1 2018-09-20 Segher Boessenkool <segher@kernel.crashing.org>
3 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete
5 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete
8 2018-09-20 Marek Polacek <polacek@redhat.com>
10 * doc/invoke.texi: Add -Wno-init-list-lifetime to C++ Language Options.
12 2018-09-20 Richard Sandiford <richard.sandiford@arm.com>
14 PR tree-optimization/87288
15 * tree-vect-loop.c (vect_analyze_loop_2): Take PEELING_FOR_GAPS
16 into account when determining PEELING_FOR_NITERS.
18 2018-09-20 Richard Sandiford <richard.sandiford@arm.com>
20 PR tree-optimization/86877
21 * tree-vect-loop.c (vect_analyze_loop_2): Call
22 vect_verify_datarefs_alignment.
24 2018-09-19 Marek Polacek <polacek@redhat.com>
26 * doc/invoke.texi: Document -Wclass-conversion.
28 2018-09-19 John David Anglin <danglin@gcc.gnu.org>
30 * config/pa/pa.c (pa_adjust_priority): Delete.
31 (TARGET_SCHED_ADJUST_PRIORITY): Delete define.
33 * config/pa/pa.md (atomic_storeqi): Restore deleted expander.
34 (atomic_storehi): Likewise.
35 (atomic_storesi): Likewise.
36 (atomic_loaddi): Restore compare and swap exchange loop code.
38 2018-09-19 Segher Boessenkool <segher@kernel.crashing.org>
40 PR rtl-optimization/86902
41 * combine.c (try_combine): When changing the CC mode used, don't change
42 an unrelated mode in other_insn to that new CC mode.
44 2018-09-19 David Malcolm <dmalcolm@redhat.com>
46 * tree-data-ref.c (runtime_alias_check_p): Use formatted printing
47 with %T in place of calls to dump_generic_expr.
48 (prune_runtime_alias_test_list): Likewise.
49 (create_runtime_alias_checks): Likewise.
50 * tree-vect-data-refs.c (vect_check_nonzero_value): Likewise.
51 (vect_analyze_data_ref_dependence): Likewise.
52 (vect_slp_analyze_data_ref_dependence): Likewise.
53 (vect_record_base_alignment): Likewise. Use %G in place of call
55 (vect_compute_data_ref_alignment): Likewise.
56 (verify_data_ref_alignment): Likewise.
57 (vect_find_same_alignment_drs): Likewise.
58 (vect_analyze_group_access_1): Likewise.
59 (vect_analyze_data_ref_accesses): Likewise.
60 (dependence_distance_ge_vf): Likewise.
61 (dump_lower_bound): Likewise.
62 (vect_prune_runtime_alias_test_list): Likewise.
63 (vect_find_stmt_data_reference): Likewise.
64 (vect_analyze_data_refs): Likewise.
65 (vect_create_addr_base_for_vector_ref): Likewise.
66 (vect_create_data_ref_ptr): Likewise.
67 * tree-vect-loop-manip.c (vect_set_loop_condition): Likewise.
68 (vect_can_advance_ivs_p): Likewise.
69 (vect_update_ivs_after_vectorizer): Likewise.
70 (vect_gen_prolog_loop_niters): Likewise.
71 (vect_prepare_for_masked_peels): Likewise.
72 * tree-vect-loop.c (vect_determine_vf_for_stmt): Likewise.
73 (vect_determine_vectorization_factor): Likewise.
74 (vect_is_simple_iv_evolution): Likewise.
75 (vect_analyze_scalar_cycles_1): Likewise.
76 (vect_analyze_loop_operations): Likewise.
77 (report_vect_op): Likewise.
78 (vect_is_slp_reduction): Likewise.
79 (check_reduction_path): Likewise.
80 (vect_is_simple_reduction): Likewise.
81 (vect_create_epilog_for_reduction): Likewise.
82 (vect_finalize_reduction:): Likewise.
83 (vectorizable_induction): Likewise.
84 (vect_transform_loop_stmt): Likewise.
85 (vect_transform_loop): Likewise.
86 (optimize_mask_stores): Likewise.
87 * tree-vect-patterns.c (vect_pattern_detected): Likewise.
88 (vect_split_statement): Likewise.
89 (vect_recog_over_widening_pattern): Likewise.
90 (vect_recog_average_pattern): Likewise.
91 (vect_determine_min_output_precision_1): Likewise.
92 (vect_determine_precisions_from_range): Likewise.
93 (vect_determine_precisions_from_users): Likewise.
94 (vect_mark_pattern_stmts): Likewise.
95 (vect_pattern_recog_1): Likewise.
96 * tree-vect-slp.c (vect_get_and_check_slp_defs): Likewise.
97 (vect_record_max_nunits): Likewise.
98 (vect_build_slp_tree_1): Likewise.
99 (vect_build_slp_tree_2): Likewise.
100 (vect_print_slp_tree): Likewise.
101 (vect_analyze_slp_instance): Likewise.
102 (vect_detect_hybrid_slp_stmts): Likewise.
103 (vect_detect_hybrid_slp_1): Likewise.
104 (vect_slp_analyze_operations): Likewise.
105 (vect_slp_analyze_bb_1): Likewise.
106 (vect_transform_slp_perm_load): Likewise.
107 (vect_schedule_slp_instance): Likewise.
108 * tree-vect-stmts.c (vect_mark_relevant): Likewise.
109 (vect_mark_stmts_to_be_vectorized): Likewise.
110 (vect_init_vector_1): Likewise.
111 (vect_get_vec_def_for_operand): Likewise.
112 (vect_finish_stmt_generation_1): Likewise.
113 (vect_check_load_store_mask): Likewise.
114 (vectorizable_call): Likewise.
115 (vectorizable_conversion): Likewise.
116 (vectorizable_operation): Likewise.
117 (vectorizable_load): Likewise.
118 (vect_analyze_stmt): Likewise.
119 (vect_is_simple_use): Likewise.
120 (vect_get_vector_types_for_stmt): Likewise.
121 (vect_get_mask_type_for_stmt): Likewise.
122 * tree-vectorizer.c (increase_alignment): Likewise.
124 2018-09-19 Andrew Stubbs <ams@codesourcery.com>
126 * doc/rtl.texi: Adjust vec_select description.
127 * simplify-rtx.c (simplify_binary_operation_1): Allow VEC_SELECT to use
128 non-constant selectors.
130 2018-09-19 Matthew Malcomson <matthew.malcomson@arm.com>
132 * config/aarch64/aarch64-protos.h
133 (aarch64_offset_9bit_signed_unscaled_p): New declaration.
134 * config/aarch64/aarch64.md (arches): New "rcpc8_4" attribute value.
135 (arch_enabled): Add check for "rcpc8_4" attribute value of "arch".
136 * config/aarch64/aarch64.h (AARCH64_FL_RCPC8_4): New bitfield.
137 (AARCH64_FL_FOR_ARCH8_4): Include AARCH64_FL_RCPC8_4.
138 (AARCH64_FL_PROFILE): Move index so flags are ordered.
139 (AARCH64_ISA_RCPC8_4): New flag.
140 * config/aarch64/aarch64.c (offset_9bit_signed_unscaled_p): Renamed
141 to aarch64_offset_9bit_signed_unscaled_p.
142 * config/aarch64/atomics.md (atomic_store<mode>): Allow offset
144 * config/aarch64/constraints.md (Ust): New constraint.
145 * config/aarch64/predicates.md.
146 (aarch64_9bit_offset_memory_operand): New predicate.
147 (aarch64_rcpc_memory_operand): New predicate.
149 2018-09-19 Eric Botcazou <ebotcazou@adacore.com>
151 PR rtl-optimization/87361
152 * rtlanal.c (nonzero_bits1): Revert accidental change.
154 2018-09-19 Richard Biener <rguenther@suse.de>
156 PR tree-optimization/87349
157 PR tree-optimization/87342
158 * tree-ssa-sccvn.c (do_rpo_vn): Iterate max_rpo computation.
160 2018-09-18 Marek Polacek <polacek@redhat.com>
162 P1064R0 - Allowing Virtual Function Calls in Constant Expressions
163 * gimple-fold.c (gimple_get_virt_method_for_vtable): Adjust assert.
165 2018-09-18 Segher Boessenkool <segher@kernel.crashing.org>
167 * config/rs6000/rs6000.md: Remove old "Cygnus sibcall" comment.
169 2018-09-18 Segher Boessenkool <segher@kernel.crashing.org>
171 PR rtl-optimization/86882
172 * rtlanal.c (reg_overlap_mentioned_p): Handle CLOBBER.
174 2018-09-18 Uros Bizjak <ubizjak@gmail.com>
176 * config/i386/i386.md (*<code>extend<mode>xf2): Macroize insn from
177 *<code>extendsfxf2 and *<code>extenddfxf2 using MODEF mode iterator.
179 2018-09-18 Jonathan Wakely <jwakely@redhat.com>
182 * doc/invoke.texi (Link Options): Fix formatting and grammar.
184 2018-09-18 Richard Biener <rguenther@suse.de>
187 * tree-ssa-coalesce.c (tree_int_map_hasher): Remove.
188 (compute_samebase_partition_bases): Likewise.
189 (coalesce_ssa_name): Always use compute_optimized_partition_bases.
190 (gimple_can_coalesce_p): Simplify.
192 2018-09-18 Hans-Peter Nilsson <hp@bitrange.com>
194 Handle a library implementation of ffs calling __builtin_ffs.
195 * config/mmix/mmix.c (TARGET_INIT_LIBFUNCS): Override with...
196 (mmix_init_libfuncs): New function: make __builtin_ffs expand
199 2018-09-17 David Malcolm <dmalcolm@redhat.com>
201 * diagnostic-show-locus.c (class layout_range): Add field
203 (layout_range::layout_range): Add "original_idx" param and use it
204 to initialize new field.
205 (make_range): Use 0 for original_idx.
206 (layout::layout): Pass in index to calls to
207 maybe_add_location_range.
208 (layout::maybe_add_location_range): Add param "original_idx" and
209 pass it on to layout_range.
210 (layout::print_any_labels): Pass on range->m_original_idx to
212 (gcc_rich_location::add_location_if_nearby): Use 0 for
214 * gcc-rich-location.h (text_range_label::get_text): Update for new
216 (range_label_for_type_mismatch::get_text): Likewise.
218 2018-09-17 Uros Bizjak <ubizjak@gmail.com>
220 * config/i386/i386.c (ix86_emit_i387_log1p): Emit fldln2 earlier.
222 2018-09-17 David Malcolm <dmalcolm@redhat.com>
224 * gimple-ssa-sprintf.c (fmtwarn): Update for introduction of
225 format_string_diagnostic_t.
226 (fmtwarn_n): Likewise.
227 * substring-locations.c
228 (format_string_diagnostic_t::format_string_diagnostic_t) New ctor.
229 (format_warning_n_va): Convert to...
230 (format_string_diagnostic_t::emit_warning_n_va): ...this.
231 (format_warning_va): Convert to...
232 (format_string_diagnostic_t::emit_warning_va): ...this.
233 (format_warning_at_substring): Convert to...
234 (format_string_diagnostic_t::emit_warning): ...this.
235 (format_warning_at_substring_n): Convert to...
236 (format_string_diagnostic_t::emit_warning_n): ...this.
237 * substring-locations.h (class format_string_diagnostic_t): New
239 (format_warning_va): Convert to
240 format_string_diagnostic_t::emit_warning_va.
241 (format_warning_n_va): Convert to
242 format_string_diagnostic_t::emit_warning_n_va.
243 (format_warning_at_substring): Convert to
244 format_string_diagnostic_t::emit_warning.
245 (format_warning_at_substring_n): Convert to
246 format_string_diagnostic_t::emit_warning_n.
248 2018-09-17 Cesar Philippidis <cesar@codesourcery.com>
249 Bernd Schmidt <bernds_cb1@t-online.de>
251 * config/nvptx/nvptx.md (atomic_fetch_<logic><mode>): Enable with
254 2018-09-17 Uros Bizjak <ubizjak@gmail.com>
256 * config/i386/i386.md (truncxf<mode>2_i387_noop_unspec): Change
257 operand 0 predicate to nonimmediate operand.
258 (rint<mode>2_frndint): Remove insn pattern.
259 (rint<mode>2): Change operand 1 predicate to general_operand.
260 Extend operand 1 to XFmode and generate rintxf2 insn.
261 (frndintxf2_<rounding>): Rename from frndint<mode>2_<rounding>.
262 Do not use X87MODEF mode macro.
263 (frndintxf2_<rounding>_i387): Rename from
264 frndint<mode>2_<rounding>_i387. Do not use X87MODEF mode macro.
265 (<rounding_insn><mode>2): For non-SSE modes, extend operand 1
266 to XFmode and generate significandxf3 insn.
268 2018-09-17 Richard Biener <rguenther@suse.de>
270 PR tree-optimization/87328
271 * tree-ssa-sccvn.c (process_bb): Remove assertion about not
272 visiting unexecutable backedges when not iterating.
273 (do_rpo_vn): Mark all edges not executable even when not
276 2018-09-17 Martin Jambor <mjambor@suse.cz>
279 * doc/invoke.texi (Warning Options): Likewise.
281 2018-09-17 Richard Biener <rguenther@suse.de>
283 PR tree-optimization/87301
284 * tree-ssa-sccvn.c (eliminate_dom_walker::eliminate_cleanup): Properly
285 clean EH info from leftover copy assignments.
287 2018-09-17 Martin Liska <mliska@suse.cz>
289 PR gcov-profile/85871
290 * gcov.c (output_intermediate_file): Fix out of bounds
293 2018-09-17 Vineet Gupta <vgupta@synopsys.com>
295 * config/arc/arc.c: Object attributes for core4 not reflected
297 * config/arc/arc.h: Don't restrict DBNZ to core3 (core4 includes
300 2018-09-17 Alexey Brodkin <abrodkin@synopsys.com>
302 * config/arc/linux.h (LINK_EH_SPEC): Add missing space.
304 2018-09-17 Martin Liska <mliska@suse.cz>
306 * doc/gcov.texi: Document new option --use-hotness-colors.
307 * gcov.c (struct source_info): Declare new field.
308 (source_info::source_info): Set default for maximum_count.
309 (print_usage): Add new -q option.
310 (process_args): Process it.
311 (accumulate_line_info): Save src->maximum_count.
312 (output_line_beginning): Make color line number if
313 flag_use_hotness_colors is set.
314 (output_line_details): Pass default argument value.
315 (output_lines): Pass src->maximum_count.
317 2018-09-17 Martin Liska <mliska@suse.cz>
319 * common/config/i386/i386-common.c (ix86_get_valid_option_values):
320 Use processor_names table.
321 * config/i386/i386.c (ix86_default_align): Use
322 processor_cost_table for alignment values.
323 (ix86_option_override_internal): Use processor_names.
324 (ix86_function_specific_print): Likewise.
325 * config/i386/i386.h (struct processor_costs):
326 Add alignment values.
327 (struct ptt): Remove and replace with const char *.
328 * config/i386/x86-tune-costs.h (struct processor_costs):
329 Declare default alignments for all costs.
331 2018-09-17 Aldy Hernandez <aldyh@redhat.com>
333 * tree-vrp.c (extract_range_from_unary_expr): Do not special case
334 symbolics or VR_VARYING ranges for ABS_EXPR.
335 * wide-int-range.cc (wide_int_range_abs): Return positive numbers
336 when range will wrap.
338 2018-09-15 Eric Botcazou <ebotcazou@adacore.com>
341 * cfgexpand.c (expand_gimple_basic_block): Be prepared for a BARRIER
342 before and after a JUMP_TABLE_DATA.
344 2018-09-14 John David Anglin <danglin@gcc.gnu.org>
347 * dojump.c (do_compare_and_jump): Canonicalize function pointers
348 when one operand is a function pointer. Use POINTER_TYPE_P and
349 FUNC_OR_METHOD_TYPE_P.
350 * expr.c (do_store_flag): Use POINTER_TYPE_P and FUNC_OR_METHOD_TYPE_P.
351 * fold-const.c (build_range_check): Likewise.
352 * match.pd (simple_comparison): Likewise.
354 2018-09-14 David Malcolm <dmalcolm@redhat.com>
357 * spellcheck.c (get_edit_distance_cutoff): New function.
358 (selftest::test_edit_distance_unit_test_oneway): Rename to...
359 (selftest::test_get_edit_distance_one_way): ...this.
360 (selftest::test_get_edit_distance_unit): Rename to...
361 (selftest::test_get_edit_distance_both_ways): ...this.
362 (selftest::test_edit_distances): Move tests to this new function,
363 and test some more pairs of strings. Update for above renaming.
364 (selftest::get_old_cutoff): New function.
365 (selftest::test_get_edit_distance_cutoff): New function.
366 (selftest::assert_suggested_for): New function.
367 (ASSERT_SUGGESTED_FOR): New macro.
368 (selftest::assert_not_suggested_for): New function.
369 (ASSERT_NOT_SUGGESTED_FOR): New macro.
370 (selftest::test_suggestions): New function.
371 (selftest::spellcheck_c_tests): Move test_get_edit_distance_unit
372 tests to selftest::test_edit_distances and call it. Add calls to
373 selftest::test_get_edit_distance_cutoff and
374 selftest::test_suggestions.
375 * spellcheck.h (get_edit_distance_cutoff): New function declaration.
376 (best_match::consider): Replace hard-coded cutoff calculation with
378 (best_match::get_cutoff): New declaration.
379 (best_match::get_best_meaningful_candidate): Likewise.
381 2018-09-14 Bernd Edlinger <bernd.edlinger@hotmail.de>
383 * builtins.c (fold_builtin_strlen): Remove TODO comment.
385 2018-09-14 Bernd Edlinger <bernd.edlinger@hotmail.de>
388 2018-07-30 Bernd Edlinger <bernd.edlinger@hotmail.de>
390 * tree-ssa-forwprop.c (simplify_builtin_call): Don't create a not NUL
391 terminated string literal.
393 2018-09-14 Martin Sebor <msebor@redhat.com>
395 * builtins.c (unterminated_array): Handle ARRAY_REF.
396 (expand_builtin_stpcpy_1): Detect unterminated char arrays.
397 * builtins.h (unterminated_array): Declare extern.
398 * gimple-fold.c (gimple_fold_builtin_stpcpy): Detect unterminated
400 (gimple_fold_builtin_sprintf): Propagate NO_WARNING to transformed
403 2018-09-14 Martin Sebor <msebor@redhat.com>
404 Jeff Law <law@redhat.com>
406 * builtins.c (unterminated_array): New.
407 (expand_builtin_strcpy): Adjust.
408 (expand_builtin_strcpy_args): Detect unterminated arrays.
409 * gimple-fold.c (get_maxval_strlen): Add argument. Detect
411 * gimple-fold.h (get_maxval_strlen): Add argument.
412 (gimple_fold_builtin_strcpy): Detec unterminated arrays.
414 * gimple-fold.c (get_range_strlen): Add argument.
415 (get_maxval_strlen): Adjust.
416 * gimple-fold.h (get_range_strlen): Add argument.
418 2018-09-14 Wei Xiao <wei3.xiao@intel.com>
420 * config/i386/movdirintrin.h: Fix copyright year.
422 2018-09-14 Uros Bizjak <ubizjak@gmail.com>
424 * reg-stack.c: Include regs.h.
425 (replace_reg): Assert that mode is MODE_FLOAT or MODE_COMPLEX_FLOAT.
426 (emit_pop_insn): Default pop insn mode to the reg_raw_mode of
427 FIRST_STACK_REG, not DFmode.
428 (emit_swap_insn): Default swap insn mode to the reg_raw_mode of
429 FIRST_STACK_REG, not XFmode. Explicitly construct swap RTX.
430 (change stack): Default register mode to the reg_raw_mode of
431 FIRST_STACK_REG, not DFmode.
432 * config/i386/i386.md (*swap<mode>): Remove insn pattern.
433 (*swapxf): Rename from swapxf.
435 2018-09-14 Carl Love <cel@us.ibm.com>
437 * config/rs6000/emmintrin.h: Add _MM_SHUFFLE2.
438 * config/rs6000/xmmintrin.h: Add _MM_SHUFFLE.
440 2018-09-14 Segher Boessenkool <segher@kernel.crashing.org>
443 * config/rs6000/rs6000.md (*mov<mode>_hardfloat64): Add Z to the Y
446 2018-09-14 Sam Tebbs <sam.tebbs@arm.com>
449 * config/aarch64/aarch64.md (*aarch64_bfxilsi_uxtw): Define.
451 2018-09-14 Jason Merrill <jason@redhat.com>
453 Fix --enable-gather-detailed-mem-stats.
454 * hash-table.c (hash_table_usage): Change from variable to function.
455 * hash-table.h: Adjust.
456 * Makefile.in: Add missing dependencies on hash-table.h.
458 2018-09-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
460 PR tree-optimization/87259
462 (pass_cse_reciprocals::execute): Run optimize_recip_sqrt after
463 execute_cse_reciprocals_1 has tried transforming.
465 2018-09-14 Aldy Hernandez <aldyh@redhat.com>
467 * tree-vrp.c (extract_range_from_binary_expr_1): Normalize
468 VR_VARYING for PLUS/MINUS_EXPR.
470 2018-09-14 Ilya Leoshkevich <iii@linux.ibm.com>
472 * config/s390/s390-passes.def (INSERT_PASS_BEFORE): Improve
475 2018-09-14 Richard Biener <rguenther@suse.de>
478 * tree-ssa-coalesce.c (ssa_conflicts_merge): Remove conflict
479 bits for the merged partition.
481 2018-09-13 Martin Sebor <msebor@redhat.com>
482 Bernd Edlinger <bernd.edlinger@hotmail.de>
484 * builtins.h (c_srlen): Add argument.
485 * builtins.c (warn_string_no_nul): New function.
486 (c_strlen): Add argument and use it. Update recursive calls.
487 Pass DECL argument to string_constant to get info on non
488 terminated strings. Update *NONSTR as needed.
489 (fold_builtin_strlen): Add argument to calls to c_strlen.
490 Warn for unterminated arrays.
491 (warn_string_no_null): Add prototype.
492 * expr.c (string_constant): Update arguments. Update recursive
493 calls appropriately. Detect missing NUL terminator and outermost
494 declaration its missing in.
495 Improve checks for arrays with nonzero lower bound or elements
496 that are not a single byte. Simplify offset computation.
497 Simplify checks for non-NUL terminated strings.
498 * gimple-fold.c (get_range_strlen): Add argument to c_strlen call.
499 * gimple-ssa-sprintf.c (get_string_length): Remove unnecessary code.
501 2018-09-13 Bernd Edlinger <bernd.edlinger@hotmail.de>
503 * builtins.c (c_strlen): Handle not zero terminated STRING_CSTs
505 * fold-const.c (c_getstr): Fix function comment. Remove unused third
506 argument. Fix range checks.
507 * fold-const.h (c_getstr): Adjust protoype.
508 * gimple-fold.c (gimple_fold_builtin_memory_op): Avoid folding when
509 string is constant but contains no NUL byte.
511 * expr.c (string_constant): Adjust function comment.
512 Remove bogus check for zero termination.
514 * fold-const.c (c_getstr): Clamp STRING_LENGTH to STRING_SIZE.
516 * varasm.c (compare_constant): Compare type size of STRING_CSTs.
517 (get_constant_size): Don't make STRING_CSTs larger than they are.
518 (check_string_literal): New check function for STRING_CSTs.
519 (output_constant): Use it.
521 2018-09-13 Eric Botcazou <ebotcazou@adacore.com>
524 * config/visium/visium.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Define.
526 2018-09-13 Richard Biener <rguenther@suse.de>
528 PR tree-optimization/87263
529 * tree-ssa-sccvn.c (visit_phi): Revert some earlier changes.
530 (struct unwind_state): Add max_rpo field.
531 (do_rpo_vn): Allow up-to-date loop state to be used when not iterating.
532 Compute max_rpo, the max RPO number a block can be backwards reached
533 from. Re-write non-iterating mode to a RPO ordered worklist approach,
534 separating it from the iterating mode.
536 2018-09-13 Vlad Lazar <vlad.lazar@arm.com>
538 * haifa-sched.c (rank_for_schedule): Schedule by INSN_COST.
539 (rfs_decision): New scheduling decision.
541 2018-09-13 Richard Biener <rguenther@suse.de>
544 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Fix assert.
545 (vn_nary_op_insert_pieces_predicated): Do not write useless
546 valid_dominated_by_p entry outside of the allocated storage.
548 2018-09-13 Omar Sandoval <osandov@osandov.com>
549 Tom de Vries <tdevries@suse.de>
552 * dwarf2out.c (is_c): New function.
553 (add_subscript_info): Add DW_AT_count of 0 for C zero-length arrays.
555 2018-09-13 Sam Tebbs <sam.tebbs@arm.com>
558 * config/aarch64/aarch64.md (*aarch64_bfxil):
560 * config/aarch64/constraints.md (Ulc): Define.
561 * config/aarch64/aarch64-protos.h (aarch64_high_bits_all_ones_p):
563 * config/aarch64/aarch64.c (aarch64_high_bits_all_ones_p):
566 2018-09-13 Vlad Lazar <vlad.lazar@arm.com>
568 * config/aarch64/aarch64.h (TARGET_COMPUTE_FRAME_LAYOUT): Define.
569 * config/aarch64/aarch64.c (aarch64_expand_prologue): Remove
570 aarch64_layout_frame call.
571 (aarch64_expand_epilogue): Likewise.
572 (aarch64_initial_elimination_offset): Likewise.
573 (aarch64_get_separate_components): Likewise.
574 (aarch64_use_return_insn_p): Likewise.
575 (aarch64_layout_frame): Remove unneeded check.
577 2018-09-13 Jakub Jelinek <jakub@redhat.com>
579 * configure.ac: Only append
580 " : (reconfigured) $TOPLEVEL_CONFIGURE_ARGUMENTS" to
581 gcc_config_arguments if it was never reconfigured or last reconfigure
582 was with different arguments.
583 * configure: Regenerated.
585 2018-09-13 Jakub Jelinek <jakub@redhat.com>
586 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
589 * expr.c (maybe_optimize_pow2p_mod_cmp): New function.
590 (maybe_optimize_mod_cmp): Use it if integer_pow2p treeop1.
592 2018-09-13 Jakub Jelinek <jakub@redhat.com>
594 PR tree-optimization/87287
595 * fold-const.c (fold_binary_loc) <case EQ_EXPR>: Move signed modulo
596 X % C == 0 to X % (unsigned) C == 0 optimization to ...
597 * match.pd (X % C == 0): ... here. New optimization.
599 2018-09-12 Jakub Jelinek <jakub@redhat.com>
602 * expr.h (maybe_optimize_mod_cmp): Declare.
603 * expr.c (mod_inv): New function.
604 (maybe_optimize_mod_cmp): New function.
605 (do_store_flag): Use it.
606 * cfgexpand.c (expand_gimple_cond): Likewise.
608 2018-09-09 Cesar Philippidis <cesar@codesourcery.com>
609 Julian Brown <julian@codesourcery.com>
612 * gimplify.c (gimplify_scan_omp_clauses): Set
613 target_firstprivatize_array_bases in OpenACC parallel and kernels
614 region contexts. Remove GOMP_MAP_FIRSTPRIVATE_REFERENCE clauses from
615 OpenACC data regions.
617 2018-09-12 Uros Bizjak <ubizjak@gmail.com>
619 * config/i386/i386.md (sqrt_extend<mode>xf3_i387): Remove.
620 (sqrt<mode>2): Extend operand 1 to XFmode and generate sqrtxf3 insn.
622 2018-09-12 Richard Biener <rguenther@suse.de>
624 PR tree-optimization/87280
625 * tree-ssa-sccvn.c (process_bb): Handle the case of executable
626 edge but unreachable target.
627 (do_rpo_vn): For conservatively handling a PHI only mark
628 the backedge executable but not the block reachable.
630 2018-09-12 Richard Biener <rguenther@suse.de>
632 PR tree-optimization/87266
633 * tree-ssa-sccvn.c (do_rpo_vn): Always iterate to not yet
636 2018-09-12 Andreas Krebbel <krebbel@linux.ibm.com>
638 * config/s390/s390.md (PFPO_RND_MODE_DFP, PFPO_RND_MODE_BFP): New
640 ("trunc<BFP:mode><DFP_ALL:mode>2")
641 ("trunc<DFP_ALL:mode><BFP:mode>2")
642 ("extend<BFP:mode><DFP_ALL:mode>2")
643 ("extend<DFP_ALL:mode><BFP:mode>2"): Set proper rounding mode
644 according to the target operand type.
646 2018-09-12 Jakub Jelinek <jakub@redhat.com>
647 Andreas Krebbel <krebbel@linux.ibm.com>
649 PR tree-optimization/86844
650 * gimple-ssa-store-merging.c
651 (imm_store_chain_info::coalesce_immediate): For overlapping stores, if
652 there are any overlapping stores in between them, make sure they are
653 also coalesced or we give up completely.
655 2018-09-12 Jakub Jelinek <jakub@redhat.com>
658 * fold-const.c (fold_ternary_loc) <case COND_EXPR>: Verify also that
659 BIT_AND_EXPR's second operand is a power of two. Formatting fix.
661 2018-09-12 Tom de Vries <tdevries@suse.de>
663 * common.opt (gdescribe-dies): Add option.
664 * dwarf2out.c (add_name_and_src_coords_attributes): Add description
665 attribute for artifical and nameless decls.
666 (dwarf2out_register_external_die): Add description attribute to
667 external reference die.
668 (add_desc_attribute): New functions.
669 (gen_subprogram_die): Add description attribute to
670 DW_TAG_call_site_parameter.
671 * tree-pretty-print.c (print_generic_expr_to_str): New function.
672 * tree-pretty-print.h (print_generic_expr_to_str): Declare.
673 * doc/invoke.texi (@item Debugging Options): Add -gdescribe-dies and
675 (@item -gdescribe-dies): Add.
677 2018-09-12 Aldy Hernandez <aldyh@redhat.com>
679 * tree-vrp.c (vrp_shift_undefined_p): Remove.
680 (extract_range_from_binary_expr_1: Call
681 wide_int_range_shift_undefined_p instead of vrp_shift_undefined_p.
682 * wide-int-range.h (wide_int_range_shift_undefined_p): Do not
685 2018-09-12 Aldy Hernandez <aldyh@redhat.com>
687 * gimple-ssa-warn-alloca.c
688 (alloca_type_and_limit::alloca_type_and_limit): Initialize limit
689 field for ALLOCA_BOUND_*_LARGE.
691 2018-09-11 Nathan Sidwell <nathan@acm.org>
693 * gcc.c (load_specs, execute, run_attempt): Use %qs not '%s'.
695 2018-09-11 Uros Bizjak <ubizjak@gmail.com>
697 * reg-stack.c (subst_asm_stack_regs): Call replace_reg also
698 for clobbers. Remove obsolete comment.
700 2018-09-11 Uros Bizjak <ubizjak@gmail.com>
702 * config/i386/i386.md (define_attr "type"): Remove mpxmov, mpxmk,
703 mpxchk, mpxld and mpxst types.
704 (define_attr length_immediate): Remove all processing of mpx types.
705 (define_attr prefix_0f): Ditto.
706 (define_attr memory): Ditto.
708 2018-09-11 Uros Bizjak <ubizjak@gmail.com>
710 * config/i386/i386.md (fyl2x_extend<mode>xf3_i387): Remove.
711 (log<mode>2): Change operand 1 predicate to general_operand.
712 Extend operand 1 to XFmode and generate logxf3 insn.
713 (log10<mode>2): Change operand 1 predicate to general_operand.
714 Extend operand 1 to XFmode and generate log10xf3 insn.
715 (log2<mode>2): Change operand 1 predicate to general_operand.
716 Extend operand 1 to XFmode and generate log2xf3 insn.
717 (fyl2xp1_extend<mode>xf3_i387): Remove.
718 (log1p<mode>2): Change operand 1 predicate to general_operand.
719 Extend operand 1 to XFmode and generate log1pxf3 insn.
720 (fxtract_extend<mode>xf3_i387): Remove.
721 (logb<mode>2): Change operand 1 predicate to general_operand.
722 Extend operand 1 to XFmode and generate logbxf3 insn.
723 (ilogb<mode>2): Change operand 1 predicate to general_operand.
724 Extend operand 1 to XFmode and generate fxtractxf3_i387 insn.
725 (significand<mode>2): Change operand 1 predicate to general_operand.
726 Extend operand 1 to XFmode and generate significandxf3 insn.
728 2018-09-11 Nathan Sidwell <nathan@acm.org>
730 * gcc.c (perror_with_name, pfatal_with_name): Delete.
731 (load_specs): Use fatal_error.
732 (DELETE_IF_ORDINARY, process_command): Use error.
733 (execute, run_attempt): Use fatal_error.
735 2018-09-11 Andrew Stubbs <ams@codesourcery.com>
737 * diagnostic-core.h (sorry_at): New prototype.
738 * diagnostic.c (sorry_at): New function.
740 2018-09-11 Aldy Hernandez <aldyh@redhat.com>
742 * tree-vrp.c (extract_range_from_binary_expr_1): Treat all divisions
743 by zero as VR_UNDEFINED.
745 2018-09-10 Uros Bizjak <ubizjak@gmail.com>
747 * config/i386/i386.md (<sincos>xf2): Rename from *<sincos>xf2_i387.
748 (*<sincos>_extend<mode>xf2_i387): Remove insn pattern.
749 (<sincos>mode2): New expander.
750 (sincos_extend<mode>xf3_i387): Remove insn pattern.
751 (sincos -> sin, cos splitters): Remove splitter patterns.
752 (sincos<mode>3): Change operand 2 predicate to general_operand.
753 Extend operand 2 to XFmode and generate sincosxf3 insn.
754 (fptanxf4_i387): Change mode of operands 0 and 3 to SFmode.
755 Change operand 3 predicate to const1_operand.
756 (fptan_extend<mode>xf4_i387): Remove insn pattern.
757 (tanxf2): Update operands in the call to fptanxf4_i387.
758 (tan<mode>2): Change operand 1 predicate to general_operand.
759 Extend operand 1 to XFmode and generate tanxf3 insn.
760 (atan2xf3): Rename from *fpatanxf3_i387.
761 (fpatan_extend<mode>xf3_i387): Remove insn pattern.
762 (atan2xf3): Remove expander.
763 (atan2<mode<3): Change operand 1 and 2 predicates to general_operand.
764 Extend operands 1 and 2 to XFmode and generate atan2xf3 insn.
765 (atan<mode>2): Change operand 1 predicate to general_operand.
766 Extend operand 1 to XFmode and generate atanxf3 insn.
768 2018-09-10 Uros Bizjak <ubizjak@gmail.com>
770 * config/i386/i386.md (x87/SSE constant load splitter): Use
771 memory_operand instead of nonimmediate_operand for input operand
774 2018-09-09 Uros Bizjak <ubizjak@gmail.com>
776 * config/i386/i386.md (float partial SSE register stall splitter): Move
777 splitter near its instruction pattern.
778 (float_extend partial SSE register stall splitter): Ditto.
779 (float_truncate partial SSE register stall splitter): Ditto.
781 2018-09-09 Hans-Peter Nilsson <hp@bitrange.com>
784 * config/mmix/mmix.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Redefine
785 to speculation_safe_value_not_needed.
788 * config/mmix/mmix.c (mmix_assemble_integer): Handle byte-size
789 non-CONST_INT rtx:es using assemble_integer_with_op ".byte".
790 (MMIX_CFUN_NEEDS_SAVED_EH_RETURN_ADDRESS): Don't call
791 leaf_function_p, instead use has_hard_reg_initial_val.
793 2018-09-09 Nathan Sidwell <nathan@acm.org>
795 * gcc.h (pfatal_with_name): Don't declare here.
796 * gcc.c (pfatal_with_name): Make static.
798 2018-09-09 Xianmiao Qu <xianmiao_qu@c-sky.com>
800 * config/csky/csky.md (*cskyv2_adddi3, *cskyv2_subdi3): Add
803 2018-09-08 John David Anglin <danglin@gcc.gnu.org>
805 PR rtl-optimization/85458
806 * sel-sched.c (sel_target_adjust_priority): Allow backend adjust
807 priority hook to reduce the priority of EXPR.
809 2018-09-07 Uros Bizjak <ubizjak@gmail.com>
811 * config/i386/i386.md (float<SWI48x:mode><MODEF:mode>2) Enable
812 DImode for x87 on 32bit targets. Conditionally disable x87 modes
813 with X87_ENABLE_FLOAT. Remove preparation code.
814 (*float<SWI48:mode><MODEF:mode>2): Rename from
815 *float<SWI48:mode><MODEF:mode>2_mixed. Handle x87, SSE and mixed
816 math using "enabled" attribute.
817 (*floatdi<MODEF:mode>2_i387): Rename from
818 *float<SWI48x:mode><MODEF:mode>2_i387. Handle only DImode and
819 enable for 32bit targets only.
820 (floatdi<X87MODEF:mode>2_i387_with_xmm pre-reload splitter): New
822 (floatdi<X87MODEF:mode>2_i387_with_xmm): Use register_operand
823 as operand 1 predicate. Rewrite as define_insn_and_split.
824 (floatdi<X87MODEF:mode>2_i387_with_xmm memory input splitter): Remove.
826 2018-09-06 Uros Bizjak <ubizjak@gmail.com>
828 * reg-stack.c (get_true_reg) <case FLOAT_TRUNCATE>: Reorder
829 to fallthru to FLOAT case.
831 2018-09-06 Will Schmidt <will_schmidt@vnet.ibm.com>
834 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Update logic
835 around folding of vec_sl to handle out of range shift values.
837 2018-09-06 Uros Bizjak <ubizjak@gmail.com>
839 * config/i386/i386.md (fix_trunc<mode>_fisttp_i387_1): Remove.
840 Update callers to gen_fix_trunc<mode>_i387_fisttp
841 (fix_trunc<mode>_i387_fisttp): Change operand 0 predicate to
842 nonimmediate_operand.
843 (fix_trunc<mode>_i387_fisttp_with_temp): Remove insn pattern
844 and corresponding splitters.
845 (*fix_trunc<mode>_i387_1): Always expand with fix_trunc<mode>_i387.
846 (fix_truncdi_i387): Change operand 0 predicate to nonimmediate_operand.
847 (fix_truncdi_i387_with_temp): Remove insn pattern
848 and corresponding splitters.
849 (fix_trunc<mode>_i387): Change operand 0 predicate to
850 nonimmediate_operand.
851 (fix_trunc<mode>_i387_with_temp): Remove insn pattern
852 and corresponding splitters.
853 (*fistdi2_1): Remove.
855 (fistdi2_with_temp): Remove insn pattern and corresponding splitters.
856 (lrintxfdi2): Remove expander. Reimplement as define_insn.
857 (*fist<mode>2_1): Remove.
858 (fist<mode>2): Ditto.
859 (fist<mode>2_with_temp): Remove insn pattern and corresponding
861 (lrintxf<mode>2): Remove expander. Reimplement as define_insn.
862 (*fist<mode>2_<rounding>_1): Always expand with fist<mode>2_<rounding>.
863 (fistdi2_<rounding>): Change operand 0 predicate to
864 nonimmediate_operand.
865 (fistdi2_<rounding>_with_temp): Remove insn pattern
866 and corresponding splitters.
867 (fist<mode>2_<rounding>): Change operand 0 predicate to
868 nonimmediate_operand.
869 (fist<mode>2_<rounding>_with_temp): Remove insn pattern
870 and corresponding splitters.
872 (*fixuns_trunc<mode>si2_avx512f_zext): Depend on TARGET_SSE_MATH.
874 2018-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
876 * varasm.c (output_constructor_regular_field): Check TYPE_SIZE_UNIT of
879 2018-09-06 Will Schmidt <will_schmidt@vnet.ibm.com>
881 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
882 early gimple folding of vec_splat().
883 * tree-vect-generic.c: Remove static from tree_vec_extract() definition.
884 * gimple-fold.h: Add an extern define for tree_vec_extract().
886 2018-09-06 Will Schmidt <will_schmidt@vnet.ibm.com>
888 * config/rs6000/rs6000.c (fold_mergehl_helper): Add types_compatible_p
889 wrappers around TREE_TYPE comparisons.
891 2018-09-06 Ilya Leoshkevich <iii@linux.ibm.com>
894 * config/s390/predicates.md: Add nonsym_memory_operand.
895 * config/s390/s390.c (s390_legitimize_cs_operand): If operand
896 contains a SYMBOL_REF, load it into an intermediate pseudo.
897 (s390_emit_compare_and_swap): Legitimize operand.
898 * config/s390/s390.md: Use the new nonsym_memory_operand
899 with UNSPECV_CAS patterns.
901 2018-09-06 Ilya Leoshkevich <iii@linux.ibm.com>
904 * config/s390/s390-passes.def: New file.
905 * config/s390/s390-protos.h (class rtl_opt_pass): Add forward
907 (make_pass_s390_early_mach): Add declaration.
908 * config/s390/s390.c (make_pass_s390_early_mach):
909 (s390_option_override): Remove dynamic registration.
910 * config/s390/t-s390: Add s390-passes.def.
912 2018-09-06 Ilya Leoshkevich <iii@linux.ibm.com>
914 * config/s390/s390.c (s390_decompose_constant_pool_ref):
915 Remove UNSPEC_LTREL_BASE check.
916 (annotate_constant_pool_refs): Likewise.
917 (find_constant_pool_ref): Likewise.
918 (find_ltrel_base): Removed.
919 (replace_ltrel_base): Removed.
920 (s390_mainpool_finish): Remove replace_ltrel_base call.
921 (s390_chunkify_start): Remove pending LTREL_BASE logic.
922 (s390_chunkify_finish): Remove replace_ltrel_base call.
923 * config/s390/s390.md: Remove UNSPEC_LTREL_BASE.
925 2018-09-06 Hans-Peter Nilsson <hp@axis.com>
928 * config/cris/cris.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Redefine
929 to speculation_safe_value_not_needed.
931 2018-09-05 Cesar Philippidis <cesar@codesourcery.com>
932 Bernd Schmidt <bernds_cb1@t-online.de>
934 * config/nvptx/nvptx-opts.h: New file.
935 * config/nvptx/nvptx.c (nvptx_file_start): Print the correct .target.
936 * config/nvptx/nvptx.h: Include "nvptx-opts.h".
938 (TARGET_SM35): New macro.
939 * config/nvptx/nvptx.md (atomic_fetch_<logic><mode>): Enable with the
941 * config/nvptx/nvptx.opt (ptx_isa, sm_30, sm_35): New enum and its
944 * doc/invoke.texi (Nvidia PTX Options): Document -misa.
946 2018-09-05 Uros Bizjak <ubizjak@gmail.com>
948 * config/i386/i386.md (truncdfsf2): Remove expander.
949 (truncdfsf2_with_temp): Ditto.
950 (truncxf<mode>2): Ditto.
951 (*truncdfsf_fast_mixed): Remove insn pattern.
952 (*truncdfsf_fast_i387): Ditto.
953 (*truncdfsf_mixed): Ditto.
954 (*truncdfsf_i387): Ditto.
955 (*truncdfsf2_i387_1): Ditto.
956 (*truncxfsf2_mixed): Ditto.
957 (*truncxfdf2_mixed): Ditto.
958 (*truncxf<mode>2_i387_noop): Ditto. Update callers
959 to call gen_truncxf<mode>2 instead.
960 (*truncxf<mode>2_i387): Remove.
961 (reg->reg splitters): Remove splitter pattern.
962 (reg->mem splitters): Ditto.
964 (truncdfsf2): New insn pattern.
965 (truncxf<mode>2): Ditto.
967 2018-09-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
969 * tree-ssa-math-opts.c (is_mult_by): New function.
970 (is_square_of): Use the above.
971 (optimize_recip_sqrt): New function.
972 (pass_cse_reciprocals::execute): Use the above.
974 2018-09-05 Richard Biener <rguenther@suse.de>
977 * tree-ssa-sccvn.c (rpo_elim::eliminate_push_avail): Make sure
978 to zero-init the emplaced vec.
980 2018-09-05 Martin Liska <mliska@suse.cz>
982 PR tree-optimization/87205
983 * tree-switch-conversion.c (pass_lower_switch::execute):
984 Group cases for switch statements.
986 2018-09-05 Richard Biener <rguenther@suse.de>
988 PR tree-optimization/87217
989 * tree-ssa-sccvn.c (vuse_valueize): New.
990 (vn_reference_lookup_pieces): Use it.
991 (vn_reference_lookup): Likewise.
993 2018-09-05 Nathan Sidwell <nathan@acm.org>
996 * stor-layout.c (place_field): Scan forwards to check last
997 bitfield when ms_bitfield_placement is in effect.
999 2018-09-05 Richard Biener <rguenther@suse.de>
1002 * tree-vect-stmts.c (vectorizable_simd_clone_call): Fix bogus
1005 2018-09-05 Siddhesh Poyarekar <siddhesh@sourceware.org>
1006 Kugan Vivekanandarajah <kugan.vivekanandarajah@linaro.org>
1008 * config/aarch64/falkor-tag-collision-avoidance.c: New file.
1009 * config.gcc (extra_objs): Build it.
1010 * config/aarch64/t-aarch64 (falkor-tag-collision-avoidance.o):
1012 * config/aarch64/aarch64-passes.def
1013 (pass_tag_collision_avoidance): New pass.
1014 * config/aarch64/aarch64.c (qdf24xx_tunings): Add
1015 AARCH64_EXTRA_TUNE_RENAME_LOAD_REGS to tuning_flags.
1016 (aarch64_classify_address): Remove static qualifier.
1017 (aarch64_address_info, aarch64_address_type): Move to...
1018 * config/aarch64/aarch64-protos.h: ... here.
1019 (make_pass_tag_collision_avoidance): New function.
1020 * config/aarch64/aarch64-tuning-flags.def (rename_load_regs):
1023 2018-09-05 Martin Liska <mliska@suse.cz>
1025 * doc/gcov.texi: Update documentation of humar
1027 * gcov.c (format_count): Print one decimal place, it provides
1028 more fine number of situations like '1G' vs. '1.4G'.
1030 2018-09-05 Martin Liska <mliska@suse.cz>
1033 * config/rs6000/rs6000.opt: Mark the option as Deprecated.
1034 * optc-gen.awk: Allow 'Var' for Deprecated options in order
1035 to generate a MASK value.
1037 2018-09-04 H.J. Lu <hongjiu.lu@intel.com>
1040 * dwarf2out.c (based_loc_descr): Allow hard frame pointer even
1041 if frame pointer isn't used.
1042 (compute_frame_pointer_to_fb_displacement): Likewise.
1044 2018-09-04 Jakub Jelinek <jakub@redhat.com>
1047 * common/config/i386/i386-common.c (OPTION_MASK_ISA_XSAVEOPT_SET,
1048 OPTION_MASK_ISA_XSAVES_SET, OPTION_MASK_ISA_XSAVEC_SET): Use
1049 OPTION_MASK_ISA_XSAVE_SET instead of OPTION_MASK_ISA_XSAVE.
1050 (OPTION_MASK_ISA_XSAVE_UNSET): Add OPTION_MASK_ISA_XSAVES_UNSET
1051 and OPTION_MASK_ISA_XSAVEC_UNSET.
1053 2018-09-04 Max Filippov <jcmvbkbc@gmail.com>
1055 * config/xtensa/xtensa.c (xtensa_expand_atomic): Reorder AND and
1056 XOR operations in NAND case.
1058 2018-09-04 Aldy Hernandez <aldyh@redhat.com>
1060 * wide-int-range.cc (wide_int_range_convert): New.
1061 * wide-int-range.h (wide_int_range_convert): New.
1062 * tree-vrp.c (extract_range_from_unary_expr): Abstract wide int
1063 code into wide_int_range_convert.
1064 (extract_range_into_wide_ints): Do not munge anti range constants
1065 into the entire domain. Just return the range back.
1067 2018-09-04 Martin Liska <mliska@suse.cz>
1069 * genmatch.c (output_line_directive): Add new argument
1071 (dt_simplify::gen_1): Encapsulate dump within __builtin_expect.
1073 2018-09-04 Jonathan Wakely <jwakely@redhat.com>
1075 * doc/invoke.texi (Option Summary): Add whitespace.
1077 * doc/invoke.texi (Option Summary): Add -Waligned-new.
1079 2018-09-04 Richard Biener <rguenther@suse.de>
1081 PR tree-optimization/87211
1082 * tree-ssa-sccvn.c (visit_phi): When value-numbering to a
1083 backedge value we're supposed to treat as VARYING also number
1084 the PHI to VARYING in case it got a different value-number already.
1086 2018-09-04 Aldy Hernandez <aldyh@redhat.com>
1088 * tree-vrp.c (vrp_can_optimize_bit_op): Remove.
1089 (extract_range_from_binary_expr_1): Do not call
1090 vrp_can_optimize_bit_op.
1091 * wide-int-range.cc (wide_int_range_can_optimize_bit_op): Make
1093 (wide_int_range_get_mask_and_bounds): New.
1094 (wide_int_range_optimize_bit_op): New.
1095 (wide_int_range_bit_ior): Call wide_int_range_optimize_bit_op.
1096 (wide_int_range_bit_and): Same.
1097 * wide-int-range.h (wide_int_range_can_optimize_bit_op): Remove.
1098 (wide_int_range_optimize_bit_op): New.
1099 (wide_int_range_get_mask_and_bounds): New.
1101 2018-09-04 Richard Biener <rguenther@suse.de>
1103 PR tree-optimization/87176
1104 * tree-ssa-sccvn.c (visit_phi): Remove redundant allsame
1105 variable. When value-numbering a virtual PHI node make sure
1106 to not value-number to the backedge value.
1108 2018-09-04 Jonathan Wakely <jwakely@redhat.com>
1110 * doc/extend.texi (Long Long, Hex Floats): Document support for
1111 long long and hex floats in more recent versions of ISO C++.
1113 2018-09-03 Richard Biener <rguenther@suse.de>
1115 PR tree-optimization/87177
1116 * tree-ssa-sccvn.c (vuse_ssa_val): Revert previous change, keep
1119 2018-09-03 Alexander Monakov <amonakov@ispras.ru>
1121 * bb-reorder.c (edge_order): Convert to C-qsort-style
1122 tri-state comparator.
1123 (reorder_basic_blocks_simple): Change std::stable_sort to gcc_stablesort.
1125 2018-09-03 Alexander Monakov <amonakov@ispras.ru>
1127 * tree-loop-distribution.c (offset_cmp): Convert to C-qsort-style
1128 tri-state comparator.
1129 (fuse_memset_builtins): Change std::stable_sort to gcc_stablesort.
1131 2018-09-03 Alexander Monakov <amonakov@ispras.ru>
1133 * sort.cc (struct sort_ctx): New field 'nlim'. Use it...
1134 (mergesort): ... here as maximum count for using netsort.
1135 (gcc_qsort): Set nlim to 3 if stable sort is requested.
1136 (gcc_stablesort): New.
1137 * system.h (gcc_stablesort): Declare.
1139 2018-09-03 Alexander Monakov <amonakov@ispras.ru>
1141 * sort.cc (gcc_qsort) [CHECKING_P]: Call qsort_chk.
1142 * system.h (qsort): Always redirect to gcc_qsort. Update comment.
1143 * vec.c (qsort_chk): Do not call gcc_qsort. Update comment.
1145 2018-09-03 Segher Boessenkool <segher@kernel.crashing.org>
1147 * config/rs6000/rs6000.md (*mov<mode>_hardfloat32): Remove %U from the
1148 lxsdx and stxsdx alternatives.
1149 (*mov<mode>_hardfloat64): Ditto.
1150 * config/rs6000/vsx.md (*vsx_extract_<mode>_store): Ditto.
1152 2018-09-03 Richard Biener <rguenther@suse.de>
1154 PR tree-optimization/87200
1155 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Valueize a
1158 2018-09-03 Martin Liska <mliska@suse.cz>
1160 PR tree-optimization/87201
1161 * tree-switch-conversion.c (switch_decision_tree::balance_case_nodes):
1162 Fix parenthesis in an expression.
1164 2018-09-03 Richard Biener <rguenther@suse.de>
1166 PR tree-optimization/87197
1167 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Mark the new def
1168 visited. CSE the VN_INFO hashtable lookup.
1170 PR tree-optimization/87169
1171 * tree-ssa-sccvn.c (do_rpo_vn): When marking loops for not
1172 iterating make sure there's no extra backedges from irreducible
1173 regions feeding the header. Mark the destination block
1176 2018-09-03 Martin Liska <mliska@suse.cz>
1179 * common/common-target.def: Add TARGET_GET_VALID_OPTION_VALUES.
1180 * common/common-targhooks.c (default_get_valid_option_values):
1182 * common/common-targhooks.h (default_get_valid_option_values):
1184 * common/config/i386/i386-common.c: Move processor_target_table
1186 (ix86_get_valid_option_values): New function.
1187 (TARGET_GET_VALID_OPTION_VALUES): New macro.
1188 * config/i386/i386.c (struct ptt): Move to i386-common.c.
1189 (PTA_*): Move all defined masks into i386-common.c.
1190 (ix86_function_specific_restore): Use new processor_cost_table.
1191 * config/i386/i386.h (struct ptt): Moved from i386.c.
1192 (struct pta): Likewise.
1193 * doc/tm.texi: Document new TARGET_GET_VALID_OPTION_VALUES.
1194 * doc/tm.texi.in: Likewise.
1195 * opt-suggestions.c (option_proposer::suggest_option):
1196 Pass prefix to build_option_suggestions.
1197 (option_proposer::get_completions): Likewise.
1198 (option_proposer::build_option_suggestions): Use the new target
1200 * opts.c (struct option_help_tuple): New struct.
1201 (print_filtered_help): Use the new target hook.
1203 2018-09-03 Martin Liska <mliska@suse.cz>
1206 * predict.c (set_even_probabilities): Add likely_edges
1207 argument and handle cases where we have precisely one
1209 (combine_predictions_for_bb): Catch also likely_edges.
1210 (tree_predict_by_opcode): Handle gswitch statements.
1211 * tree-cfg.h (find_case_label_for_value): New declaration.
1212 (find_taken_edge_switch_expr): Likewise.
1213 * tree-switch-conversion.c (switch_decision_tree::balance_case_nodes):
1214 Find pivot in decision tree based on probabily, not by number of
1217 2018-09-02 Gerald Pfeifer <gerald@pfeifer.com>
1219 * doc/standards.texi (Standards): Update Objective-C reference.
1221 2018-09-01 Gerald Pfeifer <gerald@pfeifer.com>
1223 * doc/install.texi (Prerequisites): Update link for MPC.
1225 2018-09-01 Michael Matz <matz@suse.de>
1227 PR tree-optimization/87074
1228 * gimple-loop-jam.c (unroll_jam_possible_p): Check loop exit
1229 PHIs for outer-loop uses.
1231 2018-09-01 Gerald Pfeifer <gerald@pfeifer.com>
1233 * doc/generic.texi (OpenMP): Adjust link to openmp.org.
1234 * doc/invoke.texi (C Dialect Options): Ditto.
1236 2018-09-01 Gerald Pfeifer <gerald@pfeifer.com>
1238 * doc/install.texi (Prerequisites): Adjust link mpfr.org.
1240 2018-08-31 Richard Biener <rguenther@suse.de>
1242 PR tree-optimization/87168
1243 * tree-ssa-sccvn.c (SSA_VAL): Add visited output parameter.
1244 (rpo_elim::eliminate_avail): When OP was not visited it must
1247 2018-08-31 David Malcolm <dmalcolm@redhat.com>
1249 * tree-vrp.c (copy_value_range): Convert param "from" from
1250 "value_range *" to "const value_range *".
1251 (range_is_null): Likewise for param "vr".
1252 (range_int_cst_p): Likewise.
1253 (range_int_cst_singleton_p): Likewise.
1254 (symbolic_range_p): Likewise.
1255 (value_ranges_intersect_p): Likewise for both params.
1256 (value_range_nonnegative_p): Likewise for param "vr".
1257 (value_range_constant_singleton): Likewise.
1258 (vrp_set_zero_nonzero_bits): Likewise for param "ar".
1259 (extract_range_into_wide_ints): Likewise for param "vr".
1260 (extract_range_from_multiplicative_op): Likewise for params "vr0"
1262 (vrp_can_optimize_bit_op): Likewise.
1263 (extract_range_from_binary_expr_1): Likewise for params "vr0_" and
1265 (extract_range_from_unary_expr): Likewise.
1266 (debug_value_range): Likewise for param "vr".
1267 (value_range::dump): Add "const" qualifier.
1268 (vrp_prop::check_array_ref): Convert local "vr" from
1269 "value_range *" to "const value_range *".
1270 (vrp_prop::check_mem_ref): Likewise.
1271 (vrp_prop::visit_stmt): Likewise for local "old_vr".
1272 (vrp_intersect_ranges_1): Likewise for param "vr_1".
1273 (vrp_intersect_ranges): Likewise.
1274 (simplify_stmt_for_jump_threading): Likewise for local "vr".
1275 (vrp_prop::vrp_finalize): Likewise.
1276 * tree-vrp.h (value_range::dump): Add "const" qualifier.
1277 (vrp_intersect_ranges): Add "const" qualifier to params as above.
1278 (extract_range_from_unary_expr): Likewise.
1279 (value_range_constant_singleton): Likewise.
1280 (symbolic_range_p): Likewise.
1281 (copy_value_range): Likewise.
1282 (extract_range_from_binary_expr_1): Likewise.
1283 (range_int_cst_p): Likewise.
1284 (vrp_set_zero_nonzero_bits): Likewise.
1285 (range_int_cst_singleton_p): Likewise.
1287 2018-08-31 Vlad Lazar <vlad.lazar@arm.com>
1289 * config/aarch64/arm_neon.h (vabsd_s64): New.
1290 (vnegd_s64): Likewise.
1292 2018-08-31 Martin Jambor <mjambor@suse.cz>
1294 * ipa-cp.c (estimate_local_effects): Replace wrong MAX with MIN.
1296 2018-08-31 Martin Liska <mliska@suse.cz>
1298 * ipa-icf.c (sem_item::add_type): Use
1299 sem_item::m_type_hash_cache.
1300 * ipa-icf.h: Move the cache from sem_item_optimizer
1303 2018-08-31 Nathan Sidwell <nathan@acm.org>
1305 * doc/extend.texi (Backwards Compatibility): Remove implicit
1306 extern C leeway of () being (...).
1308 2018-08-31 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1310 * ipa-inline.c (can_inline_edge_by_limits_p): Fix typos in comment.
1312 2018-08-31 Segher Boessenkool <segher@kernel.crashing.org>
1316 * config/rs6000/rs6000.md (lround<mode>di2): Gate on TARGET_FPRND.
1318 2018-08-31 Jakub Jelinek <jakub@redhat.com>
1321 * expmed.c (expand_mult_const): Use immed_wide_int_const instead of
1322 gen_int_mode. Formatting fixes.
1324 2018-08-30 Sandra Loosemore <sandra@codesourcery.com>
1326 * target.def (custom_function_descriptors): Improve documentation.
1327 * doc/tm.texi.in (Trampolines): Expand discussion of function
1328 descriptors and move TARGET_CUSTOM_FUNCTION_DESCRIPTORS to the
1329 beginning of the section.
1330 * doc/tm.texi: Regenerated.
1332 2018-08-30 Jose E. Marchesi <jose.marchesi@oracle.com>
1334 * cfg.h (class auto_edge_flag): Spell out the template-id of the
1335 base class in the initializer list. This is a workaround for
1336 building with older GCC.
1337 (class auto_bb_flag): Likewise.
1339 2018-08-30 Aaron Sawdey <acsawdey@linux.ibm.com>
1341 * config/rs6000/altivec.md (altivec_eq<mode>): Remove star.
1342 (altivec_vcmpequ<VI_char>_p): Remove star.
1343 * config/rs6000/rs6000-string.c (do_load_for_compare): Support
1345 (expand_strncmp_vec_sequence): New function.
1346 (emit_final_str_compare_vec): New function.
1347 (expand_strn_compare): Add support for vector strncmp.
1348 * config/rs6000/rs6000.opt (-mstring-compare-inline-limit): Change
1349 length specification to bytes.
1350 * config/rs6000/vsx.md (vsx_ld_elemrev_v16qi_internal): Remove star.
1351 (vcmpnezb_p): New pattern.
1352 * doc/invoke.texi (RS/6000 and PowerPC Options): Update documentation
1353 for option -mstring-compare-inline-limit.
1355 2018-08-30 Thiago Macieira <thiago.macieira@intel.com>
1357 * config/i386/i386.c (PTA_WESTMERE): Remove PTA_AES.
1358 (PTA_SKYLAKE): Add PTA_AES.
1359 (PTA_GOLDMONT): Likewise.
1361 2018-08-29 Jan Hubicka <jh@suse.cz>
1364 * lto-opts.c (lto_write_options): Always stream PIC/PIE mode.
1365 * lto-wrapper.c (merge_and_complain): Fix merging of PIC/PIE.
1367 2018-08-29 Jan Hubicka <jh@suse.cz>
1369 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not follow
1371 (hash_tree): Do not visit TYPE_STUB_DECL.
1372 * tree-streamer-out.c (write_ts_type_common_tree_pointers): Do not
1373 stream TYPE_STUB_DECL.
1374 * tree-streamer-in.c (lto_input_ts_type_common_tree_pointers): Likewise.
1375 * ipa-utils.h (type_with_linkage_p): Do not rely on TYPE_STUB_DECL
1376 after free_lang_data.
1377 (type_in_anonymous_namespace_p): Likewise.
1379 2018-08-29 Jan Hubicka <jh@suse.cz>
1381 * sreal.h (SREAL_PART_BITS): Change to 31; remove seemingly unnecessary
1382 comment that it has to be even number.
1383 (class sreal): Change m_sig type to int32_t.
1384 * sreal.c (sreal::dump, sreal::to_int, opreator+, operator-): Use
1385 int64_t for temporary calculations.
1386 (sreal_verify_basics): Drop one bit from minimum and maximum.
1388 2018-08-30 Richard Biener <rguenther@suse.de>
1390 PR tree-optimization/87147
1391 * tree-ssa-sccvn.c (SSA_VISITED): New function.
1392 (visit_phi): When the degenerate result is from the backedge and
1393 we didn't visit its definition yet drop to VARYING.
1394 (do_rpo_vn): Properly mark blocks with incoming backedges as executable.
1396 2018-08-29 Jan Hubicka <jh@suse.cz>
1398 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not walk
1400 (hash_tree): Likewise.
1402 2018-08-29 Jan Hubicka <jh@suse.cz>
1404 * tree.c (find_decls_types_r): Walk also TYPE_NEXT_PTR_TO
1405 and TYPE_NEXT_REF_TO.
1407 2018-08-29 Jan Hubicka <jh@suse.cz>
1409 * sreal.h (SREAL_PART_BITS): Change to 31; remove seemingly unnecessary
1410 comment that it has to be even number.
1411 (class sreal): Change m_sig type to int32_t.
1412 * sreal.c (sreal::dump, sreal::to_int, opreator+, operator-): Use
1413 int64_t for temporary calculations.
1414 (sreal_verify_basics): Drop one bit from minimum and maximum.
1416 2018-08-30 Tamar Christina <tamar.christina@arm.com>
1418 * config/aarch64/aarch64.c (aarch64_expand_movmem): Set TImode max.
1420 2018-08-30 Vlad Lazar <vlad.lazar@arm.com>
1423 * expmed.c (canonicalize_comparison): Use wi::sub instead of wi::add
1424 if to_add is negative.
1426 2018-08-29 Bernd Edlinger <bernd.edlinger@hotmail.de>
1429 * builtins.c (c_strlen): Improve range checks.
1431 2018-08-29 Martin Sebor <msebor@redhat.com>
1432 Jeff Law <law@redhat.com>
1434 PR tree-optimization/86714
1435 PR tree-optimization/86711
1436 * builtins.c (c_strlen): Add arguments to call to string_constant.
1437 * expr.c (string_constant): Add argument. Detect missing nul
1438 terminator and outermost declaration it's missing in.
1439 * expr.h (string_constant): Add argument.
1440 * fold-const.c (read_from_constant_string): Add arguments to call to
1442 (c_getstr): Likewise.
1443 * tree-ssa-forwprop.c (simplify_builtin_call): Likewise.
1445 * tree-ssa-strlen.c (get_stridx): Likewise.
1447 2018-08-29 Jan Hubicka <jh@suse.cz>
1449 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers):
1450 Do not stream DECL_VINDEX.
1451 * tree-streamer-out.c (write_ts_function_decl_tree_pointers): Likewise.
1452 * tree.c (free_lang_data_in_decl): Clear DECL_VINDEX.
1453 (decl_function_context): Use DECL_VIRTUAL_P rather than DECL_VINDEX.
1455 2018-08-29 Richard Biener <rguenther@suse.de>
1457 * tree-ssa-sccvn.c (vuse_ssa_val): Return NULL for unvisited
1458 virtual operands that are not default defs to honor region
1460 (rpo_vn_valueize): Remove ineffective code here.
1462 2018-08-29 Richard Biener <rguenther@suse.de>
1464 PR tree-optimization/87132
1465 * tree-ssa-alias.c (get_continuation_for_phi): Do not translate
1466 when skipping defs reachable over backedges.
1468 2018-08-29 Richard Biener <rguenther@suse.de>
1470 * tree-core.h: Document use of deprecated_flag in SSA_NAME.
1471 * tree.h (SSA_NAME_POINTS_TO_READONLY_MEMORY): Define.
1472 * tree-into-ssa.c (pass_build_ssa::execute): Initialize
1473 function parameters SSA_NAME_POINTS_TO_READONLY_MEMORY from fnspec.
1474 * tree-ssa-sccvn.c (const_parms, init_const_parms): Remove.
1475 (vn_reference_lookup_3): Remove use of const_parms.
1476 (free_rpo_vn): Do not free const_parms.
1477 (do_rpo_vn): Do not call init_const_parms.
1478 * tree-ssa-alias.c (refs_may_alias_p_1): Honor
1479 SSA_NAME_POINTS_TO_READONLY_MEMORY.
1480 (call_may_clobber_ref_p_1): Likewise.
1482 2018-08-29 Alexander Monakov <amonakov@ispras.ru>
1485 * invoke.texi (Optimization Options): List -ftree-scev-cprop.
1487 (-ftree-scev-cprop): Document.
1489 2018-08-29 Jan Hubicka <jh@suse.cz>
1491 * sreal.h (normalize, normalize_up, normalize_down): Add new_sig/new_exp
1493 (sreal constructor): Update.
1494 * sreal.c (sreal:operator+, sreal:operator-, sreal:operator*,
1495 sreal:operator/): Update.
1497 2018-08-29 Martin Liska <mliska@suse.cz>
1499 * tree-switch-conversion.c (switch_conversion::expand):
1500 Strenghten assumption about gswitch statements.
1502 2018-08-29 Richard Biener <rguenther@suse.de>
1504 PR tree-optimization/87117
1505 * tree-ssa-sccvn.c (eliminate_dom_walker::eliminate_stmt): Only
1506 re-value-number released SSA VDEFs.
1508 2018-08-29 Richard Biener <rguenther@suse.de>
1510 PR tree-optimization/87126
1511 * tree-ssa-sccvn.c (vn_reference_insert): Remove assert.
1513 2018-08-28 Jim Wilson <jimw@sifive.com>
1515 * config/riscv/pic.md: Rewrite.
1516 * config/riscv/riscv.c (riscv_address_insns): Return cost of 3 for
1518 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Delete.
1519 (SOFTF, default_load, softload, softstore): New.
1521 2018-08-28 Jeff Law <law@redhat.com>
1523 * fold-const.c (fold_binary_loc): Remove recently added assert.
1525 2018-08-28 Joern Rennecke <joern.rennecke@riscy-ip.com>
1527 * genpreds.c (write_predicate_subfunction): Also add ATTRIBUTE_UNUSED
1528 to OP parmeter of generated function.
1530 2018-08-28 MCC CS <deswurstes@users.noreply.github.com>
1532 PR tree-optimization/87009
1533 * match.pd: Add boolean optimizations.
1535 2018-08-28 Martin Sebor <msebor@redhat.com>
1538 * calls.c (alloc_max_size): Treat HOST_WIDE_INT special.
1539 * gimple-ssa-warn-alloca.c (adjusted_warn_limit): New function.
1540 (pass_walloca::gate): Use it.
1541 (alloca_call_type): Same.
1542 (pass_walloca::execute): Same.
1543 * stor-layout.c (layout_decl): Treat HOST_WIDE_INT special.
1545 2018-08-28 David Malcolm <dmalcolm@redhat.com>
1547 * dumpfile.h (ATTRIBUTE_GCC_DUMP_PRINTF): Change version check on
1548 GCC_VERSION for usage of "__gcc_dump_printf__" format from
1551 2018-08-28 Richard Biener <rguenther@suse.de>
1553 PR tree-optimization/87124
1554 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Guard against
1555 constants before looking up avail.
1557 2018-08-28 Jakub Jelinek <jakub@redhat.com>
1560 * calls.c (maybe_warn_nonstring_arg): Punt early if
1561 warn_stringop_overflow is zero. Don't call get_range_strlen
1562 on 3rd argument, keep iterating until lenrng[1] is INTEGER_CST.
1563 Swap comparison operands to have constants on rhs. Only use
1564 lenrng[1] if non-NULL and INTEGER_CST. Don't uselessly
1565 increment lenrng[0].
1567 2018-08-28 Richard Sandiford <richard.sandiford@arm.com>
1569 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Fix unguarded
1570 use of tree_to_shwi. Remove duplicated test for the size being
1571 a whole number of bytes.
1573 2018-08-28 Richard Biener <rguenther@suse.de>
1575 PR tree-optimization/87117
1576 * tree-ssa-sccvn.c (eliminate_dom_walker::eliminate_cleanup):
1577 Handle removed stmt without LHS (GIMPLE_NOP).
1579 2018-08-28 Richard Biener <rguenther@suse.de>
1581 PR tree-optimization/87117
1582 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Exclude
1583 void which is is_gimple_reg_type by checking for COMPLETE_TYPE_P.
1585 2018-08-28 Richard Biener <rguenther@suse.de>
1587 PR tree-optimization/87117
1588 * tree-ssa-pre.c (compute_avail): Do not make expressions
1589 with predicated values available.
1590 (get_expr_value_id): Assert we do not run into predicated value
1593 2018-08-28 Richard Biener <rguenther@suse.de>
1595 PR tree-optimization/87117
1596 * tree-ssa-operands.c (add_stmt_operand): STRING_CST may
1597 get virtual operands.
1598 (get_expr_operands): Handle STRING_CST like other decls.
1600 2018-08-28 Martin Liska <mliska@suse.cz>
1602 * tree.h: Update documentation of fndecl_built_in_p
1606 2018-08-27 Jeff Law <law@redhat.com>
1607 PR tree-optimization/87110
1608 * tree-ssa-dse.c (compute_trims): Handle non-constant
1611 2018-08-27 Martin Sebor <msebor@redhat.com>
1613 PR tree-optimization/86914
1614 * tree-ssa-strlen.c (maybe_set_strlen_range): Avoid MEM_REF.
1616 2018-08-27 Martin Sebor <msebor@redhat.com>
1618 PR tree-optimization/87112
1619 * builtins.c (expand_builtin_strnlen): Convert c_strlen result to
1620 the type of the bound argument.
1622 2018-08-27 Jeff Law <law@redhat.com>
1624 * tree-ssa-dse.c (compute_trims): Handle case where the reference's
1625 type does not have a TYPE_SIZE_UNIT.
1627 2018-08-27 Steve Ellcey <sellcey@cavium.com>
1629 * config/aarch64/aarch64-speculation.cc: Replace include of cfg.h
1630 with include of backend.h.
1632 2018-08-27 Richard Biener <rguenther@suse.de>
1634 PR tree-optimization/86927
1635 * tree-vect-loop.c (vect_create_epilog_for_reduction): Properly
1636 use const cond reduction code.
1638 2018-08-27 Alexander Monakov <amonakov@ispras.ru>
1640 PR tree-optimization/85758
1641 * match.pd ((X & Y) ^ Y): Add :s qualifier to inner expression.
1643 2018-08-27 David Malcolm <dmalcolm@redhat.com>
1646 * diagnostic-show-locus.c (class layout_range): Update for
1647 conversion of show_caret_p to a tri-state.
1648 (layout_range::layout_range): Likewise.
1649 (make_range): Likewise.
1650 (layout::maybe_add_location_range): Likewise.
1651 (layout::should_print_annotation_line_p): Don't show annotation
1652 lines for ranges that are SHOW_LINES_WITHOUT_RANGE.
1653 (layout::get_state_at_point): Update for conversion of
1654 show_caret_p to a tri-state. Bail out early for
1655 SHOW_LINES_WITHOUT_RANGE, so that such ranges don't affect
1656 underlining or source colorization.
1657 (gcc_rich_location::add_location_if_nearby): Update for conversion
1658 of show_caret_p to a tri-state.
1659 (selftest::test_one_liner_multiple_carets_and_ranges): Likewise.
1660 (selftest::test_one_liner_fixit_replace_equal_secondary_range):
1662 (selftest::test_one_liner_labels): Likewise.
1663 * gcc-rich-location.c (gcc_rich_location::add_expr): Update for
1664 conversion of show_caret_p to a tri-state.
1665 * pretty-print.c (text_info::set_location): Likewise.
1666 * pretty-print.h (text_info::set_location): Likewise.
1667 * substring-locations.c (format_warning_n_va): Likewise.
1668 * tree-diagnostic.c (default_tree_printer): Likewise.
1669 * tree-pretty-print.c (newline_and_indent): Likewise.
1671 2018-08-27 David Malcolm <dmalcolm@redhat.com>
1674 * diagnostic-show-locus.c (get_line_span_for_fixit_hint): Show the
1675 line above for line-insertion fix-it hints.
1676 (selftest::test_fixit_insert_containing_newline): Update the
1677 expected results, and add a test with line-numbering enabled.
1679 2018-08-27 Martin Liska <mliska@suse.cz>
1682 * sanopt.c (sanitize_rewrite_addressable_params): Ignore
1683 params with DECL_HAS_VALUE_EXPR_P.
1685 2018-08-27 Martin Liska <mliska@suse.cz>
1687 * config/i386/i386.c (ix86_expand_set_or_movmem): Dump
1688 selected expansion strategy.
1690 2018-08-27 Martin Liska <mliska@suse.cz>
1692 * builtins.h (is_builtin_fn): Remove and fndecl_built_in_p.
1693 * builtins.c (is_builtin_fn): Likewise.
1694 * attribs.c (diag_attr_exclusions): Use new function
1695 fndecl_built_in_p and remove check for FUNCTION_DECL if
1697 (builtin_mathfn_code): Likewise.
1698 (fold_builtin_expect): Likewise.
1699 (fold_call_expr): Likewise.
1700 (fold_builtin_call_array): Likewise.
1701 (fold_call_stmt): Likewise.
1702 (set_builtin_user_assembler_name): Likewise.
1703 (is_simple_builtin): Likewise.
1704 * calls.c (gimple_alloca_call_p): Likewise.
1705 (maybe_warn_nonstring_arg): Likewise.
1706 * cfgexpand.c (expand_call_stmt): Likewise.
1707 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Likewise.
1708 (cgraph_edge::verify_corresponds_to_fndecl): Likewise.
1709 (cgraph_node::verify_node): Likewise.
1710 * cgraphclones.c (build_function_decl_skip_args): Likewise.
1711 (cgraph_node::create_clone): Likewise.
1712 * config/arm/arm.c (arm_insert_attributes): Likewise.
1713 * config/i386/i386.c (ix86_gimple_fold_builtin): Likewise.
1714 * dse.c (scan_insn): Likewise.
1715 * expr.c (expand_expr_real_1): Likewise.
1716 * fold-const.c (operand_equal_p): Likewise.
1717 (fold_binary_loc): Likewise.
1718 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
1719 * gimple-low.c (lower_stmt): Likewise.
1720 * gimple-pretty-print.c (dump_gimple_call): Likewise.
1721 * gimple-ssa-warn-restrict.c (wrestrict_dom_walker::check_call): Likewise.
1722 * gimple.c (gimple_build_call_from_tree): Likewise.
1723 (gimple_call_builtin_p): Likewise.
1724 (gimple_call_combined_fn): Likewise.
1725 * gimplify.c (gimplify_call_expr): Likewise.
1726 (gimple_boolify): Likewise.
1727 (gimplify_modify_expr): Likewise.
1728 (gimplify_addr_expr): Likewise.
1729 * hsa-gen.c (gen_hsa_insns_for_call): Likewise.
1730 * ipa-cp.c (determine_versionability): Likewise.
1731 * ipa-fnsummary.c (compute_fn_summary): Likewise.
1732 * ipa-param-manipulation.c (ipa_modify_formal_parameters): Likewise.
1733 * ipa-split.c (visit_bb): Likewise.
1734 (split_function): Likewise.
1735 * ipa-visibility.c (cgraph_externally_visible_p): Likewise.
1736 * lto-cgraph.c (input_node): Likewise.
1737 * lto-streamer-out.c (write_symbol): Likewise.
1738 * omp-low.c (setjmp_or_longjmp_p): Likewise.
1739 (lower_omp_1): Likewise.
1740 * predict.c (strip_predict_hints): Likewise.
1741 * print-tree.c (print_node): Likewise.
1742 * symtab.c (symtab_node::output_to_lto_symbol_table_p): Likewise.
1743 * trans-mem.c (is_tm_irrevocable): Likewise.
1744 (is_tm_load): Likewise.
1745 (is_tm_simple_load): Likewise.
1746 (is_tm_store): Likewise.
1747 (is_tm_simple_store): Likewise.
1748 (is_tm_abort): Likewise.
1749 (tm_region_init_1): Likewise.
1750 * tree-call-cdce.c (gen_shrink_wrap_conditions): Likewise.
1751 * tree-cfg.c (verify_gimple_call): Likewise.
1752 (move_stmt_r): Likewise.
1753 (stmt_can_terminate_bb_p): Likewise.
1754 * tree-eh.c (lower_eh_constructs_2): Likewise.
1755 * tree-if-conv.c (if_convertible_stmt_p): Likewise.
1756 * tree-inline.c (remap_gimple_stmt): Likewise.
1757 (copy_bb): Likewise.
1758 (estimate_num_insns): Likewise.
1759 (fold_marked_statements): Likewise.
1760 * tree-sra.c (scan_function): Likewise.
1761 * tree-ssa-ccp.c (surely_varying_stmt_p): Likewise.
1762 (optimize_stack_restore): Likewise.
1763 (pass_fold_builtins::execute): Likewise.
1764 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
1765 (mark_all_reaching_defs_necessary_1): Likewise.
1766 * tree-ssa-dom.c (dom_opt_dom_walker::optimize_stmt): Likewise.
1767 * tree-ssa-forwprop.c (simplify_builtin_call): Likewise.
1768 (pass_forwprop::execute): Likewise.
1769 * tree-ssa-loop-im.c (stmt_cost): Likewise.
1770 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
1771 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Likewise.
1772 * tree-ssa-strlen.c (get_string_length): Likewise.
1773 * tree-ssa-structalias.c (handle_lhs_call): Likewise.
1774 (find_func_aliases_for_call): Likewise.
1775 * tree-ssa-ter.c (find_replaceable_in_bb): Likewise.
1776 * tree-stdarg.c (optimize_va_list_gpr_fpr_size): Likewise.
1777 * tree-tailcall.c (find_tail_calls): Likewise.
1778 * tree.c (need_assembler_name_p): Likewise.
1779 (free_lang_data_in_decl): Likewise.
1780 (get_call_combined_fn): Likewise.
1781 * ubsan.c (is_ubsan_builtin_p): Likewise.
1782 * varasm.c (incorporeal_function_p): Likewise.
1783 * tree.h (DECL_BUILT_IN): Remove and replace with
1785 (DECL_BUILT_IN_P): Transfort to fndecl_built_in_p.
1786 (fndecl_built_in_p): New.
1788 2018-08-27 Martin Liska <mliska@suse.cz>
1790 PR tree-optimization/86847
1791 * tree-switch-conversion.c (switch_decision_tree::dump_case_nodes):
1792 Dump also subtree probability.
1793 (switch_decision_tree::do_jump_if_equal): New function.
1794 (switch_decision_tree::emit_case_nodes): Handle special
1795 situations in balanced tree that can be emitted much simpler.
1796 Fix calculation of probabilities that happen in tree expansion.
1797 * tree-switch-conversion.h (struct cluster): Add
1799 (struct simple_cluster): Likewise.
1800 (struct case_tree_node): Add new function has_child.
1801 (do_jump_if_equal): New.
1803 2018-08-27 Martin Liska <mliska@suse.cz>
1805 * tree-switch-conversion.c (bit_test_cluster::find_bit_tests):
1806 Add new argument to bit_test_cluster constructor.
1807 (bit_test_cluster::emit): Set bits really number of values
1809 (bit_test_cluster::hoist_edge_and_branch_if_true): Add
1810 probability argument.
1811 * tree-switch-conversion.h (struct bit_test_cluster):
1812 Add m_handles_entire_switch.
1814 2018-08-27 Martin Liska <mliska@suse.cz>
1816 PR tree-optimization/86702
1817 * tree-switch-conversion.c (jump_table_cluster::emit):
1818 Make probabilities even for values in jump table
1819 according to number of cases handled.
1820 (switch_decision_tree::compute_cases_per_edge): Pass
1821 argument to reset_out_edges_aux function.
1822 (switch_decision_tree::analyze_switch_statement): Likewise.
1823 * tree-switch-conversion.h (switch_decision_tree::reset_out_edges_aux):
1826 2018-08-27 Martin Liska <mliska@suse.cz>
1828 * cfgexpand.c (expand_asm_stmt): Use label_to_block and pass
1829 cfun argument explicitly.
1830 * gimple-pretty-print.c (dump_gimple_switch): Likewise.
1831 * hsa-gen.c (gen_hsa_insns_for_switch_stmt): Use new
1832 function gimple_switch_default_bb.
1833 (convert_switch_statements):
1835 * ipa-fnsummary.c (set_switch_stmt_execution_predicate):
1836 * stmt.c (label_to_block_fn): Use label_to_block and pass
1837 cfun argument explicitly and use gimple_switch_label_bb.
1838 (expand_case): Likewise.
1839 * tree-cfg.c (lower_phi_internal_fn): Use label_to_block and pass
1840 cfun argument explicitly. Likewise.
1841 (make_edges_bb): Likewise.
1842 (make_cond_expr_edges): Likewise.
1843 (get_cases_for_edge): Likewise.
1844 (make_gimple_switch_edges): Likewise.
1845 (label_to_block_fn): Likewise.
1846 (label_to_block): Likewise.
1847 (make_goto_expr_edges): Likewise.
1848 (make_gimple_asm_edges): Likewise.
1849 (main_block_label): Likewise.
1850 (group_case_labels_stmt): Likewise.
1851 (find_taken_edge_computed_goto): Likewise.
1852 (find_taken_edge_switch_expr): Likewise.
1853 (gimple_verify_flow_info): Likewise.
1854 (gimple_redirect_edge_and_branch): Likewise.
1855 (gimple_switch_label_bb): New function.
1856 (gimple_switch_default_bb): Likewise.
1857 (gimple_switch_edge): Likewise.
1858 (gimple_switch_default_edge): Likewise.
1859 * tree-cfg.h (label_to_block_fn): Remove and replace ...
1860 (label_to_block): ... with this.
1861 (gimple_switch_label_bb): New.
1862 (gimple_switch_default_bb): Likewise.
1863 (gimple_switch_edge): Likewise.
1864 (gimple_switch_default_edge): Likewise.
1865 * tree-cfgcleanup.c (convert_single_case_switch): Use
1866 new gimple functions and pass new argument to label_to_block.
1867 (cleanup_control_flow_bb):
1868 * tree-eh.c (make_eh_dispatch_edges): Use label_to_block and pass
1869 cfun argument explicitly.
1870 (make_eh_edges): Likewise.
1871 (redirect_eh_dispatch_edge): Likewise.
1872 (lower_resx): Likewise.
1873 (lower_eh_dispatch): Likewise.
1874 (maybe_remove_unreachable_handlers): Likewise.
1875 (unsplit_eh): Likewise.
1876 (cleanup_empty_eh): Likewise.
1877 (verify_eh_edges): Likewise.
1878 (verify_eh_dispatch_edge): Likewise.
1879 * tree-ssa-dom.c (record_edge_info): Likewise.
1880 * tree-ssa-forwprop.c (simplify_gimple_switch_label_vec): Likewise.
1881 * tree-ssa-threadedge.c (thread_around_empty_blocks): Likewise.
1882 (thread_through_normal_block): Likewise.
1883 * tree-ssa-uncprop.c (associate_equivalences_with_edges): Likewise.
1884 * tree-ssa-uninit.c (convert_control_dep_chain_into_preds):
1885 * tree-switch-conversion.c (switch_conversion::collect): Use new
1887 (switch_conversion::check_final_bb): Likewise.
1888 (switch_conversion::gather_default_values): Pass new argument
1890 (switch_conversion::build_constructors): Likewise.
1891 (switch_decision_tree::compute_cases_per_edge): Use new
1892 gimple_switch_edge function.
1893 (switch_decision_tree::analyze_switch_statement): Pass new argument
1895 (switch_decision_tree::try_switch_expansion): Use
1896 gimple_switch_default_edge.
1897 * tree-vrp.c (find_switch_asserts): Pass new argument
1899 * vr-values.c (vr_values::vrp_visit_switch_stmt): Likewise.
1900 (vr_values::simplify_switch_using_ranges): Likewise.
1902 2018-08-27 Richard Biener <rguenther@suse.de>
1904 * cfganal.h (rev_post_order_and_mark_dfs_back_seme): Declare.
1905 * cfganal.c (rev_post_order_and_mark_dfs_back_seme): New function.
1907 * tree-ssa-sccvn.h (struct vn_pval): New structure.
1908 (struct vn_nary_op_s): Add unwind_to member. Add
1909 predicated_values flag and put result into a union together
1910 with a linked list of vn_pval.
1911 (struct vn_ssa_aux): Add name member to make maintaining
1912 a map of SSA name to vn_ssa_aux possible. Remove no longer
1913 needed info, dfsnum, low, visited, on_sccstack, use_processed
1914 and range_info_anti_range_p members.
1915 (run_scc_vn, vn_eliminate, free_scc_vn, vn_valueize): Remove.
1916 (do_rpo_vn, run_rpo_vn, eliminate_with_rpo_vn, free_rpo_vn):
1918 (vn_valueize): New global.
1919 (vn_context_bb): Likewise.
1920 (VN_INFO_RANGE_INFO, VN_INFO_ANTI_RANGE_P, VN_INFO_RANGE_TYPE,
1921 VN_INFO_PTR_INFO): Remove.
1922 * tree-ssa-sccvn.c: ... (rewrite)
1923 (pass_fre::execute): For -O2+ initialize loops and run
1924 RPO VN in optimistic mode (iterating). For -O1 and -Og
1925 run RPO VN in non-optimistic mode.
1926 * params.def (PARAM_SCCVN_MAX_SCC_SIZE): Remove.
1927 (PARAM_RPO_VN_MAX_LOOP_DEPTH): Add.
1928 * doc/invoke.texi (sccvn-max-scc-size): Remove.
1929 (rpo-vn-max-loop-depth): Document.
1930 * tree-ssa-alias.c (walk_non_aliased_vuses): Stop walking
1931 when valuezing the VUSE signals we walked out of the region.
1932 * tree-ssa-pre.c (phi_translate_1): Ignore predicated values.
1933 (phi_translate): Set VN context block to use for availability
1935 (compute_avail): Likewise.
1936 (pre_valueize): New function.
1937 (pass_pre::execute): Adjust to the RPO VN API.
1939 * tree-ssa-loop-ivcanon.c: Include tree-ssa-sccvn.h.
1940 (propagate_constants_for_unrolling): Remove.
1941 (tree_unroll_loops_completely): Perform value-numbering
1942 on the unrolled bodies loop parent.
1944 2018-08-27 Richard Biener <rguenther@suse.de>
1946 * tree-ssa-pre.c (compute_antic): Re-use inverted postorder
1947 for partial antic compute.
1949 2018-08-27 Jakub Jelinek <jakub@redhat.com>
1951 PR rtl-optimization/87065
1952 * combine.c (simplify_if_then_else): Formatting fix.
1953 (if_then_else_cond): Guard MULT optimization with SCALAR_INT_MODE_P
1955 (known_cond): Don't return const_true_rtx for vector modes. Use
1956 CONST0_RTX instead of const0_rtx. Formatting fixes.
1958 2018-08-27 Martin Liska <mliska@suse.cz>
1960 PR gcov-profile/87069
1961 * gcov.c (process_file): Record files already processed
1962 and warn about a file being processed multiple times.
1964 2018-08-27 Martin Liska <mliska@suse.cz>
1967 * config/aarch64/aarch64.c (aarch64_override_options_internal):
1968 Set default values for x_aarch64_*_string strings.
1969 * config/aarch64/aarch64.opt: Remove --{march,mcpu,mtune}==
1970 prefix. For -mabi do not print '=ABI' in help and use
1971 <option_value> format for -msve-vector-bits and -moverride
1974 2018-08-26 Jeff Law <law@redhat.com>
1976 * config/mips/frame-header-opt.c: Include "backend.h" rather than
1979 2018-08-26 Marek Polacek <polacek@redhat.com>
1981 PR c++/87029, Implement -Wredundant-move.
1982 * doc/invoke.texi: Document -Wredundant-move.
1984 2018-08-25 Martin Sebor <msebor@redhat.com>
1986 PR tree-optimization/87059
1987 * builtins.c (expand_builtin_strncmp): Convert MIN_EXPR operand
1988 to the same type as the other.
1989 * fold-const.c (fold_binary_loc): Assert expectation.
1991 2018-08-25 Iain Sandoe <iain@sandoe.co.uk>
1993 * config/darwin.c (machopic_legitimize_pic_address): Clean up
1994 extraneous parentheses, dead code section and formatting.
1996 2018-08-24 David Malcolm <dmalcolm@redhat.com>
1999 * diagnostic-show-locus.c (layout::layout): Ensure the margin is
2000 wide enough for jumps in the line-numbering to be visible.
2001 (layout::print_gap_in_line_numbering): New member function.
2002 (layout::calculate_line_spans): When using line numbering, merge
2003 line spans that are only 1 line apart.
2004 (diagnostic_show_locus): When printing line numbers, show gaps in
2005 line numbering directly, rather than printing headers.
2006 (selftest::test_diagnostic_show_locus_fixit_lines): Add test of
2007 line-numbering with multiple line spans.
2008 (selftest::test_fixit_insert_containing_newline_2): Add test of
2009 line-numbering, in which the spans are close enough to be merged.
2011 2018-08-24 Aldy Hernandez <aldyh@redhat.com>
2013 * gimple-ssa-evrp-analyze.c (set_ssa_range_info): Pass value_range
2014 to range_includes_zero_p. Do not special case VR_ANTI_RANGE.
2015 * tree-vrp.c (range_is_nonnull): Remove.
2016 (range_includes_zero_p): Accept value_range instead of min/max.
2017 (extract_range_from_binary_expr_1): Do not early bail on
2019 Use range_includes_zero_p instead of range_is_nonnull.
2020 (extract_range_from_unary_expr): Use range_includes_zero_p instead
2021 of range_is_nonnull.
2022 (vrp_meet_1): Pass value_range to range_includes_zero_p. Do not
2023 special case VR_ANTI_RANGE.
2024 (vrp_finalize): Same.
2025 * tree-vrp.h (range_includes_zero_p): Pass value_range as argument
2027 (range_is_nonnull): Remove.
2028 * vr-values.c (vrp_stmt_computes_nonzero): Use
2029 range_includes_zero_p instead of range_is_nonnull.
2030 (extract_range_basic): Pass value_range to range_includes_zero_p
2031 instead of range_is_nonnull.
2033 2018-08-24 Uros Bizjak <ubizjak@gmail.com>
2035 * emit-rtl.c (init_emit_once): Do not emit MODE_POINTER_BOUNDS RTXes.
2036 * emit-rtl.h (rtl_data): Remove return_bnd.
2037 * explow.c (trunc_int_for_mode): Do not handle POINTER_BOUNDS_MODE_P.
2038 * function.c (diddle_return_value): Do not handle crtl->return_bnd.
2039 * genmodes.c (complete_mode): Do not handle MODE_POINTER_BOUNDS.
2040 (POINTER_BOUNDS_MODE): Remove definition.
2041 (make_pointer_bounds_mode): Remove.
2042 (get_mode_class): Do not handle MODE_POINTER_BOUNDS.
2043 * machmode.h (POINTER_BOUNDS_MODE_P): Remove definition.
2044 (scalare_mode::includes_p): Do not handle MODE_POINTER_BOUNDS.
2045 * mode-classes.def: Do not define MODE_POINTER_BOUNDS.
2046 * stor-layout.c (int_mode_for_mode): Do not handle MODE_POINTER_BOUNDS.
2047 * tree-core.h (enum tree_index): Remove TI_POINTER_BOUNDS_TYPE.
2048 * varasm.c (output_constant_pool_2): Do not handle MODE_POINTER_BOUNDS.
2050 * config/i386/i386-modes.def (BND32, BND64): Remove.
2051 * config/i386/i386.c (dbx_register_map): Remove bound registers.
2052 (dbx64_register_map): Ditto.
2053 (svr4_dbx_register_map): Ditto.
2054 (indirect_thunk_bnd_needed): Remove.
2055 (indirect_thunks_bnd_used): Ditto.
2056 (indirect_return_bnd_needed): Ditto.
2057 (indirect_return_via_cx_bnd): Ditto.
2058 (enum indirect_thunk_prefix): Remove indirect_thunk_prefix_bnd.
2059 (indirect_thunk_name): Remove handling of indirect_thunk_prefix_bnd.
2060 (output_indirect_thunk): Ditto. Remove need_prefix argument.
2061 (output_indirect_thunk_function): Remove handling of
2062 indirect_return_bnd_needed, indirect_return_via_cx_bnd,
2063 indirect_thunk_bnd_needed and indirect_thunks_bnd_used variables.
2064 (ix86_save_reg): Remove handling of crtl->return_bnd.
2065 (ix86_legitimate_constant_p): Remove handling of POINTER_BOUNDS_MODE_P.
2066 (ix86_print_operand_address_as): Remove handling of UNSPEC_BNDMK_ADDR
2067 and UNSPEC_BNDLX_ADDR.
2068 (ix86_output_indirect_branch_via_reg): Remove handling of
2069 indirect_thunk_prefix_bnd.
2070 (ix86_output_indirect_branch_via_push): Ditto.
2071 (ix86_output_function_return): Ditto.
2072 (ix86_output_indirect_function_return): Ditto.
2073 (avoid_func_arg_motion): Do not handle UNSPEC_BNDSTX.
2074 * config/i386/i386.h (FIXED_REGISTERS): Remove bound registers.
2075 (CALL_USED_REGISTERS): Ditto.
2076 (REG_ALLOC_ORDER): Update for removal of bound registers.
2077 (HI_REGISTER_NAMES): Ditto.
2078 * config/i386/i386.md (UNSPEC_BNDMK, UNSPEC_BNDMK_ADDR, UNSPEC_BNDSTX)
2079 (UNSPEC_BNDLDX, UNSPEC_BNDLDX_ADDR, UNSPEC_BNDCL, UNSPEC_BNDCU)
2080 (UNSPEC_BNDCN, UNSPEC_MPX_FENCE): Remove.
2081 (BND0_REG, BND1_REG, BND2_REG, BND3_REG): Remove
2082 (FIRST_PSEUDO_REG): Update.
2083 (BND): Remove mode iterator.
2084 * config/i386/predicates.md (bnd_mem_operator): Remove.
2086 2018-08-24 Richard Sandiford <richard.sandiford@arm.com>
2088 * tree-vect-stmts.c (vectorizable_bswap): Handle variable-length
2091 2018-08-24 Richard Sandiford <richard.sandiford@arm.com>
2093 * tree-vect-slp.c (vect_transform_slp_perm_load): Separate out
2094 the case in which the permute needs only a single element and
2095 repeats for every vector of the result. Extend that case to
2096 handle variable-length vectors.
2097 * tree-vect-stmts.c (vectorizable_load): Update accordingly.
2099 2018-08-24 H.J. Lu <hongjiu.lu@intel.com>
2102 * dwarf2out.c (save_macinfo_strings): Call set_indirect_string
2103 on DW_MACINFO_start_file for -gsplit-dwarf -g3.
2105 2018-08-24 Richard Biener <rguenther@suse.de>
2107 * cfg.h (struct control_flow_graph): Add edge_flags_allocated and
2108 bb_flags_allocated members.
2109 (auto_flag): New RAII class for allocating flags.
2110 (auto_edge_flag): New RAII class for allocating edge flags.
2111 (auto_bb_flag): New RAII class for allocating bb flags.
2112 * cfgloop.c (verify_loop_structure): Allocate temporary edge
2114 * cfganal.c (dfs_enumerate_from): Remove use of visited sbitmap
2115 in favor of temporarily allocated BB flag.
2116 * hsa-brig.c: Re-order includes.
2117 * hsa-dump.c: Likewise.
2118 * hsa-regalloc.c: Likewise.
2119 * print-rtl.c: Likewise.
2120 * profile-count.c: Likewise.
2122 2018-08-24 Segher Boessenkool <segher@kernel.crashing.org>
2125 * config/rs6000/rs6000.c (toc_relative_expr_p): Check that the base is
2128 2018-08-24 Aldy Hernandez <aldyh@redhat.com>
2131 * wide-int-range.cc (wide_int_range_div): Do not ignore result
2132 from wide_int_range_multiplicative_op.
2134 2018-08-23 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
2136 * tree-vect-data-refs.c (vect_grouped_store_supported): Fix typo
2139 2018-08-23 Giuliano Belinassi <giuliano.belinassi@usp.br>
2141 * genmatch.c (parser::parse_operation): Fix typo 'exapnded'
2144 2018-08-23 Alexander Monakov <amonakov@ispras.ru>
2146 * tree-scalar-evolution.c (final_value_replacement_loop): Dump
2147 full GENERIC expression used for replacement.
2149 2018-08-23 Aldy Hernandez <aldyh@redhat.com>
2151 * tree-vrp.c (abs_extent_range): Remove.
2152 (extract_range_into_wide_ints): Pass wide ints by reference.
2153 (extract_range_from_binary_expr_1): Rewrite the *DIV_EXPR code.
2154 Pass wide ints by reference in all calls to
2155 extract_range_into_wide_ints.
2156 * wide-int-range.cc (wide_int_range_div): New.
2157 * wide-int-range.h (wide_int_range_div): New.
2158 (wide_int_range_includes_zero_p): New.
2159 (wide_int_range_zero_p): New.
2161 2018-08-23 Matthew Malcomson <matthew.malcomson@arm.com>
2163 * config/aarch64/aarch64.md (arches): New enum.
2164 (arch): New enum attr.
2165 (arch_enabled): New attr.
2166 (enabled): Now uses arch_enabled only.
2167 (simd, sve, fp16): Removed attribute.
2168 (fp): Attr now defined in terms of 'arch'.
2169 (*mov<mode>_aarch64, *movsi_aarch64, *movdi_aarch64, *movti_aarch64,
2170 *movhf_aarch64, <optab><fcvt_target><GPF:mode>2,
2171 <FCVT_F2FIXED:fcvt_fixed_insn><GPF:mode>3,
2172 <FCVT_FIXED2F:fcvt_fixed_insn><GPI:mode>3): Merge 'fp' and 'simd'
2173 attributes into 'arch'.
2174 (*movsf_aarch64, *movdf_aarch64, *movtf_aarch64, *add<mode>3_aarch64,
2175 subdi3, neg<mode>2, <optab><mode>3, one_cmpl<mode>2,
2176 *<NLOGICAL:optab>_one_cmpl<mode>3, *xor_one_cmpl<mode>3,
2177 *aarch64_ashl_sisd_or_int_<mode>3, *aarch64_lshr_sisd_or_int_<mode>3,
2178 *aarch64_ashr_sisd_or_int_<mode>3, *aarch64_sisd_ushl): Convert use of
2179 'simd' attribute into 'arch'.
2180 (load_pair_sw_<SX:mode><SX2:mode>, load_pair_dw_<DX:mode><DX2:mode>,
2181 store_pair_sw_<SX:mode><SX2:mode>, store_pair_dw_<DX:mode><DX2:mode>):
2182 Convert use of 'fp' attribute to 'arch'.
2183 * config/aarch64/aarch64-simd.md (move_lo_quad_internal_<mode>,
2184 move_lo_quad_internal_<mode>): (different modes) Merge 'fp' and 'simd'
2186 (move_lo_quad_internal_be_<mode>, move_lo_quad_internal_be_<mode>):
2187 (different modes) Merge 'fp' and 'simd' into 'arch'.
2188 (*aarch64_combinez<mode>, *aarch64_combinez_be<mode>): Merge 'fp' and
2191 2018-08-23 Segher Boessenkool <segher@kernel.crashing.org>
2193 PR rtl-optimization/87026
2194 * expmed.c (canonicalize_comparison): If we can no longer create
2195 pseudoregisters, don't.
2197 2018-08-23 Richard Earnshaw <rearnsha@arm.com>
2200 * config/arm/arm-protos.h (arm_emit_speculation_barrier): New
2202 * config/arm/arm.c (speculation_barrier_libfunc): New static
2204 (arm_init_libfuncs): Initialize it.
2205 (arm_emit_speculation_barrier): New function.
2206 * config/arm/arm.md (speculation_barrier): Call
2207 arm_emit_speculation_barrier for architectures that do not have
2209 (speculation_barrier_insn): Only match on Armv7 or later.
2211 2018-08-23 Richard Biener <rguenther@suse.de>
2214 * tree-inline.c (copy_bb): Drop unused __builtin_va_arg_pack_len
2217 2018-08-23 Richard Sandiford <richard.sandiford@arm.com>
2219 * config/aarch64/aarch64.c (aarch64_evpc_sve_tbl): Fix handling
2220 of single-vector TBLs.
2221 (aarch64_vectorize_vec_perm_const): Set one_vector_p when only
2224 2018-08-23 Richard Sandiford <richard.sandiford@arm.com>
2227 * config/aarch64/aarch64.c (aarch64_expand_vec_perm_const_1): Fix
2228 aarch64_evpc_tbl guard.
2230 2018-08-22 Bernd Edlinger <bernd.edlinger@hotmail.de>
2232 * tree-ssa-dse.c (compute_trims): Avoid folding away undefined
2235 2018-08-22 Martin Sebor <msebor@redhat.com>
2238 * tree-pretty-print.c (pretty_print_string): Add argument.
2239 (dump_generic_node): Call to pretty_print_string with string size.
2241 2018-08-22 Segher Boessenkool <segher@kernel.crashing.org>
2243 PR rtl-optimization/86771
2244 * combine.c (try_combine): Do not allow splitting a resulting PARALLEL
2245 of two SETs into those two SETs, one to be placed at i2, if that SETs
2246 destination is modified between i2 and i3.
2248 2018-08-22 Richard Sandiford <richard.sandiford@arm.com>
2250 PR tree-optimization/86725
2251 * tree-vect-loop.c (vect_inner_phi_in_double_reduction_p): New
2253 (vect_analyze_scalar_cycles_1): Check it.
2255 2018-08-22 Richard Sandiford <richard.sandiford@arm.com>
2257 PR tree-optimization/86725
2258 * tree-vect-loop.c (vect_is_simple_reduction): When treating
2259 an outer loop phi as a double reduction, make sure that the
2260 single user of the phi result is an inner loop phi.
2262 2018-08-22 Richard Sandiford <richard.sandiford@arm.com>
2264 * tree-vect-data-refs.c (vect_analyze_group_access_1): Convert
2265 grouped stores with gaps to a strided group.
2267 2018-08-22 Richard Sandiford <richard.sandiford@arm.com>
2269 * tree-vect-stmts.c (get_group_load_store_type)
2270 (get_load_store_type): Only test STMT_VINFO_STRIDED_P for the
2271 first statement in a group.
2273 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
2275 * config/darwin.h (LINK_COMMAND_SPEC_A): Sync LTO options with
2276 the sequence used in gcc/gcc.c.
2278 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
2281 * gcc-ar.c (main): Don’t try to invoke the plug-in if we’re not
2284 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
2286 * config/darwin10.h (LINK_GCC_C_SEQUENCE_SPEC): Adjust to use the
2287 Darwin10-specific unwinder-shim.
2288 * config/darwin12.h (LINK_GCC_C_SEQUENCE_SPEC): Remove.
2289 * config/rs6000/darwin.h (DARWIN_CRT1_SPEC, DARWIN_DYLIB1_SPEC):
2290 New to cater for Darwin10 Rosetta.
2292 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
2294 * config/i386/i386.c (ix86_output_addr_diff_elt): Move the MACH-O
2295 specific test before the one for HAVE_AS_GOTOFF_IN_DATA.
2297 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
2302 * gcc/dwarf2out.c (FUNC_SECOND_SECT_LABEL): New.
2303 (dwarf2out_switch_text_section): Generate a local label for the
2304 second function sub-section and apply it as the second FDE start
2306 * gcc/final.c (final_scan_insn_1): Emit second FDE label after the
2307 second sub-section start.
2309 2018-08-22 Richard Biener <rguenther@suse.de>
2311 PR tree-optimization/86988
2312 * tree-vrp.c (vrp_prop::check_mem_ref): Bail out on VLAs.
2314 2018-08-22 Richard Biener <rguenther@suse.de>
2316 PR tree-optimization/86945
2317 * tree-cfg.c (generate_range_test): Use unsigned arithmetic.
2319 2018-08-22 Alexandre Oliva <oliva@adacore.com>
2321 * config/rs6000/rs6000.c (SMALL_DATA_RELOC, SMALL_DATA_REG): Add
2322 a comment about how uses of r2 for .sdata2 come about.
2324 2018-08-22 Alexandre Oliva <aoliva@redhat.com>
2326 * tree-ssa-reassoc.c (is_reassociable_op): Fix cut&pasto.
2328 2018-08-21 Marek Polacek <polacek@redhat.com>
2330 PR c++/86981, Implement -Wpessimizing-move.
2331 * doc/invoke.texi: Document -Wpessimizing-move.
2333 2018-08-21 Jan Hubicka <jh@suse.cz>
2335 * tree.c (find_decls_types_r): Do not check for redundant typedefs.
2336 * tree.h (is_redundant_typedef): Remove.
2337 * dwarf2out.c (is_redundant_typedef): Turn into static function.
2339 2018-08-21 Jan Hubicka <jh@suse.cz>
2341 * tree.c (free_lang_data_in_decl): Remove types from DECL_CONTEXT
2344 2018-08-21 Tamar Christina <tamar.christina@arm.com>
2346 * expmed.c (extract_low_bits): Reject invalid subregs early.
2348 2018-08-21 Bernd Edlinger <bernd.edlinger@hotmail.de>
2351 * tree-ssa-strlen.c (adjust_last_stmt): Avoid folding away undefined
2354 2018-08-21 Rasmus Villemoes <rv@rasmusvillemoes.dk>
2356 * config/vxworks.h: Guard vxworks_asm_out_constructor and
2357 vxworks_asm_out_destructor by !HAVE_INITFINI_ARRAY_SUPPORT
2358 * config/vxworks.c: Likewise.
2360 2018-08-21 Rasmus Villemoes <rv@rasmusvillemoes.dk>
2362 * config/vxworks.c: Set targetm.have_ctors_dtors
2363 if HAVE_INITFINI_ARRAY_SUPPORT.
2364 * config/vxworks.h: Set SUPPORTS_INIT_PRIORITY
2365 if HAVE_INITFINI_ARRAY_SUPPORT.
2367 2018-08-21 Rasmus Villemoes <rv@rasmusvillemoes.dk>
2369 * config/vxworks.h: Add $(WIND_BASE)/target/h/wrn/coreip to
2370 default search path for VxWorks < 7.
2372 2018-08-21 Bernd Edlinger <bernd.edlinger@hotmail.de>
2374 * gimple-ssa-sprintf.c (decl_constant_value): Remove.
2375 (get_format_string): Refer to c_getstr.
2377 2018-08-21 Tom de Vries <tdevries@suse.de>
2379 * cgraph.h (debuginfo_early_init, debuginfo_init, debuginfo_fini)
2380 (debuginfo_start, debuginfo_stop, debuginfo_early_start)
2381 (debuginfo_early_stop): Declare.
2382 * cgraphunit.c (debuginfo_early_init, debuginfo_init, debuginfo_fini)
2383 (debuginfo_start, debuginfo_stop, debuginfo_early_start)
2384 (debuginfo_early_stop): New function.
2385 (symbol_table::finalize_compilation_unit): Call debuginfo_early_start
2386 and debuginfo_early_stop.
2387 * dwarf2out.c (dwarf2out_finish, dwarf2out_early_finish): Dump dwarf.
2388 * toplev.c (compile_file): Call debuginfo_start and debuginfo_stop.
2389 (general_init): Call debuginfo_early_init.
2390 (finalize): Call debuginfo_fini.
2391 (do_compile): Call debuginfo_init.
2392 * doc/invoke.texi (@gccoptlist): Add -fdump-debug and
2394 (@item -fdump-debug, @item -fdump-earlydebug): Add.
2396 2018-08-21 Tom de Vries <tdevries@suse.de>
2398 * dwarf2out.c (print_dw_val, print_loc_descr, print_die): Handle
2399 flag_dump_noaddr and flag_dump_unnumbered.
2401 2018-08-21 Aldy Hernandez <aldyh@redhat.com>
2403 * wide-int-range.cc (wide_int_range_abs): New.
2404 (wide_int_range_order_set): Rename from wide_int_range_min_max.
2405 * wide-int-range.h (wide_int_range_abs): New.
2406 (wide_int_range_min_max): New.
2407 * tree-vrp.c (extract_range_from_unary_expr): Rewrite ABS_EXPR
2408 case to call wide_int_range_abs.
2409 Rewrite MIN/MAX_EXPR to call wide_int_range_min_max.
2410 (extract_range_from_abs_expr): Delete.
2412 2018-08-20 Michael Meissner <meissner@linux.ibm.com>
2415 * config/rs6000/rs6000.md (extendsi<mode>2): Change constraints
2416 from 'Y' to 'YZ' to enable the LWAX instruction to be generated
2419 2018-08-20 Nathan Sidwell <nathan@acm.org>
2420 Jeff Law <law@redhat.com>
2422 * config/s390/s390-c (s390_macro_to_expand): Use cpp_macro_p.
2423 * config/spu/spu-c.c (spu_macro_to_expand): Likewise.
2425 2018-08-20 David Malcolm <dmalcolm@redhat.com>
2428 * attribs.c (diag_attr_exclusions): Add auto_diagnostic_group instance.
2429 (decl_attributes): Likewise.
2430 * calls.c (maybe_warn_nonstring_arg): Add auto_diagnostic_group
2432 * cgraphunit.c (maybe_diag_incompatible_alias): Likewise.
2433 * diagnostic-core.h (class auto_diagnostic_group): New class.
2434 * diagnostic.c (diagnostic_initialize): Initialize the new fields.
2435 (diagnostic_report_diagnostic): Handle the first diagnostics within
2437 (emit_diagnostic): Add auto_diagnostic_group instance.
2439 (inform_n): Likewise.
2440 (warning): Likewise.
2441 (warning_at): Likewise.
2442 (warning_n): Likewise.
2443 (pedwarn): Likewise.
2444 (permerror): Likewise.
2446 (error_n): Likewise.
2447 (error_at): Likewise.
2449 (fatal_error): Likewise.
2450 (internal_error): Likewise.
2451 (internal_error_no_backtrace): Likewise.
2452 (auto_diagnostic_group::auto_diagnostic_group): New ctor.
2453 (auto_diagnostic_group::~auto_diagnostic_group): New dtor.
2454 * diagnostic.h (struct diagnostic_context): Add fields
2455 "diagnostic_group_nesting_depth",
2456 "diagnostic_group_emission_count", "begin_group_cb",
2458 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior):
2459 Add auto_diagnostic_group instance(s).
2460 (find_explicit_erroneous_behavior): Likewise.
2461 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Likewise.
2462 * gimple-ssa-warn-restrict.c (maybe_diag_offset_bounds): Likewise.
2463 * gimplify.c (warn_implicit_fallthrough_r): Likewise.
2464 (gimplify_va_arg_expr): Likewise.
2465 * hsa-gen.c (HSA_SORRY_ATV): Likewise.
2466 (HSA_SORRY_AT): Likewise.
2467 * ipa-devirt.c (compare_virtual_tables): Likewise.
2468 (warn_odr): Likewise.
2469 * multiple_target.c (expand_target_clones): Likewise.
2470 * opts-common.c (cmdline_handle_error): Likewise.
2471 * reginfo.c (globalize_reg): Likewise.
2472 * substring-locations.c (format_warning_n_va): Likewise.
2473 * tree-inline.c (expand_call_inline): Likewise.
2474 * tree-ssa-ccp.c (pass_post_ipa_warn::execute): Likewise.
2475 * tree-ssa-loop-niter.c
2476 (do_warn_aggressive_loop_optimizations): Likewise.
2477 * tree-ssa-uninit.c (warn_uninit): Likewise.
2478 * tree.c (warn_deprecated_use): Likewise.
2480 2018-08-20 H.J. Lu <hongjiu.lu@intel.com>
2483 * config/i386/i386.md (eh_return): Always update EH return
2484 address in word_mode.
2486 2018-08-20 Chung-Lin Tang <cltang@codesourcery.com>
2488 * targhooks.c (std_gimplify_va_arg_expr): Properly handle case of when
2489 TARGET_SPLIT_COMPLEX_ARG is defined.
2491 2018-08-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
2493 * expr.c (store_field): Change gcc_assert to gcc_checking_assert.
2495 2018-08-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
2498 * expr.c (expand_assignment): Assert that bitpos is positive.
2499 (store_field): Likewise
2500 (expand_expr_real_1): Make sure that bitpos is positive.
2501 * config/alpha/alpha.h (CONSTANT_ADDRESS_P): Avoid signed
2504 2018-08-20 Nathan Sidwell <nathan@acm.org>
2506 * Makefile.in (CPP_ID_DATA_H): Delete.
2507 (CPP_INTERNAL_H): Don't add it.
2508 (GTFILES): Replace CPP_ID_DATA_H with CPPLIB_H.
2509 * gengtype.c (open_base_files): Replace cpp-id-data.h with cpplib.h
2511 2018-08-20 Richard Biener <rguenther@suse.de>
2513 PR tree-optimization/78655
2514 * tree-vrp.c (extract_range_from_binary_expr_1): Make
2515 pointer + offset nonnull if either operand is nonnull work.
2517 2018-08-20 Tom de Vries <tdevries@suse.de>
2519 * dwarf2out.c (add_scalar_info): Don't add reference to existing die
2520 unless the referenced die describes the added property using
2521 DW_AT_location or DW_AT_const_value. Fall back to exprloc case.
2522 Otherwise, add a DW_AT_location to the referenced die.
2524 2018-08-19 Uros Bizjak <ubizjak@gmail.com>
2527 * config/i386/i386.c (ix86_rtx_costs) [case SET]: Check source for
2528 register_operand when calling ix86_set_reg_reg_cost.
2529 [case CONST_INT, case CONST, case LABEL_REF, case SYMBOL_REF]:
2530 Set *total to 0 for operands that satisfy x86_64_immediate_operand
2531 predicate and to 1 otherwise.
2533 2018-08-18 Iain Sandoe <iain@sandoe.co.uk>
2535 * config/darwin.c (darwin_override_options): If -gsplit-dwarf is set,
2536 emit a diagnostic that it is not supported and reset the option.
2537 * config/darwin.h (DRIVER_SELF_SPECS): Note that gsplit-dwarf is not
2538 supported and consume the option. (ASM_FINAL_SPEC): New.
2540 2018-08-17 Segher Boessenkool <segher@kernel.crashing.org>
2542 * doc/md.texi (Patterns): Use @ref instead of @xref in the middle of
2545 2018-08-17 Sandra Loosemore <sandra@codesourcery.com>
2547 C-SKY port: Documentation
2549 * doc/extend.texi (C-SKY Function Attributes): New section.
2550 * doc/invoke.texi (Option Summary): Add C-SKY options.
2551 (C-SKY Options): New section.
2552 * doc/md.texi (Machine Constraints): Document C-SKY constraints.
2554 2018-08-17 Jojo <jijie_rong@c-sky.com>
2555 Huibin Wang <huibin_wang@c-sky.com>
2556 Sandra Loosemore <sandra@codesourcery.com>
2557 Chung-Lin Tang <cltang@codesourcery.com>
2559 C-SKY port: Backend implementation
2561 * config/csky/*: New.
2562 * common/config/csky/*: New.
2564 2018-08-17 Jojo <jijie_rong@c-sky.com>
2565 Huibin Wang <huibin_wang@c-sky.com>
2566 Sandra Loosemore <sandra@codesourcery.com>
2567 Chung-Lin Tang <cltang@codesourcery.com>
2568 Andrew Jenner <andrew@codesourcery.com>
2570 C-SKY port: Configury
2572 * config.gcc (csky-*-*): New.
2573 * configure.ac: Add csky to targets for dwarf2 debug_line support.
2574 * configure: Regenerated.
2576 2018-08-17 David Malcolm <dmalcolm@redhat.com>
2578 * dump-context.h: Include "dumpfile.h".
2579 (dump_context::dump_printf_va): Convert final param from va_list
2580 to va_list *. Convert from ATTRIBUTE_PRINTF to
2581 ATTRIBUTE_GCC_DUMP_PRINTF.
2582 (dump_context::dump_printf_loc_va): Likewise.
2583 * dumpfile.c: Include "stringpool.h".
2584 (make_item_for_dump_printf_va): Delete.
2585 (make_item_for_dump_printf): Delete.
2586 (class dump_pretty_printer): New class.
2587 (dump_pretty_printer::dump_pretty_printer): New ctor.
2588 (dump_pretty_printer::emit_items): New member function.
2589 (dump_pretty_printer::emit_any_pending_textual_chunks): New member
2591 (dump_pretty_printer::emit_item): New member function.
2592 (dump_pretty_printer::stash_item): New member function.
2593 (dump_pretty_printer::format_decoder_cb): New member function.
2594 (dump_pretty_printer::decode_format): New member function.
2595 (dump_context::dump_printf_va): Reimplement in terms of
2596 dump_pretty_printer.
2597 (dump_context::dump_printf_loc_va): Convert final param from va_list
2599 (dump_context::begin_scope): Reimplement call to
2600 make_item_for_dump_printf.
2601 (dump_printf): Update for change to dump_printf_va.
2602 (dump_printf_loc): Likewise.
2603 (selftest::test_capture_of_dump_calls): Convert "stmt" from
2604 greturn * to gimple *. Add a test_decl. Add tests of dump_printf
2605 with %T, %E, and %G.
2606 * dumpfile.h (ATTRIBUTE_GCC_DUMP_PRINTF): New macro.
2607 (dump_printf): Replace ATTRIBUTE_PRINTF_2 with
2608 ATTRIBUTE_GCC_DUMP_PRINTF (2, 3).
2609 (dump_printf_loc): Replace ATTRIBUTE_PRINTF_3 with
2610 ATTRIBUTE_GCC_DUMP_PRINTF (3, 0).
2611 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Convert
2612 use of HOST_WIDE_INT_PRINT_DEC on unsigned HOST_WIDE_INT "count"
2613 within a dump_printf_loc call to "%wu".
2614 (vector_alignment_reachable_p): Merge two dump_printf[_loc] calls,
2615 converting a use of HOST_WIDE_INT_PRINT_DEC to "%wd". Add a
2616 missing space after "=".
2617 * tree-vect-loop.c (vect_analyze_loop_2) Within a dump_printf
2618 call, convert use of HOST_WIDE_INT_PRINT_DEC to "%wd".
2619 * tree-vect-slp.c (vect_slp_bb): Within a dump_printf_loc call,
2620 convert use of HOST_WIDE_INT_PRINT_UNSIGNED to "%wu".
2621 * tree-vectorizer.c (try_vectorize_loop_1): Likewise. Remove
2622 duplicate "vectorized" from message.
2624 2018-08-17 Szabolcs Nagy <szabolcs.nagy@arm.com>
2626 * config/arm/arm-builtins.c (arm_init_simd_builtin_types): Clear
2627 polyNxK_t element's TYPE_STRING_FLAG.
2629 2018-08-17 Segher Boessenkool <segher@kernel.crashing.org>
2631 * config/rs6000/rs6000.md (*cbranch, *creturn): Name these patterns
2632 (they were unnamed before). Fix comments.
2634 2018-08-17 Nathan Sidwell <nathan@acm.org>
2636 * cppbuiltin.c: Include "cpplib.h", not "cpp-id-data.h".
2638 2018-08-17 Richard Biener <rguenther@suse.de>
2640 PR tree-optimization/86841
2641 * wide-int-range.cc (wide_int_range_lshift): Use to_uhwi.
2643 2018-08-17 Martin Liska <mliska@suse.cz>
2645 * common.opt: Remove Warn, Init and Report for options with
2646 Ignore/Deprecated flag. Warning is done automatically for
2648 * config/i386/i386.opt: Likewise.
2649 * config/ia64/ia64.opt: Likewise.
2650 * config/rs6000/rs6000.opt: Likewise.
2651 * cppbuiltin.c (define_builtin_macros_for_compilation_flags):
2652 Remove usage of flag_check_pointer_bounds.
2653 * lto-wrapper.c (merge_and_complain): Do not handle
2654 OPT_fcheck_pointer_bounds.
2655 (append_compiler_options): Likewise.
2656 * opt-functions.awk: Do not handle Deprecated.
2657 * optc-gen.awk: Check that Var, Report and Init are not
2658 used for an option with Ignore/Deprecated flag.
2659 * opts-common.c (decode_cmdline_option): Do not report
2661 (read_cmdline_option): Report warning for OPT_SPECIAL_deprecated
2663 * opts.h (struct cl_option): Remove cl_deprecated flag.
2664 (CL_ERR_DEPRECATED): Remove error enum value.
2666 2018-08-17 Richard Biener <rguenther@suse.de>
2669 * tree-inline.c (copy_bb): When inlining __builtin_va_arg_pack_len ()
2670 across a va-arg-pack using call adjust its return value accordingly.
2672 2018-08-16 Martin Sebor <msebor@redhat.com>
2674 PR tree-optimization/86853
2675 * gimple-ssa-sprintf.c (struct format_result): Rename member.
2676 (struct fmtresult): Add member and initialize it in ctors.
2677 (format_character): Handle %C. Extend range to NUL. Set MAYFAIL.
2678 (format_string): Handle %S the same as %ls. Set MAYFAIL.
2679 (format_directive): Set POSUNDER4K when MAYFAIL is set.
2680 (parse_directive): Handle %C same as %c.
2681 (sprintf_dom_walker::compute_format_length): Adjust.
2682 (is_call_safe): Adjust.
2684 2018-08-16 Bernd Edlinger <bernd.edlinger@hotmail.de>
2686 * builtins.c (c_strlen): Add new parameter eltsize. Use it
2687 for determining how to count the elements.
2688 * builtins.h (c_strlen): Adjust prototype.
2689 * expr.c (string_constant): Add new parameter mem_size.
2690 Set *mem_size appropriately.
2691 * expr.h (string_constant): Adjust protoype.
2692 * gimple-fold.c (get_range_strlen): Add new parameter eltsize.
2693 * gimple-fold.h (get_range_strlen): Adjust prototype.
2694 * gimple-ssa-sprintf.c (get_string_length): Add new parameter eltsize.
2695 (format_string): Call get_string_length with eltsize.
2697 2018-08-16 David Malcolm <dmalcolm@redhat.com>
2699 * diagnostic.c (default_diagnostic_start_span_fn): Call pp_string
2700 to emit the span, rather than setting it as the prefix.
2702 2018-08-16 David Malcolm <dmalcolm@redhat.com>
2704 * diagnostic-show-locus.c (layout::start_annotation_line): Add
2705 "margin_char" parameter, defaulting to space. Use it in place
2706 of pp_space for the initial part of the margin.
2707 (layout::print_leading_fixits): Use '+' when filling the margin
2708 of line-insertion fix-it hints.
2710 2018-08-16 Segher Boessenkool <segher@kernel.crashing.org>
2712 * config/rs6000/rs6000.md (two unnamed define_insn and define_split):
2715 2018-08-16 Segher Boessenkool <segher@kernel.crashing.org>
2717 * config/rs6000/altivec.md: Don't set length attribute to the default
2719 * config/rs6000/darwin.md: Ditto.
2720 * config/rs6000/dfp.md: Ditto.
2721 * config/rs6000/htm.md: Ditto.
2722 * config/rs6000/rs6000.md: Ditto.
2723 * config/rs6000/sync.md: Ditto.
2724 * config/rs6000/vsx.md: Ditto.
2726 2018-08-16 Segher Boessenkool <segher@kernel.crashing.org>
2728 * config/rs6000/altivec.md: Don't set length attribute to the default
2729 value, for branch instructions.
2730 * config/rs6000/darwin.md: Ditto.
2731 * config/rs6000/rs6000.md: Ditto.
2733 2018-08-16 Segher Boessenkool <segher@kernel.crashing.org>
2735 * config/rs6000/rs6000.md (length): Always define as const_int 4.
2736 (unnamed conditional branch define_insn): Set length to 4 or 8
2737 depending on offset.
2738 (<bd>_<mode>): Similar, for alternative 0.
2739 (<bd>tf_<mode>): Ditto.
2741 2018-08-16 Tamar Christina <tamar.christina@arm.com>
2743 * expr.c (copy_blkmode_to_reg): Perform larger copies when safe.
2745 2018-08-16 Matthew Malcomson <matthew.malcomson@arm.com>
2747 * doc/rtl.texi: Replace old RTX class names with new names.
2750 2018-08-16 Vlad Lazar <vlad.lazar@arm.com>
2752 * expmed.h (canonicalize_comparison): New declaration.
2753 * expmed.c (canonicalize_comparison, equivalent_cmp_code): New function.
2754 * expmed.c (emit_store_flag_1): Add call to canonicalize_comparison.
2755 * optabs.c (prepare_cmp_insn): Likewise.
2756 * rtl.h (unsigned_condition_p): New function which checks if a
2757 comparison operator is unsigned.
2759 2018-08-16 Nathan Sidwell <nathan@acm.org>
2761 * config/rs6000/rs6000-c.c (rs6000_macro_to_expend): Use cpp_macro_p.
2762 * config/powerpcspc/powerpcspe-c.c (rs6000_macro_to_expend): Likewise.
2764 2018-08-16 Tamar Christina <tamar.christina@arm.com>
2767 * config/arm/arm.c (arm_can_change_mode_class): Disallow subreg.
2768 * config/arm/neon.md (movv4hf, movv8hf): Refactored to..
2769 (mov<mov>): ..this and enable unconditionally.
2771 2018-08-16 Tamar Christina <tamar.christina@arm.com>
2773 * config/arm/neon.md (*neon_mov<mode>): Remove reg-to-reg alternative.
2775 2018-08-16 Sam Tebbs <sam.tebbs@arm.com>
2777 * config/aarch64/aarch64.opt (mlow-precision-recip-sqrt)
2778 (mlow-precision-sqrt, mlow-precision-div, mverbose-cost-dump): Replace
2779 "Common" with "Target".
2781 2018-08-15 Uros Bizjak <ubizjak@gmail.com>
2783 * config/i386/i386.opt (mmitigate-rop): Mark as deprecated.
2784 * doc/invoke.texi (mmitigate-rop): Remove.
2785 * config/i386/i386.c: Do not include "regrename.h".
2786 (ix86_rop_should_change_byte_p, reg_encoded_number)
2787 (ix86_get_modrm_for_rop, set_rop_modrm_reg_bits, ix86_mitigate_rop):
2789 (ix86_reorg): Remove call to ix86_mitigate_rop.
2790 * config/i386/i386.md (attr "modrm_class"): Remove.
2791 (cmp<mode>_ccno_1, mov<mode>_xor, movstrict<mode>_xor)
2792 (x86_mov<mode>cc_0_m1. x86_mov<mode>cc_0_m1_se)
2793 (x86_mov<mode>cc_0_m1_neg): Remove modrm_class attribute override.
2795 2018-08-15 Will Schmidt <will_schmidt@vnet.ibm.com>
2797 * config/rs6000/rs600.c (rs6000_gimple_fold_builtin): Add entries to
2798 allow folding of mergeh() and mergel() for the float and double types.
2799 (fold_mergehl_helper): Rework to handle building a permute tree
2802 2018-08-15 Uros Bizjak <ubizjak@gmail.com>
2804 * config/i386/i386.c (expand_vec_perm_movs): Enable V4SFmode
2807 2018-08-15 David Malcolm <dmalcolm@redhat.com>
2809 * common.opt (fdiagnostics-show-labels): New option.
2810 * diagnostic-show-locus.c (class layout_range): Add field
2812 (class layout): Add field "m_show_labels_p".
2813 (layout_range::layout_range): Add param "label" and use it to
2815 (make_range): Pass in NULL for new "label" param of layout_range's
2817 (layout::layout): Initialize m_show_labels_p.
2818 (layout::maybe_add_location_range): Pass in loc_range->m_label
2819 when constructing layout_range instances.
2820 (struct line_label): New struct.
2821 (layout::print_any_labels): New member function.
2822 (layout::print_line): Call it if label-printing is enabled.
2823 (selftest::test_one_liner_labels): New test.
2824 (selftest::test_diagnostic_show_locus_one_liner): Call it.
2825 * diagnostic.c (diagnostic_initialize): Initialize
2826 context->show_labels_p.
2827 * diagnostic.h (struct diagnostic_context): Add field
2829 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
2830 -fno-diagnostics-show-labels.
2831 * dwarf2out.c (gen_producer_string): Add
2832 OPT_fdiagnostics_show_labels to the ignored options.
2833 * gcc-rich-location.c (gcc_rich_location::add_expr): Add "label"
2835 (gcc_rich_location::maybe_add_expr): Likewise.
2836 * gcc-rich-location.h (gcc_rich_location::gcc_rich_location): Add
2837 label" param, defaulting to NULL.
2838 (gcc_rich_location::add_expr): Add "label" param.
2839 (gcc_rich_location::maybe_add_expr): Likewise.
2840 (class text_range_label): New class.
2841 (class range_label_for_type_mismatch): New class.
2842 * gimple-ssa-sprintf.c (fmtwarn): Pass NULL for new label params
2843 of format_warning_va.
2844 (fmtwarn_n): Likewise for new params of format_warning_n_va.
2845 * lto-wrapper.c (merge_and_complain): Add
2846 OPT_fdiagnostics_show_labels to the "pick one setting" options.
2847 (append_compiler_options): Likewise to the dropped options.
2848 (append_diag_options): Likewise to the passed-on options.
2849 * opts.c (common_handle_option): Handle the new option.
2850 * selftest-diagnostic.c
2851 (test_diagnostic_context::test_diagnostic_context): Enable
2853 * substring-locations.c: Include "gcc-rich-location.h".
2854 (format_warning_n_va): Add "fmt_label" and "param_label" params
2855 and use them as appropriate.
2856 (format_warning_va): Add "fmt_label" and "param_label" params,
2857 passing them on to format_warning_n_va.
2858 (format_warning_at_substring): Likewise.
2859 (format_warning_at_substring_n): Likewise.
2860 * substring-locations.h (format_warning_va): Add "fmt_label" and
2861 "param_label" params.
2862 (format_warning_n_va): Likewise.
2863 (format_warning_at_substring): Likewise.
2864 (format_warning_at_substring_n): Likewise.
2865 * toplev.c (general_init): Initialize global_dc->show_labels_p.
2867 2018-08-15 Qing Zhao <qing.zhao@oracle.com>
2870 * builtins.c (expand_builtin_memcmp): Do not expand the call
2871 when overflow is detected.
2873 2018-08-15 Martin Sebor <msebor@redhat.com>
2875 PR tree-optimization/71625
2876 * config/aarch64/aarch64-builtins.c
2877 (aarch64_init_simd_builtin_types): Clear Poly8_t's TYPE_STRING_FLAG.
2879 2018-08-15 Ilya Leoshkevich <iii@linux.ibm.com>
2881 * config/s390/s390.c (s390_reorg): Remove loop.
2883 2018-08-15 Iain Sandoe <iain@sandoe.co.uk>
2886 (darwin_function_switched_text_sections): Delete.
2887 * gcc/config/darwin.h
2888 (TARGET_ASM_FUNCTION_SWITCHED_TEXT_SECTIONS): Likewise.
2890 2018-08-15 Iain Sandoe <iain@sandoe.co.uk>
2893 * config/darwin.h: (DEBUG_STR_OFFSETS_SECTION, DEBUG_LOCLISTS_SECTION,
2894 DEBUG_RNGLISTS_SECTION) new macros. (DEBUG_PUBNAMES_SECTION,
2895 DEBUG_PUBTYPES_SECTION) update to include GNU variant.
2897 2018-08-15 Martin Liska <mliska@suse.cz>
2899 PR tree-optimization/86925
2900 * predict.c (expr_expected_value_1): When taking
2901 later predictor, assign also probability.
2902 Use fold_build2_initializer_loc in order to fold
2903 the expression in -frounding-math.
2905 2018-08-14 Allan Sandfeld Jensen <allan.jensen@qt.io>
2907 * config/i386/i386.c (expand_vec_perm_movs): New method matching movs
2909 (expand_vec_perm_1): Try the new method.
2911 2018-08-14 Ilya Leoshkevich <iii@linux.ibm.com>
2914 * lra-lives.c (remove_some_program_points_and_update_live_ranges):
2915 Check whether lra_live_max_point is 0 before dividing.
2917 2018-08-14 Martin Sebor <msebor@redhat.com>
2919 PR tree-optimization/86650
2920 * tree-vrp.c (vrp_prop::check_array_ref): Print an inform message.
2921 (vrp_prop::check_mem_ref): Same.
2923 2018-08-13 Liu Hao <lh_mouse@126.com>
2925 * pretty-print.c (eat_esc_sequence): Swap the foreground and
2926 background colors if the COMMON_LVB_REVERSE_VIDEO flag is set,
2927 and clear it thereafter, as it only works for DBCS.
2929 2018-08-13 Liu Hao <lh_mouse@126.com>
2931 * pretty-print.c (mingw_ansi_fputs): Do not call _close() on the
2932 handle returned by _get_osf_handle().
2934 2018-08-13 Will Schmidt <will_schmidt@vnet.ibm.com>
2936 * gcc/config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support
2937 for folding vec_perm.
2939 2018-08-13 Will Schmidt <will_schmidt@vnet.ibm.com>
2941 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin):
2942 Add support for gimple-folding of vec_pack() and vec_unpack()
2945 2018-08-13 Will Schmidt <will_schmidt@vnet.ibm.com>
2947 * config/rs6000/rs6000.c (rs6000_builtin_valid_without_lhs): Add
2948 vec_xst variants to the list.
2949 (rs6000_gimple_fold_builtin): Add support for folding unaligned
2950 vector loads and stores.
2952 2018-08-13 David Edelsohn <dje.gcc@gmail.com>
2954 * config.gcc (rs6000-ibm-aix4.x): Delete.
2955 (rs6000-ibm-aix5.1): Delete.
2956 (rs6000-ibm-aix5.2): Delete.
2957 (rs6000-ibm-aix5.3): Delete.
2958 * config/rs6000/aix43.h: Delete.
2959 * config/rs6000/aix51.h: Delete.
2960 * config/rs6000/aix52.h: Delete.
2961 * config/rs6000/t-aix43: Delete.
2963 2018-08-13 Ilya Leoshkevich <iii@linux.ibm.com>
2965 * config/s390/s390.c (s390_decompose_constant_pool_ref):
2967 (s390_decompose_address): Factor out constant pool ref
2970 2018-08-12 Chung-Ju Wu <jasonwucj@gmail.com>
2972 * config/nds32/nds32-predicates.c
2973 (nds32_can_use_bclr_p): Change return type as bool.
2974 (nds32_can_use_bset_p): Ditto.
2975 (nds32_can_use_btgl_p): Ditto.
2976 (nds32_can_use_bitci_p): Ditto.
2977 * config/nds32/nds32-protos.h
2978 (nds32_can_use_bclr_p): Change declaration.
2979 (nds32_can_use_bset_p): Ditto.
2980 (nds32_can_use_btgl_p): Ditto.
2981 (nds32_can_use_bitci_p): Ditto.
2983 2018-08-12 Chung-Ju Wu <jasonwucj@gmail.com>
2985 * config/nds32/nds32.c (nds32_expand_prologue, nds32_expand_epilogue):
2986 Support -msched-prolog-epilog option.
2987 * config/nds32/nds32.opt (msched-prolog-epilog): New option.
2989 2018-08-12 Chung-Ju Wu <jasonwucj@gmail.com>
2991 * common/config/nds32/nds32-common.c
2992 (nds32_option_optimization_table): Enalbe -malways-align.
2994 2018-08-12 Chung-Ju Wu <jasonwucj@gmail.com>
2996 * config.gcc (nds32*): Add nds32_isr.h and nds32_init.inc in
2998 * common/config/nds32/nds32-common.c (nds32_handle_option): Handle
2999 OPT_misr_secure_ case.
3000 * config/nds32/nds32-isr.c: Implementation of backward compatibility.
3001 * config/nds32/nds32-protos.h (nds32_isr_function_critical_p): New.
3002 * config/nds32/nds32.c (nds32_attribute_table): Add critical and
3004 * config/nds32/nds32.h (nds32_isr_nested_type): Add NDS32_CRITICAL.
3005 (nds32_isr_info): New field security_level.
3006 (TARGET_ISR_VECTOR_SIZE_4_BYTE): New macro.
3007 * config/nds32/nds32.md (return_internal): Consider critical attribute.
3008 * config/nds32/nds32.opt (misr-secure): New option.
3009 * config/nds32/nds32_init.inc: New file.
3010 * config/nds32/nds32_isr.h: New file.
3012 2018-08-11 John David Anglin <danglin@gcc.gnu.org>
3014 * config/pa/pa.md (UNSPEC_MEMORY_BARRIER): New unspec enum.
3015 Update comment for atomic instructions.
3016 (atomic_storeqi, atomic_storehi, atomic_storesi, atomic_storesf,
3017 atomic_loaddf, atomic_loaddf_1, atomic_storedf, atomic_storedf_1):
3019 (atomic_loaddi): Revise fence expansion to only emit fence prior to
3020 load for __ATOMIC_SEQ_CST model.
3021 (atomic_loaddi_1): Remove float register target.
3022 (atomic_storedi): Handle CONST_INT values.
3023 (atomic_storedi_1): Remove float register source. Add special case
3025 (memory_barrier): New expander and insn.
3027 2018-08-11 Jakub Jelinek <jakub@redhat.com>
3029 PR tree-optimization/86835
3030 * tree-ssa-math-opts.c (insert_reciprocals): Even when inserting
3031 new_stmt after def_gsi, make sure to insert new_square_stmt after
3032 that stmt, not 2 stmts before it.
3034 2018-08-10 Alexander Monakov <amonakov@ispras.ru>
3037 * config/i386/i386.md (<s>mul<mode>3_highpart): Use DWIH mode iterator
3040 2018-08-10 Martin Liska <mliska@suse.cz>
3043 * builtin-types.def (BT_FN_LONG_LONG_LONG_DOUBLE): Add new
3045 * builtins.c (expand_builtin_expect_with_probability):
3047 (expand_builtin_expect_with_probability): New function.
3048 (build_builtin_expect_predicate): Add new argumnet probability
3049 for BUILT_IN_EXPECT_WITH_PROBABILITY.
3050 (fold_builtin_expect):
3053 * builtins.def (BUILT_IN_EXPECT_WITH_PROBABILITY):
3054 * builtins.h (fold_builtin_expect): Set new argument.
3055 * doc/extend.texi: Document __builtin_expect_with_probability.
3056 * doc/invoke.texi: Likewise.
3057 * gimple-fold.c (gimple_fold_call): Pass new argument.
3058 * ipa-fnsummary.c (find_foldable_builtin_expect): Handle
3059 also BUILT_IN_EXPECT_WITH_PROBABILITY.
3060 * predict.c (get_predictor_value): New function.
3061 (expr_expected_value): Add new argument probability. Assume
3062 that predictor and probability are always non-null.
3063 (expr_expected_value_1): Likewise. For __builtin_expect and
3064 __builtin_expect_with_probability set probability. Handle
3065 combination in binary expressions.
3066 (tree_predict_by_opcode): Simplify code by simply calling
3067 get_predictor_value.
3068 (pass_strip_predict_hints::execute): Add handling of
3069 BUILT_IN_EXPECT_WITH_PROBABILITY.
3070 * predict.def (PRED_BUILTIN_EXPECT_WITH_PROBABILITY): Add
3072 * tree.h (DECL_BUILT_IN_P): New function.
3074 2018-08-10 Martin Liska <mliska@suse.cz>
3076 PR tree-optimization/85799
3077 * passes.def: Add argument for pass_strip_predict_hints.
3078 * predict.c (class pass_strip_predict_hints): Add new argument
3080 (strip_predictor_early): New function.
3081 (pass_strip_predict_hints::execute): Call the function to
3083 (strip_predict_hints): New function.
3084 * predict.def: Fix comment.
3086 2018-08-10 Thomas Preud'homme <thomas.preudhomme@linaro.org>
3088 * Makefile.in: Clarify which tm.texi to copy over to assert the
3089 right to grant a GFDL license for all.
3091 2018-08-09 Jeff Law <law@redhat.com>
3093 * config/m68k/m68k.c (m68k_adjust_decorated_operand): Remove
3096 2018-08-09 Andreas Schwab <schwab@linux-m68k.org>
3098 * config/m68k/m68k-protos.h (m68k_final_prescan_insn): Remove
3101 2018-08-09 Richard Sandiford <richard.sandiford@arm.com>
3103 * tree-vect-loop.c (vectorizable_reduction): Allow inner-loop
3104 reductions for variable-length vectors.
3106 2018-08-09 David Malcolm <dmalcolm@redhat.com>
3109 * common.opt (fdiagnostics-show-line-numbers): New option.
3110 * diagnostic-show-locus.c (class layout): Add fields
3111 "m_show_line_numbers_p" and "m_linenum_width";
3112 (num_digits): New function.
3113 (test_num_digits): New function.
3114 (layout::layout): Initialize new fields. Update m_x_offset
3115 logic to handle any left margin.
3116 (layout::print_source_line): Print line number when requested.
3117 (layout::start_annotation_line): New member function.
3118 (layout::print_annotation_line): Call it.
3119 (layout::print_leading_fixits): Likewise.
3120 (layout::print_trailing_fixits): Likewise. Update calls to
3121 move_to_column for new parameter.
3122 (layout::get_x_bound_for_row): Add "add_left_margin" param and use
3123 it to potentially call start_annotation_line.
3124 (layout::show_ruler): Call start_annotation_line.
3125 (selftest::test_line_numbers_multiline_range): New selftest.
3126 (selftest::diagnostic_show_locus_c_tests): Call test_num_digits
3127 and selftest::test_line_numbers_multiline_range.
3128 * diagnostic.c (diagnostic_initialize): Initialize
3129 show_line_numbers_p.
3130 * diagnostic.h (struct diagnostic_context): Add field
3131 "show_line_numbers_p".
3132 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
3133 -fno-diagnostics-show-line-numbers.
3134 * dwarf2out.c (gen_producer_string): Add
3135 OPT_fdiagnostics_show_line_numbers to the ignored options.
3136 * lto-wrapper.c (merge_and_complain): Likewise to the "pick
3137 one setting" options.
3138 (append_compiler_options): Likewise to the dropped options.
3139 (append_diag_options): Likewise to the passed-on options.
3140 * opts.c (common_handle_option): Handle the new option.
3141 * toplev.c (general_init): Set up global_dc->show_line_numbers_p.
3143 2018-08-09 Kelvin Nilsen <kelvin@gcc.gnu.org>
3145 * doc/extend.texi (PowerPC AltiVec Built-in Functions Available on
3146 ISA 2.07): Correct spelling of bcdsub to be __builtin_bcdsub. Add
3147 third argument of type "const signed char" to descriptions of
3148 __builtin_bcdadd, __builtin_bcdadd_lt, __builtin_bcdadd_eq,
3149 __builtin_bcdadd_gt, __builtin_bcdadd_ov, __builtin_bcdsub,
3150 __builtin_bcdsub_lt, __builtin_bcdsub_eq, __builtin_bcdsub_gt,
3151 __builtin_bcdsub_ov functions.
3153 2018-08-09 Richard Sandiford <richard.sandiford@arm.com>
3155 PR tree-optimization/86858
3156 * tree-vect-loop.c (vect_is_simple_reduction): Restore
3157 flow_bb_inside_loop_p calls.
3159 2018-08-09 Richard Sandiford <richard.sandiford@arm.com>
3161 PR tree-optimization/86871
3162 * tree-vect-stmts.c (vect_transform_stmt): Use gimple_get_lhs
3163 instead of gimple_assign_lhs.
3165 2018-08-09 Richard Earnshaw <rearnsha@arm.com>
3168 * config/aarch64/aarch64.md (add<mode>3_carryinC_zero): Add missing
3169 register constraint to operand 0.
3170 (add<mode>3_carryinC): Likewise.
3171 (add<mode>3_carryinV_zero, add<mode>3_carryinV): Likewise.
3173 2018-08-09 Martin Liska <mliska@suse.cz>
3176 * common.opt: Remove extra line.
3178 2018-08-09 Martin Liska <mliska@suse.cz>
3180 * params.def (PARAM_ALIGN_LOOP_ITERATIONS): Remove double dots
3181 at the end of a line, make first letter capital and end up
3182 a sentence with a dot.
3183 (PARAM_LOOP_INTERCHANGE_STRIDE_RATIO): Likewise.
3184 (PARAM_LOOP_BLOCK_TILE_SIZE): Likewise.
3185 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Likewise.
3186 (PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP): Likewise.
3187 (PARAM_MAX_ISL_OPERATIONS): Likewise.
3188 (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): Likewise.
3189 (PARAM_PROFILE_FUNC_INTERNAL_ID): Likewise.
3190 (PARAM_INDIR_CALL_TOPN_PROFILE): Likewise.
3191 (PARAM_SLP_MAX_INSNS_IN_BB): Likewise.
3192 (PARAM_IPA_CP_EVAL_THRESHOLD): Likewise.
3193 (PARAM_IPA_CP_RECURSION_PENALTY): Likewise.
3194 (PARAM_IPA_CP_SINGLE_CALL_PENALTY): Likewise.
3195 (PARAM_IPA_CP_LOOP_HINT_BONUS): Likewise.
3196 (PARAM_IPA_CP_ARRAY_INDEX_HINT_BONUS): Likewise.
3197 (PARAM_TREE_REASSOC_WIDTH): Likewise.
3198 (PARAM_HSA_GEN_DEBUG_STORES): Likewise.
3199 (PARAM_MAX_SPECULATIVE_DEVIRT_MAYDEFS): Likewise.
3200 (PARAM_MAX_VRP_SWITCH_ASSERTIONS): Likewise.
3202 2018-08-09 Andreas Krebbel <krebbel@linux.ibm.com>
3205 * config/s390/s390.c (s390_option_override_internal): Reduce the
3206 stack-clash-protection-probe-interval param if it would be too big
3209 2018-08-08 Andreas Schwab <schwab@linux-m68k.org>
3212 * config/m68k/m68k.h (FINAL_PRESCAN_INSN): Don't define.
3213 * config/m68k/m68k.c (handle_move_double): Don't call
3214 m68k_final_prescan_insn.
3215 (m68k_adjust_decorated_operand): Renamed from
3216 m68k_final_prescan_insn, remove first and third operand and
3218 (print_operand): Call it.
3219 (print_operand_address): Call it.
3221 2018-08-08 Nathan Sidwell <nathan@acm.org>
3223 * diagnostic.c (diagnostic_report_current_module): Use
3224 linemap_included_from & linemap_included_from_linemap.
3226 2018-08-08 Hongbo Zhang <hongbo.zhang@linaro.org>
3228 * config/aarch64/aarch64-cores.def: Add phecda core.
3229 * config/aarch64/aarch64-tune.md: Regenerate.
3230 * doc/invoke.texi: Add phecda core.
3232 2018-08-08 Andreas Krebbel <krebbel@linux.ibm.com>
3235 * config/s390/constraints.md ("NxHD0", "NxSD0"): New constraint
3237 * config/s390/s390.md ("movti"): Add more alternatives for
3238 constant to GPR copies.
3240 2018-08-08 Andreas Krebbel <krebbel@linux.ibm.com>
3242 * config/s390/s390.c: Fix whitespace damage throughout the file.
3243 * config/s390/s390.h: Likewise.
3244 * config/s390/tpf.h: Likewise.
3246 2018-08-08 Ilya Leoshkevich <iii@linux.ibm.com>
3248 * config/s390/s390.c (s390_loadrelative_operand_p):
3249 Remove TARGET_CPU_ZARCH usages.
3250 (s390_rtx_costs): Likewise.
3251 (s390_legitimate_constant_p): Likewise.
3252 (s390_cannot_force_const_mem): Likewise.
3253 (legitimate_reload_constant_p): Likewise.
3254 (s390_preferred_reload_class): Likewise.
3255 (legitimize_pic_address): Likewise.
3256 (legitimize_tls_address): Likewise.
3257 (s390_split_branches): Removed.
3258 (s390_add_execute): Removed.
3259 (s390_dump_pool): Remove TARGET_CPU_ZARCH usages.
3260 (s390_mainpool_start): Likewise.
3261 (s390_mainpool_finish): Likewise.
3262 (s390_mainpool_cancel): Removed.
3263 (s390_chunkify_start): Remove TARGET_CPU_ZARCH usages.
3264 (s390_chunkify_cancel): Likewise.
3265 (s390_return_addr_rtx): Likewise.
3266 (s390_register_info): Remove split_branches_pending_p uages.
3267 (s390_optimize_register_info): Likewise.
3268 (s390_init_frame_layout): Remove TARGET_CPU_ZARCH and
3269 split_branches_pending_p usages.
3270 (s390_can_eliminate): Remove TARGET_CPU_ZARCH usages.
3271 (s390_load_got): Likewise.
3272 (s390_expand_split_stack_prologue): Likewise.
3273 (output_asm_nops): Likewise.
3274 (s390_function_profiler): Likewise.
3275 (s390_emit_call): Likewise.
3276 (s390_conditional_register_usage): Likewise.
3277 (s390_optimize_prologue): Likewise.
3278 (s390_reorg): Remove TARGET_CPU_ZARCH and
3279 split_branches_pending_p usages.
3280 (s390_option_override_internal): Remove TARGET_CPU_ZARCH
3282 (s390_output_indirect_thunk_function): Likewise.
3283 * config/s390/s390.h (TARGET_CPU_ZARCH): Removed.
3284 (TARGET_CPU_ZARCH_P): Removed.
3285 (struct machine_function): Remove split_branches_pending_p.
3286 * config/s390/s390.md: Remove TARGET_CPU_ZARCH usages.
3288 2018-08-08 Ilya Leoshkevich <iii@linux.ibm.com>
3290 * common/config/s390/s390-common.c (processor_flags_table):
3292 * config.gcc: Remove with_arch/with_tune support.
3293 * config/s390/2064.md: Remove cpu attribute comparisons.
3294 * config/s390/driver-native.c (s390_host_detect_local_cpu):
3296 * config/s390/linux.h (ASM_SPEC):
3297 Remove -march support.
3298 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal):
3299 Use a table to get an arch level.
3300 * config/s390/s390-opts.h (enum processor_type):
3302 * config/s390/s390.c
3303 (processor_table): Remove entries, add arch_level values.
3304 (s390_issue_rate): Remove cases.
3305 (s390_option_override): Adjust
3306 s390_option_override_internal() call.
3307 (s390_option_override_internal): Remove deprecation warning.
3308 (s390_valid_target_attribute_tree): Adjust
3309 s390_option_override_internal() call.
3310 * config/s390/s390.h (struct s390_processor):
3311 Share with s390-c.c, add arch_level field.
3312 * config/s390/s390.md:
3313 Remove occurrences in cpu attribute.
3314 * config/s390/s390.opt: Remove -march/-mtune support.
3315 * config/s390/tpf.h (ASM_SPEC): Remove -march support.
3316 * doc/invoke.texi: Remove deprecation warning.
3318 2018-08-08 Luis Machado <luis.machado@linaro.org>
3320 * config/aarch64/aarch64.c (qdf24xx_vector_cost): New static
3322 (qdf24xx_tunings): Set vector cost structure to
3323 qdf24xx_vector_cost.
3325 * config/aarch64/aarch64.c (qdf24xx_addrcost_table)
3326 <register_sextend>: Set to 3.
3328 2018-08-07 Richard Sandiford <richard.sandiford@arm.com>
3331 * config/aarch64/iterators.md (FRECP, frecp_suffix): Delete.
3332 * config/aarch64/aarch64-simd.md
3333 (aarch64_frecp<FRECP:frecp_suffix><mode>): Fold FRECPE into...
3334 (@aarch64_frecpe<mode>): ...here and the move FRECPX to...
3335 (aarch64_frecpx<mode>): ...this new pattern.
3336 * config/aarch64/aarch64-simd-builtins.def: Remove comment
3337 about aarch64_frecp<FRECP:frecp_suffix><mode>.
3339 2018-08-07 Martin Liska <mliska@suse.cz>
3342 * predict.c (expr_expected_value_1): Handle DECL_IS_MALLOC,
3343 BUILT_IN_REALLOC and DECL_IS_OPERATOR_NEW.
3344 * predict.def (PRED_MALLOC_NONNULL): New predictor.
3345 * doc/extend.texi: Document that malloc attribute adds
3348 2018-08-06 John David Anglin <danglin@gcc.gnu.org>
3351 * config/pa/pa.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
3352 Define to speculation_safe_value_not_needed.
3354 2018-08-06 Jeff Law <law@redhat.com>
3356 * tree-ssa-dom.c (dom_opt_dom_walker::optimize_stmt): Pass down
3357 the vr_values instance to cprop_into_stmt.
3358 (cprop_into_stmt): Pass vr_values instance down to cprop_operand.
3359 (cprop_operand): Also query EVRP to determine if OP is a constant.
3361 2018-08-06 Nathan Sidwell <nathan@acm.org>
3363 * diagnostic.c (diagnostic_report_current_module): Reroll
3364 included-at loop. Translate text.
3366 2018-08-06 David Malcolm <dmalcolm@redhat.com>
3368 * function-tests.c (selftest::test_expansion_to_rtl): Call
3369 free_after_compilation.
3371 2018-08-06 Alan Hayward <alan.hayward@arm.com>
3373 * config/aarch64/aarch64.md: Add clobber highs to tls_desc.
3375 2018-08-06 Andreas Krebbel <krebbel@linux.ibm.com>
3377 * config/s390/s390.c (s390_loop_unroll_adjust): Prevent small
3378 loops with memory block operations from getting unrolled.
3380 2018-08-06 Ulrich Weigand <uweigand@de.ibm.com>
3383 * config/spu/spu.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
3384 Define to speculation_safe_value_not_needed.
3386 2018-08-06 Jeff Law <law@redhat.com>
3388 * reload1.c (forget_old_reloads_1): Adjust CLOBBER_HIGH
3391 2018-08-06 Jozef Lawrynowicz <jozef.l@mittosystems.com>
3394 * gcc/tree.c (build_common_tree_nodes): Initialize integer_types array
3395 with all enabled __intN types.
3397 * gcc/testsuite/gcc.target/msp430/pr86662.c: New test.
