
14924 * doc/md.texi (PDP-11): Document new D and h constraints. Update
14925 description of O constraint.
14927 2018-06-27 Jeff Law <law@redhat.com>
14928 Austin Law <austinklaw@gmail.com>
14930 * config/v850/v850.md (addsi3_set_flags): New pattern.
14931 (subsi3_set_flags, negsi2_set_flags, andsi3_set_flags): Likewise.
14932 (iorsi3_set_flags, xorsi3_set_flags, one_cmplsi2_set_flags): Likewise.
14933 (zero_extendhisi2_v850_set_flags): Likewise.
14934 (zero_extendqisi2_v850_set_flags): Likewise.
14935 (ashlsi3_set_flags, ashlsi3_v850e2_set_flags): Likewise.
14936 (lshrsi3_set_flags, lshrsi3_v850e2_set_flags): Likewise.
14937 (ashrsi3_set_flags, ashrsi3_v850e2_set_flags): Likewise.
14939 * config/v850/v850-protos.h (notice_update_cc): Remove.
14940 * config/v850/v850.c (v850_compare_op0, v850_compare_op1): Remove.
14941 (v850_print_operand): Handle 'D' and "d".
14942 (v850_select_cc_mode): Remove ATTRIBUTE_UNUSED for last argument.
14943 Add handling of arithmetic/logical operations compared against zero.
14944 (v850_gen_float_compare): Remove ATTRIBUTE_UNUSED for last argument.
14945 Do not look at v850_compare_op, instead get mode from last argument.
14946 (v850_gen_compare): Remove
14947 (increment_stack): Use addsi3_clobber_flags to avoid splitting failure
14948 after reload for prologue insns.
14949 (expand_prologue): Account for CLOBBER of CC_REGNUM in various
14951 (construct_save_jarl): Likewise.
14952 (TARGET_FLAGS_REGNUM): Define.
14953 * config/v850/v850.h (v850_compare_op0, v850_compare_op1): Remove.
14954 (NOTICE_UPDATE_CC): Remove.
14955 * config/v850/v850.md (v850_tst1): Use (reg:CCZ CC_REGNUM) rather
14956 than cc0. Conditionalize on reload_completed.
14957 (cmpsi_insn, setfcc_insn): Likewise.
14958 (tst1 splitter): Turn into define_and_split which sets the flags
14960 (cstoresi4, cbranchsf4, cbranchdf4, cbranchsi4_insn): Likewise.
14961 (cbranchsi4, branch_normal, branch_invert): Do not expose cc0 here.
14962 (cstoresf4, cstoredf4): Clobber the flags.
14963 (cmpsi, cmpsf, cmpdf): Remove expanders.
14964 (setf_insn): Remove pattern.
14965 (addsi3): Turn into define_and_split which clobbers the flags after
14966 reload and a suitable pattern (addsi3_clobber_flags) for use after
14968 (subsi3, negsi2, andsi3, iorsi3, xorsi3, one_cmplsi2) Likewise.
14969 (ashlsi3, ashlsi3_v850e2, lshrsi3, lsh4si3_v850e2): Likewise.
14970 (ashrsi3, ashrsi3_v850e2): Likewise.
14971 (bins): Clobber the flags.
14972 (movsicc_normal_cc, movsicc_normal, movsicc_tst1): Likewise.
14973 (movsicc_tst1_revesed, sasf, swap and rotate patterns): Likewise.
14974 (fix_loop_counter, call_internal_short, call_internal_long): Likewise.
14975 (call_value_internal_short, call_value_internal_long): Likewise.
14976 (callt_save_interrupt, callt_return_interrupt): Likewise.
14977 (save_interrupt, return_interrupt): Likewise.
14978 (callt_save_all_interrupt, save_all_interrupt): Likewise.
14979 (_save_all_interrupt, callt_restore_all_interrupt): Likewise.
14980 (restore_all_interrupt, _restore_all_interrupt): Likewise.
14981 (All FP comparisons): Only allow after reload has completed.
14983 (divh, divhu): Tweak output template.
14984 (branch_z_normal, branch_z_invert): Remove
14985 (branch_nz_normal, branch_nz_invert): Likewise.
14986 (extendhisi_insn, extendqisi_insn): Do not clobber flags.
14988 * config/v850/v850-modes.def (CCZ, CCNZ): Add new modes.
14989 * config/v850/v850.c (notice_update_cc): Remove.
14990 * config/v850/v850.h (CC_OVERFLOW_UNUSABLE): Remove
14991 (CC_NO_CARRY): Likewise.
14992 (NOTICE_UPDATE_CC): Define to nothing.
14993 * config/v850/v850.md: Remove block comment on cc0 handling
14994 Remove "cc" attribute from all patterns. Remove cc_status handling
14995 from all patterns. Minor formatting fixes.
14997 2018-06-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14999 * config/aarch64/aarch64-cores.def (cortex-a76): New entry.
15000 (cortex-a76.cortex-a55): Likewise.
15001 * config/aarch64/aarch64-tune.md: Regenerate.
15002 * doc/invoke.texi (AArch64 Options): Document cortex-a76 and
15003 cortex-a76.cortex-a55.
15005 2018-06-27 Jeff Law <law@redhat.com>
15007 * config/v850/t-v850 (MULTILIB_OPTIONS): Remove 8byte-align.
15008 (MULTILIB_DIRNAMES): Similarly.
15010 2018-06-27 Eric Botcazou <ebotcazou@adacore.com>
15012 * gimple.h (gimple_return_retbnd): Delete.
15013 (gimple_return_set_retbnd): Likewise.
15014 * cgraphunit.c (cgraph_node::expand_thunk): Remove call to
15015 gimple_return_set_retbnd.
15016 * gimple-pretty-print.c (dump_gimple_return): Remove call to
15017 gimple_return_retbnd and adjust.
15018 * tree-inline.h (struct copy_body_data): Remove retbnd field.
15019 * tree-inline.c (remap_gimple_stmt): Remove handling of retbnd.
15020 Explicitly return NULL in a couple more cases. Move assertion
15021 on debug statements and remove unreachable code.
15022 (reset_debug_binding): Do not test id->retbnd.
15023 (expand_call_inline): Do not set it.
15025 2018-06-27 Rasmus Villemoes <rv@rasmusvillemoes.dk>
15027 * configure.ac: Add --disable-gcov option.
15028 * configure: Regenerate.
15029 * Makefile.in: Honour @enable_gcov@.
15030 * doc/install.texi: Document --disable-gcov.
15032 2018-06-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15034 * config/arm/arm-cpus.in (cortex-a76): New entry.
15035 (cortex-a76.cortex-a55): Likewise.
15036 * config/arm/arm-tables.opt: Regenerate.
15037 * config/arm/arm-tune.md: Likewise.
15038 * config/arm/driver-arm.c (arm_cpu_table): Add Cortex-A76 entry.
15039 * doc/invoke.texi (ARM Options): Document cortex-a76 and
15040 cortex-a76.cortex-a55.
15042 2018-06-27 Tamar Christina <tamar.christina@arm.com>
15045 * config/aarch64/aarch64.md (*movhf_aarch64): Add dup v0.4h pattern.
15047 2018-06-27 Siddhesh Poyarekar <siddhesh@sourceware.org>
15049 * config/aarch64/aarch64.h (CALL_USE_REGISTERS): Fix obsolete
15051 (EPILOGUE_USES): Likewise.
15053 2018-06-26 Eric Botcazou <ebotcazou@adacore.com>
15055 * tree-inline.c (remap_location): New function extracted from...
15056 (copy_edges_for_bb): Add ID parameter. Remap goto_locus.
15057 (copy_phis_for_bb): ...here. Call remap_location.
15058 (copy_cfg_body): Adjust call to copy_edges_for_bb.
15060 2018-06-26 Aaron Sawdey <acsawdey@linux.ibm.com>
15062 * config/rs6000/rs6000-string.c (expand_block_clear): Don't use
15063 unaligned vsx for 16B memset.
15065 2018-06-26 Segher Boessenkool <segher@kernel.crashing.org>
15068 * config/rs6000/rs6000.c (rs6000_init_builtins): Do not set
15069 ieee128_float_type_node to long_double_type_node unless
15070 TARGET_LONG_DOUBLE_128 is set.
15072 2018-06-26 David Malcolm <dmalcolm@redhat.com>
15074 * cfgloop.c (get_loop_location): Convert return type from
15075 location_t to dump_user_location_t, replacing INSN_LOCATION lookups
15076 by implicit construction from rtx_insn *, and using
15077 dump_user_location_t::from_function_decl for the fallback case.
15078 * cfgloop.h (get_loop_location): Convert return type from
15079 location_t to dump_user_location_t.
15080 * cgraphunit.c (walk_polymorphic_call_targets): Update call to
15081 dump_printf_loc to pass in a dump_location_t rather than a
15082 location_t, via the gimple stmt.
15083 * coverage.c (get_coverage_counts): Update calls to
15084 dump_printf_loc to pass in dump_location_t rather than a
15086 * doc/optinfo.texi (Dump types): Convert example of
15087 dump_printf_loc from taking "locus" to taking "insn". Update
15088 description of the "_loc" calls to cover dump_location_t.
15089 * dumpfile.c: Include "backend.h", "gimple.h", "rtl.h", and
15091 (dump_user_location_t::dump_user_location_t): New constructors,
15092 from gimple *stmt and rtx_insn *.
15093 (dump_user_location_t::from_function_decl): New function.
15094 (dump_loc): Make static.
15095 (dump_gimple_stmt_loc): Convert param "loc" from location_t to
15096 const dump_location_t &.
15097 (dump_generic_expr_loc): Delete.
15098 (dump_printf_loc): Convert param "loc" from location_t to
15099 const dump_location_t &.
15100 (selftest::test_impl_location): New function.
15101 (selftest::dumpfile_c_tests): New function.
15102 * dumpfile.h: Include "profile-count.h".
15103 (class dump_user_location_t): New class.
15104 (struct dump_impl_location_t): New struct.
15105 (class dump_location_t): New class.
15106 (dump_printf_loc): Convert 2nd param from source_location to
15107 const dump_location_t &.
15108 (dump_generic_expr_loc): Delete.
15109 (dump_gimple_stmt_loc): Convert 2nd param from source_location to
15110 const dump_location_t &.
15111 * gimple-fold.c (fold_gimple_assign): Update call to
15112 dump_printf_loc to pass in a dump_location_t rather than a
15113 location_t, via the gimple stmt.
15114 (gimple_fold_call): Likewise.
15115 * gimple-loop-interchange.cc
15116 (loop_cand::analyze_iloop_reduction_var): Update for change to
15117 check_reduction_path.
15118 (tree_loop_interchange::interchange): Update for change to
15119 find_loop_location.
15120 * graphite-isl-ast-to-gimple.c (scop_to_isl_ast): Update for
15121 change in return-type of find_loop_location.
15122 (graphite_regenerate_ast_isl): Likewise.
15123 * graphite-optimize-isl.c (optimize_isl): Likewise.
15124 * graphite.c (graphite_transform_loops): Likewise.
15125 * ipa-devirt.c (ipa_devirt): Update call to dump_printf_loc to
15126 pass in a dump_location_t rather than a location_t, via the
15128 * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
15129 * ipa.c (walk_polymorphic_call_targets): Likewise.
15130 * loop-unroll.c (report_unroll): Convert "locus" param from
15131 location_t to dump_location_t.
15132 (decide_unrolling): Update for change to get_loop_location's
15134 * omp-grid.c (struct grid_prop): Convert field "target_loc" from
15135 location_t to dump_user_location_t.
15136 (grid_find_single_omp_among_assignments_1): Updates calls to
15137 dump_printf_loc to pass in a dump_location_t rather than a
15138 location_t, via the gimple stmt.
15139 (grid_parallel_clauses_gridifiable): Convert "tloc" from
15140 location_t to dump_location_t. Updates calls to dump_printf_loc
15141 to pass in a dump_location_t rather than a location_t, via the
15143 (grid_inner_loop_gridifiable_p): Likewise.
15144 (grid_dist_follows_simple_pattern): Likewise.
15145 (grid_gfor_follows_tiling_pattern): Likewise.
15146 (grid_target_follows_gridifiable_pattern): Likewise.
15147 (grid_attempt_target_gridification): Convert initialization
15148 of local "grid" from memset to zero-initialization; FIXME: does
15149 this require C++11? Update call to dump_printf_loc to pass in a
15150 optinfo_location rather than a location_t, via the gimple stmt.
15151 * profile.c (read_profile_edge_counts): Updates call to
15152 dump_printf_loc to pass in a dump_location_t rather than a
15154 (compute_branch_probabilities): Likewise.
15155 * selftest-run-tests.c (selftest::run_tests): Call
15157 * selftest.h (dumpfile_c_tests): New decl.
15158 * tree-loop-distribution.c (pass_loop_distribution::execute):
15159 Update for change in return type of find_loop_location.
15160 * tree-parloops.c (parallelize_loops): Likewise.
15161 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Convert
15162 "locus" from location_t to dump_user_location_t.
15163 (canonicalize_loop_induction_variables): Likewise.
15164 * tree-ssa-loop-ivopts.c (tree_ssa_iv_optimize_loop): Update
15165 for change in return type of find_loop_location.
15166 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update call
15167 to dump_printf_loc to pass in a dump_location_t rather than a
15168 location_t, via the stmt.
15169 * tree-ssa-sccvn.c (eliminate_dom_walker::before_dom_children):
15171 * tree-vect-loop-manip.c (find_loop_location): Convert return
15172 type from source_location to dump_user_location_t.
15173 (vect_do_peeling): Update for above change.
15174 (vect_loop_versioning): Update for change in type of
15176 * tree-vect-loop.c (check_reduction_path): Convert "loc" param
15177 from location_t to dump_user_location_t.
15178 (vect_estimate_min_profitable_iters): Update for change in type
15180 * tree-vect-slp.c (vect_print_slp_tree): Convert param "loc" from
15181 location_t to dump_location_t.
15182 (vect_slp_bb): Update for change in type of vect_location.
15183 * tree-vectorizer.c (vect_location): Convert from source_location
15184 to dump_user_location_t.
15185 (try_vectorize_loop_1): Update for change in vect_location's type.
15186 (vectorize_loops): Likewise.
15187 (increase_alignment): Likewise.
15188 * tree-vectorizer.h (vect_location): Convert from source_location
15189 to dump_user_location_t.
15190 (find_loop_location): Convert return type from source_location to
15191 dump_user_location_t.
15192 (check_reduction_path): Convert 1st param from location_t to
15193 dump_user_location_t.
15194 * value-prof.c (check_counter): Update call to dump_printf_loc to
15195 pass in a dump_user_location_t rather than a location_t; update
15196 call to error_at for change in type of "locus".
15197 (check_ic_target): Update call to dump_printf_loc to
15198 pass in a dump_user_location_t rather than a location_t, via the
15201 2018-06-26 Robin Dapp <rdapp@linux.vnet.ibm.com>
15203 * config/s390/s390.h (enum processor_flags): Do not use
15204 default tune parameter when -march was specified.
15206 2018-06-26 Jakub Jelinek <jakub@redhat.com>
15209 * config/i386/i386.md (setcc + movzbl to xor + setcc peephole2s):
15210 Check reg_overlap_mentioned_p in addition to reg_set_p with the same
15213 2018-06-26 Richard Biener <rguenther@suse.de>
15215 PR tree-optimization/86287
15217 * tree-vect-loop.c (vect_transform_loop_stmt): Fix read-after-free.
15218 (vect_analyze_loop): Initialize n_stmts.
15220 2018-06-26 Richard Biener <rguenther@suse.de>
15222 PR middle-end/86271
15223 * fold-const.c (fold_convertible_p): Pointer extension
15226 2018-06-26 Alexandre Oliva <aoliva@redhat.com>
15229 * dwarf2out.c (loc_list_has_views): Adjust comments.
15230 (dw_loc_list): Split single cross-partition range with
15233 2018-06-25 Jeff Law <law@redhat.com>
15235 * common/config/v850/v850-common.c (TARGET_DEFAULT_TARGET_FLAGS): Turn
15236 on -mbig-switch by default.
15238 * config/v850/predicates.md (const_float_1_operand): Fix match_code
15240 (const_float_0_operand): Remove unused predicate.
15241 * config/v850/v850.md (define_constants): Remove UNSPEC_LOOP.
15242 (define_c_enum unspec): Add LOOP, RCP and RSQRT constants.
15243 (recipsf2): New expander. Original pattern now called
15245 (recipdf2, recipdf2_insn): Similarly.
15246 (rsqrtsf2, rsqrtsf2_insn): Similarly
15247 (rsqrtdf2, rsqrtdf2_insn): Similarly
15249 2018-06-26 Gerald Pfeifer <gerald@pfeifer.com>
15251 * ginclude/stddef.h: Remove an obsolete comment on FreeBSD 5.
15252 Simplify logic for FreeBSD (twice).
15254 2018-06-25 Martin Sebor <msebor@redhat.com>
15256 PR tree-optimization/86204
15257 * tree-ssa-strlen.c (handle_builtin_strlen): Avoid storing
15258 a strnlen result if it's less than the length of the string.
15260 2018-06-25 Martin Sebor <msebor@redhat.com>
15262 PR tree-optimization/85700
15263 * gimple-fold.c (gimple_fold_builtin_strncat): Adjust comment.
15264 * tree-ssa-strlen.c (is_strlen_related_p): Handle integer subtraction.
15265 (maybe_diag_stxncpy_trunc): Distinguish strncat from strncpy.
15267 2018-06-25 Martin Sebor <msebor@redhat.com>
15269 * doc/extend.texi (Zero-length arrays): Update and clarify.
15271 2018-06-25 Michael Meissner <meissner@linux.ibm.com>
15273 * config.gcc (powerpc64le*): Revert January 16th, 2018 patch that
15274 added IEEE/IBM long double multilib support on PowerPC little
15275 endian Linux systems.
15276 * config/rs6000/linux64.h (MULTILIB_DEFAULTS_IEEE): Likewise.
15277 (MULTILIB_DEFAULTS): Likewise.
15278 * config/rs6000/rs6000.c (rs6000_option_override_internal):
15280 * config/rs6000/rs6000.h (TARGET_IEEEQUAD_MULTILIB): Likewise.
15281 * config/rs6000/t-ldouble-linux64le-ibm: Delete, no longer used.
15282 * config/rs6000/t-ldouble-linux64le-ieee: Delete, no longer used.
15284 2018-06-25 Alexander Monakov <amonakov@ispras.ru>
15286 PR middle-end/86311
15287 * sort.cc (REORDER_23): Avoid memcpy with same destination and source.
15288 (REORDER_45): Likewise.
15290 2018-06-25 Jeff Law <law@redhat.com>
15292 * config/v850/v850.md (divmodhi4): Make sure to sign extend the
15293 dividend to 32 bits. Adjust length.
15294 (udivmodhi4): Cleanup output template. Fix length.
15296 2018-06-25 Carl Love <cel@us.ibm.com>
15298 * config/rs6000/vsx.md: Change word selector to prefered location.
15300 2018-06-25 Richard Biener <rguenther@suse.de>
15302 PR tree-optimization/86304
15303 * tree-vectorizer.c (vectorize_loops): Walk over new possibly
15304 epilogue-if-converted loops as well.
15306 2018-06-25 Jan Hubicka <hubicka@ucw.cz>
15308 * lto-section-out.c (lto_begin_section): Do not print section
15309 name for noaddr and unnumbered dumps.
15311 2018-06-25 Richard Biener <rguenther@suse.de>
15313 * tree-vectorizer.h (struct vec_info_shared): New structure
15314 with parts split out from struct vec_info and loop_nest from
15315 struct _loop_vec_info.
15316 (struct vec_info): Adjust accordingly.
15317 (struct _loop_vec_info): Likewise.
15318 (LOOP_VINFO_LOOP_NEST): Adjust.
15319 (LOOP_VINFO_DATAREFS): Likewise.
15320 (LOOP_VINFO_DDRS): Likewise.
15321 (struct _bb_vec_info): Likewise.
15322 (BB_VINFO_DATAREFS): Likewise.
15323 (BB_VINFO_DDRS): Likewise.
15324 (struct _stmt_vec_info): Add dr_aux member.
15325 (DR_VECT_AUX): Adjust to refer to member of DR_STMTs vinfo.
15326 (DR_MISALIGNMENT_UNINITIALIZED): New.
15327 (set_dr_misalignment): Adjust.
15328 (dr_misalignment): Assert misalign isn't DR_MISALIGNMENT_UNINITIALIZED.
15329 (vect_analyze_loop): Adjust prototype.
15330 (vect_analyze_loop_form): Likewise.
15331 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences):
15332 Compute dependences lazily.
15333 (vect_record_base_alignments): Use shared datarefs/ddrs.
15334 (vect_verify_datarefs_alignment): Likewise.
15335 (vect_analyze_data_refs_alignment): Likewise.
15336 (vect_analyze_data_ref_accesses): Likewise.
15337 (vect_analyze_data_refs): Likewise.
15338 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Add
15339 constructor parameter for shared part.
15340 (vect_analyze_loop_form): Pass in shared part and adjust.
15341 (vect_analyze_loop_2): Pass in storage for the number of
15342 stmts. Move loop nest finding to the caller. Compute
15344 (vect_analyze_loop): Pass in shared part.
15345 (vect_transform_loop): Verify shared datarefs are unchanged.
15346 * tree-vect-slp.c (_bb_vec_info::_bb_vec_info): Add
15347 constructor parameter for shared part.
15348 (vect_slp_analyze_bb_1): Pass in shared part and adjust.
15349 (vect_slp_bb): Verify shared datarefs are unchanged before
15351 * tree-vect-stmts.c (ensure_base_align): Adjust for DR_AUX
15353 (new_stmt_vec_info): Initialize DR_AUX misalignment to
15354 DR_MISALIGNMENT_UNINITIALIZED.
15355 * tree-vectorizer.c (vec_info::vec_info): Add constructor
15356 parameter for shared part.
15357 (vec_info::~vec_info): Adjust.
15358 (vec_info_shared::vec_info_shared): New.
15359 (vec_info_shared::~vec_info_shared): Likewise.
15360 (vec_info_shared::save_datarefs): Likewise.
15361 (vec_info_shared::check_datarefs): Likewise.
15362 (try_vectorize_loop_1): Construct shared part live for analyses
15363 of a single loop for multiple vector sizes.
15364 * tree-parloops.c (gather_scalar_reductions): Adjust.
15366 2018-06-25 Richard Biener <rguenther@suse.de>
15368 * tree-vect-data-refs.c (vect_find_stmt_data_reference): Modify
15369 DR for SIMD lane accesses here and mark DR with (void *)-1 aux.
15370 (vect_analyze_data_refs): Remove similar code from here and
15371 simplify accordingly.
15373 2018-06-25 Richard Biener <rguenther@suse.de>
15375 * tree-vect-data-refs.c (vect_check_gather_scatter): Fail
15376 for reverse storage order accesses rather than asserting
15377 they cannot happen here.
15379 2018-06-25 Tom de Vries <tdevries@suse.de>
15382 * config/i386/i386.md (define_insn "*tls_global_dynamic_64_<mode>"):
15383 Use data16 instead of .byte for insn prefix.
15385 2018-06-25 Andreas Krebbel <krebbel@linux.ibm.com>
15388 * parser.c (make_char_string_pack): Pass this literal chars
15389 through cpp_interpret_string.
15390 (cp_parser_userdef_numeric_literal): Check the result of
15391 make_char_string_pack.
15393 2018-06-24 Maya Rashish <coypu@sdf.org>
15395 * ginclude/stddef.h: Simplify conditions around avoiding
15396 re-definition of __size_t.
15398 2018-06-22 Jan Hubicka <hubicka@ucw.cz>
15400 * lto-streamer-out.c (tree_is_indexable): Make LABEL_DECL nonindexable
15401 unless it is forced or nonlocal; assert that we stream no IMPORTED_DECL.
15403 2018-06-22 Maya Rashish <coypu@sdf.org>
15405 * doc/invoke.texi (mno-fancy-math-387): Update for changes
15406 made to OpenBSD and NetBSD through the years.
15408 2018-06-22 Kelvin Nilsen <kelvin@gcc.gnu.org>
15410 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Change
15411 behavior of vec_pack (vector double, vector double) to match
15412 behavior of vec_float2 (vector double, vector double).
15414 2018-06-22 Olivier Hainque <hainque@adacore.com>
15416 * gimplify.c (gimplify_function_tree): Prevent creation
15417 of a trampoline for the address of the current function
15418 passed to entry/exit instrumentation hooks.
15420 2018-06-22 Aaron Sawdey <acsawdey@linux.ibm.com>
15423 * config/rs6000/rs6000-string.c (expand_strn_compare): Handle -m32
15426 2018-06-22 Martin Liska <mliska@suse.cz>
15428 PR tree-optimization/86263
15429 * tree-switch-conversion.c (jump_table_cluster::find_jump_tables):
15430 Bail out if is_enabled is false.
15431 * tree-switch-conversion.h (jump_table_cluster::is_enabled):
15433 (jump_table_cluster::is_enabled): New function.
15435 2018-06-22 Jan Hubicka <hubicka@ucw.cz>
15437 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not stream
15438 BINFO_BASE_ACCESSES and BINFO_VPTR_FIELD.
15439 * tree-streamer-in.c (streamer_read_tree_bitfields): Likewise.
15440 (lto_input_ts_binfo_tree_pointers): Likewise.
15441 * tree-streamer-out.c (streamer_write_tree_bitfields,
15442 write_ts_binfo_tree_pointers): Likewise.
15443 * tree.c (free_lang_data_in_binfo): Clear BINFO_VPTR_FIELD.
15445 2018-06-22 Jan Hubicka <hubicka@ucw.cz>
15447 * tree.c (free_lang_data_in_type): Free all TYPE_VFIELDs.
15449 2018-06-22 Martin Liska <mliska@suse.cz>
15451 * symbol-summary.h (get): Make it pure and inline move
15452 functionality from ::get function.
15453 (get): Remove and inline into ::get and ::get_create.
15454 (get_create): Move code from ::get function.
15456 2018-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
15459 * config/i386/sol2.h (CPP_SPEC): Don't pass -P for
15460 -x assembler-with-cpp.
15462 2018-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
15464 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define
15465 _FILE_OFFSET_BITS=64 for C++.
15467 2018-06-21 Michael Meissner <meissner@linux.ibm.com>
15469 * config/rs6000/rs6000.md (extendtfif2): Add missing 128-bit
15470 conversion insn that shows up when pr85657-3.c is compiled using
15471 IEEE 128-bit long double.
15472 (neg<mode>2_internal): Use the correct mode to check whether the
15473 mode is IBM extended.
15474 * config/rs6000/rs6000.c (init_float128_ieee): Prevent complex
15475 multiply and divide external functions from being created more
15478 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
15480 * cfgrtl.c (fixup_reorder_chain): Do not emit NOPs in DECL_IGNORED_P
15482 (rtl_merge_blocks): Likewise. Do not emit a NOP if the location of
15483 the edge can be forwarded.
15484 (cfg_layout_merge_blocks): Likewise.
15486 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
15488 * except.c (finish_eh_generation): Commit edge insertions only after
15489 the EH edges have been redirected from post-landing to landing pads.
15491 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
15493 * tree-nested.c (get_frame_type): Use create_tmp_var_raw instead of
15494 create_tmp_var_for to create the FRAME decl.
15495 (finalize_nesting_tree_1): Do not unchain the FRAME decl.
15497 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
15499 * tree-inline.c (copy_edges_for_bb): Minor tweak.
15500 (maybe_move_debug_stmts_to_successors): Also reset the locus of the
15501 debug statement when resetting its value.
15502 (expand_call_inline): Copy the locus of the call onto the assignment
15503 of the return value, if any. Use local variable in more cases.
15505 2018-06-21 Martin Liska <mliska@suse.cz>
15507 * ipa-pure-const.c (propagate_nothrow): Use
15508 funct_state_summaries->get.
15509 (dump_malloc_lattice): Likewise.
15510 (propagate_malloc): Likewise.
15512 2018-06-21 Richard Biener <rguenther@suse.de>
15514 * lto-streamer-out.c (DFS::DFS_write_tree_body): Update outdated
15515 comment. Follow BLOCK_ABSTRACT_ORIGIN unconditionally.
15516 * tree-streamer-in.c (lto_input_ts_block_tree_pointers): Update
15518 * tree-streamer-out.c (write_ts_block_tree_pointers): Stream
15519 BLOCK_ABSTRACT_ORIGIN unconditionally.
15521 2018-06-21 David Malcolm <dmalcolm@redhat.com>
15523 * ipa-cp.c (ipcp_driver): Set edge_clone_summaries to NULL after
15525 * ipa-reference.c (ipa_reference_c_finalize): Delete
15526 ipa_ref_opt_sum_summaries and set it to NULL.
15528 2018-06-21 Tom de Vries <tdevries@suse.de>
15530 PR tree-optimization/85859
15531 * tree-ssa-tail-merge.c (stmt_local_def): Copy gimple_is_call
15532 test with comment from bb_no_side_effects_p.
15534 2018-06-21 Richard Biener <rguenther@suse.de>
15536 PR tree-optimization/86232
15537 * tree-ssa-loop-niter.c (number_of_iterations_popcount): Adjust
15538 max for constant niter.
15540 2018-06-21 Andre Vieira <andre.simoesdiasvieira@arm.com>
15542 * config/aarch64/aarch64-simd.md
15543 (*aarch64_crypto_aes<aes_op>v16qi_xor_combine): New.
15545 2018-06-21 Andre Vieira <andre.simoesdiasvieira@arm.com>
15547 * config/aarch64/aarch64-simd.md (aarch64_crypto_aes<aes_op>v16qi):
15548 Make opernads of the unspec commutative.
15550 2018-06-21 Richard Biener <rguenther@suse.de>
15552 * tree-data-ref.c (dr_step_indicator): Handle NULL DR_STEP.
15553 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
15554 Avoid calling vect_mark_for_runtime_alias_test with gathers or scatters.
15555 (vect_analyze_data_ref_dependence): Re-order checks to deal with
15557 (vect_record_base_alignments): Do not record base alignment
15558 for gathers or scatters.
15559 (vect_compute_data_ref_alignment): Drop return value that is always
15560 true. Bail out early for gathers or scatters.
15561 (vect_enhance_data_refs_alignment): Bail out early for gathers
15563 (vect_find_same_alignment_drs): Likewise.
15564 (vect_analyze_data_refs_alignment): Remove dead code.
15565 (vect_slp_analyze_and_verify_node_alignment): Likewise.
15566 (vect_analyze_data_refs): For possible gathers or scatters do
15567 not create an alternate DR, just check their possible validity
15568 and mark them. Adjust DECL_NONALIASED handling to not rely
15569 on DR_BASE_ADDRESS.
15570 * tree-vect-loop-manip.c (vect_update_inits_of_drs): Do not
15571 update inits of gathers or scatters.
15572 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern):
15573 Also copy gather/scatter flag to pattern vinfo.
15575 2018-06-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
15577 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Change
15578 behavior of vec_packsu (vector unsigned long long, vector unsigned
15579 long long) to match behavior of vec_packs with same signature.
15581 2018-06-20 Chung-Lin Tang <cltang@codesourcery.com>
15582 Thomas Schwinge <thomas@codesourcery.com>
15583 Cesar Philippidis <cesar@codesourcery.com>
15585 * gimplify.c (gimplify_scan_omp_clauses): Add support for
15586 OMP_CLAUSE_{IF_PRESENT,FINALIZE}.
15587 (gimplify_adjust_omp_clauses): Likewise.
15588 (gimplify_oacc_declare_1): Add support for GOMP_MAP_RELEASE, remove
15589 support for GOMP_MAP_FORCE_{ALLOC,TO,FROM,TOFROM}.
15590 (gimplify_omp_target_update): Update handling of acc update and
15592 * omp-low.c (install_var_field): Remove unused parameter
15593 base_pointers_restrict.
15594 (scan_sharing_clauses): Remove base_pointers_restrict parameter.
15595 Update call to install_var_field. Handle OMP_CLAUSE_{IF_PRESENT,
15597 (omp_target_base_pointers_restrict_p): Delete.
15598 (scan_omp_target): Update call to scan_sharing_clauses.
15599 * tree-core.h (enum omp_clause_code): Add OMP_CLAUSE_{IF_PRESENT,
15601 * tree-nested.c (convert_nonlocal_omp_clauses): Handle
15602 OMP_CLAUSE_{IF_PRESENT,FINALIZE}.
15603 (convert_local_omp_clauses): Likewise.
15604 * tree-pretty-print.c (dump_omp_clause): Likewise.
15605 * tree.c (omp_clause_num_ops): Add entries for OMP_CLAUSE_{IF_PRESENT,
15607 (omp_clause_code_name): Likewise.
15609 2018-06-20 Jakub Jelinek <jakub@redhat.com>
15612 * var-tracking.c (use_narrower_mode_test): Check if shift amount can
15615 PR tree-optimization/86231
15616 * tree-vrp.c (union_ranges): For ( [ ) ] or ( )[ ] range and
15617 anti-range don't overwrite *vr0min before using it to compute *vr0max.
15619 2018-06-20 Tom de Vries <tdevries@suse.de>
15621 PR tree-optimization/86097
15622 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Also convert *nit to
15623 iv type if signedness of iv type is not the same as that of *nit.
15625 2018-06-20 Jakub Jelinek <jakub@redhat.com>
15627 * cfgrtl.c (rtl_verify_edges): Formatting fix. If bb->preds has any
15628 EDGE_EH edges, verify they are all EDGE_EH.
15630 2018-06-20 Maya Rashish <coypu@sdf.org>
15632 * ginclude/stddef.h: Limit #include <machine/ansi.h> to NetBSD.
15634 2018-06-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15636 * config/aarch64/aarch64-tuning-flags.def (no_ldp_stp_qregs): New.
15637 * config/aarch64/aarch64.c (xgene1_tunings): Add
15638 AARCH64_EXTRA_TUNE_NO_LDP_STP_QREGS to tune_flags.
15639 (aarch64_mode_valid_for_sched_fusion_p):
15640 Allow 16-byte modes.
15641 (aarch64_classify_address): Allow 16-byte modes for load_store_pair_p.
15642 * config/aarch64/aarch64-ldpstp.md: Add peepholes for LDP STP of
15644 * config/aarch64/aarch64-simd.md (load_pair<VQ:mode><VQ2:mode>):
15646 (vec_store_pair<VQ:mode><VQ2:mode>): Likewise.
15647 * config/aarch64/iterators.md (VQ2): New mode iterator.
15649 2018-06-20 Martin Liska <mliska@suse.cz>
15651 * tree-switch-conversion.c (jump_table_cluster::can_be_handled):
15652 Change default ratio from 10 to 8.
15654 2018-06-20 Martin Liska <mliska@suse.cz>
15656 * tree-switch-conversion.c (jump_table_cluster::find_jump_tables):
15658 (bit_test_cluster::find_bit_tests): Likewise.
15659 (switch_decision_tree::analyze_switch_statement): Find clusters.
15660 * tree-switch-conversion.h (struct jump_table_cluster): Document
15663 2018-06-20 Martin Liska <mliska@suse.cz>
15665 * tree-switch-conversion.c (switch_conversion::collect):
15666 Record m_uniq property.
15667 (switch_conversion::expand): Bail out for special conditions.
15668 (group_cluster::~group_cluster): New.
15669 (group_cluster::group_cluster): Likewise.
15670 (group_cluster::dump): Likewise.
15671 (jump_table_cluster::emit): New.
15672 (switch_decision_tree::fix_phi_operands_for_edges): New.
15673 (struct case_node): Remove struct.
15674 (jump_table_cluster::can_be_handled): New.
15675 (case_values_threshold): Moved to header.
15676 (reset_out_edges_aux): Likewise.
15677 (jump_table_cluster::is_beneficial): New.
15678 (bit_test_cluster::can_be_handled): Likewise.
15679 (add_case_node): Remove.
15680 (bit_test_cluster::is_beneficial): New.
15681 (case_bit_test::cmp): New.
15682 (bit_test_cluster::emit): New.
15683 (expand_switch_as_decision_tree_p): Remove.
15684 (bit_test_cluster::hoist_edge_and_branch_if_true): New.
15685 (fix_phi_operands_for_edge): Likewise.
15686 (switch_decision_tree::analyze_switch_statement): New.
15687 (compute_cases_per_edge): Move ...
15688 (switch_decision_tree::compute_cases_per_edge): ... here.
15689 (try_switch_expansion): Likewise.
15690 (switch_decision_tree::try_switch_expansion): Likewise.
15691 (record_phi_operand_mapping): Likewise.
15692 (switch_decision_tree::record_phi_operand_mapping): Likewise.
15693 (emit_case_decision_tree): Likewise.
15694 (switch_decision_tree::emit): Likewise.
15695 (balance_case_nodes): Likewise.
15696 (switch_decision_tree::balance_case_nodes): Likewise.
15697 (dump_case_nodes): Likewise.
15698 (switch_decision_tree::dump_case_nodes): Likewise.
15699 (emit_jump): Likewise.
15700 (switch_decision_tree::emit_jump): Likewise.
15701 (emit_cmp_and_jump_insns): Likewise.
15702 (switch_decision_tree::emit_cmp_and_jump_insns): Likewise.
15703 (emit_case_nodes): Likewise.
15704 (switch_decision_tree::emit_case_nodes): Likewise.
15705 (conditional_probability): Remove.
15706 * tree-switch-conversion.h (enum cluster_type): New.
15708 (struct cluster): Likewise.
15709 (cluster::cluster): Likewise.
15710 (struct simple_cluster): Likewise.
15711 (simple_cluster::simple_cluster): Likewise.
15712 (struct group_cluster): Likewise.
15713 (struct jump_table_cluster): Likewise.
15714 (struct bit_test_cluster): Likewise.
15715 (struct min_cluster_item): Likewise.
15716 (struct case_tree_node): Likewise.
15717 (case_tree_node::case_tree_node): Likewise.
15718 (jump_table_cluster::case_values_threshold): Likewise.
15719 (struct case_bit_test): Likewise.
15720 (struct switch_decision_tree): Likewise.
15721 (struct switch_conversion): Likewise.
15722 (switch_decision_tree::reset_out_edges_aux): Likewise.
15724 2018-06-20 Martin Liska <mliska@suse.cz>
15726 * tree-switch-conversion.c (MAX_CASE_BIT_TESTS): Remove.
15727 (hoist_edge_and_branch_if_true): Likewise.
15728 (expand_switch_using_bit_tests_p): Likewise.
15729 (struct case_bit_test): Likewise.
15730 (case_bit_test_cmp): Likewise.
15731 (emit_case_bit_tests): Likewise.
15732 (switch_conversion::switch_conversion): New class.
15733 (struct switch_conv_info): Remove old struct.
15734 (collect_switch_conv_info): More to ...
15735 (switch_conversion::collect): ... this.
15736 (check_range): Likewise.
15737 (switch_conversion::check_range): Likewise.
15738 (check_all_empty_except_final): Likewise.
15739 (switch_conversion::check_all_empty_except_final): Likewise.
15740 (check_final_bb): Likewise.
15741 (switch_conversion::check_final_bb): Likewise.
15742 (create_temp_arrays): Likewise.
15743 (switch_conversion::create_temp_arrays): Likewise.
15744 (free_temp_arrays): Likewise.
15745 (gather_default_values): Likewise.
15746 (switch_conversion::gather_default_values): Likewise.
15747 (build_constructors): Likewise.
15748 (switch_conversion::build_constructors): Likewise.
15749 (constructor_contains_same_values_p): Likewise.
15750 (switch_conversion::contains_same_values_p): Likewise.
15751 (array_value_type): Likewise.
15752 (switch_conversion::array_value_type): Likewise.
15753 (build_one_array): Likewise.
15754 (switch_conversion::build_one_array): Likewise.
15755 (build_arrays): Likewise.
15756 (switch_conversion::build_arrays): Likewise.
15757 (gen_def_assigns): Likewise.
15758 (switch_conversion::gen_def_assigns): Likewise.
15759 (prune_bbs): Likewise.
15760 (switch_conversion::prune_bbs): Likewise.
15761 (fix_phi_nodes): Likewise.
15762 (switch_conversion::fix_phi_nodes): Likewise.
15763 (gen_inbound_check): Likewise.
15764 (switch_conversion::gen_inbound_check): Likewise.
15765 (process_switch): Use the newly created class.
15766 (switch_conversion::expand): New.
15767 (switch_conversion::~switch_conversion): New.
15768 * tree-switch-conversion.h: New file.
15770 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
15772 * tree-vectorizer.h (NUM_PATTERNS, vect_recog_func_ptr): Move to
15773 tree-vect-patterns.c.
15774 * tree-vect-patterns.c (vect_supportable_direct_optab_p): New function.
15775 (vect_recog_dot_prod_pattern): Use it. Remove the type_in argument.
15776 (vect_recog_sad_pattern): Likewise.
15777 (vect_recog_widen_sum_pattern): Likewise.
15778 (vect_recog_pow_pattern): Likewise. Check for a null vectype.
15779 (vect_recog_widen_shift_pattern): Remove the type_in argument.
15780 (vect_recog_rotate_pattern): Likewise.
15781 (vect_recog_mult_pattern): Likewise.
15782 (vect_recog_vector_vector_shift_pattern): Likewise.
15783 (vect_recog_divmod_pattern): Likewise.
15784 (vect_recog_mixed_size_cond_pattern): Likewise.
15785 (vect_recog_bool_pattern): Likewise.
15786 (vect_recog_mask_conversion_pattern): Likewise.
15787 (vect_try_gather_scatter_pattern): Likewise.
15788 (vect_recog_widen_mult_pattern): Likewise. Check for a null vectype.
15789 (vect_recog_over_widening_pattern): Likewise.
15790 (vect_recog_gather_scatter_pattern): Likewise.
15791 (vect_recog_func_ptr): Move from tree-vectorizer.h
15792 (vect_vect_recog_func_ptrs): Move further down the file.
15793 (vect_recog_func): Likewise. Remove the third argument.
15794 (NUM_PATTERNS): Define based on vect_vect_recog_func_ptrs.
15795 (vect_pattern_recog_1): Expect the pattern function to do any
15796 necessary target tests. Also expect it to provide a vector type.
15797 Remove the type_in handling.
15799 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
15801 * tree-vect-patterns.c (vect_pattern_detected): New function.
15802 (vect_recog_dot_prod_patternm, vect_recog_sad_pattern)
15803 (vect_recog_widen_mult_pattern, vect_recog_widen_sum_pattern)
15804 (vect_recog_over_widening_pattern, vect_recog_widen_shift_pattern
15805 (vect_recog_rotate_pattern, vect_recog_vector_vector_shift_pattern)
15806 (vect_recog_mult_pattern, vect_recog_divmod_pattern)
15807 (vect_recog_mixed_size_cond_pattern, vect_recog_bool_pattern)
15808 (vect_recog_mask_conversion_pattern)
15809 (vect_try_gather_scatter_pattern): Likewise.
15811 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
15813 * tree-vect-patterns.c (vect_get_internal_def): New function.
15814 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern)
15815 (vect_recog_vector_vector_shift_pattern, check_bool_pattern)
15816 (search_type_for_mask_1): Use it.
15818 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
15820 * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Remove
15821 redundant WIDEN_SUM_EXPR handling.
15822 (vect_recog_sad_pattern): Likewise.
15824 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
15826 * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Remove
15827 redundant check that the types of a PLUS_EXPR or MULT_EXPR agree.
15828 (vect_recog_sad_pattern): Likewise PLUS_EXPR, ABS_EXPR and MINUS_EXPR.
15829 (vect_recog_widen_mult_pattern): Likewise MULT_EXPR.
15830 (vect_recog_widen_sum_pattern): Likewise PLUS_EXPR.
15832 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
15834 * tree-vect-stmts.c (vectorizable_call): Make sure that we
15835 use the stmt_vec_info of the original bb statement for the
15836 new zero assignment, even if the call is part of a pattern.
15838 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
15840 * tree-vectorizer.h (_stmt_vec_info): Note above pattern_def_seq
15841 that the sequence is attached to the original statement rather
15842 than the pattern statement.
15843 * tree-vect-loop.c (vect_determine_vf_for_stmt): Take the
15844 PATTERN_DEF_SEQ from the original statement rather than
15845 the main pattern statement.
15846 * tree-vect-stmts.c (free_stmt_vec_info): Likewise.
15847 * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Likewise.
15848 (vect_mark_pattern_stmts): Don't copy the PATTERN_DEF_SEQ.
15850 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
15852 * tree-vect-stmts.c (vect_analyze_stmt): Move the handling of pattern
15853 definition statements before the early exit for statements that aren't
15855 * tree-vect-loop.c (vect_transform_loop_stmt): New function,
15857 (vect_transform_loop): ...here. Process pattern definition
15858 statements without first checking whether the main pattern
15859 statement is live or relevant.
15861 2018-06-19 Eric Botcazou <ebotcazou@adacore.com>
15863 * tree-cfgcleanup.c (tree_forwarder_block_p): Do not return false at
15864 -O0 if the locus represent UNKNOWN_LOCATION but have different values.
15866 2018-06-19 Aaron Sawdey <acsawdey@linux.ibm.com>
15868 * config/rs6000/rs6000-string.c (select_block_compare_mode): Check
15869 TARGET_EFFICIENT_OVERLAPPING_UNALIGNED here instead of in caller.
15870 (do_and3, do_and3_mask, do_compb3, do_rotl3): New functions.
15871 (expand_block_compare): Change select_block_compare_mode call.
15872 (expand_strncmp_align_check): Use new functions, fix comment.
15873 (emit_final_str_compare_gpr): New function.
15874 (expand_strn_compare): Refactor and clean up code.
15875 * config/rs6000/vsx.md (vsx_mov<mode>_64bit): Remove *.
15877 2018-06-19 Tony Reix <tony.reix@atos.com>
15878 Damien Bergamini <damien.bergamini@atos.com>
15879 David Edelsohn <dje.gcc@gmail.com>
15881 * collect2.c (static_obj): New variable.
15882 (static_libs): New variable.
15883 (is_in_list): Uncomment declaration.
15884 (main): Track AIX libraries linked statically.
15885 (is_in_list): Uncomment definition.
15886 (scan_prog_file): Don't add AIX shared libraries initializer
15887 to constructor list if linking statically.
15889 2018-06-19 Max Filippov <jcmvbkbc@gmail.com>
15891 * config/xtensa/xtensa.md (UNSPEC_FRAME_BLOCKAGE): New unspec
15893 (allocate_stack, frame_blockage, *frame_blockage): New patterns.
15895 2018-06-19 Jan Hubicka <hubicka@ucw.cz>
15897 * tree.c (find_decls_types_r): Remove all non-VAR_DECLs from
15900 2018-06-19 Martin Liska <mliska@suse.cz>
15902 * config/i386/i386.c (ix86_can_inline_p): Do not use
15903 ipa_fn_summaries::get_create.
15904 * ipa-cp.c (ipcp_cloning_candidate_p): Replace get_create with
15906 (devirtualization_time_bonus): Likewise.
15907 (ipcp_propagate_stage): Likewise.
15908 * ipa-fnsummary.c (redirect_to_unreachable): Likewise.
15909 (edge_set_predicate): Likewise.
15910 (evaluate_conditions_for_known_args): Likewise.
15911 (evaluate_properties_for_edge): Likewise.
15912 (ipa_call_summary::reset): Tranform to ...
15913 (ipa_call_summary::~ipa_call_summary): ... this.
15914 (ipa_fn_summary::reset): Transform to ...
15915 (ipa_fn_summary::~ipa_fn_summary): ... this.
15916 (ipa_fn_summary_t::remove): Rename to ...
15917 (ipa_fn_summary_t::remove_callees): ... this.
15918 (ipa_fn_summary_t::duplicate): Use placement new
15919 instead of memory copy.
15920 (ipa_call_summary_t::duplicate): Likewise.
15921 (ipa_call_summary_t::remove): Remove.
15922 (dump_ipa_call_summary): Change get_create to get.
15923 (ipa_dump_fn_summary): Dump only when summary exists.
15924 (analyze_function_body): Use symbol_summary::get instead
15926 (compute_fn_summary): Likewise.
15927 (estimate_edge_devirt_benefit): Likewise.
15928 (estimate_edge_size_and_time): Likewise.
15929 (inline_update_callee_summaries): Likewise.
15930 (remap_edge_change_prob): Likewise.
15931 (remap_edge_summaries): Likewise.
15932 (ipa_merge_fn_summary_after_inlining): Likewise.
15933 (write_ipa_call_summary): Likewise.
15934 (ipa_fn_summary_write): Likewise.
15935 (ipa_free_fn_summary): Likewise.
15936 * ipa-fnsummary.h (struct GTY): Add new ctor and copy ctor.
15937 (struct ipa_call_summary): Likewise.
15938 * ipa-icf.c (sem_function::merge): Use symbol_summary::get instead
15940 * ipa-inline-analysis.c (do_estimate_edge_time): Likewise.
15941 (estimate_size_after_inlining): Likewise.
15942 (estimate_growth): Likewise.
15943 (growth_likely_positive): Likewise.
15944 * ipa-inline-transform.c (clone_inlined_nodes): Likewise.
15945 (inline_call): Likewise.
15946 * ipa-inline.c (caller_growth_limits): Likewise.
15947 (can_inline_edge_p): Likewise.
15948 (can_inline_edge_by_limits_p): Likewise.
15949 (compute_uninlined_call_time): Likewise.
15950 (compute_inlined_call_time): Likewise.
15951 (want_inline_small_function_p): Likewise.
15952 (edge_badness): Likewise.
15953 (update_caller_keys): Likewise.
15954 (update_callee_keys): Likewise.
15955 (inline_small_functions): Likewise.
15956 (inline_to_all_callers_1): Likewise.
15957 (dump_overall_stats): Likewise.
15958 (early_inline_small_functions): Likewise.
15959 (early_inliner): Likewise.
15960 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
15961 * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
15962 * ipa-pure-const.c (malloc_candidate_p): Likewise.
15963 * ipa-split.c (execute_split_functions): Likewise.
15964 * symbol-summary.h: Likewise.
15965 * tree-sra.c (ipa_sra_preliminary_function_checks): Likewise.
15967 2018-06-19 Richard Biener <rguenther@suse.de>
15969 * tree-vectorizer.c (try_vectorize_loop_1): Split out of ...
15970 (vectorize_loops): ... here. Fix dbgcnt handling.
15971 (try_vectorize_loop): Wrap try_vectorize_loop_1.
15973 2018-06-19 Segher Boessenkool <segher@kernel.crashing.org>
15976 * config/rs6000/rs6000.md (rs6000_discover_homogeneous_aggregate): An
15977 ieee128 argument takes up only one (vector) register, not two (floating
15980 2018-06-19 Eric Botcazou <ebotcazou@adacore.com>
15982 * gimplify.c (gimplify_init_constructor): Really never clear for an
15983 incomplete constructor if CONSTRUCTOR_NO_CLEARING is set.
15985 2018-06-19 Richard Biener <rguenther@suse.de>
15987 PR tree-optimization/86179
15988 * tree-vect-patterns.c (vect_pattern_recog_1): Clean up
15989 after failed recognition.
15991 2018-06-18 Martin Sebor <msebor@redhat.com>
15993 PR middle-end/85602
15994 * calls.c (maybe_warn_nonstring_arg): Handle strncat.
15995 * tree-ssa-strlen.c (is_strlen_related_p): Make extern.
15996 Handle integer subtraction.
15997 (maybe_diag_stxncpy_trunc): Handle nonstring source arguments.
15998 * tree-ssa-strlen.h (is_strlen_related_p): Declare.
16000 2018-06-18 David Malcolm <dmalcolm@redhat.com>
16002 * config/frv/frv-protos.h (frv_ifcvt_modify_insn): Strengthen 3rd
16003 param from rtx to rtx_insn *.
16004 * config/frv/frv.c (frv_ifcvt_add_insn): Likewise for "insn"
16006 (frv_ifcvt_modify_insn): Likwise.
16007 (frv_ifcvt_modify_final): Likwise for local "existing_insn",
16008 adding an as_a <rtx_insn *> cast. Likewise for local "insn".
16009 * config/mips/mips.c (r10k_insert_cache_barriers): Add an
16010 as_a <rtx_insn *> cast to local "unprotected_region" once
16011 it's been established that it's not NULL or pc_rtx.
16012 * config/nds32/nds32-relax-opt.c (nds32_group_insns): Strengthen
16013 param "sethi" from rtx to rtx_insn *.
16014 (nds32_group_float_insns): Likewise for param "insn".
16015 * config/vax/vax-protos.h (vax_output_int_add): Likewise for 1st
16017 (vax_output_int_subtract): Likewise.
16018 * config/vax/vax.c (vax_output_int_add): Likewise for param
16020 (vax_output_int_subtract): Likewise.
16021 * emit-rtl.c (set_insn_deleted): Likewise, removing cast.
16022 (emit_pattern_after): Likewise for param "after".
16023 (emit_insn_after): Likewise.
16024 (emit_jump_insn_after): Likewise.
16025 (emit_call_insn_after): Likewise.
16026 (emit_debug_insn_after): Likewise.
16027 (emit_pattern_before): Likewise for param "before".
16028 (emit_insn_before): Likewise.
16029 (emit_jump_insn_before): Likewise.
16030 * final.c (get_insn_template): Likewise for param "insn", removing
16032 * output.h (get_insn_template): Likewise for 2nd param.
16033 * rtl.h (emit_insn_before): Likewise.
16034 (emit_jump_insn_before): Likewise.
16035 (emit_debug_insn_before_noloc): Likewise.
16036 (emit_insn_after): Likewise.
16037 (emit_jump_insn_after): Likewise.
16038 (emit_call_insn_after): Likewise.
16039 (emit_debug_insn_after): Likewise.
16040 (set_insn_deleted): Likewise for param.
16042 2018-06-18 Michael Meissner <meissner@linux.ibm.com>
16045 * config/rs6000/rs6000-modes.def (toplevel): Rework the 128-bit
16046 floating point modes, so that IFmode is numerically greater than
16047 TFmode, which is greater than KFmode using FRACTIONAL_FLOAT_MODE
16048 to declare the ordering. This prevents IFmode from being
16049 converted to TFmode when long double is IEEE 128-bit on an ISA 3.0
16050 machine. Include rs6000-modes.h to share the fractional values
16051 between genmodes* and the rest of the compiler.
16052 (IFmode): Likewise.
16053 (KFmode): Likewise.
16054 (TFmode): Likewise.
16055 * config/rs6000/rs6000-modes.h: New file.
16056 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Change the
16057 meaning of rs6000_long_double_size so that 126..128 selects an
16058 appropriate 128-bit floating point type.
16059 (rs6000_option_override_internal): Likewise.
16060 * config/rs6000/rs6000.h (toplevel): Include rs6000-modes.h.
16061 (TARGET_LONG_DOUBLE_128): Change the meaning of
16062 rs6000_long_double_size so that 126..128 selects an appropriate
16063 128-bit floating point type.
16064 (LONG_DOUBLE_TYPE_SIZE): Update comment.
16065 * config/rs6000/rs6000.md (trunciftf2): Correct the modes of the
16066 source and destination to match the standard usage.
16067 (truncifkf2): Likewise.
16068 (copysign<mode>3, IEEE iterator): Rework copysign of float128 on
16069 ISA 2.07 to use an explicit clobber, instead of passing in a
16071 (copysign<mode>3_soft): Likewise.
16073 2018-06-18 David Malcolm <dmalcolm@redhat.com>
16075 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences):
16076 Replace dump_printf_loc call with DUMP_VECT_SCOPE.
16077 (vect_slp_analyze_instance_dependence): Likewise.
16078 (vect_enhance_data_refs_alignment): Likewise.
16079 (vect_analyze_data_refs_alignment): Likewise.
16080 (vect_slp_analyze_and_verify_instance_alignment
16081 (vect_analyze_data_ref_accesses): Likewise.
16082 (vect_prune_runtime_alias_test_list): Likewise.
16083 (vect_analyze_data_refs): Likewise.
16084 * tree-vect-loop-manip.c (vect_update_inits_of_drs): Likewise.
16085 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
16086 (vect_analyze_scalar_cycles_1): Likewise.
16087 (vect_get_loop_niters): Likewise.
16088 (vect_analyze_loop_form_1): Likewise.
16089 (vect_update_vf_for_slp): Likewise.
16090 (vect_analyze_loop_operations): Likewise.
16091 (vect_analyze_loop): Likewise.
16092 (vectorizable_induction): Likewise.
16093 (vect_transform_loop): Likewise.
16094 * tree-vect-patterns.c (vect_pattern_recog): Likewise.
16095 * tree-vect-slp.c (vect_analyze_slp): Likewise.
16096 (vect_make_slp_decision): Likewise.
16097 (vect_detect_hybrid_slp): Likewise.
16098 (vect_slp_analyze_operations): Likewise.
16099 (vect_slp_bb): Likewise.
16100 * tree-vect-stmts.c (vect_mark_stmts_to_be_vectorized): Likewise.
16101 (vectorizable_bswap): Likewise.
16102 (vectorizable_call): Likewise.
16103 (vectorizable_simd_clone_call): Likewise.
16104 (vectorizable_conversion): Likewise.
16105 (vectorizable_assignment): Likewise.
16106 (vectorizable_shift): Likewise.
16107 (vectorizable_operation): Likewise.
16108 * tree-vectorizer.h (DUMP_VECT_SCOPE): New macro.
16110 2018-06-18 Martin Sebor <msebor@redhat.com>
16112 PR tree-optimization/81384
16113 * builtin-types.def (BT_FN_SIZE_CONST_STRING_SIZE): New.
16114 * builtins.c (expand_builtin_strnlen): New function.
16115 (expand_builtin): Call it.
16116 (fold_builtin_n): Avoid setting TREE_NO_WARNING.
16117 * builtins.def (BUILT_IN_STRNLEN): New.
16118 * calls.c (maybe_warn_nonstring_arg): Handle BUILT_IN_STRNLEN.
16119 Warn for bounds in excess of maximum object size.
16120 * tree-ssa-strlen.c (maybe_set_strlen_range): Return tree representing
16121 single-value ranges. Handle strnlen.
16122 (handle_builtin_strlen): Handle strnlen.
16123 (strlen_check_and_optimize_stmt): Same.
16124 * doc/extend.texi (Other Builtins): Document strnlen.
16126 2018-06-18 Maya Rashish <coypu@sdf.org>
16128 * config/alpha/openbsd.h (TARGET_DEFAULT): Define.
16129 (LINK_SPEC, STARTFILE_SPEC, ENDFILE_SPEC): Likewise.
16130 (INTMAX_TYPE, UINTMAX_TYPE, WINT_TYPE): Likewise.
16132 * config/alpha/elf.h (STARTFILE_SPEC, ENDFILE_SPEC): Move from
16134 * config/alpha/linux.h (STARTFILE_SPEC, ENDFILE_SPEC): Here.
16136 2018-06-18 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
16138 * tree.c (escaped_string::escape): Replace cast to char * by
16139 const_cast<char *> (unescaped).
16141 2018-06-18 Nick Clifton <nickc@redhat.com>
16144 * tree.c (escaped_string): New class. Converts an unescaped
16145 string into its escaped equivalent.
16146 (warn_deprecated_use): Use the new class to convert the
16147 deprecation message, if present.
16148 (test_escaped_strings): New self test.
16149 (test_c_tests): Add test_escaped_strings.
16150 * doc/extend.texi (deprecated): Add a note that the
16151 deprecation message is affected by the -fmessage-length
16152 option, and that control characters will be escaped.
16153 (#pragma GCC error): Document this pragma.
16154 (#pragma GCC warning): Likewise.
16155 * doc/invoke.texi (-fmessage-length): Document this option's
16156 effect on the #warning and #error preprocessor directives and
16157 the deprecated attribute.
16159 2018-06-18 Eric Botcazou <ebotcazou@adacore.com>
16161 * tree.c (decl_value_expr_lookup): Revert latest change.
16162 (decl_value_expr_insert): Likewise.
16164 2018-06-17 Eric Botcazou <ebotcazou@adacore.com>
16166 * gimplify.c (nonlocal_vlas): Delete.
16167 (nonlocal_vla_vars): Likewise.
16168 (gimplify_var_or_parm_decl): Do not add debug VAR_DECLs for non-local
16170 (gimplify_body): Do not create and destroy nonlocal_vlas.
16171 * tree-nested.c: Include diagnostic.h.
16172 (use_pointer_in_frame): Tweak.
16173 (lookup_field_for_decl): Add assertion and declare the transformation.
16174 (convert_nonlocal_reference_op) <PARM_DECL>: Rework and issue an
16175 internal error when the reference is in a wrong context. Do not
16176 create a debug decl by default.
16177 (note_nonlocal_block_vlas): Delete.
16178 (convert_nonlocal_reference_stmt) <GIMPLE_BIND>: Do not call it.
16179 (convert_local_reference_op) <PARM_DECL>: Skip the frame decl. Do not
16180 create a debug decl by default.
16181 (convert_gimple_call) <GIMPLE_CALL>: Issue an internal error when the
16182 call is in a wrong context.
16183 (fixup_vla_decls): New function.
16184 (finalize_nesting_tree_1): Adjust comment. Call fixup_vla_decls if no
16185 debug variables were created.
16186 * tree.c (decl_value_expr_lookup): Add checking assertion.
16187 (decl_value_expr_insert): Likewise.
16189 2018-06-16 Kugan Vivekanandarajah <kuganv@linaro.org>
16191 PR middle-end/82479
16192 * ipa-fnsummary.c (will_be_nonconstant_expr_predicate): Handle CALL_EXPR.
16193 * tree-scalar-evolution.c (interpret_expr): Likewise.
16194 (expression_expensive_p): Likewise.
16195 * tree-ssa-loop-ivopts.c (contains_abnormal_ssa_name_p): Likewise.
16196 * tree-ssa-loop-niter.c (number_of_iterations_popcount): New.
16197 (number_of_iterations_exit_assumptions): Use number_of_iterations_popcount.
16198 (ssa_defined_by_minus_one_stmt_p): New.
16200 2018-06-16 Kugan Vivekanandarajah <kuganv@linaro.org>
16202 PR middle-end/64946
16203 * cfgexpand.c (expand_debug_expr): Hande ABSU_EXPR.
16204 * config/i386/i386.c (ix86_add_stmt_cost): Likewise.
16205 * dojump.c (do_jump): Likewise.
16206 * expr.c (expand_expr_real_2): Check operand type's sign.
16207 * fold-const.c (const_unop): Handle ABSU_EXPR.
16208 (fold_abs_const): Likewise.
16209 * gimple-pretty-print.c (dump_unary_rhs): Likewise.
16210 * gimple-ssa-backprop.c (backprop::process_assign_use): Likesie.
16211 (strip_sign_op_1): Likesise.
16212 * match.pd: Add new pattern to generate ABSU_EXPR.
16213 * optabs-tree.c (optab_for_tree_code): Handle ABSU_EXPR.
16214 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
16215 * tree-eh.c (operation_could_trap_helper_p): Likewise.
16216 * tree-inline.c (estimate_operator_cost): Likewise.
16217 * tree-pretty-print.c (dump_generic_node): Likewise.
16218 * tree-vect-patterns.c (vect_recog_sad_pattern): Likewise.
16219 * tree.def (ABSU_EXPR): New.
16221 2018-06-16 Jakub Jelinek <jakub@redhat.com>
16223 PR middle-end/86095
16224 * common.opt (Wunsafe-loop-optimizations): Add Ignore, remove Var,
16225 documented as preserved for backward compatibility only.
16226 * doc/invoke.texi: Remove -Wunsafe-loop-optimizations documentation.
16228 PR rtl-optimization/86108
16229 * bb-reorder.c (create_forwarder_block): Renamed to ...
16230 (create_eh_forwarder_block): ... this. Split OLD_BB after labels and
16231 jump from new landing pad to the second part.
16232 (sjlj_fix_up_crossing_landing_pad, dw2_fix_up_crossing_landing_pad):
16235 2018-06-15 Jakub Jelinek <jakub@redhat.com>
16237 PR middle-end/85878
16238 * expr.c (expand_assignment): Remove now redundant COMPLEX_MODE_P
16239 check from first store_expr, use to_mode instead of GET_MODE (to_rtx).
16240 Only call store_expr for halves if the mode is the same.
16242 PR middle-end/86123
16243 * match.pd ((X / Y) == 0 -> X < Y): Don't transform complex divisions.
16244 Fix up comment formatting.
16246 2018-06-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
16248 * typed-splay-tree.h (typed_splay_tree::remove): New function.
16249 (typed_splay_tree::closure,
16250 typed_splay_tree::inner_foreach_fn, typed_splay_tree::m_inner): Deleted.
16251 (typed_splay_tree::typed_splay_tree,
16252 typed_splay_tree::operator =): Declared private.
16253 (typed_splay_tree::splay_tree_key, typed_splay_tree::splay_tree_value,
16254 typed_splay_tree::splay_tree_node_s, typed_splay_tree::KDEL,
16255 typed_splay_tree::VDEL, typed_splay_tree::splay_tree_delete_helper,
16256 typed_splay_tree::rotate_left, typed_splay_tree::rotate_right,
16257 typed_splay_tree::splay_tree_splay,
16258 typed_splay_tree::splay_tree_foreach_helper,
16259 typed_splay_tree::splay_tree_insert,
16260 typed_splay_tree::splay_tree_remove,
16261 typed_splay_tree::splay_tree_lookup,
16262 typed_splay_tree::splay_tree_predecessor,
16263 typed_splay_tree::splay_tree_successor,
16264 typed_splay_tree::splay_tree_min,
16265 typed_splay_tree::splay_tree_max): Took over from splay-tree.c/.h.
16266 (typed_splay_tree::root, typed_splay_tree::comp,
16267 typed_splay_tree::delete_key,
16268 typed_splay_tree::delete_value): New data members.
16269 * typed-splay-tree.c (selftest::test_str_to_int): Add a test for
16270 typed_splay_tree::remove.
16272 2018-06-15 Matthew Fortune <matthew.fortune@mips.com>
16274 * config/mips/mips.h (ASM_SPEC): Pass through -mcrc, -mno-crc,
16275 -mginv and -mno-ginv to the assembler.
16276 * config/mips/mips.opt (-mcrc): New option.
16277 (-mginv): Likewise.
16278 * doc/invoke.text (-mcrc): Document.
16279 (-mginv): Likewise.
16281 2018-06-15 Nick Clifton <nickc@redhat.com>
16284 * tree.c (escaped_string): New class. Converts an unescaped
16285 string into its escaped equivalent.
16286 (warn_deprecated_use): Use the new class to convert the
16287 deprecation message, if present.
16288 (test_escaped_strings): New self test.
16289 (test_c_tests): Add test_escaped_strings.
16290 * doc/extend.texi (deprecated): Add a note that the
16291 deprecation message is affected by the -fmessage-length
16292 option, and that control characters will be escaped.
16293 (#pragma GCC error): Document this pragma.
16294 (#pragma GCC warning): Likewise.
16295 * doc/invoke.texi (-fmessage-length): Document this option's
16296 effect on the #warning and #error preprocessor directives and
16297 the deprecated attribute.
16299 2018-06-15 Richard Biener <rguenther@suse.de>
16301 * tree-vect-slp.c (vect_slp_bb): Dump MSG_OPTIMIZED_LOCATIONS
16302 here, also noting vector size used.
16303 * tree-vectorizer.c (vectorize_loops): Adjust. Note vector
16304 size used in MSG_OPTIMIZED_LOCATIONS dump.
16305 (pass_slp_vectorize::execute): Adjust.
16307 2018-06-15 Claudiu Zissulescu <claziss@synopsys.com>
16310 * config/arc/arc.c (arc_return_address_register): Fix
16313 2018-06-15 Richard Biener <rguenther@suse.de>
16315 PR middle-end/86159
16316 * tree-cfg.c (gimplify_build3): Do not strip sign conversions,
16317 leave useless conversion stripping to force_gimple_operand_gsi.
16318 (gimplify_build2): Likewise.
16319 (gimplify_build1): Likewise.
16321 2018-06-15 Richard Biener <rguenther@suse.de>
16323 PR middle-end/86076
16324 * tree-cfg.c (move_stmt_op): unshare invariant addresses
16325 before adjusting their block.
16327 2018-06-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
16329 * config.gcc (riscv*-*-elf* | riscv*-*-rtems*): Use custom
16330 multilibs for *-*-rtems*.
16331 * config/riscv/t-rtems: New file.
16333 2018-06-14 Jakub Jelinek <jakub@redhat.com>
16335 PR middle-end/86122
16336 * match.pd ((A +- CST1) +- CST2): Punt if last resort
16337 unsigned_type_for returns NULL.
16340 * lower-subreg.c (find_decomposable_subregs): Don't decompose float
16341 subregs of multi-word pseudos unless the float mode has word size.
16343 2018-06-14 Richard Biener <rguenther@suse.de>
16345 PR middle-end/86139
16346 * tree-vect-generic.c (build_word_mode_vector_type): Remove
16347 duplicate and harmful type_hash_canon.
16348 * tree.c (type_hash_canon): Assert we didn't find ourselves.
16350 2018-06-14 Richard Biener <rguenther@suse.de>
16353 * tree-ssa-struct-alias.c (create_variable_info_for): Handle
16356 2018-06-14 Sebastian Huber <sebastian.huber@embedded-brains.de>
16358 * config/rtems.h (STDINT_LONG32): Define.
16360 2018-06-13 Matthew Fortune <matthew.fortune@mips.com>
16361 Prachi Godbole <prachi.godbole@imgtec.com>
16363 * config/mips/mips-cpus.def: Define P6600.
16364 * config/mips/mips-tables.opt: Regenerate.
16365 * config/mips/mips.c (mips_ucbranch_type): New enum.
16366 (mips_rtx_cost_data): Add support for P6600.
16367 (mips_issue_rate): Likewise.
16368 (mips_multipass_dfa_lookahead): Likewise.
16369 (mips_avoid_hazard): Likewise.
16370 (mips_reorg_process_insns): Likewise.
16371 (mips_classify_branch_p6600): New function.
16372 * config/mips/mips.h (TUNE_P6600): New define.
16373 (MIPS_ISA_LEVEL_SPEC): Infer mips64r6 from p6600.
16374 (ENABLE_LD_ST_PAIRS): Enable load/store bonding for p6600.
16375 * config/mips/mips.md: Include p6600.md.
16376 (processor): Add p6600.
16377 * config/mips/p6600.md: New file.
16378 * doc/invoke.texi: Add p6600 to supported architectures.
16380 2018-06-13 Martin Sebor <msebor@redhat.com>
16382 PR tree-optimization/86114
16383 * gimple-fold.c (gimple_fold_builtin_strlen): Only handle LHS
16385 * tree-ssa-strlen.c (maybe_set_strlen_range): Same.
16387 2018-06-13 Richard Biener <rguenther@suse.de>
16389 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern):
16390 Properly set vector type of the intermediate stmt.
16391 * tree-vect-stmts.c (vectorizable_operation): The destination
16392 var always has vectype_out type.
16394 2018-06-13 Jeff Law <law@redhat.com>
16396 * config/rl78/rl78.c (move_elim_pass): Use TDF_NONE rather than
16397 integer 0 for argument to print_rtl_with_bb.
16398 (rl78_reorg): Likewise.
16400 2018-06-13 David Malcolm <dmalcolm@redhat.com>
16402 * config/arc/arc.c (hwloop_optimize): Strengthen local "end_label"
16403 from rtx to rtx_insn *.
16404 * config/bfin/bfin.c (hwloop_optimize): Likewise for local
16406 (add_sched_insns_for_speculation): Likewise for local "target",
16407 converting usage of JUMP_LABEL to JUMP_LABEL_AS_INSN.
16408 * config/c6x/c6x.c (reorg_split_calls): Strengthen param "call_labels"
16409 from rtx_insn ** to rtx_code_label **.
16410 (reorg_emit_nops): Likewise.
16411 (c6x_reorg): Likewise for local "call_labels".
16412 * config/sh/sh-protos.h (get_dest_uid): Strengthen 1st param from
16414 * config/sh/sh.c (dump_table): Strengthen local "lab" from rtx to
16415 rtx_code_label *, adding safe_as_a <rtx_code_label *> casts to
16416 the loops over LABEL_REFS.
16417 (fixup_addr_diff_vecs): Add as_a <rtx_insn *> to usage of
16419 (barrier_align): Convert usage of JUMP_LABEL to JUMP_LABEL_AS_INSN.
16420 (get_dest_uid): Strengthen param "label" from rtx to rtx_insn *.
16421 (split_branches): Strengthen local "olabel" from rtx to
16422 rtx_insn *, adding a safe_as_a cast.
16423 * emit-rtl.c (next_real_insn): Strengthen param from "rtx"
16425 (add_insn_after): Likewise for first two params.
16426 (add_insn_before): Likewise.
16427 (remove_insn): Likewise for param.
16428 (emit_pattern_before_noloc): Likewise for second and third params.
16429 (emit_jump_insn_before_noloc): Convert NULL_RTX to NULL.
16430 (emit_call_insn_before_noloc): Likewise.
16431 (emit_debug_insn_before_noloc): Strengthen "before" param from "rtx"
16433 (emit_barrier_before): Likewise.
16434 (emit_label_before): Strengthen "label" param from "rtx" to
16435 "rtx_code_label *". Strengthen "before" param from "rtx" to
16437 (emit_insn_after_1): Strengthen "after" param from "rtx" to
16439 (emit_pattern_after_noloc): Likewise.
16440 (emit_insn_after_noloc): Likewise.
16441 (emit_jump_insn_after_noloc): Likewise.
16442 (emit_call_insn_after_noloc): Likewise.
16443 (emit_debug_insn_after_noloc): Likewise.
16444 (emit_barrier_after): Likewise.
16445 (emit_label_after): Likewise for both params.
16446 (emit_pattern_after_setloc): Likewise for "after" param. Convert
16447 "loc" param from "int" to "location_t".
16448 (emit_insn_after_setloc): Likewise.
16449 (emit_jump_insn_after_setloc): Likewise.
16450 (emit_call_insn_after_setloc): Likewise.
16451 (emit_debug_insn_after_setloc): Likewise.
16452 (emit_pattern_before_setloc): Likewise for "before" param. Convert
16453 "loc" param from "int" to "location_t".
16454 (emit_pattern_before): Convert NULL_RTX to NULL.
16455 (emit_insn_before_setloc): Convert "loc" param from "int" to
16457 (emit_jump_insn_before_setloc): Likewise.
16458 (emit_call_insn_before_setloc): Likewise.
16459 (emit_debug_insn_before_setloc): Strengthen "before" param from rtx to
16460 rtx_insn *. Convert "loc" param from "int" to "location_t".
16461 * rtl.h (emit_insn_before_setloc, emit_jump_insn_before_setloc,
16462 emit_call_insn_before_setloc, emit_debug_insn_before_setloc):
16463 Convert 3rd param from "int" to "location_t".
16464 (emit_barrier_before, emit_barrier_after, next_real_insn):
16465 Strengthen param from rtx to rtx_insn *.
16466 (emit_label_before): Strengthen 1st param from "rtx" to
16467 "rtx_code_label *". Strengthen 2nd param from "rtx" to
16469 (emit_insn_after_noloc, emit_jump_insn_after_noloc,
16470 emit_call_insn_after_noloc, emit_debug_insn_after_noloc):
16471 Strengthen 2nd param from "rtx" to "rtx_insn *".
16472 (emit_insn_after_setloc, emit_jump_insn_after_setloc)
16473 emit_call_insn_after_setloc, emit_debug_insn_after_setloc):
16474 Likewise. Convert 3rd param from "int" to "location_t".
16475 (emit_label_after): Strengthen 1st param from "rtx" to
16476 "rtx_code_label *".
16477 (next_real_insn, remove_insn): Strengthen param from "rtx" to
16479 (add_insn_before, add_insn_after): Strengthen 1st and 2nd params
16480 from "rtx" to "rtx_insn *".
16482 2018-06-13 Jan Hubicka <hubicka@gcc.gnu.org>
16484 * cgraph.c (cgraph_node::get_untransformed_body): Dump function
16485 bodies streamed in with -Q.
16486 * dumpfile.c (dump_files): Add lto-stream-out dump file.
16487 * dumpfile.h (tree_dump_index): Add lto_stream_out.
16488 * gimple-streamer-out.c: Include gimple-pretty-print.h
16489 (output_bb): Dump stmts streamed.
16490 * lto-section-out.c: Include print-tree.h
16491 (lto_begin_section): Dump sections created.
16492 (lto_output_decl_index): Dump decl encoded.
16493 * lto-streamer-out.c: Include print-tree.h
16494 (create_output_block): Dump output block created.
16495 (DFS::DFS_write_tree_body): Dump DFS SCCs streamed.
16496 (output_function): Dump function output.
16497 (output_constructor): Dump constructor streamed.
16498 (write_global_stream): Output indexes encoded.
16499 (produce_asm_for_decls): Dump streams encoded.
16500 * lto-streamer.c (streamer_dump_file): New global var.
16501 * lto-streamer.h (streamer_dump_file): Declare.
16502 * passes.c (ipa_write_summaries): Initialize streamer dump.
16503 * varpool.c (varpool_node::get_constructor): Dump constructors streamed
16506 2018-06-13 Eric Botcazou <ebotcazou@adacore.com>
16509 * config/i386/winnt.c (i386_pe_seh_cold_init): Do not emit negative
16510 offsets for register save directives. Emit a second batch of save
16511 directives, if need be, when the function accesses prior frames.
16513 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
16515 * config/arc/fpu.md (fmasf4): Force operand to register.
16516 (fnmasf4): Likewise.
16518 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
16520 * config/arc/arc-protos.h (arc_pad_return): Remove.
16521 * config/arc/arc.c (machine_function): Remove force_short_suffix
16523 (arc_print_operand): Adjust printing of '&'.
16524 (arc_verify_short): Remove conditional printing of short suffix.
16525 (arc_final_prescan_insn): Remove reference to size_reason.
16526 (pad_return): New function.
16527 (arc_reorg): Call pad_return.
16528 (arc_pad_return): Remove.
16529 (arc_init_machine_status): Remove reference to force_short_suffix.
16530 * config/arc/arc.md (vunspec): Add VUNSPEC_ARC_BLOCKAGE.
16531 (attr length): When attribute iscompact is true force to 2
16532 regardless; in the case of maybe check if we want to force the
16533 instruction to have 4 bytes length.
16534 (nopv): Change it to generate 4 byte long nop as well.
16535 (blockage): New pattern.
16536 (simple_return): Remove call to arc_pad_return.
16537 (p_return_i): Likewise.
16539 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
16541 * config/arc/elf.h (LINK_GCC_C_SEQUENCE_SPEC): Define.
16543 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
16545 * config/arc/builtins.def (SYNC): SYNC instruction is valid on all
16548 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
16550 * config/arc/arc.c (atomic_exchangesi): EX instruction is default
16551 for ARC700 and ARCv2.
16553 2018-06-13 Chenghua Xu <paul.hua.gm@gmail.com>
16556 * config/mips/loongson.md (vec_setv4hi): Gen_lowpart for
16557 operands[2] instead of operands[1].
16560 2018-06-12 Richard Sandiford <richard.sandiford@linaro.org>
16562 * lra-constraints.c (simplify_operand_subreg): In the paradoxical
16563 case, check whether the outer register overlaps an unallocatable
16564 register, not just whether it fits the required class.
16566 2018-06-12 Richard Sandiford <richard.sandiford@linaro.org>
16568 * poly-int.h (can_div_trunc_p): Add new overload in which all values
16570 * alias.c (get_addr): Extend CONST_INT handling to poly_int_rtx_p.
16571 (memrefs_conflict_p): Likewise.
16572 (init_alias_analysis): Likewise.
16573 * cfgexpand.c (expand_debug_expr): Likewise.
16574 * combine.c (combine_simplify_rtx, force_int_to_mode): Likewise.
16575 * cse.c (fold_rtx): Likewise.
16576 * explow.c (adjust_stack, anti_adjust_stack): Likewise.
16577 * expr.c (emit_block_move_hints): Likewise.
16578 (clear_storage_hints, push_block, emit_push_insn): Likewise.
16579 (store_expr_with_bounds, reduce_to_bit_field_precision): Likewise.
16580 (emit_group_load_1): Use rtx_to_poly_int64 for group offsets.
16581 (emit_group_store): Likewise.
16582 (find_args_size_adjust): Use strip_offset. Use rtx_to_poly_int64
16583 to read the PRE/POST_MODIFY increment.
16584 * calls.c (store_one_arg): Use strip_offset.
16585 * rtlanal.c (rtx_addr_can_trap_p_1): Extend CONST_INT handling to
16587 (set_noop_p): Use rtx_to_poly_int64 for the elements selected
16589 * simplify-rtx.c (avoid_constant_pool_reference): Use strip_offset.
16590 (simplify_binary_operation_1): Extend CONST_INT handling to
16592 * var-tracking.c (compute_cfa_pointer): Take a poly_int64 rather
16593 than a HOST_WIDE_INT.
16594 (hard_frame_pointer_adjustment): Change from HOST_WIDE_INT to
16596 (adjust_mems, add_stores): Update accodingly.
16597 (vt_canonicalize_addr): Track polynomial offsets.
16598 (emit_note_insn_var_location): Likewise.
16599 (vt_add_function_parameter): Likewise.
16600 (vt_initialize): Likewise.
16602 2018-06-12 Jeff Law <law@redhat.com>
16604 * config.gcc (alpha*-*-freebsd*): Remove.
16605 * config/alpha/freebsd.h: Remove.
16607 2018-06-12 David Malcolm <dmalcolm@redhat.com>
16610 * spellcheck-tree.c (levenshtein_distance): Rename to...
16611 (get_edit_distance): ...this, and update for underlying renaming.
16612 * spellcheck-tree.h (levenshtein_distance): Rename to...
16613 (get_edit_distance): ...this.
16614 * spellcheck.c (levenshtein_distance): Rename to...
16615 (get_edit_distance): ...this. Convert from Levenshtein distance
16616 to Damerau-Levenshtein distance by supporting transpositions of
16617 adjacent characters. Rename "v1" to "v_next" and "v0" to
16619 (selftest::levenshtein_distance_unit_test_oneway): Rename to...
16620 (selftest::test_edit_distance_unit_test_oneway): ...this, and
16621 update for underlying renaming.
16622 (selftest::levenshtein_distance_unit_test): Rename to...
16623 (selftest::test_get_edit_distance_unit): ...this, and update for
16624 underlying renaming.
16625 (selftest::test_find_closest_string): Add example from PR 69968
16626 where transposition helps
16627 (selftest::test_metric_conditions): Update for renaming.
16628 (selftest::test_metric_conditions): Likewise.
16629 (selftest::spellcheck_c_tests): Likewise.
16630 * spellcheck.h (levenshtein_distance): Rename both overloads to...
16631 (get_edit_distance): ...this.
16632 (best_match::consider): Update for renaming.
16634 2018-06-12 Martin Sebor <msebor@redhat.com>
16636 PR tree-optimization/85259
16637 * builtins.c (compute_objsize): Handle constant offsets.
16638 * gimple-ssa-warn-restrict.c (maybe_diag_offset_bounds): Return
16639 true iff a warning has been issued.
16640 * gimple.h (gimple_nonartificial_location): New function.
16641 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Call
16642 gimple_nonartificial_location and handle -Wno-system-headers.
16643 (handle_builtin_stxncpy): Same.
16645 2018-06-12 Martin Sebor <msebor@redhat.com>
16648 * fold-const.c (operand_equal_p): Handle SAVE_EXPR.
16650 2018-06-12 Will Schmidt <will_schmidt@vnet.ibm.com>
16652 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
16653 BUILTIN_VEC_XST entries for pointer to double and long long.
16655 2018-06-12 H.J. Lu <hongjiu.lu@intel.com>
16658 * config/i386/gnu-user.h (TARGET_THREAD_SPLIT_STACK_OFFSET):
16660 * config/i386/gnu-user64.h (TARGET_THREAD_SPLIT_STACK_OFFSET):
16663 2018-06-12 Martin Liska <mliska@suse.cz>
16665 * doc/options.texi: Document IntegerRange.
16667 2018-06-12 Martin Liska <mliska@suse.cz>
16669 * config/i386/i386.opt: Make MPX-related options as Deprecated.
16670 * opt-functions.awk: Handle Deprecated flag.
16671 * opts-common.c (decode_cmdline_option): Handle cl_deprecated
16673 (read_cmdline_option): Report warning for a deprecated option.
16674 * opts.h (struct cl_option): Add new field cl_deprecated.
16675 (CL_ERR_DEPRECATED): New.
16677 2018-06-12 Martin Liska <mliska@suse.cz>
16679 * doc/options.texi: Document Deprecated option flag.
16681 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
16683 * config/arc/arc-arch.h (arc_extras): New enum.
16684 (arc_cpu_t):Add field extra.
16685 (arc_cpu_types): Consider the extras.
16686 * config/arc/arc-cpus.def: Add extras info.
16687 * config/arc/arc-opts.h (processor_type): Consider extra field.
16688 * config/arc/arc.c (arc_override_options): Handle extra field.
16690 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
16692 * config/arc/arc-arch.h: Update ARC_OPTX macro.
16693 * config/arc/arc-options.def (ARC_OPTX): Introduce a new doc
16695 * config/arc/arc.c (arc_init): Update pic warning.
16696 (irq_range): Update irq range parsing warnings.
16697 (arc_override_options): Update various warning messages.
16698 (arc_handle_aux_attribute): Likewise.
16700 2018-06-12 Robert Suchanek <robert.suchanek@mips.com>
16702 * config/mips/i6400.md (i6400_fpu_fadd): Remove frint.
16704 2018-06-12 Jozef Lawrynowicz <jozef.l@mittosystems.com>
16706 * doc/sourcebuild.texi: Document usage of line number 0 in verify
16707 compiler messages directives.
16709 2018-06-12 Matthew Fortune <mfortune@gmail.com>
16711 * config/mips/mips-cpus.def: New MIPS_CPU for i6500.
16712 * config/mips/mips-tables.opt: Regenerate.
16713 * config/mips/mips.h (MIPS_ISA_LEVEL_SPEC): Mark i6500 as
16715 * doc/invoke.texi: Document -march=i6500.
16717 2018-06-12 Prachi Godbole <prachi.godbole@imgtec.com>
16719 * config/mips/i6400.md (i6400_gpmuldiv): Remove cpu_unit.
16720 (i6400_gpmul): Add cpu_unit.
16721 (i6400_gpdiv): Likewise.
16722 (i6400_msa_add_d): Update reservations.
16723 (i6400_msa_int_add) Likewise.
16724 (i6400_msa_short_logic3) Likewise.
16725 (i6400_msa_short_logic2) Likewise.
16726 (i6400_msa_short_logic) Likewise.
16727 (i6400_msa_move) Likewise.
16728 (i6400_msa_cmp) Likewise.
16729 (i6400_msa_short_float2) Likewise.
16730 (i6400_msa_div_d) Likewise.
16731 (i6400_msa_long_logic1) Likewise.
16732 (i6400_msa_long_logic2) Likewise.
16733 (i6400_msa_mult) Likewise.
16734 (i6400_msa_long_float2) Likewise.
16735 (i6400_msa_long_float4) Likewise.
16736 (i6400_msa_long_float5) Likewise.
16737 (i6400_msa_long_float8) Likewise.
16738 (i6400_fpu_fadd): Include frint type.
16739 (i6400_fpu_store): New define_insn_reservation.
16740 (i6400_fpu_load): Likewise.
16741 (i6400_fpu_move): Likewise.
16742 (i6400_fpu_fcmp): Likewise.
16743 (i6400_fpu_fmadd): Likewise.
16744 (i6400_int_mult): Include imul3nc type and update reservation.
16745 (i6400_int_div): Include idiv3 type and update reservation.
16746 (i6400_int_load): Update to check type not move_type.
16747 (i6400_int_store): Likewise.
16748 (i6400_int_prefetch): Set zero latency.
16750 2018-06-12 Eric Botcazou <ebotcazou@adacore.com>
16752 * gcc.c: Document new %@{...} sequence.
16753 (LINK_COMMAND_SPEC): Use it for the -L switches.
16754 (cpp_unique_options): Use it for the -I switches.
16755 (at_file_argbuf): New global variable.
16756 (in_at_file): Likewise.
16757 (alloc_args): Create at_file_argbuf.
16758 (clear_args): Truncate at_file_argbuf.
16759 (store_arg): If in_at_file, push the argument onto at_file_argbuf.
16760 (open_at_file): New function.
16761 (close_at_file): Likewise.
16762 (create_at_file): Delete.
16763 (do_spec_1) <'i'>: Use open_at_file/close_at_file.
16766 (validate_switches_from_spec): Deal with %@{...} sequence.
16767 (validate_switches): Likewise.
16768 (driver::finalize): Call clear_args.
16770 2018-06-11 Rasmus Villemoes <rasmus.villemoes@prevas.dk>
16772 * config/vx-common.h (USE_TM_CLONE_REGISTRY): #define to 0.
16774 2018-06-11 Martin Sebor <msebor@redhat.com>
16776 * doc/invoke.texi (-Wall): List -Wc++17-compat.
16777 (Wno-class-memaccess): Add @opindex.
16778 (Wno-templates, Wno-multiple-inheritance): Same.
16779 (Wno-virtual-inheritance, Wno-namespaces): Same.
16780 (Wno-pedantic, Wno-chkp, Wcoverage-mismatch): Same.
16781 (Wno-format-overflow, Wno-format-truncation): Same.
16782 (Wno-unused-local-typedefs, Walloc-size-larger-than): Same.
16783 (Wno-alloc-size-larger-than, Wframe-larger-than): Same
16784 (Wno-stack-usage, Wno-c++-compat, Wno-c++11-compat): Same.
16785 (Wno-c++14-compat, Wno-c++17-compat, Wno-openmp-simd): Same.
16786 (Wno-unsuffixed-float-constants, Wno-addr-space-convert): Same.
16787 (Wno-misspelled-isr): Same.
16789 2018-06-11 Martin Sebor <msebor@redhat.com>
16791 * PR tree-optimization/86083
16792 * tree-ssa-strlen.c (handle_char_store): Use tree_expr_nonzero_p.
16794 2018-06-11 Zhouyi Zhou <zhouzhouyi@gmail.com>
16796 * tree-eh.c (lower_eh_constructs_2): Add a comma to comment.
16798 2018-06-11 Segher Boessenkool <segher@kernel.crashing.org>
16801 * config/rs6000/rs6000.md (*movdi_internal32): Put constraint modifiers
16802 on the correct operand.
16803 (*movdi_internal64): Ditto.
16805 2018-06-11 Martin Liska <mliska@suse.cz>
16807 PR tree-optimization/86089
16808 * tree-ssa-strlen.c (get_string_length): Move back removed hunk.
16810 2018-06-11 Julia Koval <julia.koval@intel.com>
16812 * config/i386/vaesintrin.h (_mm_aesdec_epi128, _mm_aesdeclast_epi128)
16813 _mm_aesenc_epi128, _mm_aesenclast_epi128: Remove.
16814 * config/i386/vpclmulqdqintrin.h (_mm_clmulepi64_epi128): Remove.
16816 2018-06-11 Olivier Hainque <hainque@adacore.com>
16818 * dwarf2out.c (gen_compile_unit_die): Fallout to DW_LANG_Ada83
16819 for Ada with strict dwarf2.
16821 2018-06-08 Peter Bergner <bergner@vnet.ibm.com>
16824 * config/rs6000/rs6000.c (mem_operand_gpr): Enable PRE_INC and PRE_DEC
16827 2018-06-08 Jan Hubicka <hubicka@ucw.cz>
16829 * dumpfile.c (FIRST_ME_AUTO_NUMBERED_DUMP): Bump to 4.
16831 2018-06-08 David Edelsohn <dje.gcc@gmail.com>
16833 * config/rs6000/rs6000.c (rs6000_passes_ieee128): Protect with #if
16836 2018-06-08 Martin Liska <mliska@suse.cz>
16838 * tree-cfg.h (debug_function): Fix argument type to match
16841 2018-06-08 Martin Liska <mliska@suse.cz>
16843 * config/powerpcspe/powerpcspe.c (rs6000_xcoff_visibility):
16844 Remove usage of MPX-related (and removed) fields.
16845 * config/rs6000/rs6000.c (rs6000_xcoff_visibility): Likewise.
16847 2018-06-08 David Malcolm <dmalcolm@redhat.com>
16849 * cfg.c (debug): Use TDF_NONE rather than 0.
16850 * cfghooks.c (debug): Likewise.
16851 * dumpfile.c (DUMP_FILE_INFO): Likewise; also for OPTGROUP.
16852 (struct dump_option_value_info): Convert to...
16853 (struct kv_pair): ...this template type.
16854 (dump_options): Convert to kv_pair<dump_flags_t>; use TDF_NONE
16856 (optinfo_verbosity_options): Likewise.
16857 (optgroup_options): Convert to kv_pair<optgroup_flags_t>; use
16859 (gcc::dump_manager::dump_register): Use optgroup_flags_t rather
16860 than int for "optgroup_flags" param.
16861 (dump_generic_expr_loc): Use dump_flags_t rather than int for
16863 (dump_dec): Likewise.
16864 (dump_finish): Use TDF_NONE rather than 0.
16865 (gcc::dump_manager::opt_info_enable_passes): Use optgroup_flags_t
16866 rather than int for "optgroup_flags" param. Use TDF_NONE rather
16867 than 0. Update for change to option_ptr.
16868 (opt_info_switch_p_1): Convert "optgroup_flags" param from int *
16869 to optgroup_flags_t *. Use TDF_NONE and OPTGROUP_NONE rather than
16870 0. Update for changes to optinfo_verbosity_options and
16872 (opt_info_switch_p): Convert optgroup_flags from int to
16874 (dump_basic_block): Use dump_flags_t rather than int
16875 for "dump_kind" param.
16876 * dumpfile.h (TDF_ADDRESS, TDF_SLIM, TDF_RAW, TDF_DETAILS,
16877 TDF_STATS, TDF_BLOCKS, TDF_VOPS, TDF_LINENO, TDF_UID)
16878 TDF_STMTADDR, TDF_GRAPH, TDF_MEMSYMS, TDF_RHS_ONLY, TDF_ASMNAME,
16879 TDF_EH, TDF_NOUID, TDF_ALIAS, TDF_ENUMERATE_LOCALS, TDF_CSELIB,
16880 TDF_SCEV, TDF_GIMPLE, TDF_FOLDING, MSG_OPTIMIZED_LOCATIONS,
16881 MSG_MISSED_OPTIMIZATION, MSG_NOTE, MSG_ALL, TDF_COMPARE_DEBUG,
16882 TDF_NONE): Convert from macros to...
16883 (enum dump_flag): ...this new enum.
16884 (dump_flags_t): Update to use enum.
16885 (operator|, operator&, operator~, operator|=, operator&=):
16886 Implement for dump_flags_t.
16887 (OPTGROUP_NONE, OPTGROUP_IPA, OPTGROUP_LOOP, OPTGROUP_INLINE,
16888 OPTGROUP_OMP, OPTGROUP_VEC, OPTGROUP_OTHER, OPTGROUP_ALL):
16889 Convert from macros to...
16890 (enum optgroup_flag): ...this new enum.
16891 (optgroup_flags_t): New typedef.
16892 (operator|, operator|=): Implement for optgroup_flags_t.
16893 (struct dump_file_info): Convert field "alt_flags" to
16894 dump_flags_t. Convert field "optgroup_flags" to
16896 (dump_basic_block): Use dump_flags_t rather than int for param.
16897 (dump_generic_expr_loc): Likewise.
16898 (dump_dec): Likewise.
16899 (dump_register): Convert param "optgroup_flags" to
16901 (opt_info_enable_passes): Likewise.
16902 * early-remat.c (early_remat::dump_edge_list): Use TDF_NONE rather
16904 * gimple-pretty-print.c (debug): Likewise.
16905 * gimple-ssa-store-merging.c (bswap_replace): Likewise.
16906 (merged_store_group::apply_stores): Likewise.
16907 * gimple-ssa-strength-reduction.c (insert_initializers): Likewise.
16908 * gimple.c (verify_gimple_pp): Likewise.
16909 * graphite-poly.c (print_pbb_body): Likewise.
16910 * passes.c (pass_manager::register_one_dump_file): Convert
16911 local "optgroup_flags" to optgroup_flags_t.
16912 * print-tree.c (print_node): Use TDF_NONE rather than 0.
16914 (debug_body): Likewise.
16915 * tree-pass.h (struct pass_data): Convert field "optgroup_flags"
16916 to optgroup_flags_t.
16917 * tree-pretty-print.c (print_struct_decl): Use TDF_NONE rather
16919 * tree-ssa-math-opts.c (convert_mult_to_fma_1): Likewise.
16920 (convert_mult_to_fma): Likewise.
16921 * tree-ssa-reassoc.c (undistribute_ops_list): Likewise.
16922 * tree-ssa-sccvn.c (vn_eliminate): Likewise.
16923 * tree-vect-data-refs.c (dump_lower_bound): Convert param
16924 "dump_kind" to dump_flags_t.
16926 2018-06-08 Segher Boessenkool <segher@kernel.crashing.org>
16928 * config/rs6000/rs6000.c (min, max): Delete.
16930 2018-06-08 Segher Boessenkool <segher@kernel.crashing.org>
16932 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete mention of
16933 -mabi=spe and -mabi=no-spe.
16935 2018-06-08 Martin Liska <mliska@suse.cz>
16937 * ipa-pure-const.c (propagate_pure_const): Use ::get at places
16938 where we expect an existing summary.
16940 2018-06-08 Martin Liska <mliska@suse.cz>
16942 * ipa-inline-analysis.c (simple_edge_hints): Use ::get method.
16943 * ipa-inline.h (estimate_edge_growth): Likewise.
16945 2018-06-08 Martin Liska <mliska@suse.cz>
16947 * cgraph.c (function_version_hasher::hash): Use
16948 cgraph_node::get_uid ().
16949 (function_version_hasher::equal):
16950 * cgraph.h (cgraph_node::get_uid): New method.
16951 * ipa-inline.c (update_caller_keys): Use
16952 cgraph_node::get_uid ().
16953 (update_callee_keys): Likewise.
16954 * ipa-utils.c (searchc): Likewise.
16955 (ipa_reduced_postorder): Likewise.
16956 * lto-cgraph.c (input_node): Likewise.
16957 * passes.c (is_pass_explicitly_enabled_or_disabled): Likewise.
16958 * symbol-summary.h (symtab_insertion): Likewise.
16959 (symtab_removal): Likewise.
16960 (symtab_duplication): Likewise.
16961 * tree-pretty-print.c (dump_function_header): Likewise.
16962 * tree-sra.c (convert_callers_for_node): Likewise.
16964 2018-06-08 Martin Liska <mliska@suse.cz>
16966 * cgraph.c (symbol_table::create_edge): Always assign a new
16968 (symbol_table::free_edge): Do not recycle numbers.
16969 * cgraph.h (cgraph_edge::get): New method.
16970 * symbol-summary.h (symtab_removal): Use it.
16971 (symtab_duplication): Likewise.
16972 (call_summary::hashable_uid): Remove.
16974 2018-06-08 Martin Liska <mliska@suse.cz>
16976 * ipa-inline-analysis.c (inline_edge_removal_hook): Remove.
16977 (initialize_growth_caches): Remove.
16978 (free_growth_caches): Likewise.
16979 (do_estimate_edge_time): Use edge_growth_cache.
16980 (do_estimate_edge_size): Likewise.
16981 (do_estimate_edge_hints): Likewise.
16982 * ipa-inline.c (reset_edge_caches): Likewise.
16983 (recursive_inlining): Likewise.
16984 (inline_small_functions): Likewise.
16985 * ipa-inline.h (initialize_growth_caches): Remove.
16986 (estimate_edge_size): Likewise.
16987 (estimate_edge_time): Likewise.
16988 (estimate_edge_hints): Likewise.
16989 (reset_edge_growth_cache): Likewise.
16990 * symbol-summary.h (call_summary::remove): New method.
16992 2018-06-08 Martin Liska <mliska@suse.cz>
16994 * ipa-cp.c (class edge_clone_summary): New summary.
16995 (grow_edge_clone_vectors): Remove.
16996 (ipcp_edge_duplication_hook): Remove.
16997 (class edge_clone_summary_t): New call_summary class.
16998 (ipcp_edge_removal_hook): Remove.
16999 (edge_clone_summary_t::duplicate): New function.
17000 (get_next_cgraph_edge_clone): Use edge_clone_summaries.
17001 (create_specialized_node): Likewise.
17002 (ipcp_driver): Initialize edge_clone_summaries and do not
17005 2018-06-08 Martin Liska <mliska@suse.cz>
17007 * symbol-summary.h (get): New function.
17008 (call_summary::m_initialize_when_cloning): New class member.
17010 2018-06-08 Martin Liska <mliska@suse.cz>
17012 * cgraph.c (cgraph_node::remove): Do not recycle uid.
17013 * cgraph.h (symbol_table::release_symbol): Do not pass uid.
17014 (symbol_table::allocate_cgraph_symbol): Do not set uid.
17015 * passes.c (uid_hash_t): Record removed_nodes by their uids.
17016 (remove_cgraph_node_from_order): Use the removed_nodes set.
17017 (do_per_function_toporder): Likwise.
17018 * symbol-summary.h (symtab_insertion): Use cgraph_node::uid
17019 instead of summary_uid.
17020 (symtab_removal): Likewise.
17021 (symtab_duplication): Likewise.
17023 2018-06-08 Martin Liska <mliska@suse.cz>
17025 * ipa-cp.c (ipcp_store_bits_results): Use
17026 ipcp_transformation_sum.
17027 (ipcp_store_vr_results): Likewise.
17028 * ipa-prop.c (ipcp_grow_transformations_if_necessary): Renamed
17030 (ipcp_transformation_initialize): ... this.
17031 (ipa_set_node_agg_value_chain):
17032 (ipa_node_params_t::duplicate): Use ipcp_transformation_sum.
17033 (write_ipcp_transformation_info): Likewise.
17034 (read_ipcp_transformation_info): Likewise.
17035 (ipcp_update_bits): Likewise.
17036 (ipcp_update_vr): Likewise.
17037 (ipcp_transform_function): Likewise.
17038 * ipa-prop.h: Rename ipcp_transformation_summary to
17039 ipcp_transformation.
17040 (class ipcp_transformation_t): New function summary.
17041 (ipcp_get_transformation_summary): Use ipcp_transformation_sum.
17042 (ipa_get_agg_replacements_for_node): Likewise.
17044 2018-06-08 Martin Liska <mliska@suse.cz>
17046 * ipa-pure-const.c (struct funct_state_d): Do it class instead
17048 (class funct_state_summary_t): New function_summary class.
17049 (has_function_state): Remove.
17050 (get_function_state): Likewise.
17051 (set_function_state): Likewise.
17052 (add_new_function): Likewise.
17053 (funct_state_summary_t::insert): New function.
17054 (duplicate_node_data): Remove.
17055 (remove_node_data): Remove.
17056 (funct_state_summary_t::duplicate): New function.
17057 (register_hooks): Create new funct_state_summaries.
17058 (pure_const_generate_summary): Use it.
17059 (pure_const_write_summary): Likewise.
17060 (pure_const_read_summary): Likewise.
17061 (propagate_pure_const): Likewise.
17062 (propagate_nothrow): Likewise.
17063 (dump_malloc_lattice): Likewise.
17064 (propagate_malloc): Likewise.
17065 (execute): Do not register hooks, just remove summary
17067 (pass_ipa_pure_const::pass_ipa_pure_const): Simplify
17070 2018-06-08 Martin Liska <mliska@suse.cz>
17072 * ipa-reference.c (remove_node_data): Remove.
17073 (duplicate_node_data): Likewise.
17074 (class ipa_ref_var_info_summary_t): New class.
17075 (class ipa_ref_opt_summary_t): Likewise.
17076 (get_reference_vars_info): Use ipa_ref_var_info_summaries.
17077 (get_reference_optimization_summary): Use
17078 ipa_ref_opt_sum_summaries.
17079 (set_reference_vars_info): Remove.
17080 (set_reference_optimization_summary): Likewise.
17081 (ipa_init): Create summaries.
17082 (init_function_info): Use function summary.
17083 (ipa_ref_opt_summary_t::duplicate): New function.
17084 (ipa_ref_opt_summary_t::remove): New function.
17085 (get_read_write_all_from_node): Fix GNU coding style.
17086 (propagate): Use function summary.
17087 (write_node_summary_p): Fix GNU coding style.
17088 (stream_out_bitmap): Likewise.
17089 (ipa_reference_read_optimization_summary): Use function summary.
17090 (ipa_reference_c_finalize): Do not release hooks.
17092 2018-06-08 Martin Liska <mliska@suse.cz>
17094 * ipa-fnsummary.c (dump_ipa_call_summary): Use ::get method.
17095 (analyze_function_body): Extract multiple calls of get_create.
17096 * ipa-inline-analysis.c (simple_edge_hints): Likewise.
17097 * ipa-inline.c (recursive_inlining): Use ::get method.
17098 * ipa-inline.h (estimate_edge_growth): Likewise.
17100 2018-06-08 Martin Liska <mliska@suse.cz>
17102 * hsa-common.h (enum hsa_function_kind): Rename HSA_NONE to
17104 (hsa_function_summary::hsa_function_summary): Use the new enum
17106 (hsa_gpu_implementation_p): Use hsa_summaries::get.
17107 * hsa-gen.c (hsa_get_host_function): Likewise.
17108 (get_brig_function_name): Likewise.
17109 * ipa-hsa.c (process_hsa_functions): Likewise.
17110 (ipa_hsa_write_summary): Likewise.
17111 * symbol-summary.h (symtab_duplication): Use ::get function/
17112 (get): New function.
17114 2018-06-08 Martin Liska <mliska@suse.cz>
17116 * config/i386/i386.c (ix86_can_inline_p): Use get_create instead
17118 * hsa-common.c (hsa_summary_t::link_functions): Likewise.
17119 (hsa_register_kernel): Likewise.
17120 * hsa-common.h (hsa_gpu_implementation_p): Likewise.
17121 * hsa-gen.c (hsa_get_host_function): Likewise.
17122 (get_brig_function_name): Likewise.
17123 (generate_hsa): Likewise.
17124 (pass_gen_hsail::execute): Likewise.
17125 * ipa-cp.c (ipcp_cloning_candidate_p): Likewise.
17126 (devirtualization_time_bonus): Likewise.
17127 (ipcp_propagate_stage): Likewise.
17128 * ipa-fnsummary.c (redirect_to_unreachable): Likewise.
17129 (edge_set_predicate): Likewise.
17130 (evaluate_conditions_for_known_args): Likewise.
17131 (evaluate_properties_for_edge): Likewise.
17132 (ipa_fn_summary::reset): Likewise.
17133 (ipa_fn_summary_t::duplicate): Likewise.
17134 (dump_ipa_call_summary): Likewise.
17135 (ipa_dump_fn_summary): Likewise.
17136 (analyze_function_body): Likewise.
17137 (compute_fn_summary): Likewise.
17138 (estimate_edge_devirt_benefit): Likewise.
17139 (estimate_edge_size_and_time): Likewise.
17140 (estimate_calls_size_and_time): Likewise.
17141 (estimate_node_size_and_time): Likewise.
17142 (inline_update_callee_summaries): Likewise.
17143 (remap_edge_change_prob): Likewise.
17144 (remap_edge_summaries): Likewise.
17145 (ipa_merge_fn_summary_after_inlining): Likewise.
17146 (ipa_update_overall_fn_summary): Likewise.
17147 (read_ipa_call_summary): Likewise.
17148 (inline_read_section): Likewise.
17149 (write_ipa_call_summary): Likewise.
17150 (ipa_fn_summary_write): Likewise.
17151 (ipa_free_fn_summary): Likewise.
17152 * ipa-hsa.c (process_hsa_functions): Likewise.
17153 (ipa_hsa_write_summary): Likewise.
17154 (ipa_hsa_read_section): Likewise.
17155 * ipa-icf.c (sem_function::merge): Likewise.
17156 * ipa-inline-analysis.c (simple_edge_hints): Likewise.
17157 (do_estimate_edge_time): Likewise.
17158 (estimate_size_after_inlining): Likewise.
17159 (estimate_growth): Likewise.
17160 (growth_likely_positive): Likewise.
17161 * ipa-inline-transform.c (clone_inlined_nodes): Likewise.
17162 (inline_call): Likewise.
17163 * ipa-inline.c (caller_growth_limits): Likewise.
17164 (can_inline_edge_p): Likewise.