3399 2018-08-06 Alan Hayward <alan.hayward@arm.com>
3401 * alias.c (record_set): Check for clobber high.
3402 * cfgexpand.c (expand_gimple_stmt): Likewise.
3403 * combine-stack-adj.c (single_set_for_csa): Likewise.
3404 * combine.c (find_single_use_1): Likewise.
3405 (set_nonzero_bits_and_sign_copies): Likewise.
3406 (get_combine_src_dest): Likewise.
3407 (is_parallel_of_n_reg_sets): Likewise.
3408 (try_combine): Likewise.
3409 (record_dead_and_set_regs_1): Likewise.
3410 (reg_dead_at_p_1): Likewise.
3411 (reg_dead_at_p): Likewise.
3412 * dce.c (deletable_insn_p): Likewise.
3413 (mark_nonreg_stores_1): Likewise.
3414 (mark_nonreg_stores_2): Likewise.
3415 * df-scan.c (df_find_hard_reg_defs): Likewise.
3416 (df_uses_record): Likewise.
3417 (df_get_call_refs): Likewise.
3418 * dwarf2out.c (mem_loc_descriptor): Likewise.
3419 * haifa-sched.c (haifa_classify_rtx): Likewise.
3420 * ira-build.c (create_insn_allocnos): Likewise.
3421 * ira-costs.c (scan_one_insn): Likewise.
3422 * ira.c (equiv_init_movable_p): Likewise.
3423 (rtx_moveable_p): Likewise.
3424 (interesting_dest_for_shprep): Likewise.
3425 * jump.c (mark_jump_label_1): Likewise.
3426 * postreload-gcse.c (record_opr_changes): Likewise.
3427 * postreload.c (reload_cse_simplify): Likewise.
3428 (struct reg_use): Add source expr.
3429 (reload_combine): Check for clobber high.
3430 (reload_combine_note_use): Likewise.
3431 (reload_cse_move2add): Likewise.
3432 (move2add_note_store): Likewise.
3433 * print-rtl.c (print_pattern): Likewise.
3434 * recog.c (decode_asm_operands): Likewise.
3435 (store_data_bypass_p): Likewise.
3436 (if_test_bypass_p): Likewise.
3437 * regcprop.c (kill_clobbered_value): Likewise.
3438 (kill_set_value): Likewise.
3439 * reginfo.c (reg_scan_mark_refs): Likewise.
3440 * reload1.c (maybe_fix_stack_asms): Likewise.
3441 (eliminate_regs_1): Likewise.
3442 (elimination_effects): Likewise.
3443 (mark_not_eliminable): Likewise.
3444 (scan_paradoxical_subregs): Likewise.
3445 (forget_old_reloads_1): Likewise.
3446 * reorg.c (find_end_label): Likewise.
3447 (try_merge_delay_insns): Likewise.
3448 (redundant_insn): Likewise.
3449 (own_thread_p): Likewise.
3450 (fill_simple_delay_slots): Likewise.
3451 (fill_slots_from_thread): Likewise.
3452 (dbr_schedule): Likewise.
3453 * resource.c (update_live_status): Likewise.
3454 (mark_referenced_resources): Likewise.
3455 (mark_set_resources): Likewise.
3456 * rtl.c (copy_rtx): Likewise.
3457 * rtlanal.c (reg_referenced_p): Likewise.
3458 (single_set_2): Likewise.
3459 (noop_move_p): Likewise.
3460 (note_stores): Likewise.
3461 * sched-deps.c (sched_analyze_reg): Likewise.
3462 (sched_analyze_insn): Likewise.
3464 2018-08-06 Alan Hayward <alan.hayward@arm.com>
3466 * cse.c (invalidate_reg): New function extracted from...
3467 (invalidate): ...here.
3468 (canonicalize_insn): Check for clobber high.
3469 (invalidate_from_clobbers): invalidate clobber highs.
3470 (invalidate_from_sets_and_clobbers): Likewise.
3471 (count_reg_usage): Check for clobber high.
3472 (insn_live_p): Likewise.
3473 * cselib.c (cselib_expand_value_rtx_1):Likewise.
3474 (cselib_invalidate_regno): Check for clobber in setter.
3475 (cselib_invalidate_rtx): Pass through setter.
3476 (cselib_invalidate_rtx_note_stores):
3477 (cselib_process_insn): Check for clobber high.
3478 * cselib.h (cselib_invalidate_rtx): Add operand.
3480 2018-08-06 Alan Hayward <alan.hayward@arm.com>
3482 * lra-eliminations.c (lra_eliminate_regs_1): Check for clobber high.
3483 (mark_not_eliminable): Likewise.
3484 * lra-int.h (struct lra_insn_reg): Add clobber high marker.
3485 * lra-lives.c (process_bb_lives): Check for clobber high.
3486 * lra.c (new_insn_reg): Remember clobber highs.
3487 (collect_non_operand_hard_regs): Check for clobber high.
3488 (lra_set_insn_recog_data): Likewise.
3489 (add_regs_to_insn_regno_info): Likewise.
3490 (lra_update_insn_regno_info): Likewise.
3492 2018-08-06 Alan Hayward <alan.hayward@arm.com>
3494 * rtl.h (reg_is_clobbered_by_clobber_high): Add declarations.
3495 * rtlanal.c (reg_is_clobbered_by_clobber_high): Add function.
3497 2018-08-06 Alan Hayward <alan.hayward@arm.com>
3499 * emit-rtl.c (verify_rtx_sharing): Check for CLOBBER_HIGH.
3500 (copy_insn_1): Likewise.
3501 (gen_hard_reg_clobber_high): New gen function.
3502 * genconfig.c (walk_insn_part): Check for CLOBBER_HIGH.
3503 * genemit.c (gen_exp): Likewise.
3504 (gen_emit_seq): Pass through info.
3505 (gen_insn): Check for CLOBBER_HIGH.
3506 (gen_expand): Pass through info.
3507 (gen_split): Likewise.
3508 (output_add_clobbers): Likewise.
3509 * genrecog.c (validate_pattern): Check for CLOBBER_HIGH.
3510 (remove_clobbers): Likewise.
3511 * rtl.h (gen_hard_reg_clobber_high): New declaration.
3513 2018-08-06 Alan Hayward <alan.hayward@arm.com>
3515 * doc/rtl.texi (clobber_high): Add.
3516 (parallel): Add in clobber high
3517 * rtl.c (rtl_check_failed_code3): Add function.
3518 * rtl.def (CLOBBER_HIGH): Add expression.
3519 * rtl.h (RTL_CHECKC3): Add macro.
3520 (rtl_check_failed_code3): Add declaration.
3521 (XC3EXP): Add macro.
3523 2018-08-05 H.J. Lu <hongjiu.lu@intel.com>
3526 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Set
3527 cfun->machine->max_used_stack_alignment if needed.
3529 2018-08-04 Martin Sebor <msebor@redhat.com>
3531 PR tree-optimization/86571
3532 * gimple-ssa-sprintf.c (format_floating): Extend upper bound of
3535 2018-08-03 Sandra Loosemore <sandra@codesourcery.com>
3538 * config/nios2/nios2.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
3541 2018-08-03 Jeff Law <law@redhat.com>
3544 * config/mn10300/mn10300.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
3545 Define to speculation_safe_value_not_needed.
3547 2018-08-03 David Malcolm <dmalcolm@redhat.com>
3549 * doc/gcov.texi (-x): Remove duplicate "to".
3550 * doc/invoke.texi (-Wnoexcept-type): Remove duplicate "calls".
3551 (-Wif-not-aligned): Remove duplicate "is".
3552 (-flto): Remove duplicate "the".
3553 (MicroBlaze Options): In examples of "-mcpu=cpu-type", remove
3554 duplicate "v5.00.b".
3555 (MSP430 Options): Remove duplicate "and" from the description
3556 of "-mgprel-sec=regexp".
3557 (x86 Options): Remove duplicate copies of "vmldLog102" and
3558 vmlsLog104 from description of "-mveclibabi=type".
3560 2018-08-03 Richard Sandiford <richard.sandiford@arm.com>
3562 * internal-fn.h (first_commutative_argument): Declare.
3563 * internal-fn.c (first_commutative_argument): New function.
3564 * tree-vect-slp.c (vect_get_and_check_slp_defs): Remove extra
3565 restrictions for pattern statements. Use first_commutative_argument
3566 to look for commutative operands in calls to internal functions.
3568 2018-08-03 Aldy Hernandez <aldyh@redhat.com>
3570 * Makefile.in (wide-int-range.o): New.
3571 * tree-vrp.c: Move all the wide_int_* functions to...
3572 * wide-int-range.cc: ...here.
3573 * tree-vrp.h: Move all the wide_int_* prototypes to...
3574 * wide-int-range.h: ...here.
3576 2018-08-03 Tom de Vries <tdevries@suse.de>
3578 * common/config/nvptx/nvptx-common.c (nvptx_except_unwind_info): Return
3580 * config/nvptx/nvptx.c (TARGET_ASM_BYTE_OP): Remove define.
3581 * except.c (output_function_exception_table): Do early exit if
3582 targetm_common.except_unwind_info (&global_options) == UI_NONE.
3584 2018-08-03 Martin Liska <mliska@suse.cz>
3586 * predict.c (dump_prediction): Change to 2 digits
3589 2018-08-03 Siddhesh Poyarekar <siddhesh@sourceware.org>
3591 * config/aarch64/falkor.md (falkor_am_1_vxvy_vxvy): Move
3593 (falkor_am_1_gtov_gtov): ... a new insn reservation.
3595 2018-07-19 Ilya Leoshkevich <iii@linux.ibm.com>
3597 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Replace > with >=.
3598 * df-problems.c (df_remove_dead_eq_notes): Replace > with >=.
3599 * dwarf2out.c (mem_loc_descriptor): Replace > with >=.
3600 * lra-constraints.c (spill_hard_reg_in_range): Replace <= with <.
3601 * lra-remat.c (call_used_input_regno_present_p): Replace <= with <.
3603 2018-08-02 David Malcolm <dmalcolm@redhat.com>
3605 * diagnostic-show-locus.c (diagnostic_show_locus): Use
3606 pp_take_prefix when saving the existing prefix.
3607 * diagnostic.c (diagnostic_append_note): Likewise.
3608 * langhooks.c (lhd_print_error_function): Likewise.
3609 * pretty-print.c (pp_set_prefix): Drop the "const" from "prefix"
3610 param's type. Free the existing prefix.
3611 (pp_take_prefix): New function.
3612 (pretty_printer::pretty_printer): Drop the prefix parameter.
3613 Rename the length parameter to match the comment.
3614 (pretty_printer::~pretty_printer): Free the prefix.
3615 * pretty-print.h (pretty_printer::pretty_printer): Drop the prefix
3617 (struct pretty_printer): Drop the "const" from "prefix" field's
3618 type and clarify memory management.
3619 (pp_set_prefix): Drop the "const" from the 2nd param.
3620 (pp_take_prefix): New decl.
3622 2018-08-02 Aaron Sawdey <acsawdey@linux.ibm.com>
3624 * config/rs6000/rs6000-string.c (select_block_compare_mode): Move test
3625 for word_mode_ok here instead of passing as argument.
3626 (expand_block_compare): Change select_block_compare_mode() call.
3627 (expand_strncmp_gpr_sequence): New function.
3628 (expand_strn_compare): Make use of expand_strncmp_gpr_sequence.
3630 2018-08-02 Jeff Law <law@redhat.com>
3633 * config/m68k/m68k.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
3634 Define to speculation_safe_value_not_needed.
3637 * config/h8300/h8300.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
3638 Define to speculation_safe_value_not_needed.
3640 2018-08-02 Tom de Vries <tdevries@suse.de>
3643 * common/config/nvptx/nvptx-common.c (nvptx_except_unwind_info): New
3644 function. Return UI_TARGET unconditionally.
3645 (TARGET_EXCEPT_UNWIND_INFO): Redefine to nvptx_except_unwind_info.
3646 * config/nvptx/nvptx.c (TARGET_ASM_BYTE_OP): Emit commented out '.byte'.
3648 2018-08-02 Richard Sandiford <richard.sandiford@arm.com>
3650 * genemit.c (print_overload_test): Fix typo.
3652 2018-08-02 Richard Biener <rguenther@suse.de>
3654 PR tree-optimization/86816
3655 * tree-ssa-tail-merge.c (tail_merge_valueize): New function
3656 which checks for value availability before querying it.
3657 (gvn_uses_equal): Use it.
3658 (same_succ_hash): Likewise.
3659 (gimple_equal_p): Likewise.
3661 2018-08-02 Nick Clifton <nickc@redhat.com>
3664 * config/stormy16/stormy16.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
3665 Define to speculation_safe_value_not_needed.
3668 * config/v850/v850.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
3669 Define to speculation_safe_value_not_needed.
3672 * config/v850/v850.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
3673 Define to speculation_safe_value_not_needed.
3676 * config/rx/rx.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
3677 Define to speculation_safe_value_not_needed.
3680 * config/msp430/msp430.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
3681 Define to speculation_safe_value_not_needed.
3684 * config/mcore/mcore.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
3685 Define to speculation_safe_value_not_needed.
3688 * config/m32r/m32r.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
3689 Define to speculation_safe_value_not_needed.
3692 * config/iq2000/iq2000.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
3693 Define to speculation_safe_value_not_needed.
3696 * config/frv/frv.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Define to
3697 speculation_safe_value_not_needed.
3700 * config/fr30/fr30.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Define
3701 to speculation_safe_value_not_needed.
3703 2018-08-02 Richard Sandiford <richard.sandiford@arm.com>
3705 * doc/md.texi: Expand the documentation of instruction names
3706 to mention port-local uses. Document '@' in pattern names.
3707 * read-md.h (overloaded_instance, overloaded_name): New structs.
3709 (md_reader::handle_overloaded_name): New member function.
3710 (md_reader::get_overloads): Likewise.
3711 (md_reader::m_first_overload): New member variable.
3712 (md_reader::m_next_overload_ptr): Likewise.
3713 (md_reader::m_overloads_htab): Likewise.
3714 * read-md.c (md_reader::md_reader): Initialize m_first_overload,
3715 m_next_overload_ptr and m_overloads_htab.
3716 * read-rtl.c (iterator_group): Add "type" and "get_c_token" fields.
3717 (get_mode_token, get_code_token, get_int_token): New functions.
3718 (map_attr_string): Add an optional argument that passes back
3719 the associated iterator.
3720 (overloaded_name_hash, overloaded_name_eq_p, named_rtx_p):
3721 (md_reader::handle_overloaded_name, add_overload_instance): New
3723 (apply_iterators): Handle '@' names. Report an error if '@'
3724 is used without iterators.
3725 (initialize_iterators): Initialize the new iterator_group fields.
3726 * genopinit.c (handle_overloaded_code_for)
3727 (handle_overloaded_gen): New functions.
3728 (main): Use them to print declarations of maybe_code_for_* and
3729 maybe_gen_* functions, and inline definitions of code_for_* and gen_*.
3730 * genemit.c (print_overload_arguments, print_overload_test)
3731 (handle_overloaded_code_for, handle_overloaded_gen): New functions.
3732 (main): Use it to print definitions of maybe_code_for_* and
3733 maybe_gen_* functions.
3734 * config/aarch64/aarch64.c (aarch64_split_128bit_move): Use
3735 gen_aarch64_mov{low,high}_di and gen_aarch64_movdi_{low,high}
3736 instead of explicit mode checks.
3737 (aarch64_split_simd_combine): Likewise gen_aarch64_simd_combine.
3738 (aarch64_split_simd_move): Likewise gen_aarch64_split_simd_mov.
3739 (aarch64_emit_load_exclusive): Likewise gen_aarch64_load_exclusive.
3740 (aarch64_emit_store_exclusive): Likewise gen_aarch64_store_exclusive.
3741 (aarch64_expand_compare_and_swap): Likewise
3742 gen_aarch64_compare_and_swap and gen_aarch64_compare_and_swap_lse
3743 (aarch64_gen_atomic_cas): Likewise gen_aarch64_atomic_cas.
3744 (aarch64_emit_atomic_swap): Likewise gen_aarch64_atomic_swp.
3745 (aarch64_constant_pool_reload_icode): Delete.
3746 (aarch64_secondary_reload): Use code_for_aarch64_reload_movcp
3747 instead of aarch64_constant_pool_reload_icode. Use
3748 code_for_aarch64_reload_mov instead of explicit mode checks.
3749 (rsqrte_type, get_rsqrte_type, rsqrts_type, get_rsqrts_type): Delete.
3750 (aarch64_emit_approx_sqrt): Use gen_aarch64_rsqrte instead of
3751 get_rsqrte_type and gen_aarch64_rsqrts instead of gen_rqrts_type.
3752 (recpe_type, get_recpe_type, recps_type, get_recps_type): Delete.
3753 (aarch64_emit_approx_div): Use gen_aarch64_frecpe instead of
3754 get_recpe_type and gen_aarch64_frecps instead of get_recps_type.
3755 (aarch64_atomic_load_op_code): Delete.
3756 (aarch64_emit_atomic_load_op): Likewise.
3757 (aarch64_gen_atomic_ldop): Use UNSPECV_ATOMIC_* instead of
3758 aarch64_atomic_load_op_code. Use gen_aarch64_atomic_load
3759 instead of aarch64_emit_atomic_load_op.
3760 * config/aarch64/aarch64.md (aarch64_reload_movcp<GPF_TF:mode><P:mode>)
3761 (aarch64_reload_movcp<VALL:mode><P:mode>, aarch64_reload_mov<mode>)
3762 (aarch64_movdi_<mode>low, aarch64_movdi_<mode>high)
3763 (aarch64_mov<mode>high_di, aarch64_mov<mode>low_di): Add a '@'
3764 character before the pattern name.
3765 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
3766 (aarch64_rsqrte<mode>, aarch64_rsqrts<mode>)
3767 (aarch64_simd_combine<mode>, aarch64_frecpe<mode>)
3768 (aarch64_frecps<mode>): Likewise.
3769 * config/aarch64/atomics.md (atomic_compare_and_swap<mode>)
3770 (aarch64_compare_and_swap<mode>, aarch64_compare_and_swap<mode>_lse)
3771 (aarch64_load_exclusive<mode>, aarch64_store_exclusive<mode>)
3772 (aarch64_atomic_swp<mode>, aarch64_atomic_cas<mode>)
3773 (aarch64_atomic_load<atomic_ldop><mode>): Likewise.
3775 2018-08-02 Richard Sandiford <richard.sandiford@arm.com>
3777 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
3778 Allow HFmode constants if TARGET_FP_F16INST.
3780 2018-08-02 Jackson Woodruff <jackson.woodruff@arm.com>
3783 * config/aarch64/aarch64.c (aarch64_operands_adjust_ok_for_ldpstp):
3784 No longer check last store for clobber of address register.
3786 2018-08-02 Martin Liska <mliska@suse.cz>
3788 PR gcov-profile/86817
3789 * gcov.c (process_all_functions): New function.
3791 (process_file): Move functions processing to
3792 process_all_functions.
3794 2018-08-02 David Malcolm <dmalcolm@redhat.com>
3796 * dumpfile.c (dump_user_location_t::dump_user_location_t): Add
3797 "const" to the "gimple *" and "rtx_insn *" parameters.
3798 * dumpfile.h (dump_user_location_t::dump_user_location_t):
3800 (dump_location_t::dump_location_t): Likewise.
3802 2018-08-01 Martin Sebor <msebor@redhat.com>
3804 PR tree-optimization/86650
3805 * gimple-pretty-print.c (percent_G_format): Accept a "gimple *"
3806 rather than a "gcall *". Directly pass the data of interest
3807 to percent_K_format, rather than building a temporary CALL_EXPR
3809 * gimple-fold.c (gimple_fold_builtin_strncpy): Adjust.
3810 (gimple_fold_builtin_strncat): Adjust.
3811 * gimple-ssa-warn-restrict.h (check_bounds_or_overlap): Replace
3812 gcall* argument with gimple*.
3813 * gimple-ssa-warn-restrict.c (check_call): Same.
3814 (wrestrict_dom_walker::before_dom_children): Same.
3815 (builtin_access::builtin_access): Same.
3816 (check_bounds_or_overlap): Same
3817 (maybe_diag_overlap): Same.
3818 (maybe_diag_offset_bounds): Same.
3819 * tree-diagnostic.c (default_tree_printer): Move usage of
3820 EXPR_LOCATION (t) and TREE_BLOCK (t) from within percent_K_format
3822 * tree-pretty-print.c (percent_K_format): Add argument.
3823 * tree-pretty-print.h: Add argument.
3824 * tree-ssa-ccp.c (pass_post_ipa_warn::execute): Adjust.
3825 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Adjust.
3826 (maybe_diag_stxncpy_trunc): Same.
3827 (handle_builtin_stxncpy): Same.
3828 (handle_builtin_strcat): Same.
3830 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
3832 * match.pd: Optimise pointer range checks.
3834 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
3836 PR tree-optimization/86758
3837 * tree-vect-stmts.c (vectorizable_simd_clone_call): Don't try
3838 to remove pattern statements.
3840 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
3842 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Use the
3843 result of dfs_enumerate_from when constructing stmt_vec_infos,
3844 instead of additionally calling get_loop_body.
3846 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
3848 * tree-vectorizer.h (vect_create_data_ref_ptr): Remove inv_p
3850 * tree-vect-data-refs.c (vect_create_data_ref_ptr): Likewise.
3851 When creating an iv, assert that the step is not known to be zero.
3852 (vect_setup_realignment): Update call accordingly.
3853 * tree-vect-stmts.c (vectorizable_store): Likewise.
3854 (vectorizable_load): Likewise. Handle VMAT_INVARIANT separately.
3856 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
3858 * tree-vectorizer.h (vect_stmt_to_vectorize): New function.
3859 * tree-vect-loop.c (vect_update_vf_for_slp): Use it.
3860 (vectorizable_reduction): Likewise.
3861 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
3862 (vect_detect_hybrid_slp_stmts): Likewise.
3863 * tree-vect-stmts.c (vect_is_simple_use): Likewise.
3865 2018-08-01 Aldy Hernandez <aldyh@redhat.com>
3867 * tree-vrp (zero_nonzero_bits_from_bounds): Rename to...
3868 (wide_int_set_zero_nonzero_bits): ...this.
3869 (zero_nonzero_bits_from_vr): Rename to...
3870 (vrp_set_zero_nonzero_bits): ...this.
3871 (extract_range_from_multiplicative_op_1): Abstract wide int
3873 (wide_int_range_multiplicative_op): ...here.
3874 (extract_range_from_binary_expr_1): Extract wide int binary
3875 operations into their own functions.
3876 (wide_int_range_lshift): New.
3877 (wide_int_range_can_optimize_bit_op): New.
3878 (wide_int_range_shift_undefined_p): New.
3879 (wide_int_range_bit_xor): New.
3880 (wide_int_range_bit_ior): New.
3881 (wide_int_range_bit_and): New.
3882 (wide_int_range_trunc_mod): New.
3883 (extract_range_into_wide_ints): New.
3884 (vrp_shift_undefined_p): New.
3885 (extract_range_from_multiplicative_op): New.
3886 (vrp_can_optimize_bit_op): New.
3887 * tree-vrp.h (value_range::dump): New.
3888 (wide_int_range_multiplicative_op): New.
3889 (wide_int_range_lshift):New.
3890 (wide_int_range_shift_undefined_p): New.
3891 (wide_int_range_bit_xor): New.
3892 (wide_int_range_bit_ior): New.
3893 (wide_int_range_bit_and): New.
3894 (wide_int_range_trunc_mod): New.
3895 (zero_nonzero_bits_from_bounds): Rename to...
3896 (wide_int_set_zero_nonzero_bits): ...this.
3897 (zero_nonzero_bits_from_vr): Rename to...
3898 (vrp_set_zero_nonzero_bits): ...this.
3899 (range_easy_mask_min_max): Rename to...
3900 (wide_int_range_can_optimize_bit_op): this.
3901 * vr-values.c (simplify_bit_ops_using_ranges): Rename
3902 zero_nonzero_bits_from_vr into vrp_set_zero_nonzero_bits.
3904 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
3906 * tree-vectorizer.h (vect_orig_stmt): New function.
3907 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Use it.
3908 * tree-vect-loop.c (vect_model_reduction_cost): Likewise.
3909 (vect_create_epilog_for_reduction): Likewise.
3910 (vectorizable_live_operation): Likewise.
3911 * tree-vect-slp.c (vect_find_last_scalar_stmt_in_slp): Likewise.
3912 (vect_detect_hybrid_slp_stmts, vect_schedule_slp): Likewise.
3913 * tree-vect-stmts.c (vectorizable_call): Likewise.
3914 (vectorizable_simd_clone_call, vect_remove_stores): Likewise.
3916 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
3918 * tree-vectorizer.h (vect_transform_stmt): Remove grouped_store
3920 * tree-vect-stmts.c (vect_transform_stmt): Likewise.
3921 * tree-vect-loop.c (vect_transform_loop_stmt): Update call accordingly.
3922 (vect_transform_loop): Likewise.
3923 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
3925 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
3927 * tree-vectorizer.h (vect_schedule_slp): Return void.
3928 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
3929 (vect_schedule_slp): Likewise.
3931 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
3933 * tree-vect-loop.c (vect_transform_loop_stmt): Remove slp_scheduled
3935 (vect_transform_loop): Update calls accordingly. Schedule SLP
3936 instances before the main loop, if any exist.
3938 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
3940 PR tree-optimization/86749
3941 * tree-vect-patterns.c (vect_determine_min_output_precision_1):
3942 If the lhs is used in a COND_EXPR, check that it is being used
3943 as the "then" or "else" value.
3945 2018-08-01 Tom de Vries <tdevries@suse.de>
3948 * config/nvptx/nvptx.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Define to
3949 speculation_safe_value_not_needed.
3951 2018-08-01 Richard Biener <rguenther@suse.de>
3953 * tree-ssa-sccvn.c (visit_phi): Compare invariant addresses
3956 2018-08-01 Martin Liska <mliska@suse.cz>
3958 * value-prof.c (gimple_divmod_fixed_value_transform): Unify
3959 format how successful transformation is dumped.
3960 (gimple_mod_pow2_value_transform): Likewise.
3961 (gimple_mod_subtract_transform): Likewise.
3962 (gimple_stringops_transform): Likewise.
3964 2018-08-01 Martin Liska <mliska@suse.cz>
3967 * value-prof.c (interesting_stringop_to_profile_p):
3968 Simplify the code and add BUILT_IN_MEMMOVE.
3969 (gimple_stringops_transform): Likewise.
3971 2018-08-01 Sam Tebbs <sam.tebbs@arm.com>
3973 * config/aarch64/aarch64-simd.md
3974 (*aarch64_get_lane_zero_extendsi<mode>): Rename to...
3975 (*aarch64_get_lane_zero_extend<GPI:mode><VDQQH:mode>): ... This and
3976 use GPI iterator instead of SI mode.
3978 2018-08-01 Richard Earnshaw <rearnsha@arm.com>
3980 * config/rs6000/rs6000.md (speculation_barrier): Renamed from
3981 rs6000_speculation_barrier.
3982 * config/rs6000/rs6000.c (rs6000_expand_builtin): Adjust for
3983 new barrier pattern name.
3985 2018-08-01 Richard Earnshaw <rearnsha@arm.com>
3987 * config/i386/i386.md (unspecv): Add UNSPECV_SPECULATION_BARRIER.
3988 (speculation_barrier): New insn.
3990 2018-08-01 Richard Biener <rguenther@suse.de>
3993 * graphite.h: Include isl/id.h and isl/space.h to allow build
3996 2018-08-01 Jan Willem Jagersma <jwjagersma@gmail.com>
3999 * dwarf2out.c (dwarf2out_early_finish): Do not generate assembly in LTO
4000 mode for COFF targets.
4001 * defaults.h (TARGET_COFF): Define.
4002 * config/i386/djgpp.h (TARGET_ASM_LTO_START, TARGET_ASM_LTO_END,
4003 TARGET_COFF): Define.
4004 (i386_djgpp_asm_lto_start, i386_djgpp_asm_lto_end): Declare.
4005 * config/i386/djgpp.c (saved_debug_info_level): New static variable.
4006 (i386_djgpp_asm_lto_start, i386_djgpp_asm_lto_end): New functions.
4008 2018-07-31 Alexandre Oliva <oliva@adacore.com>
4010 * gimple-streamer-in.c (input_bb): Restore BB discriminator.
4011 * gimple-streamer-out.c (output_bb): Save it.
4012 * lto-streamer-in.c (input_struct_function_base): Restore
4013 instance discriminator if available. Create map on demand.
4014 * lto-streamer-out.c (output_struct_function_base): Save it if
4016 * final.c (decl_to_instance_map): Document LTO strategy.
4018 2018-07-31 Alexandre Oliva <oliva@adacore.com>
4019 Olivier Hainque <hainque@adacore.com>
4021 * debug.h (decl_to_instance_map_t): New type.
4022 (decl_to_instance_map): Declare.
4023 (maybe_create_decl_to_instance_map): New inline function.
4024 * final.c (bb_discriminator, last_bb_discriminator): New statics,
4025 to track basic block discriminators.
4026 (final_start_function_1): Initialize them.
4027 (final_scan_insn_1): On NOTE_INSN_BASIC_BLOCK, track
4029 (decl_to_instance_map): New variable.
4030 (map_decl_to_instance, maybe_set_discriminator): New functions.
4031 (notice_source_line): Set discriminator.
4033 2018-07-31 Ian Lance Taylor <iant@golang.org>
4035 * targhooks.c (default_have_speculation_safe_value): Add
4038 2018-07-31 David Malcolm <dmalcolm@redhat.com>
4040 * dump-context.h: Include "pretty-print.h".
4041 (dump_context::refresh_dumps_are_enabled): New decl.
4042 (dump_context::emit_item): New decl.
4043 (class dump_context): Add fields "m_test_pp" and
4045 (temp_dump_context::temp_dump_context): Add param "test_pp_flags".
4046 (temp_dump_context::get_dumped_text): New decl.
4047 (class temp_dump_context): Add field "m_pp".
4048 * dumpfile.c (refresh_dumps_are_enabled): Convert to...
4049 (dump_context::refresh_dumps_are_enabled): ...and add a test for
4051 (set_dump_file): Update for above change.
4052 (set_alt_dump_file): Likewise.
4053 (dump_loc): New overload, taking a pretty_printer *.
4054 (dump_context::dump_loc): Call end_any_optinfo. Dump the location
4055 to any test pretty-printer.
4056 (make_item_for_dump_gimple_stmt): New function, adapted from
4057 optinfo::add_gimple_stmt.
4058 (dump_context::dump_gimple_stmt): Call it, and use the result,
4059 eliminating the direct usage of dump_file and alt_dump_file in
4060 favor of indirectly using them via emit_item.
4061 (make_item_for_dump_gimple_expr): New function, adapted from
4062 optinfo::add_gimple_expr.
4063 (dump_context::dump_gimple_expr): Call it, and use the result,
4064 eliminating the direct usage of dump_file and alt_dump_file in
4065 favor of indirectly using them via emit_item.
4066 (make_item_for_dump_generic_expr): New function, adapted from
4068 (dump_context::dump_generic_expr): Call it, and use the result,
4069 eliminating the direct usage of dump_file and alt_dump_file in
4070 favor of indirectly using them via emit_item.
4071 (make_item_for_dump_printf_va): New function, adapted from
4072 optinfo::add_printf_va.
4073 (make_item_for_dump_printf): New function.
4074 (dump_context::dump_printf_va): Call make_item_for_dump_printf_va,
4075 and use the result, eliminating the direct usage of dump_file and
4076 alt_dump_file in favor of indirectly using them via emit_item.
4077 (make_item_for_dump_dec): New function.
4078 (dump_context::dump_dec): Call it, and use the result,
4079 eliminating the direct usage of dump_file and alt_dump_file in
4080 favor of indirectly using them via emit_item.
4081 (make_item_for_dump_symtab_node): New function, adapted from
4082 optinfo::add_symtab_node.
4083 (dump_context::dump_symtab_node): Call it, and use the result,
4084 eliminating the direct usage of dump_file and alt_dump_file in
4085 favor of indirectly using them via emit_item.
4086 (dump_context::begin_scope): Reimplement, avoiding direct usage
4087 of dump_file and alt_dump_file in favor of indirectly using them
4089 (dump_context::emit_item): New member function.
4090 (temp_dump_context::temp_dump_context): Add param "test_pp_flags".
4091 Set up test pretty-printer on the underlying context. Call
4092 refresh_dumps_are_enabled.
4093 (temp_dump_context::~temp_dump_context): Call
4094 refresh_dumps_are_enabled.
4095 (temp_dump_context::get_dumped_text): New member function.
4096 (selftest::verify_dumped_text): New function.
4097 (ASSERT_DUMPED_TEXT_EQ): New macro.
4098 (selftest::test_capture_of_dump_calls): Run all tests twice, with
4099 and then without optinfo enabled. Add uses of
4100 ASSERT_DUMPED_TEXT_EQ to all tests. Add test of nested scopes.
4101 * dumpfile.h: Update comment for the dump_* API.
4102 * optinfo-emit-json.cc
4103 (selftest::test_building_json_from_dump_calls): Update for new
4104 param for temp_dump_context ctor.
4105 * optinfo.cc (optinfo_item::optinfo_item): Remove "owned" param
4106 and "m_owned" field.
4107 (optinfo_item::~optinfo_item): Likewise.
4108 (optinfo::add_item): New member function.
4109 (optinfo::emit): Update comment.
4110 (optinfo::add_string): Delete.
4111 (optinfo::add_printf): Delete.
4112 (optinfo::add_printf_va): Delete.
4113 (optinfo::add_gimple_stmt): Delete.
4114 (optinfo::add_gimple_expr): Delete.
4115 (optinfo::add_tree): Delete.
4116 (optinfo::add_symtab_node): Delete.
4117 (optinfo::add_dec): Delete.
4118 * optinfo.h (class dump_context): New forward decl.
4119 (optinfo::add_item): New decl.
4120 (optinfo::add_string): Delete.
4121 (optinfo::add_printf): Delete.
4122 (optinfo::add_printf_va): Delete.
4123 (optinfo::add_gimple_stmt): Delete.
4124 (optinfo::add_gimple_expr): Delete.
4125 (optinfo::add_tree): Delete.
4126 (optinfo::add_symtab_node): Delete.
4127 (optinfo::add_dec): Delete.
4128 (optinfo::add_poly_int): Delete.
4129 (optinfo_item::optinfo_item): Remove "owned" param.
4130 (class optinfo_item): Remove field "m_owned".
4132 2018-07-31 Jozef Lawrynowicz <jozef.l@mittosystems.com>
4135 * gcc/cfgexpand.c (set_parm_rtl): Use the alignment of Pmode when
4136 MAX_SUPPORTED_STACK_ALIGNMENT would otherwise be exceeded by the
4137 requested variable alignment.
4138 (expand_one_ssa_partition): Likewise.
4139 (expand_one_var): Likewise.
4141 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
4143 * config/pdp11/pdp11.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Redefine
4144 to speculation_safe_value_not_needed.
4146 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
4148 * targhooks.h (speculation_safe_value_not_needed): New prototype.
4149 * targhooks.c (speculation_safe_value_not_needed): New function.
4150 * target.def (have_speculation_safe_value): Update documentation.
4151 * doc/tm.texi: Regenerated.
4153 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
4155 * config/aarch64/iterators.md (ALLI_TI): New iterator.
4156 * config/aarch64/aarch64.md (despeculate_copy<ALLI_TI:mode>): New
4158 (despeculate_copy<ALLI:mode>_insn): New insn.
4159 (despeculate_copyti_insn): New insn.
4160 (despeculate_simple<ALLI:mode>): New insn
4161 (despeculate_simpleti): New insn.
4162 * config/aarch64/aarch64.c (aarch64_speculation_safe_value): New
4164 (TARGET_SPECULATION_SAFE_VALUE): Redefine to
4165 aarch64_speculation_safe_value.
4166 (aarch64_print_operand): Handle const0_rtx in modifier 'H'.
4168 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
4170 * config/aarch64/aarch64-speculation.cc: New file.
4171 * config/aarch64/aarch64-passes.def (pass_track_speculation): Add
4172 before pass_reorder_blocks.
4173 * config/aarch64/aarch64-protos.h (make_pass_track_speculation): Add
4175 * config/aarch64/aarch64.c (aarch64_conditional_register_usage): Fix
4176 X14 and X15 when tracking speculation.
4177 * config/aarch64/aarch64.md (register name constants): Add
4178 SPECULATION_TRACKER_REGNUM and SPECULATION_SCRATCH_REGNUM.
4179 (unspec): Add UNSPEC_SPECULATION_TRACKER.
4180 (speculation_barrier): New insn attribute.
4181 (cmp<mode>): Allow SP in comparisons.
4182 (speculation_tracker): New insn.
4183 (speculation_barrier): Add speculation_barrier attribute.
4184 * config/aarch64/t-aarch64: Add make rule for aarch64-speculation.o.
4185 * config.gcc (aarch64*-*-*): Add aarch64-speculation.o to extra_objs.
4186 * doc/invoke.texi (AArch64 Options): Document -mtrack-speculation.
4188 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
4190 * config/aarch64/aarch64.md (cb<optab><mode>1): Disable when
4191 aarch64_track_speculation is true.
4192 (tb<optab><mode>1): Likewise.
4193 * config/aarch64/aarch64.c (aarch64_split_compare_regs): Do not
4194 generate CB[N]Z when tracking speculation.
4195 (aarch64_split_compare_and_swap): Likewise.
4196 (aarch64_split_atomic_op): Likewise.
4198 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
4200 * config/aarch64/aarch64.opt (mtrack-speculation): New target option.
4202 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
4204 * config/aarch64.md (unspecv): Add UNSPECV_SPECULAION_BARRIER.
4205 (speculation_barrier): New insn.
4207 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
4209 * config/arm/unspecs.md (unspecv): Add VUNSPEC_SPECULATION_BARRIER.
4210 * config/arm/arm.md (speculation_barrier): New expand.
4211 (speculation_barrier_insn): New pattern.
4213 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
4215 * builtin-types.def (BT_FN_PTR_PTR_VAR): New function type.
4216 (BT_FN_I1_I1_VAR, BT_FN_I2_I2_VAR, BT_FN_I4_I4_VAR): Likewise.
4217 (BT_FN_I8_I8_VAR, BT_FN_I16_I16_VAR): Likewise.
4218 * builtin-attrs.def (ATTR_NOVOPS_NOTHROW_LEAF_LIST): New attribute
4220 * builtins.def (BUILT_IN_SPECULATION_SAFE_VALUE_N): New builtin.
4221 (BUILT_IN_SPECULATION_SAFE_VALUE_PTR): New internal builtin.
4222 (BUILT_IN_SPECULATION_SAFE_VALUE_1): Likewise.
4223 (BUILT_IN_SPECULATION_SAFE_VALUE_2): Likewise.
4224 (BUILT_IN_SPECULATION_SAFE_VALUE_4): Likewise.
4225 (BUILT_IN_SPECULATION_SAFE_VALUE_8): Likewise.
4226 (BUILT_IN_SPECULATION_SAFE_VALUE_16): Likewise.
4227 * builtins.c (expand_speculation_safe_value): New function.
4228 (expand_builtin): Call it.
4229 * doc/cpp.texi: Document predefine __HAVE_SPECULATION_SAFE_VALUE.
4230 * doc/extend.texi: Document __builtin_speculation_safe_value.
4231 * doc/md.texi: Document "speculation_barrier" pattern.
4232 * doc/tm.texi.in: Pull in TARGET_SPECULATION_SAFE_VALUE and
4233 TARGET_HAVE_SPECULATION_SAFE_VALUE.
4234 * doc/tm.texi: Regenerated.
4235 * target.def (have_speculation_safe_value, speculation_safe_value): New
4237 * targhooks.c (default_have_speculation_safe_value): New function.
4238 (default_speculation_safe_value): New function.
4239 * targhooks.h (default_have_speculation_safe_value): Add prototype.
4240 (default_speculation_safe_value): Add prototype.
4242 2018-07-31 David Malcolm <dmalcolm@redhat.com>
4244 * dump-context.h (dump_context::dump_loc): New decl.
4245 * dumpfile.c (dump_context::dump_loc): New member function.
4246 (dump_context::dump_gimple_stmt_loc): Reimplement using dump_loc
4247 and dump_gimple_stmt.
4248 (dump_context::dump_gimple_expr_loc): Likewise, using
4250 (dump_context::dump_generic_expr_loc): Likewise, using
4252 (dump_context::dump_printf_loc_va): Likewise, using
4254 (dump_context::begin_scope): Explicitly using the global function
4255 "dump_loc", rather than the member function.
4257 2018-07-31 Martin Sebor <msebor@redhat.com>
4259 PR tree-optimization/86741
4260 * tree-vrp.c (vrp_prop::check_mem_ref): Avoid incomplete types.
4262 2018-07-31 Andreas Krebbel <krebbel@linux.ibm.com>
4264 * config/s390/s390.c (s390_expand_setmem): Make the unrolling to
4265 depend on whether prefetch instructions will be emitted or not.
4266 Use TARGET_SETMEM_PFD for checking whether prefetch instructions
4267 will be emitted or not.
4268 * config/s390/s390.h (TARGET_SETMEM_PREFETCH_DISTANCE)
4269 (TARGET_SETMEM_PFD): New macros.
4271 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4273 * tree-vectorizer.h (stmt_vec_info): Turn back into a typedef.
4274 (NULL_STMT_VEC_INFO): Delete.
4275 (stmt_vec_info::operator*): Likewise.
4276 (stmt_vec_info::operator gimple *): Likewise.
4277 * tree-vect-loop.c (vectorizable_reduction): Use NULL instead
4278 of NULL_STMT_VEC_INFO.
4279 * tree-vect-patterns.c (vect_init_pattern_stmt): Likewise.
4280 (vect_reassociating_reduction_p): Likewise.
4281 * tree-vect-stmts.c (vect_build_gather_load_calls): Likewise.
4282 (vectorizable_store): Likewise.
4283 * tree-vectorizer.c (vec_info::set_vinfo_for_stmt): Likewise.
4284 (vec_info::free_stmt_vec_infos): Likewise.
4286 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4288 * tree-vectorizer.h (vect_stmt_in_region_p): Delete.
4289 * tree-vectorizer.c (vect_stmt_in_region_p): Likewise.
4291 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4293 * tree-vectorizer.h (vec_info::new_vinfo_for_stmt)
4294 (vec_info::set_vinfo_for_stmt, vec_info::free_stmt_vec_infos)
4295 (vec_info::free_stmt_vec_info): New private member functions.
4296 (set_stmt_vec_info_vec, free_stmt_vec_infos, vinfo_for_stmt)
4297 (set_vinfo_for_stmt, new_stmt_vec_info, free_stmt_vec_info): Delete.
4298 * tree-parloops.c (gather_scalar_reductions): Remove calls to
4299 set_stmt_vec_info_vec and free_stmt_vec_infos.
4300 * tree-vect-loop.c (_loop_vec_info): Remove call to
4301 set_stmt_vec_info_vec.
4302 * tree-vect-stmts.c (new_stmt_vec_info, set_stmt_vec_info_vec)
4303 (free_stmt_vec_infos, free_stmt_vec_info): Delete in favor of...
4304 * tree-vectorizer.c (vec_info::new_stmt_vec_info)
4305 (vec_info::set_vinfo_for_stmt, vec_info::free_stmt_vec_infos)
4306 (vec_info::free_stmt_vec_info): ...these new functions. Remove
4307 assignments in {vec_info::,}new_stmt_vec_info that are redundant
4308 with the clearing in the xcalloc.
4309 (stmt_vec_info_vec): Delete.
4310 (vec_info::vec_info): Don't call set_stmt_vec_info_vec.
4311 (vectorize_loops): Likewise.
4312 (vec_info::~vec_info): Remove argument from call to
4313 free_stmt_vec_infos.
4314 (vec_info::add_stmt): Remove vinfo argument from call to
4317 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4319 * tree-vectorizer.h (free_stmt_vec_info): Take a stmt_vec_info
4320 rather than a gimple stmt.
4321 * tree-vect-stmts.c (free_stmt_vec_info): Likewise. Don't free
4322 information for pattern statements when passed the original
4323 statement; instead wait to be passed the pattern statement itself.
4324 Don't call set_vinfo_for_stmt here.
4325 (free_stmt_vec_infos): Update call to free_stmt_vec_info.
4326 * tree-vect-loop.c (_loop_vec_info::~loop_vec_info): Don't free
4327 stmt_vec_infos here.
4328 * tree-vect-slp.c (_bb_vec_info::~bb_vec_info): Likewise.
4329 * tree-vectorizer.c (vec_info::remove_stmt): Nullify the statement's
4330 stmt_vec_infos entry.
4332 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4334 * tree-vectorizer.h (vec_info::replace_stmt): Declare.
4335 * tree-vectorizer.c (vec_info::replace_stmt): New function.
4336 * tree-vect-slp.c (vect_remove_slp_scalar_calls): Use it.
4337 * tree-vect-stmts.c (vectorizable_call): Likewise.
4338 (vectorizable_simd_clone_call): Likewise.
4340 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4342 * tree-vectorizer.h (vec_info::remove_stmt): Declare.
4343 * tree-vectorizer.c (vec_info::remove_stmt): New function.
4344 * tree-vect-loop-manip.c (vect_set_loop_condition): Use it.
4345 * tree-vect-loop.c (vect_transform_loop): Likewise.
4346 * tree-vect-slp.c (vect_schedule_slp): Likewise.
4347 * tree-vect-stmts.c (vect_remove_stores): Likewise.
4349 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4351 * tree-vectorizer.h (vec_info::lookup_dr): New member function.
4352 (vect_dr_stmt): Delete.
4353 * tree-vectorizer.c (vec_info::lookup_dr): New function.
4354 * tree-vect-loop-manip.c (vect_update_inits_of_drs): Use it instead
4356 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr)
4357 (vect_analyze_data_ref_dependence, vect_record_base_alignments)
4358 (vect_verify_datarefs_alignment, vect_peeling_supportable)
4359 (vect_analyze_data_ref_accesses, vect_prune_runtime_alias_test_list)
4360 (vect_analyze_data_refs): Likewise.
4361 (vect_slp_analyze_data_ref_dependence): Likewise. Take a vec_info
4363 (vect_find_same_alignment_drs): Likewise.
4364 (vect_slp_analyze_node_dependences): Update calls accordingly.
4365 (vect_analyze_data_refs_alignment): Likewise. Use vec_info::lookup_dr
4366 instead of DR_VECT_AUX.
4367 (vect_get_peeling_costs_all_drs): Take a loop_vec_info instead
4368 of a vector data references. Use vec_info::lookup_dr instead of
4370 (vect_peeling_hash_get_lowest_cost): Update calls accordingly.
4371 (vect_enhance_data_refs_alignment): Likewise. Use vec_info::lookup_dr
4372 instead of DR_VECT_AUX.
4374 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4376 * tree-vectorizer.h (_loop_vec_info::unaligned_dr): Change to
4378 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Update
4380 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
4381 * tree-vect-loop-manip.c (get_misalign_in_elems): Likewise.
4382 (vect_gen_prolog_loop_niters): Likewise.
4384 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4386 * tree-vectorizer.h (set_dr_misalignment, dr_misalignment)
4387 (DR_TARGET_ALIGNMENT, aligned_access_p, known_alignment_for_access_p)
4388 (vect_known_alignment_in_bytes, vect_dr_behavior)
4389 (vect_get_scalar_dr_size): Take references as dr_vec_infos
4390 instead of data_references. Update calls to other routines for
4391 which the same change has been made.
4392 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Take
4393 dr_vec_infos instead of stmt_vec_infos.
4394 (vect_analyze_data_ref_dependence): Update call accordingly.
4395 (vect_slp_analyze_data_ref_dependence)
4396 (vect_record_base_alignments): Use DR_VECT_AUX.
4397 (vect_calculate_target_alignment, vect_compute_data_ref_alignment)
4398 (vect_update_misalignment_for_peel, verify_data_ref_alignment)
4399 (vector_alignment_reachable_p, vect_get_data_access_cost)
4400 (vect_peeling_supportable, vect_analyze_group_access_1)
4401 (vect_analyze_group_access, vect_analyze_data_ref_access)
4402 (vect_vfa_segment_size, vect_vfa_access_size, vect_vfa_align)
4403 (vect_compile_time_alias, vect_small_gap_p)
4404 (vectorizable_with_step_bound_p, vect_duplicate_ssa_name_ptr_info):
4405 (vect_supportable_dr_alignment): Take references as dr_vec_infos
4406 instead of data_references. Update calls to other routines for
4407 which the same change has been made.
4408 (vect_verify_datarefs_alignment, vect_get_peeling_costs_all_drs)
4409 (vect_find_same_alignment_drs, vect_analyze_data_refs_alignment)
4410 (vect_slp_analyze_and_verify_node_alignment)
4411 (vect_analyze_data_ref_accesses, vect_prune_runtime_alias_test_list)
4412 (vect_create_addr_base_for_vector_ref, vect_create_data_ref_ptr)
4413 (vect_setup_realignment): Use dr_vec_infos. Update calls after
4415 (_vect_peel_info::dr): Replace with...
4416 (_vect_peel_info::dr_info): ...this new field.
4417 (vect_peeling_hash_get_most_frequent)
4418 (vect_peeling_hash_choose_best_peeling): Update accordingly.
4419 (vect_peeling_hash_get_lowest_cost):
4420 (vect_enhance_data_refs_alignment): Likewise. Update calls to other
4421 routines for which the same change has been made.
4422 (vect_peeling_hash_insert): Likewise. Take a dr_vec_info instead of a
4424 * tree-vect-loop-manip.c (get_misalign_in_elems)
4425 (vect_gen_prolog_loop_niters): Use dr_vec_infos. Update calls after
4427 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
4428 * tree-vect-stmts.c (vect_get_store_cost, vect_get_load_cost)
4429 (vect_truncate_gather_scatter_offset, compare_step_with_zero)
4430 (get_group_load_store_type, get_negative_load_store_type)
4431 (vect_get_data_ptr_increment, vectorizable_store)
4432 (vectorizable_load): Likewise.
4433 (ensure_base_align): Take a dr_vec_info instead of a data_reference.
4434 Update calls to other routines for which the same change has been made.
4436 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4438 * tree-vectorizer.h (vec_info::move_dr): New member function.
4439 (dataref_aux): Rename to...
4440 (dr_vec_info): ...this and add "dr" and "stmt" fields.
4441 (_stmt_vec_info::dr_aux): Update accordingly.
4442 (_stmt_vec_info::data_ref_info): Delete.
4443 (STMT_VINFO_GROUPED_ACCESS, DR_GROUP_FIRST_ELEMENT)
4444 (DR_GROUP_NEXT_ELEMENT, DR_GROUP_SIZE, DR_GROUP_STORE_COUNT)
4445 (DR_GROUP_GAP, DR_GROUP_SAME_DR_STMT, REDUC_GROUP_FIRST_ELEMENT):
4446 (REDUC_GROUP_NEXT_ELEMENT, REDUC_GROUP_SIZE): Use dr_aux.dr instead
4448 (STMT_VINFO_DATA_REF): Likewise. Turn into an lvalue.
4449 (STMT_VINFO_DR_INFO): New macro.
4450 (DR_VECT_AUX): Use STMT_VINFO_DR_INKFO and vect_dr_stmt.
4451 (set_dr_misalignment): Update after rename of dataref_aux.
4452 (vect_dr_stmt): Move earlier in file. Return dr_aux.stmt.
4453 * tree-vect-stmts.c (new_stmt_vec_info): Remove redundant
4454 initialization of STMT_VINFO_DATA_REF.
4455 * tree-vectorizer.c (vec_info::move_dr): New function.
4456 * tree-vect-patterns.c (vect_recog_bool_pattern)
4457 (vect_recog_mask_conversion_pattern)
4458 (vect_recog_gather_scatter_pattern): Use it.
4459 * tree-vect-data-refs.c (vect_analyze_data_refs): Initialize
4460 the "dr" and "stmt" fields of dr_vec_info instead of
4461 STMT_VINFO_DATA_REF.
4463 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4465 * tree-vectorizer.h (_stmt_vec_info::pattern_stmt_p): New field.
4466 (is_pattern_stmt_p): Use it.
4467 * tree-vect-patterns.c (vect_init_pattern_stmt): Set pattern_stmt_p
4468 on pattern statements.
4470 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4472 * tree-vect-patterns.c (vect_mark_pattern_stmts): Take the
4473 original stmt as a stmt_vec_info rather than a gimple stmt.
4474 (vect_pattern_recog_1): Take the statement directly as a
4475 stmt_vec_info, rather than via a gimple_stmt_iterator.
4476 Update call to vect_mark_pattern_stmts.
4477 (vect_pattern_recog): Update calls accordingly.
4479 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4481 * tree-vectorizer.h (vect_get_vec_defs_for_stmt_copy)
4482 (vect_get_vec_def_for_stmt_copy): Take a vec_info rather than
4483 a vect_def_type for the first argument.
4484 * tree-vect-stmts.c (vect_get_vec_defs_for_stmt_copy): Likewise.
4485 (vect_get_vec_def_for_stmt_copy): Likewise. Return the original
4486 operand if it isn't defined by a vectorized statement.
4487 (vect_build_gather_load_calls): Remove the mask_dt argument and
4488 update calls to vect_get_vec_def_for_stmt_copy.
4489 (vectorizable_bswap): Likewise the dt argument.
4490 (vectorizable_call): Update calls to vectorizable_bswap and
4491 vect_get_vec_def_for_stmt_copy.
4492 (vectorizable_simd_clone_call, vectorizable_assignment)
4493 (vectorizable_shift, vectorizable_operation, vectorizable_condition)
4494 (vectorizable_comparison): Update calls to
4495 vect_get_vec_def_for_stmt_copy.
4496 (vectorizable_store): Likewise. Remove now-unnecessary calls to
4498 (vect_get_loop_based_defs): Remove dt argument and update call
4499 to vect_get_vec_def_for_stmt_copy.
4500 (vectorizable_conversion): Update calls to vect_get_loop_based_defs
4501 and vect_get_vec_def_for_stmt_copy.
4502 (vectorizable_load): Update calls to vect_build_gather_load_calls
4503 and vect_get_vec_def_for_stmt_copy.
4504 * tree-vect-loop.c (vect_create_epilog_for_reduction)
4505 (vectorizable_reduction, vectorizable_live_operation): Update calls
4506 to vect_get_vec_def_for_stmt_copy.
4508 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4510 * tree-vect-data-refs.c (vect_record_base_alignment): Replace vec_info
4511 and gimple stmt arguments with a stmt_vec_info.
4512 (vect_record_base_alignments): Update calls accordingly.
4513 * tree-vect-slp.c (vect_record_max_nunits): Replace vec_info
4514 and gimple stmt arguments with a stmt_vec_info.
4515 (vect_build_slp_tree_1): Remove vinfo argument and update call
4516 to vect_record_max_nunits.
4517 (vect_build_slp_tree_2): Update calls to vect_build_slp_tree_1
4518 and vect_record_max_nunits.
4520 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4522 * tree-vectorizer.h (nested_in_vect_loop_p): Move further down
4523 file and take a stmt_vec_info instead of a gimple stmt.
4524 (supportable_widening_operation, vect_finish_replace_stmt)
4525 (vect_finish_stmt_generation, vect_get_store_rhs)
4526 (vect_get_vec_def_for_operand_1, vect_get_vec_def_for_operand)
4527 (vect_get_vec_defs, vect_init_vector, vect_transform_stmt)
4528 (vect_remove_stores, vect_analyze_stmt, vectorizable_condition)
4529 (vect_get_smallest_scalar_type, vect_check_gather_scatter)
4530 (vect_create_data_ref_ptr, bump_vector_ptr)
4531 (vect_permute_store_chain, vect_setup_realignment)
4532 (vect_transform_grouped_load, vect_record_grouped_load_vectors)
4533 (vect_create_addr_base_for_vector_ref, vectorizable_live_operation)
4534 (vectorizable_reduction, vectorizable_induction)
4535 (get_initial_def_for_reduction, is_simple_and_all_uses_invariant)
4536 (vect_get_place_in_interleaving_chain): Take stmt_vec_infos rather
4537 than gimple stmts as arguments.
4538 * tree-vect-data-refs.c (vect_get_smallest_scalar_type)
4539 (vect_preserves_scalar_order_p, vect_slp_analyze_node_dependences)
4540 (can_group_stmts_p, vect_check_gather_scatter)
4541 (vect_create_addr_base_for_vector_ref, vect_create_data_ref_ptr)
4542 (bump_vector_ptr, vect_permute_store_chain, vect_setup_realignment)
4543 (vect_permute_load_chain, vect_shift_permute_load_chain)
4544 (vect_transform_grouped_load)
4545 (vect_record_grouped_load_vectors): Likewise.
4546 * tree-vect-loop.c (vect_fixup_reduc_chain)
4547 (get_initial_def_for_reduction, vect_create_epilog_for_reduction)
4548 (vectorize_fold_left_reduction, is_nonwrapping_integer_induction)
4549 (vectorizable_reduction, vectorizable_induction)
4550 (vectorizable_live_operation, vect_loop_kill_debug_uses): Likewise.
4551 * tree-vect-patterns.c (type_conversion_p, adjust_bool_stmts)
4552 (vect_get_load_store_mask): Likewise.
4553 * tree-vect-slp.c (vect_get_place_in_interleaving_chain)
4554 (vect_analyze_slp_instance, vect_mask_constant_operand_p): Likewise.
4555 * tree-vect-stmts.c (vect_mark_relevant)
4556 (is_simple_and_all_uses_invariant)
4557 (exist_non_indexing_operands_for_use_p, process_use)
4558 (vect_init_vector_1, vect_init_vector, vect_get_vec_def_for_operand_1)
4559 (vect_get_vec_def_for_operand, vect_get_vec_defs)
4560 (vect_finish_stmt_generation_1, vect_finish_replace_stmt)
4561 (vect_finish_stmt_generation, vect_truncate_gather_scatter_offset)
4562 (compare_step_with_zero, vect_get_store_rhs, get_group_load_store_type)
4563 (get_negative_load_store_type, get_load_store_type)
4564 (vect_check_load_store_mask, vect_check_store_rhs)
4565 (vect_build_gather_load_calls, vect_get_strided_load_store_ops)
4566 (vectorizable_bswap, vectorizable_call, vectorizable_simd_clone_call)
4567 (vect_create_vectorized_demotion_stmts, vectorizable_conversion)
4568 (vectorizable_assignment, vectorizable_shift, vectorizable_operation)
4569 (get_group_alias_ptr_type, vectorizable_store, hoist_defs_of_uses)
4570 (vectorizable_load, vectorizable_condition, vectorizable_comparison)
4571 (vect_analyze_stmt, vect_transform_stmt, vect_remove_stores)
4572 (supportable_widening_operation): Likewise.
4574 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4576 * tree-vect-data-refs.c (vect_describe_gather_scatter_call): Take
4577 a stmt_vec_info instead of a gcall.
4578 (vect_check_gather_scatter): Update call accordingly.
4579 * tree-vect-loop-manip.c (iv_phi_p): Take a stmt_vec_info instead
4581 (vect_can_advance_ivs_p, vect_update_ivs_after_vectorizer)
4582 (slpeel_update_phi_nodes_for_loops):): Update calls accordingly.
4583 * tree-vect-loop.c (vect_transform_loop_stmt): Take a stmt_vec_info
4584 instead of a gimple stmt.
4585 (vect_transform_loop): Update calls accordingly.
4586 * tree-vect-slp.c (vect_split_slp_store_group): Take and return
4587 stmt_vec_infos instead of gimple stmts.
4588 (vect_analyze_slp_instance): Update use accordingly.
4589 * tree-vect-stmts.c (read_vector_array, write_vector_array)
4590 (vect_clobber_variable, vect_stmt_relevant_p, permute_vec_elements)
4591 (vect_use_strided_gather_scatters_p, vect_build_all_ones_mask)
4592 (vect_build_zero_merge_argument, vect_get_gather_scatter_ops)
4593 (vect_gen_widened_results_half, vect_get_loop_based_defs)
4594 (vect_create_vectorized_promotion_stmts, can_vectorize_live_stmts):
4595 Take a stmt_vec_info instead of a gimple stmt and pass stmt_vec_infos
4596 down to subroutines.
4598 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4600 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Change the type
4601 of the worklist from a vector of gimple stmts to a vector of
4603 * tree-vect-stmts.c (vect_mark_relevant, process_use)
4604 (vect_mark_stmts_to_be_vectorized): Likewise
4606 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4608 * tree-vect-loop.c (vect_analyze_loop_operations): Look up the
4609 statement before passing it to vect_analyze_stmt.
4610 (vect_create_epilog_for_reduction): Use a stmt_vec_info to walk
4611 the chain of phi vector definitions. Track the exit phi via its
4613 (vectorizable_reduction): Set cond_stmt_vinfo directly from the
4614 STMT_VINFO_REDUC_DEF.
4615 * tree-vect-slp.c (vect_get_place_in_interleaving_chain): Use
4616 stmt_vec_infos to handle the statement chains.
4617 (vect_get_slp_defs): Record the first statement in the node
4618 using a stmt_vec_info.
4619 * tree-vect-stmts.c (vect_mark_stmts_to_be_vectorized): Look up
4620 statements here and pass their stmt_vec_info down to subroutines.
4621 (vect_init_vector_1): Hoist call to vinfo_for_stmt and pass it
4622 down to vect_finish_stmt_generation.
4623 (vect_init_vector, vect_get_vec_defs, vect_finish_replace_stmt)
4624 (vect_finish_stmt_generation): Call vinfo_for_stmt and pass
4625 stmt_vec_infos to subroutines.
4626 (vect_remove_stores): Use stmt_vec_infos to handle the statement
4629 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4631 * tree-vect-data-refs.c (vect_slp_analyze_node_dependences):
4632 (vect_check_gather_scatter, vect_create_data_ref_ptr, bump_vector_ptr)
4633 (vect_permute_store_chain, vect_setup_realignment)
4634 (vect_permute_load_chain, vect_shift_permute_load_chain)
4635 (vect_transform_grouped_load): Use stmt_vec_info rather than gimple
4636 stmts internally, and when passing values to other vectorizer routines.
4637 * tree-vect-loop-manip.c (vect_can_advance_ivs_p): Likewise.
4638 * tree-vect-loop.c (vect_analyze_scalar_cycles_1)
4639 (vect_analyze_loop_operations, get_initial_def_for_reduction)
4640 (vect_create_epilog_for_reduction, vectorize_fold_left_reduction)
4641 (vectorizable_reduction, vectorizable_induction)
4642 (vectorizable_live_operation, vect_transform_loop_stmt)
4643 (vect_transform_loop): Likewise.
4644 * tree-vect-patterns.c (vect_reassociating_reduction_p)
4645 (vect_recog_widen_op_pattern, vect_recog_mixed_size_cond_pattern)
4646 (vect_recog_bool_pattern, vect_recog_gather_scatter_pattern): Likewise.
4647 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
4648 (vect_slp_analyze_node_operations_1): Likewise.
4649 * tree-vect-stmts.c (vect_mark_relevant, process_use)
4650 (exist_non_indexing_operands_for_use_p, vect_init_vector_1)
4651 (vect_mark_stmts_to_be_vectorized, vect_get_vec_def_for_operand)
4652 (vect_finish_stmt_generation_1, get_group_load_store_type)
4653 (get_load_store_type, vect_build_gather_load_calls)
4654 (vectorizable_bswap, vectorizable_call, vectorizable_simd_clone_call)
4655 (vect_create_vectorized_demotion_stmts, vectorizable_conversion)
4656 (vectorizable_assignment, vectorizable_shift, vectorizable_operation)
4657 (vectorizable_store, vectorizable_load, vectorizable_condition)
4658 (vectorizable_comparison, vect_analyze_stmt, vect_transform_stmt)
4659 (supportable_widening_operation): Likewise.
4660 (vect_get_vector_types_for_stmt): Likewise.
4661 * tree-vectorizer.h (vect_dr_behavior): Likewise.
4663 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4665 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence)
4666 (vect_slp_analyze_node_dependences, vect_analyze_data_ref_accesses)
4667 (vect_permute_store_chain, vect_permute_load_chain)
4668 (vect_shift_permute_load_chain, vect_transform_grouped_load): Avoid
4669 repeated stmt_vec_info lookups.
4670 * tree-vect-loop-manip.c (vect_can_advance_ivs_p): Likewise.
4671 (vect_update_ivs_after_vectorizer): Likewise.
4672 * tree-vect-loop.c (vect_is_simple_reduction): Likewise.
4673 (vect_create_epilog_for_reduction, vectorizable_reduction): Likewise.
4674 * tree-vect-patterns.c (adjust_bool_stmts): Likewise.
4675 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
4676 (vect_bb_slp_scalar_cost): Likewise.
4677 * tree-vect-stmts.c (get_group_alias_ptr_type): Likewise.
4679 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4681 * tree-vect-data-refs.c (vect_check_gather_scatter): Pass the
4682 gcall rather than the generic gimple stmt to gimple_call_internal_fn.
4683 (vect_get_smallest_scalar_type, can_group_stmts_p): Use dyn_cast
4684 to get gassigns and gcalls, rather than operating on generc gimple
4686 * tree-vect-stmts.c (exist_non_indexing_operands_for_use_p)
4687 (vect_mark_stmts_to_be_vectorized, vectorizable_store)
4688 (vectorizable_load, vect_analyze_stmt): Likewise.
4689 * tree-vect-loop.c (vectorizable_reduction): Likewise gphi.
4691 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4693 * tree-vectorizer.h (get_earlier_stmt, get_later_stmt): Take and
4694 return stmt_vec_infos rather than gimple stmts. Do not accept
4696 (vect_find_last_scalar_stmt_in_slp): Return a stmt_vec_info instead
4698 * tree-vect-slp.c (vect_find_last_scalar_stmt_in_slp): Likewise.
4699 Update use of get_later_stmt.
4700 (vect_get_constant_vectors): Update call accordingly.
4701 (vect_schedule_slp_instance): Likewise
4702 * tree-vect-data-refs.c (vect_slp_analyze_node_dependences): Likewise.
4703 (vect_slp_analyze_instance_dependence): Likewise.
4704 (vect_preserves_scalar_order_p): Update use of get_earlier_stmt.
4706 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4708 * tree-vectorizer.h (stmt_info_for_cost::stmt): Replace with...
4709 (stmt_info_for_cost::stmt_info): ...this new field.
4710 (add_stmt_costs): Update accordingly.
4711 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost)
4712 (vect_get_known_peeling_cost): Likewise.
4713 (vect_estimate_min_profitable_iters): Likewise.
4714 * tree-vect-stmts.c (record_stmt_cost): Likewise.
4716 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4718 * tree-vectorizer.h (_loop_vec_info::may_misalign_stmts): Change
4719 from an auto_vec<gimple *> to an auto_vec<stmt_vec_info>.
4720 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Update
4722 * tree-vect-loop-manip.c (vect_create_cond_for_align_checks): Likewise.
4724 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4726 * tree-vectorizer.h (_stmt_vec_info::same_dr_stmt): Change from
4727 a gimple stmt to a stmt_vec_info.
4728 * tree-vect-stmts.c (vectorizable_load): Update accordingly.
4730 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4732 * tree-vectorizer.h (vec_info::grouped_stores): Change from
4733 an auto_vec<gimple *> to an auto_vec<stmt_vec_info>.
4734 (_loop_vec_info::reduction_chains): Likewise.
4735 * tree-vect-loop.c (vect_fixup_scalar_cycles_with_patterns): Update
4737 * tree-vect-slp.c (vect_analyze_slp): Likewise.
4739 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4741 * tree-vectorizer.h (_stmt_vec_info::first_element): Change from
4742 a gimple stmt to a stmt_vec_info.
4743 (_stmt_vec_info::next_element): Likewise.
4744 * tree-vect-data-refs.c (vect_update_misalignment_for_peel)
4745 (vect_slp_analyze_and_verify_node_alignment)
4746 (vect_analyze_group_access_1, vect_analyze_group_access)
4747 (vect_small_gap_p, vect_prune_runtime_alias_test_list)
4748 (vect_create_data_ref_ptr, vect_record_grouped_load_vectors)
4749 (vect_supportable_dr_alignment): Update accordingly.
4750 * tree-vect-loop.c (vect_fixup_reduc_chain): Likewise.
4751 (vect_fixup_scalar_cycles_with_patterns, vect_is_slp_reduction)
4752 (vect_is_simple_reduction, vectorizable_reduction): Likewise.
4753 * tree-vect-patterns.c (vect_reassociating_reduction_p): Likewise.
4754 * tree-vect-slp.c (vect_build_slp_tree_1)
4755 (vect_attempt_slp_rearrange_stmts, vect_supported_load_permutation_p)
4756 (vect_split_slp_store_group, vect_analyze_slp_instance)
4757 (vect_analyze_slp, vect_transform_slp_perm_load): Likewise.
4758 * tree-vect-stmts.c (vect_model_store_cost, vect_model_load_cost)
4759 (get_group_load_store_type, get_load_store_type)
4760 (get_group_alias_ptr_type, vectorizable_store, vectorizable_load)
4761 (vect_transform_stmt, vect_remove_stores): Likewise.
4763 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4765 * tree-vectorizer.h (vect_dr_stmt): Return a stmt_vec_info rather
4767 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence)
4768 (vect_slp_analyze_data_ref_dependence, vect_record_base_alignments)
4769 (vect_calculate_target_alignmentm, vect_compute_data_ref_alignment)
4770 (vect_update_misalignment_for_peel, vect_verify_datarefs_alignment)
4771 (vector_alignment_reachable_p, vect_get_data_access_cost)
4772 (vect_get_peeling_costs_all_drs, vect_peeling_hash_get_lowest_cost)
4773 (vect_peeling_supportable, vect_enhance_data_refs_alignment)
4774 (vect_find_same_alignment_drs, vect_analyze_data_refs_alignment)
4775 (vect_analyze_group_access_1, vect_analyze_group_access)
4776 (vect_analyze_data_ref_access, vect_analyze_data_ref_accesses)
4777 (vect_vfa_access_size, vect_small_gap_p, vect_analyze_data_refs)
4778 (vect_supportable_dr_alignment): Remove vinfo_for_stmt from the
4779 result of vect_dr_stmt and use the stmt_vec_info instead of
4780 the associated gimple stmt.
4781 * tree-vect-loop-manip.c (get_misalign_in_elems): Likewise.
4782 (vect_gen_prolog_loop_niters): Likewise.
4783 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
4785 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4787 * tree-vectorizer.h (_slp_tree::stmts): Change from a vec<gimple *>
4788 to a vec<stmt_vec_info>.
4789 * tree-vect-slp.c (vect_free_slp_tree): Update accordingly.
4790 (vect_create_new_slp_node): Take a vec<gimple *> instead of a
4792 (_slp_oprnd_info::def_stmts): Change from a vec<gimple *>
4793 to a vec<stmt_vec_info>.
4794 (bst_traits::value_type, bst_traits::value_type): Likewise.
4795 (bst_traits::hash): Update accordingly.
4796 (vect_get_and_check_slp_defs): Change the stmts parameter from
4797 a vec<gimple *> to a vec<stmt_vec_info>.
4798 (vect_two_operations_perm_ok_p, vect_build_slp_tree_1): Likewise.
4799 (vect_build_slp_tree): Likewise.
4800 (vect_build_slp_tree_2): Likewise. Update uses of
4801 SLP_TREE_SCALAR_STMTS.
4802 (vect_print_slp_tree): Update uses of SLP_TREE_SCALAR_STMTS.
4803 (vect_mark_slp_stmts, vect_mark_slp_stmts_relevant)
4804 (vect_slp_rearrange_stmts, vect_attempt_slp_rearrange_stmts)
4805 (vect_supported_load_permutation_p, vect_find_last_scalar_stmt_in_slp)
4806 (vect_detect_hybrid_slp_stmts, vect_slp_analyze_node_operations_1)
4807 (vect_slp_analyze_node_operations, vect_slp_analyze_operations)
4808 (vect_bb_slp_scalar_cost, vect_slp_analyze_bb_1)
4809 (vect_get_constant_vectors, vect_get_slp_defs)
4810 (vect_transform_slp_perm_load, vect_schedule_slp_instance)
4811 (vect_remove_slp_scalar_calls, vect_schedule_slp): Likewise.
4812 (vect_analyze_slp_instance): Build up a vec of stmt_vec_infos
4813 instead of gimple stmts.
4814 * tree-vect-data-refs.c (vect_slp_analyze_node_dependences): Change
4815 the stores parameter for a vec<gimple *> to a vec<stmt_vec_info>.
4816 (vect_slp_analyze_instance_dependence): Update uses of
4817 SLP_TREE_SCALAR_STMTS.
4818 (vect_slp_analyze_and_verify_node_alignment): Likewise.
4819 (vect_slp_analyze_and_verify_instance_alignment): Likewise.
4820 * tree-vect-loop.c (neutral_op_for_slp_reduction): Likewise.
4821 (get_initial_defs_for_reduction): Likewise.
4822 (vect_create_epilog_for_reduction): Likewise.
4823 (vectorize_fold_left_reduction): Likewise.
4824 * tree-vect-stmts.c (vect_prologue_cost_for_slp_op): Likewise.
4825 (vect_model_simple_cost, vectorizable_shift, vectorizable_load)
4826 (can_vectorize_live_stmts): Likewise.
4828 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4830 * tree-vectorizer.h (_loop_vec_info::reductions): Change from an
4831 auto_vec<gimple *> to an auto_vec<stmt_vec_info>.
4832 (vect_force_simple_reduction): Take and return stmt_vec_infos rather
4834 * tree-parloops.c (valid_reduction_p): Take a stmt_vec_info instead
4836 (gather_scalar_reductions): Update after above interface changes.
4837 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Likewise.
4838 (vect_is_simple_reduction): Take and return stmt_vec_infos rather
4840 (vect_force_simple_reduction): Likewise.
4841 * tree-vect-patterns.c (vect_pattern_recog_1): Update use of
4842 LOOP_VINFO_REDUCTIONS.
4843 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
4845 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4847 * tree-vectorizer.h (_stmt_vec_info::reduc_def): Change from
4848 a gimple stmt to a stmt_vec_info.
4849 * tree-vect-loop.c (vect_active_double_reduction_p)
4850 (vect_force_simple_reduction, vectorizable_reduction): Update
4853 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4855 * tree-vectorizer.h (_slp_tree::vec_stmts): Change from a
4856 vec<gimple *> to a vec<stmt_vec_info>.
4857 * tree-vect-loop.c (vect_create_epilog_for_reduction): Change
4858 the reduction_phis argument from a vec<gimple *> to a
4860 (vectorizable_reduction): Likewise the phis local variable that
4861 is passed to vect_create_epilog_for_reduction. Update for new type
4862 of SLP_TREE_VEC_STMTS.
4863 (vectorizable_induction): Update for new type of SLP_TREE_VEC_STMTS.
4864 (vectorizable_live_operation): Likewise.
4865 * tree-vect-slp.c (vect_get_slp_vect_defs): Likewise.
4866 (vect_transform_slp_perm_load, vect_schedule_slp_instance): Likewise.
4868 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4870 * tree-vectorizer.h (_stmt_vec_info::vectorized_stmt): Change from
4871 a gimple stmt to a stmt_vec_info.
4872 (vectorizable_condition, vectorizable_live_operation)
4873 (vectorizable_reduction, vectorizable_induction): Pass back the
4874 vectorized statement as a stmt_vec_info.
4875 * tree-vect-data-refs.c (vect_record_grouped_load_vectors): Update
4876 use of STMT_VINFO_VEC_STMT.
4877 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise,
4878 accumulating the inner phis that feed the STMT_VINFO_VEC_STMT
4879 as stmt_vec_infos rather than gimple stmts.
4880 (vectorize_fold_left_reduction): Change vec_stmt from a gimple stmt
4882 (vectorizable_live_operation): Likewise.
4883 (vectorizable_reduction, vectorizable_induction): Likewise,
4884 updating use of STMT_VINFO_VEC_STMT.
4885 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Update use
4886 of STMT_VINFO_VEC_STMT.
4887 (vect_build_gather_load_calls, vectorizable_bswap, vectorizable_call)
4888 (vectorizable_simd_clone_call, vectorizable_conversion)
4889 (vectorizable_assignment, vectorizable_shift, vectorizable_operation)
4890 (vectorizable_store, vectorizable_load, vectorizable_condition)
4891 (vectorizable_comparison, can_vectorize_live_stmts): Change vec_stmt
4892 from a gimple stmt to a stmt_vec_info.
4893 (vect_transform_stmt): Update use of STMT_VINFO_VEC_STMT. Pass a
4894 pointer to a stmt_vec_info to the vectorizable_* routines.
4896 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4898 * tree-vectorizer.h (_stmt_vec_info::related_stmt): Change from
4899 a gimple stmt to a stmt_vec_info.
4900 (is_pattern_stmt_p): Update accordingly.
4901 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Likewise.
4902 (vect_record_grouped_load_vectors): Likewise.
4903 * tree-vect-loop.c (vect_determine_vf_for_stmt): Likewise.
4904 (vect_fixup_reduc_chain, vect_update_vf_for_slp): Likewise.
4905 (vect_model_reduction_cost): Likewise.
4906 (vect_create_epilog_for_reduction): Likewise.
4907 (vectorizable_reduction, vectorizable_induction): Likewise.
4908 * tree-vect-patterns.c (vect_init_pattern_stmt): Likewise.
4909 Return the stmt_vec_info for the pattern statement.
4910 (vect_set_pattern_stmt): Update use of STMT_VINFO_RELATED_STMT.
4911 (vect_split_statement, vect_mark_pattern_stmts): Likewise.
4912 * tree-vect-slp.c (vect_detect_hybrid_slp_stmts): Likewise.
4913 (vect_detect_hybrid_slp, vect_get_slp_defs): Likewise.
4914 * tree-vect-stmts.c (vect_mark_relevant): Likewise.
4915 (vect_get_vec_def_for_operand_1, vectorizable_call): Likewise.
4916 (vectorizable_simd_clone_call, vect_analyze_stmt, new_stmt_vec_info)
4917 (free_stmt_vec_info, vect_is_simple_use): Likewise.
4919 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4921 * tree-vectorizer.h (vect_finish_replace_stmt): Return a stmt_vec_info
4922 (vect_finish_stmt_generation): Likewise.
4923 * tree-vect-stmts.c (vect_finish_stmt_generation_1): Likewise.
4924 (vect_finish_replace_stmt, vect_finish_stmt_generation): Likewise.
4925 (vect_build_gather_load_calls): Use the return value of the above
4926 functions instead of a separate call to vinfo_for_stmt. Use narrow
4927 scopes for the input gimple stmt and wider scopes for the associated
4928 stmt_vec_info. Use vec_info::lookup_def when setting these
4929 stmt_vec_infos from an SSA_NAME definition.
4930 (vectorizable_bswap, vectorizable_call, vectorizable_simd_clone_call)
4931 (vect_create_vectorized_demotion_stmts, vectorizable_conversion)
4932 (vectorizable_assignment, vectorizable_shift, vectorizable_operation)
4933 (vectorizable_store, vectorizable_load, vectorizable_condition)
4934 (vectorizable_comparison): Likewise.
4935 * tree-vect-loop.c (vectorize_fold_left_reduction): Likewise.
4936 (vectorizable_reduction): Likewise.
4938 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4940 * tree-vectorizer.h (vect_is_simple_use): Add an optional
4941 stmt_vec_info * parameter before the optional gimple **.
4942 * tree-vect-stmts.c (vect_is_simple_use): Likewise.
4943 (process_use, vect_get_vec_def_for_operand_1): Update callers.
4944 (vect_get_vec_def_for_operand, vectorizable_shift): Likewise.
4945 * tree-vect-loop.c (vectorizable_reduction): Likewise.
4946 (vectorizable_live_operation): Likewise.
4947 * tree-vect-patterns.c (type_conversion_p): Likewise.
4948 (vect_look_through_possible_promotion): Likewise.
4949 (vect_recog_rotate_pattern): Likewise.
4950 * tree-vect-slp.c (vect_get_and_check_slp_defs): Likewise.
4952 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4954 * tree-vectorizer.h (stmt_vec_info): Temporarily change from
4955 a typedef to a wrapper class.
4956 (NULL_STMT_VEC_INFO): New macro.
4957 (vec_info::stmt_infos): Change to vec<stmt_vec_info>.
4958 (stmt_vec_info::operator*): New function.
4959 (stmt_vec_info::operator gimple *): Likewise.
4960 (set_vinfo_for_stmt): Use NULL_STMT_VEC_INFO.
4961 (add_stmt_costs): Likewise.
4962 * tree-vect-loop-manip.c (iv_phi_p): Likewise.
4963 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost)
4964 (vect_get_known_peeling_cost): Likewise.
4965 (vect_estimate_min_profitable_iters): Likewise.
4966 * tree-vect-patterns.c (vect_init_pattern_stmt): Likewise.
4967 * tree-vect-slp.c (vect_remove_slp_scalar_calls): Likewise.
4968 * tree-vect-stmts.c (vect_build_gather_load_calls): Likewise.
4969 (vectorizable_store, free_stmt_vec_infos): Likewise.
4970 (new_stmt_vec_info): Change return type of xcalloc to
4973 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4975 * tree-vectorizer.h (vec_info::lookup_single_use): Declare.
4976 * tree-vectorizer.c (vec_info::lookup_single_use): New function.
4977 * tree-vect-loop.c (vectorizable_reduction): Use it instead of
4978 a single_imm_use-based sequence.
4979 * tree-vect-stmts.c (supportable_widening_operation): Likewise.
4981 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4983 * tree-vectorizer.h (vec_info::lookup_def): Declare.
4984 * tree-vectorizer.c (vec_info::lookup_def): New function.
4985 * tree-vect-patterns.c (vect_get_internal_def): Use it.
4986 (vect_widened_op_tree): Likewise.
4987 * tree-vect-stmts.c (vect_is_simple_use): Likewise.
4988 * tree-vect-loop.c (vect_analyze_loop_operations): Likewise.
4989 (vectorizable_reduction): Likewise.
4990 (vect_valid_reduction_input_p): Take a stmt_vec_info instead
4992 (vect_is_slp_reduction): Update calls accordingly. Use
4993 vec_info::lookup_def.
4994 (vect_is_simple_reduction): Likewise
4995 * tree-vect-slp.c (vect_detect_hybrid_slp_1): Use vec_info::lookup_def.
4997 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4999 * tree-vectorizer.h (vec_info::lookup_stmt): Declare.
5000 * tree-vectorizer.c (vec_info::lookup_stmt): New function.
5001 * tree-vect-loop.c (vect_determine_vf_for_stmt): Use it instead
5003 (vect_determine_vectorization_factor, vect_analyze_scalar_cycles_1)
5004 (vect_compute_single_scalar_iteration_cost, vect_analyze_loop_form)
5005 (vect_update_vf_for_slp, vect_analyze_loop_operations)
5006 (vect_is_slp_reduction, vectorizable_induction)
5007 (vect_transform_loop_stmt, vect_transform_loop): Likewise.
5008 * tree-vect-patterns.c (vect_init_pattern_stmt):
5009 (vect_determine_min_output_precision_1, vect_determine_precisions)
5010 (vect_pattern_recog): Likewise.
5011 * tree-vect-stmts.c (vect_analyze_stmt, vect_transform_stmt): Likewise.
5012 * config/powerpcspe/powerpcspe.c (rs6000_density_test): Likewise.
5013 * config/rs6000/rs6000.c (rs6000_density_test): Likewise.
5014 * tree-vect-slp.c (vect_detect_hybrid_slp_stmts): Likewise.
5015 (vect_detect_hybrid_slp_1, vect_detect_hybrid_slp_2)
5016 (vect_detect_hybrid_slp): Likewise. Change the walk_stmt_info
5017 info field from a loop to a loop_vec_info.
5019 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
5021 * tree-vectorizer.h (stmt_vec_info): Move typedef earlier in file.
5022 (vec_info::add_stmt): Declare.
5023 * tree-vectorizer.c (vec_info::add_stmt): New function.
5024 * tree-vect-data-refs.c (vect_create_data_ref_ptr): Use it.
5025 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Likewise.
5026 (vect_create_epilog_for_reduction, vectorizable_reduction): Likewise.
5027 (vectorizable_induction): Likewise.
5028 * tree-vect-slp.c (_bb_vec_info::_bb_vec_info): Likewise.
5029 * tree-vect-stmts.c (vect_finish_stmt_generation_1): Likewise.
5030 (vectorizable_simd_clone_call, vectorizable_store): Likewise.
5031 (vectorizable_load): Likewise.
5032 * tree-vect-patterns.c (vect_init_pattern_stmt): Likewise.
5033 (vect_recog_bool_pattern, vect_recog_mask_conversion_pattern)
5034 (vect_recog_gather_scatter_pattern): Likewise.
5035 (append_pattern_def_seq): Likewise. Remove a check that is
5036 performed by add_stmt itself.
5038 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
5040 * tree-vect-loop.c (vectorizable_reduction): Fix an instance in
5041 which make_ssa_name was called with new_stmt before new_stmt
5044 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
5046 * tree-vect-loop.c (vect_valid_reduction_input_p): New function,
5048 (vect_is_slp_reduction): ...here...
5049 (vect_is_simple_reduction): ...and here. Remove repetition of tests
5050 that are already known to be false.
5052 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
5054 * tree-vectorizer.h (vect_free_slp_instance): Add a final_p parameter.
5055 * tree-vect-slp.c (vect_free_slp_tree): Likewise. Don't update
5056 STMT_VINFO_NUM_SLP_USES when it's true.
5057 (vect_free_slp_instance): Add a final_p parameter and pass it to
5059 (vect_build_slp_tree_2): Update call to vect_free_slp_instance.
5060 (vect_analyze_slp_instance): Likewise.
5061 (vect_slp_analyze_operations): Likewise.
5062 (vect_slp_analyze_bb_1): Likewise.
5063 * tree-vectorizer.c (vec_info): Likewise.
5064 * tree-vect-loop.c (vect_transform_loop): Likewise.
5066 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
5068 * tree-vect-loop.c (vectorizable_reduction): Assert that the
5069 function is not called for second and subsequent members of
5072 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
5074 * tree-vect-loop.c (get_initial_def_for_reduction): Move special
5075 cases for nested loops from here to ...
5076 (vect_create_epilog_for_reduction): ...here. Only call
5077 vect_is_simple_use for inner-loop reductions.
5079 2018-07-31 Martin Liska <mliska@suse.cz>
5081 PR gcov-profile/85338
5082 PR gcov-profile/85350
5083 PR gcov-profile/85372
5084 * profile.c (struct location_triplet): New.
5085 (struct location_triplet_hash): Likewise.
5086 (output_location): Do not output a BB that
5087 is already recorded for a line.
5088 (branch_prob): Use streamed_locations.
5090 2018-07-31 Martin Liska <mliska@suse.cz>
5092 PR gcov-profile/85370
5093 * coverage.c (coverage_begin_function): Do not mark target
5094 clones as artificial functions.
5096 2018-07-31 Martin Liska <mliska@suse.cz>
5098 PR gcov-profile/83813
5099 PR gcov-profile/84758
5100 PR gcov-profile/85217
5101 PR gcov-profile/85332
5102 * profile.c (branch_prob): Do not record GOTO expressions
5103 for GIMPLE statements which locations are already streamed.
5105 2018-07-31 Olivier Hainque <hainque@adacore.com>
5107 * gcc.c (handle_spec_function): Accept a soft_matched_part
5108 argument, as do_spec_1. Pass it down to ...
5109 (eval_spec_function): Accept a soft_matched_part argument,
5110 and pass it down to ...
5111 (do_spec_2): Accept a soft_matched_part argument, and pass
5112 it down to do_spec_1.
5113 (do_spec_1): Pass soft_matched_part to handle_spec_function.
5114 (handle_braces): Update call to handle_spec_function.
5115 (driver::set_up_specs): Update calls to do_spec_2.
5116 (compare_debug_dump_opt_spec_function): Likewise.
5117 (compare_debug_self_opt_spec_function): Likewise.
5119 2018-07-31 Olivier Hainque <hainque@adacore.com>
5121 * common.opt (nolibc): New option.
5122 * doc/invoke.texi (Link Options): Document it.
5123 * gcc.c (LINK_GCC_C_SEQUENCE_SPEC): Honor nolibc.
5124 * config/alpha/linux.h: Likewise.
5125 * config/arc/elf.h: Likewise.
5126 * config/arm/uclinux-elf.h: Likewise.
5127 * config/arm/unknown-elf.h: Likewise.
5128 * config/avr/avrlibc.h: Likewise.
5129 * config/bfin/bfin.h: Likewise.
5130 * config/bfin/linux.h: Likewise.
5131 * config/bfin/uclinux.h: Likewise.
5132 * config/darwin.h: Likewise.
5133 * config/darwin10.h: Likewise.
5134 * config/darwin12.h: Likewise.
5135 * config/gnu-user.h: Likewise.
5136 * config/lm32/uclinux-elf.h: Likewise.
5137 * config/pa/pa-hpux11.h: Likewise.
5138 * config/pa/pa64-hpux.h: Likewise.
5139 * config/sparc/sparc.h: Likewise.
5141 2018-07-31 Olivier Hainque <hainque@adacore.com>
5143 * gcc.c (getenv_spec_function): Prepend '/' to value for allowed
5144 undefined variables.
5146 2018-07-30 Segher Boessenkool <segher@kernel.crashing.org>
5149 * config/arm/arm.c (arm_block_set_aligned_vect): Use gen_int_mode
5152 2018-07-30 Bernd Edlinger <bernd.edlinger@hotmail.de>
5154 * tree-ssa-forwprop.c (simplify_builtin_call): Don't create a not NUL
5155 terminated string literal.
5157 2018-07-30 Segher Boessenkool <segher@kernel.crashing.org>
5159 PR rtl-optimization/85160
5160 * combine.c (is_just_move): New function.
5161 (try_combine): Allow combining two instructions into two if neither of
5162 the original instructions was a move.
5164 2018-07-30 Alexander Monakov <amonakov@ispras.ru>
5167 * doc/extend.texi (Global Register Variables): Discourage use of type
5169 (Local Register Variables): Likewise.
5171 2018-07-30 Richard Sandiford <richard.sandiford@arm.com>
5173 PR tree-optimization/86506
5174 * hwint.h (ceil_log2): Resync with hwint.c implementation.
5176 2018-07-30 Ilya Leoshkevich <iii@linux.ibm.com>
5179 * lra-constraints.c (spill_hard_reg_in_range): When selecting the
5180 hard_regno, make sure no insn between `from` and `to` clobbers it.
5182 2018-07-30 Cesar Philippidis <cesar@codesourcery.com>
5183 Tom de Vries <tdevries@suse.de>
5185 * config/nvptx/nvptx.c (PTX_GANG_DEFAULT): Rename to ...
5186 (PTX_DEFAULT_RUNTIME_DIM): ... this.
5187 (nvptx_goacc_validate_dims): Set default worker and gang dims to
5188 PTX_DEFAULT_RUNTIME_DIM.
5189 (nvptx_dim_limit): Ignore GOMP_DIM_WORKER.
5191 2018-07-29 John David Anglin <danglin@gcc.gnu.org>
5193 * config/pa/pa.c (pa_output_addr_vec): Align address table.
5194 * config/pa/pa.h (JUMP_TABLES_IN_TEXT_SECTION): Revise comment.
5195 * config/pa/pa32-linux.h (JUMP_TABLES_IN_TEXT_SECTION): Define.
5197 2018-07-27 Michael Meissner <meissner@linux.ibm.com>
5199 * config/rs6000/constraints.md (wG constraint): Delete, no longer
5201 * config/rs6000/predicates.md (p9_fusion_reg_operand): Rename
5202 predicate to reflect toc fusion has been deleted.
5203 (toc_fusion_mem_raw): Delete, no longer used.
5204 (toc_fusion_mem_wrapped): Likewise.
5205 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete toc
5207 * config/rs6000/rs6000-protos.h (fusion_wrap_memory_address):
5208 Delete, no longer used.
5209 * config/rs6000/rs6000.c (struct rs6000_reg_addr): Delete fields
5210 meant to be used for toc fusion.
5211 (rs6000_debug_print_mode): Delete toc fusion debugging.
5212 (rs6000_debug_reg_global): Likewise.
5213 (rs6000_init_hard_regno_mode_ok): Delete setting up fields for toc
5214 fusion and secondary reload support that were never used.
5215 (rs6000_option_override_internal): Delete TOC fusion, that was only
5216 partially defined, and it did not work unless you also used the
5218 (rs6000_legitimate_address_p): Delete TOC fusion support.
5219 (rs6000_opt_masks): Likewise.
5220 (fusion_wrap_memory_address): Delete function, no longer used.
5221 (fusion_split_address); Delete TOC fusion support.
5222 * config/rs6000/rs6000.h (TARGET_TOC_FUSION_INT): Delete, no
5223 longer used with toc fusion being deleted.
5224 (TARGET_TOC_FUSION_FP): Likewise.
5225 * config/rs6000/rs6000.md (UNSPEC_FUSION_ADDIS): Delete TOC fusion
5227 (toc fusion spliter): Delete TOC fusion support.
5228 (toc_fusionload_<mode>): Likewise.
5229 (toc_fusionload_di): Likewise.
5230 (fusion_gpr_load_<mode>): Delete generator function, this insn no
5231 longer needs to be named. Rename predicate to delete TOC fusion.
5232 (fusion_gpr_<P:mode>_<GPR_FUSION:mode>_load): Likewise.
5233 (fusion_gpr_<P:mode>_<GPR_FUSION:mode>_store): Likewise.
5234 (fusion_vsx_<P:mode>_<GPR_FUSION:mode>_load): Likewise.
5235 (fusion_vsx_<P:mode>_<GPR_FUSION:mode>_store): Likewise.
5236 (p9 fusion peephole2s): Rename predicate to delete TOC fusion.
5238 2018-07-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
5240 * doc/extend.texi (Basic PowerPC Built-in Functions Available on
5241 ISA 2.05): Replace __uint128_t with __uint128 and __int128_t with
5242 __int128 in built-in function prototypes.
5243 (PowerPC AltiVec Built-in Functions on ISA 2.07): Likewise.
5244 (PowerPC AltiVec Built-in Functions on ISA 3.0): Likewise.
5246 2018-07-27 Martin Sebor <msebor@redhat.com>
5248 PR tree-optimization/86696
5249 * tree-ssa-strlen.c (get_min_string_length): Handle all integer
5250 types, including enums.
5251 (handle_char_store): Be prepared for the above function to fail.
5253 2018-07-26 Qing Zhao <qing.zhao@oracle.com>
5255 * builtins.c (inline_expand_builtin_string_cmp): Disable inlining
5256 when optimization level is lower than 2 or optimize for size.
5258 2018-07-26 Martin Sebor <msebor@redhat.com>
5260 PR tree-optimization/86043
5261 PR tree-optimization/86042
5262 * tree-ssa-strlen.c (handle_builtin_memcpy): Handle strict overlaps.
5263 (get_string_cst_length): Rename...
5264 (get_min_string_length): ...to this. Add argument.
5265 (handle_char_store): Extend to handle multi-character stores by
5267 * tree.c (initializer_zerop): Use new argument. Handle MEM_REF.
5268 * tree.h (initializer_zerop): Add argument.
5270 2018-07-26 Jakub Jelinek <jakub@redhat.com>
5273 * omp-low.c (scan_sharing_clauses): Don't ignore map clauses for
5274 declare target to variables if they have always,{to,from,tofrom} map
5277 2018-07-26 Martin Liska <mliska@suse.cz>
5280 * lto-wrapper.c: Add linker_output as prefix
5281 for ltrans_output_file.
5283 2018-07-26 Segher Boessenkool <segher@kernel.crashing.org>
5285 PR rtl-optimization/85805
5286 * combine.c (reg_nonzero_bits_for_combine): Only use the last set
5287 value for hard registers if that was written in the same mode.
5289 2018-07-26 Martin Liska <mliska@suse.cz>
5291 PR gcov-profile/86536
5292 * gcov.c (format_gcov): Use printf format %.*f directly
5293 and do not handle special values.
5295 2018-07-25 Claudiu Zissulescu <claziss@synopsys.com>
5297 * common/config/arc/arc-common.c (arc_option_optimization_table):
5298 Update default optimizations for size.
5300 2018-07-25 Claudiu Zissulescu <claziss@synopsys.com>
5302 * config/arc/arc.md (movsf_insn): Add short instruction selection.
5303 * config/arc/constraints.md (CfZ): New constraint.
5304 * config/arc/fpu.md (addssf3_fpu): Use CfZ constraint.
5305 (subsf3_fpu): Likewise.
5306 (cmpsf_fpu): Likewise.
5307 (cmpsf_fpu_uneq): Likewise.
5309 2018-07-25 Claudiu Zissulescu <claziss@synopsys.com>
5311 * config/arc/arc.c (compact_memory_operand_p): Check for uncached
5313 (arc_is_uncached_mem_p): uncached applies to both the variable and
5316 2018-07-25 Claudiu Zissulescu <claziss@synopsys.com>
5318 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Add additional
5321 2018-07-25 David Malcolm <dmalcolm@redhat.com>
5323 * optinfo-emit-json.cc (class optrecord_json_writer): Convert
5324 field "m_scopes" from vec to auto_vec.
5326 2018-07-25 Martin Liska <mliska@suse.cz>
5328 * config/powerpcspe/powerpcspe-protos.h (rs6000_loop_align): Fix
5331 2018-07-25 Richard Biener <rguenther@suse.de>
5334 * dwarf2out.c (dwarf2out_decl): Do not handle nested functions
5335 special wrt context_die late.
5336 (gen_subprogram_die): Re-use DIEs in local scope.
5338 2018-07-25 Richard Sandiford <richard.sandiford@arm.com>
5340 PR tree-optimization/86644
5341 * hwint.c (ceil_log2): Fix comment. Return 0 for 0.
5343 2018-07-25 Martin Liska <mliska@suse.cz>
5346 * dumpfile.c: And excluded values with TDF_ALL_VALUES.
5347 * dumpfile.h (enum dump_flag): Defince TDF_ALL_VALUES.
5349 2018-07-25 Martin Liska <mliska@suse.cz>
5352 * params.def: Explain ASan abbreviation and provide
5353 a documentation link.
5355 2018-07-24 Martin Sebor <msebor@redhat.com>
5357 PR tree-optimization/86622
5358 PR tree-optimization/86532
5359 * builtins.h (string_length): Declare.
5360 * builtins.c (c_strlen): Correct handling of non-constant offsets.
5361 (check_access): Be prepared for non-constant length ranges.
5362 (string_length): Make extern.
5363 * expr.c (string_constant): Only handle the minor non-constant
5364 array index. Use string_constant to compute the length of
5365 a generic string constant.
5367 2018-07-24 Richard Sandiford <richard.sandiford@arm.com>
5369 PR tree-optimization/86618
5370 * tree-vect-stmts.c (vectorizable_call): Don't take the address
5371 of LOOP_VINFO_MASKS (loop_vinfo) when loop_vinfo is null.
5373 2018-07-24 David Malcolm <dmalcolm@redhat.com>
5375 PR tree-optimization/86636
5376 * json.cc (json::object::set): Fix comment. Add assertions.
5377 (json::array::append): Move here from json.h. Add comment and an
5379 (json::string::string): Likewise.
5380 * json.h (json::array::append): Move to json.cc.
5381 (json::string::string): Likewise.
5382 * optinfo-emit-json.cc
5383 (optrecord_json_writer::impl_location_to_json): Assert that we
5384 aren't attempting to write out UNKNOWN_LOCATION, or an ad-hoc
5385 wrapper around it. Expand the location once, rather than three
5387 (optrecord_json_writer::inlining_chain_to_json): Fix the check for
5388 UNKNOWN_LOCATION, to use LOCATION_LOCUS to look through ad-hoc
5390 (optrecord_json_writer::optinfo_to_json): Likewise, in four
5391 places. Fix some overlong lines.
5393 2018-07-24 Matthew Malcomson <matthew.malcomson@arm.com>
5395 * config/aarch64/aarch64-simd.md
5396 (aarch64_<ANY_EXTEND:su><ADDSUB:optab>w<mode>): Split into...
5397 (aarch64_<ANY_EXTEND:su>subw<mode>): ... This...
5398 (aarch64_<ANY_EXTEND:su>addw<mode>): ... And this.
5399 (aarch64_<ANY_EXTEND:su><ADDSUB:optab>w<mode>_internal): Split into...
5400 (aarch64_<ANY_EXTEND:su>subw<mode>_internal): ... This...
5401 (aarch64_<ANY_EXTEND:su>addw<mode>_internal): ... And this.
5402 (aarch64_<ANY_EXTEND:su><ADDSUB:optab>w2<mode>_internal): Split into...
5403 (aarch64_<ANY_EXTEND:su>subw2<mode>_internal): ... This...
5404 (aarch64_<ANY_EXTEND:su>addw2<mode>_internal): ... And this.
5406 2018-07-24 Jakub Jelinek <jakub@redhat.com>
5409 * expmed.c (expand_divmod): Punt if d == HOST_WIDE_INT_MIN
5410 and size > HOST_BITS_PER_WIDE_INT. For size > HOST_BITS_PER_WIDE_INT
5411 and abs_d == d, do the power of two handling if profitable.
5413 2018-07-24 Richard Biener <rguenther@suse.de>
5415 * match.pd: Add BIT_FIELD_REF canonicalizations.
5417 2018-07-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
5420 * genmatch.c (dt_operand::gen_match_op): Avoid folding volatile values.
5422 2018-07-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
5424 * gimple-fold.c (gimple_fold_builtin_printf): Don't create a not NUL
5425 terminated STRING_CST object.
5427 2018-07-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
5429 hsa-dump.c (dump_hsa_symbol): Avoid out of scope access to buf.
5431 2018-07-23 Segher Boessenkool <segher@kernel.crashing.org>
5433 * config/rs6000/rs6000-p8swap.c (rtx_is_swappable_p): Adjust.
5434 * config/rs6000/rs6000-protos.h (rs6000_split_v4si_init): Delete.
5435 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Always force
5436 the elements into a register.
5437 (rs6000_split_v4si_init_di_reg): Delete.
5438 (rs6000_split_v4si_init): Delete.
5439 * config/rs6000/vsx.md (unspec): Delete UNSPEC_VSX_VEC_INIT.
5440 (vsx_init_v4si): Rewrite as a define_expand.
5442 2018-07-23 Segher Boessenkool <segher@kernel.crashing.org>
5444 * config/rs6000/rs6000.md (splitters for rldimi and rlwimi with the
5445 zero_extend argument from memory): New.
5447 2018-07-22 Martin Sebor <msebor@redhat.com>
5450 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Avoid
5451 diagnosing calls with unknown arguments unless -Walloca-larger-than
5452 is restricted to less than PTRDIFF_MAX bytes.
5454 2018-07-22 Gerald Pfeifer <gerald@pfeifer.com>
5456 * doc/gcov.texi (Invoking Gcov): Editorial changes.
5458 2018-07-20 David Malcolm <dmalcolm@redhat.com>
5460 * pretty-print.c (text_info::set_location): Remove redundant
5461 "line_table" parameter from call to rich_location::set_range.
5463 2018-07-20 Martin Sebor <msebor@redhat.com>
5466 * builtins.c (expand_builtin_alloca): Adjust.
5467 * calls.c (alloc_max_size): Simplify.
5468 * cgraphunit.c (cgraph_node::expand): Adjust.
5469 * common.opt (larger_than_size, warn_frame_larger_than): Remove
5471 (frame_larger_than_size): Same.
5472 (-Wframe-larger-than, -Wlarger-than, -Wstack-usage): Change options
5473 to take a HOST_WIDE_INT argument and accept a byte-size suffix.
5475 * doc/invoke.texi (GCC Command Options): Document option arguments.
5476 Explain byte-size arguments and suffixes.
5477 (-Wvla-larger-than, -Wno-alloc-size-larger-than): Update.
5478 (-Wno-alloca-larger-than, -Wno-vla-larger-than): Same.
5479 (-Wframe-larger-than, -Wlarger-than, -Wstack-usage): Same.
5480 * doc/options.texi (UInteger): Expand.
5481 (Host_Wide_Int, ByteSize): Document new properties.
5482 * final.c (final_start_function_1): Include sizes in an error message.
5483 * function.c (frame_offset_overflow): Same.
5484 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Adjust.
5485 (alloca_call_type_by_arg): Change function argument to HOST_WIDE_INT.
5486 Diagnose unbounded alloca calls only for limits of less than
5488 (alloca_call_type): Adjust. Diagnose possibly out-of-bounds alloca
5489 calls and VLA size only for limits of less than PTRDIFF_MAX. Same
5491 (pass_walloca::execute): Adjust. Diagnose alloca calls in loops
5492 only for limits of less than PTRDIFF_MAX.
5493 * langhooks-def.h (lhd_handle_option): Change function argument
5495 * langhooks.c (lhd_handle_option): Same.
5496 * langhooks.h (handle_option): Same.
5497 * opt-functions.awk (switch_bit_fields): Handle Host_Wide_Int and
5499 (var_type, var_type_struct): Same.
5500 (var_set): Handle ByteSize flag.
5501 * optc-gen.awk: Add comments to output to ease debugging. Make
5502 use of HOST_WIDE_INT where appropriate.
5503 * opts-gen-save.awk: Use %lx to format unsigned long.
5504 * opth-gen.awk: Change function argument to HOST_WIDE_INT.
5505 * opts-common.c (integral_argument): Return HOST_WIDE_INT and add
5506 arguments. Parse bytes-size suffixes.
5507 (enum_arg_to_value): Change function argument to HOST_WIDE_INT.
5508 (enum_value_to_arg): Same.
5509 (decode_cmdline_option): Handle cl_host_wide_int. Adjust.
5510 (handle_option): Adjust.
5511 (generate_option): Change function argument to HOST_WIDE_INT.
5512 (cmdline_handle_error): Adjust.
5513 (read_cmdline_option): Change function argument to HOST_WIDE_INT.
5514 (set_option): Change function argument to HOST_WIDE_INT.
5515 (option_enabled): Handle cl_host_wide_int.
5516 (get_option_state): Handle CLVC_SIZE.
5517 (control_warning_option): Same.
5518 * opts.c (common_handle_option): Change function argument to
5519 HOST_WIDE_INT. Remove handling of OPT_Walloca_larger_than_ and
5520 OPT_Wvla_larger_than_.
5521 * opts.h (enum cl_var_type): Add an enumerator.
5522 * stor-layout.c (layout_decl): Print a more meaningful warning.
5523 * toplev.c (output_stack_usage): Adjust.
5525 2018-07-20 Qing Zhao <qing.zhao@oracle.com>
5527 * builtins.c (expand_builtin_memcmp): Delete the last parameter for
5528 call to inline_expand_builtin_string_cmp.
5529 (expand_builtin_strcmp): Likewise.
5530 (expand_builtin_strncmp): Likewise.
5531 (inline_string_cmp): Delete the last parameter, change char_type_node
5532 to unsigned_char_type_node for strcmp/strncmp, add conversions to the
5534 (inline_expand_builtin_string_cmp): Delete the last parameter, give up
5535 the inlining expansion on target where the type of the call has same or
5536 narrower precision than unsigned char.
5538 2018-07-20 David Malcolm <dmalcolm@redhat.com>
5540 * Makefile.in (OBJS): Add json.o and optinfo-emit-json.o.
5541 (CFLAGS-optinfo-emit-json.o): Define TARGET_NAME.
5542 * common.opt (fsave-optimization-record): New option.
5543 * coretypes.h (struct kv_pair): Move here from dumpfile.c.
5544 * doc/invoke.texi (-fsave-optimization-record): New option.
5545 * dumpfile.c: Include "optinfo-emit-json.h".
5546 (struct kv_pair): Move to coretypes.h.
5547 (optgroup_options): Make non-static.
5548 (dump_context::end_scope): Call
5549 optimization_records_maybe_pop_dump_scope.
5550 * dumpfile.h (optgroup_options): New decl.
5551 * json.cc: New file.
5553 * optinfo-emit-json.cc: New file.
5554 * optinfo-emit-json.h: New file.
5555 * optinfo.cc: Include "optinfo-emit-json.h".
5556 (optinfo::emit): Call optimization_records_maybe_record_optinfo.
5557 (optinfo_enabled_p): Check optimization_records_enabled_p.
5558 (optinfo_wants_inlining_info_p): Likewise.
5559 * optinfo.h: Update comment.
5560 * profile-count.c (profile_quality_as_string): New function.
5561 * profile-count.h (profile_quality_as_string): New decl.
5562 (profile_count::quality): New accessor.
5563 * selftest-run-tests.c (selftest::run_tests): Call json_cc_tests
5564 and optinfo_emit_json_cc_tests.
5565 * selftest.h (selftest::json_cc_tests): New decl.
5566 (selftest::optinfo_emit_json_cc_tests): New decl.
5567 * toplev.c: Include "optinfo-emit-json.h".
5568 (compile_file): Call optimization_records_finish.
5569 (do_compile): Call optimization_records_start.
5570 * tree-ssa-live.c: Include optinfo.h.
5571 (remove_unused_scope_block_p): Retain inlining information if
5572 optinfo_wants_inlining_info_p returns true.
5574 2018-07-20 Richard Biener <rguenther@suse.de>
5577 * dwarf2out.c (dwarf2out_die_ref_for_decl): Test in_lto_p
5578 to cover -flto-partition=none.
5580 2018-07-20 Martin Liska <mliska@suse.cz>
5582 * tree.h (DECL_LOCATION_RANGE): Remove unused macro.
5583 (get_decl_source_range): Remove unused function.
5585 2018-07-20 Richard Biener <rguenther@suse.de>
5587 * tree-ssa-sccvn.h (struct vn_nary_op_s): Add next member.
5588 (struct vn_phi_s): Likewise.
5589 (struct vn_reference_s): Likewise.
5590 * tree-ssa-sccvn.c (vn_nary_op_hasher::equal): Add shortcut
5591 for searching the slot of an entry known to be in the hash itself.
5592 (vn_phi_hasher::equal): Likewise.
5593 (vn_reference_hasher::equal): Likewise.
5594 (last_inserted_ref, last_inserted_phi, last_inserted_nary): New
5596 (optimistic_info, current_info): Remove, keeping only valid_info.
5597 (vn_reference_lookup_1): Remove fallback lookup.
5598 (vn_reference_lookup_2): Likewise.
5599 (vn_nary_op_lookup_1): Likewise.
5600 (vn_phi_lookup): Likewise.
5601 (vn_nary_build_or_lookup_1): Make sure to not chain the built
5603 (vn_reference_insert): Adjust, chain the inserted hash element
5604 at last_inserted_ref.
5605 (vn_reference_insert_pieces): Likewise.
5606 (visit_reference_op_call): Likewise.
5607 (vn_nary_op_insert_into): Chain the inserted hash element at
5609 (vn_nary_op_insert_pieces): Adjust.
5610 (vn_nary_op_insert): Likewise.
5611 (vn_nary_op_insert_stmt): Likewise.
5612 (vn_phi_insert): Adjust, chain the inserted hash element at
5614 (process_scc): Remove clearing and copying the optimistic
5615 table. Instead remove elements inserted during an optimistic
5616 iteration from the single table we maintain.
5617 (init_scc_vn): Adjust.
5618 (free_scc_vn): Likewise.
5619 (sccvn_dom_walker::record_cond): Likewise.
5620 (sccvn_dom_walker::after_dom_children): Likewise.
5622 2018-07-19 Martin Sebor <msebor@redhat.com>
5624 PR tree-optimization/84047
5625 PR tree-optimization/83776
5626 * tree-vrp.c (vrp_prop::check_mem_ref): New function.
5627 (check_array_bounds): Call it.
5629 2018-07-19 Martin Sebor <msebor@redhat.com>
5631 * align.h (align_flags): Use member initialization.
5633 2018-07-19 David Malcolm <dmalcolm@redhat.com>
5635 * Makefile.in (OBJS): Add optinfo.o.
5636 * coretypes.h (class symtab_node): New forward decl.
5637 (struct cgraph_node): New forward decl.
5638 (class varpool_node): New forward decl.
5639 * dump-context.h: New file.
5640 * dumpfile.c: Include "optinfo.h", "dump-context.h", "cgraph.h",
5642 (refresh_dumps_are_enabled): Use optinfo_enabled_p.
5643 (set_dump_file): Call dumpfile_ensure_any_optinfo_are_flushed.
5644 (set_alt_dump_file): Likewise.
5645 (dump_context::~dump_context): New dtor.
5646 (dump_gimple_stmt): Move implementation to...
5647 (dump_context::dump_gimple_stmt): ...this new member function.
5648 Add the stmt to any pending optinfo, creating one if need be.
5649 (dump_gimple_stmt_loc): Move implementation to...
5650 (dump_context::dump_gimple_stmt_loc): ...this new member function.
5651 Start a new optinfo and add the stmt to it.
5652 (dump_gimple_expr): Move implementation to...
5653 (dump_context::dump_gimple_expr): ...this new member function.
5654 Add the stmt to any pending optinfo, creating one if need be.
5655 (dump_gimple_expr_loc): Move implementation to...
5656 (dump_context::dump_gimple_expr_loc): ...this new member function.
5657 Start a new optinfo and add the stmt to it.
5658 (dump_generic_expr): Move implementation to...
5659 (dump_context::dump_generic_expr): ...this new member function.
5660 Add the tree to any pending optinfo, creating one if need be.
5661 (dump_generic_expr_loc): Move implementation to...
5662 (dump_context::dump_generic_expr_loc): ...this new member
5663 function. Add the tree to any pending optinfo, creating one if
5665 (dump_printf): Move implementation to...
5666 (dump_context::dump_printf_va): ...this new member function. Add
5667 the text to any pending optinfo, creating one if need be.
5668 (dump_printf_loc): Move implementation to...
5669 (dump_context::dump_printf_loc_va): ...this new member function.
5670 Start a new optinfo and add the stmt to it.
5671 (dump_dec): Move implementation to...
5672 (dump_context::dump_dec): ...this new member function. Add the
5673 value to any pending optinfo, creating one if need be.
5674 (dump_context::dump_symtab_node): New member function.
5675 (dump_context::get_scope_depth): New member function.
5676 (dump_context::begin_scope): New member function.
5677 (dump_context::end_scope): New member function.
5678 (dump_context::ensure_pending_optinfo): New member function.
5679 (dump_context::begin_next_optinfo): New member function.
5680 (dump_context::end_any_optinfo): New member function.
5681 (dump_context::s_current): New global.
5682 (dump_context::s_default): New global.
5683 (dump_scope_depth): Delete global.
5684 (dumpfile_ensure_any_optinfo_are_flushed): New function.
5685 (dump_symtab_node): New function.
5686 (get_dump_scope_depth): Reimplement in terms of dump_context.
5687 (dump_begin_scope): Likewise.
5688 (dump_end_scope): Likewise.
5689 (selftest::temp_dump_context::temp_dump_context): New ctor.
5690 (selftest::temp_dump_context::~temp_dump_context): New dtor.
5691 (selftest::verify_item): New function.
5692 (ASSERT_IS_TEXT): New macro.
5693 (ASSERT_IS_TREE): New macro.
5694 (ASSERT_IS_GIMPLE): New macro.
5695 (selftest::test_capture_of_dump_calls): New test.
5696 (selftest::dumpfile_c_tests): Call it.
5697 * dumpfile.h (dump_printf, dump_printf_loc, dump_basic_block)
5698 (dump_generic_expr_loc, dump_generic_expr, dump_gimple_stmt_loc)
5699 (dump_gimple_stmt, dump_dec): Gather these related decls and add a
5700 descriptive comment.
5701 (dump_function, print_combine_total_stats, enable_rtl_dump_file)
5702 (dump_node, dump_bb): Move these unrelated decls.
5703 (class dump_manager): Add leading comment.
5704 * optinfo.cc: New file.
5705 * optinfo.h: New file.
5707 2018-07-19 Michael Collison <michael.collison@arm.com>
5708 Richard Henderson <rth@redhat.com>
5710 * config/aarch64/aarch64.md (subv<GPI>4, usubv<GPI>4): New patterns.
5711 (subti): Handle op1 zero.
5712 (subvti4, usub4ti4): New.
5713 (*sub<GPI>3_compare1_imm): New.
5714 (sub<GPI>3_carryinCV): New.
5715 (*sub<GPI>3_carryinCV_z1_z2, *sub<GPI>3_carryinCV_z1): New.
5716 (*sub<GPI>3_carryinCV_z2, *sub<GPI>3_carryinCV): New.
5718 2018-07-19 Michael Collison <michael.collison@arm.com>
5719 Richard Henderson <rth@redhat.com>
5721 * config/aarch64/aarch64.md: (addv<GPI>4, uaddv<GPI>4): New.
5722 (addti3): Create simpler code if low part is already known to be 0.
5723 (addvti4, uaddvti4): New.
5724 (*add<GPI>3_compareC_cconly_imm): New.
5725 (*add<GPI>3_compareC_cconly): New.
5726 (*add<GPI>3_compareC_imm): New.
5727 (*add<GPI>3_compareC): Rename from add<GPI>3_compare1; do not
5728 handle constants within this pattern..
5729 (*add<GPI>3_compareV_cconly_imm): New.
5730 (*add<GPI>3_compareV_cconly): New.
5731 (*add<GPI>3_compareV_imm): New.
5732 (add<GPI>3_compareV): New.
5733 (add<GPI>3_carryinC, add<GPI>3_carryinV): New.
5734 (*add<GPI>3_carryinC_zero, *add<GPI>3_carryinV_zero): New.
5735 (*add<GPI>3_carryinC, *add<GPI>3_carryinV): New.
5736 ((*add<GPI>3_compareC_cconly_imm): Replace 'ne' operator
5737 with 'comparison' operator.
5738 (*add<GPI>3_compareV_cconly_imm): Ditto.
5739 (*add<GPI>3_compareV_cconly): Ditto.
5740 (*add<GPI>3_compareV_imm): Ditto.
5741 (add<GPI>3_compareV): Ditto.
5742 (add<mode>3_carryinC): Ditto.
5743 (*add<mode>3_carryinC_zero): Ditto.
5744 (*add<mode>3_carryinC): Ditto.
5745 (add<mode>3_carryinV): Ditto.
5746 (*add<mode>3_carryinV_zero): Ditto.
5747 (*add<mode>3_carryinV): Ditto.
5749 2018-07-19 Michael Collison <michael.collison@arm.com>
5750 Richard Henderson <rth@redhat.com>
5752 * config/aarch64/aarch64-modes.def (CC_V): New.
5753 * config/aarch64/aarch64-protos.h
5754 (aarch64_addti_scratch_regs): Declare
5755 (aarch64_subvti_scratch_regs): Declare.
5756 (aarch64_expand_subvti): Declare.
5757 (aarch64_gen_unlikely_cbranch): Declare
5758 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Test
5759 for signed overflow using CC_Vmode.
5760 (aarch64_get_condition_code_1): Handle CC_Vmode.
5761 (aarch64_gen_unlikely_cbranch): New function.
5762 (aarch64_addti_scratch_regs): New function.
5763 (aarch64_subvti_scratch_regs): New function.
5764 (aarch64_expand_subvti): New function.
5766 2018-07-19 Andre Vieira <andre.simoesdiasvieira@arm.com>
5768 * config/aarch64/aarch64-option-extensions.def: New entry for profile
5770 * config/aarch64/aarch64.h (AARCH64_FL_PROFILE): New.
5771 * doc/invoke.texi (aarch64-feature-modifiers): New entry for profile
5774 2018-07-19 Andre Vieira <andre.simoesdiasvieira@arm.com>
5777 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Make
5778 address check not strict.
5780 2018-07-19 Andre Vieira <andre.simoesdiasvieira@arm.com>
5782 * config/aarch64/aarch64-simd.md (aarch64_simd_mov<VQ:mode>): Replace
5784 (store_pair_lanes<mode>): Likewise.
5785 * config/aarch64/aarch64-protos.h (aarch64_addr_query_type): Add new
5786 enum value 'ADDR_QUERY_LDP_STP_N'.
5787 * config/aarch64/aarch64.c (aarch64_addr_query_type): Likewise.
5788 (aarch64_print_address_internal): Add declaration.
5789 (aarch64_print_ldpstp_address): Remove.
5790 (aarch64_classify_address): Adapt mode for 'ADDR_QUERY_LDP_STP_N'.
5791 (aarch64_print_operand): Change printing of 'y'.
5792 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Use
5793 new enum value 'ADDR_QUERY_LDP_STP_N', don't hardcode mode and use
5794 'true' rather than '1'.
5795 * config/aarch64/constraints.md (Uml): Likewise.
5796 (Uml): Rename to Umn.
5799 2018-07-19 Richard Biener <rguenther@suse.de>
5801 * tree-ssa-sccvn.h (struct vn_phi_s): Make phiargs member
5803 * tree-ssa-sccvn.c: Remove alloc-pool.h use.
5804 (vn_phi_hasher): Derive from nofree_ptr_hash and remove remove method.
5805 (vn_reference_hasher): Likewise.
5806 (struct vn_tables_s): Remove obstack and alloc-pool members.
5807 (vn_tables_obstack, vn_tables_insert_obstack): New global obstacks.
5808 (vn_nary_build_or_lookup_1): Manually build in vn_tables_insert_obstack.
5809 (vn_reference_insert): Allocate from obstack instead of from alloc-pool.
5810 (vn_reference_insert_pieces): Likewise.
5811 (alloc_vn_nary_op_noinit): Adjust.
5812 (vn_nary_op_insert_stmt): Allocate phiargs in-place.
5813 (vn_phi_eq): Adjust.
5814 (shared_lookup_phiargs): Remove.
5815 (vn_phi_lookup): Allocate temporary vn_phi_s on the stack.
5816 (vn_phi_insert): Allocate from obstack instead of from alloc-pool.
5817 (visit_reference_op_call): Likewise.
5818 (copy_nary, copy_phi, copy_reference): Remove.
5819 (process_scc): Rewind the obstack when iterating. Do not
5820 copy the elements to valid_info but just move them from one
5821 hashtable to the other.
5822 (allocate_vn_table): Adjust.
5823 (free_vn_table): Likewise.
5824 (init_scc_vn): Likewise.
5825 (free_scc_vn): Likewise.
5827 2018-07-19 H.J. Lu <hongjiu.lu@intel.com>
5830 * config/i386/i386.c (rest_of_insert_endbranch): Lookup
5831 indirect_return as function type attribute.
5832 (ix86_attribute_table): Change indirect_return to function
5834 * doc/extend.texi: Update indirect_return attribute.
5836 2018-07-19 Aldy Hernandez <aldyh@redhat.com>
5838 * wide-int.h (widest2_int): New.
5839 * gimple-fold.c (arith_overflowed_p): Use it.
5840 * tree.h (widest2_int_cst): New.
5841 * tree-vrp.c (wide_int_binop_overflow): Rename from
5842 vrp_int_const_binop.
5843 Rewrite to work on trees.
5844 (extract_range_from_multiplicative_op_1): Abstract code to...
5845 (wide_int_range_min_max): ...here.
5846 (wide_int_range_cross_product): ...and here.
5847 (extract_range_from_binary_expr_1): Abstract overflow code to...
5848 (wide_int_range_mult_wrapping): ...here.
5849 * tree-vrp.h (wide_int_range_cross_product): New.
5850 (wide_int_range_mult_wrapping): New.
5852 2018-07-19 Andrew Senkevich <andrew.senkevich@intel.com>
5853 Julia Koval <julia.koval@intel.com>
5855 * config/i386/x86-tune-costs.h (skylake_memcpy,
5856 skylake_memset): Replace rep_prefix with unrolling for size 512.
5858 2018-07-18 Kugan Vivekanandarajah <kuganv@linaro.org>
5861 * tree-ssa-phiopt.c (cond_removal_in_popcount_pattern): Handle
5862 comparision with EQ_EXPR in last stmt.
5864 2018-07-18 Kelvin Nilsen <kelvin@gcc.gnu.org>
5866 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Rename
5867 this subsection to "PowerPC AltiVec/VSX Built-in Functions".
5868 (PowerPC AltiVec/VSX Built-in Functions): New name for subsection
5869 previously known as "PowerPC AltiVec Built-in Functions". Move
5870 some material to new subsubsections "PowerPC AltiVec Built-in
5871 Functions on ISA 2.06" and "PowerPC AltiVec Built-in Functions on
5873 (PowerPC Altivec Built-in Functions on ISA 2.05): New subsubsection.
5874 (PowerPC Altivec Built-in Functions on ISA 2.06): Likewise.
5875 (PowerPC Altivec Built-in Functions on ISA 2.07): Likewise.
5876 (PowerPC Altivec Built-in Functions on ISA 3.0): Likewise.
5878 2018-07-18 Richard Biener <rguenther@suse.de>
5880 PR tree-optimization/86557
5881 * tree-vect-patterns.c (vect_recog_divmod_pattern): Also handle
5884 2018-07-18 Ilya Leoshkevich <iii@linux.ibm.com>
5886 * config/s390/s390.c (s390_function_profiler): Generate CFI.
5888 2018-07-17 Jeff Law <law@redhat.com>
5890 * config/arm/arm.c (get_label_padding): Update for recent
5891 changes to label_to_alignment.
5893 PR tree-optimization/86010
5894 * tree-ssa-dse.c (compute_trims): Fix typo/thinko.
5896 * config/mips/mips.c (vr4130_align_insns): Update for recent
5897 changes to label_to_alignment.
5899 * config/frv/frv.c (frv_label_align): Update for recent changes
5900 to label_to_alignment.
5902 * config/nios2/nios2.c (nios2_label_align): Update for recent
5903 changes which dropped ALIGN_LABELS_LOG.
5905 2018-07-17 Andreas Schwab <schwab@linux-m68k.org>
5907 * config/m68k/m68k.md (umulsi3_highpart+1, const_umulsi3_highpart)
5908 (smulsi3_highpart+1, const_smulsi3_highpart): Add CC_STATUS_INIT.
5910 2018-07-17 Claudiu Zissulescu <claziss@synopsys.com>
5912 * config/arc/arc.c (arc_label_align): Use align_labels instead of
5913 deprecated align_labels_log.
5915 2018-07-17 Richard Biener <rguenther@suse.de>
5918 * dwarf2out.c (init_sections_and_labels): Always generate
5919 a debug_line_str_section for early LTO debug.
5920 (dwarf2out_finish): Reset debug_line_str_hash output early.
5921 Bump counter for extra dwarf5 .debug_loc labels to not conflict
5923 (dwarf2out_early_finish): Output debug_line_str.
5925 2018-07-17 Robin Dapp <rdapp@linux.ibm.com>
5927 * config/s390/s390.c (preferred_la_operand_p): Do not use LA with
5928 index register on z196 or later.
5930 2018-07-17 Robin Dapp <rdapp@linux.ibm.com>
5932 * config/s390/s390.c (s390_default_align): Set default function
5934 (s390_override_options_after_change): Call s390_default align.
5935 (s390_option_override_internal): Call s390_default_align.
5936 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define.
5938 2018-07-17 Jakub Jelinek <jakub@redhat.com>
5941 * omp-low.c (create_task_copyfn): Copy over also fields corresponding
5942 to _looptemp_ clauses, other than the first two.
5944 2018-07-17 Martin Liska <mliska@suse.cz>
5946 * opts.c: Do not enable OPT_falign_* for -Os.
5948 2018-07-17 Martin Liska <mliska@suse.cz>
5950 * align.h (MAX_CODE_ALIGN): New.
5951 (MAX_CODE_ALIGN_VALUE): New.
5952 * common/config/i386/i386-common.c (ix86_handle_option):
5953 (MAX_CODE_ALIGN): Moved to align.h.
5954 * final.c (MAX_CODE_ALIGN): Likewise.
5955 * opts.c (parse_and_check_align_values):
5956 (MAX_CODE_ALIGN): Likewise.
5957 (MAX_CODE_ALIGN_VALUE): Likewise.
5959 2018-07-17 Martin Liska <mliska@suse.cz>
5961 * config/i386/att.h (ASM_OUTPUT_ALIGN): Fix spacing
5962 in order to fulfil coding style.
5963 * config/i386/cygming.h (ASM_OUTPUT_ALIGN): Likewise.
5964 * config/i386/gas.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
5965 * config/i386/x86-64.h (ASM_OUTPUT_MAX_SKIP_PAD): Likewise.
5966 * config/iq2000/iq2000.h (ASM_OUTPUT_ALIGN): Likewise.
5967 * config/pa/pa.h (ASM_OUTPUT_ALIGN): Likewise.
5968 * config/sparc/sol2.h (ASM_OUTPUT_ALIGN_WITH_NOP): Likewise.
5969 * config/sparc/sparc.h (ASM_OUTPUT_ALIGN): Likewise.
5970 * config/visium/visium.h (ASM_OUTPUT_ALIGN): Likewise.
5971 (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
5973 2018-07-17 Martin Liska <mliska@suse.cz>
5975 * align.h: New file.
5976 * config/alpha/alpha.c (alpha_align_insns_1): Use align_functions
5978 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Use new return type
5979 align_flags of label_to_alignment.
5980 * config/m32r/m32r.h (LOOP_ALIGN): Wrap returned values into
5982 * config/m68k/m68k.c: Do not use removed align_labels_value and
5984 * config/nds32/nds32.h (JUMP_ALIGN): Wrap result into align_flags class.
5985 (LOOP_ALIGN): Likewise.
5986 (LABEL_ALIGN): Likewise.
5987 * config/powerpcspe/powerpcspe.c (TARGET_ASM_LOOP_ALIGN_MAX_SKIP):
5988 Remove not used macro.
5989 (rs6000_loop_align): Change return type to align_flags.
5990 (rs6000_loop_align_max_skip): Remove.
5991 * config/rs6000/rs6000-protos.h (rs6000_loop_align):
5992 Change return type to align_flags.
5993 * config/rs6000/rs6000.c (TARGET_ASM_LOOP_ALIGN_MAX_SKIP):
5994 Remove not used macro.
5995 (rs6000_loop_align): Change return type to align_flags.
5996 (rs6000_loop_align_max_skip): Remove.
5997 * config/rx/rx.h (JUMP_ALIGN): Wrap integer values
5998 * config/rx/rx-protos.h (rx_align_for_label): Make it
6000 * config/rx/rx.c (rx_align_for_label): Change return type
6002 (rx_max_skip_for_label): Remove TARGET_ASM_*_ALIGN_MAX_SKIP
6004 into align_flags class.
6005 (LABEL_ALIGN): Likewise.
6006 (LOOP_ALIGN): Likewise.
6007 * config/s390/s390.c (s390_label_align): Use align_flags
6009 (s390_asm_output_function_label): Likewise.
6010 * config/sh/sh.c (sh_override_options_after_change):
6011 Use align_flags class directly without macros.
6012 (find_barrier): Likewise.
6013 (barrier_align): Likewise.
6014 (sh_loop_align): Likewise.
6015 * config/spu/spu.c (spu_option_override):
6016 Use align_flags_tuple::get_value instead of removed macros.
6017 (spu_sched_init): Likewise.
6018 * config/spu/spu.h (GTY): Likewise.
6019 * config/visium/visium.c (visium_option_override):
6020 Set "8" as default secondary alignment.
6021 * config/visium/visium.h (SUBALIGN_LOG): Define to 3
6022 in order to guarantee secondary alignment of 8.
6023 * coretypes.h: Include align.h header file.
6024 * doc/tm.texi: Remove TARGET_ASM_JUMP_ALIGN_MAX_SKIP,
6025 TARGET_ASM_LOOP_ALIGN_MAX_SKIP, TARGET_ASM_LABEL_ALIGN_MAX_SKIP
6026 and TARGET_ASM_LABEL_ALIGN_AFTER_BARRIER_MAX_SKIP macros.
6027 * doc/tm.texi.in: Likewise.
6028 * final.c (struct label_alignment): Remove not used structure.
6029 (LABEL_ALIGN): Change type to align_flags.
6030 (LOOP_ALIGN): Likewise.
6031 (JUMP_ALIGN): Likewise.
6032 (default_loop_align_max_skip): Remove.
6033 (default_label_align_max_skip): Likewise.
6034 (default_jump_align_max_skip): Likewise.
6035 (default_label_align_after_barrier_max_skip):
6036 (LABEL_TO_ALIGNMENT): Change to access label_align vector.
6037 (LABEL_TO_MAX_SKIP): Remove.
6038 (label_to_alignment): Return align_flags type instead of integer.
6039 (label_to_max_skip): Remove.
6040 (align_fuzz): Use align_flags type.
6041 (compute_alignments): Use align_flags type and use align_flags::max
6042 to combine multiple alignments.
6043 (grow_label_align): Grow vec instead of C array.
6044 (update_alignments): Assign just LABEL_TO_ALIGNMENT.
6045 (shorten_branches): Use align_flags type and use align_flags::max
6046 to combine multiple alignments.
6047 (final_scan_insn_1): Remove usage of secondary alignment that comes
6048 from label alignment, but instead use proper secondary alignment
6049 which is computed in grow_label_align.
6050 * flags.h (struct align_flags_tuple): Move to align.h.
6051 (struct align_flags): Likewise.
6052 (state_align_loops): Rename to align_loops.
6053 (state_align_jumps): Rename to align_jumps.
6054 (state_align_labels): Rename to align_labels.
6055 (state_align_functions): Rename to align_functions.
6056 (align_loops_log): Remove.
6057 (align_jumps_log): Remove.
6058 (align_labels_log): Remove.
6059 (align_functions_log): Remove.
6060 (align_loops_max_skip): Remove.
6061 (align_jumps_max_skip): Remove.
6062 (align_labels_max_skip): Remove.
6063 (align_functions_max_skip): Remove.
6064 (align_loops_value): Remove.
6065 (align_jumps_value): Remove.
6066 (align_labels_value): Remove.
6067 (align_functions_value): Remove.
6068 * output.h (label_to_alignment): Change return type to align_flags.
6069 (label_to_max_skip): Remove.
6070 * target.def: Remove loop_align_max_skip, label_align_max_skip,
6071 jump_align_max_skip macros.
6072 * targhooks.h (default_loop_align_max_skip): Remove.
6073 (default_label_align_max_skip): Likewise.
6074 (default_jump_align_max_skip): Likewise.
6075 (default_label_align_after_barrier_max_skip): Remove.
6076 * toplev.c (read_log_maxskip): Use ::normalize function.
6077 (parse_N_M): Remove not used argument and also call ::normalize.
6078 (parse_alignment_opts): Do not pass unused arguments.
6079 * varasm.c (assemble_start_function): Use directly align_functions
6080 instead of removed macros.
6081 * system.h: Do not poison removed macros.
6083 2018-07-17 Jakub Jelinek <jakub@redhat.com>
6086 * gimplify.c (gimplify_omp_for): Ensure taskloop firstprivatized init
6087 and cond temporaries don't have reference type if iterator has
6088 pointer type. For init use &for_pre_body instead of pre_p if
6089 for_pre_body is non-empty.
6091 2018-07-16 Segher Boessenkool <segher@kernel.crashing.org>
6093 * config/rs6000/rs6000.md (trunc<mode>sf2): Expand truncates of
6094 double-double modes to SFmode directly directly.
6095 (trunc<mode>sf2_fprs): Delete.
6097 2018-07-16 Segher Boessenkool <segher@kernel.crashing.org>
6099 * config/rs6000/rs6000.c (init_float128_ibm): Use the correct names
6100 for conversions between IFmode and the decimal floating point modes.
6101 (init_float128_ieee): Use the correct names for conversions between
6102 KFmode and the decimal floating point modes.
6104 2018-07-16 Segher Boessenkool <segher@kernel.crashing.org>
6106 * config/rs6000/rs6000.c (init_float128_ibm): Use more correct names
6107 for the conversions between TDmode and IFmode.
6108 (init_float128_ieee): Use more correct names for the conversions
6109 between TDmode and KFmode.
6111 2018-07-16 Jakub Jelinek <jakub@redhat.com>
6113 PR tree-optimization/86526
6114 * builtins.c (expand_builtin_memcmp): Formatting fixes.
6115 (inline_expand_builtin_string_cmp): Likewise.
6116 (inline_string_cmp): Likewise. Use c_readstr instead of
6117 builtin_memcpy_read_str. Add unit_mode temporary.
6119 2018-07-16 Bernd Edlinger <bernd.edlinger@hotmail.de>
6122 * builtins.c (check_access): Bail out if range[0] is no INTEGER_CST.
6123 * expr.c (string_constant): Fix the element size of ARRAY_TYPE.
6125 2018-07-16 Kelvin Nilsen <kelvin@gcc.gnu.org>
6127 * doc/extend.texi (PowerPC AltiVec Built-in Functions):
6128 Alphabetize prototypes of built-in functions, separating out
6129 built-in functions that are listed in this section but should be
6130 described elsewhere.
6132 2018-07-16 Uros Bizjak <ubizjak@gmail.com>
6135 * expmed.c (emit_store_flag): Do not emit setcc followed by a
6136 conditional move when trapping comparison was split to a
6137 non-trapping one (and vice versa).
6139 2018-07-16 Ilya Leoshkevich <iii@linux.ibm.com>
6141 * config/s390/s390.c (s390_function_profiler): Generate nops
6142 instead of profiler call sequences.
6143 * config/s390/s390.opt: Add the new option.
6145 2018-07-16 Ilya Leoshkevich <iii@linux.ibm.com>
6147 * config/s390/s390.c (s390_function_profiler): Generate
6148 __mcount_loc section.
6149 * config/s390/s390.opt: Add the new option.
6151 2018-07-16 Ilya Leoshkevich <iii@linux.ibm.com>
6153 * common.opt: Add the new warning.
6154 * config/s390/s390.c (s390_function_profiler): Emit "brasl
6155 %r0,__fentry__" when -mfentry is specified.
6156 (s390_option_override_internal): Disallow -mfentry for 31-bit
6158 * config/s390/s390.opt: Add the new option.
6160 2018-07-16 Richard Biener <rguenther@suse.de>
6163 * dwarf2out.c (dwarf2out_register_external_die): Assign DIE parents
6164 for function-local FUNCTION_DECL and RESULT_DECL immediately.
6166 2018-07-16 Martin Liska <mliska@suse.cz>
6169 * ipa-pure-const.c (malloc_candidate_p): Revert ::get
6172 2017-07-16 Claudiu Zissulescu <claziss@synopsys.com>
6174 * config/arc/arcHS.md: Update ARCHS scheduling rules.
6176 2017-07-16 Claudiu Zissulescu <claziss@synopsys.com>
6178 * config/arc/arc-arch.h (arc_tune_attr): Add new tune parameters
6180 * config/arc/arc-cpus.def (hs4x): New cpu.
6182 * config/arc/arc-tables.opt: Regenerate.
6183 * config/arc/arc.c (arc_sched_issue_rate): New function.
6184 (TARGET_SCHED_ISSUE_RATE): Define.
6185 (TARGET_SCHED_EXPOSED_PIPELINE): Likewise.
6186 * config/arc/arc.md (attr type): Add fpu_fuse, fpu_sdiv, fpu_ddiv,
6188 (attr tune): Add ARCHS4x tune values.
6189 (attr tune_dspmpy): Define.
6190 (*tst): Correct instruction type.
6191 * config/arc/arcHS.md: Don't use this automaton for ARCHS4x cpus.
6192 * config/arc/arcHS4x.md: New file.
6193 * config/arc/fpu.md: Update instruction type attributes.
6194 * config/arc/t-multilib: Regenerate.
6196 2018-07-16 Tom de Vries <tdevries@suse.de>
6199 * var-tracking.c (vt_initialize): Fix pre_dec handling.
6201 2018-07-16 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
6203 * config/aarch64/atomics.md (aarch64_store_execlusive<mode>): Add
6206 2018-07-16 Eric Botcazou <ebotcazou@adacore.com>
6208 PR tree-optimization/86514
6209 * tree-ssa-reassoc.c (init_range_entry) <CASE_CONVERT>: Return for a
6210 conversion to a boolean type from a type with greater precision.
6212 2018-07-16 Tom de Vries <tdevries@suse.de>
6214 * var-tracking.c (vt_initialize): Print adjusted insn slim if
6215 dump_flags request TDF_SLIM.
6217 2018-07-16 Aldy Hernandez <aldyh@redhat.com>
6219 * fold-const.c (int_const_binop_1): Abstract...
6220 (wide_int_binop): ...wide int code here.
6221 (poly_int_binop): ...poly int code here.
6222 Abstract the rest of int_const_binop_1 into int_const_binop.
6223 * fold-const.h (wide_int_binop): New.
6224 * tree-vrp.c (vrp_int_const_binop): Call wide_int_binop.
6225 Remove useless PLUS/MINUS_EXPR case.
6226 (zero_nonzero_bits_from_vr): Move wide int code...
6227 (zero_nonzero_bits_from_bounds): ...here.
6228 (extract_range_from_binary_expr_1): Move mask optimization code...
6229 (range_easy_mask_min_max): ...here.
6230 * tree-vrp.h (zero_nonzero_bits_from_bounds): New.
6231 (range_easy_mask_min_max): New.
6233 2018-07-15 Jeff Law <law@redhat.com>
6236 * config/sh/sh.c (output_mi_thunk): Remove dead conditional
6239 2018-07-14 Jim Wilson <jimw@sifive.com>
6241 * config/riscv/linux.h (TARGET_ASM_FILE_END): New.
6243 2018-07-14 Paul Koning <ni1d@arrl.net>
6245 * config/pdp11/pdp11.c (pdp11_rtx_costs): Bugfixes.
6247 2018-07-13 Jan Hubicka <hubicka@ucw.cz>
6249 * lto-streamer-out.c (copy_function_or_variable): Dump info about
6252 2018-07-13 Bill Schmidt <wschmidt@linux.ibm.com>
6253 Steve Munroe <munroesj52@gmail.com>
6255 * config/rs6000/emmintrin.h (_mm_and_si128): New function.
6256 (_mm_andnot_si128): Likewise.
6257 (_mm_or_si128): Likewise.
6258 (_mm_xor_si128): Likewise.
6260 2018-07-13 Qing Zhao <qing.zhao@oracle.com>
6263 * builtins.c (expand_builtin_memcmp): Inline the calls first
6264 when result_eq is false.
6265 (expand_builtin_strcmp): Inline the calls first.
6266 (expand_builtin_strncmp): Likewise.
6267 (inline_string_cmp): New routine. Expand a string compare
6268 call by using a sequence of char comparison.
6269 (inline_expand_builtin_string_cmp): New routine. Inline expansion
6270 a call to str(n)cmp/memcmp.
6271 * doc/invoke.texi (--param builtin-string-cmp-inline-length):
6273 * params.def (BUILTIN_STRING_CMP_INLINE_LENGTH): New.
6275 2018-07-13 Richard Earnshaw <rearnsha@arm.com>
6277 * config/arm/driver-arm.c: Include arm-native.h.
6278 (host_detect_local_cpu): Use auto-generated data tables.
6279 (vendors, arm_cpu_table): Delete. Move part information to ...
6280 * config/arm/arm-cpus.in: ... here.
6281 * config/arm/parsecpu.awk (gen_native): New function.
6282 (vendor, part): New CPU fields.
6283 (END): Add support for building the native CPU detection tables.
6284 * config/arm/t-arm (arm-native.h): Add build rule.
6285 (driver-arm.o): Add dependency on arm-native.h.
6287 2018-07-13 Richard Biener <rguenther@suse.de>
6290 * match.pd (addr1 - addr2): Allow either of the operand to
6293 2018-07-13 Tom de Vries <tdevries@suse.de>
6295 * tree-inline.c (remap_ssa_name): Save and reuse debug exprs generated
6298 2018-07-13 Jackson Woodruff <jackson.woodruff@arm.com>
6300 * config/aarch64/aarch64.c (aarch64_operands_adjust_ok_for_ldpstp): Use
6301 arrays instead of numbered variables.
6303 2018-07-13 Eric Botcazou <ebotcazou@adacore.com>
6305 * config/sparc/sparc-protos.h (sparc_compute_frame_size): Delete.
6306 * config/sparc/sparc.c (sparc_compute_frame_size): Make static.
6308 2018-07-13 Richard Biener <rguenther@suse.de>
6311 * dwarf2out.c (gen_type_die_with_usage): Use scope_die_for
6312 instead of get_context_die.
6314 2018-07-13 Kugan Vivekanandarajah <kuganv@linaro.org>
6315 Richard Biener <rguenther@suse.de>
6318 * tree-ssa-loop-niter.c (number_of_iterations_popcount): Check
6319 that the loop latch destination where phi is defined.
6321 2018-07-12 Kito Cheng <kito.cheng@gmail.com>
6323 * config/riscv/riscv.c (enum riscv_privilege_levels): Add UNKNOWN_MODE.
6324 (riscv_expand_epilogue): Add assertion to check interrupt mode.
6325 (riscv_set_current_function): Extract getting interrupt type to new
6327 (riscv_get_interrupt_type): New function.
6328 (riscv_merge_decl_attributes): New function, checking interrupt type is
6330 (TARGET_MERGE_DECL_ATTRIBUTES): Define.
6332 2018-07-12 Paul Koning <ni1d@arrl.net>
6334 * config/pdp11/pdp11.c (pdp11_output_def): Fix typo in .set
6337 2018-07-12 Paul Koning <ni1d@arrl.net>
6339 * doc/rtl.texi (REG_NONNEG): Remove decrement and branch until
6340 zero reference, add doloop_end instead.
6341 * doc/md.texi (decrement_and_branch_until_zero): Remove.
6342 (Looping patterns): Remove decrement_and_branch_until_zero. Add
6343 detail for doloop_end.
6345 2018-07-12 Martin Sebor <msebor@redhat.com>
6348 * attribs.c (decl_attributes): Reject conflicting attributes before
6349 calling attribute handlers.
6351 2018-07-12 Jan Hubicka <hubicka@ucw.cz>
6353 * dumpfile.c (gcc::dump_manager::get_dump_file_name): Add PART
6355 (gcc::dump_manager::get_dump_file_name): likewise.
6356 (dump_begin): Likewise.
6357 * dumpfile.h (dump_begin): Update prototype.
6358 (gcc::dump_manager::get_dump_file_name,
6359 gcc::dump_manager::get_dump_file_name): Update prototype.
6361 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
6363 * internal-fn.h (vectorizable_internal_fn_p): New function.
6364 * tree-vect-slp.c (compatible_calls_p): Likewise.
6365 (vect_build_slp_tree_1): Remove nops argument. Handle calls
6366 to internal functions.
6367 (vect_build_slp_tree_2): Update call to vect_build_slp_tree_1.
6369 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
6371 * fold-const.h (inverse_conditions_p): Declare.
6372 * fold-const.c (inverse_conditions_p): New function.
6373 * match.pd: Use inverse_conditions_p. Add folds of view_converts
6374 that test the inverse condition of a conditional internal function.
6375 * internal-fn.h (vectorized_internal_fn_supported_p): Declare.
6376 * internal-fn.c (internal_fn_mask_index): Handle conditional
6378 (vectorized_internal_fn_supported_p): New function.
6379 * tree-if-conv.c: Include internal-fn.h and fold-const.h.
6380 (any_pred_load_store): Replace with...
6381 (need_to_predicate): ...this new variable.
6382 (redundant_ssa_names): New variable.
6383 (ifcvt_can_use_mask_load_store): Move initial checks to...
6384 (ifcvt_can_predicate): ...this new function. Handle tree codes
6385 for which a conditional internal function exists.
6386 (if_convertible_gimple_assign_stmt_p): Use ifcvt_can_predicate
6387 instead of ifcvt_can_use_mask_load_store. Update after variable
6389 (predicate_load_or_store): New function, split out from
6390 predicate_mem_writes.
6391 (check_redundant_cond_expr): New function.
6392 (value_available_p): Likewise.
6393 (predicate_rhs_code): Likewise.
6394 (predicate_mem_writes): Rename to...
6395 (predicate_statements): ...this. Use predicate_load_or_store
6396 and predicate_rhs_code.
6397 (combine_blocks, tree_if_conversion): Update after above name changes.
6398 (ifcvt_local_dce): Handle redundant_ssa_names.
6399 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern): Handle
6400 general conditional functions.
6401 * tree-vect-stmts.c (vectorizable_call): Likewise.
6403 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
6404 Alan Hayward <alan.hayward@arm.com>
6405 David Sherwood <david.sherwood@arm.com>
6407 * internal-fn.h (can_interpret_as_conditional_op_p): Declare.
6408 * internal-fn.c (can_interpret_as_conditional_op_p): New function.
6409 * tree-ssa-math-opts.c (convert_mult_to_fma_1): Handle conditional
6410 plus and minus and convert them into IFN_COND_FMA-based sequences.
6411 (convert_mult_to_fma): Handle conditional plus and minus.
6413 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
6415 * doc/md.texi (cond_fma, cond_fms, cond_fnma, cond_fnms): Document.
6416 * optabs.def (cond_fma_optab, cond_fms_optab, cond_fnma_optab)
6417 (cond_fnms_optab): New optabs.
6418 * internal-fn.def (COND_FMA, COND_FMS, COND_FNMA, COND_FNMS): New
6420 (FMA): Use DEF_INTERNAL_FLT_FN rather than DEF_INTERNAL_FLT_FLOATN_FN.
6421 * internal-fn.h (get_conditional_internal_fn): Declare.
6422 (get_unconditional_internal_fn): Likewise.
6423 * internal-fn.c (cond_ternary_direct): New macro.
6424 (expand_cond_ternary_optab_fn): Likewise.
6425 (direct_cond_ternary_optab_supported_p): Likewise.
6426 (FOR_EACH_COND_FN_PAIR): Likewise.
6427 (get_conditional_internal_fn): New function.
6428 (get_unconditional_internal_fn): Likewise.
6429 * gimple-match.h (gimple_match_op::MAX_NUM_OPS): Bump to 5.
6430 (gimple_match_op::gimple_match_op): Add a new overload for 5
6432 (gimple_match_op::set_op): Likewise.
6433 (gimple_resimplify5): Declare.
6434 * genmatch.c (decision_tree::gen): Generate simplifications for
6436 * gimple-match-head.c (gimple_simplify): Define an overload for
6437 5 operands. Handle calls with 5 arguments in the top-level overload.
6438 (convert_conditional_op): Handle conversions from unconditional
6439 internal functions to conditional ones.
6440 (gimple_resimplify5): New function.
6441 (build_call_internal): Pass a fifth operand.
6442 (maybe_push_res_to_seq): Likewise.
6443 (try_conditional_simplification): Try converting conditional
6444 internal functions to unconditional internal functions.
6445 Handle 3-operand unconditional forms.
6446 * match.pd (UNCOND_TERNARY, COND_TERNARY): Operator lists.
6447 Define ternary equivalents of the current rules for binary conditional
6449 * config/aarch64/aarch64.c (aarch64_preferred_else_value): Handle
6451 * config/aarch64/iterators.md (UNSPEC_COND_FMLA, UNSPEC_COND_FMLS)
6452 (UNSPEC_COND_FNMLA, UNSPEC_COND_FNMLS): New unspecs.
6453 (optab): Handle them.
6454 (SVE_COND_FP_TERNARY): New int iterator.
6455 (sve_fmla_op, sve_fmad_op): New int attributes.
6456 * config/aarch64/aarch64-sve.md (cond_<optab><mode>)
6457 (*cond_<optab><mode>_2, *cond_<optab><mode_4)
6458 (*cond_<optab><mode>_any): New SVE_COND_FP_TERNARY patterns.
6460 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
6462 * target.def (preferred_else_value): New target hook.
6463 * doc/tm.texi.in (TARGET_PREFERRED_ELSE_VALUE): New hook.
6464 * doc/tm.texi: Regenerate.
6465 * targhooks.h (default_preferred_else_value): Declare.
6466 * targhooks.c (default_preferred_else_value): New function.
6467 * internal-fn.h (conditional_internal_fn_code): Declare.
6468 * internal-fn.c (FOR_EACH_CODE_MAPPING): New macro.
6469 (get_conditional_internal_fn): Use it.
6470 (conditional_internal_fn_code): New function.
6471 * gimple-match.h (gimple_match_cond): New struct.
6472 (gimple_match_op): Add a cond member function.
6473 (gimple_match_op::gimple_match_op): Update all forms to take a
6475 * genmatch.c (expr::gen_transform): Use the same condition as res_op
6476 for the suboperation, but don't specify a particular else_value.
6477 * tree-ssa-sccvn.c (vn_nary_simplify, vn_reference_lookup_3)
6478 (visit_nary_op, visit_reference_op_load): Pass
6479 gimple_match_cond::UNCOND to the gimple_match_op constructor.
6480 * gimple-match-head.c: Include tree-eh.h
6481 (convert_conditional_op): New function.
6482 (maybe_resimplify_conditional_op): Likewise.
6483 (gimple_resimplify1): Call maybe_resimplify_conditional_op.
6484 (gimple_resimplify2): Likewise.
6485 (gimple_resimplify3): Likewise.
6486 (gimple_resimplify4): Likewise.
6487 (maybe_push_res_to_seq): Return null for conditional operations.
6488 (try_conditional_simplification): New function.
6489 (gimple_simplify): Call it. Pass conditions to the gimple_match_op
6491 * match.pd: Fold VEC_COND_EXPRs of an IFN_COND_* call to a new
6493 * config/aarch64/aarch64.c (aarch64_preferred_else_value): New
6495 (TARGET_PREFERRED_ELSE_VALUE): Redefine.
6497 2018-07-12 Jan Hubicka <hubicka@ucw.cz>
6499 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not stream
6501 (hash_tree): Do not hash DECL_FCONTEXT
6502 * tree-streamer-in.c (lto_input_ts_field_decl_tree_pointers):
6503 Do not stream DECL_FCONTEXT.
6504 * tree-streamer-out.c (write_ts_field_decl_tree_pointers): Likewise.
6505 * tree.c (free_lang_data_in_decl): Free DECL_FCONTEXT.
6507 2018-07-12 Richard Biener <rguenther@suse.de>
6510 * dwarf2out.c (gen_block_die): Only output blocks when they have
6511 at least one !DECL_IGNORED_P variable.
6513 2018-07-12 Richard Biener <rguenther@suse.de>
6516 * config/gnu-user.h (GNU_USER_TARGET_NO_PTHREADS_LIB_SPEC):
6517 Remove -mieee-fp handling.
6519 2018-07-12 Richard Biener <rguenther@suse.de>
6521 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Remove bogus
6522 left-over from last patch.
6524 2018-07-12 Jakub Jelinek <jakub@redhat.com>
6526 PR tree-optimization/86492
6527 * gimple-ssa-store-merging.c
6528 (imm_store_chain_info::coalesce_immediate_stores): Call
6529 check_no_overlap even for the merge_overlapping case. Formatting fix.
6531 2018-07-12 Richard Biener <rguenther@suse.de>
6534 * fold-const.c (fold_binary_op_with_conditional_arg): Do not
6535 move possibly trapping operations into the conditional.
6537 2018-07-12 Richard Biener <rguenther@suse.de>
6539 * tree-ssa-sccvn.c (mprts_hook_cnt): Remove.
6540 (vn_lookup_simplify_result): Remove recursion limit applied
6542 (vn_nary_build_or_lookup_1): Adjust.
6543 (try_to_simplify): Likewise.
6544 * gimple-match-head.c (gimple_resimplify1): Instead apply one
6546 (gimple_resimplify2): Likewise.
6547 (gimple_resimplify3): Likewise.
6548 (gimple_resimplify4): Likewise.
6550 2018-07-11 Jakub Jelinek <jakub@redhat.com>
6552 * config/i386/avx512bitalgintrin.h (_mm512_mask_bitshuffle_epi64_mask):
6553 Use __mmask64 type instead of __mmask8 for __M argument.
6554 * config/i386/avx512fintrin.h (_mm512_mask_xor_epi64,
6555 _mm512_maskz_xor_epi64): Use __mmask8 type instead of __mmask16 for
6557 (_mm512_mask_cmpneq_epi64_mask): Use __mmask8 type instead of
6558 __mmask16 for __M argument.
6559 (_mm512_maskz_insertf32x4, _mm512_maskz_inserti32x4,
6560 _mm512_mask_insertf32x4, _mm512_mask_inserti32x4): Cast last argument
6561 to __mmask16 instead of __mmask8.
6562 * config/i386/avx512vlintrin.h (_mm_mask_add_ps, _mm_maskz_add_ps,
6563 _mm256_mask_add_ps, _mm256_maskz_add_ps, _mm_mask_sub_ps,
6564 _mm_maskz_sub_ps, _mm256_mask_sub_ps, _mm256_maskz_sub_ps,
6565 _mm256_maskz_cvtepi32_ps, _mm_maskz_cvtepi32_ps): Use __mmask8 type
6566 instead of __mmask16 for __U argument.
6567 * config/i386/avx512vlbwintrin.h (_mm_mask_cmp_epi8_mask): Use
6568 __mmask16 instead of __mmask8 for __U argument.
6569 (_mm256_mask_cmp_epi8_mask): Use __mmask32 instead of __mmask16 for
6571 (_mm256_cmp_epi8_mask): Use __mmask32 return type instead of
6573 (_mm_mask_cmp_epu8_mask): Use __mmask16 instead of __mmask8 for __U
6575 (_mm256_mask_cmp_epu8_mask): Use __mmask32 instead of __mmask16 for
6577 (_mm256_cmp_epu8_mask): Use __mmask32 return type instead of
6579 (_mm_mask_cmp_epi16_mask): Cast last argument to __mmask8 instead
6581 (_mm256_mask_cvtepi8_epi16): Use __mmask16 instead of __mmask32 for
6583 (_mm_mask_cvtepi8_epi16): Use __mmask8 instead of __mmask32 for
6585 (_mm256_mask_cvtepu8_epi16): Use __mmask16 instead of __mmask32 for
6587 (_mm_mask_cvtepu8_epi16): Use __mmask8 instead of __mmask32 for
6589 (_mm256_mask_cmpneq_epu8_mask, _mm256_mask_cmplt_epu8_mask,
6590 _mm256_mask_cmpge_epu8_mask, _mm256_mask_cmple_epu8_mask): Change
6591 return type as well as __M argument type and all casts from __mmask8
6593 (_mm256_mask_cmpneq_epu16_mask, _mm256_mask_cmplt_epu16_mask,
6594 _mm256_mask_cmpge_epu16_mask, _mm256_mask_cmple_epu16_mask): Change
6595 return type as well as __M argument type and all casts from __mmask8
6597 (_mm256_mask_cmpneq_epi8_mask, _mm256_mask_cmplt_epi8_mask,
6598 _mm256_mask_cmpge_epi8_mask, _mm256_mask_cmple_epi8_mask): Change
6599 return type as well as __M argument type and all casts from __mmask8
6601 (_mm256_mask_cmpneq_epi16_mask, _mm256_mask_cmplt_epi16_mask,
6602 _mm256_mask_cmpge_epi16_mask, _mm256_mask_cmple_epi16_mask): Change
6603 return type as well as __M argument type and all casts from __mmask8
6605 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_shrdi_epi32,
6606 _mm_mask_shldi_epi32): Cast last argument to __mmask8 instead of
6609 2018-07-11 Grazvydas Ignotas <notasas@gmail.com>
6611 * config/i386/avx512bwintrin.h: (_mm512_mask_cmp_epi8_mask,
6612 _mm512_mask_cmp_epu8_mask): Use __mmask64 type instead of __mmask32
6615 2018-07-11 Paul Koning <ni1d@arrl.net>
6617 * doc/md.texi (define_subst): Document how multiple occurrences of
6618 the same argument in the replacement pattern are handled.
6620 2018-07-11 Paul Koning <ni1d@arrl.net>
6622 * doc/extend.texi (Common Variable Attributes): Move "mode" into
6624 (Common Type Attributes): Add "mode" attribute.
6626 2018-07-11 Jan Hubicka <hubicka@ucw.cz>
6628 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not
6629 stream DECL_ORIGINAL_TYPE.
6630 (DFS::DFS_write_tree_body): Drop hack handling local external decls.
6631 (hash_tree): Do not walk DECL_ORIGINAL_TYPE.
6632 * tree-streamer-in.c (lto_input_ts_decl_non_common_tree_pointers):
6633 Do not walk original type.
6634 * tree-streamer-out.c (streamer_write_chain): Drop hack handling
6636 (write_ts_decl_non_common_tree_pointers): Do not stream
6638 * tree.c (free_lang_data_in_decl): Clear DECL_ORIGINAL_TYPE.
6639 (find_decls_types_r): Do not walk DEC_ORIGINAL_TYPE.
6641 2018-07-11 Aldy Hernandez <aldyh@redhat.com>
6643 * tree-ssa-threadupdate.c (thread_through_all_blocks): Do not jump
6644 thread twice from the same starting edge.
6646 2018-07-11 Aldy Hernandez <aldyh@redhat.com>
6648 * vr-values.c (gimple_stmt_nonzero_p): Abstract common code to...
6649 * gimple.c (gimple_call_nonnull_result_p): ...here...
6650 (gimple_call_nonnull_arg): ...and here.
6651 * gimple.h (gimple_call_nonnull_result_p): New.
6652 (gimple_call_nonnull_arg): New.
6654 2018-07-11 Richard Earnshaw <rearnsha@arm.com>
6656 * config/arm/arm-cpus.in: Move information from fpu field of each
6657 cpu definition to the isa field.
6658 * config/arm/parsecpu.awk (fpu): Delete match rule.
6659 (gen_comm_data): Don't add bits from the CPU's FPU entry.
6661 2018-07-11 Richard Biener <rguenther@suse.de>
6664 * dwarf2out.c (init_sections_and_labels): Use
6665 output_asm_line_debug_info consistently.
6666 (dwarf2out_early_finish): Likewise.
6667 (dwarf2out_finish): Remove DW_AT_stmt_list from early generated
6670 2018-07-11 Richard Biener <rguenther@suse.de>
6672 * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely_1):
6673 Rework father_bb setting in a way to avoid propagating constants
6674 multiple times on a loop body.
6676 2018-07-10 Mark Wielaard <mark@klomp.org>
6679 * dwarf2out.c (output_macinfo_op): Fix dwarf_FORM typo in gcc_assert.
6681 2018-07-10 Richard Biener <rguenther@suse.de>
6683 * hash-map.h (hash_map::iterator::operator*): Return
6684 references to key and value.
6686 2018-07-10 Jakub Jelinek <jakub@redhat.com>
6689 * gimplify.c (find_combined_omp_for): Add DATA argument, in addition
6690 to finding the inner OMP_FOR/OMP_SIMD stmt find non-trivial wrappers,
6691 BLOCKs with BLOCK_VARs, OMP_PARALLEL in between, OMP_FOR in between.
6692 (gimplify_omp_for): For composite loops, move outer
6693 OMP_{DISTRIBUTE,TASKLOOP,FOR,PARALLEL} right around innermost
6694 OMP_FOR/OMP_SIMD if there are any non-trivial wrappers. For class
6695 iterators add any needed clauses. Allow OMP_FOR_ORIG_DECLS to contain
6696 TREE_LIST for both the original class iterator and the "last" helper
6697 var. Gimplify OMP_FOR_PRE_BODY before the outermost composite
6698 loop, remember has_decl_expr from outer composite loops for the
6699 innermost OMP_SIMD in TREE_PRIVATE bit on OMP_FOR_INIT.
6701 2018-07-09 Martin Sebor <msebor@redhat.com>
6705 * builtins.c (c_strlen): Avoid out-of-bounds warnings when
6706 accessing implicitly initialized array elements.
6707 * expr.c (string_constant): Handle string initializers of
6708 character arrays within aggregates.
6709 * gimple-fold.c (fold_array_ctor_reference): Add argument.
6710 Store element offset. As a special case, handle zero size.
6711 (fold_nonarray_ctor_reference): Same.
6712 (fold_ctor_reference): Add argument. Store subobject offset.
6713 * gimple-fold.h (fold_ctor_reference): Add argument.
6715 2018-07-09 Paul Koning <ni1d@arrl.net>
6717 * config/pdp11/pdp11.c (pdp11_addr_cost): New function.
6718 (pdp11_insn_cost): New function.
6719 (pdp11_md_asm_adjust): New function.
6720 (TARGET_INVALID_WITHIN_DOLOOP): Define.
6721 (pdp11_rtx_costs): Update to match machine better.
6722 (output_addr_const_pdp11): Correct format mismatch warnings.
6723 * config/pdp11/pdp11.h (SLOW_BYTE_ACCESS): Correct definition.
6724 * config/pdp11/pdp11.md: General change to add base_cost and/or
6725 length attributes for use by new pdp11_insn_cost function.
6726 (MIN_BRANCH): Correct definition.
6728 (doloop_end): Use standard pattern name for looping pattern.
6729 (doloop_end_nocc): New.
6730 (movsf): Add another constraint alternative.
6731 (zero_extendqihi2): Add constraint alternatives for not in place
6733 (zero_extendhisi2): Remove.
6734 (shift patterns): Add CC handling variants.
6738 (define_peephole2): New peephole to recognize mov that sets CC for
6741 2018-07-09 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
6743 * config/sparc/sparc.c (sparc_fold_builtin) <SPARC_BUILTIN_PDIST,
6744 SPARC_BUILTIN_PDISTN>: Adapt for signature change of wi::neg,
6747 2018-07-09 Jakub Jelinek <jakub@redhat.com>
6750 * real.c (real_nextafter): Return true if result is denormal.
6752 2018-07-09 Martin Liska <mliska@suse.cz>
6754 * common.opt: Add back wrongly removed attribute.
6756 2018-07-09 Richard Biener <rguenther@suse.de>
6759 * dwarf2out.c (gen_block_die): For an early generated DIE
6760 always output high/low PC attributes.
6762 2018-07-09 Tom de Vries <tdevries@suse.de>
6764 * cfgexpand.c (expand_debug_source_expr): Handle VAR_DECL.
6765 * tree-inline.c (remap_ssa_name): Handle default def ssa_name mapping
6766 onto VAR_DECL with abstract origin.
6768 2018-07-07 Jim Wilson <jimw@sifive.com>
6770 * config/riscv/riscv.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): New.
6772 2018-07-07 Tom de Vries <tdevries@suse.de>
6774 * tree-dfa.c (dump_enumerated_decls): Handle cfun->cfg == NULL.
6776 2018-07-07 Aldy Hernandez <aldyh@redhat.com>
6778 * tree-vrp.c (vrp_int_const_binop): Change overflow type to
6780 (combine_bound): Use wide-int overflow calculation instead of
6782 * calls.c (maybe_warn_alloc_args_overflow): Change overflow type to
6784 * fold-const.c (int_const_binop_2): Same.
6785 (extract_muldiv_1): Same.
6786 (fold_div_compare): Same.
6787 (fold_abs_const): Same.
6789 * poly-int.h (add): Same.
6793 * predict.c (predict_iv_comparison): Same.
6794 * profile-count.c (slow_safe_scale_64bit): Same.
6795 * simplify-rtx.c (simplify_const_binary_operation): Same.
6796 * tree-chrec.c (tree_fold_binomial): Same.
6797 * tree-data-ref.c (split_constant_offset_1): Same.
6798 * tree-if-conv.c (idx_within_array_bound): Same.
6799 * tree-scalar-evolution.c (iv_can_overflow_p): Same.
6800 * tree-ssa-phiopt.c (minmax_replacement): Same.
6801 * tree-vect-loop.c (is_nonwrapping_integer_induction): Same.
6802 * tree-vect-stmts.c (vect_truncate_gather_scatter_offset): Same.
6803 * vr-values.c (vr_values::adjust_range_with_scev): Same.
6804 * wide-int.cc (wi::add_large): Same.
6805 (wi::mul_internal): Same.
6806 (wi::sub_large): Same.
6807 (wi::divmod_internal): Same.
6808 * wide-int.h: Change overflow type to overflow_type for neg, add,
6809 mul, smul, umul, div_trunc, div_floor, div_ceil, div_round,
6810 mod_trunc, mod_ceil, mod_round, add_large, sub_large,
6811 mul_internal, divmod_internal.
6812 (overflow_type): New enum.
6813 (accumulate_overflow): New.
6815 2018-07-06 Kugan Vivekanandarajah <kugan.vivekanandarajah@linaro.org>
6817 * tree-ssa-phiopt.c (cond_removal_in_popcount_pattern): New.
6818 (tree_ssa_phiopt_worker): Call cond_removal_in_popcount_pattern.
6820 2018-07-06 Kugan Vivekanandarajah <kuganv@linaro.org>
6822 * tree-ssa-loop-niter.c (number_of_iterations_popcount): If popcount
6823 argument is checked for zero before entering loop, avoid checking again.
6825 2018-07-06 Kugan Vivekanandarajah <kuganv@linaro.org>
6827 * gimplify.h (generic_expr_could_trap_p): Set as global function.
6828 * gimplify.h (generic_expr_could_trap_p): Likwise.
6829 * tree-scalar-evolution.c (expression_expensive_p): Handle COND_EXPR.
6831 2018-07-06 Jakub Jelinek <jakub@redhat.com>
6833 PR tree-optimization/86401
6834 * fold-const.c (fold_binary_loc) <case BIT_AND_EXPR>: Move the
6835 ((A & N) + B) & M -> (A + B) & M etc. optimization into ...
6836 (fold_bit_and_mask): ... here. New helper function for match.pd.
6837 * fold-const.h (fold_bit_and_mask): Declare.
6838 * match.pd (((A & N) + B) & M -> (A + B) & M): New optimization.
6840 2018-07-06 Peter Bergner <bergner@linux.ibm.com>
6843 * target.def (translate_mode_attribute): New hook.
6844 * targhooks.h (default_translate_mode_attribute): Declare.
6845 * targhooks.c (default_translate_mode_attribute): New function.
6846 * doc/tm.texi.in (TARGET_TRANSLATE_MODE_ATTRIBUTE): New hook.
6847 * doc/tm.texi: Regenerate.
6848 * config/rs6000/rs6000.c (TARGET_TRANSLATE_MODE_ATTRIBUTE): Define.
6849 (rs6000_translate_mode_attribute): New function.
6851 2018-07-06 Paul Koning <ni1d@arrl.net>
6853 * doc/md.texi (define_split): Document DONE and FAIL.
6854 (define_peephole2): Ditto.
6856 2018-07-05 Jeff Law <law@redhat.com>
6858 PR tree-optimization/86010
6859 * tree-ssa-dse.c (compute_trims): More aggressively trim at
6860 both the head and tail of mem* and str* calls.
6862 2018-07-05 Jim Wilson <jimw@sifive.com>
6864 * config.gcc (riscv*-*-*): When setting xlen, handle riscv-*.
6866 2018-07-05 Indu Bhagat <indu.bhagat@oracle.com>
6868 * config/aarch64/aarch64-simd.md: correct flags text for
6869 MIN_EXPR replacement.
6871 2018-07-05 James Clarke <jrtc27@jrtc27.com>
6873 * configure: Regenerated.
6875 2018-07-05 Carl Love <cel@us.ibm.com>
6877 * config/rs6000/rs6000-c.c: Map ALTIVEC_BUILTIN_VEC_UNPACKH for
6878 float argument to VSX_BUILTIN_DOUBLEH_V4SF.
6879 Map ALTIVEC_BUILTIN_VEC_UNPACKL for float argument to
6880 VSX_BUILTIN_DOUBLEL_V4SF.
6882 2018-07-05 Martin Sebor <msebor@redhat.com>
6885 * tree-ssa-strlen.c (maybe_set_strlen_range): Use type size rather
6886 than its domain to compute its the upper bound of a char array.
6888 2018-07-05 Nathan Sidwell <nathan@acm.org>
6890 Replace NO_IMPLICIT_EXTERN_C with SYSTEM_IMPLICIT_EXTERN_C.
6891 * doc/cpp.texi: Update comment.
6892 * doc/tm.texi: Rebuilt.
6893 * doc/tm.texi.in (NO_IMPLICIT_EXTERN_C): Replace with ...
6894 (SYSTEM_IMPLICIT_EXTERN_C): ... this, opposite sense.
6895 * doc/extend.texi (Backwards Compatibility): Clarify it is system
6896 headers affected by extern "C".
6897 * system.h: Poison NO_IMPLICIT_EXTERN_C.
6898 * config/alpha/alpha.h, config/arm/uclinux-elf.h,
6899 config/bfin/elf.h, config/cris/cris.h, config/darwin.h,
6900 config/dragonfly.h, config/freebsd.h, config/gnu-user.h,
6901 config/i386/cygming.h, config/i386/djgpp.h, config/i386/nto.h,
6902 config/ia64/hpux.h, config/lm32/lm32.h, config/lm32/uclinux-elf.h,
6903 config/lynx.h, config/mips/elf.h, config/mmix/mmix.h,
6904 config/netbsd.h, config/pa/pa-hpux.h, config/powerpcspe/sysv4.h,
6905 config/riscv/elf.h, config/rs6000/sysv4.h, config/rtems.h,
6906 config/s390/tpf.h, config/sh/newlib.h, config/sol2.h,
6907 config/sparc/openbsd64.h, config/sparc/sp-elf.h,
6908 config/sparc/sp64-elf.h, config/spu/spu.h,
6909 config/stormy16/stormy16.h, config/v850/v850.h,
6910 config/visium/visium.h, config/vx-common.h, config/xtensa/elf.h: Don't
6911 define NO_IMPLICIT_EXTERN_C.
6912 * config/rs6000/aix.h: Set SYSTEM_IMPLICIT_EXTERN_C.
6914 2018-07-05 Tamar Christina <tamar.christina@arm.com>
6917 * config/arm/arm.c (arm_can_change_mode_class): Use GET_MODE_UNIT_SIZE
6918 instead of GET_MODE_SIZE when comparing Units.
6920 2018-07-05 Tamar Christina <tamar.christina@arm.com>
6923 * rtlanal.c (set_noop_p): Constrain on mode change,
6924 include hard-reg-set.h
6926 2018-07-05 Tamar Christina <tamar.christina@arm.com>
6928 * config/aarch64/aarch64.c (aarch64_expand_movmem): Fix mode size.
6930 2018-07-05 Jakub Jelinek <jakub@redhat.com>
6933 2018-07-04 Maxim Ostapenko <m.ostapenko@samsung.com>
6936 * config/gnu-user.h (LIBASAN_EARLY_SPEC): Pass -lstdc++ for static
6938 * gcc.c: Do not pass LIBUBSAN_SPEC if ASan is enabled with UBSan.
6940 2018-07-04 Maxim Ostapenko <m.ostapenko@samsung.com>
6943 * config/gnu-user.h (LIBASAN_EARLY_SPEC): Pass -lstdc++ for static
6945 * gcc.c: Do not pass LIBUBSAN_SPEC if ASan is enabled with UBSan.
6947 2018-07-04 Eric Botcazou <ebotcazou@adacore.com>
6950 * expmed.c (choose_multiplier): Fix incorrect comparison with mask.
6952 2018-07-04 Aldy Hernandez <aldyh@redhat.com>
6954 * tree-vrp.c (extract_range_from_binary_expr_1): Initialize
6955 neg_*_op* variables.
6957 2018-07-04 Martin Liska <mliska@suse.cz>
6959 * tree-switch-conversion.c: Define
6960 max_ratio_for_speed and max_ratio_for_size constants.
6962 2018-07-04 Denys Vlasenko <dvlasenk@redhat.com>
6963 Martin Liska <mliska@suse.cz>
6968 * common.opt: Rename align options with 'str_' prefix.
6969 * common/config/i386/i386-common.c (set_malign_value): New
6971 (ix86_handle_option): Use it to set -falign-* options/
6972 * config/aarch64/aarch64-protos.h (struct tune_params): Change
6973 type from int to string.
6974 * config/aarch64/aarch64.c: Update default values from int
6976 * config/alpha/alpha.c (alpha_override_options_after_change):
6978 * config/arm/arm.c (arm_override_options_after_change_1): Likewise.
6979 * config/i386/dragonfly.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
6980 max skip conditionally.
6981 * config/i386/freebsd.h (SUBALIGN_LOG): New.
6982 (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
6983 max skip conditionally.
6984 * config/i386/gas.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
6985 max skip conditionally.
6986 * config/i386/gnu-user.h (SUBALIGN_LOG): New.
6987 (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
6988 max skip conditionally.
6989 * config/i386/i386.c (struct ptt): Change type from int to
6991 (ix86_default_align): Set default values.
6992 * config/i386/i386.h (ASM_OUTPUT_MAX_SKIP_PAD): Print
6993 max skip conditionally.
6994 * config/i386/iamcu.h (SUBALIGN_LOG): New.
6995 (ASM_OUTPUT_MAX_SKIP_ALIGN):
6996 * config/i386/lynx.h (ASM_OUTPUT_MAX_SKIP_ALIGN):
6997 * config/i386/netbsd-elf.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
6998 max skip conditionally.
6999 * config/i386/openbsdelf.h (SUBALIGN_LOG): New.
7000 (ASM_OUTPUT_MAX_SKIP_ALIGN) Print max skip conditionally.:
7001 * config/i386/x86-64.h (SUBALIGN_LOG): New.
7002 (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
7003 max skip conditionally.
7004 (ASM_OUTPUT_MAX_SKIP_PAD): Likewise.
7005 * config/ia64/ia64.c (ia64_option_override): Set default values
7006 for alignment options.
7007 * config/m68k/m68k.c: Handle new str_align_* options.
7008 * config/mips/mips.c (mips_set_compression_mode): Change
7010 (mips_option_override): Set default values for options.
7011 * config/powerpcspe/powerpcspe.c (rs6000_option_override_internal):
7013 * config/rs6000/rs6000.c (rs6000_option_override_internal):
7015 * config/rx/rx.c (rx_option_override): Likewise.
7016 * config/rx/rx.h (JUMP_ALIGN): Use align_jumps_log.
7017 (LABEL_ALIGN): Use align_labels_log.
7018 (LOOP_ALIGN): Use align_loops_align.
7019 * config/s390/s390.c (s390_asm_output_function_label): Use new
7021 * config/sh/sh.c (sh_override_options_after_change):
7022 Change type of constants.
7023 * config/spu/spu.c (spu_sched_init): Likewise.
7024 * config/sparc/sparc.c (sparc_option_override): Set default
7026 * config/visium/visium.c (visium_option_override): Likewise.
7027 * config/visium/visium.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Do not
7028 emit p2align format with last argument if it's not needed.
7029 * doc/invoke.texi: Document extended format of -falign-*.
7030 * final.c: Use align_labels alignment.
7031 * flags.h (struct target_flag_state): Change type to use
7033 (struct align_flags_tuple): New.
7034 (struct align_flags): Likewise.
7035 (align_loops_log): Redefine macro to use new types.
7036 (align_loops_max_skip): Redefine macro to use new types.
7037 (align_jumps_log): Redefine macro to use new types.
7038 (align_jumps_max_skip): Redefine macro to use new types.
7039 (align_labels_log): Redefine macro to use new types.
7040 (align_labels_max_skip): Redefine macro to use new types.
7041 (align_functions_log): Redefine macro to use new types.
7042 (align_loops): Redefine macro to use new types.
7043 (align_jumps): Redefine macro to use new types.
7044 (align_labels): Redefine macro to use new types.
7045 (align_functions): Redefine macro to use new types.
7046 (align_functions_max_skip): Redefine macro to use new types.
7047 (align_loops_value): New macro.
7048 (align_jumps_value): New macro.
7049 (align_labels_value): New macro.
7050 (align_functions_value): New macro.
7051 * function.c (invoke_set_current_function_hook): Propagate
7052 alignment values from flags to global variables default in
7054 * ipa-icf.c (sem_function::equals_wpa): Use
7055 cl_optimization_option_eq instead of memcmp.
7056 * lto-streamer.h (cl_optimization_stream_out): Support streaming
7058 (cl_optimization_stream_in): Likewise.
7059 * optc-save-gen.awk: Support strings in cl_optimization.
7060 * opth-gen.awk: Likewise.
7061 * opts.c (finish_options): Remove error checking of invalid
7063 (MAX_CODE_ALIGN): Remove.
7064 (MAX_CODE_ALIGN_VALUE): Likewise.
7065 (parse_and_check_align_values): New function.
7066 (check_alignment_argument): Likewise.
7067 (common_handle_option): Use check_alignment_argument.
7068 * opts.h (parse_and_check_align_values): Declare.
7069 * toplev.c (init_alignments): Remove.
7070 (read_log_maxskip): New.
7071 (parse_N_M): Likewise.
7072 (parse_alignment_opts): Likewise.
7073 (backend_init_target): Remove usage of init_alignments.
7074 * toplev.h (parse_alignment_opts): Declare.
7075 * tree-streamer-in.c (streamer_read_tree_bitfields): Add new
7077 * tree-streamer-out.c (streamer_write_tree_bitfields): Likewise.
7078 * tree.c (cl_option_hasher::equal): New.
7079 * varasm.c: Use new global macros.
7081 2018-07-04 Denys Vlasenko <dvlasenk@redhat.com>
7083 * config/i386/dragonfly.h: (ASM_OUTPUT_MAX_SKIP_ALIGN):
7084 Use a simpler align directive also if MAXSKIP = ALIGN-1.
7085 * config/i386/gas.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
7086 * config/i386/lynx.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
7087 * config/i386/netbsd-elf.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
7088 * config/i386/i386.h (ASM_OUTPUT_MAX_SKIP_PAD): Likewise.
7089 * config/i386/freebsd.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Remove "If N
7090 is large, do at least 8 byte alignment" code. Add SUBALIGN_LOG
7091 define. Use a simpler align directive also if MAXSKIP = ALIGN-1.
7092 * config/i386/gnu-user.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
7093 * config/i386/iamcu.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
7094 * config/i386/openbsdelf.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
7095 * config/i386/x86-64.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
7097 2018-07-04 Martin Liska <mliska@suse.cz>
7098 Jonathan Wakely <jwakely@redhat.com>
7100 * coverage.c: Use correct type.
7101 * doc/invoke.texi: Language correction.
7103 2018-07-03 H.J. Lu <hongjiu.lu@intel.com>
7106 * config/i386/i386.c (rest_of_insert_endbranch): Also generate
7107 ENDBRANCH for non-tail call which may return via indirect branch.
7108 * doc/extend.texi: Document indirect_return attribute.
7110 2018-07-03 Martin Sebor <msebor@redhat.com>
7112 PR tree-optimization/86274
7113 * gimple-ssa-sprintf.c (fmtresult::type_max_digits): Verify
7115 (format_floating): Correct handling of infinities and NaNs.
7117 2018-07-03 Martin Sebor <msebor@redhat.com>
7119 * print-tree.c (print_real_cst): New function.
7120 (print_node_brief): Call it.
7121 (print_node): Ditto.
7123 2018-07-03 Jeff Law <law@redhat.com>
7125 * config/h8300/h8300.md (logical<mode>3_sn, logical<mode>3): Merge
7126 into a single pattern.
7128 * config/h8300/h8300.md (ors code_iterator): New.
7129 (bsetqi_msx, bnotqi_msx patterns and splitters): Consolidate into
7130 a single pattern and single splitter.
7131 (bsethi_msx, bnothi_msx patterns): Consolidate into a single pattern.
7132 (iorqi3_1, xorqi3_1): Likewise.
7133 (iorqi3, xorqi3 expanders): Similarly.
7135 * config/h8300/h8300.md (movmd_internal_normal): Consolidated with
7136 (movmd_internal) into a single pattern using the P mode iterator.
7137 (movmd splitters): Similarly.
7138 (stpcpy_internal_normal, stpcpy_internal): Similarly for thes patterns.
7139 (movsd splitters): Similarly.
7141 * config/h8300/h8300.c (h8300_insn_length_from_table): Consolidate
7142 ADDB, ADDW and ADDL into a single ADD attribute which selects the
7143 right table based on the size of the operand.
7144 * config/h8300/h8300.md (length_table): Corresponding changes. All
7145 references to "addb", "addw" and "addl" changed to "add".
7146 (btst patterns): Merge two variants into a single pattern.
7147 (tstqi, tsthi): Likewise.
7148 (addhi3_incdec, addsi3_incdec): Likewise.
7149 (subhi3_h8300hs, subsi3_h8300hs): Likewise.
7150 (mulhi3, mulsi3): Likewise.
7151 (udivhi3, udivsi3): Likewise.
7152 (divhi3, divsi3): Likewise.
7153 (andorqi3, andorhi3, andorsi3): Likewise.
7155 2018-07-03 Uros Bizjak <ubizjak@gmail.com>
7158 * config/i386/sse.md (uavg<mode>3_ceil): New expander.
7159 (<sse2_avx2>_uavg<mode>3<mask_name>): Simplify expander.
7161 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
7163 PR tree-optimization/85694
7164 * config/aarch64/iterators.md (HADD, RHADD): New int iterators.
7165 (u): Handle UNSPEC_SHADD, UNSPEC_UHADD, UNSPEC_SRHADD and
7167 * config/aarch64/aarch64-simd.md (<u>avg<mode>3_floor)
7168 (<u>avg<mode>3_ceil): New patterns.
7170 2018-07-03 David Malcolm <dmalcolm@redhat.com>
7172 * gcc.dg/vect/slp-perm-1.c: Remove "note: " prefix from
7173 scan-tree-dump directive.
7174 * gcc.dg/vect/slp-perm-2.c: Likewise.
7175 * gcc.dg/vect/slp-perm-3.c: Likewise.
7176 * gcc.dg/vect/slp-perm-5.c: Likewise.
7177 * gcc.dg/vect/slp-perm-6.c: Likewise.
7178 * gcc.dg/vect/slp-perm-7.c: Likewise.
7179 * gcc.dg/vect/slp-perm-8.c: Likewise.
7181 2018-07-03 Marek Polacek <polacek@redhat.com>
7184 * gimple-fold.c (size_must_be_zero_p): Check the type of the size.
7186 2018-07-03 Richard Biener <rguenther@suse.de>
7189 * tree-ssa-structalias.c (find_func_clobbers): Properly
7190 handle indirect calls.
7192 2018-07-03 Jeff Law <law@redhat.com>
7194 * config/h8300/h8300.md (HSI, QHSI, QHSIF): New mode iterators.
7195 (shifts): New code iterator.
7196 (movqi, movhi, movsi, movsf expanders): Consolidate into a single
7197 expander. Fix HImode handling on H8/SX.
7198 (addqi3, addhi3, addsi3 expanders): Consolidate into a single expander.
7199 (subqi3, subhi3, subsi3 expanders): Likewise.
7200 (andqi3, andhi3, andsi3 expanders): Likewise.
7201 (iorqi3, iorhi3, iorsi3 expanders): Likewise.
7202 (xorqi3, xorhi3, xorsi3 expanders): Likewise.
7203 (negqi2, neghi2, negsi2, negsf2 expanders): Likewise.
7204 (one_cmplqi2, one_cmplhi2, one_cmplsi2): Likewise.
7205 (zero_extendqihi2, zero_extendqisi2): Likewise.
7206 (extendqihi2, extendqisi2): Likewise.
7207 (rotlqi3, rotlhi3, rotlsi3): Likewise.
7208 (neghi2_h8300, negsi2_h8300): Likewise for these patterns.
7209 (rotlqi3_1, rotlhi3_1): Likewise.
7210 (logicalhi3_sn, logicalsi3_sn): Likewise.
7211 (logicalhi3, logicalsi3): Likewise.
7213 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
7215 * tree-vect-patterns.c (vect_recog_rotate_pattern)
7216 (vect_recog_vector_vector_shift_pattern, vect_recog_divmod_pattern)
7217 (vect_recog_mixed_size_cond_pattern, adjust_bool_pattern_cast)
7218 (adjust_bool_pattern, vect_recog_bool_pattern): Pass the vector
7219 type to append_pattern_def_seq instead of creating a stmt_vec_info
7221 (build_mask_conversion): Likewise. Remove vinfo argument.
7222 (vect_add_conversion_to_patterm): Likewise, renaming to...
7223 (vect_add_conversion_to_pattern): ...this.
7224 (vect_recog_mask_conversion_pattern): Update call to
7225 build_mask_conversion. Pass the vector type to
7226 append_pattern_def_seq here too.
7227 (vect_recog_gather_scatter_pattern): Update call to
7228 vect_add_conversion_to_pattern.
7230 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
7232 * tree-vect-patterns.c (new_pattern_def_seq): Delete.
7233 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern)
7234 (vect_recog_widen_op_pattern, vect_recog_over_widening_pattern)
7235 (vect_recog_rotate_pattern, vect_synth_mult_by_constant): Don't set
7236 STMT_VINFO_PATTERN_DEF_SEQ to null here.
7237 (vect_recog_pow_pattern, vect_recog_vector_vector_shift_pattern)
7238 (vect_recog_mixed_size_cond_pattern, vect_recog_bool_pattern): Use
7239 append_pattern_def_seq instead of new_pattern_def_seq.
7240 (vect_recog_divmod_pattern): Do both of the above.
7241 (vect_pattern_recog_1): Assert that STMT_VINO_PATTERN_DEF_SEQ
7244 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
7246 * tree-vect-patterns.c (vect_recog_dot_prod_pattern):
7247 (vect_recog_sad_pattern, vect_recog_widen_op_pattern)
7248 (vect_recog_widen_mult_pattern, vect_recog_pow_pattern):
7249 (vect_recog_widen_sum_pattern, vect_recog_over_widening_pattern)
7250 (vect_recog_average_pattern, vect_recog_cast_forwprop_pattern)
7251 (vect_recog_widen_shift_pattern, vect_recog_rotate_pattern)
7252 (vect_recog_vector_vector_shift_pattern, vect_synth_mult_by_constant)
7253 (vect_recog_mult_pattern, vect_recog_divmod_pattern)
7254 (vect_recog_mixed_size_cond_pattern, vect_recog_bool_pattern)
7255 (vect_recog_mask_conversion_pattern): Replace vec<gimple *>
7256 parameter with a single stmt_vec_info.
7257 (vect_recog_func_ptr): Likewise.
7258 (vect_recog_gather_scatter_pattern): Likewise, folding in...
7259 (vect_try_gather_scatter_pattern): ...this.
7260 (vect_pattern_recog_1): Remove stmts_to_replace and just pass
7261 the stmt_vec_info of the statement to be matched. Don't clear
7262 STMT_VINFO_RELATED_STMT.
7263 (vect_pattern_recog): Update call accordingly.
7265 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
7267 PR tree-optimization/85694
7268 * doc/md.texi (avgM3_floor, uavgM3_floor, avgM3_ceil)
7269 (uavgM3_ceil): Document new optabs.
7270 * doc/sourcebuild.texi (vect_avg_qi): Document new target selector.
7271 * internal-fn.def (IFN_AVG_FLOOR, IFN_AVG_CEIL): New internal
7273 * optabs.def (savg_floor_optab, uavg_floor_optab, savg_ceil_optab)
7274 (savg_ceil_optab): New optabs.
7275 * tree-vect-patterns.c (vect_recog_average_pattern): New function.
7276 (vect_vect_recog_func_ptrs): Add it.
7277 * tree-vect-stmts.c (vectorizable_call): Get the type of the zero
7278 constant directly from the associated lhs.
7280 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
7282 * tree-vect-patterns.c (vect_split_statement): New function.
7283 (vect_convert_input): Use it to try to split an existing cast.
7285 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
7287 * poly-int.h (print_hex): New function.
7288 * dumpfile.h (dump_dec, dump_hex): Declare.
7289 * dumpfile.c (dump_dec, dump_hex): New poly_wide_int functions.
7290 * tree-vectorizer.h (_stmt_vec_info): Add min_output_precision,
7291 min_input_precision, operation_precision and operation_sign.
7292 * tree-vect-patterns.c (vect_get_range_info): New function.
7293 (vect_same_loop_or_bb_p, vect_single_imm_use)
7294 (vect_operation_fits_smaller_type): Delete.
7295 (vect_look_through_possible_promotion): Add an optional
7296 single_use_p parameter.
7297 (vect_recog_over_widening_pattern): Rewrite to use new
7298 stmt_vec_info infomration. Handle one operation at a time.
7299 (vect_recog_cast_forwprop_pattern, vect_narrowable_type_p)
7300 (vect_truncatable_operation_p, vect_set_operation_type)
7301 (vect_set_min_input_precision): New functions.
7302 (vect_determine_min_output_precision_1): Likewise.
7303 (vect_determine_min_output_precision): Likewise.
7304 (vect_determine_precisions_from_range): Likewise.
7305 (vect_determine_precisions_from_users): Likewise.
7306 (vect_determine_stmt_precisions, vect_determine_precisions): Likewise.
7307 (vect_vect_recog_func_ptrs): Put over_widening first.
7309 (vect_pattern_recog): Call vect_determine_precisions.
7311 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
7313 * tree-vect-patterns.c (vect_mark_pattern_stmts): Remove pattern
7314 statements that have been replaced by further pattern statements.
7315 (vect_pattern_recog_1): Clear STMT_VINFO_PATTERN_DEF_SEQ on failure.
7317 2018-07-03 Richard Biener <rguenther@suse.de>
7319 * tree-vect-stmts.c (vect_is_simple_use): Consolidate dumping,
7320 always set *dt. Dump vectype in vectype overload.
7321 * dumpfile.h (dump_gimple_expr): New function.
7322 (dump_gimple_expr_loc): Likewise.
7323 * dumpfile.c (dump_gimple_expr): New function.
7324 (dump_gimple_expr_loc): Likewise.
7326 2018-07-02 Jeff Law <law@redhat.com>
7328 * config/h8300/h8300.md (movqi_h8300, movqi_h8300hs): Consolidate
7329 the H8/300, H8/300H and H8/S variants into a single pattern.
7330 (movhi_h8300, movqi_h8300hs): Similarly.
7331 (pushqi_h8300hs, pushhi_h8300hs): Consolidate into a single pattern.
7332 (QHI mode iterator): New.
7334 * config/h8300/h8300.md: Remove trailing whitespace.
7336 2018-07-02 Jim Wilson <jimw@sifive.com>
7338 * config/riscv/riscv.c (riscv_expand_epilogue): Use emit_jump_insn
7339 instead of emit_insn for interrupt returns.
7340 * config/riscv/riscv.md (riscv_met): Add (return) to rtl.
7341 (riscv_sret, riscv_uret): Likewise.
7343 2018-07-02 David Malcolm <dmalcolm@redhat.com>
7345 * pretty-print.c (selftest::test_pp_format): Move save and restore
7346 of quotes to class auto_fix_quotes, and add an instance.
7347 * selftest.c: Include "intl.h".
7348 (selftest::auto_fix_quotes::auto_fix_quotes): New ctor.
7349 (selftest::auto_fix_quotes::~auto_fix_quotes): New dtor.
7350 * selftest.h (selftest::auto_fix_quotes): New class.
7352 2018-07-02 Richard Henderson <richard.henderson@linaro.org>
7354 * config/aarch64/aarch64-protos.h, config/aarch64/aarch64.c
7355 (aarch64_sve_prepare_conditional_op): Remove.
7356 * config/aarch64/aarch64-sve.md (cond_<SVE_INT_BINARY><SVE_I>):
7357 Allow aarch64_simd_reg_or_zero as select operand; remove
7358 the aarch64_sve_prepare_conditional_op call.
7359 (cond_<SVE_INT_BINARY_SD><SVE_SDI>): Likewise.
7360 (cond_<SVE_COND_FP_BINARY><SVE_F>): Likewise.
7361 (*cond_<SVE_INT_BINARY><SVE_I>_z): New pattern.
7362 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_z): New pattern.
7363 (*cond_<SVE_COND_FP_BINARY><SVE_F>_z): New pattern.
7364 (*cond_<SVE_INT_BINARY><SVE_I>_any): New pattern.
7365 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_any): New pattern.
7366 (*cond_<SVE_COND_FP_BINARY><SVE_F>_any): New pattern
7367 and a splitters to match all of the *_any patterns.
7368 * config/aarch64/predicates.md (aarch64_sve_any_binary_operator): New.
7370 * config/aarch64/iterators.md (SVE_INT_BINARY_REV): Remove.
7371 (SVE_COND_FP_BINARY_REV): Remove.
7372 (sve_int_op_rev, sve_fp_op_rev): New.
7373 * config/aarch64/aarch64-sve.md (*cond_<SVE_INT_BINARY><SVE_I>_0): New.
7374 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_0): New.
7375 (*cond_<SVE_COND_FP_BINARY><SVE_F>_0): New.
7376 (*cond_<SVE_INT_BINARY><SVE_I>_2): Rename, add movprfx alternative.
7377 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_2): Similarly.
7378 (*cond_<SVE_COND_FP_BINARY><SVE_F>_2): Similarly.
7379 (*cond_<SVE_INT_BINARY><SVE_I>_3): Similarly; use sve_int_op_rev.
7380 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_3): Similarly.
7381 (*cond_<SVE_COND_FP_BINARY><SVE_F>_3): Similarly; use sve_fp_op_rev.
7383 * config/aarch64/aarch64-sve.md (cond_<SVE_COND_FP_BINARY><SVE_F>):
7384 Remove match_dup 1 from the inner unspec.
7385 (*cond_<SVE_COND_FP_BINARY><SVE_F>): Likewise.
7387 * config/aarch64/aarch64.md (movprfx): New attr.
7388 (length): Default movprfx to 8.
7389 * config/aarch64/aarch64-sve.md (*mul<SVE_I>3): Add movprfx alt.
7390 (*madd<SVE_I>, *msub<SVE_I): Likewise.
7391 (*<su>mul<SVE_I>3_highpart): Likewise.
7392 (*<SVE_INT_BINARY_SD><SVE_SDI>3): Likewise.
7393 (*v<ASHIFT><SVE_I>3): Likewise.
7394 (*<su><MAXMIN><SVE_I>3): Likewise.
7395 (*<su><MAXMIN><SVE_F>3): Likewise.
7396 (*fma<SVE_F>4, *fnma<SVE_F>4): Likewise.
7397 (*fms<SVE_F>4, *fnms<SVE_F>4): Likewise.
7398 (*div<SVE_F>4): Likewise.
7400 2018-07-02 Richard Sandiford <richard.sandiford@arm.com>
7402 * tree-vect-patterns.c (vect_recog_widen_shift_pattern): Fix typo
7405 2018-07-02 Richard Biener <rguenther@suse.de>
7407 PR tree-optimization/86363
7408 * tree-ssa-sccvn.c (vn_reference_lookup_3): Check the
7409 memset argument refers to a non-variable address.
7411 2018-07-02 Aldy Hernandez <aldyh@redhat.com>
7413 * tree-vrp.c (extract_range_from_binary_expr_1): Abstract a lot of the
7414 {PLUS,MINUS}_EXPR code to...
7415 (adjust_symbolic_bound): ...here,
7416 (combine_bound): ...here,
7417 (set_value_range_with_overflow): ...and here.
7419 2018-07-02 Aldy Hernandez <aldyh@redhat.com>
7421 * tree-vrp.c (extract_range_from_unary_expr): Abstract ABS_EXPR
7423 (extract_range_from_abs_expr): ...here.
7425 2018-07-02 Eric Botcazou <ebotcazou@adacore.com>
7427 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Do not overrule
7428 -fno-omit-frame-pointer when not optimizing.
7430 2018-07-02 Martin Liska <mliska@suse.cz>
7433 * ipa-pure-const.c (malloc_candidate_p): Revert usage of ::get.
7434 (propagate_nothrow): Likewise.
7436 2018-07-02 Martin Liska <mliska@suse.cz>
7439 * ipa-inline.c (early_inliner): Revert wrongly added ::get call.
7441 2018-07-02 David Malcolm <dmalcolm@redhat.com>
7443 * dumpfile.c (dump_generic_expr_loc): Undo removal of this
7444 function in r262149, changing "loc" param from source_location to
7445 const dump_location_t &.
7446 * dumpfile.h (dump_generic_expr_loc): Undo removal of this
7447 declaration, as above.
7449 2018-07-01 Paul Koning <ni1d@arrl.net>
7451 * common/config/pdp11/pdp11-common.c (pdp11_handle_option): Handle
7452 -munit-asm, -mgnu-asm, -mdec-asm.
7453 * config/pdp11/pdp11-protos.h (pdp11_gen_int_label): New.
7454 (pdp11_output_labelref): New.
7455 (pdp11_output_def): New.
7456 (pdp11_output_addr_vec_elt): New.
7457 * config/pdp11/pdp11.c: Use tab between opcode and operands. Use
7458 %# and %@ format codes.
7459 (pdp11_option_override): New.
7460 (TARGET_ASM_FILE_START_FILE_DIRECTIVE): Define.
7461 (pdp11_output_ident): New.
7462 (pdp11_asm_named_section): New.
7463 (pdp11_asm_init_sections): New.
7464 (pdp11_file_start): New.
7465 (pdp11_file_end): New.
7466 (output_ascii): Use .ascii/.asciz for -mdec-asm.
7467 (pdp11_asm_print_operand): Update %# and %$ for -mdec-asm. Add
7468 %o, like %c but octal.
7469 (pdp11_option_override): New.
7470 * config/pdp11/pdp11.h (TEXT_SECTION_ASM_OP): Update for
7472 (DATA_SECTION_ASM_OP): Ditto.
7473 (READONLY_DATA_SECTION_ASM_OP): New.
7474 (IS_ASM_LOGICAL_LINE_SEPARATOR): New.
7475 (ASM_GENERATE_INTERNAL_LABEL): Use new function.
7476 (ASM_OUTPUT_LABELREF): Ditto.
7477 (ASM_OUTPUT_DEF): Ditto.
7478 (ASM_OUTPUT_EXTERNAL): New.
7479 (ASM_OUTPUT_SOURCE_FILENAME): New.
7480 (ASM_OUTPUT_ADDR_VEC_ELT): Use new function.
7481 (ASM_OUTPUT_SKIP): Update for -mdec-asm.
7482 * config/pdp11/pdp11.md: Use tab between opcode and operands. Use
7483 %# and %@ format codes.
7484 * config/pdp11/pdp11.opt (mgnu-asm): New.
7485 (mdec-asm): Conflicts with -mgnu-asm and -munix-asm.
7486 (munix-asm): Conflicts with -mdec-asm and -mgnu-asm.
7487 * doc/invoke.txt (PDP-11 Options): Add -mgnu-asm.
7489 2018-07-01 Aldy Hernandez <aldyh@redhat.com>
7491 * tree-ssa-threadupdate.c (mark_threaded_blocks): Avoid
7492 dereferencing path[] beyond its length.
7494 (debug_all_paths): New.
7495 (rewire_first_differing_edge): New.
7496 (adjust_paths_after_duplication): New.
7497 (duplicate_thread_path): Call adjust_paths_after_duplication.
7499 (thread_through_all_blocks): Add new argument to
7500 duplicate_thread_path.
7502 2018-06-30 Jim Wilson <jimw@sifive.com>
7504 * config/riscv/predicates.md (p2m1_shift_operand): New.
7505 (high_mask_shift_operand): New.
7506 * config/riscv/riscv.md (lshrsi3_zero_extend_3+1): New combiner
7507 pattern using p2m1_shift_operand.
7508 (lshsi3_zero_extend_3+2): New combiner pattern using
7509 high_mask_shift_operand.
7511 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
7513 * tree-vect-patterns.c (vect_get_external_def_edge): New function,
7515 (vect_recog_rotate_pattern): ...here.
7516 (vect_convert_input): Try to insert casts of invariants in the
7518 * tree-vect-loop-manip.c (vect_loop_versioning): Don't require the
7519 preheader to be empty.
7521 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
7523 * tree-vect-patterns.c (append_pattern_def_seq): Take an optional
7524 vector type. If given, install it in the new statement's
7526 (vect_element_precision): New function.
7527 (vect_unpromoted_value): New struct.
7528 (vect_unpromoted_value::vect_unpromoted_value): New function.
7529 (vect_unpromoted_value::set_op): Likewise.
7530 (vect_look_through_possible_promotion): Likewise.
7531 (vect_joust_widened_integer, vect_joust_widened_type): Likewise.
7532 (vect_widened_op_tree, vect_convert_input): Likewise.
7533 (vect_convert_inputs, vect_convert_output): Likewise.
7534 (vect_recog_dot_prod_pattern): Use vect_look_through_possible_promotion
7535 to handle the optional cast of the multiplication result and
7536 vect_widened_op_tree to detect the widened multiplication itself.
7537 Do not require the input and output of promotion casts to have
7538 the same sign, but base the signedness of the operation on the
7539 input rather than the result. If the pattern includes two
7540 promotions, check that those promotions have the same sign.
7541 Do not restrict the MULT_EXPR handling to a double-width result;
7542 handle quadruple-width results and wider. Use vect_convert_inputs
7543 to convert the inputs to the common type.
7544 (vect_recog_sad_pattern): Use vect_look_through_possible_promotion
7545 to handle the optional cast of the ABS result. Also allow a sign
7546 change or a sign extension between the ABS and MINUS.
7547 Use vect_widened_op_tree to detect the widened subtraction and use
7548 vect_convert_inputs to convert the inputs to the common type.
7549 (vect_handle_widen_op_by_const): Delete.
7550 (vect_recog_widen_op_pattern): New function.
7551 (vect_recog_widen_mult_pattern): Use it.
7552 (vect_recog_widen_shift_pattern): Likewise.
7553 (vect_recog_widen_sum_pattern): Use
7554 vect_look_through_possible_promotion to handle the promoted
7557 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
7559 * gimple-iterator.c (gsi_for_stmt): Add a new overload that takes
7560 the containing gimple_seq *.
7561 * gimple-iterator.h (gsi_for_stmt): Declare it.
7562 * tree-vect-patterns.c (vect_recog_dot_prod_pattern)
7563 (vect_recog_sad_pattern, vect_recog_widen_sum_pattern)
7564 (vect_recog_widen_shift_pattern, vect_recog_rotate_pattern)
7565 (vect_recog_vector_vector_shift_pattern, vect_recog_divmod_pattern)
7566 (vect_recog_mask_conversion_pattern): Remove STMT_VINFO_IN_PATTERN_P
7568 (vect_init_pattern_stmt, vect_set_pattern_stmt): New functions,
7570 (vect_mark_pattern_stmts): ...here. Handle cases in which the
7571 statement being replaced is part of an existing pattern
7572 definition sequence, inserting the new pattern statements before
7574 (vect_pattern_recog_1): Don't return a bool. If the statement
7575 is already part of a pattern, instead apply pattern matching
7576 to the pattern definition statements. Don't clear the
7577 STMT_VINFO_RELATED_STMT if is_pattern_stmt_p.
7578 (vect_pattern_recog): Don't break after the first match;
7579 continue processing the pattern definition statements instead.
7580 Don't bail out for STMT_VINFO_IN_PATTERN_P here.
7582 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
7584 * tree-vect-patterns.c (vect_reassociating_reduction_p): New function.
7585 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern)
7586 (vect_recog_widen_sum_pattern): Use it.
7588 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
7590 * tree-vect-loop.c (vectorizable_reduction): Assert that the
7591 phi is not a pattern statement and has not been replaced by
7592 a pattern statement.
7593 * tree-vect-patterns.c (type_conversion_p): Don't check
7594 STMT_VINFO_IN_PATTERN_P.
7595 (vect_recog_vector_vector_shift_pattern): Likewise.
7596 (vect_recog_dot_prod_pattern): Expect vect_is_simple_use to return
7597 the pattern statement rather than the original statement; check
7598 directly for a WIDEN_MULT_EXPR here.
7599 * tree-vect-slp.c (vect_get_and_check_slp_defs): Expect
7600 vect_is_simple_use to return the pattern statement rather
7601 than the original statement; use is_pattern_stmt_p to check
7602 for such a pattern statement.
7603 * tree-vect-stmts.c (process_use): Expect vect_is_simple_use
7604 to return the pattern statement rather than the original statement;
7605 don't do the same transformation here.
7606 (vect_is_simple_use): If the defining statement has been replaced
7607 by a pattern statement, return the pattern statement instead.
7608 Remove the corresponding (local) transformation from the vectype
7611 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
7613 * tree-vectorizer.h (vect_is_simple_use): Move the gimple ** to the
7614 end and default to null.
7615 * tree-vect-loop.c (vect_create_epilog_for_reduction)
7616 (vectorizable_reduction): Update calls accordingly, dropping the
7617 gimple ** argument if the passed-back statement isn't needed.
7618 * tree-vect-patterns.c (vect_get_internal_def, type_conversion_p)
7619 (vect_recog_rotate_pattern): Likewise.
7620 (vect_recog_mask_conversion_pattern): Likewise.
7621 * tree-vect-slp.c (vect_get_and_check_slp_defs): Likewise.
7622 (vect_mask_constant_operand_p): Likewise.
7623 * tree-vect-stmts.c (is_simple_and_all_uses_invariant, process_use):
7624 (vect_model_simple_cost, vect_get_vec_def_for_operand): Likewise.
7625 (get_group_load_store_type, get_load_store_type): Likewise.
7626 (vect_check_load_store_mask, vect_check_store_rhs): Likewise.
7627 (vectorizable_call, vectorizable_simd_clone_call): Likewise.
7628 (vectorizable_conversion, vectorizable_assignment): Likewise.
7629 (vectorizable_shift, vectorizable_operation): Likewise.
7630 (vectorizable_store, vect_is_simple_cond): Likewise.
7631 (vectorizable_condition, vectorizable_comparison): Likewise.
7632 (get_same_sized_vectype, vect_get_mask_type_for_stmt): Likewise.
7633 (vect_is_simple_use): Rename the def_stmt argument to def_stmt_out
7634 and move it to the end. Cope with null def_stmt_outs.
7636 2018-06-30 Bernd Edlinger <bernd.edlinger@hotmail.de>
7638 * Makefile.in (FLAGS_TO_PASS): Add RANLIB_FOR_TARGET.
7640 2018-06-29 Jeff Law <law@redhat.com>
7642 * config/v850/v850.c (v850_legitimate_address_p): Handle large
7643 displacements for TARGET_V850E2V3 and newer.
7644 (TARGET_LRA_P): Remove. Defaults to LRA now.
7645 * config/v850/v850.md (sign23byte_load): Remove.
7646 (unsign23byte_load, sign23hword_load, unsign23hword_load): Likewise.
7647 (23word_load, 23byte_store, 23hword_store, 23word_store): Likewise.
7649 2018-06-29 Martin Liska <mliska@suse.cz>
7652 * coverage.c (coverage_init): Mangle full path name.
7653 * doc/invoke.texi: Document the change.
7654 * gcov-io.c (mangle_path): New.
7655 * gcov-io.h (mangle_path): Likewise.
7656 * gcov.c (mangle_name): Use mangle_path for path mangling.
7658 2018-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7660 * config/arm/arm.c (output_move_double): Don't allow STRD instructions
7661 if starting source register is not even.
7663 2018-06-29 Martin Liska <mliska@suse.cz>
7665 PR tree-optimization/86263
7666 * tree-switch-conversion.c (switch_decision_tree::try_switch_expansion):
7667 Make edge redirection.
7669 2018-06-29 David Malcolm <dmalcolm@redhat.com>
7671 * dumpfile.c (dump_loc): Add indentation based on scope depth.
7672 (dump_scope_depth): New variable.
7673 (get_dump_scope_depth): New function.
7674 (dump_begin_scope): New function.
7675 (dump_end_scope): New function.
7676 * dumpfile.h (get_dump_scope_depth): New declaration.
7677 (dump_begin_scope): New declaration.
7678 (dump_end_scope): New declaration.
7679 (class auto_dump_scope): New class.
7680 (AUTO_DUMP_SCOPE): New macro.
7681 * tree-vectorizer.h (DUMP_VECT_SCOPE): Reimplement in terms of
7684 2018-06-29 Richard Biener <rguenther@suse.de>
7686 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Assert
7687 compute_all_dependences succeeds.
7688 * tree-vect-loop.c (vect_get_datarefs_in_loop): Fail early if we
7689 exceed --param loop-max-datarefs-for-datadeps.
7691 2018-06-29 Jakub Jelinek <jakub@redhat.com>
7693 * config/rs6000/t-rs6000: Append rs6000-modes.h to TM_H.
7695 2018-06-28 Uros Bizjak <ubizjak@gmail.com>
7698 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): Use
7699 alternative 0 in preferred_for_speed attribute.
7701 2018-06-28 Paul Koning <ni1d@arrl.net>
7703 * config/pdp11/pdp11-protos.h (pdp11_shift_length): New function.
7704 * config/pdp11/pdp11.c (pdp11_shift_length): New function.
7705 * config/pdp11/pdp11.h (ADJUST_INSN_LENGTH): Remove.
7706 * config/pdp11/pdp11.md: Correct "length" attribute calculation
7707 for shift insn patterns.
7709 2018-06-28 David Malcolm <dmalcolm@redhat.com>
7711 * cgraph.c (cgraph_node::get_body): Replace assignments to
7712 "dump_file" with calls to set_dump_file.
7713 * dumpfile.c (alt_dump_file): Make static, and group with...
7714 (alt_flags): ...this definition.
7715 (dumps_are_enabled): New variable.
7716 (refresh_dumps_are_enabled): New function.
7717 (set_dump_file): New function.
7718 (set_alt_dump_file): New function.
7719 (gcc::dump_manager::dump_start): Replace assignments to
7720 "dump_file" and "alt_dump_file" with calls to set_dump_file and
7722 (gcc::dump_manager::dump_finish): Likewise.
7723 * dumpfile.h (alt_dump_file): Delete decl.
7724 (dumps_are_enabled): New variable decl.
7725 (set_dump_file): New function decl.
7726 (dump_enabled_p): Rewrite in terms of new "dumps_are_enabled"
7728 * tree-nested.c (lower_nested_functions): Replace assignments to
7729 "dump_file" with calls to set_dump_file.
7731 2018-06-28 Eric Botcazou <ebotcazou@adacore.com>
7733 * tree-cfg.c (verify_gimple_in_cfg): Call verify_location on the
7734 goto_locus of each outgoing edge of each basic block.
7736 2018-06-28 Richard Biener <rguenther@suse.de>
7738 * dwarf2out.c (decl_scope_table): Remove.
7739 (push_decl_scope): Likewise.
7740 (pop_decl_scope): Likewise.
7741 (gen_type_die_for_member): Do not call push/pop_decl_scope.
7742 (gen_struct_or_union_type_die): Likewise.
7743 (gen_tagged_type_die): Likewise.
7744 (dwarf2out_init): Do not initialize decl_scope_table.
7745 (dwarf2out_c_finalize): Do not free it.
7747 2018-06-28 Richard Biener <rguenther@suse.de>
7749 * dwarf2out.c (gen_subprogram_die): Use is_unit_die when
7750 deciding whether to not re-use a DIE.
7752 2018-06-28 Richard Biener <rguenther@suse.de>
7754 * dwarf2out.c (gen_subprogram_die): Always re-use DIEs with an
7755 DW_AT_abstract_origin attribute.
7757 2018-06-28 Martin Liska <mliska@suse.cz>
7759 * tree-switch-conversion.c (jump_table_cluster::can_be_handled):
7760 Use newly introduced constants.
7761 * tree-switch-conversion.h (struct jump_table_cluster):
7762 Define max_ratio_for_size and max_ratio_for_speed.
7764 2018-06-28 Martin Liska <mliska@suse.cz>
7766 * tree-switch-conversion.c (jump_table_cluster::find_jump_tables):
7767 Add new checking assert to catch invalid state.
7768 (jump_table_cluster::can_be_handled): Handle single case
7770 (jump_table_cluster::is_beneficial): Bail out for such case.
7771 (bit_test_cluster::find_bit_tests):
7772 Add new checking assert to catch invalid state.
7773 (bit_test_cluster::can_be_handled): Handle single case
7775 (bit_test_cluster::is_beneficial): Bail out for such case.
7776 (switch_decision_tree::analyze_switch_statement):
7779 2018-06-28 Martin Liska <mliska@suse.cz>
7781 * common.opt: Introduce -completion option.
7782 * gcc.c (driver_handle_option): Handle it.
7783 (driver::main): Print completions if completion
7785 * opt-suggestions.c (option_proposer::get_completions):
7787 (option_proposer::suggest_completion): Likewise.
7788 (option_proposer::find_param_completions): Likewise.
7789 (verify_autocompletions): Likewise.
7790 (test_completion_valid_options): Likewise.
7791 (test_completion_valid_params): Likewise.
7792 (in_completion_p): Likewise.
7793 (empty_completion_p): Likewise.
7794 (test_completion_partial_match): Likewise.
7795 (test_completion_garbage): Likewise.
7796 (opt_proposer_c_tests): Likewise.
7797 * opt-suggestions.h: Declare new functions.
7798 * opts.c (common_handle_option): Handle OPT__completion_.
7799 * selftest-run-tests.c (selftest::run_tests): Add
7800 opt_proposer_c_tests.
7801 * selftest.c (assert_str_startswith): New.
7802 * selftest.h (assert_str_startswith): Likewise.
7803 (opt_proposer_c_tests): New.
7804 (ASSERT_STR_STARTSWITH): Likewise.
7806 2018-06-28 Martin Liska <mliska@suse.cz>
7808 * Makefile.in: Add opt-suggestions.o.
7809 * gcc-main.c: Include opt-suggestions.h.
7810 * gcc.c (driver::driver): Likewise.
7811 (driver::~driver): Remove m_option_suggestions.
7812 (driver::build_option_suggestions): Moved to option_proposer.
7813 (driver::suggest_option): Likewise.
7814 (driver::handle_unrecognized_options): Use option_proposer.
7815 * gcc.h (class driver): Add new memver m_option_proposer.
7816 * opt-suggestions.c: New file.
7817 * opt-suggestions.h: New file.
7819 2018-06-28 Martin Liska <mliska@suse.cz>
7821 * vec.h (class auto_string_vec): New (moved from auto_argvec).
7822 (auto_string_vec::~auto_string_vec): Likewise.
7824 2018-06-28 Eric Botcazou <ebotcazou@adacore.com>
7826 * tree-inline.h (struct copy_body_data): Move remapping_type_depth and
7827 prevent_decl_creation_for_types fields up and add reset_location field.
7828 * tree-inline.c (remap_gimple_stmt): Force input_location on the new
7829 statement if id->reset_location is true.
7830 (copy_edges_for_bb): Do not set goto_locus on the new edges if
7831 id->reset_location is true.
7832 (copy_phis_for_bb): Force input_location on the arguments if
7833 id->reset_location is true.
7834 (expand_call_inline): Set id->reset_location if DECL_IGNORED_P
7835 is set on the function to be inlined.
7837 2018-06-27 Stephan Bergmann <sbergman@redhat.com>
7839 * doc/invoke.texi (Debugging Options): Add -gsplit-dwarf.
7841 2018-06-27 Dimitar Dimitrov <dimitar@dinux.eu>
7843 * lra-eliminations.c (update_reg_eliminate): Mark all spanning hard
7844 registers for Pmode.
7845 * lra-lives.c (check_pseudos_live_through_calls): Mark all spanning
7846 hard registers for the clobbered pseudo.
7848 2018-06-27 Paul Koning <ni1d@arrl.net>
7850 * common/config/pdp11/pdp11-common.c (pdp11_handle_option): Handle
7851 mutually exclusive options.
7852 * config/pdp11/constraints.md (h): New constraint.
7853 (O): Update definition to match shift code generation.
7854 (D): New constraint.
7855 * config/pdp11/pdp11-modes.def (CCNZ): Define mode.
7857 * config/pdp11/pdp11-protos.h (int_no_side_effect_operand): New
7859 (output_jump): Change arguments.
7860 (pdp11_fixed_cc_regs): New function.
7861 (pdp11_cc_mode): Ditto.
7862 (pdp11_expand_shift): Ditto.
7863 (pdp11_assemble_shift): Ditto.
7864 (pdp11_small_shift): Ditto.
7865 (pdp11_branch_cost): Remove.
7866 * config/pdp11/pdp11.c (pdp11_assemble_integer): Remove comments
7868 (pdp11_register_move_cost): Update for CC registers.
7869 (pdp11_rtx_costs): Add case for LSHIFTRT.
7870 (pdp11_output_jump): Add CCNZ mode conditional branches.
7871 (notice_update_cc_on_set): Remove.
7872 (pdp11_cc_mode): New function.
7873 (simple_memory_operand): Correct pre/post decrement case.
7874 (no_side_effect_operand): New function.
7875 (pdp11_regno_reg_class): Add CC_REGS class.
7876 (pdp11_fixed_cc_regs): New function.
7877 (pdp11_small_shift): New function.
7878 (pdp11_expand_shift): New function to expand shift insns.
7879 (pdp11_assemble_shift): New function to output shifts.
7880 (pdp11_branch_cost): Remove.
7881 (pdp11_modes_tieable_p): Make QI/HI modes tieable.
7882 * config/pdp11/pdp11.h (SIZE_TYPE): Ensure 16-bit type.
7883 (WCHAR_TYPE): Ditto.
7884 (PTRDIFF_TYPE): Ditto.
7885 (ADJUST_INSN_LENGTH): New macro.
7886 (FIXED_REGISTERS): Add CC registers.
7887 (CALL_USED_REGISTERS): Ditto.
7889 (REG_CLASS_NAMES): Ditto.
7890 (REG_CLASS_CONTENTS): Ditto.
7891 (SELECT_CC_MODE): Use new function.
7892 (TARGET_FLAGS_REGNUM): New macro.
7893 (TARGET_FIXED_CONDITION_CODE_REGS): Ditto.
7894 (cc0_reg_rtx): Remove.
7895 (CC_STATUS_MDEP): Remove.
7896 (CC_STATUS_MDEFP_INIT): Remove.
7897 (CC_IN_FPU): Remove.
7898 (NOTICE_UPDATE_CC): Remove.
7899 (REGISTER_NAMES): Add CC registers.
7900 (BRANCH_COST): Change to constant 1.
7901 * config/pdp11/pdp11.md: Rewrite for CCmode condition code
7903 * config/pdp11/pdp11.opt (mbcopy): Remove.
7904 (mbcopy-builtin): Remove.
7905 (mbranch-cheap): Remove.
7906 (mbranch-expensive): Remove.
7907 * config/pdp11/predicates.md (expand_shift_operand): Update to
7908 match shift code generation.
7909 (ccnz_operator): New predicate.
7910 * doc/invoke.texi (PDP-11 Options): Remove deleted options
7911 -mbcopy, -mbcopy-builtin, -mbranch-cheap, -mbranch-expensive.
7912 Remove non-existent option -mabshi, -mno-abshi. Document mutually
7914 * doc/md.texi (PDP-11): Document new D and h constraints. Update
7915 description of O constraint.
7917 2018-06-27 Jeff Law <law@redhat.com>
7918 Austin Law <austinklaw@gmail.com>
7920 * config/v850/v850.md (addsi3_set_flags): New pattern.
7921 (subsi3_set_flags, negsi2_set_flags, andsi3_set_flags): Likewise.
7922 (iorsi3_set_flags, xorsi3_set_flags, one_cmplsi2_set_flags): Likewise.
7923 (zero_extendhisi2_v850_set_flags): Likewise.
7924 (zero_extendqisi2_v850_set_flags): Likewise.
7925 (ashlsi3_set_flags, ashlsi3_v850e2_set_flags): Likewise.
7926 (lshrsi3_set_flags, lshrsi3_v850e2_set_flags): Likewise.
7927 (ashrsi3_set_flags, ashrsi3_v850e2_set_flags): Likewise.
7929 * config/v850/v850-protos.h (notice_update_cc): Remove.
7930 * config/v850/v850.c (v850_compare_op0, v850_compare_op1): Remove.
7931 (v850_print_operand): Handle 'D' and "d".
7932 (v850_select_cc_mode): Remove ATTRIBUTE_UNUSED for last argument.
7933 Add handling of arithmetic/logical operations compared against zero.
7934 (v850_gen_float_compare): Remove ATTRIBUTE_UNUSED for last argument.
7935 Do not look at v850_compare_op, instead get mode from last argument.
7936 (v850_gen_compare): Remove
7937 (increment_stack): Use addsi3_clobber_flags to avoid splitting failure
7938 after reload for prologue insns.
7939 (expand_prologue): Account for CLOBBER of CC_REGNUM in various
7941 (construct_save_jarl): Likewise.
7942 (TARGET_FLAGS_REGNUM): Define.
7943 * config/v850/v850.h (v850_compare_op0, v850_compare_op1): Remove.
7944 (NOTICE_UPDATE_CC): Remove.
7945 * config/v850/v850.md (v850_tst1): Use (reg:CCZ CC_REGNUM) rather
7946 than cc0. Conditionalize on reload_completed.
7947 (cmpsi_insn, setfcc_insn): Likewise.
7948 (tst1 splitter): Turn into define_and_split which sets the flags
7950 (cstoresi4, cbranchsf4, cbranchdf4, cbranchsi4_insn): Likewise.
7951 (cbranchsi4, branch_normal, branch_invert): Do not expose cc0 here.
7952 (cstoresf4, cstoredf4): Clobber the flags.
7953 (cmpsi, cmpsf, cmpdf): Remove expanders.
7954 (setf_insn): Remove pattern.
7955 (addsi3): Turn into define_and_split which clobbers the flags after
7956 reload and a suitable pattern (addsi3_clobber_flags) for use after
7958 (subsi3, negsi2, andsi3, iorsi3, xorsi3, one_cmplsi2) Likewise.
7959 (ashlsi3, ashlsi3_v850e2, lshrsi3, lsh4si3_v850e2): Likewise.
7960 (ashrsi3, ashrsi3_v850e2): Likewise.
7961 (bins): Clobber the flags.
7962 (movsicc_normal_cc, movsicc_normal, movsicc_tst1): Likewise.
7963 (movsicc_tst1_revesed, sasf, swap and rotate patterns): Likewise.
7964 (fix_loop_counter, call_internal_short, call_internal_long): Likewise.
7965 (call_value_internal_short, call_value_internal_long): Likewise.
7966 (callt_save_interrupt, callt_return_interrupt): Likewise.
7967 (save_interrupt, return_interrupt): Likewise.
7968 (callt_save_all_interrupt, save_all_interrupt): Likewise.
7969 (_save_all_interrupt, callt_restore_all_interrupt): Likewise.
7970 (restore_all_interrupt, _restore_all_interrupt): Likewise.
7971 (All FP comparisons): Only allow after reload has completed.
7973 (divh, divhu): Tweak output template.
7974 (branch_z_normal, branch_z_invert): Remove
7975 (branch_nz_normal, branch_nz_invert): Likewise.
7976 (extendhisi_insn, extendqisi_insn): Do not clobber flags.
7978 * config/v850/v850-modes.def (CCZ, CCNZ): Add new modes.
7979 * config/v850/v850.c (notice_update_cc): Remove.
7980 * config/v850/v850.h (CC_OVERFLOW_UNUSABLE): Remove
7981 (CC_NO_CARRY): Likewise.
7982 (NOTICE_UPDATE_CC): Define to nothing.
7983 * config/v850/v850.md: Remove block comment on cc0 handling
7984 Remove "cc" attribute from all patterns. Remove cc_status handling
7985 from all patterns. Minor formatting fixes.
7987 2018-06-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7989 * config/aarch64/aarch64-cores.def (cortex-a76): New entry.
7990 (cortex-a76.cortex-a55): Likewise.
7991 * config/aarch64/aarch64-tune.md: Regenerate.
7992 * doc/invoke.texi (AArch64 Options): Document cortex-a76 and
7993 cortex-a76.cortex-a55.
7995 2018-06-27 Jeff Law <law@redhat.com>
7997 * config/v850/t-v850 (MULTILIB_OPTIONS): Remove 8byte-align.
7998 (MULTILIB_DIRNAMES): Similarly.
8000 2018-06-27 Eric Botcazou <ebotcazou@adacore.com>
8002 * gimple.h (gimple_return_retbnd): Delete.
8003 (gimple_return_set_retbnd): Likewise.
8004 * cgraphunit.c (cgraph_node::expand_thunk): Remove call to
8005 gimple_return_set_retbnd.
8006 * gimple-pretty-print.c (dump_gimple_return): Remove call to
8007 gimple_return_retbnd and adjust.
8008 * tree-inline.h (struct copy_body_data): Remove retbnd field.
8009 * tree-inline.c (remap_gimple_stmt): Remove handling of retbnd.
8010 Explicitly return NULL in a couple more cases. Move assertion
8011 on debug statements and remove unreachable code.
8012 (reset_debug_binding): Do not test id->retbnd.
8013 (expand_call_inline): Do not set it.
8015 2018-06-27 Rasmus Villemoes <rv@rasmusvillemoes.dk>
8017 * configure.ac: Add --disable-gcov option.
8018 * configure: Regenerate.
8019 * Makefile.in: Honour @enable_gcov@.
8020 * doc/install.texi: Document --disable-gcov.
8022 2018-06-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8024 * config/arm/arm-cpus.in (cortex-a76): New entry.
8025 (cortex-a76.cortex-a55): Likewise.
8026 * config/arm/arm-tables.opt: Regenerate.
8027 * config/arm/arm-tune.md: Likewise.
8028 * config/arm/driver-arm.c (arm_cpu_table): Add Cortex-A76 entry.
8029 * doc/invoke.texi (ARM Options): Document cortex-a76 and
8030 cortex-a76.cortex-a55.
8032 2018-06-27 Tamar Christina <tamar.christina@arm.com>
8035 * config/aarch64/aarch64.md (*movhf_aarch64): Add dup v0.4h pattern.
8037 2018-06-27 Siddhesh Poyarekar <siddhesh@sourceware.org>
8039 * config/aarch64/aarch64.h (CALL_USE_REGISTERS): Fix obsolete
8041 (EPILOGUE_USES): Likewise.
8043 2018-06-26 Eric Botcazou <ebotcazou@adacore.com>
8045 * tree-inline.c (remap_location): New function extracted from...
8046 (copy_edges_for_bb): Add ID parameter. Remap goto_locus.
8047 (copy_phis_for_bb): ...here. Call remap_location.
8048 (copy_cfg_body): Adjust call to copy_edges_for_bb.
8050 2018-06-26 Aaron Sawdey <acsawdey@linux.ibm.com>
8052 * config/rs6000/rs6000-string.c (expand_block_clear): Don't use
8053 unaligned vsx for 16B memset.
8055 2018-06-26 Segher Boessenkool <segher@kernel.crashing.org>
8058 * config/rs6000/rs6000.c (rs6000_init_builtins): Do not set
8059 ieee128_float_type_node to long_double_type_node unless
8060 TARGET_LONG_DOUBLE_128 is set.
8062 2018-06-26 David Malcolm <dmalcolm@redhat.com>
8064 * cfgloop.c (get_loop_location): Convert return type from
8065 location_t to dump_user_location_t, replacing INSN_LOCATION lookups
8066 by implicit construction from rtx_insn *, and using
8067 dump_user_location_t::from_function_decl for the fallback case.
8068 * cfgloop.h (get_loop_location): Convert return type from
8069 location_t to dump_user_location_t.
8070 * cgraphunit.c (walk_polymorphic_call_targets): Update call to
8071 dump_printf_loc to pass in a dump_location_t rather than a
8072 location_t, via the gimple stmt.
8073 * coverage.c (get_coverage_counts): Update calls to
8074 dump_printf_loc to pass in dump_location_t rather than a
8076 * doc/optinfo.texi (Dump types): Convert example of
8077 dump_printf_loc from taking "locus" to taking "insn". Update
8078 description of the "_loc" calls to cover dump_location_t.
8079 * dumpfile.c: Include "backend.h", "gimple.h", "rtl.h", and
8081 (dump_user_location_t::dump_user_location_t): New constructors,
8082 from gimple *stmt and rtx_insn *.
8083 (dump_user_location_t::from_function_decl): New function.
8084 (dump_loc): Make static.
8085 (dump_gimple_stmt_loc): Convert param "loc" from location_t to
8086 const dump_location_t &.
8087 (dump_generic_expr_loc): Delete.
8088 (dump_printf_loc): Convert param "loc" from location_t to
8089 const dump_location_t &.
8090 (selftest::test_impl_location): New function.
8091 (selftest::dumpfile_c_tests): New function.
8092 * dumpfile.h: Include "profile-count.h".
8093 (class dump_user_location_t): New class.
8094 (struct dump_impl_location_t): New struct.
8095 (class dump_location_t): New class.
8096 (dump_printf_loc): Convert 2nd param from source_location to
8097 const dump_location_t &.
8098 (dump_generic_expr_loc): Delete.
8099 (dump_gimple_stmt_loc): Convert 2nd param from source_location to
8100 const dump_location_t &.
8101 * gimple-fold.c (fold_gimple_assign): Update call to
8102 dump_printf_loc to pass in a dump_location_t rather than a
8103 location_t, via the gimple stmt.
8104 (gimple_fold_call): Likewise.
8105 * gimple-loop-interchange.cc
8106 (loop_cand::analyze_iloop_reduction_var): Update for change to
8107 check_reduction_path.
8108 (tree_loop_interchange::interchange): Update for change to
8110 * graphite-isl-ast-to-gimple.c (scop_to_isl_ast): Update for
8111 change in return-type of find_loop_location.
8112 (graphite_regenerate_ast_isl): Likewise.
8113 * graphite-optimize-isl.c (optimize_isl): Likewise.
8114 * graphite.c (graphite_transform_loops): Likewise.
8115 * ipa-devirt.c (ipa_devirt): Update call to dump_printf_loc to
8116 pass in a dump_location_t rather than a location_t, via the
8118 * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
8119 * ipa.c (walk_polymorphic_call_targets): Likewise.
8120 * loop-unroll.c (report_unroll): Convert "locus" param from
8121 location_t to dump_location_t.
8122 (decide_unrolling): Update for change to get_loop_location's
8124 * omp-grid.c (struct grid_prop): Convert field "target_loc" from
8125 location_t to dump_user_location_t.
8126 (grid_find_single_omp_among_assignments_1): Updates calls to
8127 dump_printf_loc to pass in a dump_location_t rather than a
8128 location_t, via the gimple stmt.
8129 (grid_parallel_clauses_gridifiable): Convert "tloc" from
8130 location_t to dump_location_t. Updates calls to dump_printf_loc
8131 to pass in a dump_location_t rather than a location_t, via the
8133 (grid_inner_loop_gridifiable_p): Likewise.
8134 (grid_dist_follows_simple_pattern): Likewise.
8135 (grid_gfor_follows_tiling_pattern): Likewise.
8136 (grid_target_follows_gridifiable_pattern): Likewise.
8137 (grid_attempt_target_gridification): Convert initialization
8138 of local "grid" from memset to zero-initialization; FIXME: does
8139 this require C++11? Update call to dump_printf_loc to pass in a
8140 optinfo_location rather than a location_t, via the gimple stmt.
8141 * profile.c (read_profile_edge_counts): Updates call to
8142 dump_printf_loc to pass in a dump_location_t rather than a
8144 (compute_branch_probabilities): Likewise.
8145 * selftest-run-tests.c (selftest::run_tests): Call
8147 * selftest.h (dumpfile_c_tests): New decl.
8148 * tree-loop-distribution.c (pass_loop_distribution::execute):
8149 Update for change in return type of find_loop_location.
8150 * tree-parloops.c (parallelize_loops): Likewise.
8151 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Convert
8152 "locus" from location_t to dump_user_location_t.
8153 (canonicalize_loop_induction_variables): Likewise.
8154 * tree-ssa-loop-ivopts.c (tree_ssa_iv_optimize_loop): Update
8155 for change in return type of find_loop_location.
8156 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update call
8157 to dump_printf_loc to pass in a dump_location_t rather than a
8158 location_t, via the stmt.
8159 * tree-ssa-sccvn.c (eliminate_dom_walker::before_dom_children):
8161 * tree-vect-loop-manip.c (find_loop_location): Convert return
8162 type from source_location to dump_user_location_t.
8163 (vect_do_peeling): Update for above change.
8164 (vect_loop_versioning): Update for change in type of
8166 * tree-vect-loop.c (check_reduction_path): Convert "loc" param
8167 from location_t to dump_user_location_t.
8168 (vect_estimate_min_profitable_iters): Update for change in type
8170 * tree-vect-slp.c (vect_print_slp_tree): Convert param "loc" from
8171 location_t to dump_location_t.
8172 (vect_slp_bb): Update for change in type of vect_location.
8173 * tree-vectorizer.c (vect_location): Convert from source_location
8174 to dump_user_location_t.
8175 (try_vectorize_loop_1): Update for change in vect_location's type.
8176 (vectorize_loops): Likewise.
8177 (increase_alignment): Likewise.
8178 * tree-vectorizer.h (vect_location): Convert from source_location
8179 to dump_user_location_t.
8180 (find_loop_location): Convert return type from source_location to
8181 dump_user_location_t.
8182 (check_reduction_path): Convert 1st param from location_t to
8183 dump_user_location_t.
8184 * value-prof.c (check_counter): Update call to dump_printf_loc to
8185 pass in a dump_user_location_t rather than a location_t; update
8186 call to error_at for change in type of "locus".
8187 (check_ic_target): Update call to dump_printf_loc to
8188 pass in a dump_user_location_t rather than a location_t, via the
8191 2018-06-26 Robin Dapp <rdapp@linux.vnet.ibm.com>
8193 * config/s390/s390.h (enum processor_flags): Do not use
8194 default tune parameter when -march was specified.
8196 2018-06-26 Jakub Jelinek <jakub@redhat.com>
8199 * config/i386/i386.md (setcc + movzbl to xor + setcc peephole2s):
8200 Check reg_overlap_mentioned_p in addition to reg_set_p with the same
8203 2018-06-26 Richard Biener <rguenther@suse.de>
8205 PR tree-optimization/86287
8207 * tree-vect-loop.c (vect_transform_loop_stmt): Fix read-after-free.
8208 (vect_analyze_loop): Initialize n_stmts.
8210 2018-06-26 Richard Biener <rguenther@suse.de>
8213 * fold-const.c (fold_convertible_p): Pointer extension
8216 2018-06-26 Alexandre Oliva <aoliva@redhat.com>
8219 * dwarf2out.c (loc_list_has_views): Adjust comments.
8220 (dw_loc_list): Split single cross-partition range with
8223 2018-06-25 Jeff Law <law@redhat.com>
8225 * common/config/v850/v850-common.c (TARGET_DEFAULT_TARGET_FLAGS): Turn
8226 on -mbig-switch by default.
8228 * config/v850/predicates.md (const_float_1_operand): Fix match_code
8230 (const_float_0_operand): Remove unused predicate.
8231 * config/v850/v850.md (define_constants): Remove UNSPEC_LOOP.
8232 (define_c_enum unspec): Add LOOP, RCP and RSQRT constants.
8233 (recipsf2): New expander. Original pattern now called
8235 (recipdf2, recipdf2_insn): Similarly.
8236 (rsqrtsf2, rsqrtsf2_insn): Similarly
8237 (rsqrtdf2, rsqrtdf2_insn): Similarly
8239 2018-06-26 Gerald Pfeifer <gerald@pfeifer.com>
8241 * ginclude/stddef.h: Remove an obsolete comment on FreeBSD 5.
8242 Simplify logic for FreeBSD (twice).
8244 2018-06-25 Martin Sebor <msebor@redhat.com>
8246 PR tree-optimization/86204
8247 * tree-ssa-strlen.c (handle_builtin_strlen): Avoid storing
8248 a strnlen result if it's less than the length of the string.
8250 2018-06-25 Martin Sebor <msebor@redhat.com>
8252 PR tree-optimization/85700
8253 * gimple-fold.c (gimple_fold_builtin_strncat): Adjust comment.
8254 * tree-ssa-strlen.c (is_strlen_related_p): Handle integer subtraction.
8255 (maybe_diag_stxncpy_trunc): Distinguish strncat from strncpy.
8257 2018-06-25 Martin Sebor <msebor@redhat.com>
8259 * doc/extend.texi (Zero-length arrays): Update and clarify.
8261 2018-06-25 Michael Meissner <meissner@linux.ibm.com>
8263 * config.gcc (powerpc64le*): Revert January 16th, 2018 patch that
8264 added IEEE/IBM long double multilib support on PowerPC little
8265 endian Linux systems.
8266 * config/rs6000/linux64.h (MULTILIB_DEFAULTS_IEEE): Likewise.
8267 (MULTILIB_DEFAULTS): Likewise.
8268 * config/rs6000/rs6000.c (rs6000_option_override_internal):
8270 * config/rs6000/rs6000.h (TARGET_IEEEQUAD_MULTILIB): Likewise.
8271 * config/rs6000/t-ldouble-linux64le-ibm: Delete, no longer used.
8272 * config/rs6000/t-ldouble-linux64le-ieee: Delete, no longer used.
8274 2018-06-25 Alexander Monakov <amonakov@ispras.ru>
8277 * sort.cc (REORDER_23): Avoid memcpy with same destination and source.
8278 (REORDER_45): Likewise.
8280 2018-06-25 Jeff Law <law@redhat.com>
8282 * config/v850/v850.md (divmodhi4): Make sure to sign extend the
8283 dividend to 32 bits. Adjust length.
8284 (udivmodhi4): Cleanup output template. Fix length.
8286 2018-06-25 Carl Love <cel@us.ibm.com>
8288 * config/rs6000/vsx.md: Change word selector to prefered location.
8290 2018-06-25 Richard Biener <rguenther@suse.de>
8292 PR tree-optimization/86304
8293 * tree-vectorizer.c (vectorize_loops): Walk over new possibly
8294 epilogue-if-converted loops as well.
8296 2018-06-25 Jan Hubicka <hubicka@ucw.cz>
8298 * lto-section-out.c (lto_begin_section): Do not print section
8299 name for noaddr and unnumbered dumps.
8301 2018-06-25 Richard Biener <rguenther@suse.de>
8303 * tree-vectorizer.h (struct vec_info_shared): New structure
8304 with parts split out from struct vec_info and loop_nest from
8305 struct _loop_vec_info.
8306 (struct vec_info): Adjust accordingly.
8307 (struct _loop_vec_info): Likewise.
8308 (LOOP_VINFO_LOOP_NEST): Adjust.
8309 (LOOP_VINFO_DATAREFS): Likewise.
8310 (LOOP_VINFO_DDRS): Likewise.
8311 (struct _bb_vec_info): Likewise.
8312 (BB_VINFO_DATAREFS): Likewise.
8313 (BB_VINFO_DDRS): Likewise.
8314 (struct _stmt_vec_info): Add dr_aux member.
8315 (DR_VECT_AUX): Adjust to refer to member of DR_STMTs vinfo.
8316 (DR_MISALIGNMENT_UNINITIALIZED): New.
8317 (set_dr_misalignment): Adjust.
8318 (dr_misalignment): Assert misalign isn't DR_MISALIGNMENT_UNINITIALIZED.
8319 (vect_analyze_loop): Adjust prototype.
8320 (vect_analyze_loop_form): Likewise.
8321 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences):
8322 Compute dependences lazily.
8323 (vect_record_base_alignments): Use shared datarefs/ddrs.
8324 (vect_verify_datarefs_alignment): Likewise.
8325 (vect_analyze_data_refs_alignment): Likewise.
8326 (vect_analyze_data_ref_accesses): Likewise.
8327 (vect_analyze_data_refs): Likewise.
8328 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Add
8329 constructor parameter for shared part.
8330 (vect_analyze_loop_form): Pass in shared part and adjust.
8331 (vect_analyze_loop_2): Pass in storage for the number of
8332 stmts. Move loop nest finding to the caller. Compute
8334 (vect_analyze_loop): Pass in shared part.
8335 (vect_transform_loop): Verify shared datarefs are unchanged.
8336 * tree-vect-slp.c (_bb_vec_info::_bb_vec_info): Add
8337 constructor parameter for shared part.
8338 (vect_slp_analyze_bb_1): Pass in shared part and adjust.
8339 (vect_slp_bb): Verify shared datarefs are unchanged before
8341 * tree-vect-stmts.c (ensure_base_align): Adjust for DR_AUX
8343 (new_stmt_vec_info): Initialize DR_AUX misalignment to
8344 DR_MISALIGNMENT_UNINITIALIZED.
8345 * tree-vectorizer.c (vec_info::vec_info): Add constructor
8346 parameter for shared part.
8347 (vec_info::~vec_info): Adjust.
8348 (vec_info_shared::vec_info_shared): New.
8349 (vec_info_shared::~vec_info_shared): Likewise.
8350 (vec_info_shared::save_datarefs): Likewise.
8351 (vec_info_shared::check_datarefs): Likewise.
8352 (try_vectorize_loop_1): Construct shared part live for analyses
8353 of a single loop for multiple vector sizes.
8354 * tree-parloops.c (gather_scalar_reductions): Adjust.
8356 2018-06-25 Richard Biener <rguenther@suse.de>
8358 * tree-vect-data-refs.c (vect_find_stmt_data_reference): Modify
8359 DR for SIMD lane accesses here and mark DR with (void *)-1 aux.
8360 (vect_analyze_data_refs): Remove similar code from here and
8361 simplify accordingly.
8363 2018-06-25 Richard Biener <rguenther@suse.de>
8365 * tree-vect-data-refs.c (vect_check_gather_scatter): Fail
8366 for reverse storage order accesses rather than asserting
8367 they cannot happen here.
8369 2018-06-25 Tom de Vries <tdevries@suse.de>
8372 * config/i386/i386.md (define_insn "*tls_global_dynamic_64_<mode>"):
8373 Use data16 instead of .byte for insn prefix.
8375 2018-06-25 Andreas Krebbel <krebbel@linux.ibm.com>
8378 * parser.c (make_char_string_pack): Pass this literal chars
8379 through cpp_interpret_string.
8380 (cp_parser_userdef_numeric_literal): Check the result of
8381 make_char_string_pack.
8383 2018-06-24 Maya Rashish <coypu@sdf.org>
8385 * ginclude/stddef.h: Simplify conditions around avoiding
8386 re-definition of __size_t.
8388 2018-06-22 Jan Hubicka <hubicka@ucw.cz>
8390 * lto-streamer-out.c (tree_is_indexable): Make LABEL_DECL nonindexable
8391 unless it is forced or nonlocal; assert that we stream no IMPORTED_DECL.
8393 2018-06-22 Maya Rashish <coypu@sdf.org>
8395 * doc/invoke.texi (mno-fancy-math-387): Update for changes
8396 made to OpenBSD and NetBSD through the years.
8398 2018-06-22 Kelvin Nilsen <kelvin@gcc.gnu.org>
8400 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Change
8401 behavior of vec_pack (vector double, vector double) to match
8402 behavior of vec_float2 (vector double, vector double).
8404 2018-06-22 Olivier Hainque <hainque@adacore.com>
8406 * gimplify.c (gimplify_function_tree): Prevent creation
8407 of a trampoline for the address of the current function
8408 passed to entry/exit instrumentation hooks.
8410 2018-06-22 Aaron Sawdey <acsawdey@linux.ibm.com>
8413 * config/rs6000/rs6000-string.c (expand_strn_compare): Handle -m32
8416 2018-06-22 Martin Liska <mliska@suse.cz>
8418 PR tree-optimization/86263
8419 * tree-switch-conversion.c (jump_table_cluster::find_jump_tables):
8420 Bail out if is_enabled is false.
8421 * tree-switch-conversion.h (jump_table_cluster::is_enabled):
8423 (jump_table_cluster::is_enabled): New function.
8425 2018-06-22 Jan Hubicka <hubicka@ucw.cz>
8427 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not stream
8428 BINFO_BASE_ACCESSES and BINFO_VPTR_FIELD.
8429 * tree-streamer-in.c (streamer_read_tree_bitfields): Likewise.
8430 (lto_input_ts_binfo_tree_pointers): Likewise.
8431 * tree-streamer-out.c (streamer_write_tree_bitfields,
8432 write_ts_binfo_tree_pointers): Likewise.
8433 * tree.c (free_lang_data_in_binfo): Clear BINFO_VPTR_FIELD.
8435 2018-06-22 Jan Hubicka <hubicka@ucw.cz>
8437 * tree.c (free_lang_data_in_type): Free all TYPE_VFIELDs.
8439 2018-06-22 Martin Liska <mliska@suse.cz>
8441 * symbol-summary.h (get): Make it pure and inline move
8442 functionality from ::get function.
8443 (get): Remove and inline into ::get and ::get_create.
8444 (get_create): Move code from ::get function.
8446 2018-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
8449 * config/i386/sol2.h (CPP_SPEC): Don't pass -P for
8450 -x assembler-with-cpp.
8452 2018-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
8454 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define
8455 _FILE_OFFSET_BITS=64 for C++.
8457 2018-06-21 Michael Meissner <meissner@linux.ibm.com>
8459 * config/rs6000/rs6000.md (extendtfif2): Add missing 128-bit
8460 conversion insn that shows up when pr85657-3.c is compiled using
8461 IEEE 128-bit long double.
8462 (neg<mode>2_internal): Use the correct mode to check whether the
8463 mode is IBM extended.
8464 * config/rs6000/rs6000.c (init_float128_ieee): Prevent complex
8465 multiply and divide external functions from being created more
8468 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
8470 * cfgrtl.c (fixup_reorder_chain): Do not emit NOPs in DECL_IGNORED_P
8472 (rtl_merge_blocks): Likewise. Do not emit a NOP if the location of
8473 the edge can be forwarded.
8474 (cfg_layout_merge_blocks): Likewise.
8476 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
8478 * except.c (finish_eh_generation): Commit edge insertions only after
8479 the EH edges have been redirected from post-landing to landing pads.
8481 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
8483 * tree-nested.c (get_frame_type): Use create_tmp_var_raw instead of
8484 create_tmp_var_for to create the FRAME decl.
8485 (finalize_nesting_tree_1): Do not unchain the FRAME decl.
8487 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
8489 * tree-inline.c (copy_edges_for_bb): Minor tweak.
8490 (maybe_move_debug_stmts_to_successors): Also reset the locus of the
8491 debug statement when resetting its value.
8492 (expand_call_inline): Copy the locus of the call onto the assignment
8493 of the return value, if any. Use local variable in more cases.
8495 2018-06-21 Martin Liska <mliska@suse.cz>
8497 * ipa-pure-const.c (propagate_nothrow): Use
8498 funct_state_summaries->get.
8499 (dump_malloc_lattice): Likewise.
8500 (propagate_malloc): Likewise.
8502 2018-06-21 Richard Biener <rguenther@suse.de>
8504 * lto-streamer-out.c (DFS::DFS_write_tree_body): Update outdated
8505 comment. Follow BLOCK_ABSTRACT_ORIGIN unconditionally.
8506 * tree-streamer-in.c (lto_input_ts_block_tree_pointers): Update
8508 * tree-streamer-out.c (write_ts_block_tree_pointers): Stream
8509 BLOCK_ABSTRACT_ORIGIN unconditionally.
8511 2018-06-21 David Malcolm <dmalcolm@redhat.com>
8513 * ipa-cp.c (ipcp_driver): Set edge_clone_summaries to NULL after
8515 * ipa-reference.c (ipa_reference_c_finalize): Delete
8516 ipa_ref_opt_sum_summaries and set it to NULL.
8518 2018-06-21 Tom de Vries <tdevries@suse.de>
8520 PR tree-optimization/85859
8521 * tree-ssa-tail-merge.c (stmt_local_def): Copy gimple_is_call
8522 test with comment from bb_no_side_effects_p.
8524 2018-06-21 Richard Biener <rguenther@suse.de>
8526 PR tree-optimization/86232
8527 * tree-ssa-loop-niter.c (number_of_iterations_popcount): Adjust
8528 max for constant niter.
8530 2018-06-21 Andre Vieira <andre.simoesdiasvieira@arm.com>
8532 * config/aarch64/aarch64-simd.md
8533 (*aarch64_crypto_aes<aes_op>v16qi_xor_combine): New.
8535 2018-06-21 Andre Vieira <andre.simoesdiasvieira@arm.com>
8537 * config/aarch64/aarch64-simd.md (aarch64_crypto_aes<aes_op>v16qi):
8538 Make opernads of the unspec commutative.
8540 2018-06-21 Richard Biener <rguenther@suse.de>
8542 * tree-data-ref.c (dr_step_indicator): Handle NULL DR_STEP.
8543 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
8544 Avoid calling vect_mark_for_runtime_alias_test with gathers or scatters.
8545 (vect_analyze_data_ref_dependence): Re-order checks to deal with
8547 (vect_record_base_alignments): Do not record base alignment
8548 for gathers or scatters.
8549 (vect_compute_data_ref_alignment): Drop return value that is always
8550 true. Bail out early for gathers or scatters.
8551 (vect_enhance_data_refs_alignment): Bail out early for gathers
8553 (vect_find_same_alignment_drs): Likewise.
8554 (vect_analyze_data_refs_alignment): Remove dead code.
8555 (vect_slp_analyze_and_verify_node_alignment): Likewise.
8556 (vect_analyze_data_refs): For possible gathers or scatters do
8557 not create an alternate DR, just check their possible validity
8558 and mark them. Adjust DECL_NONALIASED handling to not rely
8560 * tree-vect-loop-manip.c (vect_update_inits_of_drs): Do not
8561 update inits of gathers or scatters.
8562 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern):
8563 Also copy gather/scatter flag to pattern vinfo.
8565 2018-06-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
8567 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Change
8568 behavior of vec_packsu (vector unsigned long long, vector unsigned
8569 long long) to match behavior of vec_packs with same signature.
8571 2018-06-20 Chung-Lin Tang <cltang@codesourcery.com>
8572 Thomas Schwinge <thomas@codesourcery.com>
8573 Cesar Philippidis <cesar@codesourcery.com>
8575 * gimplify.c (gimplify_scan_omp_clauses): Add support for
8576 OMP_CLAUSE_{IF_PRESENT,FINALIZE}.
8577 (gimplify_adjust_omp_clauses): Likewise.
8578 (gimplify_oacc_declare_1): Add support for GOMP_MAP_RELEASE, remove
8579 support for GOMP_MAP_FORCE_{ALLOC,TO,FROM,TOFROM}.
8580 (gimplify_omp_target_update): Update handling of acc update and
8582 * omp-low.c (install_var_field): Remove unused parameter
8583 base_pointers_restrict.
8584 (scan_sharing_clauses): Remove base_pointers_restrict parameter.
8585 Update call to install_var_field. Handle OMP_CLAUSE_{IF_PRESENT,
8587 (omp_target_base_pointers_restrict_p): Delete.
8588 (scan_omp_target): Update call to scan_sharing_clauses.
8589 * tree-core.h (enum omp_clause_code): Add OMP_CLAUSE_{IF_PRESENT,
8591 * tree-nested.c (convert_nonlocal_omp_clauses): Handle
8592 OMP_CLAUSE_{IF_PRESENT,FINALIZE}.
8593 (convert_local_omp_clauses): Likewise.
8594 * tree-pretty-print.c (dump_omp_clause): Likewise.
8595 * tree.c (omp_clause_num_ops): Add entries for OMP_CLAUSE_{IF_PRESENT,
8597 (omp_clause_code_name): Likewise.
8599 2018-06-20 Jakub Jelinek <jakub@redhat.com>
8602 * var-tracking.c (use_narrower_mode_test): Check if shift amount can
8605 PR tree-optimization/86231
8606 * tree-vrp.c (union_ranges): For ( [ ) ] or ( )[ ] range and
8607 anti-range don't overwrite *vr0min before using it to compute *vr0max.
8609 2018-06-20 Tom de Vries <tdevries@suse.de>
8611 PR tree-optimization/86097
8612 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Also convert *nit to
8613 iv type if signedness of iv type is not the same as that of *nit.
8615 2018-06-20 Jakub Jelinek <jakub@redhat.com>
8617 * cfgrtl.c (rtl_verify_edges): Formatting fix. If bb->preds has any
8618 EDGE_EH edges, verify they are all EDGE_EH.
8620 2018-06-20 Maya Rashish <coypu@sdf.org>
8622 * ginclude/stddef.h: Limit #include <machine/ansi.h> to NetBSD.
8624 2018-06-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8626 * config/aarch64/aarch64-tuning-flags.def (no_ldp_stp_qregs): New.
8627 * config/aarch64/aarch64.c (xgene1_tunings): Add
8628 AARCH64_EXTRA_TUNE_NO_LDP_STP_QREGS to tune_flags.
8629 (aarch64_mode_valid_for_sched_fusion_p):
8630 Allow 16-byte modes.
8631 (aarch64_classify_address): Allow 16-byte modes for load_store_pair_p.
8632 * config/aarch64/aarch64-ldpstp.md: Add peepholes for LDP STP of
8634 * config/aarch64/aarch64-simd.md (load_pair<VQ:mode><VQ2:mode>):
8636 (vec_store_pair<VQ:mode><VQ2:mode>): Likewise.
8637 * config/aarch64/iterators.md (VQ2): New mode iterator.
8639 2018-06-20 Martin Liska <mliska@suse.cz>
8641 * tree-switch-conversion.c (jump_table_cluster::can_be_handled):
8642 Change default ratio from 10 to 8.
8644 2018-06-20 Martin Liska <mliska@suse.cz>
8646 * tree-switch-conversion.c (jump_table_cluster::find_jump_tables):
8648 (bit_test_cluster::find_bit_tests): Likewise.
8649 (switch_decision_tree::analyze_switch_statement): Find clusters.
8650 * tree-switch-conversion.h (struct jump_table_cluster): Document
8653 2018-06-20 Martin Liska <mliska@suse.cz>
8655 * tree-switch-conversion.c (switch_conversion::collect):
8656 Record m_uniq property.
8657 (switch_conversion::expand): Bail out for special conditions.
8658 (group_cluster::~group_cluster): New.
8659 (group_cluster::group_cluster): Likewise.
8660 (group_cluster::dump): Likewise.
8661 (jump_table_cluster::emit): New.
8662 (switch_decision_tree::fix_phi_operands_for_edges): New.
8663 (struct case_node): Remove struct.
8664 (jump_table_cluster::can_be_handled): New.
8665 (case_values_threshold): Moved to header.
8666 (reset_out_edges_aux): Likewise.
8667 (jump_table_cluster::is_beneficial): New.
8668 (bit_test_cluster::can_be_handled): Likewise.
8669 (add_case_node): Remove.
8670 (bit_test_cluster::is_beneficial): New.
8671 (case_bit_test::cmp): New.
8672 (bit_test_cluster::emit): New.
8673 (expand_switch_as_decision_tree_p): Remove.
8674 (bit_test_cluster::hoist_edge_and_branch_if_true): New.
8675 (fix_phi_operands_for_edge): Likewise.
8676 (switch_decision_tree::analyze_switch_statement): New.
8677 (compute_cases_per_edge): Move ...
8678 (switch_decision_tree::compute_cases_per_edge): ... here.
8679 (try_switch_expansion): Likewise.
8680 (switch_decision_tree::try_switch_expansion): Likewise.
8681 (record_phi_operand_mapping): Likewise.
8682 (switch_decision_tree::record_phi_operand_mapping): Likewise.
8683 (emit_case_decision_tree): Likewise.
8684 (switch_decision_tree::emit): Likewise.
8685 (balance_case_nodes): Likewise.
8686 (switch_decision_tree::balance_case_nodes): Likewise.
8687 (dump_case_nodes): Likewise.
8688 (switch_decision_tree::dump_case_nodes): Likewise.
8689 (emit_jump): Likewise.
8690 (switch_decision_tree::emit_jump): Likewise.
8691 (emit_cmp_and_jump_insns): Likewise.
8692 (switch_decision_tree::emit_cmp_and_jump_insns): Likewise.
8693 (emit_case_nodes): Likewise.
8694 (switch_decision_tree::emit_case_nodes): Likewise.
8695 (conditional_probability): Remove.
8696 * tree-switch-conversion.h (enum cluster_type): New.
8698 (struct cluster): Likewise.
8699 (cluster::cluster): Likewise.
8700 (struct simple_cluster): Likewise.
8701 (simple_cluster::simple_cluster): Likewise.
8702 (struct group_cluster): Likewise.
8703 (struct jump_table_cluster): Likewise.
8704 (struct bit_test_cluster): Likewise.
8705 (struct min_cluster_item): Likewise.
8706 (struct case_tree_node): Likewise.
8707 (case_tree_node::case_tree_node): Likewise.
8708 (jump_table_cluster::case_values_threshold): Likewise.
8709 (struct case_bit_test): Likewise.
8710 (struct switch_decision_tree): Likewise.
8711 (struct switch_conversion): Likewise.
8712 (switch_decision_tree::reset_out_edges_aux): Likewise.
8714 2018-06-20 Martin Liska <mliska@suse.cz>
8716 * tree-switch-conversion.c (MAX_CASE_BIT_TESTS): Remove.
8717 (hoist_edge_and_branch_if_true): Likewise.
8718 (expand_switch_using_bit_tests_p): Likewise.
8719 (struct case_bit_test): Likewise.
8720 (case_bit_test_cmp): Likewise.
8721 (emit_case_bit_tests): Likewise.
8722 (switch_conversion::switch_conversion): New class.
8723 (struct switch_conv_info): Remove old struct.
8724 (collect_switch_conv_info): More to ...
8725 (switch_conversion::collect): ... this.
8726 (check_range): Likewise.
8727 (switch_conversion::check_range): Likewise.
8728 (check_all_empty_except_final): Likewise.
8729 (switch_conversion::check_all_empty_except_final): Likewise.
8730 (check_final_bb): Likewise.
8731 (switch_conversion::check_final_bb): Likewise.
8732 (create_temp_arrays): Likewise.
8733 (switch_conversion::create_temp_arrays): Likewise.
8734 (free_temp_arrays): Likewise.
8735 (gather_default_values): Likewise.
8736 (switch_conversion::gather_default_values): Likewise.
8737 (build_constructors): Likewise.
8738 (switch_conversion::build_constructors): Likewise.
8739 (constructor_contains_same_values_p): Likewise.
8740 (switch_conversion::contains_same_values_p): Likewise.
8741 (array_value_type): Likewise.
8742 (switch_conversion::array_value_type): Likewise.
8743 (build_one_array): Likewise.
8744 (switch_conversion::build_one_array): Likewise.
8745 (build_arrays): Likewise.
8746 (switch_conversion::build_arrays): Likewise.
8747 (gen_def_assigns): Likewise.
8748 (switch_conversion::gen_def_assigns): Likewise.
8749 (prune_bbs): Likewise.
8750 (switch_conversion::prune_bbs): Likewise.
8751 (fix_phi_nodes): Likewise.
8752 (switch_conversion::fix_phi_nodes): Likewise.
8753 (gen_inbound_check): Likewise.
8754 (switch_conversion::gen_inbound_check): Likewise.
8755 (process_switch): Use the newly created class.
8756 (switch_conversion::expand): New.
8757 (switch_conversion::~switch_conversion): New.
8758 * tree-switch-conversion.h: New file.
8760 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
8762 * tree-vectorizer.h (NUM_PATTERNS, vect_recog_func_ptr): Move to
8763 tree-vect-patterns.c.
8764 * tree-vect-patterns.c (vect_supportable_direct_optab_p): New function.
8765 (vect_recog_dot_prod_pattern): Use it. Remove the type_in argument.
8766 (vect_recog_sad_pattern): Likewise.
8767 (vect_recog_widen_sum_pattern): Likewise.
8768 (vect_recog_pow_pattern): Likewise. Check for a null vectype.
8769 (vect_recog_widen_shift_pattern): Remove the type_in argument.
8770 (vect_recog_rotate_pattern): Likewise.
8771 (vect_recog_mult_pattern): Likewise.
8772 (vect_recog_vector_vector_shift_pattern): Likewise.
8773 (vect_recog_divmod_pattern): Likewise.
8774 (vect_recog_mixed_size_cond_pattern): Likewise.
8775 (vect_recog_bool_pattern): Likewise.
8776 (vect_recog_mask_conversion_pattern): Likewise.
8777 (vect_try_gather_scatter_pattern): Likewise.
8778 (vect_recog_widen_mult_pattern): Likewise. Check for a null vectype.
8779 (vect_recog_over_widening_pattern): Likewise.
8780 (vect_recog_gather_scatter_pattern): Likewise.
8781 (vect_recog_func_ptr): Move from tree-vectorizer.h
8782 (vect_vect_recog_func_ptrs): Move further down the file.
8783 (vect_recog_func): Likewise. Remove the third argument.
8784 (NUM_PATTERNS): Define based on vect_vect_recog_func_ptrs.
8785 (vect_pattern_recog_1): Expect the pattern function to do any
8786 necessary target tests. Also expect it to provide a vector type.
8787 Remove the type_in handling.
8789 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
8791 * tree-vect-patterns.c (vect_pattern_detected): New function.
8792 (vect_recog_dot_prod_patternm, vect_recog_sad_pattern)
8793 (vect_recog_widen_mult_pattern, vect_recog_widen_sum_pattern)
8794 (vect_recog_over_widening_pattern, vect_recog_widen_shift_pattern
8795 (vect_recog_rotate_pattern, vect_recog_vector_vector_shift_pattern)
8796 (vect_recog_mult_pattern, vect_recog_divmod_pattern)
8797 (vect_recog_mixed_size_cond_pattern, vect_recog_bool_pattern)
8798 (vect_recog_mask_conversion_pattern)
8799 (vect_try_gather_scatter_pattern): Likewise.
8801 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
8803 * tree-vect-patterns.c (vect_get_internal_def): New function.
8804 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern)
8805 (vect_recog_vector_vector_shift_pattern, check_bool_pattern)
8806 (search_type_for_mask_1): Use it.
8808 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
8810 * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Remove
8811 redundant WIDEN_SUM_EXPR handling.
8812 (vect_recog_sad_pattern): Likewise.
8814 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
8816 * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Remove
8817 redundant check that the types of a PLUS_EXPR or MULT_EXPR agree.
8818 (vect_recog_sad_pattern): Likewise PLUS_EXPR, ABS_EXPR and MINUS_EXPR.
8819 (vect_recog_widen_mult_pattern): Likewise MULT_EXPR.
8820 (vect_recog_widen_sum_pattern): Likewise PLUS_EXPR.
8822 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
8824 * tree-vect-stmts.c (vectorizable_call): Make sure that we
8825 use the stmt_vec_info of the original bb statement for the
8826 new zero assignment, even if the call is part of a pattern.
8828 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
8830 * tree-vectorizer.h (_stmt_vec_info): Note above pattern_def_seq
8831 that the sequence is attached to the original statement rather
8832 than the pattern statement.
8833 * tree-vect-loop.c (vect_determine_vf_for_stmt): Take the
8834 PATTERN_DEF_SEQ from the original statement rather than
8835 the main pattern statement.
8836 * tree-vect-stmts.c (free_stmt_vec_info): Likewise.
8837 * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Likewise.
8838 (vect_mark_pattern_stmts): Don't copy the PATTERN_DEF_SEQ.
8840 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
8842 * tree-vect-stmts.c (vect_analyze_stmt): Move the handling of pattern
8843 definition statements before the early exit for statements that aren't
8845 * tree-vect-loop.c (vect_transform_loop_stmt): New function,
8847 (vect_transform_loop): ...here. Process pattern definition
8848 statements without first checking whether the main pattern
8849 statement is live or relevant.
8851 2018-06-19 Eric Botcazou <ebotcazou@adacore.com>
8853 * tree-cfgcleanup.c (tree_forwarder_block_p): Do not return false at
8854 -O0 if the locus represent UNKNOWN_LOCATION but have different values.
8856 2018-06-19 Aaron Sawdey <acsawdey@linux.ibm.com>
8858 * config/rs6000/rs6000-string.c (select_block_compare_mode): Check
8859 TARGET_EFFICIENT_OVERLAPPING_UNALIGNED here instead of in caller.
8860 (do_and3, do_and3_mask, do_compb3, do_rotl3): New functions.
8861 (expand_block_compare): Change select_block_compare_mode call.
8862 (expand_strncmp_align_check): Use new functions, fix comment.
8863 (emit_final_str_compare_gpr): New function.
8864 (expand_strn_compare): Refactor and clean up code.
8865 * config/rs6000/vsx.md (vsx_mov<mode>_64bit): Remove *.
8867 2018-06-19 Tony Reix <tony.reix@atos.com>
8868 Damien Bergamini <damien.bergamini@atos.com>
8869 David Edelsohn <dje.gcc@gmail.com>
8871 * collect2.c (static_obj): New variable.
8872 (static_libs): New variable.
8873 (is_in_list): Uncomment declaration.
8874 (main): Track AIX libraries linked statically.
8875 (is_in_list): Uncomment definition.
8876 (scan_prog_file): Don't add AIX shared libraries initializer
8877 to constructor list if linking statically.
8879 2018-06-19 Max Filippov <jcmvbkbc@gmail.com>
8881 * config/xtensa/xtensa.md (UNSPEC_FRAME_BLOCKAGE): New unspec
8883 (allocate_stack, frame_blockage, *frame_blockage): New patterns.
8885 2018-06-19 Jan Hubicka <hubicka@ucw.cz>
8887 * tree.c (find_decls_types_r): Remove all non-VAR_DECLs from
8890 2018-06-19 Martin Liska <mliska@suse.cz>
8892 * config/i386/i386.c (ix86_can_inline_p): Do not use
8893 ipa_fn_summaries::get_create.
8894 * ipa-cp.c (ipcp_cloning_candidate_p): Replace get_create with
8896 (devirtualization_time_bonus): Likewise.
8897 (ipcp_propagate_stage): Likewise.
8898 * ipa-fnsummary.c (redirect_to_unreachable): Likewise.
8899 (edge_set_predicate): Likewise.
8900 (evaluate_conditions_for_known_args): Likewise.
8901 (evaluate_properties_for_edge): Likewise.
8902 (ipa_call_summary::reset): Tranform to ...
8903 (ipa_call_summary::~ipa_call_summary): ... this.
8904 (ipa_fn_summary::reset): Transform to ...
8905 (ipa_fn_summary::~ipa_fn_summary): ... this.
8906 (ipa_fn_summary_t::remove): Rename to ...
8907 (ipa_fn_summary_t::remove_callees): ... this.
8908 (ipa_fn_summary_t::duplicate): Use placement new
8909 instead of memory copy.
8910 (ipa_call_summary_t::duplicate): Likewise.
8911 (ipa_call_summary_t::remove): Remove.
8912 (dump_ipa_call_summary): Change get_create to get.
8913 (ipa_dump_fn_summary): Dump only when summary exists.
8914 (analyze_function_body): Use symbol_summary::get instead
8916 (compute_fn_summary): Likewise.
8917 (estimate_edge_devirt_benefit): Likewise.
8918 (estimate_edge_size_and_time): Likewise.
8919 (inline_update_callee_summaries): Likewise.
8920 (remap_edge_change_prob): Likewise.
8921 (remap_edge_summaries): Likewise.
8922 (ipa_merge_fn_summary_after_inlining): Likewise.
8923 (write_ipa_call_summary): Likewise.
8924 (ipa_fn_summary_write): Likewise.
8925 (ipa_free_fn_summary): Likewise.
8926 * ipa-fnsummary.h (struct GTY): Add new ctor and copy ctor.
8927 (struct ipa_call_summary): Likewise.
8928 * ipa-icf.c (sem_function::merge): Use symbol_summary::get instead
8930 * ipa-inline-analysis.c (do_estimate_edge_time): Likewise.
8931 (estimate_size_after_inlining): Likewise.
8932 (estimate_growth): Likewise.
8933 (growth_likely_positive): Likewise.
8934 * ipa-inline-transform.c (clone_inlined_nodes): Likewise.
8935 (inline_call): Likewise.
8936 * ipa-inline.c (caller_growth_limits): Likewise.
8937 (can_inline_edge_p): Likewise.
8938 (can_inline_edge_by_limits_p): Likewise.
8939 (compute_uninlined_call_time): Likewise.
8940 (compute_inlined_call_time): Likewise.
8941 (want_inline_small_function_p): Likewise.
8942 (edge_badness): Likewise.
8943 (update_caller_keys): Likewise.
8944 (update_callee_keys): Likewise.
8945 (inline_small_functions): Likewise.
8946 (inline_to_all_callers_1): Likewise.
8947 (dump_overall_stats): Likewise.
8948 (early_inline_small_functions): Likewise.
8949 (early_inliner): Likewise.
8950 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
8951 * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
8952 * ipa-pure-const.c (malloc_candidate_p): Likewise.
8953 * ipa-split.c (execute_split_functions): Likewise.
8954 * symbol-summary.h: Likewise.
8955 * tree-sra.c (ipa_sra_preliminary_function_checks): Likewise.
8957 2018-06-19 Richard Biener <rguenther@suse.de>
8959 * tree-vectorizer.c (try_vectorize_loop_1): Split out of ...
8960 (vectorize_loops): ... here. Fix dbgcnt handling.
8961 (try_vectorize_loop): Wrap try_vectorize_loop_1.
8963 2018-06-19 Segher Boessenkool <segher@kernel.crashing.org>
8966 * config/rs6000/rs6000.md (rs6000_discover_homogeneous_aggregate): An
8967 ieee128 argument takes up only one (vector) register, not two (floating
8970 2018-06-19 Eric Botcazou <ebotcazou@adacore.com>
8972 * gimplify.c (gimplify_init_constructor): Really never clear for an
8973 incomplete constructor if CONSTRUCTOR_NO_CLEARING is set.
8975 2018-06-19 Richard Biener <rguenther@suse.de>
8977 PR tree-optimization/86179
8978 * tree-vect-patterns.c (vect_pattern_recog_1): Clean up
8979 after failed recognition.
8981 2018-06-18 Martin Sebor <msebor@redhat.com>
8984 * calls.c (maybe_warn_nonstring_arg): Handle strncat.
8985 * tree-ssa-strlen.c (is_strlen_related_p): Make extern.
8986 Handle integer subtraction.
8987 (maybe_diag_stxncpy_trunc): Handle nonstring source arguments.
8988 * tree-ssa-strlen.h (is_strlen_related_p): Declare.
8990 2018-06-18 David Malcolm <dmalcolm@redhat.com>
8992 * config/frv/frv-protos.h (frv_ifcvt_modify_insn): Strengthen 3rd
8993 param from rtx to rtx_insn *.
8994 * config/frv/frv.c (frv_ifcvt_add_insn): Likewise for "insn"
8996 (frv_ifcvt_modify_insn): Likwise.
8997 (frv_ifcvt_modify_final): Likwise for local "existing_insn",
8998 adding an as_a <rtx_insn *> cast. Likewise for local "insn".
8999 * config/mips/mips.c (r10k_insert_cache_barriers): Add an
9000 as_a <rtx_insn *> cast to local "unprotected_region" once
9001 it's been established that it's not NULL or pc_rtx.
9002 * config/nds32/nds32-relax-opt.c (nds32_group_insns): Strengthen
9003 param "sethi" from rtx to rtx_insn *.
9004 (nds32_group_float_insns): Likewise for param "insn".
9005 * config/vax/vax-protos.h (vax_output_int_add): Likewise for 1st
9007 (vax_output_int_subtract): Likewise.
9008 * config/vax/vax.c (vax_output_int_add): Likewise for param
9010 (vax_output_int_subtract): Likewise.
9011 * emit-rtl.c (set_insn_deleted): Likewise, removing cast.
9012 (emit_pattern_after): Likewise for param "after".
9013 (emit_insn_after): Likewise.
9014 (emit_jump_insn_after): Likewise.
9015 (emit_call_insn_after): Likewise.
9016 (emit_debug_insn_after): Likewise.
9017 (emit_pattern_before): Likewise for param "before".
9018 (emit_insn_before): Likewise.
9019 (emit_jump_insn_before): Likewise.
9020 * final.c (get_insn_template): Likewise for param "insn", removing
9022 * output.h (get_insn_template): Likewise for 2nd param.
9023 * rtl.h (emit_insn_before): Likewise.
9024 (emit_jump_insn_before): Likewise.
9025 (emit_debug_insn_before_noloc): Likewise.
9026 (emit_insn_after): Likewise.
9027 (emit_jump_insn_after): Likewise.
9028 (emit_call_insn_after): Likewise.
9029 (emit_debug_insn_after): Likewise.
9030 (set_insn_deleted): Likewise for param.
9032 2018-06-18 Michael Meissner <meissner@linux.ibm.com>
9035 * config/rs6000/rs6000-modes.def (toplevel): Rework the 128-bit
9036 floating point modes, so that IFmode is numerically greater than
9037 TFmode, which is greater than KFmode using FRACTIONAL_FLOAT_MODE
9038 to declare the ordering. This prevents IFmode from being
9039 converted to TFmode when long double is IEEE 128-bit on an ISA 3.0
9040 machine. Include rs6000-modes.h to share the fractional values
9041 between genmodes* and the rest of the compiler.
9045 * config/rs6000/rs6000-modes.h: New file.
9046 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Change the
9047 meaning of rs6000_long_double_size so that 126..128 selects an
9048 appropriate 128-bit floating point type.
9049 (rs6000_option_override_internal): Likewise.
9050 * config/rs6000/rs6000.h (toplevel): Include rs6000-modes.h.
9051 (TARGET_LONG_DOUBLE_128): Change the meaning of
9052 rs6000_long_double_size so that 126..128 selects an appropriate
9053 128-bit floating point type.
9054 (LONG_DOUBLE_TYPE_SIZE): Update comment.
9055 * config/rs6000/rs6000.md (trunciftf2): Correct the modes of the
9056 source and destination to match the standard usage.
9057 (truncifkf2): Likewise.
9058 (copysign<mode>3, IEEE iterator): Rework copysign of float128 on
9059 ISA 2.07 to use an explicit clobber, instead of passing in a
9061 (copysign<mode>3_soft): Likewise.
9063 2018-06-18 David Malcolm <dmalcolm@redhat.com>
9065 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences):
9066 Replace dump_printf_loc call with DUMP_VECT_SCOPE.
9067 (vect_slp_analyze_instance_dependence): Likewise.
9068 (vect_enhance_data_refs_alignment): Likewise.
9069 (vect_analyze_data_refs_alignment): Likewise.
9070 (vect_slp_analyze_and_verify_instance_alignment
9071 (vect_analyze_data_ref_accesses): Likewise.
9072 (vect_prune_runtime_alias_test_list): Likewise.
9073 (vect_analyze_data_refs): Likewise.
9074 * tree-vect-loop-manip.c (vect_update_inits_of_drs): Likewise.
9075 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
9076 (vect_analyze_scalar_cycles_1): Likewise.
9077 (vect_get_loop_niters): Likewise.
9078 (vect_analyze_loop_form_1): Likewise.
9079 (vect_update_vf_for_slp): Likewise.
9080 (vect_analyze_loop_operations): Likewise.
9081 (vect_analyze_loop): Likewise.
9082 (vectorizable_induction): Likewise.
9083 (vect_transform_loop): Likewise.
9084 * tree-vect-patterns.c (vect_pattern_recog): Likewise.
9085 * tree-vect-slp.c (vect_analyze_slp): Likewise.
9086 (vect_make_slp_decision): Likewise.
9087 (vect_detect_hybrid_slp): Likewise.
9088 (vect_slp_analyze_operations): Likewise.
9089 (vect_slp_bb): Likewise.
9090 * tree-vect-stmts.c (vect_mark_stmts_to_be_vectorized): Likewise.
9091 (vectorizable_bswap): Likewise.
9092 (vectorizable_call): Likewise.
9093 (vectorizable_simd_clone_call): Likewise.
9094 (vectorizable_conversion): Likewise.
9095 (vectorizable_assignment): Likewise.
9096 (vectorizable_shift): Likewise.
9097 (vectorizable_operation): Likewise.
9098 * tree-vectorizer.h (DUMP_VECT_SCOPE): New macro.
9100 2018-06-18 Martin Sebor <msebor@redhat.com>
9102 PR tree-optimization/81384
9103 * builtin-types.def (BT_FN_SIZE_CONST_STRING_SIZE): New.
9104 * builtins.c (expand_builtin_strnlen): New function.
9105 (expand_builtin): Call it.
9106 (fold_builtin_n): Avoid setting TREE_NO_WARNING.
9107 * builtins.def (BUILT_IN_STRNLEN): New.
9108 * calls.c (maybe_warn_nonstring_arg): Handle BUILT_IN_STRNLEN.
9109 Warn for bounds in excess of maximum object size.
9110 * tree-ssa-strlen.c (maybe_set_strlen_range): Return tree representing
9111 single-value ranges. Handle strnlen.
9112 (handle_builtin_strlen): Handle strnlen.
9113 (strlen_check_and_optimize_stmt): Same.
9114 * doc/extend.texi (Other Builtins): Document strnlen.
9116 2018-06-18 Maya Rashish <coypu@sdf.org>
9118 * config/alpha/openbsd.h (TARGET_DEFAULT): Define.
9119 (LINK_SPEC, STARTFILE_SPEC, ENDFILE_SPEC): Likewise.
9120 (INTMAX_TYPE, UINTMAX_TYPE, WINT_TYPE): Likewise.
9122 * config/alpha/elf.h (STARTFILE_SPEC, ENDFILE_SPEC): Move from
9124 * config/alpha/linux.h (STARTFILE_SPEC, ENDFILE_SPEC): Here.
9126 2018-06-18 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
9128 * tree.c (escaped_string::escape): Replace cast to char * by
9129 const_cast<char *> (unescaped).
9131 2018-06-18 Nick Clifton <nickc@redhat.com>
9134 * tree.c (escaped_string): New class. Converts an unescaped
9135 string into its escaped equivalent.
9136 (warn_deprecated_use): Use the new class to convert the
9137 deprecation message, if present.
9138 (test_escaped_strings): New self test.
9139 (test_c_tests): Add test_escaped_strings.
9140 * doc/extend.texi (deprecated): Add a note that the
9141 deprecation message is affected by the -fmessage-length
9142 option, and that control characters will be escaped.
9143 (#pragma GCC error): Document this pragma.
9144 (#pragma GCC warning): Likewise.
9145 * doc/invoke.texi (-fmessage-length): Document this option's
9146 effect on the #warning and #error preprocessor directives and
9147 the deprecated attribute.
9149 2018-06-18 Eric Botcazou <ebotcazou@adacore.com>
9151 * tree.c (decl_value_expr_lookup): Revert latest change.
9152 (decl_value_expr_insert): Likewise.
9154 2018-06-17 Eric Botcazou <ebotcazou@adacore.com>
9156 * gimplify.c (nonlocal_vlas): Delete.
9157 (nonlocal_vla_vars): Likewise.
9158 (gimplify_var_or_parm_decl): Do not add debug VAR_DECLs for non-local
9160 (gimplify_body): Do not create and destroy nonlocal_vlas.
9161 * tree-nested.c: Include diagnostic.h.
9162 (use_pointer_in_frame): Tweak.
9163 (lookup_field_for_decl): Add assertion and declare the transformation.
9164 (convert_nonlocal_reference_op) <PARM_DECL>: Rework and issue an
9165 internal error when the reference is in a wrong context. Do not
9166 create a debug decl by default.
9167 (note_nonlocal_block_vlas): Delete.
9168 (convert_nonlocal_reference_stmt) <GIMPLE_BIND>: Do not call it.
9169 (convert_local_reference_op) <PARM_DECL>: Skip the frame decl. Do not
9170 create a debug decl by default.
9171 (convert_gimple_call) <GIMPLE_CALL>: Issue an internal error when the
9172 call is in a wrong context.
9173 (fixup_vla_decls): New function.
9174 (finalize_nesting_tree_1): Adjust comment. Call fixup_vla_decls if no
9175 debug variables were created.
9176 * tree.c (decl_value_expr_lookup): Add checking assertion.
9177 (decl_value_expr_insert): Likewise.
9179 2018-06-16 Kugan Vivekanandarajah <kuganv@linaro.org>
9182 * ipa-fnsummary.c (will_be_nonconstant_expr_predicate): Handle CALL_EXPR.
9183 * tree-scalar-evolution.c (interpret_expr): Likewise.
9184 (expression_expensive_p): Likewise.
9185 * tree-ssa-loop-ivopts.c (contains_abnormal_ssa_name_p): Likewise.
9186 * tree-ssa-loop-niter.c (number_of_iterations_popcount): New.
9187 (number_of_iterations_exit_assumptions): Use number_of_iterations_popcount.
9188 (ssa_defined_by_minus_one_stmt_p): New.
9190 2018-06-16 Kugan Vivekanandarajah <kuganv@linaro.org>
9193 * cfgexpand.c (expand_debug_expr): Hande ABSU_EXPR.
9194 * config/i386/i386.c (ix86_add_stmt_cost): Likewise.
9195 * dojump.c (do_jump): Likewise.
9196 * expr.c (expand_expr_real_2): Check operand type's sign.
9197 * fold-const.c (const_unop): Handle ABSU_EXPR.
9198 (fold_abs_const): Likewise.
9199 * gimple-pretty-print.c (dump_unary_rhs): Likewise.
9200 * gimple-ssa-backprop.c (backprop::process_assign_use): Likesie.
9201 (strip_sign_op_1): Likesise.
9202 * match.pd: Add new pattern to generate ABSU_EXPR.
9203 * optabs-tree.c (optab_for_tree_code): Handle ABSU_EXPR.
9204 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
9205 * tree-eh.c (operation_could_trap_helper_p): Likewise.
9206 * tree-inline.c (estimate_operator_cost): Likewise.
9207 * tree-pretty-print.c (dump_generic_node): Likewise.
9208 * tree-vect-patterns.c (vect_recog_sad_pattern): Likewise.
9209 * tree.def (ABSU_EXPR): New.
9211 2018-06-16 Jakub Jelinek <jakub@redhat.com>
9214 * common.opt (Wunsafe-loop-optimizations): Add Ignore, remove Var,
9215 documented as preserved for backward compatibility only.
9216 * doc/invoke.texi: Remove -Wunsafe-loop-optimizations documentation.
9218 PR rtl-optimization/86108
9219 * bb-reorder.c (create_forwarder_block): Renamed to ...
9220 (create_eh_forwarder_block): ... this. Split OLD_BB after labels and
9221 jump from new landing pad to the second part.
9222 (sjlj_fix_up_crossing_landing_pad, dw2_fix_up_crossing_landing_pad):
9225 2018-06-15 Jakub Jelinek <jakub@redhat.com>
9228 * expr.c (expand_assignment): Remove now redundant COMPLEX_MODE_P
9229 check from first store_expr, use to_mode instead of GET_MODE (to_rtx).
9230 Only call store_expr for halves if the mode is the same.
9233 * match.pd ((X / Y) == 0 -> X < Y): Don't transform complex divisions.
9234 Fix up comment formatting.
9236 2018-06-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
9238 * typed-splay-tree.h (typed_splay_tree::remove): New function.
9239 (typed_splay_tree::closure,
9240 typed_splay_tree::inner_foreach_fn, typed_splay_tree::m_inner): Deleted.
9241 (typed_splay_tree::typed_splay_tree,
9242 typed_splay_tree::operator =): Declared private.
9243 (typed_splay_tree::splay_tree_key, typed_splay_tree::splay_tree_value,
9244 typed_splay_tree::splay_tree_node_s, typed_splay_tree::KDEL,
9245 typed_splay_tree::VDEL, typed_splay_tree::splay_tree_delete_helper,
9246 typed_splay_tree::rotate_left, typed_splay_tree::rotate_right,
9247 typed_splay_tree::splay_tree_splay,
9248 typed_splay_tree::splay_tree_foreach_helper,
9249 typed_splay_tree::splay_tree_insert,
9250 typed_splay_tree::splay_tree_remove,
9251 typed_splay_tree::splay_tree_lookup,
9252 typed_splay_tree::splay_tree_predecessor,
9253 typed_splay_tree::splay_tree_successor,
9254 typed_splay_tree::splay_tree_min,
9255 typed_splay_tree::splay_tree_max): Took over from splay-tree.c/.h.
9256 (typed_splay_tree::root, typed_splay_tree::comp,
9257 typed_splay_tree::delete_key,
9258 typed_splay_tree::delete_value): New data members.
9259 * typed-splay-tree.c (selftest::test_str_to_int): Add a test for
9260 typed_splay_tree::remove.
9262 2018-06-15 Matthew Fortune <matthew.fortune@mips.com>
9264 * config/mips/mips.h (ASM_SPEC): Pass through -mcrc, -mno-crc,
9265 -mginv and -mno-ginv to the assembler.
9266 * config/mips/mips.opt (-mcrc): New option.
9268 * doc/invoke.text (-mcrc): Document.
9271 2018-06-15 Nick Clifton <nickc@redhat.com>
9274 * tree.c (escaped_string): New class. Converts an unescaped
9275 string into its escaped equivalent.
9276 (warn_deprecated_use): Use the new class to convert the
9277 deprecation message, if present.
9278 (test_escaped_strings): New self test.
9279 (test_c_tests): Add test_escaped_strings.
9280 * doc/extend.texi (deprecated): Add a note that the
9281 deprecation message is affected by the -fmessage-length
9282 option, and that control characters will be escaped.
9283 (#pragma GCC error): Document this pragma.
9284 (#pragma GCC warning): Likewise.
9285 * doc/invoke.texi (-fmessage-length): Document this option's
9286 effect on the #warning and #error preprocessor directives and
9287 the deprecated attribute.
9289 2018-06-15 Richard Biener <rguenther@suse.de>
9291 * tree-vect-slp.c (vect_slp_bb): Dump MSG_OPTIMIZED_LOCATIONS
9292 here, also noting vector size used.
9293 * tree-vectorizer.c (vectorize_loops): Adjust. Note vector
9294 size used in MSG_OPTIMIZED_LOCATIONS dump.
9295 (pass_slp_vectorize::execute): Adjust.
9297 2018-06-15 Claudiu Zissulescu <claziss@synopsys.com>
9300 * config/arc/arc.c (arc_return_address_register): Fix
9303 2018-06-15 Richard Biener <rguenther@suse.de>
9306 * tree-cfg.c (gimplify_build3): Do not strip sign conversions,
9307 leave useless conversion stripping to force_gimple_operand_gsi.
9308 (gimplify_build2): Likewise.
9309 (gimplify_build1): Likewise.
9311 2018-06-15 Richard Biener <rguenther@suse.de>
9314 * tree-cfg.c (move_stmt_op): unshare invariant addresses
9315 before adjusting their block.
9317 2018-06-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
9319 * config.gcc (riscv*-*-elf* | riscv*-*-rtems*): Use custom
9320 multilibs for *-*-rtems*.
9321 * config/riscv/t-rtems: New file.
9323 2018-06-14 Jakub Jelinek <jakub@redhat.com>
9326 * match.pd ((A +- CST1) +- CST2): Punt if last resort
9327 unsigned_type_for returns NULL.
9330 * lower-subreg.c (find_decomposable_subregs): Don't decompose float
9331 subregs of multi-word pseudos unless the float mode has word size.
9333 2018-06-14 Richard Biener <rguenther@suse.de>
9336 * tree-vect-generic.c (build_word_mode_vector_type): Remove
9337 duplicate and harmful type_hash_canon.
9338 * tree.c (type_hash_canon): Assert we didn't find ourselves.
9340 2018-06-14 Richard Biener <rguenther@suse.de>
9343 * tree-ssa-struct-alias.c (create_variable_info_for): Handle
9346 2018-06-14 Sebastian Huber <sebastian.huber@embedded-brains.de>
9348 * config/rtems.h (STDINT_LONG32): Define.
9350 2018-06-13 Matthew Fortune <matthew.fortune@mips.com>
9351 Prachi Godbole <prachi.godbole@imgtec.com>
9353 * config/mips/mips-cpus.def: Define P6600.
9354 * config/mips/mips-tables.opt: Regenerate.
9355 * config/mips/mips.c (mips_ucbranch_type): New enum.
9356 (mips_rtx_cost_data): Add support for P6600.
9357 (mips_issue_rate): Likewise.
9358 (mips_multipass_dfa_lookahead): Likewise.
9359 (mips_avoid_hazard): Likewise.
9360 (mips_reorg_process_insns): Likewise.
9361 (mips_classify_branch_p6600): New function.
9362 * config/mips/mips.h (TUNE_P6600): New define.
9363 (MIPS_ISA_LEVEL_SPEC): Infer mips64r6 from p6600.
9364 (ENABLE_LD_ST_PAIRS): Enable load/store bonding for p6600.
9365 * config/mips/mips.md: Include p6600.md.
9366 (processor): Add p6600.
9367 * config/mips/p6600.md: New file.
9368 * doc/invoke.texi: Add p6600 to supported architectures.
9370 2018-06-13 Martin Sebor <msebor@redhat.com>
9372 PR tree-optimization/86114
9373 * gimple-fold.c (gimple_fold_builtin_strlen): Only handle LHS
9375 * tree-ssa-strlen.c (maybe_set_strlen_range): Same.
9377 2018-06-13 Richard Biener <rguenther@suse.de>
9379 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern):
9380 Properly set vector type of the intermediate stmt.
9381 * tree-vect-stmts.c (vectorizable_operation): The destination
9382 var always has vectype_out type.
9384 2018-06-13 Jeff Law <law@redhat.com>
9386 * config/rl78/rl78.c (move_elim_pass): Use TDF_NONE rather than
9387 integer 0 for argument to print_rtl_with_bb.
9388 (rl78_reorg): Likewise.
9390 2018-06-13 David Malcolm <dmalcolm@redhat.com>
9392 * config/arc/arc.c (hwloop_optimize): Strengthen local "end_label"
9393 from rtx to rtx_insn *.
9394 * config/bfin/bfin.c (hwloop_optimize): Likewise for local
9396 (add_sched_insns_for_speculation): Likewise for local "target",
9397 converting usage of JUMP_LABEL to JUMP_LABEL_AS_INSN.
9398 * config/c6x/c6x.c (reorg_split_calls): Strengthen param "call_labels"
9399 from rtx_insn ** to rtx_code_label **.
9400 (reorg_emit_nops): Likewise.
9401 (c6x_reorg): Likewise for local "call_labels".
9402 * config/sh/sh-protos.h (get_dest_uid): Strengthen 1st param from
9404 * config/sh/sh.c (dump_table): Strengthen local "lab" from rtx to
9405 rtx_code_label *, adding safe_as_a <rtx_code_label *> casts to
9406 the loops over LABEL_REFS.
9407 (fixup_addr_diff_vecs): Add as_a <rtx_insn *> to usage of
9409 (barrier_align): Convert usage of JUMP_LABEL to JUMP_LABEL_AS_INSN.
9410 (get_dest_uid): Strengthen param "label" from rtx to rtx_insn *.
9411 (split_branches): Strengthen local "olabel" from rtx to
9412 rtx_insn *, adding a safe_as_a cast.
9413 * emit-rtl.c (next_real_insn): Strengthen param from "rtx"
9415 (add_insn_after): Likewise for first two params.
9416 (add_insn_before): Likewise.
9417 (remove_insn): Likewise for param.
9418 (emit_pattern_before_noloc): Likewise for second and third params.
9419 (emit_jump_insn_before_noloc): Convert NULL_RTX to NULL.
9420 (emit_call_insn_before_noloc): Likewise.
9421 (emit_debug_insn_before_noloc): Strengthen "before" param from "rtx"
9423 (emit_barrier_before): Likewise.
9424 (emit_label_before): Strengthen "label" param from "rtx" to
9425 "rtx_code_label *". Strengthen "before" param from "rtx" to
9427 (emit_insn_after_1): Strengthen "after" param from "rtx" to
9429 (emit_pattern_after_noloc): Likewise.
9430 (emit_insn_after_noloc): Likewise.
9431 (emit_jump_insn_after_noloc): Likewise.
9432 (emit_call_insn_after_noloc): Likewise.
9433 (emit_debug_insn_after_noloc): Likewise.
9434 (emit_barrier_after): Likewise.
9435 (emit_label_after): Likewise for both params.
9436 (emit_pattern_after_setloc): Likewise for "after" param. Convert
9437 "loc" param from "int" to "location_t".
9438 (emit_insn_after_setloc): Likewise.
9439 (emit_jump_insn_after_setloc): Likewise.
9440 (emit_call_insn_after_setloc): Likewise.
9441 (emit_debug_insn_after_setloc): Likewise.
9442 (emit_pattern_before_setloc): Likewise for "before" param. Convert
9443 "loc" param from "int" to "location_t".
9444 (emit_pattern_before): Convert NULL_RTX to NULL.
9445 (emit_insn_before_setloc): Convert "loc" param from "int" to
9447 (emit_jump_insn_before_setloc): Likewise.
9448 (emit_call_insn_before_setloc): Likewise.
9449 (emit_debug_insn_before_setloc): Strengthen "before" param from rtx to
9450 rtx_insn *. Convert "loc" param from "int" to "location_t".
9451 * rtl.h (emit_insn_before_setloc, emit_jump_insn_before_setloc,
9452 emit_call_insn_before_setloc, emit_debug_insn_before_setloc):
9453 Convert 3rd param from "int" to "location_t".
9454 (emit_barrier_before, emit_barrier_after, next_real_insn):
9455 Strengthen param from rtx to rtx_insn *.
9456 (emit_label_before): Strengthen 1st param from "rtx" to
9457 "rtx_code_label *". Strengthen 2nd param from "rtx" to
9459 (emit_insn_after_noloc, emit_jump_insn_after_noloc,
9460 emit_call_insn_after_noloc, emit_debug_insn_after_noloc):
9461 Strengthen 2nd param from "rtx" to "rtx_insn *".
9462 (emit_insn_after_setloc, emit_jump_insn_after_setloc)
9463 emit_call_insn_after_setloc, emit_debug_insn_after_setloc):
9464 Likewise. Convert 3rd param from "int" to "location_t".
9465 (emit_label_after): Strengthen 1st param from "rtx" to
9467 (next_real_insn, remove_insn): Strengthen param from "rtx" to
9469 (add_insn_before, add_insn_after): Strengthen 1st and 2nd params
9470 from "rtx" to "rtx_insn *".
9472 2018-06-13 Jan Hubicka <hubicka@gcc.gnu.org>
9474 * cgraph.c (cgraph_node::get_untransformed_body): Dump function
9475 bodies streamed in with -Q.
9476 * dumpfile.c (dump_files): Add lto-stream-out dump file.
9477 * dumpfile.h (tree_dump_index): Add lto_stream_out.
9478 * gimple-streamer-out.c: Include gimple-pretty-print.h
9479 (output_bb): Dump stmts streamed.
9480 * lto-section-out.c: Include print-tree.h
9481 (lto_begin_section): Dump sections created.
9482 (lto_output_decl_index): Dump decl encoded.
9483 * lto-streamer-out.c: Include print-tree.h
9484 (create_output_block): Dump output block created.
9485 (DFS::DFS_write_tree_body): Dump DFS SCCs streamed.
9486 (output_function): Dump function output.
9487 (output_constructor): Dump constructor streamed.
9488 (write_global_stream): Output indexes encoded.
9489 (produce_asm_for_decls): Dump streams encoded.
9490 * lto-streamer.c (streamer_dump_file): New global var.
9491 * lto-streamer.h (streamer_dump_file): Declare.
9492 * passes.c (ipa_write_summaries): Initialize streamer dump.
9493 * varpool.c (varpool_node::get_constructor): Dump constructors streamed
9496 2018-06-13 Eric Botcazou <ebotcazou@adacore.com>
9499 * config/i386/winnt.c (i386_pe_seh_cold_init): Do not emit negative
9500 offsets for register save directives. Emit a second batch of save
9501 directives, if need be, when the function accesses prior frames.
9503 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
9505 * config/arc/fpu.md (fmasf4): Force operand to register.
9506 (fnmasf4): Likewise.
9508 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
9510 * config/arc/arc-protos.h (arc_pad_return): Remove.
9511 * config/arc/arc.c (machine_function): Remove force_short_suffix
9513 (arc_print_operand): Adjust printing of '&'.
9514 (arc_verify_short): Remove conditional printing of short suffix.
9515 (arc_final_prescan_insn): Remove reference to size_reason.
9516 (pad_return): New function.
9517 (arc_reorg): Call pad_return.
9518 (arc_pad_return): Remove.
9519 (arc_init_machine_status): Remove reference to force_short_suffix.
9520 * config/arc/arc.md (vunspec): Add VUNSPEC_ARC_BLOCKAGE.
9521 (attr length): When attribute iscompact is true force to 2
9522 regardless; in the case of maybe check if we want to force the
9523 instruction to have 4 bytes length.
9524 (nopv): Change it to generate 4 byte long nop as well.
9525 (blockage): New pattern.
9526 (simple_return): Remove call to arc_pad_return.
9527 (p_return_i): Likewise.
9529 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
9531 * config/arc/elf.h (LINK_GCC_C_SEQUENCE_SPEC): Define.
9533 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
9535 * config/arc/builtins.def (SYNC): SYNC instruction is valid on all
9538 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
9540 * config/arc/arc.c (atomic_exchangesi): EX instruction is default
9541 for ARC700 and ARCv2.
9543 2018-06-13 Chenghua Xu <paul.hua.gm@gmail.com>
9546 * config/mips/loongson.md (vec_setv4hi): Gen_lowpart for
9547 operands[2] instead of operands[1].
9550 2018-06-12 Richard Sandiford <richard.sandiford@linaro.org>
9552 * lra-constraints.c (simplify_operand_subreg): In the paradoxical
9553 case, check whether the outer register overlaps an unallocatable
9554 register, not just whether it fits the required class.
9556 2018-06-12 Richard Sandiford <richard.sandiford@linaro.org>
9558 * poly-int.h (can_div_trunc_p): Add new overload in which all values
9560 * alias.c (get_addr): Extend CONST_INT handling to poly_int_rtx_p.
9561 (memrefs_conflict_p): Likewise.
9562 (init_alias_analysis): Likewise.
9563 * cfgexpand.c (expand_debug_expr): Likewise.
9564 * combine.c (combine_simplify_rtx, force_int_to_mode): Likewise.
9565 * cse.c (fold_rtx): Likewise.
9566 * explow.c (adjust_stack, anti_adjust_stack): Likewise.
9567 * expr.c (emit_block_move_hints): Likewise.
9568 (clear_storage_hints, push_block, emit_push_insn): Likewise.
9569 (store_expr_with_bounds, reduce_to_bit_field_precision): Likewise.
9570 (emit_group_load_1): Use rtx_to_poly_int64 for group offsets.
9571 (emit_group_store): Likewise.
9572 (find_args_size_adjust): Use strip_offset. Use rtx_to_poly_int64
9573 to read the PRE/POST_MODIFY increment.
9574 * calls.c (store_one_arg): Use strip_offset.
9575 * rtlanal.c (rtx_addr_can_trap_p_1): Extend CONST_INT handling to
9577 (set_noop_p): Use rtx_to_poly_int64 for the elements selected
9579 * simplify-rtx.c (avoid_constant_pool_reference): Use strip_offset.
9580 (simplify_binary_operation_1): Extend CONST_INT handling to
9582 * var-tracking.c (compute_cfa_pointer): Take a poly_int64 rather
9583 than a HOST_WIDE_INT.
9584 (hard_frame_pointer_adjustment): Change from HOST_WIDE_INT to
9586 (adjust_mems, add_stores): Update accodingly.
9587 (vt_canonicalize_addr): Track polynomial offsets.
9588 (emit_note_insn_var_location): Likewise.
9589 (vt_add_function_parameter): Likewise.
9590 (vt_initialize): Likewise.
9592 2018-06-12 Jeff Law <law@redhat.com>
9594 * config.gcc (alpha*-*-freebsd*): Remove.
9595 * config/alpha/freebsd.h: Remove.
9597 2018-06-12 David Malcolm <dmalcolm@redhat.com>
9600 * spellcheck-tree.c (levenshtein_distance): Rename to...
9601 (get_edit_distance): ...this, and update for underlying renaming.
9602 * spellcheck-tree.h (levenshtein_distance): Rename to...
9603 (get_edit_distance): ...this.
9604 * spellcheck.c (levenshtein_distance): Rename to...
9605 (get_edit_distance): ...this. Convert from Levenshtein distance
9606 to Damerau-Levenshtein distance by supporting transpositions of
9607 adjacent characters. Rename "v1" to "v_next" and "v0" to
9609 (selftest::levenshtein_distance_unit_test_oneway): Rename to...
9610 (selftest::test_edit_distance_unit_test_oneway): ...this, and
9611 update for underlying renaming.
9612 (selftest::levenshtein_distance_unit_test): Rename to...
9613 (selftest::test_get_edit_distance_unit): ...this, and update for
9614 underlying renaming.
9615 (selftest::test_find_closest_string): Add example from PR 69968
9616 where transposition helps
9617 (selftest::test_metric_conditions): Update for renaming.
9618 (selftest::test_metric_conditions): Likewise.
9619 (selftest::spellcheck_c_tests): Likewise.
9620 * spellcheck.h (levenshtein_distance): Rename both overloads to...
9621 (get_edit_distance): ...this.
9622 (best_match::consider): Update for renaming.
9624 2018-06-12 Martin Sebor <msebor@redhat.com>
9626 PR tree-optimization/85259
9627 * builtins.c (compute_objsize): Handle constant offsets.
9628 * gimple-ssa-warn-restrict.c (maybe_diag_offset_bounds): Return
9629 true iff a warning has been issued.
9630 * gimple.h (gimple_nonartificial_location): New function.
9631 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Call
9632 gimple_nonartificial_location and handle -Wno-system-headers.
9633 (handle_builtin_stxncpy): Same.
9635 2018-06-12 Martin Sebor <msebor@redhat.com>
9638 * fold-const.c (operand_equal_p): Handle SAVE_EXPR.
9640 2018-06-12 Will Schmidt <will_schmidt@vnet.ibm.com>
9642 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
9643 BUILTIN_VEC_XST entries for pointer to double and long long.
9645 2018-06-12 H.J. Lu <hongjiu.lu@intel.com>
9648 * config/i386/gnu-user.h (TARGET_THREAD_SPLIT_STACK_OFFSET):
9650 * config/i386/gnu-user64.h (TARGET_THREAD_SPLIT_STACK_OFFSET):
9653 2018-06-12 Martin Liska <mliska@suse.cz>
9655 * doc/options.texi: Document IntegerRange.
9657 2018-06-12 Martin Liska <mliska@suse.cz>
9659 * config/i386/i386.opt: Make MPX-related options as Deprecated.
9660 * opt-functions.awk: Handle Deprecated flag.
9661 * opts-common.c (decode_cmdline_option): Handle cl_deprecated
9663 (read_cmdline_option): Report warning for a deprecated option.
9664 * opts.h (struct cl_option): Add new field cl_deprecated.
9665 (CL_ERR_DEPRECATED): New.
9667 2018-06-12 Martin Liska <mliska@suse.cz>
9669 * doc/options.texi: Document Deprecated option flag.
9671 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
9673 * config/arc/arc-arch.h (arc_extras): New enum.
9674 (arc_cpu_t):Add field extra.
9675 (arc_cpu_types): Consider the extras.
9676 * config/arc/arc-cpus.def: Add extras info.
9677 * config/arc/arc-opts.h (processor_type): Consider extra field.
9678 * config/arc/arc.c (arc_override_options): Handle extra field.
9680 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
9682 * config/arc/arc-arch.h: Update ARC_OPTX macro.
9683 * config/arc/arc-options.def (ARC_OPTX): Introduce a new doc
9685 * config/arc/arc.c (arc_init): Update pic warning.
9686 (irq_range): Update irq range parsing warnings.
9687 (arc_override_options): Update various warning messages.
9688 (arc_handle_aux_attribute): Likewise.
9690 2018-06-12 Robert Suchanek <robert.suchanek@mips.com>
9692 * config/mips/i6400.md (i6400_fpu_fadd): Remove frint.
9694 2018-06-12 Jozef Lawrynowicz <jozef.l@mittosystems.com>
9696 * doc/sourcebuild.texi: Document usage of line number 0 in verify
9697 compiler messages directives.
9699 2018-06-12 Matthew Fortune <mfortune@gmail.com>
9701 * config/mips/mips-cpus.def: New MIPS_CPU for i6500.
9702 * config/mips/mips-tables.opt: Regenerate.
9703 * config/mips/mips.h (MIPS_ISA_LEVEL_SPEC): Mark i6500 as
9705 * doc/invoke.texi: Document -march=i6500.
9707 2018-06-12 Prachi Godbole <prachi.godbole@imgtec.com>
9709 * config/mips/i6400.md (i6400_gpmuldiv): Remove cpu_unit.
9710 (i6400_gpmul): Add cpu_unit.
9711 (i6400_gpdiv): Likewise.
9712 (i6400_msa_add_d): Update reservations.
9713 (i6400_msa_int_add) Likewise.
9714 (i6400_msa_short_logic3) Likewise.
9715 (i6400_msa_short_logic2) Likewise.
9716 (i6400_msa_short_logic) Likewise.
9717 (i6400_msa_move) Likewise.
9718 (i6400_msa_cmp) Likewise.
9719 (i6400_msa_short_float2) Likewise.
9720 (i6400_msa_div_d) Likewise.
9721 (i6400_msa_long_logic1) Likewise.
9722 (i6400_msa_long_logic2) Likewise.
9723 (i6400_msa_mult) Likewise.
9724 (i6400_msa_long_float2) Likewise.
9725 (i6400_msa_long_float4) Likewise.
9726 (i6400_msa_long_float5) Likewise.
9727 (i6400_msa_long_float8) Likewise.
9728 (i6400_fpu_fadd): Include frint type.
9729 (i6400_fpu_store): New define_insn_reservation.
9730 (i6400_fpu_load): Likewise.
9731 (i6400_fpu_move): Likewise.
9732 (i6400_fpu_fcmp): Likewise.
9733 (i6400_fpu_fmadd): Likewise.
9734 (i6400_int_mult): Include imul3nc type and update reservation.
9735 (i6400_int_div): Include idiv3 type and update reservation.
9736 (i6400_int_load): Update to check type not move_type.
9737 (i6400_int_store): Likewise.
9738 (i6400_int_prefetch): Set zero latency.
9740 2018-06-12 Eric Botcazou <ebotcazou@adacore.com>
9742 * gcc.c: Document new %@{...} sequence.
9743 (LINK_COMMAND_SPEC): Use it for the -L switches.
9744 (cpp_unique_options): Use it for the -I switches.
9745 (at_file_argbuf): New global variable.
9746 (in_at_file): Likewise.
9747 (alloc_args): Create at_file_argbuf.
9748 (clear_args): Truncate at_file_argbuf.
9749 (store_arg): If in_at_file, push the argument onto at_file_argbuf.
9750 (open_at_file): New function.
9751 (close_at_file): Likewise.
9752 (create_at_file): Delete.
9753 (do_spec_1) <'i'>: Use open_at_file/close_at_file.
9756 (validate_switches_from_spec): Deal with %@{...} sequence.
9757 (validate_switches): Likewise.
9758 (driver::finalize): Call clear_args.
9760 2018-06-11 Rasmus Villemoes <rasmus.villemoes@prevas.dk>
9762 * config/vx-common.h (USE_TM_CLONE_REGISTRY): #define to 0.
9764 2018-06-11 Martin Sebor <msebor@redhat.com>
9766 * doc/invoke.texi (-Wall): List -Wc++17-compat.
9767 (Wno-class-memaccess): Add @opindex.
9768 (Wno-templates, Wno-multiple-inheritance): Same.
9769 (Wno-virtual-inheritance, Wno-namespaces): Same.
9770 (Wno-pedantic, Wno-chkp, Wcoverage-mismatch): Same.
9771 (Wno-format-overflow, Wno-format-truncation): Same.
9772 (Wno-unused-local-typedefs, Walloc-size-larger-than): Same.
9773 (Wno-alloc-size-larger-than, Wframe-larger-than): Same
9774 (Wno-stack-usage, Wno-c++-compat, Wno-c++11-compat): Same.