17165 (can_inline_edge_by_limits_p): Likewise.
17166 (compute_uninlined_call_time): Likewise.
17167 (compute_inlined_call_time): Likewise.
17168 (want_inline_small_function_p): Likewise.
17169 (edge_badness): Likewise.
17170 (update_caller_keys): Likewise.
17171 (update_callee_keys): Likewise.
17172 (recursive_inlining): Likewise.
17173 (inline_small_functions): Likewise.
17174 (inline_to_all_callers_1): Likewise.
17175 (dump_overall_stats): Likewise.
17176 (early_inline_small_functions): Likewise.
17177 (early_inliner): Likewise.
17178 * ipa-inline.h (estimate_edge_growth): Likewise.
17179 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
17180 * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
17181 * ipa-prop.h (IPA_NODE_REF): Likewise.
17182 (IPA_EDGE_REF): Likewise.
17183 * ipa-pure-const.c (malloc_candidate_p): Likewise.
17184 (propagate_malloc): Likewise.
17185 * ipa-split.c (execute_split_functions): Likewise.
17186 * symbol-summary.h: Rename get to get_create.
17188 (get_create): Likewise.
17189 * tree-sra.c (ipa_sra_preliminary_function_checks): Likewise.
17191 2018-06-08 Martin Liska <mliska@suse.cz>
17193 * symbol-summary.h (release): Move definition out of class
17195 (symtab_removal): Likewise.
17196 (symtab_duplication): Likewise.
17198 2018-06-08 Martin Liska <mliska@suse.cz>
17200 * symbol-summary.h (function_summary): Move constructor
17201 implementation out of class declaration.
17202 (release): Likewise.
17203 (symtab_insertion): Likewise.
17204 (symtab_removal): Likewise.
17205 (symtab_duplication): Likewise.
17208 2018-06-08 Martin Liska <mliska@suse.cz>
17210 * Makefile.in: Remove support for MPX (macros, related functions,
17211 fields in cgraph_node, ...).
17212 * builtin-types.def (BT_BND): Likewise.
17213 (BT_FN_BND_CONST_PTR): Likewise.
17214 (BT_FN_CONST_PTR_BND): Likewise.
17215 (BT_FN_VOID_PTR_BND): Likewise.
17216 (BT_FN_BND_CONST_PTR_SIZE): Likewise.
17217 (BT_FN_VOID_CONST_PTR_BND_CONST_PTR): Likewise.
17218 * builtins.c (expand_builtin_memcpy_with_bounds): Likewise.
17219 (expand_builtin_mempcpy_with_bounds): Likewise.
17220 (expand_builtin_memset_with_bounds): Likewise.
17221 (expand_builtin_memset_args): Likewise.
17222 (std_expand_builtin_va_start): Likewise.
17223 (expand_builtin): Likewise.
17224 (expand_builtin_with_bounds): Likewise.
17225 * builtins.def (DEF_BUILTIN_CHKP): Likewise.
17226 (DEF_LIB_BUILTIN_CHKP): Likewise.
17227 (DEF_EXT_LIB_BUILTIN_CHKP): Likewise.
17228 (DEF_CHKP_BUILTIN): Likewise.
17229 (BUILT_IN_MEMCPY): Likewise.
17230 (BUILT_IN_MEMMOVE): Likewise.
17231 (BUILT_IN_MEMPCPY): Likewise.
17232 (BUILT_IN_MEMSET): Likewise.
17233 (BUILT_IN_STPCPY): Likewise.
17234 (BUILT_IN_STRCAT): Likewise.
17235 (BUILT_IN_STRCHR): Likewise.
17236 (BUILT_IN_STRCPY): Likewise.
17237 (BUILT_IN_STRLEN): Likewise.
17238 (BUILT_IN_MEMCPY_CHK): Likewise.
17239 (BUILT_IN_MEMMOVE_CHK): Likewise.
17240 (BUILT_IN_MEMPCPY_CHK): Likewise.
17241 (BUILT_IN_MEMSET_CHK): Likewise.
17242 (BUILT_IN_STPCPY_CHK): Likewise.
17243 (BUILT_IN_STRCAT_CHK): Likewise.
17244 (BUILT_IN_STRCPY_CHK): Likewise.
17245 * calls.c (store_bounds): Likewise.
17246 (emit_call_1): Likewise.
17247 (special_function_p): Likewise.
17248 (maybe_warn_nonstring_arg): Likewise.
17249 (initialize_argument_information): Likewise.
17250 (finalize_must_preallocate): Likewise.
17251 (compute_argument_addresses): Likewise.
17252 (expand_call): Likewise.
17253 * cfgexpand.c (expand_call_stmt): Likewise.
17254 (expand_return): Likewise.
17255 (expand_gimple_stmt_1): Likewise.
17256 (pass_expand::execute): Likewise.
17257 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
17258 (cgraph_node::remove): Likewise.
17259 (cgraph_node::dump): Likewise.
17260 (cgraph_node::verify_node): Likewise.
17261 * cgraph.h (chkp_function_instrumented_p): Likewise.
17262 (symtab_node::get_alias_target): Likewise.
17263 (cgraph_node::can_remove_if_no_direct_calls_and_refs_p): Likewise.
17264 (cgraph_local_p): Likewise.
17265 * cgraphbuild.c (cgraph_edge::rebuild_edges): Likewise.
17266 (cgraph_edge::rebuild_references): Likewise.
17267 * cgraphunit.c (varpool_node::finalize_decl): Likewise.
17268 (walk_polymorphic_call_targets): Likewise.
17269 (cgraph_node::expand_thunk): Likewise.
17270 (symbol_table::output_weakrefs): Likewise.
17271 * common/config/i386/i386-common.c (OPTION_MASK_ISA2_GENERAL_REGS_ONLY_UNSET): Likewise.
17272 (ix86_handle_option): Likewise.
17273 * config/i386/constraints.md: Likewise.
17274 * config/i386/i386-builtin-types.def (BND): Likewise.
17278 * config/i386/i386-builtin.def (BDESC_END): Likewise.
17279 (BDESC_FIRST): Likewise.
17281 * config/i386/i386-c.c (ix86_target_macros_internal): Likewise.
17282 * config/i386/i386-protos.h (ix86_bnd_prefixed_insn_p): Likewise.
17283 * config/i386/i386.c (enum reg_class): Likewise.
17284 (ix86_target_string): Likewise.
17285 (ix86_option_override_internal): Likewise.
17286 (ix86_conditional_register_usage): Likewise.
17287 (ix86_valid_target_attribute_inner_p): Likewise.
17288 (ix86_set_indirect_branch_type): Likewise.
17289 (ix86_set_current_function): Likewise.
17290 (ix86_function_arg_regno_p): Likewise.
17291 (init_cumulative_args): Likewise.
17292 (ix86_function_arg_advance): Likewise.
17293 (ix86_function_arg): Likewise.
17294 (ix86_pass_by_reference): Likewise.
17295 (ix86_function_value_regno_p): Likewise.
17296 (ix86_function_value_1): Likewise.
17297 (ix86_function_value_bounds): Likewise.
17298 (ix86_return_in_memory): Likewise.
17299 (ix86_setup_incoming_vararg_bounds): Likewise.
17300 (ix86_va_start): Likewise.
17301 (indirect_thunk_need_prefix): Likewise.
17302 (print_reg): Likewise.
17303 (ix86_print_operand): Likewise.
17304 (ix86_expand_call): Likewise.
17305 (ix86_output_function_return): Likewise.
17306 (reg_encoded_number): Likewise.
17307 (BDESC_VERIFYS): Likewise.
17308 (ix86_init_mpx_builtins): Likewise.
17309 (ix86_init_builtins): Likewise.
17310 (ix86_emit_cmove): Likewise.
17311 (ix86_emit_move_max): Likewise.
17312 (ix86_expand_builtin): Likewise.
17313 (ix86_builtin_mpx_function): Likewise.
17314 (ix86_get_arg_address_for_bt): Likewise.
17315 (ix86_load_bounds): Likewise.
17316 (ix86_store_bounds): Likewise.
17317 (ix86_load_returned_bounds): Likewise.
17318 (ix86_store_returned_bounds): Likewise.
17319 (ix86_class_likely_spilled_p): Likewise.
17320 (ix86_hard_regno_mode_ok): Likewise.
17321 (x86_order_regs_for_local_alloc): Likewise.
17322 (ix86_mitigate_rop): Likewise.
17323 (ix86_bnd_prefixed_insn_p): Likewise.
17324 (ix86_mpx_bound_mode): Likewise.
17325 (ix86_make_bounds_constant): Likewise.
17326 (ix86_initialize_bounds): Likewise.
17327 (TARGET_LOAD_BOUNDS_FOR_ARG): Likewise.
17328 (TARGET_STORE_BOUNDS_FOR_ARG): Likewise.
17329 (TARGET_LOAD_RETURNED_BOUNDS): Likewise.
17330 (TARGET_STORE_RETURNED_BOUNDS): Likewise.
17331 (TARGET_CHKP_BOUND_MODE): Likewise.
17332 (TARGET_BUILTIN_CHKP_FUNCTION): Likewise.
17333 (TARGET_CHKP_FUNCTION_VALUE_BOUNDS): Likewise.
17334 (TARGET_CHKP_MAKE_BOUNDS_CONSTANT): Likewise.
17335 (TARGET_CHKP_INITIALIZE_BOUNDS): Likewise.
17336 * config/i386/i386.h (TARGET_MPX): Likewise.
17337 (TARGET_MPX_P): Likewise.
17338 (VALID_BND_REG_MODE): Likewise.
17339 (FIRST_BND_REG): Likewise.
17340 (LAST_BND_REG): Likewise.
17341 (enum reg_class): Likewise.
17342 (BND_REG_P): Likewise.
17343 (BND_REGNO_P): Likewise.
17344 (BNDmode): Likewise.
17345 (ADJUST_INSN_LENGTH): Likewise.
17346 * config/i386/i386.md: Likewise.
17347 * config/i386/i386.opt: Likewise.
17348 * config/i386/linux-common.h (LIBMPX_LIBS): Likewise.
17349 (defined): Likewise.
17350 (LINK_MPX): Likewise.
17351 (MPX_SPEC): Likewise.
17352 (LIBMPX_SPEC): Likewise.
17353 (LIBMPXWRAPPERS_SPEC): Likewise.
17354 (CHKP_SPEC): Likewise.
17355 * config/i386/predicates.md: Likewise.
17356 * dbxout.c (dbxout_type): Likewise.
17357 * doc/extend.texi: Likewise.
17358 * doc/invoke.texi: Likewise.
17359 * doc/md.texi: Likewise.
17360 * doc/tm.texi: Likewise.
17361 * doc/tm.texi.in: Likewise.
17362 * dwarf2out.c (is_base_type): Likewise.
17363 (gen_formal_types_die): Likewise.
17364 (gen_subprogram_die): Likewise.
17365 (gen_type_die_with_usage): Likewise.
17366 (gen_decl_die): Likewise.
17367 (dwarf2out_late_global_decl): Likewise.
17368 * expr.c (expand_assignment): Likewise.
17369 (emit_storent_insn): Likewise.
17370 (store_expr_with_bounds): Likewise.
17371 (store_expr): Likewise.
17372 (expand_expr_real_1): Likewise.
17373 * expr.h (store_expr_with_bounds): Likewise.
17374 * function.c (use_register_for_decl): Likewise.
17375 (struct bounds_parm_data): Likewise.
17376 (assign_parms_augmented_arg_list): Likewise.
17377 (assign_parm_find_entry_rtl): Likewise.
17378 (assign_parm_is_stack_parm): Likewise.
17379 (assign_parm_load_bounds): Likewise.
17380 (assign_bounds): Likewise.
17381 (assign_parms): Likewise.
17382 (expand_function_start): Likewise.
17383 * gcc.c (CHKP_SPEC): Likewise.
17384 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
17385 * gimple-ssa-warn-restrict.c (builtin_access::builtin_access): Likewise.
17386 (wrestrict_dom_walker::check_call): Likewise.
17387 * gimple.c (gimple_build_call_from_tree): Likewise.
17388 * gimple.h (enum gf_mask): Likewise.
17389 (gimple_call_with_bounds_p): Likewise.
17390 (gimple_call_set_with_bounds): Likewise.
17391 * gimplify.c (gimplify_init_constructor): Likewise.
17392 * ipa-cp.c (initialize_node_lattices): Likewise.
17393 (propagate_constants_across_call): Likewise.
17394 (find_more_scalar_values_for_callers_subset): Likewise.
17395 * ipa-hsa.c (process_hsa_functions): Likewise.
17396 * ipa-icf-gimple.c (func_checker::compare_gimple_call): Likewise.
17397 * ipa-icf.c (sem_function::merge): Likewise.
17398 * ipa-inline.c (early_inliner): Likewise.
17399 * ipa-pure-const.c (warn_function_noreturn): Likewise.
17400 (warn_function_cold): Likewise.
17401 (propagate_pure_const): Likewise.
17402 * ipa-ref.h (enum GTY): Likewise.
17403 * ipa-split.c (find_retbnd): Likewise.
17404 (consider_split): Likewise.
17405 (split_function): Likewise.
17406 * ipa-visibility.c (cgraph_externally_visible_p): Likewise.
17407 * ipa.c (walk_polymorphic_call_targets): Likewise.
17408 (symbol_table::remove_unreachable_nodes): Likewise.
17409 (process_references): Likewise.
17410 (cgraph_build_static_cdtor_1): Likewise.
17411 * lto-cgraph.c (lto_output_node): Likewise.
17412 (output_refs): Likewise.
17413 (compute_ltrans_boundary): Likewise.
17414 (input_overwrite_node): Likewise.
17415 (input_node): Likewise.
17416 (input_cgraph_1): Likewise.
17417 * params.def (PARAM_CHKP_MAX_CTOR_SIZE): Likewise.
17418 * passes.c (pass_manager::execute_early_local_passes): Likewise.
17419 (class pass_chkp_instrumentation_passes): Likewise.
17420 (make_pass_chkp_instrumentation_passes): Likewise.
17421 * passes.def: Likewise.
17422 * rtl.h (struct GTY): Likewise.
17423 (CALL_EXPR_WITH_BOUNDS_P): Likewise.
17424 * stor-layout.c (layout_type): Likewise.
17425 * symtab.c: Likewise.
17426 * target.def: Likewise.
17427 * targhooks.c (default_chkp_bound_type): Likewise.
17428 (default_chkp_bound_mode): Likewise.
17429 (default_builtin_chkp_function): Likewise.
17430 (default_chkp_function_value_bounds): Likewise.
17431 (default_chkp_make_bounds_constant): Likewise.
17432 (default_chkp_initialize_bounds): Likewise.
17433 * targhooks.h (default_chkp_bound_type): Likewise.
17434 (default_chkp_bound_mode): Likewise.
17435 (default_builtin_chkp_function): Likewise.
17436 (default_chkp_function_value_bounds): Likewise.
17437 (default_chkp_make_bounds_constant): Likewise.
17438 (default_chkp_initialize_bounds): Likewise.
17439 * toplev.c (compile_file): Likewise.
17440 (process_options): Likewise.
17441 * tree-core.h (DEF_BUILTIN): Likewise.
17442 (DEF_BUILTIN_CHKP): Likewise.
17443 * tree-inline.c (declare_return_variable): Likewise.
17444 (remap_gimple_stmt): Likewise.
17445 (copy_bb): Likewise.
17446 (initialize_inlined_parameters): Likewise.
17447 (expand_call_inline): Likewise.
17448 * tree-pass.h (make_pass_ipa_chkp_versioning): Likewise.
17449 (make_pass_ipa_chkp_early_produce_thunks): Likewise.
17450 (make_pass_ipa_chkp_produce_thunks): Likewise.
17451 (make_pass_chkp): Likewise.
17452 (make_pass_chkp_opt): Likewise.
17453 (make_pass_chkp_instrumentation_passes): Likewise.
17454 * tree-pretty-print.c (dump_generic_node): Likewise.
17455 * tree-ssa-ccp.c (insert_clobber_before_stack_restore): Likewise.
17456 * tree-ssa-dce.c (propagate_necessity): Likewise.
17457 (eliminate_unnecessary_stmts): Likewise.
17458 * tree-ssa-pre.c (create_expression_by_pieces): Likewise.
17459 * tree-ssa-sccvn.c (copy_reference_ops_from_call): Likewise.
17460 * tree-ssa-sccvn.h: Likewise.
17461 * tree-ssa-strlen.c (get_string_length): Likewise.
17462 (valid_builtin_call): Likewise.
17463 (adjust_last_stmt): Likewise.
17464 (handle_builtin_strchr): Likewise.
17465 (handle_builtin_strcpy): Likewise.
17466 (handle_builtin_stxncpy): Likewise.
17467 (handle_builtin_memcpy): Likewise.
17468 (handle_builtin_strcat): Likewise.
17469 (strlen_check_and_optimize_stmt): Likewise.
17470 * tree-stdarg.c (expand_ifn_va_arg_1): Likewise.
17471 * tree-streamer-in.c: Likewise.
17472 * tree-streamer.c (record_common_node): Likewise.
17473 * tree.c (tree_code_size): Likewise.
17474 (wide_int_to_tree_1): Likewise.
17475 (type_contains_placeholder_1): Likewise.
17476 (build_common_tree_nodes): Likewise.
17477 * tree.def (POINTER_BOUNDS_TYPE): Likewise.
17478 * tree.h (POINTER_BOUNDS_TYPE_P): Likewise.
17479 (POINTER_BOUNDS_P): Likewise.
17480 (BOUNDED_TYPE_P): Likewise.
17481 (BOUNDED_P): Likewise.
17482 (CALL_WITH_BOUNDS_P): Likewise.
17483 (pointer_bounds_type_node): Likewise.
17484 * value-prof.c (gimple_ic): Likewise.
17485 * var-tracking.c (vt_add_function_parameters): Likewise.
17486 * varasm.c (make_decl_rtl): Likewise.
17487 (assemble_start_function): Likewise.
17488 (output_constant): Likewise.
17489 (maybe_assemble_visibility): Likewise.
17490 * varpool.c (ctor_for_folding): Likewise.
17491 * chkp-builtins.def: Remove.
17492 * ipa-chkp.c: Remove.
17493 * ipa-chkp.h: Remove.
17494 * rtl-chkp.c: Remove.
17495 * rtl-chkp.h: Remove.
17496 * tree-chkp-opt.c: Remove.
17497 * tree-chkp.c: Remove.
17498 * tree-chkp.h: Remove.
17500 2018-06-07 Carl Love <cel@us.ibm.com>
17502 * config/rs6000/vsx.md (vextract_fp_from_shorth,
17503 vextract_fp_from_shortl): Add BE support.
17505 2018-06-07 Paul Koning <ni1d@arrl.net>
17507 * compare-elim.c (try_merge_compare): Don't merge compare if
17508 address contains a side effect.
17509 (try_eliminate_compare): Likewise.
17511 2018-06-07 Olga Makhotina <olga.makhotina@intel.com>
17513 * config.gcc: Support "tremont".
17514 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "tremont".
17515 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
17517 * config/i386/i386.c (m_TREMONT): Define.
17518 (processor_target_table): Add "tremont".
17519 (PTA_TREMONT): Define.
17520 (ix86_lea_outperforms): Add TARGET_TREMONT.
17521 (get_builtin_code_for_version): Handle PROCESSOR_TREMONT.
17522 (fold_builtin_cpu): Add M_INTEL_TREMONT, replace M_INTEL_GOLDMONT
17523 and M_INTEL_GOLDMONT_PLUS.
17524 (fold_builtin_cpu): Add "tremont".
17525 (ix86_add_stmt_cost): Add TARGET_TREMONT.
17526 (ix86_option_override_internal): Add "tremont".
17527 * config/i386/i386.h (processor_costs): Define TARGET_TREMONT.
17528 (processor_type): Add PROCESSOR_TREMONT.
17529 * config/i386/x86-tune.def: Add m_TREMONT.
17530 * doc/invoke.texi: Add tremont as x86 -march=/-mtune= CPU type.
17532 2018-06-07 Jozef Lawrynowicz <jozef.l@mittosystems.com>
17534 * config/msp430/msp430.c (msp430_mcu_name): Set the "i" in the
17535 symbol defined for msp430i* devices to be lower case.
17537 2018-06-07 Richard Biener <rguenther@suse.de>
17539 * graphite-sese-to-poly.c (extract_affine): Avoid unneded
17540 wrapping. Properly wrap the result of a BIT_NOT_EXPR.
17541 Properly wrap signed arithmetic if overflow wraps.
17543 2018-06-07 Jakub Jelinek <jakub@redhat.com>
17545 PR tree-optimization/69615
17546 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If rhs2 is lhs
17547 of a cast from a same precision integral SSA_NAME in a bb dominated
17548 by first_bb, retry with rhs2 set to the rhs1 of the cast. Don't emit
17549 cast to utype if rhs2 has already a compatible type.
17551 2018-06-07 Richard Biener <rguenther@suse.de>
17553 PR tree-optimization/85935
17554 * graphite-scop-detection.c (find_params_in_bb): Analyze
17555 condition operands with respect to the correct loop. Assert
17556 the analysis doesn't fail.
17558 2018-06-04 Carl Love <cel@us.ibm.com>
17560 * config/rs6000/vsx.md (first_match_index_<mode>): Calculate index
17561 using natural element order. Use gen_lshrsi3 instead of gen_ashrsi3
17562 as it is slightly cheaper.
17563 (first_match_or_eos_index_<mode>):
17564 Calculate index using natural element order.
17565 (first_match_index_<mode>):
17566 Calculate index using natural element order.
17567 (first_match_or_eos_index_<mode>):
17568 Calculate index using natural order.
17569 (define_insn vclzlsbb): Change to define_insn vclzlsbb_<mode>.
17570 for BE and LE modes.
17571 * config/rs6000/rs6000-c.c: Rename P9V_BUILTIN_VCLZLSBB,
17572 P9V_BUILTIN_VCLZLSBB_V16QI.
17573 * config/rs6000/rs6000-builtin.def: Make VCLZLSBB mode
17576 2018-06-06 Kelvin Nilsen <kelvin@gcc.gnu.org>
17578 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
17579 indentation and line wrap for many prototypes. Add missing
17580 @smallexample directives around block of prototypes for vec_xl and
17583 2018-06-05 Michael Meissner <meissner@linux.ibm.com>
17585 * config/rs6000/rs6000.c (rs6000_passes_ieee128): New boolean to
17586 track if we pass or return IEEE 128-bit floating point.
17587 (ieee128_mangling_gcc_8_1): New boolean to say whether to generate
17588 C++ mangling that is compatible with GCC 8.1.
17589 (TARGET_ASM_GLOBALIZE_DECL_NAME): Override target hook.
17590 (init_cumulative_args): Note if we pass or return IEEE 128-bit
17591 floating point types.
17592 (rs6000_function_arg_advance_1): Likewise.
17593 (rs6000_mangle_type): Optionally generate mangled names that match
17594 what GCC 8.1 generated for IEEE 128-bit floating point types.
17595 (rs6000_globalize_decl_name): If we have an external function that
17596 passes or returns IEEE 128-bit types, generate a weak reference
17597 from the mangled name used in GCC 8.1 to the current mangled
17599 (rs6000_init_builtins): Make __ibm128 use the long double type if
17600 long double is IBM extended double. Make __float128 use the long
17601 double type if long double is IEEE 128-bit.
17604 * config/rs6000/rs6000-builtin.def (BU_IBM128_2): New helper
17605 macro for __ibm128 built-in functions.
17606 (PACK_IF): Add __ibm128 pack/unpack functions.
17607 (UNPACK_IF): Likewise.
17608 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Do not
17609 enable long double built-in functions if long double is IEEE
17610 128-bit floating point.
17611 (rs6000_invalid_builtin): Update long double built-in function
17613 (rs6000_expand_builtin): For PACK_IF and UNPACK_IF built-in
17614 functions, adjust the built-in function to use the long double
17615 built-in function if __ibm128 and long double are the same type.
17616 * doc/extend.texi (PowerPC builtins): Update documention for
17617 __builtin_{,un}pack_longdouble. Add documentation for
17618 __builtin_{,un}pack_ibm128.
17620 2018-06-06 Jim Wilson <jimw@sifive.com>
17622 * config/riscv/riscv.c (enum riscv_privilege_levels): New.
17623 (struct machine_function): New field interrupt_mode.
17624 (riscv_handle_type_attribute): New function. Add forward declaration.
17625 (riscv_attribute_table) <interrupt>: Use riscv_handle_type_attribute.
17626 (riscv_expand_epilogue): Check interrupt_mode field.
17627 (riscv_set_current_function): Check interrupt attribute args and
17628 set interrupt_mode field.
17629 * config/riscv/riscv.md (UNSPECV_SRET, UNSPECV_URET): New.
17630 (riscv_sret, riscv_uret): New.
17631 * doc/extend.texi (RISC-V Function Attributes) <interrupt>: Document
17632 new arguments to interrupt attribute.
17634 2018-06-06 Peter Bergner <bergner@vnet.ibm.com>
17637 * /config/rs6000/rs6000.h (ASM_CPU_SPEC): Add support for -mpower9.
17638 Don't handle -mcpu=power8 if -mpower9-vector is also used.
17640 2018-06-06 Kelvin Nilsen <kelvin@gcc.gnu.org>
17642 * config/rs6000/rs6000-builtin.def (VSX_BUILTIN_VEC_LD,
17643 VSX_BUILTIN_VEC_ST): Add comment to explain non-traditional uses.
17644 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
17645 several redundant entries.
17647 2018-06-06 David Malcolm <dmalcolm@redhat.com>
17649 * config/i386/i386-protos.h (ix86_expand_call): Strengthen return
17650 type from "rtx" to "rtx_insn *".
17651 * config/i386/i386.c (ix86_expand_split_stack_prologue): Likewise
17652 for local "call_insn", removing cast.
17653 (ix86_expand_call): Likewise, introducing a "call_insn" local.
17655 2018-06-06 Eric Botcazou <ebotcazou@adacore.com>
17657 PR tree-optimization/86066
17658 * gimple-ssa-store-merging.c (process_store): Do not bypass BIT_NOT_EXPR
17659 for BIT_INSERT_EXPR stores.
17661 2018-06-06 Richard Biener <rguenther@suse.de>
17663 PR tree-optimization/86062
17664 * tree-ssa-sccvn.c (vn_reference_lookup_3): Handle arbitrary
17665 component refs ontop
17666 of to be offsetted base.
17668 2018-06-06 Jozef Lawrynowicz <jozef.l@mittosystems.com>
17670 * config/msp430/msp430.c (msp430_attr): Allow interrupt handlers
17671 to be static and remove check on interrupt attribute name.
17673 2018-06-05 Kelvin Nilsen <kelvin@gcc.gnu.org>
17675 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Remove
17676 volatile qualifier from vec_lvsl and vec_lvsr argument prototypes.
17678 2018-06-05 Steve Ellcey <sellcey@cavium.com>
17681 * config/aarch64/aarch64-protos.h (aarch64_err_no_fpadvsimd): Remove
17683 * config/aarch64/aarch64-protos..c (aarch64_err_no_fpadvsimd):
17684 Remove second argument, change how error is called.
17685 (aarch64_layout_arg): Remove second argument from
17686 aarch64_err_no_fpadvsimd call.
17687 (aarch64_init_cumulative_args): Ditto.
17688 (aarch64_gimplify_va_arg_expr): Ditto.
17689 * config/aarch64/aarch64.md (mov<mode>): Ditto.
17691 2018-06-05 Uros Bizjak <ubizjak@gmail.com>
17693 * config/i386/i386.md (simple_return_indirect_internal): New expander.
17694 (*simple_return_indirect_internal<mode>): Rename from
17695 simple_return_indirect_internal. Use W mode iterator.
17696 (rstorssp): New expander.
17697 (*rstorssp<mode>): Rename from rstorssp. Use P mode iterator.
17698 (clrssbsy): New expander.
17699 (*clrssbsy<mode>): Rename from clrssbsy. Use P mode iterator.
17701 2018-06-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
17703 * config/arm/arm_cmse.h (cmse_nsfptr_create): Change typeof to
17705 (cmse_check_pointed_object): Likewise.
17707 2018-06-05 Martin Liska <mliska@suse.cz>
17709 PR gcov-profile/47618
17710 * doc/invoke.texi: Document how -fprofile-dir format
17713 2018-06-05 Richard Biener <rguenther@suse.de>
17715 * tree-cfgcleanup.c (cleanup_control_flow_pre): For edge
17716 removal pretend DOM info isn't available so we do not update
17717 it and only remove edges, not dominated blocks. Actually free
17718 DOM info in case we removed something. Remove unreachable blocks.
17719 (mfb_keep_latches): Work with either DOM info or marked backedges.
17720 (cleanup_tree_cfg_noloop): Do not remove unreachable blocks
17721 first. Mark backedges if DOM info isn't available.
17722 (Re-)compute DOM info after cleanup_control_flow_pre.
17724 2018-06-05 Richard Biener <rguenther@suse.de>
17726 * tree-cfg.c (struct locus_discrim_map): Store line, not location.
17727 (locus_discrim_hasher::hash): Adjust.
17728 (locus_discrim_hasher::equal): Likewise.
17729 (next_discriminator_for_locus): Work on line directly.
17730 (same_line_p): Pass in expanded locus1 as well.
17731 (assign_discriminators): Avoid redundant location expansions.
17733 2018-06-05 Richard Biener <rguenther@suse.de>
17735 PR tree-optimization/86046
17736 * tree-ssa.c (maybe_optimize_var): Clear DECL_GIMPLE_REG_P
17737 if required after clearing TREE_ADDRESSABLE.
17739 2018-06-05 Richard Biener <rguenther@suse.de>
17741 PR tree-optimization/86047
17742 * tree-ssa-loop.c (for_each_index): Glob handling of all
17743 decls and constants and really handle all of them.
17745 2018-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17748 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
17749 qualifier_void_pointer and qualifier_const_void_pointer.
17750 (arm_ldc_qualifiers, arm_stc_qualifiers): Use the above.
17751 (arm_init_builtins): Handle the above.
17752 * config/arm/arm_acle.h (__arm_cdp, __arm_ldc, __arm_ldcl, __arm_stc,
17753 __arm_stcl, __arm_mcr, __arm_cdp2, __arm_ldc2, __arm_ldcl2, __arm_stc2,
17754 __arm_stcl2,__arm_mcr2, __arm_mcrr, __arm_mcrr2): Remove return for
17757 2018-06-05 Martin Liska <mliska@suse.cz>
17759 * auto-profile.c (read_autofdo_file): Do not use
17760 gcov_ctr_summary struct.
17761 (afdo_callsite_hot_enough_for_early_inline): Likewise.
17762 * coverage.c (struct counts_entry): Likewise.
17763 (read_counts_file): Read just single summary entry.
17764 (get_coverage_counts): Use gcov_summary struct.
17765 * coverage.h (get_coverage_counts): Likewise.
17766 * gcov-dump.c (dump_working_sets): Likewise.
17767 (tag_summary): Dump just single summary.
17768 * gcov-io.c (gcov_write_summary): Write just histogram
17770 (gcov_read_summary): Read just single summary.
17771 (compute_working_sets): Use gcov_summary struct.
17772 * gcov-io.h (GCOV_TAG_SUMMARY_LENGTH): Remove usage
17773 of GCOV_COUNTERS_SUMMABLE.
17774 (GCOV_COUNTERS_SUMMABLE): Remove.
17775 (GCOV_FIRST_VALUE_COUNTER): Replace with
17776 GCOV_COUNTER_V_INTERVAL.
17777 (struct gcov_ctr_summary): Remove.
17778 (struct gcov_summary): Directly use fields of former
17780 (compute_working_sets): Use gcov_summary struct.
17781 * gcov.c (read_count_file): Do not use ctrs fields.
17782 * lto-cgraph.c (merge_profile_summaries): Use gcov_summary
17784 * lto-streamer.h (struct GTY): Make profile_info gcov_summary
17786 * profile.c: Likewise.
17787 * profile.h: Likewise.
17789 2018-06-05 Martin Liska <mliska@suse.cz>
17791 PR gcov-profile/84846
17792 * gcov.c (output_lines): Print working directory only
17793 in intermediate format.
17795 2018-06-05 Andreas Krebbel <krebbel@linux.ibm.com>
17797 * config/s390/s390-builtin-types.def: Add void function type.
17798 * config/s390/s390-builtins.def: Use the function type for the
17801 2018-06-04 Jim Wilson <jimw@sifive.com>
17803 * config/riscv/riscv-protos.h (riscv_expand_epilogue): Change bool arg
17805 * config/riscv/riscv.c (riscv_for_each_saved_reg): New args epilogue
17806 and maybe_eh_return. Change regno to unsigned int. Use new args to
17807 handle EH_RETURN_DATA_REGNO registers properly.
17808 (riscv_expand_prologue): Pass new args to riscv_for_each_saved_reg.
17809 (riscv_expand_epilogue): Update comment. Change argument name and
17810 type. Update code to use new name and type. Pass new args to
17811 riscv_for_each_saved_reg. Only use EH_RETURN_STACKADJ_RTX when
17813 * config/riscv/riscv.md (NORMAL_RETURN): New.
17814 (SIBCALL_RETURN, EXCEPTION_RETURN): New.
17815 (epilogue, sibcall_epilogue): Update riscv_expand_epilogue arg.
17816 (eh_return): Call gen_eh_return_internal and emit barrier.
17817 (eh_return_internal): Call riscv_expand_epilogue.
17819 2018-06-04 Eric Botcazou <ebotcazou@adacore.com>
17821 * gimple-ssa-store-merging.c (struct merged_store_group): Move up
17822 bit_insertion field and declare can_be_merged_into method.
17823 (merged_store_group::can_be_merged_into): New method.
17824 (imm_store_chain_info::coalesce_immediate): Call it to decide whether
17825 consecutive non-overlapping stores can be merged. Turn MEM_REF stores
17826 into BIT_INSERT_EXPR stores if the group contains a non-MEM_REF store.
17828 2018-06-04 Richard Biener <rguenther@suse.de>
17830 PR tree-optimization/85955
17831 * builtins.c (fold_builtin_sincos): Convert pointers to
17832 destination to appropriate type before dereferencing.
17834 2018-06-04 Segher Boessenkool <segher@kernel.crashing.org>
17836 * config/rs6000/rs6000.md (abs<mode>2 for FLOAT128): Handle IFmode.
17838 2018-06-04 Richard Sandiford <richard.sandiford@linaro.org>
17840 * expr.c (expand_expr_real_1): Force the operand into memory if
17841 its TYPE_MODE is BLKmode and if there is no integer mode for
17842 the number of bits being extracted.
17844 2018-06-04 Jakub Jelinek <jakub@redhat.com>
17848 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>_1):
17849 Use vptestnm rather than vptestm in (=Yc,v,C) variant.
17851 2018-06-04 Richard Biener <rguenther@suse.de>
17853 * tree-cfgcleanup.c (cleanup_tree_cfg_1): Fold into...
17854 (cleanup_tree_cfg_noloop): ... single caller. Do
17855 start_recording_case_labels later.
17857 2018-06-04 Sebastian Peryt <sebastian.peryt@intel.com>
17859 * config/i386/cldemoteintrin.h: Change define from _X86INTRIN_H_INCLUDED
17860 to _IMMINTRIN_H_INCLUDED.
17861 * config/i386/pconfigintrin.h: Ditto.
17862 * config/i386/waitpkgintrin.h: Ditto.
17863 * config/i386/immintrin.h: Add includes for sgxintrin.h,
17864 pconfigintrin.h, waitpkgintrin.h and cldemoteintrin.h.
17865 * config/i386/x86intrin.h: Remove includes for mintrin.h, xmmintrin.h,
17866 emmintrin.h, pmmintrin.h, tmmintrin.h, smmintrin.h, wmmintrin.h,
17867 bmiintrin.h, bmi2intrin.h, lzcntintrin.h, sgxintrin.h, pconfigintrin.h,
17868 waitpkgintrin.h and cldemoteintrin.h.
17870 2018-06-04 Richard Biener <rguenther@suse.de>
17872 PR tree-optimization/86038
17873 * tracer.c (find_best_successor): Check probability for
17874 being initialized, bail out if not.
17876 2018-06-04 Richard Earnshaw <rearnsha@arm.com>
17879 * config/arm/arm-cpus.in (ALL_QUIRKS): Add xscale feature to the list
17880 of bits to ignore when comparing architectures.
17882 2018-06-04 Jakub Jelinek <jakub@redhat.com>
17884 PR tree-optimization/69615
17885 * fold-const.c (merge_ranges): If range1 is - [x, x] and x is the
17886 maximum or minimum of the type, try to merge it also as if
17887 range1 is + [-, x - 1] or + [x + 1, -].
17890 * tree.c (inchash::add_expr): Handle IDENTIFIER_NODE.
17892 2018-06-03 Eric Botcazou <ebotcazou@adacore.com>
17894 PR tree-optimization/86034
17895 * gimple-ssa-store-merging.c (output_merged_store): Convert the RHS to
17896 the unsigned bitfield type in a bit insertion sequence if it does not
17897 have a larger precision than the bitfield size.
17898 (process_store): Also bypass widening conversions for BIT_INSERT_EXPR.
17900 2018-06-03 Kito Cheng <kito.cheng@gmail.com>
17902 * config/nds32/nds32-peephole2.md: Add new patterns for code size.
17904 2018-06-03 Chung-Ju Wu <jasonwucj@gmail.com>
17906 * config/nds32/nds32-opts.h (nds32_arch_type): Add ARCH_V3J.
17907 * config/nds32/nds32.c (nds32_option_override): Consider ARCH_V3J.
17908 * config/nds32/nds32.h (TARGET_ISA_V3): Add ARCH_V3J checking.
17909 * config/nds32/nds32.opt (march): Add enum value Value(ARCH_V3J).
17911 2018-06-03 Chung-Ju Wu <jasonwucj@gmail.com>
17913 * common/config/nds32/nds32-common.c (nds32_option_optimization_table):
17914 Disable -fdelete-null-pointer-checks for ELF toolchain.
17916 2018-06-02 Chung-Ju Wu <jasonwucj@gmail.com>
17917 Kito Cheng <kito.cheng@gmail.com>
17919 * config.gcc (nds32*): Use nds32-linux.opt and nds32-elf.opt.
17920 (nds32le-*-*, nds32be-*-*): Integrate checking process.
17921 (nds32*-*-*): Add glibc and uclibc conditions.
17922 * common/config/nds32/nds32-common.c (nds32_except_unwind_info): New.
17923 (TARGET_EXCEPT_UNWIND_INFO): Define.
17924 * config/nds32/elf.h: New file.
17925 * config/nds32/linux.h: New file.
17926 * config/nds32/nds32-elf.opt: New file.
17927 * config/nds32/nds32-linux.opt: New file.
17928 * config/nds32/nds32-fp-as-gp.c
17929 (pass_nds32_fp_as_gp::gate): Consider TARGET_LINUX_ABI.
17930 * config/nds32/nds32.c (nds32_conditional_register_usage): Consider
17932 (nds32_asm_file_end): Ditto.
17933 (nds32_print_operand): Ditto.
17934 (nds32_insert_attributes): Ditto.
17935 (nds32_init_libfuncs): New function.
17936 (TARGET_HAVE_TLS): Define.
17937 (TARGET_INIT_LIBFUNCS): Define.
17938 * config/nds32/nds32.h (TARGET_DEFAULT_RELAX): Apply different relax
17940 (TARGET_ELF): Apply different mcmodel setting.
17941 (LINK_SPEC, LIB_SPEC, STARTFILE_SPEC, ENDFILE_SPEC): The content has
17942 been migrated into elf.h and linux.h files.
17943 * config/nds32/nds32.md (add_pc): Consider TARGET_LINUX_ABI.
17944 * config/nds32/nds32.opt (mvh): Consider TARGET_LINUX_ABI.
17945 (mcmodel): The content has been migrated into nds32-elf.opt and
17946 nds32-linux.opt files.
17947 * config/nds32/t-elf: New file.
17948 * config/nds32/t-linux: New file.
17950 2018-06-02 Chung-Ju Wu <jasonwucj@gmail.com>
17951 Shiva Chen <shiva0217@gmail.com>
17953 * config/nds32/constants.md (unspec_volatile_element): Add
17954 UNSPEC_VOLATILE_OMIT_FP_BEGIN and UNSPEC_VOLATILE_OMIT_FP_END.
17955 * config/nds32/nds32-fp-as-gp.c: New implementation of fp_as_gp
17957 * config/nds32/nds32-protos.h (nds32_naked_function_p): Declare.
17958 (make_pass_nds32_fp_as_gp): Declare.
17959 * config/nds32/nds32.c (nds32_register_passes): Add fp_as_gp as one
17961 (nds32_asm_function_end_prologue): Remove unused asm output.
17962 (nds32_asm_function_begin_epilogue): Remove unused asm output.
17963 (nds32_asm_file_start): Output necessary fp_as_gp information.
17964 (nds32_option_override): Adjust register usage.
17965 (nds32_expand_prologue): Consider fp_as_gp situation.
17966 (nds32_expand_prologue_v3push): Consider fp_as_gp situation.
17967 * config/nds32/nds32.md (prologue): Check fp_as_gp_p and naked_p.
17970 (simple_return): Ditto.
17971 (omit_fp_begin): Output special directive for fp_as_gp.
17972 (omit_fp_end): Output special directive for fp_as_gp.
17973 * config/nds32/nds32.opt (mfp-as-gp, mno-fp-as-gp, mforce-fp-as-gp,
17974 mforbid-fp-as-gp): New options.
17976 2018-06-01 Mark Wielaard <mark@klomp.org>
17978 * dwarf2out.c (dwarf2out_finish): Remove generation of
17979 DW_AT_loclists_base.
17981 2018-06-01 Eric Botcazou <ebotcazou@adacore.com>
17983 * gimple-ssa-store-merging.c: Include gimple-fold.h.
17984 (struct store_immediate_info): Document BIT_INSERT_EXPR stores.
17985 (struct merged_store_group): Add bit_insertion field.
17986 (dump_char_array): Use standard hexadecimal format.
17987 (merged_store_group::merged_store_group): Set bit_insertion to false.
17988 (merged_store_group::apply_stores): Use optimal buffer size. Deal
17989 with BIT_INSERT_EXPR stores. Move up code updating the mask and
17990 also print the mask in the dump file.
17991 (pass_store_merging::gate): Minor tweak.
17992 (imm_store_chain_info::coalesce_immediate): Fix wrong association
17993 of stores with groups in dump. Allow coalescing of BIT_INSERT_EXPR
17994 stores with INTEGER_CST stores.
17995 (count_multiple_uses) <BIT_INSERT_EXPR>: New case.
17996 (imm_store_chain_info::output_merged_store): Add try_bitpos variable
17997 and use it throughout. Generate bit insertion sequences if need be.
17998 (pass_store_merging::process_store): Remove redundant condition.
17999 Record stores from a SSA name to a bit-field with BIT_INSERT_EXPR.
18001 2018-06-01 Segher Boessenkool <segher@kernel.crashing.org>
18003 * config/rs6000/rs6000.c (rs6000_mangle_type): Change the mangling of
18004 the 128-bit floating point types. Fix function comment.
18006 2018-06-01 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18008 * config/aarch64/aarch64-simd.md
18009 (aarch64_simd_vec_unpack<su>_lo_<mode>): Use UXTL and SXTL assembler
18011 (aarch64_simd_vec_unpack<su>_hi_<mode>): Use UXTL2 and SXTL2 assembler
18014 2018-06-01 Richard Sandiford <richard.sandiford@linaro.org>
18016 PR tree-optimization/85989
18017 * gimple-ssa-backprop.c (backprop::m_visited_phis): New member
18019 (backprop::intersect_uses): Check it when deciding whether this
18020 is a backedge reference.
18021 (backprop::process_block): Add each phi to m_visited_phis
18022 after visiting it, then clear it at the end.
18024 2018-06-01 Richard Biener <rguenther@suse.de>
18026 * tree-vectorizer.h (vect_dr_stmt): New function.
18027 (vect_get_load_cost): Adjust.
18028 (vect_get_store_cost): Likewise.
18029 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence):
18030 Use vect_dr_stmt instead of DR_SMTT.
18031 (vect_record_base_alignments): Likewise.
18032 (vect_calculate_target_alignment): Likewise.
18033 (vect_compute_data_ref_alignment): Likewise and make static.
18034 (vect_update_misalignment_for_peel): Likewise.
18035 (vect_verify_datarefs_alignment): Likewise.
18036 (vector_alignment_reachable_p): Likewise.
18037 (vect_get_data_access_cost): Likewise. Pass down
18038 vinfo to vect_get_load_cost/vect_get_store_cost instead of DR.
18039 (vect_get_peeling_costs_all_drs): Likewise.
18040 (vect_peeling_hash_get_lowest_cost): Likewise.
18041 (vect_enhance_data_refs_alignment): Likewise.
18042 (vect_find_same_alignment_drs): Likewise.
18043 (vect_analyze_data_refs_alignment): Likewise.
18044 (vect_analyze_group_access_1): Likewise.
18045 (vect_analyze_group_access): Likewise.
18046 (vect_analyze_data_ref_access): Likewise.
18047 (vect_analyze_data_ref_accesses): Likewise.
18048 (vect_vfa_segment_size): Likewise.
18049 (vect_small_gap_p): Likewise.
18050 (vectorizable_with_step_bound_p): Likewise.
18051 (vect_prune_runtime_alias_test_list): Likewise.
18052 (vect_analyze_data_refs): Likewise.
18053 (vect_supportable_dr_alignment): Likewise.
18054 * tree-vect-loop-manip.c (get_misalign_in_elems): Likewise.
18055 (vect_gen_prolog_loop_niters): Likewise.
18056 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
18057 * tree-vect-patterns.c (vect_recog_bool_pattern): Do not
18059 (vect_recog_mask_conversion_pattern): Likewise.
18060 (vect_try_gather_scatter_pattern): Likewise.
18061 * tree-vect-stmts.c (vect_model_store_cost): Pass stmt_info
18062 to vect_get_store_cost.
18063 (vect_get_store_cost): Get stmt_info instead of DR.
18064 (vect_model_load_cost): Pass stmt_info to vect_get_load_cost.
18065 (vect_get_load_cost): Get stmt_info instead of DR.
18067 2018-06-01 Richard Biener <rguenther@suse.de>
18069 PR middle-end/86017
18070 * gimple-fold.c (var_decl_component_p): Also allow offsetted
18071 vars wrapped in MEM_REFs.
18073 2018-06-01 Richard Sandiford <richard.sandiford@linaro.org>
18075 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
18076 Fix subreg tests so that we only return a choice between
18077 GENERAL_REGS and FP_REGS if the original classes included both.
18079 2018-06-01 Richard Biener <rguenther@suse.de>
18082 * tree-ssa-structalias.c (get_function_part_constraint):
18083 Handle NULL fi->decl.
18084 (find_func_aliases_for_call): Properly handle indirect
18085 fi from direct call.
18086 (find_func_clobbers): Likewise.
18087 (ipa_pta_execute): Likewise.
18088 (create_variable_info_for): For functions that are ifunc_resolver
18089 resolve to a varinfo that contains the result of the resolver call.
18090 (associate_varinfo_to_alias): Do not treat ifunc resolvers as
18093 2018-05-31 Michael Collison <michael.collison@arm.com>
18095 * config/aarch64/aarch64.md:
18096 (*fix_to_zero_extenddfdi2): New pattern.
18097 * gcc.target/aarch64/fix_extend1.c: New testcase.
18099 2018-05-31 Qing Zhao <qing.zhao@oracle.com>
18101 PR middle-end/78809
18102 PR middle-end/83026
18103 * builtins.c (expand_builtin): Add the handling of BUILT_IN_STRCMP_EQ
18104 and BUILT_IN_STRNCMP_EQ.
18105 * builtins.def: Add new builtins BUILT_IN_STRCMP_EQ and
18106 BUILT_IN_STRNCMP_EQ.
18107 * gimple-fold.c (gimple_fold_builtin_string_compare): Add the
18108 handling of BUILTIN_IN_STRCMP_EQ and BUILT_IN_STRNCMP_EQ.
18109 (gimple_fold_builtin): Likewise.
18110 * tree-ssa-strlen.c (compute_string_length): New function.
18111 (determine_min_obsize): New function.
18112 (handle_builtin_string_cmp): New function to handle calls to
18113 string compare functions.
18114 (strlen_optimize_stmt): Add handling to builtin string compare
18116 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
18117 Add the handling of BUILT_IN_STRCMP_EQ and BUILT_IN_STRNCMP_EQ.
18118 * tree.c (build_common_builtin_nodes): Add new defines of
18119 BUILT_IN_STRNCMP_EQ and BUILT_IN_STRCMP_EQ.
18121 2018-05-31 Jakub Jelinek <jakub@redhat.com>
18124 * bb-reorder.c (pass_partition_blocks::gate): Return false for
18125 functions with naked attribute.
18127 2018-05-31 Uros Bizjak <ubizjak@gmail.com>
18129 * config/i386/sse.md (avx_vec_concat<mode>):
18130 Substitute concat_tg_mode mode attribute with xtg_mode.
18131 (<mask_codefor>avx512dq_broadcast<mode><mask_name>_1): Ditto.
18132 (concat_tg_mode): Remove mode attribute.
18134 2018-05-31 Martin Sebor <msebor@redhat.com>
18137 * calls.c (alloc_max_size): Correct a logic error/typo.
18138 Treat excessive arguments as infinite. Warn for invalid arguments.
18139 * doc/invoke.texi (-Walloc-size-larger-than): Update.
18141 2018-05-31 H.J. Lu <hongjiu.lu@intel.com>
18144 * config/i386/x86-tune.def: Re-enable partial_reg_dependency
18145 and movx for Haswell.
18147 2018-05-31 Chung-Lin Tang <cltang@codesourcery.com>
18148 Cesar Philippidis <cesar@codesourcery.com>
18150 PR middle-end/85879
18151 * gimplify.c (gimplify_adjust_omp_clauses): Add 'remove = true'
18152 when emitting error on private/firstprivate reductions.
18153 * omp-low.c (lower_omp_target): Avoid reference-type processing
18154 on pointers for firstprivate clause.
18156 2018-05-31 Sameera Deshpande <sameera.deshpande@linaro.org>
18158 * config/aarch64/aarch64-simd-builtins.def (ld1x3): New.
18161 * config/aarch64/aarch64-simd.md
18162 (aarch64_ld1x3<VALLDIF:mode>): New pattern.
18163 (aarch64_ld1_x3_<mode>): Likewise
18164 (aarch64_st1x2<VALLDIF:mode>): Likewise
18165 (aarch64_st1_x2_<mode>): Likewise
18166 (aarch64_st1x3<VALLDIF:mode>): Likewise
18167 (aarch64_st1_x3_<mode>): Likewise
18168 * config/aarch64/arm_neon.h (vld1_u8_x3): New function.
18169 (vld1_s8_x3): Likewise.
18170 (vld1_u16_x3): Likewise.
18171 (vld1_s16_x3): Likewise.
18172 (vld1_u32_x3): Likewise.
18173 (vld1_s32_x3): Likewise.
18174 (vld1_u64_x3): Likewise.
18175 (vld1_s64_x3): Likewise.
18176 (vld1_f16_x3): Likewise.
18177 (vld1_f32_x3): Likewise.
18178 (vld1_f64_x3): Likewise.
18179 (vld1_p8_x3): Likewise.
18180 (vld1_p16_x3): Likewise.
18181 (vld1_p64_x3): Likewise.
18182 (vld1q_u8_x3): Likewise.
18183 (vld1q_s8_x3): Likewise.
18184 (vld1q_u16_x3): Likewise.
18185 (vld1q_s16_x3): Likewise.
18186 (vld1q_u32_x3): Likewise.
18187 (vld1q_s32_x3): Likewise.
18188 (vld1q_u64_x3): Likewise.
18189 (vld1q_s64_x3): Likewise.
18190 (vld1q_f16_x3): Likewise.
18191 (vld1q_f32_x3): Likewise.
18192 (vld1q_f64_x3): Likewise.
18193 (vld1q_p8_x3): Likewise.
18194 (vld1q_p16_x3): Likewise.
18195 (vld1q_p64_x3): Likewise.
18196 (vst1_s64_x2): Likewise.
18197 (vst1_u64_x2): Likewise.
18198 (vst1_f64_x2): Likewise.
18199 (vst1_s8_x2): Likewise.
18200 (vst1_p8_x2): Likewise.
18201 (vst1_s16_x2): Likewise.
18202 (vst1_p16_x2): Likewise.
18203 (vst1_s32_x2): Likewise.
18204 (vst1_u8_x2): Likewise.
18205 (vst1_u16_x2): Likewise.
18206 (vst1_u32_x2): Likewise.
18207 (vst1_f16_x2): Likewise.
18208 (vst1_f32_x2): Likewise.
18209 (vst1_p64_x2): Likewise.
18210 (vst1q_s8_x2): Likewise.
18211 (vst1q_p8_x2): Likewise.
18212 (vst1q_s16_x2): Likewise.
18213 (vst1q_p16_x2): Likewise.
18214 (vst1q_s32_x2): Likewise.
18215 (vst1q_s64_x2): Likewise.
18216 (vst1q_u8_x2): Likewise.
18217 (vst1q_u16_x2): Likewise.
18218 (vst1q_u32_x2): Likewise.
18219 (vst1q_u64_x2): Likewise.
18220 (vst1q_f16_x2): Likewise.
18221 (vst1q_f32_x2): Likewise.
18222 (vst1q_f64_x2): Likewise.
18223 (vst1q_p64_x2): Likewise.
18224 (vst1_s64_x3): Likewise.
18225 (vst1_u64_x3): Likewise.
18226 (vst1_f64_x3): Likewise.
18227 (vst1_s8_x3): Likewise.
18228 (vst1_p8_x3): Likewise.
18229 (vst1_s16_x3): Likewise.
18230 (vst1_p16_x3): Likewise.
18231 (vst1_s32_x3): Likewise.
18232 (vst1_u8_x3): Likewise.
18233 (vst1_u16_x3): Likewise.
18234 (vst1_u32_x3): Likewise.
18235 (vst1_f16_x3): Likewise.
18236 (vst1_f32_x3): Likewise.
18237 (vst1_p64_x3): Likewise.
18238 (vst1q_s8_x3): Likewise.
18239 (vst1q_p8_x3): Likewise.
18240 (vst1q_s16_x3): Likewise.
18241 (vst1q_p16_x3): Likewise.
18242 (vst1q_s32_x3): Likewise.
18243 (vst1q_s64_x3): Likewise.
18244 (vst1q_u8_x3): Likewise.
18245 (vst1q_u16_x3): Likewise.
18246 (vst1q_u32_x3): Likewise.
18247 (vst1q_u64_x3): Likewise.
18248 (vst1q_f16_x3): Likewise.
18249 (vst1q_f32_x3): Likewise.
18250 (vst1q_f64_x3): Likewise.
18251 (vst1q_p64_x3): Likewise.
18253 2018-05-30 Jozef Lawrynowicz <jozef.l@mittosystems.com>
18255 * config/msp430/msp430.c (msp430_output_labelref): Prepend
18256 user_label_prefix to name.
18258 * tree-core.h: Update comment about the format of NAME string
18259 passed to handler in attribute_spec.
18261 * config/msp430/msp430.md: Remove erroneous subreg expression from
18262 zero_extendqisi2 insn pattern. Remove msp430x ISA restriction on
18263 zero_extend{q,h}isi2.
18265 2018-05-30 Borislav Petkov <bp@suse.de>
18267 * doc/extend.texi: Document some architecture specific
18268 constraints and sort entries.
18270 2018-05-30 Martin Sebor <msebor@redhat.com>
18272 PR middle-end/85369
18273 * builtins.c (expand_builtin_stpcpy_1): New function.
18274 (expand_builtin_stpcpy): Call it, and call maybe_warn_nonstring_arg
18275 only if the former succeeds.
18277 2018-05-31 Sameera Deshpande <sameera.deshpande@linaro.org>
18279 * config/aarch64/aarch64-cores.def (saphira) : Add support for ARMv8.4
18282 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
18284 * doc/invoke.texi (-flinker-output): Document
18286 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
18288 * passes.c (ipa_write_summaries): Only modify statements if body
18290 * cgraphunit.c (ipa_passes): Also produce intermeidate code when
18291 incrementally linking.
18292 (ipa_passes): Likewise.
18293 * lto-cgraph.c (lto_output_node): When incrementally linking do not
18294 pass down resolution info.
18295 * common.opt (flag_incremental_link): Update info.
18296 * gcc.c (plugin specs): Turn flinker-output=* to
18297 -plugin-opt=-linker-output-known
18298 * toplev.c (compile_file): Also cut compilation when doing incremental
18300 * flag-types. (enum lto_partition_model): Add
18301 LTO_LINKER_OUTPUT_NOLTOREL.
18302 (invoke.texi): Add -flinker-output docs.
18303 * ipa.c (symbol_table::remove_unreachable_nodes): Handle LTO incremental
18304 link same way as WPA; do not stream in dead initializers.
18306 * dwarf2out.c (dwarf2out_die_ref_for_decl,
18307 darf2out_register_external_decl): Support incremental link.
18309 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
18311 * lto-opts.c (lto_write_options): Skip OPT_dumpdir, OPT_fresolution_.
18313 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
18315 * lto-wrapper.c (debug_objcopy): Add rename parameter; pass
18316 it down to simple_object_copy_lto_debug_sections.
18317 (run_gcc): Determine incremental LTO link time and configure
18318 lto1 into non-wpa mode, disable renaming of debug sections.
18320 2018-05-30 Kelvin Nilsen <kelvin@gcc.gnu.org>
18322 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Remove
18323 descriptions of various incorrectly documented functions.
18325 2018-05-30 Andre Vieira <andre.simoesdiasvieira@arm.com>
18328 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Make
18329 address check not strict.
18331 2018-05-30 Richard Biener <rguenther@suse.de>
18333 PR tree-optimization/85964
18334 * tracer.c (better_p): Drop initialized count check, we only
18335 call the function with initialized counts now.
18336 (find_best_successor): Do find a best edge if one
18337 has uninitialized count.
18338 (find_best_predecessor): Likewise. Do BB frequency check only
18339 if count is initialized.
18341 2017-05-30 Jackson Woodruff <jackson.woodruff@arm.com>
18343 * config/aarch64/aarch64.c (aarch64_host_wide_int_compare): New.
18344 (aarch64_ldrstr_offset_compare): New.
18345 (aarch64_operands_adjust_ok_for_ldpstp): Update to consider all
18346 load/store orderings.
18347 (aarch64_gen_adjusted_ldpstp): Likewise.
18349 2018-05-30 Wilco Dijkstra <wdijkstr@arm.com>
18351 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
18352 Check for subset of GENERAL_REGS and FP_REGS.
18353 * config/aarch64/aarch64-simd.md (aarch64_get_lane): Increase cost of
18356 2018-05-30 Richard Sandiford <richard.sandiford@linaro.org>
18358 * alias.c (adjust_offset_for_component_ref): Use poly_int_tree_p
18359 and wi::to_poly_offset. Add the current offset and then check
18360 whether the sum fits, rather than using an unchecked addition of
18361 a checked term. Check for a shwi rather than a uhwi.
18362 * expr.c (get_bit_range): Use tree_to_poly_uint64.
18363 (store_constructor): Use poly_int_tree_p.
18364 (expand_expr_real_1): Likewise.
18365 * function.c (assign_temp): Likewise.
18366 * fold-const.c (const_binop): Use poly_int_tree_p and
18367 wi::to_poly_offset.
18368 (fold_indirect_ref_1): Likewise. Use multiple_p to attempt an exact
18370 * ipa-icf-gimple.c (func_checker::compare_operand): Use
18371 to_poly_offset for MEM offsets.
18372 * ipa-icf.c (sem_variable::equals): Likewise.
18373 * stor-layout.c (compute_record_mode): Use poly_int_tree_p.
18374 * tree-ssa-sccvn.c (ao_ref_init_from_vn_reference): Use
18375 wi::to_poly_offset for BIT_FIELD_REF offsets.
18376 (vn_reference_maybe_forwprop_address): Use poly_int_tree_p and
18377 wi::to_poly_offset.
18378 * var-tracking.c (emit_note_insn_var_location): Use
18379 tree_to_poly_uint64.
18381 2018-05-29 Jim Wilson <jimw@sifive.com>
18383 * config/riscv/riscv.c (riscv_interrupt_type): Fix comment typo.
18385 2018-05-29 Uros Bizjak <ubizjak@gmail.com>
18388 * config/i386/i386.md (l<rounding_insn><MODEF:mode><SWI48:mode>2):
18389 Enable for TARGET_SSE4_1 and generate rounds{s,d} and cvtts{s,d}2si{,q}
18391 (sse4_1_round<mode>2): Use nonimmediate_operand
18392 for operand 1 predicate.
18394 2018-05-29 Martin Sebor <msebor@redhat.com>
18395 Richard Biener <rguenther@suse.de>
18398 * calls.c (get_size_range): Call determine_value_range instead
18399 of get_value_range..
18400 * tree-vrp.h (determine_value_range): Declared new function.
18401 * tree-vrp.c (determine_value_range_1, determine_value_range): New.
18403 2018-05-29 Richard Biener <rguenther@suse.de>
18405 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Make
18406 sure to use non-pattern stmts for get_earlier_stmt arguments.
18407 * tree-vectorizer.h (get_earlier_stmt): Assert we do not get
18408 called on pattern stmts.
18409 (get_later_stmt): Likewise.
18411 2018-05-29 Martin Liska <mliska@suse.cz>
18413 PR gcov-profile/85759
18414 * doc/gcov.texi: Document GCOV_ERROR_FILE and GCOV_EXIT_AT_ERROR
18417 2018-05-29 Jakub Jelinek <jakub@redhat.com>
18419 * tree-cfg.c (verify_gimple_assign_unary): Add checking for
18421 (verify_gimple_assign_binary): Check TYPE_VECTOR_SUBPARTS for
18425 * tree.def (VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR,
18426 VEC_PACK_FLOAT_EXPR): New tree codes.
18427 * tree-pretty-print.c (op_code_prio): Handle
18428 VEC_UNPACK_FIX_TRUNC_HI_EXPR and VEC_UNPACK_FIX_TRUNC_LO_EXPR.
18429 (dump_generic_node): Handle VEC_UNPACK_FIX_TRUNC_HI_EXPR,
18430 VEC_UNPACK_FIX_TRUNC_LO_EXPR and VEC_PACK_FLOAT_EXPR.
18431 * tree-inline.c (estimate_operator_cost): Likewise.
18432 * gimple-pretty-print.c (dump_binary_rhs): Handle VEC_PACK_FLOAT_EXPR.
18433 * fold-const.c (const_binop): Likewise.
18434 (const_unop): Handle VEC_UNPACK_FIX_TRUNC_HI_EXPR and
18435 VEC_UNPACK_FIX_TRUNC_LO_EXPR.
18436 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
18437 (verify_gimple_assign_binary): Handle VEC_PACK_FLOAT_EXPR.
18438 * cfgexpand.c (expand_debug_expr): Handle VEC_UNPACK_FIX_TRUNC_HI_EXPR,
18439 VEC_UNPACK_FIX_TRUNC_LO_EXPR and VEC_PACK_FLOAT_EXPR.
18440 * expr.c (expand_expr_real_2): Likewise.
18441 * optabs.def (vec_packs_float_optab, vec_packu_float_optab,
18442 vec_unpack_sfix_trunc_hi_optab, vec_unpack_sfix_trunc_lo_optab,
18443 vec_unpack_ufix_trunc_hi_optab, vec_unpack_ufix_trunc_lo_optab): New
18445 * optabs.c (expand_widen_pattern_expr): For
18446 VEC_UNPACK_FIX_TRUNC_HI_EXPR and VEC_UNPACK_FIX_TRUNC_LO_EXPR use
18447 sign from result type rather than operand's type.
18448 (expand_binop_directly): For vec_packu_float_optab and
18449 vec_packs_float_optab allow result type to be different from operand's
18451 * optabs-tree.c (optab_for_tree_code): Handle
18452 VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR and
18453 VEC_PACK_FLOAT_EXPR. Formatting fixes.
18454 * tree-vect-generic.c (expand_vector_operations_1): Handle
18455 VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR and
18456 VEC_PACK_FLOAT_EXPR.
18457 * tree-vect-stmts.c (supportable_widening_operation): Handle
18459 (supportable_narrowing_operation): Handle FLOAT_EXPR.
18460 * config/i386/i386.md (fixprefix, floatprefix): New code attributes.
18461 * config/i386/sse.md (*float<floatunssuffix>v2div2sf2): Rename to ...
18462 (float<floatunssuffix>v2div2sf2): ... this. Formatting fix.
18463 (vpckfloat_concat_mode, vpckfloat_temp_mode, vpckfloat_op_mode): New
18465 (vec_pack<floatprefix>_float_<mode>): New expander.
18466 (vunpckfixt_mode, vunpckfixt_model, vunpckfixt_extract_mode): New mode
18468 (vec_unpack_<fixprefix>fix_trunc_lo_<mode>,
18469 vec_unpack_<fixprefix>fix_trunc_hi_<mode>): New expanders.
18470 * doc/md.texi (vec_packs_float_@var{m}, vec_packu_float_@var{m},
18471 vec_unpack_sfix_trunc_hi_@var{m}, vec_unpack_sfix_trunc_lo_@var{m},
18472 vec_unpack_ufix_trunc_hi_@var{m}, vec_unpack_ufix_trunc_lo_@var{m}):
18474 * doc/generic.texi (VEC_UNPACK_FLOAT_HI_EXPR,
18475 VEC_UNPACK_FLOAT_LO_EXPR): Fix pasto in description.
18476 (VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR,
18477 VEC_PACK_FLOAT_EXPR): Document.
18479 2018-05-29 Richard Biener <rguenther@suse.de>
18481 * tree-vectorizer.h (struct vec_info): Add stmt_vec_infos
18483 (stmt_vec_info_vec): Make pointer.
18484 (init_stmt_vec_info_vec): Remove.
18485 (free_stmt_vec_info_vec): Likewise.
18486 (set_stmt_vec_info_vec): New function.
18487 (free_stmt_vec_infos): Likewise.
18488 (vinfo_for_stmt): Adjust for stmt_vec_info_vec indirection.
18489 (set_vinfo_for_stmt): Likewise.
18490 (get_earlier_stmt): Likewise.
18491 (get_later_stmt): Likewise.
18492 * tree-vectorizer.c (stmt_vec_info_vec): Make pointer.
18493 (vec_info::vec_info): Allocate stmt_vec_infos and set the global.
18494 (vec_info::~vec_info): Free stmt_vec_infos.
18495 (vectorize_loops): Set the global stmt_vec_info_vec to NULL.
18496 Remove old init_stmt_vec_info_vec/free_stmt_vec_info_vec calls.
18497 (pass_slp_vectorize::execute): Likewise.
18498 * tree-vect-stmts.c (init_stmt_vec_info_vec): Remove.
18499 (free_stmt_vec_info_vec): Likewise.
18500 (set_stmt_vec_info_vec): New function.
18501 (free_stmt_vec_infos): Likewise.
18502 * tree-vect-loop.c (_loop_vec_info::~_loop_vec_info): Set
18503 the global stmt_vec_info_vec.
18504 * tree-parloops.c (gather_scalar_reductions): Use
18505 set_stmt_vec_info_vec/free_stmt_vec_infos and maintain a local
18508 2018-05-29 Richard Biener <rguenther@suse.de>
18510 * dominance.c (iterate_fix_dominators): Push/pop TV_DOMINANCE.
18512 2018-05-29 Martin Liska <mliska@suse.cz>
18513 David Malcolm <dmalcolm@redhat.com>
18515 * vec.c (test_reverse): New.
18516 (vec_c_tests): Add new test.
18517 * vec.h (vl_ptr>::reverse): New function.
18519 2018-05-29 Gerald Pfeifer <gerald@pfeifer.com>
18521 * config.gcc: Identify FreeBSD 3.x and 4.x as unsupported.
18523 * config/freebsd-spec.h (FBSD_LIB_SPEC): Only consider FreeBSD 5
18526 2018-05-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
18528 * tree-dump.c (dump_node): Use splay_tree_delete_pointers.
18530 2018-05-28 Richard Biener <rguenther@suse.de>
18532 PR tree-optimization/85933
18533 * tree-vect-data-refs.c (vect_record_base_alignments): Only
18534 look at stmts marked as vectorizable.
18536 2018-05-28 Richard Biener <rguenther@suse.de>
18538 PR tree-optimization/85934
18539 * tree-vect-generic.c (expand_vector_operations_1): Hoist
18540 vector boolean check before scalar optimization.
18542 2018-05-28 Jakub Jelinek <jakub@redhat.com>
18544 * doc/invoke.texi (ARM Options): Use @item instead of @itemx
18547 2018-05-28 Mark Wielaard <mark@klomp.org>
18549 * dwarf2asm.c (dw2_asm_output_delta_uleb128): Add brackets around lab2
18550 if it is an expression containing a minus sign.
18552 2018-05-27 John David Anglin <danglin@gcc.gnu.org>
18554 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Define to 0.
18556 2018-05-27 Paul Koning <ni1d@arrl.net>
18558 * config/pdp11/pdp11.md (truncsihi2): Remove.
18560 2018-05-27 Monk Chiang <sh.chiang04@gmail.com>
18561 Chung-Ju Wu <jasonwucj@gmail.com>
18563 * config/nds32/nds32-intrinsic.md (unaligned_storedi): Modify patterns
18565 (unaligned_store_dw): Ditto.
18566 * config/nds32/nds32-memory-manipulation.c
18567 (nds32_expand_movmemsi_loop_known_size): Refactoring implementation.
18568 (nds32_gen_dup_4_byte_to_word_value): Rename to ...
18569 (nds32_gen_dup_4_byte_to_word_value_aux): ... this.
18570 (emit_setmem_word_loop): Rename to ...
18571 (emit_setmem_doubleword_loop): ... this.
18572 (nds32_gen_dup_4_byte_to_word_value): New function.
18573 (nds32_gen_dup_8_byte_to_double_word_value): New function.
18574 (nds32_expand_setmem_loop): Refine implementation.
18575 (nds32_expand_setmem_loop_v3m): Ditto.
18576 * config/nds32/nds32-multiple.md (unaligned_store_update_base_dw): New
18579 2018-05-27 Chung-Ju Wu <jasonwucj@gmail.com>
18581 * config/nds32/nds32.md (bswapsi2, bswaphi2): New patterns.
18583 2018-05-27 Chung-Ju Wu <jasonwucj@gmail.com>
18585 * config/nds32/nds32.c (nds32_attribute_table): Add "no_prologue".
18586 (nds32_init_machine_status): Initialize machine->attr_naked_p and
18587 machine->attr_no_prologue_p.
18588 (nds32_compute_stack_frame): Check "naked" and "no_prologue" attributes.
18589 (nds32_naked_function_p): Handle "naked" and "no_prologue" attributes.
18590 (nds32_expand_epilogue): Consider attr_naked_p.
18591 (nds32_expand_epilogue_v3pop): Likewise.
18592 (nds32_can_use_return_insn): Likewise.
18593 * config/nds32/nds32.h (machine_function): Add attr_naked_p and
18594 attr_no_prologue_p fields.
18595 * config/nds32/nds32.opt (mret-in-naked-func): New option.
18597 2018-05-27 Jakub Jelinek <jakub@redhat.com>
18600 * config/i386/i386.md (fixunssuffix, floatunssuffix): New code
18602 * config/i386/sse.md
18603 (<floatsuffix>float<sseintvecmodelower><mode>2<mask_name><round_name>):
18605 (float<floatunssuffix><sseintvecmodelower><mode>2<mask_name><round_name>):
18607 (<floatsuffix>float<sselongvecmodelower><mode>2<mask_name><round_name>):
18609 (float<floatunssuffix><sselongvecmodelower><mode>2<mask_name><round_name>):
18611 (*<floatsuffix>floatv2div2sf2): Rename to ...
18612 (*float<floatunssuffix>v2div2sf2): ... this.
18613 (<floatsuffix>floatv2div2sf2_mask): Rename to ...
18614 (float<floatunssuffix>v2div2sf2_mask): ... this.
18615 (*<floatsuffix>floatv2div2sf2_mask_1): Rename to ...
18616 (*float<floatunssuffix>v2div2sf2_mask_1): ... this.
18617 (<fixsuffix>fix_truncv8dfv8si2<mask_name><round_saeonly_name>): Rename
18619 (fix<fixunssuffix>_truncv8dfv8si2<mask_name><round_saeonly_name>):
18621 (<fixsuffix>fix_trunc<mode><sseintvecmodelower>2<mask_name><round_saeonly_name>):
18623 (fix<fixunssuffix>_trunc<mode><sseintvecmodelower>2<mask_name><round_saeonly_name>):
18625 (<fixsuffix>fix_trunc<mode><sselongvecmodelower>2<mask_name><round_saeonly_name>):
18627 (fix<fixunssuffix>_trunc<mode><sselongvecmodelower>2<mask_name><round_saeonly_name>):
18629 (<fixsuffix>fix_truncv2sfv2di2<mask_name>): Rename to ...
18630 (fix<fixunssuffix>_truncv2sfv2di2<mask_name>): ... this.
18631 (vec_pack_ufix_trunc_<mode>): Use gen_fixuns_truncv8dfv8si2 instead of
18632 gen_ufix_truncv8dfv8si2.
18633 * config/i386/i386-builtin.def (__builtin_ia32_cvttpd2uqq256_mask,
18634 __builtin_ia32_cvttpd2uqq128_mask, __builtin_ia32_cvttps2uqq256_mask,
18635 __builtin_ia32_cvttps2uqq128_mask, __builtin_ia32_cvtuqq2ps256_mask,
18636 __builtin_ia32_cvtuqq2ps128_mask, __builtin_ia32_cvtuqq2pd256_mask,
18637 __builtin_ia32_cvtuqq2pd128_mask, __builtin_ia32_cvttpd2udq512_mask,
18638 __builtin_ia32_cvtuqq2ps512_mask, __builtin_ia32_cvtuqq2pd512_mask,
18639 __builtin_ia32_cvttps2uqq512_mask, __builtin_ia32_cvttpd2uqq512_mask):
18640 Use fixuns instead ufix or floatuns instead ufloat in CODE_FOR_ names.
18642 2018-05-24 H.J. Lu <hongjiu.lu@intel.com>
18646 * varasm.c (assemble_alias): Lookup ifunc attribute on error.
18648 2018-05-25 Jim Wilson <jimw@sifive.com>
18650 * config/riscv/riscv-protos.h (riscv_epilogue_uses): New.
18651 * config/riscv/riscv.c (struct machine_function): Add
18652 interrupt_handler_p and attribute_checked_p fields.
18653 (riscv_attribute_table): Add interrupt.
18654 (riscv_interrupt_type_p): New.
18655 (riscv_save_reg_p): Save extra regs for interrupt handler.
18656 (riscv_use_save_libcall): Return false for interrupt handler.
18657 (riscv_first_stack_step): Add forward declaration.
18658 (riscv_compute_frame_info): New local interrupt_save_t1. Set it
18659 for interrupt handler with large frame. Use it for saved reg list.
18660 (riscv_expand_prologue): Move flag_stack_usage_info support to
18661 eliminate duplication.
18662 (riscv_expand_epilogue): Generate mret for interrupt handler.
18663 (riscv_epilogue_uses): New.
18664 (riscv_can_use_return_insn): Return false for interrupt handler.
18665 (riscv_function_ok_for_sibcall): Likewise.
18666 (riscv_set_current_function): Add interrupt handler support.
18667 * config/riscv/riscv.h (EPILOGUE_USES): Call riscv_epilogue_uses.
18668 * config/riscv/riscv.md (UNSPECV_MRET): New.
18670 (riscv_frflags, riscv_fsflags): Use tab after opcode.
18672 * doc/extend.texi (RISC-V Function Attributes) <interrupt>: New.
18674 2018-05-25 Bill Schmidt <wschmidt@linux.ibm.com>
18676 PR tree-optimization/85712
18677 * gimple-ssa-strength-reduction.c (replace_one_candidate): Skip if
18678 this candidate has already been replaced in-situ by a copy.
18680 2018-05-25 Jason Merrill <jason@redhat.com>
18682 PR c++/80485 - inline function non-zero address.
18683 * symtab.c (nonzero_address): Check DECL_COMDAT.
18685 2018-05-25 Uros Bizjak <ubizjak@gmail.com>
18688 * config/alpha/alpha.md (ashlsi3): New insn pattern.
18689 (*ashlsi_se): Rename from *ashldi_se. Define as sign
18690 extension of SImode operation. Use const123_operand predicate.
18691 (*saddsi_1): Remove.
18692 (*saddl_se_1): Ditto.
18693 (*ssubsi_1): Ditto.
18694 (*ssubl_se_1): Ditto.
18695 * config/alpha/predicates.md (const123_operand): New predicate.
18696 * config/alpha/constraints.md (P): Use IN_RANGE.
18698 2018-05-25 Richard Biener <rguenther@suse.de>
18700 * tree-ssa-alias.h (refs_may_alias_p): Add tbaa_p bool parameter,
18702 (ref_maybe_used_by_stmt_p): Likewise.
18703 (stmt_may_clobber_ref_p): Likewise.
18704 (stmt_may_clobber_ref_p_1): Likewise.
18705 * tree-ssa-alias.c (refs_may_alias_p): Add tbaa_p bool parameter
18707 (ref_maybe_used_by_stmt_p): Likewise.
18708 (stmt_may_clobber_ref_p): Likewise.
18709 (stmt_may_clobber_ref_p_1): Likewise.
18710 * tree-vect-data-refs.c (vect_slp_analyze_node_dependences): Use
18711 the alias oracle to disambiguate DRs with stmts DR analysis
18713 (vect_analyze_data_refs): Do not give up on not analyzable
18714 DRs for BB vectorization. Remove code truncating the dataref
18717 2018-05-25 Jakub Jelinek <jakub@redhat.com>
18720 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>_1):
18721 Add (=Yk,v,C) variant using vptestm insn. Use TARGET_AVX512BW
18722 in test instead of TARGET_AVX512F for VI12_AVX512VL iterator.
18724 2018-05-25 Richard Biener <rguenther@suse.de>
18726 * tree-vect-data-refs.c (vect_find_stmt_data_reference): New
18727 function, combining stmt data ref gathering and fatal analysis
18729 (vect_analyze_data_refs): Remove now redudnant code and simplify.
18730 * tree-vect-loop.c (vect_get_datarefs_in_loop): Factor out from
18731 vect_analyze_loop_2 and use vect_find_stmt_data_reference.
18732 * tree-vect-slp.c (vect_slp_bb): Use vect_find_stmt_data_reference.
18733 * tree-vectorizer.h (vect_find_stmt_data_reference): Declare.
18735 2018-05-25 Bin Cheng <bin.cheng@arm.com>
18737 PR tree-optimization/85720
18738 * tree-loop-distribution.c (break_alias_scc_partitions): Don't merge
18739 SCC if all partitions are builtins.
18740 (version_loop_by_alias_check): New parameter. Generate cancelable
18741 runtime alias check if all partitions are builtins.
18742 (distribute_loop): Update call to above function.
18744 2018-05-25 Bin Cheng <bin.cheng@arm.com>
18746 * tree-outof-ssa.c (tree-ssa.h, tree-dfa.h): Include header files.
18747 (create_default_def, for_all_parms): Moved from tree-ssa-coalesce.c.
18748 (parm_default_def_partition_arg): Ditto.
18749 (set_parm_default_def_partition): Ditto.
18750 (get_parm_default_def_partitions): Ditto and make it static.
18751 (get_undefined_value_partitions): Ditto and make it static.
18752 (remove_ssa_form): Refactor call to init_var_map here.
18753 * tree-ssa-coalesce.c (build_ssa_conflict_graph): Support live range
18754 computation for loop region.
18755 (coalesce_partitions, compute_optimized_partition_bases): Ditto.
18756 (register_default_def): Delete.
18757 (for_all_parms, create_default_def): Move to tree-outof-ssa.c.
18758 (parm_default_def_partition_arg): Ditto.
18759 (set_parm_default_def_partition): Ditto.
18760 (get_parm_default_def_partitions): Ditto and make it static.
18761 (get_undefined_value_partitions): Ditto and make it static.
18762 (coalesce_with_default, coalesce_with_default): Update comment.
18763 (create_coalesce_list_for_region): New func factored out from
18764 create_outofssa_var_map.
18765 (populate_coalesce_list_for_outofssa): New func factored out from
18766 create_outofssa_var_map and coalesce_ssa_name.
18767 (create_outofssa_var_map): Delete.
18768 (coalesce_ssa_name): Refactor to support live range computation.
18769 * tree-ssa-coalesce.h (coalesce_ssa_name): Change decl.
18770 (get_parm_default_def_partitions): Delete.
18771 (get_undefined_value_partitions): Ditto.
18772 * tree-ssa-live.c (init_var_map, delete_var_map): Support live range
18773 computation for loop region.
18774 (new_tree_live_info, loe_visit_block): Ditto.
18775 (live_worklist, set_var_live_on_entry): Ditto.
18776 (calculate_live_on_exit, verify_live_on_entry): Ditto.
18777 * tree-ssa-live.h (struct _var_map): New fields.
18778 (init_var_map): Change decl.
18779 (region_contains_p): New.
18781 2018-05-25 Bin Cheng <bin.cheng@arm.com>
18783 * tree-ssa-live.h (live_merge_and_clear): Delete.
18785 2018-05-25 Richard Biener <rguenther@suse.de>
18788 * tree-dump.c (dequeue_and_dump): Remove access to removed
18789 operand 2 of a SWITCH_EXPR.
18791 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
18793 * doc/sourcebuild.texi (vect_double_cond_arith): Include
18794 multiplication and division.
18795 * doc/md.texi (cond_mul@var{m}, cond_div@var{m}, cond_mod@var{m})
18796 (cond_udiv@var{m}, cond_umod@var{m}): Document.
18797 * optabs.def (cond_smul_optab, cond_sdiv_optab, cond_smod_optab)
18798 (cond_udiv_optab, cond_umod_optab): New optabs.
18799 * internal-fn.def (IFN_COND_MUL, IFN_COND_DIV, IFN_COND_MOD)
18800 (IFN_COND_RDIV): New internal functions.
18801 * internal-fn.c (get_conditional_internal_fn): Handle TRUNC_DIV_EXPR,
18802 TRUNC_MOD_EXPR and RDIV_EXPR.
18803 * match.pd (UNCOND_BINARY, COND_BINARY): Handle them.
18804 * config/aarch64/iterators.md (UNSPEC_COND_MUL, UNSPEC_COND_DIV):
18806 (SVE_INT_BINARY): Include mult.
18807 (SVE_COND_FP_BINARY): Include UNSPEC_MUL and UNSPEC_DIV.
18808 (optab, sve_int_op): Handle mult.
18809 (optab, sve_fp_op, commutative): Handle UNSPEC_COND_MUL and
18811 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): New pattern
18812 for SVE_INT_BINARY_SD.
18814 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
18816 * config/aarch64/iterators.md (SVE_INT_BINARY_SD): New code iterator.
18817 (optab, sve_int_op): Handle div and udiv.
18818 * config/aarch64/aarch64-sve.md (<optab><mode>3): New expander
18819 for SVE_INT_BINARY_SD.
18820 (*<optab><mode>3): New insn for the same.
18822 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
18824 * tree-vect-patterns.c: Include predict.h.
18825 (vect_recog_divmod_pattern): Restrict check for division support
18826 to when optimizing for size.
18828 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
18830 * doc/sourcebuild.texi (vect_double_cond_arith: Document.
18831 * gimple-match.h (gimple_match_op::MAX_NUM_OPS): Bump to 4.
18832 (gimple_match_op::gimple_match_op): Add an overload for 4 operands.
18833 (gimple_match_op::set_op): Likewise.
18834 (gimple_resimplify4): Declare.
18835 * genmatch.c (get_operand_type): Handle CFN_COND_* functions.
18836 (expr::gen_transform): Likewise.
18837 (decision_tree::gen): Generate a simplification routine for 4 operands.
18838 * gimple-match-head.c (gimple_simplify): Add an overload for
18839 4 operands. In the top-level function, handle up to 4 call
18840 arguments and call gimple_resimplify4.
18841 (gimple_resimplify4): New function.
18842 (build_call_internal): Pass a fourth operand.
18843 (maybe_push_to_seq): Likewise.
18844 * match.pd (UNCOND_BINARY, COND_BINARY): New operator lists.
18845 Fold VEC_COND_EXPRs of an operation and a default value into
18846 an IFN_COND_* function if possible.
18847 * config/aarch64/iterators.md (UNSPEC_COND_MAX, UNSPEC_COND_MIN):
18849 (SVE_COND_FP_BINARY): Include them.
18850 (optab, sve_fp_op): Handle them.
18851 (SVE_INT_BINARY_REV): New code iterator.
18852 (SVE_COND_FP_BINARY_REV): New int iterator.
18853 (commutative): New int attribute.
18854 * config/aarch64/aarch64-protos.h (aarch64_sve_prepare_conditional_op):
18856 * config/aarch64/aarch64.c (aarch64_sve_prepare_conditional_op): New
18858 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): Use it.
18859 (*cond_<optab><mode>): New patterns for reversed operands.
18861 2018-05-25 Richard Biener <rguenther@suse.de>
18863 * tree-vectorizer.h (STMT_VINFO_GROUP_*, GROUP_*): Remove.
18864 (DR_GROUP_*): New, assert we have non-NULL ->data_ref_info.
18865 (REDUC_GROUP_*): New, assert we have NULL ->data_ref_info.
18866 (STMT_VINFO_GROUPED_ACCESS): Adjust.
18867 * tree-vect-data-refs.c (everywhere): Adjust users.
18868 * tree-vect-loop.c (everywhere): Likewise.
18869 * tree-vect-slp.c (everywhere): Likewise.
18870 * tree-vect-stmts.c (everywhere): Likewise.
18871 * tree-vect-patterns.c (vect_reassociating_reduction_p): Likewise.
18873 2018-05-25 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
18875 * configure.ac (gcc_cv_as_section_has_e): Move to common section.
18877 (gcc_cv_as_section_exclude): ... this.
18878 Try Solaris as #exclude syntax.
18879 * configure: Regenerate.
18880 * config.in: Regenerate.
18881 * config/i386/i386.c (i386_solaris_elf_named_section): Handle
18883 * config/sparc/sparc.c (sparc_solaris_elf_asm_named_section)
18884 [HAVE_GAS_SECTION_EXCLUDE]: Handle SECTION_EXCLUDE.
18886 * varasm.c (default_elf_asm_named_section): Don't check if
18887 HAVE_GAS_SECTION_EXCLUDE is defined.
18889 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
18891 * doc/md.texi: Update the documentation of the cond_* optabs
18892 to mention the new final operand. Fix GET_MODE_NUNITS call.
18893 Describe the scalar case too.
18894 * internal-fn.def (IFN_EXTRACT_LAST): Change type to fold_left.
18895 * internal-fn.c (expand_cond_unary_optab_fn): Expect 3 operands
18897 (expand_cond_binary_optab_fn): Expect 4 operands instead of 3.
18898 (get_conditional_internal_fn): Update comment.
18899 * tree-vect-loop.c (vectorizable_reduction): Pass the original
18900 accumulator value as a final argument to conditional functions.
18901 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): Turn into
18902 a define_expand and add an "else" operand. Assert for now that
18903 the else operand is equal to operand 2. Use SVE_INT_BINARY and
18904 SVE_COND_FP_BINARY instead of SVE_COND_INT_OP and SVE_COND_FP_OP.
18905 (*cond_<optab><mode>): New patterns.
18906 * config/aarch64/iterators.md (UNSPEC_COND_SMAX, UNSPEC_COND_UMAX)
18907 (UNSPEC_COND_SMIN, UNSPEC_COND_UMIN, UNSPEC_COND_AND, UNSPEC_COND_ORR)
18908 (UNSPEC_COND_EOR): Delete.
18909 (optab): Remove associated mappings.
18910 (SVE_INT_BINARY): New code iterator.
18911 (sve_int_op): Remove int attribute and add "minus" to the code
18913 (SVE_COND_INT_OP): Delete.
18914 (SVE_COND_FP_OP): Rename to...
18915 (SVE_COND_FP_BINARY): ...this.
18917 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
18919 * optabs.c (can_reuse_operands_p): New function.
18920 (maybe_legitimize_operands): Try to reuse the results for
18923 2018-05-24 Uros Bizjak <ubizjak@gmail.com>
18925 * config/i386/sse.md (cvtusi2<ssescalarmodesuffix>64<round_name>):
18926 Add {q} suffix to insn mnemonic.
18928 2018-05-23 Jozef Lawrynowicz <jozef.l@mittosystems.com>
18930 * config/msp430/msp430.c (TARGET_WARN_FUNC_RETURN): Define.
18931 (msp430_warn_func_return): New.
18933 2018-05-24 Roger Sayle <roger@nextmovesoftware.com>
18935 * fold-const.c (tree_nonzero_bits): New function.
18936 * fold-const.h (tree_nonzero_bits): Likewise.
18937 * match.pd (POPCOUNT): New patterns to fold BUILTIN_POPCOUNT and
18938 friends. POPCOUNT(x&1) => x&1, POPCOUNT(x)==0 => x==0, etc.
18940 2018-05-24 H.J. Lu <hongjiu.lu@intel.com>
18944 * varasm.c (assemble_alias): Check ifunc_resolver only on
18947 2018-05-24 Uros Bizjak <ubizjak@gmail.com>
18950 * config/i386/sse.md (movdi_to_sse): Do not generate pseudo
18951 when memory input operand is handled.
18953 2018-05-24 Luis Machado <luis.machado@linaro.org>
18955 * config/aarch64/aarch64.c (qdf24xx_addrcost_table): New static
18957 (qdf24xx_tunings) <addr_costs>: Set to qdf24xx_addrcost_table.
18959 2018-05-24 Richard Sandiford <richard.sandiford@linaro.org>
18961 * match.pd: Delay FMA folds until after vectorization.
18963 2018-05-24 Andre Vieira <andre.simoesdiasvieira@arm.com>
18966 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Make
18967 address check not strict.
18969 2018-05-24 Richard Sandiford <richard.sandiford@linaro.org>
18971 * gimple-match.h (gimple_match_op): New class.
18972 (mprts_hook): Replace parameters with a gimple_match_op *.
18973 (maybe_build_generic_op): Likewise.
18974 (gimple_simplified_result_is_gimple_val): Replace parameters with
18975 a const gimple_match_op *.
18976 (gimple_simplify): Replace code_helper * and tree * parameters with
18977 a gimple_match_op * parameter.
18978 (gimple_resimplify1): Replace code_helper *, tree and tree *
18979 parameters with a gimple_match_op * parameter.
18980 (gimple_resimplify2): Likewise.
18981 (gimple_resimplify3): Likewise.
18982 (maybe_push_res_to_seq): Replace code_helper, tree and tree *
18983 parameters with a gimple_match_op * parameter.
18984 * gimple-match-head.c (gimple_simplify): Change prototypes of
18985 auto-generated functions to take a gimple_match_op * instead of
18986 separate code_helper * and tree * parameters. Make the same
18987 change in the top-level overload and update calls to the
18988 gimple_resimplify routines. Update calls to the auto-generated
18989 functions and to maybe_push_res_to_seq in the publicly-facing
18990 operation-specific gimple_simplify overloads.
18991 (gimple_match_op::MAX_NUM_OPS): Define.
18992 (gimple_resimplify1): Replace rcode and ops with a single res_op
18993 parameter. Update call to gimple_simplify.
18994 (gimple_resimplify2): Likewise.
18995 (gimple_resimplify3): Likewise.
18996 (mprts_hook): Replace parameters with a gimple_match_op *.
18997 (maybe_build_generic_op): Likewise.
18998 (build_call_internal): Replace type, nargs and ops with
18999 a gimple_match_op *.
19000 (maybe_push_res_to_seq): Replace res_code, type and ops parameters
19001 with a single gimple_match_op *. Update calls to mprts_hook,
19002 build_call_internal and gimple_simplified_result_is_gimple_val.
19003 Factor out code that is common to the tree_code and combined_fn cases.
19004 * genmatch.c (expr::gen_transform): Replace tem_code and
19005 tem_ops with a gimple_match_op called tem_op. Update calls
19006 to the gimple_resimplify functions and maybe_push_res_to_seq.
19007 (dt_simplify::gen_1): Manipulate res_op instead of res_code and
19008 res_ops. Update call to the gimple_resimplify functions.
19009 (dt_simplify::gen): Pass res_op instead of res_code and res_ops.
19010 (decision_tree::gen): Make the functions take a gimple_match_op *
19011 called res_op instead of separate res_code and res_ops parameters.
19012 Update call accordingly.
19013 * gimple-fold.c (replace_stmt_with_simplification): Replace rcode
19014 and ops with a single res_op parameter. Update calls to
19015 maybe_build_generic_op and maybe_push_res_to_seq.
19016 (fold_stmt_1): Update calls to gimple_simplify and
19017 replace_stmt_with_simplification.
19018 (gimple_fold_stmt_to_constant_1): Update calls to gimple_simplify
19019 and gimple_simplified_result_is_gimple_val.
19020 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update call to
19022 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Replace parameters
19023 with a gimple_match_op *.
19024 (vn_nary_build_or_lookup): Likewise. Update call to
19025 vn_nary_build_or_lookup_1.
19026 (vn_nary_build_or_lookup_1): Replace rcode, type and ops with a
19027 gimple_match_op *. Update calls to the gimple_resimplify routines
19028 and to gimple_simplified_result_is_gimple_val.
19029 (vn_nary_simplify): Update call to vn_nary_build_or_lookup_1.
19030 Use gimple_match_op::MAX_NUM_OPS instead of a hard-coded 3.
19031 (vn_reference_lookup_3): Update call to vn_nary_build_or_lookup.
19032 (visit_nary_op): Likewise.
19033 (visit_reference_op_load): Likewise.
19035 2018-05-23 Luis Machado <luis.machado@linaro.org>
19037 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Use correct type
19038 modifier for printing the step amount.
19040 2018-05-23 Jozef Lawrynowicz <jozef.l@somniumtech.com>
19043 * tree.c (build_common_tree_nodes): Dont set TYPE_SIZE for __intN
19046 2018-05-23 Segher Boessenkool <segher@kernel.crashing.org>
19048 * doc/sourcebuild.texi (Endianness): New subsubsection.
19050 2018-05-23 Luis Machado <luis.machado@linaro.org>
19052 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
19053 <prefetch_dynamic_strides>: New const bool field.
19054 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
19055 prefetch_dynamic_strides.
19056 (exynosm1_prefetch_tune): Likewise.
19057 (thunderxt88_prefetch_tune): Likewise.
19058 (thunderx_prefetch_tune): Likewise.
19059 (thunderx2t99_prefetch_tune): Likewise.
19060 (qdf24xx_prefetch_tune): Likewise. Set prefetch_dynamic_strides to
19062 (aarch64_override_options_internal): Update to set
19063 PARAM_PREFETCH_DYNAMIC_STRIDES.
19064 * doc/invoke.texi (prefetch-dynamic-strides): Document new option.
19065 * params.def (PARAM_PREFETCH_DYNAMIC_STRIDES): New.
19066 * params.h (PARAM_PREFETCH_DYNAMIC_STRIDES): Define.
19067 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Account for
19068 prefetch-dynamic-strides setting.
19070 2018-05-23 Luis Machado <luis.machado@linaro.org>
19072 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
19073 <minimum_stride>: New const int field.
19074 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
19075 minimum_stride field defaulting to -1.
19076 (exynosm1_prefetch_tune): Likewise.
19077 (thunderxt88_prefetch_tune): Likewise.
19078 (thunderx_prefetch_tune): Likewise.
19079 (thunderx2t99_prefetch_tune): Likewise.
19080 (qdf24xx_prefetch_tune) <minimum_stride>: Set to 2048.
19081 <default_opt_level>: Set to 3.
19082 (aarch64_override_options_internal): Update to set
19083 PARAM_PREFETCH_MINIMUM_STRIDE.
19084 * doc/invoke.texi (prefetch-minimum-stride): Document new option.
19085 * params.def (PARAM_PREFETCH_MINIMUM_STRIDE): New.
19086 * params.h (PARAM_PREFETCH_MINIMUM_STRIDE): Define.
19087 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Return false if
19088 stride is constant and is below the minimum stride threshold.
19090 2018-05-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19092 * config/arm/arm-cpus.in (mode26): Delete.
19093 (armv4): Delete mode26 reference.
19094 * config/arm/arm.c (arm_configure_build_target): Delete use of
19097 2018-05-23 Uros Bizjak <ubizjak@gmail.com>
19099 * config/i386/i386.md (*floatuns<SWI48:mode><MODEF:mode>2_avx512):
19101 (floatunssi<mode>2): Also enable for AVX512F and TARGET_SSE_MATH.
19102 Rewrite expander pattern. Emit gen_floatunssi<mode>2_i387_with_xmm
19104 (floatunsdisf2): Rewrite expander pattern. Hanlde TARGET_AVX512F.
19105 (floatunsdidf2): Ditto.
19107 2018-05-23 Uros Bizjak <ubizjak@gmail.com>
19109 * config/i386/i386.md (fixuns_trunc<mode>di2): New insn pattern.
19110 (fixuns_trunc<mode>si2_avx512f): Ditto.
19111 (*fixuns_trunc<mode>si2_avx512f_zext): Ditto.
19112 (fixuns_trunc<mode>si2): Also enable for AVX512F and TARGET_SSE_MATH.
19113 Emit fixuns_trunc<mode>si2_avx512f for AVX512F targets.
19115 2018-05-23 Alexander Monakov <amonakov@ispras.ru>
19117 PR rtl-optimization/79985
19118 * df-scan.c (df_insn_refs_collect): Remove special case for
19119 global registers and asm statements.
19121 2018-05-23 Alexander Monakov <amonakov@ispras.ru>
19123 * extend.texi (Global Register Variables): Rewrite the bullet list.
19124 Note that the register is available for allocation. Note that access
19125 via inline asm must use constraints. Add note about async-signal
19126 handlers. Remove paragraph about automagic register selection.
19128 2018-05-23 Richard Biener <rguenther@suse.de>
19130 * tree-ssa-sccvn.c (vn_reference_lookup_3): Remove restriction
19131 of fixed offset from memset VN.
19133 2018-05-23 Bill Schmidt <wschmidt@linux.ibm.com>
19135 * gimple-ssa-strength-reduction.c (struct slsr_cand_d): Add
19136 first_interp field.
19137 (alloc_cand_and_find_basis): Initialize first_interp field.
19138 (slsr_process_mul): Modify first_interp field.
19139 (slsr_process_add): Likewise.
19140 (slsr_process_cast): Modify first_interp field for each new
19142 (slsr_process_copy): Likewise.
19143 (dump_candidate): Dump first_interp field.
19144 (replace_mult_candidate): Process all interpretations, not just
19146 (replace_rhs_if_not_dup): Likewise.
19147 (replace_one_candidate): Likewise.
19149 2018-05-23 Wilco Dijkstra <wdijkstr@arm.com>
19151 * config/aarch64/aarch64.c (aarch64_use_frame_pointer):
19153 (aarch64_needs_frame_chain): New function.
19154 (aarch64_parse_override_string): Set aarch64_use_frame_pointer.
19156 2018-05-23 Sudakshina Das <sudi.das@arm.com>
19159 * common/config/aarch64/aarch64-common.c (aarch64_handle_option):
19160 Check val before adding MASK_STRICT_ALIGN to opts->x_target_flags.
19161 * config/aarch64/aarch64.opt (mstrict-align): Remove RejectNegative.
19162 * config/aarch64/aarch64.c (aarch64_attributes): Mark allow_neg
19163 as true for strict-align.
19164 (aarch64_can_inline_p): Perform checks even when callee has no
19165 attributes to check for strict alignment.
19166 * doc/extend.texi (AArch64 Function Attributes): Document
19168 * doc/invoke.texi: (AArch64 Options): Likewise.
19170 2018-05-23 Richard Sandiford <richard.sandiford@linaro.org>
19172 PR tree-optimization/85853
19173 * tree-vect-slp.c (vect_slp_analyze_node_operations): Split out
19174 the handling of the root of the node to...
19175 (vect_slp_analyze_node_operations_1): ...this new function,
19176 and run the whole thing with the child nodes' def types
19177 set according to their SLP node's def type.
19179 2018-05-23 Richard Biener <rguenther@suse.de>
19181 PR middle-end/85874
19182 * tree-data-ref.c (create_runtime_alias_checks): Defer
19183 and ignore overflow warnings.
19185 2018-05-23 Yury Gribov <tetra2005@gmail.com>
19187 PR tree-optimization/85822
19188 * tree-vrp.c (is_masked_range_test): Fix handling of negative
19191 2018-05-23 Richard Biener <rguenther@suse.de>
19193 * tree-ssa-sccvn.c (vn_reference_lookup_3): Handle arbitrary
19194 memset constants via native_interpret_expr.
19196 2018-05-22 H.J. Lu <hongjiu.lu@intel.com>
19199 * cgraph.h (cgraph_node::create): Set ifunc_resolver for ifunc
19201 (cgraph_node::create_alias): Likewise.
19202 (cgraph_node::get_availability): Check ifunc_resolver instead
19203 of looking up ifunc attribute.
19204 * cgraphunit.c (maybe_diag_incompatible_alias): Likewise.
19205 * varasm.c (do_assemble_alias): Likewise.
19206 (assemble_alias): Likewise.
19207 (default_binds_local_p_3): Likewise.
19208 * cgraph.h (cgraph_node): Add ifunc_resolver.
19209 (cgraph_node::only_called_directly_or_aliased_p): Return false
19210 for IFUNC resolver.
19211 * lto-cgraph.c (input_node): Set ifunc_resolver for ifunc
19213 * symtab.c (symtab_node::verify_base): Verify that ifunc_resolver
19214 is equivalent to lookup_attribute ("ifunc", DECL_ATTRIBUTES (decl)).
19215 (symtab_node::binds_to_current_def_p): Check ifunc_resolver
19216 instead of looking up ifunc attribute.
19218 2018-05-22 Luis Machado <luis.machado@linaro.org>
19220 * config/aarch64/aarch64.md (*ashift<mode>_extv_bfiz): New pattern.
19222 2018-05-22 Martin Sebor <msebor@redhat.com>
19224 PR middle-end/85359
19225 * builtins.c (expand_builtin_strcpy): Call maybe_warn_nonstring_arg
19226 only when expasion succeeds.
19227 (expand_builtin_strcmp): Same.
19228 (expand_builtin_strncmp): Same.
19230 2018-05-22 Martin Sebor <msebor@redhat.com>
19232 * calls.c (maybe_warn_nonstring_arg): Fix a typo in a comment.
19234 2018-05-22 Jackson Woodruff <jackson.woodruff@arm.com>
19235 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19237 * config/aarch64/aarch64-ldpstp.md: Replace uses of
19238 aarch64_mem_pair_operand with memory_operand and delete operand swapping
19240 * config/aarch64/aarch64.c (aarch64_operands_ok_for_ldpstp):
19241 Add check for legitimate_address.
19242 (aarch64_gen_adjusted_ldpstp): Swap operands where appropriate.
19243 (aarch64_swap_ldrstr_operands): New.
19244 * config/aarch64/aarch64-protos.h (aarch64_swap_ldrstr_operands):
19247 2018-05-22 Jackson Woodruff <jackson.woodruff@arm.com>
19248 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19250 * config/aarch64/aarch64.md: New patterns to generate stp
19252 (store_pair_sw, store_pair_dw): New patterns to generate stp for
19253 single words and double words.
19254 (load_pair_sw, load_pair_dw): Likewise.
19255 (store_pair_sf, store_pair_df, store_pair_si, store_pair_di):
19257 (load_pair_sf, load_pair_df, load_pair_si, load_pair_di):
19259 * config/aarch64/aarch64-ldpstp.md: Modify peephole
19260 for different mode ldpstp and add peephole for merged zero stores.
19261 Likewise for loads.
19262 * config/aarch64/aarch64.c (aarch64_operands_ok_for_ldpstp):
19264 (aarch64_gen_store_pair): Rename calls to match new patterns.
19265 (aarch64_gen_load_pair): Rename calls to match new patterns.
19266 * config/aarch64/aarch64-simd.md (load_pair<mode>): Rename to...
19267 (load_pair<DREG:mode><DREG2:mode>): ... This.
19268 (store_pair<mode>): Rename to...
19269 (vec_store_pair<DREG:mode><DREG2:mode>): ... This.
19270 * config/aarch64/iterators.md (DREG, DREG2, DX2, SX, SX2, DSX):
19271 New mode iterators.
19272 (V_INT_EQUIV): Handle SImode.
19273 * config/aarch64/predicates.md (aarch64_reg_zero_or_fp_zero):
19276 2018-05-22 Martin Sebor <msebor@redhat.com>
19279 * calls.c (maybe_warn_nonstring_arg): Use string length to set
19280 or ajust the presumed bound on an operation to avoid unnecessary
19283 2018-05-22 Martin Sebor <msebor@redhat.com>
19285 PR tree-optimization/85826
19286 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Avoid
19287 assuming that a DECL necesarily has a constant size.
19289 2018-05-22 Richard Sandiford <richard.sandiford@linaro.org>
19291 PR middle-end/85862
19292 * internal-fn.c (expand_direct_optab_fn): Cope with a null lhs.
19294 2018-05-22 Richard Biener <rguenther@suse.de>
19296 PR tree-optimization/85834
19297 * tree-ssa-sccvn.c (vn_reference_lookup_3): Properly handle
19298 non-constant and non-zero memset arguments.
19300 2018-05-22 Martin Liska <mliska@suse.cz>
19303 * ipa-icf.c (sem_item::add_type): Do not ICE for incomplete types.
19305 2018-05-22 Richard Biener <rguenther@suse.de>
19307 PR tree-optimization/85863
19308 * tree-vect-stmts.c (vect_is_simple_cond): Only widen invariant
19309 comparisons when vectype is specified.
19310 (vectorizable_condition): Do not specify vectype for
19311 vect_is_simple_cond when SLP vectorizing.
19313 2018-05-21 Michael Meissner <meissner@linux.ibm.com>
19316 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Do not
19317 define __ibm128 as long double.
19318 * config/rs6000/rs6000.c (rs6000_init_builtins): Create __ibm128
19319 as a distinct type when IEEE 128-bit support is enabled.
19320 (init_float128_ieee): Fix up conversions between IFmode and IEEE
19321 128-bit types to use the correct functions.
19322 (rs6000_expand_float128_convert): Use explicit FLOAT_EXTEND to
19323 convert between 128-bit floating point types that have different
19324 modes but the same representation, instead of using gen_lowpart to
19326 * config/rs6000/rs6000.md (IFKF): New iterator for IFmode and
19328 (IFKF_reg): New attributes to give the register constraints for
19330 (extend<mode>tf2_internal): New insns to mark an explicit
19331 conversion between 128-bit floating point types that have a
19332 different mode but share the same representation.
19334 2018-05-21 Richard Sandiford <richard.sandiford@linaro.org>
19336 PR tree-optimization/85814
19337 * tree-ssa-strlen.c (get_stridx_plus_constant): Cope with
19338 a null return from get_strinfo when unsharing the next
19339 strinfo in the chain.
19341 2018-05-21 Vladimir Mezentsev <vladimir.mezentsev@oracle.com>
19344 * varasm.c (weak_finish): Clean up weak_decls.
19346 2018-05-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19348 * config/aarch64/aarch64.md ("unspec"): Define UNSPEC_SABAL,
19349 UNSPEC_SABDL2, UNSPEC_SADALP, UNSPEC_UABAL, UNSPEC_UABDL2,
19350 UNSPEC_UADALP values.
19351 * config/aarch64/iterators.md (ABAL): New int iterator.
19354 (sur): Add mappings for the above.
19355 * config/aarch64/aarch64-simd.md (aarch64_<sur>abdl2<mode>_3):
19357 (aarch64_<sur>abal<mode>_4): Likewise.
19358 (aarch64_<sur>adalp<mode>_3): Likewise.
19359 (<sur>sadv16qi): New define_expand.
19361 2018-05-21 Alexander Nesterovskiy <alexander.nesterovskiy@intel.com>
19363 * config/i386/i386.md (*movsf_internal): AVX falsedep fix.
19364 (*movdf_internal): Ditto.
19365 (*rcpsf2_sse): Ditto.
19366 (*rsqrtsf2_sse): Ditto.
19367 (*sqrt<mode>2_sse): Ditto.