9775 (Wno-c++14-compat, Wno-c++17-compat, Wno-openmp-simd): Same.
9776 (Wno-unsuffixed-float-constants, Wno-addr-space-convert): Same.
9777 (Wno-misspelled-isr): Same.
9779 2018-06-11 Martin Sebor <msebor@redhat.com>
9781 * PR tree-optimization/86083
9782 * tree-ssa-strlen.c (handle_char_store): Use tree_expr_nonzero_p.
9784 2018-06-11 Zhouyi Zhou <zhouzhouyi@gmail.com>
9786 * tree-eh.c (lower_eh_constructs_2): Add a comma to comment.
9788 2018-06-11 Segher Boessenkool <segher@kernel.crashing.org>
9791 * config/rs6000/rs6000.md (*movdi_internal32): Put constraint modifiers
9792 on the correct operand.
9793 (*movdi_internal64): Ditto.
9795 2018-06-11 Martin Liska <mliska@suse.cz>
9797 PR tree-optimization/86089
9798 * tree-ssa-strlen.c (get_string_length): Move back removed hunk.
9800 2018-06-11 Julia Koval <julia.koval@intel.com>
9802 * config/i386/vaesintrin.h (_mm_aesdec_epi128, _mm_aesdeclast_epi128)
9803 _mm_aesenc_epi128, _mm_aesenclast_epi128: Remove.
9804 * config/i386/vpclmulqdqintrin.h (_mm_clmulepi64_epi128): Remove.
9806 2018-06-11 Olivier Hainque <hainque@adacore.com>
9808 * dwarf2out.c (gen_compile_unit_die): Fallout to DW_LANG_Ada83
9809 for Ada with strict dwarf2.
9811 2018-06-08 Peter Bergner <bergner@vnet.ibm.com>
9814 * config/rs6000/rs6000.c (mem_operand_gpr): Enable PRE_INC and PRE_DEC
9817 2018-06-08 Jan Hubicka <hubicka@ucw.cz>
9819 * dumpfile.c (FIRST_ME_AUTO_NUMBERED_DUMP): Bump to 4.
9821 2018-06-08 David Edelsohn <dje.gcc@gmail.com>
9823 * config/rs6000/rs6000.c (rs6000_passes_ieee128): Protect with #if
9826 2018-06-08 Martin Liska <mliska@suse.cz>
9828 * tree-cfg.h (debug_function): Fix argument type to match
9831 2018-06-08 Martin Liska <mliska@suse.cz>
9833 * config/powerpcspe/powerpcspe.c (rs6000_xcoff_visibility):
9834 Remove usage of MPX-related (and removed) fields.
9835 * config/rs6000/rs6000.c (rs6000_xcoff_visibility): Likewise.
9837 2018-06-08 David Malcolm <dmalcolm@redhat.com>
9839 * cfg.c (debug): Use TDF_NONE rather than 0.
9840 * cfghooks.c (debug): Likewise.
9841 * dumpfile.c (DUMP_FILE_INFO): Likewise; also for OPTGROUP.
9842 (struct dump_option_value_info): Convert to...
9843 (struct kv_pair): ...this template type.
9844 (dump_options): Convert to kv_pair<dump_flags_t>; use TDF_NONE
9846 (optinfo_verbosity_options): Likewise.
9847 (optgroup_options): Convert to kv_pair<optgroup_flags_t>; use
9849 (gcc::dump_manager::dump_register): Use optgroup_flags_t rather
9850 than int for "optgroup_flags" param.
9851 (dump_generic_expr_loc): Use dump_flags_t rather than int for
9853 (dump_dec): Likewise.
9854 (dump_finish): Use TDF_NONE rather than 0.
9855 (gcc::dump_manager::opt_info_enable_passes): Use optgroup_flags_t
9856 rather than int for "optgroup_flags" param. Use TDF_NONE rather
9857 than 0. Update for change to option_ptr.
9858 (opt_info_switch_p_1): Convert "optgroup_flags" param from int *
9859 to optgroup_flags_t *. Use TDF_NONE and OPTGROUP_NONE rather than
9860 0. Update for changes to optinfo_verbosity_options and
9862 (opt_info_switch_p): Convert optgroup_flags from int to
9864 (dump_basic_block): Use dump_flags_t rather than int
9865 for "dump_kind" param.
9866 * dumpfile.h (TDF_ADDRESS, TDF_SLIM, TDF_RAW, TDF_DETAILS,
9867 TDF_STATS, TDF_BLOCKS, TDF_VOPS, TDF_LINENO, TDF_UID)
9868 TDF_STMTADDR, TDF_GRAPH, TDF_MEMSYMS, TDF_RHS_ONLY, TDF_ASMNAME,
9869 TDF_EH, TDF_NOUID, TDF_ALIAS, TDF_ENUMERATE_LOCALS, TDF_CSELIB,
9870 TDF_SCEV, TDF_GIMPLE, TDF_FOLDING, MSG_OPTIMIZED_LOCATIONS,
9871 MSG_MISSED_OPTIMIZATION, MSG_NOTE, MSG_ALL, TDF_COMPARE_DEBUG,
9872 TDF_NONE): Convert from macros to...
9873 (enum dump_flag): ...this new enum.
9874 (dump_flags_t): Update to use enum.
9875 (operator|, operator&, operator~, operator|=, operator&=):
9876 Implement for dump_flags_t.
9877 (OPTGROUP_NONE, OPTGROUP_IPA, OPTGROUP_LOOP, OPTGROUP_INLINE,
9878 OPTGROUP_OMP, OPTGROUP_VEC, OPTGROUP_OTHER, OPTGROUP_ALL):
9879 Convert from macros to...
9880 (enum optgroup_flag): ...this new enum.
9881 (optgroup_flags_t): New typedef.
9882 (operator|, operator|=): Implement for optgroup_flags_t.
9883 (struct dump_file_info): Convert field "alt_flags" to
9884 dump_flags_t. Convert field "optgroup_flags" to
9886 (dump_basic_block): Use dump_flags_t rather than int for param.
9887 (dump_generic_expr_loc): Likewise.
9888 (dump_dec): Likewise.
9889 (dump_register): Convert param "optgroup_flags" to
9891 (opt_info_enable_passes): Likewise.
9892 * early-remat.c (early_remat::dump_edge_list): Use TDF_NONE rather
9894 * gimple-pretty-print.c (debug): Likewise.
9895 * gimple-ssa-store-merging.c (bswap_replace): Likewise.
9896 (merged_store_group::apply_stores): Likewise.
9897 * gimple-ssa-strength-reduction.c (insert_initializers): Likewise.
9898 * gimple.c (verify_gimple_pp): Likewise.
9899 * graphite-poly.c (print_pbb_body): Likewise.
9900 * passes.c (pass_manager::register_one_dump_file): Convert
9901 local "optgroup_flags" to optgroup_flags_t.
9902 * print-tree.c (print_node): Use TDF_NONE rather than 0.
9904 (debug_body): Likewise.
9905 * tree-pass.h (struct pass_data): Convert field "optgroup_flags"
9906 to optgroup_flags_t.
9907 * tree-pretty-print.c (print_struct_decl): Use TDF_NONE rather
9909 * tree-ssa-math-opts.c (convert_mult_to_fma_1): Likewise.
9910 (convert_mult_to_fma): Likewise.
9911 * tree-ssa-reassoc.c (undistribute_ops_list): Likewise.
9912 * tree-ssa-sccvn.c (vn_eliminate): Likewise.
9913 * tree-vect-data-refs.c (dump_lower_bound): Convert param
9914 "dump_kind" to dump_flags_t.
9916 2018-06-08 Segher Boessenkool <segher@kernel.crashing.org>
9918 * config/rs6000/rs6000.c (min, max): Delete.
9920 2018-06-08 Segher Boessenkool <segher@kernel.crashing.org>
9922 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete mention of
9923 -mabi=spe and -mabi=no-spe.
9925 2018-06-08 Martin Liska <mliska@suse.cz>
9927 * ipa-pure-const.c (propagate_pure_const): Use ::get at places
9928 where we expect an existing summary.
9930 2018-06-08 Martin Liska <mliska@suse.cz>
9932 * ipa-inline-analysis.c (simple_edge_hints): Use ::get method.
9933 * ipa-inline.h (estimate_edge_growth): Likewise.
9935 2018-06-08 Martin Liska <mliska@suse.cz>
9937 * cgraph.c (function_version_hasher::hash): Use
9938 cgraph_node::get_uid ().
9939 (function_version_hasher::equal):
9940 * cgraph.h (cgraph_node::get_uid): New method.
9941 * ipa-inline.c (update_caller_keys): Use
9942 cgraph_node::get_uid ().
9943 (update_callee_keys): Likewise.
9944 * ipa-utils.c (searchc): Likewise.
9945 (ipa_reduced_postorder): Likewise.
9946 * lto-cgraph.c (input_node): Likewise.
9947 * passes.c (is_pass_explicitly_enabled_or_disabled): Likewise.
9948 * symbol-summary.h (symtab_insertion): Likewise.
9949 (symtab_removal): Likewise.
9950 (symtab_duplication): Likewise.
9951 * tree-pretty-print.c (dump_function_header): Likewise.
9952 * tree-sra.c (convert_callers_for_node): Likewise.
9954 2018-06-08 Martin Liska <mliska@suse.cz>
9956 * cgraph.c (symbol_table::create_edge): Always assign a new
9958 (symbol_table::free_edge): Do not recycle numbers.
9959 * cgraph.h (cgraph_edge::get): New method.
9960 * symbol-summary.h (symtab_removal): Use it.
9961 (symtab_duplication): Likewise.
9962 (call_summary::hashable_uid): Remove.
9964 2018-06-08 Martin Liska <mliska@suse.cz>
9966 * ipa-inline-analysis.c (inline_edge_removal_hook): Remove.
9967 (initialize_growth_caches): Remove.
9968 (free_growth_caches): Likewise.
9969 (do_estimate_edge_time): Use edge_growth_cache.
9970 (do_estimate_edge_size): Likewise.
9971 (do_estimate_edge_hints): Likewise.
9972 * ipa-inline.c (reset_edge_caches): Likewise.
9973 (recursive_inlining): Likewise.
9974 (inline_small_functions): Likewise.
9975 * ipa-inline.h (initialize_growth_caches): Remove.
9976 (estimate_edge_size): Likewise.
9977 (estimate_edge_time): Likewise.
9978 (estimate_edge_hints): Likewise.
9979 (reset_edge_growth_cache): Likewise.
9980 * symbol-summary.h (call_summary::remove): New method.
9982 2018-06-08 Martin Liska <mliska@suse.cz>
9984 * ipa-cp.c (class edge_clone_summary): New summary.
9985 (grow_edge_clone_vectors): Remove.
9986 (ipcp_edge_duplication_hook): Remove.
9987 (class edge_clone_summary_t): New call_summary class.
9988 (ipcp_edge_removal_hook): Remove.
9989 (edge_clone_summary_t::duplicate): New function.
9990 (get_next_cgraph_edge_clone): Use edge_clone_summaries.
9991 (create_specialized_node): Likewise.
9992 (ipcp_driver): Initialize edge_clone_summaries and do not
9995 2018-06-08 Martin Liska <mliska@suse.cz>
9997 * symbol-summary.h (get): New function.
9998 (call_summary::m_initialize_when_cloning): New class member.
10000 2018-06-08 Martin Liska <mliska@suse.cz>
10002 * cgraph.c (cgraph_node::remove): Do not recycle uid.
10003 * cgraph.h (symbol_table::release_symbol): Do not pass uid.
10004 (symbol_table::allocate_cgraph_symbol): Do not set uid.
10005 * passes.c (uid_hash_t): Record removed_nodes by their uids.
10006 (remove_cgraph_node_from_order): Use the removed_nodes set.
10007 (do_per_function_toporder): Likwise.
10008 * symbol-summary.h (symtab_insertion): Use cgraph_node::uid
10009 instead of summary_uid.
10010 (symtab_removal): Likewise.
10011 (symtab_duplication): Likewise.
10013 2018-06-08 Martin Liska <mliska@suse.cz>
10015 * ipa-cp.c (ipcp_store_bits_results): Use
10016 ipcp_transformation_sum.
10017 (ipcp_store_vr_results): Likewise.
10018 * ipa-prop.c (ipcp_grow_transformations_if_necessary): Renamed
10020 (ipcp_transformation_initialize): ... this.
10021 (ipa_set_node_agg_value_chain):
10022 (ipa_node_params_t::duplicate): Use ipcp_transformation_sum.
10023 (write_ipcp_transformation_info): Likewise.
10024 (read_ipcp_transformation_info): Likewise.
10025 (ipcp_update_bits): Likewise.
10026 (ipcp_update_vr): Likewise.
10027 (ipcp_transform_function): Likewise.
10028 * ipa-prop.h: Rename ipcp_transformation_summary to
10029 ipcp_transformation.
10030 (class ipcp_transformation_t): New function summary.
10031 (ipcp_get_transformation_summary): Use ipcp_transformation_sum.
10032 (ipa_get_agg_replacements_for_node): Likewise.
10034 2018-06-08 Martin Liska <mliska@suse.cz>
10036 * ipa-pure-const.c (struct funct_state_d): Do it class instead
10038 (class funct_state_summary_t): New function_summary class.
10039 (has_function_state): Remove.
10040 (get_function_state): Likewise.
10041 (set_function_state): Likewise.
10042 (add_new_function): Likewise.
10043 (funct_state_summary_t::insert): New function.
10044 (duplicate_node_data): Remove.
10045 (remove_node_data): Remove.
10046 (funct_state_summary_t::duplicate): New function.
10047 (register_hooks): Create new funct_state_summaries.
10048 (pure_const_generate_summary): Use it.
10049 (pure_const_write_summary): Likewise.
10050 (pure_const_read_summary): Likewise.
10051 (propagate_pure_const): Likewise.
10052 (propagate_nothrow): Likewise.
10053 (dump_malloc_lattice): Likewise.
10054 (propagate_malloc): Likewise.
10055 (execute): Do not register hooks, just remove summary
10057 (pass_ipa_pure_const::pass_ipa_pure_const): Simplify
10060 2018-06-08 Martin Liska <mliska@suse.cz>
10062 * ipa-reference.c (remove_node_data): Remove.
10063 (duplicate_node_data): Likewise.
10064 (class ipa_ref_var_info_summary_t): New class.
10065 (class ipa_ref_opt_summary_t): Likewise.
10066 (get_reference_vars_info): Use ipa_ref_var_info_summaries.
10067 (get_reference_optimization_summary): Use
10068 ipa_ref_opt_sum_summaries.
10069 (set_reference_vars_info): Remove.
10070 (set_reference_optimization_summary): Likewise.
10071 (ipa_init): Create summaries.
10072 (init_function_info): Use function summary.
10073 (ipa_ref_opt_summary_t::duplicate): New function.
10074 (ipa_ref_opt_summary_t::remove): New function.
10075 (get_read_write_all_from_node): Fix GNU coding style.
10076 (propagate): Use function summary.
10077 (write_node_summary_p): Fix GNU coding style.
10078 (stream_out_bitmap): Likewise.
10079 (ipa_reference_read_optimization_summary): Use function summary.
10080 (ipa_reference_c_finalize): Do not release hooks.
10082 2018-06-08 Martin Liska <mliska@suse.cz>
10084 * ipa-fnsummary.c (dump_ipa_call_summary): Use ::get method.
10085 (analyze_function_body): Extract multiple calls of get_create.
10086 * ipa-inline-analysis.c (simple_edge_hints): Likewise.
10087 * ipa-inline.c (recursive_inlining): Use ::get method.
10088 * ipa-inline.h (estimate_edge_growth): Likewise.
10090 2018-06-08 Martin Liska <mliska@suse.cz>
10092 * hsa-common.h (enum hsa_function_kind): Rename HSA_NONE to
10094 (hsa_function_summary::hsa_function_summary): Use the new enum
10096 (hsa_gpu_implementation_p): Use hsa_summaries::get.
10097 * hsa-gen.c (hsa_get_host_function): Likewise.
10098 (get_brig_function_name): Likewise.
10099 * ipa-hsa.c (process_hsa_functions): Likewise.
10100 (ipa_hsa_write_summary): Likewise.
10101 * symbol-summary.h (symtab_duplication): Use ::get function/
10102 (get): New function.
10104 2018-06-08 Martin Liska <mliska@suse.cz>
10106 * config/i386/i386.c (ix86_can_inline_p): Use get_create instead
10108 * hsa-common.c (hsa_summary_t::link_functions): Likewise.
10109 (hsa_register_kernel): Likewise.
10110 * hsa-common.h (hsa_gpu_implementation_p): Likewise.
10111 * hsa-gen.c (hsa_get_host_function): Likewise.
10112 (get_brig_function_name): Likewise.
10113 (generate_hsa): Likewise.
10114 (pass_gen_hsail::execute): Likewise.
10115 * ipa-cp.c (ipcp_cloning_candidate_p): Likewise.
10116 (devirtualization_time_bonus): Likewise.
10117 (ipcp_propagate_stage): Likewise.
10118 * ipa-fnsummary.c (redirect_to_unreachable): Likewise.
10119 (edge_set_predicate): Likewise.
10120 (evaluate_conditions_for_known_args): Likewise.
10121 (evaluate_properties_for_edge): Likewise.
10122 (ipa_fn_summary::reset): Likewise.
10123 (ipa_fn_summary_t::duplicate): Likewise.
10124 (dump_ipa_call_summary): Likewise.
10125 (ipa_dump_fn_summary): Likewise.
10126 (analyze_function_body): Likewise.
10127 (compute_fn_summary): Likewise.
10128 (estimate_edge_devirt_benefit): Likewise.
10129 (estimate_edge_size_and_time): Likewise.
10130 (estimate_calls_size_and_time): Likewise.
10131 (estimate_node_size_and_time): Likewise.
10132 (inline_update_callee_summaries): Likewise.
10133 (remap_edge_change_prob): Likewise.
10134 (remap_edge_summaries): Likewise.
10135 (ipa_merge_fn_summary_after_inlining): Likewise.
10136 (ipa_update_overall_fn_summary): Likewise.
10137 (read_ipa_call_summary): Likewise.
10138 (inline_read_section): Likewise.
10139 (write_ipa_call_summary): Likewise.
10140 (ipa_fn_summary_write): Likewise.
10141 (ipa_free_fn_summary): Likewise.
10142 * ipa-hsa.c (process_hsa_functions): Likewise.
10143 (ipa_hsa_write_summary): Likewise.
10144 (ipa_hsa_read_section): Likewise.
10145 * ipa-icf.c (sem_function::merge): Likewise.
10146 * ipa-inline-analysis.c (simple_edge_hints): Likewise.
10147 (do_estimate_edge_time): Likewise.
10148 (estimate_size_after_inlining): Likewise.
10149 (estimate_growth): Likewise.
10150 (growth_likely_positive): Likewise.
10151 * ipa-inline-transform.c (clone_inlined_nodes): Likewise.
10152 (inline_call): Likewise.
10153 * ipa-inline.c (caller_growth_limits): Likewise.
10154 (can_inline_edge_p): Likewise.
10155 (can_inline_edge_by_limits_p): Likewise.
10156 (compute_uninlined_call_time): Likewise.
10157 (compute_inlined_call_time): Likewise.
10158 (want_inline_small_function_p): Likewise.
10159 (edge_badness): Likewise.
10160 (update_caller_keys): Likewise.
10161 (update_callee_keys): Likewise.
10162 (recursive_inlining): Likewise.
10163 (inline_small_functions): Likewise.
10164 (inline_to_all_callers_1): Likewise.
10165 (dump_overall_stats): Likewise.
10166 (early_inline_small_functions): Likewise.
10167 (early_inliner): Likewise.
10168 * ipa-inline.h (estimate_edge_growth): Likewise.
10169 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
10170 * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
10171 * ipa-prop.h (IPA_NODE_REF): Likewise.
10172 (IPA_EDGE_REF): Likewise.
10173 * ipa-pure-const.c (malloc_candidate_p): Likewise.
10174 (propagate_malloc): Likewise.
10175 * ipa-split.c (execute_split_functions): Likewise.
10176 * symbol-summary.h: Rename get to get_create.
10178 (get_create): Likewise.
10179 * tree-sra.c (ipa_sra_preliminary_function_checks): Likewise.
10181 2018-06-08 Martin Liska <mliska@suse.cz>
10183 * symbol-summary.h (release): Move definition out of class
10185 (symtab_removal): Likewise.
10186 (symtab_duplication): Likewise.
10188 2018-06-08 Martin Liska <mliska@suse.cz>
10190 * symbol-summary.h (function_summary): Move constructor
10191 implementation out of class declaration.
10192 (release): Likewise.
10193 (symtab_insertion): Likewise.
10194 (symtab_removal): Likewise.
10195 (symtab_duplication): Likewise.
10198 2018-06-08 Martin Liska <mliska@suse.cz>
10200 * Makefile.in: Remove support for MPX (macros, related functions,
10201 fields in cgraph_node, ...).
10202 * builtin-types.def (BT_BND): Likewise.
10203 (BT_FN_BND_CONST_PTR): Likewise.
10204 (BT_FN_CONST_PTR_BND): Likewise.
10205 (BT_FN_VOID_PTR_BND): Likewise.
10206 (BT_FN_BND_CONST_PTR_SIZE): Likewise.
10207 (BT_FN_VOID_CONST_PTR_BND_CONST_PTR): Likewise.
10208 * builtins.c (expand_builtin_memcpy_with_bounds): Likewise.
10209 (expand_builtin_mempcpy_with_bounds): Likewise.
10210 (expand_builtin_memset_with_bounds): Likewise.
10211 (expand_builtin_memset_args): Likewise.
10212 (std_expand_builtin_va_start): Likewise.
10213 (expand_builtin): Likewise.
10214 (expand_builtin_with_bounds): Likewise.
10215 * builtins.def (DEF_BUILTIN_CHKP): Likewise.
10216 (DEF_LIB_BUILTIN_CHKP): Likewise.
10217 (DEF_EXT_LIB_BUILTIN_CHKP): Likewise.
10218 (DEF_CHKP_BUILTIN): Likewise.
10219 (BUILT_IN_MEMCPY): Likewise.
10220 (BUILT_IN_MEMMOVE): Likewise.
10221 (BUILT_IN_MEMPCPY): Likewise.
10222 (BUILT_IN_MEMSET): Likewise.
10223 (BUILT_IN_STPCPY): Likewise.
10224 (BUILT_IN_STRCAT): Likewise.
10225 (BUILT_IN_STRCHR): Likewise.
10226 (BUILT_IN_STRCPY): Likewise.
10227 (BUILT_IN_STRLEN): Likewise.
10228 (BUILT_IN_MEMCPY_CHK): Likewise.
10229 (BUILT_IN_MEMMOVE_CHK): Likewise.
10230 (BUILT_IN_MEMPCPY_CHK): Likewise.
10231 (BUILT_IN_MEMSET_CHK): Likewise.
10232 (BUILT_IN_STPCPY_CHK): Likewise.
10233 (BUILT_IN_STRCAT_CHK): Likewise.
10234 (BUILT_IN_STRCPY_CHK): Likewise.
10235 * calls.c (store_bounds): Likewise.
10236 (emit_call_1): Likewise.
10237 (special_function_p): Likewise.
10238 (maybe_warn_nonstring_arg): Likewise.
10239 (initialize_argument_information): Likewise.
10240 (finalize_must_preallocate): Likewise.
10241 (compute_argument_addresses): Likewise.
10242 (expand_call): Likewise.
10243 * cfgexpand.c (expand_call_stmt): Likewise.
10244 (expand_return): Likewise.
10245 (expand_gimple_stmt_1): Likewise.
10246 (pass_expand::execute): Likewise.
10247 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
10248 (cgraph_node::remove): Likewise.
10249 (cgraph_node::dump): Likewise.
10250 (cgraph_node::verify_node): Likewise.
10251 * cgraph.h (chkp_function_instrumented_p): Likewise.
10252 (symtab_node::get_alias_target): Likewise.
10253 (cgraph_node::can_remove_if_no_direct_calls_and_refs_p): Likewise.
10254 (cgraph_local_p): Likewise.
10255 * cgraphbuild.c (cgraph_edge::rebuild_edges): Likewise.
10256 (cgraph_edge::rebuild_references): Likewise.
10257 * cgraphunit.c (varpool_node::finalize_decl): Likewise.
10258 (walk_polymorphic_call_targets): Likewise.
10259 (cgraph_node::expand_thunk): Likewise.
10260 (symbol_table::output_weakrefs): Likewise.
10261 * common/config/i386/i386-common.c (OPTION_MASK_ISA2_GENERAL_REGS_ONLY_UNSET): Likewise.
10262 (ix86_handle_option): Likewise.
10263 * config/i386/constraints.md: Likewise.
10264 * config/i386/i386-builtin-types.def (BND): Likewise.
10268 * config/i386/i386-builtin.def (BDESC_END): Likewise.
10269 (BDESC_FIRST): Likewise.
10271 * config/i386/i386-c.c (ix86_target_macros_internal): Likewise.
10272 * config/i386/i386-protos.h (ix86_bnd_prefixed_insn_p): Likewise.
10273 * config/i386/i386.c (enum reg_class): Likewise.
10274 (ix86_target_string): Likewise.
10275 (ix86_option_override_internal): Likewise.
10276 (ix86_conditional_register_usage): Likewise.
10277 (ix86_valid_target_attribute_inner_p): Likewise.
10278 (ix86_set_indirect_branch_type): Likewise.
10279 (ix86_set_current_function): Likewise.
10280 (ix86_function_arg_regno_p): Likewise.
10281 (init_cumulative_args): Likewise.
10282 (ix86_function_arg_advance): Likewise.
10283 (ix86_function_arg): Likewise.
10284 (ix86_pass_by_reference): Likewise.
10285 (ix86_function_value_regno_p): Likewise.
10286 (ix86_function_value_1): Likewise.
10287 (ix86_function_value_bounds): Likewise.
10288 (ix86_return_in_memory): Likewise.
10289 (ix86_setup_incoming_vararg_bounds): Likewise.
10290 (ix86_va_start): Likewise.
10291 (indirect_thunk_need_prefix): Likewise.
10292 (print_reg): Likewise.
10293 (ix86_print_operand): Likewise.
10294 (ix86_expand_call): Likewise.
10295 (ix86_output_function_return): Likewise.
10296 (reg_encoded_number): Likewise.
10297 (BDESC_VERIFYS): Likewise.
10298 (ix86_init_mpx_builtins): Likewise.
10299 (ix86_init_builtins): Likewise.
10300 (ix86_emit_cmove): Likewise.
10301 (ix86_emit_move_max): Likewise.
10302 (ix86_expand_builtin): Likewise.
10303 (ix86_builtin_mpx_function): Likewise.
10304 (ix86_get_arg_address_for_bt): Likewise.
10305 (ix86_load_bounds): Likewise.
10306 (ix86_store_bounds): Likewise.
10307 (ix86_load_returned_bounds): Likewise.
10308 (ix86_store_returned_bounds): Likewise.
10309 (ix86_class_likely_spilled_p): Likewise.
10310 (ix86_hard_regno_mode_ok): Likewise.
10311 (x86_order_regs_for_local_alloc): Likewise.
10312 (ix86_mitigate_rop): Likewise.
10313 (ix86_bnd_prefixed_insn_p): Likewise.
10314 (ix86_mpx_bound_mode): Likewise.
10315 (ix86_make_bounds_constant): Likewise.
10316 (ix86_initialize_bounds): Likewise.
10317 (TARGET_LOAD_BOUNDS_FOR_ARG): Likewise.
10318 (TARGET_STORE_BOUNDS_FOR_ARG): Likewise.
10319 (TARGET_LOAD_RETURNED_BOUNDS): Likewise.
10320 (TARGET_STORE_RETURNED_BOUNDS): Likewise.
10321 (TARGET_CHKP_BOUND_MODE): Likewise.
10322 (TARGET_BUILTIN_CHKP_FUNCTION): Likewise.
10323 (TARGET_CHKP_FUNCTION_VALUE_BOUNDS): Likewise.
10324 (TARGET_CHKP_MAKE_BOUNDS_CONSTANT): Likewise.
10325 (TARGET_CHKP_INITIALIZE_BOUNDS): Likewise.
10326 * config/i386/i386.h (TARGET_MPX): Likewise.
10327 (TARGET_MPX_P): Likewise.
10328 (VALID_BND_REG_MODE): Likewise.
10329 (FIRST_BND_REG): Likewise.
10330 (LAST_BND_REG): Likewise.
10331 (enum reg_class): Likewise.
10332 (BND_REG_P): Likewise.
10333 (BND_REGNO_P): Likewise.
10334 (BNDmode): Likewise.
10335 (ADJUST_INSN_LENGTH): Likewise.
10336 * config/i386/i386.md: Likewise.
10337 * config/i386/i386.opt: Likewise.
10338 * config/i386/linux-common.h (LIBMPX_LIBS): Likewise.
10339 (defined): Likewise.
10340 (LINK_MPX): Likewise.
10341 (MPX_SPEC): Likewise.
10342 (LIBMPX_SPEC): Likewise.
10343 (LIBMPXWRAPPERS_SPEC): Likewise.
10344 (CHKP_SPEC): Likewise.
10345 * config/i386/predicates.md: Likewise.
10346 * dbxout.c (dbxout_type): Likewise.
10347 * doc/extend.texi: Likewise.
10348 * doc/invoke.texi: Likewise.
10349 * doc/md.texi: Likewise.
10350 * doc/tm.texi: Likewise.
10351 * doc/tm.texi.in: Likewise.
10352 * dwarf2out.c (is_base_type): Likewise.
10353 (gen_formal_types_die): Likewise.
10354 (gen_subprogram_die): Likewise.
10355 (gen_type_die_with_usage): Likewise.
10356 (gen_decl_die): Likewise.
10357 (dwarf2out_late_global_decl): Likewise.
10358 * expr.c (expand_assignment): Likewise.
10359 (emit_storent_insn): Likewise.
10360 (store_expr_with_bounds): Likewise.
10361 (store_expr): Likewise.
10362 (expand_expr_real_1): Likewise.
10363 * expr.h (store_expr_with_bounds): Likewise.
10364 * function.c (use_register_for_decl): Likewise.
10365 (struct bounds_parm_data): Likewise.
10366 (assign_parms_augmented_arg_list): Likewise.
10367 (assign_parm_find_entry_rtl): Likewise.
10368 (assign_parm_is_stack_parm): Likewise.
10369 (assign_parm_load_bounds): Likewise.
10370 (assign_bounds): Likewise.
10371 (assign_parms): Likewise.
10372 (expand_function_start): Likewise.
10373 * gcc.c (CHKP_SPEC): Likewise.
10374 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
10375 * gimple-ssa-warn-restrict.c (builtin_access::builtin_access): Likewise.
10376 (wrestrict_dom_walker::check_call): Likewise.
10377 * gimple.c (gimple_build_call_from_tree): Likewise.
10378 * gimple.h (enum gf_mask): Likewise.
10379 (gimple_call_with_bounds_p): Likewise.
10380 (gimple_call_set_with_bounds): Likewise.
10381 * gimplify.c (gimplify_init_constructor): Likewise.
10382 * ipa-cp.c (initialize_node_lattices): Likewise.
10383 (propagate_constants_across_call): Likewise.
10384 (find_more_scalar_values_for_callers_subset): Likewise.
10385 * ipa-hsa.c (process_hsa_functions): Likewise.
10386 * ipa-icf-gimple.c (func_checker::compare_gimple_call): Likewise.
10387 * ipa-icf.c (sem_function::merge): Likewise.
10388 * ipa-inline.c (early_inliner): Likewise.
10389 * ipa-pure-const.c (warn_function_noreturn): Likewise.
10390 (warn_function_cold): Likewise.
10391 (propagate_pure_const): Likewise.
10392 * ipa-ref.h (enum GTY): Likewise.
10393 * ipa-split.c (find_retbnd): Likewise.
10394 (consider_split): Likewise.
10395 (split_function): Likewise.
10396 * ipa-visibility.c (cgraph_externally_visible_p): Likewise.
10397 * ipa.c (walk_polymorphic_call_targets): Likewise.
10398 (symbol_table::remove_unreachable_nodes): Likewise.
10399 (process_references): Likewise.
10400 (cgraph_build_static_cdtor_1): Likewise.
10401 * lto-cgraph.c (lto_output_node): Likewise.
10402 (output_refs): Likewise.
10403 (compute_ltrans_boundary): Likewise.
10404 (input_overwrite_node): Likewise.
10405 (input_node): Likewise.
10406 (input_cgraph_1): Likewise.
10407 * params.def (PARAM_CHKP_MAX_CTOR_SIZE): Likewise.
10408 * passes.c (pass_manager::execute_early_local_passes): Likewise.
10409 (class pass_chkp_instrumentation_passes): Likewise.
10410 (make_pass_chkp_instrumentation_passes): Likewise.
10411 * passes.def: Likewise.
10412 * rtl.h (struct GTY): Likewise.
10413 (CALL_EXPR_WITH_BOUNDS_P): Likewise.
10414 * stor-layout.c (layout_type): Likewise.
10415 * symtab.c: Likewise.
10416 * target.def: Likewise.
10417 * targhooks.c (default_chkp_bound_type): Likewise.
10418 (default_chkp_bound_mode): Likewise.
10419 (default_builtin_chkp_function): Likewise.
10420 (default_chkp_function_value_bounds): Likewise.
10421 (default_chkp_make_bounds_constant): Likewise.
10422 (default_chkp_initialize_bounds): Likewise.
10423 * targhooks.h (default_chkp_bound_type): Likewise.
10424 (default_chkp_bound_mode): Likewise.
10425 (default_builtin_chkp_function): Likewise.
10426 (default_chkp_function_value_bounds): Likewise.
10427 (default_chkp_make_bounds_constant): Likewise.
10428 (default_chkp_initialize_bounds): Likewise.
10429 * toplev.c (compile_file): Likewise.
10430 (process_options): Likewise.
10431 * tree-core.h (DEF_BUILTIN): Likewise.
10432 (DEF_BUILTIN_CHKP): Likewise.
10433 * tree-inline.c (declare_return_variable): Likewise.
10434 (remap_gimple_stmt): Likewise.
10435 (copy_bb): Likewise.
10436 (initialize_inlined_parameters): Likewise.
10437 (expand_call_inline): Likewise.
10438 * tree-pass.h (make_pass_ipa_chkp_versioning): Likewise.
10439 (make_pass_ipa_chkp_early_produce_thunks): Likewise.
10440 (make_pass_ipa_chkp_produce_thunks): Likewise.
10441 (make_pass_chkp): Likewise.
10442 (make_pass_chkp_opt): Likewise.
10443 (make_pass_chkp_instrumentation_passes): Likewise.
10444 * tree-pretty-print.c (dump_generic_node): Likewise.
10445 * tree-ssa-ccp.c (insert_clobber_before_stack_restore): Likewise.
10446 * tree-ssa-dce.c (propagate_necessity): Likewise.
10447 (eliminate_unnecessary_stmts): Likewise.
10448 * tree-ssa-pre.c (create_expression_by_pieces): Likewise.
10449 * tree-ssa-sccvn.c (copy_reference_ops_from_call): Likewise.
10450 * tree-ssa-sccvn.h: Likewise.
10451 * tree-ssa-strlen.c (get_string_length): Likewise.
10452 (valid_builtin_call): Likewise.
10453 (adjust_last_stmt): Likewise.
10454 (handle_builtin_strchr): Likewise.
10455 (handle_builtin_strcpy): Likewise.
10456 (handle_builtin_stxncpy): Likewise.
10457 (handle_builtin_memcpy): Likewise.
10458 (handle_builtin_strcat): Likewise.
10459 (strlen_check_and_optimize_stmt): Likewise.
10460 * tree-stdarg.c (expand_ifn_va_arg_1): Likewise.
10461 * tree-streamer-in.c: Likewise.
10462 * tree-streamer.c (record_common_node): Likewise.
10463 * tree.c (tree_code_size): Likewise.
10464 (wide_int_to_tree_1): Likewise.
10465 (type_contains_placeholder_1): Likewise.
10466 (build_common_tree_nodes): Likewise.
10467 * tree.def (POINTER_BOUNDS_TYPE): Likewise.
10468 * tree.h (POINTER_BOUNDS_TYPE_P): Likewise.
10469 (POINTER_BOUNDS_P): Likewise.
10470 (BOUNDED_TYPE_P): Likewise.
10471 (BOUNDED_P): Likewise.
10472 (CALL_WITH_BOUNDS_P): Likewise.
10473 (pointer_bounds_type_node): Likewise.
10474 * value-prof.c (gimple_ic): Likewise.
10475 * var-tracking.c (vt_add_function_parameters): Likewise.
10476 * varasm.c (make_decl_rtl): Likewise.
10477 (assemble_start_function): Likewise.
10478 (output_constant): Likewise.
10479 (maybe_assemble_visibility): Likewise.
10480 * varpool.c (ctor_for_folding): Likewise.
10481 * chkp-builtins.def: Remove.
10482 * ipa-chkp.c: Remove.
10483 * ipa-chkp.h: Remove.
10484 * rtl-chkp.c: Remove.
10485 * rtl-chkp.h: Remove.
10486 * tree-chkp-opt.c: Remove.
10487 * tree-chkp.c: Remove.
10488 * tree-chkp.h: Remove.
10490 2018-06-07 Carl Love <cel@us.ibm.com>
10492 * config/rs6000/vsx.md (vextract_fp_from_shorth,
10493 vextract_fp_from_shortl): Add BE support.
10495 2018-06-07 Paul Koning <ni1d@arrl.net>
10497 * compare-elim.c (try_merge_compare): Don't merge compare if
10498 address contains a side effect.
10499 (try_eliminate_compare): Likewise.
10501 2018-06-07 Olga Makhotina <olga.makhotina@intel.com>
10503 * config.gcc: Support "tremont".
10504 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "tremont".
10505 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
10507 * config/i386/i386.c (m_TREMONT): Define.
10508 (processor_target_table): Add "tremont".
10509 (PTA_TREMONT): Define.
10510 (ix86_lea_outperforms): Add TARGET_TREMONT.
10511 (get_builtin_code_for_version): Handle PROCESSOR_TREMONT.
10512 (fold_builtin_cpu): Add M_INTEL_TREMONT, replace M_INTEL_GOLDMONT
10513 and M_INTEL_GOLDMONT_PLUS.
10514 (fold_builtin_cpu): Add "tremont".
10515 (ix86_add_stmt_cost): Add TARGET_TREMONT.
10516 (ix86_option_override_internal): Add "tremont".
10517 * config/i386/i386.h (processor_costs): Define TARGET_TREMONT.
10518 (processor_type): Add PROCESSOR_TREMONT.
10519 * config/i386/x86-tune.def: Add m_TREMONT.
10520 * doc/invoke.texi: Add tremont as x86 -march=/-mtune= CPU type.
10522 2018-06-07 Jozef Lawrynowicz <jozef.l@mittosystems.com>
10524 * config/msp430/msp430.c (msp430_mcu_name): Set the "i" in the
10525 symbol defined for msp430i* devices to be lower case.
10527 2018-06-07 Richard Biener <rguenther@suse.de>
10529 * graphite-sese-to-poly.c (extract_affine): Avoid unneded
10530 wrapping. Properly wrap the result of a BIT_NOT_EXPR.
10531 Properly wrap signed arithmetic if overflow wraps.
10533 2018-06-07 Jakub Jelinek <jakub@redhat.com>
10535 PR tree-optimization/69615
10536 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If rhs2 is lhs
10537 of a cast from a same precision integral SSA_NAME in a bb dominated
10538 by first_bb, retry with rhs2 set to the rhs1 of the cast. Don't emit
10539 cast to utype if rhs2 has already a compatible type.
10541 2018-06-07 Richard Biener <rguenther@suse.de>
10543 PR tree-optimization/85935
10544 * graphite-scop-detection.c (find_params_in_bb): Analyze
10545 condition operands with respect to the correct loop. Assert
10546 the analysis doesn't fail.
10548 2018-06-04 Carl Love <cel@us.ibm.com>
10550 * config/rs6000/vsx.md (first_match_index_<mode>): Calculate index
10551 using natural element order. Use gen_lshrsi3 instead of gen_ashrsi3
10552 as it is slightly cheaper.
10553 (first_match_or_eos_index_<mode>):
10554 Calculate index using natural element order.
10555 (first_match_index_<mode>):
10556 Calculate index using natural element order.
10557 (first_match_or_eos_index_<mode>):
10558 Calculate index using natural order.
10559 (define_insn vclzlsbb): Change to define_insn vclzlsbb_<mode>.
10560 for BE and LE modes.
10561 * config/rs6000/rs6000-c.c: Rename P9V_BUILTIN_VCLZLSBB,
10562 P9V_BUILTIN_VCLZLSBB_V16QI.
10563 * config/rs6000/rs6000-builtin.def: Make VCLZLSBB mode
10566 2018-06-06 Kelvin Nilsen <kelvin@gcc.gnu.org>
10568 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
10569 indentation and line wrap for many prototypes. Add missing
10570 @smallexample directives around block of prototypes for vec_xl and
10573 2018-06-05 Michael Meissner <meissner@linux.ibm.com>
10575 * config/rs6000/rs6000.c (rs6000_passes_ieee128): New boolean to
10576 track if we pass or return IEEE 128-bit floating point.
10577 (ieee128_mangling_gcc_8_1): New boolean to say whether to generate
10578 C++ mangling that is compatible with GCC 8.1.
10579 (TARGET_ASM_GLOBALIZE_DECL_NAME): Override target hook.
10580 (init_cumulative_args): Note if we pass or return IEEE 128-bit
10581 floating point types.
10582 (rs6000_function_arg_advance_1): Likewise.
10583 (rs6000_mangle_type): Optionally generate mangled names that match
10584 what GCC 8.1 generated for IEEE 128-bit floating point types.
10585 (rs6000_globalize_decl_name): If we have an external function that
10586 passes or returns IEEE 128-bit types, generate a weak reference
10587 from the mangled name used in GCC 8.1 to the current mangled
10589 (rs6000_init_builtins): Make __ibm128 use the long double type if
10590 long double is IBM extended double. Make __float128 use the long
10591 double type if long double is IEEE 128-bit.
10594 * config/rs6000/rs6000-builtin.def (BU_IBM128_2): New helper
10595 macro for __ibm128 built-in functions.
10596 (PACK_IF): Add __ibm128 pack/unpack functions.
10597 (UNPACK_IF): Likewise.
10598 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Do not
10599 enable long double built-in functions if long double is IEEE
10600 128-bit floating point.
10601 (rs6000_invalid_builtin): Update long double built-in function
10603 (rs6000_expand_builtin): For PACK_IF and UNPACK_IF built-in
10604 functions, adjust the built-in function to use the long double
10605 built-in function if __ibm128 and long double are the same type.
10606 * doc/extend.texi (PowerPC builtins): Update documention for
10607 __builtin_{,un}pack_longdouble. Add documentation for
10608 __builtin_{,un}pack_ibm128.
10610 2018-06-06 Jim Wilson <jimw@sifive.com>
10612 * config/riscv/riscv.c (enum riscv_privilege_levels): New.
10613 (struct machine_function): New field interrupt_mode.
10614 (riscv_handle_type_attribute): New function. Add forward declaration.
10615 (riscv_attribute_table) <interrupt>: Use riscv_handle_type_attribute.
10616 (riscv_expand_epilogue): Check interrupt_mode field.
10617 (riscv_set_current_function): Check interrupt attribute args and
10618 set interrupt_mode field.
10619 * config/riscv/riscv.md (UNSPECV_SRET, UNSPECV_URET): New.
10620 (riscv_sret, riscv_uret): New.
10621 * doc/extend.texi (RISC-V Function Attributes) <interrupt>: Document
10622 new arguments to interrupt attribute.
10624 2018-06-06 Peter Bergner <bergner@vnet.ibm.com>
10627 * /config/rs6000/rs6000.h (ASM_CPU_SPEC): Add support for -mpower9.
10628 Don't handle -mcpu=power8 if -mpower9-vector is also used.
10630 2018-06-06 Kelvin Nilsen <kelvin@gcc.gnu.org>
10632 * config/rs6000/rs6000-builtin.def (VSX_BUILTIN_VEC_LD,
10633 VSX_BUILTIN_VEC_ST): Add comment to explain non-traditional uses.
10634 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
10635 several redundant entries.
10637 2018-06-06 David Malcolm <dmalcolm@redhat.com>
10639 * config/i386/i386-protos.h (ix86_expand_call): Strengthen return
10640 type from "rtx" to "rtx_insn *".
10641 * config/i386/i386.c (ix86_expand_split_stack_prologue): Likewise
10642 for local "call_insn", removing cast.
10643 (ix86_expand_call): Likewise, introducing a "call_insn" local.
10645 2018-06-06 Eric Botcazou <ebotcazou@adacore.com>
10647 PR tree-optimization/86066
10648 * gimple-ssa-store-merging.c (process_store): Do not bypass BIT_NOT_EXPR
10649 for BIT_INSERT_EXPR stores.
10651 2018-06-06 Richard Biener <rguenther@suse.de>
10653 PR tree-optimization/86062
10654 * tree-ssa-sccvn.c (vn_reference_lookup_3): Handle arbitrary
10655 component refs ontop
10656 of to be offsetted base.
10658 2018-06-06 Jozef Lawrynowicz <jozef.l@mittosystems.com>
10660 * config/msp430/msp430.c (msp430_attr): Allow interrupt handlers
10661 to be static and remove check on interrupt attribute name.
10663 2018-06-05 Kelvin Nilsen <kelvin@gcc.gnu.org>
10665 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Remove
10666 volatile qualifier from vec_lvsl and vec_lvsr argument prototypes.
10668 2018-06-05 Steve Ellcey <sellcey@cavium.com>
10671 * config/aarch64/aarch64-protos.h (aarch64_err_no_fpadvsimd): Remove
10673 * config/aarch64/aarch64-protos..c (aarch64_err_no_fpadvsimd):
10674 Remove second argument, change how error is called.
10675 (aarch64_layout_arg): Remove second argument from
10676 aarch64_err_no_fpadvsimd call.
10677 (aarch64_init_cumulative_args): Ditto.
10678 (aarch64_gimplify_va_arg_expr): Ditto.
10679 * config/aarch64/aarch64.md (mov<mode>): Ditto.
10681 2018-06-05 Uros Bizjak <ubizjak@gmail.com>
10683 * config/i386/i386.md (simple_return_indirect_internal): New expander.
10684 (*simple_return_indirect_internal<mode>): Rename from
10685 simple_return_indirect_internal. Use W mode iterator.
10686 (rstorssp): New expander.
10687 (*rstorssp<mode>): Rename from rstorssp. Use P mode iterator.
10688 (clrssbsy): New expander.
10689 (*clrssbsy<mode>): Rename from clrssbsy. Use P mode iterator.
10691 2018-06-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
10693 * config/arm/arm_cmse.h (cmse_nsfptr_create): Change typeof to
10695 (cmse_check_pointed_object): Likewise.
10697 2018-06-05 Martin Liska <mliska@suse.cz>
10699 PR gcov-profile/47618
10700 * doc/invoke.texi: Document how -fprofile-dir format
10703 2018-06-05 Richard Biener <rguenther@suse.de>
10705 * tree-cfgcleanup.c (cleanup_control_flow_pre): For edge
10706 removal pretend DOM info isn't available so we do not update
10707 it and only remove edges, not dominated blocks. Actually free
10708 DOM info in case we removed something. Remove unreachable blocks.
10709 (mfb_keep_latches): Work with either DOM info or marked backedges.
10710 (cleanup_tree_cfg_noloop): Do not remove unreachable blocks
10711 first. Mark backedges if DOM info isn't available.
10712 (Re-)compute DOM info after cleanup_control_flow_pre.
10714 2018-06-05 Richard Biener <rguenther@suse.de>
10716 * tree-cfg.c (struct locus_discrim_map): Store line, not location.
10717 (locus_discrim_hasher::hash): Adjust.
10718 (locus_discrim_hasher::equal): Likewise.
10719 (next_discriminator_for_locus): Work on line directly.
10720 (same_line_p): Pass in expanded locus1 as well.
10721 (assign_discriminators): Avoid redundant location expansions.
10723 2018-06-05 Richard Biener <rguenther@suse.de>
10725 PR tree-optimization/86046
10726 * tree-ssa.c (maybe_optimize_var): Clear DECL_GIMPLE_REG_P
10727 if required after clearing TREE_ADDRESSABLE.
10729 2018-06-05 Richard Biener <rguenther@suse.de>
10731 PR tree-optimization/86047
10732 * tree-ssa-loop.c (for_each_index): Glob handling of all
10733 decls and constants and really handle all of them.
10735 2018-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10738 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
10739 qualifier_void_pointer and qualifier_const_void_pointer.
10740 (arm_ldc_qualifiers, arm_stc_qualifiers): Use the above.
10741 (arm_init_builtins): Handle the above.
10742 * config/arm/arm_acle.h (__arm_cdp, __arm_ldc, __arm_ldcl, __arm_stc,
10743 __arm_stcl, __arm_mcr, __arm_cdp2, __arm_ldc2, __arm_ldcl2, __arm_stc2,
10744 __arm_stcl2,__arm_mcr2, __arm_mcrr, __arm_mcrr2): Remove return for
10747 2018-06-05 Martin Liska <mliska@suse.cz>
10749 * auto-profile.c (read_autofdo_file): Do not use
10750 gcov_ctr_summary struct.
10751 (afdo_callsite_hot_enough_for_early_inline): Likewise.
10752 * coverage.c (struct counts_entry): Likewise.
10753 (read_counts_file): Read just single summary entry.
10754 (get_coverage_counts): Use gcov_summary struct.
10755 * coverage.h (get_coverage_counts): Likewise.
10756 * gcov-dump.c (dump_working_sets): Likewise.
10757 (tag_summary): Dump just single summary.
10758 * gcov-io.c (gcov_write_summary): Write just histogram
10760 (gcov_read_summary): Read just single summary.
10761 (compute_working_sets): Use gcov_summary struct.
10762 * gcov-io.h (GCOV_TAG_SUMMARY_LENGTH): Remove usage
10763 of GCOV_COUNTERS_SUMMABLE.
10764 (GCOV_COUNTERS_SUMMABLE): Remove.
10765 (GCOV_FIRST_VALUE_COUNTER): Replace with
10766 GCOV_COUNTER_V_INTERVAL.
10767 (struct gcov_ctr_summary): Remove.
10768 (struct gcov_summary): Directly use fields of former
10770 (compute_working_sets): Use gcov_summary struct.
10771 * gcov.c (read_count_file): Do not use ctrs fields.
10772 * lto-cgraph.c (merge_profile_summaries): Use gcov_summary
10774 * lto-streamer.h (struct GTY): Make profile_info gcov_summary
10776 * profile.c: Likewise.
10777 * profile.h: Likewise.
10779 2018-06-05 Martin Liska <mliska@suse.cz>
10781 PR gcov-profile/84846
10782 * gcov.c (output_lines): Print working directory only
10783 in intermediate format.
10785 2018-06-05 Andreas Krebbel <krebbel@linux.ibm.com>
10787 * config/s390/s390-builtin-types.def: Add void function type.
10788 * config/s390/s390-builtins.def: Use the function type for the
10791 2018-06-04 Jim Wilson <jimw@sifive.com>
10793 * config/riscv/riscv-protos.h (riscv_expand_epilogue): Change bool arg
10795 * config/riscv/riscv.c (riscv_for_each_saved_reg): New args epilogue
10796 and maybe_eh_return. Change regno to unsigned int. Use new args to
10797 handle EH_RETURN_DATA_REGNO registers properly.
10798 (riscv_expand_prologue): Pass new args to riscv_for_each_saved_reg.
10799 (riscv_expand_epilogue): Update comment. Change argument name and
10800 type. Update code to use new name and type. Pass new args to
10801 riscv_for_each_saved_reg. Only use EH_RETURN_STACKADJ_RTX when
10803 * config/riscv/riscv.md (NORMAL_RETURN): New.
10804 (SIBCALL_RETURN, EXCEPTION_RETURN): New.
10805 (epilogue, sibcall_epilogue): Update riscv_expand_epilogue arg.
10806 (eh_return): Call gen_eh_return_internal and emit barrier.
10807 (eh_return_internal): Call riscv_expand_epilogue.
10809 2018-06-04 Eric Botcazou <ebotcazou@adacore.com>
10811 * gimple-ssa-store-merging.c (struct merged_store_group): Move up
10812 bit_insertion field and declare can_be_merged_into method.
10813 (merged_store_group::can_be_merged_into): New method.
10814 (imm_store_chain_info::coalesce_immediate): Call it to decide whether
10815 consecutive non-overlapping stores can be merged. Turn MEM_REF stores
10816 into BIT_INSERT_EXPR stores if the group contains a non-MEM_REF store.
10818 2018-06-04 Richard Biener <rguenther@suse.de>
10820 PR tree-optimization/85955
10821 * builtins.c (fold_builtin_sincos): Convert pointers to
10822 destination to appropriate type before dereferencing.
10824 2018-06-04 Segher Boessenkool <segher@kernel.crashing.org>
10826 * config/rs6000/rs6000.md (abs<mode>2 for FLOAT128): Handle IFmode.
10828 2018-06-04 Richard Sandiford <richard.sandiford@linaro.org>
10830 * expr.c (expand_expr_real_1): Force the operand into memory if
10831 its TYPE_MODE is BLKmode and if there is no integer mode for
10832 the number of bits being extracted.
10834 2018-06-04 Jakub Jelinek <jakub@redhat.com>
10838 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>_1):
10839 Use vptestnm rather than vptestm in (=Yc,v,C) variant.
10841 2018-06-04 Richard Biener <rguenther@suse.de>
10843 * tree-cfgcleanup.c (cleanup_tree_cfg_1): Fold into...
10844 (cleanup_tree_cfg_noloop): ... single caller. Do
10845 start_recording_case_labels later.
10847 2018-06-04 Sebastian Peryt <sebastian.peryt@intel.com>
10849 * config/i386/cldemoteintrin.h: Change define from _X86INTRIN_H_INCLUDED
10850 to _IMMINTRIN_H_INCLUDED.
10851 * config/i386/pconfigintrin.h: Ditto.
10852 * config/i386/waitpkgintrin.h: Ditto.
10853 * config/i386/immintrin.h: Add includes for sgxintrin.h,
10854 pconfigintrin.h, waitpkgintrin.h and cldemoteintrin.h.
10855 * config/i386/x86intrin.h: Remove includes for mintrin.h, xmmintrin.h,
10856 emmintrin.h, pmmintrin.h, tmmintrin.h, smmintrin.h, wmmintrin.h,
10857 bmiintrin.h, bmi2intrin.h, lzcntintrin.h, sgxintrin.h, pconfigintrin.h,
10858 waitpkgintrin.h and cldemoteintrin.h.
10860 2018-06-04 Richard Biener <rguenther@suse.de>
10862 PR tree-optimization/86038
10863 * tracer.c (find_best_successor): Check probability for
10864 being initialized, bail out if not.
10866 2018-06-04 Richard Earnshaw <rearnsha@arm.com>
10869 * config/arm/arm-cpus.in (ALL_QUIRKS): Add xscale feature to the list
10870 of bits to ignore when comparing architectures.
10872 2018-06-04 Jakub Jelinek <jakub@redhat.com>
10874 PR tree-optimization/69615
10875 * fold-const.c (merge_ranges): If range1 is - [x, x] and x is the
10876 maximum or minimum of the type, try to merge it also as if
10877 range1 is + [-, x - 1] or + [x + 1, -].
10880 * tree.c (inchash::add_expr): Handle IDENTIFIER_NODE.
10882 2018-06-03 Eric Botcazou <ebotcazou@adacore.com>
10884 PR tree-optimization/86034
10885 * gimple-ssa-store-merging.c (output_merged_store): Convert the RHS to
10886 the unsigned bitfield type in a bit insertion sequence if it does not
10887 have a larger precision than the bitfield size.
10888 (process_store): Also bypass widening conversions for BIT_INSERT_EXPR.
10890 2018-06-03 Kito Cheng <kito.cheng@gmail.com>
10892 * config/nds32/nds32-peephole2.md: Add new patterns for code size.
10894 2018-06-03 Chung-Ju Wu <jasonwucj@gmail.com>
10896 * config/nds32/nds32-opts.h (nds32_arch_type): Add ARCH_V3J.
10897 * config/nds32/nds32.c (nds32_option_override): Consider ARCH_V3J.
10898 * config/nds32/nds32.h (TARGET_ISA_V3): Add ARCH_V3J checking.
10899 * config/nds32/nds32.opt (march): Add enum value Value(ARCH_V3J).
10901 2018-06-03 Chung-Ju Wu <jasonwucj@gmail.com>
10903 * common/config/nds32/nds32-common.c (nds32_option_optimization_table):
10904 Disable -fdelete-null-pointer-checks for ELF toolchain.
10906 2018-06-02 Chung-Ju Wu <jasonwucj@gmail.com>
10907 Kito Cheng <kito.cheng@gmail.com>
10909 * config.gcc (nds32*): Use nds32-linux.opt and nds32-elf.opt.
10910 (nds32le-*-*, nds32be-*-*): Integrate checking process.
10911 (nds32*-*-*): Add glibc and uclibc conditions.
10912 * common/config/nds32/nds32-common.c (nds32_except_unwind_info): New.
10913 (TARGET_EXCEPT_UNWIND_INFO): Define.
10914 * config/nds32/elf.h: New file.
10915 * config/nds32/linux.h: New file.
10916 * config/nds32/nds32-elf.opt: New file.
10917 * config/nds32/nds32-linux.opt: New file.
10918 * config/nds32/nds32-fp-as-gp.c
10919 (pass_nds32_fp_as_gp::gate): Consider TARGET_LINUX_ABI.
10920 * config/nds32/nds32.c (nds32_conditional_register_usage): Consider
10922 (nds32_asm_file_end): Ditto.
10923 (nds32_print_operand): Ditto.
10924 (nds32_insert_attributes): Ditto.
10925 (nds32_init_libfuncs): New function.
10926 (TARGET_HAVE_TLS): Define.
10927 (TARGET_INIT_LIBFUNCS): Define.
10928 * config/nds32/nds32.h (TARGET_DEFAULT_RELAX): Apply different relax
10930 (TARGET_ELF): Apply different mcmodel setting.
10931 (LINK_SPEC, LIB_SPEC, STARTFILE_SPEC, ENDFILE_SPEC): The content has
10932 been migrated into elf.h and linux.h files.
10933 * config/nds32/nds32.md (add_pc): Consider TARGET_LINUX_ABI.
10934 * config/nds32/nds32.opt (mvh): Consider TARGET_LINUX_ABI.
10935 (mcmodel): The content has been migrated into nds32-elf.opt and
10936 nds32-linux.opt files.
10937 * config/nds32/t-elf: New file.
10938 * config/nds32/t-linux: New file.
10940 2018-06-02 Chung-Ju Wu <jasonwucj@gmail.com>
10941 Shiva Chen <shiva0217@gmail.com>
10943 * config/nds32/constants.md (unspec_volatile_element): Add
10944 UNSPEC_VOLATILE_OMIT_FP_BEGIN and UNSPEC_VOLATILE_OMIT_FP_END.
10945 * config/nds32/nds32-fp-as-gp.c: New implementation of fp_as_gp
10947 * config/nds32/nds32-protos.h (nds32_naked_function_p): Declare.
10948 (make_pass_nds32_fp_as_gp): Declare.
10949 * config/nds32/nds32.c (nds32_register_passes): Add fp_as_gp as one
10951 (nds32_asm_function_end_prologue): Remove unused asm output.
10952 (nds32_asm_function_begin_epilogue): Remove unused asm output.
10953 (nds32_asm_file_start): Output necessary fp_as_gp information.
10954 (nds32_option_override): Adjust register usage.
10955 (nds32_expand_prologue): Consider fp_as_gp situation.
10956 (nds32_expand_prologue_v3push): Consider fp_as_gp situation.
10957 * config/nds32/nds32.md (prologue): Check fp_as_gp_p and naked_p.
10960 (simple_return): Ditto.
10961 (omit_fp_begin): Output special directive for fp_as_gp.
10962 (omit_fp_end): Output special directive for fp_as_gp.
10963 * config/nds32/nds32.opt (mfp-as-gp, mno-fp-as-gp, mforce-fp-as-gp,
10964 mforbid-fp-as-gp): New options.
10966 2018-06-01 Mark Wielaard <mark@klomp.org>
10968 * dwarf2out.c (dwarf2out_finish): Remove generation of
10969 DW_AT_loclists_base.
10971 2018-06-01 Eric Botcazou <ebotcazou@adacore.com>
10973 * gimple-ssa-store-merging.c: Include gimple-fold.h.
10974 (struct store_immediate_info): Document BIT_INSERT_EXPR stores.
10975 (struct merged_store_group): Add bit_insertion field.
10976 (dump_char_array): Use standard hexadecimal format.
10977 (merged_store_group::merged_store_group): Set bit_insertion to false.
10978 (merged_store_group::apply_stores): Use optimal buffer size. Deal
10979 with BIT_INSERT_EXPR stores. Move up code updating the mask and
10980 also print the mask in the dump file.
10981 (pass_store_merging::gate): Minor tweak.
10982 (imm_store_chain_info::coalesce_immediate): Fix wrong association
10983 of stores with groups in dump. Allow coalescing of BIT_INSERT_EXPR
10984 stores with INTEGER_CST stores.
10985 (count_multiple_uses) <BIT_INSERT_EXPR>: New case.
10986 (imm_store_chain_info::output_merged_store): Add try_bitpos variable
10987 and use it throughout. Generate bit insertion sequences if need be.
10988 (pass_store_merging::process_store): Remove redundant condition.
10989 Record stores from a SSA name to a bit-field with BIT_INSERT_EXPR.
10991 2018-06-01 Segher Boessenkool <segher@kernel.crashing.org>
10993 * config/rs6000/rs6000.c (rs6000_mangle_type): Change the mangling of
10994 the 128-bit floating point types. Fix function comment.
10996 2018-06-01 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10998 * config/aarch64/aarch64-simd.md
10999 (aarch64_simd_vec_unpack<su>_lo_<mode>): Use UXTL and SXTL assembler
11001 (aarch64_simd_vec_unpack<su>_hi_<mode>): Use UXTL2 and SXTL2 assembler
11004 2018-06-01 Richard Sandiford <richard.sandiford@linaro.org>
11006 PR tree-optimization/85989
11007 * gimple-ssa-backprop.c (backprop::m_visited_phis): New member
11009 (backprop::intersect_uses): Check it when deciding whether this
11010 is a backedge reference.
11011 (backprop::process_block): Add each phi to m_visited_phis
11012 after visiting it, then clear it at the end.
11014 2018-06-01 Richard Biener <rguenther@suse.de>
11016 * tree-vectorizer.h (vect_dr_stmt): New function.
11017 (vect_get_load_cost): Adjust.
11018 (vect_get_store_cost): Likewise.
11019 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence):
11020 Use vect_dr_stmt instead of DR_SMTT.
11021 (vect_record_base_alignments): Likewise.
11022 (vect_calculate_target_alignment): Likewise.
11023 (vect_compute_data_ref_alignment): Likewise and make static.
11024 (vect_update_misalignment_for_peel): Likewise.
11025 (vect_verify_datarefs_alignment): Likewise.
11026 (vector_alignment_reachable_p): Likewise.
11027 (vect_get_data_access_cost): Likewise. Pass down
11028 vinfo to vect_get_load_cost/vect_get_store_cost instead of DR.
11029 (vect_get_peeling_costs_all_drs): Likewise.
11030 (vect_peeling_hash_get_lowest_cost): Likewise.
11031 (vect_enhance_data_refs_alignment): Likewise.
11032 (vect_find_same_alignment_drs): Likewise.
11033 (vect_analyze_data_refs_alignment): Likewise.
11034 (vect_analyze_group_access_1): Likewise.
11035 (vect_analyze_group_access): Likewise.
11036 (vect_analyze_data_ref_access): Likewise.
11037 (vect_analyze_data_ref_accesses): Likewise.
11038 (vect_vfa_segment_size): Likewise.
11039 (vect_small_gap_p): Likewise.
11040 (vectorizable_with_step_bound_p): Likewise.
11041 (vect_prune_runtime_alias_test_list): Likewise.
11042 (vect_analyze_data_refs): Likewise.
11043 (vect_supportable_dr_alignment): Likewise.
11044 * tree-vect-loop-manip.c (get_misalign_in_elems): Likewise.
11045 (vect_gen_prolog_loop_niters): Likewise.
11046 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
11047 * tree-vect-patterns.c (vect_recog_bool_pattern): Do not
11049 (vect_recog_mask_conversion_pattern): Likewise.
11050 (vect_try_gather_scatter_pattern): Likewise.
11051 * tree-vect-stmts.c (vect_model_store_cost): Pass stmt_info
11052 to vect_get_store_cost.
11053 (vect_get_store_cost): Get stmt_info instead of DR.
11054 (vect_model_load_cost): Pass stmt_info to vect_get_load_cost.
11055 (vect_get_load_cost): Get stmt_info instead of DR.
11057 2018-06-01 Richard Biener <rguenther@suse.de>
11059 PR middle-end/86017
11060 * gimple-fold.c (var_decl_component_p): Also allow offsetted
11061 vars wrapped in MEM_REFs.
11063 2018-06-01 Richard Sandiford <richard.sandiford@linaro.org>
11065 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
11066 Fix subreg tests so that we only return a choice between
11067 GENERAL_REGS and FP_REGS if the original classes included both.
11069 2018-06-01 Richard Biener <rguenther@suse.de>
11072 * tree-ssa-structalias.c (get_function_part_constraint):
11073 Handle NULL fi->decl.
11074 (find_func_aliases_for_call): Properly handle indirect
11075 fi from direct call.
11076 (find_func_clobbers): Likewise.
11077 (ipa_pta_execute): Likewise.
11078 (create_variable_info_for): For functions that are ifunc_resolver
11079 resolve to a varinfo that contains the result of the resolver call.
11080 (associate_varinfo_to_alias): Do not treat ifunc resolvers as
11083 2018-05-31 Michael Collison <michael.collison@arm.com>
11085 * config/aarch64/aarch64.md:
11086 (*fix_to_zero_extenddfdi2): New pattern.
11087 * gcc.target/aarch64/fix_extend1.c: New testcase.
11089 2018-05-31 Qing Zhao <qing.zhao@oracle.com>
11091 PR middle-end/78809
11092 PR middle-end/83026
11093 * builtins.c (expand_builtin): Add the handling of BUILT_IN_STRCMP_EQ
11094 and BUILT_IN_STRNCMP_EQ.
11095 * builtins.def: Add new builtins BUILT_IN_STRCMP_EQ and
11096 BUILT_IN_STRNCMP_EQ.
11097 * gimple-fold.c (gimple_fold_builtin_string_compare): Add the
11098 handling of BUILTIN_IN_STRCMP_EQ and BUILT_IN_STRNCMP_EQ.
11099 (gimple_fold_builtin): Likewise.
11100 * tree-ssa-strlen.c (compute_string_length): New function.
11101 (determine_min_obsize): New function.
11102 (handle_builtin_string_cmp): New function to handle calls to
11103 string compare functions.
11104 (strlen_optimize_stmt): Add handling to builtin string compare
11106 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
11107 Add the handling of BUILT_IN_STRCMP_EQ and BUILT_IN_STRNCMP_EQ.
11108 * tree.c (build_common_builtin_nodes): Add new defines of
11109 BUILT_IN_STRNCMP_EQ and BUILT_IN_STRCMP_EQ.
11111 2018-05-31 Jakub Jelinek <jakub@redhat.com>
11114 * bb-reorder.c (pass_partition_blocks::gate): Return false for
11115 functions with naked attribute.
11117 2018-05-31 Uros Bizjak <ubizjak@gmail.com>
11119 * config/i386/sse.md (avx_vec_concat<mode>):
11120 Substitute concat_tg_mode mode attribute with xtg_mode.
11121 (<mask_codefor>avx512dq_broadcast<mode><mask_name>_1): Ditto.
11122 (concat_tg_mode): Remove mode attribute.
11124 2018-05-31 Martin Sebor <msebor@redhat.com>
11127 * calls.c (alloc_max_size): Correct a logic error/typo.
11128 Treat excessive arguments as infinite. Warn for invalid arguments.
11129 * doc/invoke.texi (-Walloc-size-larger-than): Update.
11131 2018-05-31 H.J. Lu <hongjiu.lu@intel.com>
11134 * config/i386/x86-tune.def: Re-enable partial_reg_dependency
11135 and movx for Haswell.
11137 2018-05-31 Chung-Lin Tang <cltang@codesourcery.com>
11138 Cesar Philippidis <cesar@codesourcery.com>
11140 PR middle-end/85879
11141 * gimplify.c (gimplify_adjust_omp_clauses): Add 'remove = true'
11142 when emitting error on private/firstprivate reductions.
11143 * omp-low.c (lower_omp_target): Avoid reference-type processing
11144 on pointers for firstprivate clause.
11146 2018-05-31 Sameera Deshpande <sameera.deshpande@linaro.org>
11148 * config/aarch64/aarch64-simd-builtins.def (ld1x3): New.
11151 * config/aarch64/aarch64-simd.md
11152 (aarch64_ld1x3<VALLDIF:mode>): New pattern.
11153 (aarch64_ld1_x3_<mode>): Likewise
11154 (aarch64_st1x2<VALLDIF:mode>): Likewise
11155 (aarch64_st1_x2_<mode>): Likewise
11156 (aarch64_st1x3<VALLDIF:mode>): Likewise
11157 (aarch64_st1_x3_<mode>): Likewise
11158 * config/aarch64/arm_neon.h (vld1_u8_x3): New function.
11159 (vld1_s8_x3): Likewise.
11160 (vld1_u16_x3): Likewise.
11161 (vld1_s16_x3): Likewise.
11162 (vld1_u32_x3): Likewise.
11163 (vld1_s32_x3): Likewise.
11164 (vld1_u64_x3): Likewise.
11165 (vld1_s64_x3): Likewise.
11166 (vld1_f16_x3): Likewise.
11167 (vld1_f32_x3): Likewise.
11168 (vld1_f64_x3): Likewise.
11169 (vld1_p8_x3): Likewise.
11170 (vld1_p16_x3): Likewise.
11171 (vld1_p64_x3): Likewise.
11172 (vld1q_u8_x3): Likewise.
11173 (vld1q_s8_x3): Likewise.
11174 (vld1q_u16_x3): Likewise.
11175 (vld1q_s16_x3): Likewise.
11176 (vld1q_u32_x3): Likewise.
11177 (vld1q_s32_x3): Likewise.
11178 (vld1q_u64_x3): Likewise.
11179 (vld1q_s64_x3): Likewise.
11180 (vld1q_f16_x3): Likewise.
11181 (vld1q_f32_x3): Likewise.
11182 (vld1q_f64_x3): Likewise.
11183 (vld1q_p8_x3): Likewise.
11184 (vld1q_p16_x3): Likewise.
11185 (vld1q_p64_x3): Likewise.
11186 (vst1_s64_x2): Likewise.
11187 (vst1_u64_x2): Likewise.
11188 (vst1_f64_x2): Likewise.
11189 (vst1_s8_x2): Likewise.
11190 (vst1_p8_x2): Likewise.
11191 (vst1_s16_x2): Likewise.
11192 (vst1_p16_x2): Likewise.
11193 (vst1_s32_x2): Likewise.
11194 (vst1_u8_x2): Likewise.
11195 (vst1_u16_x2): Likewise.
11196 (vst1_u32_x2): Likewise.
11197 (vst1_f16_x2): Likewise.
11198 (vst1_f32_x2): Likewise.
11199 (vst1_p64_x2): Likewise.
11200 (vst1q_s8_x2): Likewise.
11201 (vst1q_p8_x2): Likewise.
11202 (vst1q_s16_x2): Likewise.
11203 (vst1q_p16_x2): Likewise.
11204 (vst1q_s32_x2): Likewise.
11205 (vst1q_s64_x2): Likewise.
11206 (vst1q_u8_x2): Likewise.
11207 (vst1q_u16_x2): Likewise.
11208 (vst1q_u32_x2): Likewise.
11209 (vst1q_u64_x2): Likewise.
11210 (vst1q_f16_x2): Likewise.
11211 (vst1q_f32_x2): Likewise.
11212 (vst1q_f64_x2): Likewise.
11213 (vst1q_p64_x2): Likewise.
11214 (vst1_s64_x3): Likewise.
11215 (vst1_u64_x3): Likewise.
11216 (vst1_f64_x3): Likewise.
11217 (vst1_s8_x3): Likewise.
11218 (vst1_p8_x3): Likewise.
11219 (vst1_s16_x3): Likewise.
11220 (vst1_p16_x3): Likewise.
11221 (vst1_s32_x3): Likewise.
11222 (vst1_u8_x3): Likewise.
11223 (vst1_u16_x3): Likewise.
11224 (vst1_u32_x3): Likewise.
11225 (vst1_f16_x3): Likewise.
11226 (vst1_f32_x3): Likewise.
11227 (vst1_p64_x3): Likewise.
11228 (vst1q_s8_x3): Likewise.
11229 (vst1q_p8_x3): Likewise.
11230 (vst1q_s16_x3): Likewise.
11231 (vst1q_p16_x3): Likewise.
11232 (vst1q_s32_x3): Likewise.
11233 (vst1q_s64_x3): Likewise.
11234 (vst1q_u8_x3): Likewise.
11235 (vst1q_u16_x3): Likewise.
11236 (vst1q_u32_x3): Likewise.
11237 (vst1q_u64_x3): Likewise.
11238 (vst1q_f16_x3): Likewise.
11239 (vst1q_f32_x3): Likewise.
11240 (vst1q_f64_x3): Likewise.
11241 (vst1q_p64_x3): Likewise.
11243 2018-05-30 Jozef Lawrynowicz <jozef.l@mittosystems.com>
11245 * config/msp430/msp430.c (msp430_output_labelref): Prepend
11246 user_label_prefix to name.
11248 * tree-core.h: Update comment about the format of NAME string
11249 passed to handler in attribute_spec.
11251 * config/msp430/msp430.md: Remove erroneous subreg expression from
11252 zero_extendqisi2 insn pattern. Remove msp430x ISA restriction on
11253 zero_extend{q,h}isi2.
11255 2018-05-30 Borislav Petkov <bp@suse.de>
11257 * doc/extend.texi: Document some architecture specific
11258 constraints and sort entries.
11260 2018-05-30 Martin Sebor <msebor@redhat.com>
11262 PR middle-end/85369
11263 * builtins.c (expand_builtin_stpcpy_1): New function.
11264 (expand_builtin_stpcpy): Call it, and call maybe_warn_nonstring_arg
11265 only if the former succeeds.
11267 2018-05-31 Sameera Deshpande <sameera.deshpande@linaro.org>
11269 * config/aarch64/aarch64-cores.def (saphira) : Add support for ARMv8.4
11272 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
11274 * doc/invoke.texi (-flinker-output): Document
11276 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
11278 * passes.c (ipa_write_summaries): Only modify statements if body
11280 * cgraphunit.c (ipa_passes): Also produce intermeidate code when
11281 incrementally linking.
11282 (ipa_passes): Likewise.
11283 * lto-cgraph.c (lto_output_node): When incrementally linking do not
11284 pass down resolution info.
11285 * common.opt (flag_incremental_link): Update info.
11286 * gcc.c (plugin specs): Turn flinker-output=* to
11287 -plugin-opt=-linker-output-known
11288 * toplev.c (compile_file): Also cut compilation when doing incremental
11290 * flag-types. (enum lto_partition_model): Add
11291 LTO_LINKER_OUTPUT_NOLTOREL.
11292 (invoke.texi): Add -flinker-output docs.
11293 * ipa.c (symbol_table::remove_unreachable_nodes): Handle LTO incremental
11294 link same way as WPA; do not stream in dead initializers.
11296 * dwarf2out.c (dwarf2out_die_ref_for_decl,
11297 darf2out_register_external_decl): Support incremental link.
11299 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
11301 * lto-opts.c (lto_write_options): Skip OPT_dumpdir, OPT_fresolution_.
11303 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
11305 * lto-wrapper.c (debug_objcopy): Add rename parameter; pass
11306 it down to simple_object_copy_lto_debug_sections.
11307 (run_gcc): Determine incremental LTO link time and configure
11308 lto1 into non-wpa mode, disable renaming of debug sections.
11310 2018-05-30 Kelvin Nilsen <kelvin@gcc.gnu.org>
11312 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Remove
11313 descriptions of various incorrectly documented functions.
11315 2018-05-30 Andre Vieira <andre.simoesdiasvieira@arm.com>
11318 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Make
11319 address check not strict.
11321 2018-05-30 Richard Biener <rguenther@suse.de>
11323 PR tree-optimization/85964
11324 * tracer.c (better_p): Drop initialized count check, we only
11325 call the function with initialized counts now.
11326 (find_best_successor): Do find a best edge if one
11327 has uninitialized count.
11328 (find_best_predecessor): Likewise. Do BB frequency check only
11329 if count is initialized.
11331 2017-05-30 Jackson Woodruff <jackson.woodruff@arm.com>
11333 * config/aarch64/aarch64.c (aarch64_host_wide_int_compare): New.
11334 (aarch64_ldrstr_offset_compare): New.
11335 (aarch64_operands_adjust_ok_for_ldpstp): Update to consider all
11336 load/store orderings.
11337 (aarch64_gen_adjusted_ldpstp): Likewise.
11339 2018-05-30 Wilco Dijkstra <wdijkstr@arm.com>
11341 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
11342 Check for subset of GENERAL_REGS and FP_REGS.
11343 * config/aarch64/aarch64-simd.md (aarch64_get_lane): Increase cost of
11346 2018-05-30 Richard Sandiford <richard.sandiford@linaro.org>
11348 * alias.c (adjust_offset_for_component_ref): Use poly_int_tree_p
11349 and wi::to_poly_offset. Add the current offset and then check
11350 whether the sum fits, rather than using an unchecked addition of
11351 a checked term. Check for a shwi rather than a uhwi.
11352 * expr.c (get_bit_range): Use tree_to_poly_uint64.
11353 (store_constructor): Use poly_int_tree_p.
11354 (expand_expr_real_1): Likewise.
11355 * function.c (assign_temp): Likewise.
11356 * fold-const.c (const_binop): Use poly_int_tree_p and
11357 wi::to_poly_offset.
11358 (fold_indirect_ref_1): Likewise. Use multiple_p to attempt an exact
11360 * ipa-icf-gimple.c (func_checker::compare_operand): Use
11361 to_poly_offset for MEM offsets.
11362 * ipa-icf.c (sem_variable::equals): Likewise.
11363 * stor-layout.c (compute_record_mode): Use poly_int_tree_p.
11364 * tree-ssa-sccvn.c (ao_ref_init_from_vn_reference): Use
11365 wi::to_poly_offset for BIT_FIELD_REF offsets.
11366 (vn_reference_maybe_forwprop_address): Use poly_int_tree_p and
11367 wi::to_poly_offset.
11368 * var-tracking.c (emit_note_insn_var_location): Use
11369 tree_to_poly_uint64.
11371 2018-05-29 Jim Wilson <jimw@sifive.com>
11373 * config/riscv/riscv.c (riscv_interrupt_type): Fix comment typo.
11375 2018-05-29 Uros Bizjak <ubizjak@gmail.com>
11378 * config/i386/i386.md (l<rounding_insn><MODEF:mode><SWI48:mode>2):
11379 Enable for TARGET_SSE4_1 and generate rounds{s,d} and cvtts{s,d}2si{,q}
11381 (sse4_1_round<mode>2): Use nonimmediate_operand
11382 for operand 1 predicate.
11384 2018-05-29 Martin Sebor <msebor@redhat.com>
11385 Richard Biener <rguenther@suse.de>
11388 * calls.c (get_size_range): Call determine_value_range instead
11389 of get_value_range..
11390 * tree-vrp.h (determine_value_range): Declared new function.
11391 * tree-vrp.c (determine_value_range_1, determine_value_range): New.
11393 2018-05-29 Richard Biener <rguenther@suse.de>
11395 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Make
11396 sure to use non-pattern stmts for get_earlier_stmt arguments.
11397 * tree-vectorizer.h (get_earlier_stmt): Assert we do not get
11398 called on pattern stmts.
11399 (get_later_stmt): Likewise.
11401 2018-05-29 Martin Liska <mliska@suse.cz>
11403 PR gcov-profile/85759
11404 * doc/gcov.texi: Document GCOV_ERROR_FILE and GCOV_EXIT_AT_ERROR
11407 2018-05-29 Jakub Jelinek <jakub@redhat.com>
11409 * tree-cfg.c (verify_gimple_assign_unary): Add checking for
11411 (verify_gimple_assign_binary): Check TYPE_VECTOR_SUBPARTS for
11415 * tree.def (VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR,
11416 VEC_PACK_FLOAT_EXPR): New tree codes.
11417 * tree-pretty-print.c (op_code_prio): Handle
11418 VEC_UNPACK_FIX_TRUNC_HI_EXPR and VEC_UNPACK_FIX_TRUNC_LO_EXPR.
11419 (dump_generic_node): Handle VEC_UNPACK_FIX_TRUNC_HI_EXPR,
11420 VEC_UNPACK_FIX_TRUNC_LO_EXPR and VEC_PACK_FLOAT_EXPR.
11421 * tree-inline.c (estimate_operator_cost): Likewise.
11422 * gimple-pretty-print.c (dump_binary_rhs): Handle VEC_PACK_FLOAT_EXPR.
11423 * fold-const.c (const_binop): Likewise.
11424 (const_unop): Handle VEC_UNPACK_FIX_TRUNC_HI_EXPR and
11425 VEC_UNPACK_FIX_TRUNC_LO_EXPR.
11426 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
11427 (verify_gimple_assign_binary): Handle VEC_PACK_FLOAT_EXPR.
11428 * cfgexpand.c (expand_debug_expr): Handle VEC_UNPACK_FIX_TRUNC_HI_EXPR,
11429 VEC_UNPACK_FIX_TRUNC_LO_EXPR and VEC_PACK_FLOAT_EXPR.
11430 * expr.c (expand_expr_real_2): Likewise.
11431 * optabs.def (vec_packs_float_optab, vec_packu_float_optab,
11432 vec_unpack_sfix_trunc_hi_optab, vec_unpack_sfix_trunc_lo_optab,
11433 vec_unpack_ufix_trunc_hi_optab, vec_unpack_ufix_trunc_lo_optab): New
11435 * optabs.c (expand_widen_pattern_expr): For
11436 VEC_UNPACK_FIX_TRUNC_HI_EXPR and VEC_UNPACK_FIX_TRUNC_LO_EXPR use
11437 sign from result type rather than operand's type.
11438 (expand_binop_directly): For vec_packu_float_optab and
11439 vec_packs_float_optab allow result type to be different from operand's
11441 * optabs-tree.c (optab_for_tree_code): Handle
11442 VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR and
11443 VEC_PACK_FLOAT_EXPR. Formatting fixes.
11444 * tree-vect-generic.c (expand_vector_operations_1): Handle
11445 VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR and
11446 VEC_PACK_FLOAT_EXPR.
11447 * tree-vect-stmts.c (supportable_widening_operation): Handle
11449 (supportable_narrowing_operation): Handle FLOAT_EXPR.
11450 * config/i386/i386.md (fixprefix, floatprefix): New code attributes.
11451 * config/i386/sse.md (*float<floatunssuffix>v2div2sf2): Rename to ...
11452 (float<floatunssuffix>v2div2sf2): ... this. Formatting fix.
11453 (vpckfloat_concat_mode, vpckfloat_temp_mode, vpckfloat_op_mode): New
11455 (vec_pack<floatprefix>_float_<mode>): New expander.
11456 (vunpckfixt_mode, vunpckfixt_model, vunpckfixt_extract_mode): New mode
11458 (vec_unpack_<fixprefix>fix_trunc_lo_<mode>,
11459 vec_unpack_<fixprefix>fix_trunc_hi_<mode>): New expanders.
11460 * doc/md.texi (vec_packs_float_@var{m}, vec_packu_float_@var{m},
11461 vec_unpack_sfix_trunc_hi_@var{m}, vec_unpack_sfix_trunc_lo_@var{m},
11462 vec_unpack_ufix_trunc_hi_@var{m}, vec_unpack_ufix_trunc_lo_@var{m}):
11464 * doc/generic.texi (VEC_UNPACK_FLOAT_HI_EXPR,
11465 VEC_UNPACK_FLOAT_LO_EXPR): Fix pasto in description.
11466 (VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR,
11467 VEC_PACK_FLOAT_EXPR): Document.
11469 2018-05-29 Richard Biener <rguenther@suse.de>
11471 * tree-vectorizer.h (struct vec_info): Add stmt_vec_infos
11473 (stmt_vec_info_vec): Make pointer.
11474 (init_stmt_vec_info_vec): Remove.
11475 (free_stmt_vec_info_vec): Likewise.
11476 (set_stmt_vec_info_vec): New function.
11477 (free_stmt_vec_infos): Likewise.
11478 (vinfo_for_stmt): Adjust for stmt_vec_info_vec indirection.
11479 (set_vinfo_for_stmt): Likewise.
11480 (get_earlier_stmt): Likewise.
11481 (get_later_stmt): Likewise.
11482 * tree-vectorizer.c (stmt_vec_info_vec): Make pointer.
11483 (vec_info::vec_info): Allocate stmt_vec_infos and set the global.
11484 (vec_info::~vec_info): Free stmt_vec_infos.
11485 (vectorize_loops): Set the global stmt_vec_info_vec to NULL.
11486 Remove old init_stmt_vec_info_vec/free_stmt_vec_info_vec calls.
11487 (pass_slp_vectorize::execute): Likewise.
11488 * tree-vect-stmts.c (init_stmt_vec_info_vec): Remove.
11489 (free_stmt_vec_info_vec): Likewise.
11490 (set_stmt_vec_info_vec): New function.
11491 (free_stmt_vec_infos): Likewise.
11492 * tree-vect-loop.c (_loop_vec_info::~_loop_vec_info): Set
11493 the global stmt_vec_info_vec.
11494 * tree-parloops.c (gather_scalar_reductions): Use
11495 set_stmt_vec_info_vec/free_stmt_vec_infos and maintain a local
11498 2018-05-29 Richard Biener <rguenther@suse.de>
11500 * dominance.c (iterate_fix_dominators): Push/pop TV_DOMINANCE.
11502 2018-05-29 Martin Liska <mliska@suse.cz>
11503 David Malcolm <dmalcolm@redhat.com>
11505 * vec.c (test_reverse): New.
11506 (vec_c_tests): Add new test.
11507 * vec.h (vl_ptr>::reverse): New function.
11509 2018-05-29 Gerald Pfeifer <gerald@pfeifer.com>
11511 * config.gcc: Identify FreeBSD 3.x and 4.x as unsupported.
11513 * config/freebsd-spec.h (FBSD_LIB_SPEC): Only consider FreeBSD 5
11516 2018-05-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
11518 * tree-dump.c (dump_node): Use splay_tree_delete_pointers.
11520 2018-05-28 Richard Biener <rguenther@suse.de>
11522 PR tree-optimization/85933
11523 * tree-vect-data-refs.c (vect_record_base_alignments): Only
11524 look at stmts marked as vectorizable.
11526 2018-05-28 Richard Biener <rguenther@suse.de>
11528 PR tree-optimization/85934
11529 * tree-vect-generic.c (expand_vector_operations_1): Hoist
11530 vector boolean check before scalar optimization.
11532 2018-05-28 Jakub Jelinek <jakub@redhat.com>
11534 * doc/invoke.texi (ARM Options): Use @item instead of @itemx
11537 2018-05-28 Mark Wielaard <mark@klomp.org>
11539 * dwarf2asm.c (dw2_asm_output_delta_uleb128): Add brackets around lab2
11540 if it is an expression containing a minus sign.
11542 2018-05-27 John David Anglin <danglin@gcc.gnu.org>
11544 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Define to 0.
11546 2018-05-27 Paul Koning <ni1d@arrl.net>
11548 * config/pdp11/pdp11.md (truncsihi2): Remove.
11550 2018-05-27 Monk Chiang <sh.chiang04@gmail.com>
11551 Chung-Ju Wu <jasonwucj@gmail.com>
11553 * config/nds32/nds32-intrinsic.md (unaligned_storedi): Modify patterns
11555 (unaligned_store_dw): Ditto.
11556 * config/nds32/nds32-memory-manipulation.c
11557 (nds32_expand_movmemsi_loop_known_size): Refactoring implementation.
11558 (nds32_gen_dup_4_byte_to_word_value): Rename to ...
11559 (nds32_gen_dup_4_byte_to_word_value_aux): ... this.
11560 (emit_setmem_word_loop): Rename to ...
11561 (emit_setmem_doubleword_loop): ... this.
11562 (nds32_gen_dup_4_byte_to_word_value): New function.
11563 (nds32_gen_dup_8_byte_to_double_word_value): New function.
11564 (nds32_expand_setmem_loop): Refine implementation.
11565 (nds32_expand_setmem_loop_v3m): Ditto.
11566 * config/nds32/nds32-multiple.md (unaligned_store_update_base_dw): New
11569 2018-05-27 Chung-Ju Wu <jasonwucj@gmail.com>
11571 * config/nds32/nds32.md (bswapsi2, bswaphi2): New patterns.
11573 2018-05-27 Chung-Ju Wu <jasonwucj@gmail.com>
11575 * config/nds32/nds32.c (nds32_attribute_table): Add "no_prologue".
11576 (nds32_init_machine_status): Initialize machine->attr_naked_p and
11577 machine->attr_no_prologue_p.
11578 (nds32_compute_stack_frame): Check "naked" and "no_prologue" attributes.
11579 (nds32_naked_function_p): Handle "naked" and "no_prologue" attributes.
11580 (nds32_expand_epilogue): Consider attr_naked_p.
11581 (nds32_expand_epilogue_v3pop): Likewise.
11582 (nds32_can_use_return_insn): Likewise.
11583 * config/nds32/nds32.h (machine_function): Add attr_naked_p and
11584 attr_no_prologue_p fields.
11585 * config/nds32/nds32.opt (mret-in-naked-func): New option.
11587 2018-05-27 Jakub Jelinek <jakub@redhat.com>
11590 * config/i386/i386.md (fixunssuffix, floatunssuffix): New code
11592 * config/i386/sse.md
11593 (<floatsuffix>float<sseintvecmodelower><mode>2<mask_name><round_name>):
11595 (float<floatunssuffix><sseintvecmodelower><mode>2<mask_name><round_name>):
11597 (<floatsuffix>float<sselongvecmodelower><mode>2<mask_name><round_name>):
11599 (float<floatunssuffix><sselongvecmodelower><mode>2<mask_name><round_name>):
11601 (*<floatsuffix>floatv2div2sf2): Rename to ...
11602 (*float<floatunssuffix>v2div2sf2): ... this.
11603 (<floatsuffix>floatv2div2sf2_mask): Rename to ...
11604 (float<floatunssuffix>v2div2sf2_mask): ... this.
11605 (*<floatsuffix>floatv2div2sf2_mask_1): Rename to ...
11606 (*float<floatunssuffix>v2div2sf2_mask_1): ... this.
11607 (<fixsuffix>fix_truncv8dfv8si2<mask_name><round_saeonly_name>): Rename
11609 (fix<fixunssuffix>_truncv8dfv8si2<mask_name><round_saeonly_name>):
11611 (<fixsuffix>fix_trunc<mode><sseintvecmodelower>2<mask_name><round_saeonly_name>):
11613 (fix<fixunssuffix>_trunc<mode><sseintvecmodelower>2<mask_name><round_saeonly_name>):
11615 (<fixsuffix>fix_trunc<mode><sselongvecmodelower>2<mask_name><round_saeonly_name>):
11617 (fix<fixunssuffix>_trunc<mode><sselongvecmodelower>2<mask_name><round_saeonly_name>):
11619 (<fixsuffix>fix_truncv2sfv2di2<mask_name>): Rename to ...
11620 (fix<fixunssuffix>_truncv2sfv2di2<mask_name>): ... this.
11621 (vec_pack_ufix_trunc_<mode>): Use gen_fixuns_truncv8dfv8si2 instead of
11622 gen_ufix_truncv8dfv8si2.
11623 * config/i386/i386-builtin.def (__builtin_ia32_cvttpd2uqq256_mask,
11624 __builtin_ia32_cvttpd2uqq128_mask, __builtin_ia32_cvttps2uqq256_mask,
11625 __builtin_ia32_cvttps2uqq128_mask, __builtin_ia32_cvtuqq2ps256_mask,
11626 __builtin_ia32_cvtuqq2ps128_mask, __builtin_ia32_cvtuqq2pd256_mask,
11627 __builtin_ia32_cvtuqq2pd128_mask, __builtin_ia32_cvttpd2udq512_mask,
11628 __builtin_ia32_cvtuqq2ps512_mask, __builtin_ia32_cvtuqq2pd512_mask,
11629 __builtin_ia32_cvttps2uqq512_mask, __builtin_ia32_cvttpd2uqq512_mask):
11630 Use fixuns instead ufix or floatuns instead ufloat in CODE_FOR_ names.
11632 2018-05-24 H.J. Lu <hongjiu.lu@intel.com>
11636 * varasm.c (assemble_alias): Lookup ifunc attribute on error.
11638 2018-05-25 Jim Wilson <jimw@sifive.com>
11640 * config/riscv/riscv-protos.h (riscv_epilogue_uses): New.
11641 * config/riscv/riscv.c (struct machine_function): Add
11642 interrupt_handler_p and attribute_checked_p fields.
11643 (riscv_attribute_table): Add interrupt.
11644 (riscv_interrupt_type_p): New.
11645 (riscv_save_reg_p): Save extra regs for interrupt handler.
11646 (riscv_use_save_libcall): Return false for interrupt handler.
11647 (riscv_first_stack_step): Add forward declaration.
11648 (riscv_compute_frame_info): New local interrupt_save_t1. Set it
11649 for interrupt handler with large frame. Use it for saved reg list.
11650 (riscv_expand_prologue): Move flag_stack_usage_info support to
11651 eliminate duplication.
11652 (riscv_expand_epilogue): Generate mret for interrupt handler.
11653 (riscv_epilogue_uses): New.
11654 (riscv_can_use_return_insn): Return false for interrupt handler.
11655 (riscv_function_ok_for_sibcall): Likewise.
11656 (riscv_set_current_function): Add interrupt handler support.
11657 * config/riscv/riscv.h (EPILOGUE_USES): Call riscv_epilogue_uses.
11658 * config/riscv/riscv.md (UNSPECV_MRET): New.
11660 (riscv_frflags, riscv_fsflags): Use tab after opcode.
11662 * doc/extend.texi (RISC-V Function Attributes) <interrupt>: New.
11664 2018-05-25 Bill Schmidt <wschmidt@linux.ibm.com>
11666 PR tree-optimization/85712
11667 * gimple-ssa-strength-reduction.c (replace_one_candidate): Skip if
11668 this candidate has already been replaced in-situ by a copy.
11670 2018-05-25 Jason Merrill <jason@redhat.com>
11672 PR c++/80485 - inline function non-zero address.
11673 * symtab.c (nonzero_address): Check DECL_COMDAT.
11675 2018-05-25 Uros Bizjak <ubizjak@gmail.com>
11678 * config/alpha/alpha.md (ashlsi3): New insn pattern.
11679 (*ashlsi_se): Rename from *ashldi_se. Define as sign
11680 extension of SImode operation. Use const123_operand predicate.
11681 (*saddsi_1): Remove.
11682 (*saddl_se_1): Ditto.
11683 (*ssubsi_1): Ditto.
11684 (*ssubl_se_1): Ditto.
11685 * config/alpha/predicates.md (const123_operand): New predicate.
11686 * config/alpha/constraints.md (P): Use IN_RANGE.
11688 2018-05-25 Richard Biener <rguenther@suse.de>
11690 * tree-ssa-alias.h (refs_may_alias_p): Add tbaa_p bool parameter,
11692 (ref_maybe_used_by_stmt_p): Likewise.
11693 (stmt_may_clobber_ref_p): Likewise.
11694 (stmt_may_clobber_ref_p_1): Likewise.
11695 * tree-ssa-alias.c (refs_may_alias_p): Add tbaa_p bool parameter
11697 (ref_maybe_used_by_stmt_p): Likewise.
11698 (stmt_may_clobber_ref_p): Likewise.
11699 (stmt_may_clobber_ref_p_1): Likewise.
11700 * tree-vect-data-refs.c (vect_slp_analyze_node_dependences): Use
11701 the alias oracle to disambiguate DRs with stmts DR analysis
11703 (vect_analyze_data_refs): Do not give up on not analyzable
11704 DRs for BB vectorization. Remove code truncating the dataref
11707 2018-05-25 Jakub Jelinek <jakub@redhat.com>
11710 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>_1):
11711 Add (=Yk,v,C) variant using vptestm insn. Use TARGET_AVX512BW
11712 in test instead of TARGET_AVX512F for VI12_AVX512VL iterator.
11714 2018-05-25 Richard Biener <rguenther@suse.de>
11716 * tree-vect-data-refs.c (vect_find_stmt_data_reference): New
11717 function, combining stmt data ref gathering and fatal analysis
11719 (vect_analyze_data_refs): Remove now redudnant code and simplify.
11720 * tree-vect-loop.c (vect_get_datarefs_in_loop): Factor out from
11721 vect_analyze_loop_2 and use vect_find_stmt_data_reference.
11722 * tree-vect-slp.c (vect_slp_bb): Use vect_find_stmt_data_reference.
11723 * tree-vectorizer.h (vect_find_stmt_data_reference): Declare.
11725 2018-05-25 Bin Cheng <bin.cheng@arm.com>
11727 PR tree-optimization/85720
11728 * tree-loop-distribution.c (break_alias_scc_partitions): Don't merge
11729 SCC if all partitions are builtins.
11730 (version_loop_by_alias_check): New parameter. Generate cancelable
11731 runtime alias check if all partitions are builtins.
11732 (distribute_loop): Update call to above function.
11734 2018-05-25 Bin Cheng <bin.cheng@arm.com>
11736 * tree-outof-ssa.c (tree-ssa.h, tree-dfa.h): Include header files.
11737 (create_default_def, for_all_parms): Moved from tree-ssa-coalesce.c.
11738 (parm_default_def_partition_arg): Ditto.
11739 (set_parm_default_def_partition): Ditto.
11740 (get_parm_default_def_partitions): Ditto and make it static.
11741 (get_undefined_value_partitions): Ditto and make it static.
11742 (remove_ssa_form): Refactor call to init_var_map here.
11743 * tree-ssa-coalesce.c (build_ssa_conflict_graph): Support live range
11744 computation for loop region.
11745 (coalesce_partitions, compute_optimized_partition_bases): Ditto.
11746 (register_default_def): Delete.
11747 (for_all_parms, create_default_def): Move to tree-outof-ssa.c.
11748 (parm_default_def_partition_arg): Ditto.
11749 (set_parm_default_def_partition): Ditto.
11750 (get_parm_default_def_partitions): Ditto and make it static.
11751 (get_undefined_value_partitions): Ditto and make it static.
11752 (coalesce_with_default, coalesce_with_default): Update comment.
11753 (create_coalesce_list_for_region): New func factored out from
11754 create_outofssa_var_map.
11755 (populate_coalesce_list_for_outofssa): New func factored out from
11756 create_outofssa_var_map and coalesce_ssa_name.
11757 (create_outofssa_var_map): Delete.
11758 (coalesce_ssa_name): Refactor to support live range computation.
11759 * tree-ssa-coalesce.h (coalesce_ssa_name): Change decl.
11760 (get_parm_default_def_partitions): Delete.
11761 (get_undefined_value_partitions): Ditto.
11762 * tree-ssa-live.c (init_var_map, delete_var_map): Support live range
11763 computation for loop region.
11764 (new_tree_live_info, loe_visit_block): Ditto.
11765 (live_worklist, set_var_live_on_entry): Ditto.
11766 (calculate_live_on_exit, verify_live_on_entry): Ditto.
11767 * tree-ssa-live.h (struct _var_map): New fields.
11768 (init_var_map): Change decl.
11769 (region_contains_p): New.
11771 2018-05-25 Bin Cheng <bin.cheng@arm.com>
11773 * tree-ssa-live.h (live_merge_and_clear): Delete.
11775 2018-05-25 Richard Biener <rguenther@suse.de>
11778 * tree-dump.c (dequeue_and_dump): Remove access to removed
11779 operand 2 of a SWITCH_EXPR.
11781 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
11783 * doc/sourcebuild.texi (vect_double_cond_arith): Include
11784 multiplication and division.
11785 * doc/md.texi (cond_mul@var{m}, cond_div@var{m}, cond_mod@var{m})
11786 (cond_udiv@var{m}, cond_umod@var{m}): Document.
11787 * optabs.def (cond_smul_optab, cond_sdiv_optab, cond_smod_optab)
11788 (cond_udiv_optab, cond_umod_optab): New optabs.
11789 * internal-fn.def (IFN_COND_MUL, IFN_COND_DIV, IFN_COND_MOD)
11790 (IFN_COND_RDIV): New internal functions.
11791 * internal-fn.c (get_conditional_internal_fn): Handle TRUNC_DIV_EXPR,
11792 TRUNC_MOD_EXPR and RDIV_EXPR.
11793 * match.pd (UNCOND_BINARY, COND_BINARY): Handle them.
11794 * config/aarch64/iterators.md (UNSPEC_COND_MUL, UNSPEC_COND_DIV):
11796 (SVE_INT_BINARY): Include mult.
11797 (SVE_COND_FP_BINARY): Include UNSPEC_MUL and UNSPEC_DIV.
11798 (optab, sve_int_op): Handle mult.
11799 (optab, sve_fp_op, commutative): Handle UNSPEC_COND_MUL and
11801 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): New pattern
11802 for SVE_INT_BINARY_SD.
11804 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
11806 * config/aarch64/iterators.md (SVE_INT_BINARY_SD): New code iterator.
11807 (optab, sve_int_op): Handle div and udiv.
11808 * config/aarch64/aarch64-sve.md (<optab><mode>3): New expander
11809 for SVE_INT_BINARY_SD.
11810 (*<optab><mode>3): New insn for the same.
11812 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
11814 * tree-vect-patterns.c: Include predict.h.
11815 (vect_recog_divmod_pattern): Restrict check for division support
11816 to when optimizing for size.
11818 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
11820 * doc/sourcebuild.texi (vect_double_cond_arith: Document.
11821 * gimple-match.h (gimple_match_op::MAX_NUM_OPS): Bump to 4.
11822 (gimple_match_op::gimple_match_op): Add an overload for 4 operands.
11823 (gimple_match_op::set_op): Likewise.
11824 (gimple_resimplify4): Declare.
11825 * genmatch.c (get_operand_type): Handle CFN_COND_* functions.
11826 (expr::gen_transform): Likewise.
11827 (decision_tree::gen): Generate a simplification routine for 4 operands.
11828 * gimple-match-head.c (gimple_simplify): Add an overload for
11829 4 operands. In the top-level function, handle up to 4 call
11830 arguments and call gimple_resimplify4.
11831 (gimple_resimplify4): New function.
11832 (build_call_internal): Pass a fourth operand.
11833 (maybe_push_to_seq): Likewise.
11834 * match.pd (UNCOND_BINARY, COND_BINARY): New operator lists.
11835 Fold VEC_COND_EXPRs of an operation and a default value into
11836 an IFN_COND_* function if possible.
11837 * config/aarch64/iterators.md (UNSPEC_COND_MAX, UNSPEC_COND_MIN):
11839 (SVE_COND_FP_BINARY): Include them.
11840 (optab, sve_fp_op): Handle them.
11841 (SVE_INT_BINARY_REV): New code iterator.
11842 (SVE_COND_FP_BINARY_REV): New int iterator.
11843 (commutative): New int attribute.
11844 * config/aarch64/aarch64-protos.h (aarch64_sve_prepare_conditional_op):
11846 * config/aarch64/aarch64.c (aarch64_sve_prepare_conditional_op): New
11848 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): Use it.
11849 (*cond_<optab><mode>): New patterns for reversed operands.
11851 2018-05-25 Richard Biener <rguenther@suse.de>
11853 * tree-vectorizer.h (STMT_VINFO_GROUP_*, GROUP_*): Remove.
11854 (DR_GROUP_*): New, assert we have non-NULL ->data_ref_info.
11855 (REDUC_GROUP_*): New, assert we have NULL ->data_ref_info.
11856 (STMT_VINFO_GROUPED_ACCESS): Adjust.
11857 * tree-vect-data-refs.c (everywhere): Adjust users.
11858 * tree-vect-loop.c (everywhere): Likewise.
11859 * tree-vect-slp.c (everywhere): Likewise.
11860 * tree-vect-stmts.c (everywhere): Likewise.
11861 * tree-vect-patterns.c (vect_reassociating_reduction_p): Likewise.
11863 2018-05-25 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
11865 * configure.ac (gcc_cv_as_section_has_e): Move to common section.
11867 (gcc_cv_as_section_exclude): ... this.
11868 Try Solaris as #exclude syntax.
11869 * configure: Regenerate.
11870 * config.in: Regenerate.
11871 * config/i386/i386.c (i386_solaris_elf_named_section): Handle
11873 * config/sparc/sparc.c (sparc_solaris_elf_asm_named_section)
11874 [HAVE_GAS_SECTION_EXCLUDE]: Handle SECTION_EXCLUDE.
11876 * varasm.c (default_elf_asm_named_section): Don't check if
11877 HAVE_GAS_SECTION_EXCLUDE is defined.
11879 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
11881 * doc/md.texi: Update the documentation of the cond_* optabs
11882 to mention the new final operand. Fix GET_MODE_NUNITS call.
11883 Describe the scalar case too.
11884 * internal-fn.def (IFN_EXTRACT_LAST): Change type to fold_left.
11885 * internal-fn.c (expand_cond_unary_optab_fn): Expect 3 operands
11887 (expand_cond_binary_optab_fn): Expect 4 operands instead of 3.
11888 (get_conditional_internal_fn): Update comment.
11889 * tree-vect-loop.c (vectorizable_reduction): Pass the original
11890 accumulator value as a final argument to conditional functions.
11891 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): Turn into
11892 a define_expand and add an "else" operand. Assert for now that
11893 the else operand is equal to operand 2. Use SVE_INT_BINARY and
11894 SVE_COND_FP_BINARY instead of SVE_COND_INT_OP and SVE_COND_FP_OP.
11895 (*cond_<optab><mode>): New patterns.
11896 * config/aarch64/iterators.md (UNSPEC_COND_SMAX, UNSPEC_COND_UMAX)
11897 (UNSPEC_COND_SMIN, UNSPEC_COND_UMIN, UNSPEC_COND_AND, UNSPEC_COND_ORR)
11898 (UNSPEC_COND_EOR): Delete.
11899 (optab): Remove associated mappings.
11900 (SVE_INT_BINARY): New code iterator.
11901 (sve_int_op): Remove int attribute and add "minus" to the code
11903 (SVE_COND_INT_OP): Delete.
11904 (SVE_COND_FP_OP): Rename to...
11905 (SVE_COND_FP_BINARY): ...this.
11907 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
11909 * optabs.c (can_reuse_operands_p): New function.
11910 (maybe_legitimize_operands): Try to reuse the results for
11913 2018-05-24 Uros Bizjak <ubizjak@gmail.com>
11915 * config/i386/sse.md (cvtusi2<ssescalarmodesuffix>64<round_name>):
11916 Add {q} suffix to insn mnemonic.
11918 2018-05-23 Jozef Lawrynowicz <jozef.l@mittosystems.com>
11920 * config/msp430/msp430.c (TARGET_WARN_FUNC_RETURN): Define.
11921 (msp430_warn_func_return): New.
11923 2018-05-24 Roger Sayle <roger@nextmovesoftware.com>
11925 * fold-const.c (tree_nonzero_bits): New function.
11926 * fold-const.h (tree_nonzero_bits): Likewise.
11927 * match.pd (POPCOUNT): New patterns to fold BUILTIN_POPCOUNT and
11928 friends. POPCOUNT(x&1) => x&1, POPCOUNT(x)==0 => x==0, etc.
11930 2018-05-24 H.J. Lu <hongjiu.lu@intel.com>
11934 * varasm.c (assemble_alias): Check ifunc_resolver only on
11937 2018-05-24 Uros Bizjak <ubizjak@gmail.com>
11940 * config/i386/sse.md (movdi_to_sse): Do not generate pseudo
11941 when memory input operand is handled.
11943 2018-05-24 Luis Machado <luis.machado@linaro.org>
11945 * config/aarch64/aarch64.c (qdf24xx_addrcost_table): New static
11947 (qdf24xx_tunings) <addr_costs>: Set to qdf24xx_addrcost_table.
11949 2018-05-24 Richard Sandiford <richard.sandiford@linaro.org>
11951 * match.pd: Delay FMA folds until after vectorization.
11953 2018-05-24 Andre Vieira <andre.simoesdiasvieira@arm.com>
11956 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Make
11957 address check not strict.
11959 2018-05-24 Richard Sandiford <richard.sandiford@linaro.org>
11961 * gimple-match.h (gimple_match_op): New class.
11962 (mprts_hook): Replace parameters with a gimple_match_op *.
11963 (maybe_build_generic_op): Likewise.
11964 (gimple_simplified_result_is_gimple_val): Replace parameters with
11965 a const gimple_match_op *.
11966 (gimple_simplify): Replace code_helper * and tree * parameters with
11967 a gimple_match_op * parameter.
11968 (gimple_resimplify1): Replace code_helper *, tree and tree *
11969 parameters with a gimple_match_op * parameter.
11970 (gimple_resimplify2): Likewise.
11971 (gimple_resimplify3): Likewise.
11972 (maybe_push_res_to_seq): Replace code_helper, tree and tree *
11973 parameters with a gimple_match_op * parameter.
11974 * gimple-match-head.c (gimple_simplify): Change prototypes of
11975 auto-generated functions to take a gimple_match_op * instead of
11976 separate code_helper * and tree * parameters. Make the same
11977 change in the top-level overload and update calls to the
11978 gimple_resimplify routines. Update calls to the auto-generated
11979 functions and to maybe_push_res_to_seq in the publicly-facing
11980 operation-specific gimple_simplify overloads.
11981 (gimple_match_op::MAX_NUM_OPS): Define.
11982 (gimple_resimplify1): Replace rcode and ops with a single res_op
11983 parameter. Update call to gimple_simplify.
11984 (gimple_resimplify2): Likewise.
11985 (gimple_resimplify3): Likewise.
11986 (mprts_hook): Replace parameters with a gimple_match_op *.
11987 (maybe_build_generic_op): Likewise.
11988 (build_call_internal): Replace type, nargs and ops with
11989 a gimple_match_op *.
11990 (maybe_push_res_to_seq): Replace res_code, type and ops parameters
11991 with a single gimple_match_op *. Update calls to mprts_hook,
11992 build_call_internal and gimple_simplified_result_is_gimple_val.
11993 Factor out code that is common to the tree_code and combined_fn cases.
11994 * genmatch.c (expr::gen_transform): Replace tem_code and
11995 tem_ops with a gimple_match_op called tem_op. Update calls
11996 to the gimple_resimplify functions and maybe_push_res_to_seq.
11997 (dt_simplify::gen_1): Manipulate res_op instead of res_code and
11998 res_ops. Update call to the gimple_resimplify functions.
11999 (dt_simplify::gen): Pass res_op instead of res_code and res_ops.
12000 (decision_tree::gen): Make the functions take a gimple_match_op *
12001 called res_op instead of separate res_code and res_ops parameters.
12002 Update call accordingly.
12003 * gimple-fold.c (replace_stmt_with_simplification): Replace rcode
12004 and ops with a single res_op parameter. Update calls to
12005 maybe_build_generic_op and maybe_push_res_to_seq.
12006 (fold_stmt_1): Update calls to gimple_simplify and
12007 replace_stmt_with_simplification.
12008 (gimple_fold_stmt_to_constant_1): Update calls to gimple_simplify
12009 and gimple_simplified_result_is_gimple_val.
12010 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update call to
12012 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Replace parameters
12013 with a gimple_match_op *.
12014 (vn_nary_build_or_lookup): Likewise. Update call to
12015 vn_nary_build_or_lookup_1.
12016 (vn_nary_build_or_lookup_1): Replace rcode, type and ops with a
12017 gimple_match_op *. Update calls to the gimple_resimplify routines
12018 and to gimple_simplified_result_is_gimple_val.
12019 (vn_nary_simplify): Update call to vn_nary_build_or_lookup_1.
12020 Use gimple_match_op::MAX_NUM_OPS instead of a hard-coded 3.
12021 (vn_reference_lookup_3): Update call to vn_nary_build_or_lookup.
12022 (visit_nary_op): Likewise.
12023 (visit_reference_op_load): Likewise.
12025 2018-05-23 Luis Machado <luis.machado@linaro.org>
12027 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Use correct type
12028 modifier for printing the step amount.
12030 2018-05-23 Jozef Lawrynowicz <jozef.l@somniumtech.com>
12033 * tree.c (build_common_tree_nodes): Dont set TYPE_SIZE for __intN
12036 2018-05-23 Segher Boessenkool <segher@kernel.crashing.org>
12038 * doc/sourcebuild.texi (Endianness): New subsubsection.
12040 2018-05-23 Luis Machado <luis.machado@linaro.org>
12042 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
12043 <prefetch_dynamic_strides>: New const bool field.
12044 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
12045 prefetch_dynamic_strides.
12046 (exynosm1_prefetch_tune): Likewise.
12047 (thunderxt88_prefetch_tune): Likewise.
12048 (thunderx_prefetch_tune): Likewise.
12049 (thunderx2t99_prefetch_tune): Likewise.
12050 (qdf24xx_prefetch_tune): Likewise. Set prefetch_dynamic_strides to
12052 (aarch64_override_options_internal): Update to set
12053 PARAM_PREFETCH_DYNAMIC_STRIDES.
12054 * doc/invoke.texi (prefetch-dynamic-strides): Document new option.
12055 * params.def (PARAM_PREFETCH_DYNAMIC_STRIDES): New.
12056 * params.h (PARAM_PREFETCH_DYNAMIC_STRIDES): Define.
12057 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Account for
12058 prefetch-dynamic-strides setting.
12060 2018-05-23 Luis Machado <luis.machado@linaro.org>
12062 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
12063 <minimum_stride>: New const int field.
12064 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
12065 minimum_stride field defaulting to -1.
12066 (exynosm1_prefetch_tune): Likewise.
12067 (thunderxt88_prefetch_tune): Likewise.
12068 (thunderx_prefetch_tune): Likewise.
12069 (thunderx2t99_prefetch_tune): Likewise.
12070 (qdf24xx_prefetch_tune) <minimum_stride>: Set to 2048.
12071 <default_opt_level>: Set to 3.
12072 (aarch64_override_options_internal): Update to set
12073 PARAM_PREFETCH_MINIMUM_STRIDE.
12074 * doc/invoke.texi (prefetch-minimum-stride): Document new option.
12075 * params.def (PARAM_PREFETCH_MINIMUM_STRIDE): New.
12076 * params.h (PARAM_PREFETCH_MINIMUM_STRIDE): Define.
12077 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Return false if
12078 stride is constant and is below the minimum stride threshold.
12080 2018-05-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12082 * config/arm/arm-cpus.in (mode26): Delete.
12083 (armv4): Delete mode26 reference.
12084 * config/arm/arm.c (arm_configure_build_target): Delete use of
12087 2018-05-23 Uros Bizjak <ubizjak@gmail.com>
12089 * config/i386/i386.md (*floatuns<SWI48:mode><MODEF:mode>2_avx512):
12091 (floatunssi<mode>2): Also enable for AVX512F and TARGET_SSE_MATH.
12092 Rewrite expander pattern. Emit gen_floatunssi<mode>2_i387_with_xmm
12094 (floatunsdisf2): Rewrite expander pattern. Hanlde TARGET_AVX512F.
12095 (floatunsdidf2): Ditto.
12097 2018-05-23 Uros Bizjak <ubizjak@gmail.com>
12099 * config/i386/i386.md (fixuns_trunc<mode>di2): New insn pattern.
12100 (fixuns_trunc<mode>si2_avx512f): Ditto.
12101 (*fixuns_trunc<mode>si2_avx512f_zext): Ditto.
12102 (fixuns_trunc<mode>si2): Also enable for AVX512F and TARGET_SSE_MATH.
12103 Emit fixuns_trunc<mode>si2_avx512f for AVX512F targets.
12105 2018-05-23 Alexander Monakov <amonakov@ispras.ru>
12107 PR rtl-optimization/79985
12108 * df-scan.c (df_insn_refs_collect): Remove special case for
12109 global registers and asm statements.
12111 2018-05-23 Alexander Monakov <amonakov@ispras.ru>
12113 * extend.texi (Global Register Variables): Rewrite the bullet list.
12114 Note that the register is available for allocation. Note that access
12115 via inline asm must use constraints. Add note about async-signal
12116 handlers. Remove paragraph about automagic register selection.
12118 2018-05-23 Richard Biener <rguenther@suse.de>
12120 * tree-ssa-sccvn.c (vn_reference_lookup_3): Remove restriction
12121 of fixed offset from memset VN.
12123 2018-05-23 Bill Schmidt <wschmidt@linux.ibm.com>
12125 * gimple-ssa-strength-reduction.c (struct slsr_cand_d): Add
12126 first_interp field.
12127 (alloc_cand_and_find_basis): Initialize first_interp field.
12128 (slsr_process_mul): Modify first_interp field.
12129 (slsr_process_add): Likewise.
12130 (slsr_process_cast): Modify first_interp field for each new
12132 (slsr_process_copy): Likewise.
12133 (dump_candidate): Dump first_interp field.
12134 (replace_mult_candidate): Process all interpretations, not just
12136 (replace_rhs_if_not_dup): Likewise.
12137 (replace_one_candidate): Likewise.
12139 2018-05-23 Wilco Dijkstra <wdijkstr@arm.com>
12141 * config/aarch64/aarch64.c (aarch64_use_frame_pointer):
12143 (aarch64_needs_frame_chain): New function.
12144 (aarch64_parse_override_string): Set aarch64_use_frame_pointer.
12146 2018-05-23 Sudakshina Das <sudi.das@arm.com>
12149 * common/config/aarch64/aarch64-common.c (aarch64_handle_option):
12150 Check val before adding MASK_STRICT_ALIGN to opts->x_target_flags.
12151 * config/aarch64/aarch64.opt (mstrict-align): Remove RejectNegative.
12152 * config/aarch64/aarch64.c (aarch64_attributes): Mark allow_neg
12153 as true for strict-align.
12154 (aarch64_can_inline_p): Perform checks even when callee has no
12155 attributes to check for strict alignment.
12156 * doc/extend.texi (AArch64 Function Attributes): Document
12158 * doc/invoke.texi: (AArch64 Options): Likewise.
12160 2018-05-23 Richard Sandiford <richard.sandiford@linaro.org>
12162 PR tree-optimization/85853
12163 * tree-vect-slp.c (vect_slp_analyze_node_operations): Split out
12164 the handling of the root of the node to...
12165 (vect_slp_analyze_node_operations_1): ...this new function,
12166 and run the whole thing with the child nodes' def types
12167 set according to their SLP node's def type.
12169 2018-05-23 Richard Biener <rguenther@suse.de>
12171 PR middle-end/85874
12172 * tree-data-ref.c (create_runtime_alias_checks): Defer
12173 and ignore overflow warnings.
12175 2018-05-23 Yury Gribov <tetra2005@gmail.com>
12177 PR tree-optimization/85822
12178 * tree-vrp.c (is_masked_range_test): Fix handling of negative
12181 2018-05-23 Richard Biener <rguenther@suse.de>
12183 * tree-ssa-sccvn.c (vn_reference_lookup_3): Handle arbitrary
12184 memset constants via native_interpret_expr.
12186 2018-05-22 H.J. Lu <hongjiu.lu@intel.com>
12189 * cgraph.h (cgraph_node::create): Set ifunc_resolver for ifunc
12191 (cgraph_node::create_alias): Likewise.
12192 (cgraph_node::get_availability): Check ifunc_resolver instead
12193 of looking up ifunc attribute.
12194 * cgraphunit.c (maybe_diag_incompatible_alias): Likewise.
12195 * varasm.c (do_assemble_alias): Likewise.
12196 (assemble_alias): Likewise.
12197 (default_binds_local_p_3): Likewise.
12198 * cgraph.h (cgraph_node): Add ifunc_resolver.
12199 (cgraph_node::only_called_directly_or_aliased_p): Return false
12200 for IFUNC resolver.
12201 * lto-cgraph.c (input_node): Set ifunc_resolver for ifunc
12203 * symtab.c (symtab_node::verify_base): Verify that ifunc_resolver
12204 is equivalent to lookup_attribute ("ifunc", DECL_ATTRIBUTES (decl)).
12205 (symtab_node::binds_to_current_def_p): Check ifunc_resolver
12206 instead of looking up ifunc attribute.
12208 2018-05-22 Luis Machado <luis.machado@linaro.org>
12210 * config/aarch64/aarch64.md (*ashift<mode>_extv_bfiz): New pattern.
12212 2018-05-22 Martin Sebor <msebor@redhat.com>
12214 PR middle-end/85359
12215 * builtins.c (expand_builtin_strcpy): Call maybe_warn_nonstring_arg
12216 only when expasion succeeds.
12217 (expand_builtin_strcmp): Same.
12218 (expand_builtin_strncmp): Same.
12220 2018-05-22 Martin Sebor <msebor@redhat.com>
12222 * calls.c (maybe_warn_nonstring_arg): Fix a typo in a comment.
12224 2018-05-22 Jackson Woodruff <jackson.woodruff@arm.com>
12225 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12227 * config/aarch64/aarch64-ldpstp.md: Replace uses of
12228 aarch64_mem_pair_operand with memory_operand and delete operand swapping
12230 * config/aarch64/aarch64.c (aarch64_operands_ok_for_ldpstp):
12231 Add check for legitimate_address.
12232 (aarch64_gen_adjusted_ldpstp): Swap operands where appropriate.
12233 (aarch64_swap_ldrstr_operands): New.
12234 * config/aarch64/aarch64-protos.h (aarch64_swap_ldrstr_operands):
12237 2018-05-22 Jackson Woodruff <jackson.woodruff@arm.com>
12238 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12240 * config/aarch64/aarch64.md: New patterns to generate stp
12242 (store_pair_sw, store_pair_dw): New patterns to generate stp for
12243 single words and double words.
12244 (load_pair_sw, load_pair_dw): Likewise.
12245 (store_pair_sf, store_pair_df, store_pair_si, store_pair_di):
12247 (load_pair_sf, load_pair_df, load_pair_si, load_pair_di):
12249 * config/aarch64/aarch64-ldpstp.md: Modify peephole
12250 for different mode ldpstp and add peephole for merged zero stores.
12251 Likewise for loads.
12252 * config/aarch64/aarch64.c (aarch64_operands_ok_for_ldpstp):
12254 (aarch64_gen_store_pair): Rename calls to match new patterns.
12255 (aarch64_gen_load_pair): Rename calls to match new patterns.
12256 * config/aarch64/aarch64-simd.md (load_pair<mode>): Rename to...
12257 (load_pair<DREG:mode><DREG2:mode>): ... This.
12258 (store_pair<mode>): Rename to...
12259 (vec_store_pair<DREG:mode><DREG2:mode>): ... This.
12260 * config/aarch64/iterators.md (DREG, DREG2, DX2, SX, SX2, DSX):
12261 New mode iterators.
12262 (V_INT_EQUIV): Handle SImode.
12263 * config/aarch64/predicates.md (aarch64_reg_zero_or_fp_zero):
12266 2018-05-22 Martin Sebor <msebor@redhat.com>
12269 * calls.c (maybe_warn_nonstring_arg): Use string length to set
12270 or ajust the presumed bound on an operation to avoid unnecessary
12273 2018-05-22 Martin Sebor <msebor@redhat.com>
12275 PR tree-optimization/85826
12276 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Avoid
12277 assuming that a DECL necesarily has a constant size.
12279 2018-05-22 Richard Sandiford <richard.sandiford@linaro.org>
12281 PR middle-end/85862
12282 * internal-fn.c (expand_direct_optab_fn): Cope with a null lhs.
12284 2018-05-22 Richard Biener <rguenther@suse.de>
12286 PR tree-optimization/85834
12287 * tree-ssa-sccvn.c (vn_reference_lookup_3): Properly handle
12288 non-constant and non-zero memset arguments.
12290 2018-05-22 Martin Liska <mliska@suse.cz>
12293 * ipa-icf.c (sem_item::add_type): Do not ICE for incomplete types.
12295 2018-05-22 Richard Biener <rguenther@suse.de>
12297 PR tree-optimization/85863
12298 * tree-vect-stmts.c (vect_is_simple_cond): Only widen invariant
12299 comparisons when vectype is specified.
12300 (vectorizable_condition): Do not specify vectype for
12301 vect_is_simple_cond when SLP vectorizing.
12303 2018-05-21 Michael Meissner <meissner@linux.ibm.com>
12306 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Do not
12307 define __ibm128 as long double.
12308 * config/rs6000/rs6000.c (rs6000_init_builtins): Create __ibm128
12309 as a distinct type when IEEE 128-bit support is enabled.
12310 (init_float128_ieee): Fix up conversions between IFmode and IEEE
12311 128-bit types to use the correct functions.
12312 (rs6000_expand_float128_convert): Use explicit FLOAT_EXTEND to
12313 convert between 128-bit floating point types that have different
12314 modes but the same representation, instead of using gen_lowpart to
12316 * config/rs6000/rs6000.md (IFKF): New iterator for IFmode and
12318 (IFKF_reg): New attributes to give the register constraints for
12320 (extend<mode>tf2_internal): New insns to mark an explicit
12321 conversion between 128-bit floating point types that have a
12322 different mode but share the same representation.
12324 2018-05-21 Richard Sandiford <richard.sandiford@linaro.org>
12326 PR tree-optimization/85814
12327 * tree-ssa-strlen.c (get_stridx_plus_constant): Cope with
12328 a null return from get_strinfo when unsharing the next
12329 strinfo in the chain.
12331 2018-05-21 Vladimir Mezentsev <vladimir.mezentsev@oracle.com>
12334 * varasm.c (weak_finish): Clean up weak_decls.
12336 2018-05-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12338 * config/aarch64/aarch64.md ("unspec"): Define UNSPEC_SABAL,
12339 UNSPEC_SABDL2, UNSPEC_SADALP, UNSPEC_UABAL, UNSPEC_UABDL2,
12340 UNSPEC_UADALP values.
12341 * config/aarch64/iterators.md (ABAL): New int iterator.
12344 (sur): Add mappings for the above.
12345 * config/aarch64/aarch64-simd.md (aarch64_<sur>abdl2<mode>_3):
12347 (aarch64_<sur>abal<mode>_4): Likewise.
12348 (aarch64_<sur>adalp<mode>_3): Likewise.
12349 (<sur>sadv16qi): New define_expand.
12351 2018-05-21 Alexander Nesterovskiy <alexander.nesterovskiy@intel.com>
12353 * config/i386/i386.md (*movsf_internal): AVX falsedep fix.
12354 (*movdf_internal): Ditto.
12355 (*rcpsf2_sse): Ditto.
12356 (*rsqrtsf2_sse): Ditto.
12357 (*sqrt<mode>2_sse): Ditto.
12359 2018-05-21 Tamar Christina <tamar.christina@arm.com>
12361 * config/aarch64/aarch64-simd.md (aarch64_eor3qv8hi): Change to
12363 (aarch64_bcaxqv8hi): Change to bcaxq<mode>4.
12364 * config/aarch64/aarch64-simd-builtins.def (veor3q_u8, veor3q_u32,
12365 veor3q_u64, veor3q_s8, veor3q_s16, veor3q_s32, veor3q_s64, vbcaxq_u8,
12366 vbcaxq_u32, vbcaxq_u64, vbcaxq_s8, vbcaxq_s16, vbcaxq_s32,
12368 * config/aarch64/arm_neon.h: Likewise.
12369 * config/aarch64/iterators.md (VQ_I): New.
12371 2018-05-21 Alexey Brodkin <abrodkin@synopsys.com>
12373 * config.gcc: Add arc/t-multilib-linux to tmake_file for
12375 * config/arc/t-multilib-linux: Specify MULTILIB_OPTIONS and
12378 2018-05-20 Chung-Ju Wu <jasonwucj@gmail.com>
12380 * config/nds32/constraints.md (S): New constraint.
12381 * config/nds32/nds32.md (call_internal): Use constraint S.
12382 (call_value_internal): Likewise.
12383 (sibcall_internal): Likewise.
12384 (sibcall_value_internal): Likewise.
12386 2018-05-20 Kito Cheng <kito.cheng@gmail.com>
12387 Chung-Ju Wu <jasonwucj@gmail.com>
12389 * config/nds32/nds32.c (nds32_register_move_cost): Take garywolf cpu
12390 into consideration.
12392 2018-05-20 Kito Cheng <kito.cheng@gmail.com>
12393 Chung-Ju Wu <jasonwucj@gmail.com>
12395 * config/nds32/nds32-cost.c (rtx_cost_model_t): New structure.
12396 (insn_size_16bit, insn_size_32bit): New variables for cost evaluation.
12397 (nds32_rtx_costs_impl): Simplify.
12398 (nds32_address_cost_impl): Simplify.
12399 (nds32_init_rtx_costs): New function.
12400 (nds32_rtx_costs_speed_prefer): Likewise.
12401 (nds32_rtx_costs_size_prefer): Likewise.
12402 (nds32_address_cost_speed_prefer): Likewise.
12403 (nds32_address_cost_speed_fwprop): Likewise.
12404 (nds32_address_cost_size_prefer): Likewise.
12405 * config/nds32/nds32-protos.h (nds32_init_rtx_costs): Declare.
12406 * config/nds32/nds32.c (nds32_option_override): Use
12407 nds32_init_rtx_costs function.
12409 2018-05-20 Chung-Ju Wu <jasonwucj@gmail.com>
12411 * config/nds32/nds32.c (nds32_asm_file_start): Output pipeline model.
12412 * config/nds32/nds32.h (TARGET_PIPELINE_N7): Define.
12413 (TARGET_PIPELINE_N8): Likewise.
12414 (TARGET_PIPELINE_N10): Likewise.
12415 (TARGET_PIPELINE_N13): Likewise.
12416 (TARGET_PIPELINE_GRAYWOLF): Likewise.
12418 2018-05-19 Monk Chiang <sh.chiang04@gmail.com>
12420 * config/nds32/nds32-fpu.md: Update copyright year.
12422 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
12424 * config/nds32/nds32.h (ASM_SPEC): Adjust spec rule.
12426 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
12428 * config/nds32/nds32.c
12429 (nds32_md_asm_adjust): Consider flag_inline_asm_r15 variable.
12430 * config/nds32/nds32.opt (minline-asm-r15): New option.
12432 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
12434 * common/config/nds32/nds32-common.c (TARGET_DEFAULT_TARGET_FLAGS): Add
12436 * config/nds32/nds32.md (abssi2): New pattern.
12438 2018-05-19 Uros Bizjak <ubizjak@gmail.com>
12440 * config/i386/i386.md (rex64namesuffix): New mode attribute.
12441 * config/i386/sse.md (sse_cvtsi2ss<rex64namesuffix><round_name>):
12442 Merge insn pattern from sse_cvtsi2ss<round_name> and
12443 sse_cvtsi2ssq<round_name> using SWI48 mode iterator.
12444 (sse_cvtss2si<rex64namesuffix><round_name>): Merge insn pattern
12445 from sse_cvtss2si<round_name> and sse_cvtss2siq<round_name>
12446 using SWI48 mode iterator.
12447 (sse_cvtss2si<rex64namesuffix>_2): Merge insn pattern from
12448 sse_cvtss2si_2 and sse_cvtss2siq_2 using SWI48 mode iterator.
12449 (sse_cvttss2si<rex64namesuffix><round_saeonly_name>): Merge insn
12450 pattern from sse_cvttss2si<round_saeonly_name>
12451 and sse_cvttss2siq<round_saeonly_name> using SWI48 mode iterator.
12452 (avx512f_vcvtss2usi<rex64namesuffix><round_name>): Merge insn pattern
12453 from avx512f_vcvtss2usi<round_name> and avx512f_vcvtss2usiq<round_name>
12454 using SWI48 mode iterator.
12455 (avx512f_vcvttss2usi<rex64namesuffix><round_saeonly_name>): Merge
12456 insn pattern from avx512f_vcvttss2usi<round_saeonly_name> and
12457 avx512f_vcvttss2usiq<round_saeonly_name> using SWI48 mode iterator.
12458 (avx512f_vcvtsd2usi<rex64namesuffix><round_name>): Merge insn pattern
12459 from avx512f_vcvtsd2usi<round_name> and avx512f_vcvtsd2usiq<round_name>
12460 using SWI48 mode iterator.
12461 (avx512f_vcvttsd2usi<rex64namesuffix><round_saeonly_name>): Merge
12462 insn pattern from avx512f_vcvttsd2usi<round_saeonly_name> and
12463 avx512f_vcvttsd2usiq<round_saeonly_name> using SWI48 mode iterator.
12464 (sse2_cvtsd2si<rex64namesuffix><round_name>): Merge insn pattern from
12465 sse2_cvtsd2si<round_name> and sse2_cvtsd2siq<round_name> using
12466 SWI48 mode iterator.
12467 (sse2_cvtsd2si<rex64namesuffix>_2): Merge insn pattern from
12468 sse2_cvtsd2si_2 and sse2_cvtsd2siq_2 using SWI48 mode iterator.
12469 (sse_cvttsd2si<rex64namesuffix><round_saeonly_name>): Merge insn
12470 pattern from sse_cvttsd2si<round_saeonly_name>
12471 and sse_cvttsd2siq<round_saeonly_name> using SWI48 mode iterator.
12473 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
12475 * config/nds32/nds32-md-auxiliary.c
12476 (nds32_valid_smw_lwm_base_p): Refine.
12477 (nds32_output_smw_single_word): Refine.
12478 (nds32_output_smw_double_word): New.
12479 * config/nds32/nds32-protos.h (nds32_output_smw_double_word): New.
12481 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
12483 * config/nds32/nds32-md-auxiliary.c (nds32_output_stack_push): Refine.
12484 (nds32_output_stack_pop): Refine.
12485 (nds32_expand_unaligned_load): Refine.
12486 (nds32_expand_unaligned_store): Refine.
12488 2018-05-19 Kuan-Lin Chen <kuanlinchentw@gmail.com>
12489 Chung-Ju Wu <jasonwucj@gmail.com>
12491 * config/nds32/constants.md: Add TP_REGNUM constant.
12492 (unspec_element): Add UNSPEC_GOTINIT, UNSPEC_GOT, UNSPEC_GOTOFF,
12493 UNSPEC_PLT, UNSPEC_TLSGD, UNSPEC_TLSLD, UNSPEC_TLSIE, UNSPEC_TLSLE and
12495 * config/nds32/nds32-doubleword.md: Consider flag_pic.
12496 * config/nds32/nds32-dspext.md (mov<mode>): Expand TLS and PIC cases.
12497 * config/nds32/nds32-predicates.c (nds32_const_unspec_p): New.
12498 * config/nds32/nds32-md-auxiliary.c: Implementation that support TLS
12499 and PIC code generation.
12500 * config/nds32/nds32-protos.h: Declarations that support TLS and PIC
12502 * config/nds32/nds32-relax-opt.c: Consider TLS and PIC for relax
12504 * config/nds32/nds32.md: Support TLS and PIC.
12505 * config/nds32/nds32.c: Support TLS and PIC.
12506 * config/nds32/nds32.h (nds32_relax_insn_type): New enum type.
12507 * config/nds32/predicates.md (nds32_nonunspec_symbolic_operand): New
12510 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
12512 * config/nds32/nds32-predicates.c (const_vector_to_hwint): Use machine
12513 mode with E_ prefix.
12515 2018-05-19 Kuan-Lin Chen <kuanlinchentw@gmail.com>
12516 Chung-Ju Wu <jasonwucj@gmail.com>
12518 * config/nds32/constants.md (unspec_element): Add UNSPEC_ICT.
12519 * config/nds32/nds32-md-auxiliary.c
12520 (symbolic_reference_mentioned_p): New.
12521 (nds32_legitimize_ict_address): New.
12522 (nds32_expand_ict_move): New.
12523 (nds32_indirect_call_referenced_p): New.
12524 (nds32_symbol_binds_local_p): Delete.
12525 (nds32_long_call_p): Modify.
12526 * config/nds32/nds32-opts.h (nds32_ict_model_type): New enum type.
12527 * config/nds32/nds32-protos.h
12528 (symbolic_reference_mentioned_p): Declare.
12529 (nds32_legitimize_ict_address): Declare.
12530 (nds32_expand_ict_move): Declare.
12531 (nds32_indirect_call_referenced_p): Declare.
12532 * config/nds32/nds32-relax-opt.c (nds32_ict_const_p): New.
12533 (nds32_relax_group): Use nds32_ict_const_p as condition.
12534 * config/nds32/nds32.c (nds32_attribute_table): Add "indirect_call".
12535 (nds32_asm_file_start): Output ict_model directive in asm code.
12536 (nds32_legitimate_address_p): Consider indirect call.
12537 (nds32_print_operand): Consider indirect call.
12538 (nds32_print_operand_address): Consider indirect call.
12539 (nds32_insert_attributes): Handle "indirect_call" attribute.
12540 (TARGET_LEGITIMATE_ADDRESS_P): Define.
12541 (TARGET_LEGITIMATE_CONSTANT_P): Define.
12542 (TARGET_CANNOT_FORCE_CONST_MEM): Define.
12543 (TARGET_DELEGITIMIZE_ADDRESS): Define.
12544 (TARGET_ASM_OUTPUT_ADDR_CONST_EXTRA): Define.
12545 * config/nds32/nds32.h (SYMBOLIC_CONST_P): Define.
12546 (TARGET_ICT_MODEL_SMALL): Define.
12547 (TARGET_ICT_MODEL_LARGE): Define.
12548 * config/nds32/nds32.md (movsi): Consider ict model.
12549 (call, call_value): Consider ict model.
12550 (sibcall, sibcall_value): Consider ict model.
12551 * config/nds32/nds32.opt (mict-model): New option.
12552 * config/nds32/predicates.md (nds32_symbolic_operand): Consider ict
12555 2018-05-18 Kito Cheng <kito.cheng@gmail.com>
12556 Monk Chiang <sh.chiang04@gmail.com>
12557 Jim Wilson <jimw@sifive.com>
12559 * common/config/riscv/riscv-common.c (riscv_parse_arch_string):
12560 Add support to parse rv32e*. Clear MASK_RVE for rv32i and rv64i.
12561 * config.gcc (riscv*-*-*): Add support for rv32e* and ilp32e.
12562 * config/riscv/riscv-c.c (riscv_cpu_cpp_builtins): Define
12563 __riscv_32e when TARGET_RVE. Handle ABI_ILP32E as soft-float ABI.
12564 * config/riscv/riscv-opts.h (riscv_abi_type): Add ABI_ILP32E.
12565 * config/riscv/riscv.c (riscv_compute_frame_info): When TARGET_RVE,
12566 compute save_libcall_adjustment properly.
12567 (riscv_option_override): Call error if TARGET_RVE and not ABI_ILP32E.
12568 (riscv_conditional_register_usage): Handle TARGET_RVE and ABI_ILP32E.
12569 * config/riscv/riscv.h (UNITS_PER_FP_ARG): Handle ABI_ILP32E.
12570 (STACK_BOUNDARY, ABI_STACK_BOUNDARY): Handle TARGET_RVE.
12571 (GP_REG_LAST, MAX_ARGS_IN_REGISTERS): Likewise.
12572 (ABI_SPEC): Handle mabi=ilp32e.
12573 * config/riscv/riscv.opt (abi_type): Add ABI_ILP32E.
12574 (RVE): Add RVE mask.
12575 * doc/invoke.texi (RISC-V options) <-mabi>: Add ilp32e info.
12576 <-march>: Add rv32e as an example.
12578 2018-05-18 Marc Glisse <marc.glisse@inria.fr>
12581 * tree-ssa-structalias.c (create_variable_info_for_1): Extra argument.
12582 (intra_create_variable_infos): Handle C++ constructors.
12584 2018-05-18 Martin Liska <mliska@suse.cz>
12586 * passes.def: Remove a redundant pass.
12588 2018-05-18 Eric Botcazou <ebotcazou@adacore.com>
12591 * config/sparc/sparc.c (sparc_expand_builtin): Always initialize op[0].
12593 2018-05-18 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12595 * config/arm/arm-cpus.in (armv3m, mode32): Delete features.
12597 (ARMv2, ARMv3, ARMv3m): Delete fgroups.
12599 (armv2, armv2a, armv3, armv3m): Delete architectures.
12600 (arm2, arm250, arm3, arm6, arm60, arm600, arm610, arm620,
12601 arm7, arm7d, arm7di, arm70, arm700, arm700i, arm710, arm720,
12602 arm710c, arm7100, arm7500, arm7500fe, arm7m, arm7dm, arm7dmi):
12604 * config/arm/arm.md (maddsidi4): Remove check for arm_arch3m.
12605 (*mulsidi3adddi): Likewise.
12606 (mulsidi3): Likewise.
12607 (*mulsidi3_nov6): Likewise.
12608 (umulsidi3): Likewise.
12609 (umulsidi3_nov6): Likewise.
12610 (umaddsidi4): Likewise.
12611 (*umulsidi3adddi): Likewise.
12612 (smulsi3_highpart): Likewise.
12613 (*smulsi3_highpart_nov6): Likewise.
12614 (umulsi3_highpart): Likewise.
12615 (*umulsi3_highpart_nov6): Likewise.
12616 * config/arm/arm.h (arm_arch3m): Delete.
12617 * config/arm/arm.c (arm_arch3m): Delete.
12618 (arm_option_override_internal): Update armv3-related comment.
12619 (arm_configure_build_target): Delete use of isa_bit_mode32.
12620 (arm_option_reconfigure_globals): Delete set of arm_ach3m.
12621 (arm_rtx_costs_internal): Delete check of arm_arch3m.
12622 * config/arm/arm-fixed.md (mulsq3): Delete check for arm_arch3m.
12623 (mulsa3): Likewise.
12624 (mulusa3): Likewise.
12625 * config/arm/arm-protos.h (arm_arch3m): Delete.
12626 * config/arm/arm-tables.opt: Regenerate.
12627 * config/arm/arm-tune.md: Likewise.
12628 * config/arm/t-arm-elf (all_early_nofp): Delete mentions of
12629 deleted architectures.
12631 2018-05-18 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12633 * config/arm/arm-cpus.in (armv5, armv5e): Delete features.
12634 (armv5t, armv5te): New features.
12635 (ARMv5, ARMv5e): Delete fgroups.
12636 (ARMv5t, ARMv5te): Adjust for above changes.
12637 (ARMv6m): Likewise.
12638 (armv5, armv5e): Delete arches.
12639 * config/arm/arm.md (*call_reg_armv5): Use arm_arch5t instead of
12641 (*call_reg_arm): Likewise.
12642 (*call_value_reg_armv5): Likewise.
12643 (*call_value_reg_arm): Likewise.
12644 (*call_symbol): Likewise.
12645 (*call_value_symbol): Likewise.
12646 (*sibcall_insn): Likewise.
12647 (*sibcall_value_insn): Likewise.
12648 (clzsi2): Likewise.
12649 (prefetch): Likewise.
12650 (define_split and define_peephole2 dependent on arm_arch5):
12652 * config/arm/arm.h (TARGET_LDRD): Use arm_arch5te instead of
12654 (TARGET_ARM_QBIT): Likewise.
12655 (TARGET_DSP_MULTIPLY): Likewise.
12656 (enum base_architecture): Delete BASE_ARCH_5, BASE_ARCH_5E.
12657 (arm_arch5, arm_arch5e): Delete.
12658 (arm_arch5t, arm_arch5te): Declare.
12659 * config/arm/arm.c (arm_arch5, arm_arch5e): Delete.
12660 (arm_arch5t): Declare.
12661 (arm_option_reconfigure_globals): Update for the above.
12662 (arm_options_perform_arch_sanity_checks): Update comment, replace
12663 use of arm_arch5 with arm_arch5t.
12664 (use_return_insn): Likewise.
12665 (arm_emit_call_insn): Likewise.
12666 (output_return_instruction): Likewise.
12667 (arm_final_prescan_insn): Likewise.
12668 (arm_coproc_builtin_available): Likewise.
12669 * config/arm/arm-c.c (arm_cpu_builtins): Replace arm_arch5 and
12670 arm_arch5e with arm_arch5t and arm_arch5te.
12671 * config/arm/arm-protos.h (arm_arch5, arm_arch5e): Delete.
12672 (arm_arch5t, arm_arch5te): Declare.
12673 * config/arm/arm-tables.opt: Regenerate.
12674 * config/arm/t-arm-elf: Remove references to armv5, armv5e.
12675 * config/arm/t-multilib: Likewise.
12676 * config/arm/thumb1.md (*call_reg_thumb1_v5): Check arm_arch5t
12677 instead of arm_arch5.
12678 (*call_reg_thumb1): Likewise.
12679 (*call_value_reg_thumb1_v5): Likewise.
12680 (*call_value_reg_thumb1): Likewise.
12681 * config/arm/vxworks.h (TARGET_OS_CPP_BUILTINS): Remove now
12683 * doc/invoke.texi (ARM Options): Remove references to armv5, armv5e.
12685 2018-05-18 Martin Liska <mliska@suse.cz>
12687 PR gcov-profile/84846
12688 * doc/gcov.texi: Document -t option of gcov tool.
12690 2018-05-18 Martin Liska <mliska@suse.cz>
12692 PR gcov-profile/84846
12693 * gcov.c (print_usage): Add new -t option.
12694 (process_args): Handle the option.
12695 (generate_results): Use stdout as output when requested by
12698 2018-05-18 Martin Liska <mliska@suse.cz>
12700 PR gcov-profile/84846
12701 * coverage.c (coverage_init): Write PWD to .gcno file.
12702 * doc/gcov.texi: Document how working directory is printed.
12703 * gcov-dump.c (dump_gcov_file): Print PWD.
12704 * gcov.c (output_intermediate_file): Likewise.
12705 (read_graph_file): Read PWD string.
12706 (output_lines): Print PWD.
12708 2018-05-18 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
12710 PR middle-end/85817
12711 * ipa-pure-const.c (malloc_candidate_p): Remove the check integer_zerop
12712 for retval and return false if all args to phi are zero.
12714 2018-05-18 Richard Biener <rguenther@suse.de>
12716 * gimple-ssa-evrp.c (class evrp_folder): Add simplify_stmt_using_ranges
12718 (evrp_dom_walker::before_dom_children): Call it.
12720 2018-05-18 Richard Biener <rguenther@suse.de>
12722 * tree-dfa.c (get_ref_base_and_extent): Use range-info to refine
12723 results when processing array refs with variable index.
12725 2018-05-18 Toon Moene <toon@moene.org>
12727 * doc/invoke.texi: Move -floop-unroll-and-jam documentation
12728 directly after that of -floop-interchange. Indicate that both
12729 options are enabled by default when specifying -O3.
12731 2018-05-18 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12733 * config/aarch64/aarch64-simd.md (vec_set<mode>): Use VALL_F16 mode
12734 iterator. Delete separate integer-mode vec_set<mode> expander.
12735 (aarch64_simd_vec_setv2di): Delete.
12736 (vec_setv2di): Delete.
12737 (aarch64_simd_vec_set<mode>): Delete all other patterns with that name.
12738 Use VALL_F16 mode iterator. Add LD1 alternative and use vwcore for
12739 the "w, r" alternative.
12741 2018-05-18 Martin Liska <mliska@suse.cz>
12743 * passes.def: Add pass_lower_switch and pass_lower_switch_O0.
12744 * tree-pass.h (make_pass_lower_switch_O0): New function.
12745 * tree-switch-conversion.c (node_has_low_bound): Remove.
12746 (node_has_high_bound): Likewise.
12747 (node_is_bounded): Likewise.
12748 (class pass_lower_switch): Make it a template type and create
12750 (pass_lower_switch::execute): Add template argument.
12751 (make_pass_lower_switch): New function.
12752 (make_pass_lower_switch_O0): New function.
12753 (do_jump_if_equal): Remove.
12754 (emit_case_nodes): Simplify to just handle all 3 cases and leave
12755 all the hard work to tree optimization passes.
12757 2018-05-18 Martin Liska <mliska@suse.cz>
12759 * dbgcnt.c (limit_low): Renamed from limit.
12760 (limit_high): New variable.
12761 (dbg_cnt_is_enabled): Check for upper limit.
12762 (dbg_cnt): Adjust dumping.
12763 (dbg_cnt_set_limit_by_index): Add new argument for high
12765 (dbg_cnt_set_limit_by_name): Likewise.
12766 (dbg_cnt_process_single_pair): Parse new format.
12767 (dbg_cnt_process_opt): Use strtok.
12768 (dbg_cnt_list_all_counters): Remove 'value' and add
12770 * doc/invoke.texi: Document changes.
12772 2018-05-18 Richard Sandiford <richard.sandiford@linaro.org>
12774 * doc/sourcebuild.texi (scalar_all_fma): Document.
12775 * tree.def (FMA_EXPR): Delete.
12776 * internal-fn.def (FMA, FMS, FNMA, FNMS): New internal functions.
12777 * internal-fn.c (ternary_direct): New macro.
12778 (expand_ternary_optab_fn): Likewise.
12779 (direct_ternary_optab_supported_p): Likewise.
12780 * Makefile.in (build/genmatch.o): Depend on case-fn-macros.h.
12781 * builtins.c (fold_builtin_fma): Delete.
12782 (fold_builtin_3): Don't call it.
12783 * cfgexpand.c (expand_debug_expr): Remove FMA_EXPR handling.
12784 * expr.c (expand_expr_real_2): Likewise.
12785 * fold-const.c (operand_equal_p): Likewise.
12786 (fold_ternary_loc): Likewise.
12787 * gimple-pretty-print.c (dump_ternary_rhs): Likewise.
12788 * gimple.c (DEFTREECODE): Likewise.
12789 * gimplify.c (gimplify_expr): Likewise.
12790 * optabs-tree.c (optab_for_tree_code): Likewise.
12791 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
12792 * tree-eh.c (operation_could_trap_p): Likewise.
12793 (stmt_could_throw_1_p): Likewise.
12794 * tree-inline.c (estimate_operator_cost): Likewise.
12795 * tree-pretty-print.c (dump_generic_node): Likewise.
12796 (op_code_prio): Likewise.
12797 * tree-ssa-loop-im.c (stmt_cost): Likewise.
12798 * tree-ssa-operands.c (get_expr_operands): Likewise.
12799 * tree.c (commutative_ternary_tree_code, add_expr): Likewise.
12800 * fold-const-call.h (fold_fma): Delete.
12801 * fold-const-call.c (fold_const_call_ssss): Handle CFN_FMS,
12802 CFN_FNMA and CFN_FNMS.
12803 (fold_fma): Delete.
12804 * genmatch.c (combined_fn): New enum.
12805 (commutative_ternary_tree_code): Remove FMA_EXPR handling.
12806 (commutative_op): New function.
12807 (commutate): Use it. Handle more than 2 operands.
12808 (dt_operand::gen_gimple_expr): Use commutative_op.
12809 (parser::parse_expr): Allow :c to be used with non-binary
12810 operators if the commutative operand is known.
12811 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Handle
12812 CFN_FMS, CFN_FNMA and CFN_FNMS.
12813 (backprop::process_assign_use): Remove FMA_EXPR handling.
12814 * hsa-gen.c (gen_hsa_insns_for_operation_assignment): Likewise.
12815 (gen_hsa_fma): New function.
12816 (gen_hsa_insn_for_internal_fn_call): Use it for IFN_FMA, IFN_FMS,
12817 IFN_FNMA and IFN_FNMS.
12818 * match.pd: Add folds for IFN_FMS, IFN_FNMA and IFN_FNMS.
12819 * gimple-fold.h (follow_all_ssa_edges): Declare.
12820 * gimple-fold.c (follow_all_ssa_edges): New function.
12821 * tree-ssa-math-opts.c (convert_mult_to_fma_1): Use the
12822 gimple_build interface and use follow_all_ssa_edges to fold the result.
12823 (convert_mult_to_fma): Use direct_internal_fn_suppoerted_p
12824 instead of checking for optabs directly.
12825 * config/i386/i386.c (ix86_add_stmt_cost): Recognize FMAs as calls
12826 rather than FMA_EXPRs.
12827 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Create a
12828 call to IFN_FMA instead of an FMA_EXPR.
12830 2018-05-17 Jim Wilson <jimw@sifive.com>
12832 * expr.c (do_tablejump): When converting index to Pmode, if we have a
12833 sign extended promoted subreg, and the range does not have the sign bit
12834 set, then do a sign extend.
12836 * config/riscv/riscv.c (riscv_extend_comparands): In unsigned QImode
12837 test, check for sign extended subreg and/or constant operands, and
12838 do a sign extend in that case.
12840 2018-05-17 Steve Ellcey <sellcey@cavium.com>
12842 * config/aarch64/thunderx2t99.md (thunderx2t99_ls_both): Delete.
12843 (thunderx2t99_multiple): Delete psuedo-units from used cpus.
12845 (thunderx2t99_alu_shift): Remove alu_shift_reg, alus_shift_reg.
12846 Change logics_shift_reg to logics_shift_imm.
12847 (thunderx2t99_fp_loadpair_basic): Delete.
12848 (thunderx2t99_fp_storepair_basic): Delete.
12849 (thunderx2t99_asimd_int): Add neon_sub and neon_sub_q types.
12850 (thunderx2t99_asimd_polynomial): Delete.
12851 (thunderx2t99_asimd_fp_simple): Add neon_fp_mul_s_scalar_q
12852 and neon_fp_mul_d_scalar_q.
12853 (thunderx2t99_asimd_fp_conv): Add *int_to_fp* types.
12854 (thunderx2t99_asimd_misc): Delete neon_dup and neon_dup_q.
12855 (thunderx2t99_asimd_recip_step): Add missing *sqrt* types.
12856 (thunderx2t99_asimd_lut): Add missing tbl types.
12857 (thunderx2t99_asimd_ext): Delete.
12858 (thunderx2t99_asimd_load1_1_mult): Delete.
12859 (thunderx2t99_asimd_load1_2_mult): Delete.
12860 (thunderx2t99_asimd_load1_ldp): New.
12861 (thunderx2t99_asimd_load1): New.
12862 (thunderx2t99_asimd_load2): Add missing *load2* types.
12863 (thunderx2t99_asimd_load3): New.
12864 (thunderx2t99_asimd_load4): New.
12865 (thunderx2t99_asimd_store1_1_mult): Delete.
12866 (thunderx2t99_asimd_store1_2_mult): Delete.
12867 (thunderx2t99_asimd_store2_mult): Delete.
12868 (thunderx2t99_asimd_store2_onelane): Delete.
12869 (thunderx2t99_asimd_store_stp): New.
12870 (thunderx2t99_asimd_store1): New.
12871 (thunderx2t99_asimd_store2): New.
12872 (thunderx2t99_asimd_store3): New.
12873 (thunderx2t99_asimd_store4): New.
12875 2018-05-17 Jerome Lambourg <lambourg@adacore.com>
12877 * config/arm/arm_cmse.h (cmse_nsfptr_create, cmse_is_nsfptr): Remove
12878 #include <stdint.h>. Replace intptr_t with __INTPTR_TYPE__.