19369 2018-05-21 Tamar Christina <tamar.christina@arm.com>
19371 * config/aarch64/aarch64-simd.md (aarch64_eor3qv8hi): Change to
19373 (aarch64_bcaxqv8hi): Change to bcaxq<mode>4.
19374 * config/aarch64/aarch64-simd-builtins.def (veor3q_u8, veor3q_u32,
19375 veor3q_u64, veor3q_s8, veor3q_s16, veor3q_s32, veor3q_s64, vbcaxq_u8,
19376 vbcaxq_u32, vbcaxq_u64, vbcaxq_s8, vbcaxq_s16, vbcaxq_s32,
19378 * config/aarch64/arm_neon.h: Likewise.
19379 * config/aarch64/iterators.md (VQ_I): New.
19381 2018-05-21 Alexey Brodkin <abrodkin@synopsys.com>
19383 * config.gcc: Add arc/t-multilib-linux to tmake_file for
19385 * config/arc/t-multilib-linux: Specify MULTILIB_OPTIONS and
19388 2018-05-20 Chung-Ju Wu <jasonwucj@gmail.com>
19390 * config/nds32/constraints.md (S): New constraint.
19391 * config/nds32/nds32.md (call_internal): Use constraint S.
19392 (call_value_internal): Likewise.
19393 (sibcall_internal): Likewise.
19394 (sibcall_value_internal): Likewise.
19396 2018-05-20 Kito Cheng <kito.cheng@gmail.com>
19397 Chung-Ju Wu <jasonwucj@gmail.com>
19399 * config/nds32/nds32.c (nds32_register_move_cost): Take garywolf cpu
19400 into consideration.
19402 2018-05-20 Kito Cheng <kito.cheng@gmail.com>
19403 Chung-Ju Wu <jasonwucj@gmail.com>
19405 * config/nds32/nds32-cost.c (rtx_cost_model_t): New structure.
19406 (insn_size_16bit, insn_size_32bit): New variables for cost evaluation.
19407 (nds32_rtx_costs_impl): Simplify.
19408 (nds32_address_cost_impl): Simplify.
19409 (nds32_init_rtx_costs): New function.
19410 (nds32_rtx_costs_speed_prefer): Likewise.
19411 (nds32_rtx_costs_size_prefer): Likewise.
19412 (nds32_address_cost_speed_prefer): Likewise.
19413 (nds32_address_cost_speed_fwprop): Likewise.
19414 (nds32_address_cost_size_prefer): Likewise.
19415 * config/nds32/nds32-protos.h (nds32_init_rtx_costs): Declare.
19416 * config/nds32/nds32.c (nds32_option_override): Use
19417 nds32_init_rtx_costs function.
19419 2018-05-20 Chung-Ju Wu <jasonwucj@gmail.com>
19421 * config/nds32/nds32.c (nds32_asm_file_start): Output pipeline model.
19422 * config/nds32/nds32.h (TARGET_PIPELINE_N7): Define.
19423 (TARGET_PIPELINE_N8): Likewise.
19424 (TARGET_PIPELINE_N10): Likewise.
19425 (TARGET_PIPELINE_N13): Likewise.
19426 (TARGET_PIPELINE_GRAYWOLF): Likewise.
19428 2018-05-19 Monk Chiang <sh.chiang04@gmail.com>
19430 * config/nds32/nds32-fpu.md: Update copyright year.
19432 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
19434 * config/nds32/nds32.h (ASM_SPEC): Adjust spec rule.
19436 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
19438 * config/nds32/nds32.c
19439 (nds32_md_asm_adjust): Consider flag_inline_asm_r15 variable.
19440 * config/nds32/nds32.opt (minline-asm-r15): New option.
19442 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
19444 * common/config/nds32/nds32-common.c (TARGET_DEFAULT_TARGET_FLAGS): Add
19446 * config/nds32/nds32.md (abssi2): New pattern.
19448 2018-05-19 Uros Bizjak <ubizjak@gmail.com>
19450 * config/i386/i386.md (rex64namesuffix): New mode attribute.
19451 * config/i386/sse.md (sse_cvtsi2ss<rex64namesuffix><round_name>):
19452 Merge insn pattern from sse_cvtsi2ss<round_name> and
19453 sse_cvtsi2ssq<round_name> using SWI48 mode iterator.
19454 (sse_cvtss2si<rex64namesuffix><round_name>): Merge insn pattern
19455 from sse_cvtss2si<round_name> and sse_cvtss2siq<round_name>
19456 using SWI48 mode iterator.
19457 (sse_cvtss2si<rex64namesuffix>_2): Merge insn pattern from
19458 sse_cvtss2si_2 and sse_cvtss2siq_2 using SWI48 mode iterator.
19459 (sse_cvttss2si<rex64namesuffix><round_saeonly_name>): Merge insn
19460 pattern from sse_cvttss2si<round_saeonly_name>
19461 and sse_cvttss2siq<round_saeonly_name> using SWI48 mode iterator.
19462 (avx512f_vcvtss2usi<rex64namesuffix><round_name>): Merge insn pattern
19463 from avx512f_vcvtss2usi<round_name> and avx512f_vcvtss2usiq<round_name>
19464 using SWI48 mode iterator.
19465 (avx512f_vcvttss2usi<rex64namesuffix><round_saeonly_name>): Merge
19466 insn pattern from avx512f_vcvttss2usi<round_saeonly_name> and
19467 avx512f_vcvttss2usiq<round_saeonly_name> using SWI48 mode iterator.
19468 (avx512f_vcvtsd2usi<rex64namesuffix><round_name>): Merge insn pattern
19469 from avx512f_vcvtsd2usi<round_name> and avx512f_vcvtsd2usiq<round_name>
19470 using SWI48 mode iterator.
19471 (avx512f_vcvttsd2usi<rex64namesuffix><round_saeonly_name>): Merge
19472 insn pattern from avx512f_vcvttsd2usi<round_saeonly_name> and
19473 avx512f_vcvttsd2usiq<round_saeonly_name> using SWI48 mode iterator.
19474 (sse2_cvtsd2si<rex64namesuffix><round_name>): Merge insn pattern from
19475 sse2_cvtsd2si<round_name> and sse2_cvtsd2siq<round_name> using
19476 SWI48 mode iterator.
19477 (sse2_cvtsd2si<rex64namesuffix>_2): Merge insn pattern from
19478 sse2_cvtsd2si_2 and sse2_cvtsd2siq_2 using SWI48 mode iterator.
19479 (sse_cvttsd2si<rex64namesuffix><round_saeonly_name>): Merge insn
19480 pattern from sse_cvttsd2si<round_saeonly_name>
19481 and sse_cvttsd2siq<round_saeonly_name> using SWI48 mode iterator.
19483 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
19485 * config/nds32/nds32-md-auxiliary.c
19486 (nds32_valid_smw_lwm_base_p): Refine.
19487 (nds32_output_smw_single_word): Refine.
19488 (nds32_output_smw_double_word): New.
19489 * config/nds32/nds32-protos.h (nds32_output_smw_double_word): New.
19491 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
19493 * config/nds32/nds32-md-auxiliary.c (nds32_output_stack_push): Refine.
19494 (nds32_output_stack_pop): Refine.
19495 (nds32_expand_unaligned_load): Refine.
19496 (nds32_expand_unaligned_store): Refine.
19498 2018-05-19 Kuan-Lin Chen <kuanlinchentw@gmail.com>
19499 Chung-Ju Wu <jasonwucj@gmail.com>
19501 * config/nds32/constants.md: Add TP_REGNUM constant.
19502 (unspec_element): Add UNSPEC_GOTINIT, UNSPEC_GOT, UNSPEC_GOTOFF,
19503 UNSPEC_PLT, UNSPEC_TLSGD, UNSPEC_TLSLD, UNSPEC_TLSIE, UNSPEC_TLSLE and
19505 * config/nds32/nds32-doubleword.md: Consider flag_pic.
19506 * config/nds32/nds32-dspext.md (mov<mode>): Expand TLS and PIC cases.
19507 * config/nds32/nds32-predicates.c (nds32_const_unspec_p): New.
19508 * config/nds32/nds32-md-auxiliary.c: Implementation that support TLS
19509 and PIC code generation.
19510 * config/nds32/nds32-protos.h: Declarations that support TLS and PIC
19512 * config/nds32/nds32-relax-opt.c: Consider TLS and PIC for relax
19514 * config/nds32/nds32.md: Support TLS and PIC.
19515 * config/nds32/nds32.c: Support TLS and PIC.
19516 * config/nds32/nds32.h (nds32_relax_insn_type): New enum type.
19517 * config/nds32/predicates.md (nds32_nonunspec_symbolic_operand): New
19520 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
19522 * config/nds32/nds32-predicates.c (const_vector_to_hwint): Use machine
19523 mode with E_ prefix.
19525 2018-05-19 Kuan-Lin Chen <kuanlinchentw@gmail.com>
19526 Chung-Ju Wu <jasonwucj@gmail.com>
19528 * config/nds32/constants.md (unspec_element): Add UNSPEC_ICT.
19529 * config/nds32/nds32-md-auxiliary.c
19530 (symbolic_reference_mentioned_p): New.
19531 (nds32_legitimize_ict_address): New.
19532 (nds32_expand_ict_move): New.
19533 (nds32_indirect_call_referenced_p): New.
19534 (nds32_symbol_binds_local_p): Delete.
19535 (nds32_long_call_p): Modify.
19536 * config/nds32/nds32-opts.h (nds32_ict_model_type): New enum type.
19537 * config/nds32/nds32-protos.h
19538 (symbolic_reference_mentioned_p): Declare.
19539 (nds32_legitimize_ict_address): Declare.
19540 (nds32_expand_ict_move): Declare.
19541 (nds32_indirect_call_referenced_p): Declare.
19542 * config/nds32/nds32-relax-opt.c (nds32_ict_const_p): New.
19543 (nds32_relax_group): Use nds32_ict_const_p as condition.
19544 * config/nds32/nds32.c (nds32_attribute_table): Add "indirect_call".
19545 (nds32_asm_file_start): Output ict_model directive in asm code.
19546 (nds32_legitimate_address_p): Consider indirect call.
19547 (nds32_print_operand): Consider indirect call.
19548 (nds32_print_operand_address): Consider indirect call.
19549 (nds32_insert_attributes): Handle "indirect_call" attribute.
19550 (TARGET_LEGITIMATE_ADDRESS_P): Define.
19551 (TARGET_LEGITIMATE_CONSTANT_P): Define.
19552 (TARGET_CANNOT_FORCE_CONST_MEM): Define.
19553 (TARGET_DELEGITIMIZE_ADDRESS): Define.
19554 (TARGET_ASM_OUTPUT_ADDR_CONST_EXTRA): Define.
19555 * config/nds32/nds32.h (SYMBOLIC_CONST_P): Define.
19556 (TARGET_ICT_MODEL_SMALL): Define.
19557 (TARGET_ICT_MODEL_LARGE): Define.
19558 * config/nds32/nds32.md (movsi): Consider ict model.
19559 (call, call_value): Consider ict model.
19560 (sibcall, sibcall_value): Consider ict model.
19561 * config/nds32/nds32.opt (mict-model): New option.
19562 * config/nds32/predicates.md (nds32_symbolic_operand): Consider ict
19565 2018-05-18 Kito Cheng <kito.cheng@gmail.com>
19566 Monk Chiang <sh.chiang04@gmail.com>
19567 Jim Wilson <jimw@sifive.com>
19569 * common/config/riscv/riscv-common.c (riscv_parse_arch_string):
19570 Add support to parse rv32e*. Clear MASK_RVE for rv32i and rv64i.
19571 * config.gcc (riscv*-*-*): Add support for rv32e* and ilp32e.
19572 * config/riscv/riscv-c.c (riscv_cpu_cpp_builtins): Define
19573 __riscv_32e when TARGET_RVE. Handle ABI_ILP32E as soft-float ABI.
19574 * config/riscv/riscv-opts.h (riscv_abi_type): Add ABI_ILP32E.
19575 * config/riscv/riscv.c (riscv_compute_frame_info): When TARGET_RVE,
19576 compute save_libcall_adjustment properly.
19577 (riscv_option_override): Call error if TARGET_RVE and not ABI_ILP32E.
19578 (riscv_conditional_register_usage): Handle TARGET_RVE and ABI_ILP32E.
19579 * config/riscv/riscv.h (UNITS_PER_FP_ARG): Handle ABI_ILP32E.
19580 (STACK_BOUNDARY, ABI_STACK_BOUNDARY): Handle TARGET_RVE.
19581 (GP_REG_LAST, MAX_ARGS_IN_REGISTERS): Likewise.
19582 (ABI_SPEC): Handle mabi=ilp32e.
19583 * config/riscv/riscv.opt (abi_type): Add ABI_ILP32E.
19584 (RVE): Add RVE mask.
19585 * doc/invoke.texi (RISC-V options) <-mabi>: Add ilp32e info.
19586 <-march>: Add rv32e as an example.
19588 2018-05-18 Marc Glisse <marc.glisse@inria.fr>
19591 * tree-ssa-structalias.c (create_variable_info_for_1): Extra argument.
19592 (intra_create_variable_infos): Handle C++ constructors.
19594 2018-05-18 Martin Liska <mliska@suse.cz>
19596 * passes.def: Remove a redundant pass.
19598 2018-05-18 Eric Botcazou <ebotcazou@adacore.com>
19601 * config/sparc/sparc.c (sparc_expand_builtin): Always initialize op[0].
19603 2018-05-18 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19605 * config/arm/arm-cpus.in (armv3m, mode32): Delete features.
19607 (ARMv2, ARMv3, ARMv3m): Delete fgroups.
19609 (armv2, armv2a, armv3, armv3m): Delete architectures.
19610 (arm2, arm250, arm3, arm6, arm60, arm600, arm610, arm620,
19611 arm7, arm7d, arm7di, arm70, arm700, arm700i, arm710, arm720,
19612 arm710c, arm7100, arm7500, arm7500fe, arm7m, arm7dm, arm7dmi):
19614 * config/arm/arm.md (maddsidi4): Remove check for arm_arch3m.
19615 (*mulsidi3adddi): Likewise.
19616 (mulsidi3): Likewise.
19617 (*mulsidi3_nov6): Likewise.
19618 (umulsidi3): Likewise.
19619 (umulsidi3_nov6): Likewise.
19620 (umaddsidi4): Likewise.
19621 (*umulsidi3adddi): Likewise.
19622 (smulsi3_highpart): Likewise.
19623 (*smulsi3_highpart_nov6): Likewise.
19624 (umulsi3_highpart): Likewise.
19625 (*umulsi3_highpart_nov6): Likewise.
19626 * config/arm/arm.h (arm_arch3m): Delete.
19627 * config/arm/arm.c (arm_arch3m): Delete.
19628 (arm_option_override_internal): Update armv3-related comment.
19629 (arm_configure_build_target): Delete use of isa_bit_mode32.
19630 (arm_option_reconfigure_globals): Delete set of arm_ach3m.
19631 (arm_rtx_costs_internal): Delete check of arm_arch3m.
19632 * config/arm/arm-fixed.md (mulsq3): Delete check for arm_arch3m.
19633 (mulsa3): Likewise.
19634 (mulusa3): Likewise.
19635 * config/arm/arm-protos.h (arm_arch3m): Delete.
19636 * config/arm/arm-tables.opt: Regenerate.
19637 * config/arm/arm-tune.md: Likewise.
19638 * config/arm/t-arm-elf (all_early_nofp): Delete mentions of
19639 deleted architectures.
19641 2018-05-18 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19643 * config/arm/arm-cpus.in (armv5, armv5e): Delete features.
19644 (armv5t, armv5te): New features.
19645 (ARMv5, ARMv5e): Delete fgroups.
19646 (ARMv5t, ARMv5te): Adjust for above changes.
19647 (ARMv6m): Likewise.
19648 (armv5, armv5e): Delete arches.
19649 * config/arm/arm.md (*call_reg_armv5): Use arm_arch5t instead of
19651 (*call_reg_arm): Likewise.
19652 (*call_value_reg_armv5): Likewise.
19653 (*call_value_reg_arm): Likewise.
19654 (*call_symbol): Likewise.
19655 (*call_value_symbol): Likewise.
19656 (*sibcall_insn): Likewise.
19657 (*sibcall_value_insn): Likewise.
19658 (clzsi2): Likewise.
19659 (prefetch): Likewise.
19660 (define_split and define_peephole2 dependent on arm_arch5):
19662 * config/arm/arm.h (TARGET_LDRD): Use arm_arch5te instead of
19664 (TARGET_ARM_QBIT): Likewise.
19665 (TARGET_DSP_MULTIPLY): Likewise.
19666 (enum base_architecture): Delete BASE_ARCH_5, BASE_ARCH_5E.
19667 (arm_arch5, arm_arch5e): Delete.
19668 (arm_arch5t, arm_arch5te): Declare.
19669 * config/arm/arm.c (arm_arch5, arm_arch5e): Delete.
19670 (arm_arch5t): Declare.
19671 (arm_option_reconfigure_globals): Update for the above.
19672 (arm_options_perform_arch_sanity_checks): Update comment, replace
19673 use of arm_arch5 with arm_arch5t.
19674 (use_return_insn): Likewise.
19675 (arm_emit_call_insn): Likewise.
19676 (output_return_instruction): Likewise.
19677 (arm_final_prescan_insn): Likewise.
19678 (arm_coproc_builtin_available): Likewise.
19679 * config/arm/arm-c.c (arm_cpu_builtins): Replace arm_arch5 and
19680 arm_arch5e with arm_arch5t and arm_arch5te.
19681 * config/arm/arm-protos.h (arm_arch5, arm_arch5e): Delete.
19682 (arm_arch5t, arm_arch5te): Declare.
19683 * config/arm/arm-tables.opt: Regenerate.
19684 * config/arm/t-arm-elf: Remove references to armv5, armv5e.
19685 * config/arm/t-multilib: Likewise.
19686 * config/arm/thumb1.md (*call_reg_thumb1_v5): Check arm_arch5t
19687 instead of arm_arch5.
19688 (*call_reg_thumb1): Likewise.
19689 (*call_value_reg_thumb1_v5): Likewise.
19690 (*call_value_reg_thumb1): Likewise.
19691 * config/arm/vxworks.h (TARGET_OS_CPP_BUILTINS): Remove now
19693 * doc/invoke.texi (ARM Options): Remove references to armv5, armv5e.
19695 2018-05-18 Martin Liska <mliska@suse.cz>
19697 PR gcov-profile/84846
19698 * doc/gcov.texi: Document -t option of gcov tool.
19700 2018-05-18 Martin Liska <mliska@suse.cz>
19702 PR gcov-profile/84846
19703 * gcov.c (print_usage): Add new -t option.
19704 (process_args): Handle the option.
19705 (generate_results): Use stdout as output when requested by
19708 2018-05-18 Martin Liska <mliska@suse.cz>
19710 PR gcov-profile/84846
19711 * coverage.c (coverage_init): Write PWD to .gcno file.
19712 * doc/gcov.texi: Document how working directory is printed.
19713 * gcov-dump.c (dump_gcov_file): Print PWD.
19714 * gcov.c (output_intermediate_file): Likewise.
19715 (read_graph_file): Read PWD string.
19716 (output_lines): Print PWD.
19718 2018-05-18 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
19720 PR middle-end/85817
19721 * ipa-pure-const.c (malloc_candidate_p): Remove the check integer_zerop
19722 for retval and return false if all args to phi are zero.
19724 2018-05-18 Richard Biener <rguenther@suse.de>
19726 * gimple-ssa-evrp.c (class evrp_folder): Add simplify_stmt_using_ranges
19728 (evrp_dom_walker::before_dom_children): Call it.
19730 2018-05-18 Richard Biener <rguenther@suse.de>
19732 * tree-dfa.c (get_ref_base_and_extent): Use range-info to refine
19733 results when processing array refs with variable index.
19735 2018-05-18 Toon Moene <toon@moene.org>
19737 * doc/invoke.texi: Move -floop-unroll-and-jam documentation
19738 directly after that of -floop-interchange. Indicate that both
19739 options are enabled by default when specifying -O3.
19741 2018-05-18 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19743 * config/aarch64/aarch64-simd.md (vec_set<mode>): Use VALL_F16 mode
19744 iterator. Delete separate integer-mode vec_set<mode> expander.
19745 (aarch64_simd_vec_setv2di): Delete.
19746 (vec_setv2di): Delete.
19747 (aarch64_simd_vec_set<mode>): Delete all other patterns with that name.
19748 Use VALL_F16 mode iterator. Add LD1 alternative and use vwcore for
19749 the "w, r" alternative.
19751 2018-05-18 Martin Liska <mliska@suse.cz>
19753 * passes.def: Add pass_lower_switch and pass_lower_switch_O0.
19754 * tree-pass.h (make_pass_lower_switch_O0): New function.
19755 * tree-switch-conversion.c (node_has_low_bound): Remove.
19756 (node_has_high_bound): Likewise.
19757 (node_is_bounded): Likewise.
19758 (class pass_lower_switch): Make it a template type and create
19760 (pass_lower_switch::execute): Add template argument.
19761 (make_pass_lower_switch): New function.
19762 (make_pass_lower_switch_O0): New function.
19763 (do_jump_if_equal): Remove.
19764 (emit_case_nodes): Simplify to just handle all 3 cases and leave
19765 all the hard work to tree optimization passes.
19767 2018-05-18 Martin Liska <mliska@suse.cz>
19769 * dbgcnt.c (limit_low): Renamed from limit.
19770 (limit_high): New variable.
19771 (dbg_cnt_is_enabled): Check for upper limit.
19772 (dbg_cnt): Adjust dumping.
19773 (dbg_cnt_set_limit_by_index): Add new argument for high
19775 (dbg_cnt_set_limit_by_name): Likewise.
19776 (dbg_cnt_process_single_pair): Parse new format.
19777 (dbg_cnt_process_opt): Use strtok.
19778 (dbg_cnt_list_all_counters): Remove 'value' and add
19780 * doc/invoke.texi: Document changes.
19782 2018-05-18 Richard Sandiford <richard.sandiford@linaro.org>
19784 * doc/sourcebuild.texi (scalar_all_fma): Document.
19785 * tree.def (FMA_EXPR): Delete.
19786 * internal-fn.def (FMA, FMS, FNMA, FNMS): New internal functions.
19787 * internal-fn.c (ternary_direct): New macro.
19788 (expand_ternary_optab_fn): Likewise.
19789 (direct_ternary_optab_supported_p): Likewise.
19790 * Makefile.in (build/genmatch.o): Depend on case-fn-macros.h.
19791 * builtins.c (fold_builtin_fma): Delete.
19792 (fold_builtin_3): Don't call it.
19793 * cfgexpand.c (expand_debug_expr): Remove FMA_EXPR handling.
19794 * expr.c (expand_expr_real_2): Likewise.
19795 * fold-const.c (operand_equal_p): Likewise.
19796 (fold_ternary_loc): Likewise.
19797 * gimple-pretty-print.c (dump_ternary_rhs): Likewise.
19798 * gimple.c (DEFTREECODE): Likewise.
19799 * gimplify.c (gimplify_expr): Likewise.
19800 * optabs-tree.c (optab_for_tree_code): Likewise.
19801 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
19802 * tree-eh.c (operation_could_trap_p): Likewise.
19803 (stmt_could_throw_1_p): Likewise.
19804 * tree-inline.c (estimate_operator_cost): Likewise.
19805 * tree-pretty-print.c (dump_generic_node): Likewise.
19806 (op_code_prio): Likewise.
19807 * tree-ssa-loop-im.c (stmt_cost): Likewise.
19808 * tree-ssa-operands.c (get_expr_operands): Likewise.
19809 * tree.c (commutative_ternary_tree_code, add_expr): Likewise.
19810 * fold-const-call.h (fold_fma): Delete.
19811 * fold-const-call.c (fold_const_call_ssss): Handle CFN_FMS,
19812 CFN_FNMA and CFN_FNMS.
19813 (fold_fma): Delete.
19814 * genmatch.c (combined_fn): New enum.
19815 (commutative_ternary_tree_code): Remove FMA_EXPR handling.
19816 (commutative_op): New function.
19817 (commutate): Use it. Handle more than 2 operands.
19818 (dt_operand::gen_gimple_expr): Use commutative_op.
19819 (parser::parse_expr): Allow :c to be used with non-binary
19820 operators if the commutative operand is known.
19821 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Handle
19822 CFN_FMS, CFN_FNMA and CFN_FNMS.
19823 (backprop::process_assign_use): Remove FMA_EXPR handling.
19824 * hsa-gen.c (gen_hsa_insns_for_operation_assignment): Likewise.
19825 (gen_hsa_fma): New function.
19826 (gen_hsa_insn_for_internal_fn_call): Use it for IFN_FMA, IFN_FMS,
19827 IFN_FNMA and IFN_FNMS.
19828 * match.pd: Add folds for IFN_FMS, IFN_FNMA and IFN_FNMS.
19829 * gimple-fold.h (follow_all_ssa_edges): Declare.
19830 * gimple-fold.c (follow_all_ssa_edges): New function.
19831 * tree-ssa-math-opts.c (convert_mult_to_fma_1): Use the
19832 gimple_build interface and use follow_all_ssa_edges to fold the result.
19833 (convert_mult_to_fma): Use direct_internal_fn_suppoerted_p
19834 instead of checking for optabs directly.
19835 * config/i386/i386.c (ix86_add_stmt_cost): Recognize FMAs as calls
19836 rather than FMA_EXPRs.
19837 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Create a
19838 call to IFN_FMA instead of an FMA_EXPR.
19840 2018-05-17 Jim Wilson <jimw@sifive.com>
19842 * expr.c (do_tablejump): When converting index to Pmode, if we have a
19843 sign extended promoted subreg, and the range does not have the sign bit
19844 set, then do a sign extend.
19846 * config/riscv/riscv.c (riscv_extend_comparands): In unsigned QImode
19847 test, check for sign extended subreg and/or constant operands, and
19848 do a sign extend in that case.
19850 2018-05-17 Steve Ellcey <sellcey@cavium.com>
19852 * config/aarch64/thunderx2t99.md (thunderx2t99_ls_both): Delete.
19853 (thunderx2t99_multiple): Delete psuedo-units from used cpus.
19855 (thunderx2t99_alu_shift): Remove alu_shift_reg, alus_shift_reg.
19856 Change logics_shift_reg to logics_shift_imm.
19857 (thunderx2t99_fp_loadpair_basic): Delete.
19858 (thunderx2t99_fp_storepair_basic): Delete.
19859 (thunderx2t99_asimd_int): Add neon_sub and neon_sub_q types.
19860 (thunderx2t99_asimd_polynomial): Delete.
19861 (thunderx2t99_asimd_fp_simple): Add neon_fp_mul_s_scalar_q
19862 and neon_fp_mul_d_scalar_q.
19863 (thunderx2t99_asimd_fp_conv): Add *int_to_fp* types.
19864 (thunderx2t99_asimd_misc): Delete neon_dup and neon_dup_q.
19865 (thunderx2t99_asimd_recip_step): Add missing *sqrt* types.
19866 (thunderx2t99_asimd_lut): Add missing tbl types.
19867 (thunderx2t99_asimd_ext): Delete.
19868 (thunderx2t99_asimd_load1_1_mult): Delete.
19869 (thunderx2t99_asimd_load1_2_mult): Delete.
19870 (thunderx2t99_asimd_load1_ldp): New.
19871 (thunderx2t99_asimd_load1): New.
19872 (thunderx2t99_asimd_load2): Add missing *load2* types.
19873 (thunderx2t99_asimd_load3): New.
19874 (thunderx2t99_asimd_load4): New.
19875 (thunderx2t99_asimd_store1_1_mult): Delete.
19876 (thunderx2t99_asimd_store1_2_mult): Delete.
19877 (thunderx2t99_asimd_store2_mult): Delete.
19878 (thunderx2t99_asimd_store2_onelane): Delete.
19879 (thunderx2t99_asimd_store_stp): New.
19880 (thunderx2t99_asimd_store1): New.
19881 (thunderx2t99_asimd_store2): New.
19882 (thunderx2t99_asimd_store3): New.
19883 (thunderx2t99_asimd_store4): New.
19885 2018-05-17 Jerome Lambourg <lambourg@adacore.com>
19887 * config/arm/arm_cmse.h (cmse_nsfptr_create, cmse_is_nsfptr): Remove
19888 #include <stdint.h>. Replace intptr_t with __INTPTR_TYPE__.
19890 2018-05-17 Pat Haugen <pthaugen@us.ibm.com>
19891 Segher Boessenkool <segher@kernel.crashing.org>
19894 * config/rs6000/rs6000.c (rs6000_output_move_128bit): Check dest
19897 2018-05-17 Richard Biener <rguenther@suse.de>
19899 * tree-ssa-dse.c (dse_classify_store): Fix iterator increment
19900 for pruning loop and prune defs feeding only already visited PHIs.
19902 2018-05-17 Richard Biener <rguenther@suse.de>
19904 * tree-ssa-sccvn.c (vn_reference_lookup_3): Improve memset handling.
19906 2018-05-17 Bin Cheng <bin.cheng@arm.com>
19907 Richard Biener <rguenther@suse.de>
19909 PR tree-optimization/85793
19910 * tree-vect-stmts.c (vectorizable_load): Handle 1 element-wise load
19911 for VMAT_ELEMENTWISE.
19913 2018-05-17 Richard Sandiford <richard.sandiford@linaro.org>
19915 * internal-fn.h (lookup_internal_fn): Declare
19916 * internal-fn.c (lookup_internal_fn): New function.
19917 * gimple.c (gimple_build_call_from_tree): Handle calls to
19918 internal functions.
19919 * gimple-pretty-print.c (dump_gimple_call): Print "." before
19920 internal function names.
19921 * tree-pretty-print.c (dump_generic_node): Likewise.
19922 * tree-ssa-scopedtables.c (expr_hash_elt::print): Likewise.
19924 2018-05-17 Richard Sandiford <richard.sandiford@linaro.org>
19926 * gimple-fold.h (gimple_build): Make the function forms take
19927 combined_fn rather than built_in_function.
19928 (gimple_simplify): Likewise.
19929 * gimple-match-head.c (gimple_simplify): Likewise.
19930 * gimple-fold.c (gimple_build): Likewise.
19931 * tree-vect-loop.c (get_initial_def_for_reduction): Use gimple_build
19932 rather than gimple_build_call_internal.
19933 (get_initial_defs_for_reduction): Likewise.
19934 (vect_create_epilog_for_reduction): Likewise.
19935 (vectorizable_live_operation): Likewise.
19937 2018-05-17 Martin Liska <mliska@suse.cz>
19939 * gimple-ssa-sprintf.c (format_directive): Do not use
19940 space in between 'G_' and '('.
19942 2018-05-17 Jakub Jelinek <jakub@redhat.com>
19945 * config/i386/i386.c (ix86_fold_builtin): Handle masked shifts
19946 even if the mask is not all ones.
19949 * config/i386/i386.c (ix86_fold_builtin): Fold shift builtins by
19951 (ix86_gimple_fold_builtin): Likewise.
19954 * config/i386/i386.c: Include tree-vector-builder.h.
19955 (ix86_vector_shift_count): New function.
19956 (ix86_fold_builtin): Fold shift builtins by scalar count.
19957 (ix86_gimple_fold_builtin): Likewise.
19959 * config/i386/avx512fintrin.h (_mm512_set_epi16, _mm512_set_epi8,
19960 _mm512_setzero): New intrinsics.
19962 2018-05-17 James Greenhalgh <james.greenhalgh@arm.com>
19963 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19965 * config/aarch64/aarch64.c (aarch64_expand_vector_init): Modify
19966 code generation for cases where splatting a value is not useful.
19967 * simplify-rtx.c (simplify_ternary_operation): Simplify
19968 vec_merge across a vec_duplicate and a paradoxical subreg forming
19969 a vector mode to a vec_concat.
19971 2018-05-17 Olga Makhotina <olga.makhotina@intel.com>
19973 * config.gcc: Support "goldmont-plus".
19974 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
19976 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
19977 PROCESSOR_GOLDMONT_PLUS.
19978 * config/i386/i386.c (m_GOLDMONT_PLUS): Define.
19979 (processor_target_table): Add "goldmont-plus".
19980 (PTA_GOLDMONT_PLUS): Define.
19981 (ix86_lea_outperforms): Add TARGET_GOLDMONT_PLUS.
19982 (get_builtin_code_for_version): Handle PROCESSOR_GOLDMONT_PLUS.
19983 (fold_builtin_cpu): Add M_INTEL_GOLDMONT_PLUS.
19984 (fold_builtin_cpu): Add "goldmont-plus".
19985 (ix86_add_stmt_cost): Add TARGET_GOLDMONT_PLUS.
19986 (ix86_option_override_internal): Add "goldmont-plus".
19987 * config/i386/i386.h (processor_costs): Define TARGET_GOLDMONT_PLUS.
19988 (processor_type): Add PROCESSOR_GOLDMONT_PLUS.
19989 * config/i386/x86-tune.def: Add m_GOLDMONT_PLUS.
19990 * doc/invoke.texi: Add goldmont-plus as x86 -march=/-mtune= CPU type.
19992 2018-05-17 Richard Biener <rguenther@suse.de>
19994 PR tree-optimization/85757
19995 * tree-ssa-dse.c (dse_classify_store): Record a PHI def and
19996 remove defs that only feed that PHI from further processing.
19998 2018-05-16 Jim Wilson <jimw@sifive.com>
20000 * config/riscv/riscv.md (<optab>si3_mask, <optab>si3_mask_1): Prepend
20002 (<optab>di3_mask, <optab>di3_mask_1): Likewise.
20004 2018-05-16 Mark Wielaard <mark@klomp.org>
20006 * dwarf2out.c (count_index_strings): New function.
20007 (output_indirect_strings): Call count_index_strings and generate
20008 header for dwarf_version >= 5.
20010 2018-05-16 Mark Wielaard <mark@klomp.org>
20012 * dwarf2out.c (dwarf_FORM): New function.
20013 (set_indirect_string): Use dwarf_FORM.
20014 (reset_indirect_string): Likewise.
20015 (size_of_die): Likewise.
20016 (value_format): Likewise.
20017 (output_die): Likewise.
20018 (add_skeleton_AT_string): Likewise.
20019 (output_macinfo_op): Likewise.
20020 (index_string): Likewise.
20021 (output_index_string_offset): Likewise.
20022 (output_index_string): Likewise.
20023 (count_index_strings): Likewise.
20025 2018-05-16 Carl Love <cel@us.ibm.com>
20027 * config/rs6000/rs6000.md (prefetch): Generate ISA 2.06 instructions
20028 dcbt and dcbtstt with TH=16 if operands[2] is 0 and Power 8 or newer.
20030 2018-05-16 Martin Jambor <mjambor@suse.cz>
20032 * ipa-prop.c (ipa_free_all_edge_args): Remove.
20033 * ipa-prop.h (ipa_free_all_edge_args): Likewise.
20035 2018-05-16 Wilco Dijkstra <wdijkstr@arm.com>
20037 * config/aarch64/aarch64.md (fma<mode>4): Change into expand pattern.
20038 (fnma<mode>4): Likewise.
20039 (fms<mode>4): Likewise.
20040 (fnms<mode>4): Likewise.
20041 (aarch64_fma<mode>4): Rename insn, reorder accumulator operand.
20042 (aarch64_fnma<mode>4): Likewise.
20043 (aarch64_fms<mode>4): Likewise.
20044 (aarch64_fnms<mode>4): Likewise.
20045 (aarch64_fnmadd<mode>4): Likewise.
20047 2018-05-16 Jason Merrill <jason@redhat.com>
20049 * tree.c (warn_deprecated_use): Return bool. Simplify logic.
20051 2018-05-16 Richard Biener <rguenther@suse.de>
20053 * tree-vectorizer.h (struct stmt_info_for_cost): Add where member.
20054 (dump_stmt_cost): Declare.
20055 (add_stmt_cost): Dump cost we add.
20056 (add_stmt_costs): New function.
20057 (vect_model_simple_cost, vect_model_store_cost, vect_model_load_cost):
20058 No longer exported.
20059 (vect_analyze_stmt): Adjust prototype.
20060 (vectorizable_condition): Likewise.
20061 (vectorizable_live_operation): Likewise.
20062 (vectorizable_reduction): Likewise.
20063 (vectorizable_induction): Likewise.
20064 * tree-vect-loop.c (vect_analyze_loop_operations): Create local
20065 cost vector to pass to vectorizable_ and record afterwards.
20066 (vect_model_reduction_cost): Take cost vector argument and adjust.
20067 (vect_model_induction_cost): Likewise.
20068 (vectorizable_reduction): Likewise.
20069 (vectorizable_induction): Likewise.
20070 (vectorizable_live_operation): Likewise.
20071 * tree-vect-slp.c (vect_create_new_slp_node): Initialize
20072 SLP_TREE_NUMBER_OF_VEC_STMTS.
20073 (vect_analyze_slp_cost_1): Remove.
20074 (vect_analyze_slp_cost): Likewise.
20075 (vect_slp_analyze_node_operations): Take visited args and
20076 a target cost vector. Avoid processing already visited stmt sets.
20077 (vect_slp_analyze_operations): Use a local cost vector to gather
20078 costs and register those of non-discarded instances.
20079 (vect_bb_vectorization_profitable_p): Use add_stmt_costs.
20080 (vect_schedule_slp_instance): Remove copying of
20081 SLP_TREE_NUMBER_OF_VEC_STMTS. Instead assert that it is not
20083 * tree-vect-stmts.c (record_stmt_cost): Remove path directly
20084 adding cost. Record cost entry location.
20085 (vect_prologue_cost_for_slp_op): Function to compute cost of
20086 a constant or invariant generated for SLP vect in the prologue,
20087 split out from vect_analyze_slp_cost_1.
20088 (vect_model_simple_cost): Make static. Adjust for SLP costing.
20089 (vect_model_promotion_demotion_cost): Likewise.
20090 (vect_model_store_cost): Likewise, make static.
20091 (vect_model_load_cost): Likewise.
20092 (vectorizable_bswap): Add cost vector arg and adjust.
20093 (vectorizable_call): Likewise.
20094 (vectorizable_simd_clone_call): Likewise.
20095 (vectorizable_conversion): Likewise.
20096 (vectorizable_assignment): Likewise.
20097 (vectorizable_shift): Likewise.
20098 (vectorizable_operation): Likewise.
20099 (vectorizable_store): Likewise.
20100 (vectorizable_load): Likewise.
20101 (vectorizable_condition): Likewise.
20102 (vectorizable_comparison): Likewise.
20103 (can_vectorize_live_stmts): Likewise.
20104 (vect_analyze_stmt): Likewise.
20105 (vect_transform_stmt): Adjust calls to vectorizable_*.
20106 * tree-vectorizer.c: Include gimple-pretty-print.h.
20107 (dump_stmt_cost): New function.
20109 2018-05-16 Richard Biener <rguenther@suse.de>
20111 * params.def (PARAM_DSE_MAX_ALIAS_QUERIES_PER_STORE): New param.
20112 * doc/invoke.texi (dse-max-alias-queries-per-store): Document.
20113 * tree-ssa-dse.c: Include tree-ssa-loop.h.
20114 (check_name): New callback.
20115 (dse_classify_store): Track cycles via a visited bitmap of PHI
20116 defs and simplify handling of in-loop and across loop dead stores
20117 and properly fail for loop-variant refs. Handle byte-tracking with
20118 multiple defs. Use PARAM_DSE_MAX_ALIAS_QUERIES_PER_STORE for
20121 2018-05-16 Richard Sandiford <richard.sandiford@linaro.org>
20123 * tree-vectorizer.h (vect_get_vector_types_for_stmt): Declare.
20124 (vect_get_mask_type_for_stmt): Likewise.
20125 * tree-vect-slp.c (vect_two_operations_perm_ok_p): New function,
20127 (vect_build_slp_tree_1): ...here. Use vect_get_vector_types_for_stmt
20128 to determine the statement's vector type and the vector type that
20129 should be used for calculating nunits. Deal with cases in which
20130 the type has to be deferred.
20131 (vect_slp_analyze_node_operations): Use vect_get_vector_types_for_stmt
20132 and vect_get_mask_type_for_stmt to calculate STMT_VINFO_VECTYPE.
20133 * tree-vect-loop.c (vect_determine_vf_for_stmt_1)
20134 (vect_determine_vf_for_stmt): New functions, split out from...
20135 (vect_determine_vectorization_factor): ...here.
20136 * tree-vect-stmts.c (vect_get_vector_types_for_stmt)
20137 (vect_get_mask_type_for_stmt): New functions, split out from
20138 vect_determine_vectorization_factor.
20140 2018-05-16 Richard Biener <rguenther@suse.de>
20142 * tree-cfg.c (verify_gimple_assign_ternary): Properly
20143 verify the [VEC_]COND_EXPR embedded comparison.
20145 2018-05-15 Martin Sebor <msebor@redhat.com>
20147 PR tree-optimization/85753
20148 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Handle
20149 RECORD_TYPE in addition to ARRAY_TYPE.
20151 2018-05-15 Martin Sebor <msebor@redhat.com>
20153 PR middle-end/85643
20154 * calls.c (get_attr_nonstring_decl): Handle MEM_REF.
20156 2018-05-15 Richard Biener <rguenther@suse.de>
20158 * tree-ssa-dse.c (dse_classify_store): Remove use_stmt parameter,
20159 add by_clobber_p one. Change algorithm to collect all defs
20160 representing uses we need to walk and try reducing them to
20161 a single one before failing.
20162 (dse_dom_walker::dse_optimize_stmt): Adjust.
20164 2018-05-13 Mark Wielaard <mark@klomp.org>
20166 * dwarf2out.c (dwarf_OP): Handle DW_OP_addrx and DW_OP_constx.
20167 (size_of_loc_descr): Likewise.
20168 (output_loc_operands): Likewise.
20169 (output_loc_operands_raw): Likewise.
20170 (dw_addr_op): Use dwarf_OP () for DW_OP_constx and DW_OP_addrx.
20171 (resolve_addr_in_expr): Handle DW_OP_addrx and DW_OP_constx.
20172 (hash_loc_operands): Likewise.
20173 (compare_loc_operands): Likewise.
20175 2018-05-14 Mark Wielaard <mark@klomp.org>
20177 * dwarf2out.c (count_index_addrs): New function.
20178 (dwarf2out_finish): Use count_index_addrs to calculate addrs_length.
20180 2018-05-15 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
20182 PR tree-optimization/83648
20183 * ipa-pure-const.c (malloc_candidate_p): Allow function with NULL
20184 return value as malloc candidate.
20186 2018-05-15 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
20189 * ipa-pure-const.c (warn_function_malloc): Pass value of known_finite
20190 param as true in call to suggest_attribute.
20192 2018-05-14 Segher Boessenkool <segher@kernel.crashing.org>
20194 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete @itemx for
20195 -mreadonly-in-sdata.
20197 2018-05-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20199 * config/aarch64/aarch64-simd.md (*aarch64_crypto_aese_fused):
20201 (aarch64_crypto_aesd_fused): Likewise.
20203 2018-05-14 Wilco Dijkstra <wdijkstr@arm.com>
20205 * config/aarch64/aarch64.md (mov<mode>): Remove '*' in alternatives.
20206 (movsi_aarch64): Likewise.
20207 (load_pairsi): Likewise.
20208 (load_pairdi): Likewise.
20209 (store_pairsi): Likewise.
20210 (store_pairdi): Likewise.
20211 (load_pairsf): Likewise.
20212 (load_pairdf): Likewise.
20213 (store_pairsf): Likewise.
20214 (store_pairdf): Likewise.
20215 (zero_extend): Likewise.
20216 (trunc): Swap alternatives.
20217 (fcvt_target): Add '?' to prefer w over r.
20219 2018-05-14 Jakub Jelinek <jakub@redhat.com>
20222 * config/i386/i386.md: Disallow non-commutative arithmetics in
20223 last twpeephole for mem {+,-,&,|,^}= x; mem != 0 after cmpelim
20224 optimization. Use COMMUTATIVE_ARITH_P test rather than != MINUS
20225 in the peephole2 before it.
20227 2018-05-14 Sebastian Peryt <sebastian.peryt@intel.com>
20229 * common/config/i386/i386-common.c (OPTION_MASK_ISA_CLDEMOTE_SET,
20230 OPTION_MASK_ISA_CLDEMOTE_UNSET): New defines.
20231 (ix86_handle_option): Handle -mcldemote.
20232 * config.gcc: New header.
20233 * config/i386/cldemoteintrin.h: New file.
20234 * config/i386/cpuid.h (bit_CLDEMOTE): New bit.
20235 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
20237 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
20238 OPTION_MASK_ISA_CLDEMOTE.
20239 * config/i386/i386.c (ix86_target_string): Add -mcldemote.
20240 (ix86_valid_target_attribute_inner_p): Ditto.
20241 (enum ix86_builtins): Add IX86_BUILTIN_CLDEMOTE.
20242 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_cldemote.
20243 (ix86_expand_builtin): Expand IX86_BUILTIN_CLDEMOTE.
20244 * config/i386/i386.h (TARGET_CLDEMOTE, TARGET_CLDEMOTE_P): New.
20245 * config/i386/i386.md (UNSPECV_CLDEMOTE): New.
20247 * config/i386/i386.opt: Add -mcldemote.
20248 * config/i386/x86intrin.h: New header.
20249 * doc/invoke.texi: Add -mcldemote.
20251 2018-05-14 Richard Biener <rguenther@suse.de>
20253 * doc/match-and-simplify.texi: Adjust :s documentation.
20255 2018-05-14 Alexander Monakov <amonakov@ispras.ru>
20257 * sort.cc (REORDER_23): Pass the type for the temporaries instead of
20258 intended memcpy size.
20259 (REORDER_45): Likewise.
20261 2018-05-13 Alexander Monakov <amonakov@ispras.ru>
20263 * sort.cc: New file.
20264 * system.h [!CHECKING_P] (qsort): Redirect to gcc_qsort.
20265 * vec.c (qsort_chk): Use gcc_qsort.
20266 * Makefile.in (OBJS-libcommon): Add sort.o.
20267 (build/sort.o): New target. Use it...
20268 (BUILD_RTL): ... here, and...
20269 (build/gencfn-macros): ... here, and...
20270 (build/genmatch): ... here.
20272 2018-05-13 Kito Cheng <kito.cheng@gmail.com>
20273 Chung-Ju Wu <jasonwucj@gmail.com>
20275 * config.gcc (nds32*-*-*): Check that n15 is valid to --with-cpu.
20276 * config/nds32/nds32-graywolf.md: New file.
20277 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_GRAYWOLF.
20278 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n15
20280 * config/nds32/nds32-protos.h: More declarations for n15 pipeline.
20281 * config/nds32/nds32-utils.c: More implementations for n15 pipeline.
20282 * config/nds32/nds32.md (pipeline_model): Add graywolf.
20283 * config/nds32/nds32.opt (mcpu): Support n15 pipeline cpus.
20284 * config/nds32/pipelines.md: Include n15 settings.
20286 2018-05-13 Kito Cheng <kito.cheng@gmail.com>
20287 Chung-Ju Wu <jasonwucj@gmail.com>
20289 * config.gcc (nds32*-*-*): Check that n12/n13 are valid to --with-cpu.
20290 * config/nds32/nds32-n13.md: New file.
20291 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N12 and CPU_N13.
20292 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n12/n13
20294 * config/nds32/nds32-protos.h: More declarations for n12/n13 pipeline.
20295 * config/nds32/nds32.md (pipeline_model): Add n13.
20296 * config/nds32/nds32.opt (mcpu): Support n12/n13 pipeline cpus.
20297 * config/nds32/pipelines.md: Include n13 settings.
20299 2018-05-13 Kito Cheng <kito.cheng@gmail.com>
20300 Chung-Ju Wu <jasonwucj@gmail.com>
20302 * config.gcc (nds32*-*-*): Check that n10/d10 are valid to --with-cpu.
20303 * config/nds32/nds32-n10.md: New file.
20304 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N10.
20305 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n10
20307 * config/nds32/nds32-protos.h: More declarations for n10 pipeline.
20308 * config/nds32/nds32-utils.c: More implementations for n10 pipeline.
20309 * config/nds32/nds32.md (pipeline_model): Add n10.
20310 * config/nds32/nds32.opt (mcpu): Support n10 pipeline cpus.
20311 * config/nds32/pipelines.md: Include n10 settings.
20313 2018-05-13 Monk Chiang <sh.chiang04@gmail.com>
20314 Kito Cheng <kito.cheng@gmail.com>
20315 Chung-Ju Wu <jasonwucj@gmail.com>
20317 * config.gcc (nds32be-*-*): Handle --with-ext-dsp.
20318 * config/nds32/constants.md (unspec_element, unspec_volatile_element):
20319 Add enum values for DSP extension instructions.
20320 * config/nds32/constraints.md (Iu06, IU06, CVp5, CVs5, CVs2, CVhi):
20322 * config/nds32/iterators.md (shifts, shiftrt, sat_plus, all_plus,
20323 sat_minus, all_minus, plus_minus, extend, sumax, sumin, sumin_max):
20324 New code iterators.
20325 (su, zs, uk, opcode, add_rsub, add_sub): New code attributes.
20326 * config/nds32/nds32-dspext.md: New file for DSP implementation.
20327 * config/nds32/nds32-intrinsic.c: Implementation of DSP extension.
20328 * config/nds32/nds32-intrinsic.md: Likewise.
20329 * config/nds32/nds32_intrinsic.h: Likewise.
20330 * config/nds32/nds32-md-auxiliary.c: Likewise.
20331 * config/nds32/nds32-memory-manipulation.c: Consider DSP extension.
20332 * config/nds32/nds32-predicates.c (const_vector_to_hwint): New.
20333 (nds32_valid_CVp5_p, nds32_valid_CVs5_p): New.
20334 (nds32_valid_CVs2_p, nds32_valid_CVhi_p): New.
20335 * config/nds32/nds32-protos.h: New declarations for DSP extension.
20336 * config/nds32/nds32-utils.c (extract_mac_non_acc_rtx): New case
20337 TYPE_DMAC in switch statement.
20338 * config/nds32/nds32.c: New checking and implementation for DSP
20339 extension instructions.
20340 * config/nds32/nds32.h: Likewise.
20341 * config/nds32/nds32.md: Likewise.
20342 * config/nds32/nds32.opt (mhw-abs, mext-dsp): New options.
20343 * config/nds32/predicates.md: Implement new predicates for DSP
20346 2018-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
20348 * config/rs6000/rs6000.md (mov<mode>_softfloat, FMOVE32):
20349 Reformat alternatives and attributes so it is easier to identify
20350 which constraints/attributes go with which instruction.
20351 (mov<mode>_hardfloat32, FMOVE64): Likewise.
20352 (mov<mode>_softfloat32, FMOVE64): Likewise.
20353 (mov<mode>_hardfloat64, FMOVE64): Likewise.
20354 (mov<mode>_softfloat64, FMOVE64): Likewise.
20356 2018-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
20358 * doc/extend.texi (PowerPC Built-in Functions): Rename this
20360 (Basic PowerPC Built-in Functions): The new name of the
20361 subsection previously known as "PowerPC Built-in Functions".
20362 (Basic PowerPC Built-in Functions Available on all Configurations):
20364 (Basic PowerPC Built-in Functions Available on ISA 2.05): Likewise.
20365 (Basic PowerPC Built-in Functions Available on ISA 2.06): Likewise.
20366 (Basic PowerPC Built-in Functions Available on ISA 2.07): Likewise.
20367 (Basic PowerPC Built-in Functions Available on ISA 3.0): Likewise.
20369 2018-05-11 Martin Jambor <mjambor@suse.cz>
20372 * ipa-cp.c (intersect_with_plats): Check that the lattice contains
20375 2018-05-11 Richard Earnshaw <rearnsha@arm.com>
20378 * config/arm/arm-cpus.in (fgroup ARMv6m): Add be8 feature.
20380 2018-05-11 Sebastian Peryt <sebastian.peryt@intel.com>
20382 * common/config/i386/i386-common.c (OPTION_MASK_ISA_WAITPKG_SET,
20383 OPTION_MASK_ISA_WAITPKG_UNSET): New defines.
20384 (ix86_handle_option): Handle -mwaitpkg.
20385 * config.gcc: New header.
20386 * config/i386/cpuid.h (bit_WAITPKG): New bit.
20387 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mwaitpkg.
20388 * config/i386/i386-builtin-types.def ((UINT8, UNSIGNED, UINT64)): New
20390 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
20391 OPTION_MASK_ISA_WAITPKG.
20392 * config/i386/i386.c (ix86_target_string): Add -mwaitpkg.
20393 (ix86_option_override_internal): Add PTA_WAITPKG.
20394 (ix86_valid_target_attribute_inner_p): Add -mwaitpkg.
20395 (enum ix86_builtins): Add IX86_BUILTIN_UMONITOR, IX86_BUILTIN_UMWAIT,
20396 IX86_BUILTIN_TPAUSE.
20397 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_umonitor,
20398 __builtin_ia32_umwait and __builtin_ia32_tpause.
20399 (ix86_expand_builtin): Expand IX86_BUILTIN_UMONITOR,
20400 IX86_BUILTIN_UMWAIT, IX86_BUILTIN_TPAUSE.
20401 * config/i386/i386.h (TARGET_WAITPKG, TARGET_WAITPKG_P): New.
20402 * config/i386/i386.md (UNSPECV_UMWAIT, UNSPECV_UMONITOR,
20403 UNSPECV_TPAUSE): New.
20404 (umwait, umwait_rex64, umonitor_<mode>, tpause, tpause_rex64): New.
20405 * config/i386/i386.opt: Add -mwaitpkg.
20406 * config/i386/waitpkgintrin.h: New file.
20407 * config/i386/x86intrin.h: New header.
20408 * doc/invoke.texi: Add -mwaitpkg.
20410 2018-05-11 Richard Earnshaw <rearnsha@arm.com>
20413 * config/arm/arm-cpus.in: Add comment that ARMv6-m and ARMv6S-m are now
20415 (cortex-m0): Use armv6s-m isa.
20416 (cortex-m0plus): Likewise.
20417 (cortex-m1): Likewise.
20418 (cortex-m0.small-multiply): Likewise.
20419 (cortex-m0plus.small-multiply): Likewise.
20420 (cortex-m1.small-multiply): Likewise.
20422 2018-05-11 Allan Sandfeld Jensen <allan.jensen@qt.io>
20423 Jakub Jelinek <jakub@redhat.com>
20425 PR tree-optimization/85692
20426 * tree-ssa-forwprop.c (simplify_vector_constructor): Try two
20427 source permute as well.
20429 2018-05-11 Martin Liska <mliska@suse.cz>
20432 * doc/extend.texi: Document LLVM style format for no_sanitize
20435 2018-05-10 Michael Meissner <meissner@linux.vnet.ibm.com>
20437 * config/rs6000/rs6000.c (mode_supports_dq_form): Rename
20438 mode_supports_vsx_dform_quad to mode_supports_dq_form.
20439 (mode_supports_vsx_dform_quad): Likewise.
20440 (mode_supports_vmx_dform): Move these functions to be next to the
20441 other mode_supports functions.
20442 (mode_supports_dq_form): Likewise.
20443 (quad_address_p): Change calls of mode_supports_vsx_dform_quad to
20444 mode_supports_dq_form.
20445 (reg_offset_addressing_ok_p): Likewise.
20446 (offsettable_ok_by_alignment): Likewise.
20447 (rs6000_legitimate_offset_address_p): Likewise.
20448 (legitimate_lo_sum_address_p): Likewise.
20449 (rs6000_legitimize_address): Likewise.
20450 (rs6000_legitimize_reload_address): Likewise.
20451 (rs6000_secondary_reload_inner): Likewise.
20452 (rs6000_preferred_reload_class): Likewise.
20453 (rs6000_output_move_128bit): Likewise.
20455 2018-05-10 Uros Bizjak <ubizjak@gmail.com>
20457 * config/i386/i386.c (ix86_expand_builtin) <case IX86_BUILTIN_RDPID>:
20458 Generate SImode target register for null target.
20459 <case IX86_BUILTIN_XGETBV>: Ditto.
20460 <case IX86_BUILTIN_XSETBV>: Optimize LSHIFTRT generation.
20461 * config/i386/xsaveintrin.h (_xgetbv): Add missing return.
20463 2018-05-10 Carl Love <cel@us.ibm.com>
20465 * config/rs6000/rs6000.md (prefetch): Generate ISA 2.06 instructions
20466 dcbtt and dcbtstt if operands[2] is 0.
20468 2018-05-10 Uros Bizjak <ubizjak@gmail.com>
20471 * config/i386/sse.md (usadv64qi): New expander.
20473 2018-05-10 Segher Boessenkool <segher@kernel.crashing.org>
20475 * config/rs6000/altivec.md (altivec_vmrghb, altivec_vmrghh,
20476 altivec_vmrghw, altivec_vmrglb, altivec_vmrglh, altivec_vmrglw): Remove
20477 -maltivec=be support.
20478 (vec_widen_umult_even_v16qi, vec_widen_smult_even_v16qi,
20479 vec_widen_umult_even_v8hi, vec_widen_smult_even_v8hi,
20480 vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
20481 vec_widen_umult_odd_v16qi, vec_widen_smult_odd_v16qi,
20482 vec_widen_umult_odd_v8hi, vec_widen_smult_odd_v8hi,
20483 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si, altivec_vpkpx,
20484 altivec_vpks<VI_char>ss, altivec_vpks<VI_char>us,
20485 altivec_vpku<VI_char>us, altivec_vpku<VI_char>um, altivec_vsum2sws,
20486 altivec_vsumsws): Adjust.
20487 (altivec_vspltb *altivec_vspltb_internal, altivec_vsplth,
20488 *altivec_vsplth_internal, altivec_vspltw, *altivec_vspltw_internal,
20489 altivec_vspltsf, *altivec_vspltsf_internal): Remove -maltivec=be
20491 (altivec_vperm_<mode>, altivec_vperm_<mode>_uns,
20492 altivec_vupkhs<VU_char>, altivec_vupkls<VU_char>, altivec_vupkhpx,
20493 altivec_vupklpx, altivec_lvsl, altivec_lvsr): Adjust.
20494 (altivec_lve<VI_char>x): Delete expand.
20495 (*altivec_lve<VI_char>x_internal): Rename to...
20496 (altivec_lve<VI_char>x): ... this.
20497 (altivec_lvxl_<mode>): Delete expand.
20498 (*altivec_lvxl_<mode>_internal): Rename to ...
20499 (altivec_lvxl_<mode>): ... this.
20500 (altivec_stvxl_<mode>): Delete expand.
20501 (*altivec_stvxl_<mode>_internal): Rename to ...
20502 (altivec_stvxl_<mode>): ... this.
20503 (altivec_stve<VI_char>x): Delete expand.
20504 (*altivec_stve<VI_char>x_internal): Rename to ...
20505 (altivec_stve<VI_char>x): ... this.
20506 (doublee<mode>2, unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2,
20507 doubleh<mode>2, unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2,
20508 reduc_plus_scal_<mode>): Adjust.
20509 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Adjust
20511 (rs6000_cpu_cpp_builtins): Adjust.
20512 (altivec_resolve_overloaded_builtin): Remove -maltivec=be support.
20513 * config/rs6000/rs6000-protos.h (altivec_expand_lvx_be,
20514 altivec_expand_stvx_be, altivec_expand_stvex_be): Delete.
20515 * config/rs6000/rs6000.c (rs6000_option_override_internal): Remove
20516 -maltivec=be support.
20517 (rs6000_split_vec_extract_var): Adjust.
20518 (rs6000_split_v4si_init): Adjust.
20519 (swap_selector_for_mode): Delete.
20520 (altivec_expand_lvx_be, altivec_expand_stvx_be,
20521 altivec_expand_stvex_be): Delete.
20522 (altivec_expand_lv_builtin, altivec_expand_stv_builtin): Remove
20523 -maltivec=be support.
20524 (rs6000_gimple_fold_builtin): Ditto.
20525 (rs6000_generate_float2_double_code, rs6000_generate_float2_code):
20527 * config/rs6000/rs6000.h (VECTOR_ELT_ORDER_BIG): Delete.
20528 (TARGET_DIRECT_MOVE_64BIT): Adjust.
20529 * config/rs6000/rs6000.md (split for extendsidi2 for vectors): Adjust.
20530 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Delete.
20531 * config/rs6000/vsx.md (floate<mode>, unsfloatev2di, floato<mode>,
20532 unsfloatov2di, vsignedo_v2df, vsignede_v2df, vunsignedo_v2df,
20533 vunsignede_v2df, vsx_extract_<mode>_p9, *vsx_extract_si,
20534 *vsx_extract_<mode>_p8, *vsx_extract_si_<uns>float_df,
20535 *vsx_extract_si_<uns>float_<mode>, vsx_set_<mode>_p9, vsx_set_v4sf_p9,
20536 *vsx_insert_extract_v4sf_p9, *vsx_insert_extract_v4sf_p9_2, and an
20537 anonymous split): Adjust.
20538 (vsx_mergel_<mode>, vsx_mergeh_<mode>): Remove -maltivec=be support.
20539 (vsx_xxspltd_<mode>, extract4b, insert4b): Adjust.
20541 2018-05-10 Eric Botcazou <ebotcazou@adacore.com>
20543 * configure.ac (gcc_gxx_include_dir_add_sysroot): Set it to 1 only
20544 when --with-gxx-include-dir is also specified.
20545 * configure: Regenerate.
20547 2018-05-09 Jim Wilson <jimw@sifive.com>
20550 * config.gcc (riscv*-*-*): Handle --with-multilib-list.
20551 * config/riscv/t-withmultilib: New.
20552 * config/riscv/withmultilib.h: New.
20553 * doc/install.texi: Document RISC-V --with-multilib-list support.
20555 2018-05-09 Richard Biener <rguenther@suse.de>
20557 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Fill a cost
20559 (vect_bb_vectorization_profitable_p): Adjust. Compute
20560 actual scalar cost using the cost vector and the add_stmt_cost
20563 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
20565 PR rtl-optimization/85645
20566 * config/rs6000/rs6000.c (rs6000_emit_prologue_components): Put a SET
20567 in the REG_CFA_REGISTER note for LR, don't leave it empty.
20569 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
20571 PR rtl-optimization/85645
20572 * shrink-wrap.c (spread_components): Return a boolean saying if
20573 anything was changed.
20574 (try_shrink_wrapping_separate): Iterate spread_components until
20575 nothing changes anymore.
20577 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
20579 PR rtl-optimization/85645
20580 * regrename.c (build_def_use): Also kill the chains that include the
20581 destination of a REG_CFA_REGISTER note.
20583 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
20585 PR rtl-optimization/85645
20586 * regcprop.c (copyprop_hardreg_forward_1): Don't propagate into an
20587 insn that has a REG_CFA_REGISTER note.
20589 2018-05-09 Richard Sandiford <richard.sandiford@linaro.org>
20591 * cfgexpand.c (expand_clobber): New function.
20592 (expand_gimple_stmt_1): Use it.
20593 * tree-vect-stmts.c (vect_clobber_variable): New function,
20595 (vectorizable_simd_clone_call): ...here.
20596 (vectorizable_store): Emit a clobber either side of an
20597 IFN_STORE_LANES sequence.
20598 (vectorizable_load): Emit a clobber after an IFN_LOAD_LANES sequence.
20600 2018-05-09 Tom de Vries <tom@codesourcery.com>
20603 * config/nvptx/nvptx.md (define_insn "trap", define_insn "trap_if_true")
20604 (define_insn "trap_if_false"): Add exit after trap.
20606 2018-05-09 Eric Botcazou <ebotcazou@adacore.com>
20608 PR rtl-optimization/85638
20609 * bb-reorder.c: Include common/common-target.h.
20610 (create_forwarder_block): New function extracted from...
20611 (fix_up_crossing_landing_pad): ...here. Rename into...
20612 (dw2_fix_up_crossing_landing_pad): ...this.
20613 (sjlj_fix_up_crossing_landing_pad): New function.
20614 (find_rarely_executed_basic_blocks_and_crossing_edges): In SJLJ mode,
20615 call sjlj_fix_up_crossing_landing_pad if there are incoming EH edges
20616 from both partitions and exit the loop after one iteration.
20618 2018-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
20621 * doc/extend.texi (PowerPC Built-in Functions): Rename this
20623 (Basic PowerPC Built-in Functions): The new name of the
20624 subsection previously known as "PowerPC Built-in Functions".
20625 (Basic PowerPC Built-in Functions Available on all Configurations):
20627 (Basic PowerPC Built-in Functions Available on ISA 2.05): New
20629 (Basic PowerPC Built-in Functions Available on ISA 2.06): New
20631 (Basic PowerPC Built-in Functions Available on ISA 2.07): New
20633 (Basic PowerPC Built-in Functions Available on ISA 3.0): New
20636 2018-05-08 Jim Wilson <jimw@sifive.com>
20638 * config/riscv/linux.h (MUSL_ABI_SUFFIX): Delete unnecessary backslash.
20639 (LD_EMUL_SUFFIX): New.
20640 (LINK_SPEC): Use it.
20642 2018-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
20644 * doc/extend.texi (PowerPC Built-in Functions): Rename this
20646 (Basic PowerPC Built-in Functions): The new name of the
20647 subsection previously known as "PowerPC Built-in Functions".
20648 (Basic PowerPC Built-in Functions Available on all Configurations):
20650 (Basic PowerPC Built-in Functions Available on ISA 2.05): New
20652 (Basic PowerPC Built-in Functions Available on ISA 2.06): New
20654 (Basic PowerPC Built-in Functions Available on ISA 2.07): New
20656 (Basic PowerPC Built-in Functions Available on ISA 3.0): New
20659 2018-05-08 Jakub Jelinek <jakub@redhat.com>
20662 * config/i386/i386.md: Add peepholes for mem {+,-,&,|,^}= x; mem != 0
20663 after cmpelim optimization.
20665 2018-05-08 Olga Makhotina <olga.makhotina@intel.com>
20667 * config.gcc: Support "goldmont".
20668 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "goldmont".
20669 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
20670 PROCESSOR_GOLDMONT.
20671 * config/i386/i386.c (m_GOLDMONT): Define.
20672 (processor_target_table): Add "goldmont".
20673 (PTA_GOLDMONT): Define.
20674 (ix86_lea_outperforms): Add TARGET_GOLDMONT.
20675 (get_builtin_code_for_version): Handle PROCESSOR_GOLDMONT.
20676 (fold_builtin_cpu): Add M_INTEL_GOLDMONT.
20677 (fold_builtin_cpu): Add "goldmont".
20678 (ix86_add_stmt_cost): Add TARGET_GOLDMONT.
20679 (ix86_option_override_internal): Add "goldmont".
20680 * config/i386/i386.h (processor_costs): Define TARGET_GOLDMONT.
20681 (processor_type): Add PROCESSOR_GOLDMONT.
20682 * config/i386/i386.md: Add CPU "glm".
20683 * config/i386/glm.md: New file.
20684 * config/i386/x86-tune.def: Add m_GOLDMONT.
20685 * doc/invoke.texi: Add goldmont as x86 -march=/-mtune= CPU type.
20687 2018-05-08 Jakub Jelinek <jakub@redhat.com>
20690 * config/i386/i386.c (ix86_expand_sse2_abs): Handle E_V2DImode and
20692 * config/i386/sse.md (abs<mode>2): Use VI_AVX2 iterator instead of
20693 VI1248_AVX512VL_AVX512BW. Handle V2DImode and V4DImode if not
20694 TARGET_AVX512VL using ix86_expand_sse2_abs. Formatting fixes.
20697 * config/i386/i386.c (ix86_fold_builtin): Handle
20698 IX86_BUILTIN_{,P}MOVMSK{PS,PD,B}{,128,256}.
20701 * config/i386/sse.md (ssequaterinsnmode): New mode attribute.
20702 (*<extract_type>_vinsert<shuffletype><extract_suf>_0): New pattern.
20704 2018-05-08 Richard Earnshaw <rearnsha@arm.com>
20707 * config/arm/parsecpu.awk (check_cpu): Fix operator precedence.
20708 (check_arch): Likewise.
20709 (check_fpu): Return the result rather than printing it.
20710 (end arch): Fix operator precedence.
20711 (end cpu): Likewise.
20712 (END): Print the result from check_fpu.
20714 2018-05-08 Richard Sandiford <richard.sandiford@linaro.org>
20715 Alan Hayward <alan.hayward@arm.com>
20716 David Sherwood <david.sherwood@arm.com>
20718 * config/aarch64/aarch64-sve.md (*pred_cmp<cmp_op><mode>_combine)
20719 (*pred_cmp<cmp_op><mode>, *fcm<cmp_op><mode>_and_combine)
20720 (*fcmuo<mode>_and_combine, *fcm<cmp_op><mode>_and)
20721 (*fcmuo<mode>_and): New patterns.
20723 2018-05-08 Richard Sandiford <richard.sandiford@linaro.org>
20725 * config/aarch64/iterators.md (UNSPEC_COND_LO, UNSPEC_COND_LS)
20726 (UNSPEC_COND_HI, UNSPEC_COND_HS, UNSPEC_COND_UO): Delete.
20727 (SVE_INT_CMP, SVE_FP_CMP): New code iterators.
20728 (cmp_op, sve_imm_con): New code attributes.
20729 (SVE_COND_INT_CMP, imm_con): Delete.
20730 (cmp_op): Remove above unspecs from int attribute.
20731 * config/aarch64/aarch64-sve.md (*vec_cmp<cmp_op>_<mode>): Rename
20733 (*cmp<cmp_op><mode>): ...this. Use UNSPEC_MERGE_PTRUE instead of
20734 comparison-specific unspecs.
20735 (*vec_cmp<cmp_op>_<mode>_ptest): Rename to...
20736 (*cmp<cmp_op><mode>_ptest): ...this and adjust likewise.
20737 (*vec_cmp<cmp_op>_<mode>_cc): Rename to...
20738 (*cmp<cmp_op><mode>_cc): ...this and adjust likewise.
20739 (*vec_fcm<cmp_op><mode>): Rename to...
20740 (*fcm<cmp_op><mode>): ...this and adjust likewise.
20741 (*vec_fcmuo<mode>): Rename to...
20742 (*fcmuo<mode>): ...this and adjust likewise.
20743 (*pred_fcm<cmp_op><mode>): New pattern.
20744 * config/aarch64/aarch64.c (aarch64_emit_unop, aarch64_emit_binop)
20745 (aarch64_emit_sve_ptrue_op, aarch64_emit_sve_ptrue_op_cc): New
20747 (aarch64_unspec_cond_code): Remove handling of LTU, GTU, LEU, GEU
20749 (aarch64_gen_unspec_cond, aarch64_emit_unspec_cond): Delete.
20750 (aarch64_emit_sve_predicated_cond): New function.
20751 (aarch64_expand_sve_vec_cmp_int): Use aarch64_emit_sve_ptrue_op_cc.
20752 (aarch64_emit_unspec_cond_or): Replace with...
20753 (aarch64_emit_sve_or_conds): ...this new function. Use
20754 aarch64_emit_sve_ptrue_op for the individual comparisons and
20755 aarch64_emit_binop to OR them together.
20756 (aarch64_emit_inverted_unspec_cond): Replace with...
20757 (aarch64_emit_sve_inverted_cond): ...this new function. Use
20758 aarch64_emit_sve_ptrue_op for the comparison and
20759 aarch64_emit_unop to invert the result.
20760 (aarch64_expand_sve_vec_cmp_float): Update after the above
20761 changes. Use aarch64_emit_sve_ptrue_op for native comparisons.
20763 2018-05-07 Nathan Sidwell <nathan@acm.org>
20765 * doc/invoke.texi (C++ Dialect Options): Remove -ffor-scope.
20766 * doc/extend.texi (Deprecated Features): Remove -fno-for-scope
20767 (Backwards Compatibility): Likewise.
20769 2018-05-07 Luis Machado <luis.machado@linaro.org>
20773 2018-05-07 Luis Machado <luis.machado@linaro.org>
20775 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
20776 <prefetch_dynamic_strides>: New const bool field.
20777 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
20778 prefetch_dynamic_strides.
20779 (exynosm1_prefetch_tune): Likewise.
20780 (thunderxt88_prefetch_tune): Likewise.
20781 (thunderx_prefetch_tune): Likewise.
20782 (thunderx2t99_prefetch_tune): Likewise.
20783 (qdf24xx_prefetch_tune): Likewise. Set prefetch_dynamic_strides
20785 (aarch64_override_options_internal): Update to set
20786 PARAM_PREFETCH_DYNAMIC_STRIDES.
20787 * doc/invoke.texi (prefetch-dynamic-strides): Document new option.
20788 * params.def (PARAM_PREFETCH_DYNAMIC_STRIDES): New.
20789 * params.h (PARAM_PREFETCH_DYNAMIC_STRIDES): Define.
20790 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Account for
20791 prefetch-dynamic-strides setting.
20793 2018-05-07 Luis Machado <luis.machado@linaro.org>
20795 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
20796 <minimum_stride>: New const int field.
20797 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
20798 minimum_stride field.
20799 (exynosm1_prefetch_tune): Likewise.
20800 (thunderxt88_prefetch_tune): Likewise.
20801 (thunderx_prefetch_tune): Likewise.
20802 (thunderx2t99_prefetch_tune): Likewise.
20803 (qdf24xx_prefetch_tune): Likewise. Set minimum_stride to 2048.
20804 (aarch64_override_options_internal): Update to set
20805 PARAM_PREFETCH_MINIMUM_STRIDE.
20806 * doc/invoke.texi (prefetch-minimum-stride): Document new option.
20807 * params.def (PARAM_PREFETCH_MINIMUM_STRIDE): New.
20808 * params.h (PARAM_PREFETCH_MINIMUM_STRIDE): Define.
20809 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Return false if
20810 stride is constant and is below the minimum stride threshold.
20812 2018-05-07 Luis Machado <luis.machado@linaro.org>
20814 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune) <l2_cache_size>: Set
20817 2018-05-07 Luis Machado <luis.machado@linaro.org>
20819 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
20820 <prefetch_dynamic_strides>: New const bool field.
20821 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
20822 prefetch_dynamic_strides.
20823 (exynosm1_prefetch_tune): Likewise.
20824 (thunderxt88_prefetch_tune): Likewise.
20825 (thunderx_prefetch_tune): Likewise.
20826 (thunderx2t99_prefetch_tune): Likewise.
20827 (qdf24xx_prefetch_tune): Likewise. Set prefetch_dynamic_strides
20829 (aarch64_override_options_internal): Update to set
20830 PARAM_PREFETCH_DYNAMIC_STRIDES.
20831 * doc/invoke.texi (prefetch-dynamic-strides): Document new option.
20832 * params.def (PARAM_PREFETCH_DYNAMIC_STRIDES): New.
20833 * params.h (PARAM_PREFETCH_DYNAMIC_STRIDES): Define.
20834 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Account for
20835 prefetch-dynamic-strides setting.
20837 2018-05-07 Luis Machado <luis.machado@linaro.org>
20839 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
20840 <minimum_stride>: New const int field.
20841 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
20842 minimum_stride field.
20843 (exynosm1_prefetch_tune): Likewise.
20844 (thunderxt88_prefetch_tune): Likewise.
20845 (thunderx_prefetch_tune): Likewise.
20846 (thunderx2t99_prefetch_tune): Likewise.
20847 (qdf24xx_prefetch_tune): Likewise. Set minimum_stride to 2048.
20848 (aarch64_override_options_internal): Update to set
20849 PARAM_PREFETCH_MINIMUM_STRIDE.
20850 * doc/invoke.texi (prefetch-minimum-stride): Document new option.
20851 * params.def (PARAM_PREFETCH_MINIMUM_STRIDE): New.
20852 * params.h (PARAM_PREFETCH_MINIMUM_STRIDE): Define.
20853 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Return false if
20854 stride is constant and is below the minimum stride threshold.
20856 2018-05-06 Jakub Jelinek <jakub@redhat.com>
20859 * cfgexpand.c (expand_asm_stmt): Don't create a temporary if
20860 the type is addressable. Don't force op into register if it has
20863 2018-05-05 Roland McGrath <mcgrathr@google.com>
20866 * varasm.c (default_section_type_flags): Set SECTION_NOTYPE for
20867 any section for which we don't know a specific type it should have,
20868 regardless of name. Previously this was done only for the exact
20869 names ".init_array", ".fini_array", and ".preinit_array".
20870 (default_elf_asm_named_section): Add comment about
20871 relationship with default_section_type_flags and SECTION_NOTYPE.
20872 (get_section): Don't consider it a type conflict if one side has
20873 SECTION_NOTYPE and the other doesn't, as long as neither has the
20874 SECTION_BSS et al used in the default_section_type_flags logic.
20876 2018-05-05 Tom de Vries <tom@codesourcery.com>
20879 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG_3): Define.
20880 (workaround_barsyncs): New function.
20881 (nvptx_reorg): Use workaround_barsyncs.
20882 * config/nvptx/nvptx.md (define_c_enum "unspecv"): Add UNSPECV_MEMBAR.
20883 (define_expand "nvptx_membar_cta"): New define_expand.
20884 (define_insn "*nvptx_membar_cta"): New insn.
20886 2018-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
20888 * brig-builtins.def: Add consts to ptrs etc. in BRIG builtin defs.
20889 To improve optimization opportunities.
20890 * builtin-types.def: The new needed builtin types for the above.
20892 2018-05-04 Richard Biener <rguenther@suse.de>
20894 * bb-reorder.c (sanitize_hot_paths): Release hot_bbs_to_check.
20895 * gimple-ssa-store-merging.c
20896 (imm_store_chain_info::output_merged_store): Remove redundant create,
20897 release split_store vector contents on failure.
20898 * tree-vect-slp.c (vect_schedule_slp_instance): Avoid leaking
20899 scalar stmt vector on cache hit.
20901 2018-05-04 Segher Boessenkool <segher@kernel.crashing.org>
20903 * common/config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
20905 * config.gcc (powerpc-xilinx-eabi*): Remove.
20906 * config/rs6000/predicates.md (easy_fp_constant): Remove Xilinx FP
20908 (fusion_addis_mem_combo_load): Ditto.
20909 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove Xilinx
20911 (rs6000_cpu_cpp_builtins): Ditto.
20912 * config/rs6000/rs6000-linux.c
20913 (rs6000_linux_float_exceptions_rounding_supported_p): Ditto.
20914 * config/rs6000/rs6000-opts.h (enum fpu_type_t): Delete.
20915 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove Xilinx FP
20917 (rs6000_setup_reg_addr_masks): Ditto.
20918 (rs6000_init_hard_regno_mode_ok): Ditto.
20919 (rs6000_option_override_internal): Ditto.
20920 (legitimate_lo_sum_address_p): Ditto.
20921 (rs6000_legitimize_address): Ditto.
20922 (rs6000_legitimize_reload_address): Ditto.
20923 (rs6000_legitimate_address_p): Ditto.
20924 (abi_v4_pass_in_fpr): Ditto.
20925 (setup_incoming_varargs): Ditto.
20926 (rs6000_gimplify_va_arg): Ditto.
20927 (rs6000_split_multireg_move): Ditto.
20928 (rs6000_savres_strategy): Ditto.
20929 (rs6000_emit_prologue_components): Ditto.
20930 (rs6000_emit_epilogue_components): Ditto.
20931 (rs6000_emit_prologue): Ditto.
20932 (rs6000_emit_epilogue): Ditto.
20933 (rs6000_elf_file_end): Ditto.
20934 (rs6000_function_value): Ditto.
20935 (rs6000_libcall_value): Ditto.
20936 * config/rs6000/rs6000.h: Ditto.
20937 (TARGET_MINMAX_SF, TARGET_MINMAX_DF): Delete, merge to ...
20938 (TARGET_MINMAX): ... this. New.
20939 (TARGET_SF_FPR, TARGET_DF_FPR, TARGET_SF_INSN, TARGET_DF_INSN): Delete.
20940 * config/rs6000/rs6000.md: Remove Xilinx FP support.
20941 (*movsi_internal1_single): Delete.
20942 * config/rs6000/rs6000.opt (msingle-float, mdouble-float, msimple-fpu,
20943 mfpu=, mxilinx-fpu): Delete.
20944 * config/rs6000/singlefp.h: Delete.
20945 * config/rs6000/sysv4.h: Remove Xilinx FP support.
20946 * config/rs6000/t-rs6000: Ditto.
20947 * config/rs6000/t-xilinx: Delete.
20948 * config/rs6000/titan.md: Adjust for fp_type removal.
20949 * config/rs6000/vsx.md: Remove Xilinx FP support.
20950 (VStype_simple): Delete.
20951 (VSfptype_simple, VSfptype_mul, VSfptype_div, VSfptype_sqrt): Delete.
20952 * config/rs6000/xfpu.h: Delete.
20953 * config/rs6000/xfpu.md: Delete.
20954 * config/rs6000/xilinx.h: Delete.
20955 * config/rs6000/xilinx.opt: Delete.
20956 * doc/invoke.texi (RS/6000 and PowerPC Options): Remove
20957 -msingle-float, -mdouble-float, -msimple-fpu, -mfpu=, and -mxilinx-fpu.
20959 2018-05-04 Tom de Vries <tom@codesourcery.com>
20962 * builtins.c (expand_builtin_goacc_parlevel_id_size): Handle null target
20965 2018-05-04 Richard Biener <rguenther@suse.de>
20967 PR middle-end/85627
20968 * tree-complex.c (update_complex_assignment): We are always in SSA form.
20969 (expand_complex_div_wide): Likewise.
20970 (expand_complex_operations_1): Likewise.
20971 (expand_complex_libcall): Preserve EH info of the original stmt.
20972 (tree_lower_complex): Handle removed blocks.
20973 * tree.c (build_common_builtin_nodes): Do not set ECF_NOTRHOW
20974 on complex multiplication and division libcall builtins.
20976 2018-05-04 Richard Biener <rguenther@suse.de>
20978 PR middle-end/85574
20979 * fold-const.c (negate_expr_p): Restrict negation of operand
20980 zero of a division to when we know that can happen without
20982 (fold_negate_expr_1): Likewise.
20984 2018-05-04 Jakub Jelinek <jakub@redhat.com>
20987 * real.h (real_nextafter): Declare.
20988 * real.c (real_nextafter): New function.
20989 * fold-const-call.c (fold_const_nextafter): New function.
20990 (fold_const_call_sss): Call it for CASE_CFN_NEXTAFTER and
20991 CASE_CFN_NEXTTOWARD.
20992 (fold_const_call_1): For CASE_CFN_NEXTTOWARD call fold_const_call_sss
20993 even when arg1_mode is different from arg0_mode.
20995 2018-05-03 Nathan Sidwell <nathan@acm.org>
20997 * doc/extend.texi (Deprecated Features): Remove
20998 -ffriend-injection.
20999 (Backwards Compatibility): Likewise.
21000 * doc/invoke.texi (C++ Language Options): Likewise.
21001 (C++ Dialect Options): Likewise.
21003 2018-05-03 Jakub Jelinek <jakub@redhat.com>
21006 * config/i386/avx512fintrin.h (_mm512_mullox_epi64,
21007 _mm512_mask_mullox_epi64): New intrinsics.
21009 2018-05-03 Tom de Vries <tom@codesourcery.com>
21012 * doc/sourcebuild.texi (Commands for use in dg-final, Scan optimization
21013 dump files): Add offload-tree.
21015 2018-05-03 Richard Biener <rguenther@suse.de>
21017 PR tree-optimization/85615
21018 * tree-ssa-threadupdate.c (thread_block_1): Only allow exits
21019 to loops not nested in BBs loop father to avoid creating multi-entry
21022 2018-05-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21024 PR tree-optimization/70291
21025 * tree-complex.c (expand_complex_libcall): Add type, inplace_p
21026 arguments. Change return type to tree. Emit libcall as a new
21027 statement rather than replacing existing one when inplace_p is true.
21028 (expand_complex_multiplication_components): New function.
21029 (expand_complex_multiplication): Expand floating-point complex
21030 multiplication using the above.
21031 (expand_complex_division): Rename inner_type parameter to type.
21032 Update expand_complex_libcall call-site.
21033 (expand_complex_operations_1): Update expand_complex_multiplication
21034 and expand_complex_division call-sites.
21036 2018-05-02 Jakub Jelinek <jakub@redhat.com>
21039 * config/i386/i386.md (*ashl<dwi>3_doubleword_mask,
21040 *ashl<dwi>3_doubleword_mask_1, *<shift_insn><dwi>3_doubleword_mask,
21041 *<shift_insn><dwi>3_doubleword_mask_1): In condition require that
21042 the highest significant bit of the shift count mask is clear. In
21043 check whether and[sq]i3 is needed verify that all significant bits
21044 of the shift count other than the highest are set.
21046 2018-05-02 Tom de Vries <tom@codesourcery.com>
21049 * builtins.def (DEF_GOACC_BUILTIN_ONLY): Define.
21050 * omp-builtins.def (BUILT_IN_GOACC_PARLEVEL_ID)
21051 (BUILT_IN_GOACC_PARLEVEL_SIZE): New builtin.
21052 * builtins.c (expand_builtin_goacc_parlevel_id_size): New function.
21053 (expand_builtin): Call expand_builtin_goacc_parlevel_id_size.
21054 * doc/extend.texi (Other Builtins): Add __builtin_goacc_parlevel_id and
21055 __builtin_goacc_parlevel_size.
21057 2018-05-02 Richard Biener <rguenther@suse.de>
21059 PR tree-optimization/85597
21060 * tree-vect-stmts.c (vectorizable_operation): For ternary SLP
21061 do not use split vect_get_vec_defs call but call vect_get_slp_defs
21064 2018-05-02 Tom de Vries <tom@codesourcery.com>
21067 * doc/sourcebuild.texi (Commands for use in dg-final, Scan optimization
21068 dump files): Add ltrans-tree.
21070 2018-05-02 Tom de Vries <tom@codesourcery.com>
21073 * doc/sourcebuild.texi (Commands for use in dg-final, Scan optimization
21074 dump files): Add wpa-ipa.
21076 2018-05-02 Segher Boessenkool <segher@kernel.crashing.org>
21078 * config.gcc (powerpc*-*-*): Remove paired.h. Unsupport the
21079 powerpc*-*-linux*paired* target.
21080 * config/rs6000/750cl.h: Delete.
21081 * config/rs6000/paired.h: Delete.
21082 * config/rs6000/paired.md: Delete.
21083 * config/rs6000/predicates.md (easy_vector_constant): Remove paired
21085 * config/rs6000/rs6000-builtin.def: Remove paired float support.
21086 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
21087 comment. Remove paired float support.
21088 * config/rs6000/rs6000-modes.def: Remove V2SF and V2SI.
21089 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete
21091 * config/rs6000/rs6000-protos.h (paired_expand_vector_init,
21092 paired_emit_vector_cond_expr, paired_expand_vector_move): Delete
21094 * config/rs6000/rs6000.c: Remove paired float support.
21095 (paired_expand_vector_init, paired_expand_vector_move,
21096 paired_emit_vector_compare, paired_emit_vector_cond_expr,
21097 (paired_expand_lv_builtin, paired_expand_stv_builtin,
21098 paired_expand_builtin, paired_expand_predicate_builtin,
21099 paired_init_builtins): Delete.
21100 * config/rs6000/rs6000.h: Remove paired float support.
21101 * config/rs6000/rs6000.md: Remove paired float support.
21102 (move_from_CR_ov_bit): Delete.
21103 * config/rs6000/rs6000.opt (mpaired): Delete.
21104 * config/rs6000/t-rs6000: Remove paired.md from MD_INCLUDES.
21105 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mpaired.
21107 2018-05-02 Richard Biener <rguenther@suse.de>
21109 PR middle-end/85567
21110 * gimplify.c (gimplify_save_expr): When in SSA form allow
21111 SAVE_EXPRs to compute to SSA vars.
21113 2018-05-02 Jakub Jelinek <jakub@redhat.com>
21116 * config/i386/i386.md (*ashl<dwi>3_doubleword_mask,
21117 *ashl<dwi>3_doubleword_mask_1, *<shift_insn><dwi>3_doubleword_mask,
21118 *<shift_insn><dwi>3_doubleword_mask_1): If and[sq]i3 is needed, don't
21119 clobber operands[2], instead use a new pseudo. Formatting fixes.
21121 2018-05-02 Richard Sandiford <richard.sandiford@linaro.org>
21123 PR tree-optimization/85586
21124 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Only
21125 exit early for statements in the same group if the accesses are
21128 2018-05-02 Tom de Vries <tom@codesourcery.com>
21131 * lto-wrapper.c (compile_offload_image): Add "could not find mkoffload"
21134 2018-05-01 Marc Glisse <marc.glisse@inria.fr>
21136 PR tree-optimization/85143
21137 * match.pd (A<B&A<C): Extend to BIT_IOR_EXPR.
21139 2018-05-01 Tom de Vries <tom@codesourcery.com>
21142 * config/nvptx/mkoffload.c (main): Suggest using -B in "offload compiler
21143 not found" error message.
21145 2018-05-01 Tom de Vries <tom@codesourcery.com>
21148 * vec.h (VEC_ORDERED_REMOVE_IF, VEC_ORDERED_REMOVE_IF_FROM_TO): Define.
21149 * vec.c (test_ordered_remove_if): New function.
21150 (vec_c_tests): Call test_ordered_remove_if.
21151 * dwarf2cfi.c (connect_traces): Use VEC_ORDERED_REMOVE_IF_FROM_TO.
21152 * lto-streamer-out.c (prune_offload_funcs): Use VEC_ORDERED_REMOVE_IF.
21153 * tree-vect-patterns.c (vect_pattern_recog_1): Use
21154 VEC_ORDERED_REMOVE_IF.
21156 2018-05-01 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
21158 PR tree-optimization/82665
21159 * vr-values.c (vr_values::extract_range_from_binary_expr): Handle
21160 pointer subtraction where arguments come from a memchr call.
21162 2018-05-01 Jakub Jelinek <jakub@redhat.com>
21164 * configure.ac (LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION): Use
21165 --push-state --as-needed and --pop-state instead of --as-needed and
21166 --no-as-needed if ld supports it.
21167 * configure: Regenerated.
21170 * doc/install.texi2html: Replace _002d with - and _002a with * in
21171 generated html files using sed.
21173 2018-04-30 David Malcolm <dmalcolm@redhat.com>
21176 * gcc-rich-location.c (blank_line_before_p): New function.
21177 (use_new_line): New function.
21178 (gcc_rich_location::add_fixit_insert_formatted): New function.
21179 * gcc-rich-location.h
21180 (gcc_rich_location::add_fixit_insert_formatted): New function.
21182 2018-04-30 David Malcolm <dmalcolm@redhat.com>
21184 * selftest.c (assert_streq): Rename "expected" and "actual" to
21185 "val1" and "val2". Extend NULL-handling to cover both inputs
21186 symmetrically, while still requiring both to be non-NULL for a pass.
21187 * selftest.h (assert_streq): Rename "expected" and "actual" to
21189 (ASSERT_EQ): Likewise.
21190 (ASSERT_EQ_AT): Likewise.
21191 (ASSERT_KNOWN_EQ): Likewise.
21192 (ASSERT_KNOWN_EQ_AT): Likewise.
21193 (ASSERT_NE): Likewise.
21194 (ASSERT_MAYBE_NE): Likewise.
21195 (ASSERT_MAYBE_NE_AT): Likewise.
21196 (ASSERT_STREQ): Likewise. Clarify that both must be non-NULL for
21197 the assertion to pass.
21198 (ASSERT_STREQ_AT): Likewise.
21200 2018-04-30 Jonathan Wakely <jwakely@redhat.com>
21202 * doc/invoke.texi (-fpie, -fPIE): Fix grammar and clarify
21203 interaction with -pie.
21205 2018-04-30 David Malcolm <dmalcolm@redhat.com>
21207 * selftest.h: Fix alphabetization of per-source-file selftest
21210 2018-04-30 Jason Merrill <jason@redhat.com>
21212 PR c++/61982 - dead stores to destroyed objects.
21213 * gimplify.c (gimplify_modify_expr): Simplify complex lvalue on LHS
21216 2018-04-30 Jason Merrill <jason@redhat.com>
21218 * tree.c (build_clobber): New.
21219 * tree.h: Declare it.
21220 * gimplify.c (gimplify_bind_expr, gimplify_target_expr): Use it.
21222 2018-04-30 David Malcolm <dmalcolm@redhat.com>
21224 * diagnostic-show-locus.c (layout::layout): Update for
21225 location_get_source_line returning a char_span.
21226 (struct char_span): Move to input.h.
21227 (struct correction): Update for fields in char_span becoming
21229 (struct source_line): Update for location_get_source_line
21230 returning a char_span.
21231 (layout::print_line): Likewise.
21232 * edit-context.c (edited_file::print_content): Likewise.
21233 (edited_file::print_diff_hunk): Likewise.
21234 (edited_file::print_run_of_changed_lines): Likewise.
21235 (edited_file::get_num_lines): Likewise.
21236 (edited_line::edited_line): Likewise.
21237 * final.c (asm_show_source): Likewise.
21238 * input.c (location_get_source_line): Convert return type
21239 from const char * to char_span, losing the final "line_len"
21241 (dump_location_info): Update for the above.
21242 (get_substring_ranges_for_loc): Likewise. Use a char_span
21243 when handling the literal within the line.
21244 (test_reading_source_line): Update for location_get_source_line
21245 returning a char_span.
21246 * input.h (class char_span): Move here from
21247 diagnostic-show-locus.c, converting from a struct to a class.
21248 Make data members private.
21249 (char_span::operator bool): New.
21250 (char_span::length): New.
21251 (char_span::get_buffer): New.
21252 (char_span::operator[]): New.
21253 (char_span::subspan): Make const.
21254 (char_span::xstrdup): New.
21255 (location_get_source_line): Convert return type from const char *
21256 to char_span, losing the final "line_size" param.
21258 2018-04-30 Jan Hubicka <jh@suse.cz>
21260 * lto-wrapper.c (ltrans_priorities): New static var.
21261 (cmp_priority): New.
21262 (run_gcc): Read priorities and if doing parallel build order
21263 the Makefile by them.
21265 2018-04-30 David Malcolm <dmalcolm@redhat.com>
21267 * input.h (builtins_location_check): Convert to a STATIC_ASSERT.
21269 2018-04-30 Richard Biener <rguenther@suse.de>
21271 * tree-cfg.c (verify_address): Remove base argument, add
21272 flag whether to check TREE_ADDRESSABLE and do that.
21273 (verify_expr): Remove.
21274 (verify_types_in_gimple_reference): Add pieces from verify_expr.
21275 (verify_gimple_assign_single): Likewise.
21276 (verify_gimple_switch): Likewise.
21277 (verify_expr_location_1): Dereference tp once. Add (disabled)
21278 piece from verify_expr.
21279 (verify_gimple_in_cfg): Do not call verify_expr on all ops.
21281 2018-04-30 Claudiu Zissulescu <claziss@synopsys.com>
21283 * config/arc/linux.h (CLEAR_INSN_CACHE): Define.
21285 2018-04-30 Claudiu Zissulescu <claziss@synopsys.com>
21287 * config/arc/arc-protos.h (prepare_extend_operands): Remove.
21288 (small_data_pattern): Likewise.
21289 (arc_rewrite_small_data): Likewise.
21290 * config/arc/arc.c (LEGITIMATE_SMALL_DATA_OFFSET_P): Remove.
21291 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Likewise.
21292 (get_symbol_alignment): New function.
21293 (legitimate_small_data_address_p): Likewise.
21294 (legitimate_scaled_address): Update, call
21295 legitimate_small_data_address_p.
21296 (output_sdata): New static variable.
21297 (arc_print_operand): Update how we handle small data operands.
21298 (arc_print_operand_address): Likewise.
21299 (arc_legitimate_address_p): Update, use
21300 legitimate_small_data_address_p.
21301 (arc_rewrite_small_data_p): Remove.
21302 (arc_rewrite_small_data_1): Likewise.
21303 (arc_rewrite_small_data): Likewise.
21304 (small_data_pattern): Likewise.
21305 (compact_sda_memory_operand): Update to use
21306 legitimate_small_data_address_p and get_symbol_alignment.
21307 (prepare_move_operands): Don't rewite sdata pattern.
21308 (prepare_extend_operands): Remove.
21309 * config/arc/arc.md (zero_extendqihi2): Don't rewrite sdata
21311 (zero_extendqisi2): Likewise.
21312 (zero_extendhisi2): Likewise.
21313 (extendqihi2): Likewise.
21314 (extendqisi2): Likewise.
21315 (extendhisi2): Likewise.
21316 (addsi3): Likewise.
21317 (subsi3): Likewise.
21318 (andsi3): Likewise.
21319 * config/arc/constraints.md (Usd): Change it to memory constraint.
21321 2018-04-30 Claudiu Zissulescu <claziss@synopsys.com>
21323 * config/arc/arc.c (arc_split_move): Allow signed 6-bit constants
21324 as source of std instructions.
21325 * config/arc/arc.md (movsi_insn): Update pattern predicate to
21326 allow 6-bit constants as source for store instructions.
21327 (movdi_insn): Update instruction pattern to allow 6-bit constants
21328 as source for store instructions.
21330 2018-04-30 Jonathan Wakely <jwakely@redhat.com>
21332 * doc/invoke.texi (-fdebug-types-section): Fix grammar.
21334 2018-04-30 Nathan Sidwell <nathan@acm.org>
21335 Sandra Loosemore <sandra@codesourcery.com>
21337 * dumpfile.c (dump_open): Allow '-' for stdout.
21338 * doc/invoke.texi (Developer Options): Document dump filename
21339 determination early. Document stdin/stdout selection.
21341 2018-04-30 Andrew Sadek <andrew.sadek.se@gmail.com>
21343 Microblaze Target: PIC data text relative
21345 * config/microblaze/microblaze.opt: add new option -mpic-data-text-rel.
21346 * config/microblaze/microblaze-protos.h (microblaze_constant_address_p):
21348 * config/microblaze/microblaze.h (microblaze_constant_address_p):
21349 CONSTANT_ADDRESS_P definition to microblaze_constant_address_p.
21350 * config/microblaze/microblaze.c (TARGET_PIC_DATA_TEXT_REL):
21351 New addressing mode for data-text relative position indepenedent code.
21352 (microblaze_classify_unspec): add 'UNSPEC_TEXT' case ->
21353 'ADDRESS_SYMBOLIC_TXT_REL'.
21354 (microblaze_classify_address): Add handling for UNSPEC + CONST_INT.
21355 (microblaze_legitimate_pic_operand): Exclude function calls from
21356 pic operands in case of TARGET_PIC_DATA_TEXT_REL option.
21357 (microblaze_legitimize_address): Generate 'UNSPEC_TEXT' for all possible
21359 (microblaze_address_insns): Add 'ADDRESS_SYMBOLIC_TXT_REL' case.
21360 (print_operand): Add 'ADDRESS_SYMBOLIC_TXT_REL' case.
21361 (print_operand_address): Add 'ADDRESS_SYMBOLIC_TXT_REL' case + handling
21362 for 'address + offset'.
21363 (microblaze_expand_prologue): Add new function prologue call for
21365 (microblaze_asm_generate_pic_addr_dif_vec): Override new target hook
21366 'TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC' to disable address diff vector
21367 table in case of TARGET_PIC_DATA_TEXT_REL.
21368 (expand_pic_symbol_ref): Add handling for 'UNSPEC_TEXT'.
21369 * config/microblaze/microblaze.md (TARGET_PIC_DATA_TEXT_REL):
21370 Add new macros 'UNSPEC_TEXT',
21371 'UNSPEC_SET_TEXT' + add rule for setting r20 in function prologue
21372 + exclude function calls from 'UNSPEC_PLT' in case of data text
21374 * doc/tm.texi.in (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC): Add
21375 new target hook for generating address diff vector tables in case of
21377 * doc/tm.texi : Regenerate.
21378 * stmt.c (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC): Append new condition
21379 'targetm.asm_out.generate_pic_addr_diff_vec' to flag_pic in case
21380 of addr diff vector generation.
21381 * target.def (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC): Add
21382 target hook definition.
21383 * targhooks.h, targhooks.c (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC):
21384 Add default function for generate_pic_addr_diff_vec -> flag_pic.
21385 * doc/invoke.texi (Add new pic option): Add new microblaze pic
21386 option for data text relative.
21388 2018-04-30 Richard Biener <rguenther@suse.de>
21390 * tree-chrec.h (evolution_function_is_constant_p): Remove
21392 * tree-cfg.c (tree_node_can_be_shared): Re-order checks.
21394 2018-04-30 Richard Biener <rguenther@suse.de>
21397 * dwarf2out.c (gen_producer_string): Ignore -fchecking[=].
21399 2018-04-30 Richard Biener <rguenther@suse.de>
21401 PR tree-optimization/28364
21402 PR tree-optimization/85275
21403 * tree-ssa-loop-ch.c (ch_base::copy_headers): Stop after
21404 copying first exit test.
21406 2018-04-28 Mark Wielaard <mark@klomp.org>
21408 * dwarf2out.c (dwarf2out_finish): Add .debug_addr table header for
21409 dwarf_version >= 5.
21410 (dwarf_AT): Handle DW_AT_addr_base.
21411 (add_top_level_skeleton_die_attrs): Use dwarf_AT for DW_AT_addr_base.
21413 2018-04-28 Uros Bizjak <ubizjak@gmail.com>
21416 * config/i386/i386.md (*ashl<dwi>3_doubleword_mask): New pattern.
21417 (*ashl<dwi>3_doubleword_mask_1): Ditto.
21418 (*<shift_insn><dwi>3_doubleword_mask): Ditto.
21419 (*<shift_insn><dwi>3_doubleword_mask_1): Ditto.
21421 2018-04-28 Richard Biener <rguenther@suse.de>
21423 * tree-cfg.c (verify_gimple_phi): Take a gphi * argument.
21424 (verify_gimple_in_cfg): Rename visited_stmts to visited_throwing_stmts
21425 to reflect use. Only add interesting stmts.
21427 2018-04-27 Martin Jambor <mjambor@suse.cz>
21430 * ipa-cp.c (find_aggregate_values_for_callers_subset): Make sure
21431 the jump function allows for passing through aggregate values.
21433 2018-04-27 David Malcolm <dmalcolm@redhat.com>
21435 * input.h (in_system_header_at): Convert from macro to inline
21437 (from_macro_expansion_at): Likewise.
21438 (from_macro_definition_at): Likewise.
21440 2018-04-27 Jeff Law <law@redhat.com>
21442 * config.gcc: Mark tile* targets as deprecated/obsolete.
21444 2018-04-27 Richard Biener <rguenther@suse.de>
21446 * config/aarch64/aarch64.c: Simplify ap.__stack advance and
21449 2018-04-27 Richard Biener <rguenther@suse.de>
21451 * tree-cfg.c (verify_expr): Make dead code hit gcc_unreachable.
21453 2018-04-27 Uros Bizjak <ubizjak@gmail.com>
21455 * config/i386/i386.md (*movti_internal): Substitute Ye constraint
21456 with Yd constraint. Set "preferred_for_speed" attribute from
21457 TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC for alternatives
21458 with Yd constraint.
21459 (*movdi_internal): Ditto.
21460 (movti_interunit splitters): Remove
21461 TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC from insn condition.
21462 (movdi_interunit splitters): Ditto.
21463 * config/i386/constraints.md (Ye): Remove.
21464 (Yd): Do not depend on TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC.
21466 2018-04-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21469 * config/aarch64/constraints.md (Usg): Limit to 31.
21470 (Usj): Limit to 63.
21472 2018-04-27 Jakub Jelinek <jakub@redhat.com>
21474 PR tree-optimization/85529
21475 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Add FIRST_BB
21476 argument. Don't call get_nonzero_bits if opcode is ERROR_MARK_NODE,
21477 rhs2 def stmt's bb is dominated by first_bb and it isn't an obvious
21478 zero extension or masking of the MSB bit.
21479 (optimize_range_tests): Add FIRST_BB argument, pass it through
21480 to optimize_range_tests_var_bound.
21481 (maybe_optimize_range_tests, reassociate_bb): Adjust
21482 optimize_range_tests callers.
21484 2018-04-26 Richard Biener <rguenther@suse.de>
21485 Jakub Jelinek <jakub@redhat.com>
21487 * cgraph.h (symbol_table): Just declare debug method here.
21488 * symtab.c (symbol_table::debug): Define.
21490 2018-04-26 Eric Botcazou <ebotcazou@adacore.com>
21492 * loop-invariant.c (may_assign_reg_p): Return false for frame pointer.
21494 2018-04-26 Uros Bizjak <ubizjak@gmail.com>
21496 * config/i386/i386.md ("isa" attribute): Add x64_sse2.
21497 ("enabled" attribute): Handle x64_sse2 "isa" attribute.
21498 (*movdi_internal): Substitute Yi and Yj constraint with x
21499 and Ym and Yn constraint with y constraint. Update "isa"
21500 attribute and set "preferred_for_speed" attribute from
21501 TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC for updated alternatives.
21502 (*movsi_internal): Ditto.
21503 (*movdf_internal): Ditto.
21504 (*movsf_internal): Ditto.
21505 (*zero_extendsidi2): Ditto.
21506 * config/i386/sse.md (vec_set<mode>_0): Ditto.
21507 (sse2_loadld): Ditto.
21508 (*vec_extract<ssevecmodelower>_0): Ditto.
21509 (*vec_extractv4si_0_zext_sse4): Ditto.
21510 (vec_concatv2di): Ditto.
21511 (*vec_dup<mode>): Ditto.
21512 * config/i386/mmx.md (*mov<mode>_internal): Ditto.
21513 * config/i386/constraints.md (Yi): Remove.
21518 2018-04-26 Nathan Sidwell <nathan@acm.org>
21520 * dumpfile.c (dump_open): New.
21521 (dump_open_alternate_stream, dump_start, dump_begin): Call it.
21522 (dump_finish): Detect stdio/stderr by value not name.
21524 2018-04-26 Jonathan Wakely <jwakely@redhat.com>
21526 * doc/invoke.texi (-Wreturn-type): Document default status for C++.
21528 2018-04-26 Tom de Vries <tom@codesourcery.com>
21531 * config/nvptx/nvptx.c (verify_neutering_jumps)
21532 (verify_neutering_labels): New function
21533 (nvptx_single): Use verify_neutering_jumps and verify_neutering_labels.
21535 2018-04-26 Tom de Vries <tom@codesourcery.com>
21538 * config/nvptx/nvptx.c (needs_neutering_p): New function.
21539 (nvptx_single): Use needs_neutering_p to skip over insns that do not
21542 2018-04-26 Richard Biener <rguenther@suse.de>
21543 Tom de Vries <tom@codesourcery.com>
21546 * lto-streamer-out.c (output_function): Fixup loops if required to match
21547 discovery done in the reader.
21549 2018-04-26 Richard Biener <rguenther@suse.de>
21551 PR tree-optimization/85116
21552 * tree-ssa-loop-ch.c (do_while_loop_p): A do-while loop should
21553 have a loop exit from the single latch predecessor. Remove
21554 case of header with just condition.
21555 (ch_base::copy_headers): Exclude infinite loops from any
21557 (pass_ch::execute): Record exits.
21559 2018-04-26 Richard Biener <rguenther@suse.de>
21561 * tree-vect-data-refs.c (vect_get_data_access_cost): Get
21562 prologue cost vector and pass it to vect_get_load_cost.
21563 (vect_get_peeling_costs_all_drs): Likewise.
21564 (vect_peeling_hash_get_lowest_cost): Likewise.
21565 (vect_enhance_data_refs_alignment): Likewise.
21567 2018-04-26 Richard Biener <rguenther@suse.de>
21569 PR middle-end/85450
21570 * tree-cfg.c (verify_gimple_assign_unary): Restore proper
21571 checking of integer<->pointer conversions.
21572 * omp-expand.c (expand_omp_for_static_nochunk): Avoid
21573 sign-/zero-extending pointer types.
21574 (expand_omp_for_static_chunk): Likewise.
21576 2018-03-22 Hans-Peter Nilsson <hp@axis.com>
21577 Jean Lee <xiaoyur347@gmail.com>
21579 * config/mips/mips.c (mips_asan_shadow_offset): New function.
21580 (TARGET_ASAN_SHADOW_OFFSET): Define.
21581 * config/mips/mips.h (FRAME_GROWS_DOWNWARD): Augment to also be
21582 true for -fsanitize=address.
21584 2018-04-25 Mark Wielaard <mark@klomp.org>
21586 * dwarf2out.c (file_info_cmp): Sort longer dir prefixes before
21589 2018-04-25 Jakub Jelinek <jakub@redhat.com>
21591 * config/i386/i386.md (*x86_mov<mode>cc_0_m1): Use type "alu1" rather
21592 than "alu", remove explicit "memory" and "imm_disp" attributes.