12880 2018-05-17 Pat Haugen <pthaugen@us.ibm.com>
12881 Segher Boessenkool <segher@kernel.crashing.org>
12884 * config/rs6000/rs6000.c (rs6000_output_move_128bit): Check dest
12887 2018-05-17 Richard Biener <rguenther@suse.de>
12889 * tree-ssa-dse.c (dse_classify_store): Fix iterator increment
12890 for pruning loop and prune defs feeding only already visited PHIs.
12892 2018-05-17 Richard Biener <rguenther@suse.de>
12894 * tree-ssa-sccvn.c (vn_reference_lookup_3): Improve memset handling.
12896 2018-05-17 Bin Cheng <bin.cheng@arm.com>
12897 Richard Biener <rguenther@suse.de>
12899 PR tree-optimization/85793
12900 * tree-vect-stmts.c (vectorizable_load): Handle 1 element-wise load
12901 for VMAT_ELEMENTWISE.
12903 2018-05-17 Richard Sandiford <richard.sandiford@linaro.org>
12905 * internal-fn.h (lookup_internal_fn): Declare
12906 * internal-fn.c (lookup_internal_fn): New function.
12907 * gimple.c (gimple_build_call_from_tree): Handle calls to
12908 internal functions.
12909 * gimple-pretty-print.c (dump_gimple_call): Print "." before
12910 internal function names.
12911 * tree-pretty-print.c (dump_generic_node): Likewise.
12912 * tree-ssa-scopedtables.c (expr_hash_elt::print): Likewise.
12914 2018-05-17 Richard Sandiford <richard.sandiford@linaro.org>
12916 * gimple-fold.h (gimple_build): Make the function forms take
12917 combined_fn rather than built_in_function.
12918 (gimple_simplify): Likewise.
12919 * gimple-match-head.c (gimple_simplify): Likewise.
12920 * gimple-fold.c (gimple_build): Likewise.
12921 * tree-vect-loop.c (get_initial_def_for_reduction): Use gimple_build
12922 rather than gimple_build_call_internal.
12923 (get_initial_defs_for_reduction): Likewise.
12924 (vect_create_epilog_for_reduction): Likewise.
12925 (vectorizable_live_operation): Likewise.
12927 2018-05-17 Martin Liska <mliska@suse.cz>
12929 * gimple-ssa-sprintf.c (format_directive): Do not use
12930 space in between 'G_' and '('.
12932 2018-05-17 Jakub Jelinek <jakub@redhat.com>
12935 * config/i386/i386.c (ix86_fold_builtin): Handle masked shifts
12936 even if the mask is not all ones.
12939 * config/i386/i386.c (ix86_fold_builtin): Fold shift builtins by
12941 (ix86_gimple_fold_builtin): Likewise.
12944 * config/i386/i386.c: Include tree-vector-builder.h.
12945 (ix86_vector_shift_count): New function.
12946 (ix86_fold_builtin): Fold shift builtins by scalar count.
12947 (ix86_gimple_fold_builtin): Likewise.
12949 * config/i386/avx512fintrin.h (_mm512_set_epi16, _mm512_set_epi8,
12950 _mm512_setzero): New intrinsics.
12952 2018-05-17 James Greenhalgh <james.greenhalgh@arm.com>
12953 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12955 * config/aarch64/aarch64.c (aarch64_expand_vector_init): Modify
12956 code generation for cases where splatting a value is not useful.
12957 * simplify-rtx.c (simplify_ternary_operation): Simplify
12958 vec_merge across a vec_duplicate and a paradoxical subreg forming
12959 a vector mode to a vec_concat.
12961 2018-05-17 Olga Makhotina <olga.makhotina@intel.com>
12963 * config.gcc: Support "goldmont-plus".
12964 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
12966 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
12967 PROCESSOR_GOLDMONT_PLUS.
12968 * config/i386/i386.c (m_GOLDMONT_PLUS): Define.
12969 (processor_target_table): Add "goldmont-plus".
12970 (PTA_GOLDMONT_PLUS): Define.
12971 (ix86_lea_outperforms): Add TARGET_GOLDMONT_PLUS.
12972 (get_builtin_code_for_version): Handle PROCESSOR_GOLDMONT_PLUS.
12973 (fold_builtin_cpu): Add M_INTEL_GOLDMONT_PLUS.
12974 (fold_builtin_cpu): Add "goldmont-plus".
12975 (ix86_add_stmt_cost): Add TARGET_GOLDMONT_PLUS.
12976 (ix86_option_override_internal): Add "goldmont-plus".
12977 * config/i386/i386.h (processor_costs): Define TARGET_GOLDMONT_PLUS.
12978 (processor_type): Add PROCESSOR_GOLDMONT_PLUS.
12979 * config/i386/x86-tune.def: Add m_GOLDMONT_PLUS.
12980 * doc/invoke.texi: Add goldmont-plus as x86 -march=/-mtune= CPU type.
12982 2018-05-17 Richard Biener <rguenther@suse.de>
12984 PR tree-optimization/85757
12985 * tree-ssa-dse.c (dse_classify_store): Record a PHI def and
12986 remove defs that only feed that PHI from further processing.
12988 2018-05-16 Jim Wilson <jimw@sifive.com>
12990 * config/riscv/riscv.md (<optab>si3_mask, <optab>si3_mask_1): Prepend
12992 (<optab>di3_mask, <optab>di3_mask_1): Likewise.
12994 2018-05-16 Mark Wielaard <mark@klomp.org>
12996 * dwarf2out.c (count_index_strings): New function.
12997 (output_indirect_strings): Call count_index_strings and generate
12998 header for dwarf_version >= 5.
13000 2018-05-16 Mark Wielaard <mark@klomp.org>
13002 * dwarf2out.c (dwarf_FORM): New function.
13003 (set_indirect_string): Use dwarf_FORM.
13004 (reset_indirect_string): Likewise.
13005 (size_of_die): Likewise.
13006 (value_format): Likewise.
13007 (output_die): Likewise.
13008 (add_skeleton_AT_string): Likewise.
13009 (output_macinfo_op): Likewise.
13010 (index_string): Likewise.
13011 (output_index_string_offset): Likewise.
13012 (output_index_string): Likewise.
13013 (count_index_strings): Likewise.
13015 2018-05-16 Carl Love <cel@us.ibm.com>
13017 * config/rs6000/rs6000.md (prefetch): Generate ISA 2.06 instructions
13018 dcbt and dcbtstt with TH=16 if operands[2] is 0 and Power 8 or newer.
13020 2018-05-16 Martin Jambor <mjambor@suse.cz>
13022 * ipa-prop.c (ipa_free_all_edge_args): Remove.
13023 * ipa-prop.h (ipa_free_all_edge_args): Likewise.
13025 2018-05-16 Wilco Dijkstra <wdijkstr@arm.com>
13027 * config/aarch64/aarch64.md (fma<mode>4): Change into expand pattern.
13028 (fnma<mode>4): Likewise.
13029 (fms<mode>4): Likewise.
13030 (fnms<mode>4): Likewise.
13031 (aarch64_fma<mode>4): Rename insn, reorder accumulator operand.
13032 (aarch64_fnma<mode>4): Likewise.
13033 (aarch64_fms<mode>4): Likewise.
13034 (aarch64_fnms<mode>4): Likewise.
13035 (aarch64_fnmadd<mode>4): Likewise.
13037 2018-05-16 Jason Merrill <jason@redhat.com>
13039 * tree.c (warn_deprecated_use): Return bool. Simplify logic.
13041 2018-05-16 Richard Biener <rguenther@suse.de>
13043 * tree-vectorizer.h (struct stmt_info_for_cost): Add where member.
13044 (dump_stmt_cost): Declare.
13045 (add_stmt_cost): Dump cost we add.
13046 (add_stmt_costs): New function.
13047 (vect_model_simple_cost, vect_model_store_cost, vect_model_load_cost):
13048 No longer exported.
13049 (vect_analyze_stmt): Adjust prototype.
13050 (vectorizable_condition): Likewise.
13051 (vectorizable_live_operation): Likewise.
13052 (vectorizable_reduction): Likewise.
13053 (vectorizable_induction): Likewise.
13054 * tree-vect-loop.c (vect_analyze_loop_operations): Create local
13055 cost vector to pass to vectorizable_ and record afterwards.
13056 (vect_model_reduction_cost): Take cost vector argument and adjust.
13057 (vect_model_induction_cost): Likewise.
13058 (vectorizable_reduction): Likewise.
13059 (vectorizable_induction): Likewise.
13060 (vectorizable_live_operation): Likewise.
13061 * tree-vect-slp.c (vect_create_new_slp_node): Initialize
13062 SLP_TREE_NUMBER_OF_VEC_STMTS.
13063 (vect_analyze_slp_cost_1): Remove.
13064 (vect_analyze_slp_cost): Likewise.
13065 (vect_slp_analyze_node_operations): Take visited args and
13066 a target cost vector. Avoid processing already visited stmt sets.
13067 (vect_slp_analyze_operations): Use a local cost vector to gather
13068 costs and register those of non-discarded instances.
13069 (vect_bb_vectorization_profitable_p): Use add_stmt_costs.
13070 (vect_schedule_slp_instance): Remove copying of
13071 SLP_TREE_NUMBER_OF_VEC_STMTS. Instead assert that it is not
13073 * tree-vect-stmts.c (record_stmt_cost): Remove path directly
13074 adding cost. Record cost entry location.
13075 (vect_prologue_cost_for_slp_op): Function to compute cost of
13076 a constant or invariant generated for SLP vect in the prologue,
13077 split out from vect_analyze_slp_cost_1.
13078 (vect_model_simple_cost): Make static. Adjust for SLP costing.
13079 (vect_model_promotion_demotion_cost): Likewise.
13080 (vect_model_store_cost): Likewise, make static.
13081 (vect_model_load_cost): Likewise.
13082 (vectorizable_bswap): Add cost vector arg and adjust.
13083 (vectorizable_call): Likewise.
13084 (vectorizable_simd_clone_call): Likewise.
13085 (vectorizable_conversion): Likewise.
13086 (vectorizable_assignment): Likewise.
13087 (vectorizable_shift): Likewise.
13088 (vectorizable_operation): Likewise.
13089 (vectorizable_store): Likewise.
13090 (vectorizable_load): Likewise.
13091 (vectorizable_condition): Likewise.
13092 (vectorizable_comparison): Likewise.
13093 (can_vectorize_live_stmts): Likewise.
13094 (vect_analyze_stmt): Likewise.
13095 (vect_transform_stmt): Adjust calls to vectorizable_*.
13096 * tree-vectorizer.c: Include gimple-pretty-print.h.
13097 (dump_stmt_cost): New function.
13099 2018-05-16 Richard Biener <rguenther@suse.de>
13101 * params.def (PARAM_DSE_MAX_ALIAS_QUERIES_PER_STORE): New param.
13102 * doc/invoke.texi (dse-max-alias-queries-per-store): Document.
13103 * tree-ssa-dse.c: Include tree-ssa-loop.h.
13104 (check_name): New callback.
13105 (dse_classify_store): Track cycles via a visited bitmap of PHI
13106 defs and simplify handling of in-loop and across loop dead stores
13107 and properly fail for loop-variant refs. Handle byte-tracking with
13108 multiple defs. Use PARAM_DSE_MAX_ALIAS_QUERIES_PER_STORE for
13111 2018-05-16 Richard Sandiford <richard.sandiford@linaro.org>
13113 * tree-vectorizer.h (vect_get_vector_types_for_stmt): Declare.
13114 (vect_get_mask_type_for_stmt): Likewise.
13115 * tree-vect-slp.c (vect_two_operations_perm_ok_p): New function,
13117 (vect_build_slp_tree_1): ...here. Use vect_get_vector_types_for_stmt
13118 to determine the statement's vector type and the vector type that
13119 should be used for calculating nunits. Deal with cases in which
13120 the type has to be deferred.
13121 (vect_slp_analyze_node_operations): Use vect_get_vector_types_for_stmt
13122 and vect_get_mask_type_for_stmt to calculate STMT_VINFO_VECTYPE.
13123 * tree-vect-loop.c (vect_determine_vf_for_stmt_1)
13124 (vect_determine_vf_for_stmt): New functions, split out from...
13125 (vect_determine_vectorization_factor): ...here.
13126 * tree-vect-stmts.c (vect_get_vector_types_for_stmt)
13127 (vect_get_mask_type_for_stmt): New functions, split out from
13128 vect_determine_vectorization_factor.
13130 2018-05-16 Richard Biener <rguenther@suse.de>
13132 * tree-cfg.c (verify_gimple_assign_ternary): Properly
13133 verify the [VEC_]COND_EXPR embedded comparison.
13135 2018-05-15 Martin Sebor <msebor@redhat.com>
13137 PR tree-optimization/85753
13138 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Handle
13139 RECORD_TYPE in addition to ARRAY_TYPE.
13141 2018-05-15 Martin Sebor <msebor@redhat.com>
13143 PR middle-end/85643
13144 * calls.c (get_attr_nonstring_decl): Handle MEM_REF.
13146 2018-05-15 Richard Biener <rguenther@suse.de>
13148 * tree-ssa-dse.c (dse_classify_store): Remove use_stmt parameter,
13149 add by_clobber_p one. Change algorithm to collect all defs
13150 representing uses we need to walk and try reducing them to
13151 a single one before failing.
13152 (dse_dom_walker::dse_optimize_stmt): Adjust.
13154 2018-05-13 Mark Wielaard <mark@klomp.org>
13156 * dwarf2out.c (dwarf_OP): Handle DW_OP_addrx and DW_OP_constx.
13157 (size_of_loc_descr): Likewise.
13158 (output_loc_operands): Likewise.
13159 (output_loc_operands_raw): Likewise.
13160 (dw_addr_op): Use dwarf_OP () for DW_OP_constx and DW_OP_addrx.
13161 (resolve_addr_in_expr): Handle DW_OP_addrx and DW_OP_constx.
13162 (hash_loc_operands): Likewise.
13163 (compare_loc_operands): Likewise.
13165 2018-05-14 Mark Wielaard <mark@klomp.org>
13167 * dwarf2out.c (count_index_addrs): New function.
13168 (dwarf2out_finish): Use count_index_addrs to calculate addrs_length.
13170 2018-05-15 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
13172 PR tree-optimization/83648
13173 * ipa-pure-const.c (malloc_candidate_p): Allow function with NULL
13174 return value as malloc candidate.
13176 2018-05-15 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
13179 * ipa-pure-const.c (warn_function_malloc): Pass value of known_finite
13180 param as true in call to suggest_attribute.
13182 2018-05-14 Segher Boessenkool <segher@kernel.crashing.org>
13184 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete @itemx for
13185 -mreadonly-in-sdata.
13187 2018-05-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13189 * config/aarch64/aarch64-simd.md (*aarch64_crypto_aese_fused):
13191 (aarch64_crypto_aesd_fused): Likewise.
13193 2018-05-14 Wilco Dijkstra <wdijkstr@arm.com>
13195 * config/aarch64/aarch64.md (mov<mode>): Remove '*' in alternatives.
13196 (movsi_aarch64): Likewise.
13197 (load_pairsi): Likewise.
13198 (load_pairdi): Likewise.
13199 (store_pairsi): Likewise.
13200 (store_pairdi): Likewise.
13201 (load_pairsf): Likewise.
13202 (load_pairdf): Likewise.
13203 (store_pairsf): Likewise.
13204 (store_pairdf): Likewise.
13205 (zero_extend): Likewise.
13206 (trunc): Swap alternatives.
13207 (fcvt_target): Add '?' to prefer w over r.
13209 2018-05-14 Jakub Jelinek <jakub@redhat.com>
13212 * config/i386/i386.md: Disallow non-commutative arithmetics in
13213 last twpeephole for mem {+,-,&,|,^}= x; mem != 0 after cmpelim
13214 optimization. Use COMMUTATIVE_ARITH_P test rather than != MINUS
13215 in the peephole2 before it.
13217 2018-05-14 Sebastian Peryt <sebastian.peryt@intel.com>
13219 * common/config/i386/i386-common.c (OPTION_MASK_ISA_CLDEMOTE_SET,
13220 OPTION_MASK_ISA_CLDEMOTE_UNSET): New defines.
13221 (ix86_handle_option): Handle -mcldemote.
13222 * config.gcc: New header.
13223 * config/i386/cldemoteintrin.h: New file.
13224 * config/i386/cpuid.h (bit_CLDEMOTE): New bit.
13225 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
13227 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
13228 OPTION_MASK_ISA_CLDEMOTE.
13229 * config/i386/i386.c (ix86_target_string): Add -mcldemote.
13230 (ix86_valid_target_attribute_inner_p): Ditto.
13231 (enum ix86_builtins): Add IX86_BUILTIN_CLDEMOTE.
13232 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_cldemote.
13233 (ix86_expand_builtin): Expand IX86_BUILTIN_CLDEMOTE.
13234 * config/i386/i386.h (TARGET_CLDEMOTE, TARGET_CLDEMOTE_P): New.
13235 * config/i386/i386.md (UNSPECV_CLDEMOTE): New.
13237 * config/i386/i386.opt: Add -mcldemote.
13238 * config/i386/x86intrin.h: New header.
13239 * doc/invoke.texi: Add -mcldemote.
13241 2018-05-14 Richard Biener <rguenther@suse.de>
13243 * doc/match-and-simplify.texi: Adjust :s documentation.
13245 2018-05-14 Alexander Monakov <amonakov@ispras.ru>
13247 * sort.cc (REORDER_23): Pass the type for the temporaries instead of
13248 intended memcpy size.
13249 (REORDER_45): Likewise.
13251 2018-05-13 Alexander Monakov <amonakov@ispras.ru>
13253 * sort.cc: New file.
13254 * system.h [!CHECKING_P] (qsort): Redirect to gcc_qsort.
13255 * vec.c (qsort_chk): Use gcc_qsort.
13256 * Makefile.in (OBJS-libcommon): Add sort.o.
13257 (build/sort.o): New target. Use it...
13258 (BUILD_RTL): ... here, and...
13259 (build/gencfn-macros): ... here, and...
13260 (build/genmatch): ... here.
13262 2018-05-13 Kito Cheng <kito.cheng@gmail.com>
13263 Chung-Ju Wu <jasonwucj@gmail.com>
13265 * config.gcc (nds32*-*-*): Check that n15 is valid to --with-cpu.
13266 * config/nds32/nds32-graywolf.md: New file.
13267 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_GRAYWOLF.
13268 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n15
13270 * config/nds32/nds32-protos.h: More declarations for n15 pipeline.
13271 * config/nds32/nds32-utils.c: More implementations for n15 pipeline.
13272 * config/nds32/nds32.md (pipeline_model): Add graywolf.
13273 * config/nds32/nds32.opt (mcpu): Support n15 pipeline cpus.
13274 * config/nds32/pipelines.md: Include n15 settings.
13276 2018-05-13 Kito Cheng <kito.cheng@gmail.com>
13277 Chung-Ju Wu <jasonwucj@gmail.com>
13279 * config.gcc (nds32*-*-*): Check that n12/n13 are valid to --with-cpu.
13280 * config/nds32/nds32-n13.md: New file.
13281 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N12 and CPU_N13.
13282 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n12/n13
13284 * config/nds32/nds32-protos.h: More declarations for n12/n13 pipeline.
13285 * config/nds32/nds32.md (pipeline_model): Add n13.
13286 * config/nds32/nds32.opt (mcpu): Support n12/n13 pipeline cpus.
13287 * config/nds32/pipelines.md: Include n13 settings.
13289 2018-05-13 Kito Cheng <kito.cheng@gmail.com>
13290 Chung-Ju Wu <jasonwucj@gmail.com>
13292 * config.gcc (nds32*-*-*): Check that n10/d10 are valid to --with-cpu.
13293 * config/nds32/nds32-n10.md: New file.
13294 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N10.
13295 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n10
13297 * config/nds32/nds32-protos.h: More declarations for n10 pipeline.
13298 * config/nds32/nds32-utils.c: More implementations for n10 pipeline.
13299 * config/nds32/nds32.md (pipeline_model): Add n10.
13300 * config/nds32/nds32.opt (mcpu): Support n10 pipeline cpus.
13301 * config/nds32/pipelines.md: Include n10 settings.
13303 2018-05-13 Monk Chiang <sh.chiang04@gmail.com>
13304 Kito Cheng <kito.cheng@gmail.com>
13305 Chung-Ju Wu <jasonwucj@gmail.com>
13307 * config.gcc (nds32be-*-*): Handle --with-ext-dsp.
13308 * config/nds32/constants.md (unspec_element, unspec_volatile_element):
13309 Add enum values for DSP extension instructions.
13310 * config/nds32/constraints.md (Iu06, IU06, CVp5, CVs5, CVs2, CVhi):
13312 * config/nds32/iterators.md (shifts, shiftrt, sat_plus, all_plus,
13313 sat_minus, all_minus, plus_minus, extend, sumax, sumin, sumin_max):
13314 New code iterators.
13315 (su, zs, uk, opcode, add_rsub, add_sub): New code attributes.
13316 * config/nds32/nds32-dspext.md: New file for DSP implementation.
13317 * config/nds32/nds32-intrinsic.c: Implementation of DSP extension.
13318 * config/nds32/nds32-intrinsic.md: Likewise.
13319 * config/nds32/nds32_intrinsic.h: Likewise.
13320 * config/nds32/nds32-md-auxiliary.c: Likewise.
13321 * config/nds32/nds32-memory-manipulation.c: Consider DSP extension.
13322 * config/nds32/nds32-predicates.c (const_vector_to_hwint): New.
13323 (nds32_valid_CVp5_p, nds32_valid_CVs5_p): New.
13324 (nds32_valid_CVs2_p, nds32_valid_CVhi_p): New.
13325 * config/nds32/nds32-protos.h: New declarations for DSP extension.
13326 * config/nds32/nds32-utils.c (extract_mac_non_acc_rtx): New case
13327 TYPE_DMAC in switch statement.
13328 * config/nds32/nds32.c: New checking and implementation for DSP
13329 extension instructions.
13330 * config/nds32/nds32.h: Likewise.
13331 * config/nds32/nds32.md: Likewise.
13332 * config/nds32/nds32.opt (mhw-abs, mext-dsp): New options.
13333 * config/nds32/predicates.md: Implement new predicates for DSP
13336 2018-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
13338 * config/rs6000/rs6000.md (mov<mode>_softfloat, FMOVE32):
13339 Reformat alternatives and attributes so it is easier to identify
13340 which constraints/attributes go with which instruction.
13341 (mov<mode>_hardfloat32, FMOVE64): Likewise.
13342 (mov<mode>_softfloat32, FMOVE64): Likewise.
13343 (mov<mode>_hardfloat64, FMOVE64): Likewise.
13344 (mov<mode>_softfloat64, FMOVE64): Likewise.
13346 2018-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
13348 * doc/extend.texi (PowerPC Built-in Functions): Rename this
13350 (Basic PowerPC Built-in Functions): The new name of the
13351 subsection previously known as "PowerPC Built-in Functions".
13352 (Basic PowerPC Built-in Functions Available on all Configurations):
13354 (Basic PowerPC Built-in Functions Available on ISA 2.05): Likewise.
13355 (Basic PowerPC Built-in Functions Available on ISA 2.06): Likewise.
13356 (Basic PowerPC Built-in Functions Available on ISA 2.07): Likewise.
13357 (Basic PowerPC Built-in Functions Available on ISA 3.0): Likewise.
13359 2018-05-11 Martin Jambor <mjambor@suse.cz>
13362 * ipa-cp.c (intersect_with_plats): Check that the lattice contains
13365 2018-05-11 Richard Earnshaw <rearnsha@arm.com>
13368 * config/arm/arm-cpus.in (fgroup ARMv6m): Add be8 feature.
13370 2018-05-11 Sebastian Peryt <sebastian.peryt@intel.com>
13372 * common/config/i386/i386-common.c (OPTION_MASK_ISA_WAITPKG_SET,
13373 OPTION_MASK_ISA_WAITPKG_UNSET): New defines.
13374 (ix86_handle_option): Handle -mwaitpkg.
13375 * config.gcc: New header.
13376 * config/i386/cpuid.h (bit_WAITPKG): New bit.
13377 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mwaitpkg.
13378 * config/i386/i386-builtin-types.def ((UINT8, UNSIGNED, UINT64)): New
13380 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
13381 OPTION_MASK_ISA_WAITPKG.
13382 * config/i386/i386.c (ix86_target_string): Add -mwaitpkg.
13383 (ix86_option_override_internal): Add PTA_WAITPKG.
13384 (ix86_valid_target_attribute_inner_p): Add -mwaitpkg.
13385 (enum ix86_builtins): Add IX86_BUILTIN_UMONITOR, IX86_BUILTIN_UMWAIT,
13386 IX86_BUILTIN_TPAUSE.
13387 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_umonitor,
13388 __builtin_ia32_umwait and __builtin_ia32_tpause.
13389 (ix86_expand_builtin): Expand IX86_BUILTIN_UMONITOR,
13390 IX86_BUILTIN_UMWAIT, IX86_BUILTIN_TPAUSE.
13391 * config/i386/i386.h (TARGET_WAITPKG, TARGET_WAITPKG_P): New.
13392 * config/i386/i386.md (UNSPECV_UMWAIT, UNSPECV_UMONITOR,
13393 UNSPECV_TPAUSE): New.
13394 (umwait, umwait_rex64, umonitor_<mode>, tpause, tpause_rex64): New.
13395 * config/i386/i386.opt: Add -mwaitpkg.
13396 * config/i386/waitpkgintrin.h: New file.
13397 * config/i386/x86intrin.h: New header.
13398 * doc/invoke.texi: Add -mwaitpkg.
13400 2018-05-11 Richard Earnshaw <rearnsha@arm.com>
13403 * config/arm/arm-cpus.in: Add comment that ARMv6-m and ARMv6S-m are now
13405 (cortex-m0): Use armv6s-m isa.
13406 (cortex-m0plus): Likewise.
13407 (cortex-m1): Likewise.
13408 (cortex-m0.small-multiply): Likewise.
13409 (cortex-m0plus.small-multiply): Likewise.
13410 (cortex-m1.small-multiply): Likewise.
13412 2018-05-11 Allan Sandfeld Jensen <allan.jensen@qt.io>
13413 Jakub Jelinek <jakub@redhat.com>
13415 PR tree-optimization/85692
13416 * tree-ssa-forwprop.c (simplify_vector_constructor): Try two
13417 source permute as well.
13419 2018-05-11 Martin Liska <mliska@suse.cz>
13422 * doc/extend.texi: Document LLVM style format for no_sanitize
13425 2018-05-10 Michael Meissner <meissner@linux.vnet.ibm.com>
13427 * config/rs6000/rs6000.c (mode_supports_dq_form): Rename
13428 mode_supports_vsx_dform_quad to mode_supports_dq_form.
13429 (mode_supports_vsx_dform_quad): Likewise.
13430 (mode_supports_vmx_dform): Move these functions to be next to the
13431 other mode_supports functions.
13432 (mode_supports_dq_form): Likewise.
13433 (quad_address_p): Change calls of mode_supports_vsx_dform_quad to
13434 mode_supports_dq_form.
13435 (reg_offset_addressing_ok_p): Likewise.
13436 (offsettable_ok_by_alignment): Likewise.
13437 (rs6000_legitimate_offset_address_p): Likewise.
13438 (legitimate_lo_sum_address_p): Likewise.
13439 (rs6000_legitimize_address): Likewise.
13440 (rs6000_legitimize_reload_address): Likewise.
13441 (rs6000_secondary_reload_inner): Likewise.
13442 (rs6000_preferred_reload_class): Likewise.
13443 (rs6000_output_move_128bit): Likewise.
13445 2018-05-10 Uros Bizjak <ubizjak@gmail.com>
13447 * config/i386/i386.c (ix86_expand_builtin) <case IX86_BUILTIN_RDPID>:
13448 Generate SImode target register for null target.
13449 <case IX86_BUILTIN_XGETBV>: Ditto.
13450 <case IX86_BUILTIN_XSETBV>: Optimize LSHIFTRT generation.
13451 * config/i386/xsaveintrin.h (_xgetbv): Add missing return.
13453 2018-05-10 Carl Love <cel@us.ibm.com>
13455 * config/rs6000/rs6000.md (prefetch): Generate ISA 2.06 instructions
13456 dcbtt and dcbtstt if operands[2] is 0.
13458 2018-05-10 Uros Bizjak <ubizjak@gmail.com>
13461 * config/i386/sse.md (usadv64qi): New expander.
13463 2018-05-10 Segher Boessenkool <segher@kernel.crashing.org>
13465 * config/rs6000/altivec.md (altivec_vmrghb, altivec_vmrghh,
13466 altivec_vmrghw, altivec_vmrglb, altivec_vmrglh, altivec_vmrglw): Remove
13467 -maltivec=be support.
13468 (vec_widen_umult_even_v16qi, vec_widen_smult_even_v16qi,
13469 vec_widen_umult_even_v8hi, vec_widen_smult_even_v8hi,
13470 vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
13471 vec_widen_umult_odd_v16qi, vec_widen_smult_odd_v16qi,
13472 vec_widen_umult_odd_v8hi, vec_widen_smult_odd_v8hi,
13473 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si, altivec_vpkpx,
13474 altivec_vpks<VI_char>ss, altivec_vpks<VI_char>us,
13475 altivec_vpku<VI_char>us, altivec_vpku<VI_char>um, altivec_vsum2sws,
13476 altivec_vsumsws): Adjust.
13477 (altivec_vspltb *altivec_vspltb_internal, altivec_vsplth,
13478 *altivec_vsplth_internal, altivec_vspltw, *altivec_vspltw_internal,
13479 altivec_vspltsf, *altivec_vspltsf_internal): Remove -maltivec=be
13481 (altivec_vperm_<mode>, altivec_vperm_<mode>_uns,
13482 altivec_vupkhs<VU_char>, altivec_vupkls<VU_char>, altivec_vupkhpx,
13483 altivec_vupklpx, altivec_lvsl, altivec_lvsr): Adjust.
13484 (altivec_lve<VI_char>x): Delete expand.
13485 (*altivec_lve<VI_char>x_internal): Rename to...
13486 (altivec_lve<VI_char>x): ... this.
13487 (altivec_lvxl_<mode>): Delete expand.
13488 (*altivec_lvxl_<mode>_internal): Rename to ...
13489 (altivec_lvxl_<mode>): ... this.
13490 (altivec_stvxl_<mode>): Delete expand.
13491 (*altivec_stvxl_<mode>_internal): Rename to ...
13492 (altivec_stvxl_<mode>): ... this.
13493 (altivec_stve<VI_char>x): Delete expand.
13494 (*altivec_stve<VI_char>x_internal): Rename to ...
13495 (altivec_stve<VI_char>x): ... this.
13496 (doublee<mode>2, unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2,
13497 doubleh<mode>2, unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2,
13498 reduc_plus_scal_<mode>): Adjust.
13499 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Adjust
13501 (rs6000_cpu_cpp_builtins): Adjust.
13502 (altivec_resolve_overloaded_builtin): Remove -maltivec=be support.
13503 * config/rs6000/rs6000-protos.h (altivec_expand_lvx_be,
13504 altivec_expand_stvx_be, altivec_expand_stvex_be): Delete.
13505 * config/rs6000/rs6000.c (rs6000_option_override_internal): Remove
13506 -maltivec=be support.
13507 (rs6000_split_vec_extract_var): Adjust.
13508 (rs6000_split_v4si_init): Adjust.
13509 (swap_selector_for_mode): Delete.
13510 (altivec_expand_lvx_be, altivec_expand_stvx_be,
13511 altivec_expand_stvex_be): Delete.
13512 (altivec_expand_lv_builtin, altivec_expand_stv_builtin): Remove
13513 -maltivec=be support.
13514 (rs6000_gimple_fold_builtin): Ditto.
13515 (rs6000_generate_float2_double_code, rs6000_generate_float2_code):
13517 * config/rs6000/rs6000.h (VECTOR_ELT_ORDER_BIG): Delete.
13518 (TARGET_DIRECT_MOVE_64BIT): Adjust.
13519 * config/rs6000/rs6000.md (split for extendsidi2 for vectors): Adjust.
13520 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Delete.
13521 * config/rs6000/vsx.md (floate<mode>, unsfloatev2di, floato<mode>,
13522 unsfloatov2di, vsignedo_v2df, vsignede_v2df, vunsignedo_v2df,
13523 vunsignede_v2df, vsx_extract_<mode>_p9, *vsx_extract_si,
13524 *vsx_extract_<mode>_p8, *vsx_extract_si_<uns>float_df,
13525 *vsx_extract_si_<uns>float_<mode>, vsx_set_<mode>_p9, vsx_set_v4sf_p9,
13526 *vsx_insert_extract_v4sf_p9, *vsx_insert_extract_v4sf_p9_2, and an
13527 anonymous split): Adjust.
13528 (vsx_mergel_<mode>, vsx_mergeh_<mode>): Remove -maltivec=be support.
13529 (vsx_xxspltd_<mode>, extract4b, insert4b): Adjust.
13531 2018-05-10 Eric Botcazou <ebotcazou@adacore.com>
13533 * configure.ac (gcc_gxx_include_dir_add_sysroot): Set it to 1 only
13534 when --with-gxx-include-dir is also specified.
13535 * configure: Regenerate.
13537 2018-05-09 Jim Wilson <jimw@sifive.com>
13540 * config.gcc (riscv*-*-*): Handle --with-multilib-list.
13541 * config/riscv/t-withmultilib: New.
13542 * config/riscv/withmultilib.h: New.
13543 * doc/install.texi: Document RISC-V --with-multilib-list support.
13545 2018-05-09 Richard Biener <rguenther@suse.de>
13547 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Fill a cost
13549 (vect_bb_vectorization_profitable_p): Adjust. Compute
13550 actual scalar cost using the cost vector and the add_stmt_cost
13553 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
13555 PR rtl-optimization/85645
13556 * config/rs6000/rs6000.c (rs6000_emit_prologue_components): Put a SET
13557 in the REG_CFA_REGISTER note for LR, don't leave it empty.
13559 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
13561 PR rtl-optimization/85645
13562 * shrink-wrap.c (spread_components): Return a boolean saying if
13563 anything was changed.
13564 (try_shrink_wrapping_separate): Iterate spread_components until
13565 nothing changes anymore.
13567 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
13569 PR rtl-optimization/85645
13570 * regrename.c (build_def_use): Also kill the chains that include the
13571 destination of a REG_CFA_REGISTER note.
13573 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
13575 PR rtl-optimization/85645
13576 * regcprop.c (copyprop_hardreg_forward_1): Don't propagate into an
13577 insn that has a REG_CFA_REGISTER note.
13579 2018-05-09 Richard Sandiford <richard.sandiford@linaro.org>
13581 * cfgexpand.c (expand_clobber): New function.
13582 (expand_gimple_stmt_1): Use it.
13583 * tree-vect-stmts.c (vect_clobber_variable): New function,
13585 (vectorizable_simd_clone_call): ...here.
13586 (vectorizable_store): Emit a clobber either side of an
13587 IFN_STORE_LANES sequence.
13588 (vectorizable_load): Emit a clobber after an IFN_LOAD_LANES sequence.
13590 2018-05-09 Tom de Vries <tom@codesourcery.com>
13593 * config/nvptx/nvptx.md (define_insn "trap", define_insn "trap_if_true")
13594 (define_insn "trap_if_false"): Add exit after trap.
13596 2018-05-09 Eric Botcazou <ebotcazou@adacore.com>
13598 PR rtl-optimization/85638
13599 * bb-reorder.c: Include common/common-target.h.
13600 (create_forwarder_block): New function extracted from...
13601 (fix_up_crossing_landing_pad): ...here. Rename into...
13602 (dw2_fix_up_crossing_landing_pad): ...this.
13603 (sjlj_fix_up_crossing_landing_pad): New function.
13604 (find_rarely_executed_basic_blocks_and_crossing_edges): In SJLJ mode,
13605 call sjlj_fix_up_crossing_landing_pad if there are incoming EH edges
13606 from both partitions and exit the loop after one iteration.
13608 2018-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
13611 * doc/extend.texi (PowerPC Built-in Functions): Rename this
13613 (Basic PowerPC Built-in Functions): The new name of the
13614 subsection previously known as "PowerPC Built-in Functions".
13615 (Basic PowerPC Built-in Functions Available on all Configurations):
13617 (Basic PowerPC Built-in Functions Available on ISA 2.05): New
13619 (Basic PowerPC Built-in Functions Available on ISA 2.06): New
13621 (Basic PowerPC Built-in Functions Available on ISA 2.07): New
13623 (Basic PowerPC Built-in Functions Available on ISA 3.0): New
13626 2018-05-08 Jim Wilson <jimw@sifive.com>
13628 * config/riscv/linux.h (MUSL_ABI_SUFFIX): Delete unnecessary backslash.
13629 (LD_EMUL_SUFFIX): New.
13630 (LINK_SPEC): Use it.
13632 2018-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
13634 * doc/extend.texi (PowerPC Built-in Functions): Rename this
13636 (Basic PowerPC Built-in Functions): The new name of the
13637 subsection previously known as "PowerPC Built-in Functions".
13638 (Basic PowerPC Built-in Functions Available on all Configurations):
13640 (Basic PowerPC Built-in Functions Available on ISA 2.05): New
13642 (Basic PowerPC Built-in Functions Available on ISA 2.06): New
13644 (Basic PowerPC Built-in Functions Available on ISA 2.07): New
13646 (Basic PowerPC Built-in Functions Available on ISA 3.0): New
13649 2018-05-08 Jakub Jelinek <jakub@redhat.com>
13652 * config/i386/i386.md: Add peepholes for mem {+,-,&,|,^}= x; mem != 0
13653 after cmpelim optimization.
13655 2018-05-08 Olga Makhotina <olga.makhotina@intel.com>
13657 * config.gcc: Support "goldmont".
13658 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "goldmont".
13659 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
13660 PROCESSOR_GOLDMONT.
13661 * config/i386/i386.c (m_GOLDMONT): Define.
13662 (processor_target_table): Add "goldmont".
13663 (PTA_GOLDMONT): Define.
13664 (ix86_lea_outperforms): Add TARGET_GOLDMONT.
13665 (get_builtin_code_for_version): Handle PROCESSOR_GOLDMONT.
13666 (fold_builtin_cpu): Add M_INTEL_GOLDMONT.
13667 (fold_builtin_cpu): Add "goldmont".
13668 (ix86_add_stmt_cost): Add TARGET_GOLDMONT.
13669 (ix86_option_override_internal): Add "goldmont".
13670 * config/i386/i386.h (processor_costs): Define TARGET_GOLDMONT.
13671 (processor_type): Add PROCESSOR_GOLDMONT.
13672 * config/i386/i386.md: Add CPU "glm".
13673 * config/i386/glm.md: New file.
13674 * config/i386/x86-tune.def: Add m_GOLDMONT.
13675 * doc/invoke.texi: Add goldmont as x86 -march=/-mtune= CPU type.
13677 2018-05-08 Jakub Jelinek <jakub@redhat.com>
13680 * config/i386/i386.c (ix86_expand_sse2_abs): Handle E_V2DImode and
13682 * config/i386/sse.md (abs<mode>2): Use VI_AVX2 iterator instead of
13683 VI1248_AVX512VL_AVX512BW. Handle V2DImode and V4DImode if not
13684 TARGET_AVX512VL using ix86_expand_sse2_abs. Formatting fixes.
13687 * config/i386/i386.c (ix86_fold_builtin): Handle
13688 IX86_BUILTIN_{,P}MOVMSK{PS,PD,B}{,128,256}.
13691 * config/i386/sse.md (ssequaterinsnmode): New mode attribute.
13692 (*<extract_type>_vinsert<shuffletype><extract_suf>_0): New pattern.
13694 2018-05-08 Richard Earnshaw <rearnsha@arm.com>
13697 * config/arm/parsecpu.awk (check_cpu): Fix operator precedence.
13698 (check_arch): Likewise.
13699 (check_fpu): Return the result rather than printing it.
13700 (end arch): Fix operator precedence.
13701 (end cpu): Likewise.
13702 (END): Print the result from check_fpu.
13704 2018-05-08 Richard Sandiford <richard.sandiford@linaro.org>
13705 Alan Hayward <alan.hayward@arm.com>
13706 David Sherwood <david.sherwood@arm.com>
13708 * config/aarch64/aarch64-sve.md (*pred_cmp<cmp_op><mode>_combine)
13709 (*pred_cmp<cmp_op><mode>, *fcm<cmp_op><mode>_and_combine)
13710 (*fcmuo<mode>_and_combine, *fcm<cmp_op><mode>_and)
13711 (*fcmuo<mode>_and): New patterns.
13713 2018-05-08 Richard Sandiford <richard.sandiford@linaro.org>
13715 * config/aarch64/iterators.md (UNSPEC_COND_LO, UNSPEC_COND_LS)
13716 (UNSPEC_COND_HI, UNSPEC_COND_HS, UNSPEC_COND_UO): Delete.
13717 (SVE_INT_CMP, SVE_FP_CMP): New code iterators.
13718 (cmp_op, sve_imm_con): New code attributes.
13719 (SVE_COND_INT_CMP, imm_con): Delete.
13720 (cmp_op): Remove above unspecs from int attribute.
13721 * config/aarch64/aarch64-sve.md (*vec_cmp<cmp_op>_<mode>): Rename
13723 (*cmp<cmp_op><mode>): ...this. Use UNSPEC_MERGE_PTRUE instead of
13724 comparison-specific unspecs.
13725 (*vec_cmp<cmp_op>_<mode>_ptest): Rename to...
13726 (*cmp<cmp_op><mode>_ptest): ...this and adjust likewise.
13727 (*vec_cmp<cmp_op>_<mode>_cc): Rename to...
13728 (*cmp<cmp_op><mode>_cc): ...this and adjust likewise.
13729 (*vec_fcm<cmp_op><mode>): Rename to...
13730 (*fcm<cmp_op><mode>): ...this and adjust likewise.
13731 (*vec_fcmuo<mode>): Rename to...
13732 (*fcmuo<mode>): ...this and adjust likewise.
13733 (*pred_fcm<cmp_op><mode>): New pattern.
13734 * config/aarch64/aarch64.c (aarch64_emit_unop, aarch64_emit_binop)
13735 (aarch64_emit_sve_ptrue_op, aarch64_emit_sve_ptrue_op_cc): New
13737 (aarch64_unspec_cond_code): Remove handling of LTU, GTU, LEU, GEU
13739 (aarch64_gen_unspec_cond, aarch64_emit_unspec_cond): Delete.
13740 (aarch64_emit_sve_predicated_cond): New function.
13741 (aarch64_expand_sve_vec_cmp_int): Use aarch64_emit_sve_ptrue_op_cc.
13742 (aarch64_emit_unspec_cond_or): Replace with...
13743 (aarch64_emit_sve_or_conds): ...this new function. Use
13744 aarch64_emit_sve_ptrue_op for the individual comparisons and
13745 aarch64_emit_binop to OR them together.
13746 (aarch64_emit_inverted_unspec_cond): Replace with...
13747 (aarch64_emit_sve_inverted_cond): ...this new function. Use
13748 aarch64_emit_sve_ptrue_op for the comparison and
13749 aarch64_emit_unop to invert the result.
13750 (aarch64_expand_sve_vec_cmp_float): Update after the above
13751 changes. Use aarch64_emit_sve_ptrue_op for native comparisons.
13753 2018-05-07 Nathan Sidwell <nathan@acm.org>
13755 * doc/invoke.texi (C++ Dialect Options): Remove -ffor-scope.
13756 * doc/extend.texi (Deprecated Features): Remove -fno-for-scope
13757 (Backwards Compatibility): Likewise.
13759 2018-05-07 Luis Machado <luis.machado@linaro.org>
13763 2018-05-07 Luis Machado <luis.machado@linaro.org>
13765 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
13766 <prefetch_dynamic_strides>: New const bool field.
13767 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
13768 prefetch_dynamic_strides.
13769 (exynosm1_prefetch_tune): Likewise.
13770 (thunderxt88_prefetch_tune): Likewise.
13771 (thunderx_prefetch_tune): Likewise.
13772 (thunderx2t99_prefetch_tune): Likewise.
13773 (qdf24xx_prefetch_tune): Likewise. Set prefetch_dynamic_strides
13775 (aarch64_override_options_internal): Update to set
13776 PARAM_PREFETCH_DYNAMIC_STRIDES.
13777 * doc/invoke.texi (prefetch-dynamic-strides): Document new option.
13778 * params.def (PARAM_PREFETCH_DYNAMIC_STRIDES): New.
13779 * params.h (PARAM_PREFETCH_DYNAMIC_STRIDES): Define.
13780 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Account for
13781 prefetch-dynamic-strides setting.
13783 2018-05-07 Luis Machado <luis.machado@linaro.org>
13785 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
13786 <minimum_stride>: New const int field.
13787 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
13788 minimum_stride field.
13789 (exynosm1_prefetch_tune): Likewise.
13790 (thunderxt88_prefetch_tune): Likewise.
13791 (thunderx_prefetch_tune): Likewise.
13792 (thunderx2t99_prefetch_tune): Likewise.
13793 (qdf24xx_prefetch_tune): Likewise. Set minimum_stride to 2048.
13794 (aarch64_override_options_internal): Update to set
13795 PARAM_PREFETCH_MINIMUM_STRIDE.
13796 * doc/invoke.texi (prefetch-minimum-stride): Document new option.
13797 * params.def (PARAM_PREFETCH_MINIMUM_STRIDE): New.
13798 * params.h (PARAM_PREFETCH_MINIMUM_STRIDE): Define.
13799 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Return false if
13800 stride is constant and is below the minimum stride threshold.
13802 2018-05-07 Luis Machado <luis.machado@linaro.org>
13804 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune) <l2_cache_size>: Set
13807 2018-05-07 Luis Machado <luis.machado@linaro.org>
13809 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
13810 <prefetch_dynamic_strides>: New const bool field.
13811 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
13812 prefetch_dynamic_strides.
13813 (exynosm1_prefetch_tune): Likewise.
13814 (thunderxt88_prefetch_tune): Likewise.
13815 (thunderx_prefetch_tune): Likewise.
13816 (thunderx2t99_prefetch_tune): Likewise.
13817 (qdf24xx_prefetch_tune): Likewise. Set prefetch_dynamic_strides
13819 (aarch64_override_options_internal): Update to set
13820 PARAM_PREFETCH_DYNAMIC_STRIDES.
13821 * doc/invoke.texi (prefetch-dynamic-strides): Document new option.
13822 * params.def (PARAM_PREFETCH_DYNAMIC_STRIDES): New.
13823 * params.h (PARAM_PREFETCH_DYNAMIC_STRIDES): Define.
13824 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Account for
13825 prefetch-dynamic-strides setting.
13827 2018-05-07 Luis Machado <luis.machado@linaro.org>
13829 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
13830 <minimum_stride>: New const int field.
13831 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
13832 minimum_stride field.
13833 (exynosm1_prefetch_tune): Likewise.
13834 (thunderxt88_prefetch_tune): Likewise.
13835 (thunderx_prefetch_tune): Likewise.
13836 (thunderx2t99_prefetch_tune): Likewise.
13837 (qdf24xx_prefetch_tune): Likewise. Set minimum_stride to 2048.
13838 (aarch64_override_options_internal): Update to set
13839 PARAM_PREFETCH_MINIMUM_STRIDE.
13840 * doc/invoke.texi (prefetch-minimum-stride): Document new option.
13841 * params.def (PARAM_PREFETCH_MINIMUM_STRIDE): New.
13842 * params.h (PARAM_PREFETCH_MINIMUM_STRIDE): Define.
13843 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Return false if
13844 stride is constant and is below the minimum stride threshold.
13846 2018-05-06 Jakub Jelinek <jakub@redhat.com>
13849 * cfgexpand.c (expand_asm_stmt): Don't create a temporary if
13850 the type is addressable. Don't force op into register if it has
13853 2018-05-05 Roland McGrath <mcgrathr@google.com>
13856 * varasm.c (default_section_type_flags): Set SECTION_NOTYPE for
13857 any section for which we don't know a specific type it should have,
13858 regardless of name. Previously this was done only for the exact
13859 names ".init_array", ".fini_array", and ".preinit_array".
13860 (default_elf_asm_named_section): Add comment about
13861 relationship with default_section_type_flags and SECTION_NOTYPE.
13862 (get_section): Don't consider it a type conflict if one side has
13863 SECTION_NOTYPE and the other doesn't, as long as neither has the
13864 SECTION_BSS et al used in the default_section_type_flags logic.
13866 2018-05-05 Tom de Vries <tom@codesourcery.com>
13869 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG_3): Define.
13870 (workaround_barsyncs): New function.
13871 (nvptx_reorg): Use workaround_barsyncs.
13872 * config/nvptx/nvptx.md (define_c_enum "unspecv"): Add UNSPECV_MEMBAR.
13873 (define_expand "nvptx_membar_cta"): New define_expand.
13874 (define_insn "*nvptx_membar_cta"): New insn.
13876 2018-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
13878 * brig-builtins.def: Add consts to ptrs etc. in BRIG builtin defs.
13879 To improve optimization opportunities.
13880 * builtin-types.def: The new needed builtin types for the above.
13882 2018-05-04 Richard Biener <rguenther@suse.de>
13884 * bb-reorder.c (sanitize_hot_paths): Release hot_bbs_to_check.
13885 * gimple-ssa-store-merging.c
13886 (imm_store_chain_info::output_merged_store): Remove redundant create,
13887 release split_store vector contents on failure.
13888 * tree-vect-slp.c (vect_schedule_slp_instance): Avoid leaking
13889 scalar stmt vector on cache hit.
13891 2018-05-04 Segher Boessenkool <segher@kernel.crashing.org>
13893 * common/config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
13895 * config.gcc (powerpc-xilinx-eabi*): Remove.
13896 * config/rs6000/predicates.md (easy_fp_constant): Remove Xilinx FP
13898 (fusion_addis_mem_combo_load): Ditto.
13899 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove Xilinx
13901 (rs6000_cpu_cpp_builtins): Ditto.
13902 * config/rs6000/rs6000-linux.c
13903 (rs6000_linux_float_exceptions_rounding_supported_p): Ditto.
13904 * config/rs6000/rs6000-opts.h (enum fpu_type_t): Delete.
13905 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove Xilinx FP
13907 (rs6000_setup_reg_addr_masks): Ditto.
13908 (rs6000_init_hard_regno_mode_ok): Ditto.
13909 (rs6000_option_override_internal): Ditto.
13910 (legitimate_lo_sum_address_p): Ditto.
13911 (rs6000_legitimize_address): Ditto.
13912 (rs6000_legitimize_reload_address): Ditto.
13913 (rs6000_legitimate_address_p): Ditto.
13914 (abi_v4_pass_in_fpr): Ditto.
13915 (setup_incoming_varargs): Ditto.
13916 (rs6000_gimplify_va_arg): Ditto.
13917 (rs6000_split_multireg_move): Ditto.
13918 (rs6000_savres_strategy): Ditto.
13919 (rs6000_emit_prologue_components): Ditto.
13920 (rs6000_emit_epilogue_components): Ditto.
13921 (rs6000_emit_prologue): Ditto.
13922 (rs6000_emit_epilogue): Ditto.
13923 (rs6000_elf_file_end): Ditto.
13924 (rs6000_function_value): Ditto.
13925 (rs6000_libcall_value): Ditto.
13926 * config/rs6000/rs6000.h: Ditto.
13927 (TARGET_MINMAX_SF, TARGET_MINMAX_DF): Delete, merge to ...
13928 (TARGET_MINMAX): ... this. New.
13929 (TARGET_SF_FPR, TARGET_DF_FPR, TARGET_SF_INSN, TARGET_DF_INSN): Delete.
13930 * config/rs6000/rs6000.md: Remove Xilinx FP support.
13931 (*movsi_internal1_single): Delete.
13932 * config/rs6000/rs6000.opt (msingle-float, mdouble-float, msimple-fpu,
13933 mfpu=, mxilinx-fpu): Delete.
13934 * config/rs6000/singlefp.h: Delete.
13935 * config/rs6000/sysv4.h: Remove Xilinx FP support.
13936 * config/rs6000/t-rs6000: Ditto.
13937 * config/rs6000/t-xilinx: Delete.
13938 * config/rs6000/titan.md: Adjust for fp_type removal.
13939 * config/rs6000/vsx.md: Remove Xilinx FP support.
13940 (VStype_simple): Delete.
13941 (VSfptype_simple, VSfptype_mul, VSfptype_div, VSfptype_sqrt): Delete.
13942 * config/rs6000/xfpu.h: Delete.
13943 * config/rs6000/xfpu.md: Delete.
13944 * config/rs6000/xilinx.h: Delete.
13945 * config/rs6000/xilinx.opt: Delete.
13946 * doc/invoke.texi (RS/6000 and PowerPC Options): Remove
13947 -msingle-float, -mdouble-float, -msimple-fpu, -mfpu=, and -mxilinx-fpu.
13949 2018-05-04 Tom de Vries <tom@codesourcery.com>
13952 * builtins.c (expand_builtin_goacc_parlevel_id_size): Handle null target
13955 2018-05-04 Richard Biener <rguenther@suse.de>
13957 PR middle-end/85627
13958 * tree-complex.c (update_complex_assignment): We are always in SSA form.
13959 (expand_complex_div_wide): Likewise.
13960 (expand_complex_operations_1): Likewise.
13961 (expand_complex_libcall): Preserve EH info of the original stmt.
13962 (tree_lower_complex): Handle removed blocks.
13963 * tree.c (build_common_builtin_nodes): Do not set ECF_NOTRHOW
13964 on complex multiplication and division libcall builtins.
13966 2018-05-04 Richard Biener <rguenther@suse.de>
13968 PR middle-end/85574
13969 * fold-const.c (negate_expr_p): Restrict negation of operand
13970 zero of a division to when we know that can happen without
13972 (fold_negate_expr_1): Likewise.
13974 2018-05-04 Jakub Jelinek <jakub@redhat.com>
13977 * real.h (real_nextafter): Declare.
13978 * real.c (real_nextafter): New function.
13979 * fold-const-call.c (fold_const_nextafter): New function.
13980 (fold_const_call_sss): Call it for CASE_CFN_NEXTAFTER and
13981 CASE_CFN_NEXTTOWARD.
13982 (fold_const_call_1): For CASE_CFN_NEXTTOWARD call fold_const_call_sss
13983 even when arg1_mode is different from arg0_mode.
13985 2018-05-03 Nathan Sidwell <nathan@acm.org>
13987 * doc/extend.texi (Deprecated Features): Remove
13988 -ffriend-injection.
13989 (Backwards Compatibility): Likewise.
13990 * doc/invoke.texi (C++ Language Options): Likewise.
13991 (C++ Dialect Options): Likewise.
13993 2018-05-03 Jakub Jelinek <jakub@redhat.com>
13996 * config/i386/avx512fintrin.h (_mm512_mullox_epi64,
13997 _mm512_mask_mullox_epi64): New intrinsics.
13999 2018-05-03 Tom de Vries <tom@codesourcery.com>
14002 * doc/sourcebuild.texi (Commands for use in dg-final, Scan optimization
14003 dump files): Add offload-tree.
14005 2018-05-03 Richard Biener <rguenther@suse.de>
14007 PR tree-optimization/85615
14008 * tree-ssa-threadupdate.c (thread_block_1): Only allow exits
14009 to loops not nested in BBs loop father to avoid creating multi-entry
14012 2018-05-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14014 PR tree-optimization/70291
14015 * tree-complex.c (expand_complex_libcall): Add type, inplace_p
14016 arguments. Change return type to tree. Emit libcall as a new
14017 statement rather than replacing existing one when inplace_p is true.
14018 (expand_complex_multiplication_components): New function.
14019 (expand_complex_multiplication): Expand floating-point complex
14020 multiplication using the above.
14021 (expand_complex_division): Rename inner_type parameter to type.
14022 Update expand_complex_libcall call-site.
14023 (expand_complex_operations_1): Update expand_complex_multiplication
14024 and expand_complex_division call-sites.
14026 2018-05-02 Jakub Jelinek <jakub@redhat.com>
14029 * config/i386/i386.md (*ashl<dwi>3_doubleword_mask,
14030 *ashl<dwi>3_doubleword_mask_1, *<shift_insn><dwi>3_doubleword_mask,
14031 *<shift_insn><dwi>3_doubleword_mask_1): In condition require that
14032 the highest significant bit of the shift count mask is clear. In
14033 check whether and[sq]i3 is needed verify that all significant bits
14034 of the shift count other than the highest are set.
14036 2018-05-02 Tom de Vries <tom@codesourcery.com>
14039 * builtins.def (DEF_GOACC_BUILTIN_ONLY): Define.
14040 * omp-builtins.def (BUILT_IN_GOACC_PARLEVEL_ID)
14041 (BUILT_IN_GOACC_PARLEVEL_SIZE): New builtin.
14042 * builtins.c (expand_builtin_goacc_parlevel_id_size): New function.
14043 (expand_builtin): Call expand_builtin_goacc_parlevel_id_size.
14044 * doc/extend.texi (Other Builtins): Add __builtin_goacc_parlevel_id and
14045 __builtin_goacc_parlevel_size.
14047 2018-05-02 Richard Biener <rguenther@suse.de>
14049 PR tree-optimization/85597
14050 * tree-vect-stmts.c (vectorizable_operation): For ternary SLP
14051 do not use split vect_get_vec_defs call but call vect_get_slp_defs
14054 2018-05-02 Tom de Vries <tom@codesourcery.com>
14057 * doc/sourcebuild.texi (Commands for use in dg-final, Scan optimization
14058 dump files): Add ltrans-tree.
14060 2018-05-02 Tom de Vries <tom@codesourcery.com>
14063 * doc/sourcebuild.texi (Commands for use in dg-final, Scan optimization
14064 dump files): Add wpa-ipa.
14066 2018-05-02 Segher Boessenkool <segher@kernel.crashing.org>
14068 * config.gcc (powerpc*-*-*): Remove paired.h. Unsupport the
14069 powerpc*-*-linux*paired* target.
14070 * config/rs6000/750cl.h: Delete.
14071 * config/rs6000/paired.h: Delete.
14072 * config/rs6000/paired.md: Delete.
14073 * config/rs6000/predicates.md (easy_vector_constant): Remove paired
14075 * config/rs6000/rs6000-builtin.def: Remove paired float support.
14076 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
14077 comment. Remove paired float support.
14078 * config/rs6000/rs6000-modes.def: Remove V2SF and V2SI.
14079 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete
14081 * config/rs6000/rs6000-protos.h (paired_expand_vector_init,
14082 paired_emit_vector_cond_expr, paired_expand_vector_move): Delete
14084 * config/rs6000/rs6000.c: Remove paired float support.
14085 (paired_expand_vector_init, paired_expand_vector_move,
14086 paired_emit_vector_compare, paired_emit_vector_cond_expr,
14087 (paired_expand_lv_builtin, paired_expand_stv_builtin,
14088 paired_expand_builtin, paired_expand_predicate_builtin,
14089 paired_init_builtins): Delete.
14090 * config/rs6000/rs6000.h: Remove paired float support.
14091 * config/rs6000/rs6000.md: Remove paired float support.
14092 (move_from_CR_ov_bit): Delete.
14093 * config/rs6000/rs6000.opt (mpaired): Delete.
14094 * config/rs6000/t-rs6000: Remove paired.md from MD_INCLUDES.
14095 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mpaired.
14097 2018-05-02 Richard Biener <rguenther@suse.de>
14099 PR middle-end/85567
14100 * gimplify.c (gimplify_save_expr): When in SSA form allow
14101 SAVE_EXPRs to compute to SSA vars.
14103 2018-05-02 Jakub Jelinek <jakub@redhat.com>
14106 * config/i386/i386.md (*ashl<dwi>3_doubleword_mask,
14107 *ashl<dwi>3_doubleword_mask_1, *<shift_insn><dwi>3_doubleword_mask,
14108 *<shift_insn><dwi>3_doubleword_mask_1): If and[sq]i3 is needed, don't
14109 clobber operands[2], instead use a new pseudo. Formatting fixes.
14111 2018-05-02 Richard Sandiford <richard.sandiford@linaro.org>
14113 PR tree-optimization/85586
14114 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Only
14115 exit early for statements in the same group if the accesses are
14118 2018-05-02 Tom de Vries <tom@codesourcery.com>
14121 * lto-wrapper.c (compile_offload_image): Add "could not find mkoffload"
14124 2018-05-01 Marc Glisse <marc.glisse@inria.fr>
14126 PR tree-optimization/85143
14127 * match.pd (A<B&A<C): Extend to BIT_IOR_EXPR.
14129 2018-05-01 Tom de Vries <tom@codesourcery.com>
14132 * config/nvptx/mkoffload.c (main): Suggest using -B in "offload compiler
14133 not found" error message.
14135 2018-05-01 Tom de Vries <tom@codesourcery.com>
14138 * vec.h (VEC_ORDERED_REMOVE_IF, VEC_ORDERED_REMOVE_IF_FROM_TO): Define.
14139 * vec.c (test_ordered_remove_if): New function.
14140 (vec_c_tests): Call test_ordered_remove_if.
14141 * dwarf2cfi.c (connect_traces): Use VEC_ORDERED_REMOVE_IF_FROM_TO.
14142 * lto-streamer-out.c (prune_offload_funcs): Use VEC_ORDERED_REMOVE_IF.
14143 * tree-vect-patterns.c (vect_pattern_recog_1): Use
14144 VEC_ORDERED_REMOVE_IF.
14146 2018-05-01 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
14148 PR tree-optimization/82665
14149 * vr-values.c (vr_values::extract_range_from_binary_expr): Handle
14150 pointer subtraction where arguments come from a memchr call.
14152 2018-05-01 Jakub Jelinek <jakub@redhat.com>
14154 * configure.ac (LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION): Use
14155 --push-state --as-needed and --pop-state instead of --as-needed and
14156 --no-as-needed if ld supports it.
14157 * configure: Regenerated.
14160 * doc/install.texi2html: Replace _002d with - and _002a with * in
14161 generated html files using sed.
14163 2018-04-30 David Malcolm <dmalcolm@redhat.com>
14166 * gcc-rich-location.c (blank_line_before_p): New function.
14167 (use_new_line): New function.
14168 (gcc_rich_location::add_fixit_insert_formatted): New function.
14169 * gcc-rich-location.h
14170 (gcc_rich_location::add_fixit_insert_formatted): New function.
14172 2018-04-30 David Malcolm <dmalcolm@redhat.com>
14174 * selftest.c (assert_streq): Rename "expected" and "actual" to
14175 "val1" and "val2". Extend NULL-handling to cover both inputs
14176 symmetrically, while still requiring both to be non-NULL for a pass.
14177 * selftest.h (assert_streq): Rename "expected" and "actual" to
14179 (ASSERT_EQ): Likewise.
14180 (ASSERT_EQ_AT): Likewise.
14181 (ASSERT_KNOWN_EQ): Likewise.
14182 (ASSERT_KNOWN_EQ_AT): Likewise.
14183 (ASSERT_NE): Likewise.
14184 (ASSERT_MAYBE_NE): Likewise.
14185 (ASSERT_MAYBE_NE_AT): Likewise.
14186 (ASSERT_STREQ): Likewise. Clarify that both must be non-NULL for
14187 the assertion to pass.
14188 (ASSERT_STREQ_AT): Likewise.
14190 2018-04-30 Jonathan Wakely <jwakely@redhat.com>
14192 * doc/invoke.texi (-fpie, -fPIE): Fix grammar and clarify
14193 interaction with -pie.
14195 2018-04-30 David Malcolm <dmalcolm@redhat.com>
14197 * selftest.h: Fix alphabetization of per-source-file selftest
14200 2018-04-30 Jason Merrill <jason@redhat.com>
14202 PR c++/61982 - dead stores to destroyed objects.
14203 * gimplify.c (gimplify_modify_expr): Simplify complex lvalue on LHS
14206 2018-04-30 Jason Merrill <jason@redhat.com>
14208 * tree.c (build_clobber): New.
14209 * tree.h: Declare it.
14210 * gimplify.c (gimplify_bind_expr, gimplify_target_expr): Use it.
14212 2018-04-30 David Malcolm <dmalcolm@redhat.com>
14214 * diagnostic-show-locus.c (layout::layout): Update for
14215 location_get_source_line returning a char_span.
14216 (struct char_span): Move to input.h.
14217 (struct correction): Update for fields in char_span becoming
14219 (struct source_line): Update for location_get_source_line
14220 returning a char_span.
14221 (layout::print_line): Likewise.
14222 * edit-context.c (edited_file::print_content): Likewise.
14223 (edited_file::print_diff_hunk): Likewise.
14224 (edited_file::print_run_of_changed_lines): Likewise.
14225 (edited_file::get_num_lines): Likewise.
14226 (edited_line::edited_line): Likewise.
14227 * final.c (asm_show_source): Likewise.
14228 * input.c (location_get_source_line): Convert return type
14229 from const char * to char_span, losing the final "line_len"
14231 (dump_location_info): Update for the above.
14232 (get_substring_ranges_for_loc): Likewise. Use a char_span
14233 when handling the literal within the line.
14234 (test_reading_source_line): Update for location_get_source_line
14235 returning a char_span.
14236 * input.h (class char_span): Move here from
14237 diagnostic-show-locus.c, converting from a struct to a class.
14238 Make data members private.
14239 (char_span::operator bool): New.
14240 (char_span::length): New.
14241 (char_span::get_buffer): New.
14242 (char_span::operator[]): New.
14243 (char_span::subspan): Make const.
14244 (char_span::xstrdup): New.
14245 (location_get_source_line): Convert return type from const char *
14246 to char_span, losing the final "line_size" param.
14248 2018-04-30 Jan Hubicka <jh@suse.cz>
14250 * lto-wrapper.c (ltrans_priorities): New static var.
14251 (cmp_priority): New.
14252 (run_gcc): Read priorities and if doing parallel build order
14253 the Makefile by them.
14255 2018-04-30 David Malcolm <dmalcolm@redhat.com>
14257 * input.h (builtins_location_check): Convert to a STATIC_ASSERT.
14259 2018-04-30 Richard Biener <rguenther@suse.de>
14261 * tree-cfg.c (verify_address): Remove base argument, add
14262 flag whether to check TREE_ADDRESSABLE and do that.
14263 (verify_expr): Remove.
14264 (verify_types_in_gimple_reference): Add pieces from verify_expr.
14265 (verify_gimple_assign_single): Likewise.
14266 (verify_gimple_switch): Likewise.
14267 (verify_expr_location_1): Dereference tp once. Add (disabled)
14268 piece from verify_expr.
14269 (verify_gimple_in_cfg): Do not call verify_expr on all ops.
14271 2018-04-30 Claudiu Zissulescu <claziss@synopsys.com>
14273 * config/arc/linux.h (CLEAR_INSN_CACHE): Define.
14275 2018-04-30 Claudiu Zissulescu <claziss@synopsys.com>
14277 * config/arc/arc-protos.h (prepare_extend_operands): Remove.
14278 (small_data_pattern): Likewise.
14279 (arc_rewrite_small_data): Likewise.
14280 * config/arc/arc.c (LEGITIMATE_SMALL_DATA_OFFSET_P): Remove.
14281 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Likewise.
14282 (get_symbol_alignment): New function.
14283 (legitimate_small_data_address_p): Likewise.
14284 (legitimate_scaled_address): Update, call
14285 legitimate_small_data_address_p.
14286 (output_sdata): New static variable.
14287 (arc_print_operand): Update how we handle small data operands.
14288 (arc_print_operand_address): Likewise.
14289 (arc_legitimate_address_p): Update, use
14290 legitimate_small_data_address_p.
14291 (arc_rewrite_small_data_p): Remove.
14292 (arc_rewrite_small_data_1): Likewise.
14293 (arc_rewrite_small_data): Likewise.
14294 (small_data_pattern): Likewise.
14295 (compact_sda_memory_operand): Update to use
14296 legitimate_small_data_address_p and get_symbol_alignment.
14297 (prepare_move_operands): Don't rewite sdata pattern.
14298 (prepare_extend_operands): Remove.
14299 * config/arc/arc.md (zero_extendqihi2): Don't rewrite sdata
14301 (zero_extendqisi2): Likewise.
14302 (zero_extendhisi2): Likewise.
14303 (extendqihi2): Likewise.
14304 (extendqisi2): Likewise.
14305 (extendhisi2): Likewise.
14306 (addsi3): Likewise.
14307 (subsi3): Likewise.
14308 (andsi3): Likewise.
14309 * config/arc/constraints.md (Usd): Change it to memory constraint.
14311 2018-04-30 Claudiu Zissulescu <claziss@synopsys.com>
14313 * config/arc/arc.c (arc_split_move): Allow signed 6-bit constants
14314 as source of std instructions.
14315 * config/arc/arc.md (movsi_insn): Update pattern predicate to
14316 allow 6-bit constants as source for store instructions.
14317 (movdi_insn): Update instruction pattern to allow 6-bit constants
14318 as source for store instructions.
14320 2018-04-30 Jonathan Wakely <jwakely@redhat.com>
14322 * doc/invoke.texi (-fdebug-types-section): Fix grammar.
14324 2018-04-30 Nathan Sidwell <nathan@acm.org>
14325 Sandra Loosemore <sandra@codesourcery.com>
14327 * dumpfile.c (dump_open): Allow '-' for stdout.
14328 * doc/invoke.texi (Developer Options): Document dump filename
14329 determination early. Document stdin/stdout selection.
14331 2018-04-30 Andrew Sadek <andrew.sadek.se@gmail.com>
14333 Microblaze Target: PIC data text relative
14335 * config/microblaze/microblaze.opt: add new option -mpic-data-text-rel.
14336 * config/microblaze/microblaze-protos.h (microblaze_constant_address_p):
14338 * config/microblaze/microblaze.h (microblaze_constant_address_p):
14339 CONSTANT_ADDRESS_P definition to microblaze_constant_address_p.
14340 * config/microblaze/microblaze.c (TARGET_PIC_DATA_TEXT_REL):
14341 New addressing mode for data-text relative position indepenedent code.
14342 (microblaze_classify_unspec): add 'UNSPEC_TEXT' case ->
14343 'ADDRESS_SYMBOLIC_TXT_REL'.
14344 (microblaze_classify_address): Add handling for UNSPEC + CONST_INT.
14345 (microblaze_legitimate_pic_operand): Exclude function calls from
14346 pic operands in case of TARGET_PIC_DATA_TEXT_REL option.
14347 (microblaze_legitimize_address): Generate 'UNSPEC_TEXT' for all possible
14349 (microblaze_address_insns): Add 'ADDRESS_SYMBOLIC_TXT_REL' case.
14350 (print_operand): Add 'ADDRESS_SYMBOLIC_TXT_REL' case.
14351 (print_operand_address): Add 'ADDRESS_SYMBOLIC_TXT_REL' case + handling
14352 for 'address + offset'.
14353 (microblaze_expand_prologue): Add new function prologue call for
14355 (microblaze_asm_generate_pic_addr_dif_vec): Override new target hook
14356 'TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC' to disable address diff vector
14357 table in case of TARGET_PIC_DATA_TEXT_REL.
14358 (expand_pic_symbol_ref): Add handling for 'UNSPEC_TEXT'.
14359 * config/microblaze/microblaze.md (TARGET_PIC_DATA_TEXT_REL):
14360 Add new macros 'UNSPEC_TEXT',
14361 'UNSPEC_SET_TEXT' + add rule for setting r20 in function prologue
14362 + exclude function calls from 'UNSPEC_PLT' in case of data text
14364 * doc/tm.texi.in (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC): Add
14365 new target hook for generating address diff vector tables in case of
14367 * doc/tm.texi : Regenerate.
14368 * stmt.c (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC): Append new condition
14369 'targetm.asm_out.generate_pic_addr_diff_vec' to flag_pic in case
14370 of addr diff vector generation.
14371 * target.def (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC): Add
14372 target hook definition.
14373 * targhooks.h, targhooks.c (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC):
14374 Add default function for generate_pic_addr_diff_vec -> flag_pic.
14375 * doc/invoke.texi (Add new pic option): Add new microblaze pic
14376 option for data text relative.
14378 2018-04-30 Richard Biener <rguenther@suse.de>
14380 * tree-chrec.h (evolution_function_is_constant_p): Remove
14382 * tree-cfg.c (tree_node_can_be_shared): Re-order checks.
14384 2018-04-30 Richard Biener <rguenther@suse.de>
14387 * dwarf2out.c (gen_producer_string): Ignore -fchecking[=].
14389 2018-04-30 Richard Biener <rguenther@suse.de>
14391 PR tree-optimization/28364
14392 PR tree-optimization/85275
14393 * tree-ssa-loop-ch.c (ch_base::copy_headers): Stop after
14394 copying first exit test.
14396 2018-04-28 Mark Wielaard <mark@klomp.org>
14398 * dwarf2out.c (dwarf2out_finish): Add .debug_addr table header for
14399 dwarf_version >= 5.
14400 (dwarf_AT): Handle DW_AT_addr_base.
14401 (add_top_level_skeleton_die_attrs): Use dwarf_AT for DW_AT_addr_base.
14403 2018-04-28 Uros Bizjak <ubizjak@gmail.com>
14406 * config/i386/i386.md (*ashl<dwi>3_doubleword_mask): New pattern.
14407 (*ashl<dwi>3_doubleword_mask_1): Ditto.
14408 (*<shift_insn><dwi>3_doubleword_mask): Ditto.
14409 (*<shift_insn><dwi>3_doubleword_mask_1): Ditto.
14411 2018-04-28 Richard Biener <rguenther@suse.de>
14413 * tree-cfg.c (verify_gimple_phi): Take a gphi * argument.
14414 (verify_gimple_in_cfg): Rename visited_stmts to visited_throwing_stmts
14415 to reflect use. Only add interesting stmts.
14417 2018-04-27 Martin Jambor <mjambor@suse.cz>
14420 * ipa-cp.c (find_aggregate_values_for_callers_subset): Make sure
14421 the jump function allows for passing through aggregate values.
14423 2018-04-27 David Malcolm <dmalcolm@redhat.com>
14425 * input.h (in_system_header_at): Convert from macro to inline
14427 (from_macro_expansion_at): Likewise.
14428 (from_macro_definition_at): Likewise.
14430 2018-04-27 Jeff Law <law@redhat.com>
14432 * config.gcc: Mark tile* targets as deprecated/obsolete.
14434 2018-04-27 Richard Biener <rguenther@suse.de>
14436 * config/aarch64/aarch64.c: Simplify ap.__stack advance and
14439 2018-04-27 Richard Biener <rguenther@suse.de>
14441 * tree-cfg.c (verify_expr): Make dead code hit gcc_unreachable.
14443 2018-04-27 Uros Bizjak <ubizjak@gmail.com>
14445 * config/i386/i386.md (*movti_internal): Substitute Ye constraint
14446 with Yd constraint. Set "preferred_for_speed" attribute from
14447 TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC for alternatives
14448 with Yd constraint.
14449 (*movdi_internal): Ditto.
14450 (movti_interunit splitters): Remove
14451 TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC from insn condition.
14452 (movdi_interunit splitters): Ditto.
14453 * config/i386/constraints.md (Ye): Remove.
14454 (Yd): Do not depend on TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC.
14456 2018-04-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14459 * config/aarch64/constraints.md (Usg): Limit to 31.
14460 (Usj): Limit to 63.
14462 2018-04-27 Jakub Jelinek <jakub@redhat.com>
14464 PR tree-optimization/85529
14465 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Add FIRST_BB
14466 argument. Don't call get_nonzero_bits if opcode is ERROR_MARK_NODE,
14467 rhs2 def stmt's bb is dominated by first_bb and it isn't an obvious
14468 zero extension or masking of the MSB bit.
14469 (optimize_range_tests): Add FIRST_BB argument, pass it through
14470 to optimize_range_tests_var_bound.
14471 (maybe_optimize_range_tests, reassociate_bb): Adjust
14472 optimize_range_tests callers.
14474 2018-04-26 Richard Biener <rguenther@suse.de>
14475 Jakub Jelinek <jakub@redhat.com>
14477 * cgraph.h (symbol_table): Just declare debug method here.
14478 * symtab.c (symbol_table::debug): Define.
14480 2018-04-26 Eric Botcazou <ebotcazou@adacore.com>
14482 * loop-invariant.c (may_assign_reg_p): Return false for frame pointer.
14484 2018-04-26 Uros Bizjak <ubizjak@gmail.com>
14486 * config/i386/i386.md ("isa" attribute): Add x64_sse2.
14487 ("enabled" attribute): Handle x64_sse2 "isa" attribute.
14488 (*movdi_internal): Substitute Yi and Yj constraint with x
14489 and Ym and Yn constraint with y constraint. Update "isa"
14490 attribute and set "preferred_for_speed" attribute from
14491 TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC for updated alternatives.
14492 (*movsi_internal): Ditto.
14493 (*movdf_internal): Ditto.
14494 (*movsf_internal): Ditto.
14495 (*zero_extendsidi2): Ditto.
14496 * config/i386/sse.md (vec_set<mode>_0): Ditto.
14497 (sse2_loadld): Ditto.
14498 (*vec_extract<ssevecmodelower>_0): Ditto.
14499 (*vec_extractv4si_0_zext_sse4): Ditto.
14500 (vec_concatv2di): Ditto.
14501 (*vec_dup<mode>): Ditto.
14502 * config/i386/mmx.md (*mov<mode>_internal): Ditto.
14503 * config/i386/constraints.md (Yi): Remove.
14508 2018-04-26 Nathan Sidwell <nathan@acm.org>
14510 * dumpfile.c (dump_open): New.
14511 (dump_open_alternate_stream, dump_start, dump_begin): Call it.
14512 (dump_finish): Detect stdio/stderr by value not name.
14514 2018-04-26 Jonathan Wakely <jwakely@redhat.com>
14516 * doc/invoke.texi (-Wreturn-type): Document default status for C++.
14518 2018-04-26 Tom de Vries <tom@codesourcery.com>
14521 * config/nvptx/nvptx.c (verify_neutering_jumps)
14522 (verify_neutering_labels): New function
14523 (nvptx_single): Use verify_neutering_jumps and verify_neutering_labels.
14525 2018-04-26 Tom de Vries <tom@codesourcery.com>
14528 * config/nvptx/nvptx.c (needs_neutering_p): New function.
14529 (nvptx_single): Use needs_neutering_p to skip over insns that do not
14532 2018-04-26 Richard Biener <rguenther@suse.de>
14533 Tom de Vries <tom@codesourcery.com>
14536 * lto-streamer-out.c (output_function): Fixup loops if required to match
14537 discovery done in the reader.
14539 2018-04-26 Richard Biener <rguenther@suse.de>
14541 PR tree-optimization/85116
14542 * tree-ssa-loop-ch.c (do_while_loop_p): A do-while loop should
14543 have a loop exit from the single latch predecessor. Remove
14544 case of header with just condition.
14545 (ch_base::copy_headers): Exclude infinite loops from any
14547 (pass_ch::execute): Record exits.
14549 2018-04-26 Richard Biener <rguenther@suse.de>
14551 * tree-vect-data-refs.c (vect_get_data_access_cost): Get
14552 prologue cost vector and pass it to vect_get_load_cost.
14553 (vect_get_peeling_costs_all_drs): Likewise.
14554 (vect_peeling_hash_get_lowest_cost): Likewise.
14555 (vect_enhance_data_refs_alignment): Likewise.
14557 2018-04-26 Richard Biener <rguenther@suse.de>
14559 PR middle-end/85450
14560 * tree-cfg.c (verify_gimple_assign_unary): Restore proper
14561 checking of integer<->pointer conversions.
14562 * omp-expand.c (expand_omp_for_static_nochunk): Avoid
14563 sign-/zero-extending pointer types.
14564 (expand_omp_for_static_chunk): Likewise.
14566 2018-03-22 Hans-Peter Nilsson <hp@axis.com>
14567 Jean Lee <xiaoyur347@gmail.com>
14569 * config/mips/mips.c (mips_asan_shadow_offset): New function.
14570 (TARGET_ASAN_SHADOW_OFFSET): Define.
14571 * config/mips/mips.h (FRAME_GROWS_DOWNWARD): Augment to also be
14572 true for -fsanitize=address.
14574 2018-04-25 Mark Wielaard <mark@klomp.org>
14576 * dwarf2out.c (file_info_cmp): Sort longer dir prefixes before
14579 2018-04-25 Jakub Jelinek <jakub@redhat.com>
14581 * config/i386/i386.md (*x86_mov<mode>cc_0_m1): Use type "alu1" rather
14582 than "alu", remove explicit "memory" and "imm_disp" attributes.
14583 (*x86_mov<mode>cc_0_m1_se, *x86_mov<mode>cc_0_m1_neg): Likewise.
14585 PR middle-end/85414
14586 * simplify-rtx.c (simplify_unary_operation_1) <case SIGN_EXTEND,
14587 case ZERO_EXTEND>: Pass SUBREG_REG (op) rather than op to
14588 gen_lowpart_no_emit.
14590 2018-04-25 Sebastian Peryt <sebastian.peryt@intel.com>
14593 * config/i386/i386.c (ix86_expand_builtin): Change memory
14594 operand to XI, extend p0 to Pmode.
14595 * config/i386/i386.md: Change unspec volatile and operand
14596 1 mode to XI, change operand 0 mode to P.
14598 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
14600 * config/nds32/nds32-predicates.c (nds32_can_use_bclr_p): Mask with
14601 GET_MODE_MASK before any checking.
14602 (nds32_can_use_bset_p): Likewise.
14603 (nds32_can_use_btgl_p): Likewise.
14605 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
14607 * config/nds32/nds32-doubleword.md: New define_split pattern for
14608 illegal register number.
14610 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
14612 * config/nds32/nds32.c (nds32_print_operand): Set op_value ealier.
14614 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
14616 * config/nds32/nds32.h (ASM_APP_ON): Add missing newline character.
14618 2018-04-25 Richard Biener <rguenther@suse.de>
14620 * lto-streamer.h (LTO_major_version): Bump to 8.
14622 2018-04-25 Jakub Jelinek <jakub@redhat.com>
14624 * BASE-VER: Set to 9.0.0.
14626 2018-04-24 Segher Boessenkool <segher@kernel.crashing.org>
14628 * config/rs6000/rs6000.c (init_float128_ieee): Fix spelling mistakes
14629 in __abskf2 and __powikf2.
14631 2018-04-24 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14634 * config/aarch64/constraints.md (Usg, Usj): New constraints.
14635 * config/aarch64/iterators.md (cmode_simd): New mode attribute.
14636 * config/aarch64/aarch64.md (*aarch64_ashr_sisd_or_int_<mode>3):
14637 Use the above on operand 2. Reindent.
14638 (*aarch64_lshr_sisd_or_int_<mode>3): Likewise.
14640 2018-04-24 H.J. Lu <hongjiu.lu@intel.com>
14643 * common/config/i386/i386-common.c (ix86_handle_option): Don't
14645 * config/i386/i386.opt (mcet): Removed.
14646 * doc/install.texi: Remove -mcet documentation.
14647 * doc/invoke.texi: Likewise.
14649 2018-04-24 H.J. Lu <hongjiu.lu@intel.com>
14652 * doc/install.texi: Remove -mcet from bootstrap-cet.
14654 2018-04-24 Jakub Jelinek <jakub@redhat.com>
14657 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Don't define
14658 __builtin_ia32_readeflags_u32 and __builtin_ia32_writeeflags_u32
14662 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Punt if
14663 const_vector is not CONST_VECTOR or SYMBOL_REF for a constant pool
14664 containing a CONST_VECTOR.
14666 2018-04-24 Cesar Philippidis <cesar@codesourcery.com>
14668 * doc/install.texi: Update newlib dependency for nvptx.
14670 2018-04-24 Jakub Jelinek <jakub@redhat.com>
14673 * config/i386/i386.c (ix86_expand_vector_init_one_var): Use UINTVAL
14674 instead of INTVAL when shifting x left.
14676 2018-04-24 Andreas Krebbel <krebbel@linux.ibm.com>
14678 PR tree-optimization/85478
14679 * tree-vect-loop.c (vect_analyze_loop_2): Do not call
14680 vect_grouped_store_supported for single element vectors.
14682 2018-04-24 Richard Biener <rguenther@suse.de>
14685 * config/i386/i386.c (ix86_add_stmt_cost): Restrict strided
14686 load cost increase to the case of non-constant step.
14688 2018-04-24 Jakub Jelinek <jakub@redhat.com>
14691 * reg-stack.c (move_for_stack_reg): Don't fail assertion about dead
14692 destination if any_malformed_asm.
14694 2018-04-23 Eric Botcazou <ebotcazou@adacore.com>
14696 PR middle-end/85496
14697 * expr.c (store_field): In the bitfield case, if the value comes from
14698 a function call and is returned in registers by means of a PARALLEL,
14699 do not change the mode of the temporary unless BLKmode and VOIDmode.
14701 2018-04-23 Andrey Belevantsev <abel@ispras.ru>
14703 PR rtl-optimization/85423
14704 * sel-sched-ir.c (has_dependence_note_mem_dep): Only discard
14705 dependencies to debug insns when the previous insn is non-debug.
14707 2018-04-23 Claudiu Zissulescu <claziss@synopsys.com>
14709 * config/arc/arc.md ("vunspec"): Delete it, unify all the unspec
14710 enums into a single definition.
14711 (fls): Fix predicates and printing.
14714 2018-04-23 Claudiu Zissulescu <claziss@synopsys.com>
14716 * config/arc/arc-protos.h (check_if_valid_sleep_operand): Remove.
14717 * config/arc/arc.c (arc_expand_builtin): Sleep accepts registers
14718 and short u6 immediate.
14719 (check_if_valid_sleep_operand): Remove.
14720 * config/arc/arc.md (Sleep): Accepts registers and u6 immediates.
14722 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
14724 * config/nds32/nds32.c (nds32_compute_stack_frame): Consider
14725 flag_always_save_lp condition.
14726 * config/nds32/nds32.opt (malways-save-lp): New option.
14728 2018-04-22 Shiva Chen <shiva0217@gmail.com>
14730 * config/nds32/nds32-protos.h (nds32_use_load_post_increment): Declare.
14731 * config/nds32/nds32.c (nds32_use_load_post_increment): New.
14732 * config/nds32/nds32.h
14733 (USE_LOAD_POST_INCREMENT, USE_LOAD_POST_DECREMENT): Define.
14734 (USE_STORE_POST_INCREMENT, USE_STORE_POST_DECREMENT): Define.
14736 2018-04-22 Shiva Chen <shiva0217@gmail.com>
14738 * config/nds32/nds32-protos.h (nds32_ls_333_p): Remove.
14739 * config/nds32/nds32.c (nds32_ls_333_p): Remove.
14741 2018-04-22 Shiva Chen <shiva0217@gmail.com>
14742 Chung-Ju Wu <jasonwucj@gmail.com>
14744 * config/nds32/nds32-protos.h (nds32_case_vector_shorten_mode):
14746 * config/nds32/nds32.c (nds32_case_vector_shorten_mode): New function.
14747 * config/nds32/nds32.h (CASE_VECTOR_SHORTEN_MODE): Modify.
14749 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
14751 * config/nds32/nds32.c (nds32_compute_stack_frame): Fix wrong value.
14753 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
14755 * config/nds32/nds32-protos.h (nds32_data_alignment,
14756 nds32_local_alignment): Declare.
14757 * config/nds32/nds32.c (nds32_data_alignment, nds32_constant_alignment,
14758 nds32_local_alignment): New functions.
14759 (TARGET_CONSTANT_ALIGNMENT): Define.
14760 * config/nds32/nds32.h (DATA_ALIGNMENT, LOCAL_ALIGNMENT): Define.
14762 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
14764 * config/nds32/nds32.c
14765 (TARGET_HARD_REGNO_MODE_OK): Move to the bottom of file.
14766 (TARGET_MODES_TIEABLE_P): Likewise.
14768 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
14770 * config/nds32/nds32.c (nds32_asm_file_start): Display optimization
14771 level Ofast and Og.
14773 2018-04-22 Monk Chiang <sh.chiang04@gmail.com>
14774 Chung-Ju Wu <jasonwucj@gmail.com>
14776 * config/nds32/constants.md (unspec_volatile_element): Add enum values
14777 for unaligned access.
14778 * config/nds32/nds32-intrinsic.c: Implementation of expanding
14780 * config/nds32/nds32-intrinsic.md: Likewise.
14781 * config/nds32/nds32_intrinsic.h: Likewise.
14782 * config/nds32/nds32.h (nds32_builtins): Likewise.
14783 * config/nds32/nds32.opt (munaligned-access): New option.
14784 * config/nds32/nds32.c (nds32_asm_file_start): Display
14785 flag_unaligned_access status.
14787 2018-04-20 Kito Cheng <kito.cheng@gmail.com>
14789 * config/riscv/elf.h (LINK_SPEC): Pass --no-relax if
14790 -mno-relax is present.
14791 * config/riscv/linux.h (LINK_SPEC): Ditto.
14793 2018-04-20 Martin Sebor <msebor@redhat.com>
14796 * gimple-fold.c (gimple_fold_builtin_strcpy): Suppress -Wrestrict
14798 (gimple_fold_builtin_stxcpy_chk): Same.
14799 * gimple-ssa-warn-restrict.c (check_bounds_or_overlap): Same.
14801 2018-04-20 Michael Meissner <meissner@linux.ibm.com>
14804 * config/rs6000/rs6000.c (init_float128_ieee): Add support to call
14805 __powikf2 when long double is IEEE 128-bit.
14807 2018-04-20 Kito Cheng <kito.cheng@gmail.com>
14809 * config/riscv/riscv.c (riscv_first_stack_step): Round up min
14810 step to make sure stack always aligned.
14812 2018-04-20 Carl Love <cel@us.ibm.com>
14815 * config/rs6000/rs6000-c.c (rs6000_gimple_fold_builtin): Add
14816 size check for arg0.
14818 2018-04-20 Nathan Sidwell <nathan@codesourcery.com>
14819 Tom de Vries <tom@codesourcery.com>
14822 * config/nvptx/nvptx.c (nvptx_emit_forking, nvptx_emit_joining):
14823 Emit insns for calls too.
14824 (nvptx_find_par): Always look for worker-level predecessor insn.
14825 (nvptx_propagate): Add is_call parm, return bool. Copy frame for
14827 (nvptx_vpropagate, nvptx_wpropagate): Adjust.
14828 (nvptx_process_pars): Propagate frames for calls.
14830 2018-04-20 H.J. Lu <hongjiu.lu@intel.com>
14833 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET):
14835 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
14836 (ix86_handle_option): Don't handle OPT_mibt.
14837 * config/i386/cet.h: Check __CET__ instead of __IBT__ and
14839 * config/i386/driver-i386.c (host_detect_local_cpu): Remove
14841 * config/i386/i386-c.c (ix86_target_macros_internal): Don't
14842 check OPTION_MASK_ISA_IBT nor flag_cf_protection.
14843 (ix86_target_macros): Define __CET__ with flag_cf_protection
14844 for -fcf-protection.
14845 * config/i386/i386.c (isa2_opts): Remove -mibt.
14846 * config/i386/i386.h (TARGET_IBT): Removed.
14847 (TARGET_IBT_P): Likewise.
14848 (ix86_valid_target_attribute_inner_p): Don't check OPT_mibt.
14849 * config/i386/i386.md (nop_endbr): Don't check TARGET_IBT.
14850 * config/i386/i386.opt (mcet): Update help message.
14851 (mshstk): Likewise.
14853 * doc/invoke.texi: Remove -mibt. Document __CET__. Document
14854 -mcet as an alias for -mshstk.
14856 2018-04-20 Richard Biener <rguenther@suse.de>
14858 PR middle-end/85475
14859 * match.pd ((X * CST) * Y -> (X * Y) * CST): Avoid exponential
14860 complexity by forcing a single use of the multiply operand.
14862 2018-04-20 Martin Jambor <mjambor@suse.cz>
14865 * ipa-cp.c (cgraph_edge_brings_value_p): Move check for self-feeding
14866 recursion dependency to only apply to non-clones.
14868 2018-04-20 Martin Jambor <mjambor@suse.cz>
14871 * ipa-cp.c (create_specialized_node): Check that clones of
14872 self-recursive edges exist during IPA-CP.
14874 2018-04-19 Toon Moene <toon@moene.org>
14876 * doc/invoke.texi: Add -floop-unroll-and-jam to options enabled
14879 2018-04-19 Jakub Jelinek <jakub@redhat.com>
14881 PR tree-optimization/85467
14882 * fold-const.c (fold_ternary_loc) <case BIT_FIELD_REF>: Use
14883 VECTOR_TYPE_P macro. If type is vector type, VIEW_CONVERT_EXPR the
14884 VECTOR_CST element to type.
14886 2018-04-19 H.J. Lu <hongjiu.lu@intel.com>
14889 * config/i386/i386.h (STACK_SAVEAREA_MODE): New.
14890 * config/i386/i386.md (builtin_setjmp_setup): Removed.
14891 (builtin_longjmp): Likewise.
14892 (save_stack_nonlocal): New pattern.
14893 (restore_stack_nonlocal): Likewise.
14895 2018-04-19 H.J. Lu <hongjiu.lu@intel.com>
14898 * config/i386/cet.c (file_end_indicate_exec_stack_and_cet):
14899 Replace ASM_OUTPUT_LABEL with fprintf.
14901 2018-04-19 H.J. Lu <hongjiu.lu@intel.com>
14904 * config/i386/cet.c (file_end_indicate_exec_stack_and_cet):
14905 Check flag_cf_protection instead of TARGET_IBT and TARGET_SHSTK.
14906 * config/i386/i386-c.c (ix86_target_macros_internal): Also
14907 define __IBT__ and __SHSTK__ for -fcf-protection.
14908 * config/i386/i386.c (pass_insert_endbranch::gate): Don't check
14910 (ix86_trampoline_init): Likewise.
14911 (x86_output_mi_thunk): Likewise.
14912 (ix86_notrack_prefixed_insn_p): Likewise.
14913 (ix86_option_override_internal): Don't disallow -fcf-protection.
14914 * config/i386/i386.md (rdssp<mode>): Also enable for
14916 (incssp<mode>): Likewise.
14917 (nop_endbr): Likewise.
14918 * config/i386/i386.opt (mcet): Change help message to built-in
14921 (mshstk): Likewise.
14922 * doc/invoke.texi: Remove -mcet, -mibt and -mshstk condition
14923 on -fcf-protection. Change -mcet, -mibt and -mshstk to only
14924 enable CET built-in functions.
14926 2018-04-19 Sebastian Peryt <sebastian.peryt@intel.com>
14928 * common/config/i386/i386-common.c
14929 (OPTION_MASK_ISA_MOVDIRI_SET, OPTION_MASK_ISA_MOVDIR64B_SET,
14930 OPTION_MASK_ISA_MOVDIRI_UNSET,
14931 OPTION_MASK_ISA_MOVDIR64B_UNSET): New defines.
14932 (ix86_handle_option): Handle -mmovdiri and -mmovdir64b.
14933 * config.gcc (movdirintrin.h): New header.
14934 * config/i386/cpuid.h (bit_MOVDIRI,
14935 bit_MOVDIR64B): New bits.
14936 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mmovdiri
14938 * config/i386/i386-builtin-types.def ((VOID, PUNSIGNED, UNSIGNED),
14939 (VOID, PVOID, PCVOID)): New function types.
14940 * config/i386/i386-builtin.def (__builtin_ia32_directstoreu_u32,
14941 __builtin_ia32_directstoreu_u64,
14942 __builtin_ia32_movdir64b): New builtins.
14943 * config/i386/i386-c.c (__MOVDIRI__, __MOVDIR64B__): New.
14944 * config/i386/i386.c (ix86_target_string): Added -mmovdir64b
14946 (ix86_valid_target_attribute_inner_p): Ditto.
14947 (ix86_expand_special_args_builtin): Added VOID_FTYPE_PUNSIGNED_UNSIGNED
14948 and VOID_FTYPE_PUNSIGNED_UNSIGNED.
14949 (ix86_expand_builtin): Expand IX86_BUILTIN_MOVDIR64B.
14950 * config/i386/i386.h (TARGET_MOVDIRI, TARGET_MOVDIRI_P,
14951 TARGET_MOVDIR64B, TARGET_MOVDIR64B_P): New.
14952 * config/i386/i386.md (UNSPECV_MOVDIRI, UNSPECV_MOVDIR64B): New.
14953 (movdiri<mode>, movdir64b_<mode>): New.
14954 * config/i386/i386.opt: Add -mmovdiri and -mmovdir64b.
14955 * config/i386/immintrin.h: Include movdirintrin.h.
14956 * config/i386/movdirintrin.h: New file.
14957 * doc/invoke.texi: Added -mmovdiri and -mmovdir64b.
14959 2018-04-19 Richard Biener <rguenther@suse.de>
14961 PR middle-end/85455
14962 * cfg.c (clear_bb_flags): When loop state says we have
14963 marked irreducible regions also preserve BB_IRREDUCIBLE_LOOP.
14965 2018-04-19 Richard Biener <rguenther@suse.de>
14967 PR tree-optimization/84737
14968 * tree-vect-data-refs.c (vect_copy_ref_info): New function
14969 copying restrict info.
14970 (vect_setup_realignment): Use it.
14971 * tree-vectorizer.h (vect_copy_ref_info): Declare.
14972 * tree-vect-stmts.c (vectorizable_store): Copy ref info from
14973 the first DR to all generated stores.
14974 (vectorizable_load): Likewise for loads.
14976 2018-04-19 Jakub Jelinek <jakub@redhat.com>
14978 PR tree-optimization/85446
14979 * match.pd ((intptr_t) x eq/ne CST to x eq/ne (typeof x) cst): Require
14980 the integral and pointer types to have the same precision.
14982 * doc/install.texi: Document --disable-cet being the default and
14985 2018-04-18 Martin Liska <mliska@suse.cz>
14987 * ipa-devirt.c (odr_subtypes_equivalent_p): Fix GNU coding
14990 2018-04-18 Martin Liska <mliska@suse.cz>
14993 2018-03-02 Eric Botcazou <ebotcazou@adacore.com>
14996 * ipa-devirt.c (odr_subtypes_equivalent_p): Get the ODR type of both
14997 arguments if they are comparable.
14999 2018-04-18 Martin Liska <mliska@suse.cz>
15002 2018-03-13 Eric Botcazou <ebotcazou@adacore.com>
15005 * ipa-devirt.c (odr_subtypes_equivalent_p): Do not get the ODR type of
15008 2018-04-18 H.J. Lu <hongjiu.lu@intel.com>
15011 * config/i386/i386.c (ix86_expand_split_stack_prologue): Insert
15012 ENDBR after calling __morestack.
15014 2018-04-18 David Malcolm <dmalcolm@redhat.com>
15017 * configure.ac (gcc-driver-name.h): Honor --with-gcc-major-version
15018 by using gcc_base_ver to generate a gcc_driver_version, and use
15019 it when generating GCC_DRIVER_NAME.
15020 * configure: Regenerate.
15022 2018-04-18 Jakub Jelinek <jakub@redhat.com>
15025 * config.gcc: Obsolete powerpc*-*-*spe*.
15027 2018-04-17 Jakub Jelinek <jakub@redhat.com>
15030 * dbxout.c (dbxout_int): Perform negation in unsigned int type.
15031 (stabstr_D): Change type of unum from unsigned int to
15032 unsigned HOST_WIDE_INT. Perform negation in unsigned HOST_WIDE_INT
15035 2018-04-17 Jim Wilson <jimw@sifive.com>
15038 * config/riscv/riscv.c (riscv_compute_frame_info): Add calls to
15039 RISCV_STACK_ALIGN when using outgoing_args_size and pretend_args_size.
15040 Set arg_pointer_offset after using pretend_args_size.
15042 2018-04-17 Jakub Jelinek <jakub@redhat.com>
15044 PR rtl-optimization/85431
15045 * dse.c (record_store): Ignore zero width stores.
15048 * asan.c (handle_builtin_stack_restore): Adjust comment. Emit
15049 __asan_allocas_unpoison call and last_alloca_addr = new_sp before
15050 __builtin_stack_restore rather than after it.
15051 * builtins.c (expand_asan_emit_allocas_unpoison): Pass
15052 arg1 + (virtual_dynamic_stack_rtx - stack_pointer_rtx) as second
15053 argument instead of virtual_dynamic_stack_rtx.
15055 2018-04-17 Kelvin Nilsen <kelvin@gcc.gnu.org>
15057 * config/rs6000/rs6000-protos.h (rs6000_builtin_is_supported_p):
15059 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
15060 Add note to error message to explain internal mapping of overloaded
15061 built-in function name to non-overloaded built-in function name.
15062 * config/rs6000/rs6000.c (rs6000_builtin_is_supported_p): New
15065 2018-04-17 Michael Meissner <meissner@linux.vnet.ibm.com>
15068 * config/rs6000/rs6000.md (pack<mode>): Do not try handle a pack
15069 where the inputs overlap with the output.
15071 2018-04-17 Jakub Jelinek <jakub@redhat.com>
15073 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): Add
15074 (=v, v) alternative and explicit "memory" attribute.
15075 (vec_extract_lo_<mode><mask_name>): Likewise. Also add
15076 "type", "prefix", "prefix_extra", "length_immediate" and "mode"
15078 (vec_extract_lo_<mode><mask_name>): Add (=v, v) alternative and use
15079 "sselog1" type instead of "sselog".
15080 (vec_extract_hi_<mode><mask_name>): Use "sselog1" type instead of
15081 "sselog". Remove explicit "memory" attribute.
15082 (vec_extract_lo_v32hi): Add (=v, v) alternative and explicit "memory",
15083 "type", "prefix", "prefix_extra", "length_immediate" and "mode"
15085 (vec_extract_hi_v32hi): Merge all alternatives into one, use
15086 "sselog1" type instead of "sselog". Remove explicit "memory"
15088 (vec_extract_hi_v16hi): Merge each pair of alternatives into one,
15089 use "sselog1" type instead of "sselog". Remove explicit "memory"
15091 (vec_extract_lo_v64qi): Add (=v, v) alternative and explicit "memory",
15092 "type", "prefix", "prefix_extra", "length_immediate" and "mode"
15094 (vec_extract_hi_v64qi): Merge all alternatives into one, use
15095 "sselog1" type instead of "sselog". Remove explicit "memory"
15097 (vec_extract_hi_v32qi): Merge each pair of alternatives into one,
15098 use "sselog1" type instead of "sselog". Remove explicit "memory"
15102 * config/i386/i386.md (*ashlqi3_1_slp): Use alu1 type instead of alu.
15104 PR middle-end/85414
15105 * rtlhooks.c (gen_lowpart_if_possible): Don't call gen_lowpart_SUBREG
15108 2018-04-17 Martin Jambor <mjambor@suse.cz>
15111 * ipa-cp.c (create_specialized_node): Call
15112 expand_all_artificial_thunks if necessary.
15114 2018-04-17 Martin Liska <mliska@suse.cz>
15117 * ipa-devirt.c (odr_types_equivalent_p): Remove trailing
15118 in message, remote space in between '_G' and '('.
15120 2018-04-17 Jakub Jelinek <jakub@redhat.com>
15123 * config/i386/sse.md (reduces<mode><mask_scalar_name>,
15124 avx512f_vmcmp<mode>3<round_saeonly_name>,
15125 avx512f_vmcmp<mode>3_mask<round_saeonly_name>,
15126 avx512f_sgetexp<mode><mask_scalar_name><round_saeonly_scalar_name>,
15127 avx512f_rndscale<mode><round_saeonly_name>,
15128 avx512dq_ranges<mode><mask_scalar_name><round_saeonly_scalar_name>,
15129 avx512f_vgetmant<mode><mask_scalar_name><round_saeonly_scalar_name>):
15130 Use %<iptr>2 instead of %2 for -masm=intel.
15131 (avx512f_vcvtss2usi<round_name>, avx512f_vcvtss2usiq<round_name>,
15132 avx512f_vcvttss2usi<round_saeonly_name>,
15133 avx512f_vcvttss2usiq<round_saeonly_name>): Use %k1 instead of %1 for
15135 (avx512f_vcvtsd2usi<round_name>, avx512f_vcvtsd2usiq<round_name>,
15136 avx512f_vcvttsd2usi<round_saeonly_name>,
15137 avx512f_vcvttsd2usiq<round_saeonly_name>, ufloatv2siv2df2<mask_name>):
15138 Use %q1 instead of %1 for -masm=intel.
15139 (avx512f_sfixupimm<mode><sd_maskz_name><round_saeonly_name>,
15140 avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Use %<iptr>3 instead
15141 of %3 for -masm=intel.
15142 (sse2_shufpd_v2df_mask): Fix a typo, change %{6%} to %{%6%} for
15144 (*avx512vl_<code>v2div2qi2_store): Use %w0 instead of %0 for
15146 (*avx512vl_<code><mode>v4qi2_store): Use %k0 instead of %0 for
15148 (avx512vl_<code><mode>v4qi2_mask_store): Use a single pattern with
15149 %k0 and %1 for -masm=intel rather than two patterns, one with %0 and
15151 (*avx512vl_<code><mode>v8qi2_store): Use %q0 instead of %0 for
15153 (avx512vl_<code><mode>v8qi2_mask_store): Use a single pattern with
15154 %q0 and %1 for -masm=intel rather than two patterns, one with %0 and
15155 %g1 and one with %0 and %1.
15156 (avx512er_vmrcp28<mode><round_saeonly_name>,
15157 avx512er_vmrsqrt28<mode><round_saeonly_name>): Use %<iptr>1 instead of
15158 %1 for -masm=intel.
15159 (avx5124fmaddps_4fmaddps_mask, avx5124fmaddps_4fmaddss_mask,
15160 avx5124fmaddps_4fnmaddps_mask, avx5124fmaddps_4fnmaddss_mask,
15161 avx5124vnniw_vp4dpwssd_mask, avx5124vnniw_vp4dpwssds_mask): Swap order
15162 of %0 and %{%4%} for -masm=intel.
15163 (avx5124fmaddps_4fmaddps_maskz, avx5124fmaddps_4fmaddss_maskz,
15164 avx5124fmaddps_4fnmaddps_maskz, avx5124fmaddps_4fnmaddss_maskz,
15165 avx5124vnniw_vp4dpwssd_maskz, avx5124vnniw_vp4dpwssds_maskz): Swap
15166 order of %0 and %{%5%}%{z%} for -masm=intel.
15168 2018-04-17 Jan Hubicka <jh@suse.cz>
15171 * ipa-devirt.c (odr_types_equivalent_p): Handle bit fields.
15173 2018-04-17 Martin Liska <mliska@suse.cz>
15176 * multiple_target.c (create_dispatcher_calls): Set apostrophes
15177 for target_clone error message. Make default implementation
15178 clone to be a local declaration.
15179 (separate_attrs): Add new argument and check for an empty
15181 (expand_target_clones): Handle it.
15182 (ipa_target_clone): Make redirection just for target_clones
15185 2018-04-16 Cesar Philippidis <cesar@codesourcery.com>
15186 Tom de Vries <tom@codesourcery.com>
15188 PR middle-end/84955
15189 * omp-expand.c (expand_oacc_for): Add dummy false branch for
15190 tiled basic blocks without omp continue statements.
15192 2018-04-16 Aaron Sawdey <acsawdey@linux.ibm.com>
15195 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Mark
15196 vec_extract expression as having side effects to make sure it gets
15199 2018-04-16 H.J. Lu <hongjiu.lu@intel.com>
15202 * config/i386/i386.c (get_builtin_code_for_version): Check
15205 2018-04-16 Olga Makhotina <olga.makhotina@intel.com>
15208 * config.gcc: Support "skylake".
15209 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
15211 * config/i386/i386.c (m_SKYLAKE): Define.
15212 (processor_target_table): Add "skylake".
15213 (ix86_option_override_internal): Add "skylake".
15214 (get_builtin_code_for_version): Handle PROCESSOR_SKYLAKE,
15215 PROCESSOR_CANNONLAKE.
15216 (get_builtin_code_for_version): Fix priority for
15217 PROCESSOR_ICELAKE_CLIENT, PROCESSOR_ICELAKE_SERVER,
15218 PROCESSOR_SKYLAKE-AVX512.
15219 * config/i386/i386.h (processor_costs): Define TARGET_SKYLAKE.
15220 (processor_type): Add PROCESSOR_SKYLAKE.
15222 2018-04-16 Paolo Carlini <paolo.carlini@oracle.com>
15223 Jason Merrill <jason@redhat.com>
15226 * convert.c (convert_to_integer_1): Use direct recursion for
15227 enumeral types and types with a precision less than the number
15228 of bits in their mode.
15230 2018-04-16 Julia Koval <julia.koval@intel.com>
15233 * config/i386/x86-tune.def (X86_TUNE_SSE_UNALIGNED_LOAD_OPTIMAL,
15234 X86_TUNE_SSE_UNALIGNED_STORE_OPTIMAL): Add m_SKYLAKE_AVX512
15236 2018-04-14 Segher Boessenkool <segher@kernel.crashing.org>
15239 * config/rs6000/rs6000.opt (mdirect-move): Make deprecated.
15240 * doc/invoke.texi (RS/6000 and PowerPC Options): Remove -mdirect-move
15241 and -mno-direct-move.
15243 2018-04-13 Paul A. Clarke <pc@us.ibm.com>
15246 * config/rs6000/emmintrin.h (_mm_slli_epi{16,32,64}):
15247 Ensure that vec_splat_s32 is only called with 0 <= shift < 16.
15248 Ensure negative shifts result in {0}.
15250 2018-04-13 Vladimir Makarov <vmakarov@redhat.com>
15252 PR rtl-optimization/79916
15253 * config/rs6000/rs6000.c (rs6000_emit_move): Use assigned hard
15254 regs (if any) to define how to gnerate SD moves when LRA is in
15257 2018-04-13 Jakub Jelinek <jakub@redhat.com>
15259 PR rtl-optimization/85393
15260 * except.h (expand_dw2_landing_pad_for_region): Remove declaration.
15261 * except.c (expand_dw2_landing_pad_for_region): Make static.
15262 * bb-reorder.c (fix_up_crossing_landing_pad): In new_bb emit just
15263 a label and unconditional jump to old_bb, rather than
15264 expand_dw2_landing_pad_for_region insn(s) and jump to single_succ
15267 PR rtl-optimization/85376
15268 * simplify-rtx.c (simplify_const_unary_operation): For CLZ and CTZ and
15269 zero op0, if C?Z_DEFINED_VALUE_AT_ZERO is false, return NULL_RTX
15270 instead of a specific value.
15272 2018-04-13 Jan Hubicka <hubicka@ucw.cz>
15273 Bin Cheng <bin.cheng@arm.com>
15275 PR tree-optimization/82965
15276 PR tree-optimization/83991
15277 * cfgloopanal.c (expected_loop_iterations_unbounded): Add
15278 by_profile_only parameter.
15279 * cfgloopmanip.c (scale_loop_profile): Further scale loop's profile
15280 information if the loop was predicted to iterate too many times.
15281 * cfgloop.h (expected_loop_iterations_unbounded): Update prototype
15283 2018-04-13 Jan Hubicka <hubicka@ucw.cz>
15286 * config/i386/i386.c (ix86_can_inline_p): Allow safe transitions for
15289 2018-04-13 Martin Liska <mliska@suse.cz>
15290 Jakub Jelinek <jakub@redhat.com>
15292 PR middle-end/81657
15293 * expr.h (enum block_op_methods): Add BLOCK_OP_NO_LIBCALL_RET.
15294 * expr.c (emit_block_move_hints): Handle BLOCK_OP_NO_LIBCALL_RET.
15295 * builtins.c (expand_builtin_memory_copy_args): Use
15296 BLOCK_OP_NO_LIBCALL_RET method for mempcpy with non-ignored target,
15297 handle dest_addr == pc_rtx.
15299 2018-04-12 Segher Boessenkool <segher@kernel.crashing.org>
15302 * config/rs6000/rs6000.md (fix_trunc<mode>si2): Use legacy code if
15303 asked to not generate direct moves.
15304 (fix_trunc<mode>si2_stfiwx): Similar.
15305 (fix_trunc<mode>si2_internal): Similar.
15307 2018-04-12 Jakub Jelinek <jakub@redhat.com>
15310 * var-tracking.c (add_stores): Handle STRICT_LOW_PART SET_DEST.
15311 * cselib.c (cselib_record_sets): For STRICT_LOW_PART dest,
15312 lookup if dest in some wider mode is known to be const0_rtx and
15313 if so, record permanent equivalence for it to be ZERO_EXTEND of
15314 the narrower mode destination.
15316 2018-04-12 Cesar Philippidis <cesar@codesourcery.com>
15318 * lto-streamer-out.c (output_function): Revert 259346.
15319 * omp-expand.c (expand_oacc_for): Likewise.
15321 2018-04-12 Alexander Monakov <amonakov@ispras.ru>
15323 PR rtl-optimization/85354
15324 * sel-sched-ir.c (sel_init_pipelining): Move cfg_cleanup call...
15325 * sel-sched.c (sel_global_init): ... here.
15327 2018-04-12 Eric Botcazou <ebotcazou@adacore.com>
15330 * lto-wrapper.c (debug_objcopy): Open the files in binary mode.
15331 * dwarf2out.c (dwarf2out_early_finish): Do not generate assembly in LTO
15332 mode for PE-COFF targets.
15333 * config/i386/i386-protos.h (i386_pe_asm_lto_start): Declare.
15334 (i386_pe_asm_lto_end): Likewise.
15335 * config/i386/cygming.h (TARGET_ASM_LTO_START): Define.
15336 (TARGET_ASM_LTO_END): Likewise.
15337 * config/i386/winnt.c (saved_debug_info_level): New static variable.
15338 (i386_pe_asm_lto_start): New function.
15339 (i386_pe_asm_lto_end): Likewise.
15341 2018-04-12 Cesar Philippidis <cesar@codesourcery.com>
15342 Richard Biener <rguenther@suse.de>
15344 PR middle-end/84955
15345 * lto-streamer-out.c (output_function): Fix CFG loop state before
15347 * omp-expand.c (expand_oacc_for): Handle calls to internal
15348 functions like regular functions.
15350 2018-04-12 Richard Biener <rguenther@suse.de>
15353 * dwarf2out.c (init_sections_and_labels): Use debug_line_section[_label]
15354 for the early LTO debug to properly generate references to it
15355 during DIE emission. Do not re-use that for the skeleton for
15357 (dwarf2out_early_finish): Likewise.
15359 2018-04-12 Jakub Jelinek <jakub@redhat.com>
15362 * config/i386/sse.md
15363 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name> split,
15364 <mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name> split,
15365 vec_extract_lo_<mode><mask_name> split, vec_extract_lo_v32hi,
15366 vec_extract_lo_v64qi): For non-AVX512VL if input is xmm16+ reg
15367 and output is a reg, avoid creating invalid lowpart subreg, but
15368 instead split into a 512-bit move. Don't split if not AVX512VL,
15369 input is xmm16+ reg and output is a mem.
15370 (vec_extract_lo_<mode><mask_name>, vec_extract_lo_v32hi,
15371 vec_extract_lo_v64qi): Don't require split if not AVX512VL, input is
15372 xmm16+ reg and output is a mem.
15374 2018-04-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15376 * config/s390/s390.c (s390_output_indirect_thunk_function): Check
15377 also for flag_dwarf2_cfi_asm.
15379 2018-04-12 Jakub Jelinek <jakub@redhat.com>
15381 PR rtl-optimization/85342
15382 * regcprop.c (copyprop_hardreg_forward_1): Remove replaced array, use
15383 a bool scalar var inside of the loop instead. Don't try to update
15384 recog_data.operand after failed apply_change_group.
15386 2018-04-12 Tom de Vries <tom@codesourcery.com>
15389 * config/nvptx/nvptx.c (flexible_array_member_type_p): New function.
15390 (nvptx_assemble_decl_begin): Add undefined param. Declare undefined
15391 array with flexible array member as array without given dimension.
15392 (nvptx_assemble_undefined_decl): Set nvptx_assemble_decl_begin call
15393 argument for undefined param to true.
15395 2018-04-11 Aaron Sawdey <acsawdey@linux.ibm.com>
15398 * doc/invoke.texi (RS/6000 and PowerPC Options): Document options
15399 -mcall- and -mtraceback=. Remove options -mabi=spe and -mabi=no-spe
15400 from PowerPC section.
15401 * config/rs6000/sysv4.opt (mcall-): Improve help text.
15402 * config/rs6000/rs6000.opt (mblock-compare-inline-limit=): Trim
15403 help text that is too long.
15404 * config/rs6000/rs6000.opt (mblock-compare-inline-loop-limit=): Trim
15405 help text that is too long.
15406 * config/rs6000/rs6000.opt (mstring-compare-inline-limit=): Trim
15407 help text that is too long.
15409 2018-04-11 Uros Bizjak <ubizjak@gmail.com>
15411 * config/alpha/alpha.md (stack_probe_internal): Rename
15412 from "probe_stack". Update all callers.
15414 2018-04-11 Alexander Monakov <amonakov@ispras.ru>
15416 PR rtl-optimization/84566
15417 * sched-deps.c (sched_analyze_insn): Check deps->readonly when invoking
15418 sched_macro_fuse_insns.
15420 2018-04-11 Alexander Monakov <amonakov@ispras.ru>
15423 * sched-rgn.c (add_branch_dependences): Move sel_sched_p check here...
15424 (compute_block_dependences): ... from here.
15426 2018-04-11 Jakub Jelinek <jakub@redhat.com>
15428 PR tree-optimization/85331
15429 * vec-perm-indices.h (vec_perm_indices::clamp): Change input type
15430 from int to HOST_WIDE_INT.
15432 2018-04-11 Martin Jambor <mjambor@suse.cz>
15435 * ipa-cp.c (propagate_vals_across_pass_through): Expand comment.
15436 (cgraph_edge_brings_value_p): New parameter dest_val, check if it is
15437 not the same as the source val.
15438 (cgraph_edge_brings_value_p): New parameter.
15439 (gather_edges_for_value): Pass destination value to
15440 cgraph_edge_brings_value_p.
15441 (perhaps_add_new_callers): Likewise.
15442 (get_info_about_necessary_edges): Likewise and exclude values brought
15443 only by self-recursive edges.
15444 (create_specialized_node): Redirect only clones of self-calling edges.
15445 (+self_recursive_pass_through_p): New function.
15446 (find_more_scalar_values_for_callers_subset): Use it.
15447 (find_aggregate_values_for_callers_subset): Likewise.
15448 (known_aggs_to_agg_replacement_list): Removed.
15449 (decide_whether_version_node): Re-calculate known constants for all
15450 remaining context clones.
15452 2018-04-11 Richard Biener <rguenther@suse.de>
15455 * dwarf2out.c (dwarf2out_finish): Remove DW_AT_stmt_list attribute
15456 from early DWARF output.
15457 (dwarf2out_early_finish): Output line info unconditionally into
15458 early DWARF and add reference to it.
15460 2018-04-11 Jakub Jelinek <jakub@redhat.com>
15463 * config/i386/sse.md (iptr): Add V16SFmode and V8DFmode cases.
15464 (<avx512>_vec_dup<mode><mask_name>): Use a single pattern for modes
15465 other than V2DFmode using iptr mode attribute.
15466 (<avx512>_vec_dup<mode><mask_name>): Use iptr mode attribute.
15468 2018-04-11 Alexander Monakov <amonakov@ispras.ru>
15470 PR rtl-optimization/84659
15471 * sel-sched-ir.c (sel_init_pipelining): Invoke cleanup_cfg.
15473 2018-04-11 Jakub Jelinek <jakub@redhat.com>
15476 * dwarf2out.c (skip_loc_list_entry): Don't call size_of_locs if
15478 (output_loc_list): Pass address of a dummy size variable even in the
15479 locview handling loop.
15480 (index_location_lists): Add comment on why skip_loc_list_entry can't
15483 2018-04-11 Thomas Preud'homme <thomas.preudhomme@arm.com>
15486 * config/arm/arm-builtins.c (arm_expand_builtin): Force input operand
15489 2018-04-10 Aaron Sawdey <acsawdey@linux.ibm.com>
15492 * doc/invoke.texi (RS/6000 and PowerPC Options): Document options
15493 -mblock-compare-inline-limit, -mblock-compare-inline-loop-limit,
15494 and -mstring-compare-inline-limit.
15496 2018-04-10 Segher Boessenkool <segher@kernel.crashing.org>
15499 * config/rs6000/rs6000.md (allocate_stack): Put the residual size
15500 for stack clash protection in a register whenever we need it to be in
15503 2018-04-10 Segher Boessenkool <segher@kernel.crashing.org>
15505 * common/config/rs6000/rs6000-common.c (rs6000_option_init_struct):
15506 Enable -fasynchronous-unwind-tables by default if OBJECT_FORMAT_ELF.
15508 2018-04-10 Segher Boessenkool <segher@kernel.crashing.org>
15511 * config/rs6000/rs6000.opt (mtraceback=): Show the allowed values in
15513 (mlong-double-): Ditto.
15514 * config/rs6000/sysv4.opt (msdata=): Ditto.
15515 (mtls-size=): Ditto.
15517 2018-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
15519 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
15520 erroneous entries for
15521 "vector int vec_ldl (int, long int *)", and
15522 "vector unsigned int vec_ldl (int, unsigned long int *)".
15523 Add comments and entries for
15524 "vector bool char vec_ldl (int, bool char *)",
15525 "vector bool short vec_ldl (int, bool short *)",
15526 "vector bool int vec_ldl (int, bool int *)",
15527 "vector bool long long vec_ldl (int, bool long long *)",
15528 "vector pixel vec_ldl (int, pixel *)",
15529 "vector long long vec_ldl (int, long long *)",
15530 "vector unsigned long long vec_ldl (int, unsigned long long *)".
15531 * config/rs6000/rs6000.c (rs6000_init_builtins): Initialize new
15532 type tree bool_long_long_type_node and correct definition of
15533 bool_V2DI_type_node to make reference to this new type tree.
15534 (rs6000_mangle_type): Replace erroneous reference to
15535 bool_long_type_node with bool_long_long_type_node.
15536 * config/rs6000/rs6000.h (enum rs6000_builtin_type_index): Add
15537 comments to emphasize sign distinctions for char and int types and
15538 replace RS6000_BTI_bool_long constant with
15539 RS6000_BTI_bool_long_long constant. Also add comment to restrict
15540 use of RS6000_BTI_pixel.
15541 (bool_long_type_node): Remove this macro definition.
15542 (bool_long_long_type_node): New macro definition
15544 2018-04-10 Jakub Jelinek <jakub@redhat.com>
15546 PR rtl-optimization/85300
15547 * combine.c (subst): Handle subst of CONST_SCALAR_INT_P new_rtx also
15548 into FLOAT and UNSIGNED_FLOAT like ZERO_EXTEND, return a CLOBBER if
15549 simplify_unary_operation fails.
15551 2018-04-10 Martin Liska <mliska@suse.cz>
15553 * gdbhooks.py: Add pretty-printers for varpool_node, symtab_node,
15554 cgraph_edge and ipa_ref.
15556 2018-04-10 Jakub Jelinek <jakub@redhat.com>
15560 * config/i386/sse.md
15561 (<extract_type>_vinsert<shuffletype><extract_suf>_mask): Fix
15562 computation of the VEC_MERGE selector from mask.
15563 (<extract_type>_vinsert<shuffletype><extract_suf>_1<mask_name>):
15564 Fix decoding of the VEC_MERGE selector into mask.
15566 2018-04-10 Richard Sandiford <richard.sandiford@linaro.org>
15568 PR tree-optimization/85286
15569 * tree-vect-data-refs.c (vect_get_smallest_scalar_type):
15571 2018-04-10 Richard Sandiford <richard.sandiford@linaro.org>
15573 * final.c (final_1): Set insn_last_address as well as
15574 insn_current_address.
15576 2018-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15579 * explow.c (emit_stack_probe): Call validize_mem on memory location
15580 before passing it to gen_probe_stack. Create address operand and
15581 legitimize it for the probe_stack_address case.
15583 2018-04-09 Jan Hubicka <jh@suse.cz>
15586 * ipa-devirt.c (rebuild_type_inheritance-hash): New.
15587 * ipa-utils.h (rebuild_type_inheritance-hash): Declare.
15588 * tree.c (free_lang_data_in_type): Fix handling of binfos;
15590 (free_lang_data): Rebuild type inheritance graph.
15592 2018-04-09 Martin Sebor <msebor@redhat.com>
15594 * invoke.texi (-finline-small-functions): Mention other optimization
15596 (-findirect-inlining, -fpartial-inlining): Same.
15597 (-finline-functions-called-once): Same.
15598 (-freorder-blocks-and-partition): Same.
15600 2018-04-09 Jan Hubicka <jh@suse.cz>
15603 * cfgcleanup.c (try_forward_edges): Do not give up on crossing
15604 jumps; choose last target that matches the criteria (i.e.
15605 no partition changes for non-crossing jumps).
15606 * cfgrtl.c (cfg_layout_redirect_edge_and_branch): Add basic
15607 support for redirecting crossing jumps to non-crossing.
15609 2018-04-09 Alexey Brodkin <abrodkin@synopsys.com>
15611 * config/arc/arc.c (arc_expand_prologue): Set stack usage info
15612 also for naked functions.
15614 2018-04-09 Claudiu Zissulescu <claziss@synopsys.com>
15616 * config/arc/arc.md (add_shift): New pattern.
15617 (add_shift2): Likewise.
15618 (sub_shift): Likewise.
15619 (sub_shift_cmp0_noout): Likewise.
15620 (compare_si_ashiftsi): Likewise.
15621 (xbfu_cmp0_noout): New combine pattern.
15622 (xbfu_cmp0"): Likewise.
15623 (movsi_set_cc_insn): Place the predicable variant first.
15624 (commutative_binary_cmp0_noout): Remove clobber.
15625 (commutative_binary_cmp0): New pattern.
15626 (noncommutative_binary_cmp0): Likewise.
15627 (noncommutative_binary_cmp0_noout): Likewise.
15628 (noncommutative_binary_comparison_result_used): Removed.
15629 (rsub_cmp0): New pattern.
15630 (rsub_cmp0_noout): Likewise.
15631 (extzvsi): Changed, keep only meaningful variants.
15632 (SQH, SEZ): New iterators.
15633 (SQH_postfix): New mode attribute.
15634 (SEZ_prefix): New code attribute.
15635 (<SEZ_prefix>xt<SQH_postfix>_cmp0_noout): New instruction pattern.
15636 (<SEZ_prefix>xt<SQH_postfix>_cmp0): Likewise.
15637 * config/arc/predicates.md (cc_set_register): Use CC_REG instead
15638 of numerical value.
15639 (noncommutative_operator): Check the availability of barrel
15642 2018-04-09 Richard Biener <rguenther@suse.de>
15644 PR tree-optimization/85284
15645 * tree-ssa-loop-niter.c (number_of_iterations_exit_assumptions):
15646 Only use the niter constraining form of simple_iv when the exit
15647 is always executed.
15649 2018-04-09 Tom de Vries <tom@codesourcery.com>
15652 * config/nvptx/nvptx.md (define_c_enum "unspecv"): Add UNSPECV_MEMBAR.
15653 (define_expand "*memory_barrier"): New define_expand.
15654 (define_insn "memory_barrier"): New insn.
15656 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
15658 PR rtl-optimization/80463
15659 PR rtl-optimization/83972
15660 PR rtl-optimization/83480
15662 * sel-sched-ir.c (has_dependence_note_mem_dep): Take into account the
15663 correct producer for the insn.
15664 (tidy_control_flow): Fixup seqnos in case of debug insns.
15666 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
15668 PR rtl-optimization/83913
15670 * sel-sched-ir.c (merge_expr_data): Choose the middle between two
15671 different sched-times when merging exprs.
15673 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
15675 PR rtl-optimization/83962
15677 * sel-sched-ir.c (tidy_control_flow): Correct the order in which we call
15678 tidy_fallthru_edge and tidy_control_flow.
15680 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
15682 PR rtl-optimization/83530
15684 * sel-sched.c (force_next_insn): New global variable.
15685 (remove_insn_for_debug): When force_next_insn is true, also leave only
15686 next insn in the ready list.
15687 (sel_sched_region): When the region wasn't scheduled, make another pass
15688 over it with force_next_insn set to 1.
15690 2018-04-08 Monk Chiang <sh.chiang04@gmail.com>
15692 * config.gcc (nds32le-*-*, nds32be-*-*): Add nds32/nds32_intrinsic.h
15694 * config/nds32/constants.md (unspec_volatile_element): Add enum values
15695 for interrupt control.
15696 * config/nds32/nds32-intrinsic.c: Implementation of intrinsic
15697 functions for interrupt control.
15698 * config/nds32/nds32-intrinsic.md: Likewise.
15699 * config/nds32/nds32_intrinsic.h: Likewise.
15700 * config/nds32/nds32.h (nds32_builtins): Likewise.
15702 2018-04-08 Chung-Ju Wu <jasonwucj@gmail.com>
15704 * config/nds32/nds32.c (nds32_init_machine_status,
15705 nds32_legitimate_index_p, nds32_legitimate_address_p): Consider
15706 strict_aligned_p field.
15707 (nds32_expand_to_rtl_hook): New function.
15708 (TARGET_EXPAND_TO_RTL_HOOK): Define.
15709 * config/nds32/nds32.h (machine_function): Add strict_aligned_p field.
15711 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
15712 Chung-Ju Wu <jasonwucj@gmail.com>
15714 * config.gcc (nds32*-*-*): Check that n7 is valid to --with-cpu.
15715 * config/nds32/nds32-n7.md: New file.
15716 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N7.
15717 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n7
15719 * config/nds32/nds32-protos.h: More declarations for n7 pipeline.
15720 * config/nds32/nds32.md (pipeline_model): Add n7.
15721 * config/nds32/nds32.opt (mcpu): Support n7 pipeline cpus.
15722 * config/nds32/pipelines.md: Include n7 settings.
15724 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
15725 Chung-Ju Wu <jasonwucj@gmail.com>
15727 * config.gcc (nds32*-*-*): Check that e8 is valid to --with-cpu.
15728 * config/nds32/nds32-e8.md: New file.
15729 * config/nds32/nds32-opts.h (nds32-cpu_type): Add CPU_E8.
15730 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for e8
15732 * config/nds32/nds32-protos.h: More declarations for e8 pipeline.
15733 * config/nds32/nds32.md (pipeline_model): Add e8.
15734 * config/nds32/nds32.opt (mcpu): Support e8 pipeline cpus.
15735 * config/nds32/pipelines.md: Include e8 settings.
15737 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
15738 Chung-Ju Wu <jasonwucj@gmail.com>
15740 * config.gcc (nds32*-*-*): Check that n6/n8/s8 are valid to --with-cpu.
15741 * config/nds32/nds32-n8.md: New file.
15742 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N6 and CPU_N8.
15743 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n8
15745 * config/nds32/nds32-protos.h: More declarations for n8 pipeline.
15746 * config/nds32/nds32-utils.c: More implementations for n8 pipeline.
15747 * config/nds32/nds32.md (pipeline_model): Add n8.
15748 * config/nds32/nds32.opt (mcpu): Support n8 pipeline cpus.
15749 * config/nds32/pipelines.md: Include n8 settings.
15751 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
15752 Chung-Ju Wu <jasonwucj@gmail.com>
15754 * config.gcc (nds32*): Add nds32-utils.o into extra_objs.
15755 * config/nds32/nds32-n9-2r1w.md: New file.
15756 * config/nds32/nds32-n9-3r2w.md: New file.
15757 * config/nds32/nds32-opts.h (nds32_cpu_type, nds32_mul_type,
15758 nds32_register_ports): New or modify for cpu n9.
15759 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n9
15761 * config/nds32/nds32-protos.h: More declarations for n9 pipeline.
15762 * config/nds32/nds32-utils.c: New file.
15763 * config/nds32/nds32.h (TARGET_PIPELINE_N9, TARGET_PIPELINE_SIMPLE,
15764 TARGET_MUL_SLOW): Define.
15765 * config/nds32/nds32.md (pipeline_model): New attribute.
15766 * config/nds32/nds32.opt (mcpu, mconfig-mul, mconfig-register-ports):
15767 New options that support cpu n9.
15768 * config/nds32/pipelines.md: Include n9 settings.
15769 * config/nds32/t-nds32 (nds32-utils.o): Add dependency.
15771 2018-04-08 Chung-Ju Wu <jasonwucj@gmail.com>
15773 * config/nds32/nds32-md-auxiliary.c (output_cond_branch): Output align
15774 information if necessary.
15775 (output_cond_branch_compare_zero): Likewise.
15776 * config/nds32/nds32.c (nds32_adjust_insn_length): Consider align case.
15777 (nds32_target_alignment): Refine for alignment.
15778 * config/nds32/nds32.h (NDS32_ALIGN_P): Define.
15779 (FUNCTION_BOUNDARY): Modify.
15780 * config/nds32/nds32.md (call_internal, call_value_internal): Consider
15782 * config/nds32/nds32.opt (malways-align, malign-functions): New.
15784 2018-04-08 Monk Chiang <sh.chiang04@gmail.com>
15786 * config/nds32/constants.md (unspec_volatile_element): Add values for
15787 TLB operation and data prefetch.
15788 * config/nds32/nds32-intrinsic.c: Implementation of intrinsic
15789 functions for TLB operation and data prefetch.
15790 * config/nds32/nds32-intrinsic.md: Likewise.
15791 * config/nds32/nds32_intrinsic.h: Likewise.
15792 * config/nds32/nds32.c (nds32_dpref_names): Likewise.
15793 (nds32_print_operand): Likewise.
15794 * config/nds32/nds32.h (nds32_builtins): Likewise.
15796 2018-04-07 Thomas Koenig <tkoenig@gcc.gnu.org>
15797 Andrew Pinski <pinsika@gcc.gnu.org>
15799 PR middle-end/82976
15800 * match.pd: Use constant_boolean_node of correct type instead of
15801 boolean_true_node or boolean_false_node for simplifying
15802 pointer comparisons to zero.
15804 2018-04-07 Jakub Jelinek <jakub@redhat.com>
15806 PR tree-optimization/80021
15807 * tree.c (verify_type_variant): Make error call in verify_variant_match
15808 translatable and remove final full stop.
15810 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
15812 * config/nds32/constants.md (unspec_volatile_element): Add
15813 UNSPEC_VOLATILE_EH_RETURN.
15814 * config/nds32/nds32-md-auxiliary.c (nds32_output_stack_push,
15815 nds32_output_stack_pop): Support dwarf exception handling process.
15816 * config/nds32/nds32-protos.h (nds32_dynamic_chain_address): Declare.
15817 * config/nds32/nds32.c (nds32_init_machine_status): Support dwarf
15818 exception handling process.
15819 (nds32_compute_stack_frame): Likewise.
15820 (nds32_return_addr_rtx): Likewise.
15821 (nds32_initial_elimination_offset): Likewise.
15822 (nds32_expand_prologue): Likewise.
15823 (nds32_expand_epilogue): Likewise.
15824 (nds32_dynamic_chain_address): New function.
15825 * config/nds32/nds32.h (machine_function): Add fields for dwarf
15826 exception handling.
15827 (DYNAMIC_CHAIN_ADDRESS): Define.
15828 (EH_RETURN_DATA_REGNO): Define.
15829 (EH_RETURN_STACKADJ_RTX): Define.
15830 * config/nds32/nds32.md (eh_return, nds32_eh_return): Implement
15831 patterns for dwarf exception handling.
15833 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
15835 * config/nds32/nds32.h: Clean up obsolete macros.
15837 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
15839 * config/nds32/constants.md (unspec_element, unspec_volatile_element):
15840 Add enum values for particular instructions.
15841 * config/nds32/nds32-intrinsic.c: Implementation of expanding
15842 particular intrinsic functions.
15843 * config/nds32/nds32-intrinsic.md: Likewise.
15844 * config/nds32/nds32_intrinsic.h: Likewise.
15845 * config/nds32/nds32.h (nds32_builtins): Likewise.
15846 * config/nds32/nds32.md (type): Add pbsad and pbsada.
15847 (btst, ave): New patterns for particular instructions.
15849 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
15851 * config/nds32/constants.md (unspec_element, unspec_volatile_element):
15852 Add enum values for atomic load/store and memory sync.
15853 * config/nds32/nds32-intrinsic.c: Implementation for atomic load/store
15855 * config/nds32/nds32-intrinsic.md: Likewise.
15856 * config/nds32/nds32_intrinsic.h: Likewise.
15857 * config/nds32/nds32.h (nds32_builtins): Likewise.
15859 2018-04-07 Jakub Jelinek <jakub@redhat.com>
15861 PR tree-optimization/85257
15862 * fold-const.c (native_encode_vector): If not all elts could fit
15863 and off is -1, return 0 rather than offset.
15864 * tree-ssa-sccvn.c (vn_reference_lookup_3): Pass
15865 (offseti - offset2) / BITS_PER_UNIT as 4th argument to
15866 native_encode_expr. Verify len * BITS_PER_UNIT >= maxsizei. Don't
15867 adjust buffer in native_interpret_expr call.
15869 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
15871 * config/nds32/constants.md (unspec_volatile_element): Add cache
15872 control enum values.
15873 * config/nds32/nds32-intrinsic.c: Add cache control expand functions.
15874 * config/nds32/nds32-intrinsic.md: Add cache control patterns.
15875 * config/nds32/nds32.c (nds32_cctl_names): New.
15876 (nds32_print_operand): Handle cache control register names.
15877 * config/nds32/nds32.h (nds32_builtins): New enum values.
15878 * config/nds32/nds32_intrinsic.h: Add cache control enum types and
15880 * config/nds32/nds32.md (type): Add mmu.
15881 * config/nds32/pipelines.md (simple_insn): Add mmu.
15883 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
15885 * config/nds32/nds32.md (type): Remove call.
15886 * config/nds32/pipelines.md (simple_insn): Likewise.
15888 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
15890 * config/nds32/constants.md (unspec_volatile_element): Add
15891 UNSPEC_VOLATILE_FMFCSR, UNSPEC_VOLATILE_FMTCSR and
15892 UNSPEC_VOLATILE_FMFCFG.
15893 * config/nds32/nds32-intrinsic.c (bdesc_noarg): New builtin
15894 description for fmfcfg and fmfcsr.
15895 (bdesc_1arg): Add fmtcsr.
15896 (bdesc_2arg): Add fcpynss, fcpyss, fcpynsd and fcpysd.
15897 (nds32_expand_builtin_impl): Deal with FPU intrinsic functions.
15898 * config/nds32/nds32-intrinsic.md (unspec_fcpynsd, unspec_fcpysd,
15899 unspec_fcpynss, unspec_fcpysd, unspec_fcpyss, unspec_fmfcsr,
15900 unspec_fmfcfg): New patterns.
15901 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_FMFCFG,
15902 NDS32_BUILTIN_FMFCSR, NDS32_BUILTIN_FMTCSR, NDS32_BUILTIN_FCPYNSS,
15903 NDS32_BUILTIN_FCPYSS,NDS32_BUILTIN_FCPYNSD and NDS32_BUILTIN_FCPYSD.
15904 * config/nds32/nds32_intrinsic.h (__nds32__fcpynsd, __nds32__fcpynss,
15905 __nds32__fcpysd, __nds32__fcpyss, __nds32__fmfcsr, __nds32__fmtcsr,
15906 __nds32__fmfcfg): Define.
15908 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
15910 * config/nds32/nds32.c (nds32_intrinsic_register_names): Add more
15911 intrinsic register names.
15912 * config/nds32/nds32_intrinsic.h (nds32_intrinsic_registers): Add more
15913 intrinsic register enum values and macros.
15915 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
15917 * config/nds32/nds32.c (nds32_legitimate_index_p): Modify condition
15918 for load/store addressing form.
15919 (nds32_print_operand_address): Likewise.
15921 2018-04-06 Eric Botcazou <ebotcazou@adacore.com>
15924 * config/sparc/sparc.c (sparc_expand_move): Deal with symbolic operands
15925 based on LABEL_REF. Remove useless assertion.
15926 (pic_address_needs_scratch): Fix formatting.
15927 (sparc_legitimize_pic_address): Minor tweaks.
15928 (sparc_delegitimize_address): Adjust assertion accordingly.
15929 * config/sparc/sparc.md (movsi_pic_label_ref): Change label_ref_operand
15930 into symbolic_operand.
15931 (movsi_high_pic_label_ref): Likewise.
15932 (movsi_lo_sum_pic_label_ref): Likewise.
15933 (movdi_pic_label_ref): Likewise.
15934 (movdi_high_pic_label_ref): Likewise.
15935 (movdi_lo_sum_pic_label_ref): Likewise.
15937 2018-04-06 Amaan Cheval <amaan.cheval@gmail.com>
15939 * config.gcc (x86_64-*-rtems*): Add rtems.h to tm_file for
15940 custom LIB_SPEC setup.
15942 2018-04-06 Ruslan Bukin <br@bsdpad.com>
15943 Kito Cheng <kito.cheng@gmail.com>
15945 * config.gcc (riscv*-*-freebsd*): Add RISC-V FreeBSD support.
15946 * config/riscv/freebsd.h: New.
15948 2018-04-06 Chung-Ju Wu <jasonwucj@gmail.com>
15950 * config/nds32/nds32.c (nds32_adjust_insn_length): Refine.
15951 * config/nds32/nds32.h (ADJUST_INSN_LENGTH): Change the location in
15954 2018-04-06 Chung-Ju Wu <jasonwucj@gmail.com>
15955 Kito Cheng <kito.cheng@gmail.com>
15957 * config/nds32/nds32-md-auxiliary.c (nds32_output_return,
15958 nds32_output_call, nds32_symbol_binds_local_p): New functions.
15959 * config/nds32/nds32-protos.h (nds32_output_call,
15960 nds32_output_return): Declare.
15961 * config/nds32/nds32.md: Refine all the call and return patterns.
15963 2018-04-06 Jakub Jelinek <jakub@redhat.com>
15966 * dwarf2out.c (rtl_for_decl_init): For STRING_CST initializer only
15967 build CONST_STRING if TYPE_MAX_VALUE is non-NULL and is INTEGER_CST.
15969 PR rtl-optimization/84872
15970 * cfgloopmanip.c (create_preheader): Use make_forwarder_block even if
15971 nentry == 1 when CP_FALLTHRU_PREHEADERS and single_entry is
15972 EDGE_CROSSING edge.
15974 2018-04-06 Tamar Christina <tamar.christina@arm.com>
15976 * expr.c (copy_blkmode_to_reg): Revert 254862.
15977 * doc/sourcebuild.texi (word_mode_no_slow_unalign): Likewise.
15979 2018-04-06 Richard Biener <rguenther@suse.de>
15981 PR middle-end/85244
15982 * tree-dfa.c (get_ref_base_and_extent): Reset seen_variable_array_ref
15983 after seeing a component reference with an adjacent field. Treat
15984 refs to arrays at struct end of external decls similar to
15985 refs to unconstrained commons.
15987 2018-04-06 Jakub Jelinek <jakub@redhat.com>
15990 * fold-const.c (twoval_comparison_p): Remove SAVE_P argument and don't
15991 look through SAVE_EXPRs with non-side-effects argument. Adjust
15993 (fold_comparison): Adjust twoval_comparison_p caller, don't handle
15996 2018-04-06 Richard Biener <rguenther@suse.de>
15998 PR middle-end/85180
15999 * alias.c (find_base_term): New wrapper around find_base_term
16000 unwinding CSELIB_VAL_PTR changes.
16001 (find_base_term): Do not restore CSELIB_VAL_PTR during the
16004 2018-04-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16006 * config/s390/s390.c (s390_z10_optimize_cmp): Expand dedicated NOP
16008 * config/s390/s390.md (UNSPECV_NOP_LR_0, UNSPECV_NOP_LR_1): New
16009 constant definitions.
16010 ("nop"): lr 0,0 -> nopr r0
16011 ("nop_lr0", "nop_lr1"): New insn definitions.
16013 2018-04-06 Chung-Ju Wu <jasonwucj@gmail.com>
16015 * config/nds32/nds32.md (*stack_push, *stack_pop): Use
16016 NDS32_V3PUSH_AVAILABLE_P macro.
16018 2018-04-06 Monk Chiang <sh.chiang04@gmail.com>
16019 Chung-Ju Wu <jasonwucj@gmail.com>
16021 * config.gcc (nds32*-*-*): Add v2j v3f v3s checking.
16022 (nds32*-*-*): Add float and fpu_config into supported_defaults.
16023 * common/config/nds32/nds32-common.c (TARGET_DEFAULT_TARGET_FLAGS):
16024 Include TARGET_DEFAULT_FPU_ISA and TARGET_DEFAULT_FPU_FMA.
16025 * config/nds32/constants.md (unspec_element): Add UNSPEC_COPYSIGN,
16026 UNSPEC_FCPYNSD, UNSPEC_FCPYNSS, UNSPEC_FCPYSD and UNSPEC_FCPYSS.
16027 * config/nds32/constraints.md: New constraints and checking for hard
16028 float configuration.
16029 * config/nds32/iterators.md: New mode iterator and attribute for hard
16030 float configuration.
16031 * config/nds32/nds32-doubleword.md: Use hard float alternatives and
16033 * config/nds32/nds32-fpu.md: New file.
16034 * config/nds32/nds32-md-auxiliary.c: New functions and checkings to
16035 deal with hard float code generation.
16036 * config/nds32/nds32-opts.h (nds32_arch_type): Add ARCH_V3F and
16038 (abi_type, float_reg_number): New enum type.
16039 * config/nds32/nds32-predicates.c: New predicates for hard float.
16040 * config/nds32/nds32-protos.h: Declare functions for hard float.
16041 * config/nds32/nds32.c: Implementation for hard float configuration.
16042 * config/nds32/nds32.h: Definitions for hard float configuration.
16043 * config/nds32/nds32.md: Include hard float machine description and
16044 modify patterns for hard float configuration.
16045 * config/nds32/nds32.opt: New options for hard float configuration.
16046 * config/nds32/predicates.md: New predicates for hard float
16049 2018-04-06 Kuan-Lin Chen <kuanlinchentw@gmail.com>
16051 * common/config/nds32/nds32-common.c
16052 (nds32_option_optimization_table): Enable -mreleax-hint by default.
16054 2018-04-05 Jakub Jelinek <jakub@redhat.com>
16056 PR middle-end/85195
16057 * match.pd (BIT_FIELD_REF CONSTRUCTOR@0 @1 @2): Use view_convert around
16058 CONSTRUCTOR_ELT (ctor, ...)->value.
16060 2018-04-05 Uros Bizjak <ubizjak@gmail.com>
16063 * config/i386/i386.md (define_attr "memory"): Handle rotate1 type.
16065 2018-04-05 Tom de Vries <tom@codesourcery.com>
16068 * config/nvptx/nvptx.c (nvptx_single): Fix neutering of bb with only
16071 2018-04-05 Shiva Chen <shiva0217@gmail.com>
16072 Kito Cheng <kito.cheng@gmail.com>
16074 * config/nds32/constraints.md (U33): Fine-tune checking condition.
16075 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Ditto.
16076 * config/nds32/nds32.h (nds32_16bit_address_type): Add
16077 ADDRESS_POST_MODIFY_LO_REG_IMM3U.
16079 2018-04-05 Shiva Chen <shiva0217@gmail.com>
16080 Kito Cheng <kito.cheng@gmail.com>
16082 * config/nds32/constraints.md (Ufe): New memory constraint.
16083 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format,
16084 nds32_output_16bit_load): Consider r8 register for lwi45.fe format.
16085 * config/nds32/nds32.c (nds32_print_operand): Output lwi45.fe
16087 * config/nds32/nds32.h (nds32_16bit_address_type): Add ADDRESS_R8_IMM7U.
16088 * config/nds32/nds32.md (*mov<mode>): Adjust pattern.
16090 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
16092 * config/nds32/nds32.md: Use optimize_size in the condition for
16093 alu-shift instructions.
16095 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
16097 * config/nds32/nds32.md (divsi4, udivsi4): New patterns.
16099 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
16101 * config/nds32/nds32.md (negsi2): Refine pattern.
16103 2018-04-05 Kito Cheng <kito.cheng@gmail.com>
16104 Chung-Ju Wu <jasonwucj@gmail.com>
16106 * config/nds32/iterators.md (shift_rotate): New code iterator.
16107 (shift): New code attribute.
16108 * config/nds32/nds32-md-auxiliary.c (nds32_expand_constant): New.
16109 * config/nds32/nds32-protos.h (nds32_expand_constant): Declare.
16110 * config/nds32/nds32.c (nds32_print_operand): Deal with more cases.
16111 * config/nds32/nds32.md (addsi3, *add_srli): Refine implementation for
16112 bit-wise operations.
16113 (andsi3, *andsi3): Ditto.
16114 (iorsi3, *iorsi3, *or_slli, *or_srli): Ditto.
16115 (xorsi3, *xorsi3, *xor_slli, *xor_srli): Ditto.
16116 (<shift>si3, *ashlsi3, *ashrsi3, *lshrsi3, *rotrsi3): Ditto.
16117 * config/nds32/predicates.md (nds32_rimm5u_operand, nds32_and_operand,
16118 nds32_ior_operand, nds32_xor_operand): New predicates.
16120 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
16122 * config/nds32/nds32.md (add<mode>3, sub<mode>3): Rename to ...
16123 (addsi3, subsi3): ... this.
16125 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
16127 * config/nds32/nds32.md (*sub_srli, *and_slli): Fine-tune predicator.
16129 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
16131 * config/nds32/nds32.md: Adjust indention.
16133 2018-04-05 Kito Cheng <kito.cheng@gmail.com>
16135 * config/nds32/nds32.md (feature): New attribute.
16137 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
16139 * config/nds32/nds32.md (subtype): New attribute.
16141 2018-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
16144 * config/arm/arm-builtins.c (arm_expand_builtin): Change
16145 expansion to perform a bitwise AND of the argument followed by a
16146 boolean negation of the result.
16148 2018-04-04 Peter Bergner <bergner@vnet.ibm.com>
16150 PR rtl-optimization/84878
16151 * ddg.c (add_cross_iteration_register_deps): Use DF_REF_BB to determine
16152 the basic block. Assert the use reference is not artificial and that
16153 it has an associated insn.
16155 2018-04-04 Michael Matz <matz@suse.de>
16157 * builtins.c (compute_objsize): Pass correct operand
16158 to array_at_struct_end_p.
16160 2018-04-04 Richard Biener <rguenther@suse.de>
16163 * dwarf2out.c (dwarf2out_register_external_die): Peel namespaces
16164 from contexts for DINFO_LEVEL_TERSE and below.
16166 2018-04-04 Kito Cheng <kito.cheng@gmail.com>
16168 * config/nds32/nds32-doubleword.md (move_<mode>): Require
16169 resiter_operand condition.
16170 * config/nds32/nds32.md (*move<mode>): Ditto.
16172 2018-04-04 Kito Cheng <kito.cheng@gmail.com>
16173 Monk Chiang <sh.chiang04@gmail.com>
16175 * config/nds32/nds32.md (movmisalign<mode>): New pattern.
16177 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
16179 * config/nds32/nds32.md (movqi, movhi): Merge into mov<mode>.
16181 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
16182 Kito Cheng <kito.cheng@gmail.com>
16184 * config/nds32/nds32-md-auxiliary.c (nds32_inverse_cond_code,
16185 nds32_cond_code_str, output_cond_branch,
16186 output_cond_branch_compare_zero, nds32_expand_cbranch,
16187 nds32_expand_cstore, nds32_expand_movcc,
16188 nds32_output_cbranchsi4_equality_zero,
16189 nds32_output_cbranchsi4_equality_reg,
16190 nds32_output_cbranchsi4_equality_reg_or_const_int,
16191 nds32_output_cbranchsi4_greater_less_zero: New functions.
16192 * config/nds32/nds32-protos.h (nds32_expand_cbranch,
16193 nds32_expand_cstore, nds32_expand_movcc,
16194 nds32_output_cbranchsi4_equality_zero,
16195 nds32_output_cbranchsi4_equality_reg,
16196 nds32_output_cbranchsi4_equality_reg_or_const_int,
16197 nds32_output_cbranchsi4_greater_less_zero): Declare.
16198 * config/nds32/predicates.md (nds32_movecc_comparison_operator,
16199 nds32_rimm11s_operand): New predicates.
16200 * config/nds32/nds32.h (nds32_expand_result_type): New enum type.
16201 * config/nds32/nds32.md: Rewrite all the branch and conditional move
16204 2018-04-04 Kito Cheng <kito.cheng@gmail.com>
16206 * config/nds32/nds32-doubleword.md: Refine all the instruction type.
16207 * config/nds32/nds32.md: Ditto.
16208 * config/nds32/pipelines.md: Ditto.
16210 2018-04-04 Richard Biener <rguenther@suse.de>
16212 PR tree-optimization/85168
16213 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address): Avoid
16214 propagating abnormals.
16216 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
16218 * config/nds32/nds32.md (enabled): Use yes/no for this attribute.
16220 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
16221 Kito Cheng <kito.cheng@gmail.com>
16223 * config/nds32/nds32-md-auxiliary.c (nds32_long_call_p): New function.
16224 * config/nds32/nds32-protos.h (nds32_long_call_p): Declare.
16225 * config/nds32/nds32.c (nds32_function_ok_for_sibcall): New function.
16226 (TARGET_FUNCTION_OK_FOR_SIBCALL): Define.
16227 * config/nds32/nds32.md (sibcall_internal): New.
16228 (sibcall_register): Remove.
16229 (sibcall_immediate): Remove.
16230 (sibcall_value_internal): New.
16231 (sibcall_value_register): Remove.
16232 (sibcall_value_immediate): Remove.
16233 * config/nds32/predicates.md (nds32_general_register_operand): New.
16234 (nds32_call_address_operand): New.
16236 2018-04-03 Jakub Jelinek <jakub@redhat.com>
16238 PR rtl-optimization/85167
16239 * shrink-wrap.c (move_insn_for_shrink_wrap): Don't set bb_uses and
16240 bb_defs if *split_p, instead preinitialize it to NULL.
16242 PR tree-optimization/85156
16243 * builtins.c (fold_builtin_expect): Use save_expr on arg1 to avoid
16244 evaluating the argument multiple times.
16246 2018-04-03 Bill Schmidt <wschmidt@linux.ibm.com>
16248 * config/rs6000/emmintrin.h (_mm_cvtpd_epi32): Use __vector rather
16250 (_mm_cvtpd_ps): Likewise.
16251 (_mm_cvttpd_epi32): Likewise.
16252 * config/rs6000/mmintrin.h (_mm_unpacklo_pi8): Likewise.
16253 * config/rs6000/xmmintrin.h: For strict-ANSI C++ or C11, undefine
16254 vector, pixel, and bool following altivec.h include.
16256 2018-04-03 Martin Sebor <msebor@redhat.com>
16258 * doc/extend.texi (Common Function Attributes): Clarify.
16259 (const attribute): Likewise.
16260 (pure attribute): Likewise.
16262 2018-04-03 Jakub Jelinek <jakub@redhat.com>
16265 * config/i386/i386.c (ix86_expand_vector_set): Use
16266 HOST_WIDE_INT_1U << elt instead of 1 << elt. Formatting fix.
16268 2018-04-03 Uros Bizjak <ubizjak@gmail.com>
16270 * config/i386/i386.c (emit_i387_cw_initialization): Always use logic
16271 instructions when changing rounding bits to preserve precision bits
16272 in the x87 control word.
16274 2018-04-03 Martin Liska <mliska@suse.cz>
16276 PR tree-optimization/82491
16277 * rtl.h (strip_offset_and_add): Replace += suboffset with
16278 poly_uint64 () + suboffset.
16280 2018-03-29 Martin Liska <mliska@suse.cz>
16281 Martin Jambor <mjambor@suse.cz>
16284 * ipa-cp.c (propagate_bits_across_jump_function): Bail out if
16285 param_type is not an integral or pointer type.
16287 2018-04-03 Richard Biener <rguenther@suse.de>
16289 * sese.h (recompute_all_dominators): Remove.
16291 2018-04-02 Martin Sebor <msebor@redhat.com>
16293 * doc/invoke.texi (-Wrestrict): Fix typos.
16295 2018-04-02 Jim Wilson <jimw@sifive.com>
16297 * config/riscv/riscv.h (SHIFT_COUNT_TRUNCATED): Set to zero.
16298 * config/riscv/riscv.md (<optab>si3): Use QImode shift count.
16299 (<optab>di3, <optab>si3_extend): Likewise.
16300 (<optab>si3_mask, <optab>si3_mask_1): New.
16301 (<optab>di3_mask, <optab>di3_mask_1): New.
16302 (<optab>si3_extend_mask, <optab>si3_extend_mask_1): New.
16303 (lshrsi3_zero_extend_1): Use VOIDmode shift count.
16304 * config/riscv/sync.md (atomic_test_and_set): Emit QImode shift count.
16306 2018-04-02 Gerald Pfeifer <gerald@pfeifer.com>
16308 * doc/cpp.texi (Variadic Macros): Fix line continuation in an
16311 2018-04-02 Chung-Ju Wu <jasonwucj@gmail.com>
16313 * config/nds32/nds32.c (TARGET_CANONICALIZE_COMPARISON): Define.
16314 (nds32_canonicalize_comparison): New function.
16316 2018-04-01 Chung-Ju Wu <jasonwucj@gmail.com>
16317 Kito Cheng <kito.cheng@gmail.com>
16318 Kuan-Lin Chen <kuanlinchentw@gmail.com>
16320 * config.gcc (nds32): Add nds32-relax-opt.o into extra_objs.
16321 * config/nds32/constants.md (unspec_volatile_element): Add
16322 UNSPEC_VOLATILE_RELAX_GROUP.
16323 * config/nds32/nds32-relax-opt.c: New file.
16324 * config/nds32/nds32-predicates.c
16325 (nds32_symbol_load_store_p): New function.
16326 * config/nds32/nds32-protos.h
16327 (nds32_symbol_load_store_p): Declare function.
16328 (make_pass_nds32_relax_opt): Declare new rtl pass function.
16329 * config/nds32/nds32.c
16330 (nds32_register_pass): New function to register pass.
16331 (nds32_register_passes): New function to register passes.
16332 * config/nds32/nds32.md (relax_group): New pattern.
16333 * config/nds32/nds32.opt (mrelax-hint): New option.
16334 * config/nds32/t-nds32 (nds32-relax-opt.o): New dependency.
16336 2018-04-01 Kito Cheng <kito.cheng@gmail.com>
16338 * config/nds32/t-nds32: Modify files dependency.
16340 2018-04-01 Chung-Ju Wu <jasonwucj@gmail.com>
16342 * config/nds32/nds32.h (FUNCTION_PROFILER): Output newline character.
16343 (PROFILE_HOOK): Define its implementation.
16345 2018-04-01 Chung-Ju Wu <jasonwucj@gmail.com>
16347 * config/nds32/nds32.h (WCHAR_TYPE, WCHAR_TYPE_SIZE): Use unsigned int
16348 type and 32-bit size.
16350 2018-04-01 Jakub Jelinek <jakub@redhat.com>
16352 PR middle-end/85090
16353 * config/i386/sse.md (V): Add V64QI and V32HI for TARGET_AVX512F.
16354 (V_128_256): New mode iterator.
16355 (*avx512dq_vextract<shuffletype>64x2_1 splitter): New define_split.
16356 (*avx512f_vextract<shuffletype>32x4_1 splitter): Likewise.
16357 (xop_pcmov_<mode><avxsizesuffix>): Use V_128_256 mode iterator instead
16359 * config/i386/i386.c (ix86_expand_vector_set): Improve V32HImode and
16360 V64QImode expansion for !TARGET_AVX512BW && TARGET_AVX512F.
16362 2018-03-31 Segher Boessenkool <segher@kernel.crashing.org>
16365 * config/rs6000/xmmintrin.h (_mm_set_ps, _mm_max_ps): Handle (quiet)
16366 NaN inputs correctly.
16368 2018-03-30 Peter Bergner <bergner@vnet.ibm.com>
16371 * config/rs6000/vsx.md (??r): New mode attribute.
16372 (*vsx_mov<mode>_64bit): Use it.
16373 (*vsx_mov<mode>_32bit): Likewise.
16375 2018-03-30 Martin Sebor <msebor@redhat.com>
16377 PR tree-optimization/84818
16378 * builtins.c (check_access): Use warning_n.
16380 2018-03-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
16383 * config/rs6000/rs6000-string.c (expand_compare_loop): Fix redundant
16385 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Fix redundant
16388 2018-03-30 Julia Koval <julia.koval@intel.com>
16391 * x86-tune.def (movx, partial_reg_dependency): Enable for
16394 2018-03-29 Vladimir Makarov <vmakarov@redhat.com>
16396 PR inline-asm/84985
16397 * lra-constraints.c (process_alt_operands): Move setting
16398 this_alternative_matches below.
16400 2018-03-29 Martin Liska <mliska@suse.cz>
16403 * doc/invoke.texi: Document how LTO works with debug info.
16404 Describe auto-load support of binutils. Mention 'x86-64'
16405 as valid option value of -march option.
16407 2018-03-29 Jakub Jelinek <jakub@redhat.com>
16409 * config/i386/sse.md (<avx512>_blendm<mode>): Use <sseintprefix>.
16412 * fold-const.c (operand_equal_p): Handle DEBUG_BEGIN_STMT.
16413 For STATEMENT_LIST, pass down OEP_LEXICOGRAPHIC and maybe
16414 OEP_NO_HASH_CHECK for recursive call, to avoid exponential
16417 2018-03-28 Peter Bergner <bergner@vnet.ibm.com>
16420 * config/rs6000/rs6000.h (RS6000_BTM_POWERPC64): New define.
16421 (RS6000_BTM_COMMON): Add RS6000_BTM_POWERPC64.
16422 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add support
16423 for RS6000_BTM_POWERPC64.
16424 (rs6000_invalid_builtin): Add handling for RS6000_BTM_POWERPC64
16425 (rs6000_builtin_mask_names): Add RS6000_BTM_POWERPC64.
16426 * config/rs6000/rs6000-builtin.def (BU_P7_POWERPC64_MISC_2): New macro
16429 (DIVDEU): Likewise.
16431 2018-03-28 Carl Love <cel@us.ibm.com>
16434 2017-09-27 Carl Love <cel@us.ibm.com>
16436 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
16437 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
16438 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
16441 2018-03-28 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16443 * config/rs6000/xmmintrin.h (_mm_max_pi16): Use __vector __bool
16444 instead of __vector bool.
16445 (_mm_max_pu8): Likewise.
16446 (_mm_min_pi16): Likewise.
16448 2018-03-28 Peter Bergner <bergner@vnet.ibm.com>
16451 * config/rs6000/rs6000-builtin.def (DIVWEO): Delete macro expansion.
16452 (DIVWEUO): Likewise.
16453 (DIVDEO): Likewise.
16454 (DIVDEUO): Likewise.
16455 * config/rs6000/rs6000.c (builtin_function_type): Remove support for
16456 DIVWEUO and DIVDEUO.
16457 * config/rs6000/rs6000.md (UNSPEC_DIVEO, UNSPEC_DIVEUO): Delete unspecs.
16458 (UNSPEC_DIV_EXTEND): Remove deleted unspecs.
16459 (div_extend): Likewise.
16460 * doc/extend.texi (__builtin_divweo): Remove documentation for deleted
16462 (__builtin_divweuo): Likewise.
16463 (__builtin_divdeo): Likewise.
16464 (__builtin_divdeuo): Likewise.
16466 2018-03-28 Jakub Jelinek <jakub@redhat.com>
16469 * config/i386/i386.md (*add<mode>3_carry_0, *addsi3_carry_zext_0,
16470 *sub<mode>3_carry_0, *subsi3_carry_zext_0): New patterns.
16472 PR tree-optimization/82004
16473 * gimple-match-head.c (optimize_pow_to_exp): New function.
16474 * match.pd (pow(C,x) -> exp(log(C)*x)): Wrap with #if GIMPLE.
16475 Don't fold to exp if optimize_pow_to_exp is false.
16477 2018-03-28 Martin Liska <mliska@suse.cz>
16480 * calls.c (initialize_argument_information): Fix trailing space.
16481 * common.opt: Fix typo and provide better explanation for
16482 -fsanitize-coverage option.
16483 * config/i386/i386.opt: Fix typo.
16485 2018-03-28 Jakub Jelinek <jakub@redhat.com>
16486 Martin Liska <mliska@suse.cz>
16489 * gimplify.c (asan_poison_variable): Don't do the check for
16490 gimplify_omp_ctxp here.
16491 (gimplify_decl_expr): Do it here.
16492 (gimplify_target_expr): Likewise.
16494 2018-03-28 Martin Liska <mliska@suse.cz>
16497 * config/i386/i386.c (ix86_function_arg_advance): Do not call
16498 chkp_type_bounds_count if MPX is not enabled.
16500 2018-03-27 Chung-Ju Wu <jasonwucj@gmail.com>
16502 * config/nds32/nds32.h (BRANCH_COST): Adjust cost.
16504 2018-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
16507 * config/rs6000/rs6000.c (create_complex_muldiv): New helper
16508 function to create the function decl for complex long double
16509 multiply and divide for -mabi=ieeelongdouble.
16510 (init_float128_ieee): Call it.
16512 2018-03-27 H.J. Lu <hongjiu.lu@intel.com>
16515 * config/i386/i386.c (ix86_trampoline_init): Insert ENDBR for
16516 -fcf-protection=branch -mibt.
16517 * config/i386/i386.h (TRAMPOLINE_SIZE): Increased by 4 bytes.
16519 2018-03-27 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
16522 * config/arm/arm.c (arm_valid_symbolic_address): Handle
16523 arm_word_relocations.
16525 2018-03-27 Cesar Philippidis <cesar@codesourcery.com>
16528 * config/nvptx/nvptx.c (nvptx_assemble_decl_begin): Add '[]' to
16529 extern array declarations.
16531 2018-03-27 Richard Biener <rguenther@suse.de>
16533 PR middle-end/84067
16534 * match.pd ((A * C) +- (B * C) -> (A+-B) * C): Guard with
16535 explicit single_use checks.
16537 2018-03-27 Richard Biener <rguenther@suse.de>
16539 PR tree-optimization/85082
16540 * tree-ssa-sccvn.c (vn_reference_lookup_or_insert_for_pieces):
16543 2018-03-27 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
16545 * config.gcc (aarch64*-*-linux*): New TARGET_DEFAULT_ASYNC_UNWIND_TABLES
16546 * common/config/aarch64/aarch64-common.c (aarch64_optimization_table[]):
16547 Turn on fasynchronous-unwind-tables and funwind-tables.
16549 2018-03-26 Uros Bizjak <ubizjak@gmail.com>
16552 * config/i386/i386.md (*bmi_blsr_<mode>_cmp): New insn pattern.
16553 (*bmi_blsr_<mode>_ccz): Ditto.
16555 2018-03-26 Tom de Vries <tom@codesourcery.com>
16557 PR tree-optimization/85063
16558 * omp-general.c (offloading_function_p): New function. Factor out
16560 * omp-offload.c (pass_omp_target_link::gate): ... here.
16561 * omp-general.h (offloading_function_p): Declare.
16562 * tree-switch-conversion.c (build_one_array): Mark CSWTCH.x variable
16563 with attribute omp declare target for offloading functions.
16565 2018-03-24 Richard Sandiford <richard.sandiford@linaro.org>
16567 PR tree-optimization/84005
16568 * tree-data-ref.h (get_base_for_alignment): Declare.
16569 * tree-data-ref.c (get_base_for_alignment_1): New function.
16570 (get_base_for_alignment): Likewise.
16571 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
16572 get_base_for_alignment to find a suitable base object, instead
16573 of always using drb->base_address.
16575 2018-03-23 Jakub Jelinek <jakub@redhat.com>
16577 PR inline-asm/85022
16578 * emit-rtl.c (init_emit_regs): Indicate that VOIDmode MEMs don't have
16579 known size by default.
16581 2018-03-23 Vladimir Makarov <vmakarov@redhat.com>
16583 PR inline-asm/85030
16584 * lra-constraints.c (process_alt_operands): Don't match BLKmode
16585 and non BLKmode operands.
16587 2018-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16590 * config/arm/arm.md (unaligned_loadhis): Remove first alternative.
16591 Clean up attributes.
16593 2018-03-23 Richard Biener <rguenther@suse.de>
16596 * dwarf2out.c (rtl_for_decl_location): Do not generate RTL early when
16597 we are going to emit early debug for LTO.
16599 2018-03-23 Jakub Jelinek <jakub@redhat.com>
16601 PR inline-asm/85034
16602 * function.c (match_asm_constraints_1): Don't optimize if input
16603 doesn't satisfy general_operand predicate for output's mode.
16605 PR inline-asm/85022
16606 * alias.c (write_dependence_p): Don't require for x_canonicalized
16607 non-VOIDmode if x has VOIDmode.
16610 * sanopt.c (maybe_optimize_ubsan_ptr_ifn): If DECL_REGISTER (base),
16611 just don't try to optimize it rather than assert it never happens.
16613 2018-03-22 Kelvin Nilsen <kelvin@gcc.gnu.org>
16615 * config/rs6000/rs6000-builtin.def: Remove various BU_ALTIVEC_X
16616 macro expansions for definition of ST_INTERNAL_<mode> and
16617 LD_INTERNAL_<mode> builtins.
16618 * config/rs6000/rs6000-protos.h (rs6000_address_for_altivec):
16620 * config/rs6000/rs6000.c (altivec_expand_ld_builtin): Delete this
16622 (altivec_expand_st_builtin): Likewise.
16623 (altivec_expand_builtin): Remove calls to deleted functions.
16624 (rs6000_address_for_altivec): Delete this function.
16625 * config/rs6000/vector.md: Remove expands for
16626 vector_altivec_load_<mode> and vector_altivec_store_<mode>.
16628 2018-03-22 Sudakshina Das <sudi.das@arm.com>
16631 * config/arm/arm.h (machine_function): Add static_chain_stack_bytes.
16632 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Avoid
16633 re-computing once computed.
16634 (arm_expand_prologue): Compute machine->static_chain_stack_bytes.
16635 (arm_init_machine_status): Initialize
16636 machine->static_chain_stack_bytes.
16638 2018-03-22 Kelvin Nilsen <kelvin@gcc.gnu.org>
16641 * doc/extend.texi: Add four new prototypes for vec_ld.
16642 * config/rs6000/rs6000-builtin.def (LVX_V1TI): Reorder symbol
16643 definitions for more logical presentation.
16644 * config/rs6000/rs6000-c.c: (altivec_overloaded_builtins): Add
16645 entries for V1TI variants of __builtin_altivec_ld builtin.
16646 * config/rs6000/rs6000.c: (altivec_expand_lv_builtin): Add test and
16647 handling of V1TI variant of LVX icode pattern.
16648 (altivec_expand_builtin): Add case for ALTIVEC_BUILTIN_LVX_V1TI.
16649 (rs6000_gimple_fold_builtin): Likewise.
16650 (altivec_init_builtins): Add code to define
16651 __builtin_altivec_lvx_v1ti function.
16653 2018-03-22 Jakub Jelinek <jakub@redhat.com>
16655 PR inline-asm/84941
16656 * function.c (match_asm_constraints_1): Don't do the optimization
16657 if input isn't a REG, SUBREG, MEM or constant.
16659 2018-03-22 Tom de Vries <tom@codesourcery.com>
16661 PR tree-optimization/84956
16662 * tree-ssa-tail-merge.c (find_clusters_1): Skip bbs with
16663 bb_has_abnormal_pred.
16665 2018-03-22 Jakub Jelinek <jakub@redhat.com>
16668 * dwarf2asm.c (dw2_output_indirect_constant_1): Set
16669 DECL_INITIAL (decl) to decl at the end.
16670 * varasm.c (use_blocks_for_decl_p): Revert the 2018-03-20 change,
16671 adjust the comment.
16673 2018-03-21 Joseph Myers <joseph@codesourcery.com>
16675 * doc/extend.texi (__builtin_tgmath): Document when complex
16676 integer types are treated as _Complex _Float64.
16678 2018-03-21 Tom de Vries <tom@codesourcery.com>
16680 * doc/extend.texi (__builtin_extend_pointer): Remove pasto.
16682 2018-03-21 Jakub Jelinek <jakub@redhat.com>
16684 PR tree-optimization/84960
16685 * tree-cfg.c (remove_bb): Don't move forced labels into bb->prev_bb
16686 if it is ENTRY block, move them into single succ of ENTRY in that case.
16688 2018-03-21 Richard Sandiford <richard.sandiford@linaro.org>
16690 PR tree-optimization/84811
16691 * poly-int.h (poly_span_traits): Remove the T3 parameter and
16692 promote HOST_WIDE_INT T2 - T1 results to unsigned HOST_WIDE_INT.
16693 (maybe_in_range_p, known_in_range_p, ranges_known_overlap_p):
16694 (known_subrange_p): Update accordingly. Cast each value involved
16695 in the size comparison, rather than casting the result of the
16698 2018-03-21 Jakub Jelinek <jakub@redhat.com>
16700 PR tree-optimization/84982
16701 * gimple-ssa-store-merging.c (invert_op): Handle boolean inversion
16702 by flipping the least significant bit rather than all bits from
16703 bitpos to bitpos + bitsize - 1.
16705 2018-03-21 Nathan Sidwell <nathan@acm.org>
16707 * doc/extend.texi (Deprecated Features): Remove mention of
16708 long-deleted deprecations.
16710 2018-03-21 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
16713 * configure.ac (gcc_cv_ld_soname) <*-*-solaris2*>: Set.
16714 * configure: Regenerate.
16716 2018-03-21 Tom de Vries <tom@codesourcery.com>
16718 PR tree-optimization/83126
16719 * tree-parloops.c (num_phis): New function.
16720 (gen_parallel_loop): Detect and handle canonicalize_loop_ivs failure.
16722 2018-03-21 Nathan Sidwell <nathan@acm.org>
16724 * doc/extend.texi (Deprecated Features): Update deprecated flags,
16725 mention anon-struct/union members and trailing attributes.
16727 2018-03-21 Bin Cheng <bin.cheng@arm.com>
16729 PR tree-optimization/84969
16730 * tree-loop-distribution.c (fuse_memset_builtins): Don't reorder
16731 builtin memset partitions if they set different rhs values.
16733 2018-03-21 Jakub Jelinek <jakub@redhat.com>
16735 PR rtl-optimization/84989
16736 * simplify-rtx.c (simplify_unary_operation_1): Don't try to simplify
16737 VEC_DUPLICATE with scalar result mode.
16739 2018-03-21 Martin Liska <mliska@suse.cz>
16742 * ipa-icf.c (sem_item_optimizer::fixup_points_to_sets): Remove
16743 not intended return statement.
16745 2018-03-21 Martin Liska <mliska@suse.cz>
16748 * tree-chkp.c (CHKP_ARRAY_MAX_CHECK_STEPS): Define a new macro.
16749 (chkp_find_bound_slots_1): Limit number of iterations.
16751 2018-03-20 David H. Gutteridge <dhgutteridge@sympatico.ca>
16754 * Minor grammar fixes for x86 options.
16756 2018-03-20 Jakub Jelinek <jakub@redhat.com>
16759 * dce.c (delete_unmarked_insns): Don't remove frame related noop moves
16760 holding REG_CFA_RESTORE notes, instead turn them into a USE.
16762 2018-03-20 Peter Bergner <bergner@vnet.ibm.com>
16765 * config/rs6000/altivec.md (altivec_lvx_<mode>_2op): Delete define_insn.
16766 (altivec_lvx_<mode>_1op): Likewise.
16767 (altivec_stvx_<mode>_2op): Likewise.
16768 (altivec_stvx_<mode>_1op): Likewise.
16769 (altivec_lvx_<VM2:mode>): New define_expand.
16770 (altivec_stvx_<VM2:mode>): Likewise.
16771 (altivec_lvx_<VM2:mode>_2op_<P:mptrsize>): New define_insn.
16772 (altivec_lvx_<VM2:mode>_1op_<P:mptrsize>): Likewise.
16773 (altivec_stvx_<VM2:mode>_2op_<P:mptrsize>): Likewise.
16774 (altivec_stvx_<VM2:mode>_1op_<P:mptrsize>): Likewise.
16775 * config/rs6000/rs6000-p8swap.c (rs6000_gen_stvx): Use new expanders.
16776 (rs6000_gen_lvx): Likewise.
16777 * config/rs6000/rs6000.c (altivec_expand_lv_builtin): Likewise.
16778 (altivec_expand_stv_builtin): Likewise.
16779 (altivec_expand_builtin): Likewise.
16780 * config/rs6000/vector.md: Likewise.
16782 2018-03-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16785 * config/arm/arm.c (arm_array_mode_supported_p): Return false for
16788 2018-03-20 Richard Biener <rguenther@suse.de>
16791 * config/i386/i386.c (ix86_add_stmt_cost): Only cost
16792 sign-conversions as zero, fall back to standard scalar_stmt
16795 2018-03-20 Martin Liska <mliska@suse.cz>
16798 * predict.c (rebuild_frequencies): Handle case when we have
16799 PROFILE_ABSENT, but flag_guess_branch_prob is false.
16801 2018-03-20 Jakub Jelinek <jakub@redhat.com>
16804 * dwarf2asm.c (dw2_output_indirect_constant_1): Temporarily turn off
16805 flag_section_anchors.
16806 * varasm.c (use_blocks_for_decl_p): Remove hack for
16807 dw2_force_const_mem.
16810 * config/aarch64/aarch64.md (*aarch64_reg_<mode>3_neg_mask2): Rename
16812 (*aarch64_<optab>_reg_<mode>3_neg_mask2): ... this. If pseudos can't
16813 be created, use lowpart_subreg of operands[0] rather than operands[0]
16815 (*aarch64_reg_<mode>3_minus_mask): Rename to ...
16816 (*aarch64_ashl_reg_<mode>3_minus_mask): ... this.
16817 (*aarch64_<optab>_reg_di3_mask2): Use const_int_operand predicate
16818 and n constraint instead of aarch64_shift_imm_di and Usd.
16819 (*aarch64_reg_<optab>_minus<mode>3): Rename to ...
16820 (*aarch64_<optab>_reg_minus<mode>3): ... this.
16822 2018-03-20 Sudakshina Das <sudi.das@arm.com>
16825 * config/arm/neon.md (ashldi3_neon): Update ?s for constraints
16826 to favor GPR over NEON registers.
16827 (<shift>di3_neon): Likewise.
16829 2018-03-20 Tom de Vries <tom@codesourcery.com>
16832 * config/nvptx/nvptx.c (nvptx_single): Don't neuter bar.sync.
16833 (nvptx_process_pars): Emit bar.sync asap and alap.
16835 2018-03-20 Tom de Vries <tom@codesourcery.com>
16838 * config/nvptx/nvptx.c (prevent_branch_around_nothing): Also update
16839 seen_label if seen_label is already set.
16841 2018-03-20 Jakub Jelinek <jakub@redhat.com>
16844 * config/i386/i386.c (fold_builtin_cpu): For features above 31
16845 use __cpu_features2 variable instead of __cpu_model.__cpu_features[0].
16846 Use 1U instead of 1. Formatting fixes.
16849 * builtins.c (fold_builtin_strpbrk): For strpbrk(x, "") use type
16850 instead of TREE_TYPE (s1) for the return value.
16852 2018-03-19 Jakub Jelinek <jakub@redhat.com>
16854 PR tree-optimization/84946
16855 * gimple-ssa-store-merging.c (mem_valid_for_store_merging): Compute
16856 bitsize + bitsize in poly_uint64 rather than poly_int64.
16859 * dwarf2asm.c: Include fold-const.c.
16860 (dw2_output_indirect_constant_1): Set DECL_INITIAL (decl) to ADDR_EXPR
16861 of decl rather than decl itself.
16863 PR rtl-optimization/84643
16864 * memmodel.h (enum memmodel): Add MEMMODEL_MAX enumerator.
16866 2018-03-19 Maxim Ostapenko <m.ostapenko@samsung.com>
16869 * dwarf2asm.c (dw2_output_indirect_constant_1): Disable ASan before
16870 calling assemble_variable.
16872 2018-03-19 Sudakshina Das <sudi.das@arm.com>
16875 * config/aarch64/aarch64-simd.md (vec_cmp<mode><v_int_equiv>): Modify
16876 instructions for UNLT, UNLE, UNGT, UNGE, UNEQ, UNORDERED and ORDERED.
16878 2018-03-19 Jim Wilson <jimw@sifive.com>
16881 * config/riscv/riscv.c (riscv_function_arg_boundary): Use
16882 PREFERRED_STACK_BOUNDARY instead of STACK_BOUNDARY.
16883 (riscv_first_stack_step): Likewise.
16884 (riscv_option_override): Use STACK_BOUNDARY instead of
16885 MIN_STACK_BOUNDARY.
16886 * config/riscv/riscv.h (STACK_BOUNDARY): Renamed from
16887 MIN_STACK_BOUNDARY.
16888 (BIGGEST_ALIGNMENT): Set to 128.
16889 (PREFERRED_STACK_BOUNDARY): Renamed from STACK_BOUNDARY.
16890 (RISCV_STACK_ALIGN): Use PREFERRED_STACK_BOUNDARY instead of
16893 2018-03-19 Richard Biener <rguenther@suse.de>
16895 PR tree-optimization/84933
16896 * tree-vrp.c (set_and_canonicalize_value_range): Treat out-of-bound
16897 values as -INF/INF when canonicalizing an ANTI_RANGE to a RANGE.
16899 2018-03-19 Richard Biener <rguenther@suse.de>
16901 PR tree-optimization/84859
16902 * tree-ssa-phiopt.c (single_trailing_store_in_bb): New function.
16903 (cond_if_else_store_replacement): Perform sinking operation on
16904 single-store BBs regardless of MAX_STORES_TO_SINK setting.
16905 Generalize what a BB with a single eligible store is.
16907 2018-03-19 Richard Biener <rguenther@suse.de>
16909 PR tree-optimization/84929
16910 * tree-data-ref.c (analyze_siv_subscript_cst_affine): Guard
16911 chrec_is_positive against non-chrec arg.
16913 2018-03-19 Tamar Christina <tamar.christina@arm.com>
16916 * config/arm/arm.c (arm_can_change_mode_class): revert r258554.
16918 2018-03-18 Martin Liska <mliska@suse.cz>
16920 PR rtl-optimization/84635
16921 * regrename.c (build_def_use): Use matches_mode only when
16924 2018-03-18 Richard Sandiford <richard.sandiford@linaro.org>
16926 PR tree-optimization/84913
16927 * tree-vect-loop.c (vectorizable_reduction): Don't try to
16928 vectorize chains of COND_EXPRs.
16930 2018-03-18 Chung-Ju Wu <jasonwucj@gmail.com>
16932 * config/nds32/nds32.h (MAX_REGS_PER_ADDRESS): Fix the value.
16934 2018-03-18 Chung-Ju Wu <jasonwucj@gmail.com>
16936 * config/nds32/nds32.h (LOGICAL_OP_NON_SHORT_CIRCUIT): Define.
16938 2018-03-18 Chung-Ju Wu <jasonwucj@gmail.com>
16940 * config/nds32/nds32.h (CLZ_DEFINED_VALUE_AT_ZERO): Define.
16942 2018-03-17 Chung-Ju Wu <jasonwucj@gmail.com>
16943 Kito Cheng <kito.cheng@gmail.com>
16945 * config/nds32/nds32-protos.h (nds32_adjust_reg_alloc_order): Declare.
16946 * config/nds32/nds32.c (nds32_reg_alloc_order_for_speed): New array.
16947 (nds32_adjust_reg_alloc_order): New function.
16948 * config/nds32/nds32.h (ADJUST_REG_ALLOC_ORDER): Define.
16950 2018-03-17 Kito Cheng <kito.cheng@gmail.com>
16952 * config/nds32/nds32.c (nds32_asm_output_mi_thunk,
16953 nds32_print_operand, nds32_print_operand_address): Use
16954 HOST_WIDE_INT_PRINT_DEC instead.
16956 2018-03-17 Chung-Ju Wu <jasonwucj@gmail.com>
16958 * config/nds32/nds32.c (nds32_register_priority): Modify cost.
16960 2018-03-17 Jakub Jelinek <jakub@redhat.com>
16963 * config/i386/i386.c (initial_ix86_tune_features,
16964 initial_ix86_arch_features): Use unsigned HOST_WIDE_INT rather than
16965 unsigned long long.
16966 (set_ix86_tune_features): Change ix86_tune_mask from unsigned int
16967 to unsigned HOST_WIDE_INT, initialize to HOST_WIDE_INT_1U << ix86_tune
16968 rather than 1u << ix86_tune. Formatting fix.
16969 (ix86_option_override_internal): Change ix86_arch_mask from
16970 unsigned int to unsigned HOST_WIDE_INT, initialize to
16971 HOST_WIDE_INT_1U << ix86_arch rather than 1u << ix86_arch.
16972 (ix86_function_specific_restore): Likewise.
16974 2018-03-16 Jakub Jelinek <jakub@redhat.com>
16977 * postreload.c (reload_combine_recognize_pattern): Perform
16978 INTVAL addition in unsigned HOST_WIDE_INT type to avoid UB and
16979 truncate_int_for_mode the result for the destination's mode.
16982 * hsa-gen.c (mem_type_for_type): Fix comment typo.
16983 * tree-vect-loop-manip.c (vect_create_cond_for_niters_checks):
16985 * gimple-ssa-warn-restrict.c (builtin_memref::set_base_and_offset):
16988 2018-03-16 Vladimir Makarov <vmakarov@redhat.com>
16991 * lra-assigns.c (lra_split_hard_reg_for): Don't use
16992 regno_allocno_class_array and sorted_pseudos.
16993 * lra-constraints.c (spill_hard_reg_in_range): Ignore hard regs in
16994 insns where regno is used.
16996 2018-03-16 Martin Liska <mliska@suse.cz>
16999 * multiple_target.c (create_dispatcher_calls): Redirect
17000 reference in the symbol table.
17002 2018-03-16 Martin Liska <mliska@suse.cz>
17005 * multiple_target.c (create_dispatcher_calls): Redirect also
17008 2018-03-16 Jakub Jelinek <jakub@redhat.com>
17012 * tree.h (TARGET_EXPR_NO_ELIDE): Define.
17013 * gimplify.c (gimplify_modify_expr_rhs): Don't elide TARGET_EXPRs with
17014 TARGET_EXPR_NO_ELIDE flag set unless *expr_p is INIT_EXPR.
17016 2018-03-16 Julia Koval <julia.koval@intel.com>
17018 * doc/invoke.texi (Skylake Server): Add CLWB.
17019 Cannonlake): Remove CLWB.
17021 2018-03-16 Jakub Jelinek <jakub@redhat.com>
17023 PR tree-optimization/84841
17024 * tree-ssa-reassoc.c (INTEGER_CONST_TYPE): Change to 1 << 4 from
17026 (FLOAT_ONE_CONST_TYPE): Define.
17027 (constant_type): Return FLOAT_ONE_CONST_TYPE for -1.0 and 1.0.
17028 (sort_by_operand_rank): Put entries with higher constant_type last
17029 rather than first to match comments.
17031 2018-03-15 Sandra Loosemore <sandra@codesourcery.com>
17033 * config/nios2/nios2.md (movsi_internal): Fix thinko in
17036 2018-03-15 Jakub Jelinek <jakub@redhat.com>
17039 * calls.c (expand_call): For TREE_ADDRESSABLE rettype ignore alignment
17040 check and use address of target always.
17042 2018-03-15 H.J. Lu <hongjiu.lu@intel.com>
17045 * config/i386/i386.c (indirect_thunk_needed): Update comments.
17046 (indirect_thunk_bnd_needed): Likewise.
17047 (indirect_thunks_used): Likewise.
17048 (indirect_thunks_bnd_used): Likewise.
17049 (indirect_return_needed): New.
17050 (indirect_return_bnd_needed): Likewise.
17051 (output_indirect_thunk_function): Add a bool argument for
17053 (output_indirect_thunk_function): Don't generate alias for
17054 function return thunk.
17055 (ix86_code_end): Call output_indirect_thunk_function to generate
17056 function return thunks.
17057 (ix86_output_function_return): Set indirect_return_bnd_needed
17058 and indirect_return_needed instead of indirect_thunk_bnd_needed
17059 and indirect_thunk_needed.
17061 2018-03-15 Olga Makhotina <olga.makhotina@intel.com>
17063 * config/i386/sgxintrin.h (_enclv_u32): New intrinsic.
17064 (__enclv_bc, __enclv_cd, __enclv_generic): New definitions.
17065 (ERDINFO, ETRACKC, ELDBC, ELDUC): New leaves.
17067 2018-03-15 David Malcolm <dmalcolm@redhat.com>
17068 Paul Hua <paul.hua.gm@gmail.com>
17071 * gcc.dg/fixits-pr84852-1.c: Fix filename in dg-regexp.
17073 2018-03-15 Segher Boessenkool <segher@kernel.crashing.org>
17075 * config/rs6000/rs6000.c (abi_v4_pass_in_fpr): Add back the
17076 TARGET_DOUBLE_FLOAT and TARGET_SINGLE_FLOAT conditions on the DFmode
17077 resp. SFmode cases.
17079 2018-03-15 Tamar Christina <tamar.christina@arm.com>
17082 * config/arm/arm.c (arm_can_change_mode_class): Use GET_MODE_UNIT_SIZE
17083 instead of GET_MODE_SIZE when comparing Units.
17085 2018-03-15 Vladimir Mezentsev <vladimir.mezentsev@oracle.com>
17088 * varasm.c (hash_section): Return an unchangeble hash value
17089 * config/aarch64/aarch64.c (aarch64_use_blocks_for_constant_p):
17090 Return !aarch64_can_use_per_function_literal_pools_p ().
17092 2018-03-15 Jakub Jelinek <jakub@redhat.com>
17095 * optabs.c (emit_conditional_move): Pass address of cmode's copy
17096 rather than address of cmode as last argument to prepare_cmp_insn.
17098 2018-03-15 Julia Koval <julia.koval@intel.com>
17100 * config/i386/i386.c (F_AVX512VBMI2, F_GFNI, F_VPCLMULQDQ,
17101 F_AVX512VNNI, F_AVX512BITALG): New.
17103 2018-03-14 John David Anglin <danglin@gcc.gnu.org>
17106 * config/pa/pa.c (pa_emit_move_sequence): Always emit secondary reload
17107 insn for floating-point loads and stores.
17109 2018-03-14 Carl Love <cel@us.ibm.com>
17111 * config/rs6000/rs6000-c.c: Add macro definitions for
17112 ALTIVEC_BUILTIN_VEC_PERMXOR.
17113 * config/rs6000/rs6000.h: Add #define for vec_permxor builtin.
17114 * config/rs6000/rs6000-builtin.def: Add macro expansions for VPERMXOR.
17115 * config/rs6000/altivec.md (altivec_vpermxor): New define expand.
17116 * config/rs6000/rs6000-p8swap.c (rtx_is_swappable_p): Add case
17118 * config/doc/extend.texi: Add prototypes for vec_permxor.
17120 2018-03-14 David Malcolm <dmalcolm@redhat.com>
17123 * diagnostic-show-locus.c (class layout_point): Convert m_line
17124 from int to linenum_type.
17125 (line_span::comparator): Use linenum "compare" function when
17126 comparing line numbers.
17127 (test_line_span): New function.
17128 (layout_range::contains_point): Convert param "row" from int to
17130 (layout_range::intersects_line_p): Likewise.
17131 (layout::will_show_line_p): Likewise.
17132 (layout::print_source_line): Likewise.
17133 (layout::should_print_annotation_line_p): Likewise.
17134 (layout::print_annotation_line): Likewise.
17135 (layout::print_leading_fixits): Likewise.
17136 (layout::annotation_line_showed_range_p): Likewise.
17137 (struct line_corrections): Likewise for field m_row.
17138 (line_corrections::line_corrections): Likewise for param "row".
17139 (layout::print_trailing_fixits): Likewise.
17140 (layout::get_state_at_point): Likewise.
17141 (layout::get_x_bound_for_row): Likewise.
17142 (layout::print_line): Likewise.
17143 (diagnostic_show_locus): Likewise for locals "last_line" and "row".
17144 (selftest::diagnostic_show_locus_c_tests): Call test_line_span.
17145 * input.c (selftest::test_linenum_comparisons): New function.
17146 (selftest::input_c_tests): Call it.
17147 * selftest.c (selftest::test_assertions): Test ASSERT_GT,
17148 ASSERT_GT_AT, ASSERT_LT, and ASSERT_LT_AT.
17149 * selftest.h (ASSERT_GT): New macro.
17150 (ASSERT_GT_AT): New macro.
17151 (ASSERT_LT): New macro.
17152 (ASSERT_LT_AT): New macro.
17154 2018-03-14 Segher Boessenkool <segher@kernel.crashing.org>
17156 PR rtl-optimization/84780
17157 * combine.c (distribute_links): Don't make a link based on pc_rtx.
17159 2018-03-14 Martin Liska <mliska@suse.cz>
17161 * tree.c (record_node_allocation_statistics): Use
17162 get_stats_node_kind.
17163 (get_stats_node_kind): New function extracted from
17164 record_node_allocation_statistics.
17165 (free_node): Use get_stats_node_kind.
17167 2018-03-14 Richard Biener <rguenther@suse.de>
17169 * tree-ssa-pre.c (compute_antic_aux): Remove code that asserts
17170 that the value-set of ANTIC_IN doesn't grow.
17173 * tree-ssa-pre.c (struct bb_bitmap_sets): Add visited_with_visited_succs
17175 (BB_VISITED_WITH_VISITED_SUCCS): New define.
17176 (compute_antic): Initialize BB_VISITED_WITH_VISITED_SUCCS.
17178 2018-03-14 Julia Koval <julia.koval@intel.com>
17180 * config.gcc (icelake-client, icelake-server): New.
17182 * config/i386/i386.c (initial_ix86_tune_features): Extend to 64 bit.
17183 (initial_ix86_arch_features): Ditto.
17184 (PTA_SKYLAKE): Add SGX.
17185 (PTA_ICELAKE): Remove.
17186 (PTA_ICELAKE_CLIENT): New.
17187 (PTA_ICELAKE_SERVER): New.
17188 (ix86_option_override_internal): Split up icelake on icelake client and
17190 (get_builtin_code_for_version): Ditto.
17191 (fold_builtin_cpu): Ditto.
17192 * config/i386/driver-i386.c (config/i386/driver-i386.c): Ditto.
17193 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto
17194 * config/i386/i386.h (processor_type): Ditto.
17195 * doc/invoke.texi: Ditto.
17197 2018-03-14 Jakub Jelinek <jakub@redhat.com>
17200 * sanopt.c (maybe_optimize_ubsan_ptr_ifn): Handle also
17201 INTEGER_CST offset, add it together with bitpos / 8 and
17202 sign extend based on POINTER_SIZE.
17206 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
17209 * config/i386/constraints.md (Yc): New register constraint.
17210 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
17211 Use Yc constraint for alternative 2 of operand 0. Remove
17212 preferred_for_speed attribute.
17214 2018-03-14 Richard Biener <rguenther@suse.de>
17216 PR tree-optimization/84830
17217 * tree-ssa-pre.c (compute_antic_aux): Intersect the new ANTIC_IN
17218 with the old one to avoid oscillations.
17220 2018-03-13 Vladimir Makarov <vmakarov@redhat.com>
17223 * lra-assigns.c (find_all_spills_for): Ignore uninteresting
17225 (assign_by_spills): Return a flag of reload assignment failure.
17226 Do not process the reload assignment failures. Do not spill other
17227 reload pseudos if they has the same reg class. Update n if
17229 (lra_assign): Add a return arg. Set up from the result of
17230 assign_by_spills call.
17231 (find_reload_regno_insns, lra_split_hard_reg_for): New functions.
17232 * lra-constraints.c (split_reg): Add a new arg. Use it instead of
17233 usage_insns if it is not NULL.
17234 (spill_hard_reg_in_range): New function.
17235 (split_if_necessary, inherit_in_ebb): Pass a new arg to split_reg.
17236 * lra-int.h (spill_hard_reg_in_range, lra_split_hard_reg_for): New
17237 function prototypes.
17238 (lra_assign): Change prototype.
17239 * lra.c (lra): Add code to deal with fails by splitting hard reg
17242 2018-03-01 Palmer Dabbelt <palmer@sifive.com>
17244 * config/riscv/riscv.opt (mrelax): New option.
17245 * config/riscv/riscv.c (riscv_file_start): Emit ".option
17246 "norelax" when riscv_mrelax is disabled.
17247 * doc/invoke.texi (RISC-V): Document "-mrelax" and "-mno-relax".
17249 2018-03-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
17252 * config/rs6000/rs6000.c (rs6000_reassociation_width): Disable parallel
17253 reassociation for int modes.
17255 2018-03-13 Richard Sandiford <richard.sandiford@linaro.org>
17257 * tree-vect-loop-manip.c (vect_maybe_permute_loop_masks):
17258 Reverse the choice between VEC_UNPACK_LO_EXPR and VEC_UNPACK_HI_EXPR
17260 * config/aarch64/iterators.md (hi_lanes_optab): New int attribute.
17261 * config/aarch64/aarch64-sve.md
17262 (*aarch64_sve_<perm_insn><perm_hilo><mode>): Rename to...
17263 (aarch64_sve_<perm_insn><perm_hilo><mode>): ...this.
17264 (*extend<mode><Vwide>2): Rename to...
17265 (aarch64_sve_extend<mode><Vwide>2): ...this.
17266 (vec_unpack<su>_<perm_hilo>_<mode>): Turn into a define_expand,
17267 renaming the old pattern to...
17268 (aarch64_sve_punpk<perm_hilo>_<mode>): ...this. Only define
17270 (vec_unpack<su>_<perm_hilo>_<SVE_BHSI:mode>): Turn into a
17271 define_expand, renaming the old pattern to...
17272 (aarch64_sve_<su>unpk<perm_hilo>_<SVE_BHSI:mode>): ...this.
17273 (*vec_unpacku_<perm_hilo>_<mode>_no_convert): Delete.
17274 (vec_unpacks_<perm_hilo>_<mode>): Take BYTES_BIG_ENDIAN into
17275 account when deciding which SVE instruction the optab should use.
17276 (vec_unpack<su_optab>_float_<perm_hilo>_vnx4si): Likewise.
17278 2018-03-13 Richard Sandiford <richard.sandiford@linaro.org>
17280 * config/aarch64/aarch64.md (V4_REGNUM, V8_REGNUM, V12_REGNUM)
17281 (V20_REGNUM, V24_REGNUM, V28_REGNUM, P1_REGNUM, P2_REGNUM, P3_REGNUM)
17282 (P4_REGNUM, P5_REGNUM, P6_REGNUM, P8_REGNUM, P9_REGNUM, P10_REGNUM)
17283 (P11_REGNUM, P12_REGNUM, P13_REGNUM, P14_REGNUM): New define_constants.
17284 (tlsdesc_small_<mode>): Turn a define_expand and use
17285 tlsdesc_small_sve_<mode> for SVE. Rename original define_insn to...
17286 (tlsdesc_small_advsimd_<mode>): ...this.
17287 (tlsdesc_small_sve_<mode>): New pattern.
17289 2018-03-13 Richard Sandiford <richard.sandiford@linaro.org>
17291 * config/aarch64/iterators.md (UNSPEC_SMUL_HIGHPART)
17292 (UNSPEC_UMUL_HIGHPART): New constants.
17293 (MUL_HIGHPART): New int iteraor.
17294 (su): Handle UNSPEC_SMUL_HIGHPART and UNSPEC_UMUL_HIGHPART.
17295 * config/aarch64/aarch64-sve.md (<su>mul<mode>3_highpart): New
17297 (*<su>mul<mode>3_highpart): New define_insn.
17299 2018-03-13 Eric Botcazou <ebotcazou@adacore.com>
17302 * ipa-devirt.c (odr_subtypes_equivalent_p): Do not get the ODR type of
17305 2018-03-13 Martin Liska <mliska@suse.cz>
17308 * (sem_item_optimizer::sem_item_optimizer): Initialize new
17310 (sem_item_optimizer::~sem_item_optimizer): Release it.
17311 (sem_item_optimizer::merge_classes): Register variable aliases.
17312 (sem_item_optimizer::fixup_pt_set): New function.
17313 (sem_item_optimizer::fixup_points_to_sets): Likewise.
17314 * ipa-icf.h: Declare new variables and functions.
17316 2018-03-13 Jakub Jelinek <jakub@redhat.com>
17318 PR middle-end/84834
17319 * match.pd ((A & C) != 0 ? D : 0): Use INTEGER_CST@2 instead of
17320 integer_pow2p@2 and test integer_pow2p in condition.
17321 (A < 0 ? C : 0): Similarly for @1.
17323 PR middle-end/84831
17324 * stmt.c (parse_output_constraint): If the CONSTRAINT_LEN (*p, p)
17325 characters starting at p contain '\0' character, don't look beyond
17329 * config/i386/i386.md (round<mode>2): For 387 fancy math, disable
17330 pattern if -ftrapping-math -fno-fp-int-builtin-inexact.
17333 * reg-stack.c (change_stack): Change update_end var from int to
17334 rtx_insn *, if non-NULL don't update just BB_END (current_block), but
17335 also call set_block_for_insn on the newly added insns and rescan.
17338 * config/i386/sse.md (sse2_loadhpd): Use Yv constraint rather than v
17339 on the last operand.
17342 * tree.c (stabilize_reference_1): Return save_expr (e) for
17343 STATEMENT_LIST even if it doesn't have side-effects.
17345 2018-03-12 Jonathan Wakely <jwakely@redhat.com>
17347 * doc/invoke.texi (-mclflushopt): Fix spelling of option.
17349 2018-03-12 Renlin Li <renlin.li@arm.com>
17351 * config/aarch64/aarch64.md (movhf_aarch64): Fix mode argument to
17352 aarch64_output_scalar_simd_mov_immediate.
17354 2018-03-12 Martin Sebor <msebor@redhat.com>
17356 PR tree-optimization/83456
17357 * gimple-fold.c (gimple_fold_builtin_memory_op): Avoid warning
17358 for perfectly overlapping calls to memcpy.
17359 (gimple_fold_builtin_memory_chk): Same.
17360 (gimple_fold_builtin_strcpy): Handle no-warning.
17361 (gimple_fold_builtin_stxcpy_chk): Same.
17362 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Handle no-warning.
17364 2018-03-12 Segher Boessenkool <segher@kernel.crashing.org>
17366 * config/rs6000/rs6000.c (abi_v4_pass_in_fpr): Add bool "named"
17367 parameter. Use it for SFmode.
17368 (rs6000_function_arg_advance_1): Adjust.
17369 (rs6000_function_arg): Adjust.
17370 (rs6000_gimplify_va_arg): Pass false for that new parameter.
17372 2018-03-12 Segher Boessenkool <segher@kernel.crashing.org>
17374 PR rtl-optimization/84169
17375 PR rtl-optimization/84780
17376 * combine.c (can_combine_p): Check for a 2-insn combination whether
17377 the destination register is used between the two insns, too.
17379 2018-03-12 Richard Biener <rguenther@suse.de>
17381 PR tree-optimization/84803
17382 * tree-if-conv.c (ifcvt_memrefs_wont_trap): Don't do anything
17383 for refs DR analysis didn't process.
17385 2018-03-12 Richard Biener <rguenther@suse.de>
17387 PR tree-optimization/84777
17388 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): For
17389 force-vectorize loops ignore whether we are optimizing for size.
17391 2018-03-12 Chung-Ju Wu <jasonwucj@gmail.com>
17393 * config/nds32/nds32.c (nds32_md_asm_adjust): New function.
17394 (TARGET_MD_ASM_ADJUST): Define.
17396 2018-03-12 Monk Chiang <sh.chiang04@gmail.com>
17397 Kito Cheng <kito.cheng@gmail.com>
17398 Chung-Ju Wu <jasonwucj@gmail.com>
17400 * config/nds32/nds32.c (nds32_compute_stack_frame,
17401 nds32_emit_stack_push_multiple, nds32_emit_stack_pop_multiple,
17402 nds32_emit_stack_v3push, nds32_emit_stack_v3pop,
17403 nds32_emit_adjust_frame, nds32_expand_prologue, nds32_expand_epilogue,
17404 nds32_expand_prologue_v3push, nds32_expand_epilogue_v3pop): Refine.
17405 * config/nds32/nds32.h (NDS32_FIRST_CALLEE_SAVE_GPR_REGNUM,
17406 NDS32_LAST_CALLEE_SAVE_GPR_REGNUM, NDS32_V3PUSH_AVAILABLE_P): New.
17407 * config/nds32/nds32.md (prologue, epilogue): Use macro
17408 NDS32_V3PUSH_AVAILABLE_P to do checking.
17410 2018-03-11 Jakub Jelinek <jakub@redhat.com>
17413 * dwarf2out.c (gen_enumeration_type_die): Don't guard adding
17414 DW_AT_declaration for ENUM_IS_OPAQUE on -gdwarf-4 or -gno-strict-dwarf,
17415 but on TYPE_SIZE. Don't do anything for ENUM_IS_OPAQUE if not creating
17416 a new die. Don't set TREE_ASM_WRITTEN if ENUM_IS_OPAQUE. Guard
17417 addition of most attributes on !orig_type_die or the attribute not
17418 being present already. Assert TYPE_VALUES is NULL for ENUM_IS_OPAQUE.
17420 2018-03-11 Kito Cheng <kito.cheng@gmail.com>
17421 Chung-Ju Wu <jasonwucj@gmail.com>
17423 * config/nds32/nds32.c (nds32_cpu_cpp_builtins): Modify to define
17424 __NDS32_VH__ macro.
17425 * config/nds32/nds32.opt (mvh): New option.
17427 2018-03-11 Kito Cheng <kito.cheng@gmail.com>
17428 Chung-Ju Wu <jasonwucj@gmail.com>
17430 * config/nds32/nds32-protos.h (nds32_cpu_cpp_builtins): Declare
17432 * config/nds32/nds32.c (nds32_cpu_cpp_builtins): New function.
17433 * config/nds32/nds32.h (TARGET_CPU_CPP_BUILTINS): Modify its
17436 2018-03-11 Kito Cheng <kito.cheng@gmail.com>
17437 Chung-Ju Wu <jasonwucj@gmail.com>
17439 * config/nds32/nds32-memory-manipulation.c (nds32_expand_strlen): New
17441 * config/nds32/nds32-multiple.md (strlensi): New pattern.
17442 * config/nds32/nds32-protos.h (nds32_expand_strlen): Declare function.
17444 2018-03-11 Monk Chiang <sh.chiang04@gmail.com>
17445 Kito Cheng <kito.cheng@gmail.com>
17446 Chung-Ju Wu <jasonwucj@gmail.com>
17448 * config/nds32/constants.md (unspec_element): Add UNSPEC_FFB,
17449 UNSPEC_FFMISM and UNSPEC_FLMISM.
17450 * config/nds32/nds32-intrinsic.c (bdesc_2arg): Add builtin description
17451 for ffb, ffmism and flmism.
17452 * config/nds32/nds32-intrinsic.md (unspec_ffb): Define new pattern.
17453 (unspec_ffmism): Ditto.
17454 (unspec_flmism): Ditto.
17455 (nds32_expand_builtin_impl): Check if string extension is available.
17456 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_FFB,
17457 NDS32_BUILTIN_FFMISM and NDS32_BUILTIN_FLMISM.
17459 2018-03-10 Vladimir Makarov <vmakarov@redhat.com>
17462 2018-03-09 Vladimir Makarov <vmakarov@redhat.com>
17465 * lra-assigns.c (assign_by_spills): Return a flag of reload
17466 assignment failure. Do not process the reload assignment
17467 failures. Do not spill other reload pseudos if they has the same
17469 (lra_assign): Add a return arg. Set up from the result of
17470 assign_by_spills call.
17471 (find_reload_regno_insns, lra_split_hard_reg_for): New functions.
17472 * lra-constraints.c (split_reg): Add a new arg. Use it instead of
17473 usage_insns if it is not NULL.
17474 (spill_hard_reg_in_range): New function.
17475 (split_if_necessary, inherit_in_ebb): Pass a new arg to split_reg.
17476 * lra-int.h (spill_hard_reg_in_range, lra_split_hard_reg_for): New
17477 function prototypes.
17478 (lra_assign): Change prototype.
17479 * lra.c (lra): Add code to deal with fails by splitting hard reg
17482 2018-03-10 H.J. Lu <hongjiu.lu@intel.com>
17485 * config/i386/i386.opt: Replace Enforcment with Enforcement.
17487 2018-03-10 Alexandre Oliva <aoliva@redhat.com>
17490 * dwarf2out.h (dw_val_class): Add dw_val_class_symview.
17491 (dw_val_node): Add val_symbolic_view.
17492 * dwarf2out.c (dw_line_info_table): Add symviews_since_reset.
17493 (symview_upper_bound): New.
17494 (new_line_info_table): Initialize symviews_since_reset.
17495 (dwarf2out_source_line): Count symviews_since_reset and set
17496 symview_upper_bound.
17497 (dw_val_equal_p): Handle symview.
17498 (add_AT_symview): New.
17499 (print_dw_val): Handle symview.
17500 (attr_checksum, attr_checksum_ordered): Likewise.
17501 (same_dw_val_p, size_of_die): Likewise.
17502 (value_format, output_die): Likewise.
17503 (add_high_low_attributes): Use add_AT_symview for entry_view.
17504 (dwarf2out_finish): Reset symview_upper_bound, clear
17507 2018-03-09 Peter Bergner <bergner@vnet.ibm.com>
17510 * config/rs6000/rs6000.c (rs6000_offsettable_memref_p): New prototype.
17511 Add strict argument and use it.
17512 (rs6000_split_multireg_move): Update for new strict argument.
17513 (mem_operand_gpr): Disallow all non-offsettable addresses.
17514 * config/rs6000/rs6000.md (*movdi_internal64): Use YZ constraint.
17516 2018-03-09 Jakub Jelinek <jakub@redhat.com>
17519 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Mark va_arg_tmp
17520 temporary TREE_ADDRESSABLE before gimplification of BUILT_IN_MEMCPY.
17521 * config/powerpcspe/powerpcspe.c (rs6000_gimplify_va_arg): Likewise.
17524 * tree-inline.c (copy_tree_body_r): For INDIRECT_REF of a remapped
17525 decl, use remap_type if we want to use the type.
17527 2018-03-09 Martin Sebor <msebor@redhat.com>
17529 PR tree-optimization/84526
17530 * gimple-ssa-warn-restrict.c (builtin_memref::set_base_and_offset):
17532 (builtin_access::generic_overlap): Be prepared to handle non-array
17535 2018-03-09 Alexandre Oliva <aoliva@redhat.com>
17537 PR rtl-optimization/84682
17538 * lra-constraints.c (process_address_1): Check is_address flag
17539 for address constraints.
17540 (process_alt_operands): Likewise.
17541 * lra.c (lra_set_insn_recog_data): Pass asm operand locs to
17542 preprocess_constraints.
17543 * recog.h (preprocess_constraints): Add oploc parameter.
17545 * recog.c (preprocess_constraints): Test address_operand for
17546 CT_ADDRESS constraints.
17548 2018-03-09 Vladimir Makarov <vmakarov@redhat.com>
17551 * lra-assigns.c (assign_by_spills): Return a flag of reload
17552 assignment failure. Do not process the reload assignment
17553 failures. Do not spill other reload pseudos if they has the same
17555 (lra_assign): Add a return arg. Set up from the result of
17556 assign_by_spills call.
17557 (find_reload_regno_insns, lra_split_hard_reg_for): New functions.
17558 * lra-constraints.c (split_reg): Add a new arg. Use it instead of
17559 usage_insns if it is not NULL.
17560 (spill_hard_reg_in_range): New function.
17561 (split_if_necessary, inherit_in_ebb): Pass a new arg to split_reg.
17562 * lra-int.h (spill_hard_reg_in_range, lra_split_hard_reg_for): New
17563 function prototypes.
17564 (lra_assign): Change prototype.
17565 * lra.c (lra): Add code to deal with fails by splitting hard reg
17568 2018-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17571 * common/config/arm/arm-common.c (arm_parse_arch_option_name):
17572 Accept complain bool parameter. Only emit errors if it is true.
17573 (arm_parse_cpu_option_name): Likewise.
17574 (arm_target_thumb_only): Adjust callers of the above.
17575 * config/arm/arm-protos.h (arm_parse_cpu_option_name): Adjust
17576 prototype to take a default true bool parameter.
17577 (arm_parse_arch_option_name): Likewise.
17579 2018-03-09 David Malcolm <dmalcolm@redhat.com>
17580 Francois-Xavier Coudert <fxcoudert@gcc.gnu.org>
17584 * Makefile.in (LD_VERSION_SCRIPT_OPTION, LD_SONAME_OPTION): New.
17585 * configure: Regenerate.
17586 * configure.ac ("linker --version-script option"): New.
17587 ("linker soname option"): New.
17589 2018-03-09 Richard Biener <rguenther@suse.de>
17591 PR tree-optimization/84775
17592 * tree-if-conv.c (add_bb_predicate_gimplified_stmts): Delink
17593 immediate uses of predicate stmts and mark them modified.
17596 PR tree-optimization/84178
17597 * tree-if-conv.c (combine_blocks): Move insert_gimplified_predicates
17599 (version_loop_for_if_conversion): Delay update_ssa call.
17600 (tree_if_conversion): Delay update_ssa until after predicate
17603 2018-03-09 Eric Botcazou <ebotcazou@adacore.com>
17606 * config/i386/winnt.c (i386_pe_seh_cold_init): Use small pre-allocation
17607 when the function accesses prior frames.
17609 2018-03-08 Jakub Jelinek <jakub@redhat.com>
17612 * dwarf2out.c (dw_loc_list): If list && loc_list->first->next, call
17613 gen_llsym, otherwise call maybe_gen_llsym.
17615 PR inline-asm/84742
17616 * recog.c (asm_operand_ok): Return 0 if multi-character constraint
17617 has ',' character inside of it.
17619 2018-03-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17622 * config/aarch64/aarch64.md (*compare_cstore<mode>_insn): Mark pattern
17623 as clobbering CC_REGNUM.
17625 2018-03-08 Richard Biener <rguenther@suse.de>
17627 PR middle-end/84552
17628 * tree-scalar-evolution.c: Include tree-into-ssa.h.
17629 (follow_copies_to_constant): Do not follow SSA names registered
17632 2018-03-08 Richard Biener <rguenther@suse.de>
17634 PR tree-optimization/84178
17635 * tree-if-conv.c (combine_blocks): Move insert_gimplified_predicates
17637 (version_loop_for_if_conversion): Delay update_ssa call.
17638 (tree_if_conversion): Delay update_ssa until after predicate
17641 2018-03-08 David Malcolm <dmalcolm@redhat.com>
17643 PR tree-optimization/84178
17644 * tree-if-conv.c (release_bb_predicate): Remove the
17645 the assertion that the stmts have NULL use_ops.
17646 Discard the statements, asserting that they haven't
17647 yet been added to a BB.
17649 2018-03-08 Richard Biener <rguenther@suse.de>
17651 PR tree-optimization/84746
17652 * tree-ssa-pre.c (find_leader_in_sets): Deal with SET1 being NULL.
17653 (phi_translate): Pass in destination ANTIC_OUT set.
17654 (phi_translate_1): Likewise. For a simplified result lookup
17655 a leader in ANTIC_OUT and AVAIL_OUT, not the ANTIC_IN sets.
17656 (phi_translate_set): Adjust.
17657 (do_pre_regular_insertion): Likewise.
17658 (do_pre_partial_partial_insertion): Likewise.
17660 2018-03-08 Martin Liska <mliska@suse.cz>
17662 PR gcov-profile/84735
17663 * doc/gcov.texi: Document usage of profile files.
17664 * gcov-io.h: Document changes in the format.
17666 2018-03-08 Alexandre Oliva <aoliva@redhat.com>
17670 * dwarf2out.c (struct dw_line_info_table): Update comments for
17672 (FORCE_RESET_NEXT_VIEW): New.
17673 (FORCE_RESETTING_VIEW_P): New.
17674 (RESETTING_VIEW_P): Check for -1 too.
17675 (ZERO_VIEW_P): Likewise.
17676 (new_line_info_table): Force-reset next view.
17677 (dwarf2out_begin_function): Likewise.
17678 (dwarf2out_source_line): Simplify zero_view_p initialization.
17679 Test FORCE_RESETTING_VIEW_P and RESETTING_VIEW_P instead of
17680 view directly. Omit view when omitting .loc at line 0.
17682 2018-03-08 Jakub Jelinek <jakub@redhat.com>
17684 PR tree-optimization/84740
17685 * tree-switch-conversion.c (process_switch): Call build_constructors
17686 only if info.phi_count is non-zero.
17688 PR tree-optimization/84739
17689 * tree-tailcall.c (find_tail_calls): Check call arguments against
17690 DECL_ARGUMENTS (current_function_decl) rather than
17691 DECL_ARGUMENTS (func) when checking for tail recursion.
17693 2018-03-07 Jakub Jelinek <jakub@redhat.com>
17695 * doc/contrib.texi: Add entries for Martin Liska, David Malcolm,
17696 Marek Polacek, extend Vladimir Makarov's, Jonathan Wakely's and
17697 Volker Reichelt's entry and add entries for people that perform
17698 GCC fuzzy testing and report numerous bugs.
17700 2018-03-07 Segher Boessenkool <segher@kernel.crashing.org>
17703 * config/rs6000/rs6000.c (rs6000_elf_in_small_data_p): Don't put
17704 readonly data in sdata, if that is disabled.
17705 * config/rs6000/sysv4.opt (mreadonly-in-sdata): New option.
17706 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
17707 -mreadonly-in-sdata option.
17709 2018-03-07 Martin Sebor <msebor@redhat.com>
17711 PR tree-optimization/84468
17712 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Consider successor
17713 basic block when looking for nul assignment.
17715 2018-03-07 Eric Botcazou <ebotcazou@adacore.com>
17718 * except.h (output_function_exception_table): Adjust prototype.
17719 * except.c (output_function_exception_table): Remove FNNAME parameter
17720 and add SECTION parameter. Ouput one part of the table at a time.
17721 * final.c (final_scan_insn_1) <NOTE_INSN_SWITCH_TEXT_SECTIONS>: Output
17722 the first part of the exception table and emit unwind directives.
17723 * config/i386/i386-protos.h (i386_pe_end_cold_function): Declare.
17724 (i386_pe_seh_cold_init): Likewise.
17725 * config/i386/cygming.h (ASM_DECLARE_COLD_FUNCTION_NAME): New macro.
17726 (ASM_DECLARE_COLD_FUNCTION_SIZE): Likewise.
17727 * config/i386/i386.c (x86_expand_epilogue): Fix wording in comment.
17728 (ix86_output_call_insn): Emit a nop in one more case for SEH.
17729 * config/i386/winnt.c: Include except.h.
17730 (struct seh_frame_state): Add reg_offset, after_prologue and
17731 in_cold_section fields.
17732 (i386_pe_seh_end_prologue): Set seh->after_prologue.
17733 (i386_pe_seh_cold_init): New function.
17734 (i386_pe_seh_fini): Add COLD parameter and bail out if it is not equal
17735 to seh->in_cold_section.
17736 (seh_emit_push): Record the offset of the push.
17737 (seh_emit_save): Record the offet of the save.
17738 (i386_pe_seh_unwind_emit): Deal with NOTE_INSN_SWITCH_TEXT_SECTIONS.
17739 Test seh->after_prologue to disregard the epilogue.
17740 (i386_pe_end_function): Pass FALSE to i386_pe_seh_fini.
17741 (i386_pe_end_cold_function): New function.
17743 2018-03-07 Jakub Jelinek <jakub@redhat.com>
17746 * config/aarch64/predicates.md (aarch64_simd_reg_or_zero): Use
17747 aarch64_simd_or_scalar_imm_zero rather than aarch64_simd_imm_zero.
17750 * gimple-expr.c (create_tmp_var_raw): Set DECL_NAMELESS flag
17752 * tree-pretty-print.c (dump_decl_name): For TDF_COMPARE_DEBUG,
17753 don't print names of DECL_NAMELESS DECL_IGNORED_P decls.
17755 PR middle-end/84723
17756 * multiple_target.c: Include tree-inline.h and intl.h.
17757 (expand_target_clones): Diagnose and fail if node->definition and
17758 !tree_versionable_function_p (node->decl).
17760 2018-03-06 John David Anglin <danglin@gcc.gnu.org>
17762 * config/pa/pa.h (ASM_GENERATE_INTERNAL_LABEL): Revise to use
17764 (ASM_OUTPUT_ADDR_VEC_ELT): Revise for above change.
17765 (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
17766 * config/pa/pa64-hpux.h (ASM_GENERATE_INTERNAL_LABEL): Revise as above.
17768 2018-03-06 Jakub Jelinek <jakub@redhat.com>
17771 * combine.c (try_combine): Use reg_or_subregno instead of handling
17772 just paradoxical SUBREGs and REGs.
17774 2018-03-06 Claudiu Zissulescu <claziss@synopsys.com>
17776 * config/arc/arc.c (arc_finalize_pic): Remove function.
17777 (arc_must_save_register): We use single base PIC register, remove
17778 checks to save/restore the PIC register.
17779 (arc_expand_prologue): Likewise.
17780 * config/arc/arc-protos.h (arc_set_default_type_attributes):
17782 (arc_verify_short): Likewise.
17783 (arc_attr_type): Likewise.
17784 * config/arc/arc.c (arc_set_default_type_attributes): Remove.
17785 (walk_stores): Likewise.
17786 (arc_address_cost): Make it static.
17787 (arc_verify_short): Likewise.
17788 (branch_dest): Likewise.
17789 (arc_attr_type): Likewise.
17790 * config/arc/arc.c (TARGET_ADJUST_INSN_LENGTH): Remove.
17791 (TARGET_INSN_LENGTH_PARAMETERS): Likewise.
17792 (arc_final_prescan_insn): Remove inserting the nops due to
17793 hardware hazards. It is done in reorg step.
17794 (insn_length_variant_t): Remove.
17795 (insn_length_parameters_t): Likewise.
17796 (arc_insn_length_parameters): Likewise.
17797 (arc_get_insn_variants): Likewise.
17798 * config/arc/arc.h (TARGET_UPSIZE_DBR): Remove.
17800 2018-03-06 Jakub Jelinek <jakub@redhat.com>
17802 PR inline-asm/84683
17803 * reg-stack.c (move_for_stack_reg): If any_malformed_asm, avoid
17806 PR tree-optimization/84687
17807 * omp-simd-clone.c (simd_clone_create): Clear DECL_BUILT_IN_CLASS
17809 * match.pd (pow(C,x)*expN(y) -> expN(logN(C)*x+y)): New optimization.
17811 2018-03-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17813 * config/rs6000/rs6000-builtin.def (rs6000_speculation_barrier):
17814 Rename to ppc_speculation_barrier.
17815 * config/rs6000/rs6000.c (rs6000_init_builtins): Rename builtin to
17816 __builtin_ppc_speculation_barrier.
17818 2018-03-05 Jakub Jelinek <jakub@redhat.com>
17821 * combine.c (combine_simplify_rtx): Don't try to simplify if
17822 if_then_else_cond returned non-NULL, but either true_rtx or false_rtx
17825 2018-03-05 Segher Boessenkool <segher@kernel.crashing.org>
17827 * config/rs6000/rs6000.c (rs6000_loop_align): Don't align tiny loops
17828 to 32 bytes when compiling for POWER9.
17830 2018-03-05 Jakub Jelinek <jakub@redhat.com>
17833 * config/i386/i386.c (ix86_function_ok_for_sibcall): Check for
17834 regparm >= 3 with no arg reg available also for calls with
17835 flag_force_indirect_call. Pass decl to ix86_function_regparm.
17838 * config/i386/sse.md (*<code><mode>3): Replace <mask_prefix3> with
17840 (*<plusminus_insn><mode>3): Likewise. Remove <mask_operand3> uses.
17842 2018-03-05 Peter Bergner <bergner@vnet.ibm.com>
17845 * config/rs6000/vector.md (mov<mode>): Disallow altivec memory operands.
17847 2018-03-05 Richard Biener <rguenther@suse.de>
17849 PR tree-optimization/84486
17850 * tree-ssa-pre.c (create_expression_by_pieces): Remove dead code.
17851 When inserting a __builtin_assume_aligned call set the LHS
17852 SSA name alignment info accordingly.
17854 2018-03-05 Wilco Dijkstra <wdijkstr@arm.com>
17856 PR tree-optimization/84114
17857 * config/aarch64/aarch64.c (aarch64_reassociation_width)
17858 Avoid reassociation of FLOAT_MODE addition.
17860 2018-03-05 Olga Makhotina <olga.makhotina@intel.com>
17862 * common/config/i386/i386-common.c (OPTION_MASK_ISA_PCONFIG_SET,
17863 OPTION_MASK_ISA_PCONFIG_UNSET, OPTION_MASK_ISA_WBNOINVD_SET,
17864 OPTION_MASK_ISA_WBNOINVD_UNSET): New definitions.
17865 (ix86_handle_option): Handle -mpconfig and -mwbnoinvd.
17866 * config.gcc (pconfigintrin.h, wbnoinvdintrin.h) : Add headers.
17867 * config/i386/cpuid.h (bit_PCONFIG, bit_WBNOINVD): New.
17868 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mpconfig
17870 * config/i386/i386-builtin.def (__builtin_ia32_wbnoinvd,
17871 __builtin_ia32_wbinvd): New builtins.
17872 (SPECIAL_ARGS2): New.
17873 * config/i386/i386-c.c (__WBNOINVD__, __PCONFIG__): New.
17874 (SPECIAL_ARGS2): New.
17875 * config/i386/i386.c (ix86_target_string): Add -mpconfig and -mwbnoinvd.
17876 (ix86_valid_target_attribute_inner_p): Ditto.
17877 (ix86_init_mmx_sse_builtins): Add special_args2.
17878 * config/i386/i386.h (TARGET_PCONFIG, TARGET_PCONFIG_P, TARGET_WBNOINVD,
17879 TARGET_WBNOINVD_P): New.
17880 * config/i386/i386.md (UNSPECV_WBINVD, UNSPECV_WBNOINVD): New.
17881 (define_insn "wbinvd", define_insn "wbnoinvd"): New.
17882 * config/i386/i386.opt: Add -mpconfig and -mwbnoinvd.
17883 * config/i386/immintrin.h (_wbinvd): New intrinsic.
17884 * config/i386/pconfigintrin.h: New file.
17885 * config/i386/wbnoinvdintrin.h: Ditto.
17886 * config/i386/x86intrin.h: Add headers pconfigintrin.h and
17888 * doc/invoke.texi (-mpconfig, -mwbnoinvd): New.
17890 2018-03-05 Richard Biener <rguenther@suse.de>
17892 PR tree-optimization/84670
17893 * tree-ssa-pre.c (struct bb_bitmap_sets): Add visited_with_visited_succs
17895 (BB_VISITED_WITH_VISITED_SUCCS): New define.
17896 (compute_antic): Initialize BB_VISITED_WITH_VISITED_SUCCS.
17897 (compute_antic_aux): Only assert the number of values in ANTIC_IN
17898 doesn't grow if all successors (recursively) were visited at least
17901 2018-03-05 Richard Biener <rguenther@suse.de>
17903 PR tree-optimization/84650
17904 * tree-ssa-loop-im.c (pass_lim::execute): Reset the SCEV cache
17905 if executed in the loop pipeline.
17907 2018-03-05 Sandra Loosemore <sandra@codesourcery.com>
17909 * doc/configfiles.texi (Configuration Files): Move info about
17910 conditionalizing $target-protos.h to...
17911 * doc/sourcebuild.texi (Back End): Here. Explain how $target.h
17912 differs from $target-protos.h.
17914 2018-03-05 Kito Cheng <kito.cheng@gmail.com>
17915 Chung-Ju Wu <jasonwucj@gmail.com>
17917 * config/nds32/nds32-protos.h (nds32_expand_setmem): Declare.
17918 * config/nds32/nds32-multiple.md (setmemsi): Define.
17919 * config/nds32/nds32-memory-manipulation.c
17920 (nds32_gen_dup_4_byte_to_word_value): New.
17921 (emit_setmem_word_loop): New.
17922 (emit_setmem_byte_loop): New.
17923 (nds32_expand_setmem_loop): New.
17924 (nds32_expand_setmem_loop_v3m): New.
17925 (nds32_expand_setmem_unroll): New.
17926 (nds32_expand_setmem): New.
17928 2018-03-04 Kito Cheng <kito.cheng@gmail.com>
17929 Chung-Ju Wu <jasonwucj@gmail.com>
17931 * config/nds32/nds32-memory-manipulation.c
17932 (nds32_emit_load_store): New.
17933 (nds32_emit_post_inc_load_store): New.
17934 (nds32_emit_mem_move): New.
17935 (nds32_emit_mem_move_block): New.
17936 (nds32_expand_movmemsi_loop_unknown_size): New.
17937 (nds32_expand_movmemsi_loop_known_size): New.
17938 (nds32_expand_movmemsi_loop): New.
17939 (nds32_expand_movmemsi_unroll): New.
17940 (nds32_expand_movmemqi): Rename ...
17941 (nds32_expand_movmemsi): ... to this.
17942 * config/nds32/nds32-multiple.md (movmemqi): Rename ...
17943 (movmemsi): ... to this.
17944 * config/nds32/nds32-protos.h (nds32_expand_movmemqi): Rename ...
17945 (nds32_expand_movmemsi): ... to this.
17947 2018-03-04 Kito Cheng <kito.cheng@gmail.com>
17948 Monk Chiang <sh.chiang04@gmail.com>
17949 Chung-Ju Wu <jasonwucj@gmail.com>
17951 * config/nds32/nds32-protos.h
17952 (nds32_expand_load_multiple): New arguments.
17953 (nds32_expand_store_multiple): Ditto.
17954 (nds32_valid_multiple_load_store): Rename ...
17955 (nds32_valid_multiple_load_store_p): ... to this.
17956 * config/nds32/nds32-memory-manipulation.c
17957 (nds32_expand_load_multiple): Refine implementation.
17958 (nds32_expand_store_multiple): Ditto.
17959 * config/nds32/nds32-multiple.md
17960 (load_multiple): Update nds32_expand_load_multiple interface.
17961 (store_multiple): Update nds32_expand_store_multiple interface.
17962 * config/nds32/nds32-predicates.c
17963 (nds32_valid_multiple_load_store): Rename ...
17964 (nds32_valid_multiple_load_store_p): ... to this and refine
17966 * config/nds32/predicates.md
17967 (nds32_load_multiple_and_update_address_operation): New predicate.
17968 (nds32_store_multiple_and_update_address_operation): New predicate.
17970 2018-03-04 Kito Cheng <kito.cheng@gmail.com>
17971 Chung-Ju Wu <jasonwucj@gmail.com>
17973 * config/nds32/nds32.md (type): Add load_multiple and store_multiple.
17974 (combo): New attribute.
17975 * config/nds32/nds32-multiple.md: Refine patterns with new attributes.
17977 2018-03-03 Chung-Ju Wu <jasonwucj@gmail.com>
17979 * config/nds32/nds32.opt: Change -mcmodel= default value.
17981 2018-03-03 Kito Cheng <kito.cheng@gmail.com>
17982 Monk Chiang <sh.chiang04@gmail.com>
17983 Chung-Ju Wu <jasonwucj@gmail.com>
17985 * config/nds32/constants.md (unspec_element): New enum.
17986 * config/nds32/constraints.md (Umw): New constraint.
17987 * config/nds32/nds32-intrinsic.c: Add more builtin functions.
17988 * config/nds32/nds32-intrinsic.md: Likewise.
17989 * config/nds32/nds32-md-auxiliary.c (nds32_regno_to_enable4): New.
17990 (nds32_valid_smw_lwm_base_p): New.
17991 (nds32_output_smw_single_word): New.
17992 (nds32_output_lmw_single_word): New.
17993 (nds32_expand_unaligned_load): New.
17994 (nds32_expand_unaligned_store): New.
17995 * config/nds32/nds32-protos.h (nds32_valid_smw_lwm_base_p): Declare.
17996 (nds32_output_smw_single_word): Declare.
17997 (nds32_output_lmw_single_word): Declare.
17998 (nds32_expand_unaligned_load): Declare.
17999 (nds32_expand_unaligned_store): Declare.
18000 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_UALOAD_HW,
18001 NDS32_BUILTIN_UALOAD_W, NDS32_BUILTIN_UALOAD_DW,
18002 NDS32_BUILTIN_UASTORE_HW, NDS32_BUILTIN_UASTORE_W,
18003 NDS32_BUILTIN_UASTORE_DW.
18004 * config/nds32/predicates.md (nds32_lmw_smw_base_operand): New
18007 2018-03-03 Monk Chiang <sh.chiang04@gmail.com>
18008 Kito Cheng <kito.cheng@gmail.com>
18009 Chung-Ju Wu <jasonwucj@gmail.com>
18011 * config/nds32/nds32-intrinsic.c
18012 (nds32_expand_builtin_null_ftype_reg): Delete.
18013 (nds32_expand_builtin_reg_ftype_imm): Ditto.
18014 (nds32_expand_builtin_null_ftype_reg_imm): Ditto.
18015 (nds32_read_argument): New.
18016 (nds32_legitimize_target): Ditto.
18017 (nds32_legitimize_argument): Ditto.
18018 (nds32_check_constant_argument): Ditto.
18019 (nds32_expand_unop_builtin): Ditto.
18020 (nds32_expand_unopimm_builtin): Ditto.
18021 (nds32_expand_binop_builtin): Ditto.
18022 (nds32_builtin_decl_impl): Ditto.
18023 (builtin_description): Ditto.
18024 (nds32_expand_builtin_impl): Rewrite with new infrastructure.
18025 (nds32_init_builtins_impl): Ditto.
18026 * config/nds32/nds32.c (TARGET_BUILTIN_DECL): Define.
18027 (nds32_builtin_decl): New.
18028 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_COUNT.
18029 * config/nds32/nds32-protos.h (nds32_builtin_decl_impl): Declare.
18031 2018-03-02 Jeff Law <law@redhat.com>
18033 * reorg.c (stop_search_p): Handle DEBUG_INSN.
18034 (redundant_insn, fill_simple_delay_slots): Likewise.
18035 (fill_slots_from_thread): Likewise.
18036 * resource.c (mark_referenced_resources): Likewise.
18037 (mark_set_resources, find_dead_or_set_registers): Likewise.
18039 2018-03-02 Jakub Jelinek <jakub@redhat.com>
18041 * substring-locations.h (format_warning_va): Formatting fix for
18042 ATTRIBUTE_GCC_DIAG.
18043 (format_warning_at_substring): Fix up ATTRIBUTE_GCC_DIAG second
18045 (format_warning_n_va, format_warning_at_substring_n): New prototypes.
18046 * substring-locations.c: Include intl.h.
18047 (format_warning_va): Turned into small wrapper around
18048 format_warning_n_va, renamed to ...
18049 (format_warning_n_va): ... this, add N and PLURAL_GMSGID arguments,
18050 rename GMSGID to SINGULAR_GMSGID, if SINGULAR_GMSGID != PLURAL_GMSGID,
18052 (format_warning_at_substring_n): New function.
18053 * gimple-ssa-sprintf.c: Remove GCC diagnostic ignored pragma.
18054 (fmtwarn): Add ATTRIBUTE_GCC_DIAG. Turn into a copy of
18055 format_warning_at_substring with just a shorter name instead of
18056 const function pointer.
18057 (fmtwarn_n): New function.
18058 (maybe_warn, format_directive, parse_directive): Use fmtwarn_n where
18059 appropriate, get rid of all the fmtstr temporaries, move conditionals
18060 with G_() wrapped string literals directly into fmtwarn arguments,
18061 cast dir.len to (int), formatting fixes.
18063 2018-03-02 Thomas Schwinge <thomas@codesourcery.com>
18065 * doc/invoke.texi: Remove "Cilk Plus" references.
18067 2018-03-02 Jakub Jelinek <jakub@redhat.com>
18068 Richard Biener <rguenther@suse.de>
18071 * expr.c (expand_expr_real_1) <case CALL_EXPR>: Don't emit diagnostics
18072 for error or warning attributes if CALL_FROM_THUNK_P is set.
18075 2018-03-02 Jakub Jelinek <jakub@redhat.com>
18078 * config/pa/pa.h (TARGET_CPU_CPP_BUILTINS): Predefine
18079 __SIZEOF_128__ macro if HPUX_LONG_DOUBLE_LIBRARY.
18082 * config/ia64/ia64.h (TARGET_CPU_CPP_BUILTINS): Predefine
18083 __SIZEOF_{FPREG,FLOAT{80,128}}__ macros.
18085 * predict.c (test_prediction_value_range): Use PROB_UNINITIALIZED
18086 instead of -1U in last predictors element's probability member.
18088 2018-03-02 Eric Botcazou <ebotcazou@adacore.com>
18091 * ipa-devirt.c (odr_subtypes_equivalent_p): Get the ODR type of both
18092 arguments if they are comparable.
18094 2018-03-02 Richard Sandiford <richard.sandiford@linaro.org>
18096 PR tree-optimization/84634
18097 * tree-vect-stmts.c (vectorizable_store, vectorizable_load): Replace
18098 masks and masked_loop_p with a single loop_masks, making sure it's
18099 null for bb vectorization.
18101 2018-03-02 Richard Sandiford <richard.sandiford@linaro.org>
18103 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence)
18104 (vect_analyze_data_ref_access): Use loop->safe_len rather than
18105 loop->force_vectorize to check whether there is no alias.
18107 2018-03-02 Jakub Jelinek <jakub@redhat.com>
18110 * rtl.h (prev_real_nondebug_insn, next_real_nondebug_insn): New
18112 * emit-rtl.c (next_real_insn, prev_real_insn): Fix up function
18114 (next_real_nondebug_insn, prev_real_nondebug_insn): New functions.
18115 * cfgcleanup.c (try_head_merge_bb): Use prev_real_nondebug_insn
18116 instead of a loop around prev_real_insn.
18117 * combine.c (move_deaths): Use prev_real_nondebug_insn instead of
18120 PR inline-asm/84625
18121 * config/i386/i386.c (ix86_print_operand): Use conditional
18122 output_operand_lossage instead of gcc_assert if CONST_VECTOR is not
18125 2018-03-02 Richard Biener <rguenther@suse.de>
18127 PR tree-optimization/84427
18128 * tree-ssa-pre.c (bitmap_remove_expr_from_set): Remove.
18129 (bitmap_set_subtract_values): Rewrite to handle multiple
18132 (prune_clobbered_mems): Likewise.
18133 (phi_translate): Take edge instead of pred/phiblock.
18134 (phi_translate_1): Likewise.
18135 (phi_translate_set): Likewise. Insert all translated
18136 exprs for a value into the set, keeping possibly multiple
18137 expressions per value.
18138 (compute_antic_aux): Adjust for phi_translate changes.
18139 When intersecting union the expressions and prune those
18140 not in the final value set, keeping possibly multiple
18141 expressions per value. Do not use value-insertion
18142 for unioning ANTIC_OUT U EXP_GEN - TMP_GEN but merge
18143 all expressions. Add verification that the value-sets
18144 only shrink during iteration.
18145 (compute_partial_antic_aux): Adjust for the phi_translate changes.
18146 (do_pre_regular_insertion): Likewise.
18147 (do_pre_partial_partial_insertion): Likewise.
18149 2018-03-02 Richard Biener <rguenther@suse.de>
18152 * config/darwin.c (saved_debug_info_level): New static global.
18153 (darwin_asm_lto_start): Disable debug info generation for LTO out.
18154 (darwin_asm_lto_end): Restore debug info generation settings.
18156 2018-03-01 Martin Liska <mliska@suse.cz>
18159 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
18160 volatile arguments.
18162 2018-03-01 Richard Biener <rguenther@suse.de>
18165 * dwarf2out.c (gen_variable_die): Properly handle late VLA
18166 type annotation with LTO when debug was disabled at compile-time.
18168 2018-03-01 Matthew Fortune <mfortune@gmail.com>
18170 * config/mips/mips.c (mips_final_prescan_insn): Fix incorrect
18172 (mips_final_postscan_insn): Likewise.
18174 2018-03-01 Richard Sandiford <richard.sandiford@linaro.org>
18176 PR rtl-optimization/84528
18177 * alias.c (init_alias_target): Add commentary.
18178 (init_alias_analysis): Only give HARD_FRAME_POINTER_REGNUM
18179 a unique base value if the frame pointer is not eliminated
18180 to the stack pointer.
18182 2018-03-01 Tom de Vries <tom@codesourcery.com>
18184 PR rtl-optimization/83327
18185 * lra-int.h (hard_regs_spilled_into): Declare.
18186 * lra.c (hard_regs_spilled_into): Define.
18187 (init_reg_info): Init hard_regs_spilled_into.
18188 * lra-spills.c (assign_spill_hard_regs): Update hard_regs_spilled_into.
18189 * lra-lives.c (make_hard_regno_born, make_hard_regno_dead)
18190 (process_bb_lives): Handle hard_regs_spilled_into.
18191 (lra_create_live_ranges_1): Before doing liveness propagation, clear
18192 regs in all_hard_regs_bitmap if set in hard_regs_spilled_into.
18194 2018-02-28 David Edelsohn <dje.gcc@gmail.com>
18196 * config.gcc (powerpc-ibm-aix7.1.*): New stanza.
18197 (powerpc-ibm-aix[789]*): Default to AIX 7.2.
18198 * config/rs6000/aix71.h (TARGET_DEFAULT): Revert to Power4 ISA.
18199 * config/rs6000/aix72.h: New file.
18201 2018-02-28 Jakub Jelinek <jakub@redhat.com>
18203 * gimple-ssa-warn-restrict.c (maybe_diag_overlap): Use warning_n
18204 instead of warning_at with conditional singular and plural messages
18208 * stor-layout.c (update_alignment_for_field): For
18209 targetm.ms_bitfield_layout_p (rli->t), if !is_bitfield
18210 && !DECL_PACKED (field), do the alignment update, just use
18211 only desired_align instead of MAX (type_align, desired_align)
18213 (place_field): Don't do known_align < desired_align handling
18214 early if targetm.ms_bitfield_layout_p (rli->t) and rli->prev_field
18215 is non-NULL, instead do it after rli->prev_field handling and
18216 only if not within a bitfield word. For DECL_PACKED (field)
18217 use type_align of BITS_PER_UNIT.
18219 2018-02-28 Eric Botcazou <ebotcazou@adacore.com>
18221 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Remove
18222 superfluous parentheses and trailing spaces.
18224 2018-02-28 Richard Biener <rguenther@suse.de>
18226 PR tree-optimization/84584
18227 * graphite-scop-detection.c (scop_detection::add_scop): Discard
18228 SCoPs with fake exit edge.
18230 2018-02-28 Martin Liska <mliska@suse.cz>
18233 * timevar.c (timer::print): Fix format to properly print 100%
18236 2018-02-28 Richard Biener <rguenther@suse.de>
18238 PR middle-end/84607
18239 * genmatch.c (capture_info::walk_match): Do not mark
18240 captured expressions without operands as expr_p given
18241 they act more like predicates and should be subject to
18242 "lost tail" side-effect preserving.
18244 2018-02-28 Alexandre Oliva <aoliva@redhat.com>
18246 PR rtl-optimization/81611
18247 * auto-inc-dec.c (attempt_change): Move dead note from
18248 mem_insn if it's the next use of regno
18249 (find_address): Take address use of reg holding
18250 non-incremented value. Add parm to limit search to the named
18252 (merge_in_block): Attempt to use a mem insn that is the next
18253 use of the original regno.
18255 2018-02-27 Martin Sebor <msebor@redhat.com>
18258 * doc/invoke.texi (-Wmissing-attributes): New option.
18259 * print-tree.c (print_node): Handle DECL_UNINLINABLE.
18261 2018-02-27 Martin Sebor <msebor@redhat.com>
18263 PR translation/84207
18264 * diagnostic-core.h (warning_n, error_n, inform_n): Change
18265 n argument to unsigned HOST_WIDE_INT.
18266 * diagnostic.c (warning_n, error_n, inform_n): Ditto.
18267 (diagnostic_n_impl): Ditto. Handle arguments in excess of LONG_MAX.
18268 * gimple-ssa-sprintf.c (format_directive): Simplify inform_n call.
18269 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Use warning_n.
18271 2018-02-27 Richard Biener <rguenther@suse.de>
18273 PR tree-optimization/84512
18274 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost):
18275 Do not use the estimate returned from record_stmt_cost for
18276 the scalar iteration cost but sum properly using add_stmt_cost.
18278 2018-02-27 Richard Biener <rguenther@suse.de>
18280 PR tree-optimization/84466
18281 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
18282 Adjust last change to less strictly validate use operands.
18284 2018-02-27 Martin Liska <mliska@suse.cz>
18286 PR gcov-profile/84548
18287 * gcov.c (process_file): Allow partial overlap and consider it
18288 also as group functions.
18289 (output_lines): Properly calculate range of lines for a group.
18291 2018-02-27 Martin Liska <mliska@suse.cz>
18293 * timevar.c (timer::print_row): Remove 'usr', 'sys', 'wall' and
18294 'ggc' suffixes. Change first column width.
18295 (timer::print): Fix formatting of the column.
18297 2018-02-27 Alexandre Oliva <aoliva@redhat.com>
18299 * tree-ssa-live.c (remove_unused_scope_block_p): Do not
18300 preserve inline entry blocks for the sake of debug inline
18301 entry point markers alone.
18302 (remove_unused_locals): Suggest in comments a better place to
18303 force the preservation of inline entry blocks that are
18304 otherwise unused, but do not preserve them.
18306 2018-02-26 H.J. Lu <hongjiu.lu@intel.com>
18308 * config/i386/i386.c (ix86_output_indirect_jmp): Update comments.
18310 2018-02-26 H.J. Lu <hongjiu.lu@intel.com>
18313 * config/i386/constraints.md (Bs): Replace
18314 ix86_indirect_branch_register with
18315 TARGET_INDIRECT_BRANCH_REGISTER.
18317 * config/i386/i386.md (indirect_jump): Likewise.
18318 (tablejump): Likewise.
18319 (*sibcall_memory): Likewise.
18320 (*sibcall_value_memory): Likewise.
18321 Peepholes of indirect call and jump via memory: Likewise.
18322 (*sibcall_GOT_32): Disallowed for TARGET_INDIRECT_BRANCH_REGISTER.
18323 (*sibcall_value_GOT_32): Likewise.
18324 * config/i386/predicates.md (indirect_branch_operand): Likewise.
18325 (GOT_memory_operand): Likewise.
18326 (call_insn_operand): Likewise.
18327 (sibcall_insn_operand): Likewise.
18328 (GOT32_symbol_operand): Likewise.
18329 * config/i386/i386.h (TARGET_INDIRECT_BRANCH_REGISTER): New.
18331 2018-02-26 Eric Botcazou <ebotcazou@adacore.com>
18333 PR rtl-optimization/83496
18334 * reorg.c (steal_delay_list_from_target): Change REDUNDANT array from
18335 booleans to RTXes. Call fix_reg_dead_note on every non-null element.
18336 (steal_delay_list_from_fallthrough): Call fix_reg_dead_note on a
18337 redundant insn, if any.
18338 (relax_delay_slots): Likewise.
18339 (update_reg_unused_notes): Rename REDUNDANT_INSN to OTHER_INSN.
18341 2018-02-26 Richard Sandiford <richard.sandiford@linaro.org>
18343 PR tree-optimization/83965
18344 * tree-vect-patterns.c (vect_reassociating_reduction_p): Assume
18345 that grouped statements are part of a reduction chain. Return
18346 true if the statement is not marked as a reduction itself but
18347 is part of a group.
18348 (vect_recog_dot_prod_pattern): Don't check whether the statement
18349 is part of a group here.
18350 (vect_recog_sad_pattern): Likewise.
18351 (vect_recog_widen_sum_pattern): Likewise.
18353 2018-02-26 Eric Botcazou <ebotcazou@adacore.com>
18356 * final.c (rest_of_clean_state): Also look for calls inside sequences.
18358 2018-02-26 H.J. Lu <hongjiu.lu@intel.com>
18361 * config/i386/i386-protos.h (ix86_output_indirect_jmp): Remove
18363 (ix86_output_indirect_function_return): New prototype.
18364 (ix86_split_simple_return_pop_internal): Likewise.
18365 * config/i386/i386.c (indirect_return_via_cx): New.
18366 (indirect_return_via_cx_bnd): Likewise.
18367 (indirect_thunk_name): Handle return va CX_REG.
18368 (output_indirect_thunk_function): Create alias for
18369 __x86_return_thunk_[re]cx and __x86_return_thunk_[re]cx_bnd.
18370 (ix86_output_indirect_jmp): Remove the bool argument.
18371 (ix86_output_indirect_function_return): New function.
18372 (ix86_split_simple_return_pop_internal): Likewise.
18373 * config/i386/i386.md (*indirect_jump): Don't pass false
18374 to ix86_output_indirect_jmp.
18375 (*tablejump_1): Likewise.
18376 (simple_return_pop_internal): Change it to define_insn_and_split.
18377 Call ix86_split_simple_return_pop_internal to split it for
18378 -mfunction-return=.
18379 (simple_return_indirect_internal): Call
18380 ix86_output_indirect_function_return instead of
18381 ix86_output_indirect_jmp.
18383 2018-02-26 Jakub Jelinek <jakub@redhat.com>
18386 * vec.h (vec_default_construct): For BROKEN_VALUE_INITIALIZATION use
18387 memset and value initialization afterwards.
18389 2018-02-26 Christophe Lyon <christophe.lyon@linaro.org>
18391 * Makefile.in (lto-wrapper): Use ALL_LINKERFLAGS.
18393 2018-02-26 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
18396 * common/config/aarch64/aarch64-common.c
18397 (aarch_option_optimization_table[]): Switch
18398 off fomit-frame-pointer
18400 2018-02-26 Kito Cheng <kito.cheng@gmail.com>
18401 Chung-Ju Wu <jasonwucj@gmail.com>
18403 * config/nds32/nds32-multiple.md (load_multiple): Disallow
18405 (store_multiple): Ditto.
18407 2018-02-26 Kito Cheng <kito.cheng@gmail.com>
18409 * config.gcc: Add --with-cpu support for nds32 target.
18410 * config/nds32/nds32-opts.h (nds32_cpu_type): New.
18411 * config/nds32/nds32.opt: Add -mcpu= option.
18413 2018-02-25 Segher Boessenkool <segher@kernel.crashing.org>
18415 * config/rs6000/rs6000.opt (mvrsave=no, mvrsave=yes, isel=no,
18416 isel=yes): Warn for these deprecated options.
18418 2018-02-23 David Edelsohn <dje.gcc@gmail.com>
18420 * config/rs6000/aix71.h (TARGET_DEFAULT): Change to
18421 ISA_2_5_MASKS_EMBEDDED.
18423 2018-02-23 Jakub Jelinek <jakub@redhat.com>
18425 * ipa-prop.c (ipa_vr_ggc_hash_traits::hash): Hash p->min and
18426 p->max as pointers rather than using iterative_hash_expr.
18428 2018-02-23 Carl Love <cel@us.ibm.com>
18430 * config/rs6000/rs6000-builtin.def: Change VSIGNED2 and VUNSIGNED2
18431 macro expansions from BU_VSX_2 to BU_P8V_VSX_2 and BU_VSX_OVERLOAD_2 to
18433 * config/rs6000/rs6000-c.c: Change VSX_BUILTIN_VEC_VSIGNED2 to
18434 P8V_BUILTIN_VEC_VSIGNED2. Change VSX_BUILTIN_VEC_VUNSIGNED2 to
18435 P8V_BUILTIN_VEC_VUNSIGNED2.
18437 2018-02-22 Vladimir Makarov <vmakarov@redhat.com>
18440 * lra-int.h (LRA_UNKNOWN_ALT, LRA_NON_CLOBBERED_ALT): New macros.
18441 * lra.c (lra_set_insn_recog_data, lra_update_insn_recog_data): Use
18443 * lra-constraints.c (curr_insn_transform): Set up
18444 LRA_NON_CLOBBERED_ALT for moves processed on the fast path. Use
18446 (remove_inheritance_pseudos): Use LRA_UNKNOWN_ALT.
18447 * lra-eliminations.c (spill_pseudos): Ditto.
18448 (process_insn_for_elimination): Ditto.
18449 * lra-lives.c (reg_early_clobber_p): Use the new macros.
18450 * lra-spills.c (spill_pseudos): Use LRA_UNKNOWN_ALT and
18451 LRA_NON_CLOBBERED_ALT.
18453 2018-02-22 Martin Sebor <msebor@redhat.com>
18455 PR tree-optimization/84480
18456 * gimple-fold.c (gimple_fold_builtin_strcpy): Move warnings
18457 to maybe_diag_stxncpy_trunc. Call it.
18458 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Integrate warnings
18459 from gimple_fold_builtin_strcpy. Print inlining stack.
18460 (handle_builtin_stxncpy): Print inlining stack.
18461 * tree-ssa-strlen.h (maybe_diag_stxncpy_trunc): Declare.
18463 2018-02-22 H.J. Lu <hongjiu.lu@intel.com>
18466 * config/i386/i386.c (ix86_set_indirect_branch_type): Issue an
18467 error when -mindirect-branch=thunk-extern, -fcf-protection=branch
18468 and -fcheck-pointer-bounds are used together.
18469 (indirect_thunk_prefix): New enum.
18470 (indirect_thunk_need_prefix): New function.
18471 (indirect_thunk_name): Replace need_bnd_p with need_prefix. Use
18472 "_nt" instead of "_bnd" for NOTRACK prefix.
18473 (output_indirect_thunk): Replace need_bnd_p with need_prefix.
18474 (output_indirect_thunk_function): Likewise.
18476 (ix86_code_end): Update output_indirect_thunk_function calls.
18477 (ix86_output_indirect_branch_via_reg): Replace
18478 ix86_bnd_prefixed_insn_p with indirect_thunk_need_prefix.
18479 (ix86_output_indirect_branch_via_push): Likewise.
18480 (ix86_output_function_return): Likewise.
18481 * doc/invoke.texi: Document -mindirect-branch=thunk-extern is
18482 incompatible with -fcf-protection=branch and
18483 -fcheck-pointer-bounds.
18485 2018-02-22 Steve Ellcey <sellcey@cavium.com>
18488 * config/aarch64/aarch64.c (aarch64_print_address_internal):
18489 Change gcc_assert call to output_operand_lossage.
18491 2018-02-22 Steve Ellcey <sellcey@cavium.com>
18493 * doc/extend.texi (__builtin_extend_pointer): Document builtin.
18495 2018-02-22 DJ Delorie <dj@redhat.com>
18496 Sebastian Perta <sebastian.perta@renesas.com>
18497 Oleg Endo <olegendo@gcc.gnu.org>
18499 * config/rx/rx.c (rx_rtx_costs): New function.
18500 (TARGET_RTX_COSTS): Override to use rx_rtx_costs.
18502 2018-02-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
18504 * config/arm/t-multilib: Map Armv8-R to Armv7 multilibs.
18506 2018-02-22 Martin Liska <mliska@suse.cz>
18509 * common/config/arm/arm-common.c (arm_print_hint_for_cpu_option):
18510 Add "native" as a possible value.
18512 2018-02-22 Martin Liska <mliska@suse.cz>
18515 * config/i386/i386.c (ix86_option_override_internal):
18516 Add "native" as a possible value for -march and -mtune.
18518 2018-02-22 Jakub Jelinek <jakub@redhat.com>
18521 * stor-layout.c (finalize_type_size): Propagate TYPE_EMPTY_P flag
18522 to all type variants.
18524 PR tree-optimization/84503
18525 * gimple-ssa-store-merging.c (merged_store_group::merge_into): Compute
18526 width as info->bitpos + info->bitsize - start.
18527 (merged_store_group::merge_overlapping): Simplify width computation.
18528 (check_no_overlap): New function.
18529 (imm_store_chain_info::try_coalesce_bswap): Compute expected
18530 start + width and last_order of the group, fail if check_no_overlap
18532 (imm_store_chain_info::coalesce_immediate_stores): Don't merge info
18533 to group if check_no_overlap fails.
18535 2018-02-21 Segher Boessenkool <segher@kernel.crashing.org>
18537 * config/rs6000/altivec.md: Delete contraint arguments to
18538 define_expand, define_split, and define_peephole2, and in
18539 define_insn_and_split if always unused.
18540 * config/rs6000/darwin.md: Ditto.
18541 * config/rs6000/dfp.md: Ditto.
18542 * config/rs6000/rs6000.md: Ditto.
18543 * config/rs6000/sync.md: Ditto.
18544 * config/rs6000/vector.md: Ditto.
18545 * config/rs6000/vsx.md: Ditto.
18547 2018-02-21 Segher Boessenkool <segher@kernel.crashing.org>
18549 * config/rs6000/altivec.md: Write output control strings as braced
18550 blocks instead of double-quoted strings.
18551 * config/rs6000/darwin.md: Ditto.
18552 * config/rs6000/rs6000.md: Ditto.
18553 * config/rs6000/vector.md: Ditto.
18554 * config/rs6000/vsx.md: Ditto.
18556 2018-02-21 Jason Merrill <jason@redhat.com>
18558 PR c++/84314 - ICE with templates and fastcall attribute.
18559 * attribs.c (build_type_attribute_qual_variant): Remove assert.
18561 2018-02-21 Jan Hubicka <hubicka@ucw.cz>
18563 * ipa-cp.c (determine_versionability): Fix comment typos.
18565 2018-02-21 Jan Hubicka <hubicka@ucw.cz>
18568 * ipa-cp.c (determine_versionability): Do not version functions caling
18571 2018-02-21 Martin Liska <mliska@suse.cz>
18574 * config/aarch64/aarch64.c (aarch64_print_hint_for_core_or_arch):
18575 Add "native" as a possible value.
18576 * config/aarch64/aarch64.h (HAVE_LOCAL_CPU_DETECT): Define
18577 the macro when native cpu detection is available.
18579 2018-02-21 Martin Liska <mliska@suse.cz>
18582 * common/config/arm/arm-common.c (arm_print_hint_for_arch_option):
18583 Add "native" as a possible value.
18584 * config/arm/arm.h (HAVE_LOCAL_CPU_DETECT): Define the macro
18585 when native cpu detection is available.
18587 2018-02-21 Jakub Jelinek <jakub@redhat.com>
18588 Martin Sebor <msebor@redhat.com>
18590 PR tree-optimization/84478
18591 * gimple-fold.h (get_range_strlen): Add a bool argument defaulted to
18593 * gimple-fold.c (get_range_strlen): Make minlen const and assume it
18594 can't be NULL. Change FUZZY from bool to int, for 1 add PHI/COND_EXPR
18595 support which is conservatively correct, for 2 only stay conservative
18596 for maxlen. Formatting and comment capitalization fixes. Add STRICT
18597 argument to the 2 argument get_range_strlen, adjust 6 arg
18598 get_range_strlen caller and clear minmaxlen[0] and [1] if it returned
18600 (get_maxval_strlen): Adjust 6 arg get_range_strlen caller.
18601 (gimple_fold_builtin_strlen): Pass true as last argument to
18604 2018-02-20 Martin Sebor <msebor@redhat.com>
18606 PR middle-end/84095
18607 * gimple-ssa-warn-restrict.c (builtin_memref::extend_offset_range): New.
18608 (builtin_memref::set_base_and_offset): Same. Handle inner references.
18609 (builtin_memref::builtin_memref): Factor out parts into
18610 set_base_and_offset and call it.
18612 2018-02-20 Richard Sandiford <richard.sandiford@linaro.org>
18614 PR middle-end/84406
18615 * optabs-query.c (find_widening_optab_handler_and_mode): If from_mode
18616 is a scalar_int_mode, assert that to_mode is a scalar_int_mode with
18617 greater precision. If to_mode is a MODE_PARTIAL_INT, stop the
18618 search at the associated MODE_INT.
18620 2018-02-20 Jeff Law <law@redhat.com>
18622 PR middle-end/82123
18623 PR tree-optimization/81592
18624 PR middle-end/79257
18625 * gimple-ssa-sprintf.c (format_integer): Query EVRP range analyzer
18626 for range data rather than using global data.
18627 * gimple-ssa-sprintf.c (get_int_range): Query EVRP range analyzer for
18628 range data rather than using global data.
18629 * gimple-ssa-sprintf.c (get_int_range): Accept vr_values parameter
18630 pass it to children as needed.
18631 (struct directive::fmtresult): Similarly.
18632 (struct directive::set_width): Similarly.
18633 (struct directive::set_precision): Similarly.
18634 (format_integer, format_directive, parse_directive): Similarly.
18635 (format_none): Accept unnamed vr_values parameter.
18636 (format_percent, format_floating, format_character): Similarly.
18637 (format_string, format_plain): Similarly.
18638 * gimple-ssa-sprintf.c (sprintf_dom_walker::handle_gimple_call): Query
18639 the EVRP range analyzer for range data rather than using global data.
18640 * gimple-ssa-sprintf.c: Include alloc-pool.h, vr-values.h and
18641 gimple-ssa-evrp-analyze.h
18642 (class sprintf_dom_walker): Add after_dom_children member function.
18643 Add evrp_range_analyzer member.
18644 (sprintf_dom_walker::before_dom_children): Call into the EVRP
18645 range analyzer as needed.
18646 (sprintf_dom_walker::after_dom_children): New member function.
18647 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::enter): Do nothing
18649 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
18650 (evrp_range_analyzer::pop_to_marker): Likewise.
18652 2018-02-20 Richard Sandiford <richard.sandiford@linaro.org>
18654 PR tree-optimization/84419
18655 * internal-fn.c (expand_call_mem_ref): Create a TARGET_MEM_REF
18656 with the required type if its current type is compatible but
18659 2018-02-20 Jakub Jelinek <jakub@redhat.com>
18661 PR middle-end/82004
18662 * match.pd (pow(C,x) -> exp(log(C)*x)): Delay all folding until
18663 after vectorization.
18665 2018-02-20 Martin Liska <mliska@suse.cz>
18668 * config/aarch64/aarch64.c (aarch64_print_hint_for_core_or_arch): Print
18669 possible values if we don't have a hint.
18671 2018-02-20 Martin Liska <mliska@suse.cz>
18675 * final.c (shorten_branches): Build align_tab array with one
18677 * opts.c (finish_options): Add alignment option limit check.
18678 (MAX_CODE_ALIGN): Likewise.
18679 (MAX_CODE_ALIGN_VALUE): Likewise.
18680 * doc/invoke.texi: Document maximum allowed option value for
18681 all -falign-* options.
18683 2018-02-19 Jakub Jelinek <jakub@redhat.com>
18686 * reg-notes.def (REG_CALL_ARG_LOCATION): New reg note.
18687 * insn-notes.def (NOTE_INSN_CALL_ARG_LOCATION): Remove.
18688 * var-tracking.c (emit_note_insn_var_location): Remove all references
18689 to NOTE_INSN_CALL_ARG_LOCATION.
18690 (emit_notes_in_bb): Emit arguments as REG_CALL_ARG_LOCATION note on
18691 the CALL_INSN rather than separate NOTE_INSN_CALL_ARG_LOCATION note.
18692 Use copy_rtx_if_shared.
18693 * dwarf2out.c (gen_subprogram_die): Use XEXP with 0 instead of
18694 NOTE_VAR_LOCATION on ca_loc->call_arg_loc_note.
18695 (dwarf2out_var_location): Remove handling of
18696 NOTE_INSN_CALL_ARG_LOCATION, instead handle REG_CALL_ARG_LOCATION note
18698 * final.c (final_scan_insn): Remove all references to
18699 NOTE_INSN_CALL_ARG_LOCATION.
18700 (rest_of_clean_state): Likewise. Remove REG_CALL_ARG_LOCATION notes
18701 before dumping final insns.
18702 * except.c (emit_note_eh_region_end): Remove all references to
18703 NOTE_INSN_CALL_ARG_LOCATION.
18704 * config/alpha/alpha.c (alpha_pad_function_end): Likewise.
18705 * config/c6x/c6x.c (c6x_gen_bundles): Likewise.
18706 * config/arc/arc.c (hwloop_optimize): Likewise.
18707 * config/arm/arm.c (create_fix_barrier): Likewise.
18708 * config/s390/s390.c (s390_chunkify_start): Likewise.
18709 * config/sh/sh.c (find_barrier): Likewise.
18710 * config/i386/i386.c (rest_of_insert_endbranch,
18711 ix86_seh_fixup_eh_fallthru): Likewise.
18712 * config/xtensa/xtensa.c (hwloop_optimize): Likewise.
18713 * config/iq2000/iq2000.c (final_prescan_insn): Likewise.
18714 * config/frv/frv.c (frv_function_prologue): Likewise.
18715 * emit-rtl.c (try_split): Likewise. Copy over REG_CALL_ARG_LOCATION
18717 (note_outside_basic_block_p): Remove all references to
18718 NOTE_INSN_CALL_ARG_LOCATION.
18719 * gengtype.c (adjust_field_rtx_def): Likewise.
18720 * print-rtl.c (rtx_writer::print_rtx_operand_code_0, print_insn):
18722 * jump.c (cleanup_barriers, delete_related_insns): Likewise.
18723 * cfgrtl.c (force_nonfallthru_and_redirect): Likewise.
18726 * builtins.c (builtin_mathfn_code): Don't check if CALL_EXPR_FN (t)
18729 PR tree-optimization/84452
18730 * tree-vect-patterns.c (vect_recog_pow_pattern): Don't call
18731 expand_simd_clones if targetm.simd_clone.compute_vecsize_and_simdlen
18734 2018-02-19 Martin Liska <mliska@suse.cz>
18737 * passes.def: Put pass_sancov_O0 before pass_lower_switch with -O0.
18739 2018-02-19 Martin Liska <mliska@suse.cz>
18740 Richard Sandiford <richard.sandiford@linaro.org>
18742 PR tree-optimization/82491
18743 * gimple-fold.c (get_base_constructor): Make earlier bail out
18746 2018-02-19 Carl Love <cel@us.ibm.com>
18748 * config/rs6000/rs6000-builtin.def: Change NEG macro expansions from
18749 BU_ALTIVEC_A to BU_P8V_AV_1 and BU_ALTIVEC_OVERLOAD_1 to
18751 * config/rs6000/rs6000-c.c: Change ALTIVEC_BUILTIN_VEC_NEG to
18752 P8V_BUILTIN_VEC_NEG.
18754 2018-02-19 Sebastian Perta <sebastian.perta@renesas.com>
18756 * config/rl78/rl78.md (movdf): New define expand.
18758 2018-02-19 Martin Liska <mliska@suse.cz>
18761 * doc/invoke.texi: Fix typo.
18762 * params.def (PARAM_MAX_LOOP_HEADER_INSNS): Likewise.
18764 2018-02-18 Segher Boessenkool <segher@kernel.crashing.org>
18766 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
18767 handle rs6000_single_float and rs6000_double_float specially for
18770 2018-02-16 Jeff Law <law@redhat.com>
18772 * config/rx/rx.c (add_pop_cfi_notes): New function.;
18773 (pop_regs): Use it.
18775 2018-02-16 Jakub Jelinek <jakub@redhat.com>
18778 * ipa-inline.c (inline_small_functions): Fix a typo.
18780 2018-02-16 Nathan Sidwell <nathan@acm.org>
18782 * doc/extend.texi (Backwards Compatibility): Americanize 'behaviour'.
18784 2018-02-16 Carl Love <cel@us.ibm.com>
18786 * config/rs6000/rs6000-builtin.def: Add BU_P8V_VSX_2 macro definition.
18787 Change FLOAT2 expansions from BU_VSX_2 to BU_P8V_VSX_2 and
18788 from BU_VSX_OVERLOAD_2 to BU_P8V_OVERLOAD_2.
18789 * config/rs6000/rs6000-c.c: Changed macro VSX_BUILTIN_VEC_FLOAT2
18790 expansion to P8V_BUILTIN_VEC_FLOAT2.
18792 2018-02-16 Vladimir Makarov <vmakarov@redhat.com>
18794 PR rtl-optimization/70023
18795 * lra-constraints.c (inherit_in_ebb): Take hard reg mode of
18796 src_regno into account.
18798 2018-02-16 Carl Love <cel@us.ibm.com>
18800 * config/rs6000/altivec.h: Remove vec_vextract4b and vec_vinsert4b.
18801 * config/rs6000/rs6000-builtin.def: Remove macro expansion for
18802 VEXTRACT4B, VINSERT4B, VINSERT4B_DI and VEXTRACT4B.
18803 * config/rs6000/rs6000.c: Remove case statements for
18804 P9V_BUILTIN_VEXTRACT4B, P9V_BUILTIN_VEC_VEXTRACT4B,
18805 P9V_BUILTIN_VINSERT4B, P9V_BUILTIN_VINSERT4B_DI,
18806 and P9V_BUILTIN_VEC_VINSERT4B.
18807 * config/rs6000/rs6000-c.c (altivec_expand_builtin): Remove entries for
18808 P9V_BUILTIN_VEC_VEXTRACT4B and P9V_BUILTIN_VEC_VINSERT4B.
18809 * config/rs6000/vsx.md:
18810 * doc/extend.texi: Remove vec_vextract4b, non ABI definitions for
18813 2018-02-16 Carl Love <cel@us.ibm.com>
18815 * config/rs6000/altivec.h: Add builtin names vec_extract4b
18817 * config/rs6000/rs6000-builtin.def: Add INSERT4B and EXTRACT4B
18819 * config/rs6000/rs6000-c.c: Add the definitions for
18820 P9V_BUILTIN_VEC_EXTRACT4B and P9V_BUILTIN_VEC_INSERT4B.
18821 * config/rs6000/rs6000.c (altivec_expand_builtin): Add
18822 P9V_BUILTIN_EXTRACT4B and P9V_BUILTIN_INSERT4B case statements.
18823 * config/rs6000/vsx.md: Add define_insn extract4b. Add define_expand
18824 definition for insert4b and define insn *insert3b_internal.
18825 * doc/extend.texi: Add documentation for vec_extract4b.
18827 2018-02-16 Nathan Sidwell <nathan@acm.org>
18829 * doc/extend.texi (Backwards Compatibility): Mention friend
18830 injection. Note for-scope is deprecated.
18831 * doc/invoke.texi (-ffriend-injection): Deprecate.
18833 2018-02-16 Segher Boessenkool <segher@kernel.crashing.org>
18835 * combine.c (try_combine): When adjusting LOG_LINKS for the destination
18836 that moved to I2, also allow destinations that are a paradoxical
18837 subreg (instead of a normal reg).
18839 2018-02-16 Oleg Endo <olegendo@gcc.gnu.org>
18842 * config/rx/rx.c (rx_fuse_in_memory_bitop): Convert shift operand
18845 2018-02-16 Richard Biener <rguenther@suse.de>
18847 PR tree-optimization/84037
18848 PR tree-optimization/84016
18850 * config/i386/i386.c (ix86_builtin_vectorization_cost):
18851 Adjust vec_construct for the fact we need additional higher latency
18852 128bit inserts for AVX256 and AVX512 vector builds.
18853 (ix86_add_stmt_cost): Scale vector construction cost for
18856 2018-02-16 Richard Biener <rguenther@suse.de>
18858 PR tree-optimization/84417
18859 * tree-ssa.c (non_rewritable_mem_ref_base): Properly constrain
18860 the MEM_REF offset when conversion to BIT_FIELD_REF is desired.
18861 (non_rewritable_lvalue_p): Likewise, use poly-ints.
18863 2018-02-16 Martin Liska <mliska@suse.cz>
18866 * internal-fn.def (ASAN_CHECK): Set proper flags.
18867 (ASAN_MARK): Likewise.
18869 2018-02-16 Julia Koval <julia.koval@intel.com>
18871 * config/i386/i386.c (ix86_option_override_internal): Remove PTA_CLWB
18872 from PTA_CANNONLAKE.
18874 2018-02-16 Jakub Jelinek <jakub@redhat.com>
18877 * config/aarch64/cortex-a57-fma-steering.c (fma_forest::merge_forest):
18878 Use ++iter rather than iter++ for std::list iterators.
18879 (func_fma_steering::dfs): Likewise. Don't delete nodes right away,
18880 defer deleting them until all nodes in the forest are processed. Do
18881 free even leaf nodes. Change to_process into auto_vec.
18884 * system.h (BROKEN_VALUE_INITIALIZATION): Define for GCC < 4.3.
18885 * vec.h (vec_default_construct): Use memset instead of placement new
18886 if BROKEN_VALUE_INITIALIZATION is defined.
18887 * hash-table.h (hash_table<Descriptor, Allocator>::empty_slow): Use
18888 memset instead of value initialization if BROKEN_VALUE_INITIALIZATION
18891 PR rtl-optimization/83723
18892 * lra-int.h (lra_substitute_pseudo): Add DEBUG_P argument.
18893 * lra.c (lra_substitute_pseudo): Likewise. If true, use
18894 gen_rtx_raw_SUBREG instead of gen_rtx_SUBREG. Pass DEBUG_P to
18896 (lra_substitute_pseudo_within_insn): Adjust lra_substitute_pseudo
18898 * lra-constraints.c (inherit_reload_reg, split_reg): Likewise.
18900 2018-02-16 Eric Botcazou <ebotcazou@adacore.com>
18902 PR rtl-optimization/81443
18903 * rtlanal.c (num_sign_bit_copies1) <SUBREG>: Do not propagate results
18904 from inner REGs to paradoxical SUBREGs.
18906 2018-02-16 Richard Biener <rguenther@suse.de>
18908 PR tree-optimization/84399
18909 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
18910 For operands we can analyze at their definition make sure we can
18911 analyze them at each use as well.
18913 2018-02-16 Richard Biener <rguenther@suse.de>
18915 PR tree-optimization/84190
18916 * tree-ssa.c (non_rewritable_mem_ref_base): Do not touch
18917 volatile accesses if the decl isn't volatile.
18919 2018-02-15 Jason Merrill <jason@redhat.com>
18921 PR c++/84314 - ICE with templates and fastcall attribute.
18922 * attribs.c (build_type_attribute_qual_variant): Don't clobber
18923 TYPE_CANONICAL on an existing type.
18925 2018-02-15 Jakub Jelinek <jakub@redhat.com>
18927 PR tree-optimization/84383
18928 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Don't look at
18929 dstoff nor call operand_equal_p if dstbase is NULL.
18931 PR tree-optimization/84334
18932 * match.pd ((A +- CST1) +- CST2 -> A + CST3): If A is
18933 also a CONSTANT_CLASS_P, punt.
18935 2018-02-14 Jim Wilson <jimw@sifive.com>
18937 * config/riscv/riscv.c (riscv_first_stack_step): Move locals after
18938 first SMALL_OPERAND check. New local min_second_step. Move assert
18939 to where locals are set. Add TARGET_RVC support.
18940 * config/riscv/riscv.h (C_SxSP_BITS, SWSP_REACH, SDSP_REACH): New.
18942 2018-02-14 Indu Bhagat <indu.bhagat@oracle.com>
18944 * doc/invoke.texi: Correct -Wformat-overflow code sample.
18946 2018-02-14 Martin Sebor <msebor@redhat.com>
18948 PR tree-optimization/83698
18949 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): For
18950 arrays constrain the offset range to their bounds.
18951 (builtin_access::strcat_overlap): Adjust the bounds of overlap offset.
18952 (builtin_access::overlap): Avoid setting the size of overlap if it's
18954 (maybe_diag_overlap): Also consider arrays when deciding what values
18955 of offsets to include in diagnostics.
18957 2018-02-14 Martin Sebor <msebor@redhat.com>
18960 * attribs.c (diag_attr_exclusions): Consider the exclusion(s)
18961 that correspond to the kind of a declaration.
18963 2018-02-14 John David Anglin <danglin@gcc.gnu.org>
18966 * config/pa/pa.md: Load address of PIC label using the linkage table
18967 if the label is nonlocal.
18969 2018-02-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
18971 * config/rs6000/rs6000.c (rs6000_option_override_internal): Issue
18972 warning message if user requests -maltivec=be.
18973 * doc/invoke.texi: Document deprecation of -maltivec=be.
18975 2018-02-14 Will Schmidt <will_schmidt@vnet.ibm.com>
18978 * config/rs6000/rs6000-c.c: Update definitions for
18979 ALTIVEC_BUILTIN_VEC_SLD, ALTIVEC_BUILTIN_VEC_SLDW,
18980 VEC_XXSLDWI and ALTIVEC_BUILTIN_VEC_XXPERMDI builtins.
18982 2018-02-14 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
18985 * config/i386/cetintrin.h: Remove _rdssp[d|q] and
18986 add _get_ssp intrinsics. Remove argument from
18987 __builtin_ia32_rdssp[d|q].
18988 * config/i386/i386-builtin-types.def: Add UINT_FTYPE_VOID.
18989 * config/i386/i386-builtin.def: Remove argument from
18990 __builtin_ia32_rdssp[d|q].
18991 * config/i386/i386.c: Use UINT_FTYPE_VOID. Use
18992 ix86_expand_special_args_builtin for _rdssp[d|q].
18993 * config/i386/i386.md: Remove argument from rdssp[si|di] insn.
18994 Clear register before usage.
18995 * doc/extend.texi: Remove argument from __builtin_ia32_rdssp[d|q].
18996 Add documentation for new _get_ssp and _inc_ssp intrinsics.
18998 2018-02-14 Richard Sandiford <richard.sandiford@linaro.org>
19000 PR tree-optimization/84357
19001 * tree-data-ref.c (object_address_invariant_in_loop_p): Check
19002 operand 1 of an ARRAY_REF too.
19004 2018-02-14 Oleg Endo <olegendo@gcc.gnu.org>
19007 * config/rx/rx-protos.h (rx_reg_dead_or_unused_after_insn,
19008 rx_copy_reg_dead_or_unused_notes, rx_fuse_in_memory_bitop): New
19010 (set_of_reg): New struct.
19011 (rx_find_set_of_reg, rx_find_use_of_reg): New functions.
19012 * config/rx/rx.c (rx_reg_dead_or_unused_after_insn,
19013 rx_copy_reg_dead_or_unused_notes, rx_fuse_in_memory_bitop): New
19015 * config/rx/rx.md (andsi3, iorsi3, xorsi3): Convert to insn_and_split.
19016 Split into bitclr, bitset, bitinvert patterns if appropriate.
19017 (*bitset, *bitinvert, *bitclr): Convert to named insn_and_split and
19018 use rx_fuse_in_memory_bitop.
19019 (*bitset_in_memory, *bitinvert_in_memory, *bitclr_in_memory): Convert
19020 to named insn, correct maximum insn length.
19022 2018-02-14 Jozef Lawrynowicz <jozefl.gcc@gmail.com>
19025 * machmode.def: Define a complex mode for PARTIAL_INT.
19026 * genmodes.c (complex_class): Return MODE_COMPLEX_INT for
19028 * doc/rtl.texi: Document CSPImode.
19029 * config/msp430/msp430.c (msp430_hard_regno_nregs): Add CPSImode
19031 (msp430_hard_regno_nregs_with_padding): Likewise.
19033 2018-02-13 Peter Bergner <bergner@vnet.ibm.com>
19036 * config/rs6000/rs6000.c (mem_operand_gpr): Disallow altivec addresses.
19038 2018-02-13 Segher Boessenkool <segher@kernel.crashing.org>
19040 PR rtl-optimization/84169
19041 * combine.c (try_combine): New variable split_i2i3. Set it to true if
19042 we generated a parallel as new i3 and we split that to new i2 and i3
19043 instructions. Handle split_i2i3 similar to swap_i2i3: scan the
19044 LOG_LINKs of i3 to see which of those need to link to i2 now. Link
19045 those to i2, not i1. Partially rewrite this scan code.
19047 2018-02-13 Jakub Jelinek <jakub@redhat.com>
19050 * stor-layout.c (place_field): For variable length fields, adjust
19051 offset_align afterwards not just based on the field's alignment,
19052 but also on the size.
19054 PR middle-end/84309
19055 * match.pd (pow(C,x) -> exp(log(C)*x)): Use exp2s and log2s instead
19056 of exps and logs in the use_exp2 case.
19058 2018-02-13 Jeff Law <law@redhat.com>
19060 * config/rl/rl78.c (rl78_attribute_table): Fix terminator and
19061 entry for "vector".
19063 * config/rl78/rl78.c (rl78_handle_func_attribute): Mark
19066 2018-02-13 Alexandre Oliva <aoliva@redhat.com>
19070 * common.opt (gas-loc-support, gas-locview-support): New.
19071 (ginline-points, ginternal-reset-location-views): New.
19072 * doc/invoke.texi: Document them. Use @itemx where intended.
19073 (gvariable-location-views): Adjust.
19074 * target.def (reset_location_view): New.
19075 * doc/tm.texi.in (DWARF2_ASM_VIEW_DEBUG_INFO): New.
19076 (TARGET_RESET_LOCATION_VIEW): New.
19077 * doc/tm.texi: Rebuilt.
19078 * dwarf2out.c (dwarf2out_default_as_loc_support): New.
19079 (dwarf2out_default_as_locview_support): New.
19080 (output_asm_line_debug_info): Use option variables.
19081 (dwarf2out_maybe_output_loclist_view_pair): Likewise.
19082 (output_loc_list): Likewise.
19083 (add_high_low_attributes): Check option variables.
19084 Don't output entry view attribute in strict mode.
19085 (gen_inlined_subroutine_die): Check option variables.
19086 (dwarf2out_inline_entry): Likewise.
19087 (init_sections_and_labels): Likewise.
19088 (dwarf2out_early_finish): Likewise.
19089 (maybe_reset_location_view): New, from...
19090 (dwarf2out_var_location): ... here. Call it.
19091 * debug.h (dwarf2out_default_as_loc_support): Declare.
19092 (dwarf2out_default_as_locview_support): Declare.
19093 * hooks.c (hook_int_rtx_insn_0): New.
19094 * hooks.h (hook_int_rtx_insn_0): Declare.
19095 * toplev.c (process_options): Take -gas-loc-support and
19096 -gas-locview-support from dwarf2out. Enable
19097 -gvariable-location-views by default only with locview
19098 assembler support. Enable -ginternal-reset-location-views by
19099 default only if the target defines the corresponding hook.
19100 Enable -ginline-points by default if location views are
19101 enabled; force it disabled if statement frontiers are
19103 * tree-inline.c (expand_call_inline): Check option variables.
19104 * tree-ssa-live.c (remove_unused_scope_block_p): Likewise.
19106 2018-02-13 Richard Sandiford <richard.sandiford@linaro.org>
19108 PR tree-optimization/84321
19109 * tree-vrp.c (intersect_range_with_nonzero_bits): Fix VR_ANTI_RANGE
19110 handling. Also check whether the anti-range contains any values
19111 that satisfy the mask; switch to a VR_RANGE if not.
19113 2018-02-13 Paolo Bonzini <bonzini@gnu.org>
19116 * internal-fn.def (ASAN_CHECK, ASAN_MARK): Revert changes to fnspec.
19118 2018-02-13 Martin Jambor <mjambor@suse.cz>
19121 * ipa-param-manipulation.c (ipa_modify_call_arguments): Use location
19122 of call statements, also set location of a load to a temporary.
19124 2018-02-13 Sebastian Perta <sebastian.perta@renesas.com>
19126 * config/rl78/rl78.c (add_vector_labels): New function.
19127 * config/rl78/rl78.c (rl78_handle_vector_attribute): New function.
19128 * config/rl78/rl78.c (rl78_start_function): Call add_vector_labels.
19129 * config/rl78/rl78.c (rl78_handle_func_attribute): Removed the assert
19130 which checks that no arguments are passed.
19131 * config/rl78/rl78.c (rl78_attribute_table): Add "vector" attribute.
19132 * doc/extend.texi: Documentation for the new attribute.
19134 2018-02-13 Andreas Schwab <schwab@suse.de>
19136 * config/riscv/linux.h (CPP_SPEC): Define.
19138 2018-02-13 Jakub Jelinek <jakub@redhat.com>
19141 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Pass
19142 OPTION_MASK_ISA_AES | OPTION_MASK_ISA_SSE2 instead of
19143 OPTION_MASK_ISA_AES as first argument to def_builtin_const
19144 for AES builtins. Pass OPTION_MASK_ISA_PCLMUL | OPTION_MASK_ISA_SSE2
19145 instead of OPTION_MASK_ISA_PCLMUL as first argument to
19146 def_builtin_const for __builtin_ia32_pclmulqdq128 builtin.
19147 * config/i386/wmmintrin.h: If __SSE2__ is not defined, enable it
19148 temporarily for AES and PCLMUL builtins.
19150 PR tree-optimization/84339
19151 * gimple-fold.c (get_range_strlen): Set *FLEXP to true when handling
19152 ARRAY_REF where first operand is array_at_struct_end_p COMPONENT_REF.
19155 PR middle-end/84309
19156 * match.pd (pow(C,x) -> exp(log(C)*x)): Optimize instead into
19157 exp2(log2(C)*x) if C is a power of 2 and c99 runtime is available.
19158 * generic-match-head.c (canonicalize_math_after_vectorization_p): New
19160 * gimple-match-head.c (canonicalize_math_after_vectorization_p): New
19162 * omp-simd-clone.h: New file.
19163 * omp-simd-clone.c: Include omp-simd-clone.h.
19164 (expand_simd_clones): No longer static.
19165 * tree-vect-patterns.c: Include fold-const-call.h, attribs.h,
19166 cgraph.h and omp-simd-clone.h.
19167 (vect_recog_pow_pattern): Optimize pow(C,x) to exp(log(C)*x).
19168 (vect_recog_widen_shift_pattern): Formatting fix.
19169 (vect_pattern_recog_1): Don't check optab for calls.
19172 * config/i386/sse.md (<avx512>_vpermi2var<mode>3_mask): Force
19173 operands[2] into a REG before using gen_lowpart on it.
19175 2018-02-12 Jeff Law <law@redhat.com>
19178 * config/sh/sh.c (find_barrier): Consider a sibling call
19181 * cse.c (try_back_substitute_reg): Move any REG_ARGS_SIZE note when
19182 successfully back substituting a reg.
19184 2018-02-12 Richard Biener <rguenther@suse.de>
19186 PR tree-optimization/84037
19187 * tree-vect-slp.c (vect_analyze_slp_cost): Add visited
19188 parameter, move visited init to caller.
19189 (vect_slp_analyze_operations): Separate cost from validity
19190 check, initialize visited once for all instances.
19191 (vect_schedule_slp): Analyze map to CSE vectorized nodes once
19193 * tree-vect-stmts.c (vect_model_simple_cost): Make early
19195 (vect_model_promotion_demotion_cost): Likewise.
19196 (vectorizable_bswap): Guard cost modeling with !slp_node
19197 instead of !PURE_SLP_STMT to avoid double-counting on hybrid
19199 (vectorizable_call): Likewise.
19200 (vectorizable_conversion): Likewise.
19201 (vectorizable_assignment): Likewise.
19202 (vectorizable_shift): Likewise.
19203 (vectorizable_operation): Likewise.
19204 (vectorizable_store): Likewise.
19205 (vectorizable_load): Likewise.
19206 (vectorizable_condition): Likewise.
19207 (vectorizable_comparison): Likewise.
19209 2018-02-12 Paolo Bonzini <bonzini@gnu.org>
19212 * internal-fn.def (ASAN_CHECK): Fix fnspec to account for return value.
19213 (ASAN_MARK): Fix fnspec to account for return value, change pointer
19214 argument from 'R' to 'W' so that the pointed-to datum is clobbered.
19216 2018-02-08 Jan Hubicka <hubicka@ucw.cz>
19218 PR middle-end/83665
19219 * params.def (inline-min-speedup): Increase from 8 to 15.
19220 (max-inline-insns-auto): Decrease from 40 to 30.
19221 * ipa-split.c (consider_split): Add some buffer for function to
19222 be considered inlining candidate.
19223 * invoke.texi (max-inline-insns-auto, inline-min-speedup): UPdate
19226 2018-02-12 Richard Biener <rguenther@suse.de>
19228 PR tree-optimization/84037
19229 * tree-vect-slp.c (vect_build_slp_tree_2): Try swapping the
19230 matched stmts if we cannot swap the non-matched ones.
19232 2018-02-12 Olga Makhotina <olga.makhotina@intel.com>
19234 * config/i386/avx512fintrin.h (_mm_mask_scalef_round_sd,
19235 _mm_maskz_scalef_round_sd, _mm_mask_scalef_round_ss,
19236 _mm_maskz_scalef_round_ss): New intrinsics.
19237 (__builtin_ia32_scalefsd_round, __builtin_ia32_scalefss_round): Fix.
19238 * config/i386/i386-builtin.def (__builtin_ia32_scalefsd_round,
19239 __builtin_ia32_scalefss_round): Remove.
19240 (__builtin_ia32_scalefsd_mask_round,
19241 __builtin_ia32_scalefss_mask_round): New intrinsics.
19242 * config/i386/sse.md (vmscalef<mode><round_name>): Renamed to ...
19243 (vmscalef<mode><mask_scalar_name><round_scalar_name>): ... this.
19244 ((match_operand:VF_128 2 "<round_nimm_predicate>"
19245 "<round_constraint>")): Changed to ...
19246 ((match_operand:VF_128 2 "<round_scalar_nimm_predicate>"
19247 "<round_scalar_constraint>")): ... this.
19248 ("vscalef<ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|
19249 %0, %1, %2<round_op3>}"): Changed to ...
19250 ("vscalef<ssescalarmodesuffix>\t{<round_scalar_mask_op3>%2, %1,
19251 %0<mask_scalar_operand3>|%0<mask_scalar_operand3>, %1,
19252 %2<round_scalar_mask_op3>}"): ... this.
19253 * config/i386/subst.md (round_scalar_nimm_predicate): New.
19255 2018-02-12 Olga Makhotina <olga.makhotina@intel.com>
19257 * config/i386/avx512fintrin.h (_mm_mask_sqrt_round_sd)
19258 (_mm_maskz_sqrt_round_sd, _mm_mask_sqrt_round_ss)
19259 (_mm_maskz_sqrt_round_ss): New intrinsics.
19260 (__builtin_ia32_sqrtsd_round, __builtin_ia32_sqrtss_round): Remove.
19261 (__builtin_ia32_sqrtsd_mask_round)
19262 (__builtin_ia32_sqrtss_mask_round): New builtins.
19263 * config/i386/i386-builtin.def (__builtin_ia32_sqrtsd_round)
19264 (__builtin_ia32_sqrtss_round): Remove.
19265 (__builtin_ia32_sqrtsd_mask_round)
19266 (__builtin_ia32_sqrtss_mask_round): New builtins.
19267 * config/i386/sse.md (vmsqrt<mode>2<round_name>): Renamed to ...
19268 (vmsqrt<mode>2<mask_scalar_name><round_scalar_name>): ... this.
19269 ((match_operand:VF_128 1 "vector_operand"
19270 "xBm,<round_constraint>")): Changed to ...
19271 ((match_operand:VF_128 1 "vector_operand"
19272 "xBm,<round_scalar_constraint>")): ... this.
19273 (vsqrt<ssescalarmodesuffix>\t{<round_op3>%1, %2, %0|
19274 %0, %2, %<iptr>1<round_op3>}): Changed to ...
19275 (vsqrt<ssescalarmodesuffix>\t{<round_scalar_mask_op3>%1, %2,
19276 %0<mask_scalar_operand3>|%0<mask_scalar_operand3>, %2,
19277 %<iptr>1<round_scalar_mask_op3>}): ... this.
19278 ((set_attr "prefix" "<round_prefix>")): Changed to ...
19279 ((set_attr "prefix" "<round_scalar_prefix>")): ... this.
19281 2018-02-11 Steven Munroe <munroesj@gcc.gnu.org>
19284 * config/rs6000/mmintrin.h (_mm_cmpeq_pi32 [_ARCH_PWR9]):
19285 Cast vec_cmpeq result to correct type.
19286 * config/rs6000/mmintrin.h (_mm_cmpgt_pi32 [_ARCH_PWR9]):
19287 Cast vec_cmpgt result to correct type.
19289 2018-02-11 Alexandre Oliva <aoliva@redhat.com>
19291 * final.c (final_scan_insn_1): Renamed from...
19292 (final_scan_insn): ... this. New wrapper, to recover
19293 seen from the outermost call in recursive ones.
19294 * config/sparc/sparc.c (output_return): Drop seen from call.
19295 (output_sibcall): Likewise.
19296 * config/visium/visium.c (output_branch): Likewise.
19298 2018-02-10 John David Anglin <danglin@gcc.gnu.org>
19300 * config/pa/pa.c (hppa_profile_hook): Mark SYMBOL_REF for _mcount as
19303 2018-02-10 Alan Modra <amodra@gmail.com>
19306 * config/rs6000/rs6000.md (split_stack_return): Remove (use ..).
19307 Specify LR as an input.
19309 2018-02-10 Jakub Jelinek <jakub@redhat.com>
19312 * omp-low.c (maybe_remove_omp_member_access_dummy_vars,
19313 remove_member_access_dummy_vars): New functions.
19314 (lower_omp_for, lower_omp_taskreg, lower_omp_target,
19315 lower_omp_1, execute_lower_omp): Use them.
19317 PR rtl-optimization/84308
19318 * shrink-wrap.c (spread_components): Release todo vector.
19320 2018-02-09 Vladimir Makarov <vmakarov@redhat.com>
19322 PR rtl-optimization/57193
19323 * ira-color.c (struct allocno_color_data): Add member
19324 conflict_allocno_hard_prefs.
19325 (update_conflict_allocno_hard_prefs): New.
19326 (bucket_allocno_compare_func): Add a preference based on
19327 conflict_allocno_hard_prefs.
19328 (push_allocno_to_stack): Update conflict_allocno_hard_prefs.
19329 (color_allocnos): Remove a dead code. Initiate
19330 conflict_allocno_hard_prefs. Call update_costs_from_prefs.
19332 2018-02-09 Jakub Jelinek <jakub@redhat.com>
19335 * config/rs6000/vsx.md (p9_xxbrq_v16qi): Change input operand
19336 constraint from =wa to wa. Avoid a subreg on the output operand,
19337 instead use a pseudo and subreg it in a move.
19338 (p9_xxbrd_<mode>): Changed to ...
19339 (p9_xxbrd_v2di): ... this insn, without VSX_D iterator.
19340 (p9_xxbrd_v2df): New expander.
19341 (p9_xxbrw_<mode>): Changed to ...
19342 (p9_xxbrw_v4si): ... this insn, without VSX_W iterator.
19343 (p9_xxbrw_v4sf): New expander.
19345 2018-02-09 Sebastian Perta <sebastian.perta@renesas.com>
19347 * config/rx/rx.md (movsicc): Update expander to be matched by GCC.
19349 2018-02-09 Peter Bergner <bergner@vnet.ibm.com>
19352 * config/rs6000/vsx.md (vsx_mul_v2di): Handle generating a 64-bit
19353 multiply in 32-bit mode.
19354 (vsx_div_v2di): Handle generating a 64-bit signed divide in 32-bit mode.
19355 (vsx_udiv_v2di): Handle generating a 64-bit unsigned divide in 32-bit
19358 2018-02-09 Sebastian Perta <sebastian.perta@renesas.com>
19360 * config/rx/constraints.md (CALL_OP_SYMBOL_REF): Added new constraint
19361 to allow or block "symbol_ref" depending on the value of TARGET_JSR.
19362 * config/rx/rx.md (call_internal): Use CALL_OP_SYMBOL_REF.
19363 * config/rx/rx.md (call_value_internal): Use CALL_OP_SYMBOL_REF.
19365 2018-02-09 Pierre-Marie de Rodat <derodat@adacore.com>
19368 * dwarf2out.c (is_trivial_indirect_ref): New function.
19369 (dwarf2out_late_global_decl): Do not generate a location
19370 attribute for variables that have a non-trivial DECL_VALUE_EXPR
19371 and that are not defined in the current unit.
19373 2018-02-09 Eric Botcazou <ebotcazou@adacore.com>
19375 * optabs.c (prepare_cmp_insn): Try harder to emit a direct comparison
19376 instead of a libcall for UNORDERED.
19378 2018-02-09 Tamar Christina <tamar.christina@arm.com>
19381 * config/arm/arm-c.c (arm_cpu_builtins): Un-define __ARM_FEATURE_LDREX,
19382 __ARM_ARCH_PROFILE, __ARM_ARCH_ISA_THUMB, __ARM_FP and __ARM_NEON_FP.
19384 2018-02-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19387 * config/s390/s390.c (s390_set_current_function): Invoke
19388 s390_indirect_branch_settings also if fndecl didn't change.
19390 2018-02-09 Alexandre Oliva <aoliva@redhat.com>
19392 * config/rs6000/rs6000.md (blockage): Set length to zero.
19394 2018-02-09 Eric Botcazou <ebotcazou@adacore.com>
19396 * expr.c (optimize_bitfield_assignment_op): Remove obsolete assertion.
19398 2018-02-09 Jakub Jelinek <jakub@redhat.com>
19401 * gcc.c (STATIC_LIBASAN_LIBS, STATIC_LIBTSAN_LIBS,
19402 STATIC_LIBLSAN_LIBS, STATIC_LIBUBSAN_LIBS): Handle -static like
19406 * var-tracking.c (vt_add_function_parameter): Punt for non-onepart
19407 PARALLEL incoming that failed vt_get_decl_and_offset check.
19409 PR middle-end/84237
19410 * output.h (bss_initializer_p): Add NAMED argument, defaulted to false.
19411 * varasm.c (bss_initializer_p): Add NAMED argument, if true, ignore
19413 (get_variable_section): For decls in named .bss* sections pass true as
19414 second argument to bss_initializer_p.
19416 2018-02-09 Marek Polacek <polacek@redhat.com>
19417 Jakub Jelinek <jakub@redhat.com>
19420 * fold-const.c (fold_indirect_ref_1): Use VECTOR_TYPE_P macro.
19421 Formatting fixes. Verify first that tree_fits_poly_int64_p (op01).
19422 Sync some changes from cxx_fold_indirect_ref.
19424 2018-02-09 Alexandre Oliva <aoliva@redhat.com>
19426 * cfgexpand.c (expand_gimple_basic_block): Handle inline entry
19428 * dwarf2out.c (dwarf2_debug_hooks): Enable inline_entry hook.
19429 (BLOCK_INLINE_ENTRY_LABEL): New.
19430 (dwarf2out_var_location): Disregard inline entry markers.
19431 (inline_entry_data): New struct.
19432 (inline_entry_data_hasher): New hashtable type.
19433 (inline_entry_data_hasher::hash): New.
19434 (inline_entry_data_hasher::equal): New.
19435 (inline_entry_data_table): New variable.
19436 (add_high_low_attributes): Add DW_AT_entry_pc and
19437 DW_AT_GNU_entry_view attributes if a pending entry is found
19438 in inline_entry_data_table. Add old entry_pc attribute only
19439 if debug nonbinding markers are disabled.
19440 (gen_inlined_subroutine_die): Set BLOCK_DIE if nonbinding
19441 markers are enabled.
19442 (block_within_block_p, dwarf2out_inline_entry): New.
19443 (dwarf2out_finish): Check that no entries remained in
19444 inline_entry_data_table.
19445 * final.c (reemit_insn_block_notes): Handle inline entry notes.
19446 (final_scan_insn, notice_source_line): Likewise.
19447 (rest_of_clean_state): Skip inline entry markers.
19448 * gimple-pretty-print.c (dump_gimple_debug): Handle inline entry
19450 * gimple.c (gimple_build_debug_inline_entry): New.
19451 * gimple.h (enum gimple_debug_subcode): Add
19452 GIMPLE_DEBUG_INLINE_ENTRY.
19453 (gimple_build_debug_inline_entry): Declare.
19454 (gimple_debug_inline_entry_p): New.
19455 (gimple_debug_nonbind_marker_p): Adjust.
19456 * insn-notes.def (INLINE_ENTRY): New.
19457 * print-rtl.c (rtx_writer::print_rtx_operand_code_0): Handle
19458 inline entry marker notes.
19459 (print_insn): Likewise.
19460 * rtl.h (NOTE_MARKER_P): Add INLINE_ENTRY support.
19461 (INSN_DEBUG_MARKER_KIND): Likewise.
19462 (GEN_RTX_DEBUG_MARKER_INLINE_ENTRY_PAT): New.
19463 * tree-inline.c (expand_call_inline): Build and insert
19464 debug_inline_entry stmt.
19465 * tree-ssa-live.c (remove_unused_scope_block_p): Preserve
19466 inline entry blocks early, if nonbind markers are enabled.
19467 (dump_scope_block): Dump fragment info.
19468 * var-tracking.c (reemit_marker_as_note): Handle inline entry note.
19469 * doc/gimple.texi (gimple_debug_inline_entry_p): New.
19470 (gimple_build_debug_inline_entry): New.
19471 * doc/invoke.texi (gstatement-frontiers, gno-statement-frontiers):
19472 Enable/disable inline entry points too.
19473 * doc/rtl.texi (NOTE_INSN_INLINE_ENTRY): New.
19474 (DEBUG_INSN): Describe inline entry markers.
19476 * common.opt (gvariable-location-views): New.
19477 (gvariable-location-views=incompat5): New.
19478 * config.in: Rebuilt.
19479 * configure: Rebuilt.
19480 * configure.ac: Test assembler for view support.
19481 * dwarf2asm.c (dw2_asm_output_symname_uleb128): New.
19482 * dwarf2asm.h (dw2_asm_output_symname_uleb128): Declare.
19483 * dwarf2out.c (var_loc_view): New typedef.
19484 (struct dw_loc_list_struct): Add vl_symbol, vbegin, vend.
19485 (dwarf2out_locviews_in_attribute): New.
19486 (dwarf2out_locviews_in_loclist): New.
19487 (dw_val_equal_p): Compare val_view_list of dw_val_class_view_lists.
19488 (enum dw_line_info_opcode): Add LI_adv_address.
19489 (struct dw_line_info_table): Add view.
19490 (RESET_NEXT_VIEW, RESETTING_VIEW_P): New macros.
19491 (DWARF2_ASM_VIEW_DEBUG_INFO): Define default.
19492 (zero_view_p): New variable.
19493 (ZERO_VIEW_P): New macro.
19494 (output_asm_line_debug_info): New.
19495 (struct var_loc_node): Add view.
19496 (add_AT_view_list, AT_loc_list): New.
19497 (add_var_loc_to_decl): Add view param. Test it against last.
19498 (new_loc_list): Add view params. Record them.
19499 (AT_loc_list_ptr): Handle loc and view lists.
19500 (view_list_to_loc_list_val_node): New.
19501 (print_dw_val): Handle dw_val_class_view_list.
19502 (size_of_die): Likewise.
19503 (value_format): Likewise.
19504 (loc_list_has_views): New.
19505 (gen_llsym): Set vl_symbol too.
19506 (maybe_gen_llsym, skip_loc_list_entry): New.
19507 (dwarf2out_maybe_output_loclist_view_pair): New.
19508 (output_loc_list): Output view list or entries too.
19509 (output_view_list_offset): New.
19510 (output_die): Handle dw_val_class_view_list.
19511 (output_dwarf_version): New.
19512 (output_compilation_unit_header): Use it.
19513 (output_skeleton_debug_sections): Likewise.
19514 (output_rnglists, output_line_info): Likewise.
19515 (output_pubnames, output_aranges): Update version comments.
19516 (output_one_line_info_table): Output view numbers in asm comments.
19517 (dw_loc_list): Determine current endview, pass it to new_loc_list.
19518 Call maybe_gen_llsym.
19519 (loc_list_from_tree_1): Adjust.
19520 (add_AT_location_description): Create view list attribute if
19521 needed, check it's absent otherwise.
19522 (convert_cfa_to_fb_loc_list): Adjust.
19523 (maybe_emit_file): Call output_asm_line_debug_info for test.
19524 (dwarf2out_var_location): Reset views as needed. Precompute
19525 add_var_loc_to_decl args. Call get_attr_min_length only if we have the
19526 attribute. Set view.
19527 (new_line_info_table): Reset next view.
19528 (set_cur_line_info_table): Call output_asm_line_debug_info for test.
19529 (dwarf2out_source_line): Likewise. Output view resets and labels to
19530 the assembler, or select appropriate line info opcodes.
19531 (prune_unused_types_walk_attribs): Handle dw_val_class_view_list.
19532 (optimize_string_length): Catch it. Adjust.
19533 (resolve_addr): Copy vl_symbol along with ll_symbol. Handle
19534 dw_val_class_view_list, and remove it if no longer needed.
19535 (hash_loc_list): Hash view numbers.
19536 (loc_list_hasher::equal): Compare them.
19537 (optimize_location_lists): Check whether a view list symbol is
19538 needed, and whether the locview attribute is present, and
19539 whether they match. Remove the locview attribute if no longer
19541 (index_location_lists): Call skip_loc_list_entry for test.
19542 (dwarf2out_finish): Call output_asm_line_debug_info for test.
19543 Use output_dwarf_version.
19544 * dwarf2out.h (enum dw_val_class): Add dw_val_class_view_list.
19545 (struct dw_val_node): Add val_view_list.
19546 * final.c (SEEN_NEXT_VIEW): New.
19547 (set_next_view_needed): New.
19548 (clear_next_view_needed): New.
19549 (maybe_output_next_view): New.
19550 (final_start_function): Rename to...
19551 (final_start_function_1): ... this. Take pointer to FIRST,
19552 add SEEN parameter. Emit param bindings in the initial view.
19553 (final_start_function): Reintroduce SEEN-less interface.
19554 (final): Rename to...
19555 (final_1): ... this. Take SEEN parameter. Output final pending
19556 next view at the end.
19557 (final): Reintroduce seen-less interface.
19558 (final_scan_insn): Output pending next view before switching
19559 sections or ending a block. Mark the next view as needed when
19560 outputting variable locations. Notify debug backend of section
19561 changes, and of location view changes.
19562 (rest_of_handle_final): Adjust.
19563 * toplev.c (process_options): Autodetect value for debug variable
19564 location views option. Warn on incompat5 without -gdwarf-5.
19565 * doc/invoke.texi (gvariable-location-views): New.
19566 (gvariable-location-views=incompat5): New.
19567 (gno-variable-location-views): New.
19569 2018-02-08 David Malcolm <dmalcolm@redhat.com>
19571 PR tree-optimization/84136
19572 * tree-cfg.c (find_taken_edge_computed_goto): Remove assertion
19573 that the result of find_edge is non-NULL.
19575 2018-02-08 Sergey Shalnov <sergey.shalnov@intel.com>
19578 * config/i386/x86-tune-costs.h (skylake_cost): Fix cost of
19579 storing integer register in SImode. Fix cost of 256 and 512
19580 byte aligned SSE register store.
19582 2018-02-08 Sergey Shalnov <sergey.shalnov@intel.com>
19584 * config/i386/i386.c (ix86_multiplication_cost): Fix
19585 multiplication cost for TARGET_AVX512DQ.
19587 2018-02-08 Marek Polacek <polacek@redhat.com>
19589 PR tree-optimization/84238
19590 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Verify the result of
19593 2018-02-08 Richard Sandiford <richard.sandiford@linaro.org>
19595 PR tree-optimization/84265
19596 * tree-vect-stmts.c (vectorizable_store): Don't treat
19597 VMAT_CONTIGUOUS accesses as grouped.
19598 (vectorizable_load): Likewise.
19600 2018-02-08 Richard Sandiford <richard.sandiford@linaro.org>
19602 PR tree-optimization/81635
19603 * wide-int.h (wi::round_down_for_mask, wi::round_up_for_mask): Declare.
19604 * wide-int.cc (wi::round_down_for_mask, wi::round_up_for_mask)
19605 (test_round_for_mask): New functions.
19606 (wide_int_cc_tests): Call test_round_for_mask.
19607 * tree-vrp.h (intersect_range_with_nonzero_bits): Declare.
19608 * tree-vrp.c (intersect_range_with_nonzero_bits): New function.
19609 * tree-data-ref.c (split_constant_offset_1): Use it to refine the
19610 range returned by get_range_info.
19612 2018-02-08 Jan Hubicka <hubicka@ucw.cz>
19615 * cgraph.h (symtab_node::output_to_lto_symbol_table_p): Declare
19616 * symtab.c: Include builtins.h
19617 (symtab_node::output_to_lto_symbol_table_p): Move here
19618 from lto-streamer-out.c:output_symbol_p.
19619 * lto-streamer-out.c (write_symbol): Turn early exit to assert.
19620 (output_symbol_p): Move all logic to symtab.c
19621 (produce_symtab): Update.
19623 2018-02-08 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19625 * config/s390/s390-opts.h (enum indirect_branch): Define.
19626 * config/s390/s390-protos.h (s390_return_addr_from_memory)
19627 (s390_indirect_branch_via_thunk)
19628 (s390_indirect_branch_via_inline_thunk): Add function prototypes.
19629 (enum s390_indirect_branch_type): Define.
19630 * config/s390/s390.c (struct s390_frame_layout, struct
19631 machine_function): Remove.
19632 (indirect_branch_prez10thunk_mask, indirect_branch_z10thunk_mask)
19633 (indirect_branch_table_label_no, indirect_branch_table_name):
19635 (INDIRECT_BRANCH_NUM_OPTIONS): Define macro.
19636 (enum s390_indirect_branch_option): Define.
19637 (s390_return_addr_from_memory): New function.
19638 (s390_handle_string_attribute): New function.
19639 (s390_attribute_table): Add new attribute handler.
19640 (s390_execute_label): Handle UNSPEC_EXECUTE_JUMP patterns.
19641 (s390_indirect_branch_via_thunk): New function.
19642 (s390_indirect_branch_via_inline_thunk): New function.
19643 (s390_function_ok_for_sibcall): When jumping via thunk disallow
19644 sibling call optimization for non z10 compiles.
19645 (s390_emit_call): Force indirect branch target to be a single
19646 register. Add r1 clobber for non-z10 compiles.
19647 (s390_emit_epilogue): Emit return jump via return_use expander.
19648 (s390_reorg): Handle JUMP_INSNs as execute targets.
19649 (s390_option_override_internal): Perform validity checks for the
19650 new command line options.
19651 (s390_indirect_branch_attrvalue): New function.
19652 (s390_indirect_branch_settings): New function.
19653 (s390_set_current_function): Invoke s390_indirect_branch_settings.
19654 (s390_output_indirect_thunk_function): New function.
19655 (s390_code_end): Implement target hook.
19656 (s390_case_values_threshold): Implement target hook.
19657 (TARGET_ASM_CODE_END, TARGET_CASE_VALUES_THRESHOLD): Define target
19659 * config/s390/s390.h (struct s390_frame_layout)
19660 (struct machine_function): Move here from s390.c.
19661 (TARGET_INDIRECT_BRANCH_NOBP_RET)
19662 (TARGET_INDIRECT_BRANCH_NOBP_JUMP)
19663 (TARGET_INDIRECT_BRANCH_NOBP_JUMP_THUNK)
19664 (TARGET_INDIRECT_BRANCH_NOBP_JUMP_INLINE_THUNK)
19665 (TARGET_INDIRECT_BRANCH_NOBP_CALL)
19666 (TARGET_DEFAULT_INDIRECT_BRANCH_TABLE)
19667 (TARGET_INDIRECT_BRANCH_THUNK_NAME_EXRL)
19668 (TARGET_INDIRECT_BRANCH_THUNK_NAME_EX)
19669 (TARGET_INDIRECT_BRANCH_TABLE): Define macros.