21593 (*x86_mov<mode>cc_0_m1_se, *x86_mov<mode>cc_0_m1_neg): Likewise.
21595 PR middle-end/85414
21596 * simplify-rtx.c (simplify_unary_operation_1) <case SIGN_EXTEND,
21597 case ZERO_EXTEND>: Pass SUBREG_REG (op) rather than op to
21598 gen_lowpart_no_emit.
21600 2018-04-25 Sebastian Peryt <sebastian.peryt@intel.com>
21603 * config/i386/i386.c (ix86_expand_builtin): Change memory
21604 operand to XI, extend p0 to Pmode.
21605 * config/i386/i386.md: Change unspec volatile and operand
21606 1 mode to XI, change operand 0 mode to P.
21608 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
21610 * config/nds32/nds32-predicates.c (nds32_can_use_bclr_p): Mask with
21611 GET_MODE_MASK before any checking.
21612 (nds32_can_use_bset_p): Likewise.
21613 (nds32_can_use_btgl_p): Likewise.
21615 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
21617 * config/nds32/nds32-doubleword.md: New define_split pattern for
21618 illegal register number.
21620 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
21622 * config/nds32/nds32.c (nds32_print_operand): Set op_value ealier.
21624 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
21626 * config/nds32/nds32.h (ASM_APP_ON): Add missing newline character.
21628 2018-04-25 Richard Biener <rguenther@suse.de>
21630 * lto-streamer.h (LTO_major_version): Bump to 8.
21632 2018-04-25 Jakub Jelinek <jakub@redhat.com>
21634 * BASE-VER: Set to 9.0.0.
21636 2018-04-24 Segher Boessenkool <segher@kernel.crashing.org>
21638 * config/rs6000/rs6000.c (init_float128_ieee): Fix spelling mistakes
21639 in __abskf2 and __powikf2.
21641 2018-04-24 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21644 * config/aarch64/constraints.md (Usg, Usj): New constraints.
21645 * config/aarch64/iterators.md (cmode_simd): New mode attribute.
21646 * config/aarch64/aarch64.md (*aarch64_ashr_sisd_or_int_<mode>3):
21647 Use the above on operand 2. Reindent.
21648 (*aarch64_lshr_sisd_or_int_<mode>3): Likewise.
21650 2018-04-24 H.J. Lu <hongjiu.lu@intel.com>
21653 * common/config/i386/i386-common.c (ix86_handle_option): Don't
21655 * config/i386/i386.opt (mcet): Removed.
21656 * doc/install.texi: Remove -mcet documentation.
21657 * doc/invoke.texi: Likewise.
21659 2018-04-24 H.J. Lu <hongjiu.lu@intel.com>
21662 * doc/install.texi: Remove -mcet from bootstrap-cet.
21664 2018-04-24 Jakub Jelinek <jakub@redhat.com>
21667 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Don't define
21668 __builtin_ia32_readeflags_u32 and __builtin_ia32_writeeflags_u32
21672 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Punt if
21673 const_vector is not CONST_VECTOR or SYMBOL_REF for a constant pool
21674 containing a CONST_VECTOR.
21676 2018-04-24 Cesar Philippidis <cesar@codesourcery.com>
21678 * doc/install.texi: Update newlib dependency for nvptx.
21680 2018-04-24 Jakub Jelinek <jakub@redhat.com>
21683 * config/i386/i386.c (ix86_expand_vector_init_one_var): Use UINTVAL
21684 instead of INTVAL when shifting x left.
21686 2018-04-24 Andreas Krebbel <krebbel@linux.ibm.com>
21688 PR tree-optimization/85478
21689 * tree-vect-loop.c (vect_analyze_loop_2): Do not call
21690 vect_grouped_store_supported for single element vectors.
21692 2018-04-24 Richard Biener <rguenther@suse.de>
21695 * config/i386/i386.c (ix86_add_stmt_cost): Restrict strided
21696 load cost increase to the case of non-constant step.
21698 2018-04-24 Jakub Jelinek <jakub@redhat.com>
21701 * reg-stack.c (move_for_stack_reg): Don't fail assertion about dead
21702 destination if any_malformed_asm.
21704 2018-04-23 Eric Botcazou <ebotcazou@adacore.com>
21706 PR middle-end/85496
21707 * expr.c (store_field): In the bitfield case, if the value comes from
21708 a function call and is returned in registers by means of a PARALLEL,
21709 do not change the mode of the temporary unless BLKmode and VOIDmode.
21711 2018-04-23 Andrey Belevantsev <abel@ispras.ru>
21713 PR rtl-optimization/85423
21714 * sel-sched-ir.c (has_dependence_note_mem_dep): Only discard
21715 dependencies to debug insns when the previous insn is non-debug.
21717 2018-04-23 Claudiu Zissulescu <claziss@synopsys.com>
21719 * config/arc/arc.md ("vunspec"): Delete it, unify all the unspec
21720 enums into a single definition.
21721 (fls): Fix predicates and printing.
21724 2018-04-23 Claudiu Zissulescu <claziss@synopsys.com>
21726 * config/arc/arc-protos.h (check_if_valid_sleep_operand): Remove.
21727 * config/arc/arc.c (arc_expand_builtin): Sleep accepts registers
21728 and short u6 immediate.
21729 (check_if_valid_sleep_operand): Remove.
21730 * config/arc/arc.md (Sleep): Accepts registers and u6 immediates.
21732 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
21734 * config/nds32/nds32.c (nds32_compute_stack_frame): Consider
21735 flag_always_save_lp condition.
21736 * config/nds32/nds32.opt (malways-save-lp): New option.
21738 2018-04-22 Shiva Chen <shiva0217@gmail.com>
21740 * config/nds32/nds32-protos.h (nds32_use_load_post_increment): Declare.
21741 * config/nds32/nds32.c (nds32_use_load_post_increment): New.
21742 * config/nds32/nds32.h
21743 (USE_LOAD_POST_INCREMENT, USE_LOAD_POST_DECREMENT): Define.
21744 (USE_STORE_POST_INCREMENT, USE_STORE_POST_DECREMENT): Define.
21746 2018-04-22 Shiva Chen <shiva0217@gmail.com>
21748 * config/nds32/nds32-protos.h (nds32_ls_333_p): Remove.
21749 * config/nds32/nds32.c (nds32_ls_333_p): Remove.
21751 2018-04-22 Shiva Chen <shiva0217@gmail.com>
21752 Chung-Ju Wu <jasonwucj@gmail.com>
21754 * config/nds32/nds32-protos.h (nds32_case_vector_shorten_mode):
21756 * config/nds32/nds32.c (nds32_case_vector_shorten_mode): New function.
21757 * config/nds32/nds32.h (CASE_VECTOR_SHORTEN_MODE): Modify.
21759 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
21761 * config/nds32/nds32.c (nds32_compute_stack_frame): Fix wrong value.
21763 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
21765 * config/nds32/nds32-protos.h (nds32_data_alignment,
21766 nds32_local_alignment): Declare.
21767 * config/nds32/nds32.c (nds32_data_alignment, nds32_constant_alignment,
21768 nds32_local_alignment): New functions.
21769 (TARGET_CONSTANT_ALIGNMENT): Define.
21770 * config/nds32/nds32.h (DATA_ALIGNMENT, LOCAL_ALIGNMENT): Define.
21772 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
21774 * config/nds32/nds32.c
21775 (TARGET_HARD_REGNO_MODE_OK): Move to the bottom of file.
21776 (TARGET_MODES_TIEABLE_P): Likewise.
21778 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
21780 * config/nds32/nds32.c (nds32_asm_file_start): Display optimization
21781 level Ofast and Og.
21783 2018-04-22 Monk Chiang <sh.chiang04@gmail.com>
21784 Chung-Ju Wu <jasonwucj@gmail.com>
21786 * config/nds32/constants.md (unspec_volatile_element): Add enum values
21787 for unaligned access.
21788 * config/nds32/nds32-intrinsic.c: Implementation of expanding
21790 * config/nds32/nds32-intrinsic.md: Likewise.
21791 * config/nds32/nds32_intrinsic.h: Likewise.
21792 * config/nds32/nds32.h (nds32_builtins): Likewise.
21793 * config/nds32/nds32.opt (munaligned-access): New option.
21794 * config/nds32/nds32.c (nds32_asm_file_start): Display
21795 flag_unaligned_access status.
21797 2018-04-20 Kito Cheng <kito.cheng@gmail.com>
21799 * config/riscv/elf.h (LINK_SPEC): Pass --no-relax if
21800 -mno-relax is present.
21801 * config/riscv/linux.h (LINK_SPEC): Ditto.
21803 2018-04-20 Martin Sebor <msebor@redhat.com>
21806 * gimple-fold.c (gimple_fold_builtin_strcpy): Suppress -Wrestrict
21808 (gimple_fold_builtin_stxcpy_chk): Same.
21809 * gimple-ssa-warn-restrict.c (check_bounds_or_overlap): Same.
21811 2018-04-20 Michael Meissner <meissner@linux.ibm.com>
21814 * config/rs6000/rs6000.c (init_float128_ieee): Add support to call
21815 __powikf2 when long double is IEEE 128-bit.
21817 2018-04-20 Kito Cheng <kito.cheng@gmail.com>
21819 * config/riscv/riscv.c (riscv_first_stack_step): Round up min
21820 step to make sure stack always aligned.
21822 2018-04-20 Carl Love <cel@us.ibm.com>
21825 * config/rs6000/rs6000-c.c (rs6000_gimple_fold_builtin): Add
21826 size check for arg0.
21828 2018-04-20 Nathan Sidwell <nathan@codesourcery.com>
21829 Tom de Vries <tom@codesourcery.com>
21832 * config/nvptx/nvptx.c (nvptx_emit_forking, nvptx_emit_joining):
21833 Emit insns for calls too.
21834 (nvptx_find_par): Always look for worker-level predecessor insn.
21835 (nvptx_propagate): Add is_call parm, return bool. Copy frame for
21837 (nvptx_vpropagate, nvptx_wpropagate): Adjust.
21838 (nvptx_process_pars): Propagate frames for calls.
21840 2018-04-20 H.J. Lu <hongjiu.lu@intel.com>
21843 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET):
21845 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
21846 (ix86_handle_option): Don't handle OPT_mibt.
21847 * config/i386/cet.h: Check __CET__ instead of __IBT__ and
21849 * config/i386/driver-i386.c (host_detect_local_cpu): Remove
21851 * config/i386/i386-c.c (ix86_target_macros_internal): Don't
21852 check OPTION_MASK_ISA_IBT nor flag_cf_protection.
21853 (ix86_target_macros): Define __CET__ with flag_cf_protection
21854 for -fcf-protection.
21855 * config/i386/i386.c (isa2_opts): Remove -mibt.
21856 * config/i386/i386.h (TARGET_IBT): Removed.
21857 (TARGET_IBT_P): Likewise.
21858 (ix86_valid_target_attribute_inner_p): Don't check OPT_mibt.
21859 * config/i386/i386.md (nop_endbr): Don't check TARGET_IBT.
21860 * config/i386/i386.opt (mcet): Update help message.
21861 (mshstk): Likewise.
21863 * doc/invoke.texi: Remove -mibt. Document __CET__. Document
21864 -mcet as an alias for -mshstk.
21866 2018-04-20 Richard Biener <rguenther@suse.de>
21868 PR middle-end/85475
21869 * match.pd ((X * CST) * Y -> (X * Y) * CST): Avoid exponential
21870 complexity by forcing a single use of the multiply operand.
21872 2018-04-20 Martin Jambor <mjambor@suse.cz>
21875 * ipa-cp.c (cgraph_edge_brings_value_p): Move check for self-feeding
21876 recursion dependency to only apply to non-clones.
21878 2018-04-20 Martin Jambor <mjambor@suse.cz>
21881 * ipa-cp.c (create_specialized_node): Check that clones of
21882 self-recursive edges exist during IPA-CP.
21884 2018-04-19 Toon Moene <toon@moene.org>
21886 * doc/invoke.texi: Add -floop-unroll-and-jam to options enabled
21889 2018-04-19 Jakub Jelinek <jakub@redhat.com>
21891 PR tree-optimization/85467
21892 * fold-const.c (fold_ternary_loc) <case BIT_FIELD_REF>: Use
21893 VECTOR_TYPE_P macro. If type is vector type, VIEW_CONVERT_EXPR the
21894 VECTOR_CST element to type.
21896 2018-04-19 H.J. Lu <hongjiu.lu@intel.com>
21899 * config/i386/i386.h (STACK_SAVEAREA_MODE): New.
21900 * config/i386/i386.md (builtin_setjmp_setup): Removed.
21901 (builtin_longjmp): Likewise.
21902 (save_stack_nonlocal): New pattern.
21903 (restore_stack_nonlocal): Likewise.
21905 2018-04-19 H.J. Lu <hongjiu.lu@intel.com>
21908 * config/i386/cet.c (file_end_indicate_exec_stack_and_cet):
21909 Replace ASM_OUTPUT_LABEL with fprintf.
21911 2018-04-19 H.J. Lu <hongjiu.lu@intel.com>
21914 * config/i386/cet.c (file_end_indicate_exec_stack_and_cet):
21915 Check flag_cf_protection instead of TARGET_IBT and TARGET_SHSTK.
21916 * config/i386/i386-c.c (ix86_target_macros_internal): Also
21917 define __IBT__ and __SHSTK__ for -fcf-protection.
21918 * config/i386/i386.c (pass_insert_endbranch::gate): Don't check
21920 (ix86_trampoline_init): Likewise.
21921 (x86_output_mi_thunk): Likewise.
21922 (ix86_notrack_prefixed_insn_p): Likewise.
21923 (ix86_option_override_internal): Don't disallow -fcf-protection.
21924 * config/i386/i386.md (rdssp<mode>): Also enable for
21926 (incssp<mode>): Likewise.
21927 (nop_endbr): Likewise.
21928 * config/i386/i386.opt (mcet): Change help message to built-in
21931 (mshstk): Likewise.
21932 * doc/invoke.texi: Remove -mcet, -mibt and -mshstk condition
21933 on -fcf-protection. Change -mcet, -mibt and -mshstk to only
21934 enable CET built-in functions.
21936 2018-04-19 Sebastian Peryt <sebastian.peryt@intel.com>
21938 * common/config/i386/i386-common.c
21939 (OPTION_MASK_ISA_MOVDIRI_SET, OPTION_MASK_ISA_MOVDIR64B_SET,
21940 OPTION_MASK_ISA_MOVDIRI_UNSET,
21941 OPTION_MASK_ISA_MOVDIR64B_UNSET): New defines.
21942 (ix86_handle_option): Handle -mmovdiri and -mmovdir64b.
21943 * config.gcc (movdirintrin.h): New header.
21944 * config/i386/cpuid.h (bit_MOVDIRI,
21945 bit_MOVDIR64B): New bits.
21946 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mmovdiri
21948 * config/i386/i386-builtin-types.def ((VOID, PUNSIGNED, UNSIGNED),
21949 (VOID, PVOID, PCVOID)): New function types.
21950 * config/i386/i386-builtin.def (__builtin_ia32_directstoreu_u32,
21951 __builtin_ia32_directstoreu_u64,
21952 __builtin_ia32_movdir64b): New builtins.
21953 * config/i386/i386-c.c (__MOVDIRI__, __MOVDIR64B__): New.
21954 * config/i386/i386.c (ix86_target_string): Added -mmovdir64b
21956 (ix86_valid_target_attribute_inner_p): Ditto.
21957 (ix86_expand_special_args_builtin): Added VOID_FTYPE_PUNSIGNED_UNSIGNED
21958 and VOID_FTYPE_PUNSIGNED_UNSIGNED.
21959 (ix86_expand_builtin): Expand IX86_BUILTIN_MOVDIR64B.
21960 * config/i386/i386.h (TARGET_MOVDIRI, TARGET_MOVDIRI_P,
21961 TARGET_MOVDIR64B, TARGET_MOVDIR64B_P): New.
21962 * config/i386/i386.md (UNSPECV_MOVDIRI, UNSPECV_MOVDIR64B): New.
21963 (movdiri<mode>, movdir64b_<mode>): New.
21964 * config/i386/i386.opt: Add -mmovdiri and -mmovdir64b.
21965 * config/i386/immintrin.h: Include movdirintrin.h.
21966 * config/i386/movdirintrin.h: New file.
21967 * doc/invoke.texi: Added -mmovdiri and -mmovdir64b.
21969 2018-04-19 Richard Biener <rguenther@suse.de>
21971 PR middle-end/85455
21972 * cfg.c (clear_bb_flags): When loop state says we have
21973 marked irreducible regions also preserve BB_IRREDUCIBLE_LOOP.
21975 2018-04-19 Richard Biener <rguenther@suse.de>
21977 PR tree-optimization/84737
21978 * tree-vect-data-refs.c (vect_copy_ref_info): New function
21979 copying restrict info.
21980 (vect_setup_realignment): Use it.
21981 * tree-vectorizer.h (vect_copy_ref_info): Declare.
21982 * tree-vect-stmts.c (vectorizable_store): Copy ref info from
21983 the first DR to all generated stores.
21984 (vectorizable_load): Likewise for loads.
21986 2018-04-19 Jakub Jelinek <jakub@redhat.com>
21988 PR tree-optimization/85446
21989 * match.pd ((intptr_t) x eq/ne CST to x eq/ne (typeof x) cst): Require
21990 the integral and pointer types to have the same precision.
21992 * doc/install.texi: Document --disable-cet being the default and
21995 2018-04-18 Martin Liska <mliska@suse.cz>
21997 * ipa-devirt.c (odr_subtypes_equivalent_p): Fix GNU coding
22000 2018-04-18 Martin Liska <mliska@suse.cz>
22003 2018-03-02 Eric Botcazou <ebotcazou@adacore.com>
22006 * ipa-devirt.c (odr_subtypes_equivalent_p): Get the ODR type of both
22007 arguments if they are comparable.
22009 2018-04-18 Martin Liska <mliska@suse.cz>
22012 2018-03-13 Eric Botcazou <ebotcazou@adacore.com>
22015 * ipa-devirt.c (odr_subtypes_equivalent_p): Do not get the ODR type of
22018 2018-04-18 H.J. Lu <hongjiu.lu@intel.com>
22021 * config/i386/i386.c (ix86_expand_split_stack_prologue): Insert
22022 ENDBR after calling __morestack.
22024 2018-04-18 David Malcolm <dmalcolm@redhat.com>
22027 * configure.ac (gcc-driver-name.h): Honor --with-gcc-major-version
22028 by using gcc_base_ver to generate a gcc_driver_version, and use
22029 it when generating GCC_DRIVER_NAME.
22030 * configure: Regenerate.
22032 2018-04-18 Jakub Jelinek <jakub@redhat.com>
22035 * config.gcc: Obsolete powerpc*-*-*spe*.
22037 2018-04-17 Jakub Jelinek <jakub@redhat.com>
22040 * dbxout.c (dbxout_int): Perform negation in unsigned int type.
22041 (stabstr_D): Change type of unum from unsigned int to
22042 unsigned HOST_WIDE_INT. Perform negation in unsigned HOST_WIDE_INT
22045 2018-04-17 Jim Wilson <jimw@sifive.com>
22048 * config/riscv/riscv.c (riscv_compute_frame_info): Add calls to
22049 RISCV_STACK_ALIGN when using outgoing_args_size and pretend_args_size.
22050 Set arg_pointer_offset after using pretend_args_size.
22052 2018-04-17 Jakub Jelinek <jakub@redhat.com>
22054 PR rtl-optimization/85431
22055 * dse.c (record_store): Ignore zero width stores.
22058 * asan.c (handle_builtin_stack_restore): Adjust comment. Emit
22059 __asan_allocas_unpoison call and last_alloca_addr = new_sp before
22060 __builtin_stack_restore rather than after it.
22061 * builtins.c (expand_asan_emit_allocas_unpoison): Pass
22062 arg1 + (virtual_dynamic_stack_rtx - stack_pointer_rtx) as second
22063 argument instead of virtual_dynamic_stack_rtx.
22065 2018-04-17 Kelvin Nilsen <kelvin@gcc.gnu.org>
22067 * config/rs6000/rs6000-protos.h (rs6000_builtin_is_supported_p):
22069 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
22070 Add note to error message to explain internal mapping of overloaded
22071 built-in function name to non-overloaded built-in function name.
22072 * config/rs6000/rs6000.c (rs6000_builtin_is_supported_p): New
22075 2018-04-17 Michael Meissner <meissner@linux.vnet.ibm.com>
22078 * config/rs6000/rs6000.md (pack<mode>): Do not try handle a pack
22079 where the inputs overlap with the output.
22081 2018-04-17 Jakub Jelinek <jakub@redhat.com>
22083 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): Add
22084 (=v, v) alternative and explicit "memory" attribute.
22085 (vec_extract_lo_<mode><mask_name>): Likewise. Also add
22086 "type", "prefix", "prefix_extra", "length_immediate" and "mode"
22088 (vec_extract_lo_<mode><mask_name>): Add (=v, v) alternative and use
22089 "sselog1" type instead of "sselog".
22090 (vec_extract_hi_<mode><mask_name>): Use "sselog1" type instead of
22091 "sselog". Remove explicit "memory" attribute.
22092 (vec_extract_lo_v32hi): Add (=v, v) alternative and explicit "memory",
22093 "type", "prefix", "prefix_extra", "length_immediate" and "mode"
22095 (vec_extract_hi_v32hi): Merge all alternatives into one, use
22096 "sselog1" type instead of "sselog". Remove explicit "memory"
22098 (vec_extract_hi_v16hi): Merge each pair of alternatives into one,
22099 use "sselog1" type instead of "sselog". Remove explicit "memory"
22101 (vec_extract_lo_v64qi): Add (=v, v) alternative and explicit "memory",
22102 "type", "prefix", "prefix_extra", "length_immediate" and "mode"
22104 (vec_extract_hi_v64qi): Merge all alternatives into one, use
22105 "sselog1" type instead of "sselog". Remove explicit "memory"
22107 (vec_extract_hi_v32qi): Merge each pair of alternatives into one,
22108 use "sselog1" type instead of "sselog". Remove explicit "memory"
22112 * config/i386/i386.md (*ashlqi3_1_slp): Use alu1 type instead of alu.
22114 PR middle-end/85414
22115 * rtlhooks.c (gen_lowpart_if_possible): Don't call gen_lowpart_SUBREG
22118 2018-04-17 Martin Jambor <mjambor@suse.cz>
22121 * ipa-cp.c (create_specialized_node): Call
22122 expand_all_artificial_thunks if necessary.
22124 2018-04-17 Martin Liska <mliska@suse.cz>
22127 * ipa-devirt.c (odr_types_equivalent_p): Remove trailing
22128 in message, remote space in between '_G' and '('.
22130 2018-04-17 Jakub Jelinek <jakub@redhat.com>
22133 * config/i386/sse.md (reduces<mode><mask_scalar_name>,
22134 avx512f_vmcmp<mode>3<round_saeonly_name>,
22135 avx512f_vmcmp<mode>3_mask<round_saeonly_name>,
22136 avx512f_sgetexp<mode><mask_scalar_name><round_saeonly_scalar_name>,
22137 avx512f_rndscale<mode><round_saeonly_name>,
22138 avx512dq_ranges<mode><mask_scalar_name><round_saeonly_scalar_name>,
22139 avx512f_vgetmant<mode><mask_scalar_name><round_saeonly_scalar_name>):
22140 Use %<iptr>2 instead of %2 for -masm=intel.
22141 (avx512f_vcvtss2usi<round_name>, avx512f_vcvtss2usiq<round_name>,
22142 avx512f_vcvttss2usi<round_saeonly_name>,
22143 avx512f_vcvttss2usiq<round_saeonly_name>): Use %k1 instead of %1 for
22145 (avx512f_vcvtsd2usi<round_name>, avx512f_vcvtsd2usiq<round_name>,
22146 avx512f_vcvttsd2usi<round_saeonly_name>,
22147 avx512f_vcvttsd2usiq<round_saeonly_name>, ufloatv2siv2df2<mask_name>):
22148 Use %q1 instead of %1 for -masm=intel.
22149 (avx512f_sfixupimm<mode><sd_maskz_name><round_saeonly_name>,
22150 avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Use %<iptr>3 instead
22151 of %3 for -masm=intel.
22152 (sse2_shufpd_v2df_mask): Fix a typo, change %{6%} to %{%6%} for
22154 (*avx512vl_<code>v2div2qi2_store): Use %w0 instead of %0 for
22156 (*avx512vl_<code><mode>v4qi2_store): Use %k0 instead of %0 for
22158 (avx512vl_<code><mode>v4qi2_mask_store): Use a single pattern with
22159 %k0 and %1 for -masm=intel rather than two patterns, one with %0 and
22161 (*avx512vl_<code><mode>v8qi2_store): Use %q0 instead of %0 for
22163 (avx512vl_<code><mode>v8qi2_mask_store): Use a single pattern with
22164 %q0 and %1 for -masm=intel rather than two patterns, one with %0 and
22165 %g1 and one with %0 and %1.
22166 (avx512er_vmrcp28<mode><round_saeonly_name>,
22167 avx512er_vmrsqrt28<mode><round_saeonly_name>): Use %<iptr>1 instead of
22168 %1 for -masm=intel.
22169 (avx5124fmaddps_4fmaddps_mask, avx5124fmaddps_4fmaddss_mask,
22170 avx5124fmaddps_4fnmaddps_mask, avx5124fmaddps_4fnmaddss_mask,
22171 avx5124vnniw_vp4dpwssd_mask, avx5124vnniw_vp4dpwssds_mask): Swap order
22172 of %0 and %{%4%} for -masm=intel.
22173 (avx5124fmaddps_4fmaddps_maskz, avx5124fmaddps_4fmaddss_maskz,
22174 avx5124fmaddps_4fnmaddps_maskz, avx5124fmaddps_4fnmaddss_maskz,
22175 avx5124vnniw_vp4dpwssd_maskz, avx5124vnniw_vp4dpwssds_maskz): Swap
22176 order of %0 and %{%5%}%{z%} for -masm=intel.
22178 2018-04-17 Jan Hubicka <jh@suse.cz>
22181 * ipa-devirt.c (odr_types_equivalent_p): Handle bit fields.
22183 2018-04-17 Martin Liska <mliska@suse.cz>
22186 * multiple_target.c (create_dispatcher_calls): Set apostrophes
22187 for target_clone error message. Make default implementation
22188 clone to be a local declaration.
22189 (separate_attrs): Add new argument and check for an empty
22191 (expand_target_clones): Handle it.
22192 (ipa_target_clone): Make redirection just for target_clones
22195 2018-04-16 Cesar Philippidis <cesar@codesourcery.com>
22196 Tom de Vries <tom@codesourcery.com>
22198 PR middle-end/84955
22199 * omp-expand.c (expand_oacc_for): Add dummy false branch for
22200 tiled basic blocks without omp continue statements.
22202 2018-04-16 Aaron Sawdey <acsawdey@linux.ibm.com>
22205 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Mark
22206 vec_extract expression as having side effects to make sure it gets
22209 2018-04-16 H.J. Lu <hongjiu.lu@intel.com>
22212 * config/i386/i386.c (get_builtin_code_for_version): Check
22215 2018-04-16 Olga Makhotina <olga.makhotina@intel.com>
22218 * config.gcc: Support "skylake".
22219 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
22221 * config/i386/i386.c (m_SKYLAKE): Define.
22222 (processor_target_table): Add "skylake".
22223 (ix86_option_override_internal): Add "skylake".
22224 (get_builtin_code_for_version): Handle PROCESSOR_SKYLAKE,
22225 PROCESSOR_CANNONLAKE.
22226 (get_builtin_code_for_version): Fix priority for
22227 PROCESSOR_ICELAKE_CLIENT, PROCESSOR_ICELAKE_SERVER,
22228 PROCESSOR_SKYLAKE-AVX512.
22229 * config/i386/i386.h (processor_costs): Define TARGET_SKYLAKE.
22230 (processor_type): Add PROCESSOR_SKYLAKE.
22232 2018-04-16 Paolo Carlini <paolo.carlini@oracle.com>
22233 Jason Merrill <jason@redhat.com>
22236 * convert.c (convert_to_integer_1): Use direct recursion for
22237 enumeral types and types with a precision less than the number
22238 of bits in their mode.
22240 2018-04-16 Julia Koval <julia.koval@intel.com>
22243 * config/i386/x86-tune.def (X86_TUNE_SSE_UNALIGNED_LOAD_OPTIMAL,
22244 X86_TUNE_SSE_UNALIGNED_STORE_OPTIMAL): Add m_SKYLAKE_AVX512
22246 2018-04-14 Segher Boessenkool <segher@kernel.crashing.org>
22249 * config/rs6000/rs6000.opt (mdirect-move): Make deprecated.
22250 * doc/invoke.texi (RS/6000 and PowerPC Options): Remove -mdirect-move
22251 and -mno-direct-move.
22253 2018-04-13 Paul A. Clarke <pc@us.ibm.com>
22256 * config/rs6000/emmintrin.h (_mm_slli_epi{16,32,64}):
22257 Ensure that vec_splat_s32 is only called with 0 <= shift < 16.
22258 Ensure negative shifts result in {0}.
22260 2018-04-13 Vladimir Makarov <vmakarov@redhat.com>
22262 PR rtl-optimization/79916
22263 * config/rs6000/rs6000.c (rs6000_emit_move): Use assigned hard
22264 regs (if any) to define how to gnerate SD moves when LRA is in
22267 2018-04-13 Jakub Jelinek <jakub@redhat.com>
22269 PR rtl-optimization/85393
22270 * except.h (expand_dw2_landing_pad_for_region): Remove declaration.
22271 * except.c (expand_dw2_landing_pad_for_region): Make static.
22272 * bb-reorder.c (fix_up_crossing_landing_pad): In new_bb emit just
22273 a label and unconditional jump to old_bb, rather than
22274 expand_dw2_landing_pad_for_region insn(s) and jump to single_succ
22277 PR rtl-optimization/85376
22278 * simplify-rtx.c (simplify_const_unary_operation): For CLZ and CTZ and
22279 zero op0, if C?Z_DEFINED_VALUE_AT_ZERO is false, return NULL_RTX
22280 instead of a specific value.
22282 2018-04-13 Jan Hubicka <hubicka@ucw.cz>
22283 Bin Cheng <bin.cheng@arm.com>
22285 PR tree-optimization/82965
22286 PR tree-optimization/83991
22287 * cfgloopanal.c (expected_loop_iterations_unbounded): Add
22288 by_profile_only parameter.
22289 * cfgloopmanip.c (scale_loop_profile): Further scale loop's profile
22290 information if the loop was predicted to iterate too many times.
22291 * cfgloop.h (expected_loop_iterations_unbounded): Update prototype
22293 2018-04-13 Jan Hubicka <hubicka@ucw.cz>
22296 * config/i386/i386.c (ix86_can_inline_p): Allow safe transitions for
22299 2018-04-13 Martin Liska <mliska@suse.cz>
22300 Jakub Jelinek <jakub@redhat.com>
22302 PR middle-end/81657
22303 * expr.h (enum block_op_methods): Add BLOCK_OP_NO_LIBCALL_RET.
22304 * expr.c (emit_block_move_hints): Handle BLOCK_OP_NO_LIBCALL_RET.
22305 * builtins.c (expand_builtin_memory_copy_args): Use
22306 BLOCK_OP_NO_LIBCALL_RET method for mempcpy with non-ignored target,
22307 handle dest_addr == pc_rtx.
22309 2018-04-12 Segher Boessenkool <segher@kernel.crashing.org>
22312 * config/rs6000/rs6000.md (fix_trunc<mode>si2): Use legacy code if
22313 asked to not generate direct moves.
22314 (fix_trunc<mode>si2_stfiwx): Similar.
22315 (fix_trunc<mode>si2_internal): Similar.
22317 2018-04-12 Jakub Jelinek <jakub@redhat.com>
22320 * var-tracking.c (add_stores): Handle STRICT_LOW_PART SET_DEST.
22321 * cselib.c (cselib_record_sets): For STRICT_LOW_PART dest,
22322 lookup if dest in some wider mode is known to be const0_rtx and
22323 if so, record permanent equivalence for it to be ZERO_EXTEND of
22324 the narrower mode destination.
22326 2018-04-12 Cesar Philippidis <cesar@codesourcery.com>
22328 * lto-streamer-out.c (output_function): Revert 259346.
22329 * omp-expand.c (expand_oacc_for): Likewise.
22331 2018-04-12 Alexander Monakov <amonakov@ispras.ru>
22333 PR rtl-optimization/85354
22334 * sel-sched-ir.c (sel_init_pipelining): Move cfg_cleanup call...
22335 * sel-sched.c (sel_global_init): ... here.
22337 2018-04-12 Eric Botcazou <ebotcazou@adacore.com>
22340 * lto-wrapper.c (debug_objcopy): Open the files in binary mode.
22341 * dwarf2out.c (dwarf2out_early_finish): Do not generate assembly in LTO
22342 mode for PE-COFF targets.
22343 * config/i386/i386-protos.h (i386_pe_asm_lto_start): Declare.
22344 (i386_pe_asm_lto_end): Likewise.
22345 * config/i386/cygming.h (TARGET_ASM_LTO_START): Define.
22346 (TARGET_ASM_LTO_END): Likewise.
22347 * config/i386/winnt.c (saved_debug_info_level): New static variable.
22348 (i386_pe_asm_lto_start): New function.
22349 (i386_pe_asm_lto_end): Likewise.
22351 2018-04-12 Cesar Philippidis <cesar@codesourcery.com>
22352 Richard Biener <rguenther@suse.de>
22354 PR middle-end/84955
22355 * lto-streamer-out.c (output_function): Fix CFG loop state before
22357 * omp-expand.c (expand_oacc_for): Handle calls to internal
22358 functions like regular functions.
22360 2018-04-12 Richard Biener <rguenther@suse.de>
22363 * dwarf2out.c (init_sections_and_labels): Use debug_line_section[_label]
22364 for the early LTO debug to properly generate references to it
22365 during DIE emission. Do not re-use that for the skeleton for
22367 (dwarf2out_early_finish): Likewise.
22369 2018-04-12 Jakub Jelinek <jakub@redhat.com>
22372 * config/i386/sse.md
22373 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name> split,
22374 <mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name> split,
22375 vec_extract_lo_<mode><mask_name> split, vec_extract_lo_v32hi,
22376 vec_extract_lo_v64qi): For non-AVX512VL if input is xmm16+ reg
22377 and output is a reg, avoid creating invalid lowpart subreg, but
22378 instead split into a 512-bit move. Don't split if not AVX512VL,
22379 input is xmm16+ reg and output is a mem.
22380 (vec_extract_lo_<mode><mask_name>, vec_extract_lo_v32hi,
22381 vec_extract_lo_v64qi): Don't require split if not AVX512VL, input is
22382 xmm16+ reg and output is a mem.
22384 2018-04-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22386 * config/s390/s390.c (s390_output_indirect_thunk_function): Check
22387 also for flag_dwarf2_cfi_asm.
22389 2018-04-12 Jakub Jelinek <jakub@redhat.com>
22391 PR rtl-optimization/85342
22392 * regcprop.c (copyprop_hardreg_forward_1): Remove replaced array, use
22393 a bool scalar var inside of the loop instead. Don't try to update
22394 recog_data.operand after failed apply_change_group.
22396 2018-04-12 Tom de Vries <tom@codesourcery.com>
22399 * config/nvptx/nvptx.c (flexible_array_member_type_p): New function.
22400 (nvptx_assemble_decl_begin): Add undefined param. Declare undefined
22401 array with flexible array member as array without given dimension.
22402 (nvptx_assemble_undefined_decl): Set nvptx_assemble_decl_begin call
22403 argument for undefined param to true.
22405 2018-04-11 Aaron Sawdey <acsawdey@linux.ibm.com>
22408 * doc/invoke.texi (RS/6000 and PowerPC Options): Document options
22409 -mcall- and -mtraceback=. Remove options -mabi=spe and -mabi=no-spe
22410 from PowerPC section.
22411 * config/rs6000/sysv4.opt (mcall-): Improve help text.
22412 * config/rs6000/rs6000.opt (mblock-compare-inline-limit=): Trim
22413 help text that is too long.
22414 * config/rs6000/rs6000.opt (mblock-compare-inline-loop-limit=): Trim
22415 help text that is too long.
22416 * config/rs6000/rs6000.opt (mstring-compare-inline-limit=): Trim
22417 help text that is too long.
22419 2018-04-11 Uros Bizjak <ubizjak@gmail.com>
22421 * config/alpha/alpha.md (stack_probe_internal): Rename
22422 from "probe_stack". Update all callers.
22424 2018-04-11 Alexander Monakov <amonakov@ispras.ru>
22426 PR rtl-optimization/84566
22427 * sched-deps.c (sched_analyze_insn): Check deps->readonly when invoking
22428 sched_macro_fuse_insns.
22430 2018-04-11 Alexander Monakov <amonakov@ispras.ru>
22433 * sched-rgn.c (add_branch_dependences): Move sel_sched_p check here...
22434 (compute_block_dependences): ... from here.
22436 2018-04-11 Jakub Jelinek <jakub@redhat.com>
22438 PR tree-optimization/85331
22439 * vec-perm-indices.h (vec_perm_indices::clamp): Change input type
22440 from int to HOST_WIDE_INT.
22442 2018-04-11 Martin Jambor <mjambor@suse.cz>
22445 * ipa-cp.c (propagate_vals_across_pass_through): Expand comment.
22446 (cgraph_edge_brings_value_p): New parameter dest_val, check if it is
22447 not the same as the source val.
22448 (cgraph_edge_brings_value_p): New parameter.
22449 (gather_edges_for_value): Pass destination value to
22450 cgraph_edge_brings_value_p.
22451 (perhaps_add_new_callers): Likewise.
22452 (get_info_about_necessary_edges): Likewise and exclude values brought
22453 only by self-recursive edges.
22454 (create_specialized_node): Redirect only clones of self-calling edges.
22455 (+self_recursive_pass_through_p): New function.
22456 (find_more_scalar_values_for_callers_subset): Use it.
22457 (find_aggregate_values_for_callers_subset): Likewise.
22458 (known_aggs_to_agg_replacement_list): Removed.
22459 (decide_whether_version_node): Re-calculate known constants for all
22460 remaining context clones.
22462 2018-04-11 Richard Biener <rguenther@suse.de>
22465 * dwarf2out.c (dwarf2out_finish): Remove DW_AT_stmt_list attribute
22466 from early DWARF output.
22467 (dwarf2out_early_finish): Output line info unconditionally into
22468 early DWARF and add reference to it.
22470 2018-04-11 Jakub Jelinek <jakub@redhat.com>
22473 * config/i386/sse.md (iptr): Add V16SFmode and V8DFmode cases.
22474 (<avx512>_vec_dup<mode><mask_name>): Use a single pattern for modes
22475 other than V2DFmode using iptr mode attribute.
22476 (<avx512>_vec_dup<mode><mask_name>): Use iptr mode attribute.
22478 2018-04-11 Alexander Monakov <amonakov@ispras.ru>
22480 PR rtl-optimization/84659
22481 * sel-sched-ir.c (sel_init_pipelining): Invoke cleanup_cfg.
22483 2018-04-11 Jakub Jelinek <jakub@redhat.com>
22486 * dwarf2out.c (skip_loc_list_entry): Don't call size_of_locs if
22488 (output_loc_list): Pass address of a dummy size variable even in the
22489 locview handling loop.
22490 (index_location_lists): Add comment on why skip_loc_list_entry can't
22493 2018-04-11 Thomas Preud'homme <thomas.preudhomme@arm.com>
22496 * config/arm/arm-builtins.c (arm_expand_builtin): Force input operand
22499 2018-04-10 Aaron Sawdey <acsawdey@linux.ibm.com>
22502 * doc/invoke.texi (RS/6000 and PowerPC Options): Document options
22503 -mblock-compare-inline-limit, -mblock-compare-inline-loop-limit,
22504 and -mstring-compare-inline-limit.
22506 2018-04-10 Segher Boessenkool <segher@kernel.crashing.org>
22509 * config/rs6000/rs6000.md (allocate_stack): Put the residual size
22510 for stack clash protection in a register whenever we need it to be in
22513 2018-04-10 Segher Boessenkool <segher@kernel.crashing.org>
22515 * common/config/rs6000/rs6000-common.c (rs6000_option_init_struct):
22516 Enable -fasynchronous-unwind-tables by default if OBJECT_FORMAT_ELF.
22518 2018-04-10 Segher Boessenkool <segher@kernel.crashing.org>
22521 * config/rs6000/rs6000.opt (mtraceback=): Show the allowed values in
22523 (mlong-double-): Ditto.
22524 * config/rs6000/sysv4.opt (msdata=): Ditto.
22525 (mtls-size=): Ditto.
22527 2018-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
22529 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
22530 erroneous entries for
22531 "vector int vec_ldl (int, long int *)", and
22532 "vector unsigned int vec_ldl (int, unsigned long int *)".
22533 Add comments and entries for
22534 "vector bool char vec_ldl (int, bool char *)",
22535 "vector bool short vec_ldl (int, bool short *)",
22536 "vector bool int vec_ldl (int, bool int *)",
22537 "vector bool long long vec_ldl (int, bool long long *)",
22538 "vector pixel vec_ldl (int, pixel *)",
22539 "vector long long vec_ldl (int, long long *)",
22540 "vector unsigned long long vec_ldl (int, unsigned long long *)".
22541 * config/rs6000/rs6000.c (rs6000_init_builtins): Initialize new
22542 type tree bool_long_long_type_node and correct definition of
22543 bool_V2DI_type_node to make reference to this new type tree.
22544 (rs6000_mangle_type): Replace erroneous reference to
22545 bool_long_type_node with bool_long_long_type_node.
22546 * config/rs6000/rs6000.h (enum rs6000_builtin_type_index): Add
22547 comments to emphasize sign distinctions for char and int types and
22548 replace RS6000_BTI_bool_long constant with
22549 RS6000_BTI_bool_long_long constant. Also add comment to restrict
22550 use of RS6000_BTI_pixel.
22551 (bool_long_type_node): Remove this macro definition.
22552 (bool_long_long_type_node): New macro definition
22554 2018-04-10 Jakub Jelinek <jakub@redhat.com>
22556 PR rtl-optimization/85300
22557 * combine.c (subst): Handle subst of CONST_SCALAR_INT_P new_rtx also
22558 into FLOAT and UNSIGNED_FLOAT like ZERO_EXTEND, return a CLOBBER if
22559 simplify_unary_operation fails.
22561 2018-04-10 Martin Liska <mliska@suse.cz>
22563 * gdbhooks.py: Add pretty-printers for varpool_node, symtab_node,
22564 cgraph_edge and ipa_ref.
22566 2018-04-10 Jakub Jelinek <jakub@redhat.com>
22570 * config/i386/sse.md
22571 (<extract_type>_vinsert<shuffletype><extract_suf>_mask): Fix
22572 computation of the VEC_MERGE selector from mask.
22573 (<extract_type>_vinsert<shuffletype><extract_suf>_1<mask_name>):
22574 Fix decoding of the VEC_MERGE selector into mask.
22576 2018-04-10 Richard Sandiford <richard.sandiford@linaro.org>
22578 PR tree-optimization/85286
22579 * tree-vect-data-refs.c (vect_get_smallest_scalar_type):
22581 2018-04-10 Richard Sandiford <richard.sandiford@linaro.org>
22583 * final.c (final_1): Set insn_last_address as well as
22584 insn_current_address.
22586 2018-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22589 * explow.c (emit_stack_probe): Call validize_mem on memory location
22590 before passing it to gen_probe_stack. Create address operand and
22591 legitimize it for the probe_stack_address case.
22593 2018-04-09 Jan Hubicka <jh@suse.cz>
22596 * ipa-devirt.c (rebuild_type_inheritance-hash): New.
22597 * ipa-utils.h (rebuild_type_inheritance-hash): Declare.
22598 * tree.c (free_lang_data_in_type): Fix handling of binfos;
22600 (free_lang_data): Rebuild type inheritance graph.
22602 2018-04-09 Martin Sebor <msebor@redhat.com>
22604 * invoke.texi (-finline-small-functions): Mention other optimization
22606 (-findirect-inlining, -fpartial-inlining): Same.
22607 (-finline-functions-called-once): Same.
22608 (-freorder-blocks-and-partition): Same.
22610 2018-04-09 Jan Hubicka <jh@suse.cz>
22613 * cfgcleanup.c (try_forward_edges): Do not give up on crossing
22614 jumps; choose last target that matches the criteria (i.e.
22615 no partition changes for non-crossing jumps).
22616 * cfgrtl.c (cfg_layout_redirect_edge_and_branch): Add basic
22617 support for redirecting crossing jumps to non-crossing.
22619 2018-04-09 Alexey Brodkin <abrodkin@synopsys.com>
22621 * config/arc/arc.c (arc_expand_prologue): Set stack usage info
22622 also for naked functions.
22624 2018-04-09 Claudiu Zissulescu <claziss@synopsys.com>
22626 * config/arc/arc.md (add_shift): New pattern.
22627 (add_shift2): Likewise.
22628 (sub_shift): Likewise.
22629 (sub_shift_cmp0_noout): Likewise.
22630 (compare_si_ashiftsi): Likewise.
22631 (xbfu_cmp0_noout): New combine pattern.
22632 (xbfu_cmp0"): Likewise.
22633 (movsi_set_cc_insn): Place the predicable variant first.
22634 (commutative_binary_cmp0_noout): Remove clobber.
22635 (commutative_binary_cmp0): New pattern.
22636 (noncommutative_binary_cmp0): Likewise.
22637 (noncommutative_binary_cmp0_noout): Likewise.
22638 (noncommutative_binary_comparison_result_used): Removed.
22639 (rsub_cmp0): New pattern.
22640 (rsub_cmp0_noout): Likewise.
22641 (extzvsi): Changed, keep only meaningful variants.
22642 (SQH, SEZ): New iterators.
22643 (SQH_postfix): New mode attribute.
22644 (SEZ_prefix): New code attribute.
22645 (<SEZ_prefix>xt<SQH_postfix>_cmp0_noout): New instruction pattern.
22646 (<SEZ_prefix>xt<SQH_postfix>_cmp0): Likewise.
22647 * config/arc/predicates.md (cc_set_register): Use CC_REG instead
22648 of numerical value.
22649 (noncommutative_operator): Check the availability of barrel
22652 2018-04-09 Richard Biener <rguenther@suse.de>
22654 PR tree-optimization/85284
22655 * tree-ssa-loop-niter.c (number_of_iterations_exit_assumptions):
22656 Only use the niter constraining form of simple_iv when the exit
22657 is always executed.
22659 2018-04-09 Tom de Vries <tom@codesourcery.com>
22662 * config/nvptx/nvptx.md (define_c_enum "unspecv"): Add UNSPECV_MEMBAR.
22663 (define_expand "*memory_barrier"): New define_expand.
22664 (define_insn "memory_barrier"): New insn.
22666 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
22668 PR rtl-optimization/80463
22669 PR rtl-optimization/83972
22670 PR rtl-optimization/83480
22672 * sel-sched-ir.c (has_dependence_note_mem_dep): Take into account the
22673 correct producer for the insn.
22674 (tidy_control_flow): Fixup seqnos in case of debug insns.
22676 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
22678 PR rtl-optimization/83913
22680 * sel-sched-ir.c (merge_expr_data): Choose the middle between two
22681 different sched-times when merging exprs.
22683 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
22685 PR rtl-optimization/83962
22687 * sel-sched-ir.c (tidy_control_flow): Correct the order in which we call
22688 tidy_fallthru_edge and tidy_control_flow.
22690 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
22692 PR rtl-optimization/83530
22694 * sel-sched.c (force_next_insn): New global variable.
22695 (remove_insn_for_debug): When force_next_insn is true, also leave only
22696 next insn in the ready list.
22697 (sel_sched_region): When the region wasn't scheduled, make another pass
22698 over it with force_next_insn set to 1.
22700 2018-04-08 Monk Chiang <sh.chiang04@gmail.com>
22702 * config.gcc (nds32le-*-*, nds32be-*-*): Add nds32/nds32_intrinsic.h
22704 * config/nds32/constants.md (unspec_volatile_element): Add enum values
22705 for interrupt control.
22706 * config/nds32/nds32-intrinsic.c: Implementation of intrinsic
22707 functions for interrupt control.
22708 * config/nds32/nds32-intrinsic.md: Likewise.
22709 * config/nds32/nds32_intrinsic.h: Likewise.
22710 * config/nds32/nds32.h (nds32_builtins): Likewise.
22712 2018-04-08 Chung-Ju Wu <jasonwucj@gmail.com>
22714 * config/nds32/nds32.c (nds32_init_machine_status,
22715 nds32_legitimate_index_p, nds32_legitimate_address_p): Consider
22716 strict_aligned_p field.
22717 (nds32_expand_to_rtl_hook): New function.
22718 (TARGET_EXPAND_TO_RTL_HOOK): Define.
22719 * config/nds32/nds32.h (machine_function): Add strict_aligned_p field.
22721 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
22722 Chung-Ju Wu <jasonwucj@gmail.com>
22724 * config.gcc (nds32*-*-*): Check that n7 is valid to --with-cpu.
22725 * config/nds32/nds32-n7.md: New file.
22726 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N7.
22727 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n7
22729 * config/nds32/nds32-protos.h: More declarations for n7 pipeline.
22730 * config/nds32/nds32.md (pipeline_model): Add n7.
22731 * config/nds32/nds32.opt (mcpu): Support n7 pipeline cpus.
22732 * config/nds32/pipelines.md: Include n7 settings.
22734 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
22735 Chung-Ju Wu <jasonwucj@gmail.com>
22737 * config.gcc (nds32*-*-*): Check that e8 is valid to --with-cpu.
22738 * config/nds32/nds32-e8.md: New file.
22739 * config/nds32/nds32-opts.h (nds32-cpu_type): Add CPU_E8.
22740 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for e8
22742 * config/nds32/nds32-protos.h: More declarations for e8 pipeline.
22743 * config/nds32/nds32.md (pipeline_model): Add e8.
22744 * config/nds32/nds32.opt (mcpu): Support e8 pipeline cpus.
22745 * config/nds32/pipelines.md: Include e8 settings.
22747 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
22748 Chung-Ju Wu <jasonwucj@gmail.com>
22750 * config.gcc (nds32*-*-*): Check that n6/n8/s8 are valid to --with-cpu.
22751 * config/nds32/nds32-n8.md: New file.
22752 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N6 and CPU_N8.
22753 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n8
22755 * config/nds32/nds32-protos.h: More declarations for n8 pipeline.
22756 * config/nds32/nds32-utils.c: More implementations for n8 pipeline.
22757 * config/nds32/nds32.md (pipeline_model): Add n8.
22758 * config/nds32/nds32.opt (mcpu): Support n8 pipeline cpus.
22759 * config/nds32/pipelines.md: Include n8 settings.
22761 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
22762 Chung-Ju Wu <jasonwucj@gmail.com>
22764 * config.gcc (nds32*): Add nds32-utils.o into extra_objs.
22765 * config/nds32/nds32-n9-2r1w.md: New file.
22766 * config/nds32/nds32-n9-3r2w.md: New file.
22767 * config/nds32/nds32-opts.h (nds32_cpu_type, nds32_mul_type,
22768 nds32_register_ports): New or modify for cpu n9.
22769 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n9
22771 * config/nds32/nds32-protos.h: More declarations for n9 pipeline.
22772 * config/nds32/nds32-utils.c: New file.
22773 * config/nds32/nds32.h (TARGET_PIPELINE_N9, TARGET_PIPELINE_SIMPLE,
22774 TARGET_MUL_SLOW): Define.
22775 * config/nds32/nds32.md (pipeline_model): New attribute.
22776 * config/nds32/nds32.opt (mcpu, mconfig-mul, mconfig-register-ports):
22777 New options that support cpu n9.
22778 * config/nds32/pipelines.md: Include n9 settings.
22779 * config/nds32/t-nds32 (nds32-utils.o): Add dependency.
22781 2018-04-08 Chung-Ju Wu <jasonwucj@gmail.com>
22783 * config/nds32/nds32-md-auxiliary.c (output_cond_branch): Output align
22784 information if necessary.
22785 (output_cond_branch_compare_zero): Likewise.
22786 * config/nds32/nds32.c (nds32_adjust_insn_length): Consider align case.
22787 (nds32_target_alignment): Refine for alignment.
22788 * config/nds32/nds32.h (NDS32_ALIGN_P): Define.
22789 (FUNCTION_BOUNDARY): Modify.
22790 * config/nds32/nds32.md (call_internal, call_value_internal): Consider
22792 * config/nds32/nds32.opt (malways-align, malign-functions): New.
22794 2018-04-08 Monk Chiang <sh.chiang04@gmail.com>
22796 * config/nds32/constants.md (unspec_volatile_element): Add values for
22797 TLB operation and data prefetch.
22798 * config/nds32/nds32-intrinsic.c: Implementation of intrinsic
22799 functions for TLB operation and data prefetch.
22800 * config/nds32/nds32-intrinsic.md: Likewise.
22801 * config/nds32/nds32_intrinsic.h: Likewise.
22802 * config/nds32/nds32.c (nds32_dpref_names): Likewise.
22803 (nds32_print_operand): Likewise.
22804 * config/nds32/nds32.h (nds32_builtins): Likewise.
22806 2018-04-07 Thomas Koenig <tkoenig@gcc.gnu.org>
22807 Andrew Pinski <pinsika@gcc.gnu.org>
22809 PR middle-end/82976
22810 * match.pd: Use constant_boolean_node of correct type instead of
22811 boolean_true_node or boolean_false_node for simplifying
22812 pointer comparisons to zero.
22814 2018-04-07 Jakub Jelinek <jakub@redhat.com>
22816 PR tree-optimization/80021
22817 * tree.c (verify_type_variant): Make error call in verify_variant_match
22818 translatable and remove final full stop.
22820 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
22822 * config/nds32/constants.md (unspec_volatile_element): Add
22823 UNSPEC_VOLATILE_EH_RETURN.
22824 * config/nds32/nds32-md-auxiliary.c (nds32_output_stack_push,
22825 nds32_output_stack_pop): Support dwarf exception handling process.
22826 * config/nds32/nds32-protos.h (nds32_dynamic_chain_address): Declare.
22827 * config/nds32/nds32.c (nds32_init_machine_status): Support dwarf
22828 exception handling process.
22829 (nds32_compute_stack_frame): Likewise.
22830 (nds32_return_addr_rtx): Likewise.
22831 (nds32_initial_elimination_offset): Likewise.
22832 (nds32_expand_prologue): Likewise.
22833 (nds32_expand_epilogue): Likewise.
22834 (nds32_dynamic_chain_address): New function.
22835 * config/nds32/nds32.h (machine_function): Add fields for dwarf
22836 exception handling.
22837 (DYNAMIC_CHAIN_ADDRESS): Define.
22838 (EH_RETURN_DATA_REGNO): Define.
22839 (EH_RETURN_STACKADJ_RTX): Define.
22840 * config/nds32/nds32.md (eh_return, nds32_eh_return): Implement
22841 patterns for dwarf exception handling.
22843 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
22845 * config/nds32/nds32.h: Clean up obsolete macros.
22847 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
22849 * config/nds32/constants.md (unspec_element, unspec_volatile_element):
22850 Add enum values for particular instructions.
22851 * config/nds32/nds32-intrinsic.c: Implementation of expanding
22852 particular intrinsic functions.
22853 * config/nds32/nds32-intrinsic.md: Likewise.
22854 * config/nds32/nds32_intrinsic.h: Likewise.
22855 * config/nds32/nds32.h (nds32_builtins): Likewise.
22856 * config/nds32/nds32.md (type): Add pbsad and pbsada.
22857 (btst, ave): New patterns for particular instructions.
22859 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
22861 * config/nds32/constants.md (unspec_element, unspec_volatile_element):
22862 Add enum values for atomic load/store and memory sync.
22863 * config/nds32/nds32-intrinsic.c: Implementation for atomic load/store
22865 * config/nds32/nds32-intrinsic.md: Likewise.
22866 * config/nds32/nds32_intrinsic.h: Likewise.
22867 * config/nds32/nds32.h (nds32_builtins): Likewise.
22869 2018-04-07 Jakub Jelinek <jakub@redhat.com>
22871 PR tree-optimization/85257
22872 * fold-const.c (native_encode_vector): If not all elts could fit
22873 and off is -1, return 0 rather than offset.
22874 * tree-ssa-sccvn.c (vn_reference_lookup_3): Pass
22875 (offseti - offset2) / BITS_PER_UNIT as 4th argument to
22876 native_encode_expr. Verify len * BITS_PER_UNIT >= maxsizei. Don't
22877 adjust buffer in native_interpret_expr call.
22879 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
22881 * config/nds32/constants.md (unspec_volatile_element): Add cache
22882 control enum values.
22883 * config/nds32/nds32-intrinsic.c: Add cache control expand functions.
22884 * config/nds32/nds32-intrinsic.md: Add cache control patterns.
22885 * config/nds32/nds32.c (nds32_cctl_names): New.
22886 (nds32_print_operand): Handle cache control register names.
22887 * config/nds32/nds32.h (nds32_builtins): New enum values.
22888 * config/nds32/nds32_intrinsic.h: Add cache control enum types and
22890 * config/nds32/nds32.md (type): Add mmu.
22891 * config/nds32/pipelines.md (simple_insn): Add mmu.
22893 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
22895 * config/nds32/nds32.md (type): Remove call.
22896 * config/nds32/pipelines.md (simple_insn): Likewise.
22898 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
22900 * config/nds32/constants.md (unspec_volatile_element): Add
22901 UNSPEC_VOLATILE_FMFCSR, UNSPEC_VOLATILE_FMTCSR and
22902 UNSPEC_VOLATILE_FMFCFG.
22903 * config/nds32/nds32-intrinsic.c (bdesc_noarg): New builtin
22904 description for fmfcfg and fmfcsr.
22905 (bdesc_1arg): Add fmtcsr.
22906 (bdesc_2arg): Add fcpynss, fcpyss, fcpynsd and fcpysd.
22907 (nds32_expand_builtin_impl): Deal with FPU intrinsic functions.
22908 * config/nds32/nds32-intrinsic.md (unspec_fcpynsd, unspec_fcpysd,
22909 unspec_fcpynss, unspec_fcpysd, unspec_fcpyss, unspec_fmfcsr,
22910 unspec_fmfcfg): New patterns.
22911 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_FMFCFG,
22912 NDS32_BUILTIN_FMFCSR, NDS32_BUILTIN_FMTCSR, NDS32_BUILTIN_FCPYNSS,
22913 NDS32_BUILTIN_FCPYSS,NDS32_BUILTIN_FCPYNSD and NDS32_BUILTIN_FCPYSD.
22914 * config/nds32/nds32_intrinsic.h (__nds32__fcpynsd, __nds32__fcpynss,
22915 __nds32__fcpysd, __nds32__fcpyss, __nds32__fmfcsr, __nds32__fmtcsr,
22916 __nds32__fmfcfg): Define.
22918 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
22920 * config/nds32/nds32.c (nds32_intrinsic_register_names): Add more
22921 intrinsic register names.
22922 * config/nds32/nds32_intrinsic.h (nds32_intrinsic_registers): Add more
22923 intrinsic register enum values and macros.
22925 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
22927 * config/nds32/nds32.c (nds32_legitimate_index_p): Modify condition
22928 for load/store addressing form.
22929 (nds32_print_operand_address): Likewise.
22931 2018-04-06 Eric Botcazou <ebotcazou@adacore.com>
22934 * config/sparc/sparc.c (sparc_expand_move): Deal with symbolic operands
22935 based on LABEL_REF. Remove useless assertion.
22936 (pic_address_needs_scratch): Fix formatting.
22937 (sparc_legitimize_pic_address): Minor tweaks.
22938 (sparc_delegitimize_address): Adjust assertion accordingly.
22939 * config/sparc/sparc.md (movsi_pic_label_ref): Change label_ref_operand
22940 into symbolic_operand.
22941 (movsi_high_pic_label_ref): Likewise.
22942 (movsi_lo_sum_pic_label_ref): Likewise.
22943 (movdi_pic_label_ref): Likewise.
22944 (movdi_high_pic_label_ref): Likewise.
22945 (movdi_lo_sum_pic_label_ref): Likewise.
22947 2018-04-06 Amaan Cheval <amaan.cheval@gmail.com>
22949 * config.gcc (x86_64-*-rtems*): Add rtems.h to tm_file for
22950 custom LIB_SPEC setup.
22952 2018-04-06 Ruslan Bukin <br@bsdpad.com>
22953 Kito Cheng <kito.cheng@gmail.com>
22955 * config.gcc (riscv*-*-freebsd*): Add RISC-V FreeBSD support.
22956 * config/riscv/freebsd.h: New.
22958 2018-04-06 Chung-Ju Wu <jasonwucj@gmail.com>
22960 * config/nds32/nds32.c (nds32_adjust_insn_length): Refine.
22961 * config/nds32/nds32.h (ADJUST_INSN_LENGTH): Change the location in
22964 2018-04-06 Chung-Ju Wu <jasonwucj@gmail.com>
22965 Kito Cheng <kito.cheng@gmail.com>
22967 * config/nds32/nds32-md-auxiliary.c (nds32_output_return,
22968 nds32_output_call, nds32_symbol_binds_local_p): New functions.
22969 * config/nds32/nds32-protos.h (nds32_output_call,
22970 nds32_output_return): Declare.
22971 * config/nds32/nds32.md: Refine all the call and return patterns.
22973 2018-04-06 Jakub Jelinek <jakub@redhat.com>
22976 * dwarf2out.c (rtl_for_decl_init): For STRING_CST initializer only
22977 build CONST_STRING if TYPE_MAX_VALUE is non-NULL and is INTEGER_CST.
22979 PR rtl-optimization/84872
22980 * cfgloopmanip.c (create_preheader): Use make_forwarder_block even if
22981 nentry == 1 when CP_FALLTHRU_PREHEADERS and single_entry is
22982 EDGE_CROSSING edge.
22984 2018-04-06 Tamar Christina <tamar.christina@arm.com>
22986 * expr.c (copy_blkmode_to_reg): Revert 254862.
22987 * doc/sourcebuild.texi (word_mode_no_slow_unalign): Likewise.
22989 2018-04-06 Richard Biener <rguenther@suse.de>
22991 PR middle-end/85244
22992 * tree-dfa.c (get_ref_base_and_extent): Reset seen_variable_array_ref
22993 after seeing a component reference with an adjacent field. Treat
22994 refs to arrays at struct end of external decls similar to
22995 refs to unconstrained commons.
22997 2018-04-06 Jakub Jelinek <jakub@redhat.com>
23000 * fold-const.c (twoval_comparison_p): Remove SAVE_P argument and don't
23001 look through SAVE_EXPRs with non-side-effects argument. Adjust
23003 (fold_comparison): Adjust twoval_comparison_p caller, don't handle
23006 2018-04-06 Richard Biener <rguenther@suse.de>
23008 PR middle-end/85180
23009 * alias.c (find_base_term): New wrapper around find_base_term
23010 unwinding CSELIB_VAL_PTR changes.
23011 (find_base_term): Do not restore CSELIB_VAL_PTR during the
23014 2018-04-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23016 * config/s390/s390.c (s390_z10_optimize_cmp): Expand dedicated NOP
23018 * config/s390/s390.md (UNSPECV_NOP_LR_0, UNSPECV_NOP_LR_1): New
23019 constant definitions.
23020 ("nop"): lr 0,0 -> nopr r0
23021 ("nop_lr0", "nop_lr1"): New insn definitions.
23023 2018-04-06 Chung-Ju Wu <jasonwucj@gmail.com>
23025 * config/nds32/nds32.md (*stack_push, *stack_pop): Use
23026 NDS32_V3PUSH_AVAILABLE_P macro.
23028 2018-04-06 Monk Chiang <sh.chiang04@gmail.com>
23029 Chung-Ju Wu <jasonwucj@gmail.com>
23031 * config.gcc (nds32*-*-*): Add v2j v3f v3s checking.
23032 (nds32*-*-*): Add float and fpu_config into supported_defaults.
23033 * common/config/nds32/nds32-common.c (TARGET_DEFAULT_TARGET_FLAGS):
23034 Include TARGET_DEFAULT_FPU_ISA and TARGET_DEFAULT_FPU_FMA.
23035 * config/nds32/constants.md (unspec_element): Add UNSPEC_COPYSIGN,
23036 UNSPEC_FCPYNSD, UNSPEC_FCPYNSS, UNSPEC_FCPYSD and UNSPEC_FCPYSS.
23037 * config/nds32/constraints.md: New constraints and checking for hard
23038 float configuration.
23039 * config/nds32/iterators.md: New mode iterator and attribute for hard
23040 float configuration.
23041 * config/nds32/nds32-doubleword.md: Use hard float alternatives and
23043 * config/nds32/nds32-fpu.md: New file.
23044 * config/nds32/nds32-md-auxiliary.c: New functions and checkings to
23045 deal with hard float code generation.
23046 * config/nds32/nds32-opts.h (nds32_arch_type): Add ARCH_V3F and
23048 (abi_type, float_reg_number): New enum type.
23049 * config/nds32/nds32-predicates.c: New predicates for hard float.
23050 * config/nds32/nds32-protos.h: Declare functions for hard float.
23051 * config/nds32/nds32.c: Implementation for hard float configuration.
23052 * config/nds32/nds32.h: Definitions for hard float configuration.
23053 * config/nds32/nds32.md: Include hard float machine description and
23054 modify patterns for hard float configuration.
23055 * config/nds32/nds32.opt: New options for hard float configuration.
23056 * config/nds32/predicates.md: New predicates for hard float
23059 2018-04-06 Kuan-Lin Chen <kuanlinchentw@gmail.com>
23061 * common/config/nds32/nds32-common.c
23062 (nds32_option_optimization_table): Enable -mreleax-hint by default.
23064 2018-04-05 Jakub Jelinek <jakub@redhat.com>
23066 PR middle-end/85195
23067 * match.pd (BIT_FIELD_REF CONSTRUCTOR@0 @1 @2): Use view_convert around
23068 CONSTRUCTOR_ELT (ctor, ...)->value.
23070 2018-04-05 Uros Bizjak <ubizjak@gmail.com>
23073 * config/i386/i386.md (define_attr "memory"): Handle rotate1 type.
23075 2018-04-05 Tom de Vries <tom@codesourcery.com>
23078 * config/nvptx/nvptx.c (nvptx_single): Fix neutering of bb with only
23081 2018-04-05 Shiva Chen <shiva0217@gmail.com>
23082 Kito Cheng <kito.cheng@gmail.com>
23084 * config/nds32/constraints.md (U33): Fine-tune checking condition.
23085 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Ditto.
23086 * config/nds32/nds32.h (nds32_16bit_address_type): Add
23087 ADDRESS_POST_MODIFY_LO_REG_IMM3U.
23089 2018-04-05 Shiva Chen <shiva0217@gmail.com>
23090 Kito Cheng <kito.cheng@gmail.com>
23092 * config/nds32/constraints.md (Ufe): New memory constraint.
23093 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format,
23094 nds32_output_16bit_load): Consider r8 register for lwi45.fe format.
23095 * config/nds32/nds32.c (nds32_print_operand): Output lwi45.fe
23097 * config/nds32/nds32.h (nds32_16bit_address_type): Add ADDRESS_R8_IMM7U.
23098 * config/nds32/nds32.md (*mov<mode>): Adjust pattern.
23100 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
23102 * config/nds32/nds32.md: Use optimize_size in the condition for
23103 alu-shift instructions.
23105 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
23107 * config/nds32/nds32.md (divsi4, udivsi4): New patterns.
23109 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
23111 * config/nds32/nds32.md (negsi2): Refine pattern.
23113 2018-04-05 Kito Cheng <kito.cheng@gmail.com>
23114 Chung-Ju Wu <jasonwucj@gmail.com>
23116 * config/nds32/iterators.md (shift_rotate): New code iterator.
23117 (shift): New code attribute.
23118 * config/nds32/nds32-md-auxiliary.c (nds32_expand_constant): New.
23119 * config/nds32/nds32-protos.h (nds32_expand_constant): Declare.
23120 * config/nds32/nds32.c (nds32_print_operand): Deal with more cases.
23121 * config/nds32/nds32.md (addsi3, *add_srli): Refine implementation for
23122 bit-wise operations.
23123 (andsi3, *andsi3): Ditto.
23124 (iorsi3, *iorsi3, *or_slli, *or_srli): Ditto.
23125 (xorsi3, *xorsi3, *xor_slli, *xor_srli): Ditto.
23126 (<shift>si3, *ashlsi3, *ashrsi3, *lshrsi3, *rotrsi3): Ditto.
23127 * config/nds32/predicates.md (nds32_rimm5u_operand, nds32_and_operand,
23128 nds32_ior_operand, nds32_xor_operand): New predicates.
23130 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
23132 * config/nds32/nds32.md (add<mode>3, sub<mode>3): Rename to ...
23133 (addsi3, subsi3): ... this.
23135 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
23137 * config/nds32/nds32.md (*sub_srli, *and_slli): Fine-tune predicator.
23139 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
23141 * config/nds32/nds32.md: Adjust indention.
23143 2018-04-05 Kito Cheng <kito.cheng@gmail.com>
23145 * config/nds32/nds32.md (feature): New attribute.
23147 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
23149 * config/nds32/nds32.md (subtype): New attribute.
23151 2018-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
23154 * config/arm/arm-builtins.c (arm_expand_builtin): Change
23155 expansion to perform a bitwise AND of the argument followed by a
23156 boolean negation of the result.
23158 2018-04-04 Peter Bergner <bergner@vnet.ibm.com>
23160 PR rtl-optimization/84878
23161 * ddg.c (add_cross_iteration_register_deps): Use DF_REF_BB to determine
23162 the basic block. Assert the use reference is not artificial and that
23163 it has an associated insn.
23165 2018-04-04 Michael Matz <matz@suse.de>
23167 * builtins.c (compute_objsize): Pass correct operand
23168 to array_at_struct_end_p.
23170 2018-04-04 Richard Biener <rguenther@suse.de>
23173 * dwarf2out.c (dwarf2out_register_external_die): Peel namespaces
23174 from contexts for DINFO_LEVEL_TERSE and below.
23176 2018-04-04 Kito Cheng <kito.cheng@gmail.com>
23178 * config/nds32/nds32-doubleword.md (move_<mode>): Require
23179 resiter_operand condition.
23180 * config/nds32/nds32.md (*move<mode>): Ditto.
23182 2018-04-04 Kito Cheng <kito.cheng@gmail.com>
23183 Monk Chiang <sh.chiang04@gmail.com>
23185 * config/nds32/nds32.md (movmisalign<mode>): New pattern.
23187 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
23189 * config/nds32/nds32.md (movqi, movhi): Merge into mov<mode>.
23191 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
23192 Kito Cheng <kito.cheng@gmail.com>
23194 * config/nds32/nds32-md-auxiliary.c (nds32_inverse_cond_code,
23195 nds32_cond_code_str, output_cond_branch,
23196 output_cond_branch_compare_zero, nds32_expand_cbranch,
23197 nds32_expand_cstore, nds32_expand_movcc,
23198 nds32_output_cbranchsi4_equality_zero,
23199 nds32_output_cbranchsi4_equality_reg,
23200 nds32_output_cbranchsi4_equality_reg_or_const_int,
23201 nds32_output_cbranchsi4_greater_less_zero: New functions.
23202 * config/nds32/nds32-protos.h (nds32_expand_cbranch,
23203 nds32_expand_cstore, nds32_expand_movcc,
23204 nds32_output_cbranchsi4_equality_zero,
23205 nds32_output_cbranchsi4_equality_reg,
23206 nds32_output_cbranchsi4_equality_reg_or_const_int,
23207 nds32_output_cbranchsi4_greater_less_zero): Declare.
23208 * config/nds32/predicates.md (nds32_movecc_comparison_operator,
23209 nds32_rimm11s_operand): New predicates.
23210 * config/nds32/nds32.h (nds32_expand_result_type): New enum type.
23211 * config/nds32/nds32.md: Rewrite all the branch and conditional move
23214 2018-04-04 Kito Cheng <kito.cheng@gmail.com>
23216 * config/nds32/nds32-doubleword.md: Refine all the instruction type.
23217 * config/nds32/nds32.md: Ditto.
23218 * config/nds32/pipelines.md: Ditto.
23220 2018-04-04 Richard Biener <rguenther@suse.de>
23222 PR tree-optimization/85168
23223 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address): Avoid
23224 propagating abnormals.
23226 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
23228 * config/nds32/nds32.md (enabled): Use yes/no for this attribute.
23230 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
23231 Kito Cheng <kito.cheng@gmail.com>
23233 * config/nds32/nds32-md-auxiliary.c (nds32_long_call_p): New function.
23234 * config/nds32/nds32-protos.h (nds32_long_call_p): Declare.
23235 * config/nds32/nds32.c (nds32_function_ok_for_sibcall): New function.
23236 (TARGET_FUNCTION_OK_FOR_SIBCALL): Define.
23237 * config/nds32/nds32.md (sibcall_internal): New.
23238 (sibcall_register): Remove.
23239 (sibcall_immediate): Remove.
23240 (sibcall_value_internal): New.
23241 (sibcall_value_register): Remove.
23242 (sibcall_value_immediate): Remove.
23243 * config/nds32/predicates.md (nds32_general_register_operand): New.
23244 (nds32_call_address_operand): New.
23246 2018-04-03 Jakub Jelinek <jakub@redhat.com>
23248 PR rtl-optimization/85167
23249 * shrink-wrap.c (move_insn_for_shrink_wrap): Don't set bb_uses and
23250 bb_defs if *split_p, instead preinitialize it to NULL.
23252 PR tree-optimization/85156
23253 * builtins.c (fold_builtin_expect): Use save_expr on arg1 to avoid
23254 evaluating the argument multiple times.
23256 2018-04-03 Bill Schmidt <wschmidt@linux.ibm.com>
23258 * config/rs6000/emmintrin.h (_mm_cvtpd_epi32): Use __vector rather
23260 (_mm_cvtpd_ps): Likewise.
23261 (_mm_cvttpd_epi32): Likewise.
23262 * config/rs6000/mmintrin.h (_mm_unpacklo_pi8): Likewise.
23263 * config/rs6000/xmmintrin.h: For strict-ANSI C++ or C11, undefine
23264 vector, pixel, and bool following altivec.h include.
23266 2018-04-03 Martin Sebor <msebor@redhat.com>
23268 * doc/extend.texi (Common Function Attributes): Clarify.
23269 (const attribute): Likewise.
23270 (pure attribute): Likewise.
23272 2018-04-03 Jakub Jelinek <jakub@redhat.com>
23275 * config/i386/i386.c (ix86_expand_vector_set): Use
23276 HOST_WIDE_INT_1U << elt instead of 1 << elt. Formatting fix.
23278 2018-04-03 Uros Bizjak <ubizjak@gmail.com>
23280 * config/i386/i386.c (emit_i387_cw_initialization): Always use logic
23281 instructions when changing rounding bits to preserve precision bits
23282 in the x87 control word.
23284 2018-04-03 Martin Liska <mliska@suse.cz>
23286 PR tree-optimization/82491
23287 * rtl.h (strip_offset_and_add): Replace += suboffset with
23288 poly_uint64 () + suboffset.
23290 2018-03-29 Martin Liska <mliska@suse.cz>
23291 Martin Jambor <mjambor@suse.cz>
23294 * ipa-cp.c (propagate_bits_across_jump_function): Bail out if
23295 param_type is not an integral or pointer type.
23297 2018-04-03 Richard Biener <rguenther@suse.de>
23299 * sese.h (recompute_all_dominators): Remove.
23301 2018-04-02 Martin Sebor <msebor@redhat.com>
23303 * doc/invoke.texi (-Wrestrict): Fix typos.
23305 2018-04-02 Jim Wilson <jimw@sifive.com>
23307 * config/riscv/riscv.h (SHIFT_COUNT_TRUNCATED): Set to zero.
23308 * config/riscv/riscv.md (<optab>si3): Use QImode shift count.
23309 (<optab>di3, <optab>si3_extend): Likewise.
23310 (<optab>si3_mask, <optab>si3_mask_1): New.
23311 (<optab>di3_mask, <optab>di3_mask_1): New.
23312 (<optab>si3_extend_mask, <optab>si3_extend_mask_1): New.
23313 (lshrsi3_zero_extend_1): Use VOIDmode shift count.
23314 * config/riscv/sync.md (atomic_test_and_set): Emit QImode shift count.
23316 2018-04-02 Gerald Pfeifer <gerald@pfeifer.com>
23318 * doc/cpp.texi (Variadic Macros): Fix line continuation in an
23321 2018-04-02 Chung-Ju Wu <jasonwucj@gmail.com>
23323 * config/nds32/nds32.c (TARGET_CANONICALIZE_COMPARISON): Define.
23324 (nds32_canonicalize_comparison): New function.
23326 2018-04-01 Chung-Ju Wu <jasonwucj@gmail.com>
23327 Kito Cheng <kito.cheng@gmail.com>
23328 Kuan-Lin Chen <kuanlinchentw@gmail.com>
23330 * config.gcc (nds32): Add nds32-relax-opt.o into extra_objs.
23331 * config/nds32/constants.md (unspec_volatile_element): Add
23332 UNSPEC_VOLATILE_RELAX_GROUP.
23333 * config/nds32/nds32-relax-opt.c: New file.
23334 * config/nds32/nds32-predicates.c
23335 (nds32_symbol_load_store_p): New function.
23336 * config/nds32/nds32-protos.h
23337 (nds32_symbol_load_store_p): Declare function.
23338 (make_pass_nds32_relax_opt): Declare new rtl pass function.
23339 * config/nds32/nds32.c
23340 (nds32_register_pass): New function to register pass.
23341 (nds32_register_passes): New function to register passes.
23342 * config/nds32/nds32.md (relax_group): New pattern.
23343 * config/nds32/nds32.opt (mrelax-hint): New option.
23344 * config/nds32/t-nds32 (nds32-relax-opt.o): New dependency.
23346 2018-04-01 Kito Cheng <kito.cheng@gmail.com>
23348 * config/nds32/t-nds32: Modify files dependency.
23350 2018-04-01 Chung-Ju Wu <jasonwucj@gmail.com>
23352 * config/nds32/nds32.h (FUNCTION_PROFILER): Output newline character.
23353 (PROFILE_HOOK): Define its implementation.
23355 2018-04-01 Chung-Ju Wu <jasonwucj@gmail.com>
23357 * config/nds32/nds32.h (WCHAR_TYPE, WCHAR_TYPE_SIZE): Use unsigned int
23358 type and 32-bit size.
23360 2018-04-01 Jakub Jelinek <jakub@redhat.com>
23362 PR middle-end/85090
23363 * config/i386/sse.md (V): Add V64QI and V32HI for TARGET_AVX512F.
23364 (V_128_256): New mode iterator.
23365 (*avx512dq_vextract<shuffletype>64x2_1 splitter): New define_split.
23366 (*avx512f_vextract<shuffletype>32x4_1 splitter): Likewise.
23367 (xop_pcmov_<mode><avxsizesuffix>): Use V_128_256 mode iterator instead
23369 * config/i386/i386.c (ix86_expand_vector_set): Improve V32HImode and
23370 V64QImode expansion for !TARGET_AVX512BW && TARGET_AVX512F.
23372 2018-03-31 Segher Boessenkool <segher@kernel.crashing.org>
23375 * config/rs6000/xmmintrin.h (_mm_set_ps, _mm_max_ps): Handle (quiet)
23376 NaN inputs correctly.
23378 2018-03-30 Peter Bergner <bergner@vnet.ibm.com>
23381 * config/rs6000/vsx.md (??r): New mode attribute.
23382 (*vsx_mov<mode>_64bit): Use it.
23383 (*vsx_mov<mode>_32bit): Likewise.
23385 2018-03-30 Martin Sebor <msebor@redhat.com>
23387 PR tree-optimization/84818
23388 * builtins.c (check_access): Use warning_n.
23390 2018-03-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23393 * config/rs6000/rs6000-string.c (expand_compare_loop): Fix redundant
23395 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Fix redundant
23398 2018-03-30 Julia Koval <julia.koval@intel.com>
23401 * x86-tune.def (movx, partial_reg_dependency): Enable for
23404 2018-03-29 Vladimir Makarov <vmakarov@redhat.com>
23406 PR inline-asm/84985
23407 * lra-constraints.c (process_alt_operands): Move setting
23408 this_alternative_matches below.
23410 2018-03-29 Martin Liska <mliska@suse.cz>
23413 * doc/invoke.texi: Document how LTO works with debug info.
23414 Describe auto-load support of binutils. Mention 'x86-64'
23415 as valid option value of -march option.
23417 2018-03-29 Jakub Jelinek <jakub@redhat.com>
23419 * config/i386/sse.md (<avx512>_blendm<mode>): Use <sseintprefix>.
23422 * fold-const.c (operand_equal_p): Handle DEBUG_BEGIN_STMT.
23423 For STATEMENT_LIST, pass down OEP_LEXICOGRAPHIC and maybe
23424 OEP_NO_HASH_CHECK for recursive call, to avoid exponential
23427 2018-03-28 Peter Bergner <bergner@vnet.ibm.com>
23430 * config/rs6000/rs6000.h (RS6000_BTM_POWERPC64): New define.
23431 (RS6000_BTM_COMMON): Add RS6000_BTM_POWERPC64.
23432 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add support
23433 for RS6000_BTM_POWERPC64.
23434 (rs6000_invalid_builtin): Add handling for RS6000_BTM_POWERPC64
23435 (rs6000_builtin_mask_names): Add RS6000_BTM_POWERPC64.
23436 * config/rs6000/rs6000-builtin.def (BU_P7_POWERPC64_MISC_2): New macro
23439 (DIVDEU): Likewise.
23441 2018-03-28 Carl Love <cel@us.ibm.com>
23444 2017-09-27 Carl Love <cel@us.ibm.com>
23446 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
23447 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
23448 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
23451 2018-03-28 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23453 * config/rs6000/xmmintrin.h (_mm_max_pi16): Use __vector __bool
23454 instead of __vector bool.
23455 (_mm_max_pu8): Likewise.
23456 (_mm_min_pi16): Likewise.
23458 2018-03-28 Peter Bergner <bergner@vnet.ibm.com>
23461 * config/rs6000/rs6000-builtin.def (DIVWEO): Delete macro expansion.
23462 (DIVWEUO): Likewise.
23463 (DIVDEO): Likewise.
23464 (DIVDEUO): Likewise.
23465 * config/rs6000/rs6000.c (builtin_function_type): Remove support for
23466 DIVWEUO and DIVDEUO.
23467 * config/rs6000/rs6000.md (UNSPEC_DIVEO, UNSPEC_DIVEUO): Delete unspecs.
23468 (UNSPEC_DIV_EXTEND): Remove deleted unspecs.
23469 (div_extend): Likewise.
23470 * doc/extend.texi (__builtin_divweo): Remove documentation for deleted
23472 (__builtin_divweuo): Likewise.
23473 (__builtin_divdeo): Likewise.
23474 (__builtin_divdeuo): Likewise.
23476 2018-03-28 Jakub Jelinek <jakub@redhat.com>
23479 * config/i386/i386.md (*add<mode>3_carry_0, *addsi3_carry_zext_0,
23480 *sub<mode>3_carry_0, *subsi3_carry_zext_0): New patterns.
23482 PR tree-optimization/82004
23483 * gimple-match-head.c (optimize_pow_to_exp): New function.
23484 * match.pd (pow(C,x) -> exp(log(C)*x)): Wrap with #if GIMPLE.
23485 Don't fold to exp if optimize_pow_to_exp is false.
23487 2018-03-28 Martin Liska <mliska@suse.cz>
23490 * calls.c (initialize_argument_information): Fix trailing space.
23491 * common.opt: Fix typo and provide better explanation for
23492 -fsanitize-coverage option.
23493 * config/i386/i386.opt: Fix typo.
23495 2018-03-28 Jakub Jelinek <jakub@redhat.com>
23496 Martin Liska <mliska@suse.cz>
23499 * gimplify.c (asan_poison_variable): Don't do the check for
23500 gimplify_omp_ctxp here.
23501 (gimplify_decl_expr): Do it here.
23502 (gimplify_target_expr): Likewise.
23504 2018-03-28 Martin Liska <mliska@suse.cz>
23507 * config/i386/i386.c (ix86_function_arg_advance): Do not call
23508 chkp_type_bounds_count if MPX is not enabled.
23510 2018-03-27 Chung-Ju Wu <jasonwucj@gmail.com>
23512 * config/nds32/nds32.h (BRANCH_COST): Adjust cost.
23514 2018-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
23517 * config/rs6000/rs6000.c (create_complex_muldiv): New helper
23518 function to create the function decl for complex long double
23519 multiply and divide for -mabi=ieeelongdouble.
23520 (init_float128_ieee): Call it.
23522 2018-03-27 H.J. Lu <hongjiu.lu@intel.com>
23525 * config/i386/i386.c (ix86_trampoline_init): Insert ENDBR for
23526 -fcf-protection=branch -mibt.
23527 * config/i386/i386.h (TRAMPOLINE_SIZE): Increased by 4 bytes.
23529 2018-03-27 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
23532 * config/arm/arm.c (arm_valid_symbolic_address): Handle
23533 arm_word_relocations.
23535 2018-03-27 Cesar Philippidis <cesar@codesourcery.com>
23538 * config/nvptx/nvptx.c (nvptx_assemble_decl_begin): Add '[]' to
23539 extern array declarations.
23541 2018-03-27 Richard Biener <rguenther@suse.de>
23543 PR middle-end/84067
23544 * match.pd ((A * C) +- (B * C) -> (A+-B) * C): Guard with
23545 explicit single_use checks.
23547 2018-03-27 Richard Biener <rguenther@suse.de>
23549 PR tree-optimization/85082
23550 * tree-ssa-sccvn.c (vn_reference_lookup_or_insert_for_pieces):
23553 2018-03-27 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
23555 * config.gcc (aarch64*-*-linux*): New TARGET_DEFAULT_ASYNC_UNWIND_TABLES
23556 * common/config/aarch64/aarch64-common.c (aarch64_optimization_table[]):
23557 Turn on fasynchronous-unwind-tables and funwind-tables.
23559 2018-03-26 Uros Bizjak <ubizjak@gmail.com>
23562 * config/i386/i386.md (*bmi_blsr_<mode>_cmp): New insn pattern.
23563 (*bmi_blsr_<mode>_ccz): Ditto.
23565 2018-03-26 Tom de Vries <tom@codesourcery.com>
23567 PR tree-optimization/85063
23568 * omp-general.c (offloading_function_p): New function. Factor out
23570 * omp-offload.c (pass_omp_target_link::gate): ... here.
23571 * omp-general.h (offloading_function_p): Declare.
23572 * tree-switch-conversion.c (build_one_array): Mark CSWTCH.x variable
23573 with attribute omp declare target for offloading functions.
23575 2018-03-24 Richard Sandiford <richard.sandiford@linaro.org>
23577 PR tree-optimization/84005
23578 * tree-data-ref.h (get_base_for_alignment): Declare.
23579 * tree-data-ref.c (get_base_for_alignment_1): New function.
23580 (get_base_for_alignment): Likewise.
23581 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
23582 get_base_for_alignment to find a suitable base object, instead
23583 of always using drb->base_address.
23585 2018-03-23 Jakub Jelinek <jakub@redhat.com>
23587 PR inline-asm/85022
23588 * emit-rtl.c (init_emit_regs): Indicate that VOIDmode MEMs don't have
23589 known size by default.
23591 2018-03-23 Vladimir Makarov <vmakarov@redhat.com>
23593 PR inline-asm/85030
23594 * lra-constraints.c (process_alt_operands): Don't match BLKmode
23595 and non BLKmode operands.
23597 2018-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23600 * config/arm/arm.md (unaligned_loadhis): Remove first alternative.
23601 Clean up attributes.
23603 2018-03-23 Richard Biener <rguenther@suse.de>
23606 * dwarf2out.c (rtl_for_decl_location): Do not generate RTL early when
23607 we are going to emit early debug for LTO.
23609 2018-03-23 Jakub Jelinek <jakub@redhat.com>
23611 PR inline-asm/85034
23612 * function.c (match_asm_constraints_1): Don't optimize if input
23613 doesn't satisfy general_operand predicate for output's mode.
23615 PR inline-asm/85022
23616 * alias.c (write_dependence_p): Don't require for x_canonicalized
23617 non-VOIDmode if x has VOIDmode.
23620 * sanopt.c (maybe_optimize_ubsan_ptr_ifn): If DECL_REGISTER (base),
23621 just don't try to optimize it rather than assert it never happens.
23623 2018-03-22 Kelvin Nilsen <kelvin@gcc.gnu.org>
23625 * config/rs6000/rs6000-builtin.def: Remove various BU_ALTIVEC_X
23626 macro expansions for definition of ST_INTERNAL_<mode> and
23627 LD_INTERNAL_<mode> builtins.
23628 * config/rs6000/rs6000-protos.h (rs6000_address_for_altivec):
23630 * config/rs6000/rs6000.c (altivec_expand_ld_builtin): Delete this
23632 (altivec_expand_st_builtin): Likewise.
23633 (altivec_expand_builtin): Remove calls to deleted functions.
23634 (rs6000_address_for_altivec): Delete this function.
23635 * config/rs6000/vector.md: Remove expands for
23636 vector_altivec_load_<mode> and vector_altivec_store_<mode>.
23638 2018-03-22 Sudakshina Das <sudi.das@arm.com>
23641 * config/arm/arm.h (machine_function): Add static_chain_stack_bytes.
23642 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Avoid
23643 re-computing once computed.
23644 (arm_expand_prologue): Compute machine->static_chain_stack_bytes.
23645 (arm_init_machine_status): Initialize
23646 machine->static_chain_stack_bytes.
23648 2018-03-22 Kelvin Nilsen <kelvin@gcc.gnu.org>
23651 * doc/extend.texi: Add four new prototypes for vec_ld.
23652 * config/rs6000/rs6000-builtin.def (LVX_V1TI): Reorder symbol
23653 definitions for more logical presentation.
23654 * config/rs6000/rs6000-c.c: (altivec_overloaded_builtins): Add
23655 entries for V1TI variants of __builtin_altivec_ld builtin.
23656 * config/rs6000/rs6000.c: (altivec_expand_lv_builtin): Add test and
23657 handling of V1TI variant of LVX icode pattern.
23658 (altivec_expand_builtin): Add case for ALTIVEC_BUILTIN_LVX_V1TI.
23659 (rs6000_gimple_fold_builtin): Likewise.
23660 (altivec_init_builtins): Add code to define
23661 __builtin_altivec_lvx_v1ti function.
23663 2018-03-22 Jakub Jelinek <jakub@redhat.com>
23665 PR inline-asm/84941
23666 * function.c (match_asm_constraints_1): Don't do the optimization
23667 if input isn't a REG, SUBREG, MEM or constant.
23669 2018-03-22 Tom de Vries <tom@codesourcery.com>
23671 PR tree-optimization/84956
23672 * tree-ssa-tail-merge.c (find_clusters_1): Skip bbs with
23673 bb_has_abnormal_pred.
23675 2018-03-22 Jakub Jelinek <jakub@redhat.com>
23678 * dwarf2asm.c (dw2_output_indirect_constant_1): Set
23679 DECL_INITIAL (decl) to decl at the end.
23680 * varasm.c (use_blocks_for_decl_p): Revert the 2018-03-20 change,
23681 adjust the comment.
23683 2018-03-21 Joseph Myers <joseph@codesourcery.com>
23685 * doc/extend.texi (__builtin_tgmath): Document when complex
23686 integer types are treated as _Complex _Float64.
23688 2018-03-21 Tom de Vries <tom@codesourcery.com>
23690 * doc/extend.texi (__builtin_extend_pointer): Remove pasto.
23692 2018-03-21 Jakub Jelinek <jakub@redhat.com>
23694 PR tree-optimization/84960
23695 * tree-cfg.c (remove_bb): Don't move forced labels into bb->prev_bb
23696 if it is ENTRY block, move them into single succ of ENTRY in that case.
23698 2018-03-21 Richard Sandiford <richard.sandiford@linaro.org>
23700 PR tree-optimization/84811
23701 * poly-int.h (poly_span_traits): Remove the T3 parameter and
23702 promote HOST_WIDE_INT T2 - T1 results to unsigned HOST_WIDE_INT.
23703 (maybe_in_range_p, known_in_range_p, ranges_known_overlap_p):
23704 (known_subrange_p): Update accordingly. Cast each value involved
23705 in the size comparison, rather than casting the result of the
23708 2018-03-21 Jakub Jelinek <jakub@redhat.com>
23710 PR tree-optimization/84982
23711 * gimple-ssa-store-merging.c (invert_op): Handle boolean inversion
23712 by flipping the least significant bit rather than all bits from
23713 bitpos to bitpos + bitsize - 1.
23715 2018-03-21 Nathan Sidwell <nathan@acm.org>
23717 * doc/extend.texi (Deprecated Features): Remove mention of
23718 long-deleted deprecations.
23720 2018-03-21 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
23723 * configure.ac (gcc_cv_ld_soname) <*-*-solaris2*>: Set.
23724 * configure: Regenerate.
23726 2018-03-21 Tom de Vries <tom@codesourcery.com>
23728 PR tree-optimization/83126
23729 * tree-parloops.c (num_phis): New function.
23730 (gen_parallel_loop): Detect and handle canonicalize_loop_ivs failure.
23732 2018-03-21 Nathan Sidwell <nathan@acm.org>
23734 * doc/extend.texi (Deprecated Features): Update deprecated flags,
23735 mention anon-struct/union members and trailing attributes.
23737 2018-03-21 Bin Cheng <bin.cheng@arm.com>
23739 PR tree-optimization/84969
23740 * tree-loop-distribution.c (fuse_memset_builtins): Don't reorder
23741 builtin memset partitions if they set different rhs values.
23743 2018-03-21 Jakub Jelinek <jakub@redhat.com>
23745 PR rtl-optimization/84989
23746 * simplify-rtx.c (simplify_unary_operation_1): Don't try to simplify
23747 VEC_DUPLICATE with scalar result mode.
23749 2018-03-21 Martin Liska <mliska@suse.cz>
23752 * ipa-icf.c (sem_item_optimizer::fixup_points_to_sets): Remove
23753 not intended return statement.
23755 2018-03-21 Martin Liska <mliska@suse.cz>
23758 * tree-chkp.c (CHKP_ARRAY_MAX_CHECK_STEPS): Define a new macro.
23759 (chkp_find_bound_slots_1): Limit number of iterations.
23761 2018-03-20 David H. Gutteridge <dhgutteridge@sympatico.ca>
23764 * Minor grammar fixes for x86 options.
23766 2018-03-20 Jakub Jelinek <jakub@redhat.com>
23769 * dce.c (delete_unmarked_insns): Don't remove frame related noop moves
23770 holding REG_CFA_RESTORE notes, instead turn them into a USE.
23772 2018-03-20 Peter Bergner <bergner@vnet.ibm.com>
23775 * config/rs6000/altivec.md (altivec_lvx_<mode>_2op): Delete define_insn.
23776 (altivec_lvx_<mode>_1op): Likewise.
23777 (altivec_stvx_<mode>_2op): Likewise.
23778 (altivec_stvx_<mode>_1op): Likewise.
23779 (altivec_lvx_<VM2:mode>): New define_expand.
23780 (altivec_stvx_<VM2:mode>): Likewise.
23781 (altivec_lvx_<VM2:mode>_2op_<P:mptrsize>): New define_insn.
23782 (altivec_lvx_<VM2:mode>_1op_<P:mptrsize>): Likewise.
23783 (altivec_stvx_<VM2:mode>_2op_<P:mptrsize>): Likewise.
23784 (altivec_stvx_<VM2:mode>_1op_<P:mptrsize>): Likewise.
23785 * config/rs6000/rs6000-p8swap.c (rs6000_gen_stvx): Use new expanders.
23786 (rs6000_gen_lvx): Likewise.
23787 * config/rs6000/rs6000.c (altivec_expand_lv_builtin): Likewise.
23788 (altivec_expand_stv_builtin): Likewise.
23789 (altivec_expand_builtin): Likewise.
23790 * config/rs6000/vector.md: Likewise.
23792 2018-03-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23795 * config/arm/arm.c (arm_array_mode_supported_p): Return false for
23798 2018-03-20 Richard Biener <rguenther@suse.de>
23801 * config/i386/i386.c (ix86_add_stmt_cost): Only cost
23802 sign-conversions as zero, fall back to standard scalar_stmt
23805 2018-03-20 Martin Liska <mliska@suse.cz>
23808 * predict.c (rebuild_frequencies): Handle case when we have
23809 PROFILE_ABSENT, but flag_guess_branch_prob is false.
23811 2018-03-20 Jakub Jelinek <jakub@redhat.com>
23814 * dwarf2asm.c (dw2_output_indirect_constant_1): Temporarily turn off
23815 flag_section_anchors.
23816 * varasm.c (use_blocks_for_decl_p): Remove hack for
23817 dw2_force_const_mem.
23820 * config/aarch64/aarch64.md (*aarch64_reg_<mode>3_neg_mask2): Rename
23822 (*aarch64_<optab>_reg_<mode>3_neg_mask2): ... this. If pseudos can't
23823 be created, use lowpart_subreg of operands[0] rather than operands[0]
23825 (*aarch64_reg_<mode>3_minus_mask): Rename to ...
23826 (*aarch64_ashl_reg_<mode>3_minus_mask): ... this.
23827 (*aarch64_<optab>_reg_di3_mask2): Use const_int_operand predicate
23828 and n constraint instead of aarch64_shift_imm_di and Usd.
23829 (*aarch64_reg_<optab>_minus<mode>3): Rename to ...
23830 (*aarch64_<optab>_reg_minus<mode>3): ... this.
23832 2018-03-20 Sudakshina Das <sudi.das@arm.com>
23835 * config/arm/neon.md (ashldi3_neon): Update ?s for constraints
23836 to favor GPR over NEON registers.
23837 (<shift>di3_neon): Likewise.
23839 2018-03-20 Tom de Vries <tom@codesourcery.com>
23842 * config/nvptx/nvptx.c (nvptx_single): Don't neuter bar.sync.
23843 (nvptx_process_pars): Emit bar.sync asap and alap.
23845 2018-03-20 Tom de Vries <tom@codesourcery.com>
23848 * config/nvptx/nvptx.c (prevent_branch_around_nothing): Also update
23849 seen_label if seen_label is already set.
23851 2018-03-20 Jakub Jelinek <jakub@redhat.com>
23854 * config/i386/i386.c (fold_builtin_cpu): For features above 31
23855 use __cpu_features2 variable instead of __cpu_model.__cpu_features[0].
23856 Use 1U instead of 1. Formatting fixes.
23859 * builtins.c (fold_builtin_strpbrk): For strpbrk(x, "") use type
23860 instead of TREE_TYPE (s1) for the return value.
23862 2018-03-19 Jakub Jelinek <jakub@redhat.com>
23864 PR tree-optimization/84946
23865 * gimple-ssa-store-merging.c (mem_valid_for_store_merging): Compute
23866 bitsize + bitsize in poly_uint64 rather than poly_int64.
23869 * dwarf2asm.c: Include fold-const.c.
23870 (dw2_output_indirect_constant_1): Set DECL_INITIAL (decl) to ADDR_EXPR
23871 of decl rather than decl itself.
23873 PR rtl-optimization/84643
23874 * memmodel.h (enum memmodel): Add MEMMODEL_MAX enumerator.
23876 2018-03-19 Maxim Ostapenko <m.ostapenko@samsung.com>
23879 * dwarf2asm.c (dw2_output_indirect_constant_1): Disable ASan before
23880 calling assemble_variable.
23882 2018-03-19 Sudakshina Das <sudi.das@arm.com>
23885 * config/aarch64/aarch64-simd.md (vec_cmp<mode><v_int_equiv>): Modify
23886 instructions for UNLT, UNLE, UNGT, UNGE, UNEQ, UNORDERED and ORDERED.
23888 2018-03-19 Jim Wilson <jimw@sifive.com>
23891 * config/riscv/riscv.c (riscv_function_arg_boundary): Use
23892 PREFERRED_STACK_BOUNDARY instead of STACK_BOUNDARY.
23893 (riscv_first_stack_step): Likewise.
23894 (riscv_option_override): Use STACK_BOUNDARY instead of
23895 MIN_STACK_BOUNDARY.
23896 * config/riscv/riscv.h (STACK_BOUNDARY): Renamed from
23897 MIN_STACK_BOUNDARY.
23898 (BIGGEST_ALIGNMENT): Set to 128.
23899 (PREFERRED_STACK_BOUNDARY): Renamed from STACK_BOUNDARY.
23900 (RISCV_STACK_ALIGN): Use PREFERRED_STACK_BOUNDARY instead of
23903 2018-03-19 Richard Biener <rguenther@suse.de>
23905 PR tree-optimization/84933
23906 * tree-vrp.c (set_and_canonicalize_value_range): Treat out-of-bound
23907 values as -INF/INF when canonicalizing an ANTI_RANGE to a RANGE.
23909 2018-03-19 Richard Biener <rguenther@suse.de>
23911 PR tree-optimization/84859
23912 * tree-ssa-phiopt.c (single_trailing_store_in_bb): New function.
23913 (cond_if_else_store_replacement): Perform sinking operation on
23914 single-store BBs regardless of MAX_STORES_TO_SINK setting.
23915 Generalize what a BB with a single eligible store is.
23917 2018-03-19 Richard Biener <rguenther@suse.de>
23919 PR tree-optimization/84929
23920 * tree-data-ref.c (analyze_siv_subscript_cst_affine): Guard
23921 chrec_is_positive against non-chrec arg.
23923 2018-03-19 Tamar Christina <tamar.christina@arm.com>
23926 * config/arm/arm.c (arm_can_change_mode_class): revert r258554.
23928 2018-03-18 Martin Liska <mliska@suse.cz>
23930 PR rtl-optimization/84635
23931 * regrename.c (build_def_use): Use matches_mode only when
23934 2018-03-18 Richard Sandiford <richard.sandiford@linaro.org>
23936 PR tree-optimization/84913
23937 * tree-vect-loop.c (vectorizable_reduction): Don't try to
23938 vectorize chains of COND_EXPRs.
23940 2018-03-18 Chung-Ju Wu <jasonwucj@gmail.com>
23942 * config/nds32/nds32.h (MAX_REGS_PER_ADDRESS): Fix the value.
23944 2018-03-18 Chung-Ju Wu <jasonwucj@gmail.com>
23946 * config/nds32/nds32.h (LOGICAL_OP_NON_SHORT_CIRCUIT): Define.
23948 2018-03-18 Chung-Ju Wu <jasonwucj@gmail.com>
23950 * config/nds32/nds32.h (CLZ_DEFINED_VALUE_AT_ZERO): Define.
23952 2018-03-17 Chung-Ju Wu <jasonwucj@gmail.com>
23953 Kito Cheng <kito.cheng@gmail.com>
23955 * config/nds32/nds32-protos.h (nds32_adjust_reg_alloc_order): Declare.
23956 * config/nds32/nds32.c (nds32_reg_alloc_order_for_speed): New array.
23957 (nds32_adjust_reg_alloc_order): New function.
23958 * config/nds32/nds32.h (ADJUST_REG_ALLOC_ORDER): Define.
23960 2018-03-17 Kito Cheng <kito.cheng@gmail.com>
23962 * config/nds32/nds32.c (nds32_asm_output_mi_thunk,
23963 nds32_print_operand, nds32_print_operand_address): Use
23964 HOST_WIDE_INT_PRINT_DEC instead.
23966 2018-03-17 Chung-Ju Wu <jasonwucj@gmail.com>
23968 * config/nds32/nds32.c (nds32_register_priority): Modify cost.
23970 2018-03-17 Jakub Jelinek <jakub@redhat.com>
23973 * config/i386/i386.c (initial_ix86_tune_features,
23974 initial_ix86_arch_features): Use unsigned HOST_WIDE_INT rather than
23975 unsigned long long.
23976 (set_ix86_tune_features): Change ix86_tune_mask from unsigned int
23977 to unsigned HOST_WIDE_INT, initialize to HOST_WIDE_INT_1U << ix86_tune
23978 rather than 1u << ix86_tune. Formatting fix.
23979 (ix86_option_override_internal): Change ix86_arch_mask from
23980 unsigned int to unsigned HOST_WIDE_INT, initialize to
23981 HOST_WIDE_INT_1U << ix86_arch rather than 1u << ix86_arch.
23982 (ix86_function_specific_restore): Likewise.
23984 2018-03-16 Jakub Jelinek <jakub@redhat.com>
23987 * postreload.c (reload_combine_recognize_pattern): Perform
23988 INTVAL addition in unsigned HOST_WIDE_INT type to avoid UB and
23989 truncate_int_for_mode the result for the destination's mode.
23992 * hsa-gen.c (mem_type_for_type): Fix comment typo.
23993 * tree-vect-loop-manip.c (vect_create_cond_for_niters_checks):
23995 * gimple-ssa-warn-restrict.c (builtin_memref::set_base_and_offset):
23998 2018-03-16 Vladimir Makarov <vmakarov@redhat.com>
24001 * lra-assigns.c (lra_split_hard_reg_for): Don't use
24002 regno_allocno_class_array and sorted_pseudos.
24003 * lra-constraints.c (spill_hard_reg_in_range): Ignore hard regs in
24004 insns where regno is used.
24006 2018-03-16 Martin Liska <mliska@suse.cz>
24009 * multiple_target.c (create_dispatcher_calls): Redirect
24010 reference in the symbol table.
24012 2018-03-16 Martin Liska <mliska@suse.cz>
24015 * multiple_target.c (create_dispatcher_calls): Redirect also
24018 2018-03-16 Jakub Jelinek <jakub@redhat.com>
24022 * tree.h (TARGET_EXPR_NO_ELIDE): Define.
24023 * gimplify.c (gimplify_modify_expr_rhs): Don't elide TARGET_EXPRs with
24024 TARGET_EXPR_NO_ELIDE flag set unless *expr_p is INIT_EXPR.
24026 2018-03-16 Julia Koval <julia.koval@intel.com>
24028 * doc/invoke.texi (Skylake Server): Add CLWB.
24029 Cannonlake): Remove CLWB.
24031 2018-03-16 Jakub Jelinek <jakub@redhat.com>
24033 PR tree-optimization/84841
24034 * tree-ssa-reassoc.c (INTEGER_CONST_TYPE): Change to 1 << 4 from
24036 (FLOAT_ONE_CONST_TYPE): Define.
24037 (constant_type): Return FLOAT_ONE_CONST_TYPE for -1.0 and 1.0.
24038 (sort_by_operand_rank): Put entries with higher constant_type last
24039 rather than first to match comments.
24041 2018-03-15 Sandra Loosemore <sandra@codesourcery.com>
24043 * config/nios2/nios2.md (movsi_internal): Fix thinko in
24046 2018-03-15 Jakub Jelinek <jakub@redhat.com>
24049 * calls.c (expand_call): For TREE_ADDRESSABLE rettype ignore alignment
24050 check and use address of target always.
24052 2018-03-15 H.J. Lu <hongjiu.lu@intel.com>
24055 * config/i386/i386.c (indirect_thunk_needed): Update comments.
24056 (indirect_thunk_bnd_needed): Likewise.
24057 (indirect_thunks_used): Likewise.
24058 (indirect_thunks_bnd_used): Likewise.
24059 (indirect_return_needed): New.
24060 (indirect_return_bnd_needed): Likewise.
24061 (output_indirect_thunk_function): Add a bool argument for
24063 (output_indirect_thunk_function): Don't generate alias for
24064 function return thunk.
24065 (ix86_code_end): Call output_indirect_thunk_function to generate
24066 function return thunks.
24067 (ix86_output_function_return): Set indirect_return_bnd_needed
24068 and indirect_return_needed instead of indirect_thunk_bnd_needed
24069 and indirect_thunk_needed.
24071 2018-03-15 Olga Makhotina <olga.makhotina@intel.com>
24073 * config/i386/sgxintrin.h (_enclv_u32): New intrinsic.
24074 (__enclv_bc, __enclv_cd, __enclv_generic): New definitions.
24075 (ERDINFO, ETRACKC, ELDBC, ELDUC): New leaves.