19670 * config/s390/s390.md (UNSPEC_EXECUTE_JUMP)
19671 (INDIRECT_BRANCH_THUNK_REGNUM): Define constants.
19672 (mnemonic attribute): Add values which aren't recognized
19674 ("*cjump_long", "*icjump_long", "*basr", "*basr_r"): Disable
19675 pattern for branch conversion. Fix mnemonic attribute.
19676 ("*c<code>", "*sibcall_br", "*sibcall_value_br", "*return"): Emit
19677 indirect branch via thunk if requested.
19678 ("indirect_jump", "<code>"): Expand patterns for branch conversion.
19679 ("*indirect_jump"): Disable for branch conversion using out of
19681 ("indirect_jump_via_thunk<mode>_z10")
19682 ("indirect_jump_via_thunk<mode>")
19683 ("indirect_jump_via_inlinethunk<mode>_z10")
19684 ("indirect_jump_via_inlinethunk<mode>", "*casesi_jump")
19685 ("casesi_jump_via_thunk<mode>_z10", "casesi_jump_via_thunk<mode>")
19686 ("casesi_jump_via_inlinethunk<mode>_z10")
19687 ("casesi_jump_via_inlinethunk<mode>", "*basr_via_thunk<mode>_z10")
19688 ("*basr_via_thunk<mode>", "*basr_r_via_thunk_z10")
19689 ("*basr_r_via_thunk", "return<mode>_prez10"): New pattern.
19690 ("*indirect2_jump"): Disable for branch conversion.
19691 ("casesi_jump"): Turn into expander and expand patterns for branch
19693 ("return_use"): New expander.
19694 ("*return"): Emit return via thunk and rename it to ...
19695 ("*return<mode>"): ... this one.
19696 * config/s390/s390.opt: Add new options and and enum for the
19699 2018-02-08 Richard Sandiford <richard.sandiford@linaro.org>
19701 * lra-constraints.c (match_reload): Unconditionally use
19702 gen_lowpart_SUBREG, rather than selecting between that
19703 and equivalent gen_rtx_SUBREG code.
19705 2018-02-08 Richard Biener <rguenther@suse.de>
19707 PR tree-optimization/84233
19708 * tree-ssa-phiprop.c (propagate_with_phi): Use separate
19709 changed flag instead of boguously re-using phi_inserted.
19711 2018-02-08 Martin Jambor <mjambor@suse.cz>
19713 * hsa-gen.c (get_symbol_for_decl): Set program allocation for
19714 static local variables.
19716 2018-02-08 Richard Biener <rguenther@suse.de>
19718 PR tree-optimization/84278
19719 * tree-vect-stmts.c (vectorizable_store): When looking for
19720 smaller vector types to perform grouped strided loads/stores
19721 make sure the mode is supported by the target.
19722 (vectorizable_load): Likewise.
19724 2018-02-08 Wilco Dijkstra <wdijkstr@arm.com>
19726 * config/aarch64/aarch64.c (aarch64_components_for_bb):
19727 Increase LDP/STP opportunities by adding adjacent callee-saves.
19729 2018-02-08 Wilco Dijkstra <wdijkstr@arm.com>
19731 PR rtl-optimization/84068
19732 PR rtl-optimization/83459
19733 * haifa-sched.c (rank_for_schedule): Fix SCHED_PRESSURE_MODEL sorting.
19735 2018-02-08 Aldy Hernandez <aldyh@redhat.com>
19737 PR tree-optimization/84224
19738 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Remove assert.
19739 * calls.c (gimple_alloca_call_p): Only return TRUE when we have
19740 non-zero arguments.
19742 2018-02-07 Iain Sandoe <iain@codesourcery.com>
19745 * config/rs6000/altivec.md (*restore_world): Remove LR use.
19746 * config/rs6000/predicates.md (restore_world_operation): Adjust op
19747 count, remove one USE.
19749 2018-02-07 Michael Meissner <meissner@linux.vnet.ibm.com>
19751 * doc/install.texi (Configuration): Document the
19752 --with-long-double-format={ibm,ieee} PowerPC configuration
19756 * config/rs6000/rs6000.md (fix_trunc<SFDF:mode><QHI:mode>2):
19757 Convert from define_expand to be define_insn_and_split. Rework
19758 float/double/_Float128 conversions to QI/HI/SImode to work with
19759 both ISA 2.07 (power8) or ISA 3.0 (power9). Fix regression where
19760 conversions to QI/HImode types did a store and then a load to
19761 truncate the value. For conversions to VSX registers, don't split
19762 the insn, instead emit the code directly. Use the code iterator
19763 any_fix to combine signed and unsigned conversions.
19764 (fix<uns>_trunc<SFDF:mode>si2_p8): Likewise.
19765 (fixuns_trunc<SFDF:mode><QHI:mode>2): Likewise.
19766 (fix_trunc<IEEE128:mode><QHI:mode>2): Likewise.
19767 (fix<uns>_trunc<SFDF:mode><QHI:mode>2): Likewise.
19768 (fix_<mode>di2_hw): Likewise.
19769 (fixuns_<mode>di2_hw): Likewise.
19770 (fix_<mode>si2_hw): Likewise.
19771 (fixuns_<mode>si2_hw): Likewise.
19772 (fix<uns>_<IEEE128:mode><SDI:mode>2_hw): Likewise.
19773 (fix<uns>_trunc<IEEE128:mode><QHI:mode>2): Likewise.
19774 (fctiw<u>z_<mode>_smallint): Rename fctiw<u>z_<mode>_smallint to
19775 fix<uns>_trunc<SFDF:mode>si2_p8.
19776 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Delete, no longer used.
19777 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
19778 (fix<uns>_<mode>_mem): Likewise.
19779 (fctiw<u>z_<mode>_mem): Likewise.
19780 (fix<uns>_<mode>_mem): Likewise.
19781 (fix<uns>_trunc<SFDF:mode><QHSI:mode>2_mem): On ISA 3.0, prevent
19782 the register allocator from doing a direct move to the GPRs to do
19783 a store, and instead use the ISA 3.0 store byte/half-word from
19784 vector register instruction. For IEEE 128-bit floating point,
19785 also optimize stores of 32-bit ints.
19786 (fix<uns>_trunc<IEEE128:mode><QHSI:mode>2_mem): Likewise.
19788 2018-02-07 Alan Hayward <alan.hayward@arm.com>
19790 * genextract.c (push_pathstr_operand): New function to support
19792 (walk_rtx): Call push_pathstr_operand.
19793 (print_path): Support [a-zA-Z].
19795 2018-02-07 Richard Biener <rguenther@suse.de>
19797 PR tree-optimization/84037
19798 * tree-vectorizer.h (struct _loop_vec_info): Add ivexpr_map member.
19799 (cse_and_gimplify_to_preheader): Declare.
19800 (vect_get_place_in_interleaving_chain): Likewise.
19801 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
19803 (_loop_vec_info::~_loop_vec_info): Delete it.
19804 (cse_and_gimplify_to_preheader): New function.
19805 * tree-vect-slp.c (vect_get_place_in_interleaving_chain): Export.
19806 * tree-vect-stmts.c (vectorizable_store): CSE base and steps.
19807 (vectorizable_load): Likewise. For grouped stores always base
19808 the IV on the first element.
19809 * tree-vect-loop-manip.c (vect_loop_versioning): Unshare versioning
19810 condition before gimplifying.
19812 2018-02-07 Jakub Jelinek <jakub@redhat.com>
19814 * tree-eh.c (operation_could_trap_helper_p): Ignore honor_trapv for
19815 *DIV_EXPR and *MOD_EXPR.
19817 2018-02-07 H.J. Lu <hongjiu.lu@intel.com>
19820 * config/i386/i386.c (ix86_option_override_internal): Mask out
19821 the CF_SET bit when checking -fcf-protection.
19823 2018-02-07 Tom de Vries <tom@codesourcery.com>
19826 * omp-expand.c (expand_oacc_collapse_init): Ensure diff_type is large
19829 2018-02-07 Richard Biener <rguenther@suse.de>
19831 PR tree-optimization/84204
19832 * tree-chrec.c (chrec_fold_plus_1): Remove size limiting in
19835 PR tree-optimization/84205
19836 * graphite-isl-ast-to-gimple.c (binary_op_to_tree): Also
19837 special-case isl_ast_op_zdiv_r.
19839 PR tree-optimization/84223
19840 * graphite-scop-detection.c (gather_bbs::before_dom_children):
19841 Only add conditions from within the region.
19842 (gather_bbs::after_dom_children): Adjust.
19844 2018-02-07 Georg-Johann Lay <avr@gjlay.de>
19847 * config/avr/avr.h (GENERAL_REGNO_P, GENERAL_REG_P): New macros.
19848 * config/avr/avr.md: Only post-reload split REG-REG moves if
19849 either register is GENERAL_REG_P.
19851 2018-02-07 Jakub Jelinek <jakub@redhat.com>
19853 PR tree-optimization/84235
19854 * tree-ssa-scopedtables.c
19855 (avail_exprs_stack::simplify_binary_operation): Fir MINUS_EXPR, punt
19856 if the subtraction is performed in floating point type where NaNs are
19857 honored. For *DIV_EXPR, punt for ALL_FRACT_MODE_Ps where we can't
19858 build 1. Formatting fix.
19860 2018-02-06 Jakub Jelinek <jakub@redhat.com>
19863 * config/i386/i386.c (rest_of_insert_endbranch): Only skip
19864 NOTE_INSN_CALL_ARG_LOCATION after a call, not anything else,
19865 and skip it regardless of bb boundaries. Use CALL_P macro,
19866 don't test INSN_P (insn) together with CALL_P or JUMP_P check
19867 unnecessarily, formatting fix.
19869 2018-02-06 Michael Collison <michael.collison@arm.com>
19871 * config/arm/thumb2.md:
19872 (*thumb2_mov_negscc): Split only if TARGET_THUMB2 && !arm_restrict_it.
19873 (*thumb_mov_notscc): Ditto.
19875 2018-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
19878 * config/rs6000/rs6000.md (su code attribute): Use "u" for
19879 unsigned_fix, not "s".
19881 2018-02-06 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
19883 * configure.ac (gcc_fn_eh_frame_ro): New function.
19884 (gcc_cv_as_cfi_directive): Check both 32 and 64-bit assembler for
19885 correct .eh_frame permissions.
19886 * configure: Regenerate.
19888 2018-02-06 Andrew Jenner <andrew@codeourcery.com>
19890 * doc/invoke.texi: Add section for the PowerPC SPE backend. Remove
19891 irrelevant options.
19893 2018-02-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19895 * config/rs6000/rs6000.c (rs6000_option_override_internal):
19896 Display warning message for -mno-speculate-indirect-jumps.
19898 2018-02-06 Andrew Jenner <andrew@codesourcery.com>
19900 * config/powerpcspe/powerpcspe.opt: (msimple-fpu, mfpu) Add
19902 * config/powerpcspe/sysv4.opt (mbit-align): Likewise.
19904 2018-02-06 Aldy Hernandez <aldyh@redhat.com>
19906 PR tree-optimization/84225
19907 * tree-eh.c (find_trapping_overflow): Only call
19908 operation_no_trapping_overflow when ANY_INTEGRAL_TYPE_P.
19910 2018-02-06 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
19913 * config/i386/i386.c: Reimplement the check of possible options
19914 -mibt/-mshstk conbination. Change error messages.
19915 * doc/invoke.texi: Fix a typo: remove extra '='.
19917 2018-02-06 Marek Polacek <polacek@redhat.com>
19919 PR tree-optimization/84228
19920 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Skip debug statements.
19922 2018-02-06 Tamar Christina <tamar.christina@arm.com>
19925 * config/arm/arm.c (arm_print_asm_arch_directives): Record already
19926 emitted arch directives.
19927 * config/arm/arm-c.c (arm_cpu_builtins): Undefine __ARM_ARCH and
19928 __ARM_FEATURE_COPROC before changing architectures.
19930 2018-02-06 Richard Biener <rguenther@suse.de>
19932 * config/i386/i386.c (print_reg): Fix typo.
19933 (ix86_loop_unroll_adjust): Do not unroll beyond the original nunroll.
19935 2018-02-06 Eric Botcazou <ebotcazou@adacore.com>
19937 * configure: Regenerate.
19939 2018-02-05 Martin Sebor <msebor@redhat.com>
19941 PR tree-optimization/83369
19942 * tree-ssa-ccp.c (pass_post_ipa_warn::execute): Use %G to print
19945 2018-02-05 Martin Liska <mliska@suse.cz>
19947 * doc/invoke.texi: Cherry-pick upstream r323995.
19949 2018-02-05 Richard Sandiford <richard.sandiford@linaro.org>
19951 * ira.c (ira_init_register_move_cost): Adjust comment.
19953 2018-02-05 Martin Liska <mliska@suse.cz>
19955 PR gcov-profile/84137
19956 * doc/gcov.texi: Fix typo in documentation.
19958 2018-02-05 Martin Liska <mliska@suse.cz>
19960 PR gcov-profile/83879
19961 * doc/gcov.texi: Document necessity of --dynamic-list-data when
19962 using dlopen functionality.
19964 2018-02-05 Olga Makhotina <olga.makhotina@intel.com>
19966 * config/i386/avx512dqintrin.h (_mm_mask_range_sd, _mm_maskz_range_sd,
19967 _mm_mask_range_round_sd, _mm_maskz_range_round_sd, _mm_mask_range_ss,
19968 _mm_maskz_range_ss, _mm_mask_range_round_ss,
19969 _mm_maskz_range_round_ss): New intrinsics.
19970 (__builtin_ia32_rangesd128_round)
19971 (__builtin_ia32_rangess128_round): Remove.
19972 (__builtin_ia32_rangesd128_mask_round,
19973 __builtin_ia32_rangess128_mask_round): New builtins.
19974 * config/i386/i386-builtin.def (__builtin_ia32_rangesd128_round,
19975 __builtin_ia32_rangess128_round): Remove.
19976 (__builtin_ia32_rangesd128_mask_round,
19977 __builtin_ia32_rangess128_mask_round): New builtins.
19978 * config/i386/sse.md (ranges<mode><round_saeonly_name>): Renamed to ...
19979 (ranges<mode><mask_scalar_name><round_saeonly_scalar_name>): ... this.
19980 ((match_operand:VF_128 2 "<round_saeonly_nimm_predicate>"
19981 "<round_saeonly_constraint>")): Changed to ...
19982 ((match_operand:VF_128 2 "<round_saeonly_scalar_nimm_predicate>"
19983 "<round_saeonly_scalar_constraint>")): ... this.
19984 ("vrange<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
19985 %0, %1, %2<round_saeonly_op4>, %3}"): Changed to ...
19986 ("vrange<ssescalarmodesuffix>\t{%3, <round_saeonly_scalar_mask_op4>%2,
19987 %1, %0<mask_scalar_operand4>|%0<mask_scalar_operand4>, %1,
19988 %2<round_saeonly_scalar_mask_op4>, %3}"): ... this.
19990 2018-02-02 Andrew Jenner <andrew@codesourcery.com>
19992 * config/powerpcspe/powerpcspe.opt: Add Undocumented to irrelevant
19994 * config/powerpcspe/powerpcspe-tables.opt (rs6000_cpu_opt_value):
19995 Remove all values except native, 8540 and 8548.
19997 2018-02-02 H.J. Lu <hongjiu.lu@intel.com>
19999 * config/i386/i386.c (ix86_output_function_return): Pass
20000 INVALID_REGNUM, instead of -1, as invalid register number to
20001 indirect_thunk_name and output_indirect_thunk.
20003 2018-02-02 Julia Koval <julia.koval@intel.com>
20005 * config.gcc: Add -march=icelake.
20006 * config/i386/driver-i386.c (host_detect_local_cpu): Detect icelake.
20007 * config/i386/i386-c.c (ix86_target_macros_internal): Handle icelake.
20008 * config/i386/i386.c (processor_costs): Add m_ICELAKE.
20009 (PTA_ICELAKE, PTA_AVX512VNNI, PTA_GFNI, PTA_VAES, PTA_AVX512VBMI2,
20010 PTA_VPCLMULQDQ, PTA_RDPID, PTA_AVX512BITALG): New.
20011 (processor_target_table): Add icelake.
20012 (ix86_option_override_internal): Handle new PTAs.
20013 (get_builtin_code_for_version): Handle icelake.
20014 (M_INTEL_COREI7_ICELAKE): New.
20015 (fold_builtin_cpu): Handle icelake.
20016 * config/i386/i386.h (TARGET_ICELAKE, PROCESSOR_ICELAKE): New.
20017 * doc/invoke.texi: Add -march=icelake.
20019 2018-02-02 Julia Koval <julia.koval@intel.com>
20021 * config/i386/i386.c (ix86_option_override_internal): Change flags type
20022 to wide_int_bitmask.
20023 * wide-int-bitmask.h: New.
20025 2018-02-02 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
20028 * config/i386/i386.md: Replace Pmode with word_mode in
20029 builtin_setjmp_setup and builtin_longjmp to support x32.
20031 2018-02-01 Peter Bergner <bergner@vnet.ibm.com>
20035 * config/rs6000/driver-rs6000.c: #include "diagnostic.h".
20037 (rs6000_supported_cpu_names): New static variable.
20038 (linux_cpu_translation_table): Likewise.
20039 (elf_platform) <cpu>: Define new static variable and use it.
20040 Translate kernel AT_PLATFORM name to canonical name if needed.
20041 Error if platform name is unknown.
20043 2018-02-01 Aldy Hernandez <aldyh@redhat.com>
20046 * config/pa/predicates.md (base14_operand): Handle E_VOIDmode.
20048 2018-02-01 Jeff Law <law@redhat.com>
20051 * config/i386/i386.c (release_scratch_register_on_entry): Add new
20052 OFFSET and RELEASE_VIA_POP arguments. Use SP+OFFSET to restore
20053 the scratch if RELEASE_VIA_POP is false.
20054 (ix86_adjust_stack_and_probe_stack_clash): Un-constify SIZE.
20055 If we have to save a temporary register, decrement SIZE appropriately.
20056 Pass new arguments to release_scratch_register_on_entry.
20057 (ix86_adjust_stack_and_probe): Likewise.
20058 (ix86_emit_probe_stack_range): Pass new arguments to
20059 release_scratch_register_on_entry.
20061 2018-02-01 Uros Bizjak <ubizjak@gmail.com>
20063 PR rtl-optimization/84157
20064 * combine.c (change_zero_ext): Use REG_P predicate in
20065 front of HARD_REGISTER_P predicate.
20067 2018-02-01 Georg-Johann Lay <avr@gjlay.de>
20069 * config/avr/avr.c (avr_option_override): Move disabling of
20070 -fdelete-null-pointer-checks to...
20071 * common/config/avr/avr-common.c (avr_option_optimization_table):
20074 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
20076 PR tree-optimization/81635
20077 * tree-data-ref.c (split_constant_offset_1): For types that
20078 wrap on overflow, try to use range info to prove that wrapping
20081 2018-02-01 Renlin Li <renlin.li@arm.com>
20084 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
20085 TAILCALL_ADDR_REGS.
20086 (aarch64_register_move_cost): Likewise.
20087 * config/aarch64/aarch64.h (reg_class): Rename CALLER_SAVE_REGS to
20088 TAILCALL_ADDR_REGS.
20089 (REG_CLASS_NAMES): Likewise.
20090 (REG_CLASS_CONTENTS): Rename CALLER_SAVE_REGS to
20091 TAILCALL_ADDR_REGS. Remove IP registers.
20092 * config/aarch64/aarch64.md (Ucs): Update register constraint.
20094 2018-02-01 Richard Biener <rguenther@suse.de>
20096 * domwalk.h (dom_walker::dom_walker): Add additional constructor
20097 for specifying RPO order and allow NULL for that.
20098 * domwalk.c (dom_walker::dom_walker): Likewise.
20099 (dom_walker::walk): Handle NULL RPO order.
20100 * tree-into-ssa.c (rewrite_dom_walker): Do not walk dom children
20102 (rewrite_update_dom_walker): Likewise.
20103 (mark_def_dom_walker): Likewise.
20105 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
20107 * config/aarch64/aarch64-protos.h (aarch64_split_sve_subreg_move)
20108 (aarch64_maybe_expand_sve_subreg_move): Declare.
20109 * config/aarch64/aarch64.md (UNSPEC_REV_SUBREG): New unspec.
20110 * config/aarch64/predicates.md (aarch64_any_register_operand): New
20112 * config/aarch64/aarch64-sve.md (mov<mode>): Optimize subreg moves
20113 that are semantically a reverse operation.
20114 (*aarch64_sve_mov<mode>_subreg_be): New pattern.
20115 * config/aarch64/aarch64.c (aarch64_maybe_expand_sve_subreg_move):
20116 (aarch64_replace_reg_mode, aarch64_split_sve_subreg_move): New
20118 (aarch64_can_change_mode_class): For big-endian, forbid changes
20119 between two SVE modes if they have different element sizes.
20121 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
20123 * config/aarch64/aarch64.c (aarch64_expand_sve_const_vector): Prefer
20124 the TImode handling for big-endian targets.
20126 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
20128 * config/aarch64/aarch64-sve.md (sve_ld1rq): Replace with...
20129 (*sve_ld1rq<Vesize>): ... this new pattern. Handle all element sizes,
20131 * config/aarch64/aarch64.c (aarch64_expand_sve_widened_duplicate):
20132 Remove BSWAP handing for big-endian targets and use the form of
20133 LD1RQ appropariate for the mode.
20135 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
20137 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Handle
20138 all CONST_VECTOR_DUPLICATE_P vectors, not just those with a single
20139 duplicated element.
20141 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
20144 * config/aarch64/aarch64.c (aarch64_secondary_reload): Tighten
20145 check for operands that need to go through aarch64_sve_reload_be.
20147 2018-02-01 Jakub Jelinek <jakub@redhat.com>
20149 PR tree-optimization/81661
20150 PR tree-optimization/84117
20151 * tree-eh.h (rewrite_to_non_trapping_overflow): Declare.
20152 * tree-eh.c: Include gimplify.h.
20153 (find_trapping_overflow, replace_trapping_overflow,
20154 rewrite_to_non_trapping_overflow): New functions.
20155 * tree-vect-loop.c: Include tree-eh.h.
20156 (vect_get_loop_niters): Use rewrite_to_non_trapping_overflow.
20157 * tree-data-ref.c: Include tree-eh.h.
20158 (get_segment_min_max): Use rewrite_to_non_trapping_overflow.
20160 2018-01-31 Uros Bizjak <ubizjak@gmail.com>
20162 PR rtl-optimization/84123
20163 * combine.c (change_zero_ext): Check if hard register satisfies
20164 can_change_dest_mode before calling gen_lowpart_SUBREG.
20166 2018-01-31 Vladimir Makarov <vmakarov@redhat.com>
20169 * ira.c (ira_init_register_move_cost): Remove assert.
20171 2018-01-31 Eric Botcazou <ebotcazou@adacore.com>
20173 PR rtl-optimization/84071
20174 * doc/tm.texi.in (WORD_REGISTER_OPERATIONS): Add explicit case.
20175 * doc/tm.texi: Regenerate.
20177 2018-01-31 Richard Biener <rguenther@suse.de>
20179 PR tree-optimization/84132
20180 * tree-data-ref.c (analyze_miv_subscript): Properly
20181 check whether evolution_function_is_affine_multivariate_p
20182 before calling gcd_of_steps_may_divide_p.
20184 2018-01-31 Julia Koval <julia.koval@intel.com>
20187 * config/i386/i386.c (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
20188 * config/i386/i386.md (rdpid_rex64) New.
20189 (rdpid): Make 32bit only.
20191 2018-01-29 Aldy Hernandez <aldyh@redhat.com>
20194 * tree-pretty-print.c (dump_generic_node): Handle a TYPE_NAME with
20195 an IDENTIFIER_NODE for FUNCTION_TYPE's.
20197 2018-01-31 Eric Botcazou <ebotcazou@adacore.com>
20200 2018-01-12 Eric Botcazou <ebotcazou@adacore.com>
20202 * config/sparc/sparc.md (vxworks_load_got): Set the GOT register.
20204 2018-01-31 Eric Botcazou <ebotcazou@adacore.com>
20206 PR rtl-optimization/84071
20207 * combine.c (record_dead_and_set_regs_1): Record the source unmodified
20208 for a paradoxical SUBREG on a WORD_REGISTER_OPERATIONS target.
20210 2018-01-31 Claudiu Zissulescu <claziss@synopsys.com>
20212 * config/arc/arc.c (arc_handle_aux_attribute): New function.
20213 (arc_attribute_table): Add 'aux' attribute.
20214 (arc_in_small_data_p): Consider aux like variables.
20215 (arc_is_aux_reg_p): New function.
20216 (arc_asm_output_aligned_decl_local): Ignore 'aux' like variables.
20217 (arc_get_aux_arg): New function.
20218 (prepare_move_operands): Handle aux-register access.
20219 (arc_handle_aux_attribute): New function.
20220 * doc/extend.texi (ARC Variable attributes): Add subsection.
20222 2018-01-31 Claudiu Zissulescu <claziss@synopsys.com>
20224 * config/arc/arc-protos.h (arc_is_uncached_mem_p): Function proto.
20225 * config/arc/arc.c (arc_handle_uncached_attribute): New function.
20226 (arc_attribute_table): Add 'uncached' attribute.
20227 (arc_print_operand): Print '.di' flag for uncached memory
20229 (arc_in_small_data_p): Do not consider for small data the uncached
20231 (arc_is_uncached_mem_p): New function.
20232 * config/arc/predicates.md (compact_store_memory_operand): Check
20233 for uncached memory accesses.
20234 (nonvol_nonimm_operand): Likewise.
20235 * doc/extend.texi (ARC Type Attribute): New subsection.
20237 2018-01-31 Jakub Jelinek <jakub@redhat.com>
20240 * common.opt (falign-functions=, falign-jumps=, falign-labels=,
20241 falign-loops=): Add Optimization flag.
20243 2018-01-30 Jeff Law <law@redhat.com>
20246 * i386.c (ix86_adjust_stack_and_probe_stack_clash): New argument
20247 INT_REGISTERS_SAVED. Check it prior to calling
20248 get_scratch_register_on_entry.
20249 (ix86_adjust_stack_and_probe): Similarly.
20250 (ix86_emit_probe_stack_range): Similarly.
20251 (ix86_expand_prologue): Corresponding changes.
20253 2018-01-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
20256 * config/sol2.h (STARTFILE_ARCH_SPEC): Use -std=c*,
20257 -std=iso9899:199409 instead of -pedantic to select values-Xc.o.
20259 2018-01-30 Vladimir Makarov <vmakarov@redhat.com>
20262 * lra-constraints.c (curr_insn_transform): Process AND in the
20265 2018-01-30 Jakub Jelinek <jakub@redhat.com>
20267 PR rtl-optimization/83986
20268 * sched-deps.c (sched_analyze_insn): For frame related insns, add anti
20269 dependence against last_pending_memory_flush in addition to
20270 pending_jump_insns.
20272 2018-01-30 Alexandre Oliva <aoliva@redhat.com>
20274 PR tree-optimization/81611
20275 * tree-ssa-dom.c (simple_iv_increment_p): Skip intervening
20278 2018-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
20281 * config/rs6000/rs6000.c (rs6000_internal_arg_pointer): Only return
20284 2018-01-30 Richard Biener <rguenther@suse.de>
20285 Jakub Jelinek <jakub@redhat.com>
20287 PR tree-optimization/84111
20288 * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely_1): Skip
20289 inner loops added during recursion, as they don't have up-to-date
20292 2018-01-30 Jan Hubicka <hubicka@ucw.cz>
20295 * ipa-inline.c (can_inline_edge_p): Break out late tests to...
20296 (can_inline_edge_by_limits_p): ... here.
20297 (can_early_inline_edge_p, check_callers,
20298 update_caller_keys, update_callee_keys, recursive_inlining,
20299 add_new_edges_to_heap, speculation_useful_p,
20300 inline_small_functions,
20301 inline_small_functions, flatten_function,
20302 inline_to_all_callers_1): Update.
20304 2018-01-30 Jan Hubicka <hubicka@ucw.cz>
20306 * profile-count.c (profile_count::combine_with_ipa_count): Handle
20309 2018-01-30 Richard Biener <rguenther@suse.de>
20311 PR tree-optimization/83008
20312 * tree-vect-slp.c (vect_analyze_slp_cost_1): Properly cost
20313 invariant and constant vector uses in stmts when they need
20314 more than one stmt.
20316 2018-01-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
20319 * configure.ac (gcc_cv_as_shf_merge): Disable on Solaris 10/x86.
20320 * configure: Regenerate.
20322 2018-01-30 Richard Sandiford <richard.sandiford@linaro.org>
20324 * config/aarch64/aarch64-sve.md (*vec_extract<mode><Vel>_0): New
20326 (*vec_extract<mode><Vel>_v128): Require a nonzero lane number.
20327 Use gen_rtx_REG rather than gen_lowpart.
20329 2018-01-30 Richard Sandiford <richard.sandiford@linaro.org>
20331 * lra-constraints.c (match_reload): Use subreg_lowpart_offset
20332 rather than 0 when creating partial subregs.
20334 2018-01-30 Richard Sandiford <richard.sandiford@linaro.org>
20336 * vec-perm-indices.c (vec_perm_indices::series_p): Give examples
20339 2018-01-29 Michael Meissner <meissner@linux.vnet.ibm.com>
20342 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): If DFmode
20343 and SFmode can go in Altivec registers (-mcpu=power7 for DFmode,
20344 -mcpu=power8 for SFmode) don't set the PRE_INCDEC or PRE_MODIFY
20345 flags. This restores the settings used before the 2017-07-24.
20346 Turning off pre increment/decrement/modify allows IVOPTS to
20347 optimize DF/SF loops where the index is an int.
20349 2018-01-29 Richard Biener <rguenther@suse.de>
20350 Kelvin Nilsen <kelvin@gcc.gnu.org>
20353 * tree-vect-stmts.c (vectorizable_call): Don't call
20354 targetm.vectorize_builtin_md_vectorized_function if callee is
20357 2018-01-22 Carl Love <cel@us.ibm.com>
20359 * doc/extend.tex: Fix typo in second arg in
20360 __builtin_bcdadd_{lt|eq|gt|ov} and __builtin_bcdsub_{lt|eq|gt|ov}.
20362 2018-01-29 Richard Biener <rguenther@suse.de>
20364 PR tree-optimization/84086
20365 * tree-ssanames.c: Include cfgloop.h and tree-scalar-evolution.h.
20366 (flush_ssaname_freelist): When SSA names were released reset
20367 the SCEV hash table.
20369 2018-01-29 Richard Biener <rguenther@suse.de>
20371 PR tree-optimization/84057
20372 * tree-ssa-loop-ivcanon.c (unloop_loops): Deal with already
20373 removed paths when removing edges.
20375 2018-01-27 H.J. Lu <hongjiu.lu@intel.com>
20377 * doc/invoke.texi: Replace -mfunction-return==@var{choice} with
20378 -mfunction-return=@var{choice}.
20380 2018-01-27 Bernd Edlinger <bernd.edlinger@hotmail.de>
20382 PR diagnostic/84034
20383 * diagnostic-show-locus.c (get_line_width_without_trailing_whitespace):
20384 Handle CR like TAB.
20385 (layout::print_source_line): Likewise.
20386 (test_get_line_width_without_trailing_whitespace): Add test cases.
20388 2018-01-27 Jakub Jelinek <jakub@redhat.com>
20390 PR middle-end/84040
20391 * sched-deps.c (sched_macro_fuse_insns): Return immediately for
20394 2018-01-26 Jim Wilson <jimw@sifive.com>
20396 * config/riscv/riscv.h (MAX_FIXED_MODE_SIZE): New.
20398 * config/riscv/elf.h (LIB_SPEC): Don't include -lgloss when nosys.specs
20401 2018-01-26 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20403 * config/aarch64/aarch64.md: Add peepholes for CMP + SUB -> SUBS
20404 and CMP + SUB-immediate -> SUBS.
20406 2018-01-26 Martin Sebor <msebor@redhat.com>
20408 PR tree-optimization/83896
20409 * tree-ssa-strlen.c (get_string_len): Rename...
20410 (get_string_cst_length): ...to this. Return HOST_WIDE_INT.
20411 Avoid assuming length is constant.
20412 (handle_char_store): Use HOST_WIDE_INT for string length.
20414 2018-01-26 Uros Bizjak <ubizjak@gmail.com>
20417 * config/i386/i386.md (*andndi3_doubleword): Add earlyclobber
20418 to (=&r,r,rm) alternative. Add (=r,0,rm) and (=r,r,0) alternatives.
20420 2018-01-26 Richard Biener <rguenther@suse.de>
20422 PR rtl-optimization/84003
20423 * dse.c (record_store): Only record redundant stores when
20424 the earlier store aliases at least all accesses the later one does.
20426 2018-01-26 Jakub Jelinek <jakub@redhat.com>
20428 PR rtl-optimization/83985
20429 * dce.c (deletable_insn_p): Return false for separate shrink wrapping
20430 REG_CFA_RESTORE insns.
20431 (delete_unmarked_insns): Don't ignore separate shrink wrapping
20432 REG_CFA_RESTORE insns here.
20435 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Don't
20436 use SSA_NAME_VAR as base for SSA_NAMEs with non-NULL SSA_NAME_VAR.
20438 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
20440 * config/arc/arc-arch.h (arc_tune_attr): Add ARC_TUNE_CORE_3.
20441 * config/arc/arc.c (arc_sched_issue_rate): Use ARC_TUNE_... .
20442 (arc_init): Likewise.
20443 (arc_override_options): Likewise.
20444 (arc_file_start): Choose Tag_ARC_CPU_variation based on arc_tune
20446 (hwloop_fail): Use TARGET_DBNZ when we want to check for dbnz insn
20448 * config/arc/arc.h (TARGET_DBNZ): Define.
20449 * config/arc/arc.md (attr tune): Add core_3, use ARC_TUNE_... to
20450 properly set the tune attribute.
20451 (dbnz): Use TARGET_DBNZ guard.
20452 * config/arc/arc.opt (mtune): Add core3 option.
20454 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
20456 * config/arc/arc.c (arc_delegitimize_address_0): Refactored to
20457 recognize new pic like addresses.
20458 (arc_delegitimize_address): Clean up.
20460 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
20462 * config/arc/arc-arches.def: Option mrf16 valid for all
20464 * config/arc/arc-c.def (__ARC_RF16__): New predefined macro.
20465 * config/arc/arc-cpus.def (em_mini): New cpu with rf16 on.
20466 * config/arc/arc-options.def (FL_RF16): Add mrf16 option.
20467 * config/arc/arc-tables.opt: Regenerate.
20468 * config/arc/arc.c (arc_conditional_register_usage): Handle
20469 reduced register file case.
20470 (arc_file_start): Set must have build attributes.
20471 * config/arc/arc.h (MAX_ARC_PARM_REGS): Conditional define using
20472 mrf16 option value.
20473 * config/arc/arc.opt (mrf16): Add new option.
20474 * config/arc/elf.h (ATTRIBUTE_PCS): Define.
20475 * config/arc/genmultilib.awk: Handle new mrf16 option.
20476 * config/arc/linux.h (ATTRIBUTE_PCS): Define.
20477 * config/arc/t-multilib: Regenerate.
20478 * doc/invoke.texi (ARC Options): Document mrf16 option.
20480 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
20482 * config/arc/arc-protos.h: Add arc_is_secure_call_p proto.
20483 * config/arc/arc.c (arc_handle_secure_attribute): New function.
20484 (arc_attribute_table): Add 'secure_call' attribute.
20485 (arc_print_operand): Print secure call operand.
20486 (arc_function_ok_for_sibcall): Don't optimize tail calls when
20488 (arc_is_secure_call_p): New function. * config/arc/arc.md
20489 (call_i): Add support for sjli instruction.
20490 (call_value_i): Likewise.
20491 * config/arc/constraints.md (Csc): New constraint.
20493 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
20494 John Eric Martin <John.Martin@emmicro-us.com>
20496 * config/arc/arc-protos.h: Add arc_is_jli_call_p proto.
20497 * config/arc/arc.c (_arc_jli_section): New struct.
20498 (arc_jli_section): New type.
20499 (rc_jli_sections): New static variable.
20500 (arc_handle_jli_attribute): New function.
20501 (arc_attribute_table): Add jli_always and jli_fixed attribute.
20502 (arc_file_end): New function.
20503 (TARGET_ASM_FILE_END): Define.
20504 (arc_print_operand): Reuse 'S' letter for JLI output instruction.
20505 (arc_add_jli_section): New function.
20506 (jli_call_scan): Likewise.
20507 (arc_reorg): Call jli_call_scan.
20508 (arc_output_addsi): Remove 'S' from printing asm operand.
20509 (arc_is_jli_call_p): New function.
20510 * config/arc/arc.md (movqi_insn): Remove 'S' from printing asm
20512 (movhi_insn): Likewise.
20513 (movsi_insn): Likewise.
20514 (movsi_set_cc_insn): Likewise.
20515 (loadqi_update): Likewise.
20516 (load_zeroextendqisi_update): Likewise.
20517 (load_signextendqisi_update): Likewise.
20518 (loadhi_update): Likewise.
20519 (load_zeroextendhisi_update): Likewise.
20520 (load_signextendhisi_update): Likewise.
20521 (loadsi_update): Likewise.
20522 (loadsf_update): Likewise.
20523 (movsicc_insn): Likewise.
20524 (bset_insn): Likewise.
20525 (bxor_insn): Likewise.
20526 (bclr_insn): Likewise.
20527 (bmsk_insn): Likewise.
20528 (bicsi3_insn): Likewise.
20529 (cmpsi_cc_c_insn): Likewise.
20530 (movsi_ne): Likewise.
20531 (movsi_cond_exec): Likewise.
20532 (clrsbsi2): Likewise.
20533 (norm_f): Likewise.
20543 (call_i): Remove 'S' asm letter, add jli instruction.
20544 (call_value_i): Likewise.
20545 * config/arc/arc.op (mjli-always): New option.
20546 * config/arc/constraints.md (Cji): New constraint.
20547 * config/arc/fpx.md (addsf3_fpx): Remove 'S' from printing asm
20549 (subsf3_fpx): Likewise.
20550 (mulsf3_fpx): Likewise.
20551 * config/arc/simdext.md (vendrec_insn): Remove 'S' from printing
20553 * doc/extend.texi (ARC): Document 'jli-always' and 'jli-fixed'
20554 function attrbutes.
20555 * doc/invoke.texi (ARC): Document mjli-always option.
20557 2018-01-26 Sebastian Perta <sebastian.perta@renesas.com>
20559 * config/rl78/rl78.c (rl78_addsi3_internal): If operand 2 is const
20560 avoid addition with 0 and use incw and decw where possible.
20562 2018-01-26 Richard Biener <rguenther@suse.de>
20564 PR tree-optimization/81082
20565 * fold-const.c (fold_plusminus_mult_expr): Do not perform the
20566 association if it requires casting to unsigned.
20567 * match.pd ((A * C) +- (B * C) -> (A+-B)): New patterns derived
20568 from fold_plusminus_mult_expr to catch important cases late when
20569 range info is available.
20571 2018-01-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
20573 * config/i386/sol2.h (USE_HIDDEN_LINKONCE): Remove.
20574 * configure.ac (hidden_linkonce): New test.
20575 * configure: Regenerate.
20576 * config.in: Regenerate.
20578 2018-01-26 Julia Koval <julia.koval@intel.com>
20580 * config/i386/avx512bitalgintrin.h (_mm512_bitshuffle_epi64_mask,
20581 _mm512_mask_bitshuffle_epi64_mask, _mm256_bitshuffle_epi64_mask,
20582 _mm256_mask_bitshuffle_epi64_mask, _mm_bitshuffle_epi64_mask,
20583 _mm_mask_bitshuffle_epi64_mask): Fix type.
20584 * config/i386/i386-builtin-types.def (UHI_FTYPE_V2DI_V2DI_UHI,
20585 USI_FTYPE_V4DI_V4DI_USI): Remove.
20586 * config/i386/i386-builtin.def (__builtin_ia32_vpshufbitqmb512_mask,
20587 __builtin_ia32_vpshufbitqmb256_mask,
20588 __builtin_ia32_vpshufbitqmb128_mask): Fix types.
20589 * config/i386/i386.c (ix86_expand_args_builtin): Remove old types.
20590 * config/i386/sse.md (VI1_AVX512VLBW): Change types.
20592 2018-01-26 Alan Modra <amodra@gmail.com>
20595 * config/rs6000/rs6000-p8swap.c (rtx_is_swappable_p): Exclude
20596 UNSPEC_VBPERMQ. Sort other unspecs.
20598 2018-01-25 David Edelsohn <dje.gcc@gmail.com>
20600 * doc/invoke.texi (PowerPC Options): Document 'native' cpu type.
20602 2018-01-25 Jan Hubicka <hubicka@ucw.cz>
20604 PR middle-end/83055
20605 * predict.c (drop_profile): Do not push/pop cfun; update also
20607 (handle_missing_profiles): Fix logic looking for zero profiles.
20609 2018-01-25 Jakub Jelinek <jakub@redhat.com>
20611 PR middle-end/83977
20612 * ipa-fnsummary.c (compute_fn_summary): Clear can_change_signature
20613 on functions with #pragma omp declare simd or functions with simd
20615 * omp-simd-clone.c (expand_simd_clones): Revert 2018-01-24 change.
20616 * config/i386/i386.c (ix86_simd_clone_compute_vecsize_and_simdlen):
20617 Remove trailing \n from warning_at calls.
20619 2018-01-25 Tom de Vries <tom@codesourcery.com>
20622 * config/nvptx/nvptx.c (nvptx_single): Add exit insn after noreturn call
20623 for neutered workers.
20625 2018-01-24 Joseph Myers <joseph@codesourcery.com>
20628 * config/m68k/m68k.c (m68k_promote_function_mode): New function.
20629 (TARGET_PROMOTE_FUNCTION_MODE): New macro.
20631 2018-01-24 Jeff Law <law@redhat.com>
20634 * i386.c (get_probe_interval): Move to earlier point.
20635 (ix86_compute_frame_layout): If -fstack-clash-protection and
20636 the frame is larger than the probe interval, then use pushes
20637 to save registers rather than reg->mem moves.
20638 (ix86_expand_prologue): Remove conditional for int_registers_saved
20641 2018-01-24 Vladimir Makarov <vmakarov@redhat.com>
20644 * ira-build.c (setup_min_max_allocno_live_range_point): Set up
20645 min/max for never referenced object.
20647 2018-01-24 Jakub Jelinek <jakub@redhat.com>
20649 PR middle-end/83977
20650 * tree.c (free_lang_data_in_decl): Don't clear DECL_ABSTRACT_ORIGIN
20652 * omp-low.c (create_omp_child_function): Remove "omp declare simd"
20653 attributes from DECL_ATTRIBUTES (decl) without affecting
20654 DECL_ATTRIBUTES (current_function_decl).
20655 * omp-simd-clone.c (expand_simd_clones): Ignore DECL_ARTIFICIAL
20656 functions with non-NULL DECL_ABSTRACT_ORIGIN.
20658 2018-01-24 Richard Sandiford <richard.sandiford@linaro.org>
20660 PR tree-optimization/83979
20661 * fold-const.c (fold_comparison): Use constant_boolean_node
20662 instead of boolean_{true,false}_node.
20664 2018-01-24 Jan Hubicka <hubicka@ucw.cz>
20666 * ipa-profile.c (ipa_propagate_frequency_1): Fix logic skipping calls
20669 2018-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20671 * config/rs6000/rs6000.md (*call_indirect_nonlocal_sysv<mode>):
20672 Simplify the clause that sets the length attribute.
20673 (*call_value_indirect_nonlocal_sysv<mode>): Likewise.
20674 (*sibcall_nonlocal_sysv<mode>): Clean up code block; simplify the
20675 clause that sets the length attribute.
20676 (*sibcall_value_nonlocal_sysv<mode>): Likewise.
20678 2018-01-24 Tom de Vries <tom@codesourcery.com>
20681 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG_2): Define to 1.
20682 (nvptx_pc_set, nvptx_condjump_label): New function. Copy from jump.c.
20683 Add strict parameter.
20684 (prevent_branch_around_nothing): Insert dummy insn between branch to
20685 label and label with no ptx insn inbetween.
20686 * config/nvptx/nvptx.md (define_insn "fake_nop"): New insn.
20688 2018-01-24 Tom de Vries <tom@codesourcery.com>
20691 * config/nvptx/nvptx.c (nvptx_single): Add exit insn after noreturn call
20692 for neutered threads in warp.
20693 * config/nvptx/nvptx.md (define_insn "exit"): New insn.
20695 2018-01-24 Richard Biener <rguenther@suse.de>
20697 PR tree-optimization/83176
20698 * tree-chrec.c (chrec_fold_plus_1): Handle (signed T){(T) .. }
20701 2018-01-24 Richard Biener <rguenther@suse.de>
20703 PR tree-optimization/82819
20704 * graphite-isl-ast-to-gimple.c (binary_op_to_tree): Avoid
20705 code generating pluses that are no-ops in the target precision.
20707 2018-01-24 Richard Biener <rguenther@suse.de>
20709 PR middle-end/84000
20710 * tree-cfg.c (replace_loop_annotate): Handle annot_expr_parallel_kind.
20712 2018-01-23 Jan Hubicka <hubicka@ucw.cz>
20714 * cfgcleanup.c (try_crossjump_to_edge): Use combine_with_count
20715 to merge probabilities.
20716 * predict.c (probably_never_executed): Also mark as cold functions
20717 with global 0 profile and guessed local profile.
20718 * profile-count.c (profile_probability::combine_with_count): New
20720 * profile-count.h (profile_probability::operator*,
20721 profile_probability::operator*=, profile_probability::operator/,
20722 profile_probability::operator/=): Reduce precision to adjusted
20723 and set value to guessed on contradictory divisions.
20724 (profile_probability::combine_with_freq): Remove.
20725 (profile_probability::combine_wiht_count): Declare.
20726 (profile_count::force_nonzero):: Set to adjusted.
20727 (profile_count::probability_in):: Set quality to adjusted.
20728 * tree-ssa-tail-merge.c (replace_block_by): Use
20729 combine_with_count.
20731 2018-01-23 Andrew Waterman <andrew@sifive.com>
20732 Jim Wilson <jimw@sifive.com>
20734 * config/riscv/riscv.c (riscv_stack_boundary): New.
20735 (riscv_option_override): Set riscv_stack_boundary. Handle
20736 riscv_preferred_stack_boundary_arg.
20737 * config/riscv/riscv.h (MIN_STACK_BOUNDARY, ABI_STACK_BOUNDARY): New.
20738 (BIGGEST_ALIGNMENT): Set to STACK_BOUNDARY.
20739 (STACK_BOUNDARY): Set to riscv_stack_boundary.
20740 (RISCV_STACK_ALIGN): Use STACK_BOUNDARY.
20741 * config/riscv/riscv.opt (mpreferred-stack-boundary): New.
20742 * doc/invoke.tex (RISC-V Options): Add -mpreferred-stack-boundary.
20744 2018-01-23 H.J. Lu <hongjiu.lu@intel.com>
20747 * config/i386/i386.c (ix86_expand_prologue): Use cost reference
20748 of struct ix86_frame.
20749 (ix86_expand_epilogue): Likewise. Add a local variable for
20750 the reg_save_offset field in struct ix86_frame.
20752 2018-01-23 Bin Cheng <bin.cheng@arm.com>
20754 PR tree-optimization/82604
20755 * tree-loop-distribution.c (enum partition_kind): New enum item
20756 PKIND_PARTIAL_MEMSET.
20757 (partition_builtin_p): Support above new enum item.
20758 (generate_code_for_partition): Ditto.
20759 (compute_access_range): Differentiate cases that equality can be
20760 proven at all loops, the innermost loops or no loops.
20761 (classify_builtin_st, classify_builtin_ldst): Adjust call to above
20762 function. Set PKIND_PARTIAL_MEMSET for partition appropriately.
20763 (finalize_partitions, distribute_loop): Don't fuse partition of
20764 PKIND_PARTIAL_MEMSET kind when distributing 3-level loop nest.
20765 (prepare_perfect_loop_nest): Distribute 3-level loop nest only if
20766 parloop is enabled.
20768 2018-01-23 Martin Liska <mliska@suse.cz>
20770 * predict.def (PRED_INDIR_CALL): Set probability to PROB_EVEN in
20771 order to ignore the predictor.
20772 (PRED_POLYMORPHIC_CALL): Likewise.
20773 (PRED_RECURSIVE_CALL): Likewise.
20775 2018-01-23 Martin Liska <mliska@suse.cz>
20777 * tree-profile.c (tree_profiling): Print function header to
20778 aware reader which function we are working on.
20779 * value-prof.c (gimple_find_values_to_profile): Do not print
20780 not interesting value histograms.
20782 2018-01-23 Martin Liska <mliska@suse.cz>
20784 * profile-count.h (enum profile_quality): Add
20785 profile_uninitialized as the first value. Do not number values
20786 as they are zero based.
20787 (profile_count::verify): Update sanity check.
20788 (profile_probability::verify): Likewise.
20790 2018-01-23 Nathan Sidwell <nathan@acm.org>
20792 * doc/invoke.texi (ffor-scope): Deprecate.
20794 2018-01-23 David Malcolm <dmalcolm@redhat.com>
20796 PR tree-optimization/83510
20797 * domwalk.c (set_all_edges_as_executable): New function.
20798 (dom_walker::dom_walker): Convert bool param
20799 "skip_unreachable_blocks" to enum reachability. Move setup of
20800 edge flags to set_all_edges_as_executable and only do it when
20801 reachability is REACHABLE_BLOCKS.
20802 * domwalk.h (enum dom_walker::reachability): New enum.
20803 (dom_walker::dom_walker): Convert bool param
20804 "skip_unreachable_blocks" to enum reachability.
20805 (set_all_edges_as_executable): New decl.
20806 * graphite-scop-detection.c (gather_bbs::gather_bbs): Convert
20807 from false for "skip_unreachable_blocks" to ALL_BLOCKS for
20809 * tree-ssa-dom.c (dom_opt_dom_walker::dom_opt_dom_walker): Likewise,
20810 but converting true to REACHABLE_BLOCKS.
20811 * tree-ssa-sccvn.c (sccvn_dom_walker::sccvn_dom_walker): Likewise.
20813 (check_array_bounds_dom_walker::check_array_bounds_dom_walker):
20814 Likewise, but converting it to REACHABLE_BLOCKS_PRESERVING_FLAGS.
20815 (vrp_dom_walker::vrp_dom_walker): Likewise, but converting it to
20817 (vrp_prop::vrp_finalize): Call set_all_edges_as_executable
20818 if check_all_array_refs will be called.
20820 2018-01-23 David Malcolm <dmalcolm@redhat.com>
20822 * tree.c (selftest::test_location_wrappers): Add more test
20825 2018-01-23 David Malcolm <dmalcolm@redhat.com>
20827 * sbitmap.c (selftest::test_set_range): Fix memory leaks.
20828 (selftest::test_bit_in_range): Likewise.
20830 2018-01-23 Richard Sandiford <richard.sandiford@linaro.org>
20833 * doc/sourcebuild.texi (vect_float): Say that the selector
20834 only describes the situation when -funsafe-math-optimizations is on.
20835 (vect_float_strict): Document.
20837 2018-01-23 Richard Sandiford <richard.sandiford@linaro.org>
20839 PR tree-optimization/83965
20840 * tree-vect-patterns.c (vect_reassociating_reduction_p): New function.
20841 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern): Use it
20842 instead of checking only for a reduction.
20843 (vect_recog_widen_sum_pattern): Likewise.
20845 2018-01-23 Jan Hubicka <hubicka@ucw.cz>
20847 * predict.c (probably_never_executed): Only use precise profile info.
20848 (compute_function_frequency): Skip after inlining hack since we now
20849 have quality checking.
20851 2018-01-23 Jan Hubicka <hubicka@ucw.cz>
20853 * profile-count.h (profile_probability::very_unlikely,
20854 profile_probability::unlikely, profile_probability::even): Set
20855 precision to guessed.
20857 2018-01-23 Richard Biener <rguenther@suse.de>
20859 PR tree-optimization/83963
20860 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
20861 Properly terminate dominator walk when crossing the exit edge not
20862 when visiting its source block.
20864 2018-01-23 Jakub Jelinek <jakub@redhat.com>
20867 * tree.c (maybe_wrap_with_location): Use NON_LVALUE_EXPR rather than
20868 VIEW_CONVERT_EXPR to wrap CONST_DECLs.
20870 2018-01-22 Jakub Jelinek <jakub@redhat.com>
20872 PR tree-optimization/83957
20873 * omp-expand.c (expand_omp_for_generic): Ignore virtual PHIs. Remove
20874 semicolon after for body surrounded by braces.
20876 PR tree-optimization/83081
20877 * profile-count.h (profile_probability::split): New method.
20878 * dojump.c (do_jump_1) <case TRUTH_ANDIF_EXPR, case TRUTH_ORIF_EXPR>:
20879 Use profile_probability::split.
20880 (do_compare_rtx_and_jump): Fix adjustment of probabilities
20881 when splitting a single conditional jump into 2.
20883 2018-01-22 David Malcolm <dmalcolm@redhat.com>
20885 PR tree-optimization/69452
20886 * tree-ssa-loop-im.c (class move_computations_dom_walker): Remove
20889 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
20891 * config/rl78/rl78-expand.md (bswaphi2): New define_expand.
20892 * config/rl78/rl78-virt.md (*bswaphi2_virt): New define_insn.
20893 * config/rl78/rl78-real.md (*bswaphi2_real): New define_insn.
20895 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
20897 * config/rl78/rl78-protos.h (rl78_split_movdi): New function
20899 * config/rl78/rl78.md (movdi): New define_expand.
20900 * config/rl78/rl78.c (rl78_split_movdi): New function.
20902 2018-01-22 Michael Meissner <meissner@linux.vnet.ibm.com>
20905 * config/rs6000/rs6000-protos.h (rs6000_split_signbit): Delete,
20907 * config/rs6000/rs6000.c (rs6000_split_signbit): Likewise.
20908 * config/rs6000/rs6000.md (signbit<mode>2): Change code for IEEE
20909 128-bit to produce an UNSPEC move to get the double word with the
20910 signbit and then a shift directly to do signbit.
20911 (signbit<mode>2_dm): Replace old IEEE 128-bit signbit
20912 implementation with a new version that just does either a direct
20913 move or a regular move. Move memory interface to separate insns.
20914 Move insns so they are next to the expander.
20915 (signbit<mode>2_dm_mem_be): New combiner insns to combine load
20916 with signbit move. Split big and little endian case.
20917 (signbit<mode>2_dm_mem_le): Likewise.
20918 (signbit<mode>2_dm_<su>ext): Delete, no longer used.
20919 (signbit<mode>2_dm2): Likewise.
20921 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
20923 * config/rl78/rl78.md (anddi3): New define_expand.
20925 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
20927 * config/rl78/rl78.md (umindi3): New define_expand.
20929 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
20931 * config/rl78/rl78.md (smindi3): New define_expand.
20933 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
20935 * config/rl78/rl78.md (smaxdi3): New define_expand.
20937 2018-01-22 Carl Love <cel@us.ibm.com>
20939 * config/rs6000/rs6000-builtin.def (ST_ELEMREV_V1TI, LD_ELEMREV_V1TI,
20940 LVX_V1TI): Add macro expansion.
20941 * config/rs6000/rs6000-c.c (altivec_builtin_types): Add argument
20942 definitions for VSX_BUILTIN_VEC_XST_BE, VSX_BUILTIN_VEC_ST,
20943 VSX_BUILTIN_VEC_XL, LD_ELEMREV_V1TI builtins.
20944 * config/rs6000/rs6000-p8swap.c (insn_is_swappable_p);
20945 Change check to determine if the instruction is a byte reversing
20946 entry. Fix typo in comment.
20947 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case entry
20948 for VSX_BUILTIN_ST_ELEMREV_V1TI and VSX_BUILTIN_LD_ELEMREV_V1TI.
20949 Add def_builtin calls for new builtins.
20950 * config/rs6000/vsx.md (vsx_st_elemrev_v1ti, vsx_ld_elemrev_v1ti):
20951 Add define_insn expansion.
20953 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
20955 * config/rl78/rl78.md (umaxdi3): New define_expand.
20957 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
20959 * config/rl78/rl78.c (rl78_note_reg_set): Fixed dead reg check
20960 for non-QImode registers.
20962 2018-01-22 Richard Biener <rguenther@suse.de>
20964 PR tree-optimization/83963
20965 * graphite-scop-detection.c (scop_detection::get_sese): Delay
20966 including the loop exit block.
20967 (scop_detection::merge_sese): Likewise.
20968 (scop_detection::add_scop): Do it here instead.
20970 2018-01-22 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20972 * doc/sourcebuild.texi (arm_softfloat): Document.
20974 2018-01-21 John David Anglin <danglin@gcc.gnu.org>
20977 * config/pa/pa.c (pa_function_ok_for_sibcall): Use
20978 targetm.binds_local_p instead of TREE_PUBLIC to check local binding.
20979 Move TARGET_PORTABLE_RUNTIME check after TARGET_64BIT check.
20981 2018-01-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20982 David Edelsohn <dje.gcc@gmail.com>
20985 * config/rs6000/rs6000.md (*call_indirect_nonlocal_sysv<mode>):
20986 Change "crset eq" to "crset 2".
20987 (*call_value_indirect_nonlocal_sysv<mode>): Likewise.
20988 (*call_indirect_aix<mode>_nospec): Likewise.
20989 (*call_value_indirect_aix<mode>_nospec): Likewise.
20990 (*call_indirect_elfv2<mode>_nospec): Likewise.
20991 (*call_value_indirect_elfv2<mode>_nospec): Likewise.
20992 (*sibcall_nonlocal_sysv<mode>): Change "crset eq" to "crset 2";
20993 change assembly output from . to $.
20994 (*sibcall_value_nonlocal_sysv<mode>): Likewise.
20995 (indirect_jump<mode>_nospec): Change assembly output from . to $.
20996 (*tablejump<mode>_internal1_nospec): Likewise.
20998 2018-01-21 Oleg Endo <olegendo@gcc.gnu.org>
21001 * config/sh/sh_optimize_sett_clrt.cc:
21002 Use INCLUDE_ALGORITHM and INCLUDE_VECTOR instead of direct includes.
21004 2018-01-20 Richard Sandiford <richard.sandiford@linaro.org>
21006 PR tree-optimization/83940
21007 * tree-vect-stmts.c (vect_truncate_gather_scatter_offset): Set
21008 offset_dt to vect_constant_def rather than vect_unknown_def_type.
21009 (vect_check_load_store_mask): Add a mask_dt_out parameter and
21010 use it to pass back the definition type.
21011 (vect_check_store_rhs): Likewise rhs_dt_out.
21012 (vect_build_gather_load_calls): Add a mask_dt argument and use
21013 it instead of a call to vect_is_simple_use.
21014 (vectorizable_store): Update calls to vect_check_load_store_mask
21015 and vect_check_store_rhs. Use the dt returned by the latter instead
21016 of scatter_src_dt. Use the cached mask_dt and gs_info.offset_dt
21017 instead of calls to vect_is_simple_use. Pass the scalar rather
21018 than the vector operand to vect_is_simple_use when handling
21019 second and subsequent copies of an rhs value.
21020 (vectorizable_load): Update calls to vect_check_load_store_mask
21021 and vect_build_gather_load_calls. Use the cached mask_dt and
21022 gs_info.offset_dt instead of calls to vect_is_simple_use.
21024 2018-01-20 Jakub Jelinek <jakub@redhat.com>
21026 PR middle-end/83945
21027 * tree-emutls.c: Include gimplify.h.
21028 (lower_emutls_2): New function.
21029 (lower_emutls_1): If ADDR_EXPR is a gimple invariant and walk_tree
21030 with lower_emutls_2 callback finds some TLS decl in it, unshare_expr
21031 it before further processing.
21034 * simplify-rtx.c (simplify_binary_operation_1) <case UMOD>: Use
21035 UINTVAL (trueop1) instead of INTVAL (op1).
21037 2018-01-19 Jakub Jelinek <jakub@redhat.com>
21041 * dwarf2cfi.c (DEFAULT_INCOMING_FRAME_SP_OFFSET): Define to
21042 INCOMING_FRAME_SP_OFFSET if not defined.
21043 (scan_trace): Add ENTRY argument. If true and
21044 DEFAULT_INCOMING_FRAME_SP_OFFSET != INCOMING_FRAME_SP_OFFSET,
21045 emit a note to adjust the CFA offset.
21046 (create_cfi_notes): Adjust scan_trace callers.
21047 (create_cie_data): Use DEFAULT_INCOMING_FRAME_SP_OFFSET rather than
21048 INCOMING_FRAME_SP_OFFSET in the CIE.
21049 * config/i386/i386.h (DEFAULT_INCOMING_FRAME_SP_OFFSET): Define.
21050 * config/stormy16/stormy16.h (DEFAULT_INCOMING_FRAME_SP_OFFSET):
21052 * doc/tm.texi.in (DEFAULT_INCOMING_FRAME_SP_OFFSET): Document.
21053 * doc/tm.texi: Regenerated.
21055 2018-01-19 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21057 PR rtl-optimization/83147
21058 * lra-constraints.c (remove_inheritance_pseudos): Use
21059 lra_substitute_pseudo_within_insn.
21061 2018-01-19 Tom de Vries <tom@codesourcery.com>
21062 Cesar Philippidis <cesar@codesourcery.com>
21065 * config/nvptx/nvptx.c (nvptx_single): Fix jit workaround.
21067 2018-01-19 Cesar Philippidis <cesar@codesourcery.com>
21070 * config/nvptx/nvptx.c (output_init_frag): Don't use generic address
21071 spaces for function labels.
21073 2018-01-19 Martin Liska <mliska@suse.cz>
21075 * predict.def (PRED_LOOP_EXIT): Change from 85 to 89.
21076 (PRED_LOOP_EXIT_WITH_RECURSION): Change from 72 to 78.
21077 (PRED_LOOP_EXTRA_EXIT): Change from 83 to 67.
21078 (PRED_OPCODE_POSITIVE): Change from 64 to 59.
21079 (PRED_TREE_OPCODE_POSITIVE): Change from 64 to 59.
21080 (PRED_CONST_RETURN): Change from 69 to 65.
21081 (PRED_NULL_RETURN): Change from 91 to 71.
21082 (PRED_LOOP_IV_COMPARE_GUESS): Change from 98 to 64.
21083 (PRED_LOOP_GUARD): Change from 66 to 73.
21085 2018-01-19 Martin Liska <mliska@suse.cz>
21087 * predict.c (predict_insn_def): Add new assert.
21088 (struct branch_predictor): Change type to signed integer.
21089 (test_prediction_value_range): Amend test to cover
21090 PROB_UNINITIALIZED.
21091 * predict.def (PRED_LOOP_ITERATIONS): Use the new constant.
21092 (PRED_LOOP_ITERATIONS_GUESSED): Likewise.
21093 (PRED_LOOP_ITERATIONS_MAX): Likewise.
21094 (PRED_LOOP_IV_COMPARE): Likewise.
21095 * predict.h (PROB_UNINITIALIZED): Define new constant.
21097 2018-01-19 Martin Liska <mliska@suse.cz>
21099 * predict.c (dump_prediction): Add new format for
21100 analyze_brprob.py script which is enabled with -details
21102 * profile-count.h (precise_p): New function.
21104 2018-01-19 Richard Sandiford <richard.sandiford@linaro.org>
21106 PR tree-optimization/83922
21107 * tree-vect-loop.c (vect_verify_full_masking): Return false if
21108 there are no statements that need masking.
21109 (vect_active_double_reduction_p): New function.
21110 (vect_analyze_loop_operations): Use it when handling phis that
21111 are not in the loop header.
21113 2018-01-19 Richard Sandiford <richard.sandiford@linaro.org>
21115 PR tree-optimization/83914
21116 * tree-vect-loop.c (vectorizable_induction): Don't convert
21117 init_expr or apply the peeling adjustment for inductions
21118 that are nested within the vectorized loop.
21120 2018-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21122 * config/arm/thumb2.md (*thumb2_negsi2_short): Use RSB mnemonic
21125 2018-01-18 Jakub Jelinek <jakub@redhat.com>
21129 * function.h (gimplify_parameters): Add gimple_seq * argument.
21130 * function.c: Include gimple.h and options.h.
21131 (gimplify_parameters): Add cleanup argument, add CLOBBER stmts
21132 for the added local temporaries if needed.
21133 * gimplify.c (gimplify_body): Adjust gimplify_parameters caller,
21134 if there are any parameter cleanups, wrap whole body into a
21135 try/finally with the cleanups.
21137 2018-01-18 Wilco Dijkstra <wdijkstr@arm.com>
21140 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
21141 Use GET_MODE_CLASS for scalar floating point.
21143 2018-01-18 Jan Hubicka <hubicka@ucw.cz>
21147 * cgraphclones.c (cgraph_node::create_version_clone_with_body):
21148 Fix call of call_cgraph_insertion_hooks.
21150 2018-01-18 Martin Sebor <msebor@redhat.com>
21152 * doc/invoke.texi (-Wclass-memaccess): Tweak text.
21154 2018-01-18 Jan Hubicka <hubicka@ucw.cz>
21157 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Update edge
21160 2018-01-18 Boris Kolpackov <boris@codesynthesis.com>
21163 * common.opt: (-ffile-prefix-map): New option.
21164 * opts.c (common_handle_option): Defer it.
21165 * opts-global.c (handle_common_deferred_options): Handle it.
21166 * debug.h (remap_debug_filename, add_debug_prefix_map): Move to...
21167 * file-prefix-map.h: New file.
21168 (remap_debug_filename, add_debug_prefix_map): ...here.
21169 (add_macro_prefix_map, add_file_prefix_map, remap_macro_filename): New.
21170 * final.c (debug_prefix_map, add_debug_prefix_map
21171 remap_debug_filename): Move to...
21172 * file-prefix-map.c: New file.
21173 (file_prefix_map, add_prefix_map, remap_filename) ...here and rename,
21174 generalize, get rid of alloca(), use strrchr() instead of strchr().
21175 (add_macro_prefix_map, add_debug_prefix_map, add_file_prefix_map):
21176 Implement in terms of add_prefix_map().
21177 (remap_macro_filename, remap_debug_filename): Implement in term of
21179 * Makefile.in (OBJS, PLUGIN_HEADERS): Add new files.
21180 * builtins.c (fold_builtin_FILE): Call remap_macro_filename().
21181 * dbxout.c: Include file-prefix-map.h.
21182 * varasm.c: Likewise.
21183 * vmsdbgout.c: Likewise.
21184 * xcoffout.c: Likewise.
21185 * dwarf2out.c: Likewise plus omit new options from DW_AT_producer.
21186 * doc/cppopts.texi (-fmacro-prefix-map): Document.
21187 * doc/invoke.texi (-ffile-prefix-map): Document.
21188 (-fdebug-prefix-map): Update description.
21190 2018-01-18 Martin Liska <mliska@suse.cz>
21192 * config/i386/i386.c (indirect_thunk_name): Document that also
21194 (output_indirect_thunk): Document why both instructions
21195 (pause and lfence) are generated.
21197 2018-01-18 Richard Biener <rguenther@suse.de>
21199 PR tree-optimization/83887
21200 * graphite-scop-detection.c
21201 (scop_detection::get_nearest_dom_with_single_entry): Remove.
21202 (scop_detection::get_nearest_pdom_with_single_exit): Likewise.
21203 (scop_detection::merge_sese): Re-implement with a flood-fill
21204 algorithm that properly finds a SESE region if it exists.
21206 2018-01-18 Jakub Jelinek <jakub@redhat.com>
21209 * match.pd ((P + A) - P, P - (P + A), (P + A) - (P + B)): For
21210 pointer_diff optimizations use view_convert instead of convert.
21212 2018-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21214 * config/rs6000/rs6000.md (*call_indirect_nonlocal_sysv<mode>):
21215 Generate different code for -mno-speculate-indirect-jumps.
21216 (*call_value_indirect_nonlocal_sysv<mode>): Likewise.
21217 (*call_indirect_aix<mode>): Disable for
21218 -mno-speculate-indirect-jumps.
21219 (*call_indirect_aix<mode>_nospec): New define_insn.
21220 (*call_value_indirect_aix<mode>): Disable for
21221 -mno-speculate-indirect-jumps.
21222 (*call_value_indirect_aix<mode>_nospec): New define_insn.
21223 (*sibcall_nonlocal_sysv<mode>): Generate different code for
21224 -mno-speculate-indirect-jumps.
21225 (*sibcall_value_nonlocal_sysv<mode>): Likewise.
21227 2018-01-17 Michael Meissner <meissner@linux.vnet.ibm.com>
21229 * config/rs6000/rs6000.c (rs6000_emit_move): If we load or store a
21230 long double type, set the flags for noting the default long double
21231 type, even if we don't pass or return a long double type.
21233 2018-01-17 Jan Hubicka <hubicka@ucw.cz>
21236 * ipa-inline.c (flatten_function): Do not overwrite final inlining
21239 2018-01-17 Will Schmidt <will_schmidt@vnet.ibm.com>
21241 * config/rs6000/rs6000.c (rs6000_gimple_builtin): Add gimple folding
21242 support for merge[hl].
21243 (fold_mergehl_helper): New helper function.
21244 (tree-vector-builder.h): New #include for tree_vector_builder usage.
21245 * config/rs6000/altivec.md (altivec_vmrghw_direct): Add xxmrghw insn.
21246 (altivec_vmrglw_direct): Add xxmrglw insn.
21248 2018-01-17 Andrew Waterman <andrew@sifive.com>
21250 * config/riscv/riscv.c (riscv_conditional_register_usage): If
21251 UNITS_PER_FP_ARG is 0, set call_used_regs to 1 for all FP regs.
21253 2018-01-17 David Malcolm <dmalcolm@redhat.com>
21256 * ipa-devirt.c (add_type_duplicate): When comparing memory layout,
21257 call the lto_location_cache before reading the
21258 DECL_SOURCE_LOCATION of the types.
21260 2018-01-17 Wilco Dijkstra <wdijkstr@arm.com>
21261 Richard Sandiford <richard.sandiford@linaro.org>
21263 * config/aarch64/aarch64.md (movti_aarch64): Use Uti constraint.
21264 * config/aarch64/aarch64.c (aarch64_mov128_immediate): New function.
21265 (aarch64_legitimate_constant_p): Just support CONST_DOUBLE
21266 SF/DF/TF mode to avoid creating illegal CONST_WIDE_INT immediates.
21267 * config/aarch64/aarch64-protos.h (aarch64_mov128_immediate):
21269 * config/aarch64/constraints.md (aarch64_movti_operand):
21271 * config/aarch64/predicates.md (Uti): Add new constraint.
21273 2018-01-17 Carl Love <cel@us.ibm.com>
21275 * config/rs6000/vsx.md (define_expand xl_len_r,
21276 define_expand stxvl, define_expand *stxvl): Add match_dup argument.
21277 (define_insn): Add, match_dup 1 argument to define_insn stxvll and
21279 (define_expand, define_insn): Move the shift left from the
21280 define_insn to the define_expand for lxvl and stxvl instructions.
21281 * config/rs6000/rs6000-builtin.def (BU_P9V_64BIT_VSX_2): Change LXVL
21282 and XL_LEN_R definitions to PURE.
21284 2018-01-17 Uros Bizjak <ubizjak@gmail.com>
21286 * config/i386/i386.c (indirect_thunk_name): Declare regno
21287 as unsigned int. Compare regno with INVALID_REGNUM.
21288 (output_indirect_thunk): Ditto.
21289 (output_indirect_thunk_function): Ditto.
21290 (ix86_code_end): Declare regno as unsigned int. Use INVALID_REGNUM
21291 in the call to output_indirect_thunk_function.
21293 2018-01-17 Richard Sandiford <richard.sandiford@linaro.org>
21295 PR middle-end/83884
21296 * expr.c (expand_expr_real_1): Use the size of GET_MODE (op0)
21297 rather than the size of inner_type to determine the stack slot size
21298 when handling VIEW_CONVERT_EXPRs on strict-alignment targets.
21300 2018-01-16 Sebastian Peryt <sebastian.peryt@intel.com>
21303 * config/i386/i386.c (ix86_option_override_internal): Add PTA_RDRND
21306 2018-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
21308 * config.gcc (powerpc*-linux*-*): Add support for 64-bit little
21309 endian Linux systems to optionally enable multilibs for selecting
21310 the long double type if the user configured an explicit type.
21311 * config/rs6000/rs6000.h (TARGET_IEEEQUAD_MULTILIB): Indicate we
21312 have no long double multilibs if not defined.
21313 * config/rs6000/rs6000.c (rs6000_option_override_internal): Do not
21314 warn if the user used -mabi={ieee,ibm}longdouble and we built
21315 multilibs for long double.
21316 * config/rs6000/linux64.h (MULTILIB_DEFAULTS_IEEE): Define as the
21317 appropriate multilib option.
21318 (MULTILIB_DEFAULTS): Add MULTILIB_DEFAULTS_IEEE to the default
21320 * config/rs6000/t-ldouble-linux64le-ibm: New configuration files
21321 for building long double multilibs.
21322 * config/rs6000/t-ldouble-linux64le-ieee: Likewise.
21324 2018-01-16 John David Anglin <danglin@gcc.gnu.org>
21326 * config.gcc (hppa*-*-linux*): Change callee copies ABI to caller
21329 * config/pa.h (MALLOC_ABI_ALIGNMENT): Set 32-bit alignment default to
21331 * config/pa/pa32-linux.h (MALLOC_ABI_ALIGNMENT): Set alignment to
21334 * config/pa/som.h (ASM_DECLARE_FUNCTION_NAME): Cleanup type and mode
21337 * config/pa/pa.c (pa_function_arg_size): Apply CEIL to GET_MODE_SIZE
21340 2018-01-16 Eric Botcazou <ebotcazou@adacore.com>
21342 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): For an
21343 ADDR_EXPR, do not count the offset of a COMPONENT_REF twice.
21345 2018-01-16 Kelvin Nilsen <kelvin@gcc.gnu.org>
21347 * config/rs6000/rs6000-p8swap.c (rs6000_gen_stvx): Generate
21348 different rtl trees depending on TARGET_64BIT.
21349 (rs6000_gen_lvx): Likewise.
21351 2018-01-16 Eric Botcazou <ebotcazou@adacore.com>
21353 * config/visium/visium.md (nop): Tweak comment.
21354 (hazard_nop): Likewise.
21356 2018-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21358 * config/rs6000/rs6000.c (rs6000_opt_vars): Add entry for
21359 -mspeculate-indirect-jumps.
21360 * config/rs6000/rs6000.md (*call_indirect_elfv2<mode>): Disable
21361 for -mno-speculate-indirect-jumps.
21362 (*call_indirect_elfv2<mode>_nospec): New define_insn.
21363 (*call_value_indirect_elfv2<mode>): Disable for
21364 -mno-speculate-indirect-jumps.
21365 (*call_value_indirect_elfv2<mode>_nospec): New define_insn.
21366 (indirect_jump): Emit different RTL for
21367 -mno-speculate-indirect-jumps.
21368 (*indirect_jump<mode>): Disable for
21369 -mno-speculate-indirect-jumps.
21370 (*indirect_jump<mode>_nospec): New define_insn.
21371 (tablejump): Emit different RTL for
21372 -mno-speculate-indirect-jumps.
21373 (tablejumpsi): Disable for -mno-speculate-indirect-jumps.
21374 (tablejumpsi_nospec): New define_expand.
21375 (tablejumpdi): Disable for -mno-speculate-indirect-jumps.
21376 (tablejumpdi_nospec): New define_expand.
21377 (*tablejump<mode>_internal1): Disable for
21378 -mno-speculate-indirect-jumps.
21379 (*tablejump<mode>_internal1_nospec): New define_insn.
21380 * config/rs6000/rs6000.opt (mspeculate-indirect-jumps): New
21383 2018-01-16 Artyom Skrobov tyomitch@gmail.com
21385 * caller-save.c (insert_save): Drop unnecessary parameter. All
21388 2018-01-16 Jakub Jelinek <jakub@redhat.com>
21389 Richard Biener <rguenth@suse.de>
21392 * gimplify.c (gimplify_one_sizepos): For is_gimple_constant (expr)
21393 return early, inline manually is_gimple_sizepos. Make sure if we
21394 call gimplify_expr we don't end up with a gimple constant.
21395 * tree.c (variably_modified_type_p): Don't return true for
21396 is_gimple_constant (_t). Inline manually is_gimple_sizepos.
21397 * gimplify.h (is_gimple_sizepos): Remove.
21399 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
21401 PR tree-optimization/83857
21402 * tree-vect-loop.c (vect_analyze_loop_operations): Don't call
21403 vectorizable_live_operation for pure SLP statements.
21404 (vectorizable_live_operation): Handle PHIs.
21406 2018-01-16 Richard Biener <rguenther@suse.de>
21408 PR tree-optimization/83867
21409 * tree-vect-stmts.c (vect_transform_stmt): Precompute
21410 nested_in_vect_loop_p since the scalar stmt may get invalidated.
21412 2018-01-16 Jakub Jelinek <jakub@redhat.com>
21415 * stor-layout.c (handle_warn_if_not_align): Use byte_position and
21416 multiple_of_p instead of unchecked tree_to_uhwi and UHWI check.
21417 If off is not INTEGER_CST, issue a may not be aligned warning
21418 rather than isn't aligned. Use isn%'t rather than isn't.
21419 * fold-const.c (multiple_of_p) <case BIT_AND_EXPR>: Don't fall through
21421 <case MULT_EXPR>: Improve the case when bottom and one of the
21422 MULT_EXPR operands are INTEGER_CSTs and bottom is multiple of that
21423 operand, in that case check if the other operand is multiple of
21424 bottom divided by the INTEGER_CST operand.
21426 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
21429 * config/pa/pa.h (FUNCTION_ARG_SIZE): Delete.
21430 * config/pa/pa-protos.h (pa_function_arg_size): Declare.
21431 * config/pa/som.h (ASM_DECLARE_FUNCTION_NAME): Use
21432 pa_function_arg_size instead of FUNCTION_ARG_SIZE.
21433 * config/pa/pa.c (pa_function_arg_advance): Likewise.
21434 (pa_function_arg, pa_arg_partial_bytes): Likewise.
21435 (pa_function_arg_size): New function.
21437 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
21439 * fold-const.c (fold_ternary_loc): Construct the vec_perm_indices
21440 in a separate statement.
21442 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
21444 PR tree-optimization/83847
21445 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Don't
21446 group gathers and scatters.
21448 2018-01-16 Jakub Jelinek <jakub@redhat.com>
21450 PR rtl-optimization/86620
21451 * params.def (max-sched-ready-insns): Bump minimum value to 1.
21453 PR rtl-optimization/83213
21454 * recog.c (peep2_attempt): Copy over CROSSING_JUMP_P from peepinsn
21455 to last if both are JUMP_INSNs.
21457 PR tree-optimization/83843
21458 * gimple-ssa-store-merging.c
21459 (imm_store_chain_info::output_merged_store): Handle bit_not_p on
21460 store_immediate_info for bswap/nop orig_stores.
21462 2018-01-15 Andrew Waterman <andrew@sifive.com>
21464 * config/riscv/riscv.c (riscv_rtx_costs) <MULT>: Increase cost if
21466 <UDIV>: Increase cost if !TARGET_DIV.
21468 2018-01-15 Segher Boessenkool <segher@kernel.crashing.org>
21470 * config/rs6000/rs6000.md (define_attr "type"): Remove delayed_cr.
21471 (define_attr "cr_logical_3op"): New.
21472 (cceq_ior_compare): Adjust.
21473 (cceq_ior_compare_complement): Adjust.
21474 (*cceq_rev_compare): Adjust.
21475 * config/rs6000/rs6000.c (rs6000_adjust_cost): Adjust.
21476 (is_cracked_insn): Adjust.
21477 (insn_must_be_first_in_group): Adjust.
21478 * config/rs6000/40x.md: Adjust.
21479 * config/rs6000/440.md: Adjust.
21480 * config/rs6000/476.md: Adjust.
21481 * config/rs6000/601.md: Adjust.
21482 * config/rs6000/603.md: Adjust.
21483 * config/rs6000/6xx.md: Adjust.
21484 * config/rs6000/7450.md: Adjust.
21485 * config/rs6000/7xx.md: Adjust.
21486 * config/rs6000/8540.md: Adjust.
21487 * config/rs6000/cell.md: Adjust.
21488 * config/rs6000/e300c2c3.md: Adjust.
21489 * config/rs6000/e500mc.md: Adjust.
21490 * config/rs6000/e500mc64.md: Adjust.
21491 * config/rs6000/e5500.md: Adjust.
21492 * config/rs6000/e6500.md: Adjust.
21493 * config/rs6000/mpc.md: Adjust.
21494 * config/rs6000/power4.md: Adjust.
21495 * config/rs6000/power5.md: Adjust.
21496 * config/rs6000/power6.md: Adjust.
21497 * config/rs6000/power7.md: Adjust.
21498 * config/rs6000/power8.md: Adjust.
21499 * config/rs6000/power9.md: Adjust.
21500 * config/rs6000/rs64.md: Adjust.
21501 * config/rs6000/titan.md: Adjust.
21503 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
21505 * config/i386/predicates.md (indirect_branch_operand): Rewrite
21506 ix86_indirect_branch_register logic.
21508 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
21510 * config/i386/constraints.md (Bs): Update
21511 ix86_indirect_branch_register check. Don't check
21512 ix86_indirect_branch_register with GOT_memory_operand.
21514 * config/i386/predicates.md (GOT_memory_operand): Don't check
21515 ix86_indirect_branch_register here.
21516 (GOT32_symbol_operand): Likewise.
21518 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
21520 * config/i386/predicates.md (constant_call_address_operand):
21521 Rewrite ix86_indirect_branch_register logic.
21522 (sibcall_insn_operand): Likewise.
21524 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
21526 * config/i386/constraints.md (Bs): Replace
21527 ix86_indirect_branch_thunk_register with
21528 ix86_indirect_branch_register.
21530 * config/i386/i386.md (indirect_jump): Likewise.
21531 (tablejump): Likewise.
21532 (*sibcall_memory): Likewise.
21533 (*sibcall_value_memory): Likewise.
21534 Peepholes of indirect call and jump via memory: Likewise.
21535 * config/i386/i386.opt: Likewise.
21536 * config/i386/predicates.md (indirect_branch_operand): Likewise.
21537 (GOT_memory_operand): Likewise.
21538 (call_insn_operand): Likewise.
21539 (sibcall_insn_operand): Likewise.
21540 (GOT32_symbol_operand): Likewise.
21542 2018-01-15 Jakub Jelinek <jakub@redhat.com>
21544 PR middle-end/83837
21545 * omp-expand.c (expand_omp_atomic_pipeline): Use loaded_val
21546 type rather than type addr's type points to.
21547 (expand_omp_atomic_mutex): Likewise.
21548 (expand_omp_atomic): Likewise.
21550 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
21553 * config/i386/i386.c (output_indirect_thunk_function): Use
21554 ASM_OUTPUT_LABEL, instead of ASM_OUTPUT_DEF, for TARGET_MACHO
21555 for __x86_return_thunk.
21557 2018-01-15 Richard Biener <rguenther@suse.de>
21559 PR middle-end/83850
21560 * expmed.c (extract_bit_field_1): Fix typo.
21562 2018-01-15 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21565 * config/arm/iterators.md (VF): New mode iterator.
21566 * config/arm/neon.md (neon_vabd<mode>_2): Use the above.
21567 Remove integer-related logic from pattern.
21568 (neon_vabd<mode>_3): Likewise.
21570 2018-01-15 Jakub Jelinek <jakub@redhat.com>
21572 PR middle-end/82694
21573 * common.opt (fstrict-overflow): No longer an alias.
21574 (fwrapv-pointer): New option.
21575 * tree.h (TYPE_OVERFLOW_WRAPS, TYPE_OVERFLOW_UNDEFINED): Define
21576 also for pointer types based on flag_wrapv_pointer.
21577 * opts.c (common_handle_option) <case OPT_fstrict_overflow>: Set
21578 opts->x_flag_wrap[pv] to !value, clear opts->x_flag_trapv if
21579 opts->x_flag_wrapv got set.
21580 * fold-const.c (fold_comparison, fold_binary_loc): Revert 2017-08-01
21581 changes, just use TYPE_OVERFLOW_UNDEFINED on pointer type instead of
21582 POINTER_TYPE_OVERFLOW_UNDEFINED.
21583 * match.pd: Likewise in address comparison pattern.
21584 * doc/invoke.texi: Document -fwrapv and -fstrict-overflow.
21586 2018-01-15 Richard Biener <rguenther@suse.de>
21589 * tree.c (free_lang_data_in_type): Always unlink TYPE_DECLs
21590 from TYPE_FIELDS. Free TYPE_BINFO if not used by devirtualization.
21591 Reset type names to their identifier if their TYPE_DECL doesn't
21592 have linkage (and thus is used for ODR and devirt).
21593 (save_debug_info_for_decl): Remove.
21594 (save_debug_info_for_type): Likewise.
21595 (add_tree_to_fld_list): Adjust.
21596 * tree-pretty-print.c (dump_generic_node): Make dumping of
21597 type names more robust.
21599 2018-01-15 Richard Biener <rguenther@suse.de>
21601 * BASE-VER: Bump to 8.0.1.
21603 2018-01-14 Martin Sebor <msebor@redhat.com>
21606 * builtins.c (check_access): Avoid warning when the no-warning bit
21609 2018-01-14 Cory Fields <cory-nospam-@coryfields.com>
21611 * tree-ssa-loop-im.c (sort_bbs_in_loop_postorder_cmp): Stabilize sort.
21612 * ira-color (allocno_hard_regs_compare): Likewise.
21614 2018-01-14 Nathan Rossi <nathan@nathanrossi.com>
21617 * config/microblaze/microblaze.c (microblaze_asm_output_ident):
21618 Use .pushsection/.popsection.
21620 2018-01-14 Martin Sebor <msebor@redhat.com>
21623 * doc/invoke.texi (-Wlass-memaccess): Document suppression by casting.
21625 2018-01-14 Jakub Jelinek <jakub@redhat.com>
21627 * config.gcc (i[34567]86-*-*): Remove one duplicate gfniintrin.h
21628 entry from extra_headers.
21629 (x86_64-*-*): Remove two duplicate gfniintrin.h entries from
21630 extra_headers, make the list bitwise identical to the i?86-*-* one.
21632 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
21634 * config/i386/i386.c (ix86_set_indirect_branch_type): Disallow
21635 -mcmodel=large with -mindirect-branch=thunk,
21636 -mindirect-branch=thunk-extern, -mfunction-return=thunk and
21637 -mfunction-return=thunk-extern.
21638 * doc/invoke.texi: Document -mcmodel=large is incompatible with
21639 -mindirect-branch=thunk, -mindirect-branch=thunk-extern,
21640 -mfunction-return=thunk and -mfunction-return=thunk-extern.
21642 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
21644 * config/i386/i386.c (print_reg): Print the name of the full
21645 integer register without '%'.
21646 (ix86_print_operand): Handle 'V'.
21647 * doc/extend.texi: Document 'V' modifier.
21649 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
21651 * config/i386/constraints.md (Bs): Disallow memory operand for
21652 -mindirect-branch-register.
21654 * config/i386/predicates.md (indirect_branch_operand): Likewise.
21655 (GOT_memory_operand): Likewise.
21656 (call_insn_operand): Likewise.
21657 (sibcall_insn_operand): Likewise.
21658 (GOT32_symbol_operand): Likewise.
21659 * config/i386/i386.md (indirect_jump): Call convert_memory_address
21660 for -mindirect-branch-register.
21661 (tablejump): Likewise.
21662 (*sibcall_memory): Likewise.
21663 (*sibcall_value_memory): Likewise.
21664 Disallow peepholes of indirect call and jump via memory for
21665 -mindirect-branch-register.
21666 (*call_pop): Replace m with Bw.
21667 (*call_value_pop): Likewise.
21668 (*sibcall_pop_memory): Replace m with Bs.
21669 * config/i386/i386.opt (mindirect-branch-register): New option.
21670 * doc/invoke.texi: Document -mindirect-branch-register option.
21672 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
21674 * config/i386/i386-protos.h (ix86_output_function_return): New.
21675 * config/i386/i386.c (ix86_set_indirect_branch_type): Also
21676 set function_return_type.
21677 (indirect_thunk_name): Add ret_p to indicate thunk for function
21679 (output_indirect_thunk_function): Pass false to
21680 indirect_thunk_name.
21681 (ix86_output_indirect_branch_via_reg): Likewise.
21682 (ix86_output_indirect_branch_via_push): Likewise.
21683 (output_indirect_thunk_function): Create alias for function
21684 return thunk if regno < 0.
21685 (ix86_output_function_return): New function.
21686 (ix86_handle_fndecl_attribute): Handle function_return.
21687 (ix86_attribute_table): Add function_return.
21688 * config/i386/i386.h (machine_function): Add
21689 function_return_type.
21690 * config/i386/i386.md (simple_return_internal): Use
21691 ix86_output_function_return.
21692 (simple_return_internal_long): Likewise.
21693 * config/i386/i386.opt (mfunction-return=): New option.
21694 (indirect_branch): Mention -mfunction-return=.
21695 * doc/extend.texi: Document function_return function attribute.
21696 * doc/invoke.texi: Document -mfunction-return= option.
21698 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
21700 * config/i386/i386-opts.h (indirect_branch): New.
21701 * config/i386/i386-protos.h (ix86_output_indirect_jmp): Likewise.
21702 * config/i386/i386.c (ix86_using_red_zone): Disallow red-zone
21703 with local indirect jump when converting indirect call and jump.
21704 (ix86_set_indirect_branch_type): New.
21705 (ix86_set_current_function): Call ix86_set_indirect_branch_type.
21706 (indirectlabelno): New.
21707 (indirect_thunk_needed): Likewise.
21708 (indirect_thunk_bnd_needed): Likewise.
21709 (indirect_thunks_used): Likewise.
21710 (indirect_thunks_bnd_used): Likewise.
21711 (INDIRECT_LABEL): Likewise.
21712 (indirect_thunk_name): Likewise.
21713 (output_indirect_thunk): Likewise.
21714 (output_indirect_thunk_function): Likewise.
21715 (ix86_output_indirect_branch_via_reg): Likewise.
21716 (ix86_output_indirect_branch_via_push): Likewise.
21717 (ix86_output_indirect_branch): Likewise.
21718 (ix86_output_indirect_jmp): Likewise.
21719 (ix86_code_end): Call output_indirect_thunk_function if needed.
21720 (ix86_output_call_insn): Call ix86_output_indirect_branch if
21722 (ix86_handle_fndecl_attribute): Handle indirect_branch.
21723 (ix86_attribute_table): Add indirect_branch.
21724 * config/i386/i386.h (machine_function): Add indirect_branch_type
21725 and has_local_indirect_jump.
21726 * config/i386/i386.md (indirect_jump): Set has_local_indirect_jump
21728 (tablejump): Likewise.
21729 (*indirect_jump): Use ix86_output_indirect_jmp.
21730 (*tablejump_1): Likewise.
21731 (simple_return_indirect_internal): Likewise.
21732 * config/i386/i386.opt (mindirect-branch=): New option.
21733 (indirect_branch): New.
21736 (thunk-inline): Likewise.
21737 (thunk-extern): Likewise.
21738 * doc/extend.texi: Document indirect_branch function attribute.
21739 * doc/invoke.texi: Document -mindirect-branch= option.
21741 2018-01-14 Jan Hubicka <hubicka@ucw.cz>
21744 * ipa-inline.c (edge_badness): Tolerate roundoff errors.
21746 2018-01-14 Richard Sandiford <richard.sandiford@linaro.org>
21748 * ipa-inline.c (want_inline_small_function_p): Return false if
21749 inlining has already failed with CIF_FINAL_ERROR.
21750 (update_caller_keys): Call want_inline_small_function_p before
21752 (update_callee_keys): Likewise.
21754 2018-01-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
21756 * config/rs6000/rs6000-p8swap.c (rs6000_sum_of_two_registers_p):
21758 (rs6000_quadword_masked_address_p): Likewise.
21759 (quad_aligned_load_p): Likewise.
21760 (quad_aligned_store_p): Likewise.
21761 (const_load_sequence_p): Add comment to describe the outer-most loop.
21762 (mimic_memory_attributes_and_flags): New function.
21763 (rs6000_gen_stvx): Likewise.
21764 (replace_swapped_aligned_store): Likewise.
21765 (rs6000_gen_lvx): Likewise.
21766 (replace_swapped_aligned_load): Likewise.
21767 (replace_swapped_load_constant): Capitalize argument name in
21768 comment describing this function.
21769 (rs6000_analyze_swaps): Add a third pass to search for vector loads
21770 and stores that access quad-word aligned addresses and replace
21771 with stvx or lvx instructions when appropriate.
21772 * config/rs6000/rs6000-protos.h (rs6000_sum_of_two_registers_p):
21773 New function prototype.
21774 (rs6000_quadword_masked_address_p): Likewise.
21775 (rs6000_gen_lvx): Likewise.
21776 (rs6000_gen_stvx): Likewise.
21777 * config/rs6000/vsx.md (*vsx_le_perm_load_<mode>): For modes
21778 VSX_D (V2DF, V2DI), modify this split to select lvx instruction
21779 when memory address is aligned.
21780 (*vsx_le_perm_load_<mode>): For modes VSX_W (V4SF, V4SI), modify
21781 this split to select lvx instruction when memory address is aligned.
21782 (*vsx_le_perm_load_v8hi): Modify this split to select lvx
21783 instruction when memory address is aligned.
21784 (*vsx_le_perm_load_v16qi): Likewise.
21785 (four unnamed splitters): Modify to select the stvx instruction
21786 when memory is aligned.
21788 2018-01-13 Jan Hubicka <hubicka@ucw.cz>
21790 * predict.c (determine_unlikely_bbs): Handle correctly BBs
21791 which appears in the queue multiple times.
21793 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
21794 Alan Hayward <alan.hayward@arm.com>
21795 David Sherwood <david.sherwood@arm.com>
21797 * tree-vectorizer.h (vec_lower_bound): New structure.
21798 (_loop_vec_info): Add check_nonzero and lower_bounds.
21799 (LOOP_VINFO_CHECK_NONZERO): New macro.
21800 (LOOP_VINFO_LOWER_BOUNDS): Likewise.
21801 (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Check lower_bounds too.
21802 * tree-data-ref.h (dr_with_seg_len): Add access_size and align
21803 fields. Make seg_len the distance travelled, not including the
21805 (dr_direction_indicator): Declare.
21806 (dr_zero_step_indicator): Likewise.
21807 (dr_known_forward_stride_p): Likewise.
21808 * tree-data-ref.c: Include stringpool.h, tree-vrp.h and
21810 (runtime_alias_check_p): Allow runtime alias checks with
21812 (operator ==): Compare access_size and align.
21813 (prune_runtime_alias_test_list): Rework for new distinction between
21814 the access_size and seg_len.
21815 (create_intersect_range_checks_index): Likewise. Cope with polynomial
21817 (get_segment_min_max): New function.
21818 (create_intersect_range_checks): Use it.
21819 (dr_step_indicator): New function.
21820 (dr_direction_indicator): Likewise.
21821 (dr_zero_step_indicator): Likewise.
21822 (dr_known_forward_stride_p): Likewise.
21823 * tree-loop-distribution.c (data_ref_segment_size): Return
21824 DR_STEP * (niters - 1).
21825 (compute_alias_check_pairs): Update call to the dr_with_seg_len
21827 * tree-vect-data-refs.c (vect_check_nonzero_value): New function.
21828 (vect_preserves_scalar_order_p): New function, split out from...
21829 (vect_analyze_data_ref_dependence): ...here. Check for zero steps.
21830 (vect_vfa_segment_size): Return DR_STEP * (length_factor - 1).
21831 (vect_vfa_access_size): New function.
21832 (vect_vfa_align): Likewise.
21833 (vect_compile_time_alias): Take access_size_a and access_b arguments.
21834 (dump_lower_bound): New function.
21835 (vect_check_lower_bound): Likewise.
21836 (vect_small_gap_p): Likewise.
21837 (vectorizable_with_step_bound_p): Likewise.
21838 (vect_prune_runtime_alias_test_list): Ignore cross-iteration
21839 depencies if the vectorization factor is 1. Convert the checks
21840 for nonzero steps into checks on the bounds of DR_STEP. Try using
21841 a bunds check for variable steps if the minimum required step is
21842 relatively small. Update calls to the dr_with_seg_len
21843 constructor and to vect_compile_time_alias.
21844 * tree-vect-loop-manip.c (vect_create_cond_for_lower_bounds): New
21846 (vect_loop_versioning): Call it.
21847 * tree-vect-loop.c (vect_analyze_loop_2): Clear LOOP_VINFO_LOWER_BOUNDS
21849 (vect_estimate_min_profitable_iters): Account for any bounds checks.
21851 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
21852 Alan Hayward <alan.hayward@arm.com>
21853 David Sherwood <david.sherwood@arm.com>
21855 * doc/sourcebuild.texi (vect_scatter_store): Document.
21856 * optabs.def (scatter_store_optab, mask_scatter_store_optab): New
21858 * doc/md.texi (scatter_store@var{m}, mask_scatter_store@var{m}):
21860 * genopinit.c (main): Add supports_vec_scatter_store and
21861 supports_vec_scatter_store_cached to target_optabs.
21862 * gimple.h (gimple_expr_type): Handle IFN_SCATTER_STORE and
21863 IFN_MASK_SCATTER_STORE.
21864 * internal-fn.def (SCATTER_STORE, MASK_SCATTER_STORE): New internal
21866 * internal-fn.h (internal_store_fn_p): Declare.
21867 (internal_fn_stored_value_index): Likewise.
21868 * internal-fn.c (scatter_store_direct): New macro.
21869 (expand_scatter_store_optab_fn): New function.
21870 (direct_scatter_store_optab_supported_p): New macro.
21871 (internal_store_fn_p): New function.
21872 (internal_gather_scatter_fn_p): Handle IFN_SCATTER_STORE and
21873 IFN_MASK_SCATTER_STORE.
21874 (internal_fn_mask_index): Likewise.
21875 (internal_fn_stored_value_index): New function.
21876 (internal_gather_scatter_fn_supported_p): Adjust operand numbers
21877 for scatter stores.
21878 * optabs-query.h (supports_vec_scatter_store_p): Declare.
21879 * optabs-query.c (supports_vec_scatter_store_p): New function.
21880 * tree-vectorizer.h (vect_get_store_rhs): Declare.
21881 * tree-vect-data-refs.c (vect_analyze_data_ref_access): Return
21882 true for scatter stores.
21883 (vect_gather_scatter_fn_p): Handle scatter stores too.
21884 (vect_check_gather_scatter): Consider using scatter stores if
21885 supports_vec_scatter_store_p.
21886 * tree-vect-patterns.c (vect_try_gather_scatter_pattern): Handle
21887 scatter stores too.
21888 * tree-vect-stmts.c (exist_non_indexing_operands_for_use_p): Use
21889 internal_fn_stored_value_index.
21890 (check_load_store_masking): Handle scatter stores too.
21891 (vect_get_store_rhs): Make public.
21892 (vectorizable_call): Use internal_store_fn_p.
21893 (vectorizable_store): Handle scatter store internal functions.
21894 (vect_transform_stmt): Compare GROUP_STORE_COUNT with GROUP_SIZE
21895 when deciding whether the end of the group has been reached.
21896 * config/aarch64/aarch64.md (UNSPEC_ST1_SCATTER): New unspec.
21897 * config/aarch64/aarch64-sve.md (scatter_store<mode>): New expander.
21898 (mask_scatter_store<mode>): New insns.
21900 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
21901 Alan Hayward <alan.hayward@arm.com>
21902 David Sherwood <david.sherwood@arm.com>
21904 * tree-vectorizer.h (vect_gather_scatter_fn_p): Declare.
21905 * tree-vect-data-refs.c (vect_gather_scatter_fn_p): Make public.
21906 * tree-vect-stmts.c (vect_truncate_gather_scatter_offset): New
21908 (vect_use_strided_gather_scatters_p): Take a masked_p argument.
21909 Use vect_truncate_gather_scatter_offset if we can't treat the
21910 operation as a normal gather load or scatter store.
21911 (get_group_load_store_type): Take the gather_scatter_info
21912 as argument. Try using a gather load or scatter store for
21913 single-element groups.
21914 (get_load_store_type): Update calls to get_group_load_store_type
21915 and vect_use_strided_gather_scatters_p.
21917 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
21918 Alan Hayward <alan.hayward@arm.com>
21919 David Sherwood <david.sherwood@arm.com>
21921 * tree-vectorizer.h (vect_create_data_ref_ptr): Take an extra
21922 optional tree argument.
21923 * tree-vect-data-refs.c (vect_check_gather_scatter): Check for
21925 (vect_create_data_ref_ptr): Take the iv_step as an optional argument,
21926 but continue to use the current value as a fallback.
21927 (bump_vector_ptr): Use operand_equal_p rather than tree_int_cst_compare
21928 to compare the updates.
21929 * tree-vect-stmts.c (vect_use_strided_gather_scatters_p): New function.
21930 (get_load_store_type): Use it when handling a strided access.
21931 (vect_get_strided_load_store_ops): New function.
21932 (vect_get_data_ptr_increment): Likewise.
21933 (vectorizable_load): Handle strided gather loads. Always pass
21934 a step to vect_create_data_ref_ptr and bump_vector_ptr.
21936 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
21937 Alan Hayward <alan.hayward@arm.com>
21938 David Sherwood <david.sherwood@arm.com>
21940 * doc/md.texi (gather_load@var{m}): Document.
21941 (mask_gather_load@var{m}): Likewise.
21942 * genopinit.c (main): Add supports_vec_gather_load and
21943 supports_vec_gather_load_cached to target_optabs.
21944 * optabs-tree.c (init_tree_optimization_optabs): Use
21945 ggc_cleared_alloc to allocate target_optabs.
21946 * optabs.def (gather_load_optab, mask_gather_laod_optab): New optabs.
21947 * internal-fn.def (GATHER_LOAD, MASK_GATHER_LOAD): New internal
21949 * internal-fn.h (internal_load_fn_p): Declare.
21950 (internal_gather_scatter_fn_p): Likewise.
21951 (internal_fn_mask_index): Likewise.
21952 (internal_gather_scatter_fn_supported_p): Likewise.
21953 * internal-fn.c (gather_load_direct): New macro.
21954 (expand_gather_load_optab_fn): New function.
21955 (direct_gather_load_optab_supported_p): New macro.
21956 (direct_internal_fn_optab): New function.
21957 (internal_load_fn_p): Likewise.
21958 (internal_gather_scatter_fn_p): Likewise.
21959 (internal_fn_mask_index): Likewise.
21960 (internal_gather_scatter_fn_supported_p): Likewise.
21961 * optabs-query.c (supports_at_least_one_mode_p): New function.
21962 (supports_vec_gather_load_p): Likewise.
21963 * optabs-query.h (supports_vec_gather_load_p): Declare.
21964 * tree-vectorizer.h (gather_scatter_info): Add ifn, element_type
21965 and memory_type field.
21966 (NUM_PATTERNS): Bump to 15.
21967 * tree-vect-data-refs.c: Include internal-fn.h.
21968 (vect_gather_scatter_fn_p): New function.
21969 (vect_describe_gather_scatter_call): Likewise.
21970 (vect_check_gather_scatter): Try using internal functions for
21971 gather loads. Recognize existing calls to a gather load function.
21972 (vect_analyze_data_refs): Consider using gather loads if
21973 supports_vec_gather_load_p.
21974 * tree-vect-patterns.c (vect_get_load_store_mask): New function.
21975 (vect_get_gather_scatter_offset_type): Likewise.
21976 (vect_convert_mask_for_vectype): Likewise.
21977 (vect_add_conversion_to_patterm): Likewise.
21978 (vect_try_gather_scatter_pattern): Likewise.
21979 (vect_recog_gather_scatter_pattern): New pattern recognizer.
21980 (vect_vect_recog_func_ptrs): Add it.
21981 * tree-vect-stmts.c (exist_non_indexing_operands_for_use_p): Use
21982 internal_fn_mask_index and internal_gather_scatter_fn_p.
21983 (check_load_store_masking): Take the gather_scatter_info as an
21984 argument and handle gather loads.
21985 (vect_get_gather_scatter_ops): New function.
21986 (vectorizable_call): Check internal_load_fn_p.
21987 (vectorizable_load): Likewise. Handle gather load internal
21989 (vectorizable_store): Update call to check_load_store_masking.
21990 * config/aarch64/aarch64.md (UNSPEC_LD1_GATHER): New unspec.
21991 * config/aarch64/iterators.md (SVE_S, SVE_D): New mode iterators.
21992 * config/aarch64/predicates.md (aarch64_gather_scale_operand_w)
21993 (aarch64_gather_scale_operand_d): New predicates.
21994 * config/aarch64/aarch64-sve.md (gather_load<mode>): New expander.
21995 (mask_gather_load<mode>): New insns.
21997 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
21998 Alan Hayward <alan.hayward@arm.com>
21999 David Sherwood <david.sherwood@arm.com>
22001 * optabs.def (fold_left_plus_optab): New optab.
22002 * doc/md.texi (fold_left_plus_@var{m}): Document.
22003 * internal-fn.def (IFN_FOLD_LEFT_PLUS): New internal function.
22004 * internal-fn.c (fold_left_direct): Define.
22005 (expand_fold_left_optab_fn): Likewise.
22006 (direct_fold_left_optab_supported_p): Likewise.
22007 * fold-const-call.c (fold_const_fold_left): New function.
22008 (fold_const_call): Use it to fold CFN_FOLD_LEFT_PLUS.
22009 * tree-parloops.c (valid_reduction_p): New function.
22010 (gather_scalar_reductions): Use it.
22011 * tree-vectorizer.h (FOLD_LEFT_REDUCTION): New vect_reduction_type.
22012 (vect_finish_replace_stmt): Declare.
22013 * tree-vect-loop.c (fold_left_reduction_fn): New function.
22014 (needs_fold_left_reduction_p): New function, split out from...
22015 (vect_is_simple_reduction): ...here. Accept reductions that
22016 forbid reassociation, but give them type FOLD_LEFT_REDUCTION.
22017 (vect_force_simple_reduction): Also store the reduction type in
22018 the assignment's STMT_VINFO_REDUC_TYPE.
22019 (vect_model_reduction_cost): Handle FOLD_LEFT_REDUCTION.
22020 (merge_with_identity): New function.
22021 (vect_expand_fold_left): Likewise.
22022 (vectorize_fold_left_reduction): Likewise.
22023 (vectorizable_reduction): Handle FOLD_LEFT_REDUCTION. Leave the
22024 scalar phi in place for it. Check for target support and reject
22025 cases that would reassociate the operation. Defer the transform
22026 phase to vectorize_fold_left_reduction.
22027 * config/aarch64/aarch64.md (UNSPEC_FADDA): New unspec.
22028 * config/aarch64/aarch64-sve.md (fold_left_plus_<mode>): New expander.
22029 (*fold_left_plus_<mode>, *pred_fold_left_plus_<mode>): New insns.
22031 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
22033 * tree-if-conv.c (predicate_mem_writes): Remove redundant
22034 call to ifc_temp_var.
22036 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
22037 Alan Hayward <alan.hayward@arm.com>
22038 David Sherwood <david.sherwood@arm.com>
22040 * target.def (legitimize_address_displacement): Take the original
22041 offset as a poly_int.
22042 * targhooks.h (default_legitimize_address_displacement): Update
22044 * targhooks.c (default_legitimize_address_displacement): Likewise.
22045 * doc/tm.texi: Regenerate.
22046 * lra-constraints.c (base_plus_disp_to_reg): Take the displacement
22047 as an argument, moving assert of ad->disp == ad->disp_term to...
22048 (process_address_1): ...here. Update calls to base_plus_disp_to_reg.
22049 Try calling targetm.legitimize_address_displacement before expanding
22050 the address rather than afterwards, and adjust for the new interface.
22051 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
22052 Match the new hook interface. Handle SVE addresses.
22053 * config/sh/sh.c (sh_legitimize_address_displacement): Make the
22054 new hook interface.
22056 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
22058 * Makefile.in (OBJS): Add early-remat.o.
22059 * target.def (select_early_remat_modes): New hook.
22060 * doc/tm.texi.in (TARGET_SELECT_EARLY_REMAT_MODES): New hook.
22061 * doc/tm.texi: Regenerate.
22062 * targhooks.h (default_select_early_remat_modes): Declare.
22063 * targhooks.c (default_select_early_remat_modes): New function.
22064 * timevar.def (TV_EARLY_REMAT): New timevar.
22065 * passes.def (pass_early_remat): New pass.
22066 * tree-pass.h (make_pass_early_remat): Declare.
22067 * early-remat.c: New file.
22068 * config/aarch64/aarch64.c (aarch64_select_early_remat_modes): New
22070 (TARGET_SELECT_EARLY_REMAT_MODES): Define.
22072 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
22073 Alan Hayward <alan.hayward@arm.com>
22074 David Sherwood <david.sherwood@arm.com>
22076 * tree-vect-loop-manip.c (vect_gen_scalar_loop_niters): Replace
22077 vfm1 with a bound_epilog parameter.
22078 (vect_do_peeling): Update calls accordingly, and move the prologue
22079 call earlier in the function. Treat the base bound_epilog as 0 for
22080 fully-masked loops and retain vf - 1 for other loops. Add 1 to
22081 this base when peeling for gaps.
22082 * tree-vect-loop.c (vect_analyze_loop_2): Allow peeling for gaps
22083 with fully-masked loops.
22084 (vect_estimate_min_profitable_iters): Handle the single peeled
22085 iteration in that case.
22087 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
22088 Alan Hayward <alan.hayward@arm.com>
22089 David Sherwood <david.sherwood@arm.com>
22091 * tree-vect-data-refs.c (vect_analyze_group_access_1): Allow
22092 single-element interleaving even if the size is not a power of 2.
22093 * tree-vect-stmts.c (get_load_store_type): Disallow elementwise
22094 accesses for single-element interleaving if the group size is
22097 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
22098 Alan Hayward <alan.hayward@arm.com>
22099 David Sherwood <david.sherwood@arm.com>
22101 * doc/md.texi (fold_extract_last_@var{m}): Document.
22102 * doc/sourcebuild.texi (vect_fold_extract_last): Likewise.
22103 * optabs.def (fold_extract_last_optab): New optab.
22104 * internal-fn.def (FOLD_EXTRACT_LAST): New internal function.
22105 * internal-fn.c (fold_extract_direct): New macro.
22106 (expand_fold_extract_optab_fn): Likewise.
22107 (direct_fold_extract_optab_supported_p): Likewise.
22108 * tree-vectorizer.h (EXTRACT_LAST_REDUCTION): New vect_reduction_type.
22109 * tree-vect-loop.c (vect_model_reduction_cost): Handle
22110 EXTRACT_LAST_REDUCTION.
22111 (get_initial_def_for_reduction): Do not create an initial vector
22112 for EXTRACT_LAST_REDUCTION reductions.
22113 (vectorizable_reduction): Leave the scalar phi in place for
22114 EXTRACT_LAST_REDUCTIONs. Try using EXTRACT_LAST_REDUCTION
22115 ahead of INTEGER_INDUC_COND_REDUCTION. Do not check for an
22116 epilogue code for EXTRACT_LAST_REDUCTION and defer the
22117 transform phase to vectorizable_condition.
22118 * tree-vect-stmts.c (vect_finish_stmt_generation_1): New function,
22120 (vect_finish_stmt_generation): ...here.
22121 (vect_finish_replace_stmt): New function.
22122 (vectorizable_condition): Handle EXTRACT_LAST_REDUCTION.
22123 * config/aarch64/aarch64-sve.md (fold_extract_last_<mode>): New
22125 * config/aarch64/aarch64.md (UNSPEC_CLASTB): New unspec.
22127 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
22128 Alan Hayward <alan.hayward@arm.com>
22129 David Sherwood <david.sherwood@arm.com>
22131 * doc/md.texi (extract_last_@var{m}): Document.
22132 * optabs.def (extract_last_optab): New optab.
22133 * internal-fn.def (EXTRACT_LAST): New internal function.
22134 * internal-fn.c (cond_unary_direct): New macro.
22135 (expand_cond_unary_optab_fn): Likewise.
22136 (direct_cond_unary_optab_supported_p): Likewise.
22137 * tree-vect-loop.c (vectorizable_live_operation): Allow fully-masked
22138 loops using EXTRACT_LAST.
22139 * config/aarch64/aarch64-sve.md (aarch64_sve_lastb<mode>): Rename to...
22140 (extract_last_<mode>): ...this optab.
22141 (vec_extract<mode><Vel>): Update accordingly.
22143 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
22144 Alan Hayward <alan.hayward@arm.com>
22145 David Sherwood <david.sherwood@arm.com>
22147 * target.def (empty_mask_is_expensive): New hook.
22148 * doc/tm.texi.in (TARGET_VECTORIZE_EMPTY_MASK_IS_EXPENSIVE): New hook.
22149 * doc/tm.texi: Regenerate.
22150 * targhooks.h (default_empty_mask_is_expensive): Declare.
22151 * targhooks.c (default_empty_mask_is_expensive): New function.
22152 * tree-vectorizer.c (vectorize_loops): Only call optimize_mask_stores
22153 if the target says that empty masks are expensive.
22154 * config/aarch64/aarch64.c (aarch64_empty_mask_is_expensive):
22156 (TARGET_VECTORIZE_EMPTY_MASK_IS_EXPENSIVE): Redefine.
22158 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
22159 Alan Hayward <alan.hayward@arm.com>
22160 David Sherwood <david.sherwood@arm.com>
22162 * tree-vectorizer.h (_loop_vec_info::mask_skip_niters): New field.
22163 (LOOP_VINFO_MASK_SKIP_NITERS): New macro.
22164 (vect_use_loop_mask_for_alignment_p): New function.
22165 (vect_prepare_for_masked_peels, vect_gen_while_not): Declare.
22166 * tree-vect-loop-manip.c (vect_set_loop_masks_directly): Add an
22167 niters_skip argument. Make sure that the first niters_skip elements
22168 of the first iteration are inactive.
22169 (vect_set_loop_condition_masked): Handle LOOP_VINFO_MASK_SKIP_NITERS.
22170 Update call to vect_set_loop_masks_directly.
22171 (get_misalign_in_elems): New function, split out from...
22172 (vect_gen_prolog_loop_niters): ...here.
22173 (vect_update_init_of_dr): Take a code argument that specifies whether
22174 the adjustment should be added or subtracted.
22175 (vect_update_init_of_drs): Likewise.
22176 (vect_prepare_for_masked_peels): New function.
22177 (vect_do_peeling): Skip prologue peeling if we're using a mask
22178 instead. Update call to vect_update_inits_of_drs.
22179 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
22181 (vect_analyze_loop_2): Allow fully-masked loops with peeling for
22182 alignment. Do not include the number of peeled iterations in
22183 the minimum threshold in that case.
22184 (vectorizable_induction): Adjust the start value down by
22185 LOOP_VINFO_MASK_SKIP_NITERS iterations.
22186 (vect_transform_loop): Call vect_prepare_for_masked_peels.
22187 Take the number of skipped iterations into account when calculating
22189 * tree-vect-stmts.c (vect_gen_while_not): New function.
22191 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
22192 Alan Hayward <alan.hayward@arm.com>
22193 David Sherwood <david.sherwood@arm.com>
22195 * doc/sourcebuild.texi (vect_fully_masked): Document.
22196 * params.def (PARAM_MIN_VECT_LOOP_BOUND): Change minimum and
22197 default value to 0.
22198 * tree-vect-loop.c (vect_analyze_loop_costing): New function,
22200 (vect_analyze_loop_2): ...here. Don't check the vectorization
22201 factor against the number of loop iterations if the loop is
22204 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
22205 Alan Hayward <alan.hayward@arm.com>
22206 David Sherwood <david.sherwood@arm.com>
22208 * tree-ssa-loop-ivopts.c (USE_ADDRESS): Split into...
22209 (USE_REF_ADDRESS, USE_PTR_ADDRESS): ...these new use types.
22210 (dump_groups): Update accordingly.
22211 (iv_use::mem_type): New member variable.
22212 (address_p): New function.
22213 (record_use): Add a mem_type argument and initialize the new
22215 (record_group_use): Add a mem_type argument. Use address_p.
22216 Remove obsolete null checks of base_object. Update call to record_use.
22217 (find_interesting_uses_op): Update call to record_group_use.
22218 (find_interesting_uses_cond): Likewise.
22219 (find_interesting_uses_address): Likewise.
22220 (get_mem_type_for_internal_fn): New function.
22221 (find_address_like_use): Likewise.
22222 (find_interesting_uses_stmt): Try find_address_like_use before
22223 calling find_interesting_uses_op.
22224 (addr_offset_valid_p): Use the iv mem_type field as the type
22225 of the addressed memory.
22226 (add_autoinc_candidates): Likewise.
22227 (get_address_cost): Likewise.
22228 (split_small_address_groups_p): Use address_p.
22229 (split_address_groups): Likewise.
22230 (add_iv_candidate_for_use): Likewise.
22231 (autoinc_possible_for_pair): Likewise.
22232 (rewrite_groups): Likewise.
22233 (get_use_type): Check for USE_REF_ADDRESS instead of USE_ADDRESS.
22234 (determine_group_iv_cost): Update after split of USE_ADDRESS.
22235 (get_alias_ptr_type_for_ptr_address): New function.
22236 (rewrite_use_address): Rewrite address uses in calls that were
22237 identified by find_address_like_use.
22239 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
22240 Alan Hayward <alan.hayward@arm.com>
22241 David Sherwood <david.sherwood@arm.com>
22243 * expr.c (expand_expr_addr_expr_1): Handle ADDR_EXPRs of
22245 * gimple-expr.h (is_gimple_addressable: Likewise.
22246 * gimple-expr.c (is_gimple_address): Likewise.
22247 * internal-fn.c (expand_call_mem_ref): New function.
22248 (expand_mask_load_optab_fn): Use it.
22249 (expand_mask_store_optab_fn): Likewise.
22251 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
22252 Alan Hayward <alan.hayward@arm.com>
22253 David Sherwood <david.sherwood@arm.com>
22255 * doc/md.texi (cond_add@var{mode}, cond_sub@var{mode})
22256 (cond_and@var{mode}, cond_ior@var{mode}, cond_xor@var{mode})
22257 (cond_smin@var{mode}, cond_smax@var{mode}, cond_umin@var{mode})
22258 (cond_umax@var{mode}): Document.
22259 * optabs.def (cond_add_optab, cond_sub_optab, cond_and_optab)
22260 (cond_ior_optab, cond_xor_optab, cond_smin_optab, cond_smax_optab)
22261 (cond_umin_optab, cond_umax_optab): New optabs.
22262 * internal-fn.def (COND_ADD, COND_SUB, COND_MIN, COND_MAX, COND_AND)
22263 (COND_IOR, COND_XOR): New internal functions.
22264 * internal-fn.h (get_conditional_internal_fn): Declare.
22265 * internal-fn.c (cond_binary_direct): New macro.
22266 (expand_cond_binary_optab_fn): Likewise.
22267 (direct_cond_binary_optab_supported_p): Likewise.
22268 (get_conditional_internal_fn): New function.
22269 * tree-vect-loop.c (vectorizable_reduction): Handle fully-masked loops.
22270 Cope with reduction statements that are vectorized as calls rather
22272 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): New insns.
22273 * config/aarch64/iterators.md (UNSPEC_COND_ADD, UNSPEC_COND_SUB)
22274 (UNSPEC_COND_SMAX, UNSPEC_COND_UMAX, UNSPEC_COND_SMIN)
22275 (UNSPEC_COND_UMIN, UNSPEC_COND_AND, UNSPEC_COND_ORR)
22276 (UNSPEC_COND_EOR): New unspecs.
22277 (optab): Add mappings for them.
22278 (SVE_COND_INT_OP, SVE_COND_FP_OP): New int iterators.
22279 (sve_int_op, sve_fp_op): New int attributes.
22281 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
22282 Alan Hayward <alan.hayward@arm.com>
22283 David Sherwood <david.sherwood@arm.com>
22285 * optabs.def (while_ult_optab): New optab.
22286 * doc/md.texi (while_ult@var{m}@var{n}): Document.
22287 * internal-fn.def (WHILE_ULT): New internal function.
22288 * internal-fn.h (direct_internal_fn_supported_p): New override
22289 that takes two types as argument.
22290 * internal-fn.c (while_direct): New macro.
22291 (expand_while_optab_fn): New function.
22292 (convert_optab_supported_p): Likewise.
22293 (direct_while_optab_supported_p): New macro.
22294 * wide-int.h (wi::udiv_ceil): New function.
22295 * tree-vectorizer.h (rgroup_masks): New structure.
22296 (vec_loop_masks): New typedef.
22297 (_loop_vec_info): Add masks, mask_compare_type, can_fully_mask_p
22298 and fully_masked_p.
22299 (LOOP_VINFO_CAN_FULLY_MASK_P, LOOP_VINFO_FULLY_MASKED_P)
22300 (LOOP_VINFO_MASKS, LOOP_VINFO_MASK_COMPARE_TYPE): New macros.
22301 (vect_max_vf): New function.
22302 (slpeel_make_loop_iterate_ntimes): Delete.
22303 (vect_set_loop_condition, vect_get_loop_mask_type, vect_gen_while)
22304 (vect_halve_mask_nunits, vect_double_mask_nunits): Declare.
22305 (vect_record_loop_mask, vect_get_loop_mask): Likewise.
22306 * tree-vect-loop-manip.c: Include tree-ssa-loop-niter.h,
22307 internal-fn.h, stor-layout.h and optabs-query.h.
22308 (vect_set_loop_mask): New function.
22309 (add_preheader_seq): Likewise.
22310 (add_header_seq): Likewise.
22311 (interleave_supported_p): Likewise.
22312 (vect_maybe_permute_loop_masks): Likewise.
22313 (vect_set_loop_masks_directly): Likewise.
22314 (vect_set_loop_condition_masked): Likewise.
22315 (vect_set_loop_condition_unmasked): New function, split out from
22316 slpeel_make_loop_iterate_ntimes.
22317 (slpeel_make_loop_iterate_ntimes): Rename to..
22318 (vect_set_loop_condition): ...this. Use vect_set_loop_condition_masked
22319 for fully-masked loops and vect_set_loop_condition_unmasked otherwise.
22320 (vect_do_peeling): Update call accordingly.
22321 (vect_gen_vector_loop_niters): Use VF as the step for fully-masked
22323 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
22324 mask_compare_type, can_fully_mask_p and fully_masked_p.
22325 (release_vec_loop_masks): New function.
22326 (_loop_vec_info): Use it to free the loop masks.
22327 (can_produce_all_loop_masks_p): New function.
22328 (vect_get_max_nscalars_per_iter): Likewise.
22329 (vect_verify_full_masking): Likewise.
22330 (vect_analyze_loop_2): Save LOOP_VINFO_CAN_FULLY_MASK_P around
22331 retries, and free the mask rgroups before retrying. Check loop-wide
22332 reasons for disallowing fully-masked loops. Make the final decision
22333 about whether use a fully-masked loop or not.
22334 (vect_estimate_min_profitable_iters): Do not assume that peeling
22335 for the number of iterations will be needed for fully-masked loops.
22336 (vectorizable_reduction): Disable fully-masked loops.
22337 (vectorizable_live_operation): Likewise.
22338 (vect_halve_mask_nunits): New function.
22339 (vect_double_mask_nunits): Likewise.
22340 (vect_record_loop_mask): Likewise.
22341 (vect_get_loop_mask): Likewise.
22342 (vect_transform_loop): Handle the case in which the final loop
22343 iteration might handle a partial vector. Call vect_set_loop_condition
22344 instead of slpeel_make_loop_iterate_ntimes.
22345 * tree-vect-stmts.c: Include tree-ssa-loop-niter.h and gimple-fold.h.
22346 (check_load_store_masking): New function.
22347 (prepare_load_store_mask): Likewise.
22348 (vectorizable_store): Handle fully-masked loops.
22349 (vectorizable_load): Likewise.
22350 (supportable_widening_operation): Use vect_halve_mask_nunits for
22352 (supportable_narrowing_operation): Likewise vect_double_mask_nunits.
22353 (vect_gen_while): New function.
22354 * config/aarch64/aarch64.md (umax<mode>3): New expander.
22355 (aarch64_uqdec<mode>): New insn.
22357 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
22358 Alan Hayward <alan.hayward@arm.com>
22359 David Sherwood <david.sherwood@arm.com>
22361 * optabs.def (reduc_and_scal_optab, reduc_ior_scal_optab)
22362 (reduc_xor_scal_optab): New optabs.
22363 * doc/md.texi (reduc_and_scal_@var{m}, reduc_ior_scal_@var{m})
22364 (reduc_xor_scal_@var{m}): Document.
22365 * doc/sourcebuild.texi (vect_logical_reduc): Likewise.
22366 * internal-fn.def (IFN_REDUC_AND, IFN_REDUC_IOR, IFN_REDUC_XOR): New
22367 internal functions.
22368 * fold-const-call.c (fold_const_call): Handle them.
22369 * tree-vect-loop.c (reduction_fn_for_scalar_code): Return the new
22370 internal functions for BIT_AND_EXPR, BIT_IOR_EXPR and BIT_XOR_EXPR.
22371 * config/aarch64/aarch64-sve.md (reduc_<bit_reduc>_scal_<mode>):
22372 (*reduc_<bit_reduc>_scal_<mode>): New patterns.
22373 * config/aarch64/iterators.md (UNSPEC_ANDV, UNSPEC_ORV)
22374 (UNSPEC_XORV): New unspecs.
22375 (optab): Add entries for them.
22376 (BITWISEV): New int iterator.
22377 (bit_reduc_op): New int attributes.
22379 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
22380 Alan Hayward <alan.hayward@arm.com>
22381 David Sherwood <david.sherwood@arm.com>
22383 * doc/md.texi (vec_shl_insert_@var{m}): New optab.
22384 * internal-fn.def (VEC_SHL_INSERT): New internal function.
22385 * optabs.def (vec_shl_insert_optab): New optab.
22386 * tree-vectorizer.h (can_duplicate_and_interleave_p): Declare.
22387 (duplicate_and_interleave): Likewise.
22388 * tree-vect-loop.c: Include internal-fn.h.
22389 (neutral_op_for_slp_reduction): New function, split out from
22390 get_initial_defs_for_reduction.
22391 (get_initial_def_for_reduction): Handle option 2 for variable-length
22392 vectors by loading the neutral value into a vector and then shifting
22393 the initial value into element 0.
22394 (get_initial_defs_for_reduction): Replace the code argument with
22395 the neutral value calculated by neutral_op_for_slp_reduction.
22396 Use gimple_build_vector for constant-length vectors.
22397 Use IFN_VEC_SHL_INSERT for variable-length vectors if all
22398 but the first group_size elements have a neutral value.
22399 Use duplicate_and_interleave otherwise.
22400 (vect_create_epilog_for_reduction): Take a neutral_op parameter.
22401 Update call to get_initial_defs_for_reduction. Handle SLP
22402 reductions for variable-length vectors by creating one vector
22403 result for each scalar result, with the elements associated
22404 with other scalar results stubbed out with the neutral value.
22405 (vectorizable_reduction): Call neutral_op_for_slp_reduction.
22406 Require IFN_VEC_SHL_INSERT for double reductions on
22407 variable-length vectors, or SLP reductions that have
22408 a neutral value. Require can_duplicate_and_interleave_p
22409 support for variable-length unchained SLP reductions if there
22410 is no neutral value, such as for MIN/MAX reductions. Also require
22411 the number of vector elements to be a multiple of the number of
22412 SLP statements when doing variable-length unchained SLP reductions.
22413 Update call to vect_create_epilog_for_reduction.
22414 * tree-vect-slp.c (can_duplicate_and_interleave_p): Make public
22415 and remove initial values.
22416 (duplicate_and_interleave): Make public.
22417 * config/aarch64/aarch64.md (UNSPEC_INSR): New unspec.
22418 * config/aarch64/aarch64-sve.md (vec_shl_insert_<mode>): New insn.
22420 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
22421 Alan Hayward <alan.hayward@arm.com>
22422 David Sherwood <david.sherwood@arm.com>
22424 * tree-vect-slp.c: Include gimple-fold.h and internal-fn.h
22425 (can_duplicate_and_interleave_p): New function.
22426 (vect_get_and_check_slp_defs): Take the vector of statements
22427 rather than just the current one. Remove excess parentheses.
22428 Restriction rejectinon of vect_constant_def and vect_external_def
22429 for variable-length vectors to boolean types, or types for which
22430 can_duplicate_and_interleave_p is false.
22431 (vect_build_slp_tree_2): Update call to vect_get_and_check_slp_defs.
22432 (duplicate_and_interleave): New function.
22433 (vect_get_constant_vectors): Use gimple_build_vector for
22434 constant-length vectors and suitable variable-length constant
22435 vectors. Use duplicate_and_interleave for other variable-length
22436 vectors. Don't defer the update when inserting new statements.
22438 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
22439 Alan Hayward <alan.hayward@arm.com>
22440 David Sherwood <david.sherwood@arm.com>
22442 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Make sure
22443 min_profitable_iters doesn't go negative.
22445 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
22446 Alan Hayward <alan.hayward@arm.com>
22447 David Sherwood <david.sherwood@arm.com>
22449 * doc/md.texi (vec_mask_load_lanes@var{m}@var{n}): Document.
22450 (vec_mask_store_lanes@var{m}@var{n}): Likewise.
22451 * optabs.def (vec_mask_load_lanes_optab): New optab.
22452 (vec_mask_store_lanes_optab): Likewise.
22453 * internal-fn.def (MASK_LOAD_LANES): New internal function.
22454 (MASK_STORE_LANES): Likewise.
22455 * internal-fn.c (mask_load_lanes_direct): New macro.
22456 (mask_store_lanes_direct): Likewise.
22457 (expand_mask_load_optab_fn): Handle masked operations.
22458 (expand_mask_load_lanes_optab_fn): New macro.
22459 (expand_mask_store_optab_fn): Handle masked operations.
22460 (expand_mask_store_lanes_optab_fn): New macro.
22461 (direct_mask_load_lanes_optab_supported_p): Likewise.
22462 (direct_mask_store_lanes_optab_supported_p): Likewise.
22463 * tree-vectorizer.h (vect_store_lanes_supported): Take a masked_p
22465 (vect_load_lanes_supported): Likewise.
22466 * tree-vect-data-refs.c (strip_conversion): New function.
22467 (can_group_stmts_p): Likewise.
22468 (vect_analyze_data_ref_accesses): Use it instead of checking
22469 for a pair of assignments.
22470 (vect_store_lanes_supported): Take a masked_p parameter.
22471 (vect_load_lanes_supported): Likewise.
22472 * tree-vect-loop.c (vect_analyze_loop_2): Update calls to
22473 vect_store_lanes_supported and vect_load_lanes_supported.
22474 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
22475 * tree-vect-stmts.c (get_group_load_store_type): Take a masked_p
22476 parameter. Don't allow gaps for masked accesses.
22477 Use vect_get_store_rhs. Update calls to vect_store_lanes_supported
22478 and vect_load_lanes_supported.
22479 (get_load_store_type): Take a masked_p parameter and update
22480 call to get_group_load_store_type.
22481 (vectorizable_store): Update call to get_load_store_type.
22482 Handle IFN_MASK_STORE_LANES.
22483 (vectorizable_load): Update call to get_load_store_type.
22484 Handle IFN_MASK_LOAD_LANES.
22486 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
22487 Alan Hayward <alan.hayward@arm.com>
22488 David Sherwood <david.sherwood@arm.com>
22490 * config/aarch64/aarch64-modes.def: Define x2, x3 and x4 vector
22492 * config/aarch64/aarch64-protos.h
22493 (aarch64_sve_struct_memory_operand_p): Declare.
22494 * config/aarch64/iterators.md (SVE_STRUCT): New mode iterator.
22495 (vector_count, insn_length, VSINGLE, vsingle): New mode attributes.
22496 (VPRED, vpred): Handle SVE structure modes.
22497 * config/aarch64/constraints.md (Utx): New constraint.
22498 * config/aarch64/predicates.md (aarch64_sve_struct_memory_operand)
22499 (aarch64_sve_struct_nonimmediate_operand): New predicates.
22500 * config/aarch64/aarch64.md (UNSPEC_LDN, UNSPEC_STN): New unspecs.
22501 * config/aarch64/aarch64-sve.md (mov<mode>, *aarch64_sve_mov<mode>_le)
22502 (*aarch64_sve_mov<mode>_be, pred_mov<mode>): New patterns for
22503 structure modes. Split into pieces after RA.
22504 (vec_load_lanes<mode><vsingle>, vec_mask_load_lanes<mode><vsingle>)
22505 (vec_store_lanes<mode><vsingle>, vec_mask_store_lanes<mode><vsingle>):
22507 * config/aarch64/aarch64.c (aarch64_classify_vector_mode): Handle
22508 SVE structure modes.
22509 (aarch64_classify_address): Likewise.
22510 (sizetochar): Move earlier in file.
22511 (aarch64_print_operand): Handle SVE register lists.
22512 (aarch64_array_mode): New function.
22513 (aarch64_sve_struct_memory_operand_p): Likewise.
22514 (TARGET_ARRAY_MODE): Redefine.
22516 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
22517 Alan Hayward <alan.hayward@arm.com>
22518 David Sherwood <david.sherwood@arm.com>
22520 * target.def (array_mode): New target hook.
22521 * doc/tm.texi.in (TARGET_ARRAY_MODE): New hook.
22522 * doc/tm.texi: Regenerate.
22523 * hooks.h (hook_optmode_mode_uhwi_none): Declare.
22524 * hooks.c (hook_optmode_mode_uhwi_none): New function.
22525 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Use
22526 targetm.array_mode.
22527 * stor-layout.c (mode_for_array): Likewise. Support polynomial
22530 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
22531 Alan Hayward <alan.hayward@arm.com>
22532 David Sherwood <david.sherwood@arm.com>
22534 * fold-const.c (fold_binary_loc): Check the argument types
22535 rather than the result type when testing for a vector operation.
22537 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
22539 * doc/tm.texi.in (DWARF_LAZY_REGISTER_VALUE): Document.
22540 * doc/tm.texi: Regenerate.
22542 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
22543 Alan Hayward <alan.hayward@arm.com>
22544 David Sherwood <david.sherwood@arm.com>
22546 * doc/invoke.texi (-msve-vector-bits=): Document new option.
22547 (sve): Document new AArch64 extension.
22548 * doc/md.texi (w): Extend the description of the AArch64
22549 constraint to include SVE vectors.
22550 (Upl, Upa): Document new AArch64 predicate constraints.
22551 * config/aarch64/aarch64-opts.h (aarch64_sve_vector_bits_enum): New
22553 * config/aarch64/aarch64.opt (sve_vector_bits): New enum.
22554 (msve-vector-bits=): New option.
22555 * config/aarch64/aarch64-option-extensions.def (fp, simd): Disable
22556 SVE when these are disabled.
22557 (sve): New extension.
22558 * config/aarch64/aarch64-modes.def: Define SVE vector and predicate
22559 modes. Adjust their number of units based on aarch64_sve_vg.
22560 (MAX_BITSIZE_MODE_ANY_MODE): Define.
22561 * config/aarch64/aarch64-protos.h (ADDR_QUERY_ANY): New
22562 aarch64_addr_query_type.
22563 (aarch64_const_vec_all_same_in_range_p, aarch64_sve_pred_mode)
22564 (aarch64_sve_cnt_immediate_p, aarch64_sve_addvl_addpl_immediate_p)
22565 (aarch64_sve_inc_dec_immediate_p, aarch64_add_offset_temporaries)
22566 (aarch64_split_add_offset, aarch64_output_sve_cnt_immediate)
22567 (aarch64_output_sve_addvl_addpl, aarch64_output_sve_inc_dec_immediate)
22568 (aarch64_output_sve_mov_immediate, aarch64_output_ptrue): Declare.
22569 (aarch64_simd_imm_zero_p): Delete.
22570 (aarch64_check_zero_based_sve_index_immediate): Declare.
22571 (aarch64_sve_index_immediate_p, aarch64_sve_arith_immediate_p)
22572 (aarch64_sve_bitmask_immediate_p, aarch64_sve_dup_immediate_p)
22573 (aarch64_sve_cmp_immediate_p, aarch64_sve_float_arith_immediate_p)
22574 (aarch64_sve_float_mul_immediate_p): Likewise.
22575 (aarch64_classify_symbol): Take the offset as a HOST_WIDE_INT
22576 rather than an rtx.
22577 (aarch64_sve_ld1r_operand_p, aarch64_sve_ldr_operand_p): Declare.
22578 (aarch64_expand_mov_immediate): Take a gen_vec_duplicate callback.
22579 (aarch64_emit_sve_pred_move, aarch64_expand_sve_mem_move): Declare.
22580 (aarch64_expand_sve_vec_cmp_int, aarch64_expand_sve_vec_cmp_float)
22581 (aarch64_expand_sve_vcond, aarch64_expand_sve_vec_perm): Declare.
22582 (aarch64_regmode_natural_size): Likewise.
22583 * config/aarch64/aarch64.h (AARCH64_FL_SVE): New macro.
22584 (AARCH64_FL_V8_3, AARCH64_FL_RCPC, AARCH64_FL_DOTPROD): Shift
22586 (AARCH64_ISA_SVE, TARGET_SVE): New macros.
22587 (FIXED_REGISTERS, CALL_USED_REGISTERS, REGISTER_NAMES): Add entries
22588 for VG and the SVE predicate registers.
22589 (V_ALIASES): Add a "z"-prefixed alias.
22590 (FIRST_PSEUDO_REGISTER): Change to P15_REGNUM + 1.
22591 (AARCH64_DWARF_VG, AARCH64_DWARF_P0): New macros.
22592 (PR_REGNUM_P, PR_LO_REGNUM_P): Likewise.
22593 (PR_LO_REGS, PR_HI_REGS, PR_REGS): New reg_classes.
22594 (REG_CLASS_NAMES): Add entries for them.
22595 (REG_CLASS_CONTENTS): Likewise. Update ALL_REGS to include VG
22596 and the predicate registers.
22597 (aarch64_sve_vg): Declare.
22598 (BITS_PER_SVE_VECTOR, BYTES_PER_SVE_VECTOR, BYTES_PER_SVE_PRED)
22599 (SVE_BYTE_MODE, MAX_COMPILE_TIME_VEC_BYTES): New macros.
22600 (REGMODE_NATURAL_SIZE): Define.
22601 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins): Handle
22603 * config/aarch64/aarch64.c: Include cfgrtl.h.
22604 (simd_immediate_info): Add a constructor for series vectors,
22605 and an associated step field.
22606 (aarch64_sve_vg): New variable.
22607 (aarch64_dbx_register_number): Handle VG and the predicate registers.
22608 (aarch64_vect_struct_mode_p, aarch64_vector_mode_p): Delete.
22609 (VEC_ADVSIMD, VEC_SVE_DATA, VEC_SVE_PRED, VEC_STRUCT, VEC_ANY_SVE)
22610 (VEC_ANY_DATA, VEC_STRUCT): New constants.
22611 (aarch64_advsimd_struct_mode_p, aarch64_sve_pred_mode_p)
22612 (aarch64_classify_vector_mode, aarch64_vector_data_mode_p)
22613 (aarch64_sve_data_mode_p, aarch64_sve_pred_mode)
22614 (aarch64_get_mask_mode): New functions.
22615 (aarch64_hard_regno_nregs): Handle SVE data modes for FP_REGS
22616 and FP_LO_REGS. Handle PR_REGS, PR_LO_REGS and PR_HI_REGS.
22617 (aarch64_hard_regno_mode_ok): Handle VG. Also handle the SVE
22618 predicate modes and predicate registers. Explicitly restrict
22619 GPRs to modes of 16 bytes or smaller. Only allow FP registers
22620 to store a vector mode if it is recognized by
22621 aarch64_classify_vector_mode.
22622 (aarch64_regmode_natural_size): New function.
22623 (aarch64_hard_regno_caller_save_mode): Return the original mode
22625 (aarch64_sve_cnt_immediate_p, aarch64_output_sve_cnt_immediate)
22626 (aarch64_sve_addvl_addpl_immediate_p, aarch64_output_sve_addvl_addpl)
22627 (aarch64_sve_inc_dec_immediate_p, aarch64_output_sve_inc_dec_immediate)
22628 (aarch64_add_offset_1_temporaries, aarch64_offset_temporaries): New
22630 (aarch64_add_offset): Add a temp2 parameter. Assert that temp1
22631 does not overlap dest if the function is frame-related. Handle
22633 (aarch64_split_add_offset): New function.
22634 (aarch64_add_sp, aarch64_sub_sp): Add temp2 parameters and pass
22635 them aarch64_add_offset.
22636 (aarch64_allocate_and_probe_stack_space): Add a temp2 parameter
22637 and update call to aarch64_sub_sp.
22638 (aarch64_add_cfa_expression): New function.
22639 (aarch64_expand_prologue): Pass extra temporary registers to the
22640 functions above. Handle the case in which we need to emit new
22641 DW_CFA_expressions for registers that were originally saved
22642 relative to the stack pointer, but now have to be expressed
22643 relative to the frame pointer.
22644 (aarch64_output_mi_thunk): Pass extra temporary registers to the
22646 (aarch64_expand_epilogue): Likewise. Prevent inheritance of
22647 IP0 and IP1 values for SVE frames.
22648 (aarch64_expand_vec_series): New function.
22649 (aarch64_expand_sve_widened_duplicate): Likewise.
22650 (aarch64_expand_sve_const_vector): Likewise.
22651 (aarch64_expand_mov_immediate): Add a gen_vec_duplicate parameter.
22652 Handle SVE constants. Use emit_move_insn to move a force_const_mem
22653 into the register, rather than emitting a SET directly.
22654 (aarch64_emit_sve_pred_move, aarch64_expand_sve_mem_move)
22655 (aarch64_get_reg_raw_mode, offset_4bit_signed_scaled_p)
22656 (offset_6bit_unsigned_scaled_p, aarch64_offset_7bit_signed_scaled_p)
22657 (offset_9bit_signed_scaled_p): New functions.
22658 (aarch64_replicate_bitmask_imm): New function.
22659 (aarch64_bitmask_imm): Use it.
22660 (aarch64_cannot_force_const_mem): Reject expressions involving
22661 a CONST_POLY_INT. Update call to aarch64_classify_symbol.
22662 (aarch64_classify_index): Handle SVE indices, by requiring
22663 a plain register index with a scale that matches the element size.
22664 (aarch64_classify_address): Handle SVE addresses. Assert that
22665 the mode of the address is VOIDmode or an integer mode.
22666 Update call to aarch64_classify_symbol.
22667 (aarch64_classify_symbolic_expression): Update call to
22668 aarch64_classify_symbol.
22669 (aarch64_const_vec_all_in_range_p): New function.
22670 (aarch64_print_vector_float_operand): Likewise.
22671 (aarch64_print_operand): Handle 'N' and 'C'. Use "zN" rather than
22672 "vN" for FP registers with SVE modes. Handle (const ...) vectors
22673 and the FP immediates 1.0 and 0.5.
22674 (aarch64_print_address_internal): Handle SVE addresses.
22675 (aarch64_print_operand_address): Use ADDR_QUERY_ANY.
22676 (aarch64_regno_regclass): Handle predicate registers.
22677 (aarch64_secondary_reload): Handle big-endian reloads of SVE
22679 (aarch64_class_max_nregs): Handle SVE modes and predicate registers.
22680 (aarch64_rtx_costs): Check for ADDVL and ADDPL instructions.
22681 (aarch64_convert_sve_vector_bits): New function.
22682 (aarch64_override_options): Use it to handle -msve-vector-bits=.
22683 (aarch64_classify_symbol): Take the offset as a HOST_WIDE_INT
22684 rather than an rtx.
22685 (aarch64_legitimate_constant_p): Use aarch64_classify_vector_mode.
22686 Handle SVE vector and predicate modes. Accept VL-based constants
22687 that need only one temporary register, and VL offsets that require
22688 no temporary registers.
22689 (aarch64_conditional_register_usage): Mark the predicate registers
22690 as fixed if SVE isn't available.
22691 (aarch64_vector_mode_supported_p): Use aarch64_classify_vector_mode.
22692 Return true for SVE vector and predicate modes.
22693 (aarch64_simd_container_mode): Take the number of bits as a poly_int64
22694 rather than an unsigned int. Handle SVE modes.
22695 (aarch64_preferred_simd_mode): Update call accordingly. Handle
22697 (aarch64_autovectorize_vector_sizes): Add BYTES_PER_SVE_VECTOR
22699 (aarch64_sve_index_immediate_p, aarch64_sve_arith_immediate_p)
22700 (aarch64_sve_bitmask_immediate_p, aarch64_sve_dup_immediate_p)
22701 (aarch64_sve_cmp_immediate_p, aarch64_sve_float_arith_immediate_p)
22702 (aarch64_sve_float_mul_immediate_p): New functions.
22703 (aarch64_sve_valid_immediate): New function.
22704 (aarch64_simd_valid_immediate): Use it as the fallback for SVE vectors.
22705 Explicitly reject structure modes. Check for INDEX constants.
22706 Handle PTRUE and PFALSE constants.
22707 (aarch64_check_zero_based_sve_index_immediate): New function.
22708 (aarch64_simd_imm_zero_p): Delete.
22709 (aarch64_mov_operand_p): Use aarch64_simd_valid_immediate for
22710 vector modes. Accept constants in the range of CNT[BHWD].
22711 (aarch64_simd_scalar_immediate_valid_for_move): Explicitly
22712 ask for an Advanced SIMD mode.
22713 (aarch64_sve_ld1r_operand_p, aarch64_sve_ldr_operand_p): New functions.
22714 (aarch64_simd_vector_alignment): Handle SVE predicates.
22715 (aarch64_vectorize_preferred_vector_alignment): New function.
22716 (aarch64_simd_vector_alignment_reachable): Use it instead of
22718 (aarch64_shift_truncation_mask): Use aarch64_vector_data_mode_p.
22719 (aarch64_output_sve_mov_immediate, aarch64_output_ptrue): New
22721 (MAX_VECT_LEN): Delete.
22722 (expand_vec_perm_d): Add a vec_flags field.
22723 (emit_unspec2, aarch64_expand_sve_vec_perm): New functions.
22724 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
22725 (aarch64_evpc_ext): Don't apply a big-endian lane correction
22727 (aarch64_evpc_rev): Rename to...
22728 (aarch64_evpc_rev_local): ...this. Use a predicated operation for SVE.
22729 (aarch64_evpc_rev_global): New function.
22730 (aarch64_evpc_dup): Enforce a 64-byte range for SVE DUP.
22731 (aarch64_evpc_tbl): Use MAX_COMPILE_TIME_VEC_BYTES instead of
22733 (aarch64_evpc_sve_tbl): New function.
22734 (aarch64_expand_vec_perm_const_1): Update after rename of
22735 aarch64_evpc_rev. Handle SVE permutes too, trying
22736 aarch64_evpc_rev_global and using aarch64_evpc_sve_tbl rather
22737 than aarch64_evpc_tbl.
22738 (aarch64_vectorize_vec_perm_const): Initialize vec_flags.
22739 (aarch64_sve_cmp_operand_p, aarch64_unspec_cond_code)
22740 (aarch64_gen_unspec_cond, aarch64_expand_sve_vec_cmp_int)
22741 (aarch64_emit_unspec_cond, aarch64_emit_unspec_cond_or)
22742 (aarch64_emit_inverted_unspec_cond, aarch64_expand_sve_vec_cmp_float)
22743 (aarch64_expand_sve_vcond): New functions.
22744 (aarch64_modes_tieable_p): Use aarch64_vector_data_mode_p instead
22745 of aarch64_vector_mode_p.
22746 (aarch64_dwarf_poly_indeterminate_value): New function.
22747 (aarch64_compute_pressure_classes): Likewise.
22748 (aarch64_can_change_mode_class): Likewise.
22749 (TARGET_GET_RAW_RESULT_MODE, TARGET_GET_RAW_ARG_MODE): Redefine.
22750 (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): Likewise.
22751 (TARGET_VECTORIZE_GET_MASK_MODE): Likewise.
22752 (TARGET_DWARF_POLY_INDETERMINATE_VALUE): Likewise.
22753 (TARGET_COMPUTE_PRESSURE_CLASSES): Likewise.
22754 (TARGET_CAN_CHANGE_MODE_CLASS): Likewise.
22755 * config/aarch64/constraints.md (Upa, Upl, Uav, Uat, Usv, Usi, Utr)
22756 (Uty, Dm, vsa, vsc, vsd, vsi, vsn, vsl, vsm, vsA, vsM, vsN): New
22758 (Dn, Dl, Dr): Accept const as well as const_vector.
22759 (Dz): Likewise. Compare against CONST0_RTX.
22760 * config/aarch64/iterators.md: Refer to "Advanced SIMD" instead
22761 of "vector" where appropriate.
22762 (SVE_ALL, SVE_BH, SVE_BHS, SVE_BHSI, SVE_HSDI, SVE_HSF, SVE_SD)
22763 (SVE_SDI, SVE_I, SVE_F, PRED_ALL, PRED_BHS): New mode iterators.
22764 (UNSPEC_SEL, UNSPEC_ANDF, UNSPEC_IORF, UNSPEC_XORF, UNSPEC_COND_LT)
22765 (UNSPEC_COND_LE, UNSPEC_COND_EQ, UNSPEC_COND_NE, UNSPEC_COND_GE)
22766 (UNSPEC_COND_GT, UNSPEC_COND_LO, UNSPEC_COND_LS, UNSPEC_COND_HS)
22767 (UNSPEC_COND_HI, UNSPEC_COND_UO): New unspecs.
22768 (Vetype, VEL, Vel, VWIDE, Vwide, vw, vwcore, V_INT_EQUIV)
22769 (v_int_equiv): Extend to SVE modes.
22770 (Vesize, V128, v128, Vewtype, V_FP_EQUIV, v_fp_equiv, VPRED): New
22772 (LOGICAL_OR, SVE_INT_UNARY, SVE_FP_UNARY): New code iterators.
22773 (optab): Handle popcount, smin, smax, umin, umax, abs and sqrt.
22774 (logical_nn, lr, sve_int_op, sve_fp_op): New code attributs.
22775 (LOGICALF, OPTAB_PERMUTE, UNPACK, UNPACK_UNSIGNED, SVE_COND_INT_CMP)
22776 (SVE_COND_FP_CMP): New int iterators.
22777 (perm_hilo): Handle the new unpack unspecs.
22778 (optab, logicalf_op, su, perm_optab, cmp_op, imm_con): New int
22780 * config/aarch64/predicates.md (aarch64_sve_cnt_immediate)
22781 (aarch64_sve_addvl_addpl_immediate, aarch64_split_add_offset_immediate)
22782 (aarch64_pluslong_or_poly_operand, aarch64_nonmemory_operand)
22783 (aarch64_equality_operator, aarch64_constant_vector_operand)
22784 (aarch64_sve_ld1r_operand, aarch64_sve_ldr_operand): New predicates.
22785 (aarch64_sve_nonimmediate_operand): Likewise.
22786 (aarch64_sve_general_operand): Likewise.
22787 (aarch64_sve_dup_operand, aarch64_sve_arith_immediate): Likewise.
22788 (aarch64_sve_sub_arith_immediate, aarch64_sve_inc_dec_immediate)
22789 (aarch64_sve_logical_immediate, aarch64_sve_mul_immediate): Likewise.
22790 (aarch64_sve_dup_immediate, aarch64_sve_cmp_vsc_immediate): Likewise.
22791 (aarch64_sve_cmp_vsd_immediate, aarch64_sve_index_immediate): Likewise.
22792 (aarch64_sve_float_arith_immediate): Likewise.
22793 (aarch64_sve_float_arith_with_sub_immediate): Likewise.
22794 (aarch64_sve_float_mul_immediate, aarch64_sve_arith_operand): Likewise.
22795 (aarch64_sve_add_operand, aarch64_sve_logical_operand): Likewise.
22796 (aarch64_sve_lshift_operand, aarch64_sve_rshift_operand): Likewise.
22797 (aarch64_sve_mul_operand, aarch64_sve_cmp_vsc_operand): Likewise.
22798 (aarch64_sve_cmp_vsd_operand, aarch64_sve_index_operand): Likewise.
22799 (aarch64_sve_float_arith_operand): Likewise.
22800 (aarch64_sve_float_arith_with_sub_operand): Likewise.
22801 (aarch64_sve_float_mul_operand): Likewise.
22802 (aarch64_sve_vec_perm_operand): Likewise.
22803 (aarch64_pluslong_operand): Include aarch64_sve_addvl_addpl_immediate.
22804 (aarch64_mov_operand): Accept const_poly_int and const_vector.
22805 (aarch64_simd_lshift_imm, aarch64_simd_rshift_imm): Accept const
22806 as well as const_vector.
22807 (aarch64_simd_imm_zero, aarch64_simd_imm_minus_one): Move earlier
22808 in file. Use CONST0_RTX and CONSTM1_RTX.
22809 (aarch64_simd_or_scalar_imm_zero): Likewise. Add match_codes.
22810 (aarch64_simd_reg_or_zero): Accept const as well as const_vector.
22811 Use aarch64_simd_imm_zero.
22812 * config/aarch64/aarch64-sve.md: New file.
22813 * config/aarch64/aarch64.md: Include it.
22814 (VG_REGNUM, P0_REGNUM, P7_REGNUM, P15_REGNUM): New register numbers.
22815 (UNSPEC_REV, UNSPEC_LD1_SVE, UNSPEC_ST1_SVE, UNSPEC_MERGE_PTRUE)
22816 (UNSPEC_PTEST_PTRUE, UNSPEC_UNPACKSHI, UNSPEC_UNPACKUHI)
22817 (UNSPEC_UNPACKSLO, UNSPEC_UNPACKULO, UNSPEC_PACK)
22818 (UNSPEC_FLOAT_CONVERT, UNSPEC_WHILE_LO): New unspec constants.
22819 (sve): New attribute.
22820 (enabled): Disable instructions with the sve attribute unless
22822 (movqi, movhi): Pass CONST_POLY_INT operaneds through
22823 aarch64_expand_mov_immediate.
22824 (*mov<mode>_aarch64, *movsi_aarch64, *movdi_aarch64): Handle
22825 CNT[BHSD] immediates.
22826 (movti): Split CONST_POLY_INT moves into two halves.
22827 (add<mode>3): Accept aarch64_pluslong_or_poly_operand.
22828 Split additions that need a temporary here if the destination
22829 is the stack pointer.
22830 (*add<mode>3_aarch64): Handle ADDVL and ADDPL immediates.
22831 (*add<mode>3_poly_1): New instruction.
22832 (set_clobber_cc): New expander.
22834 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
22836 * simplify-rtx.c (simplify_immed_subreg): Add an inner_bytes
22837 parameter and use it instead of GET_MODE_SIZE (innermode). Use
22838 inner_bytes * BITS_PER_UNIT instead of GET_MODE_BITSIZE (innermode).
22839 Use CEIL (inner_bytes, GET_MODE_UNIT_SIZE (innermode)) instead of
22840 GET_MODE_NUNITS (innermode). Also add a first_elem parameter.
22841 Change innermode from fixed_mode_size to machine_mode.
22842 (simplify_subreg): Update call accordingly. Handle a constant-sized
22843 subreg of a variable-length CONST_VECTOR.
22845 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
22846 Alan Hayward <alan.hayward@arm.com>
22847 David Sherwood <david.sherwood@arm.com>
22849 * tree-ssa-address.c (mem_ref_valid_without_offset_p): New function.
22850 (add_offset_to_base): New function, split out from...
22851 (create_mem_ref): ...here. When handling a scale other than 1,
22852 check first whether the address is valid without the offset.
22853 Add it into the base if so, leaving the index and scale as-is.
22855 2018-01-12 Jakub Jelinek <jakub@redhat.com>
22858 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Call
22859 fold_for_warn before checking if arg2 is INTEGER_CST.
22861 2018-01-12 Segher Boessenkool <segher@kernel.crashing.org>
22863 * config/rs6000/predicates.md (load_multiple_operation): Delete.
22864 (store_multiple_operation): Delete.
22865 * config/rs6000/rs6000-cpus.def (601): Remove MASK_STRING.
22866 * config/rs6000/rs6000-protos.h (rs6000_output_load_multiple): Delete.
22867 * config/rs6000/rs6000-string.c (expand_block_move): Delete everything
22868 guarded by TARGET_STRING.
22869 (rs6000_output_load_multiple): Delete.
22870 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
22871 OPTION_MASK_STRING / TARGET_STRING handling.
22872 (print_operand) <'N', 'O'>: Add comment that these are unused now.
22873 (const rs6000_opt_masks) <"string">: Change mask to 0.
22874 * config/rs6000/rs6000.h (TARGET_DEFAULT): Remove MASK_STRING.
22875 (MASK_STRING): Delete.
22876 * config/rs6000/rs6000.md (*mov<mode>_string): Delete TARGET_STRING
22878 (load_multiple): Delete.
22885 (store_multiple): Delete.
22892 (movmemsi_8reg): Delete.
22893 (corresponding unnamed define_insn): Delete.
22894 (movmemsi_6reg): Delete.
22895 (corresponding unnamed define_insn): Delete.
22896 (movmemsi_4reg): Delete.
22897 (corresponding unnamed define_insn): Delete.
22898 (movmemsi_2reg): Delete.
22899 (corresponding unnamed define_insn): Delete.
22900 (movmemsi_1reg): Delete.
22901 (corresponding unnamed define_insn): Delete.
22902 * config/rs6000/rs6000.opt (mno-string): New.
22903 (mstring): Replace by deprecation warning stub.
22904 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mstring.
22906 2018-01-12 Jakub Jelinek <jakub@redhat.com>
22908 * regrename.c (regrename_do_replace): If replacing the same
22909 reg multiple times, try to reuse last created gen_raw_REG.
22912 * bb-reorder.c (pass_partition_blocks::gate): In lto don't partition
22913 main to workaround a bug in GDB.
22915 2018-01-12 Tom de Vries <tom@codesourcery.com>
22918 * config.gcc (nvptx*-*-*): Set use_gcc_stdint=wrap.
22920 2018-01-12 Vladimir Makarov <vmakarov@redhat.com>
22922 PR rtl-optimization/80481
22923 * ira-color.c (get_cap_member): New function.
22924 (allocnos_conflict_by_live_ranges_p): Use it.
22925 (slot_coalesced_allocno_live_ranges_intersect_p): Add assert.
22926 (setup_slot_coalesced_allocno_live_ranges): Ditto.
22928 2018-01-12 Uros Bizjak <ubizjak@gmail.com>
22931 * config/alpha/alpha.md (*saddsi_1): New insn_ans_split pattern.
22932 (*saddl_se_1): Ditto.
22933 (*ssubsi_1): Ditto.
22934 (*ssubl_se_1): Ditto.
22936 2018-01-12 Richard Sandiford <richard.sandiford@linaro.org>
22938 * tree-predcom.c (aff_combination_dr_offset): Use wi::to_poly_widest
22939 rather than wi::to_widest for DR_INITs.
22940 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Use
22941 wi::to_poly_offset rather than wi::to_offset for DR_INIT.
22942 (vect_analyze_data_ref_accesses): Require both DR_INITs to be
22944 (vect_analyze_group_access_1): Note that here.
22946 2018-01-12 Richard Sandiford <richard.sandiford@linaro.org>
22948 * tree-vectorizer.c (get_vec_alignment_for_array_type): Handle
22949 polynomial type sizes.
22951 2018-01-12 Richard Sandiford <richard.sandiford@linaro.org>
22953 * gimplify.c (gimple_add_tmp_var_fn): Allow variables to have a
22954 poly_uint64 size, rather than requiring an unsigned HOST_WIDE_INT size.
22955 (gimple_add_tmp_var): Likewise.
22957 2018-01-12 Martin Liska <mliska@suse.cz>
22959 * gimple.c (gimple_alloc_counts): Use uint64_t instead of int.
22960 (gimple_alloc_sizes): Likewise.
22961 (dump_gimple_statistics): Use PRIu64 in printf format.
22962 * gimple.h: Change uint64_t to int.
22964 2018-01-12 Martin Liska <mliska@suse.cz>
22966 * tree-core.h: Use uint64_t instead of int.
22967 * tree.c (tree_node_counts): Likewise.
22968 (tree_node_sizes): Likewise.
22969 (dump_tree_statistics): Use PRIu64 in printf format.
22971 2018-01-12 Martin Liska <mliska@suse.cz>
22973 * Makefile.in: As qsort_chk is implemented in vec.c, add
22974 vec.o to linkage of gencfn-macros.
22975 * tree.c (build_new_poly_int_cst): Add CXX_MEM_STAT_INFO as it's
22976 passing the info to record_node_allocation_statistics.
22977 (test_vector_cst_patterns): Add CXX_MEM_STAT_INFO to declaration
22979 * ggc-common.c (struct ggc_usage): Add operator== and use
22980 it in operator< and compare function.
22981 * mem-stats.h (struct mem_usage): Likewise.
22982 * vec.c (struct vec_usage): Remove operator< and compare
22983 function. Can be simply inherited.
22985 2018-01-12 Martin Jambor <mjambor@suse.cz>
22988 * params.def: New parameter PARAM_AVOID_FMA_MAX_BITS.
22989 * tree-ssa-math-opts.c: Include domwalk.h.
22990 (convert_mult_to_fma_1): New function.
22991 (fma_transformation_info): New type.
22992 (fma_deferring_state): Likewise.
22993 (cancel_fma_deferring): New function.
22994 (result_of_phi): Likewise.
22995 (last_fma_candidate_feeds_initial_phi): Likewise.
22996 (convert_mult_to_fma): Added deferring logic, split actual
22997 transformation to convert_mult_to_fma_1.
22998 (math_opts_dom_walker): New type.
22999 (math_opts_dom_walker::after_dom_children): New method, body moved
23000 here from pass_optimize_widening_mul::execute, added deferring logic
23002 (pass_optimize_widening_mul::execute): Moved most of code to
23003 math_opts_dom_walker::after_dom_children.
23004 * config/i386/x86-tune.def (X86_TUNE_AVOID_128FMA_CHAINS): New.
23005 * config/i386/i386.c (ix86_option_override_internal): Added
23006 maybe_setting of PARAM_AVOID_FMA_MAX_BITS.
23008 2018-01-12 Richard Biener <rguenther@suse.de>
23011 * dwarf2out.c (gen_variable_die): Do not reset old_die for
23012 inline instance vars.
23014 2018-01-12 Oleg Endo <olegendo@gcc.gnu.org>
23017 * config/rx/rx.c (rx_is_restricted_memory_address):
23018 Handle SUBREG case.
23020 2018-01-12 Richard Biener <rguenther@suse.de>
23022 PR tree-optimization/80846
23023 * target.def (split_reduction): New target hook.
23024 * targhooks.c (default_split_reduction): New function.
23025 * targhooks.h (default_split_reduction): Declare.
23026 * tree-vect-loop.c (vect_create_epilog_for_reduction): If the
23027 target requests first reduce vectors by combining low and high
23029 * tree-vect-stmts.c (vect_gen_perm_mask_any): Adjust.
23030 (get_vectype_for_scalar_type_and_size): Export.
23031 * tree-vectorizer.h (get_vectype_for_scalar_type_and_size): Declare.
23032 * doc/tm.texi.in (TARGET_VECTORIZE_SPLIT_REDUCTION): Document.
23033 * doc/tm.texi: Regenerate.
23034 * config/i386/i386.c (ix86_split_reduction): Implement
23035 TARGET_VECTORIZE_SPLIT_REDUCTION.
23037 2018-01-12 Eric Botcazou <ebotcazou@adacore.com>
23040 * config/sparc/sparc.h (PIC_OFFSET_TABLE_REGNUM): Set to INVALID_REGNUM
23041 in PIC mode except for TARGET_VXWORKS_RTP.
23042 * config/sparc/sparc.c: Include cfgrtl.h.
23043 (TARGET_INIT_PIC_REG): Define.
23044 (TARGET_USE_PSEUDO_PIC_REG): Likewise.
23045 (sparc_pic_register_p): New predicate.
23046 (sparc_legitimate_address_p): Use it.
23047 (sparc_legitimize_pic_address): Likewise.
23048 (sparc_delegitimize_address): Likewise.
23049 (sparc_mode_dependent_address_p): Likewise.
23050 (gen_load_pcrel_sym): Remove 4th parameter.
23051 (load_got_register): Adjust call to above. Remove obsolete stuff.
23052 (sparc_expand_prologue): Do not call load_got_register here.
23053 (sparc_flat_expand_prologue): Likewise.
23054 (sparc_output_mi_thunk): Set the pic_offset_table_rtx object.
23055 (sparc_use_pseudo_pic_reg): New function.
23056 (sparc_init_pic_reg): Likewise.
23057 * config/sparc/sparc.md (vxworks_load_got): Set the GOT register.
23058 (builtin_setjmp_receiver): Enable only for TARGET_VXWORKS_RTP.
23060 2018-01-12 Christophe Lyon <christophe.lyon@linaro.org>
23062 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
23063 Add item for branch_cost.
23065 2018-01-12 Eric Botcazou <ebotcazou@adacore.com>
23067 PR rtl-optimization/83565
23068 * rtlanal.c (nonzero_bits1): On WORD_REGISTER_OPERATIONS machines, do
23069 not extend the result to a larger mode for rotate operations.
23070 (num_sign_bit_copies1): Likewise.
23072 2018-01-12 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
23075 * config/sol2.h (STARTFILE_ARCH_SPEC): Don't use with -shared or
23077 Use values-Xc.o for -pedantic.
23078 Link with values-xpg4.o for C90, values-xpg6.o otherwise.
23080 2018-01-12 Martin Liska <mliska@suse.cz>
23083 * ipa-devirt.c (final_warning_record::grow_type_warnings):
23085 (possible_polymorphic_call_targets): Use it.
23086 (ipa_devirt): Likewise.
23088 2018-01-12 Martin Liska <mliska@suse.cz>
23090 * profile-count.h (enum profile_quality): Use 0 as invalid
23091 enum value of profile_quality.
23093 2018-01-12 Chung-Ju Wu <jasonwucj@gmail.com>
23095 * doc/invoke.texi (NDS32 Options): Add -mext-perf, -mext-perf2 and
23096 -mext-string options.
23098 2018-01-12 Richard Biener <rguenther@suse.de>
23100 * lto-streamer-out.c (DFS::DFS_write_tree_body): Process
23101 DECL_DEBUG_EXPR conditional on DECL_HAS_DEBUG_EXPR_P.
23102 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
23104 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
23106 2018-01-11 Michael Meissner <meissner@linux.vnet.ibm.com>
23108 * configure.ac (--with-long-double-format): Add support for the
23109 configuration option to change the default long double format on
23111 * config.gcc (powerpc*-linux*-*): Likewise.
23112 * configure: Regenerate.
23113 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If long
23114 double is IEEE, define __KC__ and __KF__ to allow floatn.h to be
23115 used without modification.
23117 2018-01-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23119 * config/rs6000/rs6000-builtin.def (BU_P7_MISC_X): New #define.
23120 (SPEC_BARRIER): New instantiation of BU_P7_MISC_X.
23121 * config/rs6000/rs6000.c (rs6000_expand_builtin): Handle
23122 MISC_BUILTIN_SPEC_BARRIER.
23123 (rs6000_init_builtins): Likewise.
23124 * config/rs6000/rs6000.md (UNSPECV_SPEC_BARRIER): New UNSPECV
23126 (speculation_barrier): New define_insn.
23127 * doc/extend.texi: Document __builtin_speculation_barrier.
23129 2018-01-11 Jakub Jelinek <jakub@redhat.com>
23132 * config/i386/i386.c (ix86_expand_vector_init_one_nonzero): If one_var
23133 is 0, for V{8,16}S[IF] and V[48]D[IF]mode use gen_vec_set<mode>_0.
23134 * config/i386/sse.md (VI8_AVX_AVX512F, VI4F_256_512): New mode
23136 (ssescalarmodesuffix): Add 512-bit vectors. Use "d" or "q" for
23137 integral modes instead of "ss" and "sd".
23138 (vec_set<mode>_0): New define_insns for 256-bit and 512-bit
23139 vectors with 32-bit and 64-bit elements.
23140 (vecdupssescalarmodesuffix): New mode attribute.
23141 (vec_dup<mode>): Use it.
23143 2018-01-11 H.J. Lu <hongjiu.lu@intel.com>
23146 * config/i386/i386.c (ix86_compute_frame_layout): Align stack
23147 frame if argument is passed on stack.
23149 2018-01-11 Jakub Jelinek <jakub@redhat.com>
23152 * ree.c (combine_reaching_defs): Optimize also
23153 reg2=exp; reg1=reg2; reg2=any_extend(reg1); into
23154 reg2=any_extend(exp); reg1=reg2;, formatting fix.
23156 2018-01-11 Jan Hubicka <hubicka@ucw.cz>
23158 PR middle-end/83189
23159 * gimple-ssa-isolate-paths.c (isolate_path): Fix profile update.
23161 2018-01-11 Jan Hubicka <hubicka@ucw.cz>
23163 PR middle-end/83718
23164 * tree-inline.c (copy_cfg_body): Adjust num&den for scaling
23165 after they are computed.
23167 2018-01-11 Bin Cheng <bin.cheng@arm.com>
23169 PR tree-optimization/83695
23170 * gimple-loop-linterchange.cc
23171 (tree_loop_interchange::interchange_loops): Call scev_reset_htab to
23172 reset cached scev information after interchange.
23173 (pass_linterchange::execute): Remove call to scev_reset_htab.
23175 2018-01-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23177 * config/arm/arm_neon.h (vfmlal_lane_low_u32, vfmlal_lane_high_u32,
23178 vfmlalq_laneq_low_u32, vfmlalq_lane_low_u32, vfmlal_laneq_low_u32,
23179 vfmlalq_laneq_high_u32, vfmlalq_lane_high_u32, vfmlal_laneq_high_u32,
23180 vfmlsl_lane_low_u32, vfmlsl_lane_high_u32, vfmlslq_laneq_low_u32,
23181 vfmlslq_lane_low_u32, vfmlsl_laneq_low_u32, vfmlslq_laneq_high_u32,
23182 vfmlslq_lane_high_u32, vfmlsl_laneq_high_u32): Define.
23183 * config/arm/arm_neon_builtins.def (vfmal_lane_low,
23184 vfmal_lane_lowv4hf, vfmal_lane_lowv8hf, vfmal_lane_high,
23185 vfmal_lane_highv4hf, vfmal_lane_highv8hf, vfmsl_lane_low,
23186 vfmsl_lane_lowv4hf, vfmsl_lane_lowv8hf, vfmsl_lane_high,
23187 vfmsl_lane_highv4hf, vfmsl_lane_highv8hf): New sets of builtins.
23188 * config/arm/iterators.md (VFMLSEL2, vfmlsel2): New mode attributes.
23189 (V_lane_reg): Likewise.
23190 * config/arm/neon.md (neon_vfm<vfml_op>l_lane_<vfml_half><VCVTF:mode>):
23192 (neon_vfm<vfml_op>l_lane_<vfml_half><vfmlsel2><mode>): Likewise.
23193 (vfmal_lane_low<mode>_intrinsic,
23194 vfmal_lane_low<vfmlsel2><mode>_intrinsic,
23195 vfmal_lane_high<vfmlsel2><mode>_intrinsic,
23196 vfmal_lane_high<mode>_intrinsic, vfmsl_lane_low<mode>_intrinsic,
23197 vfmsl_lane_low<vfmlsel2><mode>_intrinsic,
23198 vfmsl_lane_high<vfmlsel2><mode>_intrinsic,
23199 vfmsl_lane_high<mode>_intrinsic): New define_insns.
23201 2018-01-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23203 * config/arm/arm-cpus.in (fp16fml): New feature.
23204 (ALL_SIMD): Add fp16fml.
23205 (armv8.2-a): Add fp16fml as an option.
23206 (armv8.3-a): Likewise.
23207 (armv8.4-a): Add fp16fml as part of fp16.
23208 * config/arm/arm.h (TARGET_FP16FML): Define.
23209 * config/arm/arm-c.c (arm_cpu_builtins): Define __ARM_FEATURE_FP16_FML
23211 * config/arm/arm-modes.def (V2HF): Define.
23212 * config/arm/arm_neon.h (vfmlal_low_u32, vfmlsl_low_u32,
23213 vfmlal_high_u32, vfmlsl_high_u32, vfmlalq_low_u32,
23214 vfmlslq_low_u32, vfmlalq_high_u32, vfmlslq_high_u32): Define.
23215 * config/arm/arm_neon_builtins.def (vfmal_low, vfmal_high,
23216 vfmsl_low, vfmsl_high): New set of builtins.
23217 * config/arm/iterators.md (PLUSMINUS): New code iterator.
23218 (vfml_op): New code attribute.
23219 (VFMLHALVES): New int iterator.
23220 (VFML, VFMLSEL): New mode attributes.
23221 (V_reg): Define mapping for V2HF.
23222 (V_hi, V_lo): New mode attributes.
23223 (VF_constraint): Likewise.
23224 (vfml_half, vfml_half_selector): New int attributes.
23225 * config/arm/neon.md (neon_vfm<vfml_op>l_<vfml_half><mode>): New
23227 (vfmal_low<mode>_intrinsic, vfmsl_high<mode>_intrinsic,
23228 vfmal_high<mode>_intrinsic, vfmsl_low<mode>_intrinsic):
23230 * config/arm/t-arm-elf (v8_fps): Add fp16fml.
23231 * config/arm/t-multilib (v8_2_a_simd_variants): Add fp16fml.
23232 * config/arm/unspecs.md (UNSPEC_VFML_LO, UNSPEC_VFML_HI): New unspecs.
23233 * doc/invoke.texi (ARM Options): Document fp16fml. Update armv8.4-a
23235 * doc/sourcebuild.texi (arm_fp16fml_neon_ok, arm_fp16fml_neon):
23236 Document new effective target and option set.
23238 2018-01-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23240 * config/arm/arm-cpus.in (armv8_4): New feature.
23241 (ARMv8_4a): New fgroup.
23242 (armv8.4-a): New arch.
23243 * config/arm/arm-tables.opt: Regenerate.
23244 * config/arm/t-aprofile: Add matching rules for -march=armv8.4-a.
23245 * config/arm/t-arm-elf (all_v8_archs): Add armv8.4-a.
23246 * config/arm/t-multilib (v8_4_a_simd_variants): New variable.
23247 Add matching rules for -march=armv8.4-a and extensions.
23248 * doc/invoke.texi (ARM Options): Document -march=armv8.4-a.
23250 2018-01-11 Oleg Endo <olegendo@gcc.gnu.org>
23253 * config/rx/rx.md (BW): New mode attribute.
23254 (sync_lock_test_and_setsi): Add mode suffix to insn output.
23256 2018-01-11 Richard Biener <rguenther@suse.de>
23258 PR tree-optimization/83435
23259 * graphite.c (canonicalize_loop_form): Ignore fake loop exit edges.
23260 * graphite-scop-detection.c (scop_detection::get_sese): Likewise.
23261 * tree-vrp.c (add_assert_info): Drop TREE_OVERFLOW if they appear.
23263 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
23264 Alan Hayward <alan.hayward@arm.com>
23265 David Sherwood <david.sherwood@arm.com>
23267 * config/aarch64/aarch64.c (aarch64_address_info): Add a const_offset
23269 (aarch64_classify_address): Initialize it. Track polynomial offsets.
23270 (aarch64_print_address_internal): Use it to check for a zero offset.
23272 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
23273 Alan Hayward <alan.hayward@arm.com>
23274 David Sherwood <david.sherwood@arm.com>
23276 * config/aarch64/aarch64-modes.def (NUM_POLY_INT_COEFFS): Set to 2.
23277 * config/aarch64/aarch64-protos.h (aarch64_initial_elimination_offset):
23278 Return a poly_int64 rather than a HOST_WIDE_INT.
23279 (aarch64_offset_7bit_signed_scaled_p): Take the offset as a poly_int64
23280 rather than a HOST_WIDE_INT.
23281 * config/aarch64/aarch64.h (aarch64_frame): Protect with
23282 HAVE_POLY_INT_H rather than HOST_WIDE_INT. Change locals_offset,
23283 hard_fp_offset, frame_size, initial_adjust, callee_offset and
23284 final_offset from HOST_WIDE_INT to poly_int64.
23285 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args): Use
23286 to_constant when getting the number of units in an Advanced SIMD
23288 (aarch64_builtin_vectorized_function): Check for a constant number
23290 * config/aarch64/aarch64-simd.md (mov<mode>): Handle polynomial
23292 (aarch64_ld<VSTRUCT:nregs>_lane<VALLDIF:mode>): Use the nunits
23293 attribute instead of GET_MODE_NUNITS.
23294 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs)
23295 (aarch64_class_max_nregs): Use the constant_lowest_bound of the
23296 GET_MODE_SIZE for fixed-size registers.
23297 (aarch64_const_vec_all_same_in_range_p): Use const_vec_duplicate_p.
23298 (aarch64_hard_regno_call_part_clobbered, aarch64_classify_index)
23299 (aarch64_mode_valid_for_sched_fusion_p, aarch64_classify_address)
23300 (aarch64_legitimize_address_displacement, aarch64_secondary_reload)
23301 (aarch64_print_operand, aarch64_print_address_internal)
23302 (aarch64_address_cost, aarch64_rtx_costs, aarch64_register_move_cost)
23303 (aarch64_short_vector_p, aapcs_vfp_sub_candidate)
23304 (aarch64_simd_attr_length_rglist, aarch64_operands_ok_for_ldpstp):
23305 Handle polynomial GET_MODE_SIZE.
23306 (aarch64_hard_regno_caller_save_mode): Likewise. Return modes
23307 wider than SImode without modification.
23308 (tls_symbolic_operand_type): Use strip_offset instead of split_const.
23309 (aarch64_pass_by_reference, aarch64_layout_arg, aarch64_pad_reg_upward)
23310 (aarch64_gimplify_va_arg_expr): Assert that we don't yet handle
23311 passing and returning SVE modes.
23312 (aarch64_function_value, aarch64_layout_arg): Use gen_int_mode
23313 rather than GEN_INT.
23314 (aarch64_emit_probe_stack_range): Take the size as a poly_int64
23315 rather than a HOST_WIDE_INT, but call sorry if it isn't constant.
23316 (aarch64_allocate_and_probe_stack_space): Likewise.
23317 (aarch64_layout_frame): Cope with polynomial offsets.
23318 (aarch64_save_callee_saves, aarch64_restore_callee_saves): Take the
23319 start_offset as a poly_int64 rather than a HOST_WIDE_INT. Track
23320 polynomial offsets.
23321 (offset_9bit_signed_unscaled_p, offset_12bit_unsigned_scaled_p)
23322 (aarch64_offset_7bit_signed_scaled_p): Take the offset as a
23323 poly_int64 rather than a HOST_WIDE_INT.
23324 (aarch64_get_separate_components, aarch64_process_components)
23325 (aarch64_expand_prologue, aarch64_expand_epilogue)
23326 (aarch64_use_return_insn_p): Handle polynomial frame offsets.
23327 (aarch64_anchor_offset): New function, split out from...
23328 (aarch64_legitimize_address): ...here.
23329 (aarch64_builtin_vectorization_cost): Handle polynomial
23330 TYPE_VECTOR_SUBPARTS.
23331 (aarch64_simd_check_vect_par_cnst_half): Handle polynomial
23333 (aarch64_simd_make_constant, aarch64_expand_vector_init): Get the
23334 number of elements from the PARALLEL rather than the mode.
23335 (aarch64_shift_truncation_mask): Use GET_MODE_UNIT_BITSIZE
23336 rather than GET_MODE_BITSIZE.
23337 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_ext)
23338 (aarch64_evpc_rev, aarch64_evpc_dup, aarch64_evpc_zip)
23339 (aarch64_expand_vec_perm_const_1): Handle polynomial
23340 d->perm.length () and d->perm elements.
23341 (aarch64_evpc_tbl): Likewise. Use nelt rather than GET_MODE_NUNITS.
23342 Apply to_constant to d->perm elements.
23343 (aarch64_simd_valid_immediate, aarch64_vec_fpconst_pow_of_2): Handle
23344 polynomial CONST_VECTOR_NUNITS.
23345 (aarch64_move_pointer): Take amount as a poly_int64 rather
23347 (aarch64_progress_pointer): Avoid temporary variable.
23348 * config/aarch64/aarch64.md (aarch64_<crc_variant>): Use
23349 the mode attribute instead of GET_MODE.
23351 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
23352 Alan Hayward <alan.hayward@arm.com>
23353 David Sherwood <david.sherwood@arm.com>
23355 * config/aarch64/aarch64.c (aarch64_force_temporary): Assert that
23356 x exists before using it.
23357 (aarch64_add_constant_internal): Rename to...
23358 (aarch64_add_offset_1): ...this. Replace regnum with separate
23359 src and dest rtxes. Handle the case in which they're different,
23360 including when the offset is zero. Replace scratchreg with an rtx.
23361 Use 2 additions if there is no spare register into which we can
23362 move a 16-bit constant.
23363 (aarch64_add_constant): Delete.
23364 (aarch64_add_offset): Replace reg with separate src and dest
23365 rtxes. Take a poly_int64 offset instead of a HOST_WIDE_INT.
23366 Use aarch64_add_offset_1.
23367 (aarch64_add_sp, aarch64_sub_sp): Take the scratch register as
23368 an rtx rather than an int. Take the delta as a poly_int64
23369 rather than a HOST_WIDE_INT. Use aarch64_add_offset.
23370 (aarch64_expand_mov_immediate): Update uses of aarch64_add_offset.
23371 (aarch64_expand_prologue): Update calls to aarch64_sub_sp,
23372 aarch64_allocate_and_probe_stack_space and aarch64_add_offset.
23373 (aarch64_expand_epilogue): Update calls to aarch64_add_offset
23374 and aarch64_add_sp.
23375 (aarch64_output_mi_thunk): Use aarch64_add_offset rather than
23376 aarch64_add_constant.
23378 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
23380 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
23381 Use scalar_float_mode.
23383 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
23385 * config/aarch64/aarch64-simd.md
23386 (aarch64_fml<f16mac1>l<f16quad>_low<mode>): Avoid GET_MODE_NUNITS.
23387 (aarch64_fml<f16mac1>l<f16quad>_high<mode>): Likewise.
23388 (aarch64_fml<f16mac1>l_lane_lowv2sf): Likewise.
23389 (aarch64_fml<f16mac1>l_lane_highv2sf): Likewise.
23390 (aarch64_fml<f16mac1>lq_laneq_lowv4sf): Likewise.
23391 (aarch64_fml<f16mac1>lq_laneq_highv4sf): Likewise.
23392 (aarch64_fml<f16mac1>l_laneq_lowv2sf): Likewise.
23393 (aarch64_fml<f16mac1>l_laneq_highv2sf): Likewise.
23394 (aarch64_fml<f16mac1>lq_lane_lowv4sf): Likewise.
23395 (aarch64_fml<f16mac1>lq_lane_highv4sf): Likewise.
23397 2018-01-11 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
23400 * config/arm/arm.c (arm_declare_function_name): Set arch_to_print if
23401 targ_options->x_arm_arch_string is non NULL.
23403 2018-01-11 Tamar Christina <tamar.christina@arm.com>
23405 * config/aarch64/aarch64.h
23406 (AARCH64_FL_FOR_ARCH8_4): Add AARCH64_FL_DOTPROD.
23408 2018-01-11 Sudakshina Das <sudi.das@arm.com>
23411 * expmed.c (emit_store_flag_force): Swap if const op0
23412 and change VOIDmode to mode of op0.
23414 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
23416 PR rtl-optimization/83761
23417 * caller-save.c (replace_reg_with_saved_mem): Pass bits rather
23418 than bytes to mode_for_size.
23420 2018-01-10 Jan Hubicka <hubicka@ucw.cz>
23422 PR middle-end/83189
23423 * gfortran.fortran-torture/compile/pr83189.f90: New testcase.
23424 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Handle zero
23427 2018-01-10 Jan Hubicka <hubicka@ucw.cz>
23429 PR middle-end/83575
23430 * cfgrtl.c (rtl_verify_edges): Only verify fixability of partition
23431 when in layout mode.
23432 (cfg_layout_finalize): Do not verify cfg before we are out of layout.
23433 * cfgcleanup.c (try_optimize_cfg): Only verify flow info when doing
23436 2018-01-10 Michael Collison <michael.collison@arm.com>
23438 * config/aarch64/aarch64-modes.def (V2HF): New VECTOR_MODE.
23439 * config/aarch64/aarch64-option-extension.def: Add
23440 AARCH64_OPT_EXTENSION of 'fp16fml'.
23441 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
23442 (__ARM_FEATURE_FP16_FML): Define if TARGET_F16FML is true.
23443 * config/aarch64/predicates.md (aarch64_lane_imm3): New predicate.
23444 * config/aarch64/constraints.md (Ui7): New constraint.
23445 * config/aarch64/iterators.md (VFMLA_W): New mode iterator.
23446 (VFMLA_SEL_W): Ditto.
23449 (VFMLA16_LOW): New int iterator.
23450 (VFMLA16_HIGH): Ditto.
23451 (UNSPEC_FMLAL): New unspec.
23452 (UNSPEC_FMLSL): Ditto.
23453 (UNSPEC_FMLAL2): Ditto.
23454 (UNSPEC_FMLSL2): Ditto.
23455 (f16mac): New code attribute.
23456 * config/aarch64/aarch64-simd-builtins.def
23457 (aarch64_fmlal_lowv2sf): Ditto.
23458 (aarch64_fmlsl_lowv2sf): Ditto.
23459 (aarch64_fmlalq_lowv4sf): Ditto.
23460 (aarch64_fmlslq_lowv4sf): Ditto.
23461 (aarch64_fmlal_highv2sf): Ditto.
23462 (aarch64_fmlsl_highv2sf): Ditto.
23463 (aarch64_fmlalq_highv4sf): Ditto.
23464 (aarch64_fmlslq_highv4sf): Ditto.
23465 (aarch64_fmlal_lane_lowv2sf): Ditto.
23466 (aarch64_fmlsl_lane_lowv2sf): Ditto.
23467 (aarch64_fmlal_laneq_lowv2sf): Ditto.
23468 (aarch64_fmlsl_laneq_lowv2sf): Ditto.
23469 (aarch64_fmlalq_lane_lowv4sf): Ditto.
23470 (aarch64_fmlsl_lane_lowv4sf): Ditto.
23471 (aarch64_fmlalq_laneq_lowv4sf): Ditto.
23472 (aarch64_fmlsl_laneq_lowv4sf): Ditto.
23473 (aarch64_fmlal_lane_highv2sf): Ditto.
23474 (aarch64_fmlsl_lane_highv2sf): Ditto.
23475 (aarch64_fmlal_laneq_highv2sf): Ditto.
23476 (aarch64_fmlsl_laneq_highv2sf): Ditto.
23477 (aarch64_fmlalq_lane_highv4sf): Ditto.
23478 (aarch64_fmlsl_lane_highv4sf): Ditto.
23479 (aarch64_fmlalq_laneq_highv4sf): Ditto.
23480 (aarch64_fmlsl_laneq_highv4sf): Ditto.
23481 * config/aarch64/aarch64-simd.md:
23482 (aarch64_fml<f16mac1>l<f16quad>_low<mode>): New pattern.
23483 (aarch64_fml<f16mac1>l<f16quad>_high<mode>): Ditto.
23484 (aarch64_simd_fml<f16mac1>l<f16quad>_low<mode>): Ditto.
23485 (aarch64_simd_fml<f16mac1>l<f16quad>_high<mode>): Ditto.
23486 (aarch64_fml<f16mac1>l_lane_lowv2sf): Ditto.
23487 (aarch64_fml<f16mac1>l_lane_highv2sf): Ditto.
23488 (aarch64_simd_fml<f16mac>l_lane_lowv2sf): Ditto.
23489 (aarch64_simd_fml<f16mac>l_lane_highv2sf): Ditto.
23490 (aarch64_fml<f16mac1>lq_laneq_lowv4sf): Ditto.
23491 (aarch64_fml<f16mac1>lq_laneq_highv4sf): Ditto.
23492 (aarch64_simd_fml<f16mac>lq_laneq_lowv4sf): Ditto.
23493 (aarch64_simd_fml<f16mac>lq_laneq_highv4sf): Ditto.
23494 (aarch64_fml<f16mac1>l_laneq_lowv2sf): Ditto.
23495 (aarch64_fml<f16mac1>l_laneq_highv2sf): Ditto.
23496 (aarch64_simd_fml<f16mac>l_laneq_lowv2sf): Ditto.
23497 (aarch64_simd_fml<f16mac>l_laneq_highv2sf): Ditto.
23498 (aarch64_fml<f16mac1>lq_lane_lowv4sf): Ditto.
23499 (aarch64_fml<f16mac1>lq_lane_highv4sf): Ditto.
23500 (aarch64_simd_fml<f16mac>lq_lane_lowv4sf): Ditto.
23501 (aarch64_simd_fml<f16mac>lq_lane_highv4sf): Ditto.
23502 * config/aarch64/arm_neon.h (vfmlal_low_u32): New intrinsic.
23503 (vfmlsl_low_u32): Ditto.
23504 (vfmlalq_low_u32): Ditto.
23505 (vfmlslq_low_u32): Ditto.
23506 (vfmlal_high_u32): Ditto.
23507 (vfmlsl_high_u32): Ditto.
23508 (vfmlalq_high_u32): Ditto.
23509 (vfmlslq_high_u32): Ditto.
23510 (vfmlal_lane_low_u32): Ditto.
23511 (vfmlsl_lane_low_u32): Ditto.
23512 (vfmlal_laneq_low_u32): Ditto.
23513 (vfmlsl_laneq_low_u32): Ditto.
23514 (vfmlalq_lane_low_u32): Ditto.
23515 (vfmlslq_lane_low_u32): Ditto.
23516 (vfmlalq_laneq_low_u32): Ditto.
23517 (vfmlslq_laneq_low_u32): Ditto.
23518 (vfmlal_lane_high_u32): Ditto.
23519 (vfmlsl_lane_high_u32): Ditto.
23520 (vfmlal_laneq_high_u32): Ditto.
23521 (vfmlsl_laneq_high_u32): Ditto.
23522 (vfmlalq_lane_high_u32): Ditto.
23523 (vfmlslq_lane_high_u32): Ditto.
23524 (vfmlalq_laneq_high_u32): Ditto.
23525 (vfmlslq_laneq_high_u32): Ditto.
23526 * config/aarch64/aarch64.h (AARCH64_FL_F16SML): New flag.
23527 (AARCH64_FL_FOR_ARCH8_4): New.
23528 (AARCH64_ISA_F16FML): New ISA flag.
23529 (TARGET_F16FML): New feature flag for fp16fml.
23530 (doc/invoke.texi): Document new fp16fml option.
23532 2018-01-10 Michael Collison <michael.collison@arm.com>
23534 * config/aarch64/aarch64-builtins.c:
23535 (aarch64_types_ternopu_imm_qualifiers, TYPES_TERNOPUI): New.
23536 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
23537 (__ARM_FEATURE_SHA3): Define if TARGET_SHA3 is true.
23538 * config/aarch64/aarch64.h (AARCH64_FL_SHA3): New flags.
23539 (AARCH64_ISA_SHA3): New ISA flag.
23540 (TARGET_SHA3): New feature flag for sha3.
23541 * config/aarch64/iterators.md (sha512_op): New int attribute.
23542 (CRYPTO_SHA512): New int iterator.
23543 (UNSPEC_SHA512H): New unspec.
23544 (UNSPEC_SHA512H2): Ditto.
23545 (UNSPEC_SHA512SU0): Ditto.
23546 (UNSPEC_SHA512SU1): Ditto.
23547 * config/aarch64/aarch64-simd-builtins.def
23548 (aarch64_crypto_sha512hqv2di): New builtin.
23549 (aarch64_crypto_sha512h2qv2di): Ditto.
23550 (aarch64_crypto_sha512su0qv2di): Ditto.
23551 (aarch64_crypto_sha512su1qv2di): Ditto.
23552 (aarch64_eor3qv8hi): Ditto.
23553 (aarch64_rax1qv2di): Ditto.
23554 (aarch64_xarqv2di): Ditto.
23555 (aarch64_bcaxqv8hi): Ditto.
23556 * config/aarch64/aarch64-simd.md:
23557 (aarch64_crypto_sha512h<sha512_op>qv2di): New pattern.
23558 (aarch64_crypto_sha512su0qv2di): Ditto.
23559 (aarch64_crypto_sha512su1qv2di): Ditto.
23560 (aarch64_eor3qv8hi): Ditto.
23561 (aarch64_rax1qv2di): Ditto.
23562 (aarch64_xarqv2di): Ditto.
23563 (aarch64_bcaxqv8hi): Ditto.
23564 * config/aarch64/arm_neon.h (vsha512hq_u64): New intrinsic.
23565 (vsha512h2q_u64): Ditto.
23566 (vsha512su0q_u64): Ditto.
23567 (vsha512su1q_u64): Ditto.
23568 (veor3q_u16): Ditto.
23569 (vrax1q_u64): Ditto.
23570 (vxarq_u64): Ditto.
23571 (vbcaxq_u16): Ditto.
23572 * config/arm/types.md (crypto_sha512): New type attribute.
23573 (crypto_sha3): Ditto.
23574 (doc/invoke.texi): Document new sha3 option.
23576 2018-01-10 Michael Collison <michael.collison@arm.com>
23578 * config/aarch64/aarch64-builtins.c:
23579 (aarch64_types_quadopu_imm_qualifiers, TYPES_QUADOPUI): New.
23580 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
23581 (__ARM_FEATURE_SM3): Define if TARGET_SM4 is true.
23582 (__ARM_FEATURE_SM4): Define if TARGET_SM4 is true.
23583 * config/aarch64/aarch64.h (AARCH64_FL_SM4): New flags.
23584 (AARCH64_ISA_SM4): New ISA flag.
23585 (TARGET_SM4): New feature flag for sm4.
23586 * config/aarch64/aarch64-simd-builtins.def
23587 (aarch64_sm3ss1qv4si): Ditto.
23588 (aarch64_sm3tt1aq4si): Ditto.
23589 (aarch64_sm3tt1bq4si): Ditto.
23590 (aarch64_sm3tt2aq4si): Ditto.
23591 (aarch64_sm3tt2bq4si): Ditto.
23592 (aarch64_sm3partw1qv4si): Ditto.
23593 (aarch64_sm3partw2qv4si): Ditto.
23594 (aarch64_sm4eqv4si): Ditto.
23595 (aarch64_sm4ekeyqv4si): Ditto.
23596 * config/aarch64/aarch64-simd.md:
23597 (aarch64_sm3ss1qv4si): Ditto.
23598 (aarch64_sm3tt<sm3tt_op>qv4si): Ditto.
23599 (aarch64_sm3partw<sm3part_op>qv4si): Ditto.
23600 (aarch64_sm4eqv4si): Ditto.
23601 (aarch64_sm4ekeyqv4si): Ditto.
23602 * config/aarch64/iterators.md (sm3tt_op): New int iterator.
23603 (sm3part_op): Ditto.
23604 (CRYPTO_SM3TT): Ditto.
23605 (CRYPTO_SM3PART): Ditto.
23606 (UNSPEC_SM3SS1): New unspec.
23607 (UNSPEC_SM3TT1A): Ditto.
23608 (UNSPEC_SM3TT1B): Ditto.
23609 (UNSPEC_SM3TT2A): Ditto.
23610 (UNSPEC_SM3TT2B): Ditto.
23611 (UNSPEC_SM3PARTW1): Ditto.
23612 (UNSPEC_SM3PARTW2): Ditto.
23613 (UNSPEC_SM4E): Ditto.
23614 (UNSPEC_SM4EKEY): Ditto.
23615 * config/aarch64/constraints.md (Ui2): New constraint.
23616 * config/aarch64/predicates.md (aarch64_imm2): New predicate.
23617 * config/arm/types.md (crypto_sm3): New type attribute.
23618 (crypto_sm4): Ditto.
23619 * config/aarch64/arm_neon.h (vsm3ss1q_u32): New intrinsic.
23620 (vsm3tt1aq_u32): Ditto.
23621 (vsm3tt1bq_u32): Ditto.
23622 (vsm3tt2aq_u32): Ditto.
23623 (vsm3tt2bq_u32): Ditto.
23624 (vsm3partw1q_u32): Ditto.
23625 (vsm3partw2q_u32): Ditto.
23626 (vsm4eq_u32): Ditto.
23627 (vsm4ekeyq_u32): Ditto.
23628 (doc/invoke.texi): Document new sm4 option.
23630 2018-01-10 Michael Collison <michael.collison@arm.com>
23632 * config/aarch64/aarch64-arches.def (armv8.4-a): New architecture.
23633 * config/aarch64/aarch64.h (AARCH64_ISA_V8_4): New ISA flag.
23634 (AARCH64_FL_FOR_ARCH8_4): New.
23635 (AARCH64_FL_V8_4): New flag.
23636 (doc/invoke.texi): Document new armv8.4-a option.
23638 2018-01-10 Michael Collison <michael.collison@arm.com>
23640 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
23641 (__ARM_FEATURE_AES): Define if TARGET_AES is true.
23642 (__ARM_FEATURE_SHA2): Define if TARGET_SHA2 is true.
23643 * config/aarch64/aarch64-option-extension.def: Add
23644 AARCH64_OPT_EXTENSION of 'sha2'.
23645 (aes): Add AARCH64_OPT_EXTENSION of 'aes'.
23646 (crypto): Disable sha2 and aes if crypto disabled.
23647 (crypto): Enable aes and sha2 if enabled.
23648 (simd): Disable sha2 and aes if simd disabled.
23649 * config/aarch64/aarch64.h (AARCH64_FL_AES, AARCH64_FL_SHA2):
23651 (AARCH64_ISA_AES, AARCH64_ISA_SHA2): New ISA flags.
23652 (TARGET_SHA2): New feature flag for sha2.
23653 (TARGET_AES): New feature flag for aes.
23654 * config/aarch64/aarch64-simd.md:
23655 (aarch64_crypto_aes<aes_op>v16qi): Make pattern
23656 conditional on TARGET_AES.
23657 (aarch64_crypto_aes<aesmc_op>v16qi): Ditto.
23658 (aarch64_crypto_sha1hsi): Make pattern conditional
23660 (aarch64_crypto_sha1hv4si): Ditto.
23661 (aarch64_be_crypto_sha1hv4si): Ditto.
23662 (aarch64_crypto_sha1su1v4si): Ditto.
23663 (aarch64_crypto_sha1<sha1_op>v4si): Ditto.
23664 (aarch64_crypto_sha1su0v4si): Ditto.
23665 (aarch64_crypto_sha256h<sha256_op>v4si): Ditto.
23666 (aarch64_crypto_sha256su0v4si): Ditto.
23667 (aarch64_crypto_sha256su1v4si): Ditto.
23668 (doc/invoke.texi): Document new aes and sha2 options.
23670 2018-01-10 Martin Sebor <msebor@redhat.com>
23672 PR tree-optimization/83781
23673 * gimple-fold.c (get_range_strlen): Avoid treating arrays of pointers
23676 2018-01-11 Martin Sebor <msebor@gmail.com>
23677 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
23679 PR tree-optimization/83501
23680 PR tree-optimization/81703
23682 * tree-ssa-strlen.c (get_string_cst): Rename...
23683 (get_string_len): ...to this. Handle global constants.
23684 (handle_char_store): Adjust.
23686 2018-01-10 Kito Cheng <kito.cheng@gmail.com>
23687 Jim Wilson <jimw@sifive.com>
23689 * config/riscv/riscv-protos.h (riscv_output_return): New.
23690 * config/riscv/riscv.c (struct machine_function): New naked_p field.
23691 (riscv_attribute_table, riscv_output_return),
23692 (riscv_handle_fndecl_attribute, riscv_naked_function_p),
23693 (riscv_allocate_stack_slots_for_args, riscv_warn_func_return): New.
23694 (riscv_compute_frame_info): Only compute frame->mask if not a naked
23696 (riscv_expand_prologue): Add early return for naked function.
23697 (riscv_expand_epilogue): Likewise.
23698 (riscv_function_ok_for_sibcall): Return false for naked function.
23699 (riscv_set_current_function): New.
23700 (TARGET_SET_CURRENT_FUNCTION, TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS),
23701 (TARGET_ATTRIBUTE_TABLE, TARGET_WARN_FUNC_RETURN): New.
23702 * config/riscv/riscv.md (simple_return): Call riscv_output_return.
23703 * doc/extend.texi (RISC-V Function Attributes): New.
23705 2018-01-10 Michael Meissner <meissner@linux.vnet.ibm.com>
23707 * config/rs6000/rs6000.c (is_complex_IBM_long_double): Explicitly
23708 check for 128-bit long double before checking TCmode.
23709 * config/rs6000/rs6000.h (FLOAT128_IEEE_P): Explicitly check for
23710 128-bit long doubles before checking TFmode or TCmode.
23711 (FLOAT128_IBM_P): Likewise.
23713 2018-01-10 Martin Sebor <msebor@redhat.com>
23715 PR tree-optimization/83671
23716 * builtins.c (c_strlen): Unconditionally return zero for the empty
23718 Use -Warray-bounds for warnings.
23719 * gimple-fold.c (get_range_strlen): Handle non-constant lengths
23720 for non-constant array indices with COMPONENT_REF, arrays of
23721 arrays, and pointers to arrays.
23722 (gimple_fold_builtin_strlen): Determine and set length range for
23723 non-constant character arrays.
23725 2018-01-10 Aldy Hernandez <aldyh@redhat.com>
23727 PR middle-end/81897
23728 * tree-ssa-uninit.c (convert_control_dep_chain_into_preds): Skip
23731 2018-01-10 Eric Botcazou <ebotcazou@adacore.com>
23733 * dwarf2out.c (dwarf2out_var_location): Do not pass NULL to fprintf.
23735 2018-01-10 Peter Bergner <bergner@vnet.ibm.com>
23738 * config/rs6000/rs6000.c (print_operand) <'y'>: Use
23739 VECTOR_MEM_ALTIVEC_OR_VSX_P.
23740 * config/rs6000/vsx.md (*vsx_le_perm_load_<mode> for VSX_D): Use
23741 indexed_or_indirect_operand predicate.
23742 (*vsx_le_perm_load_<mode> for VSX_W): Likewise.
23743 (*vsx_le_perm_load_v8hi): Likewise.
23744 (*vsx_le_perm_load_v16qi): Likewise.
23745 (*vsx_le_perm_store_<mode> for VSX_D): Likewise.
23746 (*vsx_le_perm_store_<mode> for VSX_W): Likewise.
23747 (*vsx_le_perm_store_v8hi): Likewise.
23748 (*vsx_le_perm_store_v16qi): Likewise.
23749 (eight unnamed splitters): Likewise.
23751 2018-01-10 Peter Bergner <bergner@vnet.ibm.com>
23753 * config/rs6000/x86intrin.h: Change #warning to #error. Update message.
23754 * config/rs6000/emmintrin.h: Likewise.
23755 * config/rs6000/mmintrin.h: Likewise.
23756 * config/rs6000/xmmintrin.h: Likewise.
23758 2018-01-10 David Malcolm <dmalcolm@redhat.com>
23761 * tree-core.h: Document EXPR_LOCATION_WRAPPER_P's usage of
23763 * tree.c (tree_nop_conversion): Return true for location wrapper
23765 (maybe_wrap_with_location): New function.
23766 (selftest::check_strip_nops): New function.
23767 (selftest::test_location_wrappers): New function.
23768 (selftest::tree_c_tests): Call it.
23769 * tree.h (STRIP_ANY_LOCATION_WRAPPER): New macro.
23770 (maybe_wrap_with_location): New decl.
23771 (EXPR_LOCATION_WRAPPER_P): New macro.
23772 (location_wrapper_p): New inline function.
23773 (tree_strip_any_location_wrapper): New inline function.
23775 2018-01-10 H.J. Lu <hongjiu.lu@intel.com>
23778 * config/i386/i386.c (ix86_compute_frame_layout): Always adjust
23779 stack_realign_offset for the largest alignment of stack slot
23781 (ix86_find_max_used_stack_alignment): New function.
23782 (ix86_finalize_stack_frame_flags): Use it. Set
23783 max_used_stack_alignment if we don't realign stack.
23784 * config/i386/i386.h (machine_function): Add
23785 max_used_stack_alignment.
23787 2018-01-10 Christophe Lyon <christophe.lyon@linaro.org>
23789 * config/arm/arm.opt (-mbranch-cost): New option.
23790 * config/arm/arm.h (BRANCH_COST): Take arm_branch_cost into
23793 2018-01-10 Segher Boessenkool <segher@kernel.crashing.org>
23796 * config/rs6000/rs6000.md (load_toc_v4_PIC_2, load_toc_v4_PIC_3b,
23797 load_toc_v4_PIC_3c): Wrap const term in CONST RTL.
23799 2018-01-10 Richard Biener <rguenther@suse.de>
23802 * dwarf2out.c (gen_subprogram_die): Hoist old_die && declaration
23803 early out so it also covers the case where we have a non-NULL
23806 2018-01-10 Richard Sandiford <richard.sandiford@linaro.org>
23808 PR tree-optimization/83753
23809 * tree-vect-stmts.c (get_group_load_store_type): Use VMAT_CONTIGUOUS
23810 for non-strided grouped accesses if the number of elements is 1.
23812 2018-01-10 Jan Hubicka <hubicka@ucw.cz>
23815 * i386.c (ix86_vectorize_builtin_gather): Check TARGET_USE_GATHER.
23816 * i386.h (TARGET_USE_GATHER): Define.
23817 * x86-tune.def (X86_TUNE_USE_GATHER): New.
23819 2018-01-10 Martin Liska <mliska@suse.cz>
23822 * basic-block.h (CLEANUP_NO_PARTITIONING): New define.
23823 * bb-reorder.c (pass_reorder_blocks::execute): Do not clean up
23825 * cfgcleanup.c (try_optimize_cfg): Fix up partitioning if
23826 CLEANUP_NO_PARTITIONING is not set.
23828 2018-01-10 Richard Sandiford <richard.sandiford@linaro.org>
23830 * doc/rtl.texi: Remove documentation of (const ...) wrappers
23831 for vectors, as a partial revert of r254296.
23832 * rtl.h (const_vec_p): Delete.
23833 (const_vec_duplicate_p): Don't test for vector CONSTs.
23834 (unwrap_const_vec_duplicate, const_vec_series_p): Likewise.
23835 * expmed.c (make_tree): Likewise.
23838 * common.md (E, F): Use CONSTANT_P instead of checking for
23840 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
23841 checking for CONST_VECTOR.
23843 2018-01-09 Jan Hubicka <hubicka@ucw.cz>
23845 PR middle-end/83575
23846 * predict.c (force_edge_cold): Handle in more sane way edges
23847 with no prediction.
23849 2018-01-09 Carl Love <cel@us.ibm.com>
23851 * config/rs6002/altivec.md (p8_vmrgow): Add support for V2DI, V2DF,
23853 (p8_vmrgew): Add support for V2DI, V2DF, V4SF types.
23854 * config/rs6000/rs6000-builtin.def: Add definitions for FLOAT2_V2DF,
23855 VMRGEW_V2DI, VMRGEW_V2DF, VMRGEW_V4SF, VMRGOW_V4SI, VMRGOW_V4SF,
23856 VMRGOW_V2DI, VMRGOW_V2DF. Remove definition for VMRGOW.
23857 * config/rs6000/rs6000-c.c (VSX_BUILTIN_VEC_FLOAT2,
23858 P8V_BUILTIN_VEC_VMRGEW, P8V_BUILTIN_VEC_VMRGOW): Add definitions.
23859 * config/rs6000/rs6000-protos.h: Add extern defition for
23860 rs6000_generate_float2_double_code.
23861 * config/rs6000/rs6000.c (rs6000_generate_float2_double_code): Add
23863 * config/rs6000/vsx.md (vsx_xvcdpsp): Add define_insn.
23864 (float2_v2df): Add define_expand.
23866 2018-01-09 Uros Bizjak <ubizjak@gmail.com>
23869 * combine.c (force_int_to_mode) <case ASHIFT>: Use mode instead of
23870 op_mode in the force_to_mode call.
23872 2018-01-09 Richard Sandiford <richard.sandiford@linaro.org>
23874 * config/aarch64/aarch64.c (aarch64_evpc_trn): Use d.perm.series_p
23875 instead of checking each element individually.
23876 (aarch64_evpc_uzp): Likewise.
23877 (aarch64_evpc_zip): Likewise.
23878 (aarch64_evpc_ext): Likewise.
23879 (aarch64_evpc_rev): Likewise.
23880 (aarch64_evpc_dup): Test the encoding for a single duplicated element,
23881 instead of checking each element individually. Return true without
23883 (aarch64_vectorize_vec_perm_const): Use all_from_input_p to test
23884 whether all selected elements come from the same input, instead of
23885 checking each element individually. Remove calls to gen_rtx_REG,
23886 start_sequence and end_sequence and instead assert that no rtl is
23889 2018-01-09 Richard Sandiford <richard.sandiford@linaro.org>
23891 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p): Fix
23892 order of HIGH and CONST checks.
23894 2018-01-09 Richard Sandiford <richard.sandiford@linaro.org>
23896 * tree-vect-stmts.c (permute_vec_elements): Create a fresh variable
23897 if the destination isn't an SSA_NAME.
23899 2018-01-09 Richard Biener <rguenther@suse.de>
23901 PR tree-optimization/83668
23902 * graphite.c (canonicalize_loop_closed_ssa): Add edge argument,
23904 (canonicalize_loop_form): ... here, renamed from ...
23905 (canonicalize_loop_closed_ssa_form): ... this and amended to
23906 swap successor edges for loop exit blocks to make us use
23907 the RPO order we need for initial schedule generation.
23909 2018-01-09 Joseph Myers <joseph@codesourcery.com>
23911 PR tree-optimization/64811
23912 * match.pd: When optimizing comparisons with Inf, avoid
23913 introducing or losing exceptions from comparisons with NaN.
23915 2018-01-09 Martin Liska <mliska@suse.cz>
23918 * asan.c (shadow_mem_size): Add gcc_assert.
23920 2018-01-09 Georg-Johann Lay <avr@gjlay.de>
23922 Don't save registers in main().
23925 * doc/invoke.texi (AVR Options) [-mmain-is-OS_task]: Document it.
23926 * config/avr/avr.opt (-mmain-is-OS_task): New target option.
23927 * config/avr/avr.c (avr_set_current_function): Don't error if
23928 naked, OS_task or OS_main are specified at the same time.
23929 (avr_function_ok_for_sibcall): Don't disable sibcalls for OS_task,
23931 (avr_insert_attributes) [-mmain-is-OS_task] <main>: Add OS_task
23933 * common/config/avr/avr-common.c (avr_option_optimization_table):
23934 Switch on -mmain-is-OS_task for optimizing compilations.
23936 2018-01-09 Richard Biener <rguenther@suse.de>
23938 PR tree-optimization/83572
23939 * graphite.c: Include cfganal.h.
23940 (graphite_transform_loops): Connect infinite loops to exit
23941 and remove fake edges at the end.
23943 2018-01-09 Jan Hubicka <hubicka@ucw.cz>
23945 * ipa-inline.c (edge_badness): Revert accidental checkin.
23947 2018-01-09 Jan Hubicka <hubicka@ucw.cz>
23950 * ipa-comdats.c (set_comdat_group): Only set comdat group of real
23951 symbols; not inline clones.
23953 2018-01-09 Jakub Jelinek <jakub@redhat.com>
23956 * modulo-sched.c (schedule_reg_moves): Punt if we'd need to move
23957 hard registers. Formatting fixes.
23959 PR preprocessor/83722
23960 * gcc.c (try_generate_repro): Pass
23961 &temp_stderr_files[RETRY_ICE_ATTEMPTS - 1] rather than
23962 &temp_stdout_files[RETRY_ICE_ATTEMPTS - 1] as last argument to
23965 2018-01-08 Monk Chiang <sh.chiang04@gmail.com>
23966 Kito Cheng <kito.cheng@gmail.com>
23968 * config/riscv/riscv.c (machine_function::is_leaf): Remove field.
23969 (riscv_leaf_function_p): Delete.
23970 (riscv_function_ok_for_sibcall): Return false when TARGET_SAVE_RESTORE.
23972 2018-01-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23974 * config/rs6000/rs6000-string.c (do_load_for_compare_from_addr): New
23976 (do_ifelse): New function.
23977 (do_isel): New function.
23978 (do_sub3): New function.
23979 (do_add3): New function.
23980 (do_load_mask_compare): New function.
23981 (do_overlap_load_compare): New function.
23982 (expand_compare_loop): New function.
23983 (expand_block_compare): Call expand_compare_loop() when appropriate.
23984 * config/rs6000/rs6000.opt (-mblock-compare-inline-limit): Change
23985 option description.
23986 (-mblock-compare-inline-loop-limit): New option.
23988 2018-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23991 * config/rs6000/altivec.md (*altivec_vpermr_<mode>_internal):
23992 Reverse order of second and third operands in first alternative.
23993 * config/rs6000/rs6000.c (rs6000_expand_vector_set): Reverse order
23994 of first and second elements in UNSPEC_VPERMR vector.
23995 (altivec_expand_vec_perm_le): Likewise.
23997 2018-01-08 Jeff Law <law@redhat.com>
23999 PR rtl-optimizatin/81308
24000 * tree-switch-conversion.c (cfg_altered): New file scoped static.
24001 (process_switch): If group_case_labels makes a change, then set
24003 (pass_convert_switch::execute): If a switch is converted, then
24004 set cfg_altered. Return TODO_cfg_cleanup if cfg_altered is true.
24006 PR rtl-optimization/81308
24007 * recog.c (split_all_insns): Conditionally cleanup the CFG after
24010 2018-01-08 Vidya Praveen <vidyapraveen@arm.com>
24012 PR target/83663 - Revert r255946
24013 * config/aarch64/aarch64.c (aarch64_expand_vector_init): Modify code
24014 generation for cases where splatting a value is not useful.
24015 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
24016 across a vec_duplicate and a paradoxical subreg forming a vector
24017 mode to a vec_concat.
24019 2018-01-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24021 * config/arm/t-aprofile (MULTILIB_MATCHES): Add mapping rules for
24022 -march=armv8.3-a variants.
24023 * config/arm/t-multilib: Likewise.
24024 * config/arm/t-arm-elf: Likewise. Handle dotprod extension.
24026 2018-01-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
24028 * config/rs6000/rs6000.md (cceq_ior_compare): Remove * so I can use it
24030 (cceq_ior_compare_complement): Give it a name so I can use it, and
24031 change boolean_or_operator predicate to boolean_operator so it can
24032 be used to generate a crand.
24033 (eqne): New code iterator.
24034 (bd/bd_neg): New code_attrs.
24035 (<bd>_<mode>): New name for ctr<mode>_internal[12] now combined into
24036 a single define_insn.
24037 (<bd>tf_<mode>): A new insn pattern for the conditional form branch
24038 decrement (bdnzt/bdnzf/bdzt/bdzf).
24039 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Updated
24040 with the new names of the branch decrement patterns, and added the
24041 names of the branch decrement conditional patterns.
24043 2018-01-08 Richard Biener <rguenther@suse.de>
24045 PR tree-optimization/83563
24046 * graphite.c (canonicalize_loop_closed_ssa_form): Reset the SCEV
24049 2018-01-08 Richard Biener <rguenther@suse.de>
24051 PR middle-end/83713
24052 * convert.c (do_narrow): Properly guard TYPE_OVERFLOW_WRAPS checks.
24054 2018-01-08 Richard Biener <rguenther@suse.de>
24056 PR tree-optimization/83685
24057 * tree-ssa-pre.c (create_expression_by_pieces): Do not insert
24058 references to abnormals.
24060 2018-01-08 Richard Biener <rguenther@suse.de>
24063 * dwarf2out.c (output_indirect_strings): Handle empty
24064 skeleton_debug_str_hash.
24065 (dwarf2out_early_finish): Index strings for -gsplit-dwarf.
24067 2018-01-08 Claudiu Zissulescu <claziss@synopsys.com>
24069 * config/arc/arc.c (TARGET_TRAMPOLINE_ADJUST_ADDRESS): Delete.
24070 (emit_store_direct): Likewise.
24071 (arc_trampoline_adjust_address): Likewise.
24072 (arc_asm_trampoline_template): New function.
24073 (arc_initialize_trampoline): Use asm_trampoline_template.
24074 (TARGET_ASM_TRAMPOLINE_TEMPLATE): Define.
24075 * config/arc/arc.h (TRAMPOLINE_SIZE): Adjust to 16.
24076 * config/arc/arc.md (flush_icache): Delete pattern.
24078 2018-01-08 Claudiu Zissulescu <claziss@synopsys.com>
24080 * config/arc/arc-c.def (__ARC_UNALIGNED__): New define.
24081 * config/arc/arc.h (STRICT_ALIGNMENT): Control this macro using
24084 2018-01-08 Sebastian Huber <sebastian.huber@embedded-brains.de>
24087 * config/epiphany/epiphany.h (make_pass_mode_switch_use): Guard
24088 by not USED_FOR_TARGET.
24089 (make_pass_resolve_sw_modes): Likewise.
24091 2018-01-08 Sebastian Huber <sebastian.huber@embedded-brains.de>
24093 * config/nios2/nios2.h (nios2_section_threshold): Guard by not
24096 2018-01-08 Richard Biener <rguenther@suse.de>
24098 PR middle-end/83580
24099 * tree-data-ref.c (split_constant_offset): Remove STRIP_NOPS.
24101 2018-01-08 Richard Biener <rguenther@suse.de>
24103 PR middle-end/83517
24104 * match.pd ((t * 2) / 2) -> t): Add missing :c.
24106 2018-01-06 Aldy Hernandez <aldyh@redhat.com>
24108 PR middle-end/81897
24109 * tree-ssa-uninit.c (compute_control_dep_chain): Do not bail on
24110 basic blocks with a small number of successors.
24111 (convert_control_dep_chain_into_preds): Improve handling of
24113 (dump_predicates): Split apart into...
24114 (dump_pred_chain): ...here...
24115 (dump_pred_info): ...and here.
24116 (can_one_predicate_be_invalidated_p): Add debugging printfs.
24117 (can_chain_union_be_invalidated_p): Improve check for invalidation
24119 (uninit_uses_cannot_happen): Avoid unnecessary if
24120 convert_control_dep_chain_into_preds yielded nothing.
24122 2018-01-06 Martin Sebor <msebor@redhat.com>
24124 PR tree-optimization/83640
24125 * gimple-ssa-warn-restrict.c (builtin_access::builtin_access): Avoid
24126 subtracting negative offset from size.
24127 (builtin_access::overlap): Adjust offset bounds of the access to fall
24128 within the size of the object if possible.
24130 2018-01-06 Richard Sandiford <richard.sandiford@linaro.org>
24132 PR rtl-optimization/83699
24133 * expmed.c (extract_bit_field_1): Restrict the vector usage of
24134 extract_bit_field_as_subreg to cases in which the extracted
24135 value is also a vector.
24137 * lra-constraints.c (process_alt_operands): Test for the equivalence
24138 substitutions when detecting a possible reload cycle.
24140 2018-01-06 Jakub Jelinek <jakub@redhat.com>
24143 * toplev.c (process_options): Don't enable debug_nonbind_markers_p
24144 by default if flag_selective_schedling{,2}. Formatting fixes.
24146 PR rtl-optimization/83682
24147 * rtl.h (const_vec_duplicate_p): Only return true for VEC_DUPLICATE
24148 if it has non-VECTOR_MODE element mode.
24149 (vec_duplicate_p): Likewise.
24151 PR middle-end/83694
24152 * cfgexpand.c (expand_debug_expr): Punt if mode1 is VOIDmode
24153 and bitsize might be greater than MAX_BITSIZE_MODE_ANY_INT.
24155 2018-01-05 Jakub Jelinek <jakub@redhat.com>
24158 * config/i386/i386-builtin.def
24159 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
24160 __builtin_ia32_vgf2p8affineqb_v64qi, __builtin_ia32_vgf2p8mulb_v64qi):
24161 Require also OPTION_MASK_ISA_AVX512F in addition to
24162 OPTION_MASK_ISA_GFNI.
24163 (__builtin_ia32_vgf2p8affineinvqb_v16qi_mask,
24164 __builtin_ia32_vgf2p8affineqb_v16qi_mask): Require
24165 OPTION_MASK_ISA_AVX512VL instead of OPTION_MASK_ISA_SSE in addition
24166 to OPTION_MASK_ISA_GFNI.
24167 (__builtin_ia32_vgf2p8mulb_v32qi_mask): Require
24168 OPTION_MASK_ISA_AVX512VL in addition to OPTION_MASK_ISA_GFNI and
24169 OPTION_MASK_ISA_AVX512BW.
24170 (__builtin_ia32_vgf2p8mulb_v16qi_mask): Require
24171 OPTION_MASK_ISA_AVX512VL instead of OPTION_MASK_ISA_AVX512BW in
24172 addition to OPTION_MASK_ISA_GFNI.
24173 (__builtin_ia32_vgf2p8affineinvqb_v16qi,
24174 __builtin_ia32_vgf2p8affineqb_v16qi, __builtin_ia32_vgf2p8mulb_v16qi):
24175 Require OPTION_MASK_ISA_SSE2 instead of OPTION_MASK_ISA_SSE in addition
24176 to OPTION_MASK_ISA_GFNI.
24177 * config/i386/i386.c (def_builtin): Change to builtin isa/isa2 being
24178 a requirement for all ISAs rather than any of them with a few
24180 (ix86_add_new_builtins): Clear OPTION_MASK_ISA_64BIT from isa before
24182 (ix86_expand_builtin): Require all ISAs from builtin's isa and isa2
24183 bitmasks to be enabled with 3 exceptions, instead of requiring any
24184 enabled ISA with lots of exceptions.
24185 * config/i386/sse.md (vgf2p8affineinvqb_<mode><mask_name>,
24186 vgf2p8affineqb_<mode><mask_name>, vgf2p8mulb_<mode><mask_name>):
24187 Change avx512bw in isa attribute to avx512f.
24188 * config/i386/sgxintrin.h: Add license boilerplate.
24189 * config/i386/vaesintrin.h: Likewise. Fix macro spelling __AVX512F
24190 to __AVX512F__ and __AVX512VL to __AVX512VL__.
24191 (_mm256_aesdec_epi128, _mm256_aesdeclast_epi128, _mm256_aesenc_epi128,
24192 _mm256_aesenclast_epi128): Enable temporarily avx if __AVX__ is not
24194 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8,
24195 _mm_gf2p8affineinv_epi64_epi8, _mm_gf2p8affine_epi64_epi8): Enable
24196 temporarily sse2 rather than sse if not enabled already.
24199 * config/i386/sse.md (VI248_VLBW): Rename to ...
24200 (VI248_AVX512VL): ... this. Don't guard V32HI with TARGET_AVX512BW.
24201 (vpshrd_<mode><mask_name>, vpshld_<mode><mask_name>,
24202 vpshrdv_<mode>, vpshrdv_<mode>_mask, vpshrdv_<mode>_maskz,
24203 vpshrdv_<mode>_maskz_1, vpshldv_<mode>, vpshldv_<mode>_mask,
24204 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1): Use VI248_AVX512VL
24205 mode iterator instead of VI248_VLBW.
24207 2018-01-05 Jan Hubicka <hubicka@ucw.cz>
24209 * ipa-fnsummary.c (record_modified_bb_info): Add OP.
24210 (record_modified): Skip clobbers; add debug output.
24211 (param_change_prob): Use sreal frequencies.
24213 2018-01-05 Richard Sandiford <richard.sandiford@linaro.org>
24215 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
24216 punt for user-aligned variables.
24218 2018-01-05 Richard Sandiford <richard.sandiford@linaro.org>
24220 * tree-chrec.c (chrec_contains_symbols): Return true for
24223 2018-01-05 Sudakshina Das <sudi.das@arm.com>
24226 * simplify-rtx.c (simplify_relational_operation_1): Add simplifications
24227 of (x|y) == x for BICS pattern.
24229 2018-01-05 Jakub Jelinek <jakub@redhat.com>
24231 PR tree-optimization/83605
24232 * gimple-ssa-strength-reduction.c: Include tree-eh.h.
24233 (find_candidates_dom_walker::before_dom_children): Ignore stmts that
24236 2018-01-05 Sebastian Huber <sebastian.huber@embedded-brains.de>
24238 * config.gcc (epiphany-*-elf*): Add (epiphany-*-rtems*) configuration.
24239 * config/epiphany/rtems.h: New file.
24241 2018-01-04 Jakub Jelinek <jakub@redhat.com>
24242 Uros Bizjak <ubizjak@gmail.com>
24245 * config/i386/i386.md (*<rotate_insn>hi3_1 splitter): Use
24246 QIreg_operand instead of register_operand predicate.
24247 * config/i386/i386.c (ix86_rop_should_change_byte_p,
24248 set_rop_modrm_reg_bits, ix86_mitigate_rop): Use -mmitigate-rop in
24249 comments instead of -fmitigate[-_]rop.
24251 2018-01-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
24254 * cgraphunit.c (symbol_table::compile): Switch to text_section
24255 before calling assembly_start debug hook.
24256 * run-rtl-passes.c (run_rtl_passes): Likewise.
24259 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
24261 * tree-vrp.c (extract_range_from_binary_expr_1): Check
24262 range_int_cst_p rather than !symbolic_range_p before calling
24263 extract_range_from_multiplicative_op_1.
24265 2018-01-04 Jeff Law <law@redhat.com>
24267 * tree-ssa-math-opts.c (execute_cse_reciprocals_1): Remove
24268 redundant test in assertion.
24270 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
24272 * doc/rtl.texi: Document machine_mode wrapper classes.
24274 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
24276 * fold-const.c (fold_ternary_loc): Check tree_fits_uhwi_p before
24277 using tree_to_uhwi.
24279 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
24281 * tree-ssa-forwprop.c (is_combined_permutation_identity): Allow
24282 the VEC_PERM_EXPR fold to fail.
24284 2018-01-04 Jakub Jelinek <jakub@redhat.com>
24287 * bb-reorder.c (insert_section_boundary_note): Set has_bb_partition
24288 to switched_sections.
24290 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
24293 * config/arm/arm.c (arm_vectorize_vec_perm_const): Fix inverted
24294 test for d.testing.
24296 2018-01-04 Peter Bergner <bergner@vnet.ibm.com>
24299 * config/rs6000/rs6000.c (rs6000_discover_homogeneous_aggregate): Do not
24300 allow arguments in FP registers if TARGET_HARD_FLOAT is false.
24302 2018-01-04 Jakub Jelinek <jakub@redhat.com>
24305 * cfgexpand.c (expand_debug_expr) <case BIT_FIELD_REF>: Punt if mode
24306 is BLKmode and bitpos not zero or mode change is needed.
24308 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
24311 * config/sparc/sparc.c (sparc_vectorize_vec_perm_const): Require
24314 2018-01-04 Uros Bizjak <ubizjak@gmail.com>
24317 * config/alpha/alpha.md (*sadd<modesuffix>): Use ASHIFT
24318 instead of MULT rtx. Update all corresponding splitters.
24319 (*saddl_se): Ditto.
24320 (*ssub<modesuffix>): Ditto.
24321 (*ssubl_se): Ditto.
24322 (*cmp_sadd_di): Update split patterns.
24323 (*cmp_sadd_si): Ditto.
24324 (*cmp_sadd_sidi): Ditto.
24325 (*cmp_ssub_di): Ditto.
24326 (*cmp_ssub_si): Ditto.
24327 (*cmp_ssub_sidi): Ditto.
24328 * config/alpha/predicates.md (const23_operand): New predicate.
24329 * config/alpha/alpha.c (alpha_rtx_costs) [PLUS, MINUS]:
24330 Look for ASHIFT, not MULT inner operand.
24331 (alpha_split_conditional_move): Update for *sadd<modesuffix> change.
24333 2018-01-04 Martin Liska <mliska@suse.cz>
24335 PR gcov-profile/83669
24336 * gcov.c (output_intermediate_file): Add version to intermediate
24338 * doc/gcov.texi: Document new field 'version' in intermediate
24339 file format. Fix location of '-k' option of gcov command.
24341 2018-01-04 Martin Liska <mliska@suse.cz>
24344 * ipa-icf.c (sem_function::merge): Do not cross comdat boundary.
24346 2018-01-04 Jakub Jelinek <jakub@redhat.com>
24348 * gimple-ssa-sprintf.c (parse_directive): Cast second dir.len to uhwi.
24350 2018-01-03 Martin Sebor <msebor@redhat.com>
24352 PR tree-optimization/83655
24353 * gimple-ssa-warn-restrict.c (wrestrict_dom_walker::check_call): Avoid
24354 checking calls with invalid arguments.
24356 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24358 * tree-vect-stmts.c (vect_get_store_rhs): New function.
24359 (vectorizable_mask_load_store): Delete.
24360 (vectorizable_call): Return false for masked loads and stores.
24361 (vectorizable_store): Handle IFN_MASK_STORE. Use vect_get_store_rhs
24362 instead of gimple_assign_rhs1.
24363 (vectorizable_load): Handle IFN_MASK_LOAD.
24364 (vect_transform_stmt): Don't set is_store for call_vec_info_type.
24366 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24368 * tree-vect-stmts.c (vect_build_gather_load_calls): New function,
24370 (vectorizable_mask_load_store): ...here.
24371 (vectorizable_load): ...and here.
24373 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24375 * tree-vect-stmts.c (vect_build_all_ones_mask)
24376 (vect_build_zero_merge_argument): New functions, split out from...
24377 (vectorizable_load): ...here.
24379 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24381 * tree-vect-stmts.c (vect_check_store_rhs): New function,
24383 (vectorizable_mask_load_store): ...here.
24384 (vectorizable_store): ...and here.
24386 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24388 * tree-vect-stmts.c (vect_check_load_store_mask): New function,
24390 (vectorizable_mask_load_store): ...here.
24392 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24394 * tree-vectorizer.h (vec_load_store_type): Moved from tree-vec-stmts.c
24395 (vect_model_store_cost): Take a vec_load_store_type instead of a
24397 * tree-vect-stmts.c (vec_load_store_type): Move to tree-vectorizer.h.
24398 (vect_model_store_cost): Take a vec_load_store_type instead of a
24400 (vectorizable_mask_load_store): Update accordingly.
24401 (vectorizable_store): Likewise.
24402 * tree-vect-slp.c (vect_analyze_slp_cost_1): Update accordingly.
24404 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24406 * tree-vect-loop.c (vect_transform_loop): Stub out scalar
24407 IFN_MASK_LOAD calls here rather than...
24408 * tree-vect-stmts.c (vectorizable_mask_load_store): ...here.
24410 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24411 Alan Hayward <alan.hayward@arm.com>
24412 David Sherwood <david.sherwood@arm.com>
24414 * expmed.c (extract_bit_field_1): For vector extracts,
24415 fall back to extract_bit_field_as_subreg if vec_extract
24418 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24419 Alan Hayward <alan.hayward@arm.com>
24420 David Sherwood <david.sherwood@arm.com>
24422 * lra-spills.c (pseudo_reg_slot_compare): Sort slots by whether
24423 they are variable or constant sized.
24424 (assign_stack_slot_num_and_sort_pseudos): Don't reuse variable-sized
24425 slots for constant-sized data.
24427 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24428 Alan Hayward <alan.hayward@arm.com>
24429 David Sherwood <david.sherwood@arm.com>
24431 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern): When
24432 handling COND_EXPRs with boolean comparisons, try to find a better
24433 basis for the mask type than the boolean itself.
24435 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24437 * doc/rtl.texi (MAX_BITSIZE_MODE_ANY_MODE): Describe how the default
24438 is calculated and how it can be overridden.
24439 * genmodes.c (max_bitsize_mode_any_mode): New variable.
24440 (create_modes): Initialize it from MAX_BITSIZE_MODE_ANY_MODE,
24442 (emit_max_int): Use it to set the output MAX_BITSIZE_MODE_ANY_MODE,
24445 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24446 Alan Hayward <alan.hayward@arm.com>
24447 David Sherwood <david.sherwood@arm.com>
24449 * config/aarch64/aarch64-protos.h (aarch64_output_simd_mov_immediate):
24450 Remove the mode argument.
24451 (aarch64_simd_valid_immediate): Remove the mode and inverse
24453 * config/aarch64/iterators.md (bitsize): New iterator.
24454 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>, and<mode>3)
24455 (ior<mode>3): Update calls to aarch64_output_simd_mov_immediate.
24456 * config/aarch64/constraints.md (Do, Db, Dn): Update calls to
24457 aarch64_simd_valid_immediate.
24458 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Likewise.
24459 (aarch64_reg_or_bic_imm): Likewise.
24460 * config/aarch64/aarch64.c (simd_immediate_info): Replace mvn
24461 with an insn_type enum and msl with a modifier_type enum.
24462 Replace element_width with a scalar_mode. Change the shift
24463 to unsigned int. Add constructors for scalar_float_mode and
24464 scalar_int_mode elements.
24465 (aarch64_vect_float_const_representable_p): Delete.
24466 (aarch64_can_const_movi_rtx_p)
24467 (aarch64_simd_scalar_immediate_valid_for_move)
24468 (aarch64_simd_make_constant): Update call to
24469 aarch64_simd_valid_immediate.
24470 (aarch64_advsimd_valid_immediate_hs): New function.
24471 (aarch64_advsimd_valid_immediate): Likewise.
24472 (aarch64_simd_valid_immediate): Remove mode and inverse
24473 arguments. Rewrite to use the above. Use const_vec_duplicate_p
24474 to detect duplicated constants and use aarch64_float_const_zero_rtx_p
24475 and aarch64_float_const_representable_p on the result.
24476 (aarch64_output_simd_mov_immediate): Remove mode argument.
24477 Update call to aarch64_simd_valid_immediate and use of
24478 simd_immediate_info.
24479 (aarch64_output_scalar_simd_mov_immediate): Update call
24482 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24483 Alan Hayward <alan.hayward@arm.com>
24484 David Sherwood <david.sherwood@arm.com>
24486 * machmode.h (mode_precision): Prefix with CONST_MODE_PRECISION.
24487 (mode_nunits): Likewise CONST_MODE_NUNITS.
24488 * machmode.def (ADJUST_NUNITS): Document.
24489 * genmodes.c (mode_data::need_nunits_adj): New field.
24490 (blank_mode): Update accordingly.
24491 (adj_nunits): New variable.
24492 (print_maybe_const_decl): Replace CATEGORY with a NEEDS_ADJ
24494 (emit_mode_size_inline): Set need_bytesize_adj for all modes
24495 listed in adj_nunits.
24496 (emit_mode_nunits_inline): Set need_nunits_adj for all modes
24497 listed in adj_nunits. Don't emit case statements for such modes.
24498 (emit_insn_modes_h): Emit definitions of CONST_MODE_NUNITS
24499 and CONST_MODE_PRECISION. Make CONST_MODE_SIZE expand to
24500 nothing if adj_nunits is nonnull.
24501 (emit_mode_precision, emit_mode_nunits): Use print_maybe_const_decl.
24502 (emit_mode_unit_size, emit_mode_base_align, emit_mode_ibit)
24503 (emit_mode_fbit): Update use of print_maybe_const_decl.
24504 (emit_move_size): Likewise. Treat the array as non-const
24506 (emit_mode_adjustments): Handle adj_nunits.
24508 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24510 * machmode.def (VECTOR_MODES_WITH_PREFIX): Document.
24511 * genmodes.c (VECTOR_MODES_WITH_PREFIX): New macro.
24512 (VECTOR_MODES): Use it.
24513 (make_vector_modes): Take the prefix as an argument.
24515 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24516 Alan Hayward <alan.hayward@arm.com>
24517 David Sherwood <david.sherwood@arm.com>
24519 * mode-classes.def (MODE_VECTOR_BOOL): New mode class.
24520 * machmode.h (INTEGRAL_MODE_P, VECTOR_MODE_P): Return true
24521 for MODE_VECTOR_BOOL.
24522 * machmode.def (VECTOR_BOOL_MODE): Document.
24523 * genmodes.c (VECTOR_BOOL_MODE): New macro.
24524 (make_vector_bool_mode): New function.
24525 (complete_mode, emit_mode_wider, emit_mode_adjustments): Handle
24527 * lto-streamer-in.c (lto_input_mode_table): Likewise.
24528 * rtx-vector-builder.c (rtx_vector_builder::find_cached_value):
24530 * stor-layout.c (int_mode_for_mode): Likewise.
24531 * tree.c (build_vector_type_for_mode): Likewise.
24532 * varasm.c (output_constant_pool_2): Likewise.
24533 * emit-rtl.c (init_emit_once): Make sure that CONST1_RTX (BImode) and
24534 CONSTM1_RTX (BImode) are the same thing. Initialize const_tiny_rtx
24535 for MODE_VECTOR_BOOL.
24536 * expr.c (expand_expr_real_1): Use VECTOR_MODE_P instead of a list
24537 of mode class checks.
24538 * tree-vect-generic.c (expand_vector_operation): Use VECTOR_MODE_P
24539 instead of a list of mode class checks.
24540 (expand_vector_scalar_condition): Likewise.
24541 (type_for_widest_vector_mode): Handle BImode as an inner mode.
24543 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24544 Alan Hayward <alan.hayward@arm.com>
24545 David Sherwood <david.sherwood@arm.com>
24547 * machmode.h (mode_size): Change from unsigned short to
24549 (mode_to_bytes): Return a poly_uint16 rather than an unsigned short.
24550 (GET_MODE_SIZE): Return a constant if ONLY_FIXED_SIZE_MODES,
24551 or if measurement_type is not polynomial.
24552 (fixed_size_mode::includes_p): Check for constant-sized modes.
24553 * genmodes.c (emit_mode_size_inline): Make mode_size_inline
24554 return a poly_uint16 rather than an unsigned short.
24555 (emit_mode_size): Change the type of mode_size from unsigned short
24556 to poly_uint16_pod. Use ZERO_COEFFS for the initializer.
24557 (emit_mode_adjustments): Cope with polynomial vector sizes.
24558 * lto-streamer-in.c (lto_input_mode_table): Use bp_unpack_poly_value
24560 * lto-streamer-out.c (lto_write_mode_table): Use bp_pack_poly_value
24562 * auto-inc-dec.c (try_merge): Treat GET_MODE_SIZE as polynomial.
24563 * builtins.c (expand_ifn_atomic_compare_exchange_into_call): Likewise.
24564 * caller-save.c (setup_save_areas): Likewise.
24565 (replace_reg_with_saved_mem): Likewise.
24566 * calls.c (emit_library_call_value_1): Likewise.
24567 * combine-stack-adj.c (combine_stack_adjustments_for_block): Likewise.
24568 * combine.c (simplify_set, make_extraction, simplify_shift_const_1)
24569 (gen_lowpart_for_combine): Likewise.
24570 * convert.c (convert_to_integer_1): Likewise.
24571 * cse.c (equiv_constant, cse_insn): Likewise.
24572 * cselib.c (autoinc_split, cselib_hash_rtx): Likewise.
24573 (cselib_subst_to_values): Likewise.
24574 * dce.c (word_dce_process_block): Likewise.
24575 * df-problems.c (df_word_lr_mark_ref): Likewise.
24576 * dwarf2cfi.c (init_one_dwarf_reg_size): Likewise.
24577 * dwarf2out.c (multiple_reg_loc_descriptor, mem_loc_descriptor)
24578 (concat_loc_descriptor, concatn_loc_descriptor, loc_descriptor)
24579 (rtl_for_decl_location): Likewise.
24580 * emit-rtl.c (gen_highpart, widen_memory_access): Likewise.
24581 * expmed.c (extract_bit_field_1, extract_integral_bit_field): Likewise.
24582 * expr.c (emit_group_load_1, clear_storage_hints): Likewise.
24583 (emit_move_complex, emit_move_multi_word, emit_push_insn): Likewise.
24584 (expand_expr_real_1): Likewise.
24585 * function.c (assign_parm_setup_block_p, assign_parm_setup_block)
24586 (pad_below): Likewise.
24587 * gimple-fold.c (optimize_atomic_compare_exchange_p): Likewise.
24588 * gimple-ssa-store-merging.c (rhs_valid_for_store_merging_p): Likewise.
24589 * ira.c (get_subreg_tracking_sizes): Likewise.
24590 * ira-build.c (ira_create_allocno_objects): Likewise.
24591 * ira-color.c (coalesced_pseudo_reg_slot_compare): Likewise.
24592 (ira_sort_regnos_for_alter_reg): Likewise.
24593 * ira-costs.c (record_operand_costs): Likewise.
24594 * lower-subreg.c (interesting_mode_p, simplify_gen_subreg_concatn)
24595 (resolve_simple_move): Likewise.
24596 * lra-constraints.c (get_reload_reg, operands_match_p): Likewise.
24597 (process_addr_reg, simplify_operand_subreg, curr_insn_transform)
24598 (lra_constraints): Likewise.
24599 (CONST_POOL_OK_P): Reject variable-sized modes.
24600 * lra-spills.c (slot, assign_mem_slot, pseudo_reg_slot_compare)
24601 (add_pseudo_to_slot, lra_spill): Likewise.
24602 * omp-low.c (omp_clause_aligned_alignment): Likewise.
24603 * optabs-query.c (get_best_extraction_insn): Likewise.
24604 * optabs-tree.c (expand_vec_cond_expr_p): Likewise.
24605 * optabs.c (expand_vec_perm_var, expand_vec_cond_expr): Likewise.
24606 (expand_mult_highpart, valid_multiword_target_p): Likewise.
24607 * recog.c (offsettable_address_addr_space_p): Likewise.
24608 * regcprop.c (maybe_mode_change): Likewise.
24609 * reginfo.c (choose_hard_reg_mode, record_subregs_of_mode): Likewise.
24610 * regrename.c (build_def_use): Likewise.
24611 * regstat.c (dump_reg_info): Likewise.
24612 * reload.c (complex_word_subreg_p, push_reload, find_dummy_reload)
24613 (find_reloads, find_reloads_subreg_address): Likewise.
24614 * reload1.c (eliminate_regs_1): Likewise.
24615 * rtlanal.c (for_each_inc_dec_find_inc_dec, rtx_cost): Likewise.
24616 * simplify-rtx.c (avoid_constant_pool_reference): Likewise.
24617 (simplify_binary_operation_1, simplify_subreg): Likewise.
24618 * targhooks.c (default_function_arg_padding): Likewise.
24619 (default_hard_regno_nregs, default_class_max_nregs): Likewise.
24620 * tree-cfg.c (verify_gimple_assign_binary): Likewise.
24621 (verify_gimple_assign_ternary): Likewise.
24622 * tree-inline.c (estimate_move_cost): Likewise.
24623 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
24624 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Likewise.
24625 (get_address_cost_ainc): Likewise.
24626 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
24627 (vect_supportable_dr_alignment): Likewise.
24628 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
24629 (vectorizable_reduction): Likewise.
24630 * tree-vect-stmts.c (vectorizable_assignment, vectorizable_shift)
24631 (vectorizable_operation, vectorizable_load): Likewise.
24632 * tree.c (build_same_sized_truth_vector_type): Likewise.
24633 * valtrack.c (cleanup_auto_inc_dec): Likewise.
24634 * var-tracking.c (emit_note_insn_var_location): Likewise.
24635 * config/arc/arc.h (ASM_OUTPUT_CASE_END): Use as_a <scalar_int_mode>.
24636 (ADDR_VEC_ALIGN): Likewise.
24638 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24639 Alan Hayward <alan.hayward@arm.com>
24640 David Sherwood <david.sherwood@arm.com>
24642 * machmode.h (mode_to_bits): Return a poly_uint16 rather than an
24644 (GET_MODE_BITSIZE): Return a constant if ONLY_FIXED_SIZE_MODES,
24645 or if measurement_type is polynomial.
24646 * calls.c (shift_return_value): Treat GET_MODE_BITSIZE as polynomial.
24647 * combine.c (make_extraction): Likewise.
24648 * dse.c (find_shift_sequence): Likewise.
24649 * dwarf2out.c (mem_loc_descriptor): Likewise.
24650 * expmed.c (store_integral_bit_field, extract_bit_field_1): Likewise.
24651 (extract_bit_field, extract_low_bits): Likewise.
24652 * expr.c (convert_move, convert_modes, emit_move_insn_1): Likewise.
24653 (optimize_bitfield_assignment_op, expand_assignment): Likewise.
24654 (store_expr_with_bounds, store_field, expand_expr_real_1): Likewise.
24655 * fold-const.c (optimize_bit_field_compare, merge_ranges): Likewise.
24656 * gimple-fold.c (optimize_atomic_compare_exchange_p): Likewise.
24657 * reload.c (find_reloads): Likewise.
24658 * reload1.c (alter_reg): Likewise.
24659 * stor-layout.c (bitwise_mode_for_mode, compute_record_mode): Likewise.
24660 * targhooks.c (default_secondary_memory_needed_mode): Likewise.
24661 * tree-if-conv.c (predicate_mem_writes): Likewise.
24662 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
24663 * tree-vect-patterns.c (adjust_bool_pattern): Likewise.
24664 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
24665 * valtrack.c (dead_debug_insert_temp): Likewise.
24666 * varasm.c (mergeable_constant_section): Likewise.
24667 * config/sh/sh.h (LOCAL_ALIGNMENT): Use as_a <fixed_size_mode>.
24669 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24670 Alan Hayward <alan.hayward@arm.com>
24671 David Sherwood <david.sherwood@arm.com>
24673 * expr.c (expand_assignment): Cope with polynomial mode sizes
24674 when assigning to a CONCAT.
24676 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24677 Alan Hayward <alan.hayward@arm.com>
24678 David Sherwood <david.sherwood@arm.com>
24680 * machmode.h (mode_precision): Change from unsigned short to
24682 (mode_to_precision): Return a poly_uint16 rather than an unsigned
24684 (GET_MODE_PRECISION): Return a constant if ONLY_FIXED_SIZE_MODES,
24685 or if measurement_type is not polynomial.
24686 (HWI_COMPUTABLE_MODE_P): Turn into a function. Optimize the case
24687 in which the mode is already known to be a scalar_int_mode.
24688 * genmodes.c (emit_mode_precision): Change the type of mode_precision
24689 from unsigned short to poly_uint16_pod. Use ZERO_COEFFS for the
24691 * lto-streamer-in.c (lto_input_mode_table): Use bp_unpack_poly_value
24692 for GET_MODE_PRECISION.
24693 * lto-streamer-out.c (lto_write_mode_table): Use bp_pack_poly_value
24694 for GET_MODE_PRECISION.
24695 * combine.c (update_rsp_from_reg_equal): Treat GET_MODE_PRECISION
24697 (try_combine, find_split_point, combine_simplify_rtx): Likewise.
24698 (expand_field_assignment, make_extraction): Likewise.
24699 (make_compound_operation_int, record_dead_and_set_regs_1): Likewise.
24700 (get_last_value): Likewise.
24701 * convert.c (convert_to_integer_1): Likewise.
24702 * cse.c (cse_insn): Likewise.
24703 * expr.c (expand_expr_real_1): Likewise.
24704 * lra-constraints.c (simplify_operand_subreg): Likewise.
24705 * optabs-query.c (can_atomic_load_p): Likewise.
24706 * optabs.c (expand_atomic_load): Likewise.
24707 (expand_atomic_store): Likewise.
24708 * ree.c (combine_reaching_defs): Likewise.
24709 * rtl.h (partial_subreg_p, paradoxical_subreg_p): Likewise.
24710 * rtlanal.c (nonzero_bits1, lsb_bitfield_op_p): Likewise.
24711 * tree.h (type_has_mode_precision_p): Likewise.
24712 * ubsan.c (instrument_si_overflow): Likewise.
24714 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24715 Alan Hayward <alan.hayward@arm.com>
24716 David Sherwood <david.sherwood@arm.com>
24718 * tree.h (TYPE_VECTOR_SUBPARTS): Turn into a function and handle
24719 polynomial numbers of units.
24720 (SET_TYPE_VECTOR_SUBPARTS): Likewise.
24721 (valid_vector_subparts_p): New function.
24722 (build_vector_type): Remove temporary shim and take the number
24723 of units as a poly_uint64 rather than an int.
24724 (build_opaque_vector_type): Take the number of units as a
24725 poly_uint64 rather than an int.
24726 * tree.c (build_vector_from_ctor): Handle polynomial
24727 TYPE_VECTOR_SUBPARTS.
24728 (type_hash_canon_hash, type_cache_hasher::equal): Likewise.
24729 (uniform_vector_p, vector_type_mode, build_vector): Likewise.
24730 (build_vector_from_val): If the number of units is variable,
24731 use build_vec_duplicate_cst for constant operands and
24732 VEC_DUPLICATE_EXPR otherwise.
24733 (make_vector_type): Remove temporary is_constant ().
24734 (build_vector_type, build_opaque_vector_type): Take the number of
24735 units as a poly_uint64 rather than an int.
24736 (check_vector_cst): Handle polynomial TYPE_VECTOR_SUBPARTS and
24738 * cfgexpand.c (expand_debug_expr): Likewise.
24739 * expr.c (count_type_elements, categorize_ctor_elements_1): Likewise.
24740 (store_constructor, expand_expr_real_1): Likewise.
24741 (const_scalar_mask_from_tree): Likewise.
24742 * fold-const-call.c (fold_const_reduction): Likewise.
24743 * fold-const.c (const_binop, const_unop, fold_convert_const): Likewise.
24744 (operand_equal_p, fold_vec_perm, fold_ternary_loc): Likewise.
24745 (native_encode_vector, vec_cst_ctor_to_array): Likewise.
24746 (fold_relational_const): Likewise.
24747 (native_interpret_vector): Likewise. Change the size from an
24748 int to an unsigned int.
24749 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Handle polynomial
24750 TYPE_VECTOR_SUBPARTS.
24751 (gimple_fold_indirect_ref, gimple_build_vector): Likewise.
24752 (gimple_build_vector_from_val): Use VEC_DUPLICATE_EXPR when
24753 duplicating a non-constant operand into a variable-length vector.
24754 * hsa-brig.c (hsa_op_immed::emit_to_buffer): Handle polynomial
24755 TYPE_VECTOR_SUBPARTS and VECTOR_CST_NELTS.
24756 * ipa-icf.c (sem_variable::equals): Likewise.
24757 * match.pd: Likewise.
24758 * omp-simd-clone.c (simd_clone_subparts): Likewise.
24759 * print-tree.c (print_node): Likewise.
24760 * stor-layout.c (layout_type): Likewise.
24761 * targhooks.c (default_builtin_vectorization_cost): Likewise.
24762 * tree-cfg.c (verify_gimple_comparison): Likewise.
24763 (verify_gimple_assign_binary): Likewise.
24764 (verify_gimple_assign_ternary): Likewise.
24765 (verify_gimple_assign_single): Likewise.
24766 * tree-pretty-print.c (dump_generic_node): Likewise.
24767 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
24768 (simplify_bitfield_ref, is_combined_permutation_identity): Likewise.
24769 * tree-vect-data-refs.c (vect_permute_store_chain): Likewise.
24770 (vect_grouped_load_supported, vect_permute_load_chain): Likewise.
24771 (vect_shift_permute_load_chain): Likewise.
24772 * tree-vect-generic.c (nunits_for_known_piecewise_op): Likewise.
24773 (expand_vector_condition, optimize_vector_constructor): Likewise.
24774 (lower_vec_perm, get_compute_type): Likewise.
24775 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
24776 (get_initial_defs_for_reduction, vect_transform_loop): Likewise.
24777 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
24778 (vect_recog_mask_conversion_pattern): Likewise.
24779 * tree-vect-slp.c (vect_supported_load_permutation_p): Likewise.
24780 (vect_get_constant_vectors, vect_transform_slp_perm_load): Likewise.
24781 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
24782 (get_group_load_store_type, vectorizable_mask_load_store): Likewise.
24783 (vectorizable_bswap, simd_clone_subparts, vectorizable_assignment)
24784 (vectorizable_shift, vectorizable_operation, vectorizable_store)
24785 (vectorizable_load, vect_is_simple_cond, vectorizable_comparison)
24786 (supportable_widening_operation): Likewise.
24787 (supportable_narrowing_operation): Likewise.
24788 * tree-vector-builder.c (tree_vector_builder::binary_encoded_nelts):
24790 * varasm.c (output_constant): Likewise.
24792 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24793 Alan Hayward <alan.hayward@arm.com>
24794 David Sherwood <david.sherwood@arm.com>
24796 * tree-vect-data-refs.c (vect_permute_store_chain): Reorganize
24797 so that both the length == 3 and length != 3 cases set up their
24798 own permute vectors. Add comments explaining why we know the
24799 number of elements is constant.
24800 (vect_permute_load_chain): Likewise.
24802 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24803 Alan Hayward <alan.hayward@arm.com>
24804 David Sherwood <david.sherwood@arm.com>
24806 * machmode.h (mode_nunits): Change from unsigned char to
24808 (ONLY_FIXED_SIZE_MODES): New macro.
24809 (pod_mode::measurement_type, scalar_int_mode::measurement_type)
24810 (scalar_float_mode::measurement_type, scalar_mode::measurement_type)
24811 (complex_mode::measurement_type, fixed_size_mode::measurement_type):
24813 (mode_to_nunits): Return a poly_uint16 rather than an unsigned short.
24814 (GET_MODE_NUNITS): Return a constant if ONLY_FIXED_SIZE_MODES,
24815 or if measurement_type is not polynomial.
24816 * genmodes.c (ZERO_COEFFS): New macro.
24817 (emit_mode_nunits_inline): Make mode_nunits_inline return a
24819 (emit_mode_nunits): Change the type of mode_nunits to poly_uint16_pod.
24820 Use ZERO_COEFFS when emitting initializers.
24821 * data-streamer.h (bp_pack_poly_value): New function.
24822 (bp_unpack_poly_value): Likewise.
24823 * lto-streamer-in.c (lto_input_mode_table): Use bp_unpack_poly_value
24824 for GET_MODE_NUNITS.
24825 * lto-streamer-out.c (lto_write_mode_table): Use bp_pack_poly_value
24826 for GET_MODE_NUNITS.
24827 * tree.c (make_vector_type): Remove temporary shim and make
24828 the real function take the number of units as a poly_uint64
24829 rather than an int.
24830 (build_vector_type_for_mode): Handle polynomial nunits.
24831 * dwarf2out.c (loc_descriptor, add_const_value_attribute): Likewise.
24832 * emit-rtl.c (const_vec_series_p_1): Likewise.
24833 (gen_rtx_CONST_VECTOR): Likewise.
24834 * fold-const.c (test_vec_duplicate_folding): Likewise.
24835 * genrecog.c (validate_pattern): Likewise.
24836 * optabs-query.c (can_vec_perm_var_p, can_mult_highpart_p): Likewise.
24837 * optabs-tree.c (expand_vec_cond_expr_p): Likewise.
24838 * optabs.c (expand_vector_broadcast, expand_binop_directly): Likewise.
24839 (shift_amt_for_vec_perm_mask, expand_vec_perm_var): Likewise.
24840 (expand_vec_cond_expr, expand_mult_highpart): Likewise.
24841 * rtlanal.c (subreg_get_info): Likewise.
24842 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
24843 (vect_grouped_load_supported): Likewise.
24844 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
24845 * tree-vect-loop.c (have_whole_vector_shift): Likewise.
24846 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
24847 (simplify_const_unary_operation, simplify_binary_operation_1)
24848 (simplify_const_binary_operation, simplify_ternary_operation)
24849 (test_vector_ops_duplicate, test_vector_ops): Likewise.
24850 (simplify_immed_subreg): Use GET_MODE_NUNITS on a fixed_size_mode
24851 instead of CONST_VECTOR_NUNITS.
24852 * varasm.c (output_constant_pool_2): Likewise.
24853 * rtx-vector-builder.c (rtx_vector_builder::build): Only include the
24854 explicit-encoded elements in the XVEC for variable-length vectors.
24856 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24858 * lra-constraints.c (curr_insn_transform): Use partial_subreg_p.
24860 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24861 Alan Hayward <alan.hayward@arm.com>
24862 David Sherwood <david.sherwood@arm.com>
24864 * coretypes.h (fixed_size_mode): Declare.
24865 (fixed_size_mode_pod): New typedef.
24866 * builtins.h (target_builtins::x_apply_args_mode)
24867 (target_builtins::x_apply_result_mode): Change type to
24868 fixed_size_mode_pod.
24869 * builtins.c (apply_args_size, apply_result_size, result_vector)
24870 (expand_builtin_apply_args_1, expand_builtin_apply)
24871 (expand_builtin_return): Update accordingly.
24873 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24875 * cse.c (hash_rtx_cb): Hash only the encoded elements.
24876 * cselib.c (cselib_hash_rtx): Likewise.
24877 * expmed.c (make_tree): Build VECTOR_CSTs directly from the
24878 CONST_VECTOR encoding.
24880 2018-01-03 Jakub Jelinek <jakub@redhat.com>
24881 Jeff Law <law@redhat.com>
24884 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): For
24885 noreturn probe, use gen_pop instead of ix86_emit_restore_reg_using_pop,
24886 only set RTX_FRAME_RELATED_P on both the push and pop if cfa_reg is sp
24887 and add REG_CFA_ADJUST_CFA notes in that case to both insns.
24890 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Do not
24891 explicitly probe *sp in a noreturn function if there were any callee
24892 register saves or frame pointer is needed.
24894 2018-01-03 Jakub Jelinek <jakub@redhat.com>
24897 * cfgexpand.c (expand_debug_expr): Return NULL if mode is
24898 BLKmode for ternary, binary or unary expressions.
24901 * var-tracking.c (delete_vta_debug_insn): New inline function.
24902 (delete_vta_debug_insns): Add USE_CFG argument, if true, walk just
24903 insns from get_insns () to NULL instead of each bb separately.
24904 Use delete_vta_debug_insn. No longer static.
24905 (vt_debug_insns_local, variable_tracking_main_1): Adjust
24906 delete_vta_debug_insns callers.
24907 * rtl.h (delete_vta_debug_insns): Declare.
24908 * final.c (rest_of_handle_final): Call delete_vta_debug_insns
24909 instead of variable_tracking_main.
24911 2018-01-03 Martin Sebor <msebor@redhat.com>
24913 PR tree-optimization/83603
24914 * calls.c (maybe_warn_nonstring_arg): Avoid accessing function
24915 arguments past the endof the argument list in functions declared
24916 without a prototype.
24917 * gimple-ssa-warn-restrict.c (wrestrict_dom_walker::check_call):
24918 Avoid checking when arguments are null.
24920 2018-01-03 Martin Sebor <msebor@redhat.com>
24923 * doc/extend.texi (attribute const): Fix a typo.
24924 * ipa-pure-const.c ((warn_function_const, warn_function_pure): Avoid
24925 issuing -Wsuggest-attribute for void functions.
24927 2018-01-03 Martin Sebor <msebor@redhat.com>
24929 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Use
24930 offset_int::from instead of wide_int::to_shwi.
24931 (maybe_diag_overlap): Remove assertion.
24932 Use HOST_WIDE_INT_PRINT_DEC instead of %lli.
24933 * gimple-ssa-sprintf.c (format_directive): Same.
24934 (parse_directive): Same.
24935 (sprintf_dom_walker::compute_format_length): Same.
24936 (try_substitute_return_value): Same.
24938 2018-01-03 Jeff Law <law@redhat.com>
24940 PR middle-end/83654
24941 * explow.c (anti_adjust_stack_and_probe_stack_clash): Test a
24942 non-constant residual for zero at runtime and avoid probing in
24943 that case. Reorganize code for trailing problem to mirror handling
24946 2018-01-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
24948 PR tree-optimization/83501
24949 * tree-ssa-strlen.c (get_string_cst): New.
24950 (handle_char_store): Call get_string_cst.
24952 2018-01-03 Martin Liska <mliska@suse.cz>
24954 PR tree-optimization/83593
24955 * tree-ssa-strlen.c: Include tree-cfg.h.
24956 (strlen_check_and_optimize_stmt): Add new argument cleanup_eh.
24957 (strlen_dom_walker): Add new member variable m_cleanup_cfg.
24958 (strlen_dom_walker::strlen_dom_walker): Initialize m_cleanup_cfg
24960 (strlen_dom_walker::before_dom_children): Call
24961 gimple_purge_dead_eh_edges. Dump tranformation with details
24963 (strlen_dom_walker::before_dom_children): Update call by adding
24964 new argument cleanup_eh.
24965 (pass_strlen::execute): Return TODO_cleanup_cfg if needed.
24967 2018-01-03 Martin Liska <mliska@suse.cz>
24970 * cif-code.def (VARIADIC_THUNK): New enum value.
24971 * ipa-fnsummary.c (compute_fn_summary): Do not inline variadic
24974 2018-01-03 Jan Beulich <jbeulich@suse.com>
24976 * sse.md (mov<mode>_internal): Tighten condition for when to use
24977 vmovdqu<ssescalarsize> for TI and OI modes.
24979 2018-01-03 Jakub Jelinek <jakub@redhat.com>
24981 Update copyright years.
24983 2018-01-03 Martin Liska <mliska@suse.cz>
24986 * ipa-visibility.c (function_and_variable_visibility): Skip
24987 functions with noipa attribure.
24989 2018-01-03 Jakub Jelinek <jakub@redhat.com>
24991 * gcc.c (process_command): Update copyright notice dates.
24992 * gcov-dump.c (print_version): Ditto.
24993 * gcov.c (print_version): Ditto.
24994 * gcov-tool.c (print_version): Ditto.
24995 * gengtype.c (create_file): Ditto.
24996 * doc/cpp.texi: Bump @copying's copyright year.
24997 * doc/cppinternals.texi: Ditto.
24998 * doc/gcc.texi: Ditto.
24999 * doc/gccint.texi: Ditto.
25000 * doc/gcov.texi: Ditto.
25001 * doc/install.texi: Ditto.
25002 * doc/invoke.texi: Ditto.
25004 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25006 * vector-builder.h (vector_builder::m_full_nelts): Change from
25007 unsigned int to poly_uint64.
25008 (vector_builder::full_nelts): Update prototype accordingly.
25009 (vector_builder::new_vector): Likewise.
25010 (vector_builder::encoded_full_vector_p): Handle polynomial full_nelts.
25011 (vector_builder::operator ==): Likewise.
25012 (vector_builder::finalize): Likewise.
25013 * int-vector-builder.h (int_vector_builder::int_vector_builder):
25014 Take the number of elements as a poly_uint64 rather than an
25016 * vec-perm-indices.h (vec_perm_indices::m_nelts_per_input): Change
25017 from unsigned int to poly_uint64.
25018 (vec_perm_indices::vec_perm_indices): Update prototype accordingly.
25019 (vec_perm_indices::new_vector): Likewise.
25020 (vec_perm_indices::length): Likewise.
25021 (vec_perm_indices::nelts_per_input): Likewise.
25022 (vec_perm_indices::input_nelts): Likewise.
25023 * vec-perm-indices.c (vec_perm_indices::new_vector): Take the
25024 number of elements per input as a poly_uint64 rather than an
25025 unsigned int. Use the original encoding for variable-length
25026 vectors, rather than clamping each individual element.
25027 For the second and subsequent elements in each pattern,
25028 clamp the step and base before clamping their sum.
25029 (vec_perm_indices::series_p): Handle polynomial element counts.
25030 (vec_perm_indices::all_in_range_p): Likewise.
25031 (vec_perm_indices_to_tree): Likewise.
25032 (vec_perm_indices_to_rtx): Likewise.
25033 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
25034 * tree-vector-builder.c (tree_vector_builder::new_unary_operation)
25035 (tree_vector_builder::new_binary_operation): Handle polynomial
25036 element counts. Return false if we need to know the number
25037 of elements at compile time.
25038 * fold-const.c (fold_vec_perm): Punt if the number of elements
25039 isn't known at compile time.
25041 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25043 * vec-perm-indices.h (vec_perm_builder): Change element type
25044 from HOST_WIDE_INT to poly_int64.
25045 (vec_perm_indices::element_type): Update accordingly.
25046 (vec_perm_indices::clamp): Handle polynomial element_types.
25047 * vec-perm-indices.c (vec_perm_indices::series_p): Likewise.
25048 (vec_perm_indices::all_in_range_p): Likewise.
25049 (tree_to_vec_perm_builder): Check for poly_int64 trees rather
25051 * vector-builder.h (vector_builder::stepped_sequence_p): Handle
25052 polynomial vec_perm_indices element types.
25053 * int-vector-builder.h (int_vector_builder::equal_p): Likewise.
25054 * fold-const.c (fold_vec_perm): Likewise.
25055 * optabs.c (shift_amt_for_vec_perm_mask): Likewise.
25056 * tree-vect-generic.c (lower_vec_perm): Likewise.
25057 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
25058 * config/aarch64/aarch64.c (aarch64_evpc_tbl): Cast d->perm
25059 element type to HOST_WIDE_INT.
25061 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25062 Alan Hayward <alan.hayward@arm.com>
25063 David Sherwood <david.sherwood@arm.com>
25065 * alias.c (addr_side_effect_eval): Take the size as a poly_int64
25066 rather than an int. Use plus_constant.
25067 (memrefs_conflict_p): Take the sizes as poly_int64s rather than ints.
25068 Take the offset "c" as a poly_int64 rather than a HOST_WIDE_INT.
25070 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25071 Alan Hayward <alan.hayward@arm.com>
25072 David Sherwood <david.sherwood@arm.com>
25074 * calls.c (emit_call_1, expand_call): Change struct_value_size from
25075 a HOST_WIDE_INT to a poly_int64.
25077 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25078 Alan Hayward <alan.hayward@arm.com>
25079 David Sherwood <david.sherwood@arm.com>
25081 * calls.c (load_register_parameters): Cope with polynomial
25082 mode sizes. Require a constant size for BLKmode parameters
25083 that aren't described by a PARALLEL. If BLOCK_REG_PADDING
25084 forces a parameter to be padded at the lsb end in order to
25085 fill a complete number of words, require the parameter size
25086 to be ordered wrt UNITS_PER_WORD.
25088 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25089 Alan Hayward <alan.hayward@arm.com>
25090 David Sherwood <david.sherwood@arm.com>
25092 * reload1.c (spill_stack_slot_width): Change element type
25093 from unsigned int to poly_uint64_pod.
25094 (alter_reg): Treat mode sizes as polynomial.
25096 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25097 Alan Hayward <alan.hayward@arm.com>
25098 David Sherwood <david.sherwood@arm.com>
25100 * reload.c (complex_word_subreg_p): New function.
25101 (reload_inner_reg_of_subreg, push_reload): Use it.
25103 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25104 Alan Hayward <alan.hayward@arm.com>
25105 David Sherwood <david.sherwood@arm.com>
25107 * lra-constraints.c (process_alt_operands): Reject matched
25108 operands whose sizes aren't ordered.
25109 (match_reload): Refer to this check here.
25111 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25112 Alan Hayward <alan.hayward@arm.com>
25113 David Sherwood <david.sherwood@arm.com>
25115 * builtins.c (expand_ifn_atomic_compare_exchange_into_call): Assert
25116 that the mode size is in the set {1, 2, 4, 8, 16}.
25118 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25119 Alan Hayward <alan.hayward@arm.com>
25120 David Sherwood <david.sherwood@arm.com>
25122 * var-tracking.c (adjust_mems): Treat mode sizes as polynomial.
25123 Use plus_constant instead of gen_rtx_PLUS.
25125 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25126 Alan Hayward <alan.hayward@arm.com>
25127 David Sherwood <david.sherwood@arm.com>
25129 * config/cr16/cr16-protos.h (cr16_push_rounding): Declare.
25130 * config/cr16/cr16.h (PUSH_ROUNDING): Move implementation to...
25131 * config/cr16/cr16.c (cr16_push_rounding): ...this new function.
25132 * config/h8300/h8300-protos.h (h8300_push_rounding): Declare.
25133 * config/h8300/h8300.h (PUSH_ROUNDING): Move implementation to...
25134 * config/h8300/h8300.c (h8300_push_rounding): ...this new function.
25135 * config/i386/i386-protos.h (ix86_push_rounding): Declare.
25136 * config/i386/i386.h (PUSH_ROUNDING): Move implementation to...
25137 * config/i386/i386.c (ix86_push_rounding): ...this new function.
25138 * config/m32c/m32c-protos.h (m32c_push_rounding): Take and return
25140 * config/m32c/m32c.c (m32c_push_rounding): Likewise.
25141 * config/m68k/m68k-protos.h (m68k_push_rounding): Declare.
25142 * config/m68k/m68k.h (PUSH_ROUNDING): Move implementation to...
25143 * config/m68k/m68k.c (m68k_push_rounding): ...this new function.
25144 * config/pdp11/pdp11-protos.h (pdp11_push_rounding): Declare.
25145 * config/pdp11/pdp11.h (PUSH_ROUNDING): Move implementation to...
25146 * config/pdp11/pdp11.c (pdp11_push_rounding): ...this new function.
25147 * config/stormy16/stormy16-protos.h (xstormy16_push_rounding): Declare.
25148 * config/stormy16/stormy16.h (PUSH_ROUNDING): Move implementation to...
25149 * config/stormy16/stormy16.c (xstormy16_push_rounding): ...this new
25151 * expr.c (emit_move_resolve_push): Treat the input and result
25152 of PUSH_ROUNDING as a poly_int64.
25153 (emit_move_complex_push, emit_single_push_insn_1): Likewise.
25154 (emit_push_insn): Likewise.
25155 * lra-eliminations.c (mark_not_eliminable): Likewise.
25156 * recog.c (push_operand): Likewise.
25157 * reload1.c (elimination_effects): Likewise.
25158 * rtlanal.c (nonzero_bits1): Likewise.
25159 * calls.c (store_one_arg): Likewise. Require the padding to be
25160 known at compile time.
25162 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25163 Alan Hayward <alan.hayward@arm.com>
25164 David Sherwood <david.sherwood@arm.com>
25166 * expr.c (emit_single_push_insn_1): Treat mode sizes as polynomial.
25167 Use plus_constant instead of gen_rtx_PLUS.
25169 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25170 Alan Hayward <alan.hayward@arm.com>
25171 David Sherwood <david.sherwood@arm.com>
25173 * auto-inc-dec.c (set_inc_state): Take the mode size as a poly_int64
25174 rather than an int.
25176 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25177 Alan Hayward <alan.hayward@arm.com>
25178 David Sherwood <david.sherwood@arm.com>
25180 * expr.c (expand_expr_real_1): Use tree_to_poly_uint64
25181 instead of int_size_in_bytes when handling VIEW_CONVERT_EXPRs
25182 via stack temporaries. Treat the mode size as polynomial too.
25184 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25185 Alan Hayward <alan.hayward@arm.com>
25186 David Sherwood <david.sherwood@arm.com>
25188 * expr.c (expand_expr_real_2): When handling conversions involving
25189 unions, apply tree_to_poly_uint64 to the TYPE_SIZE rather than
25190 multiplying int_size_in_bytes by BITS_PER_UNIT. Treat GET_MODE_BISIZE
25191 as a poly_uint64 too.
25193 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25194 Alan Hayward <alan.hayward@arm.com>
25195 David Sherwood <david.sherwood@arm.com>
25197 * rtlanal.c (subreg_get_info): Handle polynomial mode sizes.
25199 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25200 Alan Hayward <alan.hayward@arm.com>
25201 David Sherwood <david.sherwood@arm.com>
25203 * combine.c (can_change_dest_mode): Handle polynomial
25204 REGMODE_NATURAL_SIZE.
25205 * expmed.c (store_bit_field_1): Likewise.
25206 * expr.c (store_constructor): Likewise.
25207 * emit-rtl.c (validate_subreg): Operate on polynomial mode sizes
25208 and polynomial REGMODE_NATURAL_SIZE.
25209 (gen_lowpart_common): Likewise.
25210 * reginfo.c (record_subregs_of_mode): Likewise.
25211 * rtlanal.c (read_modify_subreg_p): Likewise.
25213 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25214 Alan Hayward <alan.hayward@arm.com>
25215 David Sherwood <david.sherwood@arm.com>
25217 * internal-fn.c (expand_vector_ubsan_overflow): Handle polynomial
25218 numbers of elements.
25220 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25221 Alan Hayward <alan.hayward@arm.com>
25222 David Sherwood <david.sherwood@arm.com>
25224 * match.pd: Cope with polynomial numbers of vector elements.
25226 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25227 Alan Hayward <alan.hayward@arm.com>
25228 David Sherwood <david.sherwood@arm.com>
25230 * fold-const.c (fold_indirect_ref_1): Handle polynomial offsets
25231 in a POINTER_PLUS_EXPR.
25233 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25234 Alan Hayward <alan.hayward@arm.com>
25235 David Sherwood <david.sherwood@arm.com>
25237 * omp-simd-clone.c (simd_clone_subparts): New function.
25238 (simd_clone_init_simd_arrays): Use it instead of TYPE_VECTOR_SUBPARTS.
25239 (ipa_simd_modify_function_body): Likewise.
25241 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25242 Alan Hayward <alan.hayward@arm.com>
25243 David Sherwood <david.sherwood@arm.com>
25245 * tree-vect-generic.c (nunits_for_known_piecewise_op): New function.
25246 (expand_vector_piecewise): Use it instead of TYPE_VECTOR_SUBPARTS.
25247 (expand_vector_addition, add_rshift, expand_vector_divmod): Likewise.
25248 (expand_vector_condition, vector_element): Likewise.
25249 (subparts_gt): New function.
25250 (get_compute_type): Use subparts_gt.
25251 (count_type_subparts): Delete.
25252 (expand_vector_operations_1): Use subparts_gt instead of
25253 count_type_subparts.
25255 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25256 Alan Hayward <alan.hayward@arm.com>
25257 David Sherwood <david.sherwood@arm.com>
25259 * tree-vect-data-refs.c (vect_no_alias_p): Replace with...
25260 (vect_compile_time_alias): ...this new function. Do the calculation
25261 on poly_ints rather than trees.
25262 (vect_prune_runtime_alias_test_list): Update call accordingly.
25264 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25265 Alan Hayward <alan.hayward@arm.com>
25266 David Sherwood <david.sherwood@arm.com>
25268 * tree-vect-slp.c (vect_build_slp_tree_1): Handle polynomial
25270 (vect_schedule_slp_instance): Likewise.
25272 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25273 Alan Hayward <alan.hayward@arm.com>
25274 David Sherwood <david.sherwood@arm.com>
25276 * tree-vect-slp.c (vect_get_and_check_slp_defs): Reject
25277 constant and extern definitions for variable-length vectors.
25278 (vect_get_constant_vectors): Note that the number of units
25279 is known to be constant.
25281 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25282 Alan Hayward <alan.hayward@arm.com>
25283 David Sherwood <david.sherwood@arm.com>
25285 * tree-vect-stmts.c (vectorizable_conversion): Treat the number
25286 of units as polynomial. Choose between WIDE and NARROW based
25289 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25290 Alan Hayward <alan.hayward@arm.com>
25291 David Sherwood <david.sherwood@arm.com>
25293 * tree-vect-stmts.c (simd_clone_subparts): New function.
25294 (vectorizable_simd_clone_call): Use it instead of TYPE_VECTOR_SUBPARTS.
25296 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25297 Alan Hayward <alan.hayward@arm.com>
25298 David Sherwood <david.sherwood@arm.com>
25300 * tree-vect-stmts.c (vectorizable_call): Treat the number of
25301 vectors as polynomial. Use build_index_vector for
25302 IFN_GOMP_SIMD_LANE.
25304 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25305 Alan Hayward <alan.hayward@arm.com>
25306 David Sherwood <david.sherwood@arm.com>
25308 * tree-vect-stmts.c (get_load_store_type): Treat the number of
25309 units as polynomial. Reject VMAT_ELEMENTWISE and VMAT_STRIDED_SLP
25310 for variable-length vectors.
25311 (vectorizable_mask_load_store): Treat the number of units as
25312 polynomial, asserting that it is constant if the condition has
25313 already been enforced.
25314 (vectorizable_store, vectorizable_load): Likewise.
25316 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25317 Alan Hayward <alan.hayward@arm.com>
25318 David Sherwood <david.sherwood@arm.com>
25320 * tree-vect-loop.c (vectorizable_live_operation): Treat the number
25321 of units as polynomial. Punt if we can't tell at compile time
25322 which vector contains the final result.
25324 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25325 Alan Hayward <alan.hayward@arm.com>
25326 David Sherwood <david.sherwood@arm.com>
25328 * tree-vect-loop.c (vectorizable_induction): Treat the number
25329 of units as polynomial. Punt on SLP inductions. Use an integer
25330 VEC_SERIES_EXPR for variable-length integer reductions. Use a
25331 cast of such a series for variable-length floating-point
25334 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25335 Alan Hayward <alan.hayward@arm.com>
25336 David Sherwood <david.sherwood@arm.com>
25338 * tree.h (build_index_vector): Declare.
25339 * tree.c (build_index_vector): New function.
25340 * tree-vect-loop.c (get_initial_defs_for_reduction): Treat the number
25341 of units as polynomial, forcibly converting it to a constant if
25342 vectorizable_reduction has already enforced the condition.
25343 (vect_create_epilog_for_reduction): Likewise. Use build_index_vector
25344 to create a {1,2,3,...} vector.
25345 (vectorizable_reduction): Treat the number of units as polynomial.
25346 Choose vectype_in based on the largest scalar element size rather
25347 than the smallest number of units. Enforce the restrictions
25350 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25351 Alan Hayward <alan.hayward@arm.com>
25352 David Sherwood <david.sherwood@arm.com>
25354 * tree-vect-data-refs.c (vector_alignment_reachable_p): Treat the
25355 number of units as polynomial.
25357 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25358 Alan Hayward <alan.hayward@arm.com>
25359 David Sherwood <david.sherwood@arm.com>
25361 * target.h (vector_sizes, auto_vector_sizes): New typedefs.
25362 * target.def (autovectorize_vector_sizes): Return the vector sizes
25363 by pointer, using vector_sizes rather than a bitmask.
25364 * targhooks.h (default_autovectorize_vector_sizes): Update accordingly.
25365 * targhooks.c (default_autovectorize_vector_sizes): Likewise.
25366 * config/aarch64/aarch64.c (aarch64_autovectorize_vector_sizes):
25368 * config/arc/arc.c (arc_autovectorize_vector_sizes): Likewise.
25369 * config/arm/arm.c (arm_autovectorize_vector_sizes): Likewise.
25370 * config/i386/i386.c (ix86_autovectorize_vector_sizes): Likewise.
25371 * config/mips/mips.c (mips_autovectorize_vector_sizes): Likewise.
25372 * omp-general.c (omp_max_vf): Likewise.
25373 * omp-low.c (omp_clause_aligned_alignment): Likewise.
25374 * optabs-query.c (can_vec_mask_load_store_p): Likewise.
25375 * tree-vect-loop.c (vect_analyze_loop): Likewise.
25376 * tree-vect-slp.c (vect_slp_bb): Likewise.
25377 * doc/tm.texi: Regenerate.
25378 * tree-vectorizer.h (current_vector_size): Change from an unsigned int
25380 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Take
25381 the vector size as a poly_uint64 rather than an unsigned int.
25382 (current_vector_size): Change from an unsigned int to a poly_uint64.
25383 (get_vectype_for_scalar_type): Update accordingly.
25384 * tree.h (build_truth_vector_type): Take the size and number of
25385 units as a poly_uint64 rather than an unsigned int.
25386 (build_vector_type): Add a temporary overload that takes
25387 the number of units as a poly_uint64 rather than an unsigned int.
25388 * tree.c (make_vector_type): Likewise.
25389 (build_truth_vector_type): Take the number of units as a poly_uint64
25390 rather than an unsigned int.
25392 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25393 Alan Hayward <alan.hayward@arm.com>
25394 David Sherwood <david.sherwood@arm.com>
25396 * target.def (get_mask_mode): Take the number of units and length
25397 as poly_uint64s rather than unsigned ints.
25398 * targhooks.h (default_get_mask_mode): Update accordingly.
25399 * targhooks.c (default_get_mask_mode): Likewise.
25400 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
25401 * doc/tm.texi: Regenerate.
25403 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25404 Alan Hayward <alan.hayward@arm.com>
25405 David Sherwood <david.sherwood@arm.com>
25407 * omp-general.h (omp_max_vf): Return a poly_uint64 instead of an int.
25408 * omp-general.c (omp_max_vf): Likewise.
25409 * omp-expand.c (omp_adjust_chunk_size): Update call to omp_max_vf.
25410 (expand_omp_simd): Handle polynomial safelen.
25411 * omp-low.c (omplow_simd_context): Add a default constructor.
25412 (omplow_simd_context::max_vf): Change from int to poly_uint64.
25413 (lower_rec_simd_input_clauses): Update accordingly.
25414 (lower_rec_input_clauses): Likewise.
25416 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25417 Alan Hayward <alan.hayward@arm.com>
25418 David Sherwood <david.sherwood@arm.com>
25420 * tree-vectorizer.h (vect_nunits_for_cost): New function.
25421 * tree-vect-loop.c (vect_model_reduction_cost): Use it.
25422 * tree-vect-slp.c (vect_analyze_slp_cost_1): Likewise.
25423 (vect_analyze_slp_cost): Likewise.
25424 * tree-vect-stmts.c (vect_model_store_cost): Likewise.
25425 (vect_model_load_cost): Likewise.
25427 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25428 Alan Hayward <alan.hayward@arm.com>
25429 David Sherwood <david.sherwood@arm.com>
25431 * tree-vect-slp.c (vect_record_max_nunits, vect_build_slp_tree_1)
25432 (vect_build_slp_tree_2, vect_build_slp_tree): Change max_nunits
25433 from an unsigned int * to a poly_uint64_pod *.
25434 (calculate_unrolling_factor): New function.
25435 (vect_analyze_slp_instance): Use it. Track polynomial max_nunits.
25437 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25438 Alan Hayward <alan.hayward@arm.com>
25439 David Sherwood <david.sherwood@arm.com>
25441 * tree-vectorizer.h (_slp_instance::unrolling_factor): Change
25442 from an unsigned int to a poly_uint64.
25443 (_loop_vec_info::slp_unrolling_factor): Likewise.
25444 (_loop_vec_info::vectorization_factor): Change from an int
25446 (MAX_VECTORIZATION_FACTOR): Bump from 64 to INT_MAX.
25447 (vect_get_num_vectors): New function.
25448 (vect_update_max_nunits, vect_vf_for_cost): Likewise.
25449 (vect_get_num_copies): Use vect_get_num_vectors.
25450 (vect_analyze_data_ref_dependences): Change max_vf from an int *
25451 to an unsigned int *.
25452 (vect_analyze_data_refs): Change min_vf from an int * to a
25454 (vect_transform_slp_perm_load): Take the vf as a poly_uint64 rather
25455 than an unsigned HOST_WIDE_INT.
25456 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr)
25457 (vect_analyze_data_ref_dependence): Change max_vf from an int *
25458 to an unsigned int *.
25459 (vect_analyze_data_ref_dependences): Likewise.
25460 (vect_compute_data_ref_alignment): Handle polynomial vf.
25461 (vect_enhance_data_refs_alignment): Likewise.
25462 (vect_prune_runtime_alias_test_list): Likewise.
25463 (vect_shift_permute_load_chain): Likewise.
25464 (vect_supportable_dr_alignment): Likewise.
25465 (dependence_distance_ge_vf): Take the vectorization factor as a
25466 poly_uint64 rather than an unsigned HOST_WIDE_INT.
25467 (vect_analyze_data_refs): Change min_vf from an int * to a
25469 * tree-vect-loop-manip.c (vect_gen_scalar_loop_niters): Take
25470 vfm1 as a poly_uint64 rather than an int. Make the same change
25471 for the returned bound_scalar.
25472 (vect_gen_vector_loop_niters): Handle polynomial vf.
25473 (vect_do_peeling): Likewise. Update call to
25474 vect_gen_scalar_loop_niters and handle polynomial bound_scalars.
25475 (vect_gen_vector_loop_niters_mult_vf): Assert that the vf must
25477 * tree-vect-loop.c (vect_determine_vectorization_factor)
25478 (vect_update_vf_for_slp, vect_analyze_loop_2): Handle polynomial vf.
25479 (vect_get_known_peeling_cost): Likewise.
25480 (vect_estimate_min_profitable_iters, vectorizable_reduction): Likewise.
25481 (vect_worthwhile_without_simd_p, vectorizable_induction): Likewise.
25482 (vect_transform_loop): Likewise. Use the lowest possible VF when
25483 updating the upper bounds of the loop.
25484 (vect_min_worthwhile_factor): Make static. Return an unsigned int
25485 rather than an int.
25486 * tree-vect-slp.c (vect_attempt_slp_rearrange_stmts): Cope with
25487 polynomial unroll factors.
25488 (vect_analyze_slp_cost_1, vect_analyze_slp_instance): Likewise.
25489 (vect_make_slp_decision): Likewise.
25490 (vect_supported_load_permutation_p): Likewise, and polynomial
25492 (vect_analyze_slp_cost): Handle polynomial vf.
25493 (vect_slp_analyze_node_operations): Likewise.
25494 (vect_slp_analyze_bb_1): Likewise.
25495 (vect_transform_slp_perm_load): Take the vf as a poly_uint64 rather
25496 than an unsigned HOST_WIDE_INT.
25497 * tree-vect-stmts.c (vectorizable_simd_clone_call, vectorizable_store)
25498 (vectorizable_load): Handle polynomial vf.
25499 * tree-vectorizer.c (simduid_to_vf::vf): Change from an int to
25501 (adjust_simduid_builtins, shrink_simd_arrays): Update accordingly.
25503 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25504 Alan Hayward <alan.hayward@arm.com>
25505 David Sherwood <david.sherwood@arm.com>
25507 * match.pd: Handle bit operations involving three constants
25508 and try to fold one pair.
25510 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25512 * tree-vect-loop-manip.c: Include gimple-fold.h.
25513 (slpeel_make_loop_iterate_ntimes): Add step, final_iv and
25514 niters_maybe_zero parameters. Handle other cases besides a step of 1.
25515 (vect_gen_vector_loop_niters): Add a step_vector_ptr parameter.
25516 Add a path that uses a step of VF instead of 1, but disable it
25518 (vect_do_peeling): Add step_vector, niters_vector_mult_vf_var
25519 and niters_no_overflow parameters. Update calls to
25520 slpeel_make_loop_iterate_ntimes and vect_gen_vector_loop_niters.
25521 Create a new SSA name if the latter choses to use a ste other
25522 than zero, and return it via niters_vector_mult_vf_var.
25523 * tree-vect-loop.c (vect_transform_loop): Update calls to
25524 vect_do_peeling, vect_gen_vector_loop_niters and
25525 slpeel_make_loop_iterate_ntimes.
25526 * tree-vectorizer.h (slpeel_make_loop_iterate_ntimes, vect_do_peeling)
25527 (vect_gen_vector_loop_niters): Update declarations after above changes.
25529 2018-01-02 Michael Meissner <meissner@linux.vnet.ibm.com>
25531 * config/rs6000/rs6000.md (floor<mode>2): Add support for IEEE
25532 128-bit round to integer instructions.
25533 (ceil<mode>2): Likewise.
25534 (btrunc<mode>2): Likewise.
25535 (round<mode>2): Likewise.
25537 2018-01-02 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25539 * config/rs6000/rs6000-string.c (expand_block_move): Allow the use of
25540 unaligned VSX load/store on P8/P9.
25541 (expand_block_clear): Allow the use of unaligned VSX
25542 load/store on P8/P9.
25544 2018-01-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25546 * config/rs6000/rs6000-p8swap.c (swap_feeds_both_load_and_store):
25548 (rs6000_analyze_swaps): Mark a web unoptimizable if it contains a
25549 swap associated with both a load and a store.
25551 2018-01-02 Andrew Waterman <andrew@sifive.com>
25553 * config/riscv/linux.h (ICACHE_FLUSH_FUNC): New.
25554 * config/riscv/riscv.md (clear_cache): Use it.
25556 2018-01-02 Artyom Skrobov <tyomitch@gmail.com>
25558 * web.c: Remove out-of-date comment.
25560 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
25562 * expr.c (fixup_args_size_notes): Check that any existing
25563 REG_ARGS_SIZE notes are correct, and don't try to re-add them.
25564 (emit_single_push_insn_1): Move stack_pointer_delta adjustment to...
25565 (emit_single_push_insn): ...here.
25567 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
25569 * rtl.h (CONST_VECTOR_ELT): Redefine to const_vector_elt.
25570 (const_vector_encoded_nelts): New function.
25571 (CONST_VECTOR_NUNITS): Redefine to use GET_MODE_NUNITS.
25572 (const_vector_int_elt, const_vector_elt): Declare.
25573 * emit-rtl.c (const_vector_int_elt_1): New function.
25574 (const_vector_elt): Likewise.
25575 * simplify-rtx.c (simplify_immed_subreg): Avoid taking the address
25576 of CONST_VECTOR_ELT.
25578 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
25580 * expr.c: Include rtx-vector-builder.h.
25581 (const_vector_mask_from_tree): Use rtx_vector_builder and operate
25582 directly on the tree encoding.
25583 (const_vector_from_tree): Likewise.
25584 * optabs.c: Include rtx-vector-builder.h.
25585 (expand_vec_perm_var): Use rtx_vector_builder and create a repeating
25586 sequence of "u" values.
25587 * vec-perm-indices.c: Include rtx-vector-builder.h.
25588 (vec_perm_indices_to_rtx): Use rtx_vector_builder and operate
25589 directly on the vec_perm_indices encoding.
25591 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
25593 * doc/rtl.texi (const_vector): Describe new encoding scheme.
25594 * Makefile.in (OBJS): Add rtx-vector-builder.o.
25595 * rtx-vector-builder.h: New file.
25596 * rtx-vector-builder.c: Likewise.
25597 * rtl.h (rtx_def::u2): Add a const_vector field.
25598 (CONST_VECTOR_NPATTERNS): New macro.
25599 (CONST_VECTOR_NELTS_PER_PATTERN): Likewise.
25600 (CONST_VECTOR_DUPLICATE_P): Likewise.
25601 (CONST_VECTOR_STEPPED_P): Likewise.
25602 (CONST_VECTOR_ENCODED_ELT): Likewise.
25603 (const_vec_duplicate_p): Check for a duplicated vector encoding.
25604 (unwrap_const_vec_duplicate): Likewise.
25605 (const_vec_series_p): Check for a non-duplicated vector encoding.
25606 Say that the function only returns true for integer vectors.
25607 * emit-rtl.c: Include rtx-vector-builder.h.
25608 (gen_const_vec_duplicate_1): Delete.
25609 (gen_const_vector): Call gen_const_vec_duplicate instead of
25610 gen_const_vec_duplicate_1.
25611 (const_vec_series_p_1): Operate directly on the CONST_VECTOR encoding.
25612 (gen_const_vec_duplicate): Use rtx_vector_builder.
25613 (gen_const_vec_series): Likewise.
25614 (gen_rtx_CONST_VECTOR): Likewise.
25615 * config/powerpcspe/powerpcspe.c: Include rtx-vector-builder.h.
25616 (swap_const_vector_halves): Take an rtx pointer rather than rtx.
25617 Build a new vector rather than modifying a CONST_VECTOR in-place.
25618 (handle_special_swappables): Update call accordingly.
25619 * config/rs6000/rs6000-p8swap.c: Include rtx-vector-builder.h.
25620 (swap_const_vector_halves): Take an rtx pointer rather than rtx.
25621 Build a new vector rather than modifying a CONST_VECTOR in-place.
25622 (handle_special_swappables): Update call accordingly.
25624 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
25626 * simplify-rtx.c (simplify_const_binary_operation): Use
25627 CONST_VECTOR_ELT instead of XVECEXP.
25629 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
25631 * tree-cfg.c (verify_gimple_assign_ternary): Allow the size of
25632 the selector elements to be different from the data elements
25633 if the selector is a VECTOR_CST.
25634 * tree-vect-stmts.c (vect_gen_perm_mask_any): Use a vector of
25635 ssizetype for the selector.
25637 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
25639 * optabs.c (shift_amt_for_vec_perm_mask): Try using series_p
25640 before testing each element individually.
25641 * tree-vect-generic.c (lower_vec_perm): Likewise.
25643 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
25645 * selftest.h (selftest::vec_perm_indices_c_tests): Declare.
25646 * selftest-run-tests.c (selftest::run_tests): Call it.
25647 * vector-builder.h (vector_builder::operator ==): New function.
25648 (vector_builder::operator !=): Likewise.
25649 * vec-perm-indices.h (vec_perm_indices::series_p): Declare.
25650 (vec_perm_indices::all_from_input_p): New function.
25651 * vec-perm-indices.c (vec_perm_indices::series_p): Likewise.
25652 (test_vec_perm_12, selftest::vec_perm_indices_c_tests): Likewise.
25653 * fold-const.c (fold_ternary_loc): Use tree_to_vec_perm_builder
25654 instead of reading the VECTOR_CST directly. Detect whether both
25655 vector inputs are the same before constructing the vec_perm_indices,
25656 and update the number of inputs argument accordingly. Use the
25657 utility functions added above. Only construct sel2 if we need to.
25659 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
25661 * optabs.c (expand_vec_perm_var): Use an explicit encoding for
25662 the broadcast of the low byte.
25663 (expand_mult_highpart): Use an explicit encoding for the permutes.
25664 * optabs-query.c (can_mult_highpart_p): Likewise.
25665 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
25666 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
25667 (vectorizable_bswap): Likewise.
25668 * tree-vect-data-refs.c (vect_grouped_store_supported): Use an
25669 explicit encoding for the power-of-2 permutes.
25670 (vect_permute_store_chain): Likewise.
25671 (vect_grouped_load_supported): Likewise.
25672 (vect_permute_load_chain): Likewise.
25674 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
25676 * vec-perm-indices.h (vec_perm_indices_to_tree): Declare.
25677 * vec-perm-indices.c (vec_perm_indices_to_tree): New function.
25678 * tree-ssa-forwprop.c (simplify_vector_constructor): Use it.
25679 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
25680 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
25681 (vect_gen_perm_mask_any): Likewise.
25683 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
25685 * int-vector-builder.h: New file.
25686 * vec-perm-indices.h: Include int-vector-builder.h.
25687 (vec_perm_indices): Redefine as an int_vector_builder.
25688 (auto_vec_perm_indices): Delete.
25689 (vec_perm_builder): Redefine as a stand-alone class.
25690 (vec_perm_indices::vec_perm_indices): New function.
25691 (vec_perm_indices::clamp): Likewise.
25692 * vec-perm-indices.c: Include fold-const.h and tree-vector-builder.h.
25693 (vec_perm_indices::new_vector): New function.
25694 (vec_perm_indices::new_expanded_vector): Update for new
25695 vec_perm_indices class.
25696 (vec_perm_indices::rotate_inputs): New function.
25697 (vec_perm_indices::all_in_range_p): Operate directly on the
25698 encoded form, without computing elided elements.
25699 (tree_to_vec_perm_builder): Operate directly on the VECTOR_CST
25700 encoding. Update for new vec_perm_indices class.
25701 * optabs.c (expand_vec_perm_const): Create a vec_perm_indices for
25702 the given vec_perm_builder.
25703 (expand_vec_perm_var): Update vec_perm_builder constructor.
25704 (expand_mult_highpart): Use vec_perm_builder instead of
25705 auto_vec_perm_indices.
25706 * optabs-query.c (can_mult_highpart_p): Use vec_perm_builder and
25707 vec_perm_indices instead of auto_vec_perm_indices. Use a single
25708 or double series encoding as appropriate.
25709 * fold-const.c (fold_ternary_loc): Use vec_perm_builder and
25710 vec_perm_indices instead of auto_vec_perm_indices.
25711 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
25712 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
25713 (vect_permute_store_chain): Likewise.
25714 (vect_grouped_load_supported): Likewise.
25715 (vect_permute_load_chain): Likewise.
25716 (vect_shift_permute_load_chain): Likewise.
25717 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
25718 (vect_transform_slp_perm_load): Likewise.
25719 (vect_schedule_slp_instance): Likewise.
25720 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
25721 (vectorizable_mask_load_store): Likewise.
25722 (vectorizable_bswap): Likewise.
25723 (vectorizable_store): Likewise.
25724 (vectorizable_load): Likewise.
25725 * tree-vect-generic.c (lower_vec_perm): Use vec_perm_builder and
25726 vec_perm_indices instead of auto_vec_perm_indices. Use
25727 tree_to_vec_perm_builder to read the vector from a tree.
25728 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Take a
25729 vec_perm_builder instead of a vec_perm_indices.
25730 (have_whole_vector_shift): Use vec_perm_builder and
25731 vec_perm_indices instead of auto_vec_perm_indices. Leave the
25732 truncation to calc_vec_perm_mask_for_shift.
25733 (vect_create_epilog_for_reduction): Likewise.
25734 * config/aarch64/aarch64.c (expand_vec_perm_d::perm): Change
25735 from auto_vec_perm_indices to vec_perm_indices.
25736 (aarch64_expand_vec_perm_const_1): Use rotate_inputs on d.perm
25737 instead of changing individual elements.
25738 (aarch64_vectorize_vec_perm_const): Use new_vector to install
25739 the vector in d.perm.
25740 * config/arm/arm.c (expand_vec_perm_d::perm): Change
25741 from auto_vec_perm_indices to vec_perm_indices.
25742 (arm_expand_vec_perm_const_1): Use rotate_inputs on d.perm
25743 instead of changing individual elements.
25744 (arm_vectorize_vec_perm_const): Use new_vector to install
25745 the vector in d.perm.
25746 * config/powerpcspe/powerpcspe.c (rs6000_expand_extract_even):
25747 Update vec_perm_builder constructor.
25748 (rs6000_expand_interleave): Likewise.
25749 * config/rs6000/rs6000.c (rs6000_expand_extract_even): Likewise.
25750 (rs6000_expand_interleave): Likewise.
25752 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
25754 * optabs-query.c (can_vec_perm_var_p): Check whether lowering
25755 to qimode could truncate the indices.
25756 * optabs.c (expand_vec_perm_var): Likewise.
25758 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
25760 * Makefile.in (OBJS): Add vec-perm-indices.o.
25761 * vec-perm-indices.h: New file.
25762 * vec-perm-indices.c: Likewise.
25763 * target.h (vec_perm_indices): Replace with a forward class
25765 (auto_vec_perm_indices): Move to vec-perm-indices.h.
25766 * optabs.h: Include vec-perm-indices.h.
25767 (expand_vec_perm): Delete.
25768 (selector_fits_mode_p, expand_vec_perm_var): Declare.
25769 (expand_vec_perm_const): Declare.
25770 * target.def (vec_perm_const_ok): Replace with...
25771 (vec_perm_const): ...this new hook.
25772 * doc/tm.texi.in (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Replace with...
25773 (TARGET_VECTORIZE_VEC_PERM_CONST): ...this new hook.
25774 * doc/tm.texi: Regenerate.
25775 * optabs.def (vec_perm_const): Delete.
25776 * doc/md.texi (vec_perm_const): Likewise.
25777 (vec_perm): Refer to TARGET_VECTORIZE_VEC_PERM_CONST.
25778 * expr.c (expand_expr_real_2): Use expand_vec_perm_const rather than
25779 expand_vec_perm for constant permutation vectors. Assert that
25780 the mode of variable permutation vectors is the integer equivalent
25781 of the mode that is being permuted.
25782 * optabs-query.h (selector_fits_mode_p): Declare.
25783 * optabs-query.c: Include vec-perm-indices.h.
25784 (selector_fits_mode_p): New function.
25785 (can_vec_perm_const_p): Check whether targetm.vectorize.vec_perm_const
25786 is defined, instead of checking whether the vec_perm_const_optab
25787 exists. Use targetm.vectorize.vec_perm_const instead of
25788 targetm.vectorize.vec_perm_const_ok. Check whether the indices
25789 fit in the vector mode before using a variable permute.
25790 * optabs.c (shift_amt_for_vec_perm_mask): Take a mode and a
25791 vec_perm_indices instead of an rtx.
25792 (expand_vec_perm): Replace with...
25793 (expand_vec_perm_const): ...this new function. Take the selector
25794 as a vec_perm_indices rather than an rtx. Also take the mode of
25795 the selector. Update call to shift_amt_for_vec_perm_mask.
25796 Use targetm.vectorize.vec_perm_const instead of vec_perm_const_optab.
25797 Use vec_perm_indices::new_expanded_vector to expand the original
25798 selector into bytes. Check whether the indices fit in the vector
25799 mode before using a variable permute.
25800 (expand_vec_perm_var): Make global.
25801 (expand_mult_highpart): Use expand_vec_perm_const.
25802 * fold-const.c: Includes vec-perm-indices.h.
25803 * tree-ssa-forwprop.c: Likewise.
25804 * tree-vect-data-refs.c: Likewise.
25805 * tree-vect-generic.c: Likewise.
25806 * tree-vect-loop.c: Likewise.
25807 * tree-vect-slp.c: Likewise.
25808 * tree-vect-stmts.c: Likewise.
25809 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm_const):
25811 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>): Delete.
25812 * config/aarch64/aarch64.c (aarch64_expand_vec_perm_const)
25813 (aarch64_vectorize_vec_perm_const_ok): Fuse into...
25814 (aarch64_vectorize_vec_perm_const): ...this new function.
25815 (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
25816 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
25817 * config/arm/arm-protos.h (arm_expand_vec_perm_const): Delete.
25818 * config/arm/vec-common.md (vec_perm_const<mode>): Delete.
25819 * config/arm/arm.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
25820 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
25821 (arm_expand_vec_perm_const, arm_vectorize_vec_perm_const_ok): Merge
25823 (arm_vectorize_vec_perm_const): ...this new function. Explicitly
25824 check for NEON modes.
25825 * config/i386/i386-protos.h (ix86_expand_vec_perm_const): Delete.
25826 * config/i386/sse.md (VEC_PERM_CONST, vec_perm_const<mode>): Delete.
25827 * config/i386/i386.c (ix86_expand_vec_perm_const_1): Update comment.
25828 (ix86_expand_vec_perm_const, ix86_vectorize_vec_perm_const_ok): Merge
25830 (ix86_vectorize_vec_perm_const): ...this new function. Incorporate
25831 the old VEC_PERM_CONST conditions.
25832 * config/ia64/ia64-protos.h (ia64_expand_vec_perm_const): Delete.
25833 * config/ia64/vect.md (vec_perm_const<mode>): Delete.
25834 * config/ia64/ia64.c (ia64_expand_vec_perm_const)
25835 (ia64_vectorize_vec_perm_const_ok): Merge into...
25836 (ia64_vectorize_vec_perm_const): ...this new function.
25837 * config/mips/loongson.md (vec_perm_const<mode>): Delete.
25838 * config/mips/mips-msa.md (vec_perm_const<mode>): Delete.
25839 * config/mips/mips-ps-3d.md (vec_perm_constv2sf): Delete.
25840 * config/mips/mips-protos.h (mips_expand_vec_perm_const): Delete.
25841 * config/mips/mips.c (mips_expand_vec_perm_const)
25842 (mips_vectorize_vec_perm_const_ok): Merge into...
25843 (mips_vectorize_vec_perm_const): ...this new function.
25844 * config/powerpcspe/altivec.md (vec_perm_constv16qi): Delete.
25845 * config/powerpcspe/paired.md (vec_perm_constv2sf): Delete.
25846 * config/powerpcspe/spe.md (vec_perm_constv2si): Delete.
25847 * config/powerpcspe/vsx.md (vec_perm_const<mode>): Delete.
25848 * config/powerpcspe/powerpcspe-protos.h (altivec_expand_vec_perm_const)
25849 (rs6000_expand_vec_perm_const): Delete.
25850 * config/powerpcspe/powerpcspe.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK):
25852 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
25853 (altivec_expand_vec_perm_const_le): Take each operand individually.
25854 Operate on constant selectors rather than rtxes.
25855 (altivec_expand_vec_perm_const): Likewise. Update call to
25856 altivec_expand_vec_perm_const_le.
25857 (rs6000_expand_vec_perm_const): Delete.
25858 (rs6000_vectorize_vec_perm_const_ok): Delete.
25859 (rs6000_vectorize_vec_perm_const): New function.
25860 (rs6000_do_expand_vec_perm): Take a vec_perm_builder instead of
25861 an element count and rtx array.
25862 (rs6000_expand_extract_even): Update call accordingly.
25863 (rs6000_expand_interleave): Likewise.
25864 * config/rs6000/altivec.md (vec_perm_constv16qi): Delete.
25865 * config/rs6000/paired.md (vec_perm_constv2sf): Delete.
25866 * config/rs6000/vsx.md (vec_perm_const<mode>): Delete.
25867 * config/rs6000/rs6000-protos.h (altivec_expand_vec_perm_const)
25868 (rs6000_expand_vec_perm_const): Delete.
25869 * config/rs6000/rs6000.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
25870 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
25871 (altivec_expand_vec_perm_const_le): Take each operand individually.
25872 Operate on constant selectors rather than rtxes.
25873 (altivec_expand_vec_perm_const): Likewise. Update call to
25874 altivec_expand_vec_perm_const_le.
25875 (rs6000_expand_vec_perm_const): Delete.
25876 (rs6000_vectorize_vec_perm_const_ok): Delete.
25877 (rs6000_vectorize_vec_perm_const): New function. Remove stray
25878 reference to the SPE evmerge intructions.
25879 (rs6000_do_expand_vec_perm): Take a vec_perm_builder instead of
25880 an element count and rtx array.
25881 (rs6000_expand_extract_even): Update call accordingly.
25882 (rs6000_expand_interleave): Likewise.
25883 * config/sparc/sparc.md (vec_perm_constv8qi): Delete in favor of...
25884 * config/sparc/sparc.c (sparc_vectorize_vec_perm_const): ...this
25886 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
25888 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
25890 * optabs.c (expand_vec_perm_1): Assert that SEL has an integer
25891 vector mode and that that mode matches the mode of the data
25893 (expand_vec_perm): Split handling of non-CONST_VECTOR selectors
25894 out into expand_vec_perm_var. Do all CONST_VECTOR handling here,
25895 directly using expand_vec_perm_1 when forcing selectors into
25897 (expand_vec_perm_var): New function, split out from expand_vec_perm.
25899 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
25901 * optabs-query.h (can_vec_perm_p): Delete.
25902 (can_vec_perm_var_p, can_vec_perm_const_p): Declare.
25903 * optabs-query.c (can_vec_perm_p): Split into...
25904 (can_vec_perm_var_p, can_vec_perm_const_p): ...these two functions.
25905 (can_mult_highpart_p): Use can_vec_perm_const_p to test whether a
25906 particular selector is valid.
25907 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
25908 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
25909 (vect_grouped_load_supported): Likewise.
25910 (vect_shift_permute_load_chain): Likewise.
25911 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
25912 (vect_transform_slp_perm_load): Likewise.
25913 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
25914 (vectorizable_bswap): Likewise.
25915 (vect_gen_perm_mask_checked): Likewise.
25916 * fold-const.c (fold_ternary_loc): Likewise. Don't take
25917 implementations of variable permutation vectors into account
25918 when deciding which selector to use.
25919 * tree-vect-loop.c (have_whole_vector_shift): Don't check whether
25920 vec_perm_const_optab is supported; instead use can_vec_perm_const_p
25921 with a false third argument.
25922 * tree-vect-generic.c (lower_vec_perm): Use can_vec_perm_const_p
25923 to test whether the constant selector is valid and can_vec_perm_var_p
25924 to test whether a variable selector is valid.
25926 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
25928 * optabs-query.h (can_vec_perm_p): Take a const vec_perm_indices *.
25929 * optabs-query.c (can_vec_perm_p): Likewise.
25930 * fold-const.c (fold_vec_perm): Take a const vec_perm_indices &
25931 instead of vec_perm_indices.
25932 * tree-vectorizer.h (vect_gen_perm_mask_any): Likewise,
25933 (vect_gen_perm_mask_checked): Likewise,
25934 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise,
25935 (vect_gen_perm_mask_checked): Likewise,
25937 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
25939 * optabs-query.h (qimode_for_vec_perm): Declare.
25940 * optabs-query.c (can_vec_perm_p): Split out qimode search to...
25941 (qimode_for_vec_perm): ...this new function.
25942 * optabs.c (expand_vec_perm): Use qimode_for_vec_perm.
25944 2018-01-02 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25946 * rtlanal.c (canonicalize_condition): Return 0 if final rtx
25947 does not have a conditional at the top.
25949 2018-01-02 Richard Biener <rguenther@suse.de>
25951 * ipa-inline.c (big_speedup_p): Fix expression.
25953 2018-01-02 Jan Hubicka <hubicka@ucw.cz>
25956 * config/i386/x86-tune-costs.h: Increase cost of integer load costs
25959 2018-01-02 Jan Hubicka <hubicka@ucw.cz>
25963 * x86-tune-costs.h (generic_cost): Reduce cost of FDIV 20->17,
25964 cost of sqrt 20->14, DIVSS 18->13, DIVSD 32->17, SQRtSS 30->14
25965 and SQRTsD 58->18, cond_not_taken_branch_cost. 2->1. Increase
25966 cond_taken_branch_cost 3->4.
25968 2018-01-01 Jakub Jelinek <jakub@redhat.com>
25970 PR tree-optimization/83581
25971 * tree-loop-distribution.c (pass_loop_distribution::execute): Return
25972 TODO_cleanup_cfg if any changes have been made.
25974 PR middle-end/83608
25975 * expr.c (store_expr_with_bounds): Use simplify_gen_subreg instead of
25976 convert_modes if target mode has the right side, but different mode
25979 PR middle-end/83609
25980 * expr.c (expand_assignment): Fix up a typo in simplify_gen_subreg
25981 last argument when extracting from CONCAT. If either from_real or
25982 from_imag is NULL, use expansion through memory. If result is not
25983 a CONCAT and simplify_gen_subreg fails, try to simplify_gen_subreg
25984 the parts directly to inner mode, if even that fails, use expansion
25987 PR middle-end/83623
25988 * expmed.c (expand_shift_1): For 2-byte rotates by BITS_PER_UNIT,
25989 check for bswap in mode rather than HImode and use that in expand_unop
25992 Copyright (C) 2018 Free Software Foundation, Inc.
25994 Copying and distribution of this file, with or without modification,
25995 are permitted in any medium without royalty provided the copyright
25996 notice and this notice are preserved.