24077 2018-03-15 David Malcolm <dmalcolm@redhat.com>
24078 Paul Hua <paul.hua.gm@gmail.com>
24081 * gcc.dg/fixits-pr84852-1.c: Fix filename in dg-regexp.
24083 2018-03-15 Segher Boessenkool <segher@kernel.crashing.org>
24085 * config/rs6000/rs6000.c (abi_v4_pass_in_fpr): Add back the
24086 TARGET_DOUBLE_FLOAT and TARGET_SINGLE_FLOAT conditions on the DFmode
24087 resp. SFmode cases.
24089 2018-03-15 Tamar Christina <tamar.christina@arm.com>
24092 * config/arm/arm.c (arm_can_change_mode_class): Use GET_MODE_UNIT_SIZE
24093 instead of GET_MODE_SIZE when comparing Units.
24095 2018-03-15 Vladimir Mezentsev <vladimir.mezentsev@oracle.com>
24098 * varasm.c (hash_section): Return an unchangeble hash value
24099 * config/aarch64/aarch64.c (aarch64_use_blocks_for_constant_p):
24100 Return !aarch64_can_use_per_function_literal_pools_p ().
24102 2018-03-15 Jakub Jelinek <jakub@redhat.com>
24105 * optabs.c (emit_conditional_move): Pass address of cmode's copy
24106 rather than address of cmode as last argument to prepare_cmp_insn.
24108 2018-03-15 Julia Koval <julia.koval@intel.com>
24110 * config/i386/i386.c (F_AVX512VBMI2, F_GFNI, F_VPCLMULQDQ,
24111 F_AVX512VNNI, F_AVX512BITALG): New.
24113 2018-03-14 John David Anglin <danglin@gcc.gnu.org>
24116 * config/pa/pa.c (pa_emit_move_sequence): Always emit secondary reload
24117 insn for floating-point loads and stores.
24119 2018-03-14 Carl Love <cel@us.ibm.com>
24121 * config/rs6000/rs6000-c.c: Add macro definitions for
24122 ALTIVEC_BUILTIN_VEC_PERMXOR.
24123 * config/rs6000/rs6000.h: Add #define for vec_permxor builtin.
24124 * config/rs6000/rs6000-builtin.def: Add macro expansions for VPERMXOR.
24125 * config/rs6000/altivec.md (altivec_vpermxor): New define expand.
24126 * config/rs6000/rs6000-p8swap.c (rtx_is_swappable_p): Add case
24128 * config/doc/extend.texi: Add prototypes for vec_permxor.
24130 2018-03-14 David Malcolm <dmalcolm@redhat.com>
24133 * diagnostic-show-locus.c (class layout_point): Convert m_line
24134 from int to linenum_type.
24135 (line_span::comparator): Use linenum "compare" function when
24136 comparing line numbers.
24137 (test_line_span): New function.
24138 (layout_range::contains_point): Convert param "row" from int to
24140 (layout_range::intersects_line_p): Likewise.
24141 (layout::will_show_line_p): Likewise.
24142 (layout::print_source_line): Likewise.
24143 (layout::should_print_annotation_line_p): Likewise.
24144 (layout::print_annotation_line): Likewise.
24145 (layout::print_leading_fixits): Likewise.
24146 (layout::annotation_line_showed_range_p): Likewise.
24147 (struct line_corrections): Likewise for field m_row.
24148 (line_corrections::line_corrections): Likewise for param "row".
24149 (layout::print_trailing_fixits): Likewise.
24150 (layout::get_state_at_point): Likewise.
24151 (layout::get_x_bound_for_row): Likewise.
24152 (layout::print_line): Likewise.
24153 (diagnostic_show_locus): Likewise for locals "last_line" and "row".
24154 (selftest::diagnostic_show_locus_c_tests): Call test_line_span.
24155 * input.c (selftest::test_linenum_comparisons): New function.
24156 (selftest::input_c_tests): Call it.
24157 * selftest.c (selftest::test_assertions): Test ASSERT_GT,
24158 ASSERT_GT_AT, ASSERT_LT, and ASSERT_LT_AT.
24159 * selftest.h (ASSERT_GT): New macro.
24160 (ASSERT_GT_AT): New macro.
24161 (ASSERT_LT): New macro.
24162 (ASSERT_LT_AT): New macro.
24164 2018-03-14 Segher Boessenkool <segher@kernel.crashing.org>
24166 PR rtl-optimization/84780
24167 * combine.c (distribute_links): Don't make a link based on pc_rtx.
24169 2018-03-14 Martin Liska <mliska@suse.cz>
24171 * tree.c (record_node_allocation_statistics): Use
24172 get_stats_node_kind.
24173 (get_stats_node_kind): New function extracted from
24174 record_node_allocation_statistics.
24175 (free_node): Use get_stats_node_kind.
24177 2018-03-14 Richard Biener <rguenther@suse.de>
24179 * tree-ssa-pre.c (compute_antic_aux): Remove code that asserts
24180 that the value-set of ANTIC_IN doesn't grow.
24183 * tree-ssa-pre.c (struct bb_bitmap_sets): Add visited_with_visited_succs
24185 (BB_VISITED_WITH_VISITED_SUCCS): New define.
24186 (compute_antic): Initialize BB_VISITED_WITH_VISITED_SUCCS.
24188 2018-03-14 Julia Koval <julia.koval@intel.com>
24190 * config.gcc (icelake-client, icelake-server): New.
24192 * config/i386/i386.c (initial_ix86_tune_features): Extend to 64 bit.
24193 (initial_ix86_arch_features): Ditto.
24194 (PTA_SKYLAKE): Add SGX.
24195 (PTA_ICELAKE): Remove.
24196 (PTA_ICELAKE_CLIENT): New.
24197 (PTA_ICELAKE_SERVER): New.
24198 (ix86_option_override_internal): Split up icelake on icelake client and
24200 (get_builtin_code_for_version): Ditto.
24201 (fold_builtin_cpu): Ditto.
24202 * config/i386/driver-i386.c (config/i386/driver-i386.c): Ditto.
24203 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto
24204 * config/i386/i386.h (processor_type): Ditto.
24205 * doc/invoke.texi: Ditto.
24207 2018-03-14 Jakub Jelinek <jakub@redhat.com>
24210 * sanopt.c (maybe_optimize_ubsan_ptr_ifn): Handle also
24211 INTEGER_CST offset, add it together with bitpos / 8 and
24212 sign extend based on POINTER_SIZE.
24216 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
24219 * config/i386/constraints.md (Yc): New register constraint.
24220 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
24221 Use Yc constraint for alternative 2 of operand 0. Remove
24222 preferred_for_speed attribute.
24224 2018-03-14 Richard Biener <rguenther@suse.de>
24226 PR tree-optimization/84830
24227 * tree-ssa-pre.c (compute_antic_aux): Intersect the new ANTIC_IN
24228 with the old one to avoid oscillations.
24230 2018-03-13 Vladimir Makarov <vmakarov@redhat.com>
24233 * lra-assigns.c (find_all_spills_for): Ignore uninteresting
24235 (assign_by_spills): Return a flag of reload assignment failure.
24236 Do not process the reload assignment failures. Do not spill other
24237 reload pseudos if they has the same reg class. Update n if
24239 (lra_assign): Add a return arg. Set up from the result of
24240 assign_by_spills call.
24241 (find_reload_regno_insns, lra_split_hard_reg_for): New functions.
24242 * lra-constraints.c (split_reg): Add a new arg. Use it instead of
24243 usage_insns if it is not NULL.
24244 (spill_hard_reg_in_range): New function.
24245 (split_if_necessary, inherit_in_ebb): Pass a new arg to split_reg.
24246 * lra-int.h (spill_hard_reg_in_range, lra_split_hard_reg_for): New
24247 function prototypes.
24248 (lra_assign): Change prototype.
24249 * lra.c (lra): Add code to deal with fails by splitting hard reg
24252 2018-03-01 Palmer Dabbelt <palmer@sifive.com>
24254 * config/riscv/riscv.opt (mrelax): New option.
24255 * config/riscv/riscv.c (riscv_file_start): Emit ".option
24256 "norelax" when riscv_mrelax is disabled.
24257 * doc/invoke.texi (RISC-V): Document "-mrelax" and "-mno-relax".
24259 2018-03-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
24262 * config/rs6000/rs6000.c (rs6000_reassociation_width): Disable parallel
24263 reassociation for int modes.
24265 2018-03-13 Richard Sandiford <richard.sandiford@linaro.org>
24267 * tree-vect-loop-manip.c (vect_maybe_permute_loop_masks):
24268 Reverse the choice between VEC_UNPACK_LO_EXPR and VEC_UNPACK_HI_EXPR
24270 * config/aarch64/iterators.md (hi_lanes_optab): New int attribute.
24271 * config/aarch64/aarch64-sve.md
24272 (*aarch64_sve_<perm_insn><perm_hilo><mode>): Rename to...
24273 (aarch64_sve_<perm_insn><perm_hilo><mode>): ...this.
24274 (*extend<mode><Vwide>2): Rename to...
24275 (aarch64_sve_extend<mode><Vwide>2): ...this.
24276 (vec_unpack<su>_<perm_hilo>_<mode>): Turn into a define_expand,
24277 renaming the old pattern to...
24278 (aarch64_sve_punpk<perm_hilo>_<mode>): ...this. Only define
24280 (vec_unpack<su>_<perm_hilo>_<SVE_BHSI:mode>): Turn into a
24281 define_expand, renaming the old pattern to...
24282 (aarch64_sve_<su>unpk<perm_hilo>_<SVE_BHSI:mode>): ...this.
24283 (*vec_unpacku_<perm_hilo>_<mode>_no_convert): Delete.
24284 (vec_unpacks_<perm_hilo>_<mode>): Take BYTES_BIG_ENDIAN into
24285 account when deciding which SVE instruction the optab should use.
24286 (vec_unpack<su_optab>_float_<perm_hilo>_vnx4si): Likewise.
24288 2018-03-13 Richard Sandiford <richard.sandiford@linaro.org>
24290 * config/aarch64/aarch64.md (V4_REGNUM, V8_REGNUM, V12_REGNUM)
24291 (V20_REGNUM, V24_REGNUM, V28_REGNUM, P1_REGNUM, P2_REGNUM, P3_REGNUM)
24292 (P4_REGNUM, P5_REGNUM, P6_REGNUM, P8_REGNUM, P9_REGNUM, P10_REGNUM)
24293 (P11_REGNUM, P12_REGNUM, P13_REGNUM, P14_REGNUM): New define_constants.
24294 (tlsdesc_small_<mode>): Turn a define_expand and use
24295 tlsdesc_small_sve_<mode> for SVE. Rename original define_insn to...
24296 (tlsdesc_small_advsimd_<mode>): ...this.
24297 (tlsdesc_small_sve_<mode>): New pattern.
24299 2018-03-13 Richard Sandiford <richard.sandiford@linaro.org>
24301 * config/aarch64/iterators.md (UNSPEC_SMUL_HIGHPART)
24302 (UNSPEC_UMUL_HIGHPART): New constants.
24303 (MUL_HIGHPART): New int iteraor.
24304 (su): Handle UNSPEC_SMUL_HIGHPART and UNSPEC_UMUL_HIGHPART.
24305 * config/aarch64/aarch64-sve.md (<su>mul<mode>3_highpart): New
24307 (*<su>mul<mode>3_highpart): New define_insn.
24309 2018-03-13 Eric Botcazou <ebotcazou@adacore.com>
24312 * ipa-devirt.c (odr_subtypes_equivalent_p): Do not get the ODR type of
24315 2018-03-13 Martin Liska <mliska@suse.cz>
24318 * (sem_item_optimizer::sem_item_optimizer): Initialize new
24320 (sem_item_optimizer::~sem_item_optimizer): Release it.
24321 (sem_item_optimizer::merge_classes): Register variable aliases.
24322 (sem_item_optimizer::fixup_pt_set): New function.
24323 (sem_item_optimizer::fixup_points_to_sets): Likewise.
24324 * ipa-icf.h: Declare new variables and functions.
24326 2018-03-13 Jakub Jelinek <jakub@redhat.com>
24328 PR middle-end/84834
24329 * match.pd ((A & C) != 0 ? D : 0): Use INTEGER_CST@2 instead of
24330 integer_pow2p@2 and test integer_pow2p in condition.
24331 (A < 0 ? C : 0): Similarly for @1.
24333 PR middle-end/84831
24334 * stmt.c (parse_output_constraint): If the CONSTRAINT_LEN (*p, p)
24335 characters starting at p contain '\0' character, don't look beyond
24339 * config/i386/i386.md (round<mode>2): For 387 fancy math, disable
24340 pattern if -ftrapping-math -fno-fp-int-builtin-inexact.
24343 * reg-stack.c (change_stack): Change update_end var from int to
24344 rtx_insn *, if non-NULL don't update just BB_END (current_block), but
24345 also call set_block_for_insn on the newly added insns and rescan.
24348 * config/i386/sse.md (sse2_loadhpd): Use Yv constraint rather than v
24349 on the last operand.
24352 * tree.c (stabilize_reference_1): Return save_expr (e) for
24353 STATEMENT_LIST even if it doesn't have side-effects.
24355 2018-03-12 Jonathan Wakely <jwakely@redhat.com>
24357 * doc/invoke.texi (-mclflushopt): Fix spelling of option.
24359 2018-03-12 Renlin Li <renlin.li@arm.com>
24361 * config/aarch64/aarch64.md (movhf_aarch64): Fix mode argument to
24362 aarch64_output_scalar_simd_mov_immediate.
24364 2018-03-12 Martin Sebor <msebor@redhat.com>
24366 PR tree-optimization/83456
24367 * gimple-fold.c (gimple_fold_builtin_memory_op): Avoid warning
24368 for perfectly overlapping calls to memcpy.
24369 (gimple_fold_builtin_memory_chk): Same.
24370 (gimple_fold_builtin_strcpy): Handle no-warning.
24371 (gimple_fold_builtin_stxcpy_chk): Same.
24372 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Handle no-warning.
24374 2018-03-12 Segher Boessenkool <segher@kernel.crashing.org>
24376 * config/rs6000/rs6000.c (abi_v4_pass_in_fpr): Add bool "named"
24377 parameter. Use it for SFmode.
24378 (rs6000_function_arg_advance_1): Adjust.
24379 (rs6000_function_arg): Adjust.
24380 (rs6000_gimplify_va_arg): Pass false for that new parameter.
24382 2018-03-12 Segher Boessenkool <segher@kernel.crashing.org>
24384 PR rtl-optimization/84169
24385 PR rtl-optimization/84780
24386 * combine.c (can_combine_p): Check for a 2-insn combination whether
24387 the destination register is used between the two insns, too.
24389 2018-03-12 Richard Biener <rguenther@suse.de>
24391 PR tree-optimization/84803
24392 * tree-if-conv.c (ifcvt_memrefs_wont_trap): Don't do anything
24393 for refs DR analysis didn't process.
24395 2018-03-12 Richard Biener <rguenther@suse.de>
24397 PR tree-optimization/84777
24398 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): For
24399 force-vectorize loops ignore whether we are optimizing for size.
24401 2018-03-12 Chung-Ju Wu <jasonwucj@gmail.com>
24403 * config/nds32/nds32.c (nds32_md_asm_adjust): New function.
24404 (TARGET_MD_ASM_ADJUST): Define.
24406 2018-03-12 Monk Chiang <sh.chiang04@gmail.com>
24407 Kito Cheng <kito.cheng@gmail.com>
24408 Chung-Ju Wu <jasonwucj@gmail.com>
24410 * config/nds32/nds32.c (nds32_compute_stack_frame,
24411 nds32_emit_stack_push_multiple, nds32_emit_stack_pop_multiple,
24412 nds32_emit_stack_v3push, nds32_emit_stack_v3pop,
24413 nds32_emit_adjust_frame, nds32_expand_prologue, nds32_expand_epilogue,
24414 nds32_expand_prologue_v3push, nds32_expand_epilogue_v3pop): Refine.
24415 * config/nds32/nds32.h (NDS32_FIRST_CALLEE_SAVE_GPR_REGNUM,
24416 NDS32_LAST_CALLEE_SAVE_GPR_REGNUM, NDS32_V3PUSH_AVAILABLE_P): New.
24417 * config/nds32/nds32.md (prologue, epilogue): Use macro
24418 NDS32_V3PUSH_AVAILABLE_P to do checking.
24420 2018-03-11 Jakub Jelinek <jakub@redhat.com>
24423 * dwarf2out.c (gen_enumeration_type_die): Don't guard adding
24424 DW_AT_declaration for ENUM_IS_OPAQUE on -gdwarf-4 or -gno-strict-dwarf,
24425 but on TYPE_SIZE. Don't do anything for ENUM_IS_OPAQUE if not creating
24426 a new die. Don't set TREE_ASM_WRITTEN if ENUM_IS_OPAQUE. Guard
24427 addition of most attributes on !orig_type_die or the attribute not
24428 being present already. Assert TYPE_VALUES is NULL for ENUM_IS_OPAQUE.
24430 2018-03-11 Kito Cheng <kito.cheng@gmail.com>
24431 Chung-Ju Wu <jasonwucj@gmail.com>
24433 * config/nds32/nds32.c (nds32_cpu_cpp_builtins): Modify to define
24434 __NDS32_VH__ macro.
24435 * config/nds32/nds32.opt (mvh): New option.
24437 2018-03-11 Kito Cheng <kito.cheng@gmail.com>
24438 Chung-Ju Wu <jasonwucj@gmail.com>
24440 * config/nds32/nds32-protos.h (nds32_cpu_cpp_builtins): Declare
24442 * config/nds32/nds32.c (nds32_cpu_cpp_builtins): New function.
24443 * config/nds32/nds32.h (TARGET_CPU_CPP_BUILTINS): Modify its
24446 2018-03-11 Kito Cheng <kito.cheng@gmail.com>
24447 Chung-Ju Wu <jasonwucj@gmail.com>
24449 * config/nds32/nds32-memory-manipulation.c (nds32_expand_strlen): New
24451 * config/nds32/nds32-multiple.md (strlensi): New pattern.
24452 * config/nds32/nds32-protos.h (nds32_expand_strlen): Declare function.
24454 2018-03-11 Monk Chiang <sh.chiang04@gmail.com>
24455 Kito Cheng <kito.cheng@gmail.com>
24456 Chung-Ju Wu <jasonwucj@gmail.com>
24458 * config/nds32/constants.md (unspec_element): Add UNSPEC_FFB,
24459 UNSPEC_FFMISM and UNSPEC_FLMISM.
24460 * config/nds32/nds32-intrinsic.c (bdesc_2arg): Add builtin description
24461 for ffb, ffmism and flmism.
24462 * config/nds32/nds32-intrinsic.md (unspec_ffb): Define new pattern.
24463 (unspec_ffmism): Ditto.
24464 (unspec_flmism): Ditto.
24465 (nds32_expand_builtin_impl): Check if string extension is available.
24466 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_FFB,
24467 NDS32_BUILTIN_FFMISM and NDS32_BUILTIN_FLMISM.
24469 2018-03-10 Vladimir Makarov <vmakarov@redhat.com>
24472 2018-03-09 Vladimir Makarov <vmakarov@redhat.com>
24475 * lra-assigns.c (assign_by_spills): Return a flag of reload
24476 assignment failure. Do not process the reload assignment
24477 failures. Do not spill other reload pseudos if they has the same
24479 (lra_assign): Add a return arg. Set up from the result of
24480 assign_by_spills call.
24481 (find_reload_regno_insns, lra_split_hard_reg_for): New functions.
24482 * lra-constraints.c (split_reg): Add a new arg. Use it instead of
24483 usage_insns if it is not NULL.
24484 (spill_hard_reg_in_range): New function.
24485 (split_if_necessary, inherit_in_ebb): Pass a new arg to split_reg.
24486 * lra-int.h (spill_hard_reg_in_range, lra_split_hard_reg_for): New
24487 function prototypes.
24488 (lra_assign): Change prototype.
24489 * lra.c (lra): Add code to deal with fails by splitting hard reg
24492 2018-03-10 H.J. Lu <hongjiu.lu@intel.com>
24495 * config/i386/i386.opt: Replace Enforcment with Enforcement.
24497 2018-03-10 Alexandre Oliva <aoliva@redhat.com>
24500 * dwarf2out.h (dw_val_class): Add dw_val_class_symview.
24501 (dw_val_node): Add val_symbolic_view.
24502 * dwarf2out.c (dw_line_info_table): Add symviews_since_reset.
24503 (symview_upper_bound): New.
24504 (new_line_info_table): Initialize symviews_since_reset.
24505 (dwarf2out_source_line): Count symviews_since_reset and set
24506 symview_upper_bound.
24507 (dw_val_equal_p): Handle symview.
24508 (add_AT_symview): New.
24509 (print_dw_val): Handle symview.
24510 (attr_checksum, attr_checksum_ordered): Likewise.
24511 (same_dw_val_p, size_of_die): Likewise.
24512 (value_format, output_die): Likewise.
24513 (add_high_low_attributes): Use add_AT_symview for entry_view.
24514 (dwarf2out_finish): Reset symview_upper_bound, clear
24517 2018-03-09 Peter Bergner <bergner@vnet.ibm.com>
24520 * config/rs6000/rs6000.c (rs6000_offsettable_memref_p): New prototype.
24521 Add strict argument and use it.
24522 (rs6000_split_multireg_move): Update for new strict argument.
24523 (mem_operand_gpr): Disallow all non-offsettable addresses.
24524 * config/rs6000/rs6000.md (*movdi_internal64): Use YZ constraint.
24526 2018-03-09 Jakub Jelinek <jakub@redhat.com>
24529 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Mark va_arg_tmp
24530 temporary TREE_ADDRESSABLE before gimplification of BUILT_IN_MEMCPY.
24531 * config/powerpcspe/powerpcspe.c (rs6000_gimplify_va_arg): Likewise.
24534 * tree-inline.c (copy_tree_body_r): For INDIRECT_REF of a remapped
24535 decl, use remap_type if we want to use the type.
24537 2018-03-09 Martin Sebor <msebor@redhat.com>
24539 PR tree-optimization/84526
24540 * gimple-ssa-warn-restrict.c (builtin_memref::set_base_and_offset):
24542 (builtin_access::generic_overlap): Be prepared to handle non-array
24545 2018-03-09 Alexandre Oliva <aoliva@redhat.com>
24547 PR rtl-optimization/84682
24548 * lra-constraints.c (process_address_1): Check is_address flag
24549 for address constraints.
24550 (process_alt_operands): Likewise.
24551 * lra.c (lra_set_insn_recog_data): Pass asm operand locs to
24552 preprocess_constraints.
24553 * recog.h (preprocess_constraints): Add oploc parameter.
24555 * recog.c (preprocess_constraints): Test address_operand for
24556 CT_ADDRESS constraints.
24558 2018-03-09 Vladimir Makarov <vmakarov@redhat.com>
24561 * lra-assigns.c (assign_by_spills): Return a flag of reload
24562 assignment failure. Do not process the reload assignment
24563 failures. Do not spill other reload pseudos if they has the same
24565 (lra_assign): Add a return arg. Set up from the result of
24566 assign_by_spills call.
24567 (find_reload_regno_insns, lra_split_hard_reg_for): New functions.
24568 * lra-constraints.c (split_reg): Add a new arg. Use it instead of
24569 usage_insns if it is not NULL.
24570 (spill_hard_reg_in_range): New function.
24571 (split_if_necessary, inherit_in_ebb): Pass a new arg to split_reg.
24572 * lra-int.h (spill_hard_reg_in_range, lra_split_hard_reg_for): New
24573 function prototypes.
24574 (lra_assign): Change prototype.
24575 * lra.c (lra): Add code to deal with fails by splitting hard reg
24578 2018-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24581 * common/config/arm/arm-common.c (arm_parse_arch_option_name):
24582 Accept complain bool parameter. Only emit errors if it is true.
24583 (arm_parse_cpu_option_name): Likewise.
24584 (arm_target_thumb_only): Adjust callers of the above.
24585 * config/arm/arm-protos.h (arm_parse_cpu_option_name): Adjust
24586 prototype to take a default true bool parameter.
24587 (arm_parse_arch_option_name): Likewise.
24589 2018-03-09 David Malcolm <dmalcolm@redhat.com>
24590 Francois-Xavier Coudert <fxcoudert@gcc.gnu.org>
24594 * Makefile.in (LD_VERSION_SCRIPT_OPTION, LD_SONAME_OPTION): New.
24595 * configure: Regenerate.
24596 * configure.ac ("linker --version-script option"): New.
24597 ("linker soname option"): New.
24599 2018-03-09 Richard Biener <rguenther@suse.de>
24601 PR tree-optimization/84775
24602 * tree-if-conv.c (add_bb_predicate_gimplified_stmts): Delink
24603 immediate uses of predicate stmts and mark them modified.
24606 PR tree-optimization/84178
24607 * tree-if-conv.c (combine_blocks): Move insert_gimplified_predicates
24609 (version_loop_for_if_conversion): Delay update_ssa call.
24610 (tree_if_conversion): Delay update_ssa until after predicate
24613 2018-03-09 Eric Botcazou <ebotcazou@adacore.com>
24616 * config/i386/winnt.c (i386_pe_seh_cold_init): Use small pre-allocation
24617 when the function accesses prior frames.
24619 2018-03-08 Jakub Jelinek <jakub@redhat.com>
24622 * dwarf2out.c (dw_loc_list): If list && loc_list->first->next, call
24623 gen_llsym, otherwise call maybe_gen_llsym.
24625 PR inline-asm/84742
24626 * recog.c (asm_operand_ok): Return 0 if multi-character constraint
24627 has ',' character inside of it.
24629 2018-03-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24632 * config/aarch64/aarch64.md (*compare_cstore<mode>_insn): Mark pattern
24633 as clobbering CC_REGNUM.
24635 2018-03-08 Richard Biener <rguenther@suse.de>
24637 PR middle-end/84552
24638 * tree-scalar-evolution.c: Include tree-into-ssa.h.
24639 (follow_copies_to_constant): Do not follow SSA names registered
24642 2018-03-08 Richard Biener <rguenther@suse.de>
24644 PR tree-optimization/84178
24645 * tree-if-conv.c (combine_blocks): Move insert_gimplified_predicates
24647 (version_loop_for_if_conversion): Delay update_ssa call.
24648 (tree_if_conversion): Delay update_ssa until after predicate
24651 2018-03-08 David Malcolm <dmalcolm@redhat.com>
24653 PR tree-optimization/84178
24654 * tree-if-conv.c (release_bb_predicate): Remove the
24655 the assertion that the stmts have NULL use_ops.
24656 Discard the statements, asserting that they haven't
24657 yet been added to a BB.
24659 2018-03-08 Richard Biener <rguenther@suse.de>
24661 PR tree-optimization/84746
24662 * tree-ssa-pre.c (find_leader_in_sets): Deal with SET1 being NULL.
24663 (phi_translate): Pass in destination ANTIC_OUT set.
24664 (phi_translate_1): Likewise. For a simplified result lookup
24665 a leader in ANTIC_OUT and AVAIL_OUT, not the ANTIC_IN sets.
24666 (phi_translate_set): Adjust.
24667 (do_pre_regular_insertion): Likewise.
24668 (do_pre_partial_partial_insertion): Likewise.
24670 2018-03-08 Martin Liska <mliska@suse.cz>
24672 PR gcov-profile/84735
24673 * doc/gcov.texi: Document usage of profile files.
24674 * gcov-io.h: Document changes in the format.
24676 2018-03-08 Alexandre Oliva <aoliva@redhat.com>
24680 * dwarf2out.c (struct dw_line_info_table): Update comments for
24682 (FORCE_RESET_NEXT_VIEW): New.
24683 (FORCE_RESETTING_VIEW_P): New.
24684 (RESETTING_VIEW_P): Check for -1 too.
24685 (ZERO_VIEW_P): Likewise.
24686 (new_line_info_table): Force-reset next view.
24687 (dwarf2out_begin_function): Likewise.
24688 (dwarf2out_source_line): Simplify zero_view_p initialization.
24689 Test FORCE_RESETTING_VIEW_P and RESETTING_VIEW_P instead of
24690 view directly. Omit view when omitting .loc at line 0.
24692 2018-03-08 Jakub Jelinek <jakub@redhat.com>
24694 PR tree-optimization/84740
24695 * tree-switch-conversion.c (process_switch): Call build_constructors
24696 only if info.phi_count is non-zero.
24698 PR tree-optimization/84739
24699 * tree-tailcall.c (find_tail_calls): Check call arguments against
24700 DECL_ARGUMENTS (current_function_decl) rather than
24701 DECL_ARGUMENTS (func) when checking for tail recursion.
24703 2018-03-07 Jakub Jelinek <jakub@redhat.com>
24705 * doc/contrib.texi: Add entries for Martin Liska, David Malcolm,
24706 Marek Polacek, extend Vladimir Makarov's, Jonathan Wakely's and
24707 Volker Reichelt's entry and add entries for people that perform
24708 GCC fuzzy testing and report numerous bugs.
24710 2018-03-07 Segher Boessenkool <segher@kernel.crashing.org>
24713 * config/rs6000/rs6000.c (rs6000_elf_in_small_data_p): Don't put
24714 readonly data in sdata, if that is disabled.
24715 * config/rs6000/sysv4.opt (mreadonly-in-sdata): New option.
24716 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
24717 -mreadonly-in-sdata option.
24719 2018-03-07 Martin Sebor <msebor@redhat.com>
24721 PR tree-optimization/84468
24722 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Consider successor
24723 basic block when looking for nul assignment.
24725 2018-03-07 Eric Botcazou <ebotcazou@adacore.com>
24728 * except.h (output_function_exception_table): Adjust prototype.
24729 * except.c (output_function_exception_table): Remove FNNAME parameter
24730 and add SECTION parameter. Ouput one part of the table at a time.
24731 * final.c (final_scan_insn_1) <NOTE_INSN_SWITCH_TEXT_SECTIONS>: Output
24732 the first part of the exception table and emit unwind directives.
24733 * config/i386/i386-protos.h (i386_pe_end_cold_function): Declare.
24734 (i386_pe_seh_cold_init): Likewise.
24735 * config/i386/cygming.h (ASM_DECLARE_COLD_FUNCTION_NAME): New macro.
24736 (ASM_DECLARE_COLD_FUNCTION_SIZE): Likewise.
24737 * config/i386/i386.c (x86_expand_epilogue): Fix wording in comment.
24738 (ix86_output_call_insn): Emit a nop in one more case for SEH.
24739 * config/i386/winnt.c: Include except.h.
24740 (struct seh_frame_state): Add reg_offset, after_prologue and
24741 in_cold_section fields.
24742 (i386_pe_seh_end_prologue): Set seh->after_prologue.
24743 (i386_pe_seh_cold_init): New function.
24744 (i386_pe_seh_fini): Add COLD parameter and bail out if it is not equal
24745 to seh->in_cold_section.
24746 (seh_emit_push): Record the offset of the push.
24747 (seh_emit_save): Record the offet of the save.
24748 (i386_pe_seh_unwind_emit): Deal with NOTE_INSN_SWITCH_TEXT_SECTIONS.
24749 Test seh->after_prologue to disregard the epilogue.
24750 (i386_pe_end_function): Pass FALSE to i386_pe_seh_fini.
24751 (i386_pe_end_cold_function): New function.
24753 2018-03-07 Jakub Jelinek <jakub@redhat.com>
24756 * config/aarch64/predicates.md (aarch64_simd_reg_or_zero): Use
24757 aarch64_simd_or_scalar_imm_zero rather than aarch64_simd_imm_zero.
24760 * gimple-expr.c (create_tmp_var_raw): Set DECL_NAMELESS flag
24762 * tree-pretty-print.c (dump_decl_name): For TDF_COMPARE_DEBUG,
24763 don't print names of DECL_NAMELESS DECL_IGNORED_P decls.
24765 PR middle-end/84723
24766 * multiple_target.c: Include tree-inline.h and intl.h.
24767 (expand_target_clones): Diagnose and fail if node->definition and
24768 !tree_versionable_function_p (node->decl).
24770 2018-03-06 John David Anglin <danglin@gcc.gnu.org>
24772 * config/pa/pa.h (ASM_GENERATE_INTERNAL_LABEL): Revise to use
24774 (ASM_OUTPUT_ADDR_VEC_ELT): Revise for above change.
24775 (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
24776 * config/pa/pa64-hpux.h (ASM_GENERATE_INTERNAL_LABEL): Revise as above.
24778 2018-03-06 Jakub Jelinek <jakub@redhat.com>
24781 * combine.c (try_combine): Use reg_or_subregno instead of handling
24782 just paradoxical SUBREGs and REGs.
24784 2018-03-06 Claudiu Zissulescu <claziss@synopsys.com>
24786 * config/arc/arc.c (arc_finalize_pic): Remove function.
24787 (arc_must_save_register): We use single base PIC register, remove
24788 checks to save/restore the PIC register.
24789 (arc_expand_prologue): Likewise.
24790 * config/arc/arc-protos.h (arc_set_default_type_attributes):
24792 (arc_verify_short): Likewise.
24793 (arc_attr_type): Likewise.
24794 * config/arc/arc.c (arc_set_default_type_attributes): Remove.
24795 (walk_stores): Likewise.
24796 (arc_address_cost): Make it static.
24797 (arc_verify_short): Likewise.
24798 (branch_dest): Likewise.
24799 (arc_attr_type): Likewise.
24800 * config/arc/arc.c (TARGET_ADJUST_INSN_LENGTH): Remove.
24801 (TARGET_INSN_LENGTH_PARAMETERS): Likewise.
24802 (arc_final_prescan_insn): Remove inserting the nops due to
24803 hardware hazards. It is done in reorg step.
24804 (insn_length_variant_t): Remove.
24805 (insn_length_parameters_t): Likewise.
24806 (arc_insn_length_parameters): Likewise.
24807 (arc_get_insn_variants): Likewise.
24808 * config/arc/arc.h (TARGET_UPSIZE_DBR): Remove.
24810 2018-03-06 Jakub Jelinek <jakub@redhat.com>
24812 PR inline-asm/84683
24813 * reg-stack.c (move_for_stack_reg): If any_malformed_asm, avoid
24816 PR tree-optimization/84687
24817 * omp-simd-clone.c (simd_clone_create): Clear DECL_BUILT_IN_CLASS
24819 * match.pd (pow(C,x)*expN(y) -> expN(logN(C)*x+y)): New optimization.
24821 2018-03-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24823 * config/rs6000/rs6000-builtin.def (rs6000_speculation_barrier):
24824 Rename to ppc_speculation_barrier.
24825 * config/rs6000/rs6000.c (rs6000_init_builtins): Rename builtin to
24826 __builtin_ppc_speculation_barrier.
24828 2018-03-05 Jakub Jelinek <jakub@redhat.com>
24831 * combine.c (combine_simplify_rtx): Don't try to simplify if
24832 if_then_else_cond returned non-NULL, but either true_rtx or false_rtx
24835 2018-03-05 Segher Boessenkool <segher@kernel.crashing.org>
24837 * config/rs6000/rs6000.c (rs6000_loop_align): Don't align tiny loops
24838 to 32 bytes when compiling for POWER9.
24840 2018-03-05 Jakub Jelinek <jakub@redhat.com>
24843 * config/i386/i386.c (ix86_function_ok_for_sibcall): Check for
24844 regparm >= 3 with no arg reg available also for calls with
24845 flag_force_indirect_call. Pass decl to ix86_function_regparm.
24848 * config/i386/sse.md (*<code><mode>3): Replace <mask_prefix3> with
24850 (*<plusminus_insn><mode>3): Likewise. Remove <mask_operand3> uses.
24852 2018-03-05 Peter Bergner <bergner@vnet.ibm.com>
24855 * config/rs6000/vector.md (mov<mode>): Disallow altivec memory operands.
24857 2018-03-05 Richard Biener <rguenther@suse.de>
24859 PR tree-optimization/84486
24860 * tree-ssa-pre.c (create_expression_by_pieces): Remove dead code.
24861 When inserting a __builtin_assume_aligned call set the LHS
24862 SSA name alignment info accordingly.
24864 2018-03-05 Wilco Dijkstra <wdijkstr@arm.com>
24866 PR tree-optimization/84114
24867 * config/aarch64/aarch64.c (aarch64_reassociation_width)
24868 Avoid reassociation of FLOAT_MODE addition.
24870 2018-03-05 Olga Makhotina <olga.makhotina@intel.com>
24872 * common/config/i386/i386-common.c (OPTION_MASK_ISA_PCONFIG_SET,
24873 OPTION_MASK_ISA_PCONFIG_UNSET, OPTION_MASK_ISA_WBNOINVD_SET,
24874 OPTION_MASK_ISA_WBNOINVD_UNSET): New definitions.
24875 (ix86_handle_option): Handle -mpconfig and -mwbnoinvd.
24876 * config.gcc (pconfigintrin.h, wbnoinvdintrin.h) : Add headers.
24877 * config/i386/cpuid.h (bit_PCONFIG, bit_WBNOINVD): New.
24878 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mpconfig
24880 * config/i386/i386-builtin.def (__builtin_ia32_wbnoinvd,
24881 __builtin_ia32_wbinvd): New builtins.
24882 (SPECIAL_ARGS2): New.
24883 * config/i386/i386-c.c (__WBNOINVD__, __PCONFIG__): New.
24884 (SPECIAL_ARGS2): New.
24885 * config/i386/i386.c (ix86_target_string): Add -mpconfig and -mwbnoinvd.
24886 (ix86_valid_target_attribute_inner_p): Ditto.
24887 (ix86_init_mmx_sse_builtins): Add special_args2.
24888 * config/i386/i386.h (TARGET_PCONFIG, TARGET_PCONFIG_P, TARGET_WBNOINVD,
24889 TARGET_WBNOINVD_P): New.
24890 * config/i386/i386.md (UNSPECV_WBINVD, UNSPECV_WBNOINVD): New.
24891 (define_insn "wbinvd", define_insn "wbnoinvd"): New.
24892 * config/i386/i386.opt: Add -mpconfig and -mwbnoinvd.
24893 * config/i386/immintrin.h (_wbinvd): New intrinsic.
24894 * config/i386/pconfigintrin.h: New file.
24895 * config/i386/wbnoinvdintrin.h: Ditto.
24896 * config/i386/x86intrin.h: Add headers pconfigintrin.h and
24898 * doc/invoke.texi (-mpconfig, -mwbnoinvd): New.
24900 2018-03-05 Richard Biener <rguenther@suse.de>
24902 PR tree-optimization/84670
24903 * tree-ssa-pre.c (struct bb_bitmap_sets): Add visited_with_visited_succs
24905 (BB_VISITED_WITH_VISITED_SUCCS): New define.
24906 (compute_antic): Initialize BB_VISITED_WITH_VISITED_SUCCS.
24907 (compute_antic_aux): Only assert the number of values in ANTIC_IN
24908 doesn't grow if all successors (recursively) were visited at least
24911 2018-03-05 Richard Biener <rguenther@suse.de>
24913 PR tree-optimization/84650
24914 * tree-ssa-loop-im.c (pass_lim::execute): Reset the SCEV cache
24915 if executed in the loop pipeline.
24917 2018-03-05 Sandra Loosemore <sandra@codesourcery.com>
24919 * doc/configfiles.texi (Configuration Files): Move info about
24920 conditionalizing $target-protos.h to...
24921 * doc/sourcebuild.texi (Back End): Here. Explain how $target.h
24922 differs from $target-protos.h.
24924 2018-03-05 Kito Cheng <kito.cheng@gmail.com>
24925 Chung-Ju Wu <jasonwucj@gmail.com>
24927 * config/nds32/nds32-protos.h (nds32_expand_setmem): Declare.
24928 * config/nds32/nds32-multiple.md (setmemsi): Define.
24929 * config/nds32/nds32-memory-manipulation.c
24930 (nds32_gen_dup_4_byte_to_word_value): New.
24931 (emit_setmem_word_loop): New.
24932 (emit_setmem_byte_loop): New.
24933 (nds32_expand_setmem_loop): New.
24934 (nds32_expand_setmem_loop_v3m): New.
24935 (nds32_expand_setmem_unroll): New.
24936 (nds32_expand_setmem): New.
24938 2018-03-04 Kito Cheng <kito.cheng@gmail.com>
24939 Chung-Ju Wu <jasonwucj@gmail.com>
24941 * config/nds32/nds32-memory-manipulation.c
24942 (nds32_emit_load_store): New.
24943 (nds32_emit_post_inc_load_store): New.
24944 (nds32_emit_mem_move): New.
24945 (nds32_emit_mem_move_block): New.
24946 (nds32_expand_movmemsi_loop_unknown_size): New.
24947 (nds32_expand_movmemsi_loop_known_size): New.
24948 (nds32_expand_movmemsi_loop): New.
24949 (nds32_expand_movmemsi_unroll): New.
24950 (nds32_expand_movmemqi): Rename ...
24951 (nds32_expand_movmemsi): ... to this.
24952 * config/nds32/nds32-multiple.md (movmemqi): Rename ...
24953 (movmemsi): ... to this.
24954 * config/nds32/nds32-protos.h (nds32_expand_movmemqi): Rename ...
24955 (nds32_expand_movmemsi): ... to this.
24957 2018-03-04 Kito Cheng <kito.cheng@gmail.com>
24958 Monk Chiang <sh.chiang04@gmail.com>
24959 Chung-Ju Wu <jasonwucj@gmail.com>
24961 * config/nds32/nds32-protos.h
24962 (nds32_expand_load_multiple): New arguments.
24963 (nds32_expand_store_multiple): Ditto.
24964 (nds32_valid_multiple_load_store): Rename ...
24965 (nds32_valid_multiple_load_store_p): ... to this.
24966 * config/nds32/nds32-memory-manipulation.c
24967 (nds32_expand_load_multiple): Refine implementation.
24968 (nds32_expand_store_multiple): Ditto.
24969 * config/nds32/nds32-multiple.md
24970 (load_multiple): Update nds32_expand_load_multiple interface.
24971 (store_multiple): Update nds32_expand_store_multiple interface.
24972 * config/nds32/nds32-predicates.c
24973 (nds32_valid_multiple_load_store): Rename ...
24974 (nds32_valid_multiple_load_store_p): ... to this and refine
24976 * config/nds32/predicates.md
24977 (nds32_load_multiple_and_update_address_operation): New predicate.
24978 (nds32_store_multiple_and_update_address_operation): New predicate.
24980 2018-03-04 Kito Cheng <kito.cheng@gmail.com>
24981 Chung-Ju Wu <jasonwucj@gmail.com>
24983 * config/nds32/nds32.md (type): Add load_multiple and store_multiple.
24984 (combo): New attribute.
24985 * config/nds32/nds32-multiple.md: Refine patterns with new attributes.
24987 2018-03-03 Chung-Ju Wu <jasonwucj@gmail.com>
24989 * config/nds32/nds32.opt: Change -mcmodel= default value.
24991 2018-03-03 Kito Cheng <kito.cheng@gmail.com>
24992 Monk Chiang <sh.chiang04@gmail.com>
24993 Chung-Ju Wu <jasonwucj@gmail.com>
24995 * config/nds32/constants.md (unspec_element): New enum.
24996 * config/nds32/constraints.md (Umw): New constraint.
24997 * config/nds32/nds32-intrinsic.c: Add more builtin functions.
24998 * config/nds32/nds32-intrinsic.md: Likewise.
24999 * config/nds32/nds32-md-auxiliary.c (nds32_regno_to_enable4): New.
25000 (nds32_valid_smw_lwm_base_p): New.
25001 (nds32_output_smw_single_word): New.
25002 (nds32_output_lmw_single_word): New.
25003 (nds32_expand_unaligned_load): New.
25004 (nds32_expand_unaligned_store): New.
25005 * config/nds32/nds32-protos.h (nds32_valid_smw_lwm_base_p): Declare.
25006 (nds32_output_smw_single_word): Declare.
25007 (nds32_output_lmw_single_word): Declare.
25008 (nds32_expand_unaligned_load): Declare.
25009 (nds32_expand_unaligned_store): Declare.
25010 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_UALOAD_HW,
25011 NDS32_BUILTIN_UALOAD_W, NDS32_BUILTIN_UALOAD_DW,
25012 NDS32_BUILTIN_UASTORE_HW, NDS32_BUILTIN_UASTORE_W,
25013 NDS32_BUILTIN_UASTORE_DW.
25014 * config/nds32/predicates.md (nds32_lmw_smw_base_operand): New
25017 2018-03-03 Monk Chiang <sh.chiang04@gmail.com>
25018 Kito Cheng <kito.cheng@gmail.com>
25019 Chung-Ju Wu <jasonwucj@gmail.com>
25021 * config/nds32/nds32-intrinsic.c
25022 (nds32_expand_builtin_null_ftype_reg): Delete.
25023 (nds32_expand_builtin_reg_ftype_imm): Ditto.
25024 (nds32_expand_builtin_null_ftype_reg_imm): Ditto.
25025 (nds32_read_argument): New.
25026 (nds32_legitimize_target): Ditto.
25027 (nds32_legitimize_argument): Ditto.
25028 (nds32_check_constant_argument): Ditto.
25029 (nds32_expand_unop_builtin): Ditto.
25030 (nds32_expand_unopimm_builtin): Ditto.
25031 (nds32_expand_binop_builtin): Ditto.
25032 (nds32_builtin_decl_impl): Ditto.
25033 (builtin_description): Ditto.
25034 (nds32_expand_builtin_impl): Rewrite with new infrastructure.
25035 (nds32_init_builtins_impl): Ditto.
25036 * config/nds32/nds32.c (TARGET_BUILTIN_DECL): Define.
25037 (nds32_builtin_decl): New.
25038 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_COUNT.
25039 * config/nds32/nds32-protos.h (nds32_builtin_decl_impl): Declare.
25041 2018-03-02 Jeff Law <law@redhat.com>
25043 * reorg.c (stop_search_p): Handle DEBUG_INSN.
25044 (redundant_insn, fill_simple_delay_slots): Likewise.
25045 (fill_slots_from_thread): Likewise.
25046 * resource.c (mark_referenced_resources): Likewise.
25047 (mark_set_resources, find_dead_or_set_registers): Likewise.
25049 2018-03-02 Jakub Jelinek <jakub@redhat.com>
25051 * substring-locations.h (format_warning_va): Formatting fix for
25052 ATTRIBUTE_GCC_DIAG.
25053 (format_warning_at_substring): Fix up ATTRIBUTE_GCC_DIAG second
25055 (format_warning_n_va, format_warning_at_substring_n): New prototypes.
25056 * substring-locations.c: Include intl.h.
25057 (format_warning_va): Turned into small wrapper around
25058 format_warning_n_va, renamed to ...
25059 (format_warning_n_va): ... this, add N and PLURAL_GMSGID arguments,
25060 rename GMSGID to SINGULAR_GMSGID, if SINGULAR_GMSGID != PLURAL_GMSGID,
25062 (format_warning_at_substring_n): New function.
25063 * gimple-ssa-sprintf.c: Remove GCC diagnostic ignored pragma.
25064 (fmtwarn): Add ATTRIBUTE_GCC_DIAG. Turn into a copy of
25065 format_warning_at_substring with just a shorter name instead of
25066 const function pointer.
25067 (fmtwarn_n): New function.
25068 (maybe_warn, format_directive, parse_directive): Use fmtwarn_n where
25069 appropriate, get rid of all the fmtstr temporaries, move conditionals
25070 with G_() wrapped string literals directly into fmtwarn arguments,
25071 cast dir.len to (int), formatting fixes.
25073 2018-03-02 Thomas Schwinge <thomas@codesourcery.com>
25075 * doc/invoke.texi: Remove "Cilk Plus" references.
25077 2018-03-02 Jakub Jelinek <jakub@redhat.com>
25078 Richard Biener <rguenther@suse.de>
25081 * expr.c (expand_expr_real_1) <case CALL_EXPR>: Don't emit diagnostics
25082 for error or warning attributes if CALL_FROM_THUNK_P is set.
25085 2018-03-02 Jakub Jelinek <jakub@redhat.com>
25088 * config/pa/pa.h (TARGET_CPU_CPP_BUILTINS): Predefine
25089 __SIZEOF_128__ macro if HPUX_LONG_DOUBLE_LIBRARY.
25092 * config/ia64/ia64.h (TARGET_CPU_CPP_BUILTINS): Predefine
25093 __SIZEOF_{FPREG,FLOAT{80,128}}__ macros.
25095 * predict.c (test_prediction_value_range): Use PROB_UNINITIALIZED
25096 instead of -1U in last predictors element's probability member.
25098 2018-03-02 Eric Botcazou <ebotcazou@adacore.com>
25101 * ipa-devirt.c (odr_subtypes_equivalent_p): Get the ODR type of both
25102 arguments if they are comparable.
25104 2018-03-02 Richard Sandiford <richard.sandiford@linaro.org>
25106 PR tree-optimization/84634
25107 * tree-vect-stmts.c (vectorizable_store, vectorizable_load): Replace
25108 masks and masked_loop_p with a single loop_masks, making sure it's
25109 null for bb vectorization.
25111 2018-03-02 Richard Sandiford <richard.sandiford@linaro.org>
25113 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence)
25114 (vect_analyze_data_ref_access): Use loop->safe_len rather than
25115 loop->force_vectorize to check whether there is no alias.
25117 2018-03-02 Jakub Jelinek <jakub@redhat.com>
25120 * rtl.h (prev_real_nondebug_insn, next_real_nondebug_insn): New
25122 * emit-rtl.c (next_real_insn, prev_real_insn): Fix up function
25124 (next_real_nondebug_insn, prev_real_nondebug_insn): New functions.
25125 * cfgcleanup.c (try_head_merge_bb): Use prev_real_nondebug_insn
25126 instead of a loop around prev_real_insn.
25127 * combine.c (move_deaths): Use prev_real_nondebug_insn instead of
25130 PR inline-asm/84625
25131 * config/i386/i386.c (ix86_print_operand): Use conditional
25132 output_operand_lossage instead of gcc_assert if CONST_VECTOR is not
25135 2018-03-02 Richard Biener <rguenther@suse.de>
25137 PR tree-optimization/84427
25138 * tree-ssa-pre.c (bitmap_remove_expr_from_set): Remove.
25139 (bitmap_set_subtract_values): Rewrite to handle multiple
25142 (prune_clobbered_mems): Likewise.
25143 (phi_translate): Take edge instead of pred/phiblock.
25144 (phi_translate_1): Likewise.
25145 (phi_translate_set): Likewise. Insert all translated
25146 exprs for a value into the set, keeping possibly multiple
25147 expressions per value.
25148 (compute_antic_aux): Adjust for phi_translate changes.
25149 When intersecting union the expressions and prune those
25150 not in the final value set, keeping possibly multiple
25151 expressions per value. Do not use value-insertion
25152 for unioning ANTIC_OUT U EXP_GEN - TMP_GEN but merge
25153 all expressions. Add verification that the value-sets
25154 only shrink during iteration.
25155 (compute_partial_antic_aux): Adjust for the phi_translate changes.
25156 (do_pre_regular_insertion): Likewise.
25157 (do_pre_partial_partial_insertion): Likewise.
25159 2018-03-02 Richard Biener <rguenther@suse.de>
25162 * config/darwin.c (saved_debug_info_level): New static global.
25163 (darwin_asm_lto_start): Disable debug info generation for LTO out.
25164 (darwin_asm_lto_end): Restore debug info generation settings.
25166 2018-03-01 Martin Liska <mliska@suse.cz>
25169 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
25170 volatile arguments.
25172 2018-03-01 Richard Biener <rguenther@suse.de>
25175 * dwarf2out.c (gen_variable_die): Properly handle late VLA
25176 type annotation with LTO when debug was disabled at compile-time.
25178 2018-03-01 Matthew Fortune <mfortune@gmail.com>
25180 * config/mips/mips.c (mips_final_prescan_insn): Fix incorrect
25182 (mips_final_postscan_insn): Likewise.
25184 2018-03-01 Richard Sandiford <richard.sandiford@linaro.org>
25186 PR rtl-optimization/84528
25187 * alias.c (init_alias_target): Add commentary.
25188 (init_alias_analysis): Only give HARD_FRAME_POINTER_REGNUM
25189 a unique base value if the frame pointer is not eliminated
25190 to the stack pointer.
25192 2018-03-01 Tom de Vries <tom@codesourcery.com>
25194 PR rtl-optimization/83327
25195 * lra-int.h (hard_regs_spilled_into): Declare.
25196 * lra.c (hard_regs_spilled_into): Define.
25197 (init_reg_info): Init hard_regs_spilled_into.
25198 * lra-spills.c (assign_spill_hard_regs): Update hard_regs_spilled_into.
25199 * lra-lives.c (make_hard_regno_born, make_hard_regno_dead)
25200 (process_bb_lives): Handle hard_regs_spilled_into.
25201 (lra_create_live_ranges_1): Before doing liveness propagation, clear
25202 regs in all_hard_regs_bitmap if set in hard_regs_spilled_into.
25204 2018-02-28 David Edelsohn <dje.gcc@gmail.com>
25206 * config.gcc (powerpc-ibm-aix7.1.*): New stanza.
25207 (powerpc-ibm-aix[789]*): Default to AIX 7.2.
25208 * config/rs6000/aix71.h (TARGET_DEFAULT): Revert to Power4 ISA.
25209 * config/rs6000/aix72.h: New file.
25211 2018-02-28 Jakub Jelinek <jakub@redhat.com>
25213 * gimple-ssa-warn-restrict.c (maybe_diag_overlap): Use warning_n
25214 instead of warning_at with conditional singular and plural messages
25218 * stor-layout.c (update_alignment_for_field): For
25219 targetm.ms_bitfield_layout_p (rli->t), if !is_bitfield
25220 && !DECL_PACKED (field), do the alignment update, just use
25221 only desired_align instead of MAX (type_align, desired_align)
25223 (place_field): Don't do known_align < desired_align handling
25224 early if targetm.ms_bitfield_layout_p (rli->t) and rli->prev_field
25225 is non-NULL, instead do it after rli->prev_field handling and
25226 only if not within a bitfield word. For DECL_PACKED (field)
25227 use type_align of BITS_PER_UNIT.
25229 2018-02-28 Eric Botcazou <ebotcazou@adacore.com>
25231 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Remove
25232 superfluous parentheses and trailing spaces.
25234 2018-02-28 Richard Biener <rguenther@suse.de>
25236 PR tree-optimization/84584
25237 * graphite-scop-detection.c (scop_detection::add_scop): Discard
25238 SCoPs with fake exit edge.
25240 2018-02-28 Martin Liska <mliska@suse.cz>
25243 * timevar.c (timer::print): Fix format to properly print 100%
25246 2018-02-28 Richard Biener <rguenther@suse.de>
25248 PR middle-end/84607
25249 * genmatch.c (capture_info::walk_match): Do not mark
25250 captured expressions without operands as expr_p given
25251 they act more like predicates and should be subject to
25252 "lost tail" side-effect preserving.
25254 2018-02-28 Alexandre Oliva <aoliva@redhat.com>
25256 PR rtl-optimization/81611
25257 * auto-inc-dec.c (attempt_change): Move dead note from
25258 mem_insn if it's the next use of regno
25259 (find_address): Take address use of reg holding
25260 non-incremented value. Add parm to limit search to the named
25262 (merge_in_block): Attempt to use a mem insn that is the next
25263 use of the original regno.
25265 2018-02-27 Martin Sebor <msebor@redhat.com>
25268 * doc/invoke.texi (-Wmissing-attributes): New option.
25269 * print-tree.c (print_node): Handle DECL_UNINLINABLE.
25271 2018-02-27 Martin Sebor <msebor@redhat.com>
25273 PR translation/84207
25274 * diagnostic-core.h (warning_n, error_n, inform_n): Change
25275 n argument to unsigned HOST_WIDE_INT.
25276 * diagnostic.c (warning_n, error_n, inform_n): Ditto.
25277 (diagnostic_n_impl): Ditto. Handle arguments in excess of LONG_MAX.
25278 * gimple-ssa-sprintf.c (format_directive): Simplify inform_n call.
25279 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Use warning_n.
25281 2018-02-27 Richard Biener <rguenther@suse.de>
25283 PR tree-optimization/84512
25284 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost):
25285 Do not use the estimate returned from record_stmt_cost for
25286 the scalar iteration cost but sum properly using add_stmt_cost.
25288 2018-02-27 Richard Biener <rguenther@suse.de>
25290 PR tree-optimization/84466
25291 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
25292 Adjust last change to less strictly validate use operands.
25294 2018-02-27 Martin Liska <mliska@suse.cz>
25296 PR gcov-profile/84548
25297 * gcov.c (process_file): Allow partial overlap and consider it
25298 also as group functions.
25299 (output_lines): Properly calculate range of lines for a group.
25301 2018-02-27 Martin Liska <mliska@suse.cz>
25303 * timevar.c (timer::print_row): Remove 'usr', 'sys', 'wall' and
25304 'ggc' suffixes. Change first column width.
25305 (timer::print): Fix formatting of the column.
25307 2018-02-27 Alexandre Oliva <aoliva@redhat.com>
25309 * tree-ssa-live.c (remove_unused_scope_block_p): Do not
25310 preserve inline entry blocks for the sake of debug inline
25311 entry point markers alone.
25312 (remove_unused_locals): Suggest in comments a better place to
25313 force the preservation of inline entry blocks that are
25314 otherwise unused, but do not preserve them.
25316 2018-02-26 H.J. Lu <hongjiu.lu@intel.com>
25318 * config/i386/i386.c (ix86_output_indirect_jmp): Update comments.
25320 2018-02-26 H.J. Lu <hongjiu.lu@intel.com>
25323 * config/i386/constraints.md (Bs): Replace
25324 ix86_indirect_branch_register with
25325 TARGET_INDIRECT_BRANCH_REGISTER.
25327 * config/i386/i386.md (indirect_jump): Likewise.
25328 (tablejump): Likewise.
25329 (*sibcall_memory): Likewise.
25330 (*sibcall_value_memory): Likewise.
25331 Peepholes of indirect call and jump via memory: Likewise.
25332 (*sibcall_GOT_32): Disallowed for TARGET_INDIRECT_BRANCH_REGISTER.
25333 (*sibcall_value_GOT_32): Likewise.
25334 * config/i386/predicates.md (indirect_branch_operand): Likewise.
25335 (GOT_memory_operand): Likewise.
25336 (call_insn_operand): Likewise.
25337 (sibcall_insn_operand): Likewise.
25338 (GOT32_symbol_operand): Likewise.
25339 * config/i386/i386.h (TARGET_INDIRECT_BRANCH_REGISTER): New.
25341 2018-02-26 Eric Botcazou <ebotcazou@adacore.com>
25343 PR rtl-optimization/83496
25344 * reorg.c (steal_delay_list_from_target): Change REDUNDANT array from
25345 booleans to RTXes. Call fix_reg_dead_note on every non-null element.
25346 (steal_delay_list_from_fallthrough): Call fix_reg_dead_note on a
25347 redundant insn, if any.
25348 (relax_delay_slots): Likewise.
25349 (update_reg_unused_notes): Rename REDUNDANT_INSN to OTHER_INSN.
25351 2018-02-26 Richard Sandiford <richard.sandiford@linaro.org>
25353 PR tree-optimization/83965
25354 * tree-vect-patterns.c (vect_reassociating_reduction_p): Assume
25355 that grouped statements are part of a reduction chain. Return
25356 true if the statement is not marked as a reduction itself but
25357 is part of a group.
25358 (vect_recog_dot_prod_pattern): Don't check whether the statement
25359 is part of a group here.
25360 (vect_recog_sad_pattern): Likewise.
25361 (vect_recog_widen_sum_pattern): Likewise.
25363 2018-02-26 Eric Botcazou <ebotcazou@adacore.com>
25366 * final.c (rest_of_clean_state): Also look for calls inside sequences.
25368 2018-02-26 H.J. Lu <hongjiu.lu@intel.com>
25371 * config/i386/i386-protos.h (ix86_output_indirect_jmp): Remove
25373 (ix86_output_indirect_function_return): New prototype.
25374 (ix86_split_simple_return_pop_internal): Likewise.
25375 * config/i386/i386.c (indirect_return_via_cx): New.
25376 (indirect_return_via_cx_bnd): Likewise.
25377 (indirect_thunk_name): Handle return va CX_REG.
25378 (output_indirect_thunk_function): Create alias for
25379 __x86_return_thunk_[re]cx and __x86_return_thunk_[re]cx_bnd.
25380 (ix86_output_indirect_jmp): Remove the bool argument.
25381 (ix86_output_indirect_function_return): New function.
25382 (ix86_split_simple_return_pop_internal): Likewise.
25383 * config/i386/i386.md (*indirect_jump): Don't pass false
25384 to ix86_output_indirect_jmp.
25385 (*tablejump_1): Likewise.
25386 (simple_return_pop_internal): Change it to define_insn_and_split.
25387 Call ix86_split_simple_return_pop_internal to split it for
25388 -mfunction-return=.
25389 (simple_return_indirect_internal): Call
25390 ix86_output_indirect_function_return instead of
25391 ix86_output_indirect_jmp.
25393 2018-02-26 Jakub Jelinek <jakub@redhat.com>
25396 * vec.h (vec_default_construct): For BROKEN_VALUE_INITIALIZATION use
25397 memset and value initialization afterwards.
25399 2018-02-26 Christophe Lyon <christophe.lyon@linaro.org>
25401 * Makefile.in (lto-wrapper): Use ALL_LINKERFLAGS.
25403 2018-02-26 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
25406 * common/config/aarch64/aarch64-common.c
25407 (aarch_option_optimization_table[]): Switch
25408 off fomit-frame-pointer
25410 2018-02-26 Kito Cheng <kito.cheng@gmail.com>
25411 Chung-Ju Wu <jasonwucj@gmail.com>
25413 * config/nds32/nds32-multiple.md (load_multiple): Disallow
25415 (store_multiple): Ditto.
25417 2018-02-26 Kito Cheng <kito.cheng@gmail.com>
25419 * config.gcc: Add --with-cpu support for nds32 target.
25420 * config/nds32/nds32-opts.h (nds32_cpu_type): New.
25421 * config/nds32/nds32.opt: Add -mcpu= option.
25423 2018-02-25 Segher Boessenkool <segher@kernel.crashing.org>
25425 * config/rs6000/rs6000.opt (mvrsave=no, mvrsave=yes, isel=no,
25426 isel=yes): Warn for these deprecated options.
25428 2018-02-23 David Edelsohn <dje.gcc@gmail.com>
25430 * config/rs6000/aix71.h (TARGET_DEFAULT): Change to
25431 ISA_2_5_MASKS_EMBEDDED.
25433 2018-02-23 Jakub Jelinek <jakub@redhat.com>
25435 * ipa-prop.c (ipa_vr_ggc_hash_traits::hash): Hash p->min and
25436 p->max as pointers rather than using iterative_hash_expr.
25438 2018-02-23 Carl Love <cel@us.ibm.com>
25440 * config/rs6000/rs6000-builtin.def: Change VSIGNED2 and VUNSIGNED2
25441 macro expansions from BU_VSX_2 to BU_P8V_VSX_2 and BU_VSX_OVERLOAD_2 to
25443 * config/rs6000/rs6000-c.c: Change VSX_BUILTIN_VEC_VSIGNED2 to
25444 P8V_BUILTIN_VEC_VSIGNED2. Change VSX_BUILTIN_VEC_VUNSIGNED2 to
25445 P8V_BUILTIN_VEC_VUNSIGNED2.
25447 2018-02-22 Vladimir Makarov <vmakarov@redhat.com>
25450 * lra-int.h (LRA_UNKNOWN_ALT, LRA_NON_CLOBBERED_ALT): New macros.
25451 * lra.c (lra_set_insn_recog_data, lra_update_insn_recog_data): Use
25453 * lra-constraints.c (curr_insn_transform): Set up
25454 LRA_NON_CLOBBERED_ALT for moves processed on the fast path. Use
25456 (remove_inheritance_pseudos): Use LRA_UNKNOWN_ALT.
25457 * lra-eliminations.c (spill_pseudos): Ditto.
25458 (process_insn_for_elimination): Ditto.
25459 * lra-lives.c (reg_early_clobber_p): Use the new macros.
25460 * lra-spills.c (spill_pseudos): Use LRA_UNKNOWN_ALT and
25461 LRA_NON_CLOBBERED_ALT.
25463 2018-02-22 Martin Sebor <msebor@redhat.com>
25465 PR tree-optimization/84480
25466 * gimple-fold.c (gimple_fold_builtin_strcpy): Move warnings
25467 to maybe_diag_stxncpy_trunc. Call it.
25468 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Integrate warnings
25469 from gimple_fold_builtin_strcpy. Print inlining stack.
25470 (handle_builtin_stxncpy): Print inlining stack.
25471 * tree-ssa-strlen.h (maybe_diag_stxncpy_trunc): Declare.
25473 2018-02-22 H.J. Lu <hongjiu.lu@intel.com>
25476 * config/i386/i386.c (ix86_set_indirect_branch_type): Issue an
25477 error when -mindirect-branch=thunk-extern, -fcf-protection=branch
25478 and -fcheck-pointer-bounds are used together.
25479 (indirect_thunk_prefix): New enum.
25480 (indirect_thunk_need_prefix): New function.
25481 (indirect_thunk_name): Replace need_bnd_p with need_prefix. Use
25482 "_nt" instead of "_bnd" for NOTRACK prefix.
25483 (output_indirect_thunk): Replace need_bnd_p with need_prefix.
25484 (output_indirect_thunk_function): Likewise.
25486 (ix86_code_end): Update output_indirect_thunk_function calls.
25487 (ix86_output_indirect_branch_via_reg): Replace
25488 ix86_bnd_prefixed_insn_p with indirect_thunk_need_prefix.
25489 (ix86_output_indirect_branch_via_push): Likewise.
25490 (ix86_output_function_return): Likewise.
25491 * doc/invoke.texi: Document -mindirect-branch=thunk-extern is
25492 incompatible with -fcf-protection=branch and
25493 -fcheck-pointer-bounds.
25495 2018-02-22 Steve Ellcey <sellcey@cavium.com>
25498 * config/aarch64/aarch64.c (aarch64_print_address_internal):
25499 Change gcc_assert call to output_operand_lossage.
25501 2018-02-22 Steve Ellcey <sellcey@cavium.com>
25503 * doc/extend.texi (__builtin_extend_pointer): Document builtin.
25505 2018-02-22 DJ Delorie <dj@redhat.com>
25506 Sebastian Perta <sebastian.perta@renesas.com>
25507 Oleg Endo <olegendo@gcc.gnu.org>
25509 * config/rx/rx.c (rx_rtx_costs): New function.
25510 (TARGET_RTX_COSTS): Override to use rx_rtx_costs.
25512 2018-02-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
25514 * config/arm/t-multilib: Map Armv8-R to Armv7 multilibs.
25516 2018-02-22 Martin Liska <mliska@suse.cz>
25519 * common/config/arm/arm-common.c (arm_print_hint_for_cpu_option):
25520 Add "native" as a possible value.
25522 2018-02-22 Martin Liska <mliska@suse.cz>
25525 * config/i386/i386.c (ix86_option_override_internal):
25526 Add "native" as a possible value for -march and -mtune.
25528 2018-02-22 Jakub Jelinek <jakub@redhat.com>
25531 * stor-layout.c (finalize_type_size): Propagate TYPE_EMPTY_P flag
25532 to all type variants.
25534 PR tree-optimization/84503
25535 * gimple-ssa-store-merging.c (merged_store_group::merge_into): Compute
25536 width as info->bitpos + info->bitsize - start.
25537 (merged_store_group::merge_overlapping): Simplify width computation.
25538 (check_no_overlap): New function.
25539 (imm_store_chain_info::try_coalesce_bswap): Compute expected
25540 start + width and last_order of the group, fail if check_no_overlap
25542 (imm_store_chain_info::coalesce_immediate_stores): Don't merge info
25543 to group if check_no_overlap fails.
25545 2018-02-21 Segher Boessenkool <segher@kernel.crashing.org>
25547 * config/rs6000/altivec.md: Delete contraint arguments to
25548 define_expand, define_split, and define_peephole2, and in
25549 define_insn_and_split if always unused.
25550 * config/rs6000/darwin.md: Ditto.
25551 * config/rs6000/dfp.md: Ditto.
25552 * config/rs6000/rs6000.md: Ditto.
25553 * config/rs6000/sync.md: Ditto.
25554 * config/rs6000/vector.md: Ditto.
25555 * config/rs6000/vsx.md: Ditto.
25557 2018-02-21 Segher Boessenkool <segher@kernel.crashing.org>
25559 * config/rs6000/altivec.md: Write output control strings as braced
25560 blocks instead of double-quoted strings.
25561 * config/rs6000/darwin.md: Ditto.
25562 * config/rs6000/rs6000.md: Ditto.
25563 * config/rs6000/vector.md: Ditto.
25564 * config/rs6000/vsx.md: Ditto.
25566 2018-02-21 Jason Merrill <jason@redhat.com>
25568 PR c++/84314 - ICE with templates and fastcall attribute.
25569 * attribs.c (build_type_attribute_qual_variant): Remove assert.
25571 2018-02-21 Jan Hubicka <hubicka@ucw.cz>
25573 * ipa-cp.c (determine_versionability): Fix comment typos.
25575 2018-02-21 Jan Hubicka <hubicka@ucw.cz>
25578 * ipa-cp.c (determine_versionability): Do not version functions caling
25581 2018-02-21 Martin Liska <mliska@suse.cz>
25584 * config/aarch64/aarch64.c (aarch64_print_hint_for_core_or_arch):
25585 Add "native" as a possible value.
25586 * config/aarch64/aarch64.h (HAVE_LOCAL_CPU_DETECT): Define
25587 the macro when native cpu detection is available.
25589 2018-02-21 Martin Liska <mliska@suse.cz>
25592 * common/config/arm/arm-common.c (arm_print_hint_for_arch_option):
25593 Add "native" as a possible value.
25594 * config/arm/arm.h (HAVE_LOCAL_CPU_DETECT): Define the macro
25595 when native cpu detection is available.
25597 2018-02-21 Jakub Jelinek <jakub@redhat.com>
25598 Martin Sebor <msebor@redhat.com>
25600 PR tree-optimization/84478
25601 * gimple-fold.h (get_range_strlen): Add a bool argument defaulted to
25603 * gimple-fold.c (get_range_strlen): Make minlen const and assume it
25604 can't be NULL. Change FUZZY from bool to int, for 1 add PHI/COND_EXPR
25605 support which is conservatively correct, for 2 only stay conservative
25606 for maxlen. Formatting and comment capitalization fixes. Add STRICT
25607 argument to the 2 argument get_range_strlen, adjust 6 arg
25608 get_range_strlen caller and clear minmaxlen[0] and [1] if it returned
25610 (get_maxval_strlen): Adjust 6 arg get_range_strlen caller.
25611 (gimple_fold_builtin_strlen): Pass true as last argument to
25614 2018-02-20 Martin Sebor <msebor@redhat.com>
25616 PR middle-end/84095
25617 * gimple-ssa-warn-restrict.c (builtin_memref::extend_offset_range): New.
25618 (builtin_memref::set_base_and_offset): Same. Handle inner references.
25619 (builtin_memref::builtin_memref): Factor out parts into
25620 set_base_and_offset and call it.
25622 2018-02-20 Richard Sandiford <richard.sandiford@linaro.org>
25624 PR middle-end/84406
25625 * optabs-query.c (find_widening_optab_handler_and_mode): If from_mode
25626 is a scalar_int_mode, assert that to_mode is a scalar_int_mode with
25627 greater precision. If to_mode is a MODE_PARTIAL_INT, stop the
25628 search at the associated MODE_INT.
25630 2018-02-20 Jeff Law <law@redhat.com>
25632 PR middle-end/82123
25633 PR tree-optimization/81592
25634 PR middle-end/79257
25635 * gimple-ssa-sprintf.c (format_integer): Query EVRP range analyzer
25636 for range data rather than using global data.
25637 * gimple-ssa-sprintf.c (get_int_range): Query EVRP range analyzer for
25638 range data rather than using global data.
25639 * gimple-ssa-sprintf.c (get_int_range): Accept vr_values parameter
25640 pass it to children as needed.
25641 (struct directive::fmtresult): Similarly.
25642 (struct directive::set_width): Similarly.
25643 (struct directive::set_precision): Similarly.
25644 (format_integer, format_directive, parse_directive): Similarly.
25645 (format_none): Accept unnamed vr_values parameter.
25646 (format_percent, format_floating, format_character): Similarly.
25647 (format_string, format_plain): Similarly.
25648 * gimple-ssa-sprintf.c (sprintf_dom_walker::handle_gimple_call): Query
25649 the EVRP range analyzer for range data rather than using global data.
25650 * gimple-ssa-sprintf.c: Include alloc-pool.h, vr-values.h and
25651 gimple-ssa-evrp-analyze.h
25652 (class sprintf_dom_walker): Add after_dom_children member function.
25653 Add evrp_range_analyzer member.
25654 (sprintf_dom_walker::before_dom_children): Call into the EVRP
25655 range analyzer as needed.
25656 (sprintf_dom_walker::after_dom_children): New member function.
25657 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::enter): Do nothing
25659 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
25660 (evrp_range_analyzer::pop_to_marker): Likewise.
25662 2018-02-20 Richard Sandiford <richard.sandiford@linaro.org>
25664 PR tree-optimization/84419
25665 * internal-fn.c (expand_call_mem_ref): Create a TARGET_MEM_REF
25666 with the required type if its current type is compatible but
25669 2018-02-20 Jakub Jelinek <jakub@redhat.com>
25671 PR middle-end/82004
25672 * match.pd (pow(C,x) -> exp(log(C)*x)): Delay all folding until
25673 after vectorization.
25675 2018-02-20 Martin Liska <mliska@suse.cz>
25678 * config/aarch64/aarch64.c (aarch64_print_hint_for_core_or_arch): Print
25679 possible values if we don't have a hint.
25681 2018-02-20 Martin Liska <mliska@suse.cz>
25685 * final.c (shorten_branches): Build align_tab array with one
25687 * opts.c (finish_options): Add alignment option limit check.
25688 (MAX_CODE_ALIGN): Likewise.
25689 (MAX_CODE_ALIGN_VALUE): Likewise.
25690 * doc/invoke.texi: Document maximum allowed option value for
25691 all -falign-* options.
25693 2018-02-19 Jakub Jelinek <jakub@redhat.com>
25696 * reg-notes.def (REG_CALL_ARG_LOCATION): New reg note.
25697 * insn-notes.def (NOTE_INSN_CALL_ARG_LOCATION): Remove.
25698 * var-tracking.c (emit_note_insn_var_location): Remove all references
25699 to NOTE_INSN_CALL_ARG_LOCATION.
25700 (emit_notes_in_bb): Emit arguments as REG_CALL_ARG_LOCATION note on
25701 the CALL_INSN rather than separate NOTE_INSN_CALL_ARG_LOCATION note.
25702 Use copy_rtx_if_shared.
25703 * dwarf2out.c (gen_subprogram_die): Use XEXP with 0 instead of
25704 NOTE_VAR_LOCATION on ca_loc->call_arg_loc_note.
25705 (dwarf2out_var_location): Remove handling of
25706 NOTE_INSN_CALL_ARG_LOCATION, instead handle REG_CALL_ARG_LOCATION note
25708 * final.c (final_scan_insn): Remove all references to
25709 NOTE_INSN_CALL_ARG_LOCATION.
25710 (rest_of_clean_state): Likewise. Remove REG_CALL_ARG_LOCATION notes
25711 before dumping final insns.
25712 * except.c (emit_note_eh_region_end): Remove all references to
25713 NOTE_INSN_CALL_ARG_LOCATION.
25714 * config/alpha/alpha.c (alpha_pad_function_end): Likewise.
25715 * config/c6x/c6x.c (c6x_gen_bundles): Likewise.
25716 * config/arc/arc.c (hwloop_optimize): Likewise.
25717 * config/arm/arm.c (create_fix_barrier): Likewise.
25718 * config/s390/s390.c (s390_chunkify_start): Likewise.
25719 * config/sh/sh.c (find_barrier): Likewise.
25720 * config/i386/i386.c (rest_of_insert_endbranch,
25721 ix86_seh_fixup_eh_fallthru): Likewise.
25722 * config/xtensa/xtensa.c (hwloop_optimize): Likewise.
25723 * config/iq2000/iq2000.c (final_prescan_insn): Likewise.
25724 * config/frv/frv.c (frv_function_prologue): Likewise.
25725 * emit-rtl.c (try_split): Likewise. Copy over REG_CALL_ARG_LOCATION
25727 (note_outside_basic_block_p): Remove all references to
25728 NOTE_INSN_CALL_ARG_LOCATION.
25729 * gengtype.c (adjust_field_rtx_def): Likewise.
25730 * print-rtl.c (rtx_writer::print_rtx_operand_code_0, print_insn):
25732 * jump.c (cleanup_barriers, delete_related_insns): Likewise.
25733 * cfgrtl.c (force_nonfallthru_and_redirect): Likewise.
25736 * builtins.c (builtin_mathfn_code): Don't check if CALL_EXPR_FN (t)
25739 PR tree-optimization/84452
25740 * tree-vect-patterns.c (vect_recog_pow_pattern): Don't call
25741 expand_simd_clones if targetm.simd_clone.compute_vecsize_and_simdlen
25744 2018-02-19 Martin Liska <mliska@suse.cz>
25747 * passes.def: Put pass_sancov_O0 before pass_lower_switch with -O0.
25749 2018-02-19 Martin Liska <mliska@suse.cz>
25750 Richard Sandiford <richard.sandiford@linaro.org>
25752 PR tree-optimization/82491
25753 * gimple-fold.c (get_base_constructor): Make earlier bail out
25756 2018-02-19 Carl Love <cel@us.ibm.com>
25758 * config/rs6000/rs6000-builtin.def: Change NEG macro expansions from
25759 BU_ALTIVEC_A to BU_P8V_AV_1 and BU_ALTIVEC_OVERLOAD_1 to
25761 * config/rs6000/rs6000-c.c: Change ALTIVEC_BUILTIN_VEC_NEG to
25762 P8V_BUILTIN_VEC_NEG.
25764 2018-02-19 Sebastian Perta <sebastian.perta@renesas.com>
25766 * config/rl78/rl78.md (movdf): New define expand.
25768 2018-02-19 Martin Liska <mliska@suse.cz>
25771 * doc/invoke.texi: Fix typo.
25772 * params.def (PARAM_MAX_LOOP_HEADER_INSNS): Likewise.
25774 2018-02-18 Segher Boessenkool <segher@kernel.crashing.org>
25776 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
25777 handle rs6000_single_float and rs6000_double_float specially for
25780 2018-02-16 Jeff Law <law@redhat.com>
25782 * config/rx/rx.c (add_pop_cfi_notes): New function.;
25783 (pop_regs): Use it.
25785 2018-02-16 Jakub Jelinek <jakub@redhat.com>
25788 * ipa-inline.c (inline_small_functions): Fix a typo.
25790 2018-02-16 Nathan Sidwell <nathan@acm.org>
25792 * doc/extend.texi (Backwards Compatibility): Americanize 'behaviour'.
25794 2018-02-16 Carl Love <cel@us.ibm.com>
25796 * config/rs6000/rs6000-builtin.def: Add BU_P8V_VSX_2 macro definition.
25797 Change FLOAT2 expansions from BU_VSX_2 to BU_P8V_VSX_2 and
25798 from BU_VSX_OVERLOAD_2 to BU_P8V_OVERLOAD_2.
25799 * config/rs6000/rs6000-c.c: Changed macro VSX_BUILTIN_VEC_FLOAT2
25800 expansion to P8V_BUILTIN_VEC_FLOAT2.
25802 2018-02-16 Vladimir Makarov <vmakarov@redhat.com>
25804 PR rtl-optimization/70023
25805 * lra-constraints.c (inherit_in_ebb): Take hard reg mode of
25806 src_regno into account.
25808 2018-02-16 Carl Love <cel@us.ibm.com>
25810 * config/rs6000/altivec.h: Remove vec_vextract4b and vec_vinsert4b.
25811 * config/rs6000/rs6000-builtin.def: Remove macro expansion for
25812 VEXTRACT4B, VINSERT4B, VINSERT4B_DI and VEXTRACT4B.
25813 * config/rs6000/rs6000.c: Remove case statements for
25814 P9V_BUILTIN_VEXTRACT4B, P9V_BUILTIN_VEC_VEXTRACT4B,
25815 P9V_BUILTIN_VINSERT4B, P9V_BUILTIN_VINSERT4B_DI,
25816 and P9V_BUILTIN_VEC_VINSERT4B.
25817 * config/rs6000/rs6000-c.c (altivec_expand_builtin): Remove entries for
25818 P9V_BUILTIN_VEC_VEXTRACT4B and P9V_BUILTIN_VEC_VINSERT4B.
25819 * config/rs6000/vsx.md:
25820 * doc/extend.texi: Remove vec_vextract4b, non ABI definitions for
25823 2018-02-16 Carl Love <cel@us.ibm.com>
25825 * config/rs6000/altivec.h: Add builtin names vec_extract4b
25827 * config/rs6000/rs6000-builtin.def: Add INSERT4B and EXTRACT4B
25829 * config/rs6000/rs6000-c.c: Add the definitions for
25830 P9V_BUILTIN_VEC_EXTRACT4B and P9V_BUILTIN_VEC_INSERT4B.
25831 * config/rs6000/rs6000.c (altivec_expand_builtin): Add
25832 P9V_BUILTIN_EXTRACT4B and P9V_BUILTIN_INSERT4B case statements.
25833 * config/rs6000/vsx.md: Add define_insn extract4b. Add define_expand
25834 definition for insert4b and define insn *insert3b_internal.
25835 * doc/extend.texi: Add documentation for vec_extract4b.
25837 2018-02-16 Nathan Sidwell <nathan@acm.org>
25839 * doc/extend.texi (Backwards Compatibility): Mention friend
25840 injection. Note for-scope is deprecated.
25841 * doc/invoke.texi (-ffriend-injection): Deprecate.
25843 2018-02-16 Segher Boessenkool <segher@kernel.crashing.org>
25845 * combine.c (try_combine): When adjusting LOG_LINKS for the destination
25846 that moved to I2, also allow destinations that are a paradoxical
25847 subreg (instead of a normal reg).
25849 2018-02-16 Oleg Endo <olegendo@gcc.gnu.org>
25852 * config/rx/rx.c (rx_fuse_in_memory_bitop): Convert shift operand
25855 2018-02-16 Richard Biener <rguenther@suse.de>
25857 PR tree-optimization/84037
25858 PR tree-optimization/84016
25860 * config/i386/i386.c (ix86_builtin_vectorization_cost):
25861 Adjust vec_construct for the fact we need additional higher latency
25862 128bit inserts for AVX256 and AVX512 vector builds.
25863 (ix86_add_stmt_cost): Scale vector construction cost for
25866 2018-02-16 Richard Biener <rguenther@suse.de>
25868 PR tree-optimization/84417
25869 * tree-ssa.c (non_rewritable_mem_ref_base): Properly constrain
25870 the MEM_REF offset when conversion to BIT_FIELD_REF is desired.
25871 (non_rewritable_lvalue_p): Likewise, use poly-ints.
25873 2018-02-16 Martin Liska <mliska@suse.cz>
25876 * internal-fn.def (ASAN_CHECK): Set proper flags.
25877 (ASAN_MARK): Likewise.
25879 2018-02-16 Julia Koval <julia.koval@intel.com>
25881 * config/i386/i386.c (ix86_option_override_internal): Remove PTA_CLWB
25882 from PTA_CANNONLAKE.
25884 2018-02-16 Jakub Jelinek <jakub@redhat.com>
25887 * config/aarch64/cortex-a57-fma-steering.c (fma_forest::merge_forest):
25888 Use ++iter rather than iter++ for std::list iterators.
25889 (func_fma_steering::dfs): Likewise. Don't delete nodes right away,
25890 defer deleting them until all nodes in the forest are processed. Do
25891 free even leaf nodes. Change to_process into auto_vec.
25894 * system.h (BROKEN_VALUE_INITIALIZATION): Define for GCC < 4.3.
25895 * vec.h (vec_default_construct): Use memset instead of placement new
25896 if BROKEN_VALUE_INITIALIZATION is defined.
25897 * hash-table.h (hash_table<Descriptor, Allocator>::empty_slow): Use
25898 memset instead of value initialization if BROKEN_VALUE_INITIALIZATION
25901 PR rtl-optimization/83723
25902 * lra-int.h (lra_substitute_pseudo): Add DEBUG_P argument.
25903 * lra.c (lra_substitute_pseudo): Likewise. If true, use
25904 gen_rtx_raw_SUBREG instead of gen_rtx_SUBREG. Pass DEBUG_P to
25906 (lra_substitute_pseudo_within_insn): Adjust lra_substitute_pseudo
25908 * lra-constraints.c (inherit_reload_reg, split_reg): Likewise.
25910 2018-02-16 Eric Botcazou <ebotcazou@adacore.com>
25912 PR rtl-optimization/81443
25913 * rtlanal.c (num_sign_bit_copies1) <SUBREG>: Do not propagate results
25914 from inner REGs to paradoxical SUBREGs.
25916 2018-02-16 Richard Biener <rguenther@suse.de>
25918 PR tree-optimization/84399
25919 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
25920 For operands we can analyze at their definition make sure we can
25921 analyze them at each use as well.
25923 2018-02-16 Richard Biener <rguenther@suse.de>
25925 PR tree-optimization/84190
25926 * tree-ssa.c (non_rewritable_mem_ref_base): Do not touch
25927 volatile accesses if the decl isn't volatile.
25929 2018-02-15 Jason Merrill <jason@redhat.com>
25931 PR c++/84314 - ICE with templates and fastcall attribute.
25932 * attribs.c (build_type_attribute_qual_variant): Don't clobber
25933 TYPE_CANONICAL on an existing type.
25935 2018-02-15 Jakub Jelinek <jakub@redhat.com>
25937 PR tree-optimization/84383
25938 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Don't look at
25939 dstoff nor call operand_equal_p if dstbase is NULL.
25941 PR tree-optimization/84334
25942 * match.pd ((A +- CST1) +- CST2 -> A + CST3): If A is
25943 also a CONSTANT_CLASS_P, punt.
25945 2018-02-14 Jim Wilson <jimw@sifive.com>
25947 * config/riscv/riscv.c (riscv_first_stack_step): Move locals after
25948 first SMALL_OPERAND check. New local min_second_step. Move assert
25949 to where locals are set. Add TARGET_RVC support.
25950 * config/riscv/riscv.h (C_SxSP_BITS, SWSP_REACH, SDSP_REACH): New.
25952 2018-02-14 Indu Bhagat <indu.bhagat@oracle.com>
25954 * doc/invoke.texi: Correct -Wformat-overflow code sample.
25956 2018-02-14 Martin Sebor <msebor@redhat.com>
25958 PR tree-optimization/83698
25959 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): For
25960 arrays constrain the offset range to their bounds.
25961 (builtin_access::strcat_overlap): Adjust the bounds of overlap offset.
25962 (builtin_access::overlap): Avoid setting the size of overlap if it's
25964 (maybe_diag_overlap): Also consider arrays when deciding what values
25965 of offsets to include in diagnostics.
25967 2018-02-14 Martin Sebor <msebor@redhat.com>
25970 * attribs.c (diag_attr_exclusions): Consider the exclusion(s)
25971 that correspond to the kind of a declaration.
25973 2018-02-14 John David Anglin <danglin@gcc.gnu.org>
25976 * config/pa/pa.md: Load address of PIC label using the linkage table
25977 if the label is nonlocal.
25979 2018-02-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
25981 * config/rs6000/rs6000.c (rs6000_option_override_internal): Issue
25982 warning message if user requests -maltivec=be.
25983 * doc/invoke.texi: Document deprecation of -maltivec=be.
25985 2018-02-14 Will Schmidt <will_schmidt@vnet.ibm.com>
25988 * config/rs6000/rs6000-c.c: Update definitions for
25989 ALTIVEC_BUILTIN_VEC_SLD, ALTIVEC_BUILTIN_VEC_SLDW,
25990 VEC_XXSLDWI and ALTIVEC_BUILTIN_VEC_XXPERMDI builtins.
25992 2018-02-14 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
25995 * config/i386/cetintrin.h: Remove _rdssp[d|q] and
25996 add _get_ssp intrinsics. Remove argument from
25997 __builtin_ia32_rdssp[d|q].
25998 * config/i386/i386-builtin-types.def: Add UINT_FTYPE_VOID.
25999 * config/i386/i386-builtin.def: Remove argument from
26000 __builtin_ia32_rdssp[d|q].
26001 * config/i386/i386.c: Use UINT_FTYPE_VOID. Use
26002 ix86_expand_special_args_builtin for _rdssp[d|q].
26003 * config/i386/i386.md: Remove argument from rdssp[si|di] insn.
26004 Clear register before usage.
26005 * doc/extend.texi: Remove argument from __builtin_ia32_rdssp[d|q].
26006 Add documentation for new _get_ssp and _inc_ssp intrinsics.
26008 2018-02-14 Richard Sandiford <richard.sandiford@linaro.org>
26010 PR tree-optimization/84357
26011 * tree-data-ref.c (object_address_invariant_in_loop_p): Check
26012 operand 1 of an ARRAY_REF too.
26014 2018-02-14 Oleg Endo <olegendo@gcc.gnu.org>
26017 * config/rx/rx-protos.h (rx_reg_dead_or_unused_after_insn,
26018 rx_copy_reg_dead_or_unused_notes, rx_fuse_in_memory_bitop): New
26020 (set_of_reg): New struct.
26021 (rx_find_set_of_reg, rx_find_use_of_reg): New functions.
26022 * config/rx/rx.c (rx_reg_dead_or_unused_after_insn,
26023 rx_copy_reg_dead_or_unused_notes, rx_fuse_in_memory_bitop): New
26025 * config/rx/rx.md (andsi3, iorsi3, xorsi3): Convert to insn_and_split.
26026 Split into bitclr, bitset, bitinvert patterns if appropriate.
26027 (*bitset, *bitinvert, *bitclr): Convert to named insn_and_split and
26028 use rx_fuse_in_memory_bitop.
26029 (*bitset_in_memory, *bitinvert_in_memory, *bitclr_in_memory): Convert
26030 to named insn, correct maximum insn length.
26032 2018-02-14 Jozef Lawrynowicz <jozefl.gcc@gmail.com>
26035 * machmode.def: Define a complex mode for PARTIAL_INT.
26036 * genmodes.c (complex_class): Return MODE_COMPLEX_INT for
26038 * doc/rtl.texi: Document CSPImode.
26039 * config/msp430/msp430.c (msp430_hard_regno_nregs): Add CPSImode
26041 (msp430_hard_regno_nregs_with_padding): Likewise.
26043 2018-02-13 Peter Bergner <bergner@vnet.ibm.com>
26046 * config/rs6000/rs6000.c (mem_operand_gpr): Disallow altivec addresses.
26048 2018-02-13 Segher Boessenkool <segher@kernel.crashing.org>
26050 PR rtl-optimization/84169
26051 * combine.c (try_combine): New variable split_i2i3. Set it to true if
26052 we generated a parallel as new i3 and we split that to new i2 and i3
26053 instructions. Handle split_i2i3 similar to swap_i2i3: scan the
26054 LOG_LINKs of i3 to see which of those need to link to i2 now. Link
26055 those to i2, not i1. Partially rewrite this scan code.
26057 2018-02-13 Jakub Jelinek <jakub@redhat.com>
26060 * stor-layout.c (place_field): For variable length fields, adjust
26061 offset_align afterwards not just based on the field's alignment,
26062 but also on the size.
26064 PR middle-end/84309
26065 * match.pd (pow(C,x) -> exp(log(C)*x)): Use exp2s and log2s instead
26066 of exps and logs in the use_exp2 case.
26068 2018-02-13 Jeff Law <law@redhat.com>
26070 * config/rl/rl78.c (rl78_attribute_table): Fix terminator and
26071 entry for "vector".
26073 * config/rl78/rl78.c (rl78_handle_func_attribute): Mark
26076 2018-02-13 Alexandre Oliva <aoliva@redhat.com>
26080 * common.opt (gas-loc-support, gas-locview-support): New.
26081 (ginline-points, ginternal-reset-location-views): New.
26082 * doc/invoke.texi: Document them. Use @itemx where intended.
26083 (gvariable-location-views): Adjust.
26084 * target.def (reset_location_view): New.
26085 * doc/tm.texi.in (DWARF2_ASM_VIEW_DEBUG_INFO): New.
26086 (TARGET_RESET_LOCATION_VIEW): New.
26087 * doc/tm.texi: Rebuilt.
26088 * dwarf2out.c (dwarf2out_default_as_loc_support): New.
26089 (dwarf2out_default_as_locview_support): New.
26090 (output_asm_line_debug_info): Use option variables.
26091 (dwarf2out_maybe_output_loclist_view_pair): Likewise.
26092 (output_loc_list): Likewise.
26093 (add_high_low_attributes): Check option variables.
26094 Don't output entry view attribute in strict mode.
26095 (gen_inlined_subroutine_die): Check option variables.
26096 (dwarf2out_inline_entry): Likewise.
26097 (init_sections_and_labels): Likewise.
26098 (dwarf2out_early_finish): Likewise.
26099 (maybe_reset_location_view): New, from...
26100 (dwarf2out_var_location): ... here. Call it.
26101 * debug.h (dwarf2out_default_as_loc_support): Declare.
26102 (dwarf2out_default_as_locview_support): Declare.
26103 * hooks.c (hook_int_rtx_insn_0): New.
26104 * hooks.h (hook_int_rtx_insn_0): Declare.
26105 * toplev.c (process_options): Take -gas-loc-support and
26106 -gas-locview-support from dwarf2out. Enable
26107 -gvariable-location-views by default only with locview
26108 assembler support. Enable -ginternal-reset-location-views by
26109 default only if the target defines the corresponding hook.
26110 Enable -ginline-points by default if location views are
26111 enabled; force it disabled if statement frontiers are
26113 * tree-inline.c (expand_call_inline): Check option variables.
26114 * tree-ssa-live.c (remove_unused_scope_block_p): Likewise.
26116 2018-02-13 Richard Sandiford <richard.sandiford@linaro.org>
26118 PR tree-optimization/84321
26119 * tree-vrp.c (intersect_range_with_nonzero_bits): Fix VR_ANTI_RANGE
26120 handling. Also check whether the anti-range contains any values
26121 that satisfy the mask; switch to a VR_RANGE if not.
26123 2018-02-13 Paolo Bonzini <bonzini@gnu.org>
26126 * internal-fn.def (ASAN_CHECK, ASAN_MARK): Revert changes to fnspec.
26128 2018-02-13 Martin Jambor <mjambor@suse.cz>
26131 * ipa-param-manipulation.c (ipa_modify_call_arguments): Use location
26132 of call statements, also set location of a load to a temporary.
26134 2018-02-13 Sebastian Perta <sebastian.perta@renesas.com>
26136 * config/rl78/rl78.c (add_vector_labels): New function.
26137 * config/rl78/rl78.c (rl78_handle_vector_attribute): New function.
26138 * config/rl78/rl78.c (rl78_start_function): Call add_vector_labels.
26139 * config/rl78/rl78.c (rl78_handle_func_attribute): Removed the assert
26140 which checks that no arguments are passed.
26141 * config/rl78/rl78.c (rl78_attribute_table): Add "vector" attribute.
26142 * doc/extend.texi: Documentation for the new attribute.
26144 2018-02-13 Andreas Schwab <schwab@suse.de>
26146 * config/riscv/linux.h (CPP_SPEC): Define.
26148 2018-02-13 Jakub Jelinek <jakub@redhat.com>
26151 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Pass
26152 OPTION_MASK_ISA_AES | OPTION_MASK_ISA_SSE2 instead of
26153 OPTION_MASK_ISA_AES as first argument to def_builtin_const
26154 for AES builtins. Pass OPTION_MASK_ISA_PCLMUL | OPTION_MASK_ISA_SSE2
26155 instead of OPTION_MASK_ISA_PCLMUL as first argument to
26156 def_builtin_const for __builtin_ia32_pclmulqdq128 builtin.
26157 * config/i386/wmmintrin.h: If __SSE2__ is not defined, enable it
26158 temporarily for AES and PCLMUL builtins.
26160 PR tree-optimization/84339
26161 * gimple-fold.c (get_range_strlen): Set *FLEXP to true when handling
26162 ARRAY_REF where first operand is array_at_struct_end_p COMPONENT_REF.
26165 PR middle-end/84309
26166 * match.pd (pow(C,x) -> exp(log(C)*x)): Optimize instead into
26167 exp2(log2(C)*x) if C is a power of 2 and c99 runtime is available.
26168 * generic-match-head.c (canonicalize_math_after_vectorization_p): New
26170 * gimple-match-head.c (canonicalize_math_after_vectorization_p): New
26172 * omp-simd-clone.h: New file.
26173 * omp-simd-clone.c: Include omp-simd-clone.h.
26174 (expand_simd_clones): No longer static.
26175 * tree-vect-patterns.c: Include fold-const-call.h, attribs.h,
26176 cgraph.h and omp-simd-clone.h.
26177 (vect_recog_pow_pattern): Optimize pow(C,x) to exp(log(C)*x).
26178 (vect_recog_widen_shift_pattern): Formatting fix.
26179 (vect_pattern_recog_1): Don't check optab for calls.
26182 * config/i386/sse.md (<avx512>_vpermi2var<mode>3_mask): Force
26183 operands[2] into a REG before using gen_lowpart on it.
26185 2018-02-12 Jeff Law <law@redhat.com>
26188 * config/sh/sh.c (find_barrier): Consider a sibling call
26191 * cse.c (try_back_substitute_reg): Move any REG_ARGS_SIZE note when
26192 successfully back substituting a reg.
26194 2018-02-12 Richard Biener <rguenther@suse.de>
26196 PR tree-optimization/84037
26197 * tree-vect-slp.c (vect_analyze_slp_cost): Add visited
26198 parameter, move visited init to caller.
26199 (vect_slp_analyze_operations): Separate cost from validity
26200 check, initialize visited once for all instances.
26201 (vect_schedule_slp): Analyze map to CSE vectorized nodes once
26203 * tree-vect-stmts.c (vect_model_simple_cost): Make early
26205 (vect_model_promotion_demotion_cost): Likewise.
26206 (vectorizable_bswap): Guard cost modeling with !slp_node
26207 instead of !PURE_SLP_STMT to avoid double-counting on hybrid
26209 (vectorizable_call): Likewise.
26210 (vectorizable_conversion): Likewise.
26211 (vectorizable_assignment): Likewise.
26212 (vectorizable_shift): Likewise.
26213 (vectorizable_operation): Likewise.
26214 (vectorizable_store): Likewise.
26215 (vectorizable_load): Likewise.
26216 (vectorizable_condition): Likewise.
26217 (vectorizable_comparison): Likewise.
26219 2018-02-12 Paolo Bonzini <bonzini@gnu.org>
26222 * internal-fn.def (ASAN_CHECK): Fix fnspec to account for return value.
26223 (ASAN_MARK): Fix fnspec to account for return value, change pointer
26224 argument from 'R' to 'W' so that the pointed-to datum is clobbered.
26226 2018-02-08 Jan Hubicka <hubicka@ucw.cz>
26228 PR middle-end/83665
26229 * params.def (inline-min-speedup): Increase from 8 to 15.
26230 (max-inline-insns-auto): Decrease from 40 to 30.
26231 * ipa-split.c (consider_split): Add some buffer for function to
26232 be considered inlining candidate.
26233 * invoke.texi (max-inline-insns-auto, inline-min-speedup): UPdate
26236 2018-02-12 Richard Biener <rguenther@suse.de>
26238 PR tree-optimization/84037
26239 * tree-vect-slp.c (vect_build_slp_tree_2): Try swapping the
26240 matched stmts if we cannot swap the non-matched ones.
26242 2018-02-12 Olga Makhotina <olga.makhotina@intel.com>
26244 * config/i386/avx512fintrin.h (_mm_mask_scalef_round_sd,
26245 _mm_maskz_scalef_round_sd, _mm_mask_scalef_round_ss,
26246 _mm_maskz_scalef_round_ss): New intrinsics.
26247 (__builtin_ia32_scalefsd_round, __builtin_ia32_scalefss_round): Fix.
26248 * config/i386/i386-builtin.def (__builtin_ia32_scalefsd_round,
26249 __builtin_ia32_scalefss_round): Remove.
26250 (__builtin_ia32_scalefsd_mask_round,
26251 __builtin_ia32_scalefss_mask_round): New intrinsics.
26252 * config/i386/sse.md (vmscalef<mode><round_name>): Renamed to ...
26253 (vmscalef<mode><mask_scalar_name><round_scalar_name>): ... this.
26254 ((match_operand:VF_128 2 "<round_nimm_predicate>"
26255 "<round_constraint>")): Changed to ...
26256 ((match_operand:VF_128 2 "<round_scalar_nimm_predicate>"
26257 "<round_scalar_constraint>")): ... this.
26258 ("vscalef<ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|
26259 %0, %1, %2<round_op3>}"): Changed to ...
26260 ("vscalef<ssescalarmodesuffix>\t{<round_scalar_mask_op3>%2, %1,
26261 %0<mask_scalar_operand3>|%0<mask_scalar_operand3>, %1,
26262 %2<round_scalar_mask_op3>}"): ... this.
26263 * config/i386/subst.md (round_scalar_nimm_predicate): New.
26265 2018-02-12 Olga Makhotina <olga.makhotina@intel.com>
26267 * config/i386/avx512fintrin.h (_mm_mask_sqrt_round_sd)
26268 (_mm_maskz_sqrt_round_sd, _mm_mask_sqrt_round_ss)
26269 (_mm_maskz_sqrt_round_ss): New intrinsics.
26270 (__builtin_ia32_sqrtsd_round, __builtin_ia32_sqrtss_round): Remove.
26271 (__builtin_ia32_sqrtsd_mask_round)
26272 (__builtin_ia32_sqrtss_mask_round): New builtins.
26273 * config/i386/i386-builtin.def (__builtin_ia32_sqrtsd_round)
26274 (__builtin_ia32_sqrtss_round): Remove.
26275 (__builtin_ia32_sqrtsd_mask_round)
26276 (__builtin_ia32_sqrtss_mask_round): New builtins.
26277 * config/i386/sse.md (vmsqrt<mode>2<round_name>): Renamed to ...
26278 (vmsqrt<mode>2<mask_scalar_name><round_scalar_name>): ... this.
26279 ((match_operand:VF_128 1 "vector_operand"
26280 "xBm,<round_constraint>")): Changed to ...
26281 ((match_operand:VF_128 1 "vector_operand"
26282 "xBm,<round_scalar_constraint>")): ... this.
26283 (vsqrt<ssescalarmodesuffix>\t{<round_op3>%1, %2, %0|
26284 %0, %2, %<iptr>1<round_op3>}): Changed to ...
26285 (vsqrt<ssescalarmodesuffix>\t{<round_scalar_mask_op3>%1, %2,
26286 %0<mask_scalar_operand3>|%0<mask_scalar_operand3>, %2,
26287 %<iptr>1<round_scalar_mask_op3>}): ... this.
26288 ((set_attr "prefix" "<round_prefix>")): Changed to ...
26289 ((set_attr "prefix" "<round_scalar_prefix>")): ... this.
26291 2018-02-11 Steven Munroe <munroesj@gcc.gnu.org>
26294 * config/rs6000/mmintrin.h (_mm_cmpeq_pi32 [_ARCH_PWR9]):
26295 Cast vec_cmpeq result to correct type.
26296 * config/rs6000/mmintrin.h (_mm_cmpgt_pi32 [_ARCH_PWR9]):
26297 Cast vec_cmpgt result to correct type.
26299 2018-02-11 Alexandre Oliva <aoliva@redhat.com>
26301 * final.c (final_scan_insn_1): Renamed from...
26302 (final_scan_insn): ... this. New wrapper, to recover
26303 seen from the outermost call in recursive ones.
26304 * config/sparc/sparc.c (output_return): Drop seen from call.
26305 (output_sibcall): Likewise.
26306 * config/visium/visium.c (output_branch): Likewise.
26308 2018-02-10 John David Anglin <danglin@gcc.gnu.org>
26310 * config/pa/pa.c (hppa_profile_hook): Mark SYMBOL_REF for _mcount as
26313 2018-02-10 Alan Modra <amodra@gmail.com>
26316 * config/rs6000/rs6000.md (split_stack_return): Remove (use ..).
26317 Specify LR as an input.
26319 2018-02-10 Jakub Jelinek <jakub@redhat.com>
26322 * omp-low.c (maybe_remove_omp_member_access_dummy_vars,
26323 remove_member_access_dummy_vars): New functions.
26324 (lower_omp_for, lower_omp_taskreg, lower_omp_target,
26325 lower_omp_1, execute_lower_omp): Use them.
26327 PR rtl-optimization/84308
26328 * shrink-wrap.c (spread_components): Release todo vector.
26330 2018-02-09 Vladimir Makarov <vmakarov@redhat.com>
26332 PR rtl-optimization/57193
26333 * ira-color.c (struct allocno_color_data): Add member
26334 conflict_allocno_hard_prefs.
26335 (update_conflict_allocno_hard_prefs): New.
26336 (bucket_allocno_compare_func): Add a preference based on
26337 conflict_allocno_hard_prefs.
26338 (push_allocno_to_stack): Update conflict_allocno_hard_prefs.
26339 (color_allocnos): Remove a dead code. Initiate
26340 conflict_allocno_hard_prefs. Call update_costs_from_prefs.
26342 2018-02-09 Jakub Jelinek <jakub@redhat.com>
26345 * config/rs6000/vsx.md (p9_xxbrq_v16qi): Change input operand
26346 constraint from =wa to wa. Avoid a subreg on the output operand,
26347 instead use a pseudo and subreg it in a move.
26348 (p9_xxbrd_<mode>): Changed to ...
26349 (p9_xxbrd_v2di): ... this insn, without VSX_D iterator.
26350 (p9_xxbrd_v2df): New expander.
26351 (p9_xxbrw_<mode>): Changed to ...
26352 (p9_xxbrw_v4si): ... this insn, without VSX_W iterator.
26353 (p9_xxbrw_v4sf): New expander.
26355 2018-02-09 Sebastian Perta <sebastian.perta@renesas.com>
26357 * config/rx/rx.md (movsicc): Update expander to be matched by GCC.
26359 2018-02-09 Peter Bergner <bergner@vnet.ibm.com>
26362 * config/rs6000/vsx.md (vsx_mul_v2di): Handle generating a 64-bit
26363 multiply in 32-bit mode.
26364 (vsx_div_v2di): Handle generating a 64-bit signed divide in 32-bit mode.
26365 (vsx_udiv_v2di): Handle generating a 64-bit unsigned divide in 32-bit
26368 2018-02-09 Sebastian Perta <sebastian.perta@renesas.com>
26370 * config/rx/constraints.md (CALL_OP_SYMBOL_REF): Added new constraint
26371 to allow or block "symbol_ref" depending on the value of TARGET_JSR.
26372 * config/rx/rx.md (call_internal): Use CALL_OP_SYMBOL_REF.
26373 * config/rx/rx.md (call_value_internal): Use CALL_OP_SYMBOL_REF.
26375 2018-02-09 Pierre-Marie de Rodat <derodat@adacore.com>
26378 * dwarf2out.c (is_trivial_indirect_ref): New function.
26379 (dwarf2out_late_global_decl): Do not generate a location
26380 attribute for variables that have a non-trivial DECL_VALUE_EXPR
26381 and that are not defined in the current unit.
26383 2018-02-09 Eric Botcazou <ebotcazou@adacore.com>
26385 * optabs.c (prepare_cmp_insn): Try harder to emit a direct comparison
26386 instead of a libcall for UNORDERED.
26388 2018-02-09 Tamar Christina <tamar.christina@arm.com>
26391 * config/arm/arm-c.c (arm_cpu_builtins): Un-define __ARM_FEATURE_LDREX,
26392 __ARM_ARCH_PROFILE, __ARM_ARCH_ISA_THUMB, __ARM_FP and __ARM_NEON_FP.
26394 2018-02-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26397 * config/s390/s390.c (s390_set_current_function): Invoke
26398 s390_indirect_branch_settings also if fndecl didn't change.
26400 2018-02-09 Alexandre Oliva <aoliva@redhat.com>
26402 * config/rs6000/rs6000.md (blockage): Set length to zero.
26404 2018-02-09 Eric Botcazou <ebotcazou@adacore.com>
26406 * expr.c (optimize_bitfield_assignment_op): Remove obsolete assertion.
26408 2018-02-09 Jakub Jelinek <jakub@redhat.com>
26411 * gcc.c (STATIC_LIBASAN_LIBS, STATIC_LIBTSAN_LIBS,
26412 STATIC_LIBLSAN_LIBS, STATIC_LIBUBSAN_LIBS): Handle -static like
26416 * var-tracking.c (vt_add_function_parameter): Punt for non-onepart
26417 PARALLEL incoming that failed vt_get_decl_and_offset check.
26419 PR middle-end/84237
26420 * output.h (bss_initializer_p): Add NAMED argument, defaulted to false.
26421 * varasm.c (bss_initializer_p): Add NAMED argument, if true, ignore
26423 (get_variable_section): For decls in named .bss* sections pass true as
26424 second argument to bss_initializer_p.
26426 2018-02-09 Marek Polacek <polacek@redhat.com>
26427 Jakub Jelinek <jakub@redhat.com>
26430 * fold-const.c (fold_indirect_ref_1): Use VECTOR_TYPE_P macro.
26431 Formatting fixes. Verify first that tree_fits_poly_int64_p (op01).
26432 Sync some changes from cxx_fold_indirect_ref.
26434 2018-02-09 Alexandre Oliva <aoliva@redhat.com>
26436 * cfgexpand.c (expand_gimple_basic_block): Handle inline entry
26438 * dwarf2out.c (dwarf2_debug_hooks): Enable inline_entry hook.
26439 (BLOCK_INLINE_ENTRY_LABEL): New.
26440 (dwarf2out_var_location): Disregard inline entry markers.
26441 (inline_entry_data): New struct.
26442 (inline_entry_data_hasher): New hashtable type.
26443 (inline_entry_data_hasher::hash): New.
26444 (inline_entry_data_hasher::equal): New.
26445 (inline_entry_data_table): New variable.
26446 (add_high_low_attributes): Add DW_AT_entry_pc and
26447 DW_AT_GNU_entry_view attributes if a pending entry is found
26448 in inline_entry_data_table. Add old entry_pc attribute only
26449 if debug nonbinding markers are disabled.
26450 (gen_inlined_subroutine_die): Set BLOCK_DIE if nonbinding
26451 markers are enabled.
26452 (block_within_block_p, dwarf2out_inline_entry): New.
26453 (dwarf2out_finish): Check that no entries remained in
26454 inline_entry_data_table.
26455 * final.c (reemit_insn_block_notes): Handle inline entry notes.
26456 (final_scan_insn, notice_source_line): Likewise.
26457 (rest_of_clean_state): Skip inline entry markers.
26458 * gimple-pretty-print.c (dump_gimple_debug): Handle inline entry
26460 * gimple.c (gimple_build_debug_inline_entry): New.
26461 * gimple.h (enum gimple_debug_subcode): Add
26462 GIMPLE_DEBUG_INLINE_ENTRY.
26463 (gimple_build_debug_inline_entry): Declare.
26464 (gimple_debug_inline_entry_p): New.
26465 (gimple_debug_nonbind_marker_p): Adjust.
26466 * insn-notes.def (INLINE_ENTRY): New.
26467 * print-rtl.c (rtx_writer::print_rtx_operand_code_0): Handle
26468 inline entry marker notes.
26469 (print_insn): Likewise.
26470 * rtl.h (NOTE_MARKER_P): Add INLINE_ENTRY support.
26471 (INSN_DEBUG_MARKER_KIND): Likewise.
26472 (GEN_RTX_DEBUG_MARKER_INLINE_ENTRY_PAT): New.
26473 * tree-inline.c (expand_call_inline): Build and insert
26474 debug_inline_entry stmt.
26475 * tree-ssa-live.c (remove_unused_scope_block_p): Preserve
26476 inline entry blocks early, if nonbind markers are enabled.
26477 (dump_scope_block): Dump fragment info.
26478 * var-tracking.c (reemit_marker_as_note): Handle inline entry note.
26479 * doc/gimple.texi (gimple_debug_inline_entry_p): New.
26480 (gimple_build_debug_inline_entry): New.
26481 * doc/invoke.texi (gstatement-frontiers, gno-statement-frontiers):
26482 Enable/disable inline entry points too.
26483 * doc/rtl.texi (NOTE_INSN_INLINE_ENTRY): New.
26484 (DEBUG_INSN): Describe inline entry markers.
26486 * common.opt (gvariable-location-views): New.
26487 (gvariable-location-views=incompat5): New.
26488 * config.in: Rebuilt.
26489 * configure: Rebuilt.
26490 * configure.ac: Test assembler for view support.
26491 * dwarf2asm.c (dw2_asm_output_symname_uleb128): New.
26492 * dwarf2asm.h (dw2_asm_output_symname_uleb128): Declare.
26493 * dwarf2out.c (var_loc_view): New typedef.
26494 (struct dw_loc_list_struct): Add vl_symbol, vbegin, vend.
26495 (dwarf2out_locviews_in_attribute): New.
26496 (dwarf2out_locviews_in_loclist): New.
26497 (dw_val_equal_p): Compare val_view_list of dw_val_class_view_lists.
26498 (enum dw_line_info_opcode): Add LI_adv_address.
26499 (struct dw_line_info_table): Add view.
26500 (RESET_NEXT_VIEW, RESETTING_VIEW_P): New macros.
26501 (DWARF2_ASM_VIEW_DEBUG_INFO): Define default.
26502 (zero_view_p): New variable.
26503 (ZERO_VIEW_P): New macro.
26504 (output_asm_line_debug_info): New.
26505 (struct var_loc_node): Add view.
26506 (add_AT_view_list, AT_loc_list): New.
26507 (add_var_loc_to_decl): Add view param. Test it against last.
26508 (new_loc_list): Add view params. Record them.
26509 (AT_loc_list_ptr): Handle loc and view lists.
26510 (view_list_to_loc_list_val_node): New.
26511 (print_dw_val): Handle dw_val_class_view_list.
26512 (size_of_die): Likewise.
26513 (value_format): Likewise.
26514 (loc_list_has_views): New.
26515 (gen_llsym): Set vl_symbol too.
26516 (maybe_gen_llsym, skip_loc_list_entry): New.
26517 (dwarf2out_maybe_output_loclist_view_pair): New.
26518 (output_loc_list): Output view list or entries too.
26519 (output_view_list_offset): New.
26520 (output_die): Handle dw_val_class_view_list.
26521 (output_dwarf_version): New.
26522 (output_compilation_unit_header): Use it.
26523 (output_skeleton_debug_sections): Likewise.
26524 (output_rnglists, output_line_info): Likewise.
26525 (output_pubnames, output_aranges): Update version comments.
26526 (output_one_line_info_table): Output view numbers in asm comments.
26527 (dw_loc_list): Determine current endview, pass it to new_loc_list.
26528 Call maybe_gen_llsym.
26529 (loc_list_from_tree_1): Adjust.
26530 (add_AT_location_description): Create view list attribute if
26531 needed, check it's absent otherwise.
26532 (convert_cfa_to_fb_loc_list): Adjust.
26533 (maybe_emit_file): Call output_asm_line_debug_info for test.
26534 (dwarf2out_var_location): Reset views as needed. Precompute
26535 add_var_loc_to_decl args. Call get_attr_min_length only if we have the
26536 attribute. Set view.
26537 (new_line_info_table): Reset next view.
26538 (set_cur_line_info_table): Call output_asm_line_debug_info for test.
26539 (dwarf2out_source_line): Likewise. Output view resets and labels to
26540 the assembler, or select appropriate line info opcodes.
26541 (prune_unused_types_walk_attribs): Handle dw_val_class_view_list.
26542 (optimize_string_length): Catch it. Adjust.
26543 (resolve_addr): Copy vl_symbol along with ll_symbol. Handle
26544 dw_val_class_view_list, and remove it if no longer needed.
26545 (hash_loc_list): Hash view numbers.
26546 (loc_list_hasher::equal): Compare them.
26547 (optimize_location_lists): Check whether a view list symbol is
26548 needed, and whether the locview attribute is present, and
26549 whether they match. Remove the locview attribute if no longer
26551 (index_location_lists): Call skip_loc_list_entry for test.
26552 (dwarf2out_finish): Call output_asm_line_debug_info for test.
26553 Use output_dwarf_version.
26554 * dwarf2out.h (enum dw_val_class): Add dw_val_class_view_list.
26555 (struct dw_val_node): Add val_view_list.
26556 * final.c (SEEN_NEXT_VIEW): New.
26557 (set_next_view_needed): New.
26558 (clear_next_view_needed): New.
26559 (maybe_output_next_view): New.
26560 (final_start_function): Rename to...
26561 (final_start_function_1): ... this. Take pointer to FIRST,
26562 add SEEN parameter. Emit param bindings in the initial view.
26563 (final_start_function): Reintroduce SEEN-less interface.
26564 (final): Rename to...
26565 (final_1): ... this. Take SEEN parameter. Output final pending
26566 next view at the end.
26567 (final): Reintroduce seen-less interface.
26568 (final_scan_insn): Output pending next view before switching
26569 sections or ending a block. Mark the next view as needed when
26570 outputting variable locations. Notify debug backend of section
26571 changes, and of location view changes.
26572 (rest_of_handle_final): Adjust.
26573 * toplev.c (process_options): Autodetect value for debug variable
26574 location views option. Warn on incompat5 without -gdwarf-5.
26575 * doc/invoke.texi (gvariable-location-views): New.
26576 (gvariable-location-views=incompat5): New.
26577 (gno-variable-location-views): New.
26579 2018-02-08 David Malcolm <dmalcolm@redhat.com>
26581 PR tree-optimization/84136
26582 * tree-cfg.c (find_taken_edge_computed_goto): Remove assertion
26583 that the result of find_edge is non-NULL.
26585 2018-02-08 Sergey Shalnov <sergey.shalnov@intel.com>
26588 * config/i386/x86-tune-costs.h (skylake_cost): Fix cost of
26589 storing integer register in SImode. Fix cost of 256 and 512
26590 byte aligned SSE register store.
26592 2018-02-08 Sergey Shalnov <sergey.shalnov@intel.com>
26594 * config/i386/i386.c (ix86_multiplication_cost): Fix
26595 multiplication cost for TARGET_AVX512DQ.
26597 2018-02-08 Marek Polacek <polacek@redhat.com>
26599 PR tree-optimization/84238
26600 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Verify the result of
26603 2018-02-08 Richard Sandiford <richard.sandiford@linaro.org>
26605 PR tree-optimization/84265
26606 * tree-vect-stmts.c (vectorizable_store): Don't treat
26607 VMAT_CONTIGUOUS accesses as grouped.
26608 (vectorizable_load): Likewise.
26610 2018-02-08 Richard Sandiford <richard.sandiford@linaro.org>
26612 PR tree-optimization/81635
26613 * wide-int.h (wi::round_down_for_mask, wi::round_up_for_mask): Declare.
26614 * wide-int.cc (wi::round_down_for_mask, wi::round_up_for_mask)
26615 (test_round_for_mask): New functions.
26616 (wide_int_cc_tests): Call test_round_for_mask.
26617 * tree-vrp.h (intersect_range_with_nonzero_bits): Declare.
26618 * tree-vrp.c (intersect_range_with_nonzero_bits): New function.
26619 * tree-data-ref.c (split_constant_offset_1): Use it to refine the
26620 range returned by get_range_info.
26622 2018-02-08 Jan Hubicka <hubicka@ucw.cz>
26625 * cgraph.h (symtab_node::output_to_lto_symbol_table_p): Declare
26626 * symtab.c: Include builtins.h
26627 (symtab_node::output_to_lto_symbol_table_p): Move here
26628 from lto-streamer-out.c:output_symbol_p.
26629 * lto-streamer-out.c (write_symbol): Turn early exit to assert.
26630 (output_symbol_p): Move all logic to symtab.c
26631 (produce_symtab): Update.
26633 2018-02-08 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26635 * config/s390/s390-opts.h (enum indirect_branch): Define.
26636 * config/s390/s390-protos.h (s390_return_addr_from_memory)
26637 (s390_indirect_branch_via_thunk)
26638 (s390_indirect_branch_via_inline_thunk): Add function prototypes.
26639 (enum s390_indirect_branch_type): Define.
26640 * config/s390/s390.c (struct s390_frame_layout, struct
26641 machine_function): Remove.
26642 (indirect_branch_prez10thunk_mask, indirect_branch_z10thunk_mask)
26643 (indirect_branch_table_label_no, indirect_branch_table_name):
26645 (INDIRECT_BRANCH_NUM_OPTIONS): Define macro.
26646 (enum s390_indirect_branch_option): Define.
26647 (s390_return_addr_from_memory): New function.
26648 (s390_handle_string_attribute): New function.
26649 (s390_attribute_table): Add new attribute handler.
26650 (s390_execute_label): Handle UNSPEC_EXECUTE_JUMP patterns.
26651 (s390_indirect_branch_via_thunk): New function.
26652 (s390_indirect_branch_via_inline_thunk): New function.
26653 (s390_function_ok_for_sibcall): When jumping via thunk disallow
26654 sibling call optimization for non z10 compiles.
26655 (s390_emit_call): Force indirect branch target to be a single
26656 register. Add r1 clobber for non-z10 compiles.
26657 (s390_emit_epilogue): Emit return jump via return_use expander.
26658 (s390_reorg): Handle JUMP_INSNs as execute targets.
26659 (s390_option_override_internal): Perform validity checks for the
26660 new command line options.
26661 (s390_indirect_branch_attrvalue): New function.
26662 (s390_indirect_branch_settings): New function.
26663 (s390_set_current_function): Invoke s390_indirect_branch_settings.
26664 (s390_output_indirect_thunk_function): New function.
26665 (s390_code_end): Implement target hook.
26666 (s390_case_values_threshold): Implement target hook.
26667 (TARGET_ASM_CODE_END, TARGET_CASE_VALUES_THRESHOLD): Define target
26669 * config/s390/s390.h (struct s390_frame_layout)
26670 (struct machine_function): Move here from s390.c.
26671 (TARGET_INDIRECT_BRANCH_NOBP_RET)
26672 (TARGET_INDIRECT_BRANCH_NOBP_JUMP)
26673 (TARGET_INDIRECT_BRANCH_NOBP_JUMP_THUNK)
26674 (TARGET_INDIRECT_BRANCH_NOBP_JUMP_INLINE_THUNK)
26675 (TARGET_INDIRECT_BRANCH_NOBP_CALL)
26676 (TARGET_DEFAULT_INDIRECT_BRANCH_TABLE)
26677 (TARGET_INDIRECT_BRANCH_THUNK_NAME_EXRL)
26678 (TARGET_INDIRECT_BRANCH_THUNK_NAME_EX)
26679 (TARGET_INDIRECT_BRANCH_TABLE): Define macros.
26680 * config/s390/s390.md (UNSPEC_EXECUTE_JUMP)
26681 (INDIRECT_BRANCH_THUNK_REGNUM): Define constants.
26682 (mnemonic attribute): Add values which aren't recognized
26684 ("*cjump_long", "*icjump_long", "*basr", "*basr_r"): Disable
26685 pattern for branch conversion. Fix mnemonic attribute.
26686 ("*c<code>", "*sibcall_br", "*sibcall_value_br", "*return"): Emit
26687 indirect branch via thunk if requested.
26688 ("indirect_jump", "<code>"): Expand patterns for branch conversion.
26689 ("*indirect_jump"): Disable for branch conversion using out of
26691 ("indirect_jump_via_thunk<mode>_z10")
26692 ("indirect_jump_via_thunk<mode>")
26693 ("indirect_jump_via_inlinethunk<mode>_z10")
26694 ("indirect_jump_via_inlinethunk<mode>", "*casesi_jump")
26695 ("casesi_jump_via_thunk<mode>_z10", "casesi_jump_via_thunk<mode>")
26696 ("casesi_jump_via_inlinethunk<mode>_z10")
26697 ("casesi_jump_via_inlinethunk<mode>", "*basr_via_thunk<mode>_z10")
26698 ("*basr_via_thunk<mode>", "*basr_r_via_thunk_z10")
26699 ("*basr_r_via_thunk", "return<mode>_prez10"): New pattern.
26700 ("*indirect2_jump"): Disable for branch conversion.
26701 ("casesi_jump"): Turn into expander and expand patterns for branch
26703 ("return_use"): New expander.
26704 ("*return"): Emit return via thunk and rename it to ...
26705 ("*return<mode>"): ... this one.
26706 * config/s390/s390.opt: Add new options and and enum for the
26709 2018-02-08 Richard Sandiford <richard.sandiford@linaro.org>
26711 * lra-constraints.c (match_reload): Unconditionally use
26712 gen_lowpart_SUBREG, rather than selecting between that
26713 and equivalent gen_rtx_SUBREG code.
26715 2018-02-08 Richard Biener <rguenther@suse.de>
26717 PR tree-optimization/84233
26718 * tree-ssa-phiprop.c (propagate_with_phi): Use separate
26719 changed flag instead of boguously re-using phi_inserted.
26721 2018-02-08 Martin Jambor <mjambor@suse.cz>
26723 * hsa-gen.c (get_symbol_for_decl): Set program allocation for
26724 static local variables.
26726 2018-02-08 Richard Biener <rguenther@suse.de>
26728 PR tree-optimization/84278
26729 * tree-vect-stmts.c (vectorizable_store): When looking for
26730 smaller vector types to perform grouped strided loads/stores
26731 make sure the mode is supported by the target.
26732 (vectorizable_load): Likewise.
26734 2018-02-08 Wilco Dijkstra <wdijkstr@arm.com>
26736 * config/aarch64/aarch64.c (aarch64_components_for_bb):
26737 Increase LDP/STP opportunities by adding adjacent callee-saves.
26739 2018-02-08 Wilco Dijkstra <wdijkstr@arm.com>
26741 PR rtl-optimization/84068
26742 PR rtl-optimization/83459
26743 * haifa-sched.c (rank_for_schedule): Fix SCHED_PRESSURE_MODEL sorting.
26745 2018-02-08 Aldy Hernandez <aldyh@redhat.com>
26747 PR tree-optimization/84224
26748 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Remove assert.
26749 * calls.c (gimple_alloca_call_p): Only return TRUE when we have
26750 non-zero arguments.
26752 2018-02-07 Iain Sandoe <iain@codesourcery.com>
26755 * config/rs6000/altivec.md (*restore_world): Remove LR use.
26756 * config/rs6000/predicates.md (restore_world_operation): Adjust op
26757 count, remove one USE.
26759 2018-02-07 Michael Meissner <meissner@linux.vnet.ibm.com>
26761 * doc/install.texi (Configuration): Document the
26762 --with-long-double-format={ibm,ieee} PowerPC configuration
26766 * config/rs6000/rs6000.md (fix_trunc<SFDF:mode><QHI:mode>2):
26767 Convert from define_expand to be define_insn_and_split. Rework
26768 float/double/_Float128 conversions to QI/HI/SImode to work with
26769 both ISA 2.07 (power8) or ISA 3.0 (power9). Fix regression where
26770 conversions to QI/HImode types did a store and then a load to
26771 truncate the value. For conversions to VSX registers, don't split
26772 the insn, instead emit the code directly. Use the code iterator
26773 any_fix to combine signed and unsigned conversions.
26774 (fix<uns>_trunc<SFDF:mode>si2_p8): Likewise.
26775 (fixuns_trunc<SFDF:mode><QHI:mode>2): Likewise.
26776 (fix_trunc<IEEE128:mode><QHI:mode>2): Likewise.
26777 (fix<uns>_trunc<SFDF:mode><QHI:mode>2): Likewise.
26778 (fix_<mode>di2_hw): Likewise.
26779 (fixuns_<mode>di2_hw): Likewise.
26780 (fix_<mode>si2_hw): Likewise.
26781 (fixuns_<mode>si2_hw): Likewise.
26782 (fix<uns>_<IEEE128:mode><SDI:mode>2_hw): Likewise.
26783 (fix<uns>_trunc<IEEE128:mode><QHI:mode>2): Likewise.
26784 (fctiw<u>z_<mode>_smallint): Rename fctiw<u>z_<mode>_smallint to
26785 fix<uns>_trunc<SFDF:mode>si2_p8.
26786 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Delete, no longer used.
26787 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
26788 (fix<uns>_<mode>_mem): Likewise.
26789 (fctiw<u>z_<mode>_mem): Likewise.
26790 (fix<uns>_<mode>_mem): Likewise.
26791 (fix<uns>_trunc<SFDF:mode><QHSI:mode>2_mem): On ISA 3.0, prevent
26792 the register allocator from doing a direct move to the GPRs to do
26793 a store, and instead use the ISA 3.0 store byte/half-word from
26794 vector register instruction. For IEEE 128-bit floating point,
26795 also optimize stores of 32-bit ints.
26796 (fix<uns>_trunc<IEEE128:mode><QHSI:mode>2_mem): Likewise.
26798 2018-02-07 Alan Hayward <alan.hayward@arm.com>
26800 * genextract.c (push_pathstr_operand): New function to support
26802 (walk_rtx): Call push_pathstr_operand.
26803 (print_path): Support [a-zA-Z].
26805 2018-02-07 Richard Biener <rguenther@suse.de>
26807 PR tree-optimization/84037
26808 * tree-vectorizer.h (struct _loop_vec_info): Add ivexpr_map member.
26809 (cse_and_gimplify_to_preheader): Declare.
26810 (vect_get_place_in_interleaving_chain): Likewise.
26811 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
26813 (_loop_vec_info::~_loop_vec_info): Delete it.
26814 (cse_and_gimplify_to_preheader): New function.
26815 * tree-vect-slp.c (vect_get_place_in_interleaving_chain): Export.
26816 * tree-vect-stmts.c (vectorizable_store): CSE base and steps.
26817 (vectorizable_load): Likewise. For grouped stores always base
26818 the IV on the first element.
26819 * tree-vect-loop-manip.c (vect_loop_versioning): Unshare versioning
26820 condition before gimplifying.
26822 2018-02-07 Jakub Jelinek <jakub@redhat.com>
26824 * tree-eh.c (operation_could_trap_helper_p): Ignore honor_trapv for
26825 *DIV_EXPR and *MOD_EXPR.
26827 2018-02-07 H.J. Lu <hongjiu.lu@intel.com>
26830 * config/i386/i386.c (ix86_option_override_internal): Mask out
26831 the CF_SET bit when checking -fcf-protection.
26833 2018-02-07 Tom de Vries <tom@codesourcery.com>
26836 * omp-expand.c (expand_oacc_collapse_init): Ensure diff_type is large
26839 2018-02-07 Richard Biener <rguenther@suse.de>
26841 PR tree-optimization/84204
26842 * tree-chrec.c (chrec_fold_plus_1): Remove size limiting in
26845 PR tree-optimization/84205
26846 * graphite-isl-ast-to-gimple.c (binary_op_to_tree): Also
26847 special-case isl_ast_op_zdiv_r.
26849 PR tree-optimization/84223
26850 * graphite-scop-detection.c (gather_bbs::before_dom_children):
26851 Only add conditions from within the region.
26852 (gather_bbs::after_dom_children): Adjust.
26854 2018-02-07 Georg-Johann Lay <avr@gjlay.de>
26857 * config/avr/avr.h (GENERAL_REGNO_P, GENERAL_REG_P): New macros.
26858 * config/avr/avr.md: Only post-reload split REG-REG moves if
26859 either register is GENERAL_REG_P.
26861 2018-02-07 Jakub Jelinek <jakub@redhat.com>
26863 PR tree-optimization/84235
26864 * tree-ssa-scopedtables.c
26865 (avail_exprs_stack::simplify_binary_operation): Fir MINUS_EXPR, punt
26866 if the subtraction is performed in floating point type where NaNs are
26867 honored. For *DIV_EXPR, punt for ALL_FRACT_MODE_Ps where we can't
26868 build 1. Formatting fix.
26870 2018-02-06 Jakub Jelinek <jakub@redhat.com>
26873 * config/i386/i386.c (rest_of_insert_endbranch): Only skip
26874 NOTE_INSN_CALL_ARG_LOCATION after a call, not anything else,
26875 and skip it regardless of bb boundaries. Use CALL_P macro,
26876 don't test INSN_P (insn) together with CALL_P or JUMP_P check
26877 unnecessarily, formatting fix.
26879 2018-02-06 Michael Collison <michael.collison@arm.com>
26881 * config/arm/thumb2.md:
26882 (*thumb2_mov_negscc): Split only if TARGET_THUMB2 && !arm_restrict_it.
26883 (*thumb_mov_notscc): Ditto.
26885 2018-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
26888 * config/rs6000/rs6000.md (su code attribute): Use "u" for
26889 unsigned_fix, not "s".
26891 2018-02-06 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
26893 * configure.ac (gcc_fn_eh_frame_ro): New function.
26894 (gcc_cv_as_cfi_directive): Check both 32 and 64-bit assembler for
26895 correct .eh_frame permissions.
26896 * configure: Regenerate.
26898 2018-02-06 Andrew Jenner <andrew@codeourcery.com>
26900 * doc/invoke.texi: Add section for the PowerPC SPE backend. Remove
26901 irrelevant options.
26903 2018-02-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26905 * config/rs6000/rs6000.c (rs6000_option_override_internal):
26906 Display warning message for -mno-speculate-indirect-jumps.
26908 2018-02-06 Andrew Jenner <andrew@codesourcery.com>
26910 * config/powerpcspe/powerpcspe.opt: (msimple-fpu, mfpu) Add
26912 * config/powerpcspe/sysv4.opt (mbit-align): Likewise.
26914 2018-02-06 Aldy Hernandez <aldyh@redhat.com>
26916 PR tree-optimization/84225
26917 * tree-eh.c (find_trapping_overflow): Only call
26918 operation_no_trapping_overflow when ANY_INTEGRAL_TYPE_P.
26920 2018-02-06 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
26923 * config/i386/i386.c: Reimplement the check of possible options
26924 -mibt/-mshstk conbination. Change error messages.
26925 * doc/invoke.texi: Fix a typo: remove extra '='.
26927 2018-02-06 Marek Polacek <polacek@redhat.com>
26929 PR tree-optimization/84228
26930 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Skip debug statements.
26932 2018-02-06 Tamar Christina <tamar.christina@arm.com>
26935 * config/arm/arm.c (arm_print_asm_arch_directives): Record already
26936 emitted arch directives.
26937 * config/arm/arm-c.c (arm_cpu_builtins): Undefine __ARM_ARCH and
26938 __ARM_FEATURE_COPROC before changing architectures.
26940 2018-02-06 Richard Biener <rguenther@suse.de>
26942 * config/i386/i386.c (print_reg): Fix typo.
26943 (ix86_loop_unroll_adjust): Do not unroll beyond the original nunroll.
26945 2018-02-06 Eric Botcazou <ebotcazou@adacore.com>
26947 * configure: Regenerate.
26949 2018-02-05 Martin Sebor <msebor@redhat.com>
26951 PR tree-optimization/83369
26952 * tree-ssa-ccp.c (pass_post_ipa_warn::execute): Use %G to print
26955 2018-02-05 Martin Liska <mliska@suse.cz>
26957 * doc/invoke.texi: Cherry-pick upstream r323995.
26959 2018-02-05 Richard Sandiford <richard.sandiford@linaro.org>
26961 * ira.c (ira_init_register_move_cost): Adjust comment.
26963 2018-02-05 Martin Liska <mliska@suse.cz>
26965 PR gcov-profile/84137
26966 * doc/gcov.texi: Fix typo in documentation.
26968 2018-02-05 Martin Liska <mliska@suse.cz>
26970 PR gcov-profile/83879
26971 * doc/gcov.texi: Document necessity of --dynamic-list-data when
26972 using dlopen functionality.
26974 2018-02-05 Olga Makhotina <olga.makhotina@intel.com>
26976 * config/i386/avx512dqintrin.h (_mm_mask_range_sd, _mm_maskz_range_sd,
26977 _mm_mask_range_round_sd, _mm_maskz_range_round_sd, _mm_mask_range_ss,
26978 _mm_maskz_range_ss, _mm_mask_range_round_ss,
26979 _mm_maskz_range_round_ss): New intrinsics.
26980 (__builtin_ia32_rangesd128_round)
26981 (__builtin_ia32_rangess128_round): Remove.
26982 (__builtin_ia32_rangesd128_mask_round,
26983 __builtin_ia32_rangess128_mask_round): New builtins.
26984 * config/i386/i386-builtin.def (__builtin_ia32_rangesd128_round,
26985 __builtin_ia32_rangess128_round): Remove.
26986 (__builtin_ia32_rangesd128_mask_round,
26987 __builtin_ia32_rangess128_mask_round): New builtins.
26988 * config/i386/sse.md (ranges<mode><round_saeonly_name>): Renamed to ...
26989 (ranges<mode><mask_scalar_name><round_saeonly_scalar_name>): ... this.
26990 ((match_operand:VF_128 2 "<round_saeonly_nimm_predicate>"
26991 "<round_saeonly_constraint>")): Changed to ...
26992 ((match_operand:VF_128 2 "<round_saeonly_scalar_nimm_predicate>"
26993 "<round_saeonly_scalar_constraint>")): ... this.
26994 ("vrange<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
26995 %0, %1, %2<round_saeonly_op4>, %3}"): Changed to ...
26996 ("vrange<ssescalarmodesuffix>\t{%3, <round_saeonly_scalar_mask_op4>%2,
26997 %1, %0<mask_scalar_operand4>|%0<mask_scalar_operand4>, %1,
26998 %2<round_saeonly_scalar_mask_op4>, %3}"): ... this.
27000 2018-02-02 Andrew Jenner <andrew@codesourcery.com>
27002 * config/powerpcspe/powerpcspe.opt: Add Undocumented to irrelevant
27004 * config/powerpcspe/powerpcspe-tables.opt (rs6000_cpu_opt_value):
27005 Remove all values except native, 8540 and 8548.
27007 2018-02-02 H.J. Lu <hongjiu.lu@intel.com>
27009 * config/i386/i386.c (ix86_output_function_return): Pass
27010 INVALID_REGNUM, instead of -1, as invalid register number to
27011 indirect_thunk_name and output_indirect_thunk.
27013 2018-02-02 Julia Koval <julia.koval@intel.com>
27015 * config.gcc: Add -march=icelake.
27016 * config/i386/driver-i386.c (host_detect_local_cpu): Detect icelake.
27017 * config/i386/i386-c.c (ix86_target_macros_internal): Handle icelake.
27018 * config/i386/i386.c (processor_costs): Add m_ICELAKE.
27019 (PTA_ICELAKE, PTA_AVX512VNNI, PTA_GFNI, PTA_VAES, PTA_AVX512VBMI2,
27020 PTA_VPCLMULQDQ, PTA_RDPID, PTA_AVX512BITALG): New.
27021 (processor_target_table): Add icelake.
27022 (ix86_option_override_internal): Handle new PTAs.
27023 (get_builtin_code_for_version): Handle icelake.
27024 (M_INTEL_COREI7_ICELAKE): New.
27025 (fold_builtin_cpu): Handle icelake.
27026 * config/i386/i386.h (TARGET_ICELAKE, PROCESSOR_ICELAKE): New.
27027 * doc/invoke.texi: Add -march=icelake.
27029 2018-02-02 Julia Koval <julia.koval@intel.com>
27031 * config/i386/i386.c (ix86_option_override_internal): Change flags type
27032 to wide_int_bitmask.
27033 * wide-int-bitmask.h: New.
27035 2018-02-02 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
27038 * config/i386/i386.md: Replace Pmode with word_mode in
27039 builtin_setjmp_setup and builtin_longjmp to support x32.
27041 2018-02-01 Peter Bergner <bergner@vnet.ibm.com>
27045 * config/rs6000/driver-rs6000.c: #include "diagnostic.h".
27047 (rs6000_supported_cpu_names): New static variable.
27048 (linux_cpu_translation_table): Likewise.
27049 (elf_platform) <cpu>: Define new static variable and use it.
27050 Translate kernel AT_PLATFORM name to canonical name if needed.
27051 Error if platform name is unknown.
27053 2018-02-01 Aldy Hernandez <aldyh@redhat.com>
27056 * config/pa/predicates.md (base14_operand): Handle E_VOIDmode.
27058 2018-02-01 Jeff Law <law@redhat.com>
27061 * config/i386/i386.c (release_scratch_register_on_entry): Add new
27062 OFFSET and RELEASE_VIA_POP arguments. Use SP+OFFSET to restore
27063 the scratch if RELEASE_VIA_POP is false.
27064 (ix86_adjust_stack_and_probe_stack_clash): Un-constify SIZE.
27065 If we have to save a temporary register, decrement SIZE appropriately.
27066 Pass new arguments to release_scratch_register_on_entry.
27067 (ix86_adjust_stack_and_probe): Likewise.
27068 (ix86_emit_probe_stack_range): Pass new arguments to
27069 release_scratch_register_on_entry.
27071 2018-02-01 Uros Bizjak <ubizjak@gmail.com>
27073 PR rtl-optimization/84157
27074 * combine.c (change_zero_ext): Use REG_P predicate in
27075 front of HARD_REGISTER_P predicate.
27077 2018-02-01 Georg-Johann Lay <avr@gjlay.de>
27079 * config/avr/avr.c (avr_option_override): Move disabling of
27080 -fdelete-null-pointer-checks to...
27081 * common/config/avr/avr-common.c (avr_option_optimization_table):
27084 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
27086 PR tree-optimization/81635
27087 * tree-data-ref.c (split_constant_offset_1): For types that
27088 wrap on overflow, try to use range info to prove that wrapping
27091 2018-02-01 Renlin Li <renlin.li@arm.com>
27094 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
27095 TAILCALL_ADDR_REGS.
27096 (aarch64_register_move_cost): Likewise.
27097 * config/aarch64/aarch64.h (reg_class): Rename CALLER_SAVE_REGS to
27098 TAILCALL_ADDR_REGS.
27099 (REG_CLASS_NAMES): Likewise.
27100 (REG_CLASS_CONTENTS): Rename CALLER_SAVE_REGS to
27101 TAILCALL_ADDR_REGS. Remove IP registers.
27102 * config/aarch64/aarch64.md (Ucs): Update register constraint.
27104 2018-02-01 Richard Biener <rguenther@suse.de>
27106 * domwalk.h (dom_walker::dom_walker): Add additional constructor
27107 for specifying RPO order and allow NULL for that.
27108 * domwalk.c (dom_walker::dom_walker): Likewise.
27109 (dom_walker::walk): Handle NULL RPO order.
27110 * tree-into-ssa.c (rewrite_dom_walker): Do not walk dom children
27112 (rewrite_update_dom_walker): Likewise.
27113 (mark_def_dom_walker): Likewise.
27115 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
27117 * config/aarch64/aarch64-protos.h (aarch64_split_sve_subreg_move)
27118 (aarch64_maybe_expand_sve_subreg_move): Declare.
27119 * config/aarch64/aarch64.md (UNSPEC_REV_SUBREG): New unspec.
27120 * config/aarch64/predicates.md (aarch64_any_register_operand): New
27122 * config/aarch64/aarch64-sve.md (mov<mode>): Optimize subreg moves
27123 that are semantically a reverse operation.
27124 (*aarch64_sve_mov<mode>_subreg_be): New pattern.
27125 * config/aarch64/aarch64.c (aarch64_maybe_expand_sve_subreg_move):
27126 (aarch64_replace_reg_mode, aarch64_split_sve_subreg_move): New
27128 (aarch64_can_change_mode_class): For big-endian, forbid changes
27129 between two SVE modes if they have different element sizes.
27131 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
27133 * config/aarch64/aarch64.c (aarch64_expand_sve_const_vector): Prefer
27134 the TImode handling for big-endian targets.
27136 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
27138 * config/aarch64/aarch64-sve.md (sve_ld1rq): Replace with...
27139 (*sve_ld1rq<Vesize>): ... this new pattern. Handle all element sizes,
27141 * config/aarch64/aarch64.c (aarch64_expand_sve_widened_duplicate):
27142 Remove BSWAP handing for big-endian targets and use the form of
27143 LD1RQ appropariate for the mode.
27145 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
27147 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Handle
27148 all CONST_VECTOR_DUPLICATE_P vectors, not just those with a single
27149 duplicated element.
27151 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
27154 * config/aarch64/aarch64.c (aarch64_secondary_reload): Tighten
27155 check for operands that need to go through aarch64_sve_reload_be.
27157 2018-02-01 Jakub Jelinek <jakub@redhat.com>
27159 PR tree-optimization/81661
27160 PR tree-optimization/84117
27161 * tree-eh.h (rewrite_to_non_trapping_overflow): Declare.
27162 * tree-eh.c: Include gimplify.h.
27163 (find_trapping_overflow, replace_trapping_overflow,
27164 rewrite_to_non_trapping_overflow): New functions.
27165 * tree-vect-loop.c: Include tree-eh.h.
27166 (vect_get_loop_niters): Use rewrite_to_non_trapping_overflow.
27167 * tree-data-ref.c: Include tree-eh.h.
27168 (get_segment_min_max): Use rewrite_to_non_trapping_overflow.
27170 2018-01-31 Uros Bizjak <ubizjak@gmail.com>
27172 PR rtl-optimization/84123
27173 * combine.c (change_zero_ext): Check if hard register satisfies
27174 can_change_dest_mode before calling gen_lowpart_SUBREG.
27176 2018-01-31 Vladimir Makarov <vmakarov@redhat.com>
27179 * ira.c (ira_init_register_move_cost): Remove assert.
27181 2018-01-31 Eric Botcazou <ebotcazou@adacore.com>
27183 PR rtl-optimization/84071
27184 * doc/tm.texi.in (WORD_REGISTER_OPERATIONS): Add explicit case.
27185 * doc/tm.texi: Regenerate.
27187 2018-01-31 Richard Biener <rguenther@suse.de>
27189 PR tree-optimization/84132
27190 * tree-data-ref.c (analyze_miv_subscript): Properly
27191 check whether evolution_function_is_affine_multivariate_p
27192 before calling gcd_of_steps_may_divide_p.
27194 2018-01-31 Julia Koval <julia.koval@intel.com>
27197 * config/i386/i386.c (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
27198 * config/i386/i386.md (rdpid_rex64) New.
27199 (rdpid): Make 32bit only.
27201 2018-01-29 Aldy Hernandez <aldyh@redhat.com>
27204 * tree-pretty-print.c (dump_generic_node): Handle a TYPE_NAME with
27205 an IDENTIFIER_NODE for FUNCTION_TYPE's.
27207 2018-01-31 Eric Botcazou <ebotcazou@adacore.com>
27210 2018-01-12 Eric Botcazou <ebotcazou@adacore.com>
27212 * config/sparc/sparc.md (vxworks_load_got): Set the GOT register.
27214 2018-01-31 Eric Botcazou <ebotcazou@adacore.com>
27216 PR rtl-optimization/84071
27217 * combine.c (record_dead_and_set_regs_1): Record the source unmodified
27218 for a paradoxical SUBREG on a WORD_REGISTER_OPERATIONS target.
27220 2018-01-31 Claudiu Zissulescu <claziss@synopsys.com>
27222 * config/arc/arc.c (arc_handle_aux_attribute): New function.
27223 (arc_attribute_table): Add 'aux' attribute.
27224 (arc_in_small_data_p): Consider aux like variables.
27225 (arc_is_aux_reg_p): New function.
27226 (arc_asm_output_aligned_decl_local): Ignore 'aux' like variables.
27227 (arc_get_aux_arg): New function.
27228 (prepare_move_operands): Handle aux-register access.
27229 (arc_handle_aux_attribute): New function.
27230 * doc/extend.texi (ARC Variable attributes): Add subsection.
27232 2018-01-31 Claudiu Zissulescu <claziss@synopsys.com>
27234 * config/arc/arc-protos.h (arc_is_uncached_mem_p): Function proto.
27235 * config/arc/arc.c (arc_handle_uncached_attribute): New function.
27236 (arc_attribute_table): Add 'uncached' attribute.
27237 (arc_print_operand): Print '.di' flag for uncached memory
27239 (arc_in_small_data_p): Do not consider for small data the uncached
27241 (arc_is_uncached_mem_p): New function.
27242 * config/arc/predicates.md (compact_store_memory_operand): Check
27243 for uncached memory accesses.
27244 (nonvol_nonimm_operand): Likewise.
27245 * doc/extend.texi (ARC Type Attribute): New subsection.
27247 2018-01-31 Jakub Jelinek <jakub@redhat.com>
27250 * common.opt (falign-functions=, falign-jumps=, falign-labels=,
27251 falign-loops=): Add Optimization flag.
27253 2018-01-30 Jeff Law <law@redhat.com>
27256 * i386.c (ix86_adjust_stack_and_probe_stack_clash): New argument
27257 INT_REGISTERS_SAVED. Check it prior to calling
27258 get_scratch_register_on_entry.
27259 (ix86_adjust_stack_and_probe): Similarly.
27260 (ix86_emit_probe_stack_range): Similarly.
27261 (ix86_expand_prologue): Corresponding changes.
27263 2018-01-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
27266 * config/sol2.h (STARTFILE_ARCH_SPEC): Use -std=c*,
27267 -std=iso9899:199409 instead of -pedantic to select values-Xc.o.
27269 2018-01-30 Vladimir Makarov <vmakarov@redhat.com>
27272 * lra-constraints.c (curr_insn_transform): Process AND in the
27275 2018-01-30 Jakub Jelinek <jakub@redhat.com>
27277 PR rtl-optimization/83986
27278 * sched-deps.c (sched_analyze_insn): For frame related insns, add anti
27279 dependence against last_pending_memory_flush in addition to
27280 pending_jump_insns.
27282 2018-01-30 Alexandre Oliva <aoliva@redhat.com>
27284 PR tree-optimization/81611
27285 * tree-ssa-dom.c (simple_iv_increment_p): Skip intervening
27288 2018-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27291 * config/rs6000/rs6000.c (rs6000_internal_arg_pointer): Only return
27294 2018-01-30 Richard Biener <rguenther@suse.de>
27295 Jakub Jelinek <jakub@redhat.com>
27297 PR tree-optimization/84111
27298 * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely_1): Skip
27299 inner loops added during recursion, as they don't have up-to-date
27302 2018-01-30 Jan Hubicka <hubicka@ucw.cz>
27305 * ipa-inline.c (can_inline_edge_p): Break out late tests to...
27306 (can_inline_edge_by_limits_p): ... here.
27307 (can_early_inline_edge_p, check_callers,
27308 update_caller_keys, update_callee_keys, recursive_inlining,
27309 add_new_edges_to_heap, speculation_useful_p,
27310 inline_small_functions,
27311 inline_small_functions, flatten_function,
27312 inline_to_all_callers_1): Update.
27314 2018-01-30 Jan Hubicka <hubicka@ucw.cz>
27316 * profile-count.c (profile_count::combine_with_ipa_count): Handle
27319 2018-01-30 Richard Biener <rguenther@suse.de>
27321 PR tree-optimization/83008
27322 * tree-vect-slp.c (vect_analyze_slp_cost_1): Properly cost
27323 invariant and constant vector uses in stmts when they need
27324 more than one stmt.
27326 2018-01-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
27329 * configure.ac (gcc_cv_as_shf_merge): Disable on Solaris 10/x86.
27330 * configure: Regenerate.
27332 2018-01-30 Richard Sandiford <richard.sandiford@linaro.org>
27334 * config/aarch64/aarch64-sve.md (*vec_extract<mode><Vel>_0): New
27336 (*vec_extract<mode><Vel>_v128): Require a nonzero lane number.
27337 Use gen_rtx_REG rather than gen_lowpart.
27339 2018-01-30 Richard Sandiford <richard.sandiford@linaro.org>
27341 * lra-constraints.c (match_reload): Use subreg_lowpart_offset
27342 rather than 0 when creating partial subregs.
27344 2018-01-30 Richard Sandiford <richard.sandiford@linaro.org>
27346 * vec-perm-indices.c (vec_perm_indices::series_p): Give examples
27349 2018-01-29 Michael Meissner <meissner@linux.vnet.ibm.com>
27352 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): If DFmode
27353 and SFmode can go in Altivec registers (-mcpu=power7 for DFmode,
27354 -mcpu=power8 for SFmode) don't set the PRE_INCDEC or PRE_MODIFY
27355 flags. This restores the settings used before the 2017-07-24.
27356 Turning off pre increment/decrement/modify allows IVOPTS to
27357 optimize DF/SF loops where the index is an int.
27359 2018-01-29 Richard Biener <rguenther@suse.de>
27360 Kelvin Nilsen <kelvin@gcc.gnu.org>
27363 * tree-vect-stmts.c (vectorizable_call): Don't call
27364 targetm.vectorize_builtin_md_vectorized_function if callee is
27367 2018-01-22 Carl Love <cel@us.ibm.com>
27369 * doc/extend.tex: Fix typo in second arg in
27370 __builtin_bcdadd_{lt|eq|gt|ov} and __builtin_bcdsub_{lt|eq|gt|ov}.
27372 2018-01-29 Richard Biener <rguenther@suse.de>
27374 PR tree-optimization/84086
27375 * tree-ssanames.c: Include cfgloop.h and tree-scalar-evolution.h.
27376 (flush_ssaname_freelist): When SSA names were released reset
27377 the SCEV hash table.
27379 2018-01-29 Richard Biener <rguenther@suse.de>
27381 PR tree-optimization/84057
27382 * tree-ssa-loop-ivcanon.c (unloop_loops): Deal with already
27383 removed paths when removing edges.
27385 2018-01-27 H.J. Lu <hongjiu.lu@intel.com>
27387 * doc/invoke.texi: Replace -mfunction-return==@var{choice} with
27388 -mfunction-return=@var{choice}.
27390 2018-01-27 Bernd Edlinger <bernd.edlinger@hotmail.de>
27392 PR diagnostic/84034
27393 * diagnostic-show-locus.c (get_line_width_without_trailing_whitespace):
27394 Handle CR like TAB.
27395 (layout::print_source_line): Likewise.
27396 (test_get_line_width_without_trailing_whitespace): Add test cases.
27398 2018-01-27 Jakub Jelinek <jakub@redhat.com>
27400 PR middle-end/84040
27401 * sched-deps.c (sched_macro_fuse_insns): Return immediately for
27404 2018-01-26 Jim Wilson <jimw@sifive.com>
27406 * config/riscv/riscv.h (MAX_FIXED_MODE_SIZE): New.
27408 * config/riscv/elf.h (LIB_SPEC): Don't include -lgloss when nosys.specs
27411 2018-01-26 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27413 * config/aarch64/aarch64.md: Add peepholes for CMP + SUB -> SUBS
27414 and CMP + SUB-immediate -> SUBS.
27416 2018-01-26 Martin Sebor <msebor@redhat.com>
27418 PR tree-optimization/83896
27419 * tree-ssa-strlen.c (get_string_len): Rename...
27420 (get_string_cst_length): ...to this. Return HOST_WIDE_INT.
27421 Avoid assuming length is constant.
27422 (handle_char_store): Use HOST_WIDE_INT for string length.
27424 2018-01-26 Uros Bizjak <ubizjak@gmail.com>
27427 * config/i386/i386.md (*andndi3_doubleword): Add earlyclobber
27428 to (=&r,r,rm) alternative. Add (=r,0,rm) and (=r,r,0) alternatives.
27430 2018-01-26 Richard Biener <rguenther@suse.de>
27432 PR rtl-optimization/84003
27433 * dse.c (record_store): Only record redundant stores when
27434 the earlier store aliases at least all accesses the later one does.
27436 2018-01-26 Jakub Jelinek <jakub@redhat.com>
27438 PR rtl-optimization/83985
27439 * dce.c (deletable_insn_p): Return false for separate shrink wrapping
27440 REG_CFA_RESTORE insns.
27441 (delete_unmarked_insns): Don't ignore separate shrink wrapping
27442 REG_CFA_RESTORE insns here.
27445 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Don't
27446 use SSA_NAME_VAR as base for SSA_NAMEs with non-NULL SSA_NAME_VAR.
27448 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
27450 * config/arc/arc-arch.h (arc_tune_attr): Add ARC_TUNE_CORE_3.
27451 * config/arc/arc.c (arc_sched_issue_rate): Use ARC_TUNE_... .
27452 (arc_init): Likewise.
27453 (arc_override_options): Likewise.
27454 (arc_file_start): Choose Tag_ARC_CPU_variation based on arc_tune
27456 (hwloop_fail): Use TARGET_DBNZ when we want to check for dbnz insn
27458 * config/arc/arc.h (TARGET_DBNZ): Define.
27459 * config/arc/arc.md (attr tune): Add core_3, use ARC_TUNE_... to
27460 properly set the tune attribute.
27461 (dbnz): Use TARGET_DBNZ guard.
27462 * config/arc/arc.opt (mtune): Add core3 option.
27464 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
27466 * config/arc/arc.c (arc_delegitimize_address_0): Refactored to
27467 recognize new pic like addresses.
27468 (arc_delegitimize_address): Clean up.
27470 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
27472 * config/arc/arc-arches.def: Option mrf16 valid for all
27474 * config/arc/arc-c.def (__ARC_RF16__): New predefined macro.
27475 * config/arc/arc-cpus.def (em_mini): New cpu with rf16 on.
27476 * config/arc/arc-options.def (FL_RF16): Add mrf16 option.
27477 * config/arc/arc-tables.opt: Regenerate.
27478 * config/arc/arc.c (arc_conditional_register_usage): Handle
27479 reduced register file case.
27480 (arc_file_start): Set must have build attributes.
27481 * config/arc/arc.h (MAX_ARC_PARM_REGS): Conditional define using
27482 mrf16 option value.
27483 * config/arc/arc.opt (mrf16): Add new option.
27484 * config/arc/elf.h (ATTRIBUTE_PCS): Define.
27485 * config/arc/genmultilib.awk: Handle new mrf16 option.
27486 * config/arc/linux.h (ATTRIBUTE_PCS): Define.
27487 * config/arc/t-multilib: Regenerate.
27488 * doc/invoke.texi (ARC Options): Document mrf16 option.
27490 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
27492 * config/arc/arc-protos.h: Add arc_is_secure_call_p proto.
27493 * config/arc/arc.c (arc_handle_secure_attribute): New function.
27494 (arc_attribute_table): Add 'secure_call' attribute.
27495 (arc_print_operand): Print secure call operand.
27496 (arc_function_ok_for_sibcall): Don't optimize tail calls when
27498 (arc_is_secure_call_p): New function. * config/arc/arc.md
27499 (call_i): Add support for sjli instruction.
27500 (call_value_i): Likewise.
27501 * config/arc/constraints.md (Csc): New constraint.
27503 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
27504 John Eric Martin <John.Martin@emmicro-us.com>
27506 * config/arc/arc-protos.h: Add arc_is_jli_call_p proto.
27507 * config/arc/arc.c (_arc_jli_section): New struct.
27508 (arc_jli_section): New type.
27509 (rc_jli_sections): New static variable.
27510 (arc_handle_jli_attribute): New function.
27511 (arc_attribute_table): Add jli_always and jli_fixed attribute.
27512 (arc_file_end): New function.
27513 (TARGET_ASM_FILE_END): Define.
27514 (arc_print_operand): Reuse 'S' letter for JLI output instruction.
27515 (arc_add_jli_section): New function.
27516 (jli_call_scan): Likewise.
27517 (arc_reorg): Call jli_call_scan.
27518 (arc_output_addsi): Remove 'S' from printing asm operand.
27519 (arc_is_jli_call_p): New function.
27520 * config/arc/arc.md (movqi_insn): Remove 'S' from printing asm
27522 (movhi_insn): Likewise.
27523 (movsi_insn): Likewise.
27524 (movsi_set_cc_insn): Likewise.
27525 (loadqi_update): Likewise.
27526 (load_zeroextendqisi_update): Likewise.
27527 (load_signextendqisi_update): Likewise.
27528 (loadhi_update): Likewise.
27529 (load_zeroextendhisi_update): Likewise.
27530 (load_signextendhisi_update): Likewise.
27531 (loadsi_update): Likewise.
27532 (loadsf_update): Likewise.
27533 (movsicc_insn): Likewise.
27534 (bset_insn): Likewise.
27535 (bxor_insn): Likewise.
27536 (bclr_insn): Likewise.
27537 (bmsk_insn): Likewise.
27538 (bicsi3_insn): Likewise.
27539 (cmpsi_cc_c_insn): Likewise.
27540 (movsi_ne): Likewise.
27541 (movsi_cond_exec): Likewise.
27542 (clrsbsi2): Likewise.
27543 (norm_f): Likewise.
27553 (call_i): Remove 'S' asm letter, add jli instruction.
27554 (call_value_i): Likewise.
27555 * config/arc/arc.op (mjli-always): New option.
27556 * config/arc/constraints.md (Cji): New constraint.
27557 * config/arc/fpx.md (addsf3_fpx): Remove 'S' from printing asm
27559 (subsf3_fpx): Likewise.
27560 (mulsf3_fpx): Likewise.
27561 * config/arc/simdext.md (vendrec_insn): Remove 'S' from printing
27563 * doc/extend.texi (ARC): Document 'jli-always' and 'jli-fixed'
27564 function attrbutes.
27565 * doc/invoke.texi (ARC): Document mjli-always option.
27567 2018-01-26 Sebastian Perta <sebastian.perta@renesas.com>
27569 * config/rl78/rl78.c (rl78_addsi3_internal): If operand 2 is const
27570 avoid addition with 0 and use incw and decw where possible.
27572 2018-01-26 Richard Biener <rguenther@suse.de>
27574 PR tree-optimization/81082
27575 * fold-const.c (fold_plusminus_mult_expr): Do not perform the
27576 association if it requires casting to unsigned.
27577 * match.pd ((A * C) +- (B * C) -> (A+-B)): New patterns derived
27578 from fold_plusminus_mult_expr to catch important cases late when
27579 range info is available.
27581 2018-01-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
27583 * config/i386/sol2.h (USE_HIDDEN_LINKONCE): Remove.
27584 * configure.ac (hidden_linkonce): New test.
27585 * configure: Regenerate.
27586 * config.in: Regenerate.
27588 2018-01-26 Julia Koval <julia.koval@intel.com>
27590 * config/i386/avx512bitalgintrin.h (_mm512_bitshuffle_epi64_mask,
27591 _mm512_mask_bitshuffle_epi64_mask, _mm256_bitshuffle_epi64_mask,
27592 _mm256_mask_bitshuffle_epi64_mask, _mm_bitshuffle_epi64_mask,
27593 _mm_mask_bitshuffle_epi64_mask): Fix type.
27594 * config/i386/i386-builtin-types.def (UHI_FTYPE_V2DI_V2DI_UHI,
27595 USI_FTYPE_V4DI_V4DI_USI): Remove.
27596 * config/i386/i386-builtin.def (__builtin_ia32_vpshufbitqmb512_mask,
27597 __builtin_ia32_vpshufbitqmb256_mask,
27598 __builtin_ia32_vpshufbitqmb128_mask): Fix types.
27599 * config/i386/i386.c (ix86_expand_args_builtin): Remove old types.
27600 * config/i386/sse.md (VI1_AVX512VLBW): Change types.
27602 2018-01-26 Alan Modra <amodra@gmail.com>
27605 * config/rs6000/rs6000-p8swap.c (rtx_is_swappable_p): Exclude
27606 UNSPEC_VBPERMQ. Sort other unspecs.
27608 2018-01-25 David Edelsohn <dje.gcc@gmail.com>
27610 * doc/invoke.texi (PowerPC Options): Document 'native' cpu type.
27612 2018-01-25 Jan Hubicka <hubicka@ucw.cz>
27614 PR middle-end/83055
27615 * predict.c (drop_profile): Do not push/pop cfun; update also
27617 (handle_missing_profiles): Fix logic looking for zero profiles.
27619 2018-01-25 Jakub Jelinek <jakub@redhat.com>
27621 PR middle-end/83977
27622 * ipa-fnsummary.c (compute_fn_summary): Clear can_change_signature
27623 on functions with #pragma omp declare simd or functions with simd
27625 * omp-simd-clone.c (expand_simd_clones): Revert 2018-01-24 change.
27626 * config/i386/i386.c (ix86_simd_clone_compute_vecsize_and_simdlen):
27627 Remove trailing \n from warning_at calls.
27629 2018-01-25 Tom de Vries <tom@codesourcery.com>
27632 * config/nvptx/nvptx.c (nvptx_single): Add exit insn after noreturn call
27633 for neutered workers.
27635 2018-01-24 Joseph Myers <joseph@codesourcery.com>
27638 * config/m68k/m68k.c (m68k_promote_function_mode): New function.
27639 (TARGET_PROMOTE_FUNCTION_MODE): New macro.
27641 2018-01-24 Jeff Law <law@redhat.com>
27644 * i386.c (get_probe_interval): Move to earlier point.
27645 (ix86_compute_frame_layout): If -fstack-clash-protection and
27646 the frame is larger than the probe interval, then use pushes
27647 to save registers rather than reg->mem moves.
27648 (ix86_expand_prologue): Remove conditional for int_registers_saved
27651 2018-01-24 Vladimir Makarov <vmakarov@redhat.com>
27654 * ira-build.c (setup_min_max_allocno_live_range_point): Set up
27655 min/max for never referenced object.
27657 2018-01-24 Jakub Jelinek <jakub@redhat.com>
27659 PR middle-end/83977
27660 * tree.c (free_lang_data_in_decl): Don't clear DECL_ABSTRACT_ORIGIN
27662 * omp-low.c (create_omp_child_function): Remove "omp declare simd"
27663 attributes from DECL_ATTRIBUTES (decl) without affecting
27664 DECL_ATTRIBUTES (current_function_decl).
27665 * omp-simd-clone.c (expand_simd_clones): Ignore DECL_ARTIFICIAL
27666 functions with non-NULL DECL_ABSTRACT_ORIGIN.
27668 2018-01-24 Richard Sandiford <richard.sandiford@linaro.org>
27670 PR tree-optimization/83979
27671 * fold-const.c (fold_comparison): Use constant_boolean_node
27672 instead of boolean_{true,false}_node.
27674 2018-01-24 Jan Hubicka <hubicka@ucw.cz>
27676 * ipa-profile.c (ipa_propagate_frequency_1): Fix logic skipping calls
27679 2018-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27681 * config/rs6000/rs6000.md (*call_indirect_nonlocal_sysv<mode>):
27682 Simplify the clause that sets the length attribute.
27683 (*call_value_indirect_nonlocal_sysv<mode>): Likewise.
27684 (*sibcall_nonlocal_sysv<mode>): Clean up code block; simplify the
27685 clause that sets the length attribute.
27686 (*sibcall_value_nonlocal_sysv<mode>): Likewise.
27688 2018-01-24 Tom de Vries <tom@codesourcery.com>
27691 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG_2): Define to 1.
27692 (nvptx_pc_set, nvptx_condjump_label): New function. Copy from jump.c.
27693 Add strict parameter.
27694 (prevent_branch_around_nothing): Insert dummy insn between branch to
27695 label and label with no ptx insn inbetween.
27696 * config/nvptx/nvptx.md (define_insn "fake_nop"): New insn.
27698 2018-01-24 Tom de Vries <tom@codesourcery.com>
27701 * config/nvptx/nvptx.c (nvptx_single): Add exit insn after noreturn call
27702 for neutered threads in warp.
27703 * config/nvptx/nvptx.md (define_insn "exit"): New insn.
27705 2018-01-24 Richard Biener <rguenther@suse.de>
27707 PR tree-optimization/83176
27708 * tree-chrec.c (chrec_fold_plus_1): Handle (signed T){(T) .. }
27711 2018-01-24 Richard Biener <rguenther@suse.de>
27713 PR tree-optimization/82819
27714 * graphite-isl-ast-to-gimple.c (binary_op_to_tree): Avoid
27715 code generating pluses that are no-ops in the target precision.
27717 2018-01-24 Richard Biener <rguenther@suse.de>
27719 PR middle-end/84000
27720 * tree-cfg.c (replace_loop_annotate): Handle annot_expr_parallel_kind.
27722 2018-01-23 Jan Hubicka <hubicka@ucw.cz>
27724 * cfgcleanup.c (try_crossjump_to_edge): Use combine_with_count
27725 to merge probabilities.
27726 * predict.c (probably_never_executed): Also mark as cold functions
27727 with global 0 profile and guessed local profile.
27728 * profile-count.c (profile_probability::combine_with_count): New
27730 * profile-count.h (profile_probability::operator*,
27731 profile_probability::operator*=, profile_probability::operator/,
27732 profile_probability::operator/=): Reduce precision to adjusted
27733 and set value to guessed on contradictory divisions.
27734 (profile_probability::combine_with_freq): Remove.
27735 (profile_probability::combine_wiht_count): Declare.
27736 (profile_count::force_nonzero):: Set to adjusted.
27737 (profile_count::probability_in):: Set quality to adjusted.
27738 * tree-ssa-tail-merge.c (replace_block_by): Use
27739 combine_with_count.
27741 2018-01-23 Andrew Waterman <andrew@sifive.com>
27742 Jim Wilson <jimw@sifive.com>
27744 * config/riscv/riscv.c (riscv_stack_boundary): New.
27745 (riscv_option_override): Set riscv_stack_boundary. Handle
27746 riscv_preferred_stack_boundary_arg.
27747 * config/riscv/riscv.h (MIN_STACK_BOUNDARY, ABI_STACK_BOUNDARY): New.
27748 (BIGGEST_ALIGNMENT): Set to STACK_BOUNDARY.
27749 (STACK_BOUNDARY): Set to riscv_stack_boundary.
27750 (RISCV_STACK_ALIGN): Use STACK_BOUNDARY.
27751 * config/riscv/riscv.opt (mpreferred-stack-boundary): New.
27752 * doc/invoke.tex (RISC-V Options): Add -mpreferred-stack-boundary.
27754 2018-01-23 H.J. Lu <hongjiu.lu@intel.com>
27757 * config/i386/i386.c (ix86_expand_prologue): Use cost reference
27758 of struct ix86_frame.
27759 (ix86_expand_epilogue): Likewise. Add a local variable for
27760 the reg_save_offset field in struct ix86_frame.
27762 2018-01-23 Bin Cheng <bin.cheng@arm.com>
27764 PR tree-optimization/82604
27765 * tree-loop-distribution.c (enum partition_kind): New enum item
27766 PKIND_PARTIAL_MEMSET.
27767 (partition_builtin_p): Support above new enum item.
27768 (generate_code_for_partition): Ditto.
27769 (compute_access_range): Differentiate cases that equality can be
27770 proven at all loops, the innermost loops or no loops.
27771 (classify_builtin_st, classify_builtin_ldst): Adjust call to above
27772 function. Set PKIND_PARTIAL_MEMSET for partition appropriately.
27773 (finalize_partitions, distribute_loop): Don't fuse partition of
27774 PKIND_PARTIAL_MEMSET kind when distributing 3-level loop nest.
27775 (prepare_perfect_loop_nest): Distribute 3-level loop nest only if
27776 parloop is enabled.
27778 2018-01-23 Martin Liska <mliska@suse.cz>
27780 * predict.def (PRED_INDIR_CALL): Set probability to PROB_EVEN in
27781 order to ignore the predictor.
27782 (PRED_POLYMORPHIC_CALL): Likewise.
27783 (PRED_RECURSIVE_CALL): Likewise.
27785 2018-01-23 Martin Liska <mliska@suse.cz>
27787 * tree-profile.c (tree_profiling): Print function header to
27788 aware reader which function we are working on.
27789 * value-prof.c (gimple_find_values_to_profile): Do not print
27790 not interesting value histograms.
27792 2018-01-23 Martin Liska <mliska@suse.cz>
27794 * profile-count.h (enum profile_quality): Add
27795 profile_uninitialized as the first value. Do not number values
27796 as they are zero based.
27797 (profile_count::verify): Update sanity check.
27798 (profile_probability::verify): Likewise.
27800 2018-01-23 Nathan Sidwell <nathan@acm.org>
27802 * doc/invoke.texi (ffor-scope): Deprecate.
27804 2018-01-23 David Malcolm <dmalcolm@redhat.com>
27806 PR tree-optimization/83510
27807 * domwalk.c (set_all_edges_as_executable): New function.
27808 (dom_walker::dom_walker): Convert bool param
27809 "skip_unreachable_blocks" to enum reachability. Move setup of
27810 edge flags to set_all_edges_as_executable and only do it when
27811 reachability is REACHABLE_BLOCKS.
27812 * domwalk.h (enum dom_walker::reachability): New enum.
27813 (dom_walker::dom_walker): Convert bool param
27814 "skip_unreachable_blocks" to enum reachability.
27815 (set_all_edges_as_executable): New decl.
27816 * graphite-scop-detection.c (gather_bbs::gather_bbs): Convert
27817 from false for "skip_unreachable_blocks" to ALL_BLOCKS for
27819 * tree-ssa-dom.c (dom_opt_dom_walker::dom_opt_dom_walker): Likewise,
27820 but converting true to REACHABLE_BLOCKS.
27821 * tree-ssa-sccvn.c (sccvn_dom_walker::sccvn_dom_walker): Likewise.
27823 (check_array_bounds_dom_walker::check_array_bounds_dom_walker):
27824 Likewise, but converting it to REACHABLE_BLOCKS_PRESERVING_FLAGS.
27825 (vrp_dom_walker::vrp_dom_walker): Likewise, but converting it to
27827 (vrp_prop::vrp_finalize): Call set_all_edges_as_executable
27828 if check_all_array_refs will be called.
27830 2018-01-23 David Malcolm <dmalcolm@redhat.com>
27832 * tree.c (selftest::test_location_wrappers): Add more test
27835 2018-01-23 David Malcolm <dmalcolm@redhat.com>
27837 * sbitmap.c (selftest::test_set_range): Fix memory leaks.
27838 (selftest::test_bit_in_range): Likewise.
27840 2018-01-23 Richard Sandiford <richard.sandiford@linaro.org>
27843 * doc/sourcebuild.texi (vect_float): Say that the selector
27844 only describes the situation when -funsafe-math-optimizations is on.
27845 (vect_float_strict): Document.
27847 2018-01-23 Richard Sandiford <richard.sandiford@linaro.org>
27849 PR tree-optimization/83965
27850 * tree-vect-patterns.c (vect_reassociating_reduction_p): New function.
27851 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern): Use it
27852 instead of checking only for a reduction.
27853 (vect_recog_widen_sum_pattern): Likewise.
27855 2018-01-23 Jan Hubicka <hubicka@ucw.cz>
27857 * predict.c (probably_never_executed): Only use precise profile info.
27858 (compute_function_frequency): Skip after inlining hack since we now
27859 have quality checking.
27861 2018-01-23 Jan Hubicka <hubicka@ucw.cz>
27863 * profile-count.h (profile_probability::very_unlikely,
27864 profile_probability::unlikely, profile_probability::even): Set
27865 precision to guessed.
27867 2018-01-23 Richard Biener <rguenther@suse.de>
27869 PR tree-optimization/83963
27870 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
27871 Properly terminate dominator walk when crossing the exit edge not
27872 when visiting its source block.
27874 2018-01-23 Jakub Jelinek <jakub@redhat.com>
27877 * tree.c (maybe_wrap_with_location): Use NON_LVALUE_EXPR rather than
27878 VIEW_CONVERT_EXPR to wrap CONST_DECLs.
27880 2018-01-22 Jakub Jelinek <jakub@redhat.com>
27882 PR tree-optimization/83957
27883 * omp-expand.c (expand_omp_for_generic): Ignore virtual PHIs. Remove
27884 semicolon after for body surrounded by braces.
27886 PR tree-optimization/83081
27887 * profile-count.h (profile_probability::split): New method.
27888 * dojump.c (do_jump_1) <case TRUTH_ANDIF_EXPR, case TRUTH_ORIF_EXPR>:
27889 Use profile_probability::split.
27890 (do_compare_rtx_and_jump): Fix adjustment of probabilities
27891 when splitting a single conditional jump into 2.
27893 2018-01-22 David Malcolm <dmalcolm@redhat.com>
27895 PR tree-optimization/69452
27896 * tree-ssa-loop-im.c (class move_computations_dom_walker): Remove
27899 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
27901 * config/rl78/rl78-expand.md (bswaphi2): New define_expand.
27902 * config/rl78/rl78-virt.md (*bswaphi2_virt): New define_insn.
27903 * config/rl78/rl78-real.md (*bswaphi2_real): New define_insn.
27905 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
27907 * config/rl78/rl78-protos.h (rl78_split_movdi): New function
27909 * config/rl78/rl78.md (movdi): New define_expand.
27910 * config/rl78/rl78.c (rl78_split_movdi): New function.
27912 2018-01-22 Michael Meissner <meissner@linux.vnet.ibm.com>
27915 * config/rs6000/rs6000-protos.h (rs6000_split_signbit): Delete,
27917 * config/rs6000/rs6000.c (rs6000_split_signbit): Likewise.
27918 * config/rs6000/rs6000.md (signbit<mode>2): Change code for IEEE
27919 128-bit to produce an UNSPEC move to get the double word with the
27920 signbit and then a shift directly to do signbit.
27921 (signbit<mode>2_dm): Replace old IEEE 128-bit signbit
27922 implementation with a new version that just does either a direct
27923 move or a regular move. Move memory interface to separate insns.
27924 Move insns so they are next to the expander.
27925 (signbit<mode>2_dm_mem_be): New combiner insns to combine load
27926 with signbit move. Split big and little endian case.
27927 (signbit<mode>2_dm_mem_le): Likewise.
27928 (signbit<mode>2_dm_<su>ext): Delete, no longer used.
27929 (signbit<mode>2_dm2): Likewise.
27931 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
27933 * config/rl78/rl78.md (anddi3): New define_expand.
27935 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
27937 * config/rl78/rl78.md (umindi3): New define_expand.
27939 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
27941 * config/rl78/rl78.md (smindi3): New define_expand.
27943 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
27945 * config/rl78/rl78.md (smaxdi3): New define_expand.
27947 2018-01-22 Carl Love <cel@us.ibm.com>
27949 * config/rs6000/rs6000-builtin.def (ST_ELEMREV_V1TI, LD_ELEMREV_V1TI,
27950 LVX_V1TI): Add macro expansion.
27951 * config/rs6000/rs6000-c.c (altivec_builtin_types): Add argument
27952 definitions for VSX_BUILTIN_VEC_XST_BE, VSX_BUILTIN_VEC_ST,
27953 VSX_BUILTIN_VEC_XL, LD_ELEMREV_V1TI builtins.
27954 * config/rs6000/rs6000-p8swap.c (insn_is_swappable_p);
27955 Change check to determine if the instruction is a byte reversing
27956 entry. Fix typo in comment.
27957 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case entry
27958 for VSX_BUILTIN_ST_ELEMREV_V1TI and VSX_BUILTIN_LD_ELEMREV_V1TI.
27959 Add def_builtin calls for new builtins.
27960 * config/rs6000/vsx.md (vsx_st_elemrev_v1ti, vsx_ld_elemrev_v1ti):
27961 Add define_insn expansion.
27963 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
27965 * config/rl78/rl78.md (umaxdi3): New define_expand.
27967 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
27969 * config/rl78/rl78.c (rl78_note_reg_set): Fixed dead reg check
27970 for non-QImode registers.
27972 2018-01-22 Richard Biener <rguenther@suse.de>
27974 PR tree-optimization/83963
27975 * graphite-scop-detection.c (scop_detection::get_sese): Delay
27976 including the loop exit block.
27977 (scop_detection::merge_sese): Likewise.
27978 (scop_detection::add_scop): Do it here instead.
27980 2018-01-22 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27982 * doc/sourcebuild.texi (arm_softfloat): Document.
27984 2018-01-21 John David Anglin <danglin@gcc.gnu.org>
27987 * config/pa/pa.c (pa_function_ok_for_sibcall): Use
27988 targetm.binds_local_p instead of TREE_PUBLIC to check local binding.
27989 Move TARGET_PORTABLE_RUNTIME check after TARGET_64BIT check.
27991 2018-01-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27992 David Edelsohn <dje.gcc@gmail.com>
27995 * config/rs6000/rs6000.md (*call_indirect_nonlocal_sysv<mode>):
27996 Change "crset eq" to "crset 2".
27997 (*call_value_indirect_nonlocal_sysv<mode>): Likewise.
27998 (*call_indirect_aix<mode>_nospec): Likewise.
27999 (*call_value_indirect_aix<mode>_nospec): Likewise.
28000 (*call_indirect_elfv2<mode>_nospec): Likewise.
28001 (*call_value_indirect_elfv2<mode>_nospec): Likewise.
28002 (*sibcall_nonlocal_sysv<mode>): Change "crset eq" to "crset 2";
28003 change assembly output from . to $.
28004 (*sibcall_value_nonlocal_sysv<mode>): Likewise.
28005 (indirect_jump<mode>_nospec): Change assembly output from . to $.
28006 (*tablejump<mode>_internal1_nospec): Likewise.
28008 2018-01-21 Oleg Endo <olegendo@gcc.gnu.org>
28011 * config/sh/sh_optimize_sett_clrt.cc:
28012 Use INCLUDE_ALGORITHM and INCLUDE_VECTOR instead of direct includes.
28014 2018-01-20 Richard Sandiford <richard.sandiford@linaro.org>
28016 PR tree-optimization/83940
28017 * tree-vect-stmts.c (vect_truncate_gather_scatter_offset): Set
28018 offset_dt to vect_constant_def rather than vect_unknown_def_type.
28019 (vect_check_load_store_mask): Add a mask_dt_out parameter and
28020 use it to pass back the definition type.
28021 (vect_check_store_rhs): Likewise rhs_dt_out.
28022 (vect_build_gather_load_calls): Add a mask_dt argument and use
28023 it instead of a call to vect_is_simple_use.
28024 (vectorizable_store): Update calls to vect_check_load_store_mask
28025 and vect_check_store_rhs. Use the dt returned by the latter instead
28026 of scatter_src_dt. Use the cached mask_dt and gs_info.offset_dt
28027 instead of calls to vect_is_simple_use. Pass the scalar rather
28028 than the vector operand to vect_is_simple_use when handling
28029 second and subsequent copies of an rhs value.
28030 (vectorizable_load): Update calls to vect_check_load_store_mask
28031 and vect_build_gather_load_calls. Use the cached mask_dt and
28032 gs_info.offset_dt instead of calls to vect_is_simple_use.
28034 2018-01-20 Jakub Jelinek <jakub@redhat.com>
28036 PR middle-end/83945
28037 * tree-emutls.c: Include gimplify.h.
28038 (lower_emutls_2): New function.
28039 (lower_emutls_1): If ADDR_EXPR is a gimple invariant and walk_tree
28040 with lower_emutls_2 callback finds some TLS decl in it, unshare_expr
28041 it before further processing.
28044 * simplify-rtx.c (simplify_binary_operation_1) <case UMOD>: Use
28045 UINTVAL (trueop1) instead of INTVAL (op1).
28047 2018-01-19 Jakub Jelinek <jakub@redhat.com>
28051 * dwarf2cfi.c (DEFAULT_INCOMING_FRAME_SP_OFFSET): Define to
28052 INCOMING_FRAME_SP_OFFSET if not defined.
28053 (scan_trace): Add ENTRY argument. If true and
28054 DEFAULT_INCOMING_FRAME_SP_OFFSET != INCOMING_FRAME_SP_OFFSET,
28055 emit a note to adjust the CFA offset.
28056 (create_cfi_notes): Adjust scan_trace callers.
28057 (create_cie_data): Use DEFAULT_INCOMING_FRAME_SP_OFFSET rather than
28058 INCOMING_FRAME_SP_OFFSET in the CIE.
28059 * config/i386/i386.h (DEFAULT_INCOMING_FRAME_SP_OFFSET): Define.
28060 * config/stormy16/stormy16.h (DEFAULT_INCOMING_FRAME_SP_OFFSET):
28062 * doc/tm.texi.in (DEFAULT_INCOMING_FRAME_SP_OFFSET): Document.
28063 * doc/tm.texi: Regenerated.
28065 2018-01-19 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28067 PR rtl-optimization/83147
28068 * lra-constraints.c (remove_inheritance_pseudos): Use
28069 lra_substitute_pseudo_within_insn.
28071 2018-01-19 Tom de Vries <tom@codesourcery.com>
28072 Cesar Philippidis <cesar@codesourcery.com>
28075 * config/nvptx/nvptx.c (nvptx_single): Fix jit workaround.
28077 2018-01-19 Cesar Philippidis <cesar@codesourcery.com>
28080 * config/nvptx/nvptx.c (output_init_frag): Don't use generic address
28081 spaces for function labels.
28083 2018-01-19 Martin Liska <mliska@suse.cz>
28085 * predict.def (PRED_LOOP_EXIT): Change from 85 to 89.
28086 (PRED_LOOP_EXIT_WITH_RECURSION): Change from 72 to 78.
28087 (PRED_LOOP_EXTRA_EXIT): Change from 83 to 67.
28088 (PRED_OPCODE_POSITIVE): Change from 64 to 59.
28089 (PRED_TREE_OPCODE_POSITIVE): Change from 64 to 59.
28090 (PRED_CONST_RETURN): Change from 69 to 65.
28091 (PRED_NULL_RETURN): Change from 91 to 71.
28092 (PRED_LOOP_IV_COMPARE_GUESS): Change from 98 to 64.
28093 (PRED_LOOP_GUARD): Change from 66 to 73.
28095 2018-01-19 Martin Liska <mliska@suse.cz>
28097 * predict.c (predict_insn_def): Add new assert.
28098 (struct branch_predictor): Change type to signed integer.
28099 (test_prediction_value_range): Amend test to cover
28100 PROB_UNINITIALIZED.
28101 * predict.def (PRED_LOOP_ITERATIONS): Use the new constant.
28102 (PRED_LOOP_ITERATIONS_GUESSED): Likewise.
28103 (PRED_LOOP_ITERATIONS_MAX): Likewise.
28104 (PRED_LOOP_IV_COMPARE): Likewise.
28105 * predict.h (PROB_UNINITIALIZED): Define new constant.
28107 2018-01-19 Martin Liska <mliska@suse.cz>
28109 * predict.c (dump_prediction): Add new format for
28110 analyze_brprob.py script which is enabled with -details
28112 * profile-count.h (precise_p): New function.
28114 2018-01-19 Richard Sandiford <richard.sandiford@linaro.org>
28116 PR tree-optimization/83922
28117 * tree-vect-loop.c (vect_verify_full_masking): Return false if
28118 there are no statements that need masking.
28119 (vect_active_double_reduction_p): New function.
28120 (vect_analyze_loop_operations): Use it when handling phis that
28121 are not in the loop header.
28123 2018-01-19 Richard Sandiford <richard.sandiford@linaro.org>
28125 PR tree-optimization/83914
28126 * tree-vect-loop.c (vectorizable_induction): Don't convert
28127 init_expr or apply the peeling adjustment for inductions
28128 that are nested within the vectorized loop.
28130 2018-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28132 * config/arm/thumb2.md (*thumb2_negsi2_short): Use RSB mnemonic
28135 2018-01-18 Jakub Jelinek <jakub@redhat.com>
28139 * function.h (gimplify_parameters): Add gimple_seq * argument.
28140 * function.c: Include gimple.h and options.h.
28141 (gimplify_parameters): Add cleanup argument, add CLOBBER stmts
28142 for the added local temporaries if needed.
28143 * gimplify.c (gimplify_body): Adjust gimplify_parameters caller,
28144 if there are any parameter cleanups, wrap whole body into a
28145 try/finally with the cleanups.
28147 2018-01-18 Wilco Dijkstra <wdijkstr@arm.com>
28150 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
28151 Use GET_MODE_CLASS for scalar floating point.
28153 2018-01-18 Jan Hubicka <hubicka@ucw.cz>
28157 * cgraphclones.c (cgraph_node::create_version_clone_with_body):
28158 Fix call of call_cgraph_insertion_hooks.
28160 2018-01-18 Martin Sebor <msebor@redhat.com>
28162 * doc/invoke.texi (-Wclass-memaccess): Tweak text.
28164 2018-01-18 Jan Hubicka <hubicka@ucw.cz>
28167 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Update edge
28170 2018-01-18 Boris Kolpackov <boris@codesynthesis.com>
28173 * common.opt: (-ffile-prefix-map): New option.
28174 * opts.c (common_handle_option): Defer it.
28175 * opts-global.c (handle_common_deferred_options): Handle it.
28176 * debug.h (remap_debug_filename, add_debug_prefix_map): Move to...
28177 * file-prefix-map.h: New file.
28178 (remap_debug_filename, add_debug_prefix_map): ...here.
28179 (add_macro_prefix_map, add_file_prefix_map, remap_macro_filename): New.
28180 * final.c (debug_prefix_map, add_debug_prefix_map
28181 remap_debug_filename): Move to...
28182 * file-prefix-map.c: New file.
28183 (file_prefix_map, add_prefix_map, remap_filename) ...here and rename,
28184 generalize, get rid of alloca(), use strrchr() instead of strchr().
28185 (add_macro_prefix_map, add_debug_prefix_map, add_file_prefix_map):
28186 Implement in terms of add_prefix_map().
28187 (remap_macro_filename, remap_debug_filename): Implement in term of
28189 * Makefile.in (OBJS, PLUGIN_HEADERS): Add new files.
28190 * builtins.c (fold_builtin_FILE): Call remap_macro_filename().
28191 * dbxout.c: Include file-prefix-map.h.
28192 * varasm.c: Likewise.
28193 * vmsdbgout.c: Likewise.
28194 * xcoffout.c: Likewise.
28195 * dwarf2out.c: Likewise plus omit new options from DW_AT_producer.
28196 * doc/cppopts.texi (-fmacro-prefix-map): Document.
28197 * doc/invoke.texi (-ffile-prefix-map): Document.
28198 (-fdebug-prefix-map): Update description.
28200 2018-01-18 Martin Liska <mliska@suse.cz>
28202 * config/i386/i386.c (indirect_thunk_name): Document that also
28204 (output_indirect_thunk): Document why both instructions
28205 (pause and lfence) are generated.
28207 2018-01-18 Richard Biener <rguenther@suse.de>
28209 PR tree-optimization/83887
28210 * graphite-scop-detection.c
28211 (scop_detection::get_nearest_dom_with_single_entry): Remove.
28212 (scop_detection::get_nearest_pdom_with_single_exit): Likewise.
28213 (scop_detection::merge_sese): Re-implement with a flood-fill
28214 algorithm that properly finds a SESE region if it exists.
28216 2018-01-18 Jakub Jelinek <jakub@redhat.com>
28219 * match.pd ((P + A) - P, P - (P + A), (P + A) - (P + B)): For
28220 pointer_diff optimizations use view_convert instead of convert.
28222 2018-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28224 * config/rs6000/rs6000.md (*call_indirect_nonlocal_sysv<mode>):
28225 Generate different code for -mno-speculate-indirect-jumps.
28226 (*call_value_indirect_nonlocal_sysv<mode>): Likewise.
28227 (*call_indirect_aix<mode>): Disable for
28228 -mno-speculate-indirect-jumps.
28229 (*call_indirect_aix<mode>_nospec): New define_insn.
28230 (*call_value_indirect_aix<mode>): Disable for
28231 -mno-speculate-indirect-jumps.
28232 (*call_value_indirect_aix<mode>_nospec): New define_insn.
28233 (*sibcall_nonlocal_sysv<mode>): Generate different code for
28234 -mno-speculate-indirect-jumps.
28235 (*sibcall_value_nonlocal_sysv<mode>): Likewise.
28237 2018-01-17 Michael Meissner <meissner@linux.vnet.ibm.com>
28239 * config/rs6000/rs6000.c (rs6000_emit_move): If we load or store a
28240 long double type, set the flags for noting the default long double
28241 type, even if we don't pass or return a long double type.
28243 2018-01-17 Jan Hubicka <hubicka@ucw.cz>
28246 * ipa-inline.c (flatten_function): Do not overwrite final inlining
28249 2018-01-17 Will Schmidt <will_schmidt@vnet.ibm.com>
28251 * config/rs6000/rs6000.c (rs6000_gimple_builtin): Add gimple folding
28252 support for merge[hl].
28253 (fold_mergehl_helper): New helper function.
28254 (tree-vector-builder.h): New #include for tree_vector_builder usage.
28255 * config/rs6000/altivec.md (altivec_vmrghw_direct): Add xxmrghw insn.
28256 (altivec_vmrglw_direct): Add xxmrglw insn.
28258 2018-01-17 Andrew Waterman <andrew@sifive.com>
28260 * config/riscv/riscv.c (riscv_conditional_register_usage): If
28261 UNITS_PER_FP_ARG is 0, set call_used_regs to 1 for all FP regs.
28263 2018-01-17 David Malcolm <dmalcolm@redhat.com>
28266 * ipa-devirt.c (add_type_duplicate): When comparing memory layout,
28267 call the lto_location_cache before reading the
28268 DECL_SOURCE_LOCATION of the types.
28270 2018-01-17 Wilco Dijkstra <wdijkstr@arm.com>
28271 Richard Sandiford <richard.sandiford@linaro.org>
28273 * config/aarch64/aarch64.md (movti_aarch64): Use Uti constraint.
28274 * config/aarch64/aarch64.c (aarch64_mov128_immediate): New function.
28275 (aarch64_legitimate_constant_p): Just support CONST_DOUBLE
28276 SF/DF/TF mode to avoid creating illegal CONST_WIDE_INT immediates.
28277 * config/aarch64/aarch64-protos.h (aarch64_mov128_immediate):
28279 * config/aarch64/constraints.md (aarch64_movti_operand):
28281 * config/aarch64/predicates.md (Uti): Add new constraint.
28283 2018-01-17 Carl Love <cel@us.ibm.com>
28285 * config/rs6000/vsx.md (define_expand xl_len_r,
28286 define_expand stxvl, define_expand *stxvl): Add match_dup argument.
28287 (define_insn): Add, match_dup 1 argument to define_insn stxvll and
28289 (define_expand, define_insn): Move the shift left from the
28290 define_insn to the define_expand for lxvl and stxvl instructions.
28291 * config/rs6000/rs6000-builtin.def (BU_P9V_64BIT_VSX_2): Change LXVL
28292 and XL_LEN_R definitions to PURE.
28294 2018-01-17 Uros Bizjak <ubizjak@gmail.com>
28296 * config/i386/i386.c (indirect_thunk_name): Declare regno
28297 as unsigned int. Compare regno with INVALID_REGNUM.
28298 (output_indirect_thunk): Ditto.
28299 (output_indirect_thunk_function): Ditto.
28300 (ix86_code_end): Declare regno as unsigned int. Use INVALID_REGNUM
28301 in the call to output_indirect_thunk_function.
28303 2018-01-17 Richard Sandiford <richard.sandiford@linaro.org>
28305 PR middle-end/83884
28306 * expr.c (expand_expr_real_1): Use the size of GET_MODE (op0)
28307 rather than the size of inner_type to determine the stack slot size
28308 when handling VIEW_CONVERT_EXPRs on strict-alignment targets.
28310 2018-01-16 Sebastian Peryt <sebastian.peryt@intel.com>
28313 * config/i386/i386.c (ix86_option_override_internal): Add PTA_RDRND
28316 2018-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
28318 * config.gcc (powerpc*-linux*-*): Add support for 64-bit little
28319 endian Linux systems to optionally enable multilibs for selecting
28320 the long double type if the user configured an explicit type.
28321 * config/rs6000/rs6000.h (TARGET_IEEEQUAD_MULTILIB): Indicate we
28322 have no long double multilibs if not defined.
28323 * config/rs6000/rs6000.c (rs6000_option_override_internal): Do not
28324 warn if the user used -mabi={ieee,ibm}longdouble and we built
28325 multilibs for long double.
28326 * config/rs6000/linux64.h (MULTILIB_DEFAULTS_IEEE): Define as the
28327 appropriate multilib option.
28328 (MULTILIB_DEFAULTS): Add MULTILIB_DEFAULTS_IEEE to the default
28330 * config/rs6000/t-ldouble-linux64le-ibm: New configuration files
28331 for building long double multilibs.
28332 * config/rs6000/t-ldouble-linux64le-ieee: Likewise.
28334 2018-01-16 John David Anglin <danglin@gcc.gnu.org>
28336 * config.gcc (hppa*-*-linux*): Change callee copies ABI to caller
28339 * config/pa.h (MALLOC_ABI_ALIGNMENT): Set 32-bit alignment default to
28341 * config/pa/pa32-linux.h (MALLOC_ABI_ALIGNMENT): Set alignment to
28344 * config/pa/som.h (ASM_DECLARE_FUNCTION_NAME): Cleanup type and mode
28347 * config/pa/pa.c (pa_function_arg_size): Apply CEIL to GET_MODE_SIZE
28350 2018-01-16 Eric Botcazou <ebotcazou@adacore.com>
28352 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): For an
28353 ADDR_EXPR, do not count the offset of a COMPONENT_REF twice.
28355 2018-01-16 Kelvin Nilsen <kelvin@gcc.gnu.org>
28357 * config/rs6000/rs6000-p8swap.c (rs6000_gen_stvx): Generate
28358 different rtl trees depending on TARGET_64BIT.
28359 (rs6000_gen_lvx): Likewise.
28361 2018-01-16 Eric Botcazou <ebotcazou@adacore.com>
28363 * config/visium/visium.md (nop): Tweak comment.
28364 (hazard_nop): Likewise.
28366 2018-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28368 * config/rs6000/rs6000.c (rs6000_opt_vars): Add entry for
28369 -mspeculate-indirect-jumps.
28370 * config/rs6000/rs6000.md (*call_indirect_elfv2<mode>): Disable
28371 for -mno-speculate-indirect-jumps.
28372 (*call_indirect_elfv2<mode>_nospec): New define_insn.
28373 (*call_value_indirect_elfv2<mode>): Disable for
28374 -mno-speculate-indirect-jumps.
28375 (*call_value_indirect_elfv2<mode>_nospec): New define_insn.
28376 (indirect_jump): Emit different RTL for
28377 -mno-speculate-indirect-jumps.
28378 (*indirect_jump<mode>): Disable for
28379 -mno-speculate-indirect-jumps.
28380 (*indirect_jump<mode>_nospec): New define_insn.
28381 (tablejump): Emit different RTL for
28382 -mno-speculate-indirect-jumps.
28383 (tablejumpsi): Disable for -mno-speculate-indirect-jumps.
28384 (tablejumpsi_nospec): New define_expand.
28385 (tablejumpdi): Disable for -mno-speculate-indirect-jumps.
28386 (tablejumpdi_nospec): New define_expand.
28387 (*tablejump<mode>_internal1): Disable for
28388 -mno-speculate-indirect-jumps.
28389 (*tablejump<mode>_internal1_nospec): New define_insn.
28390 * config/rs6000/rs6000.opt (mspeculate-indirect-jumps): New
28393 2018-01-16 Artyom Skrobov tyomitch@gmail.com
28395 * caller-save.c (insert_save): Drop unnecessary parameter. All
28398 2018-01-16 Jakub Jelinek <jakub@redhat.com>
28399 Richard Biener <rguenth@suse.de>
28402 * gimplify.c (gimplify_one_sizepos): For is_gimple_constant (expr)
28403 return early, inline manually is_gimple_sizepos. Make sure if we
28404 call gimplify_expr we don't end up with a gimple constant.
28405 * tree.c (variably_modified_type_p): Don't return true for
28406 is_gimple_constant (_t). Inline manually is_gimple_sizepos.
28407 * gimplify.h (is_gimple_sizepos): Remove.
28409 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
28411 PR tree-optimization/83857
28412 * tree-vect-loop.c (vect_analyze_loop_operations): Don't call
28413 vectorizable_live_operation for pure SLP statements.
28414 (vectorizable_live_operation): Handle PHIs.
28416 2018-01-16 Richard Biener <rguenther@suse.de>
28418 PR tree-optimization/83867
28419 * tree-vect-stmts.c (vect_transform_stmt): Precompute
28420 nested_in_vect_loop_p since the scalar stmt may get invalidated.
28422 2018-01-16 Jakub Jelinek <jakub@redhat.com>
28425 * stor-layout.c (handle_warn_if_not_align): Use byte_position and
28426 multiple_of_p instead of unchecked tree_to_uhwi and UHWI check.
28427 If off is not INTEGER_CST, issue a may not be aligned warning
28428 rather than isn't aligned. Use isn%'t rather than isn't.
28429 * fold-const.c (multiple_of_p) <case BIT_AND_EXPR>: Don't fall through
28431 <case MULT_EXPR>: Improve the case when bottom and one of the
28432 MULT_EXPR operands are INTEGER_CSTs and bottom is multiple of that
28433 operand, in that case check if the other operand is multiple of
28434 bottom divided by the INTEGER_CST operand.
28436 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
28439 * config/pa/pa.h (FUNCTION_ARG_SIZE): Delete.
28440 * config/pa/pa-protos.h (pa_function_arg_size): Declare.
28441 * config/pa/som.h (ASM_DECLARE_FUNCTION_NAME): Use
28442 pa_function_arg_size instead of FUNCTION_ARG_SIZE.
28443 * config/pa/pa.c (pa_function_arg_advance): Likewise.
28444 (pa_function_arg, pa_arg_partial_bytes): Likewise.
28445 (pa_function_arg_size): New function.
28447 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
28449 * fold-const.c (fold_ternary_loc): Construct the vec_perm_indices
28450 in a separate statement.
28452 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
28454 PR tree-optimization/83847
28455 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Don't
28456 group gathers and scatters.
28458 2018-01-16 Jakub Jelinek <jakub@redhat.com>
28460 PR rtl-optimization/86620
28461 * params.def (max-sched-ready-insns): Bump minimum value to 1.
28463 PR rtl-optimization/83213
28464 * recog.c (peep2_attempt): Copy over CROSSING_JUMP_P from peepinsn
28465 to last if both are JUMP_INSNs.
28467 PR tree-optimization/83843
28468 * gimple-ssa-store-merging.c
28469 (imm_store_chain_info::output_merged_store): Handle bit_not_p on
28470 store_immediate_info for bswap/nop orig_stores.
28472 2018-01-15 Andrew Waterman <andrew@sifive.com>
28474 * config/riscv/riscv.c (riscv_rtx_costs) <MULT>: Increase cost if
28476 <UDIV>: Increase cost if !TARGET_DIV.
28478 2018-01-15 Segher Boessenkool <segher@kernel.crashing.org>
28480 * config/rs6000/rs6000.md (define_attr "type"): Remove delayed_cr.
28481 (define_attr "cr_logical_3op"): New.
28482 (cceq_ior_compare): Adjust.
28483 (cceq_ior_compare_complement): Adjust.
28484 (*cceq_rev_compare): Adjust.
28485 * config/rs6000/rs6000.c (rs6000_adjust_cost): Adjust.
28486 (is_cracked_insn): Adjust.
28487 (insn_must_be_first_in_group): Adjust.
28488 * config/rs6000/40x.md: Adjust.
28489 * config/rs6000/440.md: Adjust.
28490 * config/rs6000/476.md: Adjust.
28491 * config/rs6000/601.md: Adjust.
28492 * config/rs6000/603.md: Adjust.
28493 * config/rs6000/6xx.md: Adjust.
28494 * config/rs6000/7450.md: Adjust.
28495 * config/rs6000/7xx.md: Adjust.
28496 * config/rs6000/8540.md: Adjust.
28497 * config/rs6000/cell.md: Adjust.
28498 * config/rs6000/e300c2c3.md: Adjust.
28499 * config/rs6000/e500mc.md: Adjust.
28500 * config/rs6000/e500mc64.md: Adjust.
28501 * config/rs6000/e5500.md: Adjust.
28502 * config/rs6000/e6500.md: Adjust.
28503 * config/rs6000/mpc.md: Adjust.
28504 * config/rs6000/power4.md: Adjust.
28505 * config/rs6000/power5.md: Adjust.
28506 * config/rs6000/power6.md: Adjust.
28507 * config/rs6000/power7.md: Adjust.
28508 * config/rs6000/power8.md: Adjust.
28509 * config/rs6000/power9.md: Adjust.
28510 * config/rs6000/rs64.md: Adjust.
28511 * config/rs6000/titan.md: Adjust.
28513 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
28515 * config/i386/predicates.md (indirect_branch_operand): Rewrite
28516 ix86_indirect_branch_register logic.
28518 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
28520 * config/i386/constraints.md (Bs): Update
28521 ix86_indirect_branch_register check. Don't check
28522 ix86_indirect_branch_register with GOT_memory_operand.
28524 * config/i386/predicates.md (GOT_memory_operand): Don't check
28525 ix86_indirect_branch_register here.
28526 (GOT32_symbol_operand): Likewise.
28528 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
28530 * config/i386/predicates.md (constant_call_address_operand):
28531 Rewrite ix86_indirect_branch_register logic.
28532 (sibcall_insn_operand): Likewise.
28534 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
28536 * config/i386/constraints.md (Bs): Replace
28537 ix86_indirect_branch_thunk_register with
28538 ix86_indirect_branch_register.
28540 * config/i386/i386.md (indirect_jump): Likewise.
28541 (tablejump): Likewise.
28542 (*sibcall_memory): Likewise.
28543 (*sibcall_value_memory): Likewise.
28544 Peepholes of indirect call and jump via memory: Likewise.
28545 * config/i386/i386.opt: Likewise.
28546 * config/i386/predicates.md (indirect_branch_operand): Likewise.
28547 (GOT_memory_operand): Likewise.
28548 (call_insn_operand): Likewise.
28549 (sibcall_insn_operand): Likewise.
28550 (GOT32_symbol_operand): Likewise.
28552 2018-01-15 Jakub Jelinek <jakub@redhat.com>
28554 PR middle-end/83837
28555 * omp-expand.c (expand_omp_atomic_pipeline): Use loaded_val
28556 type rather than type addr's type points to.
28557 (expand_omp_atomic_mutex): Likewise.
28558 (expand_omp_atomic): Likewise.
28560 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
28563 * config/i386/i386.c (output_indirect_thunk_function): Use
28564 ASM_OUTPUT_LABEL, instead of ASM_OUTPUT_DEF, for TARGET_MACHO
28565 for __x86_return_thunk.
28567 2018-01-15 Richard Biener <rguenther@suse.de>
28569 PR middle-end/83850
28570 * expmed.c (extract_bit_field_1): Fix typo.
28572 2018-01-15 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28575 * config/arm/iterators.md (VF): New mode iterator.
28576 * config/arm/neon.md (neon_vabd<mode>_2): Use the above.
28577 Remove integer-related logic from pattern.
28578 (neon_vabd<mode>_3): Likewise.
28580 2018-01-15 Jakub Jelinek <jakub@redhat.com>
28582 PR middle-end/82694
28583 * common.opt (fstrict-overflow): No longer an alias.
28584 (fwrapv-pointer): New option.
28585 * tree.h (TYPE_OVERFLOW_WRAPS, TYPE_OVERFLOW_UNDEFINED): Define
28586 also for pointer types based on flag_wrapv_pointer.
28587 * opts.c (common_handle_option) <case OPT_fstrict_overflow>: Set
28588 opts->x_flag_wrap[pv] to !value, clear opts->x_flag_trapv if
28589 opts->x_flag_wrapv got set.
28590 * fold-const.c (fold_comparison, fold_binary_loc): Revert 2017-08-01
28591 changes, just use TYPE_OVERFLOW_UNDEFINED on pointer type instead of
28592 POINTER_TYPE_OVERFLOW_UNDEFINED.
28593 * match.pd: Likewise in address comparison pattern.
28594 * doc/invoke.texi: Document -fwrapv and -fstrict-overflow.
28596 2018-01-15 Richard Biener <rguenther@suse.de>
28599 * tree.c (free_lang_data_in_type): Always unlink TYPE_DECLs
28600 from TYPE_FIELDS. Free TYPE_BINFO if not used by devirtualization.
28601 Reset type names to their identifier if their TYPE_DECL doesn't
28602 have linkage (and thus is used for ODR and devirt).
28603 (save_debug_info_for_decl): Remove.
28604 (save_debug_info_for_type): Likewise.
28605 (add_tree_to_fld_list): Adjust.
28606 * tree-pretty-print.c (dump_generic_node): Make dumping of
28607 type names more robust.
28609 2018-01-15 Richard Biener <rguenther@suse.de>
28611 * BASE-VER: Bump to 8.0.1.
28613 2018-01-14 Martin Sebor <msebor@redhat.com>
28616 * builtins.c (check_access): Avoid warning when the no-warning bit
28619 2018-01-14 Cory Fields <cory-nospam-@coryfields.com>
28621 * tree-ssa-loop-im.c (sort_bbs_in_loop_postorder_cmp): Stabilize sort.
28622 * ira-color (allocno_hard_regs_compare): Likewise.
28624 2018-01-14 Nathan Rossi <nathan@nathanrossi.com>
28627 * config/microblaze/microblaze.c (microblaze_asm_output_ident):
28628 Use .pushsection/.popsection.
28630 2018-01-14 Martin Sebor <msebor@redhat.com>
28633 * doc/invoke.texi (-Wlass-memaccess): Document suppression by casting.
28635 2018-01-14 Jakub Jelinek <jakub@redhat.com>
28637 * config.gcc (i[34567]86-*-*): Remove one duplicate gfniintrin.h
28638 entry from extra_headers.
28639 (x86_64-*-*): Remove two duplicate gfniintrin.h entries from
28640 extra_headers, make the list bitwise identical to the i?86-*-* one.
28642 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
28644 * config/i386/i386.c (ix86_set_indirect_branch_type): Disallow
28645 -mcmodel=large with -mindirect-branch=thunk,
28646 -mindirect-branch=thunk-extern, -mfunction-return=thunk and
28647 -mfunction-return=thunk-extern.
28648 * doc/invoke.texi: Document -mcmodel=large is incompatible with
28649 -mindirect-branch=thunk, -mindirect-branch=thunk-extern,
28650 -mfunction-return=thunk and -mfunction-return=thunk-extern.
28652 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
28654 * config/i386/i386.c (print_reg): Print the name of the full
28655 integer register without '%'.
28656 (ix86_print_operand): Handle 'V'.
28657 * doc/extend.texi: Document 'V' modifier.
28659 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
28661 * config/i386/constraints.md (Bs): Disallow memory operand for
28662 -mindirect-branch-register.
28664 * config/i386/predicates.md (indirect_branch_operand): Likewise.
28665 (GOT_memory_operand): Likewise.
28666 (call_insn_operand): Likewise.
28667 (sibcall_insn_operand): Likewise.
28668 (GOT32_symbol_operand): Likewise.
28669 * config/i386/i386.md (indirect_jump): Call convert_memory_address
28670 for -mindirect-branch-register.
28671 (tablejump): Likewise.
28672 (*sibcall_memory): Likewise.
28673 (*sibcall_value_memory): Likewise.
28674 Disallow peepholes of indirect call and jump via memory for
28675 -mindirect-branch-register.
28676 (*call_pop): Replace m with Bw.
28677 (*call_value_pop): Likewise.
28678 (*sibcall_pop_memory): Replace m with Bs.
28679 * config/i386/i386.opt (mindirect-branch-register): New option.
28680 * doc/invoke.texi: Document -mindirect-branch-register option.
28682 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
28684 * config/i386/i386-protos.h (ix86_output_function_return): New.
28685 * config/i386/i386.c (ix86_set_indirect_branch_type): Also
28686 set function_return_type.
28687 (indirect_thunk_name): Add ret_p to indicate thunk for function
28689 (output_indirect_thunk_function): Pass false to
28690 indirect_thunk_name.
28691 (ix86_output_indirect_branch_via_reg): Likewise.
28692 (ix86_output_indirect_branch_via_push): Likewise.
28693 (output_indirect_thunk_function): Create alias for function
28694 return thunk if regno < 0.
28695 (ix86_output_function_return): New function.
28696 (ix86_handle_fndecl_attribute): Handle function_return.
28697 (ix86_attribute_table): Add function_return.
28698 * config/i386/i386.h (machine_function): Add
28699 function_return_type.
28700 * config/i386/i386.md (simple_return_internal): Use
28701 ix86_output_function_return.
28702 (simple_return_internal_long): Likewise.
28703 * config/i386/i386.opt (mfunction-return=): New option.
28704 (indirect_branch): Mention -mfunction-return=.
28705 * doc/extend.texi: Document function_return function attribute.
28706 * doc/invoke.texi: Document -mfunction-return= option.
28708 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
28710 * config/i386/i386-opts.h (indirect_branch): New.
28711 * config/i386/i386-protos.h (ix86_output_indirect_jmp): Likewise.
28712 * config/i386/i386.c (ix86_using_red_zone): Disallow red-zone
28713 with local indirect jump when converting indirect call and jump.
28714 (ix86_set_indirect_branch_type): New.
28715 (ix86_set_current_function): Call ix86_set_indirect_branch_type.
28716 (indirectlabelno): New.
28717 (indirect_thunk_needed): Likewise.
28718 (indirect_thunk_bnd_needed): Likewise.
28719 (indirect_thunks_used): Likewise.
28720 (indirect_thunks_bnd_used): Likewise.
28721 (INDIRECT_LABEL): Likewise.
28722 (indirect_thunk_name): Likewise.
28723 (output_indirect_thunk): Likewise.
28724 (output_indirect_thunk_function): Likewise.
28725 (ix86_output_indirect_branch_via_reg): Likewise.
28726 (ix86_output_indirect_branch_via_push): Likewise.
28727 (ix86_output_indirect_branch): Likewise.
28728 (ix86_output_indirect_jmp): Likewise.
28729 (ix86_code_end): Call output_indirect_thunk_function if needed.
28730 (ix86_output_call_insn): Call ix86_output_indirect_branch if
28732 (ix86_handle_fndecl_attribute): Handle indirect_branch.
28733 (ix86_attribute_table): Add indirect_branch.
28734 * config/i386/i386.h (machine_function): Add indirect_branch_type
28735 and has_local_indirect_jump.
28736 * config/i386/i386.md (indirect_jump): Set has_local_indirect_jump
28738 (tablejump): Likewise.
28739 (*indirect_jump): Use ix86_output_indirect_jmp.
28740 (*tablejump_1): Likewise.
28741 (simple_return_indirect_internal): Likewise.
28742 * config/i386/i386.opt (mindirect-branch=): New option.
28743 (indirect_branch): New.
28746 (thunk-inline): Likewise.
28747 (thunk-extern): Likewise.
28748 * doc/extend.texi: Document indirect_branch function attribute.
28749 * doc/invoke.texi: Document -mindirect-branch= option.
28751 2018-01-14 Jan Hubicka <hubicka@ucw.cz>
28754 * ipa-inline.c (edge_badness): Tolerate roundoff errors.
28756 2018-01-14 Richard Sandiford <richard.sandiford@linaro.org>
28758 * ipa-inline.c (want_inline_small_function_p): Return false if
28759 inlining has already failed with CIF_FINAL_ERROR.
28760 (update_caller_keys): Call want_inline_small_function_p before
28762 (update_callee_keys): Likewise.
28764 2018-01-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
28766 * config/rs6000/rs6000-p8swap.c (rs6000_sum_of_two_registers_p):
28768 (rs6000_quadword_masked_address_p): Likewise.
28769 (quad_aligned_load_p): Likewise.
28770 (quad_aligned_store_p): Likewise.
28771 (const_load_sequence_p): Add comment to describe the outer-most loop.
28772 (mimic_memory_attributes_and_flags): New function.
28773 (rs6000_gen_stvx): Likewise.
28774 (replace_swapped_aligned_store): Likewise.
28775 (rs6000_gen_lvx): Likewise.
28776 (replace_swapped_aligned_load): Likewise.
28777 (replace_swapped_load_constant): Capitalize argument name in
28778 comment describing this function.
28779 (rs6000_analyze_swaps): Add a third pass to search for vector loads
28780 and stores that access quad-word aligned addresses and replace
28781 with stvx or lvx instructions when appropriate.
28782 * config/rs6000/rs6000-protos.h (rs6000_sum_of_two_registers_p):
28783 New function prototype.
28784 (rs6000_quadword_masked_address_p): Likewise.
28785 (rs6000_gen_lvx): Likewise.
28786 (rs6000_gen_stvx): Likewise.
28787 * config/rs6000/vsx.md (*vsx_le_perm_load_<mode>): For modes
28788 VSX_D (V2DF, V2DI), modify this split to select lvx instruction
28789 when memory address is aligned.
28790 (*vsx_le_perm_load_<mode>): For modes VSX_W (V4SF, V4SI), modify
28791 this split to select lvx instruction when memory address is aligned.
28792 (*vsx_le_perm_load_v8hi): Modify this split to select lvx
28793 instruction when memory address is aligned.
28794 (*vsx_le_perm_load_v16qi): Likewise.
28795 (four unnamed splitters): Modify to select the stvx instruction
28796 when memory is aligned.
28798 2018-01-13 Jan Hubicka <hubicka@ucw.cz>
28800 * predict.c (determine_unlikely_bbs): Handle correctly BBs
28801 which appears in the queue multiple times.
28803 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
28804 Alan Hayward <alan.hayward@arm.com>
28805 David Sherwood <david.sherwood@arm.com>
28807 * tree-vectorizer.h (vec_lower_bound): New structure.
28808 (_loop_vec_info): Add check_nonzero and lower_bounds.
28809 (LOOP_VINFO_CHECK_NONZERO): New macro.
28810 (LOOP_VINFO_LOWER_BOUNDS): Likewise.
28811 (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Check lower_bounds too.
28812 * tree-data-ref.h (dr_with_seg_len): Add access_size and align
28813 fields. Make seg_len the distance travelled, not including the
28815 (dr_direction_indicator): Declare.
28816 (dr_zero_step_indicator): Likewise.
28817 (dr_known_forward_stride_p): Likewise.
28818 * tree-data-ref.c: Include stringpool.h, tree-vrp.h and
28820 (runtime_alias_check_p): Allow runtime alias checks with
28822 (operator ==): Compare access_size and align.
28823 (prune_runtime_alias_test_list): Rework for new distinction between
28824 the access_size and seg_len.
28825 (create_intersect_range_checks_index): Likewise. Cope with polynomial
28827 (get_segment_min_max): New function.
28828 (create_intersect_range_checks): Use it.
28829 (dr_step_indicator): New function.
28830 (dr_direction_indicator): Likewise.
28831 (dr_zero_step_indicator): Likewise.
28832 (dr_known_forward_stride_p): Likewise.
28833 * tree-loop-distribution.c (data_ref_segment_size): Return
28834 DR_STEP * (niters - 1).
28835 (compute_alias_check_pairs): Update call to the dr_with_seg_len
28837 * tree-vect-data-refs.c (vect_check_nonzero_value): New function.
28838 (vect_preserves_scalar_order_p): New function, split out from...
28839 (vect_analyze_data_ref_dependence): ...here. Check for zero steps.
28840 (vect_vfa_segment_size): Return DR_STEP * (length_factor - 1).
28841 (vect_vfa_access_size): New function.
28842 (vect_vfa_align): Likewise.
28843 (vect_compile_time_alias): Take access_size_a and access_b arguments.
28844 (dump_lower_bound): New function.
28845 (vect_check_lower_bound): Likewise.
28846 (vect_small_gap_p): Likewise.
28847 (vectorizable_with_step_bound_p): Likewise.
28848 (vect_prune_runtime_alias_test_list): Ignore cross-iteration
28849 depencies if the vectorization factor is 1. Convert the checks
28850 for nonzero steps into checks on the bounds of DR_STEP. Try using
28851 a bunds check for variable steps if the minimum required step is
28852 relatively small. Update calls to the dr_with_seg_len
28853 constructor and to vect_compile_time_alias.
28854 * tree-vect-loop-manip.c (vect_create_cond_for_lower_bounds): New
28856 (vect_loop_versioning): Call it.
28857 * tree-vect-loop.c (vect_analyze_loop_2): Clear LOOP_VINFO_LOWER_BOUNDS
28859 (vect_estimate_min_profitable_iters): Account for any bounds checks.
28861 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
28862 Alan Hayward <alan.hayward@arm.com>
28863 David Sherwood <david.sherwood@arm.com>
28865 * doc/sourcebuild.texi (vect_scatter_store): Document.
28866 * optabs.def (scatter_store_optab, mask_scatter_store_optab): New
28868 * doc/md.texi (scatter_store@var{m}, mask_scatter_store@var{m}):
28870 * genopinit.c (main): Add supports_vec_scatter_store and
28871 supports_vec_scatter_store_cached to target_optabs.
28872 * gimple.h (gimple_expr_type): Handle IFN_SCATTER_STORE and
28873 IFN_MASK_SCATTER_STORE.
28874 * internal-fn.def (SCATTER_STORE, MASK_SCATTER_STORE): New internal
28876 * internal-fn.h (internal_store_fn_p): Declare.
28877 (internal_fn_stored_value_index): Likewise.
28878 * internal-fn.c (scatter_store_direct): New macro.
28879 (expand_scatter_store_optab_fn): New function.
28880 (direct_scatter_store_optab_supported_p): New macro.
28881 (internal_store_fn_p): New function.
28882 (internal_gather_scatter_fn_p): Handle IFN_SCATTER_STORE and
28883 IFN_MASK_SCATTER_STORE.
28884 (internal_fn_mask_index): Likewise.
28885 (internal_fn_stored_value_index): New function.
28886 (internal_gather_scatter_fn_supported_p): Adjust operand numbers
28887 for scatter stores.
28888 * optabs-query.h (supports_vec_scatter_store_p): Declare.
28889 * optabs-query.c (supports_vec_scatter_store_p): New function.
28890 * tree-vectorizer.h (vect_get_store_rhs): Declare.
28891 * tree-vect-data-refs.c (vect_analyze_data_ref_access): Return
28892 true for scatter stores.
28893 (vect_gather_scatter_fn_p): Handle scatter stores too.
28894 (vect_check_gather_scatter): Consider using scatter stores if
28895 supports_vec_scatter_store_p.
28896 * tree-vect-patterns.c (vect_try_gather_scatter_pattern): Handle
28897 scatter stores too.
28898 * tree-vect-stmts.c (exist_non_indexing_operands_for_use_p): Use
28899 internal_fn_stored_value_index.
28900 (check_load_store_masking): Handle scatter stores too.
28901 (vect_get_store_rhs): Make public.
28902 (vectorizable_call): Use internal_store_fn_p.
28903 (vectorizable_store): Handle scatter store internal functions.
28904 (vect_transform_stmt): Compare GROUP_STORE_COUNT with GROUP_SIZE
28905 when deciding whether the end of the group has been reached.
28906 * config/aarch64/aarch64.md (UNSPEC_ST1_SCATTER): New unspec.
28907 * config/aarch64/aarch64-sve.md (scatter_store<mode>): New expander.
28908 (mask_scatter_store<mode>): New insns.
28910 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
28911 Alan Hayward <alan.hayward@arm.com>
28912 David Sherwood <david.sherwood@arm.com>
28914 * tree-vectorizer.h (vect_gather_scatter_fn_p): Declare.
28915 * tree-vect-data-refs.c (vect_gather_scatter_fn_p): Make public.
28916 * tree-vect-stmts.c (vect_truncate_gather_scatter_offset): New
28918 (vect_use_strided_gather_scatters_p): Take a masked_p argument.
28919 Use vect_truncate_gather_scatter_offset if we can't treat the
28920 operation as a normal gather load or scatter store.
28921 (get_group_load_store_type): Take the gather_scatter_info
28922 as argument. Try using a gather load or scatter store for
28923 single-element groups.
28924 (get_load_store_type): Update calls to get_group_load_store_type
28925 and vect_use_strided_gather_scatters_p.
28927 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
28928 Alan Hayward <alan.hayward@arm.com>
28929 David Sherwood <david.sherwood@arm.com>
28931 * tree-vectorizer.h (vect_create_data_ref_ptr): Take an extra
28932 optional tree argument.
28933 * tree-vect-data-refs.c (vect_check_gather_scatter): Check for
28935 (vect_create_data_ref_ptr): Take the iv_step as an optional argument,
28936 but continue to use the current value as a fallback.
28937 (bump_vector_ptr): Use operand_equal_p rather than tree_int_cst_compare
28938 to compare the updates.
28939 * tree-vect-stmts.c (vect_use_strided_gather_scatters_p): New function.
28940 (get_load_store_type): Use it when handling a strided access.
28941 (vect_get_strided_load_store_ops): New function.
28942 (vect_get_data_ptr_increment): Likewise.
28943 (vectorizable_load): Handle strided gather loads. Always pass
28944 a step to vect_create_data_ref_ptr and bump_vector_ptr.
28946 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
28947 Alan Hayward <alan.hayward@arm.com>
28948 David Sherwood <david.sherwood@arm.com>
28950 * doc/md.texi (gather_load@var{m}): Document.
28951 (mask_gather_load@var{m}): Likewise.
28952 * genopinit.c (main): Add supports_vec_gather_load and
28953 supports_vec_gather_load_cached to target_optabs.
28954 * optabs-tree.c (init_tree_optimization_optabs): Use
28955 ggc_cleared_alloc to allocate target_optabs.
28956 * optabs.def (gather_load_optab, mask_gather_laod_optab): New optabs.
28957 * internal-fn.def (GATHER_LOAD, MASK_GATHER_LOAD): New internal
28959 * internal-fn.h (internal_load_fn_p): Declare.
28960 (internal_gather_scatter_fn_p): Likewise.
28961 (internal_fn_mask_index): Likewise.
28962 (internal_gather_scatter_fn_supported_p): Likewise.
28963 * internal-fn.c (gather_load_direct): New macro.
28964 (expand_gather_load_optab_fn): New function.
28965 (direct_gather_load_optab_supported_p): New macro.
28966 (direct_internal_fn_optab): New function.
28967 (internal_load_fn_p): Likewise.
28968 (internal_gather_scatter_fn_p): Likewise.
28969 (internal_fn_mask_index): Likewise.
28970 (internal_gather_scatter_fn_supported_p): Likewise.
28971 * optabs-query.c (supports_at_least_one_mode_p): New function.
28972 (supports_vec_gather_load_p): Likewise.
28973 * optabs-query.h (supports_vec_gather_load_p): Declare.
28974 * tree-vectorizer.h (gather_scatter_info): Add ifn, element_type
28975 and memory_type field.
28976 (NUM_PATTERNS): Bump to 15.
28977 * tree-vect-data-refs.c: Include internal-fn.h.
28978 (vect_gather_scatter_fn_p): New function.
28979 (vect_describe_gather_scatter_call): Likewise.
28980 (vect_check_gather_scatter): Try using internal functions for
28981 gather loads. Recognize existing calls to a gather load function.
28982 (vect_analyze_data_refs): Consider using gather loads if
28983 supports_vec_gather_load_p.
28984 * tree-vect-patterns.c (vect_get_load_store_mask): New function.
28985 (vect_get_gather_scatter_offset_type): Likewise.
28986 (vect_convert_mask_for_vectype): Likewise.
28987 (vect_add_conversion_to_patterm): Likewise.
28988 (vect_try_gather_scatter_pattern): Likewise.
28989 (vect_recog_gather_scatter_pattern): New pattern recognizer.
28990 (vect_vect_recog_func_ptrs): Add it.
28991 * tree-vect-stmts.c (exist_non_indexing_operands_for_use_p): Use
28992 internal_fn_mask_index and internal_gather_scatter_fn_p.
28993 (check_load_store_masking): Take the gather_scatter_info as an
28994 argument and handle gather loads.
28995 (vect_get_gather_scatter_ops): New function.
28996 (vectorizable_call): Check internal_load_fn_p.
28997 (vectorizable_load): Likewise. Handle gather load internal
28999 (vectorizable_store): Update call to check_load_store_masking.
29000 * config/aarch64/aarch64.md (UNSPEC_LD1_GATHER): New unspec.
29001 * config/aarch64/iterators.md (SVE_S, SVE_D): New mode iterators.
29002 * config/aarch64/predicates.md (aarch64_gather_scale_operand_w)
29003 (aarch64_gather_scale_operand_d): New predicates.
29004 * config/aarch64/aarch64-sve.md (gather_load<mode>): New expander.
29005 (mask_gather_load<mode>): New insns.
29007 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
29008 Alan Hayward <alan.hayward@arm.com>
29009 David Sherwood <david.sherwood@arm.com>
29011 * optabs.def (fold_left_plus_optab): New optab.
29012 * doc/md.texi (fold_left_plus_@var{m}): Document.
29013 * internal-fn.def (IFN_FOLD_LEFT_PLUS): New internal function.
29014 * internal-fn.c (fold_left_direct): Define.
29015 (expand_fold_left_optab_fn): Likewise.
29016 (direct_fold_left_optab_supported_p): Likewise.
29017 * fold-const-call.c (fold_const_fold_left): New function.
29018 (fold_const_call): Use it to fold CFN_FOLD_LEFT_PLUS.
29019 * tree-parloops.c (valid_reduction_p): New function.
29020 (gather_scalar_reductions): Use it.
29021 * tree-vectorizer.h (FOLD_LEFT_REDUCTION): New vect_reduction_type.
29022 (vect_finish_replace_stmt): Declare.
29023 * tree-vect-loop.c (fold_left_reduction_fn): New function.
29024 (needs_fold_left_reduction_p): New function, split out from...
29025 (vect_is_simple_reduction): ...here. Accept reductions that
29026 forbid reassociation, but give them type FOLD_LEFT_REDUCTION.
29027 (vect_force_simple_reduction): Also store the reduction type in
29028 the assignment's STMT_VINFO_REDUC_TYPE.
29029 (vect_model_reduction_cost): Handle FOLD_LEFT_REDUCTION.
29030 (merge_with_identity): New function.
29031 (vect_expand_fold_left): Likewise.
29032 (vectorize_fold_left_reduction): Likewise.
29033 (vectorizable_reduction): Handle FOLD_LEFT_REDUCTION. Leave the
29034 scalar phi in place for it. Check for target support and reject
29035 cases that would reassociate the operation. Defer the transform
29036 phase to vectorize_fold_left_reduction.
29037 * config/aarch64/aarch64.md (UNSPEC_FADDA): New unspec.
29038 * config/aarch64/aarch64-sve.md (fold_left_plus_<mode>): New expander.
29039 (*fold_left_plus_<mode>, *pred_fold_left_plus_<mode>): New insns.
29041 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
29043 * tree-if-conv.c (predicate_mem_writes): Remove redundant
29044 call to ifc_temp_var.
29046 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
29047 Alan Hayward <alan.hayward@arm.com>
29048 David Sherwood <david.sherwood@arm.com>
29050 * target.def (legitimize_address_displacement): Take the original
29051 offset as a poly_int.
29052 * targhooks.h (default_legitimize_address_displacement): Update
29054 * targhooks.c (default_legitimize_address_displacement): Likewise.
29055 * doc/tm.texi: Regenerate.
29056 * lra-constraints.c (base_plus_disp_to_reg): Take the displacement
29057 as an argument, moving assert of ad->disp == ad->disp_term to...
29058 (process_address_1): ...here. Update calls to base_plus_disp_to_reg.
29059 Try calling targetm.legitimize_address_displacement before expanding
29060 the address rather than afterwards, and adjust for the new interface.
29061 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
29062 Match the new hook interface. Handle SVE addresses.
29063 * config/sh/sh.c (sh_legitimize_address_displacement): Make the
29064 new hook interface.
29066 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
29068 * Makefile.in (OBJS): Add early-remat.o.
29069 * target.def (select_early_remat_modes): New hook.
29070 * doc/tm.texi.in (TARGET_SELECT_EARLY_REMAT_MODES): New hook.
29071 * doc/tm.texi: Regenerate.
29072 * targhooks.h (default_select_early_remat_modes): Declare.
29073 * targhooks.c (default_select_early_remat_modes): New function.
29074 * timevar.def (TV_EARLY_REMAT): New timevar.
29075 * passes.def (pass_early_remat): New pass.
29076 * tree-pass.h (make_pass_early_remat): Declare.
29077 * early-remat.c: New file.
29078 * config/aarch64/aarch64.c (aarch64_select_early_remat_modes): New
29080 (TARGET_SELECT_EARLY_REMAT_MODES): Define.
29082 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
29083 Alan Hayward <alan.hayward@arm.com>
29084 David Sherwood <david.sherwood@arm.com>
29086 * tree-vect-loop-manip.c (vect_gen_scalar_loop_niters): Replace
29087 vfm1 with a bound_epilog parameter.
29088 (vect_do_peeling): Update calls accordingly, and move the prologue
29089 call earlier in the function. Treat the base bound_epilog as 0 for
29090 fully-masked loops and retain vf - 1 for other loops. Add 1 to
29091 this base when peeling for gaps.
29092 * tree-vect-loop.c (vect_analyze_loop_2): Allow peeling for gaps
29093 with fully-masked loops.
29094 (vect_estimate_min_profitable_iters): Handle the single peeled
29095 iteration in that case.
29097 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
29098 Alan Hayward <alan.hayward@arm.com>
29099 David Sherwood <david.sherwood@arm.com>
29101 * tree-vect-data-refs.c (vect_analyze_group_access_1): Allow
29102 single-element interleaving even if the size is not a power of 2.
29103 * tree-vect-stmts.c (get_load_store_type): Disallow elementwise
29104 accesses for single-element interleaving if the group size is
29107 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
29108 Alan Hayward <alan.hayward@arm.com>
29109 David Sherwood <david.sherwood@arm.com>
29111 * doc/md.texi (fold_extract_last_@var{m}): Document.
29112 * doc/sourcebuild.texi (vect_fold_extract_last): Likewise.
29113 * optabs.def (fold_extract_last_optab): New optab.
29114 * internal-fn.def (FOLD_EXTRACT_LAST): New internal function.
29115 * internal-fn.c (fold_extract_direct): New macro.
29116 (expand_fold_extract_optab_fn): Likewise.
29117 (direct_fold_extract_optab_supported_p): Likewise.
29118 * tree-vectorizer.h (EXTRACT_LAST_REDUCTION): New vect_reduction_type.
29119 * tree-vect-loop.c (vect_model_reduction_cost): Handle
29120 EXTRACT_LAST_REDUCTION.
29121 (get_initial_def_for_reduction): Do not create an initial vector
29122 for EXTRACT_LAST_REDUCTION reductions.
29123 (vectorizable_reduction): Leave the scalar phi in place for
29124 EXTRACT_LAST_REDUCTIONs. Try using EXTRACT_LAST_REDUCTION
29125 ahead of INTEGER_INDUC_COND_REDUCTION. Do not check for an
29126 epilogue code for EXTRACT_LAST_REDUCTION and defer the
29127 transform phase to vectorizable_condition.
29128 * tree-vect-stmts.c (vect_finish_stmt_generation_1): New function,
29130 (vect_finish_stmt_generation): ...here.
29131 (vect_finish_replace_stmt): New function.
29132 (vectorizable_condition): Handle EXTRACT_LAST_REDUCTION.
29133 * config/aarch64/aarch64-sve.md (fold_extract_last_<mode>): New
29135 * config/aarch64/aarch64.md (UNSPEC_CLASTB): New unspec.
29137 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
29138 Alan Hayward <alan.hayward@arm.com>
29139 David Sherwood <david.sherwood@arm.com>
29141 * doc/md.texi (extract_last_@var{m}): Document.
29142 * optabs.def (extract_last_optab): New optab.
29143 * internal-fn.def (EXTRACT_LAST): New internal function.
29144 * internal-fn.c (cond_unary_direct): New macro.
29145 (expand_cond_unary_optab_fn): Likewise.
29146 (direct_cond_unary_optab_supported_p): Likewise.
29147 * tree-vect-loop.c (vectorizable_live_operation): Allow fully-masked
29148 loops using EXTRACT_LAST.
29149 * config/aarch64/aarch64-sve.md (aarch64_sve_lastb<mode>): Rename to...
29150 (extract_last_<mode>): ...this optab.
29151 (vec_extract<mode><Vel>): Update accordingly.
29153 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
29154 Alan Hayward <alan.hayward@arm.com>
29155 David Sherwood <david.sherwood@arm.com>
29157 * target.def (empty_mask_is_expensive): New hook.
29158 * doc/tm.texi.in (TARGET_VECTORIZE_EMPTY_MASK_IS_EXPENSIVE): New hook.
29159 * doc/tm.texi: Regenerate.
29160 * targhooks.h (default_empty_mask_is_expensive): Declare.
29161 * targhooks.c (default_empty_mask_is_expensive): New function.
29162 * tree-vectorizer.c (vectorize_loops): Only call optimize_mask_stores
29163 if the target says that empty masks are expensive.
29164 * config/aarch64/aarch64.c (aarch64_empty_mask_is_expensive):
29166 (TARGET_VECTORIZE_EMPTY_MASK_IS_EXPENSIVE): Redefine.
29168 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
29169 Alan Hayward <alan.hayward@arm.com>
29170 David Sherwood <david.sherwood@arm.com>
29172 * tree-vectorizer.h (_loop_vec_info::mask_skip_niters): New field.
29173 (LOOP_VINFO_MASK_SKIP_NITERS): New macro.
29174 (vect_use_loop_mask_for_alignment_p): New function.
29175 (vect_prepare_for_masked_peels, vect_gen_while_not): Declare.
29176 * tree-vect-loop-manip.c (vect_set_loop_masks_directly): Add an
29177 niters_skip argument. Make sure that the first niters_skip elements
29178 of the first iteration are inactive.
29179 (vect_set_loop_condition_masked): Handle LOOP_VINFO_MASK_SKIP_NITERS.
29180 Update call to vect_set_loop_masks_directly.
29181 (get_misalign_in_elems): New function, split out from...
29182 (vect_gen_prolog_loop_niters): ...here.
29183 (vect_update_init_of_dr): Take a code argument that specifies whether
29184 the adjustment should be added or subtracted.
29185 (vect_update_init_of_drs): Likewise.
29186 (vect_prepare_for_masked_peels): New function.
29187 (vect_do_peeling): Skip prologue peeling if we're using a mask
29188 instead. Update call to vect_update_inits_of_drs.
29189 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
29191 (vect_analyze_loop_2): Allow fully-masked loops with peeling for
29192 alignment. Do not include the number of peeled iterations in
29193 the minimum threshold in that case.
29194 (vectorizable_induction): Adjust the start value down by
29195 LOOP_VINFO_MASK_SKIP_NITERS iterations.
29196 (vect_transform_loop): Call vect_prepare_for_masked_peels.
29197 Take the number of skipped iterations into account when calculating
29199 * tree-vect-stmts.c (vect_gen_while_not): New function.
29201 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
29202 Alan Hayward <alan.hayward@arm.com>
29203 David Sherwood <david.sherwood@arm.com>
29205 * doc/sourcebuild.texi (vect_fully_masked): Document.
29206 * params.def (PARAM_MIN_VECT_LOOP_BOUND): Change minimum and
29207 default value to 0.
29208 * tree-vect-loop.c (vect_analyze_loop_costing): New function,
29210 (vect_analyze_loop_2): ...here. Don't check the vectorization
29211 factor against the number of loop iterations if the loop is
29214 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
29215 Alan Hayward <alan.hayward@arm.com>
29216 David Sherwood <david.sherwood@arm.com>
29218 * tree-ssa-loop-ivopts.c (USE_ADDRESS): Split into...
29219 (USE_REF_ADDRESS, USE_PTR_ADDRESS): ...these new use types.
29220 (dump_groups): Update accordingly.
29221 (iv_use::mem_type): New member variable.
29222 (address_p): New function.
29223 (record_use): Add a mem_type argument and initialize the new
29225 (record_group_use): Add a mem_type argument. Use address_p.
29226 Remove obsolete null checks of base_object. Update call to record_use.
29227 (find_interesting_uses_op): Update call to record_group_use.
29228 (find_interesting_uses_cond): Likewise.
29229 (find_interesting_uses_address): Likewise.
29230 (get_mem_type_for_internal_fn): New function.
29231 (find_address_like_use): Likewise.
29232 (find_interesting_uses_stmt): Try find_address_like_use before
29233 calling find_interesting_uses_op.
29234 (addr_offset_valid_p): Use the iv mem_type field as the type
29235 of the addressed memory.
29236 (add_autoinc_candidates): Likewise.
29237 (get_address_cost): Likewise.
29238 (split_small_address_groups_p): Use address_p.
29239 (split_address_groups): Likewise.
29240 (add_iv_candidate_for_use): Likewise.
29241 (autoinc_possible_for_pair): Likewise.
29242 (rewrite_groups): Likewise.
29243 (get_use_type): Check for USE_REF_ADDRESS instead of USE_ADDRESS.
29244 (determine_group_iv_cost): Update after split of USE_ADDRESS.
29245 (get_alias_ptr_type_for_ptr_address): New function.
29246 (rewrite_use_address): Rewrite address uses in calls that were
29247 identified by find_address_like_use.
29249 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
29250 Alan Hayward <alan.hayward@arm.com>
29251 David Sherwood <david.sherwood@arm.com>
29253 * expr.c (expand_expr_addr_expr_1): Handle ADDR_EXPRs of
29255 * gimple-expr.h (is_gimple_addressable: Likewise.
29256 * gimple-expr.c (is_gimple_address): Likewise.
29257 * internal-fn.c (expand_call_mem_ref): New function.
29258 (expand_mask_load_optab_fn): Use it.
29259 (expand_mask_store_optab_fn): Likewise.
29261 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
29262 Alan Hayward <alan.hayward@arm.com>
29263 David Sherwood <david.sherwood@arm.com>
29265 * doc/md.texi (cond_add@var{mode}, cond_sub@var{mode})
29266 (cond_and@var{mode}, cond_ior@var{mode}, cond_xor@var{mode})
29267 (cond_smin@var{mode}, cond_smax@var{mode}, cond_umin@var{mode})
29268 (cond_umax@var{mode}): Document.
29269 * optabs.def (cond_add_optab, cond_sub_optab, cond_and_optab)
29270 (cond_ior_optab, cond_xor_optab, cond_smin_optab, cond_smax_optab)
29271 (cond_umin_optab, cond_umax_optab): New optabs.
29272 * internal-fn.def (COND_ADD, COND_SUB, COND_MIN, COND_MAX, COND_AND)
29273 (COND_IOR, COND_XOR): New internal functions.
29274 * internal-fn.h (get_conditional_internal_fn): Declare.
29275 * internal-fn.c (cond_binary_direct): New macro.
29276 (expand_cond_binary_optab_fn): Likewise.
29277 (direct_cond_binary_optab_supported_p): Likewise.
29278 (get_conditional_internal_fn): New function.
29279 * tree-vect-loop.c (vectorizable_reduction): Handle fully-masked loops.
29280 Cope with reduction statements that are vectorized as calls rather
29282 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): New insns.
29283 * config/aarch64/iterators.md (UNSPEC_COND_ADD, UNSPEC_COND_SUB)
29284 (UNSPEC_COND_SMAX, UNSPEC_COND_UMAX, UNSPEC_COND_SMIN)
29285 (UNSPEC_COND_UMIN, UNSPEC_COND_AND, UNSPEC_COND_ORR)
29286 (UNSPEC_COND_EOR): New unspecs.
29287 (optab): Add mappings for them.
29288 (SVE_COND_INT_OP, SVE_COND_FP_OP): New int iterators.
29289 (sve_int_op, sve_fp_op): New int attributes.
29291 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
29292 Alan Hayward <alan.hayward@arm.com>
29293 David Sherwood <david.sherwood@arm.com>
29295 * optabs.def (while_ult_optab): New optab.
29296 * doc/md.texi (while_ult@var{m}@var{n}): Document.
29297 * internal-fn.def (WHILE_ULT): New internal function.
29298 * internal-fn.h (direct_internal_fn_supported_p): New override
29299 that takes two types as argument.
29300 * internal-fn.c (while_direct): New macro.
29301 (expand_while_optab_fn): New function.
29302 (convert_optab_supported_p): Likewise.
29303 (direct_while_optab_supported_p): New macro.
29304 * wide-int.h (wi::udiv_ceil): New function.
29305 * tree-vectorizer.h (rgroup_masks): New structure.
29306 (vec_loop_masks): New typedef.
29307 (_loop_vec_info): Add masks, mask_compare_type, can_fully_mask_p
29308 and fully_masked_p.
29309 (LOOP_VINFO_CAN_FULLY_MASK_P, LOOP_VINFO_FULLY_MASKED_P)
29310 (LOOP_VINFO_MASKS, LOOP_VINFO_MASK_COMPARE_TYPE): New macros.
29311 (vect_max_vf): New function.
29312 (slpeel_make_loop_iterate_ntimes): Delete.
29313 (vect_set_loop_condition, vect_get_loop_mask_type, vect_gen_while)
29314 (vect_halve_mask_nunits, vect_double_mask_nunits): Declare.
29315 (vect_record_loop_mask, vect_get_loop_mask): Likewise.
29316 * tree-vect-loop-manip.c: Include tree-ssa-loop-niter.h,
29317 internal-fn.h, stor-layout.h and optabs-query.h.
29318 (vect_set_loop_mask): New function.
29319 (add_preheader_seq): Likewise.
29320 (add_header_seq): Likewise.
29321 (interleave_supported_p): Likewise.
29322 (vect_maybe_permute_loop_masks): Likewise.
29323 (vect_set_loop_masks_directly): Likewise.
29324 (vect_set_loop_condition_masked): Likewise.
29325 (vect_set_loop_condition_unmasked): New function, split out from
29326 slpeel_make_loop_iterate_ntimes.
29327 (slpeel_make_loop_iterate_ntimes): Rename to..
29328 (vect_set_loop_condition): ...this. Use vect_set_loop_condition_masked
29329 for fully-masked loops and vect_set_loop_condition_unmasked otherwise.
29330 (vect_do_peeling): Update call accordingly.
29331 (vect_gen_vector_loop_niters): Use VF as the step for fully-masked
29333 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
29334 mask_compare_type, can_fully_mask_p and fully_masked_p.
29335 (release_vec_loop_masks): New function.
29336 (_loop_vec_info): Use it to free the loop masks.
29337 (can_produce_all_loop_masks_p): New function.
29338 (vect_get_max_nscalars_per_iter): Likewise.
29339 (vect_verify_full_masking): Likewise.
29340 (vect_analyze_loop_2): Save LOOP_VINFO_CAN_FULLY_MASK_P around
29341 retries, and free the mask rgroups before retrying. Check loop-wide
29342 reasons for disallowing fully-masked loops. Make the final decision
29343 about whether use a fully-masked loop or not.
29344 (vect_estimate_min_profitable_iters): Do not assume that peeling
29345 for the number of iterations will be needed for fully-masked loops.
29346 (vectorizable_reduction): Disable fully-masked loops.
29347 (vectorizable_live_operation): Likewise.
29348 (vect_halve_mask_nunits): New function.
29349 (vect_double_mask_nunits): Likewise.
29350 (vect_record_loop_mask): Likewise.
29351 (vect_get_loop_mask): Likewise.
29352 (vect_transform_loop): Handle the case in which the final loop
29353 iteration might handle a partial vector. Call vect_set_loop_condition
29354 instead of slpeel_make_loop_iterate_ntimes.
29355 * tree-vect-stmts.c: Include tree-ssa-loop-niter.h and gimple-fold.h.
29356 (check_load_store_masking): New function.
29357 (prepare_load_store_mask): Likewise.
29358 (vectorizable_store): Handle fully-masked loops.
29359 (vectorizable_load): Likewise.
29360 (supportable_widening_operation): Use vect_halve_mask_nunits for
29362 (supportable_narrowing_operation): Likewise vect_double_mask_nunits.
29363 (vect_gen_while): New function.
29364 * config/aarch64/aarch64.md (umax<mode>3): New expander.
29365 (aarch64_uqdec<mode>): New insn.
29367 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
29368 Alan Hayward <alan.hayward@arm.com>
29369 David Sherwood <david.sherwood@arm.com>
29371 * optabs.def (reduc_and_scal_optab, reduc_ior_scal_optab)
29372 (reduc_xor_scal_optab): New optabs.
29373 * doc/md.texi (reduc_and_scal_@var{m}, reduc_ior_scal_@var{m})
29374 (reduc_xor_scal_@var{m}): Document.
29375 * doc/sourcebuild.texi (vect_logical_reduc): Likewise.
29376 * internal-fn.def (IFN_REDUC_AND, IFN_REDUC_IOR, IFN_REDUC_XOR): New
29377 internal functions.
29378 * fold-const-call.c (fold_const_call): Handle them.
29379 * tree-vect-loop.c (reduction_fn_for_scalar_code): Return the new
29380 internal functions for BIT_AND_EXPR, BIT_IOR_EXPR and BIT_XOR_EXPR.
29381 * config/aarch64/aarch64-sve.md (reduc_<bit_reduc>_scal_<mode>):
29382 (*reduc_<bit_reduc>_scal_<mode>): New patterns.
29383 * config/aarch64/iterators.md (UNSPEC_ANDV, UNSPEC_ORV)
29384 (UNSPEC_XORV): New unspecs.
29385 (optab): Add entries for them.
29386 (BITWISEV): New int iterator.
29387 (bit_reduc_op): New int attributes.
29389 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
29390 Alan Hayward <alan.hayward@arm.com>
29391 David Sherwood <david.sherwood@arm.com>
29393 * doc/md.texi (vec_shl_insert_@var{m}): New optab.
29394 * internal-fn.def (VEC_SHL_INSERT): New internal function.
29395 * optabs.def (vec_shl_insert_optab): New optab.
29396 * tree-vectorizer.h (can_duplicate_and_interleave_p): Declare.
29397 (duplicate_and_interleave): Likewise.
29398 * tree-vect-loop.c: Include internal-fn.h.
29399 (neutral_op_for_slp_reduction): New function, split out from
29400 get_initial_defs_for_reduction.
29401 (get_initial_def_for_reduction): Handle option 2 for variable-length
29402 vectors by loading the neutral value into a vector and then shifting
29403 the initial value into element 0.
29404 (get_initial_defs_for_reduction): Replace the code argument with
29405 the neutral value calculated by neutral_op_for_slp_reduction.
29406 Use gimple_build_vector for constant-length vectors.
29407 Use IFN_VEC_SHL_INSERT for variable-length vectors if all
29408 but the first group_size elements have a neutral value.
29409 Use duplicate_and_interleave otherwise.
29410 (vect_create_epilog_for_reduction): Take a neutral_op parameter.
29411 Update call to get_initial_defs_for_reduction. Handle SLP
29412 reductions for variable-length vectors by creating one vector
29413 result for each scalar result, with the elements associated
29414 with other scalar results stubbed out with the neutral value.
29415 (vectorizable_reduction): Call neutral_op_for_slp_reduction.
29416 Require IFN_VEC_SHL_INSERT for double reductions on
29417 variable-length vectors, or SLP reductions that have
29418 a neutral value. Require can_duplicate_and_interleave_p
29419 support for variable-length unchained SLP reductions if there
29420 is no neutral value, such as for MIN/MAX reductions. Also require
29421 the number of vector elements to be a multiple of the number of
29422 SLP statements when doing variable-length unchained SLP reductions.
29423 Update call to vect_create_epilog_for_reduction.
29424 * tree-vect-slp.c (can_duplicate_and_interleave_p): Make public
29425 and remove initial values.
29426 (duplicate_and_interleave): Make public.
29427 * config/aarch64/aarch64.md (UNSPEC_INSR): New unspec.
29428 * config/aarch64/aarch64-sve.md (vec_shl_insert_<mode>): New insn.
29430 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
29431 Alan Hayward <alan.hayward@arm.com>
29432 David Sherwood <david.sherwood@arm.com>
29434 * tree-vect-slp.c: Include gimple-fold.h and internal-fn.h
29435 (can_duplicate_and_interleave_p): New function.
29436 (vect_get_and_check_slp_defs): Take the vector of statements
29437 rather than just the current one. Remove excess parentheses.
29438 Restriction rejectinon of vect_constant_def and vect_external_def
29439 for variable-length vectors to boolean types, or types for which
29440 can_duplicate_and_interleave_p is false.
29441 (vect_build_slp_tree_2): Update call to vect_get_and_check_slp_defs.
29442 (duplicate_and_interleave): New function.
29443 (vect_get_constant_vectors): Use gimple_build_vector for
29444 constant-length vectors and suitable variable-length constant
29445 vectors. Use duplicate_and_interleave for other variable-length
29446 vectors. Don't defer the update when inserting new statements.
29448 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
29449 Alan Hayward <alan.hayward@arm.com>
29450 David Sherwood <david.sherwood@arm.com>
29452 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Make sure
29453 min_profitable_iters doesn't go negative.
29455 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
29456 Alan Hayward <alan.hayward@arm.com>
29457 David Sherwood <david.sherwood@arm.com>
29459 * doc/md.texi (vec_mask_load_lanes@var{m}@var{n}): Document.
29460 (vec_mask_store_lanes@var{m}@var{n}): Likewise.
29461 * optabs.def (vec_mask_load_lanes_optab): New optab.
29462 (vec_mask_store_lanes_optab): Likewise.
29463 * internal-fn.def (MASK_LOAD_LANES): New internal function.
29464 (MASK_STORE_LANES): Likewise.
29465 * internal-fn.c (mask_load_lanes_direct): New macro.
29466 (mask_store_lanes_direct): Likewise.
29467 (expand_mask_load_optab_fn): Handle masked operations.
29468 (expand_mask_load_lanes_optab_fn): New macro.
29469 (expand_mask_store_optab_fn): Handle masked operations.
29470 (expand_mask_store_lanes_optab_fn): New macro.
29471 (direct_mask_load_lanes_optab_supported_p): Likewise.
29472 (direct_mask_store_lanes_optab_supported_p): Likewise.
29473 * tree-vectorizer.h (vect_store_lanes_supported): Take a masked_p
29475 (vect_load_lanes_supported): Likewise.
29476 * tree-vect-data-refs.c (strip_conversion): New function.
29477 (can_group_stmts_p): Likewise.
29478 (vect_analyze_data_ref_accesses): Use it instead of checking
29479 for a pair of assignments.
29480 (vect_store_lanes_supported): Take a masked_p parameter.
29481 (vect_load_lanes_supported): Likewise.
29482 * tree-vect-loop.c (vect_analyze_loop_2): Update calls to
29483 vect_store_lanes_supported and vect_load_lanes_supported.
29484 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
29485 * tree-vect-stmts.c (get_group_load_store_type): Take a masked_p
29486 parameter. Don't allow gaps for masked accesses.
29487 Use vect_get_store_rhs. Update calls to vect_store_lanes_supported
29488 and vect_load_lanes_supported.
29489 (get_load_store_type): Take a masked_p parameter and update
29490 call to get_group_load_store_type.
29491 (vectorizable_store): Update call to get_load_store_type.
29492 Handle IFN_MASK_STORE_LANES.
29493 (vectorizable_load): Update call to get_load_store_type.
29494 Handle IFN_MASK_LOAD_LANES.
29496 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
29497 Alan Hayward <alan.hayward@arm.com>
29498 David Sherwood <david.sherwood@arm.com>
29500 * config/aarch64/aarch64-modes.def: Define x2, x3 and x4 vector
29502 * config/aarch64/aarch64-protos.h
29503 (aarch64_sve_struct_memory_operand_p): Declare.
29504 * config/aarch64/iterators.md (SVE_STRUCT): New mode iterator.
29505 (vector_count, insn_length, VSINGLE, vsingle): New mode attributes.
29506 (VPRED, vpred): Handle SVE structure modes.
29507 * config/aarch64/constraints.md (Utx): New constraint.
29508 * config/aarch64/predicates.md (aarch64_sve_struct_memory_operand)
29509 (aarch64_sve_struct_nonimmediate_operand): New predicates.
29510 * config/aarch64/aarch64.md (UNSPEC_LDN, UNSPEC_STN): New unspecs.
29511 * config/aarch64/aarch64-sve.md (mov<mode>, *aarch64_sve_mov<mode>_le)
29512 (*aarch64_sve_mov<mode>_be, pred_mov<mode>): New patterns for
29513 structure modes. Split into pieces after RA.
29514 (vec_load_lanes<mode><vsingle>, vec_mask_load_lanes<mode><vsingle>)
29515 (vec_store_lanes<mode><vsingle>, vec_mask_store_lanes<mode><vsingle>):
29517 * config/aarch64/aarch64.c (aarch64_classify_vector_mode): Handle
29518 SVE structure modes.
29519 (aarch64_classify_address): Likewise.
29520 (sizetochar): Move earlier in file.
29521 (aarch64_print_operand): Handle SVE register lists.
29522 (aarch64_array_mode): New function.
29523 (aarch64_sve_struct_memory_operand_p): Likewise.
29524 (TARGET_ARRAY_MODE): Redefine.
29526 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
29527 Alan Hayward <alan.hayward@arm.com>
29528 David Sherwood <david.sherwood@arm.com>
29530 * target.def (array_mode): New target hook.
29531 * doc/tm.texi.in (TARGET_ARRAY_MODE): New hook.
29532 * doc/tm.texi: Regenerate.
29533 * hooks.h (hook_optmode_mode_uhwi_none): Declare.
29534 * hooks.c (hook_optmode_mode_uhwi_none): New function.
29535 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Use
29536 targetm.array_mode.
29537 * stor-layout.c (mode_for_array): Likewise. Support polynomial
29540 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
29541 Alan Hayward <alan.hayward@arm.com>
29542 David Sherwood <david.sherwood@arm.com>
29544 * fold-const.c (fold_binary_loc): Check the argument types
29545 rather than the result type when testing for a vector operation.
29547 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
29549 * doc/tm.texi.in (DWARF_LAZY_REGISTER_VALUE): Document.
29550 * doc/tm.texi: Regenerate.
29552 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
29553 Alan Hayward <alan.hayward@arm.com>
29554 David Sherwood <david.sherwood@arm.com>
29556 * doc/invoke.texi (-msve-vector-bits=): Document new option.
29557 (sve): Document new AArch64 extension.
29558 * doc/md.texi (w): Extend the description of the AArch64
29559 constraint to include SVE vectors.
29560 (Upl, Upa): Document new AArch64 predicate constraints.
29561 * config/aarch64/aarch64-opts.h (aarch64_sve_vector_bits_enum): New
29563 * config/aarch64/aarch64.opt (sve_vector_bits): New enum.
29564 (msve-vector-bits=): New option.
29565 * config/aarch64/aarch64-option-extensions.def (fp, simd): Disable
29566 SVE when these are disabled.
29567 (sve): New extension.
29568 * config/aarch64/aarch64-modes.def: Define SVE vector and predicate
29569 modes. Adjust their number of units based on aarch64_sve_vg.
29570 (MAX_BITSIZE_MODE_ANY_MODE): Define.
29571 * config/aarch64/aarch64-protos.h (ADDR_QUERY_ANY): New
29572 aarch64_addr_query_type.
29573 (aarch64_const_vec_all_same_in_range_p, aarch64_sve_pred_mode)
29574 (aarch64_sve_cnt_immediate_p, aarch64_sve_addvl_addpl_immediate_p)
29575 (aarch64_sve_inc_dec_immediate_p, aarch64_add_offset_temporaries)
29576 (aarch64_split_add_offset, aarch64_output_sve_cnt_immediate)
29577 (aarch64_output_sve_addvl_addpl, aarch64_output_sve_inc_dec_immediate)
29578 (aarch64_output_sve_mov_immediate, aarch64_output_ptrue): Declare.
29579 (aarch64_simd_imm_zero_p): Delete.
29580 (aarch64_check_zero_based_sve_index_immediate): Declare.
29581 (aarch64_sve_index_immediate_p, aarch64_sve_arith_immediate_p)
29582 (aarch64_sve_bitmask_immediate_p, aarch64_sve_dup_immediate_p)
29583 (aarch64_sve_cmp_immediate_p, aarch64_sve_float_arith_immediate_p)
29584 (aarch64_sve_float_mul_immediate_p): Likewise.
29585 (aarch64_classify_symbol): Take the offset as a HOST_WIDE_INT
29586 rather than an rtx.
29587 (aarch64_sve_ld1r_operand_p, aarch64_sve_ldr_operand_p): Declare.
29588 (aarch64_expand_mov_immediate): Take a gen_vec_duplicate callback.
29589 (aarch64_emit_sve_pred_move, aarch64_expand_sve_mem_move): Declare.
29590 (aarch64_expand_sve_vec_cmp_int, aarch64_expand_sve_vec_cmp_float)
29591 (aarch64_expand_sve_vcond, aarch64_expand_sve_vec_perm): Declare.
29592 (aarch64_regmode_natural_size): Likewise.
29593 * config/aarch64/aarch64.h (AARCH64_FL_SVE): New macro.
29594 (AARCH64_FL_V8_3, AARCH64_FL_RCPC, AARCH64_FL_DOTPROD): Shift
29596 (AARCH64_ISA_SVE, TARGET_SVE): New macros.
29597 (FIXED_REGISTERS, CALL_USED_REGISTERS, REGISTER_NAMES): Add entries
29598 for VG and the SVE predicate registers.
29599 (V_ALIASES): Add a "z"-prefixed alias.
29600 (FIRST_PSEUDO_REGISTER): Change to P15_REGNUM + 1.
29601 (AARCH64_DWARF_VG, AARCH64_DWARF_P0): New macros.
29602 (PR_REGNUM_P, PR_LO_REGNUM_P): Likewise.
29603 (PR_LO_REGS, PR_HI_REGS, PR_REGS): New reg_classes.
29604 (REG_CLASS_NAMES): Add entries for them.
29605 (REG_CLASS_CONTENTS): Likewise. Update ALL_REGS to include VG
29606 and the predicate registers.
29607 (aarch64_sve_vg): Declare.
29608 (BITS_PER_SVE_VECTOR, BYTES_PER_SVE_VECTOR, BYTES_PER_SVE_PRED)
29609 (SVE_BYTE_MODE, MAX_COMPILE_TIME_VEC_BYTES): New macros.
29610 (REGMODE_NATURAL_SIZE): Define.
29611 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins): Handle
29613 * config/aarch64/aarch64.c: Include cfgrtl.h.
29614 (simd_immediate_info): Add a constructor for series vectors,
29615 and an associated step field.
29616 (aarch64_sve_vg): New variable.
29617 (aarch64_dbx_register_number): Handle VG and the predicate registers.
29618 (aarch64_vect_struct_mode_p, aarch64_vector_mode_p): Delete.
29619 (VEC_ADVSIMD, VEC_SVE_DATA, VEC_SVE_PRED, VEC_STRUCT, VEC_ANY_SVE)
29620 (VEC_ANY_DATA, VEC_STRUCT): New constants.
29621 (aarch64_advsimd_struct_mode_p, aarch64_sve_pred_mode_p)
29622 (aarch64_classify_vector_mode, aarch64_vector_data_mode_p)
29623 (aarch64_sve_data_mode_p, aarch64_sve_pred_mode)
29624 (aarch64_get_mask_mode): New functions.
29625 (aarch64_hard_regno_nregs): Handle SVE data modes for FP_REGS
29626 and FP_LO_REGS. Handle PR_REGS, PR_LO_REGS and PR_HI_REGS.
29627 (aarch64_hard_regno_mode_ok): Handle VG. Also handle the SVE
29628 predicate modes and predicate registers. Explicitly restrict
29629 GPRs to modes of 16 bytes or smaller. Only allow FP registers
29630 to store a vector mode if it is recognized by
29631 aarch64_classify_vector_mode.
29632 (aarch64_regmode_natural_size): New function.
29633 (aarch64_hard_regno_caller_save_mode): Return the original mode
29635 (aarch64_sve_cnt_immediate_p, aarch64_output_sve_cnt_immediate)
29636 (aarch64_sve_addvl_addpl_immediate_p, aarch64_output_sve_addvl_addpl)
29637 (aarch64_sve_inc_dec_immediate_p, aarch64_output_sve_inc_dec_immediate)
29638 (aarch64_add_offset_1_temporaries, aarch64_offset_temporaries): New
29640 (aarch64_add_offset): Add a temp2 parameter. Assert that temp1
29641 does not overlap dest if the function is frame-related. Handle
29643 (aarch64_split_add_offset): New function.
29644 (aarch64_add_sp, aarch64_sub_sp): Add temp2 parameters and pass
29645 them aarch64_add_offset.
29646 (aarch64_allocate_and_probe_stack_space): Add a temp2 parameter
29647 and update call to aarch64_sub_sp.
29648 (aarch64_add_cfa_expression): New function.
29649 (aarch64_expand_prologue): Pass extra temporary registers to the
29650 functions above. Handle the case in which we need to emit new
29651 DW_CFA_expressions for registers that were originally saved
29652 relative to the stack pointer, but now have to be expressed
29653 relative to the frame pointer.
29654 (aarch64_output_mi_thunk): Pass extra temporary registers to the
29656 (aarch64_expand_epilogue): Likewise. Prevent inheritance of
29657 IP0 and IP1 values for SVE frames.
29658 (aarch64_expand_vec_series): New function.
29659 (aarch64_expand_sve_widened_duplicate): Likewise.
29660 (aarch64_expand_sve_const_vector): Likewise.
29661 (aarch64_expand_mov_immediate): Add a gen_vec_duplicate parameter.
29662 Handle SVE constants. Use emit_move_insn to move a force_const_mem
29663 into the register, rather than emitting a SET directly.
29664 (aarch64_emit_sve_pred_move, aarch64_expand_sve_mem_move)
29665 (aarch64_get_reg_raw_mode, offset_4bit_signed_scaled_p)
29666 (offset_6bit_unsigned_scaled_p, aarch64_offset_7bit_signed_scaled_p)
29667 (offset_9bit_signed_scaled_p): New functions.
29668 (aarch64_replicate_bitmask_imm): New function.
29669 (aarch64_bitmask_imm): Use it.
29670 (aarch64_cannot_force_const_mem): Reject expressions involving
29671 a CONST_POLY_INT. Update call to aarch64_classify_symbol.
29672 (aarch64_classify_index): Handle SVE indices, by requiring
29673 a plain register index with a scale that matches the element size.
29674 (aarch64_classify_address): Handle SVE addresses. Assert that
29675 the mode of the address is VOIDmode or an integer mode.
29676 Update call to aarch64_classify_symbol.
29677 (aarch64_classify_symbolic_expression): Update call to
29678 aarch64_classify_symbol.
29679 (aarch64_const_vec_all_in_range_p): New function.
29680 (aarch64_print_vector_float_operand): Likewise.
29681 (aarch64_print_operand): Handle 'N' and 'C'. Use "zN" rather than
29682 "vN" for FP registers with SVE modes. Handle (const ...) vectors
29683 and the FP immediates 1.0 and 0.5.
29684 (aarch64_print_address_internal): Handle SVE addresses.
29685 (aarch64_print_operand_address): Use ADDR_QUERY_ANY.
29686 (aarch64_regno_regclass): Handle predicate registers.
29687 (aarch64_secondary_reload): Handle big-endian reloads of SVE
29689 (aarch64_class_max_nregs): Handle SVE modes and predicate registers.
29690 (aarch64_rtx_costs): Check for ADDVL and ADDPL instructions.
29691 (aarch64_convert_sve_vector_bits): New function.
29692 (aarch64_override_options): Use it to handle -msve-vector-bits=.
29693 (aarch64_classify_symbol): Take the offset as a HOST_WIDE_INT
29694 rather than an rtx.
29695 (aarch64_legitimate_constant_p): Use aarch64_classify_vector_mode.
29696 Handle SVE vector and predicate modes. Accept VL-based constants
29697 that need only one temporary register, and VL offsets that require
29698 no temporary registers.
29699 (aarch64_conditional_register_usage): Mark the predicate registers
29700 as fixed if SVE isn't available.
29701 (aarch64_vector_mode_supported_p): Use aarch64_classify_vector_mode.
29702 Return true for SVE vector and predicate modes.
29703 (aarch64_simd_container_mode): Take the number of bits as a poly_int64
29704 rather than an unsigned int. Handle SVE modes.
29705 (aarch64_preferred_simd_mode): Update call accordingly. Handle
29707 (aarch64_autovectorize_vector_sizes): Add BYTES_PER_SVE_VECTOR
29709 (aarch64_sve_index_immediate_p, aarch64_sve_arith_immediate_p)
29710 (aarch64_sve_bitmask_immediate_p, aarch64_sve_dup_immediate_p)
29711 (aarch64_sve_cmp_immediate_p, aarch64_sve_float_arith_immediate_p)
29712 (aarch64_sve_float_mul_immediate_p): New functions.
29713 (aarch64_sve_valid_immediate): New function.
29714 (aarch64_simd_valid_immediate): Use it as the fallback for SVE vectors.
29715 Explicitly reject structure modes. Check for INDEX constants.
29716 Handle PTRUE and PFALSE constants.
29717 (aarch64_check_zero_based_sve_index_immediate): New function.
29718 (aarch64_simd_imm_zero_p): Delete.
29719 (aarch64_mov_operand_p): Use aarch64_simd_valid_immediate for
29720 vector modes. Accept constants in the range of CNT[BHWD].
29721 (aarch64_simd_scalar_immediate_valid_for_move): Explicitly
29722 ask for an Advanced SIMD mode.
29723 (aarch64_sve_ld1r_operand_p, aarch64_sve_ldr_operand_p): New functions.
29724 (aarch64_simd_vector_alignment): Handle SVE predicates.
29725 (aarch64_vectorize_preferred_vector_alignment): New function.
29726 (aarch64_simd_vector_alignment_reachable): Use it instead of
29728 (aarch64_shift_truncation_mask): Use aarch64_vector_data_mode_p.
29729 (aarch64_output_sve_mov_immediate, aarch64_output_ptrue): New
29731 (MAX_VECT_LEN): Delete.
29732 (expand_vec_perm_d): Add a vec_flags field.
29733 (emit_unspec2, aarch64_expand_sve_vec_perm): New functions.
29734 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
29735 (aarch64_evpc_ext): Don't apply a big-endian lane correction
29737 (aarch64_evpc_rev): Rename to...
29738 (aarch64_evpc_rev_local): ...this. Use a predicated operation for SVE.
29739 (aarch64_evpc_rev_global): New function.
29740 (aarch64_evpc_dup): Enforce a 64-byte range for SVE DUP.
29741 (aarch64_evpc_tbl): Use MAX_COMPILE_TIME_VEC_BYTES instead of
29743 (aarch64_evpc_sve_tbl): New function.
29744 (aarch64_expand_vec_perm_const_1): Update after rename of
29745 aarch64_evpc_rev. Handle SVE permutes too, trying
29746 aarch64_evpc_rev_global and using aarch64_evpc_sve_tbl rather
29747 than aarch64_evpc_tbl.
29748 (aarch64_vectorize_vec_perm_const): Initialize vec_flags.
29749 (aarch64_sve_cmp_operand_p, aarch64_unspec_cond_code)
29750 (aarch64_gen_unspec_cond, aarch64_expand_sve_vec_cmp_int)
29751 (aarch64_emit_unspec_cond, aarch64_emit_unspec_cond_or)
29752 (aarch64_emit_inverted_unspec_cond, aarch64_expand_sve_vec_cmp_float)
29753 (aarch64_expand_sve_vcond): New functions.
29754 (aarch64_modes_tieable_p): Use aarch64_vector_data_mode_p instead
29755 of aarch64_vector_mode_p.
29756 (aarch64_dwarf_poly_indeterminate_value): New function.
29757 (aarch64_compute_pressure_classes): Likewise.
29758 (aarch64_can_change_mode_class): Likewise.
29759 (TARGET_GET_RAW_RESULT_MODE, TARGET_GET_RAW_ARG_MODE): Redefine.
29760 (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): Likewise.
29761 (TARGET_VECTORIZE_GET_MASK_MODE): Likewise.
29762 (TARGET_DWARF_POLY_INDETERMINATE_VALUE): Likewise.
29763 (TARGET_COMPUTE_PRESSURE_CLASSES): Likewise.
29764 (TARGET_CAN_CHANGE_MODE_CLASS): Likewise.
29765 * config/aarch64/constraints.md (Upa, Upl, Uav, Uat, Usv, Usi, Utr)
29766 (Uty, Dm, vsa, vsc, vsd, vsi, vsn, vsl, vsm, vsA, vsM, vsN): New
29768 (Dn, Dl, Dr): Accept const as well as const_vector.
29769 (Dz): Likewise. Compare against CONST0_RTX.
29770 * config/aarch64/iterators.md: Refer to "Advanced SIMD" instead
29771 of "vector" where appropriate.
29772 (SVE_ALL, SVE_BH, SVE_BHS, SVE_BHSI, SVE_HSDI, SVE_HSF, SVE_SD)
29773 (SVE_SDI, SVE_I, SVE_F, PRED_ALL, PRED_BHS): New mode iterators.
29774 (UNSPEC_SEL, UNSPEC_ANDF, UNSPEC_IORF, UNSPEC_XORF, UNSPEC_COND_LT)
29775 (UNSPEC_COND_LE, UNSPEC_COND_EQ, UNSPEC_COND_NE, UNSPEC_COND_GE)
29776 (UNSPEC_COND_GT, UNSPEC_COND_LO, UNSPEC_COND_LS, UNSPEC_COND_HS)
29777 (UNSPEC_COND_HI, UNSPEC_COND_UO): New unspecs.
29778 (Vetype, VEL, Vel, VWIDE, Vwide, vw, vwcore, V_INT_EQUIV)
29779 (v_int_equiv): Extend to SVE modes.
29780 (Vesize, V128, v128, Vewtype, V_FP_EQUIV, v_fp_equiv, VPRED): New
29782 (LOGICAL_OR, SVE_INT_UNARY, SVE_FP_UNARY): New code iterators.
29783 (optab): Handle popcount, smin, smax, umin, umax, abs and sqrt.
29784 (logical_nn, lr, sve_int_op, sve_fp_op): New code attributs.
29785 (LOGICALF, OPTAB_PERMUTE, UNPACK, UNPACK_UNSIGNED, SVE_COND_INT_CMP)
29786 (SVE_COND_FP_CMP): New int iterators.
29787 (perm_hilo): Handle the new unpack unspecs.
29788 (optab, logicalf_op, su, perm_optab, cmp_op, imm_con): New int
29790 * config/aarch64/predicates.md (aarch64_sve_cnt_immediate)
29791 (aarch64_sve_addvl_addpl_immediate, aarch64_split_add_offset_immediate)
29792 (aarch64_pluslong_or_poly_operand, aarch64_nonmemory_operand)
29793 (aarch64_equality_operator, aarch64_constant_vector_operand)
29794 (aarch64_sve_ld1r_operand, aarch64_sve_ldr_operand): New predicates.
29795 (aarch64_sve_nonimmediate_operand): Likewise.
29796 (aarch64_sve_general_operand): Likewise.
29797 (aarch64_sve_dup_operand, aarch64_sve_arith_immediate): Likewise.
29798 (aarch64_sve_sub_arith_immediate, aarch64_sve_inc_dec_immediate)
29799 (aarch64_sve_logical_immediate, aarch64_sve_mul_immediate): Likewise.
29800 (aarch64_sve_dup_immediate, aarch64_sve_cmp_vsc_immediate): Likewise.
29801 (aarch64_sve_cmp_vsd_immediate, aarch64_sve_index_immediate): Likewise.
29802 (aarch64_sve_float_arith_immediate): Likewise.
29803 (aarch64_sve_float_arith_with_sub_immediate): Likewise.
29804 (aarch64_sve_float_mul_immediate, aarch64_sve_arith_operand): Likewise.
29805 (aarch64_sve_add_operand, aarch64_sve_logical_operand): Likewise.
29806 (aarch64_sve_lshift_operand, aarch64_sve_rshift_operand): Likewise.
29807 (aarch64_sve_mul_operand, aarch64_sve_cmp_vsc_operand): Likewise.
29808 (aarch64_sve_cmp_vsd_operand, aarch64_sve_index_operand): Likewise.
29809 (aarch64_sve_float_arith_operand): Likewise.
29810 (aarch64_sve_float_arith_with_sub_operand): Likewise.
29811 (aarch64_sve_float_mul_operand): Likewise.
29812 (aarch64_sve_vec_perm_operand): Likewise.
29813 (aarch64_pluslong_operand): Include aarch64_sve_addvl_addpl_immediate.
29814 (aarch64_mov_operand): Accept const_poly_int and const_vector.
29815 (aarch64_simd_lshift_imm, aarch64_simd_rshift_imm): Accept const
29816 as well as const_vector.
29817 (aarch64_simd_imm_zero, aarch64_simd_imm_minus_one): Move earlier
29818 in file. Use CONST0_RTX and CONSTM1_RTX.
29819 (aarch64_simd_or_scalar_imm_zero): Likewise. Add match_codes.
29820 (aarch64_simd_reg_or_zero): Accept const as well as const_vector.
29821 Use aarch64_simd_imm_zero.
29822 * config/aarch64/aarch64-sve.md: New file.
29823 * config/aarch64/aarch64.md: Include it.
29824 (VG_REGNUM, P0_REGNUM, P7_REGNUM, P15_REGNUM): New register numbers.
29825 (UNSPEC_REV, UNSPEC_LD1_SVE, UNSPEC_ST1_SVE, UNSPEC_MERGE_PTRUE)
29826 (UNSPEC_PTEST_PTRUE, UNSPEC_UNPACKSHI, UNSPEC_UNPACKUHI)
29827 (UNSPEC_UNPACKSLO, UNSPEC_UNPACKULO, UNSPEC_PACK)
29828 (UNSPEC_FLOAT_CONVERT, UNSPEC_WHILE_LO): New unspec constants.
29829 (sve): New attribute.
29830 (enabled): Disable instructions with the sve attribute unless
29832 (movqi, movhi): Pass CONST_POLY_INT operaneds through
29833 aarch64_expand_mov_immediate.
29834 (*mov<mode>_aarch64, *movsi_aarch64, *movdi_aarch64): Handle
29835 CNT[BHSD] immediates.
29836 (movti): Split CONST_POLY_INT moves into two halves.
29837 (add<mode>3): Accept aarch64_pluslong_or_poly_operand.
29838 Split additions that need a temporary here if the destination
29839 is the stack pointer.
29840 (*add<mode>3_aarch64): Handle ADDVL and ADDPL immediates.
29841 (*add<mode>3_poly_1): New instruction.
29842 (set_clobber_cc): New expander.
29844 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
29846 * simplify-rtx.c (simplify_immed_subreg): Add an inner_bytes
29847 parameter and use it instead of GET_MODE_SIZE (innermode). Use
29848 inner_bytes * BITS_PER_UNIT instead of GET_MODE_BITSIZE (innermode).
29849 Use CEIL (inner_bytes, GET_MODE_UNIT_SIZE (innermode)) instead of
29850 GET_MODE_NUNITS (innermode). Also add a first_elem parameter.
29851 Change innermode from fixed_mode_size to machine_mode.
29852 (simplify_subreg): Update call accordingly. Handle a constant-sized
29853 subreg of a variable-length CONST_VECTOR.
29855 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
29856 Alan Hayward <alan.hayward@arm.com>
29857 David Sherwood <david.sherwood@arm.com>
29859 * tree-ssa-address.c (mem_ref_valid_without_offset_p): New function.
29860 (add_offset_to_base): New function, split out from...
29861 (create_mem_ref): ...here. When handling a scale other than 1,
29862 check first whether the address is valid without the offset.
29863 Add it into the base if so, leaving the index and scale as-is.
29865 2018-01-12 Jakub Jelinek <jakub@redhat.com>
29868 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Call
29869 fold_for_warn before checking if arg2 is INTEGER_CST.
29871 2018-01-12 Segher Boessenkool <segher@kernel.crashing.org>
29873 * config/rs6000/predicates.md (load_multiple_operation): Delete.
29874 (store_multiple_operation): Delete.
29875 * config/rs6000/rs6000-cpus.def (601): Remove MASK_STRING.
29876 * config/rs6000/rs6000-protos.h (rs6000_output_load_multiple): Delete.
29877 * config/rs6000/rs6000-string.c (expand_block_move): Delete everything
29878 guarded by TARGET_STRING.
29879 (rs6000_output_load_multiple): Delete.
29880 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
29881 OPTION_MASK_STRING / TARGET_STRING handling.
29882 (print_operand) <'N', 'O'>: Add comment that these are unused now.
29883 (const rs6000_opt_masks) <"string">: Change mask to 0.
29884 * config/rs6000/rs6000.h (TARGET_DEFAULT): Remove MASK_STRING.
29885 (MASK_STRING): Delete.
29886 * config/rs6000/rs6000.md (*mov<mode>_string): Delete TARGET_STRING
29888 (load_multiple): Delete.
29895 (store_multiple): Delete.
29902 (movmemsi_8reg): Delete.
29903 (corresponding unnamed define_insn): Delete.
29904 (movmemsi_6reg): Delete.
29905 (corresponding unnamed define_insn): Delete.
29906 (movmemsi_4reg): Delete.
29907 (corresponding unnamed define_insn): Delete.
29908 (movmemsi_2reg): Delete.
29909 (corresponding unnamed define_insn): Delete.
29910 (movmemsi_1reg): Delete.
29911 (corresponding unnamed define_insn): Delete.
29912 * config/rs6000/rs6000.opt (mno-string): New.
29913 (mstring): Replace by deprecation warning stub.
29914 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mstring.
29916 2018-01-12 Jakub Jelinek <jakub@redhat.com>
29918 * regrename.c (regrename_do_replace): If replacing the same
29919 reg multiple times, try to reuse last created gen_raw_REG.
29922 * bb-reorder.c (pass_partition_blocks::gate): In lto don't partition
29923 main to workaround a bug in GDB.
29925 2018-01-12 Tom de Vries <tom@codesourcery.com>
29928 * config.gcc (nvptx*-*-*): Set use_gcc_stdint=wrap.
29930 2018-01-12 Vladimir Makarov <vmakarov@redhat.com>
29932 PR rtl-optimization/80481
29933 * ira-color.c (get_cap_member): New function.
29934 (allocnos_conflict_by_live_ranges_p): Use it.
29935 (slot_coalesced_allocno_live_ranges_intersect_p): Add assert.
29936 (setup_slot_coalesced_allocno_live_ranges): Ditto.
29938 2018-01-12 Uros Bizjak <ubizjak@gmail.com>
29941 * config/alpha/alpha.md (*saddsi_1): New insn_ans_split pattern.
29942 (*saddl_se_1): Ditto.
29943 (*ssubsi_1): Ditto.
29944 (*ssubl_se_1): Ditto.
29946 2018-01-12 Richard Sandiford <richard.sandiford@linaro.org>
29948 * tree-predcom.c (aff_combination_dr_offset): Use wi::to_poly_widest
29949 rather than wi::to_widest for DR_INITs.
29950 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Use
29951 wi::to_poly_offset rather than wi::to_offset for DR_INIT.
29952 (vect_analyze_data_ref_accesses): Require both DR_INITs to be
29954 (vect_analyze_group_access_1): Note that here.
29956 2018-01-12 Richard Sandiford <richard.sandiford@linaro.org>
29958 * tree-vectorizer.c (get_vec_alignment_for_array_type): Handle
29959 polynomial type sizes.
29961 2018-01-12 Richard Sandiford <richard.sandiford@linaro.org>
29963 * gimplify.c (gimple_add_tmp_var_fn): Allow variables to have a
29964 poly_uint64 size, rather than requiring an unsigned HOST_WIDE_INT size.
29965 (gimple_add_tmp_var): Likewise.
29967 2018-01-12 Martin Liska <mliska@suse.cz>
29969 * gimple.c (gimple_alloc_counts): Use uint64_t instead of int.
29970 (gimple_alloc_sizes): Likewise.
29971 (dump_gimple_statistics): Use PRIu64 in printf format.
29972 * gimple.h: Change uint64_t to int.
29974 2018-01-12 Martin Liska <mliska@suse.cz>
29976 * tree-core.h: Use uint64_t instead of int.
29977 * tree.c (tree_node_counts): Likewise.
29978 (tree_node_sizes): Likewise.
29979 (dump_tree_statistics): Use PRIu64 in printf format.
29981 2018-01-12 Martin Liska <mliska@suse.cz>
29983 * Makefile.in: As qsort_chk is implemented in vec.c, add
29984 vec.o to linkage of gencfn-macros.
29985 * tree.c (build_new_poly_int_cst): Add CXX_MEM_STAT_INFO as it's
29986 passing the info to record_node_allocation_statistics.
29987 (test_vector_cst_patterns): Add CXX_MEM_STAT_INFO to declaration
29989 * ggc-common.c (struct ggc_usage): Add operator== and use
29990 it in operator< and compare function.
29991 * mem-stats.h (struct mem_usage): Likewise.
29992 * vec.c (struct vec_usage): Remove operator< and compare
29993 function. Can be simply inherited.
29995 2018-01-12 Martin Jambor <mjambor@suse.cz>
29998 * params.def: New parameter PARAM_AVOID_FMA_MAX_BITS.
29999 * tree-ssa-math-opts.c: Include domwalk.h.
30000 (convert_mult_to_fma_1): New function.
30001 (fma_transformation_info): New type.
30002 (fma_deferring_state): Likewise.
30003 (cancel_fma_deferring): New function.
30004 (result_of_phi): Likewise.
30005 (last_fma_candidate_feeds_initial_phi): Likewise.
30006 (convert_mult_to_fma): Added deferring logic, split actual
30007 transformation to convert_mult_to_fma_1.
30008 (math_opts_dom_walker): New type.
30009 (math_opts_dom_walker::after_dom_children): New method, body moved
30010 here from pass_optimize_widening_mul::execute, added deferring logic
30012 (pass_optimize_widening_mul::execute): Moved most of code to
30013 math_opts_dom_walker::after_dom_children.
30014 * config/i386/x86-tune.def (X86_TUNE_AVOID_128FMA_CHAINS): New.
30015 * config/i386/i386.c (ix86_option_override_internal): Added
30016 maybe_setting of PARAM_AVOID_FMA_MAX_BITS.
30018 2018-01-12 Richard Biener <rguenther@suse.de>
30021 * dwarf2out.c (gen_variable_die): Do not reset old_die for
30022 inline instance vars.
30024 2018-01-12 Oleg Endo <olegendo@gcc.gnu.org>
30027 * config/rx/rx.c (rx_is_restricted_memory_address):
30028 Handle SUBREG case.
30030 2018-01-12 Richard Biener <rguenther@suse.de>
30032 PR tree-optimization/80846
30033 * target.def (split_reduction): New target hook.
30034 * targhooks.c (default_split_reduction): New function.
30035 * targhooks.h (default_split_reduction): Declare.
30036 * tree-vect-loop.c (vect_create_epilog_for_reduction): If the
30037 target requests first reduce vectors by combining low and high
30039 * tree-vect-stmts.c (vect_gen_perm_mask_any): Adjust.
30040 (get_vectype_for_scalar_type_and_size): Export.
30041 * tree-vectorizer.h (get_vectype_for_scalar_type_and_size): Declare.
30042 * doc/tm.texi.in (TARGET_VECTORIZE_SPLIT_REDUCTION): Document.
30043 * doc/tm.texi: Regenerate.
30044 * config/i386/i386.c (ix86_split_reduction): Implement
30045 TARGET_VECTORIZE_SPLIT_REDUCTION.
30047 2018-01-12 Eric Botcazou <ebotcazou@adacore.com>
30050 * config/sparc/sparc.h (PIC_OFFSET_TABLE_REGNUM): Set to INVALID_REGNUM
30051 in PIC mode except for TARGET_VXWORKS_RTP.
30052 * config/sparc/sparc.c: Include cfgrtl.h.
30053 (TARGET_INIT_PIC_REG): Define.
30054 (TARGET_USE_PSEUDO_PIC_REG): Likewise.
30055 (sparc_pic_register_p): New predicate.
30056 (sparc_legitimate_address_p): Use it.
30057 (sparc_legitimize_pic_address): Likewise.
30058 (sparc_delegitimize_address): Likewise.
30059 (sparc_mode_dependent_address_p): Likewise.
30060 (gen_load_pcrel_sym): Remove 4th parameter.
30061 (load_got_register): Adjust call to above. Remove obsolete stuff.
30062 (sparc_expand_prologue): Do not call load_got_register here.
30063 (sparc_flat_expand_prologue): Likewise.
30064 (sparc_output_mi_thunk): Set the pic_offset_table_rtx object.
30065 (sparc_use_pseudo_pic_reg): New function.
30066 (sparc_init_pic_reg): Likewise.
30067 * config/sparc/sparc.md (vxworks_load_got): Set the GOT register.
30068 (builtin_setjmp_receiver): Enable only for TARGET_VXWORKS_RTP.
30070 2018-01-12 Christophe Lyon <christophe.lyon@linaro.org>
30072 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
30073 Add item for branch_cost.
30075 2018-01-12 Eric Botcazou <ebotcazou@adacore.com>
30077 PR rtl-optimization/83565
30078 * rtlanal.c (nonzero_bits1): On WORD_REGISTER_OPERATIONS machines, do
30079 not extend the result to a larger mode for rotate operations.
30080 (num_sign_bit_copies1): Likewise.
30082 2018-01-12 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
30085 * config/sol2.h (STARTFILE_ARCH_SPEC): Don't use with -shared or
30087 Use values-Xc.o for -pedantic.
30088 Link with values-xpg4.o for C90, values-xpg6.o otherwise.
30090 2018-01-12 Martin Liska <mliska@suse.cz>
30093 * ipa-devirt.c (final_warning_record::grow_type_warnings):
30095 (possible_polymorphic_call_targets): Use it.
30096 (ipa_devirt): Likewise.
30098 2018-01-12 Martin Liska <mliska@suse.cz>
30100 * profile-count.h (enum profile_quality): Use 0 as invalid
30101 enum value of profile_quality.
30103 2018-01-12 Chung-Ju Wu <jasonwucj@gmail.com>
30105 * doc/invoke.texi (NDS32 Options): Add -mext-perf, -mext-perf2 and
30106 -mext-string options.
30108 2018-01-12 Richard Biener <rguenther@suse.de>
30110 * lto-streamer-out.c (DFS::DFS_write_tree_body): Process
30111 DECL_DEBUG_EXPR conditional on DECL_HAS_DEBUG_EXPR_P.
30112 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
30114 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
30116 2018-01-11 Michael Meissner <meissner@linux.vnet.ibm.com>
30118 * configure.ac (--with-long-double-format): Add support for the
30119 configuration option to change the default long double format on
30121 * config.gcc (powerpc*-linux*-*): Likewise.
30122 * configure: Regenerate.
30123 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If long
30124 double is IEEE, define __KC__ and __KF__ to allow floatn.h to be
30125 used without modification.
30127 2018-01-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30129 * config/rs6000/rs6000-builtin.def (BU_P7_MISC_X): New #define.
30130 (SPEC_BARRIER): New instantiation of BU_P7_MISC_X.
30131 * config/rs6000/rs6000.c (rs6000_expand_builtin): Handle
30132 MISC_BUILTIN_SPEC_BARRIER.
30133 (rs6000_init_builtins): Likewise.
30134 * config/rs6000/rs6000.md (UNSPECV_SPEC_BARRIER): New UNSPECV
30136 (speculation_barrier): New define_insn.
30137 * doc/extend.texi: Document __builtin_speculation_barrier.
30139 2018-01-11 Jakub Jelinek <jakub@redhat.com>
30142 * config/i386/i386.c (ix86_expand_vector_init_one_nonzero): If one_var
30143 is 0, for V{8,16}S[IF] and V[48]D[IF]mode use gen_vec_set<mode>_0.
30144 * config/i386/sse.md (VI8_AVX_AVX512F, VI4F_256_512): New mode
30146 (ssescalarmodesuffix): Add 512-bit vectors. Use "d" or "q" for
30147 integral modes instead of "ss" and "sd".
30148 (vec_set<mode>_0): New define_insns for 256-bit and 512-bit
30149 vectors with 32-bit and 64-bit elements.
30150 (vecdupssescalarmodesuffix): New mode attribute.
30151 (vec_dup<mode>): Use it.
30153 2018-01-11 H.J. Lu <hongjiu.lu@intel.com>
30156 * config/i386/i386.c (ix86_compute_frame_layout): Align stack
30157 frame if argument is passed on stack.
30159 2018-01-11 Jakub Jelinek <jakub@redhat.com>
30162 * ree.c (combine_reaching_defs): Optimize also
30163 reg2=exp; reg1=reg2; reg2=any_extend(reg1); into
30164 reg2=any_extend(exp); reg1=reg2;, formatting fix.
30166 2018-01-11 Jan Hubicka <hubicka@ucw.cz>
30168 PR middle-end/83189
30169 * gimple-ssa-isolate-paths.c (isolate_path): Fix profile update.
30171 2018-01-11 Jan Hubicka <hubicka@ucw.cz>
30173 PR middle-end/83718
30174 * tree-inline.c (copy_cfg_body): Adjust num&den for scaling
30175 after they are computed.
30177 2018-01-11 Bin Cheng <bin.cheng@arm.com>
30179 PR tree-optimization/83695
30180 * gimple-loop-linterchange.cc
30181 (tree_loop_interchange::interchange_loops): Call scev_reset_htab to
30182 reset cached scev information after interchange.
30183 (pass_linterchange::execute): Remove call to scev_reset_htab.
30185 2018-01-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30187 * config/arm/arm_neon.h (vfmlal_lane_low_u32, vfmlal_lane_high_u32,
30188 vfmlalq_laneq_low_u32, vfmlalq_lane_low_u32, vfmlal_laneq_low_u32,
30189 vfmlalq_laneq_high_u32, vfmlalq_lane_high_u32, vfmlal_laneq_high_u32,
30190 vfmlsl_lane_low_u32, vfmlsl_lane_high_u32, vfmlslq_laneq_low_u32,
30191 vfmlslq_lane_low_u32, vfmlsl_laneq_low_u32, vfmlslq_laneq_high_u32,
30192 vfmlslq_lane_high_u32, vfmlsl_laneq_high_u32): Define.
30193 * config/arm/arm_neon_builtins.def (vfmal_lane_low,
30194 vfmal_lane_lowv4hf, vfmal_lane_lowv8hf, vfmal_lane_high,
30195 vfmal_lane_highv4hf, vfmal_lane_highv8hf, vfmsl_lane_low,
30196 vfmsl_lane_lowv4hf, vfmsl_lane_lowv8hf, vfmsl_lane_high,
30197 vfmsl_lane_highv4hf, vfmsl_lane_highv8hf): New sets of builtins.
30198 * config/arm/iterators.md (VFMLSEL2, vfmlsel2): New mode attributes.
30199 (V_lane_reg): Likewise.
30200 * config/arm/neon.md (neon_vfm<vfml_op>l_lane_<vfml_half><VCVTF:mode>):
30202 (neon_vfm<vfml_op>l_lane_<vfml_half><vfmlsel2><mode>): Likewise.
30203 (vfmal_lane_low<mode>_intrinsic,
30204 vfmal_lane_low<vfmlsel2><mode>_intrinsic,
30205 vfmal_lane_high<vfmlsel2><mode>_intrinsic,
30206 vfmal_lane_high<mode>_intrinsic, vfmsl_lane_low<mode>_intrinsic,
30207 vfmsl_lane_low<vfmlsel2><mode>_intrinsic,
30208 vfmsl_lane_high<vfmlsel2><mode>_intrinsic,
30209 vfmsl_lane_high<mode>_intrinsic): New define_insns.
30211 2018-01-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30213 * config/arm/arm-cpus.in (fp16fml): New feature.
30214 (ALL_SIMD): Add fp16fml.
30215 (armv8.2-a): Add fp16fml as an option.
30216 (armv8.3-a): Likewise.
30217 (armv8.4-a): Add fp16fml as part of fp16.
30218 * config/arm/arm.h (TARGET_FP16FML): Define.
30219 * config/arm/arm-c.c (arm_cpu_builtins): Define __ARM_FEATURE_FP16_FML
30221 * config/arm/arm-modes.def (V2HF): Define.
30222 * config/arm/arm_neon.h (vfmlal_low_u32, vfmlsl_low_u32,
30223 vfmlal_high_u32, vfmlsl_high_u32, vfmlalq_low_u32,
30224 vfmlslq_low_u32, vfmlalq_high_u32, vfmlslq_high_u32): Define.
30225 * config/arm/arm_neon_builtins.def (vfmal_low, vfmal_high,
30226 vfmsl_low, vfmsl_high): New set of builtins.
30227 * config/arm/iterators.md (PLUSMINUS): New code iterator.
30228 (vfml_op): New code attribute.
30229 (VFMLHALVES): New int iterator.
30230 (VFML, VFMLSEL): New mode attributes.
30231 (V_reg): Define mapping for V2HF.
30232 (V_hi, V_lo): New mode attributes.
30233 (VF_constraint): Likewise.
30234 (vfml_half, vfml_half_selector): New int attributes.
30235 * config/arm/neon.md (neon_vfm<vfml_op>l_<vfml_half><mode>): New
30237 (vfmal_low<mode>_intrinsic, vfmsl_high<mode>_intrinsic,
30238 vfmal_high<mode>_intrinsic, vfmsl_low<mode>_intrinsic):
30240 * config/arm/t-arm-elf (v8_fps): Add fp16fml.
30241 * config/arm/t-multilib (v8_2_a_simd_variants): Add fp16fml.
30242 * config/arm/unspecs.md (UNSPEC_VFML_LO, UNSPEC_VFML_HI): New unspecs.
30243 * doc/invoke.texi (ARM Options): Document fp16fml. Update armv8.4-a
30245 * doc/sourcebuild.texi (arm_fp16fml_neon_ok, arm_fp16fml_neon):
30246 Document new effective target and option set.
30248 2018-01-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30250 * config/arm/arm-cpus.in (armv8_4): New feature.
30251 (ARMv8_4a): New fgroup.
30252 (armv8.4-a): New arch.
30253 * config/arm/arm-tables.opt: Regenerate.
30254 * config/arm/t-aprofile: Add matching rules for -march=armv8.4-a.
30255 * config/arm/t-arm-elf (all_v8_archs): Add armv8.4-a.
30256 * config/arm/t-multilib (v8_4_a_simd_variants): New variable.
30257 Add matching rules for -march=armv8.4-a and extensions.
30258 * doc/invoke.texi (ARM Options): Document -march=armv8.4-a.
30260 2018-01-11 Oleg Endo <olegendo@gcc.gnu.org>
30263 * config/rx/rx.md (BW): New mode attribute.
30264 (sync_lock_test_and_setsi): Add mode suffix to insn output.
30266 2018-01-11 Richard Biener <rguenther@suse.de>
30268 PR tree-optimization/83435
30269 * graphite.c (canonicalize_loop_form): Ignore fake loop exit edges.
30270 * graphite-scop-detection.c (scop_detection::get_sese): Likewise.
30271 * tree-vrp.c (add_assert_info): Drop TREE_OVERFLOW if they appear.
30273 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
30274 Alan Hayward <alan.hayward@arm.com>
30275 David Sherwood <david.sherwood@arm.com>
30277 * config/aarch64/aarch64.c (aarch64_address_info): Add a const_offset
30279 (aarch64_classify_address): Initialize it. Track polynomial offsets.
30280 (aarch64_print_address_internal): Use it to check for a zero offset.
30282 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
30283 Alan Hayward <alan.hayward@arm.com>
30284 David Sherwood <david.sherwood@arm.com>
30286 * config/aarch64/aarch64-modes.def (NUM_POLY_INT_COEFFS): Set to 2.
30287 * config/aarch64/aarch64-protos.h (aarch64_initial_elimination_offset):
30288 Return a poly_int64 rather than a HOST_WIDE_INT.
30289 (aarch64_offset_7bit_signed_scaled_p): Take the offset as a poly_int64
30290 rather than a HOST_WIDE_INT.
30291 * config/aarch64/aarch64.h (aarch64_frame): Protect with
30292 HAVE_POLY_INT_H rather than HOST_WIDE_INT. Change locals_offset,
30293 hard_fp_offset, frame_size, initial_adjust, callee_offset and
30294 final_offset from HOST_WIDE_INT to poly_int64.
30295 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args): Use
30296 to_constant when getting the number of units in an Advanced SIMD
30298 (aarch64_builtin_vectorized_function): Check for a constant number
30300 * config/aarch64/aarch64-simd.md (mov<mode>): Handle polynomial
30302 (aarch64_ld<VSTRUCT:nregs>_lane<VALLDIF:mode>): Use the nunits
30303 attribute instead of GET_MODE_NUNITS.
30304 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs)
30305 (aarch64_class_max_nregs): Use the constant_lowest_bound of the
30306 GET_MODE_SIZE for fixed-size registers.
30307 (aarch64_const_vec_all_same_in_range_p): Use const_vec_duplicate_p.
30308 (aarch64_hard_regno_call_part_clobbered, aarch64_classify_index)
30309 (aarch64_mode_valid_for_sched_fusion_p, aarch64_classify_address)
30310 (aarch64_legitimize_address_displacement, aarch64_secondary_reload)
30311 (aarch64_print_operand, aarch64_print_address_internal)
30312 (aarch64_address_cost, aarch64_rtx_costs, aarch64_register_move_cost)
30313 (aarch64_short_vector_p, aapcs_vfp_sub_candidate)
30314 (aarch64_simd_attr_length_rglist, aarch64_operands_ok_for_ldpstp):
30315 Handle polynomial GET_MODE_SIZE.
30316 (aarch64_hard_regno_caller_save_mode): Likewise. Return modes
30317 wider than SImode without modification.
30318 (tls_symbolic_operand_type): Use strip_offset instead of split_const.
30319 (aarch64_pass_by_reference, aarch64_layout_arg, aarch64_pad_reg_upward)
30320 (aarch64_gimplify_va_arg_expr): Assert that we don't yet handle
30321 passing and returning SVE modes.
30322 (aarch64_function_value, aarch64_layout_arg): Use gen_int_mode
30323 rather than GEN_INT.
30324 (aarch64_emit_probe_stack_range): Take the size as a poly_int64
30325 rather than a HOST_WIDE_INT, but call sorry if it isn't constant.
30326 (aarch64_allocate_and_probe_stack_space): Likewise.
30327 (aarch64_layout_frame): Cope with polynomial offsets.
30328 (aarch64_save_callee_saves, aarch64_restore_callee_saves): Take the
30329 start_offset as a poly_int64 rather than a HOST_WIDE_INT. Track
30330 polynomial offsets.
30331 (offset_9bit_signed_unscaled_p, offset_12bit_unsigned_scaled_p)
30332 (aarch64_offset_7bit_signed_scaled_p): Take the offset as a
30333 poly_int64 rather than a HOST_WIDE_INT.
30334 (aarch64_get_separate_components, aarch64_process_components)
30335 (aarch64_expand_prologue, aarch64_expand_epilogue)
30336 (aarch64_use_return_insn_p): Handle polynomial frame offsets.
30337 (aarch64_anchor_offset): New function, split out from...
30338 (aarch64_legitimize_address): ...here.
30339 (aarch64_builtin_vectorization_cost): Handle polynomial
30340 TYPE_VECTOR_SUBPARTS.
30341 (aarch64_simd_check_vect_par_cnst_half): Handle polynomial
30343 (aarch64_simd_make_constant, aarch64_expand_vector_init): Get the
30344 number of elements from the PARALLEL rather than the mode.
30345 (aarch64_shift_truncation_mask): Use GET_MODE_UNIT_BITSIZE
30346 rather than GET_MODE_BITSIZE.
30347 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_ext)
30348 (aarch64_evpc_rev, aarch64_evpc_dup, aarch64_evpc_zip)
30349 (aarch64_expand_vec_perm_const_1): Handle polynomial
30350 d->perm.length () and d->perm elements.
30351 (aarch64_evpc_tbl): Likewise. Use nelt rather than GET_MODE_NUNITS.
30352 Apply to_constant to d->perm elements.
30353 (aarch64_simd_valid_immediate, aarch64_vec_fpconst_pow_of_2): Handle
30354 polynomial CONST_VECTOR_NUNITS.
30355 (aarch64_move_pointer): Take amount as a poly_int64 rather
30357 (aarch64_progress_pointer): Avoid temporary variable.
30358 * config/aarch64/aarch64.md (aarch64_<crc_variant>): Use
30359 the mode attribute instead of GET_MODE.
30361 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
30362 Alan Hayward <alan.hayward@arm.com>
30363 David Sherwood <david.sherwood@arm.com>
30365 * config/aarch64/aarch64.c (aarch64_force_temporary): Assert that
30366 x exists before using it.
30367 (aarch64_add_constant_internal): Rename to...
30368 (aarch64_add_offset_1): ...this. Replace regnum with separate
30369 src and dest rtxes. Handle the case in which they're different,
30370 including when the offset is zero. Replace scratchreg with an rtx.
30371 Use 2 additions if there is no spare register into which we can
30372 move a 16-bit constant.
30373 (aarch64_add_constant): Delete.
30374 (aarch64_add_offset): Replace reg with separate src and dest
30375 rtxes. Take a poly_int64 offset instead of a HOST_WIDE_INT.
30376 Use aarch64_add_offset_1.
30377 (aarch64_add_sp, aarch64_sub_sp): Take the scratch register as
30378 an rtx rather than an int. Take the delta as a poly_int64
30379 rather than a HOST_WIDE_INT. Use aarch64_add_offset.
30380 (aarch64_expand_mov_immediate): Update uses of aarch64_add_offset.
30381 (aarch64_expand_prologue): Update calls to aarch64_sub_sp,
30382 aarch64_allocate_and_probe_stack_space and aarch64_add_offset.
30383 (aarch64_expand_epilogue): Update calls to aarch64_add_offset
30384 and aarch64_add_sp.
30385 (aarch64_output_mi_thunk): Use aarch64_add_offset rather than
30386 aarch64_add_constant.
30388 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
30390 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
30391 Use scalar_float_mode.
30393 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
30395 * config/aarch64/aarch64-simd.md
30396 (aarch64_fml<f16mac1>l<f16quad>_low<mode>): Avoid GET_MODE_NUNITS.
30397 (aarch64_fml<f16mac1>l<f16quad>_high<mode>): Likewise.
30398 (aarch64_fml<f16mac1>l_lane_lowv2sf): Likewise.
30399 (aarch64_fml<f16mac1>l_lane_highv2sf): Likewise.
30400 (aarch64_fml<f16mac1>lq_laneq_lowv4sf): Likewise.
30401 (aarch64_fml<f16mac1>lq_laneq_highv4sf): Likewise.
30402 (aarch64_fml<f16mac1>l_laneq_lowv2sf): Likewise.
30403 (aarch64_fml<f16mac1>l_laneq_highv2sf): Likewise.
30404 (aarch64_fml<f16mac1>lq_lane_lowv4sf): Likewise.
30405 (aarch64_fml<f16mac1>lq_lane_highv4sf): Likewise.
30407 2018-01-11 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
30410 * config/arm/arm.c (arm_declare_function_name): Set arch_to_print if
30411 targ_options->x_arm_arch_string is non NULL.
30413 2018-01-11 Tamar Christina <tamar.christina@arm.com>
30415 * config/aarch64/aarch64.h
30416 (AARCH64_FL_FOR_ARCH8_4): Add AARCH64_FL_DOTPROD.
30418 2018-01-11 Sudakshina Das <sudi.das@arm.com>
30421 * expmed.c (emit_store_flag_force): Swap if const op0
30422 and change VOIDmode to mode of op0.
30424 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
30426 PR rtl-optimization/83761
30427 * caller-save.c (replace_reg_with_saved_mem): Pass bits rather
30428 than bytes to mode_for_size.
30430 2018-01-10 Jan Hubicka <hubicka@ucw.cz>
30432 PR middle-end/83189
30433 * gfortran.fortran-torture/compile/pr83189.f90: New testcase.
30434 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Handle zero
30437 2018-01-10 Jan Hubicka <hubicka@ucw.cz>
30439 PR middle-end/83575
30440 * cfgrtl.c (rtl_verify_edges): Only verify fixability of partition
30441 when in layout mode.
30442 (cfg_layout_finalize): Do not verify cfg before we are out of layout.
30443 * cfgcleanup.c (try_optimize_cfg): Only verify flow info when doing
30446 2018-01-10 Michael Collison <michael.collison@arm.com>
30448 * config/aarch64/aarch64-modes.def (V2HF): New VECTOR_MODE.
30449 * config/aarch64/aarch64-option-extension.def: Add
30450 AARCH64_OPT_EXTENSION of 'fp16fml'.
30451 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
30452 (__ARM_FEATURE_FP16_FML): Define if TARGET_F16FML is true.
30453 * config/aarch64/predicates.md (aarch64_lane_imm3): New predicate.
30454 * config/aarch64/constraints.md (Ui7): New constraint.
30455 * config/aarch64/iterators.md (VFMLA_W): New mode iterator.
30456 (VFMLA_SEL_W): Ditto.
30459 (VFMLA16_LOW): New int iterator.
30460 (VFMLA16_HIGH): Ditto.
30461 (UNSPEC_FMLAL): New unspec.
30462 (UNSPEC_FMLSL): Ditto.
30463 (UNSPEC_FMLAL2): Ditto.
30464 (UNSPEC_FMLSL2): Ditto.
30465 (f16mac): New code attribute.
30466 * config/aarch64/aarch64-simd-builtins.def
30467 (aarch64_fmlal_lowv2sf): Ditto.
30468 (aarch64_fmlsl_lowv2sf): Ditto.
30469 (aarch64_fmlalq_lowv4sf): Ditto.
30470 (aarch64_fmlslq_lowv4sf): Ditto.
30471 (aarch64_fmlal_highv2sf): Ditto.
30472 (aarch64_fmlsl_highv2sf): Ditto.
30473 (aarch64_fmlalq_highv4sf): Ditto.
30474 (aarch64_fmlslq_highv4sf): Ditto.
30475 (aarch64_fmlal_lane_lowv2sf): Ditto.
30476 (aarch64_fmlsl_lane_lowv2sf): Ditto.
30477 (aarch64_fmlal_laneq_lowv2sf): Ditto.
30478 (aarch64_fmlsl_laneq_lowv2sf): Ditto.
30479 (aarch64_fmlalq_lane_lowv4sf): Ditto.
30480 (aarch64_fmlsl_lane_lowv4sf): Ditto.
30481 (aarch64_fmlalq_laneq_lowv4sf): Ditto.
30482 (aarch64_fmlsl_laneq_lowv4sf): Ditto.
30483 (aarch64_fmlal_lane_highv2sf): Ditto.
30484 (aarch64_fmlsl_lane_highv2sf): Ditto.
30485 (aarch64_fmlal_laneq_highv2sf): Ditto.
30486 (aarch64_fmlsl_laneq_highv2sf): Ditto.
30487 (aarch64_fmlalq_lane_highv4sf): Ditto.
30488 (aarch64_fmlsl_lane_highv4sf): Ditto.
30489 (aarch64_fmlalq_laneq_highv4sf): Ditto.
30490 (aarch64_fmlsl_laneq_highv4sf): Ditto.
30491 * config/aarch64/aarch64-simd.md:
30492 (aarch64_fml<f16mac1>l<f16quad>_low<mode>): New pattern.
30493 (aarch64_fml<f16mac1>l<f16quad>_high<mode>): Ditto.
30494 (aarch64_simd_fml<f16mac1>l<f16quad>_low<mode>): Ditto.
30495 (aarch64_simd_fml<f16mac1>l<f16quad>_high<mode>): Ditto.
30496 (aarch64_fml<f16mac1>l_lane_lowv2sf): Ditto.
30497 (aarch64_fml<f16mac1>l_lane_highv2sf): Ditto.
30498 (aarch64_simd_fml<f16mac>l_lane_lowv2sf): Ditto.
30499 (aarch64_simd_fml<f16mac>l_lane_highv2sf): Ditto.
30500 (aarch64_fml<f16mac1>lq_laneq_lowv4sf): Ditto.
30501 (aarch64_fml<f16mac1>lq_laneq_highv4sf): Ditto.
30502 (aarch64_simd_fml<f16mac>lq_laneq_lowv4sf): Ditto.
30503 (aarch64_simd_fml<f16mac>lq_laneq_highv4sf): Ditto.
30504 (aarch64_fml<f16mac1>l_laneq_lowv2sf): Ditto.
30505 (aarch64_fml<f16mac1>l_laneq_highv2sf): Ditto.
30506 (aarch64_simd_fml<f16mac>l_laneq_lowv2sf): Ditto.
30507 (aarch64_simd_fml<f16mac>l_laneq_highv2sf): Ditto.
30508 (aarch64_fml<f16mac1>lq_lane_lowv4sf): Ditto.
30509 (aarch64_fml<f16mac1>lq_lane_highv4sf): Ditto.
30510 (aarch64_simd_fml<f16mac>lq_lane_lowv4sf): Ditto.
30511 (aarch64_simd_fml<f16mac>lq_lane_highv4sf): Ditto.
30512 * config/aarch64/arm_neon.h (vfmlal_low_u32): New intrinsic.
30513 (vfmlsl_low_u32): Ditto.
30514 (vfmlalq_low_u32): Ditto.
30515 (vfmlslq_low_u32): Ditto.
30516 (vfmlal_high_u32): Ditto.
30517 (vfmlsl_high_u32): Ditto.
30518 (vfmlalq_high_u32): Ditto.
30519 (vfmlslq_high_u32): Ditto.
30520 (vfmlal_lane_low_u32): Ditto.
30521 (vfmlsl_lane_low_u32): Ditto.
30522 (vfmlal_laneq_low_u32): Ditto.
30523 (vfmlsl_laneq_low_u32): Ditto.
30524 (vfmlalq_lane_low_u32): Ditto.
30525 (vfmlslq_lane_low_u32): Ditto.
30526 (vfmlalq_laneq_low_u32): Ditto.
30527 (vfmlslq_laneq_low_u32): Ditto.
30528 (vfmlal_lane_high_u32): Ditto.
30529 (vfmlsl_lane_high_u32): Ditto.
30530 (vfmlal_laneq_high_u32): Ditto.
30531 (vfmlsl_laneq_high_u32): Ditto.
30532 (vfmlalq_lane_high_u32): Ditto.
30533 (vfmlslq_lane_high_u32): Ditto.
30534 (vfmlalq_laneq_high_u32): Ditto.
30535 (vfmlslq_laneq_high_u32): Ditto.
30536 * config/aarch64/aarch64.h (AARCH64_FL_F16SML): New flag.
30537 (AARCH64_FL_FOR_ARCH8_4): New.
30538 (AARCH64_ISA_F16FML): New ISA flag.
30539 (TARGET_F16FML): New feature flag for fp16fml.
30540 (doc/invoke.texi): Document new fp16fml option.
30542 2018-01-10 Michael Collison <michael.collison@arm.com>
30544 * config/aarch64/aarch64-builtins.c:
30545 (aarch64_types_ternopu_imm_qualifiers, TYPES_TERNOPUI): New.
30546 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
30547 (__ARM_FEATURE_SHA3): Define if TARGET_SHA3 is true.
30548 * config/aarch64/aarch64.h (AARCH64_FL_SHA3): New flags.
30549 (AARCH64_ISA_SHA3): New ISA flag.
30550 (TARGET_SHA3): New feature flag for sha3.
30551 * config/aarch64/iterators.md (sha512_op): New int attribute.
30552 (CRYPTO_SHA512): New int iterator.
30553 (UNSPEC_SHA512H): New unspec.
30554 (UNSPEC_SHA512H2): Ditto.
30555 (UNSPEC_SHA512SU0): Ditto.
30556 (UNSPEC_SHA512SU1): Ditto.
30557 * config/aarch64/aarch64-simd-builtins.def
30558 (aarch64_crypto_sha512hqv2di): New builtin.
30559 (aarch64_crypto_sha512h2qv2di): Ditto.
30560 (aarch64_crypto_sha512su0qv2di): Ditto.
30561 (aarch64_crypto_sha512su1qv2di): Ditto.
30562 (aarch64_eor3qv8hi): Ditto.
30563 (aarch64_rax1qv2di): Ditto.
30564 (aarch64_xarqv2di): Ditto.
30565 (aarch64_bcaxqv8hi): Ditto.
30566 * config/aarch64/aarch64-simd.md:
30567 (aarch64_crypto_sha512h<sha512_op>qv2di): New pattern.
30568 (aarch64_crypto_sha512su0qv2di): Ditto.
30569 (aarch64_crypto_sha512su1qv2di): Ditto.
30570 (aarch64_eor3qv8hi): Ditto.
30571 (aarch64_rax1qv2di): Ditto.
30572 (aarch64_xarqv2di): Ditto.
30573 (aarch64_bcaxqv8hi): Ditto.
30574 * config/aarch64/arm_neon.h (vsha512hq_u64): New intrinsic.
30575 (vsha512h2q_u64): Ditto.
30576 (vsha512su0q_u64): Ditto.
30577 (vsha512su1q_u64): Ditto.
30578 (veor3q_u16): Ditto.
30579 (vrax1q_u64): Ditto.
30580 (vxarq_u64): Ditto.
30581 (vbcaxq_u16): Ditto.
30582 * config/arm/types.md (crypto_sha512): New type attribute.
30583 (crypto_sha3): Ditto.
30584 (doc/invoke.texi): Document new sha3 option.
30586 2018-01-10 Michael Collison <michael.collison@arm.com>
30588 * config/aarch64/aarch64-builtins.c:
30589 (aarch64_types_quadopu_imm_qualifiers, TYPES_QUADOPUI): New.
30590 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
30591 (__ARM_FEATURE_SM3): Define if TARGET_SM4 is true.
30592 (__ARM_FEATURE_SM4): Define if TARGET_SM4 is true.
30593 * config/aarch64/aarch64.h (AARCH64_FL_SM4): New flags.
30594 (AARCH64_ISA_SM4): New ISA flag.
30595 (TARGET_SM4): New feature flag for sm4.
30596 * config/aarch64/aarch64-simd-builtins.def
30597 (aarch64_sm3ss1qv4si): Ditto.
30598 (aarch64_sm3tt1aq4si): Ditto.
30599 (aarch64_sm3tt1bq4si): Ditto.
30600 (aarch64_sm3tt2aq4si): Ditto.
30601 (aarch64_sm3tt2bq4si): Ditto.
30602 (aarch64_sm3partw1qv4si): Ditto.
30603 (aarch64_sm3partw2qv4si): Ditto.
30604 (aarch64_sm4eqv4si): Ditto.
30605 (aarch64_sm4ekeyqv4si): Ditto.
30606 * config/aarch64/aarch64-simd.md:
30607 (aarch64_sm3ss1qv4si): Ditto.
30608 (aarch64_sm3tt<sm3tt_op>qv4si): Ditto.
30609 (aarch64_sm3partw<sm3part_op>qv4si): Ditto.
30610 (aarch64_sm4eqv4si): Ditto.
30611 (aarch64_sm4ekeyqv4si): Ditto.
30612 * config/aarch64/iterators.md (sm3tt_op): New int iterator.
30613 (sm3part_op): Ditto.
30614 (CRYPTO_SM3TT): Ditto.
30615 (CRYPTO_SM3PART): Ditto.
30616 (UNSPEC_SM3SS1): New unspec.
30617 (UNSPEC_SM3TT1A): Ditto.
30618 (UNSPEC_SM3TT1B): Ditto.
30619 (UNSPEC_SM3TT2A): Ditto.
30620 (UNSPEC_SM3TT2B): Ditto.
30621 (UNSPEC_SM3PARTW1): Ditto.
30622 (UNSPEC_SM3PARTW2): Ditto.
30623 (UNSPEC_SM4E): Ditto.
30624 (UNSPEC_SM4EKEY): Ditto.
30625 * config/aarch64/constraints.md (Ui2): New constraint.
30626 * config/aarch64/predicates.md (aarch64_imm2): New predicate.
30627 * config/arm/types.md (crypto_sm3): New type attribute.
30628 (crypto_sm4): Ditto.
30629 * config/aarch64/arm_neon.h (vsm3ss1q_u32): New intrinsic.
30630 (vsm3tt1aq_u32): Ditto.
30631 (vsm3tt1bq_u32): Ditto.
30632 (vsm3tt2aq_u32): Ditto.
30633 (vsm3tt2bq_u32): Ditto.
30634 (vsm3partw1q_u32): Ditto.
30635 (vsm3partw2q_u32): Ditto.
30636 (vsm4eq_u32): Ditto.
30637 (vsm4ekeyq_u32): Ditto.
30638 (doc/invoke.texi): Document new sm4 option.
30640 2018-01-10 Michael Collison <michael.collison@arm.com>
30642 * config/aarch64/aarch64-arches.def (armv8.4-a): New architecture.
30643 * config/aarch64/aarch64.h (AARCH64_ISA_V8_4): New ISA flag.
30644 (AARCH64_FL_FOR_ARCH8_4): New.
30645 (AARCH64_FL_V8_4): New flag.
30646 (doc/invoke.texi): Document new armv8.4-a option.
30648 2018-01-10 Michael Collison <michael.collison@arm.com>
30650 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
30651 (__ARM_FEATURE_AES): Define if TARGET_AES is true.
30652 (__ARM_FEATURE_SHA2): Define if TARGET_SHA2 is true.
30653 * config/aarch64/aarch64-option-extension.def: Add
30654 AARCH64_OPT_EXTENSION of 'sha2'.
30655 (aes): Add AARCH64_OPT_EXTENSION of 'aes'.
30656 (crypto): Disable sha2 and aes if crypto disabled.
30657 (crypto): Enable aes and sha2 if enabled.
30658 (simd): Disable sha2 and aes if simd disabled.
30659 * config/aarch64/aarch64.h (AARCH64_FL_AES, AARCH64_FL_SHA2):
30661 (AARCH64_ISA_AES, AARCH64_ISA_SHA2): New ISA flags.
30662 (TARGET_SHA2): New feature flag for sha2.
30663 (TARGET_AES): New feature flag for aes.
30664 * config/aarch64/aarch64-simd.md:
30665 (aarch64_crypto_aes<aes_op>v16qi): Make pattern
30666 conditional on TARGET_AES.
30667 (aarch64_crypto_aes<aesmc_op>v16qi): Ditto.
30668 (aarch64_crypto_sha1hsi): Make pattern conditional
30670 (aarch64_crypto_sha1hv4si): Ditto.
30671 (aarch64_be_crypto_sha1hv4si): Ditto.
30672 (aarch64_crypto_sha1su1v4si): Ditto.
30673 (aarch64_crypto_sha1<sha1_op>v4si): Ditto.
30674 (aarch64_crypto_sha1su0v4si): Ditto.
30675 (aarch64_crypto_sha256h<sha256_op>v4si): Ditto.
30676 (aarch64_crypto_sha256su0v4si): Ditto.
30677 (aarch64_crypto_sha256su1v4si): Ditto.
30678 (doc/invoke.texi): Document new aes and sha2 options.
30680 2018-01-10 Martin Sebor <msebor@redhat.com>
30682 PR tree-optimization/83781
30683 * gimple-fold.c (get_range_strlen): Avoid treating arrays of pointers
30686 2018-01-11 Martin Sebor <msebor@gmail.com>
30687 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
30689 PR tree-optimization/83501
30690 PR tree-optimization/81703
30692 * tree-ssa-strlen.c (get_string_cst): Rename...
30693 (get_string_len): ...to this. Handle global constants.
30694 (handle_char_store): Adjust.
30696 2018-01-10 Kito Cheng <kito.cheng@gmail.com>
30697 Jim Wilson <jimw@sifive.com>
30699 * config/riscv/riscv-protos.h (riscv_output_return): New.
30700 * config/riscv/riscv.c (struct machine_function): New naked_p field.
30701 (riscv_attribute_table, riscv_output_return),
30702 (riscv_handle_fndecl_attribute, riscv_naked_function_p),
30703 (riscv_allocate_stack_slots_for_args, riscv_warn_func_return): New.
30704 (riscv_compute_frame_info): Only compute frame->mask if not a naked
30706 (riscv_expand_prologue): Add early return for naked function.
30707 (riscv_expand_epilogue): Likewise.
30708 (riscv_function_ok_for_sibcall): Return false for naked function.
30709 (riscv_set_current_function): New.
30710 (TARGET_SET_CURRENT_FUNCTION, TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS),
30711 (TARGET_ATTRIBUTE_TABLE, TARGET_WARN_FUNC_RETURN): New.
30712 * config/riscv/riscv.md (simple_return): Call riscv_output_return.
30713 * doc/extend.texi (RISC-V Function Attributes): New.
30715 2018-01-10 Michael Meissner <meissner@linux.vnet.ibm.com>
30717 * config/rs6000/rs6000.c (is_complex_IBM_long_double): Explicitly
30718 check for 128-bit long double before checking TCmode.
30719 * config/rs6000/rs6000.h (FLOAT128_IEEE_P): Explicitly check for
30720 128-bit long doubles before checking TFmode or TCmode.
30721 (FLOAT128_IBM_P): Likewise.
30723 2018-01-10 Martin Sebor <msebor@redhat.com>
30725 PR tree-optimization/83671
30726 * builtins.c (c_strlen): Unconditionally return zero for the empty
30728 Use -Warray-bounds for warnings.
30729 * gimple-fold.c (get_range_strlen): Handle non-constant lengths
30730 for non-constant array indices with COMPONENT_REF, arrays of
30731 arrays, and pointers to arrays.
30732 (gimple_fold_builtin_strlen): Determine and set length range for
30733 non-constant character arrays.
30735 2018-01-10 Aldy Hernandez <aldyh@redhat.com>
30737 PR middle-end/81897
30738 * tree-ssa-uninit.c (convert_control_dep_chain_into_preds): Skip
30741 2018-01-10 Eric Botcazou <ebotcazou@adacore.com>
30743 * dwarf2out.c (dwarf2out_var_location): Do not pass NULL to fprintf.
30745 2018-01-10 Peter Bergner <bergner@vnet.ibm.com>
30748 * config/rs6000/rs6000.c (print_operand) <'y'>: Use
30749 VECTOR_MEM_ALTIVEC_OR_VSX_P.
30750 * config/rs6000/vsx.md (*vsx_le_perm_load_<mode> for VSX_D): Use
30751 indexed_or_indirect_operand predicate.
30752 (*vsx_le_perm_load_<mode> for VSX_W): Likewise.
30753 (*vsx_le_perm_load_v8hi): Likewise.
30754 (*vsx_le_perm_load_v16qi): Likewise.
30755 (*vsx_le_perm_store_<mode> for VSX_D): Likewise.
30756 (*vsx_le_perm_store_<mode> for VSX_W): Likewise.
30757 (*vsx_le_perm_store_v8hi): Likewise.
30758 (*vsx_le_perm_store_v16qi): Likewise.
30759 (eight unnamed splitters): Likewise.
30761 2018-01-10 Peter Bergner <bergner@vnet.ibm.com>
30763 * config/rs6000/x86intrin.h: Change #warning to #error. Update message.
30764 * config/rs6000/emmintrin.h: Likewise.
30765 * config/rs6000/mmintrin.h: Likewise.
30766 * config/rs6000/xmmintrin.h: Likewise.
30768 2018-01-10 David Malcolm <dmalcolm@redhat.com>
30771 * tree-core.h: Document EXPR_LOCATION_WRAPPER_P's usage of
30773 * tree.c (tree_nop_conversion): Return true for location wrapper
30775 (maybe_wrap_with_location): New function.
30776 (selftest::check_strip_nops): New function.
30777 (selftest::test_location_wrappers): New function.
30778 (selftest::tree_c_tests): Call it.
30779 * tree.h (STRIP_ANY_LOCATION_WRAPPER): New macro.
30780 (maybe_wrap_with_location): New decl.
30781 (EXPR_LOCATION_WRAPPER_P): New macro.
30782 (location_wrapper_p): New inline function.
30783 (tree_strip_any_location_wrapper): New inline function.
30785 2018-01-10 H.J. Lu <hongjiu.lu@intel.com>
30788 * config/i386/i386.c (ix86_compute_frame_layout): Always adjust
30789 stack_realign_offset for the largest alignment of stack slot
30791 (ix86_find_max_used_stack_alignment): New function.
30792 (ix86_finalize_stack_frame_flags): Use it. Set
30793 max_used_stack_alignment if we don't realign stack.
30794 * config/i386/i386.h (machine_function): Add
30795 max_used_stack_alignment.
30797 2018-01-10 Christophe Lyon <christophe.lyon@linaro.org>
30799 * config/arm/arm.opt (-mbranch-cost): New option.
30800 * config/arm/arm.h (BRANCH_COST): Take arm_branch_cost into
30803 2018-01-10 Segher Boessenkool <segher@kernel.crashing.org>
30806 * config/rs6000/rs6000.md (load_toc_v4_PIC_2, load_toc_v4_PIC_3b,
30807 load_toc_v4_PIC_3c): Wrap const term in CONST RTL.
30809 2018-01-10 Richard Biener <rguenther@suse.de>
30812 * dwarf2out.c (gen_subprogram_die): Hoist old_die && declaration
30813 early out so it also covers the case where we have a non-NULL
30816 2018-01-10 Richard Sandiford <richard.sandiford@linaro.org>
30818 PR tree-optimization/83753
30819 * tree-vect-stmts.c (get_group_load_store_type): Use VMAT_CONTIGUOUS
30820 for non-strided grouped accesses if the number of elements is 1.
30822 2018-01-10 Jan Hubicka <hubicka@ucw.cz>
30825 * i386.c (ix86_vectorize_builtin_gather): Check TARGET_USE_GATHER.
30826 * i386.h (TARGET_USE_GATHER): Define.
30827 * x86-tune.def (X86_TUNE_USE_GATHER): New.
30829 2018-01-10 Martin Liska <mliska@suse.cz>
30832 * basic-block.h (CLEANUP_NO_PARTITIONING): New define.
30833 * bb-reorder.c (pass_reorder_blocks::execute): Do not clean up
30835 * cfgcleanup.c (try_optimize_cfg): Fix up partitioning if
30836 CLEANUP_NO_PARTITIONING is not set.
30838 2018-01-10 Richard Sandiford <richard.sandiford@linaro.org>
30840 * doc/rtl.texi: Remove documentation of (const ...) wrappers
30841 for vectors, as a partial revert of r254296.
30842 * rtl.h (const_vec_p): Delete.
30843 (const_vec_duplicate_p): Don't test for vector CONSTs.
30844 (unwrap_const_vec_duplicate, const_vec_series_p): Likewise.
30845 * expmed.c (make_tree): Likewise.
30848 * common.md (E, F): Use CONSTANT_P instead of checking for
30850 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
30851 checking for CONST_VECTOR.
30853 2018-01-09 Jan Hubicka <hubicka@ucw.cz>
30855 PR middle-end/83575
30856 * predict.c (force_edge_cold): Handle in more sane way edges
30857 with no prediction.
30859 2018-01-09 Carl Love <cel@us.ibm.com>
30861 * config/rs6002/altivec.md (p8_vmrgow): Add support for V2DI, V2DF,
30863 (p8_vmrgew): Add support for V2DI, V2DF, V4SF types.
30864 * config/rs6000/rs6000-builtin.def: Add definitions for FLOAT2_V2DF,
30865 VMRGEW_V2DI, VMRGEW_V2DF, VMRGEW_V4SF, VMRGOW_V4SI, VMRGOW_V4SF,
30866 VMRGOW_V2DI, VMRGOW_V2DF. Remove definition for VMRGOW.
30867 * config/rs6000/rs6000-c.c (VSX_BUILTIN_VEC_FLOAT2,
30868 P8V_BUILTIN_VEC_VMRGEW, P8V_BUILTIN_VEC_VMRGOW): Add definitions.
30869 * config/rs6000/rs6000-protos.h: Add extern defition for
30870 rs6000_generate_float2_double_code.
30871 * config/rs6000/rs6000.c (rs6000_generate_float2_double_code): Add
30873 * config/rs6000/vsx.md (vsx_xvcdpsp): Add define_insn.
30874 (float2_v2df): Add define_expand.
30876 2018-01-09 Uros Bizjak <ubizjak@gmail.com>
30879 * combine.c (force_int_to_mode) <case ASHIFT>: Use mode instead of
30880 op_mode in the force_to_mode call.
30882 2018-01-09 Richard Sandiford <richard.sandiford@linaro.org>
30884 * config/aarch64/aarch64.c (aarch64_evpc_trn): Use d.perm.series_p
30885 instead of checking each element individually.
30886 (aarch64_evpc_uzp): Likewise.
30887 (aarch64_evpc_zip): Likewise.
30888 (aarch64_evpc_ext): Likewise.
30889 (aarch64_evpc_rev): Likewise.
30890 (aarch64_evpc_dup): Test the encoding for a single duplicated element,
30891 instead of checking each element individually. Return true without
30893 (aarch64_vectorize_vec_perm_const): Use all_from_input_p to test
30894 whether all selected elements come from the same input, instead of
30895 checking each element individually. Remove calls to gen_rtx_REG,
30896 start_sequence and end_sequence and instead assert that no rtl is
30899 2018-01-09 Richard Sandiford <richard.sandiford@linaro.org>
30901 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p): Fix
30902 order of HIGH and CONST checks.
30904 2018-01-09 Richard Sandiford <richard.sandiford@linaro.org>
30906 * tree-vect-stmts.c (permute_vec_elements): Create a fresh variable
30907 if the destination isn't an SSA_NAME.
30909 2018-01-09 Richard Biener <rguenther@suse.de>
30911 PR tree-optimization/83668
30912 * graphite.c (canonicalize_loop_closed_ssa): Add edge argument,
30914 (canonicalize_loop_form): ... here, renamed from ...
30915 (canonicalize_loop_closed_ssa_form): ... this and amended to
30916 swap successor edges for loop exit blocks to make us use
30917 the RPO order we need for initial schedule generation.
30919 2018-01-09 Joseph Myers <joseph@codesourcery.com>
30921 PR tree-optimization/64811
30922 * match.pd: When optimizing comparisons with Inf, avoid
30923 introducing or losing exceptions from comparisons with NaN.
30925 2018-01-09 Martin Liska <mliska@suse.cz>
30928 * asan.c (shadow_mem_size): Add gcc_assert.
30930 2018-01-09 Georg-Johann Lay <avr@gjlay.de>
30932 Don't save registers in main().
30935 * doc/invoke.texi (AVR Options) [-mmain-is-OS_task]: Document it.
30936 * config/avr/avr.opt (-mmain-is-OS_task): New target option.
30937 * config/avr/avr.c (avr_set_current_function): Don't error if
30938 naked, OS_task or OS_main are specified at the same time.
30939 (avr_function_ok_for_sibcall): Don't disable sibcalls for OS_task,
30941 (avr_insert_attributes) [-mmain-is-OS_task] <main>: Add OS_task
30943 * common/config/avr/avr-common.c (avr_option_optimization_table):
30944 Switch on -mmain-is-OS_task for optimizing compilations.
30946 2018-01-09 Richard Biener <rguenther@suse.de>
30948 PR tree-optimization/83572
30949 * graphite.c: Include cfganal.h.
30950 (graphite_transform_loops): Connect infinite loops to exit
30951 and remove fake edges at the end.
30953 2018-01-09 Jan Hubicka <hubicka@ucw.cz>
30955 * ipa-inline.c (edge_badness): Revert accidental checkin.
30957 2018-01-09 Jan Hubicka <hubicka@ucw.cz>
30960 * ipa-comdats.c (set_comdat_group): Only set comdat group of real
30961 symbols; not inline clones.
30963 2018-01-09 Jakub Jelinek <jakub@redhat.com>
30966 * modulo-sched.c (schedule_reg_moves): Punt if we'd need to move
30967 hard registers. Formatting fixes.
30969 PR preprocessor/83722
30970 * gcc.c (try_generate_repro): Pass
30971 &temp_stderr_files[RETRY_ICE_ATTEMPTS - 1] rather than
30972 &temp_stdout_files[RETRY_ICE_ATTEMPTS - 1] as last argument to
30975 2018-01-08 Monk Chiang <sh.chiang04@gmail.com>
30976 Kito Cheng <kito.cheng@gmail.com>
30978 * config/riscv/riscv.c (machine_function::is_leaf): Remove field.
30979 (riscv_leaf_function_p): Delete.
30980 (riscv_function_ok_for_sibcall): Return false when TARGET_SAVE_RESTORE.
30982 2018-01-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
30984 * config/rs6000/rs6000-string.c (do_load_for_compare_from_addr): New
30986 (do_ifelse): New function.
30987 (do_isel): New function.
30988 (do_sub3): New function.
30989 (do_add3): New function.
30990 (do_load_mask_compare): New function.
30991 (do_overlap_load_compare): New function.
30992 (expand_compare_loop): New function.
30993 (expand_block_compare): Call expand_compare_loop() when appropriate.
30994 * config/rs6000/rs6000.opt (-mblock-compare-inline-limit): Change
30995 option description.
30996 (-mblock-compare-inline-loop-limit): New option.
30998 2018-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31001 * config/rs6000/altivec.md (*altivec_vpermr_<mode>_internal):
31002 Reverse order of second and third operands in first alternative.
31003 * config/rs6000/rs6000.c (rs6000_expand_vector_set): Reverse order
31004 of first and second elements in UNSPEC_VPERMR vector.
31005 (altivec_expand_vec_perm_le): Likewise.
31007 2018-01-08 Jeff Law <law@redhat.com>
31009 PR rtl-optimizatin/81308
31010 * tree-switch-conversion.c (cfg_altered): New file scoped static.
31011 (process_switch): If group_case_labels makes a change, then set
31013 (pass_convert_switch::execute): If a switch is converted, then
31014 set cfg_altered. Return TODO_cfg_cleanup if cfg_altered is true.
31016 PR rtl-optimization/81308
31017 * recog.c (split_all_insns): Conditionally cleanup the CFG after
31020 2018-01-08 Vidya Praveen <vidyapraveen@arm.com>
31022 PR target/83663 - Revert r255946
31023 * config/aarch64/aarch64.c (aarch64_expand_vector_init): Modify code
31024 generation for cases where splatting a value is not useful.
31025 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
31026 across a vec_duplicate and a paradoxical subreg forming a vector
31027 mode to a vec_concat.
31029 2018-01-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31031 * config/arm/t-aprofile (MULTILIB_MATCHES): Add mapping rules for
31032 -march=armv8.3-a variants.
31033 * config/arm/t-multilib: Likewise.
31034 * config/arm/t-arm-elf: Likewise. Handle dotprod extension.
31036 2018-01-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
31038 * config/rs6000/rs6000.md (cceq_ior_compare): Remove * so I can use it
31040 (cceq_ior_compare_complement): Give it a name so I can use it, and
31041 change boolean_or_operator predicate to boolean_operator so it can
31042 be used to generate a crand.
31043 (eqne): New code iterator.
31044 (bd/bd_neg): New code_attrs.
31045 (<bd>_<mode>): New name for ctr<mode>_internal[12] now combined into
31046 a single define_insn.
31047 (<bd>tf_<mode>): A new insn pattern for the conditional form branch
31048 decrement (bdnzt/bdnzf/bdzt/bdzf).
31049 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Updated
31050 with the new names of the branch decrement patterns, and added the
31051 names of the branch decrement conditional patterns.
31053 2018-01-08 Richard Biener <rguenther@suse.de>
31055 PR tree-optimization/83563
31056 * graphite.c (canonicalize_loop_closed_ssa_form): Reset the SCEV
31059 2018-01-08 Richard Biener <rguenther@suse.de>
31061 PR middle-end/83713
31062 * convert.c (do_narrow): Properly guard TYPE_OVERFLOW_WRAPS checks.
31064 2018-01-08 Richard Biener <rguenther@suse.de>
31066 PR tree-optimization/83685
31067 * tree-ssa-pre.c (create_expression_by_pieces): Do not insert
31068 references to abnormals.
31070 2018-01-08 Richard Biener <rguenther@suse.de>
31073 * dwarf2out.c (output_indirect_strings): Handle empty
31074 skeleton_debug_str_hash.
31075 (dwarf2out_early_finish): Index strings for -gsplit-dwarf.
31077 2018-01-08 Claudiu Zissulescu <claziss@synopsys.com>
31079 * config/arc/arc.c (TARGET_TRAMPOLINE_ADJUST_ADDRESS): Delete.
31080 (emit_store_direct): Likewise.
31081 (arc_trampoline_adjust_address): Likewise.
31082 (arc_asm_trampoline_template): New function.
31083 (arc_initialize_trampoline): Use asm_trampoline_template.
31084 (TARGET_ASM_TRAMPOLINE_TEMPLATE): Define.
31085 * config/arc/arc.h (TRAMPOLINE_SIZE): Adjust to 16.
31086 * config/arc/arc.md (flush_icache): Delete pattern.
31088 2018-01-08 Claudiu Zissulescu <claziss@synopsys.com>
31090 * config/arc/arc-c.def (__ARC_UNALIGNED__): New define.
31091 * config/arc/arc.h (STRICT_ALIGNMENT): Control this macro using
31094 2018-01-08 Sebastian Huber <sebastian.huber@embedded-brains.de>
31097 * config/epiphany/epiphany.h (make_pass_mode_switch_use): Guard
31098 by not USED_FOR_TARGET.
31099 (make_pass_resolve_sw_modes): Likewise.
31101 2018-01-08 Sebastian Huber <sebastian.huber@embedded-brains.de>
31103 * config/nios2/nios2.h (nios2_section_threshold): Guard by not
31106 2018-01-08 Richard Biener <rguenther@suse.de>
31108 PR middle-end/83580
31109 * tree-data-ref.c (split_constant_offset): Remove STRIP_NOPS.
31111 2018-01-08 Richard Biener <rguenther@suse.de>
31113 PR middle-end/83517
31114 * match.pd ((t * 2) / 2) -> t): Add missing :c.
31116 2018-01-06 Aldy Hernandez <aldyh@redhat.com>
31118 PR middle-end/81897
31119 * tree-ssa-uninit.c (compute_control_dep_chain): Do not bail on
31120 basic blocks with a small number of successors.
31121 (convert_control_dep_chain_into_preds): Improve handling of
31123 (dump_predicates): Split apart into...
31124 (dump_pred_chain): ...here...
31125 (dump_pred_info): ...and here.
31126 (can_one_predicate_be_invalidated_p): Add debugging printfs.
31127 (can_chain_union_be_invalidated_p): Improve check for invalidation
31129 (uninit_uses_cannot_happen): Avoid unnecessary if
31130 convert_control_dep_chain_into_preds yielded nothing.
31132 2018-01-06 Martin Sebor <msebor@redhat.com>
31134 PR tree-optimization/83640
31135 * gimple-ssa-warn-restrict.c (builtin_access::builtin_access): Avoid
31136 subtracting negative offset from size.
31137 (builtin_access::overlap): Adjust offset bounds of the access to fall
31138 within the size of the object if possible.
31140 2018-01-06 Richard Sandiford <richard.sandiford@linaro.org>
31142 PR rtl-optimization/83699
31143 * expmed.c (extract_bit_field_1): Restrict the vector usage of
31144 extract_bit_field_as_subreg to cases in which the extracted
31145 value is also a vector.
31147 * lra-constraints.c (process_alt_operands): Test for the equivalence
31148 substitutions when detecting a possible reload cycle.
31150 2018-01-06 Jakub Jelinek <jakub@redhat.com>
31153 * toplev.c (process_options): Don't enable debug_nonbind_markers_p
31154 by default if flag_selective_schedling{,2}. Formatting fixes.
31156 PR rtl-optimization/83682
31157 * rtl.h (const_vec_duplicate_p): Only return true for VEC_DUPLICATE
31158 if it has non-VECTOR_MODE element mode.
31159 (vec_duplicate_p): Likewise.
31161 PR middle-end/83694
31162 * cfgexpand.c (expand_debug_expr): Punt if mode1 is VOIDmode
31163 and bitsize might be greater than MAX_BITSIZE_MODE_ANY_INT.
31165 2018-01-05 Jakub Jelinek <jakub@redhat.com>
31168 * config/i386/i386-builtin.def
31169 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
31170 __builtin_ia32_vgf2p8affineqb_v64qi, __builtin_ia32_vgf2p8mulb_v64qi):
31171 Require also OPTION_MASK_ISA_AVX512F in addition to
31172 OPTION_MASK_ISA_GFNI.
31173 (__builtin_ia32_vgf2p8affineinvqb_v16qi_mask,
31174 __builtin_ia32_vgf2p8affineqb_v16qi_mask): Require
31175 OPTION_MASK_ISA_AVX512VL instead of OPTION_MASK_ISA_SSE in addition
31176 to OPTION_MASK_ISA_GFNI.
31177 (__builtin_ia32_vgf2p8mulb_v32qi_mask): Require
31178 OPTION_MASK_ISA_AVX512VL in addition to OPTION_MASK_ISA_GFNI and
31179 OPTION_MASK_ISA_AVX512BW.
31180 (__builtin_ia32_vgf2p8mulb_v16qi_mask): Require
31181 OPTION_MASK_ISA_AVX512VL instead of OPTION_MASK_ISA_AVX512BW in
31182 addition to OPTION_MASK_ISA_GFNI.
31183 (__builtin_ia32_vgf2p8affineinvqb_v16qi,
31184 __builtin_ia32_vgf2p8affineqb_v16qi, __builtin_ia32_vgf2p8mulb_v16qi):
31185 Require OPTION_MASK_ISA_SSE2 instead of OPTION_MASK_ISA_SSE in addition
31186 to OPTION_MASK_ISA_GFNI.
31187 * config/i386/i386.c (def_builtin): Change to builtin isa/isa2 being
31188 a requirement for all ISAs rather than any of them with a few
31190 (ix86_add_new_builtins): Clear OPTION_MASK_ISA_64BIT from isa before
31192 (ix86_expand_builtin): Require all ISAs from builtin's isa and isa2
31193 bitmasks to be enabled with 3 exceptions, instead of requiring any
31194 enabled ISA with lots of exceptions.
31195 * config/i386/sse.md (vgf2p8affineinvqb_<mode><mask_name>,
31196 vgf2p8affineqb_<mode><mask_name>, vgf2p8mulb_<mode><mask_name>):
31197 Change avx512bw in isa attribute to avx512f.
31198 * config/i386/sgxintrin.h: Add license boilerplate.
31199 * config/i386/vaesintrin.h: Likewise. Fix macro spelling __AVX512F
31200 to __AVX512F__ and __AVX512VL to __AVX512VL__.
31201 (_mm256_aesdec_epi128, _mm256_aesdeclast_epi128, _mm256_aesenc_epi128,
31202 _mm256_aesenclast_epi128): Enable temporarily avx if __AVX__ is not
31204 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8,
31205 _mm_gf2p8affineinv_epi64_epi8, _mm_gf2p8affine_epi64_epi8): Enable
31206 temporarily sse2 rather than sse if not enabled already.
31209 * config/i386/sse.md (VI248_VLBW): Rename to ...
31210 (VI248_AVX512VL): ... this. Don't guard V32HI with TARGET_AVX512BW.
31211 (vpshrd_<mode><mask_name>, vpshld_<mode><mask_name>,
31212 vpshrdv_<mode>, vpshrdv_<mode>_mask, vpshrdv_<mode>_maskz,
31213 vpshrdv_<mode>_maskz_1, vpshldv_<mode>, vpshldv_<mode>_mask,
31214 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1): Use VI248_AVX512VL
31215 mode iterator instead of VI248_VLBW.
31217 2018-01-05 Jan Hubicka <hubicka@ucw.cz>
31219 * ipa-fnsummary.c (record_modified_bb_info): Add OP.
31220 (record_modified): Skip clobbers; add debug output.
31221 (param_change_prob): Use sreal frequencies.
31223 2018-01-05 Richard Sandiford <richard.sandiford@linaro.org>
31225 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
31226 punt for user-aligned variables.
31228 2018-01-05 Richard Sandiford <richard.sandiford@linaro.org>
31230 * tree-chrec.c (chrec_contains_symbols): Return true for
31233 2018-01-05 Sudakshina Das <sudi.das@arm.com>
31236 * simplify-rtx.c (simplify_relational_operation_1): Add simplifications
31237 of (x|y) == x for BICS pattern.
31239 2018-01-05 Jakub Jelinek <jakub@redhat.com>
31241 PR tree-optimization/83605
31242 * gimple-ssa-strength-reduction.c: Include tree-eh.h.
31243 (find_candidates_dom_walker::before_dom_children): Ignore stmts that
31246 2018-01-05 Sebastian Huber <sebastian.huber@embedded-brains.de>
31248 * config.gcc (epiphany-*-elf*): Add (epiphany-*-rtems*) configuration.
31249 * config/epiphany/rtems.h: New file.
31251 2018-01-04 Jakub Jelinek <jakub@redhat.com>
31252 Uros Bizjak <ubizjak@gmail.com>
31255 * config/i386/i386.md (*<rotate_insn>hi3_1 splitter): Use
31256 QIreg_operand instead of register_operand predicate.
31257 * config/i386/i386.c (ix86_rop_should_change_byte_p,
31258 set_rop_modrm_reg_bits, ix86_mitigate_rop): Use -mmitigate-rop in
31259 comments instead of -fmitigate[-_]rop.
31261 2018-01-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
31264 * cgraphunit.c (symbol_table::compile): Switch to text_section
31265 before calling assembly_start debug hook.
31266 * run-rtl-passes.c (run_rtl_passes): Likewise.
31269 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
31271 * tree-vrp.c (extract_range_from_binary_expr_1): Check
31272 range_int_cst_p rather than !symbolic_range_p before calling
31273 extract_range_from_multiplicative_op_1.
31275 2018-01-04 Jeff Law <law@redhat.com>
31277 * tree-ssa-math-opts.c (execute_cse_reciprocals_1): Remove
31278 redundant test in assertion.
31280 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
31282 * doc/rtl.texi: Document machine_mode wrapper classes.
31284 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
31286 * fold-const.c (fold_ternary_loc): Check tree_fits_uhwi_p before
31287 using tree_to_uhwi.
31289 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
31291 * tree-ssa-forwprop.c (is_combined_permutation_identity): Allow
31292 the VEC_PERM_EXPR fold to fail.
31294 2018-01-04 Jakub Jelinek <jakub@redhat.com>
31297 * bb-reorder.c (insert_section_boundary_note): Set has_bb_partition
31298 to switched_sections.
31300 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
31303 * config/arm/arm.c (arm_vectorize_vec_perm_const): Fix inverted
31304 test for d.testing.
31306 2018-01-04 Peter Bergner <bergner@vnet.ibm.com>
31309 * config/rs6000/rs6000.c (rs6000_discover_homogeneous_aggregate): Do not
31310 allow arguments in FP registers if TARGET_HARD_FLOAT is false.
31312 2018-01-04 Jakub Jelinek <jakub@redhat.com>
31315 * cfgexpand.c (expand_debug_expr) <case BIT_FIELD_REF>: Punt if mode
31316 is BLKmode and bitpos not zero or mode change is needed.
31318 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
31321 * config/sparc/sparc.c (sparc_vectorize_vec_perm_const): Require
31324 2018-01-04 Uros Bizjak <ubizjak@gmail.com>
31327 * config/alpha/alpha.md (*sadd<modesuffix>): Use ASHIFT
31328 instead of MULT rtx. Update all corresponding splitters.
31329 (*saddl_se): Ditto.
31330 (*ssub<modesuffix>): Ditto.
31331 (*ssubl_se): Ditto.
31332 (*cmp_sadd_di): Update split patterns.
31333 (*cmp_sadd_si): Ditto.
31334 (*cmp_sadd_sidi): Ditto.
31335 (*cmp_ssub_di): Ditto.
31336 (*cmp_ssub_si): Ditto.
31337 (*cmp_ssub_sidi): Ditto.
31338 * config/alpha/predicates.md (const23_operand): New predicate.
31339 * config/alpha/alpha.c (alpha_rtx_costs) [PLUS, MINUS]:
31340 Look for ASHIFT, not MULT inner operand.
31341 (alpha_split_conditional_move): Update for *sadd<modesuffix> change.
31343 2018-01-04 Martin Liska <mliska@suse.cz>
31345 PR gcov-profile/83669
31346 * gcov.c (output_intermediate_file): Add version to intermediate
31348 * doc/gcov.texi: Document new field 'version' in intermediate
31349 file format. Fix location of '-k' option of gcov command.
31351 2018-01-04 Martin Liska <mliska@suse.cz>
31354 * ipa-icf.c (sem_function::merge): Do not cross comdat boundary.
31356 2018-01-04 Jakub Jelinek <jakub@redhat.com>
31358 * gimple-ssa-sprintf.c (parse_directive): Cast second dir.len to uhwi.
31360 2018-01-03 Martin Sebor <msebor@redhat.com>
31362 PR tree-optimization/83655
31363 * gimple-ssa-warn-restrict.c (wrestrict_dom_walker::check_call): Avoid
31364 checking calls with invalid arguments.
31366 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31368 * tree-vect-stmts.c (vect_get_store_rhs): New function.
31369 (vectorizable_mask_load_store): Delete.
31370 (vectorizable_call): Return false for masked loads and stores.
31371 (vectorizable_store): Handle IFN_MASK_STORE. Use vect_get_store_rhs
31372 instead of gimple_assign_rhs1.
31373 (vectorizable_load): Handle IFN_MASK_LOAD.
31374 (vect_transform_stmt): Don't set is_store for call_vec_info_type.
31376 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31378 * tree-vect-stmts.c (vect_build_gather_load_calls): New function,
31380 (vectorizable_mask_load_store): ...here.
31381 (vectorizable_load): ...and here.
31383 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31385 * tree-vect-stmts.c (vect_build_all_ones_mask)
31386 (vect_build_zero_merge_argument): New functions, split out from...
31387 (vectorizable_load): ...here.
31389 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31391 * tree-vect-stmts.c (vect_check_store_rhs): New function,
31393 (vectorizable_mask_load_store): ...here.
31394 (vectorizable_store): ...and here.
31396 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31398 * tree-vect-stmts.c (vect_check_load_store_mask): New function,
31400 (vectorizable_mask_load_store): ...here.
31402 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31404 * tree-vectorizer.h (vec_load_store_type): Moved from tree-vec-stmts.c
31405 (vect_model_store_cost): Take a vec_load_store_type instead of a
31407 * tree-vect-stmts.c (vec_load_store_type): Move to tree-vectorizer.h.
31408 (vect_model_store_cost): Take a vec_load_store_type instead of a
31410 (vectorizable_mask_load_store): Update accordingly.
31411 (vectorizable_store): Likewise.
31412 * tree-vect-slp.c (vect_analyze_slp_cost_1): Update accordingly.
31414 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31416 * tree-vect-loop.c (vect_transform_loop): Stub out scalar
31417 IFN_MASK_LOAD calls here rather than...
31418 * tree-vect-stmts.c (vectorizable_mask_load_store): ...here.
31420 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31421 Alan Hayward <alan.hayward@arm.com>
31422 David Sherwood <david.sherwood@arm.com>
31424 * expmed.c (extract_bit_field_1): For vector extracts,
31425 fall back to extract_bit_field_as_subreg if vec_extract
31428 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31429 Alan Hayward <alan.hayward@arm.com>
31430 David Sherwood <david.sherwood@arm.com>
31432 * lra-spills.c (pseudo_reg_slot_compare): Sort slots by whether
31433 they are variable or constant sized.
31434 (assign_stack_slot_num_and_sort_pseudos): Don't reuse variable-sized
31435 slots for constant-sized data.
31437 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31438 Alan Hayward <alan.hayward@arm.com>
31439 David Sherwood <david.sherwood@arm.com>
31441 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern): When
31442 handling COND_EXPRs with boolean comparisons, try to find a better
31443 basis for the mask type than the boolean itself.
31445 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31447 * doc/rtl.texi (MAX_BITSIZE_MODE_ANY_MODE): Describe how the default
31448 is calculated and how it can be overridden.
31449 * genmodes.c (max_bitsize_mode_any_mode): New variable.
31450 (create_modes): Initialize it from MAX_BITSIZE_MODE_ANY_MODE,
31452 (emit_max_int): Use it to set the output MAX_BITSIZE_MODE_ANY_MODE,
31455 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31456 Alan Hayward <alan.hayward@arm.com>
31457 David Sherwood <david.sherwood@arm.com>
31459 * config/aarch64/aarch64-protos.h (aarch64_output_simd_mov_immediate):
31460 Remove the mode argument.
31461 (aarch64_simd_valid_immediate): Remove the mode and inverse
31463 * config/aarch64/iterators.md (bitsize): New iterator.
31464 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>, and<mode>3)
31465 (ior<mode>3): Update calls to aarch64_output_simd_mov_immediate.
31466 * config/aarch64/constraints.md (Do, Db, Dn): Update calls to
31467 aarch64_simd_valid_immediate.
31468 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Likewise.
31469 (aarch64_reg_or_bic_imm): Likewise.
31470 * config/aarch64/aarch64.c (simd_immediate_info): Replace mvn
31471 with an insn_type enum and msl with a modifier_type enum.
31472 Replace element_width with a scalar_mode. Change the shift
31473 to unsigned int. Add constructors for scalar_float_mode and
31474 scalar_int_mode elements.
31475 (aarch64_vect_float_const_representable_p): Delete.
31476 (aarch64_can_const_movi_rtx_p)
31477 (aarch64_simd_scalar_immediate_valid_for_move)
31478 (aarch64_simd_make_constant): Update call to
31479 aarch64_simd_valid_immediate.
31480 (aarch64_advsimd_valid_immediate_hs): New function.
31481 (aarch64_advsimd_valid_immediate): Likewise.
31482 (aarch64_simd_valid_immediate): Remove mode and inverse
31483 arguments. Rewrite to use the above. Use const_vec_duplicate_p
31484 to detect duplicated constants and use aarch64_float_const_zero_rtx_p
31485 and aarch64_float_const_representable_p on the result.
31486 (aarch64_output_simd_mov_immediate): Remove mode argument.
31487 Update call to aarch64_simd_valid_immediate and use of
31488 simd_immediate_info.
31489 (aarch64_output_scalar_simd_mov_immediate): Update call
31492 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31493 Alan Hayward <alan.hayward@arm.com>
31494 David Sherwood <david.sherwood@arm.com>
31496 * machmode.h (mode_precision): Prefix with CONST_MODE_PRECISION.
31497 (mode_nunits): Likewise CONST_MODE_NUNITS.
31498 * machmode.def (ADJUST_NUNITS): Document.
31499 * genmodes.c (mode_data::need_nunits_adj): New field.
31500 (blank_mode): Update accordingly.
31501 (adj_nunits): New variable.
31502 (print_maybe_const_decl): Replace CATEGORY with a NEEDS_ADJ
31504 (emit_mode_size_inline): Set need_bytesize_adj for all modes
31505 listed in adj_nunits.
31506 (emit_mode_nunits_inline): Set need_nunits_adj for all modes
31507 listed in adj_nunits. Don't emit case statements for such modes.
31508 (emit_insn_modes_h): Emit definitions of CONST_MODE_NUNITS
31509 and CONST_MODE_PRECISION. Make CONST_MODE_SIZE expand to
31510 nothing if adj_nunits is nonnull.
31511 (emit_mode_precision, emit_mode_nunits): Use print_maybe_const_decl.
31512 (emit_mode_unit_size, emit_mode_base_align, emit_mode_ibit)
31513 (emit_mode_fbit): Update use of print_maybe_const_decl.
31514 (emit_move_size): Likewise. Treat the array as non-const
31516 (emit_mode_adjustments): Handle adj_nunits.
31518 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31520 * machmode.def (VECTOR_MODES_WITH_PREFIX): Document.
31521 * genmodes.c (VECTOR_MODES_WITH_PREFIX): New macro.
31522 (VECTOR_MODES): Use it.
31523 (make_vector_modes): Take the prefix as an argument.
31525 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31526 Alan Hayward <alan.hayward@arm.com>
31527 David Sherwood <david.sherwood@arm.com>
31529 * mode-classes.def (MODE_VECTOR_BOOL): New mode class.
31530 * machmode.h (INTEGRAL_MODE_P, VECTOR_MODE_P): Return true
31531 for MODE_VECTOR_BOOL.
31532 * machmode.def (VECTOR_BOOL_MODE): Document.
31533 * genmodes.c (VECTOR_BOOL_MODE): New macro.
31534 (make_vector_bool_mode): New function.
31535 (complete_mode, emit_mode_wider, emit_mode_adjustments): Handle
31537 * lto-streamer-in.c (lto_input_mode_table): Likewise.
31538 * rtx-vector-builder.c (rtx_vector_builder::find_cached_value):
31540 * stor-layout.c (int_mode_for_mode): Likewise.
31541 * tree.c (build_vector_type_for_mode): Likewise.
31542 * varasm.c (output_constant_pool_2): Likewise.
31543 * emit-rtl.c (init_emit_once): Make sure that CONST1_RTX (BImode) and
31544 CONSTM1_RTX (BImode) are the same thing. Initialize const_tiny_rtx
31545 for MODE_VECTOR_BOOL.
31546 * expr.c (expand_expr_real_1): Use VECTOR_MODE_P instead of a list
31547 of mode class checks.
31548 * tree-vect-generic.c (expand_vector_operation): Use VECTOR_MODE_P
31549 instead of a list of mode class checks.
31550 (expand_vector_scalar_condition): Likewise.
31551 (type_for_widest_vector_mode): Handle BImode as an inner mode.
31553 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31554 Alan Hayward <alan.hayward@arm.com>
31555 David Sherwood <david.sherwood@arm.com>
31557 * machmode.h (mode_size): Change from unsigned short to
31559 (mode_to_bytes): Return a poly_uint16 rather than an unsigned short.
31560 (GET_MODE_SIZE): Return a constant if ONLY_FIXED_SIZE_MODES,
31561 or if measurement_type is not polynomial.
31562 (fixed_size_mode::includes_p): Check for constant-sized modes.
31563 * genmodes.c (emit_mode_size_inline): Make mode_size_inline
31564 return a poly_uint16 rather than an unsigned short.
31565 (emit_mode_size): Change the type of mode_size from unsigned short
31566 to poly_uint16_pod. Use ZERO_COEFFS for the initializer.
31567 (emit_mode_adjustments): Cope with polynomial vector sizes.
31568 * lto-streamer-in.c (lto_input_mode_table): Use bp_unpack_poly_value
31570 * lto-streamer-out.c (lto_write_mode_table): Use bp_pack_poly_value
31572 * auto-inc-dec.c (try_merge): Treat GET_MODE_SIZE as polynomial.
31573 * builtins.c (expand_ifn_atomic_compare_exchange_into_call): Likewise.
31574 * caller-save.c (setup_save_areas): Likewise.
31575 (replace_reg_with_saved_mem): Likewise.
31576 * calls.c (emit_library_call_value_1): Likewise.
31577 * combine-stack-adj.c (combine_stack_adjustments_for_block): Likewise.
31578 * combine.c (simplify_set, make_extraction, simplify_shift_const_1)
31579 (gen_lowpart_for_combine): Likewise.
31580 * convert.c (convert_to_integer_1): Likewise.
31581 * cse.c (equiv_constant, cse_insn): Likewise.
31582 * cselib.c (autoinc_split, cselib_hash_rtx): Likewise.
31583 (cselib_subst_to_values): Likewise.
31584 * dce.c (word_dce_process_block): Likewise.
31585 * df-problems.c (df_word_lr_mark_ref): Likewise.
31586 * dwarf2cfi.c (init_one_dwarf_reg_size): Likewise.
31587 * dwarf2out.c (multiple_reg_loc_descriptor, mem_loc_descriptor)
31588 (concat_loc_descriptor, concatn_loc_descriptor, loc_descriptor)
31589 (rtl_for_decl_location): Likewise.
31590 * emit-rtl.c (gen_highpart, widen_memory_access): Likewise.
31591 * expmed.c (extract_bit_field_1, extract_integral_bit_field): Likewise.
31592 * expr.c (emit_group_load_1, clear_storage_hints): Likewise.
31593 (emit_move_complex, emit_move_multi_word, emit_push_insn): Likewise.
31594 (expand_expr_real_1): Likewise.
31595 * function.c (assign_parm_setup_block_p, assign_parm_setup_block)
31596 (pad_below): Likewise.
31597 * gimple-fold.c (optimize_atomic_compare_exchange_p): Likewise.
31598 * gimple-ssa-store-merging.c (rhs_valid_for_store_merging_p): Likewise.
31599 * ira.c (get_subreg_tracking_sizes): Likewise.
31600 * ira-build.c (ira_create_allocno_objects): Likewise.
31601 * ira-color.c (coalesced_pseudo_reg_slot_compare): Likewise.
31602 (ira_sort_regnos_for_alter_reg): Likewise.
31603 * ira-costs.c (record_operand_costs): Likewise.
31604 * lower-subreg.c (interesting_mode_p, simplify_gen_subreg_concatn)
31605 (resolve_simple_move): Likewise.
31606 * lra-constraints.c (get_reload_reg, operands_match_p): Likewise.
31607 (process_addr_reg, simplify_operand_subreg, curr_insn_transform)
31608 (lra_constraints): Likewise.
31609 (CONST_POOL_OK_P): Reject variable-sized modes.
31610 * lra-spills.c (slot, assign_mem_slot, pseudo_reg_slot_compare)
31611 (add_pseudo_to_slot, lra_spill): Likewise.
31612 * omp-low.c (omp_clause_aligned_alignment): Likewise.
31613 * optabs-query.c (get_best_extraction_insn): Likewise.
31614 * optabs-tree.c (expand_vec_cond_expr_p): Likewise.
31615 * optabs.c (expand_vec_perm_var, expand_vec_cond_expr): Likewise.
31616 (expand_mult_highpart, valid_multiword_target_p): Likewise.
31617 * recog.c (offsettable_address_addr_space_p): Likewise.
31618 * regcprop.c (maybe_mode_change): Likewise.
31619 * reginfo.c (choose_hard_reg_mode, record_subregs_of_mode): Likewise.
31620 * regrename.c (build_def_use): Likewise.
31621 * regstat.c (dump_reg_info): Likewise.
31622 * reload.c (complex_word_subreg_p, push_reload, find_dummy_reload)
31623 (find_reloads, find_reloads_subreg_address): Likewise.
31624 * reload1.c (eliminate_regs_1): Likewise.
31625 * rtlanal.c (for_each_inc_dec_find_inc_dec, rtx_cost): Likewise.
31626 * simplify-rtx.c (avoid_constant_pool_reference): Likewise.
31627 (simplify_binary_operation_1, simplify_subreg): Likewise.
31628 * targhooks.c (default_function_arg_padding): Likewise.
31629 (default_hard_regno_nregs, default_class_max_nregs): Likewise.
31630 * tree-cfg.c (verify_gimple_assign_binary): Likewise.
31631 (verify_gimple_assign_ternary): Likewise.
31632 * tree-inline.c (estimate_move_cost): Likewise.
31633 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
31634 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Likewise.
31635 (get_address_cost_ainc): Likewise.
31636 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
31637 (vect_supportable_dr_alignment): Likewise.
31638 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
31639 (vectorizable_reduction): Likewise.
31640 * tree-vect-stmts.c (vectorizable_assignment, vectorizable_shift)
31641 (vectorizable_operation, vectorizable_load): Likewise.
31642 * tree.c (build_same_sized_truth_vector_type): Likewise.
31643 * valtrack.c (cleanup_auto_inc_dec): Likewise.
31644 * var-tracking.c (emit_note_insn_var_location): Likewise.
31645 * config/arc/arc.h (ASM_OUTPUT_CASE_END): Use as_a <scalar_int_mode>.
31646 (ADDR_VEC_ALIGN): Likewise.
31648 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31649 Alan Hayward <alan.hayward@arm.com>
31650 David Sherwood <david.sherwood@arm.com>
31652 * machmode.h (mode_to_bits): Return a poly_uint16 rather than an
31654 (GET_MODE_BITSIZE): Return a constant if ONLY_FIXED_SIZE_MODES,
31655 or if measurement_type is polynomial.
31656 * calls.c (shift_return_value): Treat GET_MODE_BITSIZE as polynomial.
31657 * combine.c (make_extraction): Likewise.
31658 * dse.c (find_shift_sequence): Likewise.
31659 * dwarf2out.c (mem_loc_descriptor): Likewise.
31660 * expmed.c (store_integral_bit_field, extract_bit_field_1): Likewise.
31661 (extract_bit_field, extract_low_bits): Likewise.
31662 * expr.c (convert_move, convert_modes, emit_move_insn_1): Likewise.
31663 (optimize_bitfield_assignment_op, expand_assignment): Likewise.
31664 (store_expr_with_bounds, store_field, expand_expr_real_1): Likewise.
31665 * fold-const.c (optimize_bit_field_compare, merge_ranges): Likewise.
31666 * gimple-fold.c (optimize_atomic_compare_exchange_p): Likewise.
31667 * reload.c (find_reloads): Likewise.
31668 * reload1.c (alter_reg): Likewise.
31669 * stor-layout.c (bitwise_mode_for_mode, compute_record_mode): Likewise.
31670 * targhooks.c (default_secondary_memory_needed_mode): Likewise.
31671 * tree-if-conv.c (predicate_mem_writes): Likewise.
31672 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
31673 * tree-vect-patterns.c (adjust_bool_pattern): Likewise.
31674 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
31675 * valtrack.c (dead_debug_insert_temp): Likewise.
31676 * varasm.c (mergeable_constant_section): Likewise.
31677 * config/sh/sh.h (LOCAL_ALIGNMENT): Use as_a <fixed_size_mode>.
31679 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31680 Alan Hayward <alan.hayward@arm.com>
31681 David Sherwood <david.sherwood@arm.com>
31683 * expr.c (expand_assignment): Cope with polynomial mode sizes
31684 when assigning to a CONCAT.
31686 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31687 Alan Hayward <alan.hayward@arm.com>
31688 David Sherwood <david.sherwood@arm.com>
31690 * machmode.h (mode_precision): Change from unsigned short to
31692 (mode_to_precision): Return a poly_uint16 rather than an unsigned
31694 (GET_MODE_PRECISION): Return a constant if ONLY_FIXED_SIZE_MODES,
31695 or if measurement_type is not polynomial.
31696 (HWI_COMPUTABLE_MODE_P): Turn into a function. Optimize the case
31697 in which the mode is already known to be a scalar_int_mode.
31698 * genmodes.c (emit_mode_precision): Change the type of mode_precision
31699 from unsigned short to poly_uint16_pod. Use ZERO_COEFFS for the
31701 * lto-streamer-in.c (lto_input_mode_table): Use bp_unpack_poly_value
31702 for GET_MODE_PRECISION.
31703 * lto-streamer-out.c (lto_write_mode_table): Use bp_pack_poly_value
31704 for GET_MODE_PRECISION.
31705 * combine.c (update_rsp_from_reg_equal): Treat GET_MODE_PRECISION
31707 (try_combine, find_split_point, combine_simplify_rtx): Likewise.
31708 (expand_field_assignment, make_extraction): Likewise.
31709 (make_compound_operation_int, record_dead_and_set_regs_1): Likewise.
31710 (get_last_value): Likewise.
31711 * convert.c (convert_to_integer_1): Likewise.
31712 * cse.c (cse_insn): Likewise.
31713 * expr.c (expand_expr_real_1): Likewise.
31714 * lra-constraints.c (simplify_operand_subreg): Likewise.
31715 * optabs-query.c (can_atomic_load_p): Likewise.
31716 * optabs.c (expand_atomic_load): Likewise.
31717 (expand_atomic_store): Likewise.
31718 * ree.c (combine_reaching_defs): Likewise.
31719 * rtl.h (partial_subreg_p, paradoxical_subreg_p): Likewise.
31720 * rtlanal.c (nonzero_bits1, lsb_bitfield_op_p): Likewise.
31721 * tree.h (type_has_mode_precision_p): Likewise.
31722 * ubsan.c (instrument_si_overflow): Likewise.
31724 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31725 Alan Hayward <alan.hayward@arm.com>
31726 David Sherwood <david.sherwood@arm.com>
31728 * tree.h (TYPE_VECTOR_SUBPARTS): Turn into a function and handle
31729 polynomial numbers of units.
31730 (SET_TYPE_VECTOR_SUBPARTS): Likewise.
31731 (valid_vector_subparts_p): New function.
31732 (build_vector_type): Remove temporary shim and take the number
31733 of units as a poly_uint64 rather than an int.
31734 (build_opaque_vector_type): Take the number of units as a
31735 poly_uint64 rather than an int.
31736 * tree.c (build_vector_from_ctor): Handle polynomial
31737 TYPE_VECTOR_SUBPARTS.
31738 (type_hash_canon_hash, type_cache_hasher::equal): Likewise.
31739 (uniform_vector_p, vector_type_mode, build_vector): Likewise.
31740 (build_vector_from_val): If the number of units is variable,
31741 use build_vec_duplicate_cst for constant operands and
31742 VEC_DUPLICATE_EXPR otherwise.
31743 (make_vector_type): Remove temporary is_constant ().
31744 (build_vector_type, build_opaque_vector_type): Take the number of
31745 units as a poly_uint64 rather than an int.
31746 (check_vector_cst): Handle polynomial TYPE_VECTOR_SUBPARTS and
31748 * cfgexpand.c (expand_debug_expr): Likewise.
31749 * expr.c (count_type_elements, categorize_ctor_elements_1): Likewise.
31750 (store_constructor, expand_expr_real_1): Likewise.
31751 (const_scalar_mask_from_tree): Likewise.
31752 * fold-const-call.c (fold_const_reduction): Likewise.
31753 * fold-const.c (const_binop, const_unop, fold_convert_const): Likewise.
31754 (operand_equal_p, fold_vec_perm, fold_ternary_loc): Likewise.
31755 (native_encode_vector, vec_cst_ctor_to_array): Likewise.
31756 (fold_relational_const): Likewise.
31757 (native_interpret_vector): Likewise. Change the size from an
31758 int to an unsigned int.
31759 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Handle polynomial
31760 TYPE_VECTOR_SUBPARTS.
31761 (gimple_fold_indirect_ref, gimple_build_vector): Likewise.
31762 (gimple_build_vector_from_val): Use VEC_DUPLICATE_EXPR when
31763 duplicating a non-constant operand into a variable-length vector.
31764 * hsa-brig.c (hsa_op_immed::emit_to_buffer): Handle polynomial
31765 TYPE_VECTOR_SUBPARTS and VECTOR_CST_NELTS.
31766 * ipa-icf.c (sem_variable::equals): Likewise.
31767 * match.pd: Likewise.
31768 * omp-simd-clone.c (simd_clone_subparts): Likewise.
31769 * print-tree.c (print_node): Likewise.
31770 * stor-layout.c (layout_type): Likewise.
31771 * targhooks.c (default_builtin_vectorization_cost): Likewise.
31772 * tree-cfg.c (verify_gimple_comparison): Likewise.
31773 (verify_gimple_assign_binary): Likewise.
31774 (verify_gimple_assign_ternary): Likewise.
31775 (verify_gimple_assign_single): Likewise.
31776 * tree-pretty-print.c (dump_generic_node): Likewise.
31777 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
31778 (simplify_bitfield_ref, is_combined_permutation_identity): Likewise.
31779 * tree-vect-data-refs.c (vect_permute_store_chain): Likewise.
31780 (vect_grouped_load_supported, vect_permute_load_chain): Likewise.
31781 (vect_shift_permute_load_chain): Likewise.
31782 * tree-vect-generic.c (nunits_for_known_piecewise_op): Likewise.
31783 (expand_vector_condition, optimize_vector_constructor): Likewise.
31784 (lower_vec_perm, get_compute_type): Likewise.
31785 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
31786 (get_initial_defs_for_reduction, vect_transform_loop): Likewise.
31787 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
31788 (vect_recog_mask_conversion_pattern): Likewise.
31789 * tree-vect-slp.c (vect_supported_load_permutation_p): Likewise.
31790 (vect_get_constant_vectors, vect_transform_slp_perm_load): Likewise.
31791 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
31792 (get_group_load_store_type, vectorizable_mask_load_store): Likewise.
31793 (vectorizable_bswap, simd_clone_subparts, vectorizable_assignment)
31794 (vectorizable_shift, vectorizable_operation, vectorizable_store)
31795 (vectorizable_load, vect_is_simple_cond, vectorizable_comparison)
31796 (supportable_widening_operation): Likewise.
31797 (supportable_narrowing_operation): Likewise.
31798 * tree-vector-builder.c (tree_vector_builder::binary_encoded_nelts):
31800 * varasm.c (output_constant): Likewise.
31802 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31803 Alan Hayward <alan.hayward@arm.com>
31804 David Sherwood <david.sherwood@arm.com>
31806 * tree-vect-data-refs.c (vect_permute_store_chain): Reorganize
31807 so that both the length == 3 and length != 3 cases set up their
31808 own permute vectors. Add comments explaining why we know the
31809 number of elements is constant.
31810 (vect_permute_load_chain): Likewise.
31812 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31813 Alan Hayward <alan.hayward@arm.com>
31814 David Sherwood <david.sherwood@arm.com>
31816 * machmode.h (mode_nunits): Change from unsigned char to
31818 (ONLY_FIXED_SIZE_MODES): New macro.
31819 (pod_mode::measurement_type, scalar_int_mode::measurement_type)
31820 (scalar_float_mode::measurement_type, scalar_mode::measurement_type)
31821 (complex_mode::measurement_type, fixed_size_mode::measurement_type):
31823 (mode_to_nunits): Return a poly_uint16 rather than an unsigned short.
31824 (GET_MODE_NUNITS): Return a constant if ONLY_FIXED_SIZE_MODES,
31825 or if measurement_type is not polynomial.
31826 * genmodes.c (ZERO_COEFFS): New macro.
31827 (emit_mode_nunits_inline): Make mode_nunits_inline return a
31829 (emit_mode_nunits): Change the type of mode_nunits to poly_uint16_pod.
31830 Use ZERO_COEFFS when emitting initializers.
31831 * data-streamer.h (bp_pack_poly_value): New function.
31832 (bp_unpack_poly_value): Likewise.
31833 * lto-streamer-in.c (lto_input_mode_table): Use bp_unpack_poly_value
31834 for GET_MODE_NUNITS.
31835 * lto-streamer-out.c (lto_write_mode_table): Use bp_pack_poly_value
31836 for GET_MODE_NUNITS.
31837 * tree.c (make_vector_type): Remove temporary shim and make
31838 the real function take the number of units as a poly_uint64
31839 rather than an int.
31840 (build_vector_type_for_mode): Handle polynomial nunits.
31841 * dwarf2out.c (loc_descriptor, add_const_value_attribute): Likewise.
31842 * emit-rtl.c (const_vec_series_p_1): Likewise.
31843 (gen_rtx_CONST_VECTOR): Likewise.
31844 * fold-const.c (test_vec_duplicate_folding): Likewise.
31845 * genrecog.c (validate_pattern): Likewise.
31846 * optabs-query.c (can_vec_perm_var_p, can_mult_highpart_p): Likewise.
31847 * optabs-tree.c (expand_vec_cond_expr_p): Likewise.
31848 * optabs.c (expand_vector_broadcast, expand_binop_directly): Likewise.
31849 (shift_amt_for_vec_perm_mask, expand_vec_perm_var): Likewise.
31850 (expand_vec_cond_expr, expand_mult_highpart): Likewise.
31851 * rtlanal.c (subreg_get_info): Likewise.
31852 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
31853 (vect_grouped_load_supported): Likewise.
31854 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
31855 * tree-vect-loop.c (have_whole_vector_shift): Likewise.
31856 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
31857 (simplify_const_unary_operation, simplify_binary_operation_1)
31858 (simplify_const_binary_operation, simplify_ternary_operation)
31859 (test_vector_ops_duplicate, test_vector_ops): Likewise.
31860 (simplify_immed_subreg): Use GET_MODE_NUNITS on a fixed_size_mode
31861 instead of CONST_VECTOR_NUNITS.
31862 * varasm.c (output_constant_pool_2): Likewise.
31863 * rtx-vector-builder.c (rtx_vector_builder::build): Only include the
31864 explicit-encoded elements in the XVEC for variable-length vectors.
31866 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31868 * lra-constraints.c (curr_insn_transform): Use partial_subreg_p.
31870 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31871 Alan Hayward <alan.hayward@arm.com>
31872 David Sherwood <david.sherwood@arm.com>
31874 * coretypes.h (fixed_size_mode): Declare.
31875 (fixed_size_mode_pod): New typedef.
31876 * builtins.h (target_builtins::x_apply_args_mode)
31877 (target_builtins::x_apply_result_mode): Change type to
31878 fixed_size_mode_pod.
31879 * builtins.c (apply_args_size, apply_result_size, result_vector)
31880 (expand_builtin_apply_args_1, expand_builtin_apply)
31881 (expand_builtin_return): Update accordingly.
31883 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31885 * cse.c (hash_rtx_cb): Hash only the encoded elements.
31886 * cselib.c (cselib_hash_rtx): Likewise.
31887 * expmed.c (make_tree): Build VECTOR_CSTs directly from the
31888 CONST_VECTOR encoding.
31890 2018-01-03 Jakub Jelinek <jakub@redhat.com>
31891 Jeff Law <law@redhat.com>
31894 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): For
31895 noreturn probe, use gen_pop instead of ix86_emit_restore_reg_using_pop,
31896 only set RTX_FRAME_RELATED_P on both the push and pop if cfa_reg is sp
31897 and add REG_CFA_ADJUST_CFA notes in that case to both insns.
31900 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Do not
31901 explicitly probe *sp in a noreturn function if there were any callee
31902 register saves or frame pointer is needed.
31904 2018-01-03 Jakub Jelinek <jakub@redhat.com>
31907 * cfgexpand.c (expand_debug_expr): Return NULL if mode is
31908 BLKmode for ternary, binary or unary expressions.
31911 * var-tracking.c (delete_vta_debug_insn): New inline function.
31912 (delete_vta_debug_insns): Add USE_CFG argument, if true, walk just
31913 insns from get_insns () to NULL instead of each bb separately.
31914 Use delete_vta_debug_insn. No longer static.
31915 (vt_debug_insns_local, variable_tracking_main_1): Adjust
31916 delete_vta_debug_insns callers.
31917 * rtl.h (delete_vta_debug_insns): Declare.
31918 * final.c (rest_of_handle_final): Call delete_vta_debug_insns
31919 instead of variable_tracking_main.
31921 2018-01-03 Martin Sebor <msebor@redhat.com>
31923 PR tree-optimization/83603
31924 * calls.c (maybe_warn_nonstring_arg): Avoid accessing function
31925 arguments past the endof the argument list in functions declared
31926 without a prototype.
31927 * gimple-ssa-warn-restrict.c (wrestrict_dom_walker::check_call):
31928 Avoid checking when arguments are null.
31930 2018-01-03 Martin Sebor <msebor@redhat.com>
31933 * doc/extend.texi (attribute const): Fix a typo.
31934 * ipa-pure-const.c ((warn_function_const, warn_function_pure): Avoid
31935 issuing -Wsuggest-attribute for void functions.
31937 2018-01-03 Martin Sebor <msebor@redhat.com>
31939 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Use
31940 offset_int::from instead of wide_int::to_shwi.
31941 (maybe_diag_overlap): Remove assertion.
31942 Use HOST_WIDE_INT_PRINT_DEC instead of %lli.
31943 * gimple-ssa-sprintf.c (format_directive): Same.
31944 (parse_directive): Same.
31945 (sprintf_dom_walker::compute_format_length): Same.
31946 (try_substitute_return_value): Same.
31948 2018-01-03 Jeff Law <law@redhat.com>
31950 PR middle-end/83654
31951 * explow.c (anti_adjust_stack_and_probe_stack_clash): Test a
31952 non-constant residual for zero at runtime and avoid probing in
31953 that case. Reorganize code for trailing problem to mirror handling
31956 2018-01-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
31958 PR tree-optimization/83501
31959 * tree-ssa-strlen.c (get_string_cst): New.
31960 (handle_char_store): Call get_string_cst.
31962 2018-01-03 Martin Liska <mliska@suse.cz>
31964 PR tree-optimization/83593
31965 * tree-ssa-strlen.c: Include tree-cfg.h.
31966 (strlen_check_and_optimize_stmt): Add new argument cleanup_eh.
31967 (strlen_dom_walker): Add new member variable m_cleanup_cfg.
31968 (strlen_dom_walker::strlen_dom_walker): Initialize m_cleanup_cfg
31970 (strlen_dom_walker::before_dom_children): Call
31971 gimple_purge_dead_eh_edges. Dump tranformation with details
31973 (strlen_dom_walker::before_dom_children): Update call by adding
31974 new argument cleanup_eh.
31975 (pass_strlen::execute): Return TODO_cleanup_cfg if needed.
31977 2018-01-03 Martin Liska <mliska@suse.cz>
31980 * cif-code.def (VARIADIC_THUNK): New enum value.
31981 * ipa-fnsummary.c (compute_fn_summary): Do not inline variadic
31984 2018-01-03 Jan Beulich <jbeulich@suse.com>
31986 * sse.md (mov<mode>_internal): Tighten condition for when to use
31987 vmovdqu<ssescalarsize> for TI and OI modes.
31989 2018-01-03 Jakub Jelinek <jakub@redhat.com>
31991 Update copyright years.
31993 2018-01-03 Martin Liska <mliska@suse.cz>
31996 * ipa-visibility.c (function_and_variable_visibility): Skip
31997 functions with noipa attribure.
31999 2018-01-03 Jakub Jelinek <jakub@redhat.com>
32001 * gcc.c (process_command): Update copyright notice dates.
32002 * gcov-dump.c (print_version): Ditto.
32003 * gcov.c (print_version): Ditto.
32004 * gcov-tool.c (print_version): Ditto.
32005 * gengtype.c (create_file): Ditto.
32006 * doc/cpp.texi: Bump @copying's copyright year.
32007 * doc/cppinternals.texi: Ditto.
32008 * doc/gcc.texi: Ditto.
32009 * doc/gccint.texi: Ditto.
32010 * doc/gcov.texi: Ditto.
32011 * doc/install.texi: Ditto.
32012 * doc/invoke.texi: Ditto.
32014 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32016 * vector-builder.h (vector_builder::m_full_nelts): Change from
32017 unsigned int to poly_uint64.
32018 (vector_builder::full_nelts): Update prototype accordingly.
32019 (vector_builder::new_vector): Likewise.
32020 (vector_builder::encoded_full_vector_p): Handle polynomial full_nelts.
32021 (vector_builder::operator ==): Likewise.
32022 (vector_builder::finalize): Likewise.
32023 * int-vector-builder.h (int_vector_builder::int_vector_builder):
32024 Take the number of elements as a poly_uint64 rather than an
32026 * vec-perm-indices.h (vec_perm_indices::m_nelts_per_input): Change
32027 from unsigned int to poly_uint64.
32028 (vec_perm_indices::vec_perm_indices): Update prototype accordingly.
32029 (vec_perm_indices::new_vector): Likewise.
32030 (vec_perm_indices::length): Likewise.
32031 (vec_perm_indices::nelts_per_input): Likewise.
32032 (vec_perm_indices::input_nelts): Likewise.
32033 * vec-perm-indices.c (vec_perm_indices::new_vector): Take the
32034 number of elements per input as a poly_uint64 rather than an
32035 unsigned int. Use the original encoding for variable-length
32036 vectors, rather than clamping each individual element.
32037 For the second and subsequent elements in each pattern,
32038 clamp the step and base before clamping their sum.
32039 (vec_perm_indices::series_p): Handle polynomial element counts.
32040 (vec_perm_indices::all_in_range_p): Likewise.
32041 (vec_perm_indices_to_tree): Likewise.
32042 (vec_perm_indices_to_rtx): Likewise.
32043 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
32044 * tree-vector-builder.c (tree_vector_builder::new_unary_operation)
32045 (tree_vector_builder::new_binary_operation): Handle polynomial
32046 element counts. Return false if we need to know the number
32047 of elements at compile time.
32048 * fold-const.c (fold_vec_perm): Punt if the number of elements
32049 isn't known at compile time.
32051 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32053 * vec-perm-indices.h (vec_perm_builder): Change element type
32054 from HOST_WIDE_INT to poly_int64.
32055 (vec_perm_indices::element_type): Update accordingly.
32056 (vec_perm_indices::clamp): Handle polynomial element_types.
32057 * vec-perm-indices.c (vec_perm_indices::series_p): Likewise.
32058 (vec_perm_indices::all_in_range_p): Likewise.
32059 (tree_to_vec_perm_builder): Check for poly_int64 trees rather
32061 * vector-builder.h (vector_builder::stepped_sequence_p): Handle
32062 polynomial vec_perm_indices element types.
32063 * int-vector-builder.h (int_vector_builder::equal_p): Likewise.
32064 * fold-const.c (fold_vec_perm): Likewise.
32065 * optabs.c (shift_amt_for_vec_perm_mask): Likewise.
32066 * tree-vect-generic.c (lower_vec_perm): Likewise.
32067 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
32068 * config/aarch64/aarch64.c (aarch64_evpc_tbl): Cast d->perm
32069 element type to HOST_WIDE_INT.
32071 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32072 Alan Hayward <alan.hayward@arm.com>
32073 David Sherwood <david.sherwood@arm.com>
32075 * alias.c (addr_side_effect_eval): Take the size as a poly_int64
32076 rather than an int. Use plus_constant.
32077 (memrefs_conflict_p): Take the sizes as poly_int64s rather than ints.
32078 Take the offset "c" as a poly_int64 rather than a HOST_WIDE_INT.
32080 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32081 Alan Hayward <alan.hayward@arm.com>
32082 David Sherwood <david.sherwood@arm.com>
32084 * calls.c (emit_call_1, expand_call): Change struct_value_size from
32085 a HOST_WIDE_INT to a poly_int64.
32087 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32088 Alan Hayward <alan.hayward@arm.com>
32089 David Sherwood <david.sherwood@arm.com>
32091 * calls.c (load_register_parameters): Cope with polynomial
32092 mode sizes. Require a constant size for BLKmode parameters
32093 that aren't described by a PARALLEL. If BLOCK_REG_PADDING
32094 forces a parameter to be padded at the lsb end in order to
32095 fill a complete number of words, require the parameter size
32096 to be ordered wrt UNITS_PER_WORD.
32098 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32099 Alan Hayward <alan.hayward@arm.com>
32100 David Sherwood <david.sherwood@arm.com>
32102 * reload1.c (spill_stack_slot_width): Change element type
32103 from unsigned int to poly_uint64_pod.
32104 (alter_reg): Treat mode sizes as polynomial.
32106 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32107 Alan Hayward <alan.hayward@arm.com>
32108 David Sherwood <david.sherwood@arm.com>
32110 * reload.c (complex_word_subreg_p): New function.
32111 (reload_inner_reg_of_subreg, push_reload): Use it.
32113 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32114 Alan Hayward <alan.hayward@arm.com>
32115 David Sherwood <david.sherwood@arm.com>
32117 * lra-constraints.c (process_alt_operands): Reject matched
32118 operands whose sizes aren't ordered.
32119 (match_reload): Refer to this check here.
32121 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32122 Alan Hayward <alan.hayward@arm.com>
32123 David Sherwood <david.sherwood@arm.com>
32125 * builtins.c (expand_ifn_atomic_compare_exchange_into_call): Assert
32126 that the mode size is in the set {1, 2, 4, 8, 16}.
32128 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32129 Alan Hayward <alan.hayward@arm.com>
32130 David Sherwood <david.sherwood@arm.com>
32132 * var-tracking.c (adjust_mems): Treat mode sizes as polynomial.
32133 Use plus_constant instead of gen_rtx_PLUS.
32135 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32136 Alan Hayward <alan.hayward@arm.com>
32137 David Sherwood <david.sherwood@arm.com>
32139 * config/cr16/cr16-protos.h (cr16_push_rounding): Declare.
32140 * config/cr16/cr16.h (PUSH_ROUNDING): Move implementation to...
32141 * config/cr16/cr16.c (cr16_push_rounding): ...this new function.
32142 * config/h8300/h8300-protos.h (h8300_push_rounding): Declare.
32143 * config/h8300/h8300.h (PUSH_ROUNDING): Move implementation to...
32144 * config/h8300/h8300.c (h8300_push_rounding): ...this new function.
32145 * config/i386/i386-protos.h (ix86_push_rounding): Declare.
32146 * config/i386/i386.h (PUSH_ROUNDING): Move implementation to...
32147 * config/i386/i386.c (ix86_push_rounding): ...this new function.
32148 * config/m32c/m32c-protos.h (m32c_push_rounding): Take and return
32150 * config/m32c/m32c.c (m32c_push_rounding): Likewise.
32151 * config/m68k/m68k-protos.h (m68k_push_rounding): Declare.
32152 * config/m68k/m68k.h (PUSH_ROUNDING): Move implementation to...
32153 * config/m68k/m68k.c (m68k_push_rounding): ...this new function.
32154 * config/pdp11/pdp11-protos.h (pdp11_push_rounding): Declare.
32155 * config/pdp11/pdp11.h (PUSH_ROUNDING): Move implementation to...
32156 * config/pdp11/pdp11.c (pdp11_push_rounding): ...this new function.
32157 * config/stormy16/stormy16-protos.h (xstormy16_push_rounding): Declare.
32158 * config/stormy16/stormy16.h (PUSH_ROUNDING): Move implementation to...
32159 * config/stormy16/stormy16.c (xstormy16_push_rounding): ...this new
32161 * expr.c (emit_move_resolve_push): Treat the input and result
32162 of PUSH_ROUNDING as a poly_int64.
32163 (emit_move_complex_push, emit_single_push_insn_1): Likewise.
32164 (emit_push_insn): Likewise.
32165 * lra-eliminations.c (mark_not_eliminable): Likewise.
32166 * recog.c (push_operand): Likewise.
32167 * reload1.c (elimination_effects): Likewise.
32168 * rtlanal.c (nonzero_bits1): Likewise.
32169 * calls.c (store_one_arg): Likewise. Require the padding to be
32170 known at compile time.
32172 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32173 Alan Hayward <alan.hayward@arm.com>
32174 David Sherwood <david.sherwood@arm.com>
32176 * expr.c (emit_single_push_insn_1): Treat mode sizes as polynomial.
32177 Use plus_constant instead of gen_rtx_PLUS.
32179 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32180 Alan Hayward <alan.hayward@arm.com>
32181 David Sherwood <david.sherwood@arm.com>
32183 * auto-inc-dec.c (set_inc_state): Take the mode size as a poly_int64
32184 rather than an int.
32186 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32187 Alan Hayward <alan.hayward@arm.com>
32188 David Sherwood <david.sherwood@arm.com>
32190 * expr.c (expand_expr_real_1): Use tree_to_poly_uint64
32191 instead of int_size_in_bytes when handling VIEW_CONVERT_EXPRs
32192 via stack temporaries. Treat the mode size as polynomial too.
32194 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32195 Alan Hayward <alan.hayward@arm.com>
32196 David Sherwood <david.sherwood@arm.com>
32198 * expr.c (expand_expr_real_2): When handling conversions involving
32199 unions, apply tree_to_poly_uint64 to the TYPE_SIZE rather than
32200 multiplying int_size_in_bytes by BITS_PER_UNIT. Treat GET_MODE_BISIZE
32201 as a poly_uint64 too.
32203 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32204 Alan Hayward <alan.hayward@arm.com>
32205 David Sherwood <david.sherwood@arm.com>
32207 * rtlanal.c (subreg_get_info): Handle polynomial mode sizes.
32209 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32210 Alan Hayward <alan.hayward@arm.com>
32211 David Sherwood <david.sherwood@arm.com>
32213 * combine.c (can_change_dest_mode): Handle polynomial
32214 REGMODE_NATURAL_SIZE.
32215 * expmed.c (store_bit_field_1): Likewise.
32216 * expr.c (store_constructor): Likewise.
32217 * emit-rtl.c (validate_subreg): Operate on polynomial mode sizes
32218 and polynomial REGMODE_NATURAL_SIZE.
32219 (gen_lowpart_common): Likewise.
32220 * reginfo.c (record_subregs_of_mode): Likewise.
32221 * rtlanal.c (read_modify_subreg_p): Likewise.
32223 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32224 Alan Hayward <alan.hayward@arm.com>
32225 David Sherwood <david.sherwood@arm.com>
32227 * internal-fn.c (expand_vector_ubsan_overflow): Handle polynomial
32228 numbers of elements.
32230 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32231 Alan Hayward <alan.hayward@arm.com>
32232 David Sherwood <david.sherwood@arm.com>
32234 * match.pd: Cope with polynomial numbers of vector elements.
32236 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32237 Alan Hayward <alan.hayward@arm.com>
32238 David Sherwood <david.sherwood@arm.com>
32240 * fold-const.c (fold_indirect_ref_1): Handle polynomial offsets
32241 in a POINTER_PLUS_EXPR.
32243 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32244 Alan Hayward <alan.hayward@arm.com>
32245 David Sherwood <david.sherwood@arm.com>
32247 * omp-simd-clone.c (simd_clone_subparts): New function.
32248 (simd_clone_init_simd_arrays): Use it instead of TYPE_VECTOR_SUBPARTS.
32249 (ipa_simd_modify_function_body): Likewise.
32251 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32252 Alan Hayward <alan.hayward@arm.com>
32253 David Sherwood <david.sherwood@arm.com>
32255 * tree-vect-generic.c (nunits_for_known_piecewise_op): New function.
32256 (expand_vector_piecewise): Use it instead of TYPE_VECTOR_SUBPARTS.
32257 (expand_vector_addition, add_rshift, expand_vector_divmod): Likewise.
32258 (expand_vector_condition, vector_element): Likewise.
32259 (subparts_gt): New function.
32260 (get_compute_type): Use subparts_gt.
32261 (count_type_subparts): Delete.
32262 (expand_vector_operations_1): Use subparts_gt instead of
32263 count_type_subparts.
32265 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32266 Alan Hayward <alan.hayward@arm.com>
32267 David Sherwood <david.sherwood@arm.com>
32269 * tree-vect-data-refs.c (vect_no_alias_p): Replace with...
32270 (vect_compile_time_alias): ...this new function. Do the calculation
32271 on poly_ints rather than trees.
32272 (vect_prune_runtime_alias_test_list): Update call accordingly.
32274 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32275 Alan Hayward <alan.hayward@arm.com>
32276 David Sherwood <david.sherwood@arm.com>
32278 * tree-vect-slp.c (vect_build_slp_tree_1): Handle polynomial
32280 (vect_schedule_slp_instance): Likewise.
32282 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32283 Alan Hayward <alan.hayward@arm.com>
32284 David Sherwood <david.sherwood@arm.com>
32286 * tree-vect-slp.c (vect_get_and_check_slp_defs): Reject
32287 constant and extern definitions for variable-length vectors.
32288 (vect_get_constant_vectors): Note that the number of units
32289 is known to be constant.
32291 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32292 Alan Hayward <alan.hayward@arm.com>
32293 David Sherwood <david.sherwood@arm.com>
32295 * tree-vect-stmts.c (vectorizable_conversion): Treat the number
32296 of units as polynomial. Choose between WIDE and NARROW based
32299 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32300 Alan Hayward <alan.hayward@arm.com>
32301 David Sherwood <david.sherwood@arm.com>
32303 * tree-vect-stmts.c (simd_clone_subparts): New function.
32304 (vectorizable_simd_clone_call): Use it instead of TYPE_VECTOR_SUBPARTS.
32306 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32307 Alan Hayward <alan.hayward@arm.com>
32308 David Sherwood <david.sherwood@arm.com>
32310 * tree-vect-stmts.c (vectorizable_call): Treat the number of
32311 vectors as polynomial. Use build_index_vector for
32312 IFN_GOMP_SIMD_LANE.
32314 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32315 Alan Hayward <alan.hayward@arm.com>
32316 David Sherwood <david.sherwood@arm.com>
32318 * tree-vect-stmts.c (get_load_store_type): Treat the number of
32319 units as polynomial. Reject VMAT_ELEMENTWISE and VMAT_STRIDED_SLP
32320 for variable-length vectors.
32321 (vectorizable_mask_load_store): Treat the number of units as
32322 polynomial, asserting that it is constant if the condition has
32323 already been enforced.
32324 (vectorizable_store, vectorizable_load): Likewise.
32326 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32327 Alan Hayward <alan.hayward@arm.com>
32328 David Sherwood <david.sherwood@arm.com>
32330 * tree-vect-loop.c (vectorizable_live_operation): Treat the number
32331 of units as polynomial. Punt if we can't tell at compile time
32332 which vector contains the final result.
32334 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32335 Alan Hayward <alan.hayward@arm.com>
32336 David Sherwood <david.sherwood@arm.com>
32338 * tree-vect-loop.c (vectorizable_induction): Treat the number
32339 of units as polynomial. Punt on SLP inductions. Use an integer
32340 VEC_SERIES_EXPR for variable-length integer reductions. Use a
32341 cast of such a series for variable-length floating-point
32344 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32345 Alan Hayward <alan.hayward@arm.com>
32346 David Sherwood <david.sherwood@arm.com>
32348 * tree.h (build_index_vector): Declare.
32349 * tree.c (build_index_vector): New function.
32350 * tree-vect-loop.c (get_initial_defs_for_reduction): Treat the number
32351 of units as polynomial, forcibly converting it to a constant if
32352 vectorizable_reduction has already enforced the condition.
32353 (vect_create_epilog_for_reduction): Likewise. Use build_index_vector
32354 to create a {1,2,3,...} vector.
32355 (vectorizable_reduction): Treat the number of units as polynomial.
32356 Choose vectype_in based on the largest scalar element size rather
32357 than the smallest number of units. Enforce the restrictions
32360 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32361 Alan Hayward <alan.hayward@arm.com>
32362 David Sherwood <david.sherwood@arm.com>
32364 * tree-vect-data-refs.c (vector_alignment_reachable_p): Treat the
32365 number of units as polynomial.
32367 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32368 Alan Hayward <alan.hayward@arm.com>
32369 David Sherwood <david.sherwood@arm.com>
32371 * target.h (vector_sizes, auto_vector_sizes): New typedefs.
32372 * target.def (autovectorize_vector_sizes): Return the vector sizes
32373 by pointer, using vector_sizes rather than a bitmask.
32374 * targhooks.h (default_autovectorize_vector_sizes): Update accordingly.
32375 * targhooks.c (default_autovectorize_vector_sizes): Likewise.
32376 * config/aarch64/aarch64.c (aarch64_autovectorize_vector_sizes):
32378 * config/arc/arc.c (arc_autovectorize_vector_sizes): Likewise.
32379 * config/arm/arm.c (arm_autovectorize_vector_sizes): Likewise.
32380 * config/i386/i386.c (ix86_autovectorize_vector_sizes): Likewise.
32381 * config/mips/mips.c (mips_autovectorize_vector_sizes): Likewise.
32382 * omp-general.c (omp_max_vf): Likewise.
32383 * omp-low.c (omp_clause_aligned_alignment): Likewise.
32384 * optabs-query.c (can_vec_mask_load_store_p): Likewise.
32385 * tree-vect-loop.c (vect_analyze_loop): Likewise.
32386 * tree-vect-slp.c (vect_slp_bb): Likewise.
32387 * doc/tm.texi: Regenerate.
32388 * tree-vectorizer.h (current_vector_size): Change from an unsigned int
32390 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Take
32391 the vector size as a poly_uint64 rather than an unsigned int.
32392 (current_vector_size): Change from an unsigned int to a poly_uint64.
32393 (get_vectype_for_scalar_type): Update accordingly.
32394 * tree.h (build_truth_vector_type): Take the size and number of
32395 units as a poly_uint64 rather than an unsigned int.
32396 (build_vector_type): Add a temporary overload that takes
32397 the number of units as a poly_uint64 rather than an unsigned int.
32398 * tree.c (make_vector_type): Likewise.
32399 (build_truth_vector_type): Take the number of units as a poly_uint64
32400 rather than an unsigned int.
32402 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32403 Alan Hayward <alan.hayward@arm.com>
32404 David Sherwood <david.sherwood@arm.com>
32406 * target.def (get_mask_mode): Take the number of units and length
32407 as poly_uint64s rather than unsigned ints.
32408 * targhooks.h (default_get_mask_mode): Update accordingly.
32409 * targhooks.c (default_get_mask_mode): Likewise.
32410 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
32411 * doc/tm.texi: Regenerate.
32413 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32414 Alan Hayward <alan.hayward@arm.com>
32415 David Sherwood <david.sherwood@arm.com>
32417 * omp-general.h (omp_max_vf): Return a poly_uint64 instead of an int.
32418 * omp-general.c (omp_max_vf): Likewise.
32419 * omp-expand.c (omp_adjust_chunk_size): Update call to omp_max_vf.
32420 (expand_omp_simd): Handle polynomial safelen.
32421 * omp-low.c (omplow_simd_context): Add a default constructor.
32422 (omplow_simd_context::max_vf): Change from int to poly_uint64.
32423 (lower_rec_simd_input_clauses): Update accordingly.
32424 (lower_rec_input_clauses): Likewise.
32426 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32427 Alan Hayward <alan.hayward@arm.com>
32428 David Sherwood <david.sherwood@arm.com>
32430 * tree-vectorizer.h (vect_nunits_for_cost): New function.
32431 * tree-vect-loop.c (vect_model_reduction_cost): Use it.
32432 * tree-vect-slp.c (vect_analyze_slp_cost_1): Likewise.
32433 (vect_analyze_slp_cost): Likewise.
32434 * tree-vect-stmts.c (vect_model_store_cost): Likewise.
32435 (vect_model_load_cost): Likewise.
32437 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32438 Alan Hayward <alan.hayward@arm.com>
32439 David Sherwood <david.sherwood@arm.com>
32441 * tree-vect-slp.c (vect_record_max_nunits, vect_build_slp_tree_1)
32442 (vect_build_slp_tree_2, vect_build_slp_tree): Change max_nunits
32443 from an unsigned int * to a poly_uint64_pod *.
32444 (calculate_unrolling_factor): New function.
32445 (vect_analyze_slp_instance): Use it. Track polynomial max_nunits.
32447 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32448 Alan Hayward <alan.hayward@arm.com>
32449 David Sherwood <david.sherwood@arm.com>
32451 * tree-vectorizer.h (_slp_instance::unrolling_factor): Change
32452 from an unsigned int to a poly_uint64.
32453 (_loop_vec_info::slp_unrolling_factor): Likewise.
32454 (_loop_vec_info::vectorization_factor): Change from an int
32456 (MAX_VECTORIZATION_FACTOR): Bump from 64 to INT_MAX.
32457 (vect_get_num_vectors): New function.
32458 (vect_update_max_nunits, vect_vf_for_cost): Likewise.
32459 (vect_get_num_copies): Use vect_get_num_vectors.
32460 (vect_analyze_data_ref_dependences): Change max_vf from an int *
32461 to an unsigned int *.
32462 (vect_analyze_data_refs): Change min_vf from an int * to a
32464 (vect_transform_slp_perm_load): Take the vf as a poly_uint64 rather
32465 than an unsigned HOST_WIDE_INT.
32466 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr)
32467 (vect_analyze_data_ref_dependence): Change max_vf from an int *
32468 to an unsigned int *.
32469 (vect_analyze_data_ref_dependences): Likewise.
32470 (vect_compute_data_ref_alignment): Handle polynomial vf.
32471 (vect_enhance_data_refs_alignment): Likewise.
32472 (vect_prune_runtime_alias_test_list): Likewise.
32473 (vect_shift_permute_load_chain): Likewise.
32474 (vect_supportable_dr_alignment): Likewise.
32475 (dependence_distance_ge_vf): Take the vectorization factor as a
32476 poly_uint64 rather than an unsigned HOST_WIDE_INT.
32477 (vect_analyze_data_refs): Change min_vf from an int * to a
32479 * tree-vect-loop-manip.c (vect_gen_scalar_loop_niters): Take
32480 vfm1 as a poly_uint64 rather than an int. Make the same change
32481 for the returned bound_scalar.
32482 (vect_gen_vector_loop_niters): Handle polynomial vf.
32483 (vect_do_peeling): Likewise. Update call to
32484 vect_gen_scalar_loop_niters and handle polynomial bound_scalars.
32485 (vect_gen_vector_loop_niters_mult_vf): Assert that the vf must
32487 * tree-vect-loop.c (vect_determine_vectorization_factor)
32488 (vect_update_vf_for_slp, vect_analyze_loop_2): Handle polynomial vf.
32489 (vect_get_known_peeling_cost): Likewise.
32490 (vect_estimate_min_profitable_iters, vectorizable_reduction): Likewise.
32491 (vect_worthwhile_without_simd_p, vectorizable_induction): Likewise.
32492 (vect_transform_loop): Likewise. Use the lowest possible VF when
32493 updating the upper bounds of the loop.
32494 (vect_min_worthwhile_factor): Make static. Return an unsigned int
32495 rather than an int.
32496 * tree-vect-slp.c (vect_attempt_slp_rearrange_stmts): Cope with
32497 polynomial unroll factors.
32498 (vect_analyze_slp_cost_1, vect_analyze_slp_instance): Likewise.
32499 (vect_make_slp_decision): Likewise.
32500 (vect_supported_load_permutation_p): Likewise, and polynomial
32502 (vect_analyze_slp_cost): Handle polynomial vf.
32503 (vect_slp_analyze_node_operations): Likewise.
32504 (vect_slp_analyze_bb_1): Likewise.
32505 (vect_transform_slp_perm_load): Take the vf as a poly_uint64 rather
32506 than an unsigned HOST_WIDE_INT.
32507 * tree-vect-stmts.c (vectorizable_simd_clone_call, vectorizable_store)
32508 (vectorizable_load): Handle polynomial vf.
32509 * tree-vectorizer.c (simduid_to_vf::vf): Change from an int to
32511 (adjust_simduid_builtins, shrink_simd_arrays): Update accordingly.
32513 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32514 Alan Hayward <alan.hayward@arm.com>
32515 David Sherwood <david.sherwood@arm.com>
32517 * match.pd: Handle bit operations involving three constants
32518 and try to fold one pair.
32520 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32522 * tree-vect-loop-manip.c: Include gimple-fold.h.
32523 (slpeel_make_loop_iterate_ntimes): Add step, final_iv and
32524 niters_maybe_zero parameters. Handle other cases besides a step of 1.
32525 (vect_gen_vector_loop_niters): Add a step_vector_ptr parameter.
32526 Add a path that uses a step of VF instead of 1, but disable it
32528 (vect_do_peeling): Add step_vector, niters_vector_mult_vf_var
32529 and niters_no_overflow parameters. Update calls to
32530 slpeel_make_loop_iterate_ntimes and vect_gen_vector_loop_niters.
32531 Create a new SSA name if the latter choses to use a ste other
32532 than zero, and return it via niters_vector_mult_vf_var.
32533 * tree-vect-loop.c (vect_transform_loop): Update calls to
32534 vect_do_peeling, vect_gen_vector_loop_niters and
32535 slpeel_make_loop_iterate_ntimes.
32536 * tree-vectorizer.h (slpeel_make_loop_iterate_ntimes, vect_do_peeling)
32537 (vect_gen_vector_loop_niters): Update declarations after above changes.
32539 2018-01-02 Michael Meissner <meissner@linux.vnet.ibm.com>
32541 * config/rs6000/rs6000.md (floor<mode>2): Add support for IEEE
32542 128-bit round to integer instructions.
32543 (ceil<mode>2): Likewise.
32544 (btrunc<mode>2): Likewise.
32545 (round<mode>2): Likewise.
32547 2018-01-02 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
32549 * config/rs6000/rs6000-string.c (expand_block_move): Allow the use of
32550 unaligned VSX load/store on P8/P9.
32551 (expand_block_clear): Allow the use of unaligned VSX
32552 load/store on P8/P9.
32554 2018-01-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32556 * config/rs6000/rs6000-p8swap.c (swap_feeds_both_load_and_store):
32558 (rs6000_analyze_swaps): Mark a web unoptimizable if it contains a
32559 swap associated with both a load and a store.
32561 2018-01-02 Andrew Waterman <andrew@sifive.com>
32563 * config/riscv/linux.h (ICACHE_FLUSH_FUNC): New.
32564 * config/riscv/riscv.md (clear_cache): Use it.
32566 2018-01-02 Artyom Skrobov <tyomitch@gmail.com>
32568 * web.c: Remove out-of-date comment.
32570 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
32572 * expr.c (fixup_args_size_notes): Check that any existing
32573 REG_ARGS_SIZE notes are correct, and don't try to re-add them.
32574 (emit_single_push_insn_1): Move stack_pointer_delta adjustment to...
32575 (emit_single_push_insn): ...here.
32577 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
32579 * rtl.h (CONST_VECTOR_ELT): Redefine to const_vector_elt.
32580 (const_vector_encoded_nelts): New function.
32581 (CONST_VECTOR_NUNITS): Redefine to use GET_MODE_NUNITS.
32582 (const_vector_int_elt, const_vector_elt): Declare.
32583 * emit-rtl.c (const_vector_int_elt_1): New function.
32584 (const_vector_elt): Likewise.
32585 * simplify-rtx.c (simplify_immed_subreg): Avoid taking the address
32586 of CONST_VECTOR_ELT.
32588 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
32590 * expr.c: Include rtx-vector-builder.h.
32591 (const_vector_mask_from_tree): Use rtx_vector_builder and operate
32592 directly on the tree encoding.
32593 (const_vector_from_tree): Likewise.
32594 * optabs.c: Include rtx-vector-builder.h.
32595 (expand_vec_perm_var): Use rtx_vector_builder and create a repeating
32596 sequence of "u" values.
32597 * vec-perm-indices.c: Include rtx-vector-builder.h.
32598 (vec_perm_indices_to_rtx): Use rtx_vector_builder and operate
32599 directly on the vec_perm_indices encoding.
32601 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
32603 * doc/rtl.texi (const_vector): Describe new encoding scheme.
32604 * Makefile.in (OBJS): Add rtx-vector-builder.o.
32605 * rtx-vector-builder.h: New file.
32606 * rtx-vector-builder.c: Likewise.
32607 * rtl.h (rtx_def::u2): Add a const_vector field.
32608 (CONST_VECTOR_NPATTERNS): New macro.
32609 (CONST_VECTOR_NELTS_PER_PATTERN): Likewise.
32610 (CONST_VECTOR_DUPLICATE_P): Likewise.
32611 (CONST_VECTOR_STEPPED_P): Likewise.
32612 (CONST_VECTOR_ENCODED_ELT): Likewise.
32613 (const_vec_duplicate_p): Check for a duplicated vector encoding.
32614 (unwrap_const_vec_duplicate): Likewise.
32615 (const_vec_series_p): Check for a non-duplicated vector encoding.
32616 Say that the function only returns true for integer vectors.
32617 * emit-rtl.c: Include rtx-vector-builder.h.
32618 (gen_const_vec_duplicate_1): Delete.
32619 (gen_const_vector): Call gen_const_vec_duplicate instead of
32620 gen_const_vec_duplicate_1.
32621 (const_vec_series_p_1): Operate directly on the CONST_VECTOR encoding.
32622 (gen_const_vec_duplicate): Use rtx_vector_builder.
32623 (gen_const_vec_series): Likewise.
32624 (gen_rtx_CONST_VECTOR): Likewise.
32625 * config/powerpcspe/powerpcspe.c: Include rtx-vector-builder.h.
32626 (swap_const_vector_halves): Take an rtx pointer rather than rtx.
32627 Build a new vector rather than modifying a CONST_VECTOR in-place.
32628 (handle_special_swappables): Update call accordingly.
32629 * config/rs6000/rs6000-p8swap.c: Include rtx-vector-builder.h.
32630 (swap_const_vector_halves): Take an rtx pointer rather than rtx.
32631 Build a new vector rather than modifying a CONST_VECTOR in-place.
32632 (handle_special_swappables): Update call accordingly.
32634 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
32636 * simplify-rtx.c (simplify_const_binary_operation): Use
32637 CONST_VECTOR_ELT instead of XVECEXP.
32639 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
32641 * tree-cfg.c (verify_gimple_assign_ternary): Allow the size of
32642 the selector elements to be different from the data elements
32643 if the selector is a VECTOR_CST.
32644 * tree-vect-stmts.c (vect_gen_perm_mask_any): Use a vector of
32645 ssizetype for the selector.
32647 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
32649 * optabs.c (shift_amt_for_vec_perm_mask): Try using series_p
32650 before testing each element individually.
32651 * tree-vect-generic.c (lower_vec_perm): Likewise.
32653 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
32655 * selftest.h (selftest::vec_perm_indices_c_tests): Declare.
32656 * selftest-run-tests.c (selftest::run_tests): Call it.
32657 * vector-builder.h (vector_builder::operator ==): New function.
32658 (vector_builder::operator !=): Likewise.
32659 * vec-perm-indices.h (vec_perm_indices::series_p): Declare.
32660 (vec_perm_indices::all_from_input_p): New function.
32661 * vec-perm-indices.c (vec_perm_indices::series_p): Likewise.
32662 (test_vec_perm_12, selftest::vec_perm_indices_c_tests): Likewise.
32663 * fold-const.c (fold_ternary_loc): Use tree_to_vec_perm_builder
32664 instead of reading the VECTOR_CST directly. Detect whether both
32665 vector inputs are the same before constructing the vec_perm_indices,
32666 and update the number of inputs argument accordingly. Use the
32667 utility functions added above. Only construct sel2 if we need to.
32669 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
32671 * optabs.c (expand_vec_perm_var): Use an explicit encoding for
32672 the broadcast of the low byte.
32673 (expand_mult_highpart): Use an explicit encoding for the permutes.
32674 * optabs-query.c (can_mult_highpart_p): Likewise.
32675 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
32676 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
32677 (vectorizable_bswap): Likewise.
32678 * tree-vect-data-refs.c (vect_grouped_store_supported): Use an
32679 explicit encoding for the power-of-2 permutes.
32680 (vect_permute_store_chain): Likewise.
32681 (vect_grouped_load_supported): Likewise.
32682 (vect_permute_load_chain): Likewise.
32684 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
32686 * vec-perm-indices.h (vec_perm_indices_to_tree): Declare.
32687 * vec-perm-indices.c (vec_perm_indices_to_tree): New function.
32688 * tree-ssa-forwprop.c (simplify_vector_constructor): Use it.
32689 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
32690 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
32691 (vect_gen_perm_mask_any): Likewise.
32693 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
32695 * int-vector-builder.h: New file.
32696 * vec-perm-indices.h: Include int-vector-builder.h.
32697 (vec_perm_indices): Redefine as an int_vector_builder.
32698 (auto_vec_perm_indices): Delete.
32699 (vec_perm_builder): Redefine as a stand-alone class.
32700 (vec_perm_indices::vec_perm_indices): New function.
32701 (vec_perm_indices::clamp): Likewise.
32702 * vec-perm-indices.c: Include fold-const.h and tree-vector-builder.h.
32703 (vec_perm_indices::new_vector): New function.
32704 (vec_perm_indices::new_expanded_vector): Update for new
32705 vec_perm_indices class.
32706 (vec_perm_indices::rotate_inputs): New function.
32707 (vec_perm_indices::all_in_range_p): Operate directly on the
32708 encoded form, without computing elided elements.
32709 (tree_to_vec_perm_builder): Operate directly on the VECTOR_CST
32710 encoding. Update for new vec_perm_indices class.
32711 * optabs.c (expand_vec_perm_const): Create a vec_perm_indices for
32712 the given vec_perm_builder.
32713 (expand_vec_perm_var): Update vec_perm_builder constructor.
32714 (expand_mult_highpart): Use vec_perm_builder instead of
32715 auto_vec_perm_indices.
32716 * optabs-query.c (can_mult_highpart_p): Use vec_perm_builder and
32717 vec_perm_indices instead of auto_vec_perm_indices. Use a single
32718 or double series encoding as appropriate.
32719 * fold-const.c (fold_ternary_loc): Use vec_perm_builder and
32720 vec_perm_indices instead of auto_vec_perm_indices.
32721 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
32722 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
32723 (vect_permute_store_chain): Likewise.
32724 (vect_grouped_load_supported): Likewise.
32725 (vect_permute_load_chain): Likewise.
32726 (vect_shift_permute_load_chain): Likewise.
32727 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
32728 (vect_transform_slp_perm_load): Likewise.
32729 (vect_schedule_slp_instance): Likewise.
32730 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
32731 (vectorizable_mask_load_store): Likewise.
32732 (vectorizable_bswap): Likewise.
32733 (vectorizable_store): Likewise.
32734 (vectorizable_load): Likewise.
32735 * tree-vect-generic.c (lower_vec_perm): Use vec_perm_builder and
32736 vec_perm_indices instead of auto_vec_perm_indices. Use
32737 tree_to_vec_perm_builder to read the vector from a tree.
32738 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Take a
32739 vec_perm_builder instead of a vec_perm_indices.
32740 (have_whole_vector_shift): Use vec_perm_builder and
32741 vec_perm_indices instead of auto_vec_perm_indices. Leave the
32742 truncation to calc_vec_perm_mask_for_shift.
32743 (vect_create_epilog_for_reduction): Likewise.
32744 * config/aarch64/aarch64.c (expand_vec_perm_d::perm): Change
32745 from auto_vec_perm_indices to vec_perm_indices.
32746 (aarch64_expand_vec_perm_const_1): Use rotate_inputs on d.perm
32747 instead of changing individual elements.
32748 (aarch64_vectorize_vec_perm_const): Use new_vector to install
32749 the vector in d.perm.
32750 * config/arm/arm.c (expand_vec_perm_d::perm): Change
32751 from auto_vec_perm_indices to vec_perm_indices.
32752 (arm_expand_vec_perm_const_1): Use rotate_inputs on d.perm
32753 instead of changing individual elements.
32754 (arm_vectorize_vec_perm_const): Use new_vector to install
32755 the vector in d.perm.
32756 * config/powerpcspe/powerpcspe.c (rs6000_expand_extract_even):
32757 Update vec_perm_builder constructor.
32758 (rs6000_expand_interleave): Likewise.
32759 * config/rs6000/rs6000.c (rs6000_expand_extract_even): Likewise.
32760 (rs6000_expand_interleave): Likewise.
32762 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
32764 * optabs-query.c (can_vec_perm_var_p): Check whether lowering
32765 to qimode could truncate the indices.
32766 * optabs.c (expand_vec_perm_var): Likewise.
32768 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
32770 * Makefile.in (OBJS): Add vec-perm-indices.o.
32771 * vec-perm-indices.h: New file.
32772 * vec-perm-indices.c: Likewise.
32773 * target.h (vec_perm_indices): Replace with a forward class
32775 (auto_vec_perm_indices): Move to vec-perm-indices.h.
32776 * optabs.h: Include vec-perm-indices.h.
32777 (expand_vec_perm): Delete.
32778 (selector_fits_mode_p, expand_vec_perm_var): Declare.
32779 (expand_vec_perm_const): Declare.
32780 * target.def (vec_perm_const_ok): Replace with...
32781 (vec_perm_const): ...this new hook.
32782 * doc/tm.texi.in (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Replace with...
32783 (TARGET_VECTORIZE_VEC_PERM_CONST): ...this new hook.
32784 * doc/tm.texi: Regenerate.
32785 * optabs.def (vec_perm_const): Delete.
32786 * doc/md.texi (vec_perm_const): Likewise.
32787 (vec_perm): Refer to TARGET_VECTORIZE_VEC_PERM_CONST.
32788 * expr.c (expand_expr_real_2): Use expand_vec_perm_const rather than
32789 expand_vec_perm for constant permutation vectors. Assert that
32790 the mode of variable permutation vectors is the integer equivalent
32791 of the mode that is being permuted.
32792 * optabs-query.h (selector_fits_mode_p): Declare.
32793 * optabs-query.c: Include vec-perm-indices.h.
32794 (selector_fits_mode_p): New function.
32795 (can_vec_perm_const_p): Check whether targetm.vectorize.vec_perm_const
32796 is defined, instead of checking whether the vec_perm_const_optab
32797 exists. Use targetm.vectorize.vec_perm_const instead of
32798 targetm.vectorize.vec_perm_const_ok. Check whether the indices
32799 fit in the vector mode before using a variable permute.
32800 * optabs.c (shift_amt_for_vec_perm_mask): Take a mode and a
32801 vec_perm_indices instead of an rtx.
32802 (expand_vec_perm): Replace with...
32803 (expand_vec_perm_const): ...this new function. Take the selector
32804 as a vec_perm_indices rather than an rtx. Also take the mode of
32805 the selector. Update call to shift_amt_for_vec_perm_mask.
32806 Use targetm.vectorize.vec_perm_const instead of vec_perm_const_optab.
32807 Use vec_perm_indices::new_expanded_vector to expand the original
32808 selector into bytes. Check whether the indices fit in the vector
32809 mode before using a variable permute.
32810 (expand_vec_perm_var): Make global.
32811 (expand_mult_highpart): Use expand_vec_perm_const.
32812 * fold-const.c: Includes vec-perm-indices.h.
32813 * tree-ssa-forwprop.c: Likewise.
32814 * tree-vect-data-refs.c: Likewise.
32815 * tree-vect-generic.c: Likewise.
32816 * tree-vect-loop.c: Likewise.
32817 * tree-vect-slp.c: Likewise.
32818 * tree-vect-stmts.c: Likewise.
32819 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm_const):
32821 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>): Delete.
32822 * config/aarch64/aarch64.c (aarch64_expand_vec_perm_const)
32823 (aarch64_vectorize_vec_perm_const_ok): Fuse into...
32824 (aarch64_vectorize_vec_perm_const): ...this new function.
32825 (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
32826 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
32827 * config/arm/arm-protos.h (arm_expand_vec_perm_const): Delete.
32828 * config/arm/vec-common.md (vec_perm_const<mode>): Delete.
32829 * config/arm/arm.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
32830 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
32831 (arm_expand_vec_perm_const, arm_vectorize_vec_perm_const_ok): Merge
32833 (arm_vectorize_vec_perm_const): ...this new function. Explicitly
32834 check for NEON modes.
32835 * config/i386/i386-protos.h (ix86_expand_vec_perm_const): Delete.
32836 * config/i386/sse.md (VEC_PERM_CONST, vec_perm_const<mode>): Delete.
32837 * config/i386/i386.c (ix86_expand_vec_perm_const_1): Update comment.
32838 (ix86_expand_vec_perm_const, ix86_vectorize_vec_perm_const_ok): Merge
32840 (ix86_vectorize_vec_perm_const): ...this new function. Incorporate
32841 the old VEC_PERM_CONST conditions.
32842 * config/ia64/ia64-protos.h (ia64_expand_vec_perm_const): Delete.
32843 * config/ia64/vect.md (vec_perm_const<mode>): Delete.
32844 * config/ia64/ia64.c (ia64_expand_vec_perm_const)
32845 (ia64_vectorize_vec_perm_const_ok): Merge into...
32846 (ia64_vectorize_vec_perm_const): ...this new function.
32847 * config/mips/loongson.md (vec_perm_const<mode>): Delete.
32848 * config/mips/mips-msa.md (vec_perm_const<mode>): Delete.
32849 * config/mips/mips-ps-3d.md (vec_perm_constv2sf): Delete.
32850 * config/mips/mips-protos.h (mips_expand_vec_perm_const): Delete.
32851 * config/mips/mips.c (mips_expand_vec_perm_const)
32852 (mips_vectorize_vec_perm_const_ok): Merge into...
32853 (mips_vectorize_vec_perm_const): ...this new function.
32854 * config/powerpcspe/altivec.md (vec_perm_constv16qi): Delete.
32855 * config/powerpcspe/paired.md (vec_perm_constv2sf): Delete.
32856 * config/powerpcspe/spe.md (vec_perm_constv2si): Delete.
32857 * config/powerpcspe/vsx.md (vec_perm_const<mode>): Delete.
32858 * config/powerpcspe/powerpcspe-protos.h (altivec_expand_vec_perm_const)
32859 (rs6000_expand_vec_perm_const): Delete.
32860 * config/powerpcspe/powerpcspe.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK):
32862 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
32863 (altivec_expand_vec_perm_const_le): Take each operand individually.
32864 Operate on constant selectors rather than rtxes.
32865 (altivec_expand_vec_perm_const): Likewise. Update call to
32866 altivec_expand_vec_perm_const_le.
32867 (rs6000_expand_vec_perm_const): Delete.
32868 (rs6000_vectorize_vec_perm_const_ok): Delete.
32869 (rs6000_vectorize_vec_perm_const): New function.
32870 (rs6000_do_expand_vec_perm): Take a vec_perm_builder instead of
32871 an element count and rtx array.
32872 (rs6000_expand_extract_even): Update call accordingly.
32873 (rs6000_expand_interleave): Likewise.
32874 * config/rs6000/altivec.md (vec_perm_constv16qi): Delete.
32875 * config/rs6000/paired.md (vec_perm_constv2sf): Delete.
32876 * config/rs6000/vsx.md (vec_perm_const<mode>): Delete.
32877 * config/rs6000/rs6000-protos.h (altivec_expand_vec_perm_const)
32878 (rs6000_expand_vec_perm_const): Delete.
32879 * config/rs6000/rs6000.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
32880 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
32881 (altivec_expand_vec_perm_const_le): Take each operand individually.
32882 Operate on constant selectors rather than rtxes.
32883 (altivec_expand_vec_perm_const): Likewise. Update call to
32884 altivec_expand_vec_perm_const_le.
32885 (rs6000_expand_vec_perm_const): Delete.
32886 (rs6000_vectorize_vec_perm_const_ok): Delete.
32887 (rs6000_vectorize_vec_perm_const): New function. Remove stray
32888 reference to the SPE evmerge intructions.
32889 (rs6000_do_expand_vec_perm): Take a vec_perm_builder instead of
32890 an element count and rtx array.
32891 (rs6000_expand_extract_even): Update call accordingly.
32892 (rs6000_expand_interleave): Likewise.
32893 * config/sparc/sparc.md (vec_perm_constv8qi): Delete in favor of...
32894 * config/sparc/sparc.c (sparc_vectorize_vec_perm_const): ...this
32896 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
32898 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
32900 * optabs.c (expand_vec_perm_1): Assert that SEL has an integer
32901 vector mode and that that mode matches the mode of the data
32903 (expand_vec_perm): Split handling of non-CONST_VECTOR selectors
32904 out into expand_vec_perm_var. Do all CONST_VECTOR handling here,
32905 directly using expand_vec_perm_1 when forcing selectors into
32907 (expand_vec_perm_var): New function, split out from expand_vec_perm.
32909 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
32911 * optabs-query.h (can_vec_perm_p): Delete.
32912 (can_vec_perm_var_p, can_vec_perm_const_p): Declare.
32913 * optabs-query.c (can_vec_perm_p): Split into...
32914 (can_vec_perm_var_p, can_vec_perm_const_p): ...these two functions.
32915 (can_mult_highpart_p): Use can_vec_perm_const_p to test whether a
32916 particular selector is valid.
32917 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
32918 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
32919 (vect_grouped_load_supported): Likewise.
32920 (vect_shift_permute_load_chain): Likewise.
32921 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
32922 (vect_transform_slp_perm_load): Likewise.
32923 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
32924 (vectorizable_bswap): Likewise.
32925 (vect_gen_perm_mask_checked): Likewise.
32926 * fold-const.c (fold_ternary_loc): Likewise. Don't take
32927 implementations of variable permutation vectors into account
32928 when deciding which selector to use.
32929 * tree-vect-loop.c (have_whole_vector_shift): Don't check whether
32930 vec_perm_const_optab is supported; instead use can_vec_perm_const_p
32931 with a false third argument.
32932 * tree-vect-generic.c (lower_vec_perm): Use can_vec_perm_const_p
32933 to test whether the constant selector is valid and can_vec_perm_var_p
32934 to test whether a variable selector is valid.
32936 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
32938 * optabs-query.h (can_vec_perm_p): Take a const vec_perm_indices *.
32939 * optabs-query.c (can_vec_perm_p): Likewise.
32940 * fold-const.c (fold_vec_perm): Take a const vec_perm_indices &
32941 instead of vec_perm_indices.
32942 * tree-vectorizer.h (vect_gen_perm_mask_any): Likewise,
32943 (vect_gen_perm_mask_checked): Likewise,
32944 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise,
32945 (vect_gen_perm_mask_checked): Likewise,
32947 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
32949 * optabs-query.h (qimode_for_vec_perm): Declare.
32950 * optabs-query.c (can_vec_perm_p): Split out qimode search to...
32951 (qimode_for_vec_perm): ...this new function.
32952 * optabs.c (expand_vec_perm): Use qimode_for_vec_perm.
32954 2018-01-02 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
32956 * rtlanal.c (canonicalize_condition): Return 0 if final rtx
32957 does not have a conditional at the top.
32959 2018-01-02 Richard Biener <rguenther@suse.de>
32961 * ipa-inline.c (big_speedup_p): Fix expression.
32963 2018-01-02 Jan Hubicka <hubicka@ucw.cz>
32966 * config/i386/x86-tune-costs.h: Increase cost of integer load costs
32969 2018-01-02 Jan Hubicka <hubicka@ucw.cz>
32973 * x86-tune-costs.h (generic_cost): Reduce cost of FDIV 20->17,
32974 cost of sqrt 20->14, DIVSS 18->13, DIVSD 32->17, SQRtSS 30->14
32975 and SQRTsD 58->18, cond_not_taken_branch_cost. 2->1. Increase
32976 cond_taken_branch_cost 3->4.
32978 2018-01-01 Jakub Jelinek <jakub@redhat.com>
32980 PR tree-optimization/83581
32981 * tree-loop-distribution.c (pass_loop_distribution::execute): Return
32982 TODO_cleanup_cfg if any changes have been made.
32984 PR middle-end/83608
32985 * expr.c (store_expr_with_bounds): Use simplify_gen_subreg instead of
32986 convert_modes if target mode has the right side, but different mode
32989 PR middle-end/83609
32990 * expr.c (expand_assignment): Fix up a typo in simplify_gen_subreg
32991 last argument when extracting from CONCAT. If either from_real or
32992 from_imag is NULL, use expansion through memory. If result is not
32993 a CONCAT and simplify_gen_subreg fails, try to simplify_gen_subreg
32994 the parts directly to inner mode, if even that fails, use expansion
32997 PR middle-end/83623
32998 * expmed.c (expand_shift_1): For 2-byte rotates by BITS_PER_UNIT,
32999 check for bswap in mode rather than HImode and use that in expand_unop
33002 Copyright (C) 2018 Free Software Foundation, Inc.
33004 Copying and distribution of this file, with or without modification,
33005 are permitted in any medium without royalty provided the copyright
33006 notice and this notice are preserved.